Mechanistic study of the influence of surface acidity on lean NO 2 reduction by propane in HZSM-5

Article abstract of DOI:10.1016/j.jcat.2005.02.022

This study focuses on the mechanism of lean NO_x reduction by propane, over acidic zeolites (HZSM-5), and the influence of surface acidity. In situ FTIR measurements of ammonia adsorption indicate a higher number of Bronsted acid sites for a sample with a low SiO₂/Al ₂O₃ ratio. The activity for NO_x reduction and the selectivity for N₂ formation correlate well with the Bronsted acidity. Step-response experiments with NO₂ and propane show the formation of surface-bound NO⁺, isocyanate, unsaturated hydrocarbons, and amine species. Formation of the latter two seems to be closely related to the Bronsted acidity. In the NO₂ reaction with propane, the NO⁺ species seem to play a vital role, probably reacting with carbenium ions (from propane cracking) to form isocyanates, which may be hydrolysed to amine species. Step-response experiments with isopropylamine and NO₂ indicate a fast reaction, where the amine and NO⁺ species react over Bronsted acid sites. Hence, amine species are possible reaction intermediates in the lean reduction of NO₂ by saturated hydrocarbons, such as propane, over HZSM-5.

Full text of DOI:10.1016/j.jcat.2005.02.022

Journal of Catalysis 232 (2005) 68–79
www.elsevier.com/locate/jcat
Mechanistic study of the influence of surface acidity on lean NO2
reduction by propane in HZSM-5
a,b,∗
a,c
a,b
a,b
Hanna Härelind Ingelsten
, Dongmei Zhao , Anders Palmqvist , Magnus Skoglundh
a
Competence Centre for Catalysis, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
b
Applied Surface Chemistry, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
c
Department of Chemistry, Göteborg University, SE-412 96 Göteborg, Sweden
Received 22 December 2004; revised 24 February 2005; accepted 28 February 2005
Available online 7 April 2005
Abstract
This study focuses on the mechanism of lean NOx reduction by propane, over acidic zeolites (HZSM-5), and the influence of surface acid-
ity. In situ FTIR measurements of ammonia adsorption indicate a higher number of Brønsted acid sites for a sample with a low SiO /Al O
2
2 3
ratio. The activity for NOx reduction and the selectivity for N formation correlate well with the Brønsted acidity. Step-response experiments
2
+
with NO and propane show the formation of surface-bound NO , isocyanate, unsaturated hydrocarbons, and amine species. Formation
of the latter two seems to be closely related to the Brønsted acidity. In the NO reaction with propane, the NO species seem to play a
2
+
2
vital role, probably reacting with carbenium ions (from propane cracking) to form isocyanates, which may be hydrolysed to amine species.
+
Step-response experiments with isopropylamine and NO indicate a fast reaction, where the amine and NO species react over Brønsted
2
acid sites. Hence, amine species are possible reaction intermediates in the lean reduction of NO by saturated hydrocarbons, such as propane,
2
over HZSM-5.

2005 Elsevier Inc. All rights reserved.
Keywords: HZSM-5; Lean NOx reduction; Propane; Isopropylamine; Surface acidity; FTIR
1
. Introduction
Platinum-based catalysts have been shown to be highly
active towards NOx reduction [15,16], with the use of HC-
SCR. The high activity, unfortunately, is accompanied by
low selectivity for N2 formation [15]. In the presence of
acidic groups on the surface of the support material, the N2
selectivity can be enhanced [17]. Furthermore, it has been
found that saturated hydrocarbons, such as propane, can be
activated on the catalyst surface if the support material is
modified to contain strong acidic sites [18,19] or by the in-
troduction of surface sulphates [20–25]. Acke et al. studied
the lean reduction of NOx by C3H6, C3H8, NH3, and HNCO
over Pt supported on γ -Al2O3, ZSM-5, or SiC [7–11] and
found that the NOx reduction is affected by the platinum
support material. Both the activity for NOx reduction and the
selectivity for N2 formation were lowest for the samples sup-
ported on SiC, that is, the most inert support material. Previ-
Diesel and lean-burning engines are favourable engine
concepts because of their relatively good fuel economy,
which leads to less CO2 formation. However, the lean en-
vironment in the exhaust obstructs the reduction of nitro-
gen oxides (NOx) with the conventional three-way catalyst
(
TWC), since this technique requires stoichiometric condi-
tions to achieve satisfactory NOx abatement. This calls for
more efficient techniques for lean NOx reduction. Several
solutions are possible, such as NOx storage [1–3] and se-
lective catalytic reduction with ammonia (NH3-SCR) [4–6]
or hydrocarbons (HC-SCR) [7–14] as the reducing agent for
NOx.
*
ously we have investigated lean NO reduction, by propene
Corresponding author. Fax: +46 31 16 00 62.
x
E-mail address: hannahi@chem.chalmers.se (H.H. Ingelsten).
or propane, over platinum supported on aluminum-silicates
0021-9517/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.02.022

H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
69
[
26] and over (Pt free) aluminum-silicates [27]. When plat-
fused into the tube 28 cm from the upper end). The reactor
was heated by a tubular furnace (controlled by Eurotherm
818S), and the temperature was measured with a thermo-
couple (K-type) located inside the reactor, 2 mm above
the catalyst bed. Gases, He (99.9996%), O2 (99.9996%),
NO2 (0.983% in He), and C3H8 (2% in He), were intro-
duced via separate mass flow controllers. Reactant and prod-
uct gases were analysed on-line. NO and NO2 were de-
termined with an Eco Physics, CLD 569 EL ht chemilu-
minescence analyser. An Agilent 6850 gas chromatograph
fitted with a HP-PLOT Molesieve column and a thermal
conductivity detector was used to quantify N2. Water, CO2,
NOx, and unreacted hydrocarbons were removed by a liquid
nitrogen-cooled cold trap, located upstream of the gas chro-
matograph. CO, CO2, and N2O were measured by photo-
acoustic FTIR spectroscopy, with a Bruel & Kjaer Type 1301
analyser. Other possible nitrogen-containing reaction prod-
ucts, if present, were detected by the same technique. The
catalyst samples were pretreated in 10% O2 balanced with
inum was present in the samples we found a strong connec-
tion between the Brønsted site density and the selectivity
for N2 formation [26], even though the relation to the ac-
tivity for NOx reduction was not as clear. However, for the
aluminium-silicate samples we found a clear correlation be-
tween the amount of Brønsted acid sites and the NOx reduc-
tion activity, for propane as the reducing agent [27]. Thus,
previous results illustrate the important role of the support
material for lean NOx reduction and emphasise the connec-
tion between the activity for NOx reduction, N2 selectivity,
and surface acidity.
Since NO2 is more easily reduced than NO, one possible
NOx reduction route involves an initial oxidation of NO to
NO2. Some NOx reduction systems, such as the dual pore
system, utilise this concept, starting by selective oxidation of
NO [28,29]. The subsequent NO2 reduction step in the dual
pore system occurs over acidic zeolites, where the acidity
seems to be the key factor for the process to occur.
◦
In this work we have focused on the continuous catalytic
reduction of NO2 by propane over acidic zeolites (HZSM-5)
with varying SiO2/Al2O3 ratio. The objective was to study
the mechanism of lean NOx reduction by propane and its
correlation with the surface acidity.
He (30 min at 500 C), and the gas composition used in the
experiments was 10% O2, 1040 ppm NO2, and 300 ppm
C3H8, balanced with He to maintain a total gas flow rate of
1500 ml/min, corresponding to a space velocity of 50,000
−
1
h
5
. Steady-state measurements were performed at 650,
◦
50, 450, and 350 C.
2
. Experimental
2.2.2. FTIR spectroscopy
The in situ FTIR (Fourier transform infrared) spec-
troscopy measurements were carried out with a BioRad FTS
2
.1. Catalyst preparation
6
000 spectrometer in diffuse reflectance (DRIFT) mode.
The catalysts used in this study were HZSM-5 zeolites
Akzo Nobel Catalysts BV) with varying SiO2/Al2O3 molar
The setup is described elsewhere [27,30]. The calcined ze-
olite samples were placed in the DRIFT cell. The gases,
Ar (99.995%), O2 (99.95%), NO2 (5000 ppm in Ar), C3H8
(1% in Ar), and NH3 (4% in Ar), were introduced via mass
flow controllers (Bronkhorst Hi-Tech) to the DRIFT cell.
The samples were initially pretreated in nitrogen dioxide
and oxygen (1000 ppm NO2 and 10% O2, balanced with Ar,
(
ratio (Table 1). To vary only the surface acidity, the catalytic
samples are chosen to have similar textural properties, with
the exception of the SiO2/Al2O3 molar ratio. The zeolite
powders were pressed into discs, which were ground and
sieved to obtain a size fraction of 300 to 500 µm. The sam-
◦
◦
◦
ples where then calcined in air at 500 C for 1 h.
500 C, 40 min) followed by oxygen (10% O2 in Ar, 500 C,
◦
1
h) and then evacuated in pure Ar (500 C, 15 min) at a to-
tal flow rate of 300 ml/min (the flow rate was kept constant
2
2
.2. Activity and characterisation
throughout the experiments). Background spectra (40 scans
−1
at a resolution of 1 cm ) were collected for each temper-
ature, under Ar exposure. Fresh samples were used for all
experiments.
.2.1. NOx reduction activity
The activity and selectivity for NOx reduction with C3H8
were investigated in a vertical flow reactor (21 mm i.d., 58-
cm-long quartz tube with a thin sintered quartz filter disc
Ammonia adsorption experiments were carried out at
◦
4
50, 350, and 250 C to distinguish the acidity of the zeo-
lites. The samples were exposed to 1000 ppm NH3 in Ar
until saturation (about 10 min) and then flushed with pure
Table 1
Textural properties of the HZSM-5 zeolites used in the experiments
−1
Ar for 10 min. Spectra (20 scans at a resolution of 1 cm
were taken continuously during the experiments.
)
HZSM-5 40
HZSM-5 234
HZSM-5 569
Crystallinity (%)
SiO /Al O
95
98
100
569
0.03
0.183
377
Step-response experiments (with propane as a reducing
agent) were performed for two of the samples, HZSM-5 40
39.7
< 0.05
0.177
405
234
0.01
0.183
407
2
2 3
Na O (wt%)
2
◦
◦
a
MiPV (ml/g)
at 450 and 350 C and HZSM-5 234 at 450 C. The exper-
iment consisted of six sequences with different gas compo-
b
2
SA (m /g)
−1
a
sitions, and spectra (five scans at a resolution of 1 cm
where taken after 0, 20 s, and 1, 2, 3, 5, 7, 10, 15, and
)
Micro pore volume.
BET surface area.
b

70
H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
2
0 min (or until saturation was reached) for each sequence.
Sequence 1 started with a fresh sample (pretreated as men-
tioned above), and a mixture of 1000 ppm NO2 and 10% O2
(
0
balanced with Ar) was introduced to the sample cell (at time
s). When saturation was reached the gas composition was
changed and new spectra were collected. The subsequent se-
quences proceeded as follows:
Sequence 2 300 ppm propane was added (1000 ppm NO2,
3
00 ppm C3H8, 10% O2, balanced with Ar).
Sequence 3 The NO2 was turned off (300 ppm C3H8, 10%
O2, balanced with Ar).
Sequence 4 The NO2 was turned on (1000 ppm NO2, 300
ppm C3H8, 10% O2, balanced with Ar).
Sequence 5 The propane was turned off (1000 ppm NO2,
1
0% O2, balanced with Ar).
Sequence 6 The propane was turned on (1000 ppm NO2,
00 ppm C3H8, 10% O2, balanced with Ar).
3
◦
Fig. 1. FTIR spectra collected at steady-state during Ar flow at 250 C. The
measured signals of the baselines are separated by −0.02, 0.03 and 0.1 V,
For the HZSM-5 40 sample, step-response experiments
were also performed, with isopropylamine as the reducing
agent. The background spectra were collected under reac-
tion conditions (20 scans at a resolution of 4 cm ), and the
difference spectra were recorded, when either NO2 or iso-
respectively.
−1
◦
40 sample (steady state, 250 C) peaks at 3374, 3283, and
◦
−1
propylamine was turned off and on, at 450, 350, and 250 C,
1483 cm appear, which are attributed to asymmetric and
symmetric NH stretching and bending vibrations of NH4 at
−1
+
respectively (1 scan/s, 4 cm resolution). The experimental
setup was the same as that for the previous experiments, with
the exception of one additional Ar flow, which was saturated
with isopropylamine before the reaction cell. In addition,
the Brønsted acid sites [6,34,38,39]. The origin of the 1483
−1
cm peak is somewhat uncertain, however, since there are
−1
possible bands in this region (below 2000 cm ) connected
−
1
spectra were taken (five scans at a resolution of 1 cm
)
to perturbed vibration frequencies of the zeolite framework
when isopropylamine was pre-adsorbed on the sample, and
a mixture of 1000 ppm NO2 and 10% O2, balanced with Ar,
was subsequently introduced. Background spectra, 40 scans
+
[
6,37,40]. The presence of NH4 species at the Brønsted
acid sites is also strongly supported by the pronounced neg-
−1
ative peak that appears at 3607 cm , which is assigned to
−1
at a resolution of 1 cm , were collected under Ar exposure
directly after the pretreatment.
OH stretching vibrations of the Brønsted acid sites [6,34,39].
−1
Bands at 3248, 3208, and 2736 cm are also observed that
+
most likely are due to NH stretching vibrations of NH4
[
6,41] and in the latter case to weakly bound NH3 [6]. Dur-
3
. Results
ing the ammonia adsorption the presence of trace amounts
of water cannot be excluded, and the broad feature observed
in the region of 3500–2200 cm may be related to OH
stretching vibrations of water present at the Lewis acid sites
in the sample [6,37]. Increasing the temperature results in
an overall decrease in the peak heights. The peaks related
to NH4 species at Brønsted acid sites (3374, 3283, and
1
sites (3607 cm ) decrease slightly with increasing temper-
ature.
3
.1. Acidity characterisation
−1
The background spectrum for the HZSM-5 40 sample
(
collected under Ar) shows absorption peaks at 3737 and
−1
3
600 cm , which are assigned to OH stretching vibra-
+
tions originating from silanol groups (Si–OH) and Si(OH)Al
bridging hydroxyl groups, respectively [6,31–37] (Fig. 1).
The latter peak (3600 cm ) is typical for Brønsted acid sites
in zeolites. This peak is significantly weaker for the HZSM-
−1
483 cm ) and OH stretching vibrations of Brønsted acid
−1
−1
The intensities of the peaks in the NH stretching region
5
234 sample and is absent for the HZSM-5 569 sample.
−1
(
3400–3200 cm ), compared with the HZSM-5 40 sam-
The background spectra are indicative of the differences in
acidity between the different samples, where the HZSM-5 40
sample contains significant amounts of Brønsted acid sites in
the zeolite framework.
ple, are significantly lower for the HZSM-5 234 sample and
absent for the HZSM-5 569 sample. Furthermore, the peak
originating from OH stretching vibrations at 3607 cm
−1
To further investigate the acidity of the samples, FTIR
spectra were recorded during ammonia exposure at 450, 350,
and 250 C; the results are shown in Fig. 2. For the HZSM-5
(Brønsted acid sites) is significantly weaker for the HZSM-
5 234 sample and negligible for the HZSM-5 569 sam-
ple.
◦

H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
71
◦
Fig. 2. FTIR spectra collected at steady-state conditions during a flow of 1000 ppm NH in Ar (a) at 450 C over HZSM-5 40, HZSM-5 234, and HZSM-5
3
◦
569, respectively, and (b) at 250, 350, and 450 C over HZSM-5 40.
Fig. 3. (a) N formation (percentage of the sum of all N-containing species in the outlet gas flow), designated by open symbols and NOx reduction (percentage
2
of the sum of all N-containing species in the outlet gas flow) designated by filled symbols, during steady-state NO reduction experiments in oxygen excess,
2
using propane as reducing agent: circle, HZSM-5 40; square, HZSM-5 234; and triangle, HZSM-5 569. (b) C H consumption (percentage of the sum of all
3
8
C-containing species in the outlet gas flow) during steady-state NO reduction experiments in oxygen excess, using propane as reducing agent: circle, HZSM-5
2
40; square, HZSM-5 234; and triangle, HZSM-5 569.
3
.2. Activity studies
above. The formation of N2O is low for all of the samples
and does not exceed 1.5% at any temperature.
The influence of mass transfer limitations on the reduc-
tion of NOx is considered by estimation of the Weiszs modu-
lus (see Appendix A), and the results show that the catalytic
reaction is likely not influenced by mass transport limita-
tions.
Fig. 3 shows the NOx reduction, the N2 formation (in per-
centage of the sum of all N-containing species in the outlet
gas flow), and the C3H8 consumption (in percentage of the
sum of all C-containing species in the outlet gas flow) for the
different samples as a function of the reactor temperature.
The highest activities for NOx reduction and the formation
3.3. Mechanistic studies
◦
of N2 are obtained with the HZSM-5 40 sample at 450 C,
followed, in decreasing order of activity, by HZSM-5 234
and HZSM-5 569. For the latter two samples the maximum
N2 formation is achieved at higher temperatures, 550 C and
The results from the step-response experiments, per-
formed with the in situ FTIR equipment, are presented in
Figs. 4–6 and 8. Below 2000 cm the background noise
◦
−1

72
H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
◦
Fig. 4. Step-response experiments at 450 C, over HZSM-5 40 zeolite, using propane as the reducing agent. Light grey corresponds to direct after the gas
switch and time is then increased until steady-state is reached, which is illustrated with black. Sequence 1: 1000 ppm NO , 10% O in Ar; sequence 2: 1000
2
2
ppm NO , 300 ppm C H , 10% O in Ar; sequence 3: 300 ppm C H , 10% O in Ar; sequence 4: 1000 ppm NO , 300 ppm C H , 10% O in Ar; sequence
2
3
8
2
3
8
2
2
3
8
2
5
: 1000 ppm NO , 10% O in Ar; sequence 6: 1000 ppm NO , 300 ppm C H , 10% O in Ar.
2 2 2 3 8 2
level is significant, which is probably related to perturbed
vibration frequencies of the zeolite framework [6,37,40],
which make peak observation and assignment difficult in
this region.
the sample (sequence 2: 1000 ppm NO2, 300 ppm C3H8,
◦
+
and 10% O ), several peaks appear. At 450 C the NO
2
−1
peak decreases slightly with time, and a peak at 2260 cm
−1
with a small shoulder at 2275 cm appears. These peaks
are most probably due, respectively, to isocyanate (–NCO)
species bound to Al and Si in the zeolite framework [35,
3
.3.1. Reduction of NO2 by propane over HZSM-5 40
Step-response experiments for HZSM-5 40, with propane
◦
4
2
3–50] and/or nitrile species [35,43,46,47,51]. At 350 C the
−1
260 cm peak is present only as a small shoulder on the
as the reducing agent, are shown in Figs. 4 and 5. During the
first sequence (1000 ppm NO2 and 10% O2), one peak is ob-
+
−1
NO peak, and there is no sign of the shoulder at 2275 cm
Several peaks are observed in the 3000–2800 cm region,
which most likely can be assigned to CH stretching vibra-
tions [35,42,47]. A peak at 2968 cm in particular can be
attributed to adsorbed and/or gas-phase propane [52]. These
peaks (3000–2800 cm ) appear already after ca. 20 s and
seem to be constant throughout the sequence (for both tem-
.
−1
◦
◦
−1
served at 2133 cm (for both 450 and 350 C). At 350 C
this peak increases during the first minute of exposure, after
which it gradually decreases and reaches a steady-state level
after about 20 min, which has also been observed by Xin
et al. [42]. According to Hadjiivanov and co-workers [31],
the 2133 cm peak corresponds to NO species present at
the Brønsted acid sites in the zeolite. Since weak negative
peaks are observed around 3610 and 3745 cm , which are
assigned to OH stretching vibrations originating from Brøn-
sted acid sites and silanol groups, respectively [6,31–37], it
−1
−1
−1
+
−1
peratures). After about 1 min, a peak around 3130 cm
−1
◦
starts to grow (450 C) and increases until saturation is
reached. This peak may be due to CH stretching vibrations
of unsaturated hydrocarbons [46,47]. For the lower tempe-
+
◦
is probable that the NO species are present at the Brøn-
sted acid sites. When propane is brought into contact with
rature (350 C) this peak is significantly weaker. In the NH
−1
stretching region of 3400–3200 cm [35,46,47,53,54] a

H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
73
◦
Fig. 5. Step-response experiments at 350 C, over HZSM-5 40 zeolite, using propane as the reducing agent. Light grey corresponds to direct after the gas
switch and time is then increased until steady-state is reached, which is illustrated with black. Sequence 1: 1000 ppm NO , 10% O in Ar; sequence 2: 1000
2
2
ppm NO , 300 ppm C H , 10% O in Ar; sequence 3: 300 ppm C H , 10% O in Ar; sequence 4: 1000 ppm NO , 300 ppm C H , 10% O in Ar; sequence
2
3
8
2
3
8
2
2
3
8
2
5
: 1000 ppm NO , 10% O in Ar; sequence 6: 1000 ppm NO , 300 ppm C H , 10% O in Ar.
2 2 2 3 8 2
◦
small shoulder is observed at 450 C, which possibly can
be attributed to amine species. The negative peak around
3
state (for both temperatures). In the CH region the peaks de-
◦
◦
crease markedly at 350 C and disappear at 450 C, and the
propane peak disappears completely (for both temperatures).
−1
610 cm gradually becomes more negative during the
sequence with propane (more negative for 450 than for
50 C). This is indicative of blocking of the Brønsted acid
−
1
The peak at 3130 cm decreases but does not disappear,
◦
−1
3
and the 3610 cm peak becomes less negative. In the final
sites by adsorbed species, most likely hydrocarbons. In se-
quence 3, when NO2 is turned off (300 ppm propane and
sequence (sequence 6) propane is turned on again, and the
bands in the CH-region, including the propane peak and the
+
−1
−1
1
0% O2), the NO peak rapidly disappears and the –NCO
peak at 3130 cm , increase. The 3610 cm peak gradu-
ally becomes more negative, the isocyanate peak increases,
−1
peak becomes less intense. The 3610 cm peak is constant.
In the CH vibration region the magnitude of the peaks de-
creases, but the relative heights of the separate peaks are
constant. This may be due to less water formation, since
NO2 is a stronger oxidising agent than O2. The OH stretch-
ing vibrations of water, hydrogen-bonded to Lewis acid sites,
appear in the 3500–2200 cm region [6,37]. The changes
in the spectra are similar for the two temperatures, except
for the NH stretching region, where a shoulder is observed
◦
and the shoulder in the NH-region, which appears at 450 C,
becomes more distinct.
Sequences 2 and 6 should be similar since, in both cases,
propane is turned on after a sequence with only NO2 and
oxygen. There are some differences in the results, however,
which may be explained by the presence of C-containing
species on the surface before the sixth sequence starts. At
−1
◦
+
450 C the –NCO peak is higher than the NO peak in se-
◦
+
at 450 C, whereas there are no signs of NH vibrations at
50 C. When NO2 is turned on again (sequence 4), the NO
quence 6; however, in sequence 2 the NO peak is higher
◦
+
3
than the –NCO peak. Moreover, the shoulder in the NH re-
◦
peak develops and the –NCO peak increases slightly. The
signal in the CH stretching region increases (with the rela-
tive peak heights constant), likely because of increased water
formation. The shoulder in the NH region becomes more
gion (450 C) is more pronounced in sequence 6 than in
−1
sequence 2. The shoulder at 2260 cm , which appears in
sequence 2 at 350 C, becomes a small peak in sequence 6,
and the 3130 cm peak is increased slightly in sequence 6
compared with sequence 2.
◦
−
1
◦
−1
pronounced (450 C), and the 3610 cm peak is constant.
+
During sequence 5 (propane is turned off) the NO peak
is constant at 450 C but decreases slightly at 350 C. The
isocyanate peak decreases but is still significant at steady
The differences between the two temperatures indicate
more interaction on the surface at the higher temperature
◦
◦
◦
(450 C). The unsaturated hydrocarbon peak seems more

74
H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
◦
Fig. 6. Step-response experiments at 450 C, over HZSM-5 234 zeolite, using propane as the reducing agent. Light grey corresponds to direct after the gas
switch and time is then increased until steady-state is reached, which is illustrated with black. Sequence 1: 1000 ppm NO , 10% O in Ar; sequence 2:
2
2
1
000 ppm NO , 300 ppm C H , 10% O in Ar; sequence 3: 300 ppm C H , 10% O in Ar; sequence 4: 1000 ppm NO , 300 ppm C H , 10% O in Ar;
2 3 8 2 3 8 2 2 3 8 2
sequence 5: 1000 ppm NO , 10% O in Ar; sequence 6: 1000 ppm NO , 300 ppm C H , 10% O in Ar.
2
2
2
3
8
2
intense, and the Brønsted peak is more negative for this
temperature. The isocyanate peak is visible only as a small
(sequence 6) some small negative peaks can be observed
around 3610 and 3745 cm , indicating blocking of OH
sites.
−1
◦
shoulder, which develops to a small peak at 350 C, whereas
◦
at 450 C this peak is significantly more pronounced. Fur-
An overall observation for this sample is that the total
amount of adsorbates is much less than for the HZSM-5 40
thermore, there are no indications of NH vibrations at
◦
◦
−1
3
50 C. At 450 C, on the other hand, a small shoulder is
sample. In particular, the 3130 cm peak and the peaks in
observed in the NH region. These results correlate well with
the OH region are significantly less intense or do not appear
at all for some of the sequences. An interesting observa-
tion is that the –NCO peak increases (to approximately three
times the initial size) during the entire experiment. During
this experiment there are no signs of NH stretching vibra-
tions, which are observed at this temperature for the HZSM-
5 40 sample.
the activity study, where the NOx reduction activity is found
◦
to be higher at 450 than at 350 C.
3
.3.2. Reduction of NO2 by propane over HZSM-5 234
In Fig. 6 the results from the step-response experiments
◦
with propane at 450 C over the HZSM-5 234 sample can
be seen. The first sequence (with only NO2 and oxygen) is
similar to the experiment for HZSM-5 40, that is, one peak
3.3.3. Reduction of NO2 by isopropylamine over
HZSM-5 40
−
1
at around 2130 cm is observed. When propane is turned
−1
on (sequence 2), peaks at 2261, 3000–2800, and 3130 cm
evolve. The –NCO peak is relatively high compared with
the NO peak. In sequence 3, NO2 is turned off, the NO
Since there are signs of NH stretching vibrations dur-
ing the step-response experiments over HZSM-5 40, and the
presence of amine intermediates in the HC-SCR reaction is
debated in the literature [45,47,53–57], step-response exper-
iments with isopropylamine as the reducing agent for NO2
are performed. The results are summarised in Figs. 7 and 8.
Fig. 7 shows the reaction of pre-adsorbed isopropylamine
+
+
peak gradually disappears, and the isocyanate peak seems
constant. Turning on the NO2 again (sequence 4) results in
+
a developing NO peak and a small increase in intensity of
the peaks in the CH region. In sequence 5 the propane is
turned off and the CH stretching vibrations gradually dis-
appear. When the propane is turned on (sequence 6) these
peaks start to grow again. The signal in the OH region is very
noisy in these experiments; however, in the last sequence
◦
with NO2 and O2 (at 500 C). Initially, before the NO2 and
O2 are turned on, four main groups of peaks can be observed.
−1
The peaks at 2786, 2910, 2942, and 2987 cm are attributed
to (saturated) CH stretching vibrations [35,42,46,47], and

H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
75
◦
Fig. 7. Reaction of pre-adsorbed isopropylamine with 1000 ppm NO and 10% O (balanced with Ar) at 500 C. The time separation between the spectra is
2
2
approximately 10 min.
◦
Fig. 8. Step-response experiments at 450 C, over HZSM-5 40 zeolite, using isopropylamine as the reducing agent. Light grey corresponds to direct after the
gas switch and time is then increased until steady-state is reached, which is illustrated with black.
the peak at 3130 cm 1 originates most likely from (unsat-
−
When NO2 and oxygen are introduced into the sample cell,
the intensities of the CH stretching vibrations and the NH
stretching region decrease and eventually disappear. The OH
peaks gradually become less negative, approaching the zero
level; that is, the species that block the sites disappear. Peaks
−1
urated) CH vibrations [46,47]. One peak at 3275 cm , with
a shoulder around 3230 cm , is probably due to NH stretch-
ing vibrations from amine species [35,46,47,53,54], and the
negative peaks at 3603 and 3745 cm indicate blocking of
the Brønsted acid sites and silanol groups [6,31–37], pre-
sumably by hydrocarbon and/or amine species, respectively.
−1
−1
−1
at 2258, 2293, and 2133 cm start to grow. The latter is
+
likely due to NO species [31] formed on the surface as the

76
H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
isopropylamine is consumed. The peak at 2258 cm 1 proba-
−
The intensity of the NO peak is monitored as a func-
+
−1
bly relates to isocyanate species, and the peak at 2293 cm
to CO2 (in gas phase) [58], which is formed during the reac-
tion between NO2 and isopropylamine.
,
tion of time (Fig. 9), and the change is used to calculate the
apparent activation energy (Ea) for the reaction, from the
Arrhenius equation,
Step-response experiments with isopropylamine as a re-
ducing agent are shown in Fig. 8. The background spec-
tra are collected at reaction conditions (for each tempera-
ture). Nitrogen dioxide and isopropylamine are alternatively
turned on and off. Spectra are recorded every second during
k = Ae−
Ea/RT
,
(1)
where k is the rate constant, A is the pre-exponential fac-
tor, Ea is the activation energy, R is the gas constant, and T
is the temperature. With this expression the apparent activa-
tion energy (Ea) for the reaction, when NO2 is turned on,
is found to equal approximately 25 kJ/mol. The correspond-
ing value for Ea, when isopropylamine is turned on, is ca. 10
kJ/mol. These activation energies are relatively low, which
is also expected, since the reaction seems to proceed quickly.
◦
this process. When NO2 is turned off at 450 C, a negative
−
1
peak develops at 2133 cm , corresponding to the disap-
+
+
pearance of NO species. The NO peak returns to the zero
level (complete reaction) again when NO2 is turned on. Iso-
propylamine is then switched off and positive peaks appear
−1
at 2133, 3610, and 3745 cm . These peaks are related to the
+
formation of NO species and formation and/or unblocking
4
. Discussion
of Brønsted acid sites and silanol groups at the sample sur-
face, respectively. It is likely that the Brønsted acid sites are
occupied by hydrocarbon and/or amine species, which are
consumed in the reaction when the isopropylamine is turned
off. Isopropylamine is then turned on again and a negative
In previous studies of lean NOx reduction by hydrocar-
bons over Pt/aluminum-silicates [26] and over aluminum-
silicates [27], we have found a clear correlation between the
activity for NO reduction and the Brønsted-site density, as
well as between the N2 selectivity and the Brønsted-site den-
sity. In the present work we focus on the mechanism for NOx
reduction by saturated hydrocarbons and the influence of the
acidity of the catalyst samples.
The ammonia adsorption experiments with FTIR indi-
cate strong Brønsted acid sites for the sample with low
SiO2/Al2O3 ratio, whereas the samples with a high SiO2/
Al2O3 ratio show low or negligible amounts of Brønsted
acid sites. The acidity is found to correlate well with the
NOx reduction activity and with the N2 formation, which
is highest for the HZSM-5 40 sample.
−
1
x
peak at 2133 cm is initially formed together with negative
−1
peaks in the OH region. The negative peak at 2133 cm
reaches a maximum value (in less than 1 min), after which
it decreases in magnitude and eventually reaches the zero
level (complete reaction). The results from the step-response
◦
experiments at 350 and 250 C are similar. However, the
reaction seems to proceed more slowly at the lower tempera-
tures, and turning off the isopropylamine results in negligible
changes (compared with the complete reaction).
To study the lean reduction of NOx in more detail, step-
response experiments with FTIR, where NO2 and propane
are turned off and on, are performed over two of the samples
(
HZSM-5 40 and HZSM-5 234). When the samples are ex-
+
posed to NO2, NO species are formed, likely on Brønsted
acid sites. The possible formation of surface-bound nitrates
(
− −
NO3 ) and nitrites (NO2 ), which should be indicated by
−1
IR absorption bands in the region of 1650–1000 cm , has
not been studied in the present work because of disturbances
related to perturbed vibrations of the zeolite framework (be-
low 2000 cm ). However, nitrates are formed mainly on
basic oxides; for instance, Svedberg and co-workers [59] in-
vestigated the ability to store NOx (as nitrates) for several
materials and found that the NOx storage capacity increased
with increasing basicity of the support material. For the most
acidic sample investigated (an aluminium-silicate sample),
−1
◦
the amounts of stored NOx at temperatures above 300 C
were negligible. Furthermore, Hadjiivanov et al. [31] have
performed NO + O2 adsorption on HZSM-5 zeolites, and
+
they state that the NO species is the only stable NOx sur-
Fig. 9. Intensity change during isopropylamine reaction when NO is turned
2
+
−1
◦
face species on HZSM-5. Hence we conclude that the for-
mation of nitrates on the acidic zeolites used in this study is
less likely.
on. (a) Intensity of the NO peak (2133 cm ) vs time (min) at 450 C and
b) ln(1/t) vs 1/T , where t is the time consumed for the intensity to de-
crease from 90 to 10%. The slope corresponds to the activation energy, Ea.
(

H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
77
For both samples (HZSM-5 40 and HZSM-5 234) iso-
cyanate (and/or nitrile) species are formed when both NO2
and propane are present in the feed gas. Especially for the
less acidic sample (HZSM-5 234), isocyanate species seem
to be accumulated rather than consumed. However, for the
more acidic sample (HZSM-5 40) these species seem to
be accumulated but are partly consumed when either NO2
or propane is turned off. This accumulation of isocyanate
species is probably related to the water formation that oc-
curs during the reaction. The NCO species are likely hydrol-
ysed to amine species [55], and when all water is consumed
the hydrolysis will decrease and eventually stop. Since the
propane conversion and the NOx reduction are higher for the
HZSM-5 40 sample (compared with the less acidic sample),
it is likely that more water is also formed over this sample,
and hence the accumulation of NCO species is less for this
sample compared with the HZSM-5 234 sample. During the
sequences with propane, CH vibrations probably related to
gas-phase propane and/or loosely bound hydrocarbons are
observed. After ca. 1 min (of propane present in the feed
gas), a peak, which can be due to unsaturated hydrocarbons
on the surface, appears for the more acidic sample. This peak
grows as the negative OH vibration peak, typical for Brøn-
sted acid sites in zeolites, becomes more negative. Hence, it
seems likely that the unsaturated hydrocarbons are formed
on the Brønsted acid sites, since the OH peak becomes more
negative when the unsaturated hydrocarbon peak increases,
that is, the Brønsted acid sites are blocked. This is further
illustrated by the fact that this peak is small or negligible
for the HZSM-5 234 sample (which contains significantly
lower amounts of Brønsted acid sites). From the hydrocar-
bon cracking literature it is well known that Brønsted acid
sites catalyse the cracking of hydrocarbons. For instance,
Jentoft et al. [60] describe this phenomenon and suggest, as
one possible route, that strong proton–donor sites of the cat-
alyst (i.e., Brønsted acid sites) can protonate the alkane to
form alkanium ions, which subsequently collapse to form
carbenium ions and then further react to form alkenes. From
these results it is evident that the surface acidity plays an im-
portant role for the NO2 reduction by propane over this type
of catalyst sample. The Brønsted acid sites seem to be in-
dispensable for the activation of the propane (i.e., to form
some kind of carbenium ions), so that further reactions can
proceed.
sults indicate a rapid reaction between the isopropylamine
and NOx species. In this process the NO species seem
+
to play a vital role. When the NO2 is switched off (in the
step-response experiments), a negative peak corresponding
+
to removal of the NO species occurs. Similarly, when the
+
isopropylamine is turned on, the negative NO peak imme-
diately develops, which after a short period of time decreases
in magnitude and approaches the zero level (reaction condi-
◦
tions). When the isopropylamine is switched off (at 450 C)
−1
positive peaks at 2133, 3610, and 3745 cm become vis-
ible. These peaks are probably related to the formation of
NO species on the surface and unblocking of Brønsted
+
acid sites, respectively. During reaction conditions it is con-
ceivable that hydrocarbon and/or amine species, originat-
ing from the isopropylamine, are present at the Brønsted
acid sites. These species are consumed in the reaction, and
the Brønsted sites appear as a positive peak at the same
+
time that the NO species start to accumulate on the sur-
+
face. Hence, it seems likely that the NO species react with
isopropylamine-derived species and that the reaction pro-
ceeds over Brønsted acid sites in the zeolite.
5. Conclusions
The reactants are most likely adsorbed surface species for
which the functional group (isocyanate, amine, etc.) is the
key in the reaction. The exact nature of the adsorbed mole-
cule is less certain and most probably also of minor impor-
tance for the overall reaction pathway. We can conclude that
the propane is likely activated over the Brønsted acid sites in
HZSM-5, forming some kind of carbenium ions, and that the
+
NO2 forms NO species, probably situated on Brønsted acid
+
sites. It is conceivable that the carbenium ions and the NO
react to form isocyanates, since these species are present
at the sample surface. For the acidic sample (HZSM-5 40)
◦
at higher temperature (450 C), NH vibrations are observed
(with propane as a reducing agent for NO2), and it is proba-
ble that the isocyanate species are hydrolysed to form some
type of amine species.
The reaction between isopropylamine and NO2 is fast
compared with the corresponding reaction with propane. In
+
this reaction the NO species are likely to react with the
amine species over Brønsted acid sites. These results indi-
cate that amine species are possible intermediates in the SCR
reaction with saturated hydrocarbons as a reducing agent,
over acidic zeolites. However, for the reaction to occur the
activation of the hydrocarbon, forming some kind of carbe-
The step-response experiments also show indications of
a shoulder in the NH region for the HZSM-5 40 sample at
◦
4
50 C. The shoulder is more pronounced when both NO2
and propane are present in the feed gas, but is also ob-
served when either NO2 or propane is turned off. The latter
may be related to accumulation of isocyanate (and/or nitrile)
species, which can be readily hydrolysed to amine species
+
nium ions, and the formation of NO species seem to be
indispensable.
[
11,48,50,61,62]. For the less acidic sample there are no
signs of NH vibrations. Consequently, the NH vibrations
also seem to correlate with the Brønsted acidity.
Acknowledgments
To further investigate the surface reactions, isopropy-
lamine is used as the reducing agent for NOx, and the re-
The authors thank Prof. Bengt Andersson, Chalmers Uni-
versity of Technology, for valuable discussions concerning

78
H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
mass transfer limitations and Albemarle Catalysts BV for
providing the zeolites. The KK foundation is kindly ac-
knowledged for financial support via the materials research
school (MARCHAL). This work was performed at the Com-
petence Centre for Catalysis, which is hosted by Chalmers
University of Technology and financially supported by the
Swedish Energy Agency and the member companies: AB
Volvo, Johnson Matthey-CSD, Saab Automobile Powertrain
AB, Perstorp AB, AVL-MTC AB, Albemarle Catalysts and
the Swedish Space Corporation.
Table A.1
◦
The values used for estimation of the Weisz-modulus at 450
C
r
(mol/(s kgcat))
Φ
obs
HZSM-5 40
NO
−
−
−
−
4
4
4
4
4.0 × 10
6.1 × 10
2.1 × 10
3.2 × 10
0.43
0.52
0.77
0.95
2
cat.part.
NO
2 zeolite.part.
C H
3
8 cat.part.
C H
3
8 zeolite.part.
HZSM-5 234
NO
−
−
−
−
4
4
4
4
1.7 × 10
2.6 × 10
1.2 × 10
1.8 × 10
0.18
0.23
0.44
0.54
2
2
cat.part.
NO
zeolite.part.
8 cat.part.
C H
3
C H
3
8 zeolite.part.
Appendix A. Influence of Internal diffusion limitations:
Weisz–Prater method
HZSM-5 569
NO2 cat.part.
NO
−
−
−
−
4
4
5
5
1.4 × 10
2.1 × 10
3.5 × 10
5.4 × 10
0.15
0.18
0.13
0.16
2
zeolite.part.
C H
3
8 cat.part.
The influence of mass transfer limitations in a porous cat-
alyst can be estimated by calculation of the Weiszs–modulus
C H
3
8 zeolite.part.
2
rvr
p
References
Φ =
,
(A.1)
Deffcas
[
1] N. Takahashi, H. Shinjoh, T. Iijima, T. Suzuki, K. Yamazaki, K.
Yokota, H. Suzuki, N. Miyoshi, S. Matsumoto, T. Tanizawa, T. Tanaka,
S. Tateishi, K. Kasahara, Catal. Today 27 (1996) 63.
where rv is the reaction rate per volume of the catalyst, rp
is the particle radius, Deff is the effective diffusivity, and
cas is the surface concentration of the reactant. The Weisz–
Prater criterion can then be used; that is,for an isothermal
spherical catalyst particle and a first-order reaction, Φ < 1
indicates a particle free from internal concentration gradi-
ents [63].
[2] E. Fridell, M. Skoglundh, B. Westerberg, S. Johansson, G. Smedler,
J. Catal. 183 (1999) 196.
[
3] E. Fridell, M. Skoglundh, S. Johansson, B. Westerberg, A. Törncrona,
G. Smedler, Stud. Surf. Sci. Catal. 116 (1998) 537.
4] V.I. Parvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233,
and references herein.
[
If the concentration at the surface of the catalyst particle
is well approximated with the concentration in the gas bulk,
the Weisz modulus may be estimated with the following ex-
pression:
[5] M. Wallin, C.J. Karlsson, M. Skoglundh, A. Palmqvist, J. Catal. 218
2003) 354.
(
[6] M. Wallin, C.-J. Karlsson, A. Palmqvist, M. Skoglundh, Top. Ca-
tal. 30/31 (2004) 107.
[
[
7] F. Acke, M. Skoglundh, Appl. Catal. B 22 (1999) L1.
8] F. Acke, M. Skoglundh, Appl. Catal. B 20 (1999) 235.
2
robsρcatr
p
[9] F. Acke, M. Skoglundh, Appl. Catal. B 20 (1999) 133.
10] F. Acke, B. Westerberg, L. Eriksson, S. Johansson, M. Skoglundh, E.
Fridell, G. Smedler, Stud. Surf. Sci. Catal. 116 (1998) 285.
11] F. Acke, B. Westerberg, M. Skoglundh, J. Catal. 179 (1998) 528.
12] R. Burch, A.A. Shestov, J.A. Sullivan, J. Catal. 182 (1999) 497.
13] R. Burch, J.A. Sullivan, J. Catal. 182 (1999) 489.
14] R. Burch, T.C. Watling, J. Catal. 169 (1997) 45.
15] R. Burch, P.J. Millington, Catal. Today 26 (1995) 185.
[16] H. Hamada, Catal. Today 22 (1994) 21.
17] G. Zhang, T. Yamaguchi, H. Kawakami, T. Suzuki, Appl. Catal. B 1
Φ =
.
[
Deffcr
[
[
[
[
[
However, for the present case the catalyst particle consists
of small zeolite particles, and hence the Weisz-modulus for
the small zeolite particles must also be taken into account.
The effective diffusivity for the catalyst particle is estimated
−
5
as Deff,cat.part. = εpDbulk/3 = 0.35 × 2 × 10 /3 = 2 ×
−6
2
[
[
[
1
0
m /s, and the effective diffusivity for the zeolite par-
−11
2
(1992) L15.
ticle (Deff,zeolite.part.) is approximately 10
m /s [64,65].
18] H. Hattori, O. Takahashi, M. Takagi, K. Tanabe, J. Catal. 68 (1981)
For the zeolite particles the pore volume must be included
in the observed reaction rate (robs); hence robs,zeolite.part. =
robs/(1 − εp). The density of the catalyst (ρcat) is ap-
132.
19] M. Hino, K. Arata, Catal. Lett. 30 (1995) 25.
[20] R. Burch, E. Halpin, M. Hayes, K. Ruth, J.A. Sullivan, Appl. Catal.
B 19 (1998) 199.
3
proximately 600 kg/m , the radius of the zeolite particle
[
21] C.P. Hubbard, K. Otto, H.S. Gandhi, K.Y.S. Ng, Catal. Lett. 30 (1995)
1.
(
rp,zeolite) is estimated to be 0.5 µm, and the radius of
the catalyst particle (rp,cat) is estimated at 250 µm. The
concentrations of reactants (cNO and cC H ) are 18 and
4
[
22] A.F. Lee, K. Wilson, R.M. Lambert, C.P. Hubbard, R.G. Hurley, R.W.
McCabe, H.S. Gandhi, J. Catal. 184 (1999) 491.
[23] M. Skoglundh, A. Ljungqvist, M. Petersson, E. Fridell, N. Cruise, O.
Augustsson, E. Jobson, Appl. Catal. B 30 (2001) 315.
2
3 8
3
5
.1 mmol/m , respectively. The observed reaction rates and
the estimation of the Weisz modulus are presented in Ta-
ble A.1.
For all samples the calculated values of the Weiszs mod-
ulus are below 1 (see Table A.1); hence the catalytic reaction
is likely not influenced by mass transport limitations.
[
24] K. Wilson, C. Hardacre, R.M. Lambert, J. Phys. Chem. 99 (1995)
13755.
[
[
25] H.C. Yao, H.K. Stephen, H.S. Gandhi, J. Catal. 67 (1981) 231.
26] H.H. Ingelsten, M. Skoglundh, E. Fridell, Appl. Catal. B 41 (2003)
287.

H.H. Ingelsten et al. / Journal of Catalysis 232 (2005) 68–79
79
[
[
[
[
[
27] H.H. Ingelsten, Å. Hildesson, E. Fridell, M. Skoglundh, J. Mol. Catal.
[46] H.Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 180 (1998)
171.
[47] H.Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, Catal. Today 54
(1999) 483.
[48] M. Yamaguchi, J. Chem. Soc.-Faraday Trans. 93 (1997) 3581.
[49] F. Solymosi, T. Bansagi, J. Catal. 156 (1995) 75.
[50] S.K. Park, H. Choo, L. Kevan, Phys. Chem. Chem. Phys. 3 (2001)
3247.
[51] H.Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 186 (1999)
91.
A 209 (2004) 199.
28] J.A. Martens, A. Cauvel, F. Jayat, S. Vergne, E. Jobson, Appl. Catal.
B 29 (2001) 299.
29] T. Holma, A. Palmqvist, M. Skoglundh, E. Jobson, Appl. Catal. B 48
(
2004) 95.
30] A. Hinz, M. Skoglundh, E. Fridell, A. Andersson, J. Catal. 201 (2001)
47.
31] K. Hadjiivanov, J. Saussey, J.L. Freysz, J.C. Lavalley, Catal. Lett. 52
1998) 103.
2
(
[
32] J. Szanyi, M.T. Paffett, J. Catal. 164 (1996) 232.
[52] T. Shimanouchi, in: Tables of Molecular Vibrational Frequences Con-
solidated, vol. 1, National Bureau of Standards, Washington, 1972,
pp. 1–160.
[33] M. Trombetta, G. Busca, S. Rossini, V. Piccoli, U. Cornaro, A. Guer-
cio, R. Catani, R.J. Willey, J. Catal. 179 (1998) 581.
[
[
[
34] G. Boskovic, T. Vulic, E. Kis, P. Putanov, Chem. Eng. Technol. 24
[53] F. Poignant, J. Saussey, J.C. Lavalley, G. Mabilon, Catal. Today 29
(1996) 93.
[54] S.K. Park, Y.K. Park, S.E. Park, L. Kevan, Phys. Chem. Chem. Phys. 2
(2000) 5500.
[55] R. Burch, J.P. Breen, F.C. Meunier, Appl. Catal. B 39 (2002) 283.
[56] I.O.Y. Liu, N.W. Cant, B.S. Haynes, P.F. Nelson, J. Catal. 203 (2001)
487.
[57] R. Brosius, J.A. Martens, Top. Catal. 28 (2004) 119.
[58] NIST Standard Reference Database 69, Coblentz No. 8753, http://
webbook.nist.gov/chemistry/ (18 October, 2004).
[59] P. Svedberg, E. Jobson, S. Erkfeldt, B. Andersson, M. Larsson, M.
Skoglundh, Top. Catal. 30–31 (2004) 199.
(
2001) 269.
35] F. Poignant, J.L. Freysz, M. Daturi, J. Saussey, Catal. Today 70 (2001)
97.
1
36] B.I. Mosqueda-Jimenez, A. Jentys, K. Seshan, J.A. Lercher, Appl.
Catal. B 43 (2003) 105.
[
[
37] P.E. Fanning, M.A. Vannice, J. Catal. 207 (2002) 166.
38] E.H. Teunissen, R.A. Vansanten, A.P.J. Jansen, F.B. Vanduijneveldt,
J. Phys. Chem. 97 (1993) 203.
[
[
39] Q. Sun, Z.X. Gao, B. Wen, W.M.H. Sachtler, Catal. Lett. 78 (2002) 1.
40] R. Szostak, Molecular Sieves, Van Nostrand Reinhold, New York,
1989.
[
[
[
[
41] J. Eng, C.H. Bartholomew, J. Catal. 171 (1997) 27.
42] M. Xin, I.C. Hwang, S.I. Woo, J. Phys. Chem. B 101 (1997) 9005.
43] K.I. Hadjiivanov, Catal. Rev.-Sci. Eng. 42 (2000) 71.
[60] F.C. Jentoft, B.C. Gates, Top. Catal. 4 (1997) 1.
[61] A.D. Cowan, R. Dumpelmann, N.W. Cant, J. Catal. 151 (1995) 356.
[62] A.D. Cowan, N.W. Cant, B.S. Haynes, P.F. Nelson, J. Catal. 176 (1998)
329.
44] S.A. Beloshapkin, E.A. Paukshtis, V.A. Sadykov, J. Mol. Catal. A 158
[
[
[
63] P.B. Weiszs, C.D. Prater, Adv. Catal. 6 (1954) 143.
64] S.F. Garcia, P.B. Weisz, J. Catal. 142 (1993) 691.
65] A.S. Chiang, A.G. Dixon, Y.H. Ma, Chem. Eng. Sci. 39 (1984) 1461.
(
2000) 355.
45] A. Obuchi, C. Wogerbauer, R. Koppel, A. Baiker, Appl. Catal. B 19
1998) 9.
[
(