Study on metal-MFI/cordierite as promising catalysts for selective catalytic reduction of nitric oxide by propane in excess oxygen

Article abstract of DOI:10.1016/j.jcat.2004.08.016

ZSM-5 and AlTS-1 zeolites were successfully in situ synthesized on the surface of honeycomb cordierite substrate, certified by XRD and SEM techniques. Strong interaction between zeolite and substrate developed during in situ synthesis, entailing improved hydrothermal stabilities of the zeolites. A series of nonnoble metal ion-exchanged ZSM-5/cordierite and AlTS-1/cordierite were studied as catalysts for the selective reduction of nitric oxide by propane under dry and wet conditions. In the deNO_x processes, NO_x (ads) and C_x H_yO_z (ads) were thought as the important reaction intermediates and the formation of them on the catalysts was a key step in deNO_x reactions. Based on this, the factors that influence the activities of the catalysts were discussed considering not only the active components but also the supports. As for the active components, the ability to adsorb gas reactants and the oxidative activity are two dominant factors that determined the deNO_x activities. As for the supports, the Bronsted acidity and oxidative activity are two important factors that influenced the deNO_x activities. From the catalytic testing results, Cu-ZSM-5/cordierite is selected as a promising SCR deNO_x catalyst due to its superior deNO_x activity and high selectivity. For practical purposes, the catalytic testing of Cu-ZSM-5/cordierite was also preformed on a real lean-burn engine. Hydrocarbons and carbon monoxide in the exhaust were directly used as reductants for NO_x reduction. Thus, three main pollutants in the exhaust could be removed simultaneously. As expected, Cu-ZSM-5/cordierite also exhibited a rather good durability due to the in situ synthesis method.

Full text of DOI:10.1016/j.jcat.2004.08.016

Journal of Catalysis 228 (2004) 12–22
www.elsevier.com/locate/jcat
Study on metal-MFI/cordierite as promising catalysts for selective
catalytic reduction of nitric oxide by propane in excess oxygen
Landong Li a, Jixin Chen a, Shujuan Zhang a, Naijia Guan a,∗, Manfred Richter b,
Reinhard Eckelt b, Rolf Fricke b
a
Institute of New Catalytic Materials Science, College of Chemistry, Nankai University Cooperative Institute of Nankai & Tianjin University,
Tianjin 300071, People’s Republic of China
b
Institute for Applied Chemistry, Berlin-Adlershof, D-12489 Berlin, Germany
Received 19 May 2004; revised 2 August 2004; accepted 12 August 2004
Available online 22 September 2004
Abstract
ZSM-5 and AlTS-1 zeolites were successfully in situ synthesized on the surface of honeycomb cordierite substrate, certified by XRD
and SEM techniques. Strong interaction between zeolite and substrate developed during in situ synthesis, entailing improved hydrothermal
stabilities of the zeolites. A series of nonnoble metal ion-exchanged ZSM-5/cordierite and AlTS-1/cordierite were studied as catalysts for
the selective reduction of nitric oxide by propane under dry and wet conditions. In the deNO processes, NO (ads) and C H O (ads) were
x
x
x y z
thought as the important reaction intermediates and the formation of them on the catalysts was a key step in deNO reactions. Based on this,
x
the factors that influence the activities of the catalysts were discussed considering not only the active components but also the supports. As
for the active components, the ability to adsorb gas reactants and the oxidative activity are two dominant factors that determined the deNO
x
activities. As for the supports, the Brønsted acidity and oxidative activity are two important factors that influenced the deNO activities. From
x
the catalytic testing results, Cu-ZSM-5/cordierite is selected as a promising SCR deNO catalyst due to its superior deNO activity and high
x
x
selectivity. For practical purposes, the catalytic testing of Cu-ZSM-5/cordierite was also preformed on a real lean-burn engine. Hydrocarbons
and carbon monoxide in the exhaust were directly used as reductants for NO reduction. Thus, three main pollutants in the exhaust could be
x
removed simultaneously. As expected, Cu-ZSM-5/cordierite also exhibited a rather good durability due to the in situ synthesis method.
2004 Elsevier Inc. All rights reserved.
Keywords: NO ; SCR; MFI/cordierite; In situ synthesis; Adsorption; Oxidative activity; Brønsted acidity
x
1. Introduction
burn gasoline engines. Thus, selective catalytic reduction
(SCR) of NOx by hydrocarbons in the presence of oxygen
becomes a potential method for removing NOx from exhaust
[5–7]. Much work has been focused on zeolite-based materi-
als, particularly noble metal-exchanged zeolites [8–12] and
copper ion-exchanged zeolites [13–15], due to their high cat-
alytic activities.
A breakthrough of the SCR technology is hindered by the
insufficient low activity, narrow temperature window, and in-
sufficient durability of the catalysts [16]. Moreover, for prac-
tical purposes, catalytic powders must be fixed on the shaped
substrate. Honeycomb cordierite (2MgO · 2Al2O3 · 5SiO2)
is the substrate in common use for its superior hydrother-
mal stability and plasticity as well as economic reasons. The
Nitrogen oxides (NOx) are major air pollutants that
greatly contribute to the formation of photochemical smog
and acid rain [1]. Nearly all NOx (95%) derives from trans-
portation (49%) and power plants (46%) [2]. Three-way
catalysts can effectively convert NOx to nitrogen under the
condition of stoichiometric air-to-fuel ratios between 14.7
and 14.2 [3,4]. However, the three-way catalysts suffer from
severe loss of activity for NOx reduction in the presence of
excess oxygen, which is the prevalent condition for lean-
*
Corresponding author. Fax: +86-22-2350-0341.
E-mail address: guanj@public.tpt.tj.cn (N. Guan).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.08.016
2004 Elsevier Inc. All rights reserved.

L. Li et al. / Journal of Catalysis 228 (2004) 12–22
13
usual coating techniques are dip coating, slip coating, and
Table 1
Ion-exchange processes and results
spin coating [17]. However, it is well known that crystalline
zeolite materials are difficult to washcoat, and that excessive
use of binders often influences the activity of the catalyst and
even results in a “second pollution” because of its friabil-
ity. To solve the problems, it has already been proposed that
the catalytic zeolite component be prepared from crystallites
grown on substrate surfaces [18]. The main advantage of in
situ synthesis is that the substrate is used as a base for nucle-
ation and that under specific conditions a chemical bonding
between crystals and substrate layer is formed and no binder
or glue is needed [19]. Thus, the preparation course is greatly
simplified and under the condition of high space velocities,
the loss of catalyst can be prevented effectively.
Samples
Solution
(mM)
pH
Time
(h)
Metal loading
(wt%)
Cu-ZSM-5/cordierite
In-ZSM-5/cordierite
La-ZSM-5/cordierite
20 Cu(Ac)
15 In(NO )
15 La(NO )
5.5–6 20 × 4 Cu 0.70%
5.5–6 20 × 4 In 0.35%
5–5.5 20 × 4 La 0.14%
5.5–6 15 × 4 Co 0.06%
5.5–6 15 × 4 Cu 0.90%
5.5–6 15 × 4 Ni 0.07%
5.5–6 15 × 4 Cu 0.88%
4.5–5 15 × 4 Ag 0.98%
5.5–6 15 × 4 Cu 1.00%
5.5–6 25 × 4 Cu 0.65%
2
3 3
3 3
CoCu-ZSM-5/cordierite 25 Co(NO )
3 3
20 Cu(Ac)
2
NiCu-ZSM-5/cordierite 25 Ni(NO )
3 2
20 Cu(Ac)
2
AgCu-ZSM-5/cordierite 15 AgNO
3
20 Cu(Ac)
20 Cu(Ac)
2
2
Cu-AlTS-1/cordierite
In the previous work, zeolites with MFI structures such
as ZSM-5 and TS-1 had been successfully in situ syn-
thesized on the cordierite substrate and formed monolithic
MFI/cordierite [20,21]. The aim of this work is to present
MFI/cordierite monolith by an in situ synthesis method and
present a systematic study of nonnoble metal-exchanged
MFI/cordierite for SCR of NO with propane in the presence
of oxygen in a microreactor. The deNOx process is described
in detail and some factors that influence deNOx activities of
the catalysts are discussed. Through a comparison of vari-
ous metal-MFI/cordierites, Cu-ZSM-5/cordierite stands out
as a promising SCR catalyst due to its superior catalytic
activity and nitrogen selectivity. For practical purposes, Cu-
ZSM-5/cordierite was used to treat the exhaust from a real
lean-burn engine. Main pollutants in the exhaust—NOx, hy-
drocarbons, and CO—were abated simultaneously.
After 6 h vigorous stirring, the gained transparent liquid and
cordierite were put together into a PTFE-lined autoclave for
static crystallization at 180 ◦C for 90 h. Then the samples
were taken out, washed by distilled water, dried at 100 ◦C,
and then calcined at 550 ◦C in the air for 4 h to remove the
template.
2.1.3. Ion exchange
Metal-MFI/cordierite samples were prepared from as-
synthesized MFI/cordierites by a conventional solution ion-
exchange method described by Iwamoto et al. [22] at room
temperature. The concrete conditions of preparation and re-
sulting metal loadings of samples are given in Table 1. The
ion-exchanged catalysts were washed 5 times by an ultra-
sonic generator, 10 min each time, to remove the metal ions
adhering on the surface of catalysts, and then calcined at
550 ◦C for 4 h in air.
2. Experimental
2.2. Catalyst characterization
2.1. Catalysts preparation
The crystallinity of the synthesized MFI/cordierite was
given by XRD data collected with a Rigaku D/max 2500 dif-
fractometer, equipped with a graphite monochromator and
using Cu-Kα radiation.
SEM from a HITACHI X-650 scanning electron micro-
scope detected the surface features and degree of accumu-
lated homogeneity of cordierite substrate and MFI/cordierite
monolith. Surface areas and pore diameters were determined
by nitrogen adsorption on the ASAP 2010M facility (Mi-
cromeritics).
2.1.1. In situ synthesis of ZSM-5/cordierite
Silica sol was used as silicon source, and aluminum sul-
fate as aluminum source; no template was used. Silica sol,
sodium hydroxide, aluminum sulfate, and water were mixed
with this proportion: 1 Al2O3:84 SiO2:10 Na2O:3500 H2O.
After 2 h vigorous stirring, the gained transparent liquid and
cordierite (60 cells cm−2, 0.3 mm average wall thickness)
were put into a polytetrafluoroethylene (PTFE)-lined auto-
clave together for static crystallization at 180 ◦C for 16 h.
Then the samples were taken out, washed by distilled water,
and dried at 100 ◦C overnight.
The concentration of active metal components in the as-
synthesized samples was measured by ICP (IRIS Advantage,
TJA solution).
Chemical valence states of active metal components in
the catalysts before reaction were measured by XPS with a
PHI-5300 instrument with a Mg-Kα X-ray excitation source.
Binding energies accurate by 0.1 eV were determined with
respect to the position of the adventitious C 1s peak at
284.8 eV. The residual pressure in the analysis chamber was
maintained below 10−7 Pa during data acquisition.
The oxidative activities of active sites in the catalysts
were characterized by temperature-programmed reduction
2.1.2. In situ synthesis of AlTS-1/cordierite
Silica sol was used as silicon source and tetrabutylorthoti-
tanate (TBOT) as titanium source. Aluminum was leached
into the synthesis system from cordierite. Tetrapropylammo-
nium hydroxide (TPAOH) was used as template. Silica sol,
template, and water were mixed together and stirred for an
hour before TBOT was dropwise added in. The molar com-
position of the reagent is 30 SiO2:TiO2:3 TPAOH:600 H2O.

14
L. Li et al. / Journal of Catalysis 228 (2004) 12–22
using 5 vol% H2/Ar as reductant heating at 10 K min−1 from
100 to 800 ◦C. Prior to the reduction procedure the samples
were in situ calcined at 500 ◦C for 2 h. The consumption
of the reducing agent was measured by an on-line gas chro-
matograph equipped with a TCD.
2.3. Catalytic testing in microreactor
The SCR reaction of NO by propane at atmospheric pres-
sure was carried out in a fixed-bed flow microreactor. The
typical reactant gas mixture consists of NO (1000 ppm),
C3H8 (1000 ppm), H2O (0 or 10%), and O2 (5.0%) bal-
ance with He to 100%. The total flow of the inlet gas was
set at 60 mL min−1. The gas hourly space velocity (GHSV)
Fig. 1. Equipment of catalytic assessment experiments on real lean-burn
engine. Basic status of the engine in the evaluation experiment: Load =
−1
21 Nm; rotate speed = 1800 r min
.
employed in our experiment is about 36,000 mL g−1
h
−1
cat
(1.0 g sample containing about 10% catalyst powder and
90% cordierite substrate).
corresponding to the characteristic MFI structure peaks of
zeolites, which suggests that zeolite crystallites with MFI
structure have grown on the cordierite substrate in high qual-
ity.
The products were analyzed on-line using a gas chro-
matograph (HP 6890 series) equipped with a TCD detector.
A molecular sieve 5A column served for separation of N2,
O2, and CO, and a porapak Q column for separation of CO2,
C3H8, N2O, NO, and H2O. Simultaneously, the products
were continuously analyzed by a multigas sensor (Multor
610, Maihak GmbH, Germany) equipped with nondisper-
sive infrared channels for NO, CO, and SO2 registration and
a channel for electrochemical O2 detection as well as with
a NO2/NO converter reducing NO2 to NO over a molybde-
num catalyst. The difference between overall NOx and NO
concentration corresponds to the NO2 percentage.
In the real mobile exhaust environment, which contains
about 10% water vapor, the catalysts must endure the tem-
perature from the ambient to above 700 ◦C. Therefore, their
hydrothermal stability is one of the key factors for indus-
trial uses. For study, the synthesized MFI/cordierite samples
together with the pure MFI zeolite powders were treated in
high-temperature steam at 750 ◦C for 3 h and the XRD re-
sults are also displayed in Fig. 2. From the results, we can
see that the crystallinity of pure ZSM-5 zeolite powders
showed a decline of about 10% after hydrothermal treat-
ment, while no obvious change in crystallinity is observed
on ZSM-5/cordierite under the same hydrothermal treat-
ment conditions. As far as AlTS-1 zeolite powders are con-
cerned, about a 5% decline in crystallinity after hydrother-
mal treatment can be observed. Similar to ZSM-5/cordierite,
AlTS-1/cordierite showed good hydrothermal stability and
no decline in crystallinity observed after hydrothermal treat-
ment. As is known, the decline in crystallinity of MFI zeolite
during hydrothermal treatment is caused by a dealumina-
tion process and is highly influenced by the Si/Al ratio of
the zeolite. With much higher Si/Al ratios, AlTS-1 zeo-
lite powders show better hydrothermal stability than ZSM-5
zeolite powders studied. Compared to the pure zeolite pow-
ders, both ZSM-5/cordierite and AlTS-1/cordierite show
very good hydrothermal stability. The good hydrothermal
stability of zeolites on cordierite substrate comes from the
in situ synthesis method. Under our synthesis conditions,
some aluminum atoms of the cordierite become the base
for zeolite nucleation and may go into the framework of
the zeolites. (It is worthwhile to mention, no aluminum
source besides cordierite substrate was added in the synthe-
sis course of AlTS-1/cordierite. That is, all aluminum atoms
in AlTS-1 zeolite come from the aluminum in cordierite sub-
strate, which can be seen as a proof that aluminum atoms
of cordierite go into the framework of zeolite.) Strong in-
teractions exist not only between the aluminum atoms and
the zeolite but also between the aluminum atoms and the
2.4. Catalytic testing on lean-burn engine
The catalytic evaluation experiment was also carried out
on a 4-valve S.I. (spark ignited) gasoline-fueled engine, in
which, the controllable quasi-homogeneous mixture inside
the cylinder made by controllable injection realized fast
quasi-homogeneous mixture combustion. The catalyst was
fixed inside the vent pipe of the engine, heated by an elec-
trical heater. The products after catalytic reaction were ana-
lyzed on-line by an exhaust analyzer (MW56-AVL DIGAS
4000 LIGHT). The equipment of the catalytic assessment
experiments is shown in Fig. 1.
3. Results and discussion
3.1. Characterization of catalysts
The XRD method allows an easy and straightforward
judgment about the crystallinity and structure of the zeolite.
Both zeolites and cordierite substrates have their own typical
diffraction pattern. From the comparison of in situ synthe-
sized zeolite/cordierite, zeolite powder, and blank cordierite
samples shown in Fig. 2, we can see typical peaks at 7.88,
8.76, 23.04, 23.88, and 24.36◦ (d = 10.98, 9.86, 3.83, 3.70,
3.63 Å) in the XRD pattern of zeolite/cordierite samples

L. Li et al. / Journal of Catalysis 228 (2004) 12–22
15
Fig. 2. XRD patterns: (a) Blank cordierite; (b) ZSM-5 powder; (c) ZSM-5 after hydrothermal treatment; (d) ZSM-5/cordierite; (e) ZSM-5/cordierite after
hydrothermal treatment; (f) AlTS-1 powder; (g) AlTS-1/cordierite after hydrothermal treatment; (h) AlTS-1/cordierite; (i) AlTS-1/cordierite after hydrothermal
◦
treatment. Hydrothermal treatment condition: 750 C, 3 h.
cordierite substrate. The aluminum atoms are then greatly
stabilized. Thus, the dealumination process can be sup-
pressed and the hydrothermal stability of the MFI/cordierite
monolith is greatly improved.
zeolite load on support still involves synthesis of a thick ze-
olite layer where crystals are highly intergrown. Diffusion
rates of reactants into and product out of the zeolite layer
could be strongly lowered, therefore decreasing the accessi-
bility [23]. In this research, MFI zeolite loadings were con-
trolled at about 10% (wt%, obtained by the weight increase
after hydrothermal synthesis, template removed) under the
hydrothermal synthesis conditions. Thus, a layer of closely
packed individual zeolite crystals (as seen in Fig. 3) with
good accessibility could be preserved.
The surface of blank cordierite before in situ synthesis
and the surface of zeolite/cordierite after in situ synthesis
can be compared in the SEM images. It is found that sand-
wich ZSM-5 zeolite crystalloid (Fig. 3B) of about 5 µm and
cubical AlTS-1 zeolite crystalloid (Fig. 3C) of about 4 µm
has grown on the irregular external surface of cordierite
(Fig. 3A) compactly and homogeneously. In the internal
surface SEM images of ZSM-5/cordierite (Fig. 3D) and
AlTS-1/cordierite (Fig. 3E), we can clearly see small homo-
geneous crystalloids (< 2 µm) grown on the internal surface
of the zeolite substrate. The zeolite and the substrate phases
coupled tightly and became a whole. The cordierite sub-
strate itself has an irregular macroporous structure, but its
surface area is very small, less than 1 m2 g−1. This struc-
ture is utilized efficiently for the crystallization process by
application of the in situ synthesis method. The substrate is
used as a base for nucleation, and zeolites grow not only on
the surface of cordierite substrate but also inside the macrop-
orous. After the process of in situ synthesis, the surface areas
of samples increase greatly, while the surface area of zeo-
lite/cordierite monolith is about 110 m2 g−1. The increase of
the BET surface areas is essential to the catalytic reaction
and it provides more positions for the active metal compo-
nents.
3.2. Catalytic results
Catalytic properties of Cu-ZSM-5/cordierite, In-ZSM-5/
cordierite, and La-ZSM-5/cordier−it1e under dry conditions at
the space velocity of 36,000 mL gcat h
−1 within the tempera-
ture range 300–500◦C are shown in Fig. 4. Remarkably, the
only product from NO was nitrogen, neither N2O nor NO2
were formed in the whole reaction process for all the three
catalysts. There was some CO yield because of the incom-
plete oxidation of propane and the concentrations were also
shown in one of the figures. The SCR deNOx activities of
three catalysts all increased with the reaction temperature.
The highest NO conversion was obtained on Cu-ZSM-5/
cordierite followed by In-ZSM-5/cordierite and La-ZSM-5/
cordierite. Notable, the metal loadings were different in
Me-ZSM-5/cordierite. Metal loading (ion-exchange level)
may have an effect on the activity of metal ion-exchanged
ZSM-5 catalyst in deNOx reactions, but the effect was not so
great, especially at high ion-exchange levels. For example,
the catalytic activity, as indicated by maximum NO con-
version to N2, of Cu-ZSM-5 only decreased slightly with
the increasing Cu ion-exchange level [24,25]. In our re-
search, the Metal-ZSM-5/cordierite catalysts were prepared
from ZSM-5/cordierite by solution ion-exchange method for
a long time to get high ion-exchange level. Compared with
The accessibility in a gas-phase reaction is very important
for catalysts. Preparation of catalyst coatings by deposition
or using binder materials will result in a lack of accessibil-
ity unavoidably, originating diffusion limitations under re-
action conditions. By in situ synthesis, the zeolite crystals
nucleate directly on cordierite substrates and grew mainly
along certain axes. The order in crystal growth greatly im-
proves the accessibility of monolith. However, increasing

16
L. Li et al. / Journal of Catalysis 228 (2004) 12–22
Fig. 3. SEM images of the samples: (A) Surface of blank cordierite, (B) external surface of ZSM-5/cordierite, (C) external surface of AlTS-1/cordierite,
(D) internal surface of ZSM-5/cordierite, (E) internal surface of AlTS-1/cordierite.
Fig. 4. SCR of NO by propane over Cu-ZSM-5/cordierite ("), In-ZSM-5/cordierite (Q), and La-ZSM-5/cordierite (2) vs reaction temperatures. Dry conditions:
−1
−1
−1
.
1000 ppm NO, 1000 ppm C H , 5% O , balance He. Total flow rate = 60 mL min . GHSV = 36,000 mL g
h
3
8
2
cat
the great gap in catalytic behaviors of the three catalysts
(seen in Fig. 4), the effect of difference in cation loadings on
the catalytic activity became just a minor factor that could
be nearly ignored. So, we obtained the catalytic activity
order as Cu-ZSM-5/cordierite > In-ZSM-5/cordierite > La-
ZSM-5/cordierite, which is consistent with most of the work
reported [26]. Under dry conditions, Cu-ZSM-5/cordierite is
the best catalyst for its high deNOx activity and CO2 selec-
tivity.
Steam is one of the main components in flue gas and leads
to catalyst deactivation. Therefore, resistance of deNOx cat-
alysts to deactivation by water vapor is very important for
practical applications. The impact of steam on NO conver-
sion over the three catalysts is shown in Fig. 5. For Cu-
ZSM-5/cordierite and La-ZSM-5/cordierite, a decrease of
about 30% in NO conversion can be observed at 500 ◦C,
when 10% steam is added into the reaction system, while
for In-ZSM-5/cordierite, the impact of steam is much more

L. Li et al. / Journal of Catalysis 228 (2004) 12–22
17
Fig. 5. Catalytic performance of Cu-ZSM-5/cordierite (", !), In-ZSM-5/cordierite (Q, P), and La-ZSM-5/cordierite (2, 1) vs reaction temperatures in dry
(open symbols) and wet (filled symbols) conditions. Dry conditions: 1000 ppm NO, 1000 ppm C H , 5% O , balance He; Wet conditions: 1000 ppm NO,
3
8
2
−1
−1
−1
1000 ppm C H , 10% H O, 5% O , balance He. Total flow rate = 60 mL min . GHSV = 36,000 mL g
h
.
3
8
2
2
cat
Fig. 6. XPS spectra of as-synthesized Cu-ZSM-5/cordierite, In-ZSM-5/cordierite, and La-ZSM-5/cordierite.
serious and NO conversion shows a decrease of as much as
50%. It should be emphasized, however, that the overall de-
gree of NO conversion of In-ZSM-5/cordierite is still higher
than that of La-ZSM-5/cordierite.
According to preceding investigations [5–7,27–31] and
our experimental results, a supposed reaction mechanism
will be outlined as follows. From the preparation method and
posttreatments of the catalysts, active metal components In,
Cu, and La exist as In(III), Cu(II), and La(III) in the cata-
lysts which could easily be proven by XPS analysis (Fig. 6).
We suggested that these are the active sites in the three cata-
lysts studied. NOx(ads) and CxHyOz(ads) on the active sites
are thought as two important reaction intermediates. Once
formed, the NOx(ads) and CxHyOz(ads) will react with each
other immediately to form organonitrogen species [32–34]
and eventually give off nitrogen, H2O, and carbon oxides
(carbon monoxide or carbon dioxide) as seen in Fig. 7.
In the first step of deNOx reaction, NO and oxygen adsorb
on the active sites to produce O(ads) and NO(ads). Then,
strongly bound NOx species (mainly NO2) are formed on
the active sites from O(ads) and NO(ads). C3H8 cannot ad-
sorb on the active sites of the catalyst as strong as NO or
O2 and it will react with O(ads) to form CxHyOz(ads). The
formation processes of CxHyOz(ads) are simple and the for-
Fig. 7. Scheme of proposed reaction mechanism for NO/C H /O reaction
on Me-MFI/cordierite catalysts.
3
8
2
mation rates are nearly the same on all Me-ZSM-5/cordierite
samples. So the formation rates of NOx(ads) will have the
dominant influence on deNOx rates for it has been found
that the oxidation of NO(ads) (NO + O → NO2) is a slow
reaction [35]. But what influences and determines the rate
of NOx(ads) formation? The ability to adsorb NO and oxy-
gen is an important factor. Obviously, strong adsorption of
the gas reactants NO and O2 is necessary for NOx(ads) for-
mation. TPR measurements reveal that the oxidative activity
of the active metal component is a comparably important

18
L. Li et al. / Journal of Catalysis 228 (2004) 12–22
ite, the formation process of NOx(ads) on In(III) might pro-
ceed as it did on Cu(II). But the oxidative activity of In(III)
is not as strong as Cu(II), so the NOx(ads) formation rate
on In(III) was slower than that on Cu(II). As a consequence,
the catalytic activity of In-ZSM-5/cordierite is between that
of Cu-ZSM-5/cordierite and La-ZSM-5/cordierite. From the
discussion above, we can see that the active site must have
a rather strong oxidative activity, which can at least oxidize
NO(ads) to NOx(ads). However, a too strong oxidative ac-
tivity is not preferred. In that case, the reduced active site
would not be oxidized to the initial state by O(ads). Then
the formation process of NOx(ads) could not proceed con-
tinuously in the way of Cu-ZSM-5/cordierite. As discussed,
a proper redox activity of the active metal component is pre-
ferred to the formation NOx(ads), which is essential for the
deNOx process.
Fig. 8. H -TPR profiles of Cu-ZSM-5, In-ZSM-5, and La-ZSM-5.
2
When excess steam (10%) is added into the reaction sys-
tem, it can adsorb on the active site competing with the
small molecule gas reactants and hydrates the active sites.
Thus, the existent state of the active site is changed, leading
not only to a change in adsorption properties of gas reac-
tants but also in the oxidative activity. Generally speaking,
the adsorption of NO and O2 becomes more difficult and
the oxidative activity of the active site decreases, which sup-
presses the NOx(ads) formation and finally brings about the
decrease in NO conversion. As far as In-ZSM-5/cordierite
is concerned, In(III) active sites are easily hydrated forming
(In(OH)2)+ or InOOH species [42–44]. These newly formed
species have a much lower oxidative activity compared to
the original In(III) species. The hydrated In(III) species
(In(OH)2)+ or InOOH would not be able to oxidize NO(ads)
to NOx(ads), and, hence, NO(ads) could only be oxidized
by O(ads). The existence of excessive water vapor in the
reaction system causes the modification of NOx(ads) for-
mation over In-ZSM-5/cordierite, so that the effect of steam
on In-ZSM-5/cordierite is much stronger than observed for
the other two catalysts. Hence, In-ZSM-5/cordierite is not a
good catalyst for deNOx application.
Since the discovery of titanosilicate molecular sieve TS-1
in the early 1980s, there has been a large scientific ef-
fort on the synthesis, characterization, and utilization of
titanium-containing molecular sieves [45,46]. AlTS-1 has
the same MFI structure as ZSM-5, but nevertheless, it pos-
sesses some particular properties. Fig. 9 compares the con-
version of propane and NO to N2 for C3H8-SCR of NO
over Cu-AlTS-1/cordierite and Cu-ZSM-5/cordierite un-
der dry and wet conditions. As seen from the figure, Cu-
ZSM-5/cordierite exhibited higher NO and propane con-
version than Cu-AlTS-1/cordierite under both dry and wet
conditions. Notable, no by-products (N2O, NO2, and CO)
were detected in all processes for the two catalysts. To
make sure what causes the difference between the two Cu-
MFI/cordierite catalysts, comparison of their textural prop-
erties is shown in Table 3.
Table 2
Comparison of different processes of NO (ads) formation
x
Catalyst
NO (ads) formation process
x
Cu-ZSM-5/
cordierite
Cu(II) + O → Cu(II)–O
(1)
(2)
(3)
2
Cu(II) + NO → Cu(II)–NO → Cu(I)–NO
2
Cu(I)–NO + Cu(II)–O → Cu(II)–NO + Cu(II)
2
2
La-ZSM-5/
cordierite
La(III) + O → La(III)–O
(4)
(5)
(6)
2
La(III) + NO → La(III)–NO
La(III)–NO + La(III)–O → La(III)–NO + La(III)
2
factor. From H2-TPR profiles the different oxidative activ-
ities of active sites are obvious (Fig. 8). Cu-ZSM-5 shows
two peaks at 240 and 480 ◦C corresponding to the reduc-
tion of introzeolite Cu(II) to Cu(I) and Cu(I) to metallic Cu
[36,37]. In-ZSM-5 shows a reduction peak of introzeolite
In(III) at 390 ◦C [38] and La-ZSM-5 shows a reduction peak
of La(III) at 630 ◦C. Since a low reduction temperature is
equivalent to high oxidation capacity of the species involved,
obviously Cu(II) possesses the highest oxidation activity of
the three elements. Owing to the great difference between
the oxidative activities of active component in catalyst, two
different processes of NOx(ads) formation were supposed
and described in Table 2.
As far as Cu-ZSM-5/cordierite was concerned, NO(ads)
could be oxidized to NOx(ads) by Cu(II) and simultaneously
Cu(II) was reduced to Cu(I) (Eq. (2)). The formed Cu(I)
could be reoxidized to Cu(II) by the function of O(ads).
As a support of our opinion, the appearance of Cu(I) has
been detected in the SCR reaction by various in situ char-
acterization techniques [28,39–41]. On the other hand, for
La-ZSM-5/cordierite, the oxidative activity of La(III) was
not strong enough to oxidize NO(ads) to NOx(ads), so that
NO(ads) could only be oxidized by O(ads) as shown in
Eq. (6). Comparing the two processes of NOx(ads) forma-
tion, it is believed that NOx(ads) formed much more eas-
ily on Cu-ZSM-5/cordierite than on La-ZSM-5/cordierite.
Thus, Cu-ZSM-5/cordierite exhibited much higher catalytic
activity than La-ZSM-5/cordierite. As for In-ZSM-5/cordier-
It is clear that the two samples have similar BET surfaces
and copper loadings. The main difference of the two samples

L. Li et al. / Journal of Catalysis 228 (2004) 12–22
19
Fig. 9. SCR of NO by propane over Cu-ZSM-5/cordierite (", !), Cu-AlTS-1/cordierite (2, 1) vs reaction temperatures in dry (open symbols) and wet con-
−1
−1
−1
.
ditions (filled symbols): 1000 ppm NO, 1000 ppm C H , 0 or 10% H O, 5% O , balance He. Total flow rate = 60 mL min . GHSV = 36,000 mL g
h
3
8
2
2
cat
Table 3
Comparison of textural properties of Cu-MFI/cordierite
Sample Elemental BET Cu
surface area loading
Acid
a
b
analysis
site
2
−1
)
−1
(m g
(%)
(mmol g
)
Cu-AlTS-1/cordierite Si/Ti = 58 115
0.65
0.44
Si/Al = 72
Cu-ZSM-5/cordierite Si/Al = 30 106
0.70
0.18
a
Data gained from ICP, cordierite substrate excluded. Samples for ICP
analysis were zeolite powders separated from cordierite substrate by ultra-
sound.
b
Data determined by NH -TPD, cordierite substrate excluded.
3
consists in the incorporation of different atoms (Al and Ti)
into zeolitic framework and associated change of Brønsted
acidity. Brønsted acidity is considered as an essential feature
of good catalysts for HC-SCR and the carbonaceous deposits
will be favored by Brønsted acid sites [47–50]. In our ex-
periments, the formation rate of the reaction intermediate
CxHyOz(ads) will be accelerated by higher Brønsted acid-
ity. As a consequence, Cu-ZSM-5/cordierite shows much
higher deNOx catalytic activity than Cu-AlTS-1/cordierite
under the same reaction conditions.
Fig. 10. Ratio between converted NO and propane on Cu-ZSM-5/cordierite
(", !), Cu-AlTS-1/cordierite (2, 1) in dry (open symbols) and wet condi-
tions (filled symbols): 1000 ppm NO, 1000 ppm C H , 10% H O, 5% O ,
3
8
2
2
−1
−1
−1
balance He. Total flow rate = 60 mL min . GHSV = 36,000 mL g
h
.
cat
ZSM-5/cordierite is given in Fig. 10 to certify it. Cu-AlTS-1/
cordierite shows a lower ratio of NO/C3H8 converted than
Cu-ZSM-5/cordierite within the entire temperature range in-
vestigated under both dry and wet conditions due to the
oxidation activity provided by titanium in the zeolite frame-
work.
In our deNOx system over Cu-MFI/cordierite catalysts,
there are two reaction pathways competing for the propane
oxidation:
In the design of deNOx catalysts, the selection of support
materials for main active components is very important. As
discussed above, the Brønsted acidity and oxidative activity
of the supports are two key factors that can influence the final
deNOx activities of the catalysts.
The bimetal catalyst concept was introduced in the SCR
deNOx study to improve the performance of deNOx cata-
lysts by adding certain promoters [55–57]. Recently, some
publications have appeared which reported the enhancement
of deNOx activity by adding transition metal cocations into
catalysts [58–60]. Therefore, various transition metals (Ni,
Ag, and Co) were added into Cu-ZSM-5/cordierite through
the solution ion-exchange method (Table 1). The catalytic
performance of the obtained bimetallic catalysts is shown
C3H8 + 2NO + 4O2 → N2 + 3CO2 + 4H2O,
C3H8 + 5O2 → 3CO2 + 4H2O.
(1)
(2)
Reaction (1) is the main reaction in the deNOx process and
reaction (2) is the side reaction to be avoided. Titanium
in the framework of TS-1 can provide oxidation centers in
many reactions [51–53] and the oxidation ability of differ-
ent Ti species in TS-1 zeolite has recently been proved by
Zhuang et al. [54]. Obviously, oxidation centers also exist
in AlTS-1 zeolites. With the oxidation centers provided by
AlTS-1, the reaction intermediate CxHyOz(ads) tends to be
further oxidized by oxygen (probably adsorbed on oxidation
centers) according to reaction (2). The ratio between con-
verted NO and propane over Cu-AlTS-1/cordierite and Cu-

20
L. Li et al. / Journal of Catalysis 228 (2004) 12–22
in Fig. 11. Similar propane conversion levels were achieved
over the three bimetallic catalysts but different NOx conver-
sions. The highest NOx conversion was obtained over NiCu-
ZSM-5/cordierite followed by AgCu-ZSM-5/cordierite and
CoCu-ZSM-5/cordierite. Unexpectedly, only the cocation Ni
showed a weak promotion effect on NOx conversion in the
temperature range of 350 to 450 ◦C when added into Cu-
ZSM-5/cordierite. The function of cocation in the catalysts is
quite complex. At least, the oxidative activity and adsorption
ability to gas reactants as well as the acidity of the supports
can be influenced by the addition of cocations. The latter
conclusion is supported at least for Ni by recent reports in the
literature [49,61]. Further studies on the bimetallic deNOx
catalysts are being done.
and their concentrations are also displayed on the left top
of the figure. Hydrocarbons and carbon monoxide in the
exhaust acted as the reductants directly to selectively re-
duce NOx. An average NOx conversion of more than 30%
is attained between 250 and 550 ◦C, with maximum perfor-
mance of 45% between 375 and 450 ◦C. This means that a
purification of the exhaust from main pollutants like NOx,
hydrocarbons, and carbon monoxide is achieved to an ap-
preciable extent under these “road” conditions. The NOx
conversion is lower than that in the microreactor over the
same Cu-ZSM-5/cordierite catalyst even at a lower space
velocity. In our opinion, the concentration of hydrocarbons,
which plays a primary role in reducing NOx at low temper-
atures, has a great effect on the NOx conversion. Lack of
reductants causes a corresponding decline of NOx conver-
sion. Thus, to make full use of the reductants in the exhaust,
carbon monoxide is especially of great significance to prac-
tical SCR deNOx techniques.
For practical purposes, catalyst testing was also per-
formed with lean-burn engine exhaust. Fig. 12, left, displays
the status of the lean-burn engine exhaust treated by cata-
lyst Cu-ZSM-5/cordierite. The components of the exhaust
Fig. 11. Catalytic performance of Cu-ZSM-5/cordierite (!), CoCu-ZSM-5/cordierite ("), AgCu-ZSM-5/cordierite (Q), and NiCu-ZSM-5/cordierite (2) vs
−1
reaction temperatures. Wet conditions: 1000 ppm NO, 1000 ppm C H , 10% H O, 5% O , balance He. Total flow rate = 60 mL min . GHSV =
3
8
2
2
−1
−1
.
36,000 mL g
h
cat
Fig. 12. Conversions of three main components in exhaust gas vs temperature, catalyzed by Cu-ZSM-5/cordierite (left) and results of a duration study of
Cu-ZSM-5/cordierite on real lean-burn engine (right). (a) Catalyst after 5 h reaction; (b) catalyst after 10 h reaction; (c) catalyst after 15 h reaction; (d) catalyst
−1
after 20 h. Working status of engine: A/F = 21; GHSV = 25,000 h
.

L. Li et al. / Journal of Catalysis 228 (2004) 12–22
21
Durability and poison-resistant properties are important
for catalysts. The Cu-ZSM-5 catalyst usually suffers from
high deactivation in engine tests [62,63]. Therefore, we com-
pared the curves of NOx conversion changing with temper-
atures after a period of continuous operation (Fig. 12, right).
The four curves (a, b, c, d) were quite similar. The NOx
conversion and the temperature at which maximal NOx con-
version was reached obviously did not change. Compared
with Cu-ZSM-5 powders, monolith Cu-ZSM-5/cordierite
has much better durability and poison-resistant capabili-
ties. The differences between Cu-ZSM-5/cordierite and Cu-
ZSM-5 powders are associated with the cordierite substrate
and the in situ synthesis method. It is known that copper
species in Cu-ZSM-5 tend to aggregate into clusters with
concomitant dealumination upon thermal stress in excess
water vapor [64]. As discussed above (Fig. 2), the dealumi-
nation process could be suppressed efficiently by the in situ
synthesis method. Thus, agglomeration of copper species
might be prevented effectively and high dispersion could
be obtained. XRD patterns of Cu-ZSM-5/cordierite after re-
action did not exhibit peaks at 2.52 and 2.32 Å, typical of
CuO crystallites, clearly indicating a good dispersion of the
active phase [65]. With high dispersion of active metal com-
on a lean-burn engine. Hydrocarbons and carbon monoxide
in the exhaust were directly used as reductants for NOx re-
duction. Consequently, NOx, residual hydrocarbons and car-
bon monoxide are removed simultaneously to an appreciable
extent. Good durability and poison resistance of the catalyst
were attained. Cu-ZSM-5/cordierite is a promising catalyst
for NOx abatement in automobile exhaust under permanent
lean conditions.
Acknowledgments
This work was financially supported by DAAD (Ger-
many) and National Natural Science Foundation of China
(20233030).
References
[1] H. Bosch, F. Jassen, Catal. Today 2 (1988) 369.
[2] H. Scheider, U. Scharf, A. Wokaun, A. Baiker, J. Catal. 147 (1994)
545.
[3] M.M. Zwinkels, S.G. Jara, P.G. Menon, Catal. Rev.-Sci. Eng. 35
(1993) 319.
[4] Taylor, Catal. Rev.-Sci. Eng. 35 (1993) 457.
[5] M. Shelef, Chem. Rev. 95 (1995) 209.
[6] D.B. Lukyanov, E.A. Lombardo, G.A. Sill, J.L. d’Itri, W.K. Hall,
J. Catal. 163 (1996) 447.
ponents, the deactivation by sulfur (CuO + SO2 + 1 O2 +
2
nH2O → CuSO4 · nH2O) [66] might effectively be pre-
vented.
[7] R. Hernández-Huesca, J. Santamaría-González, P. Braos-García,
P. Maireles-Torres, E. Rodríguez-Castellón, A. Jiménez-López,
Appl. Catal. B 29 (2001) 1.
4. Conclusion
[8] H. Uchida, K. Yamaseki, I. Takahashi, Catal. Today 29 (1996) 99.
[9] H. Ohtsuka, T. Tabata, Appl. Catal. B 21 (1999) 133.
[10] M. Misono, Y. Nishizaka, M. Kawamoto, H. Kato, Stud. Surf. Sci.
Catal. 105 (1997) 1501.
[11] C. Descorme, P. Gélin, C. Lécuyer, M. Primet, J. Catal. 177 (1998)
352.
[12] S.C. Shen, S. Kawi, Appl. Catal. B 45 (2003) 63.
[13] M. Iwamoto, N. Mizuno, J. Auto. Eng. 207 (1993) 23.
[14] A.Z. Ma, M. Muhler, W. Grünert, Appl. Catal. B 27 (2000) 37.
[15] Z. Schay, L. Guczi, A. Becka, I. Nagy, V. Samuel, S.P. Mirajkar, A.V.
Ramaswamy, G. Pál-Borbély, Catal. Today 75 (2002) 393.
[16] H. Hamada, Catal. Surv. 1 (1997) 53.
[17] R.M. Heck, S. Gulati, R.J. Farrauto, Chem. Eng. J. 82 (2001) 149.
[18] A. Danner, K.K. Unger, Chem. Ing. Techn. 62 (1990) 487.
[19] G.B.F. Seijger, O.L. Oudshoorn, A. Boekhorst, H. van Bekkum, C.M.
van den Bleek, H.P.A. Calis, Chem. Eng. Sci. 56 (2001) 849.
[20] N.J. Guan, X.L. Shan, K. Zhang, D.S. Wang, S.H. Xiang, in: Proceed-
ings of the 12th International Zeolite Conference, Baltimore, 1998,
p. 2803.
By in situ synthesis, ZSM-5 and AlTS-1 zeolites with
MFI structure grew on the surface of honeycomb cordierite
substrate and formed MFI/cordierite monoliths. Strong in-
teraction was shown to develop between the zeolite and
the substrate, leading to improved hydrothermal stability
of the zeolite. A series of nonnoble metal ion-exchanged
MFI/cordierite materials were studied as catalysts for the
selective reduction of nitric oxide under lab-scale and bench-
scale conditions with good results. In the study of deNOx
processes, NOx(ads) and CxHyOz(ads) were identified to be
important reaction intermediates and the formation of these
intermediates could be shown to be a key step, determining
the deNOx rates. From this picture, the following four fac-
tors of the catalyst predominantly influence or determine the
catalytic activity:
[21] N.J. Guan, Y.S. Han, Chem. Lett. (2000) 1084.
[22] M. Iwamoto, H. Yahiro, Y. Mine, S. Kagawa, Chem. Lett. (1989) 213.
[23] M.A. Ulla, E. Miro, R. Mallada, J. Coronas, J. Santamaría, Chem.
Commun. (2004) 528.
[24] C. Torre-Abreu, M.F. Ribeiro, C. Henriques, F.R. Rabeiro, Appl. Catal.
B 11 (1997) 383.
[25] F. Seyedeyn-Azad, D. Zhang, Catal. Today 68 (2001) 161.
[26] Y. Traa, B. Burger, J. Weitkamp, Micropor. Mesopor. Mater. 30
(1999) 3.
[27] J.Y. Yan, H.H. Kung, W.M.H. Sachtler, M.C. Kung, J. Catal. 175
(1998) 294.
1. Adsorption ability of the active metal component toward
gaseous reactants.
2. Oxidative activity of the active metal components.
3. Brønsted acidities of the supports.
4. Oxidative activities of the supports.
Considering these factors, Cu-ZSM-5/cordierite, among
all investigated catalysts, revealed the most outstanding ac-
tivity and high selectivity both to N2 and to CO2. For practi-
cal relevance, Cu-ZSM-5/cordierite was also used as catalyst
[28] D.J. Liu, H.J. Robota, J. Phys. Chem. B 103 (1999) 2755.

22
L. Li et al. / Journal of Catalysis 228 (2004) 12–22
[29] N. Mongkolsiri, P. Praserthdam, P.L. Silveston, R.R. Hudgins, Chem.
Eng. Sci. 55 (2000) 2249.
[30] R. Burch, J.P. Breen, F.C. Meunier, Appl. Catal. B 39 (2002) 283.
[31] S. Bennici, A. Gervasini, N. Ravasio, F. Zaccheria, J. Phys. Chem.
B 107 (2003) 5168.
[47] G.P. Ansell, A.F. Diwell, S.E. Golunski, J.W. Hayes, R.R. Rajaram,
T.J. Truex, A.P. Walker, Appl. Catal. B 2 (1993) 81.
[48] A.P. Walker, Catal. Today 26 (1995) 107.
[49] B.I. Mosqueda-Jiménez, A. Jentys, K. Seshan, J.A. Lercher, J. Catal.
218 (2003) 348.
[50] B.I. Mosqueda-Jiménez, A. Jentys, K. Seshan, J.A. Lercher, Appl.
Catal. B 43 (2003) 105.
[51] A.M. Prakash, Larry Kevan, J. Catal. 78 (1998) 586.
[52] G.X. Lu, H.X. Gao, J.H. Suo, S.B. Li, J. Chem. Soc., Chem. Commun.
(1994) 2423.
[53] S.V.N. Raju, T.T. Upadhya, S. Ponrathnam, T. Daniel, A. Sudalai,
Chem. Commun. (1996) 1969.
[54] J.Q. Zhuang, D. Ma, Z.M. Yan, F. Deng, X.M. Liu, X.W. Han, X.H.
Bao, X.W. Liu, X.W. Guo, X. Wang, J. Catal. 221 (2004) 670.
[55] M.J. Rokosz, A.V. Kucherov, H.-W. Jen, M. Shelef, Catal. Today 35
(1997) 65.
[56] J.Y. Yan, W.M.H. Sachtler, H.H. Kung, Catal. Today 33 (1997) 279.
[57] Y. Yokomichi, T. Yamabe, T. Kakumoto, O. Okada, H. Ishikawa,
Y. Nakamura, H. Kimura, I. Yasuda, Appl. Catal. B 28 (2000) 1.
[58] T. Liese, W. Grünert, J. Catal. 172 (1997) 34.
[59] M. Haneda, Y. Kintaichi, H. Hamada, Catal. Today 54 (1999) 391.
[60] Z.M. Liu, J.M. Hao, L.X. Fu, T.L. Zhu, J.H. Li, X.Y. Cui, Appl. Catal.
B 48 (2004) 37.
[32] C. Yokoyama, M. Misono, J. Catal. 150 (1994) 9.
[33] B.J. Adelman, T. Beutel, G.D. Lei, W.M.H. Sachtler, J. Catal. 158
(1996) 327.
[34] T. Beutel, B. Adelman, W.M.H. Sachtler, Catal. Lett. 37 (1996) 125.
[35] M. Iwamoto, T. Zengyo, A.M. Hernandez, H. Araki, Appl. Catal. B 17
(1998) 259.
[36] J. Sarkany, J.L. d’Itri, W.M.H. Sachtler, Catal. Lett. 16 (1992) 241.
[37] C. Torre-Abreu, M.F. Ribeiro, C. Henriques, G. Delahay, Appl. Catal.
B 14 (1997) 261.
[38] E.E. Miró, L. Gutiérrez, J.M. Ramallo López, F.G. Requejoy, J. Catal.
188 (1999) 375.
[39] A.V. Kucherov, J.L. Gerlock, H.W. Jen, M. Shelef, J. Catal. 152 (1995)
63.
[40] H. Praliaud, S. Mikhailenko, Z. Chajar, M. Primet, Appl. Catal. B 16
(1998) 359.
[41] C. Dossi, A. Fusi, G. Moretti, S. Recchia, R. Psaro, Appl. Catal. A 188
(1999) 107.
[42] H. Berndt, F.W. Schutze, M. Richter, T. Sowade, W. Grünert, Appl.
Catal. B 40 (2003) 51.
[61] B.I. Mosqueda-Jiménez, A. Jentys, K. Seshan, J.A. Lercher, Appl.
Catal. B 46 (2003) 189.
[43] F.W. Schütze, H. Berndt, M. Richter, B. Lücke, C. Schmidt, T. Sowade,
W. Grünert, Stud. Surf. Sci. Catal. 135 (2001) 135.
[44] T. Sowade, C. Schmidt, F.W. Schütze, H. Berndt, W. Grünert, J. Ca-
tal. 214 (2003) 100.
[45] B. Notari, Adv. Catal. 41 (1996) 253.
[46] G.N. Vayssilov, Catal. Rev.-Sci. Eng. 39 (1997) 209.
[62] M.J. Heimrich, M.L. Deviney, SAE Paper (1994) 930736.
[63] J.N. Armor, Appl. Catal. B 4 (1994) N18.
[64] R.A. Grinsted, H.W. Jen, C.N. Montreuil, Zeolites 13 (1993) 602.
[65] J.A. Anderson, C. Márquez-Álvarez, M.J. López-Muñoz, I. Rodrý-
guez-Ramos, A. Guerrero-Ruýz, Appl. Catal. B 14 (1997) 189.
[66] F.C. Meunier, J.R.H. Ross, Appl. Catal. B 24 (2001) 23.