Lean hydrocarbon selective catalytic reduction over dual pore system zeolite mixtures

Article abstract of DOI:10.1016/j.jcat.2005.07.018

Physical mixtures of CuFER and HZSM-5, CuFER and AgMOR, and CoFER and AgMOR zeolites were investigated as potential dual pore system catalysts for continuous lean NOx reduction with propene or isooctane as reducing agents. The effects of variations in temperature and oxygen concentrations were studied using steady-state experiments. Despite lower rates of external and internal mass transfer, the mixture of CuFER and HZSM-5 was found to have the highest NOx reduction activity with isooctane as the reducing agent. A maximum NOx reduction of about 40% was found for this catalyst at 350°C and 12% oxygen. In this case, the CuFER extrudate acted as an NO oxidation catalyst, and the NO₂ formed was reduced by the isooctane in the HZSM-5 extrudate. The better performance of the CuFER and HZSM-5 mixture compared with the other mixtures was attributed to a higher selectivity for NO₂ to form nitrogen in the reduction reaction. When propene was used, the two mixtures containing CuFER showed similar reduction activity with a maximum conversion of about 65% at 370°C and 2% oxygen. In contrast with the isooctane, the NOx reduction with propene occurred primarily in the extrudate containing ferrierite. Again, the selectivity of NO₂ to form nitrogen governed the overall NOx reduction.

Full text of DOI:10.1016/j.jcat.2005.07.018

Journal of Catalysis 235 (2005) 114–127
www.elsevier.com/locate/jcat
Lean hydrocarbon selective catalytic reduction over
dual pore system zeolite mixtures
Martin Petersson a,b,c, Thomas Holma a,d, Bengt Andersson a,b, Edward Jobson a,b,c
Anders Palmqvist a,c,d,∗
,
a
Competence Centre for Catalysis, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
b
Department of Chemical Engineering and Environmental Science, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
c
Volvo Technology AB, Chalmers Science Park, SE-412 88 Göteborg, Sweden
d
Department of Materials and Surface Chemistry, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
Received 10 May 2005; revised 15 July 2005; accepted 18 July 2005
Available online 22 August 2005
Abstract
Physical mixtures of CuFER and HZSM-5, CuFER and AgMOR, and CoFER and AgMOR zeolites were investigated as potential dual pore
system catalysts for continuous lean NO reduction with propene or isooctane as reducing agents. The effects of variations in temperature and
x
oxygen concentrations were studied using steady-state experiments. Despite lower rates of external and internal mass transfer, the mixture of
CuFER and HZSM-5 was found to have the highest NO reduction activity with isooctane as the reducing agent. A maximum NO reduction
x
x
◦
of about 40% was found for this catalyst at 350 C and 12% oxygen. In this case, the CuFER extrudate acted as an NO oxidation catalyst,
and the NO formed was reduced by the isooctane in the HZSM-5 extrudate. The better performance of the CuFER and HZSM-5 mixture
2
compared with the other mixtures was attributed to a higher selectivity for NO to form nitrogen in the reduction reaction. When propene
2
◦
was used, the two mixtures containing CuFER showed similar reduction activity with a maximum conversion of about 65% at 370 C and
2% oxygen. In contrast with the isooctane, the NO reduction with propene occurred primarily in the extrudate containing ferrierite. Again,
x
the selectivity of NO to form nitrogen governed the overall NO reduction.
x
2
2005 Elsevier Inc. All rights reserved.
Keywords: HC-SCR; Zeolites; Propene; Isooctane; NO reduction; Dual pore system catalyst
x
1. Introduction
ment to reduce the average carbon dioxide emissions for new
cars to 140 g/km in 2008, a 25% reduction from the 1995
level [4].
Global warming is probably the greatest environmen-
tal threat today. This fact is internationally recognized
by the United Nations Framework Convention of Climate
Change [1], a document currently ratified by 193 nations [2].
To meet this threat, the nations that have ratified the Kyoto
Protocol [3] have pledged to reduce their collective emis-
sions of greenhouse gases by at least 5% below the 1990
level in the time period 2008–2012. To contribute to this
target, the European vehicle industry has made the commit-
Diesel and lean burn technologies have a high potential to
lower fuel consumption and thus lower carbon dioxide emis-
sions. The drawbacks to these technologies are the higher
emissions of particulate matter and the problems in reduc-
ing the emissions of NOx. To control particulate emissions,
particulate traps such as the CRT system [5] have been im-
plemented in vehicles. For NOx emission control, several
techniques are currently under development [6]. A poten-
tially useful technology is hydrocarbon-selective catalytic
reduction (HC-SCR). This technique can use the fuel as a
reducing agent, precluding the need for an additional re-
ducing agent on-board, as in urea SCR. In addition, high-
*
Corresponding author. Fax: +46-31-160062.
E-mail address: adde@chem.chalmers.se (A. Palmqvist).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.07.018
2005 Elsevier Inc. All rights reserved.

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
115
oxygen (reaction (3)) competes with the reduction reaction.
Although both NO oxidation and NO2 reduction to nitrogen
(reactions (1) and (2a)) may occur in a single zeolite com-
ponent (for example [23]), the selectivity toward these reac-
tions is not typically optimal in a single zeolite component,
which is why the addition of a proper second component is
beneficial. To minimize competition for the reducing agent
between reactions (2a) and (3), it is desirable to suppress ox-
idation of the hydrocarbon with oxygen. This can be done by
placing the oxidation catalyst for reaction (1) in a component
with a low affinity for the hydrocarbon. In addition, the other
component needs to have a high selectivity for reaction (2a)
in comparison with reactions (2b) and (3), to achieve a high
overall nitrogen formation.
In the present study, three new zeolite mixtures were eval-
uated as dual pore system catalysts. The mixtures consisted
of a combination of three zeolite structures: ferrierite, mor-
denite, and ZSM-5. Of these three structures, ferrierite has
the narrowest pores, followed by ZSM-5 and mordenite.
The ferrierite structure was ion-exchanged with either cop-
per or cobalt, mordenite was silver-exchanged, and ZSM-5
was in acidic form. Copper and cobalt was selected because
they were expected to have suitable activity for NO oxida-
tion.
The effect of temperature and oxygen concentration was
investigated under stationary conditions with either propene
or isooctane as the reducing agent. Propene was expected to
readily access even the small-pore zeolite, with the bulkier
isooctane able to enter only the large-pore zeolite. The dual
pore system effect was thus expected to be detected only for
isooctane. The results are discussed in the light of the sug-
gested simplified reaction scheme shown in Fig. 1. When-
ever zeolite samples are compared, the effect of diffusion
limitation is of vital importance. In the current study, the
zeolite crystal size and the average extrudate pore radius var-
ied with crystal structure. The impacts of these co-variations
were taken into account in the evaluation of the results.
NO + (1/2)O ꢀ NO
(1)
(2a)
(2b)
(3)
2
2
NO + HC→/ N + oxidized HC + CO + CO + H O
2
2
2
2
NO + HC→/ NO + oxidized HC + CO + CO + H O
2
2
2
O + HC→/ oxidized HC + CO + CO + H O
2
2
2
Fig. 1. Simplified reaction scheme for NO reduction over a dual pore sys-
tem catalyst. Crossed over arrows indicate a non-balanced reaction.
x
temperature sulfur regeneration is not required, as for NOx
storage catalysts.
Several catalyst systems are active for HC-SCR [6].
Copper-exchanged zeolite ZSM-5 attracted attention in the
early 1990s. However, this material has insufficient long-
term stability and activity as well as a narrow operating
temperature window, and water diminish its performance
[7–9]. High oxygen concentrations also lower catalyst per-
formance. It has been reported that a gas mixture consisting
of 1000 ppm NO, 1000 ppm propene, and an oxygen con-
centration of 2% yields the highest NOx reduction over
CuZSM-5 [10]. In addition to CuZSM-5, other zeolite-based
SCR systems have been evaluated. Metal-containing zeo-
lites, such as CoZSM-5 [11,12] and FeZSM-5 [13,14], and
acidic zeolites [15–19] have shown NOx reduction activity
with hydrocarbons. CoZSM-5 was found to be SCR ac-
tive with methane, and FeZSM-5 showed good results with
isobutane, even in the presence of sulfur and water. There
have also been reports that mixtures of zeolites and metal
oxides [20] can be used for the reduction of NOx; for ex-
ample, a mixture of Mn2O3 and SnZSM-5 was reported to
give better NOx reduction in the presence of water than in
the absence of water.
Mixtures of zeolites with differing pore structures, termed
dual pore systems, are also reported to be more active for
NOx reduction than the single components alone [21–23].
Combining platinum ferrierite with silver mordenite en-
hanced the NOx reduction with isooctane as the reducing
agent [21], and a mixture of cobalt ferrierite and acidic
ZSM-5 was active for NOx reduction with isobutane [23].
The mechanism suggested in the former article is that NO
was selectively oxidized to NO2 in the small-pore zeolite
and, subsequently, reduced to nitrogen by the hydrocarbon
in the large-pore zeolite [21].
2. Experiments
2.1. Sample preparation and characterization
Several reaction mechanisms for NO reduction with hy-
drocarbons over zeolite systems have been suggested, but
most evidence currently supports that NO2 acts as an inter-
mediate in the reduction [9,19,24,25]. Based on these pre-
vious studies, the overall reduction reaction in dual pore
system catalysts can be visualized by the simplified reac-
tion scheme shown in Fig. 1. In reaction (1), NO is oxidized
to NO2. Depending on the oxygen concentration, this re-
action is thermodynamically limited at high temperatures.
The NO2 is then reduced to either nitrogen (reaction (2a))
or back to NO with a hydrocarbon species (reaction (2b)).
Some form of oxidized hydrocarbon species is also formed
in these reactions. The oxidation of the hydrocarbon with
Three zeolite types were used in this study: ferrierite,
mordenite, and ZSM-5. All three of these zeolites were pro-
vided from their manufacturer in acidic form as 1.5-mm-
diameter extrudates. The ferrierite extrudates were provided
by Tosoh, and the mordenite and ZSM-5 were from Eka
Chemicals. The ZSM-5 sample was used as received and
is denoted as HZSM-5 throughout this paper. From the fer-
rierite extrudates, one batch of copper-exchanged samples
and another batch of cobalt-exchanged samples were pre-
pared; these samples are denoted as CuFER and CoFER. The
mordenite extrudates were silver-exchanged and are denoted
as AgMOR.

116
M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
Table 1
Table 2
Zeolite extrudate properties
Zeolite mixture properties
Property
CuFER CoFER HZSM-5 AgMOR
Property
CuFER +
CuFER +
CoFER +
AgMOR
HZSM-5
AgMOR
Manufacturer
Tosoh
16.5
0.3433 0.0272
Tosoh
16.5
Eka
33
–
Eka
30
0.193
SiO /Al O ratio
Weight (g)
Weight ratio between
components
8.9657
1:1
8.0906
1:1
8.5191
1:1
2
2 3
Ion/Al ratio
(Ion = Cu, Co or Ag)
Extrudate diameter (mm)
Average extrudate length (mm)
Extrudate density (kg/m )
Binder type
Binder (wt%)
Crystal density (kg/m )
Average crystal size (µm)
BET surface area (m /g)
Total pore volume (ml/g)
Micropore area (m /g)
Micropore volume (ml/g)
Average pore radius
excluding micropores (Å)
Extrudate porosity
3
1.5
6
1.5
6
1.5
6
1.5
6
Density (kg/m )
713
644
678
3
1110
Silica
30
1110
Silica
30
1048
1106
Alumina Alumina
30
tor. The average inlet temperatures were maintained within
3 ◦C of the setpoint.
30
3
2110
1
2110
1
1894
10
2083
0.5
Four separate mass flow controllers (Brooks Instrument
models 5850S and 5850E) were used to mix the reactor feed
stream of nitrogen (99.996%), oxygen (99.95%), nitric oxide
(99%), and propene (99.5%). A liquid mass flow controller
(Bronkhorst Hi-Tec µ-Flow) was used to inject isooctane
(99.5%) into the 120 ◦C hot gas feed just before the capil-
lary preheater. To avoid condensation of isooctane, all tubes
between the reactor and the analysis instruments were kept
at 120 ◦C. The reactant and product gases were analyzed on-
line using a chemiluminiscence detector (Eco Physics CLD
700 RE ht) for NO and NO2, a gas phase Fourier trans-
form infrared spectroscope (Gasmet FT-IR) for CO and CO2,
and a flame ionization detector (FID) (Signal Hydrocarbon
Analyser Model 3000) for total hydrocarbon concentration.
The zeolite sample mixtures were heated to 500 ◦C in 6%
oxygen and kept constant at that temperature for 30 min to
condition the sample before the experiments were started.
Continuous NOx reduction with hydrocarbon was studied at
steady-state temperatures of 500, 370, 350, 330, and 200 ◦C.
The feed gas consisted of oxygen (varied between 0 and
12%), nitric oxide (1000 ppm), nitrogen (balance), and either
propene (1067 ppm) or isooctane (400 ppm). The oxygen
concentration was varied in the order 6, 10, 4, 2, 12, 0, and
8%. At both 0 and 6%, the system needed 20 min to stabilize,
whereas 10 min was sufficient for the other concentrations.
A total flow rate of 3240 ml/min was used, corresponding to
2
322.0
322.0
366.7
463.0
0.3368
350.1
0.1388
35.1
0.3217 0.3217 0.2815
276.6 276.6 266.5
0.1083 0.1083 0.1149
94.0 94.0 33.3
2
0.2369 0.2369 0.1747
0.2190
excluding micropores (–)
The metal ion exchange was conducted by first passing a
1 mol/dm3 NaNO3 solution through a 10 g extrudate bed in
an ion-exchange column until the pH of the solution passing
the bed was equal to the pH in the sodium solution. There-
after, the sodium-exchanged samples were ion-exchanged
with the desired ion. All of the metal salt solutions were
0.025 mol/dm3 and consisted of Cu(CH3COO)2, Co(NO3)2,
or AgNO3. In each ion exchange, 3 dm3 of the solution was
used. After the exchange, the ion-exchange ratio was deter-
mined spectroscopically. The resulting metal ion/aluminum
ratio is given in Table 1. In addition, SEM analysis (us-
ing a JEOL JSM-5200) was done to determine the zeolite
crystal sizes, and nitrogen adsorption at −196 ◦C (using an
ASAP 2010) was done to investigate the extrudate pore sys-
tem sizes and BET area. The results of these evaluations are
given in Table 1. For the ferrierite samples, the crystal size
specification from the supplier is listed.
a space velocity of 15 500 h−1
.
Temperature-programmed reaction ramps were per-
formed immediately after the steady-state reaction measure-
ments. The samples were cooled from 500 to 120 ◦C and
saturated at 120 ◦C in a gas mixture consisting of nitric ox-
ide (1000 ppm), propene (1067 ppm), oxygen (6%), and
nitrogen (balance). The samples were degassed at 120 ◦C
for 10 min and heated to 500 ◦C at 10 ◦C/min in a gas con-
sisting of 6% oxygen with nitrogen as balance. The total
flow rate was 3240 ml/min throughout these experiments.
From these four zeolite extrudate batches, three zeolite
mixtures were prepared: CuFER + AgMOR, CuFER +
HZSM-5, and CoFER + AgMOR. Each mixture was pre-
pared by mixing the two zeolite extrudate batches at a
50:50 wt% ratio (see Table 2).
2.2. Reactor experiments
The NOx reduction experiments were performed in a
vertically mounted metal tube reactor. The zeolite bed was
placed between two sintered quartz frits, to distribute the gas
evenly over the reactor cross-section. The gas entering the
reactor was heated by a capillary pre-heater, and the reactor
wall was heated by a metallic heating coil. The temperature
was measured by thermocouples before and after the zeolite
bed; separate temperature control units were used to regulate
the temperature in the capillary preheater and the tube reac-
3. Results and discussion
3.1. Impact of test conditions, external and intrapellet mass
transfer
The NOx reduction was evaluated for the three zeolite
mixtures, using propene or isooctane as the reducing agent
at different oxygen concentrations and temperatures. The

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
117
◦
Fig. 2. NO concentration at the reactor outlet with propene at 200–500 C. CuFER + AgMOR (!); CuFER + HZSM-5 (1); CoFER + AgMOR (P). Gas
x
inlet composition: 1000 ppm NO, 1067 ppm C H and 0–12% O . Balanced with N .
3
6
2
2
NOx concentrations at the reactor outlet with propene and
isooctane are shown in Figs. 2 and 3. When the NOx reduc-
tion was compared for the same sample with isooctane and
propene, propene was the more effective reducing agent in
all cases. With propene as the reducing agent, very similar
results were obtained for CuFER + AgMOR and CuFER
+ HZSM-5, but somewhat higher NOx concentrations were
observed for CoFER + AgMOR. This order of performance
was changed somewhat for isooctane. At 200 and 500 ◦C,
no significant differences were seen among the samples. At
the intermediate temperatures, CuFER + HZSM-5 had the
lowest NOx concentrations, followed by CoFER + AgMOR
and then CuFER + AgMOR. Based on these observations,
the overall reaction rate of NOx reduction with propene ap-
pears to be governed by a process in the ferrierite extrudate.
The second zeolite in the mixture influenced the results only
when isooctane was used.
Before the differences in sample performance can be at-
tributed to variations in intracrystalline diffusivity or catalyst
activity, the effects of deviations in test conditions and in
physical properties of the samples (Tables 1 and 2) must be
taken into account. Identical test conditions (i.e., gas com-
position, gas flow, equilibrium time, and inlet gas temper-
ature) were used. Because the hydrocarbon concentrations
were rather high, the exotherms over the reactor bed were
significant (up to 30 ◦C). However, these large exotherms
were connected with high conversions of hydrocarbon; that
is, they were observed only at 500 ◦C for isooctane. In the
isooctane experiments at lower temperatures, the exotherms
were <4 ◦C. This means that large exotherms were observed
only when there were negligible variations in NOx conver-
sion between CuFER + AgMOR and CuFER + HZSM-5,
and thus the differences in conversion cannot be attributed to
deviations in test conditions.

118
M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
◦
Fig. 3. NO concentration at the reactor outlet with isooctane at 200–500 C. CuFER + AgMOR (!); CuFER + HZSM-5 (1); CoFER + AgMOR (P). Gas
x
inlet composition: 1000 ppm NO, 400 ppm i-C H and 0–12% O . Balanced with N .
8
18
2
2
The bed volumes were the same for all samples, but due
to differences in packing, the catalyst weights were not the
same. Because of the deviations in catalyst amounts, the
rate of external mass transfer varied for the samples. The
weight of CuFER + HZSM-5 was 90.2% of the weight
of CuFER + AgMOR. Estimating the external mass trans-
fer coefficient with Colburn jD factors [26], the coefficient
was approximately 3.7% higher for CuFER + HZSM-5 than
for CuFER + AgMOR. However, a lower amount of sam-
ple also corresponded to a lower external surface area. The
external surface area of the extrudates in the CuFER +
HZSM-5 sample was 92.7% of that in the CuFER + Ag-
MOR sample. The higher external mass transfer coefficient
in the CuFER + HZSM-5 sample was completely offset
by the higher external surface area in CuFER + AgMOR.
All other factors being equal, the observed conversion thus
would be lower for the CuFER + HZSM-5. Hence the differ-
ences in NOx conversion cannot be attributed to differences
in catalyst weight.
Apart from the difference in sample weights evaluated
earlier, there were also some physical variations between
the four types of extrudates (see Table 1). The difference
between AgMOR and HZSM-5 is of particular interest, be-
cause the variation in NOx conversion with isooctane has
been attributed to this difference. In comparison, AgMOR
has larger pores and higher porosity in the external pore sys-
tem (i.e., the pore system outside the zeolite framework),
which corresponds to a higher effective diffusivity in the
extrudate. Thus the observed conversion would be higher
for the sample containing AgMOR if all other factors were
equal. The AgMOR extrudates also have smaller zeolite
crystallites than the HZSM-5 extrudates. With equal effec-
tive intracrystalline diffusivity, the intracrystalline pore dif-
fusion limitation would be greater for the HZSM-5 extru-

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
119
◦
Fig. 4. CO and CO concentrations at the reactor outlet with propene at 200–500 C. CuFER + AgMOR with CO (!) or CO ("); CuFER + HZSM-5 with
2
2
CO (1) or CO (2); CoFER + AgMOR with CO (P) or CO (Q). Gas inlet composition: 1000 ppm NO, 1067 ppm C H and 0–12% O . Balanced with N .
2
2
3
6
2
2
dates, because of longer diffusion paths, and the NOx con-
version would be higher for the mixture containing AgMOR
if all other factors were equal.
From this discussion, it can be concluded that CuFER +
HZSM-5 had the highest NOx conversion with isooctane,
despite the physical differences in Tables 1 and 2. Because
the differences in NOx conversion do not correlate with the
differences in physical properties of the extrudates, these
findings must be due to variations in catalyst activity or the
intracrystalline diffusion coefficient.
Starting with the propene experiments, the concentra-
tions of CO2 and CO are shown in Fig. 4, and the corre-
sponding hydrocarbon concentrations are shown in Fig. 6.
CO2 production rises with increasing temperature and in-
creasing oxygen concentration. At 200 ◦C, no formation of
CO2 was observed for any sample. At 330 ◦C and above,
CuFER + AgMOR and CuFER + HZSM-5 had high CO2
concentrations. The CoFER + AgMOR yielded modest CO2
concentrations at 350 and 370 ◦C and high concentrations
at 500 ◦C. The total hydrocarbon concentrations showed an
opposite trend to that of CO2 production, with increasing
temperature and oxygen concentration resulting in decreased
hydrocarbon concentration. Note that even at the highest
temperature, with maximum oxygen concentration and with
propene, some hydrocarbon was still reaching the reactor
outlet. This indicates an external mass transfer limitation.
Based on the total hydrocarbon and CO2 concentra-
tions, the propene conversion was identical for CuFER +
3.2. Hydrocarbon oxidation
The gas composition in these experiments was chosen so
there would be excess oxygen (when present) compared with
hydrocarbon and NOx and excess hydrocarbon compared
with NOx. This makes it possible to examine hydrocarbon
oxidation separately from NOx reduction.

120
M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
◦
Fig. 5. CO and CO concentrations at the reactor outlet with isooctane at 200–500 C. CuFER + AgMOR with CO (!) or CO ("); CuFER + HZSM-5
2
2
with CO (1) or CO (2); CoFER + AgMOR with CO (P) or CO (Q). Gas inlet composition: 1000 ppm NO, 400 ppm i-C H and 0–12% O . Balanced
2
2
8
18
2
with N .
2
AgMOR and CuFER + HZSM-5, and lower for CoFER +
AgMOR. This indicates that these variations was due to dif-
ferences between the CuFER and CoFER extrudates. Apart
from different ions, the ion concentration in the CoFER
extrudate was only 7.9% of the concentration in CuFER.
Based on the observed CO2 concentrations and assuming a
plug flow reactor and a first-order reaction, the apparent rate
constant for propene oxidation over these two samples can
be estimated. For propene conversions up to 70%, the ap-
parent rate constant for the sample containing CoFER was
between 6.3 and 11.3% of the constants observed for the
sample containing CuFER. This indicates that the major dif-
ference in propene oxidation was due to the differing number
of oxidation sites in these samples. Although CoFER ap-
pears to be as good a propene oxidation catalyst as CuFER,
it was harder to introduce sufficient numbers of counter
ions.
The production of CO was generally significantly lower
than that of CO2. As shown in Fig. 4, CO production was
observed over the CoFER + AgMOR sample at 350 ◦C and
above, whereas all three samples had some CO production
at 330 ◦C. Adsorbed CO is assumed to be formed as an in-
termediate in the complete oxidation of the hydrocarbon. If
the rate of CO oxidation is too slow, then the adsorbed CO
can desorb and cause CO emissions, as was observed. The
observed CO concentrations over the CoFER + AgMOR
sample at 350 ◦C and above thus appears to indicate that the
number of metal sites in the CoFER extrudate was too low
to achieve complete oxidation of the hydrocarbons at these
temperatures. The CO observed for all samples at 330 ◦C in-
dicate that not even the CuFER extrudate was sufficiently
active for the CO oxidation.
The results from the isooctane experiments are shown in
Figs. 5 (for CO and CO2 concentration) and 7 (for total hy-

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
121
◦
Fig. 6. Hydrocarbon concentration at the reactor outlet with propene at 200–500 C. CuFER + AgMOR (!); CuFER + HZSM-5 (1); CoFER + AgMOR (P).
Gas inlet composition: 1000 ppm NO, 1067 ppm C H and 0–12% O . Balanced with N .
3
6
2
2
drocarbon concentration). Because the FID detector was cal-
ibrated for C3 equivalence, the 400-ppm feed concentration
of isooctane corresponds to a total hydrocarbon concentra-
tion of 1067 ppm C3 in Fig. 7. Comparing these results with
those for propene shows that CO2 formation was signifi-
cantly lower at all temperatures, indicating a lower rate of
oxidation for isooctane.
ples in the propene oxidation experiments. Thus the trapped
reaction product is assumed to be a partially oxidized hy-
drocarbon species. A possible explanation for this behavior
is that when the isooctane was partially oxidized, it broke
down into smaller species that were adsorbed in the ferrierite
pore system.
The results from the isooctane experiments given in Fig. 5
show low concentrations of CO at all temperatures and low
concentrations of CO2 below 500 ◦C. But despite this find-
ing, the total hydrocarbon concentration in Fig. 7 decreased
with rising oxygen concentration at temperatures between
330 and 370 ◦C. This oxygen dependence indicates that an
oxidation was occurring. Because the carbon balance does
not add up, at least a portion of the oxidation product must be
trapped in the sample. This trapped oxidation product cannot
be CO2, because CO2 was readily emitted from the sam-
3.3. NOx reduction
Returning to the NOx concentration results during the
propene experiments in Fig. 2, a minimum was found for
the CuFER + AgMOR and CuFER + HZSM-5 samples at
2% oxygen at 350 ◦C and above. At 330 ◦C, the same sam-
ples had a wide minimum around 6% oxygen, whereas at
200 ◦C, the NOx concentration decreased continuously with
increasing oxygen concentration, within the range studied.
For the CoFER + AgMOR mixture, the oxygen concentra-

122
M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
◦
Fig. 7. Hydrocarbon concentration at the reactor outlet with isooctane at 200–500 C. CuFER + AgMOR (!); CuFER + HZSM-5 (1); CoFER + Ag-
MOR (P). Gas inlet composition: 1000 ppm NO, 400 ppm i-C H and 0–12% O . Balanced with N .
8
18
2
2
tion had a minor effect at 500 ◦C when oxygen was present.
At lower temperatures, increasing the oxygen concentration
decreased the NOx concentration. The corresponding con-
centrations of NO and NO2 are shown in Fig. 8. Almost all
of the NOx observed was in the form of NO. Elevated con-
centrations of NO2 were found for all samples at 200 ◦C and
for CuFER + AgMOR and CuFER + HZSM-5 at 500 ◦C.
The oxygen-dependent minimum in the NOx concen-
tration observed for CuFER + AgMOR and CuFER +
HZSM-5 with propene can be attributed to the rates of reac-
tions (1) and (3) in Fig. 1. Low oxygen concentrations result
in a lower rate of NO2 formation (reaction (1)), whereas
high oxygen concentrations result in high hydrocarbon con-
sumption (reaction (3)). The effect of limited oxygen con-
centration on reaction (1) was most pronounced at 200 ◦C, at
which maximum NOx reduction was observed at the high-
est oxygen concentration. At higher temperatures, the rate of
hydrocarbon oxidation increases, and the hydrocarbon con-
centration limits the NOx reduction. The lower NOx reduc-
tion for CoFER + AgMOR compared with the other samples
is attributed to a lower selectivity for nitrogen formation (re-
actions (2a) and (2b)). The alternative explanation, that the
NO2 formation (reaction (1)) was lower for CoFER + Ag-
MOR, is not likely, because it was found to have a high NO2
formation already at 200 ◦C (Fig. 9). The reason why no
minimum in NOx concentration was found for this sample
is that the hydrocarbon oxidation (reaction (3)) was slower
due to the lower number of active sites, as discussed earlier.
The limited oxidation of isooctane observed means that
there was no limitation in hydrocarbon supply for the ni-
trogen formation reaction (reaction (2a)). Hence, when the
oxygen concentration was varied, no minimum in the NOx
concentration could be observed in Fig. 3. The NOx con-
centrations were higher with isooctane than with propene,

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
123
◦
Fig. 8. NO and NO concentrations at the reactor outlet with propene at 200–500 C. CuFER + AgMOR with NO (!) or NO ("); CuFER + HZSM-5 with
2
2
NO (1) or NO (2); CoFER + AgMOR with NO (P) or NO (Q). Gas inlet composition: 1000 ppm NO, 1067 ppm C H and 0–12% O . Balanced with N .
2
2
3
6
2
2
however. This means that reaction (2a) must be slower
with isooctane than with propene. Still, almost no NO2
was observed at temperatures >200 ◦C (Fig. 9), because the
NO formation reaction (reaction (2b)) was consuming the
NO2. Thus the variations in NOx reduction performance
between the samples can be attributed to the selectivity
for NO2 to form nitrogen instead of NO when isooctane
was used. When combined with CuFER, the structure of
HZSM-5 appears to be better suited for reaction (2a) than
AgMOR. An alternative explanation, that the entire reac-
tion scheme was governed by the reaction rates in a sin-
gle component (i.e., the CuFER component did not affect
the overall performance), is not likely based on the single-
component studies reported previously [21,23]. Thus with
isooctane, the mixtures behaved as dual pore system cata-
lysts.
three-dimensional pore system, the framework of AgMOR is
one-dimensional. Strongly adsorbing species inside the Ag-
MOR crystallites could thus block the diffusion paths for
other species. Other possible explanations for the better per-
formance of the sample containing HZSM-5 include a higher
affinity for one or more of the reactants, higher diffusivities,
and a higher number of active sites.
At 330 ◦C with isooctane, the NOx concentration was
lower in CoFER + AgMOR than in CuFER + AgMOR.
This is assumed to be due to the higher NO2 formation
for CoFER + AgMOR than for CuFER + AgMOR (reac-
tion (1)) and thus the higher rate of reaction (2a). The same
effect was not found with propene, for which CuFER + Ag-
MOR was more active for NOx reduction than CoFER +
AgMOR. The reason for this difference is that the NOx re-
duction reactions ((2a) and (2b)) occurred in the ferrierite
extrudate in the presence of propene and in AgMOR (or
HZSM-5) in the presence of isooctane.
An important difference between these structures is that
whereas the framework of HZSM-5 (the MFI structure) has a

124
M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
◦
Fig. 9. NO and NO concentrations at the reactor outlet with isooctane at 200–500 C. CuFER + AgMOR with NO (!) or NO ("); CuFER + HZSM-5
2
2
with NO (1) or NO (2); CoFER + AgMOR with NO (P) or NO (Q). Gas inlet composition: 1000 ppm NO, 400 ppm i-C H and 0–12% O . Balanced
2
2
8
18
2
with N .
2
For both reducing agents, the NOx concentrations were
lower at 370 ◦C than at 500 ◦C (Figs. 2 and 3). At the higher
temperature, the formation of NO2 (reaction (1)) is limited
by the thermodynamic equilibrium. This explains the in-
crease in NOx concentration observed for all samples with
isooctane and for CoFER + AgMOR with propene. How-
ever, the thermodynamic limitation cannot explain the in-
crease in NOx concentration seen for CuFER + AgMOR
and CuFER + HZSM-5 with propene, because significant
concentrations of NO2 were observed at 500 ◦C (Fig. 8). The
NOx reduction for these samples at 500 ◦C in the presence
of propene was thus limited by the hydrocarbon concentra-
tion rather than by the NO2 was not. Supporting this theory
is the earlier observation that propene oxidation was limited
by external mass transfer.
The foregoing model cannot explain the decrease in NOx
concentration in the absence of oxygen at 500 ◦C for propene
and isooctane. Instead, this is attributed to a direct reaction
between NO and the hydrocarbon. Because no NOx reduc-
tion was observed in the absence of oxygen below 500 ◦C,
the activation energy must be so high as to make the rate of
this reaction negligible for this temperature range.
In the envisaged application for dual pore system cat-
alysts, SO2 and water will be present. Water has been
shown to cause a reversible inhibition on related metal-
containing zeolite systems (CuZSM-5 [24], CoFER [27],
and Ce/AgZSM-5 [28]), and a similar inhibition may be
expected on the systems studied here. It is well known
that more severe hydrothermal conditions cause dealumina-
tion of zeolites that becomes more severe with decreasing

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
125
Fig. 10. Molar release rates from samples during temperature programmed reaction ramps after saturation in 1000 ppm NO, 1067 ppm C H and 6% O at
3
6
2
◦
◦
◦
◦
120 C. Samples were degassed for 10 min at 120 C and heated to 500 C in 6% O at 10 C/min. CuFER + AgMOR after 12.5 h on stream (!); CuFER +
2
HZSM-5 after 8.5 h on stream (1); CoFER + AgMOR after 12.5 h on stream (P).
SiO2/Al2O3 ratio of the zeolite. We thus would expect the
ZSM-5 and MOR components to have similar hydrothermal
stabilities and the FER components to have lower stability.
A recent study found that a CuFER sample was hydrother-
mally stable for at least 20 h at 600 ◦C [29]. Thus, hydrother-
mal deactivation would not be expected within the tempera-
ture range studied.
The sensitivity of Co-exchanged zeolites toward SO2 in
the absence of water has been found to depend on the zeo-
lite structure, whereas the lean NO reduction with methane
was more affected over CoFER than over CoZSM-5. When
combining SO2 and water, both samples were significantly
deactivated [27]. Martens et al. showed that the deNOx per-
formance of Ag/H-MOR and Ag/H-FER zeolites remained
unaffected by the presence of 150 ppm SO2 [30]. SO2 has
been shown to reversibly poison CuZSM-5 catalysts [24],
and a similar effect may be expected for the CuFER. To our
knowledge, the effect of SO2 on the HC reduction of NOx
over HZSM-5 has not been studied; however, in combina-
tion with Pt/ZrO2, the NO reduction with hydrogen was only
slightly affected by SO2 [31]. Based on these observations,
we would expect the presence of water and SO2 to inhibit
the dual pore system catalysts studied here but not alter their
relative order of performance.
3.4. Temperature-programmed studies
Fig. 10 shows temperature-programmed reaction ramps
of the zeolite mixtures after saturation in reactant gas mix-
ture at 120 ◦C. Starting with the hydrocarbon graph, there
clearly was only a small difference between the CuFER and
CoFER samples containing AgMOR, whereas the HZSM-5-

126
M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
Fig. 11. Molar release rates from samples during temperature programmed reaction ramps after saturation in 1000 ppm NO, 1067 ppm C H and 6% O at
3
6
2
◦
◦
◦
◦
120 C. Samples were degassed for 10 min at 120 C and heated to 500 C in 6% O at 10 C/min. CuFER + HZSM-5 after 8.5 h on stream (1) and after
2
28 h on stream (2).
containing sample had a significantly higher release. Note-
worthy is that the CoFER + AgMOR sample released some
hydrocarbons above 450 ◦C, in contrast to the other two
samples. For CO and CO2, there was only a small differ-
ence between the two samples containing CuFER, whereas
the CoFER-containing sample released significantly lower
amounts of these oxidation products, presumably due to the
lower number of oxidation sites in CoFER as discussed ear-
lier.
NOx was released in two temperature regions, one around
200–300 ◦C because the hydrocarbon was not activated at
that temperature, and the other around 420–500 ◦C due to
the NO/NO2 equilibrium favoring the formation of the less-
reactive NO. At high temperature, the CoFER + AgMOR
sample released more NO than the two CuFER-containing
samples, suggesting that NOx was more strongly bound to
CoFER than to CuFER. This suggestion is supported by the
lower release of NOx observed from the CoFER sample at
low temperatures. In addition, the CoFER + AgMOR sam-
ple released NOx with a significantly higher NO2/NO ratio
than CuFER + AgMOR, which in turn had a higher ratio
than CuFER + HZSM-5. This supports our previous ob-
servation that CoFER has a higher NO2 formation rate than
CuFER. The difference between the two CuFER-containing
samples was attributed to a higher rate of NO2 reduction over
the HZSM-5 containing sample. The higher rate is due to a
lower required temperature for hydrocarbon activation over
the HZSM-5-containing sample, as is evident from the CO
and CO2 graphs. The TPR results thus support the earlier
discussion regarding the reaction mechanism.
Fig. 11 shows the TPR results for the CuFER + HZSM-5
sample after different times on stream. The most conspicu-

M. Petersson et al. / Journal of Catalysis 235 (2005) 114–127
127
ous difference is seen in the hydrocarbon graph. Even though
the CO and CO2 curves appeared less affected, hydrocarbon
release was significantly lower after 28 h on stream; in par-
ticular, the low-temperature peak almost disappeared. The
NO2 release peak was almost doubled, due to either a de-
crease in NO2 reduction rate or an increase in NO oxidation
rate. But this had only a minor effect on the NOx and NO re-
lease curves, where the peaks at low temperature were only
slightly broadened. In the tested time and temperature re-
gion, the NOx reduction performance of this sample thus
appears stable.
Finally, the NOx reduction studies were performed using
a 1:1 weight ratio of the individual components. Because
there was no breakthrough of NO2 in the most active sys-
tems, increasing the NO oxidation component in the mixture
would likely further increase nitrogen formation.
for providing the zeolite samples. The work reported in this
paper was conducted at the Competence Centre for Cataly-
sis, which is supported by the Swedish Energy Agency and
member companies AB Volvo, Saab Automobile AB, John-
son Matthey CSD, Perstorp AB, AVL-MTC AB, Albemarle
Catalysts, and the Swedish Space Administration.
References
[1] United Nations, United Nations Framework Convention on Climate
Change, May 1992.
[2] United Nations, http://unfccc.int/resource/conv/ratlist.pdf,
2005.
9 May
[3] United Nations, Kyoto Protocol to the United Nations Framework
Convention on Climate Change, December 1997.
[4] Association des constructeurs européens d’automobiles, ACEA’s CO
Commitment, December 2002.
2
[5] A. König, G. Herding, B. Hupfeld, Th. Richter, K. Weidmann, Top.
Catal. 16/17 (2001) 23.
4. Conclusions
[6] E. Jobson, Top. Catal. 28 (2004) 191.
[7] M.D. Amiridis, T. Zhang, R.J. Farrauto, Appl. Catal. B 10 (1996) 203.
[8] V.I. Parvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233.
[9] Y. Traa, B. Burger, J. Weitkamp, Micropor. Mesopor. Mater. 30 (1999)
3.
[10] M. Iwamoto, N. Mizuno, H. Yahiro, Stud. Surf. Sci. Catal. 75 (1992)
1285.
[11] X. Wang, H.Y. Chen, W.M.H. Sachtler, J. Catal. 197 (2001) 281.
[12] T. Tabata, H. Ohtsuka, L.M. F Sabatino, G. Bellussi, Micropor. Meso-
por. Mater. 21 (1998) 517.
[13] X. Feng, H.K. Hall, J. Catal. 166 (1997) 368.
[14] F. Heinrich, C. Schmidt, E. Löffler, W. Grunert, Catal. Commun. 2
(2001) 317.
[15] H. Hamada, Y. Kintaichi, M. Sasaki, T. Ito, Appl. Catal. 64 (1990) L1.
[16] I. Halasz, A. Brenner, K.Y.S. Ng, Y. Hou, J. Catal. 161 (1996) 359.
[17] H. Hamada, Y. Kintaichi, M. Sasaki, T. Ito, M. Tabata, Appl. Catal. 70
(1991) L15.
[18] H.H. Ingelsten, D. Zhao, A. Palmqvist, M. Skoglundh, J. Catal. 232
(2005) 68.
[19] J.O. Petunchi, G. Sill, W.K. Hall, Appl. Catal. B 2 (1993) 303.
[20] M. Misono, Y. Hirao, C. Yokoyama, Catal. Today 38 (1997) 157.
[21] J.A. Martens, A. Cauvel, F. Jayat, S. Vergne, E. Jobson, Appl. Catal.
B 29 (2001) 299.
[22] E. Jobson, J.A. Martens, A. Cauvel, F. Jayat, European Patent
1044056, to Volvo Car Corporation.
Physical mixtures of zeolite extrudates were found to ex-
hibit lean NOx reduction capability with propene and isooc-
tane. For propene, the overall rate of NOx reduction was de-
termined by a process in the extrudate containing ferrierite,
whereas NOx reduction with isooctane was influenced by
both the extrudates. These mixtures were thus found to act
as dual pore system catalysts when isooctane was used as the
reducing agent.
For both reducing agents, NO2 acted as an intermediate
for the NOx reduction. With propene, the CuFER-containing
mixtures were more active in the reduction than the CoFER-
containing mixture, due to higher selectivity for NO2 to form
nitrogen rather than NO. The samples containing CuFER
showed a maximum in NOx conversion at an intermediate
oxygen concentration, due to competition between NO2 for-
mation and consumption of the reducing agent.
When isooctane was used, the CuFER + HZSM-5 mix-
ture had the highest NOx conversion, despite this sample’s
more severe external and internal mass transfer limitations.
Generally, all samples had a lower selectivity for NO2 re-
acting to nitrogen with isooctane than with propene. The
differences in NOx reduction between the samples for isooc-
tane are again attributed to differing selectivity for complete
reduction. The reason for this higher selectivity awaits fur-
ther investigation.
[23] T. Holma, A. Palmqvist, M. Skoglundh, E. Jobson, Appl. Catal. B 48
(2004) 95.
[24] A.P. Walker, Catal. Today 26 (1995) 107.
[25] R. Brosius, J.A. Martens, Top. Catal. 28 (2004) 119.
[26] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, Wiley,
New York, 1960, p. 647.
[27] Y. Li, J.N. Armor, Appl. Catal. B 5 (1995) L257.
[28] Z. Li, M. Flytzani-Stephanopoulos, Appl. Catal. B 22 (1999) 35.
[29] K. Rahkamaa-Tolonen, T. Maunula, M. Lomma, M. Huuhtanen, R.L.
Keiski, Catal. Today 100 (2005) 217.
[30] J.A. Martens, A. Cauvel, A. Francis, C. Hermans, F. Jayat, M. Remy,
M. Keung, J. Lievens, P.A. Jacobs, Angew. Chem. 37 (1998) 1901.
[31] T. Nanba, K. Sugawara, S. Masukawa, J. Uchisawa, A. Obuchi, Ind.
Eng. Chem. Res. 44 (2005) 3426.
Acknowledgments
The authors are pleased to acknowledge financial sup-
port from the Swedish Research Council for Engineering
Sciences and from the KK Foundation through the MAR-
CHAL Program. They also thank Eka Chemicals and Tosoh