Structure sensitivity of NO reduction over iridium catalysts in HC-SCR

Article abstract of DOI:10.1006/jcat.2001.3437

Ir black with different crystallite sizes and Ir-H-ZSM-5 prepared using two different precursors were investigated for their behavior in selective catalytic reduction using hydrocarbons as reducing agents (HC-SCR). Emphasis was given to the study of the influence of time onstream on N₂ yield, the change in Ir crystallite size, and the change in the ratio of Ir:IrO₂. Under reaction conditions at 450°C Ir-H-ZSM-5 did not reach steady-state catalytic behavior within 32 h. In contrast, unsupported Ir black showed higher initial yields of nitrogen and approached steady-state considerably faster. Ir black with the largest crystallite size (45 nm) required the shortest time for reaching steady-state behavior. The influence of crystallite size on the reaction of Ir with O₂ and NO was addressed and related to the increase in N₂ yields with increasing Ir crystallite size in the reduction of NO using propene as a reducing agent. Comparative pulse thermoanalysis studies of NO and O₂ adsorption on Ir black with different crystallite sizes (5-45 nm) revealed that the relative uptake of NO (m_NO/m_O2) increases strongly with increasing crystallite size. This behavior is due to a strong structure sensitivity of NO adsorption, whereas O₂ adsorption is relatively insensitive to crystallite size. The improved yield of N₂ with increasing crystallite size under HC-SCR conditions is traced to the higher surface concentration of NO relative to O₂ with increasing crystallite size.

Full text of DOI:10.1006/jcat.2001.3437

Journal of Catalysis 205, 157–167 (2002)
doi:10.1006/jcat.2001.3437, available online at http://www.idealibrary.com on
Structure Sensitivity of NO Reduction over Iridium Catalysts
in HC–SCR
C. Wo¨ gerbauer, M. Maciejewski, and A. Baiker1
Laboratory of Technical Chemistry, Swiss Federal Institute of Technology, ETH–Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland
Received June 25, 2001; revised October 3, 2001; accepted October 4, 2001
logical need to prevent or at least reduce the emission of
NOx .
With the invention of the three-way catalyst there is a
Ir black with different crystallite sizes and Ir–H–ZSM-5 prepared
using two different precursors were investigated for their behav-
ior in selective catalytic reduction using hydrocarbons as reducing
agents (HC–SCR). Emphasis was given to the study of the influ-
ence of time onstream on N2 yield, the change in Ir crystallite size,
and the change in the ratio of Ir : IrO2. Under reaction conditions
at 450 C Ir–H–ZSM-5 did not reach steady-state catalytic behav-
ior within 32 h. In contrast, unsupported Ir black showed higher
initial yields of nitrogen and approached steady-state considerably
faster. Ir black with the largest crystallite size (45 nm) required the
shortest time for reaching steady-state behavior. The influence of
crystallite size on the reaction of Ir with O2 and NO was addressed
and related to the increase in N2 yields with increasing Ir crystal-
lite size in the reduction of NO using propene as a reducing agent.
Comparative pulse thermoanalysis studies of NO and O2 adsorp-
tion on Ir black with different crystallite sizes (5–45 nm) revealed
that the relative uptake of NO (mNO/mO2) increases strongly with
increasing crystallite size. This behavior is due to a strong structure
sensitivity of NO adsorption, whereas O2 adsorption is relatively in-
sensitive to crystallite size. The improved yield of N2 with increasing
crystallite size under HC–SCR conditions is traced to the higher sur-
face concentration of NO relative to O2 with increasing crystallite
c
very powerful and in fact successful tool at hand for the
reduction of NOx produced by engines which work under
stoichiometric air to fuel conditions. On the other hand
the development of a novel system for catalytic exhaust-
gas after-treatment capable of reducing NOx under lean
gas conditions still remains a cardinal problem for those
involved in automotive pollution control. Especially under
the pressure of ever more stringent NOx emissions limits,
in combination with the need for a reduction in fuel con-
sumption and hence CO2 emissions, researchers have been
striving to find a solution to the NOx emissions problems
related to lean-burn engines (4). Legislative pressure to de-
velop an operative exhaust-gas after-treatment system for
lean-burn engines might be exemplified by the fact that
some countries, such as Brazil and Taiwan, even ban the
sale of new cars equipped with diesel engines entirely (5).
Selective catalytic reduction using hydrocarbons as re-
ducing agents (HC–SCR) has long been discussed as a fea-
sible solution to the problem of removing NOx in excess
oxygen (6–9, and references therein). Unfortunately no
catalytic system could be developed which fulfills all the
needs, such as high selective NOx conversion to N2, hy-
drothermal stability, low light-off temperature, and a broad
temperature window. Platinum group metals seem to be a
favorable choice as active catalysts in this reaction (4). Most
studies on platinum group metals concerning the reduction
of NO were performed using Pt, Pd, and Rh; Ir so far has
received comparatively little attention and was usually only
used in multicomponent systems (10, 11). The limited use of
Ir (injection engines from Mitsubishi) can be traced to sev-
eral factors, among which the scarcity of Ir, problems with
Ir loss due to the formation of volatile iridium chlorides
and oxychlorides, and Ir’s underestimated catalytic poten-
tial may be the most crucial ones. Only recently Ir was found
to exhibit high activity in the reduction of NO in excess oxy-
gen (12–14), and especially unsupported Ir (Ir black) was
shown to exhibit high yields of N2 (15–17). Most of the
studies concerning NO decomposition and reduction have
dealt with supported catalysts but only a few studies on NO
size.
2002 Elsevier Science
Key Words: unsupported iridium; iridium black; IrO2; platinum
group metals; reduction of NO; HC–SCR; lean DeNOx; NOx reduc-
tion; dispersion; crystallite size; pulse thermal analysis.
1. INTRODUCTION
NOx (the sum of NO and NO2) belongs to the so-called
group of classical air pollutants and it is also termed by
the World Health Organization (WHO) a key air pollu-
tant (1). Its negative effects on the environment range from
the contribution to eutrophication of aquatic systems and
soil, to acidification, to the contribution to the formation of
tropospheric ozone, and therefore it constitutes an impor-
tant precursor substance for the formation of summer smog
(1–3). As in addition NO and NO2 are toxic and hazardous
compounds per se (2, and references therein), there is a
1 To whom correspondence should be addressed. Fax: +41 1 632 11 63.
E-mail: baiker@tech.chem.ethz.ch.
157
0021-9517/02 $35.00
2002 Elsevier Science
All rights reserved.
c

¨
WOGERBAUER, MACIEJEWSKI, AND BAIKER
158
decomposition were performed using unsupported plat- lyst B is the product of refluxing an aqueous solution of
inum group metals, such as Pt wire (18, 19), Pt and Rh wire IrCl3 3H2O with 25% NH3 for several hours, subsequent
(20, 21), and Pt foil (22). More recently van den Broek removal of excess NH3, and dilution with water.
et al. (23) reported on the low-temperature oxidation of
After impregnation of the zeolite, the resulting catalysts
NH3 to N2 over Pt and Ir sponge and found Ir sponge to be were dried in air at 120 C for 12 h; catalysts A and B were
highly selective to N2 due to the high activity of Ir for NO either calcined in air at 500 C for 2 h (oxidized form) or
dissociation. Acke and Skoglundh studied NO reduction reduced with 6% H2 at 450 C for 6 h, heating, and cooling
using propene, propane, and NH3 as reducing agents over in N2 (reduced form).
Pt black (24, 25). Except for Ir sponge in NH3 oxidation, as
Ir black and IrO2 were supplied by Alfa Aesar. These ma-
mentioned above (23), and our own studies on Ir black in terials were either used as received or, to produce a certain
the HC–SCR process (15–17), unsupported Ir seems not to crystallite size, presintered by heating in He to the corre-
have been studied before. The use of unsupported material sponding set temperature with a heating rate of 10 K/min
has many advantages. It enables the detailed investigation and then quenched to room temperature. Samples of Ir
of the intrinsic catalytic activity of the respective metal, and black were prereduced at 300 C with 6% H2 prior to the
the influence of metal dispersion can be assessed without catalytictests. Energy-dispersiveX-rayspectrometry, X-ray
interference from the support. Furthermore the synergies photoelectron spectroscopy, and X-ray fluorescence mea-
between the active metal and the mechanical mixing com- surements did not reveal any significant impurities in the
ponents (Al2O3, SiO2, H–ZSM-5) can be explored, and the samples.
comparison between supported and unsupported materi-
To distinguish supported catalysts from mechanical mix-
als allows an estimation of the influence of metal carrier tures, for supported catalysts a dash was used to separate
interaction on catalytic activity. Single reactions such as the metal and support (Ir–H–ZSM-5) and for mixtures a slash
reaction of NO, O2, and the reducing agent with the metal was used (Ir/H–ZSM-5).
can be studied without any disturbance from other compo-
2.2. Mixing Material
nents and reactions (such as adsorption and reaction of the
reducing agent on the support, spillover of species from the
metal to the support, etc.).
The silica gel, which was used as mixing material, was pre-
pared using the sol–gel method. The gel was prepared as xe-
rogel using a solution of tetramethylorthosilicate (22.83 g)
in i-propanol (105 ml) as starting material. A solution of
H2O (13.01 g) and HNO3 (1.45 g) in i-propanol (40 ml)
was added under vigorous stirring for hydrolysis. After 6 h
of hydrolysis trihexylamine (6.06 g in 40 ml of i-propanol)
was added and the gel was left to age for 6 days, dried in a
vacuum oven at 50 C for another 5 days, and then calcined
in air at 750 C for 3 h.
Here we investigate the behavior of Ir black under HC–
SCR conditions and we focus on the relation between Ir
crystallite size, Ir : IrO2 ratio, and catalytic activity. Sup-
ported Ir–H–ZSM-5 systems, prepared according to a stan-
dard preparation procedure, are used to establish a relation
between supported and unsupported Ir catalysts. Emphasis
is put on the development of catalytic activity with time on-
stream and the effects on crystallite size and the oxidation
state of Ir. As larger crystallites (catalysts with lower metal
dispersion) were found to exhibit significantly higher N2
yields (12, 13, 15), we provide evidence that the increase in
N2 yield with increasing Ir crystallite size could be closely
related to a change in the relative adsorption properties
of O2 and NO on Ir. Furthermore we studied the redox
behavior of the systems Ir–O2 and Ir–NO.
2.3. Catalytic Tests
For activity tests, 150 mg of the catalyst (particle fraction
between 0.066 and 0.177 m) was held between two quartz
wool plugs in a quartz reactor (6 mm i.d.). The mechani-
cal mixtures were prepared by dilution of Ir black with the
mixing material, SiO2. A reactor filling with a mixture of
unsupported Ir black contained 7.5 mg of Ir black and sil-
2. EXPERIMENTAL
1
ica to achieve a space velocity of approximately 80,000 h
.
The feed gas contained 1800 ppm propene, 450 ppm CO,
300 ppm NO, 10% H2O, 10.7% CO2, 8% O2, and N2 as
2.1. Catalyst Preparation
Ir catalysts used in this study were prepared by wet balance gas at a flow rate of 3 Nl/g/min. The term in situ
impregnation of H–ZSM-5 (Si/Al = 34, 400 m2/g, Chemie conditioning represents the treatment of the catalysts for
Uetikon). Two different aqueous solutions of Ir salts 4 h at 450 C with the above feed gas (heating in reac-
were used for impregnation: (A) IrCl3 3H2O, and (B) tion atmosphere). Online analysis was performed using a
Ir(NH3)x Cl3 yH2O. Catalysts are accordingly denomi- Bruker IFS 66 Fourier transform IR spectrometer; details
nated catalyst A and catalyst B, with actual metal loadings of the apparatus are described in Ref. (26). During steady-
of 4.2, and 4.6 wt%, respectively. Iridium chloride (cata- state experiments the temperature was gradually increased
lyst A) was obtain from Alfa Aesar and was used as the in steps of 50 C, starting at 160 C, with a holding time
starting material for synthesis of the latter compound. Cata- at each step of 1 h. For testing the influence of the time

STRUCTURE SENSITIVITY OF NO REDUCTION
159
onstream the catalysts were heated in N2 to the set temper-
ature, which was held for several hours after switching on
the synthetic exhaust gas. After the experiments the cata-
lysts were quenched in N2 to room temperature.
The total consumption of NO is given as XNO and NO
conversion to N2 (yield, YN2) was calculated according to
the following equation (superscript 0 denotes inlet flows):
YN2
=
FN0 O (FNO + FNO2 + 2 FN2O) /F0NO 100.
2.4. Characterization
FIG. 1. DevelopmentofyieldofN2 ofcatalystsAandB(solidsymbols
reduced, open symbols oxidized) over time. The catalysts were kept at
450 C for 16 h under reaction conditions. Catalyst A was kept onstream
for 32 h.
X-raydiffraction(XRD). X-rayanalysiswascarriedout
on a Siemens D 5000 powder X-ray diffractometer using
the CuK radiation in step mode between 20 and 80 22,
with a step size of 0.01 and 4 s/step. The Ir crystallite size
was estimated using XRD line broadening of the Ir(111) activity of a well-defined material at the specified temper-
reflection and the Scherrer equation, taking 0.9 as a shape ature. Figure 1 shows the development of YN2 over time.
factor value.
It can be seen that catalyst B starts at far higher YN2 than
does catalyst A and that the reduced form of catalyst B
is initially superior to the oxidized form. The yield of N2
increased with time onstream for both catalysts A and B,
but a steady-state could not be reached after 16 h onstream.
The oxidized form of catalyst A was treated 32 h onstream
and the activity still improved slightly. The accompany-
ing changes in Ir crystallite size and Ir0 : IrO2 are listed in
Table 1. The reduced and the oxidized form of catalyst A be-
haved rather similarly, whereas with catalyst B the oxidized
form was distinctly inferior, especially during the first 5 h of
conditioning, to the reduced form. From Table 1 it emerges
that catalyst A in all cases showed a lower Ir crystallite size
and a lower content of Ir0 compared to catalyst B. Irrespec-
tive of the initial oxidation state the treatment onstream led
to the establishment of a certain Ir : IrO2 ratio; this means
Thermal Analysis (TA). Thermoanalytical experiments
were carried out both isothermally and nonisothermally
(heating rate 5 or 10 K/min) on a Netzsch STA 409 ther-
moanalyzer equipped with a gas injector (PulseTA box,
Netzsch), which allows for the injection of a defined sample
gas volume into the carrier gas stream (27). If not stated oth-
erwise, the amount of injected gases (hydrogen, oxygen, ni-
tric oxide) was 1.0 ml. Gases evolved during reaction and/or
injected into the system were monitored online with a Balz-
ers QMG 420 quadrupole mass spectrometer connected to
the thermoanalyzer by a heated (ca. 200 C) capillary.
3. RESULTS
3.1. Change in Activity during Onstream Treatment
of Ir–H–ZSM-5
TABLE 1
It was found that the usually applied conditioning period
of 4 h at 450 C (12) was not enough for Ir–H–ZSM-5 cata-
lysts to reach their most active state. To study the influence
of onstream conditioning on Ir–H–ZSM-5 in more detail,
two Ir–H–ZSM-5 catalysts (catalysts A and B) with differ-
ent crystallite sizes, prepared using two different Ir precur-
sors, were treated onstream at 450 C for 16 h. This experi-
mentwascarriedoutwiththereducedandoxidizedformsof
the respective catalysts. During this onstream treatment the
development of NO conversion to N2 (YN2) over time was
monitored. The oxidized form of catalyst A was addition-
ally kept onstream for 32 h to check whether steady-state
could be reached by a further prolongation of the condition-
ing procedure. In contrast to the conditioning described in
(12), in the experiments performed here the catalysts were
heated slowly not in the reaction atmosphere but in N2, to a
final temperature of 450 C, and then the synthetic exhaust
feed gas was switched on. This was to avoid any bias due to
oxidation or reduction which could occur during heating in
the synthetic exhaust gas and to be able to record the initial
Ir Crystallite Size and Ir : IrO2 Ratio of Ir–H–ZSM-5
Catalysts A and B after Conditioning at 450 Ca
Crystallite
size before
Crystallite
size after
Time
Ir : IrO2 after
onstream conditioning conditioning conditioning
Material
(h)
(nm)
(nm)
(mol%)
Ir–H–ZSM-5 A
Reduced
Oxidized
Oxidized
Ir–H–ZSM-5 B
Reduced
16
16
32
4b
9c
9c
8b/8c
4b/10c
9b/9c
40 : 60
34 : 66
37 : 63
16
16
14b
17b/9c
19b/7c
66 : 34
73 : 27
Oxidized
16b/14c
a The samples were heated in N2 to 450 C, where the synthetic exhaust
gas was switched on. The oxidized form of catalyst B (15 h at 500 C in
air) initially contained 58% IrO2. The Ir : IrO2 ratio was determined by
reduction of the catalyst samples after conditioning by pulses of hydrogen
at 250 C.
b Ir crystallite size.
c IrO2 crystallite size.

¨
WOGERBAUER, MACIEJEWSKI, AND BAIKER
160
increase in temperature seemed to shorten the time to reach
steady-state but for the samples with 3 nm even at 450 C
steady-state was not achieved after 11 h. The results pre-
sented in Fig. 3 indicate that YN2 of the sample with 45 nm
did barely change over time (the slope of the interpolated
line is near zero), whereas the 3-nm sample showed a large
increase in YN2 (as evidenced by the large positive slope
of the interpolated line for 3 nm). The slope of the inter-
polated line for the sample with 19-nm size lies between
the values of the two other samples. Table 2 describes the
development of the crystallite size and the IrO2 content of
the investigated samples after activation at different tem-
peratures. It clearly illustrates the fact that, as with the sup-
ported catalysts, the lager the Ir crystallite size, the smaller
the ability to oxidize the catalyst during onstream treat-
ment. Catalysts with an initial crystallite size of 19 nm or
above did not or only very slightly sintered under reaction
conditions. As expected, the sample with the lowest crystal-
lite size exhibited the most pronounced changes. It showed
the highest content of IrO2 after all experiments and the
crystallite size increased from 3 to 15 nm at 450 C. In con-
trast to the other samples this sample showed a decrease in
oxide content with increasing reaction temperature. This is
in line with the general dependence of oxidation on crystal-
lite size: the higher the temperature, the more pronounced
the Ir sintering and the lower the corresponding final oxide
content. Figure 4 depicts the time necessary to reach 80%
of the total change in YN2, which was achieved after 11 h at
400 C. The dependence that with larger crystallite size less
time is necessary to reach steady-state holds true for 450 C
as well, but the t80% time was expectedly smaller than at
400 C (not shown).
FIG. 2. Development of yield of N2 and Ir : IrO2 ratio of catalysts A
and B (in reduced and oxidized forms) during conditioning at 450 C for
16 h under reaction conditions.
that under reaction conditions at 450 C, metallic Ir was ox-
idized and IrO2 reduced. In all cases the treatment led to
an increase in Ir crystallite size. This increase is somewhat
smaller compared to the experiments in Ref. (12), where
the catalysts were heated slowly in the reaction mixture to
450 C. Figure 2 summarizes the results for catalysts A and
B concerning the development of the oxidation state and
selective reduction of NO to N2 over time. Irrespective of
starting from the reduced or the oxidized form, the cata-
lysts seem to approach a certain equilibrium in the Ir : IrO2
ratio, which is approximately 1 : 2 for catalyst A and 2 : 1 for
catalyst B. YN2 before conditioning was usually somewhat
lower for the oxidized catalysts but after conditioning the
same values for YN2 and similar values for the Ir : IrO2 ratio
were reached.
Additionally, activity tests without SiO2, using a catalyst
bed of 100 mg of Ir black sintered to 27 nm, were performed
according to the experiments described above; ordinary ac-
tivity tests were as described in Refs. (12, 15). These ex-
periments delivered results similar to those of experiments
using mixtures of Ir and SiO2, proving that the observed
catalytic behavior is due to the Ir component alone and not
influenced by some kind of interaction of Ir with the mixing
material.
3.2. Change of Activity during Onstream Treatment
of Ir Black Mixed with SiO2
The influence of time onstream, initial crystallite size, and
temperature on YN2 and the Ir : IrO2 ratio as well as crystal-
lite growth for samples of Ir black were studied. The Ir black
samples were mixed with silica to insure plug flow condi-
tions. The samples were prereduced in 6% H2 at 300 C for
30minandthenheatedtotherespectivesettemperaturesof
350, 400, and 450 C; the final temperature was held for 11 h
and the development of YN2 was recorded. Figure 3 depicts
the development of YN2 vs time for samples with crystallite
sizes of 3, 19, and 45 nm at 400 and 450 C. It emerges from
Fig. 3 that the lower the crystallite size, the longer the time
FIG. 3. Development of yield of N2 for Ir black of different crystallite
to reach a steady-state and the lower the values of YN2. An sizes (3, 19, and 45 nm) at 400 (a) and 450 C (b).

STRUCTURE SENSITIVITY OF NO REDUCTION
161
TABLE 2
Crystallite Size and IrO2 Content of Samples of Ir Black with Different Initial Crystallite Sizes after 11 h at 350, 400, and 450 C
Initial crystallite
size (nm)
Crystallite size
after 350 C (nm)
IrO2 content after
350 C (mol%)
Crystallite size
after 400 C (nm)
IrO2 content after
400 C (mol%)
Crystallite size
after 450 C (nm)
IrO2 content after
450 C (mol%)
3
5
13
19
27
45
9
10
13
19
27
45
13.1
12
13
15
19
27
45
12.4
15
11.8
7.5
0
8.7
5.4
0.9
21
27
45
10.8
8.2
2.0
Note. The IrO2 content was determined by reduction of the catalyst samples after conditioning by pulses of hydrogen at 250 C.
3.3. Influence of Ir Crystallite Size on Reactions
Occurring during HC–SCR
of NOx reduction to N2. This behavior was only found with
prereduced samples; conditioned catalysts showed usually
a simultaneous reductant and NOx reduction light-off and
exhibited higher yields of N2. This coincidence of reduc-
tant light-off and NOx reduction light-off has often been
described before with platinum group metals in HC–SCR
(for instance, Refs. 7, 28). In Figures 5C–F the situation for
nitrogen-containing compounds is depicted. With increas-
ing crystallite size the tendency to form N2O at 210 C is
reduced, NO2 formation was totally suppressed on larger
crystals, XNO decreased with increasing crystallite size, and
YN2 at 360 C developed in the sequence 5 < 45 < 19 nm,
indicating that the increase in selectivity over larger Ir crys-
tallites is partly offset by a reduction in XNO, which is very
likely due to the loss of specific surface area.
Steady-state activity tests up to 360 C, where usually a
large difference in activity between the samples of different
Ir crystallite size is found (NOx light-off region), confirmed
the differences in Ir : IrO2 ratio during reaction for Ir black
samples with different crystallite sizes. As in the usual activ-
ity tests (i.e., heating to a final temperature of 460 C), the
samples were mixed with SiO2 and heated in steps of 50 C
from 160 to 360 C; at this temperature the reaction was
stopped and the samples were quenched to room tempera-
ture in N2. From Table 3 it can be deduced that even in the
light-off region for NOx reduction the usual tendencies are
found: the larger the crystallite size, the lower the degree
of Ir oxidation, and the less the tendency for further sinter-
ing. Figures 5A and 5B depict the conversion curves for the
reductants, CO and propene. Note that with increasing crys-
tallite size the light-off for the reductants is shifted to higher
temperatures and that both reducing agents are completely
consumed from 250 C on. It is interesting to note that the
onset of propene consumption occurred at the same tem-
perature where production of N2O set in and the total con-
sumptionofthereductantsoccurredwellbelowthelight-off
3.4. Interaction of NO with Iridium
To better understand possible elementary steps which
could occur in the HC–SCR process over Ir, the interac-
tions of NO and Ir were studied more deeply. Though the
Ir–NO system seems to be rather simple compared to a
complete synthetic exhaust gas and Ir/SiO2, as used above,
it turned out that it exhibits rather complicated behavior.
In the following, interactions between NO and crystalline
IrO2 are neglected because only weak adsorption of NO
on crystalline IrO2 was found and in Ref. (15) it was shown
that IrO2 was inactive for NO reduction. Furthermore, we
found no interaction of NO with an XRD-amorphous IrO2
precursor, which is formed during HC–SCR conditions
(15) or by low-temperature adsorption of O2 on Ir. Three
TABLE 3
Crystallite Size and IrO2 Content after Activity Tests up to 360 C
of Ir Black with Different Initial Crystallite Sizes
Initial crystallite
size (nm)
Crystallite size
after 350 C (nm)
IrO2 content after
350 C (mol%)
5
19
45
9
19
45
9
5
0
FIG. 4. Time required to reach 80% of the increase in yield (t80%) at
400 C vs crystallite size of Ir black.

¨
WOGERBAUER, MACIEJEWSKI, AND BAIKER
162
pulses, respectively, allows the estimation of the contri-
bution of the various reactions to the overall process.
As reactions 1–3 depend differently on temperature, with
increasing temperature the fraction of NO decomposing
over Ir increases, as depicted in Fig. 6C. At 500 C almost
all NO in contact with Ir0 was decomposed to N2 and oxi-
dized Ir0 to IrO2. Figure 6D illustrates that this increase in
the contribution of the Ir oxidation to the overall mass-gain
was mainly due to a decrease in the amount of chemisorbed
NO and only to a relatively small increase in the oxidation
of Ir. This implies reversible adsorption of NO on Ir and
its strong temperature dependence. As adsorption of NO
is a prerequisite for Ir oxidation by NO, the much smaller
amount of adsorbed NO at higher temperature has the con-
sequence that the amount of oxidized Ir due to a pulse of
NO does not increase much with temperature. NO decom-
position or NO dissociation over platinum group metals
has often been interpreted as the main process leading to
N2 and to a poisoning of the active sites for NO decompo-
sition with oxygen from NO (28, 29). Figure 6A shows the
increase in N2 and N2O yield with temperature due to NO
decomposition over Ir, and Fig. 6B illustrates the decrease
in produced N2/N2O after reaction of the chemisorbed NO
with a pulse of H2, indicating the decrease in adsorbed NO.
Due to three contributions to the intensity of the mass spec-
trometric signal of m/e = 28 (i.e., the presence of residual
amounts of nitrogen in the TA-MS system, the fragmenta-
tion of N2O, and the main signal originating from formed
N2), the concentration of N2 and N2O could be only roughly
estimated. The maximum concentration of N2O, detected
at ca. 300 C, did not exceed 20–30% of the amount of N2
formed. In the following, where significant contribution of
FIG. 5. Conversion of propene (A) and CO (B), production of N2O
(C) and NO2 (D), total conversion of NO (E), and yield of N2 (F). Exper-
iments where carried out using Ir black with 5-, 19-, or 45-nm Ir crystallite
size up to 360 C and using the complete synthetic exhaust gas. The tem-
perature was increased stepwise up to 360 C from 160 C in steps of 50 C,
with a holding time of 1 h per step.
different processes can occur during interaction of NO and
metallic Ir:
1. decomposition of NO and formation of N2:
Ir0 + 2NO → Ir–Oad + N2,
2. decomposition of NO and formation of N2O:
Ir0 + 4NO → Ir–Oad + 2N2O,
3. adsorption of NO on the Ir surface:
Ir0 + NO → Ir–NOad,
where Ir–Oad and Ir–NOad correspond to adsorbed and/or
reacted oxygen and NO, respectively. These processes dif-
ferently depend on temperature, as emerges from the re-
sults summarized in Fig. 6. At room temperature a large
fraction of NO chemisorbed on the Ir surface without re-
action and only a small part of NO decomposed, leav-
ing oxygen behind on the Ir surface. The contribution of
processes 1 and 2 could be quantified by the determina-
tion of the amount of N2 and N2O formed during the
pulse of NO and by quantifying the content of Ir–Oad
by reduction of the samples with H2. The contribution
of process 3 can be estimated from the quantification of
the amount of N2/N2O formed due to a pulse of H2 af-
ter the NO pulse (i.e., by the reduction of chemisorbed
NO by H2). This information together with the mass-
FIG. 6. Ir black with 13-nm crystallite size exposed to 1-ml pulses
of NO at different temperatures (results for 5 nm are included as open
circles): (A) amount of N2 formed due to NO decomposition after a 1-ml
pulse of NO; (B) evolution of N2 due to pulses of H2 after the NO pulse;
(C) increase in NO decomposition compared to molecular adsorption with
temperature. Ir black with 5 nm: (D) contribution of oxidation of Ir by NO
gain and mass-loss of the sample due to NO and H2 (1mO) and adsorption of NO on Ir to the overall mass-gain (1mA+O).

STRUCTURE SENSITIVITY OF NO REDUCTION
163
desorption. About 66% of the NO chemisorbed at 100 C
desorbed without reaction; the rest was decomposed during
the desorption (ca. 26% NO reacted to N2O and 8% to N2),
oxidizing the part of the Ir surface which was not oxidized
during adsorption. The partial decomposition of NO on the
Ir surface during desorption was reflected by the existence
of a minimum centered at 250 C in the mass spectrometric
signal of evolved NO (m/e = 30) accompanied by maxima
in m/e = 44 and m/e = 28 signals (not shown). The amount
of oxygen irreversibly adsorbed on the Ir surface during ad-
sorption and desorption of NO was quantified using reduc-
tion by a pulse of H2 at 400 C (mass-loss depicted as 1mR
in Fig. 7B). Desorption of NO was complete at ca. 380 C.
3.5. Influence of Iridium Crystallite Size on Reactions
of Iridium with NO and O2
The competition between NO and O2 for adsorbing on
and oxidizing Ir as well as their different affinities for oxi-
dation of the reducing agent are key factors for explaining
the selective reduction of NO over Ir. Triggered by the ob-
servation that large Ir crystallite size favors high yields of
N2 (Fig. 3), a closer look was taken at the changes in the
reactions of NO and O2 with Ir which are due to a variation
in crystallite size.
The oxidation behavior of unsupported Ir was investi-
gated at temperatures between ca. 200 and 450 C by in-
jecting 1 ml of O2. Figure 8 summarizes the oxidation of
Ir black with different crystallite sizes. In Fig. 8A the pulse
sequence at 350 C is depicted which was used to study the
oxidationprocess. Thepulseof1mlofO2, which ledtoacer-
tain mass-uptake of the catalyst sample, was followed by a
pulse of H2 removing the oxygen taken up before. This pro-
FIG. 7. (A) Pulses of 1 ml of NO over Ir black at 100 C illustrating NO
decomposition (mass-gain accompanied by the evolution of N2) during the
first pulse and deactivation and saturation of the Ir surface by the following
pulses. (B) Desorption of NO adsorbed at 100 C (heating rate 10 K/min)
followed by reduction of the present oxygen by a pulse of 2 ml of H2 at ca.
400 C.
N2O to the N2 signal could be excluded, m/e = 28 was con- cedure was repeated several times to check reproducibility.
sidered representative for N2. The corresponding TG curves for samples with 5, 13, 27,
Figure 7A depicts the adsorption of NO over Ir at 100 C. and 45 nm Ir crystallite size show that the mass-uptake due
Results presented in Fig. 6 indicate that ca. 30% of the ob- to the O2 pulse distinctly decreases with increasing crystal-
served mass-gain was due to the deposition of oxygen on lite size. Figure 8B shows the progress of oxidation of Ir
the Ir surface and 70% was due to the chemisorption of samples having crystallite sizes of 13, 27, and 46 nm at dif-
NO. It can be clearly seen that only the first pulse of NO ferent temperatures due to pulses of 1 ml of O2. After each
(2 ml) led to significant formation of N2. This indeed in- oxygen pulse, samples were reduced by hydrogen pulses.
dicates the poisoning of active sites by oxygen produced The results are in accordance with the tendencies observed
from the decomposition of NO and the blocking of the rest in the catalytic tests where samples with the largest crys-
of the reaction sites by NO adsorbed during the first pulse. tallite sizes were oxidized to the lowest degree, and the
Thesecondpulseledtoaverysmallfurthermass-uptakebut higher the temperature, the higher the oxide content. The
the third pulse corroborated the fact that the surface had sample with 45-nm crystallite size showed only a small but
been saturated and all adsorption sites had already been linear increase in mass-gain with temperature, whereas the
occupied before. Further pulses of NO only led to a small sample with the smallest crystallite size exhibited a more
temporary mass-gain, which is typical for weak adsorption pronounced increase in the slope of the interpolated oxida-
and subsequent desorption. The temperature-programmed tion rate at temperatures above 350 C. The inset in Fig. 8B
desorption of preadsorbed NO is depicted in Fig. 7B; this depicts the relationship between the oxidation extent and
proves that NO, to a certain extent, is reversibly adsorbed crystallite size at 350 C. The sample with 13-nm size was
on Ir. The mass-loss due to NO desorption (1mD in Fig. 7B) oxidized to ca. 4%, whereas the sample with 45-nm size
and integral intensities of m/e = 30 (NO), m/e = 44 (N2O), was only oxidized to less than 1%. With O2 no reversible
and m/e = 28 (N2 and N2O fragmentation, not shown) al- adsorption could be witnessed. Adsorbed O2 formed an
low the quantification of the processes occurring during amorphous precursor of IrO2 which crystallized between

¨
WOGERBAUER, MACIEJEWSKI, AND BAIKER
164
FIG. 9. (A) Mass-gain due to nonisothermal (heating rate 2 K/min)
oxidation of Ir black by pulses of 1 ml of O2 and subsequent reduc-
tion by H2. (B) Mass-gain due to nonisothermal (2 K/min) adsorp-
tion/decomposition of NO over Ir and subsequent reduction by H2.
(C) Mass-gain due to adsorption/oxidation resulting from the injection
of NO (2-ml pulse) and oxygen (1-ml pulse) over 27 nm of Ir at 350 C.
FIG. 8. (A) Pulse sequence used to investigate the mass changes of Ir
black with crystallite sizes of 5, 13, 27, and 45 nm due to 1-ml pulses of O2
and H2. (B) Amount of irreversibly adsorbed oxygen on Ir black expressed
as oxidation extent due to 1-ml pulses of O2 as a function of temperature
and Ir crystallite size (13, 27, and 45 nm). The inset represents the oxidation
extent at 350 C as a function of crystallite size.
400 and 500 C to crystalline IrO2, as detected by XRD.
Only at temperatures above 800 C could decomposition of
IrO2 to Ir and O2 be witnessed. The characterization of
the phase formed due to adsorption of oxygen on Ir below
400 C is presently underway.
The differences described above in the interaction of NO
and O2 with Ir are further illustrated in Fig. 9. With increas-
ing temperature, oxidation with O2 led to a pronounced
mass-gain, whereas with NO, processes 1–3 were contribut-
ing differently, depending on the temperature, to the over-
all process, leading to a net decrease in mass-uptake with
temperature. Figure 9C depicts the interesting feature that
after two 2-ml pulses of NO over Ir black, which is enough
to saturate the Ir surface, as shown in Fig. 7A, a following
pulse of 1 ml of O2 led to a pronounced further mass-gain
in the sample (see also Ref. 16). This could indicate that
there are more adsorption sites available for O2 than for
NO. To check how this behavior depends on crystallite size,
samples with various Ir crystallite sizes were subjected to
the same series of pulses to compare their ability to re-
act with Ir. To estimate the influence of temperature these
experiments were performed over the range between 100
FIG. 10. (A) Influence of the crystallite size of Ir black on the ratio
of the mass-gain due to the pulses of 2 ml of NO and 1 ml of O2 at 350 C.
(B) Dependence of the mass-gain due to deposition of oxygen on the Ir
surface resulting from pulses of 2 ml of NO or 1 ml of O2 on the crystallite
and 500 C. From Fig. 10A it emerges that with increasing size of Ir and on temperature.

STRUCTURE SENSITIVITY OF NO REDUCTION
165
crystallite size at 350 C the mass-gain due to oxidation by onstream at 350, 400, and 450 C, the sample with 3-nm ini-
either reactant (NO or O2) decreased but the decrease for tialcrystallitesizehadthehighestcontentofIrO2 compared
oxidation by O2 was far more pronounced. The inset in to all other samples, but due to the growing crystallite size
Fig. 10A presents the change in the ratio of the amount of this sample, the IrO2 content after reaction decreased
of oxygen adsorbed during O2 pulses to the amount ad- from the 350–450 C experiment. With the 19- and 45-nm-
sorbed after NO decomposition as a function of the crystal- sized samples, the IrO2 content increased with increas-
lite size of Ir. In Fig. 10B this ratio is depicted as a function ing temperature and no further crystallite size growth was
of temperature for samples with 5-, 19-, and 45-nm size. It is observed.
evident that this ratio is significantly smaller for larger crys-
The experiments depicted above confirm the strong in-
tals than for smaller ones. The slower increase in this ratio fluence of Ir crystallite size on the HC–SCR process. A de-
for the 5-nm sample at temperatures above 250 C is due pendence of DeNOx -activity of Pt-based catalysts on par-
to the increase in its crystallite size at higher temperatures. ticle size has been mentioned before, in Refs. (7, 28), which
The larger the Ir crystallite size, the smaller the difference claim that catalysts with higher Pt dispersion exhibit lower
between the oxidation ability of both gases. All results pre- activity compared to catalysts with lower dispersion. For Rh
sented in Fig. 10 indicate that the size of active adsorption catalysts similar findings were published on the reduction
centeres for NO is distinctly smaller than for O2, but for of NO by CO (30–32), but it seems that for Ir catalysts in
larger Ir crystallite size a relative increase in NO adsorp- HC–SCR this effect of dispersion is most pronounced (12,
tion centeres is found. This observation could explain the 13, 15). It influences redox behavior and hence the activity
increased selectivity of the NO reduction process over Ir and selectivity of the reactions involved. The observed ten-
with increasing crystallite size.
dencies are quite clear: increasing crystallite size decreases
the degree of oxidation of the samples and the time to reach
steady-state; it also increases YN2 as well as selectivity to N2
and strongly disfavors oxidation of NO to NO2. It is inter-
4. DISCUSSION
For both supported and unsupported Ir catalysts, condi- esting that catalysts with the lowest degree of oxidation are
tioning onstream proved to be beneficial, as reported previ- the most active and are even more active than is completely
ously in Ref. (12). This process leads to the establishment of reduced, metallic Ir black. It seems that a certain degree of
a certain Ir : IrO2 ratio dependent on Ir crystallite size, time surface oxidation, with the bulk of the material remaining
onstream, and temperature. For catalysts with small initial metallic Ir, enhances activity for NO reduction (see the re-
crystallite size it causes significant agglomeration/sintering. sults for unsupported Ir with 45-nm size). Assuming an NO
Although the mechanism of this agglomeration process of decomposition mechanism as proposed for Pt by Burch and
the iridium component may be influenced by the presence coworkers (28, 33) this may indicate that the reconstruction
of chlorine on the supported Ir–H–ZSM5 catylysts (volatal- of the Ir surface by oxygen favors NO decomposition (34),
ization/redeposition of formed oxychlorides), with unsup- but it may also indicate that hydrocarbon activation is facil-
ported Ir black sintering is the most likely process leading itated (35). The described tendencies are confirmed by ex-
to crystal growth under the conditions used. Morpholog- periments using pure Ir black without mixing material. The
ical changes accompanied by crystal growth have been results proved that the activity development we observed
followed using XRD and electron microscopy (15). With with supported Ir catalysts and in mixtures of Ir black with
unsupported Ir black ca. 20 nm seems to be a stable Ir SiO2 is mainly due to the intrinsic catalytic activity of the
crystallite size under the conditions applied in these ex- Ir component and not to an interaction between support or
periments. All samples with smaller crystallite sizes were mixing material and Ir or Ir black, respectively.
readily sintered up to values near 20 nm, while crystallite
Crystallite size was found to have an effect not only on
size of samples above 20 nm was not influenced markedly oxidation behavior using O2 as oxidation agent but also
by onstream treatment. As expected, on supported Ir–H– on the reaction of NO with Ir. These findings appear to
ZSM-5 conditioning is far less effective compared to that be important for the explanation of the catalytic behavior
on Ir black. This is reflected by the fact that YN2 of the sup- of Ir. The results of the experiments of the oxidation of
ported catalysts never reached values comparable to those Ir by O2 and NO differ considerably. Oxygen seems to be
of Ir black (ca. 50% compared to ca. 80%) and did not lead adsorbed irreversibly on the Ir surface whereas NO adsorp-
to steady-state even after 32 h onstream at 450 C compared tion is partially reversible. The temperature dependence of
to less than 1 h for Ir with 45-nm size at the same tempera- irreversibleoxygenadsorptiononIrismorepronouncedfor
ture. With the unsupported catalysts only the samples of Ir O2 than for NO. The dependence of oxygen uptake on crys-
blackwithaverysmallinitialcrystallitesize(3nm)werestill tallitesizealsoshowsdistinctdifferencesforthesereactants.
far away from steady-state conditions after 11 h onstream. The reactions of NO with Ir can be considered the superpo-
With the Ir black sample of 3-nm Ir crystallite size, the sition of three processes which differently depend on tem-
superposition of the effects of crystallite growth and Ir perature: (i) (partially reversible) adsorption of NO on Ir,
oxidation can be observed. After keeping the samples 11 h (ii) decomposition of NO and formation of Ir–Oad leading

¨
WOGERBAUER, MACIEJEWSKI, AND BAIKER
166
to N2, and (iii) decomposition of NO and formation of Ir– catalytic tests confirm this behavior: for Ir with larger crys-
Oad leading to N2O. It was found that with increasing tem- tallite size we observe less production of NO2 and higher
perature the percentage of NO which decomposes largely yields of N2.
increases compared to the amount of NO adsorbed, leading
to an increase in N2 yield. This decomposition of NO over Ir
results in self-poisoning, as has been confirmed before for
Pt catalysts (28), and hence oxygen from NO is blocking
active sites, leading to fast deactivation of the catalyst. This
decomposition of NO was also described by Savkin and
Kislyuk (34), who found NO decomposition to be more ef-
fective on Ir than on Pt. They also stated, similar to our
findings, that with increasing temperature the adsorption
rate of NO is rather low due to a very short lifetime of NO
molecules on the Ir surface. This also explains why with in-
creasing temperature the degree of oxidation of Ir by NO
lags behind the degree of oxidation with O2. Whereas O2
is irreversibly adsorbed on Ir, not all NO molecules are ca-
pable of dissociating during the short lifetime on the Ir sur-
face due to the fast desorption process. The temperature
dependence of adsorption/desorption and decomposition
of NO could also explain why the combustion of propene
is strongly inhibited at lower temperatures (below 230 C)
in the presence of NO, as described in Ref. (16). NO seems
to be competing for the same surface sites as propene and
at lower temperatures a major part of NO is chemisorbed
on Ir and thereby blocks adsorption sites for propene. Only
when enough NO is decomposed or desorbed again can
propene oxidation proceed on the newly liberated adsorp-
tion sites. This also indicates that at lower temperatures no
direct reaction between propene and adsorbed NO occurs.
The fact that after saturation of the catalyst with NO by
pulsing several pulses of NO over Ir black a pulse of O2 still
leads to a further mass-gain (i.e., oxidation of the sample) is
an indication that NO and O2 can adsorb on the Ir surface at
different sites. NO seems to only selectively adsorb and re-
act on certain surface sites, whereas O2 adsorption is rather
indifferent to the nature of sites. With increasing crystallite
size the ratio of NO adsorption sites to O2 adsorption sites
changes in favor of NO adsorption and/or reaction. The rel-
ative amount of sites on which only O2 can react seems to
decrease. This could be rationalized by the disappearance
of edges and steps with increasing crystallite size on which
O2 could preferentially dissociate, or by an increase in the
relative abundance of certain crystal faces which are active
for NO adsorption/decomposition. These assumptions ap-
pear reasonable, as NO decomposition belongs to the group
of structure-sensitive reactions (34) (i.e., the reaction rate
is dependent on the crystal face). The above-described dif-
ference between O2 and NO is a significant finding for the
question of why with increasing crystallite size NO reduc-
tion is favored over NO oxidation in an oxygen-containing
atmosphere. The decrease in oxygen concentration has two
important consequences: it reduces the chance for NO to be
oxidized to NO2 and it diminishes unselective hydrocarbon
consumption, affording increasing reductant efficiency. The
5. CONCLUSIONS
Studies on the structure sensitivity of HC–SCR of NO
over Ir catalysts have been combined with the investiga-
tion of the adsorption behavior of O2 and NO. The effi-
ciency of the NO reduction was found to be strongly influ-
enced by Ir crystallite size. Ir crystallite size determines the
time needed for the catalyst to reach steady-state under re-
action conditions. The larger the crystallite size, the faster
steady-state is achieved and the faster a certain Ir : IrO2 ra-
tio, together with high yields of N2, is established. For sup-
ported catalysts, prepared according to standard catalyst
preparation procedures, this process of reaching steady-
state seems to be retarded by the metal-support interac-
tion; unsupported Ir black reaches higher yields of N2 in
the reduction of NO by propene compared to supported
catalysts. Ir black with a crystallite size below 20 nm sin-
ters under HC–SCR conditions. NO adsorption on Ir0 was
found to be partially reversible, whereas O2 adsorption is ir-
reversible. Furthermore NO selectively adsorbs on certain
surface sites, whereas O2 adsorption seems to be indifferent
to structural differences on the Ir surface. With increasing
crystallite size the number of sites on which only O2 can
adsorb decreases and thus the ratio of the surface concen-
tration of NO to O2 changes in favor of NO. This leads
to a reduced probability of NO oxidation and unselective
hydrocarbon consumption. This seems to be a key factor
responsible for the pronounced improvement in selective
conversion of NO to N2 with increasing Ir crystallite size.
ACKNOWLEDGMENT
We thank dmc2 Degussa Metals Catalyst Cerdec AG for financial sup-
port of this research project.
REFERENCES
1. “Guidlines for Air Quality,” available at http://www.who.int/ peh/.
2. Fritz, A., and Pitchon, V., Appl. Catal. B 13, 1 (1997).
3. “Air Pollution in Europe,” EEA, Environmental Monograph No. 4.,
(A. Jol and B. Kielland, Eds.), Office for Official Publications of the
European Community, Luxenburg, 1977.
4. Ko¨nig, A., Herding, G., Hupfeld, B., Richter, T., and Weidmann, K.,
Top. Catal. 16/17, 23 (2001).
5. Walsh, M. P., SAE P-990107 (1999).
6. Akama, H., and Matsushita, K., Catal. Surv. Jpn. 3, 139 (1999).
7. Burch, R., and Millington, P. J., Catal. Today 26, 185 (1995).
8. Farrauto, R. J., and Heck, R. M., Catal. Today 51, 351 (1999).
9. Paˆrvulescu, V. I., Grange, P., and Delmon, B., Catal. Today 46, 253
(1998).
10. Nakatsuji, T., Yasukawa, R., Tabata, K., Ueda, K., and Niwa, M., Appl.
Catal. 21, 121 (1999).
11. Takami, A., Takemoto, T., Iwakuni, H., Yamada, K., Shigetsu, M., and
Komatsu, K., Catal. Today 35, 75 (1997).

STRUCTURE SENSITIVITY OF NO REDUCTION
167
12. Wo¨gerbauer, C., Maciejewski, M., Baiker, A., and Go¨bel, U., Top.
24. Acke, F., and Skoglundh, M., Appl. Catal. 22, L1 (1999).
25. Acke, F., and Skoglundh, M., Appl. Catal. 20, 133 (1999).
Catal. 16/17, 181 (2001).
13. Nakatsuji, T., Appl. Catal. 25, 163 (2000).
14. Nawdali, M., Praliaud, H., and Primet, M., Top. Catal. 16/17, 199
26. Radtke, F., Ko¨ppel, R., and Baiker, A., Catal. Today 26, 159
(1995).
(2001).
27. Maciejewski, M., Mu¨ller, C. A., Tschan, R., Emmerich, W. D., and
Baiker, A., Thermo. Chim. Acta 295, 167 (1997).
28. Burch, R., Millington, P. J., and Walker, A. P., Appl. Catal. B 4, 65
(1994).
15. Wo¨gerbauer, C., Maciejewski, M., Baiker, A., and Go¨bel, U., J. Catal.
201, 113 (2001).
16. Wo¨gerbauer, C., Maciejewski, M., and Baiker, A., Appl. Catal. B 34,
11 (2001).
29. Burch, R., and Watling, T. C., Catal. Lett. 37, 51 (1996).
17. Wo¨gerbauer, C., Maciejewski, M., Schubert, M. M., and Baiker, A., 30. Oh, S. H., and Eickel, C. C., J. Catal. 128, 526 (1990).
Catal. Lett. 74, 1 (2001).
31. Hecker, W. C., and Breneman, R. B., in “Proceedings, Catalysis and
Automotive Pollution Control, Brussels” (A. Crucq and A. Frennet,
Eds.), Vol. 30, p. 257. 1986.
32. Kaspar, J., de Leitenburg, C., Fornasiero, P., Trovarelli, A., and
Graziani, M., J. Catal. 146, 136 (1994).
33. Burch, R., and Sullivan, J. A., J. Catal. 182, 489 (1999).
34. Savkin, V. V., and Kislyuk, M. U., Kinet. Catal. 37, 555 (1996).
35. Obuchi, A., Ohi, A., Nakamura, M., Ogata, A., Mizuno, K., and
Ohuchi, H., Appl. Catal. B 2, 71 (1993).
18. Green, T. E., and Hinshelwood, C. N., J. Chem. Soc. 128, 1709 (1926).
19. Pancharatnum, S., Lim, K. J., and Manson, D. M., Chem. Eng. Sci. 30,
781 (1975).
20. Bachmann, P. W., and Taylor, G. B., J. Phys. Chem. 33, 447 (1929).
21. Zawadski, J., and Perlinsky, G., Compt. Rend. 198, 260 (1934).
22. Amirnazmi, A., and Boudart, M., J. Catal. 39, 383 (1975).
23. van den Broek, A. C. M., van Grondelle, J., and van Santen, R. A.,
J. Catal. 185, 297 (1999).