Selective catalytic reduction of NOx with NH3 over zeolite H-ZSM-5: Influence of transient ammonia supply

Article abstract of DOI:10.1016/S0021-9517(03)00148-9

The effects of transient NH₃ supply on the selective catalytic reduction of NO_x over H-ZSM-5 were studied using flow reactor experiments between 200°C and 500°C. With continuous NH₃ supply, the highest NH₃-SCR activity was observed with equimolar amounts of NO and NO₂ in the feed. For inlet NO/NO₂ molar ratios > 1:1, the activity for NO_x reduction was enhanced up to five times compared to continuous supply of NH₃. For NO/NO₂ ratios ≤ 1:1, however, the same experiment showed a decreased NO_x reduction activity. The oxidation of NO was the rate-determining step in the NH₃-SCR reaction over H-ZSM-5 for NO/NO₂ ratios > 1:1. The presence of NH₃ in the zeolite suppressed the activity for NO oxidation and consequently the SCR activity. A minimum in the activity for NO_x reduction was observed at 130°C independent of the ammonia preadsorption temperature.

Full text of DOI:10.1016/S0021-9517(03)00148-9

Journal of Catalysis 218 (2003) 354–364
www.elsevier.com/locate/jcat
Selective catalytic reduction of NOx with NH3 over zeolite H–ZSM-5:
influence of transient ammonia supply
Mikaela Wallin,a,b,∗ Carl-Johan Karlsson,a,b,1 Magnus Skoglundh,a,b and Anders Palmqvist a,b,c
a
Competence Centre for Catalysis, Chalmers University of Technology, SE-41296, Göteborg, Sweden
b
Department of Applied Surface Chemistry, Chalmers University of Technology, SE-41296, Göteborg, Sweden
c
Volvo Technology Corporation, SE-41288, Göteborg, Sweden
Received 20 November 2002; revised 20 March 2003; accepted 21 March 2003
Abstract
The effect of ammonia supply on the selective catalytic reduction of NO over zeolite H–ZSM-5 was investigated using step response
x
◦
experiments between 200 and 500 C. For inlet NO:NO ratios > 1, the activity for NO reduction transiently increased when NH was
x
2
3
removed from the feed. For NO:NO ratios ꢀ 1, the NO reduction however decreased. By pulsing NH to the feed, the activity for NO
x
2
3
reduction was enhanced up to five times compared to continuous supply of ammonia. For NO:NO ratios exceeding one, also the selectivity
2
towards N O formation was lower with transient ammonia supply. Temperature programmed reaction experiments with preadsorbed NH
2
3
◦
◦
showed highest initial NO reduction activity when ammonia had been adsorbed at 300 or 250 C compared to 200 C. A minimum in
x
◦
NO reduction was observed at 130 C independent of the ammonia adsorption temperature. For NO:NO ratios > 1, the results strongly
2
indicate that NO oxidation is the rate determining step in the ammonia selective catalytic reduction (NH -SCR) reaction over H–ZSM-5.
3
2003 Elsevier Inc. All rights reserved.
Keywords: H–ZSM-5; Selective catalytic reduction; NO reduction; Ammonia transients; Lean deNO
x
x
1. Introduction
and the most commonly used catalysts are vanadia–titania-
based materials, which have high activity and selectivity
between 270 and 450 ◦C [2]. At temperatures higher than
420 ◦C these materials catalyse the oxidation of NH3 by
oxygen, which lead to a loss in SCR activity [3]. Vanadia-
based catalysts are also known to catalyse the oxidation
of SO2 to SO3, which contributes to the formation of par-
ticulates [3]. For applications in automotive exhaust after
treatment systems, there has been significant interest in de-
veloping catalysts that can operate in a wider temperature
range from below 200 ◦C up to 600 ◦C. Several studies have
focused on zeolite materials, which are typically active in
a wider temperature range and less of a disposal problem
compared to poisonous vanadia-based materials. Although
most reports have been devoted to different ion exchanged
zeolites [4–9], several authors have reported that zeolites in
their acidic form, such as H–mordernite and H–ZSM-5, are
also active as NH3-SCR catalysts, particularly at high tem-
peratures [10–16], but also under ambient conditions [17].
Recently, very high NOx conversions have also been re-
ported at temperatures around 200 ◦C over H-form zeolites
with preadsorbed NH3 and for zeolites in their ammonium
Today diesel and lean-burn engines are attractive alter-
natives to Otto engines due to lower fuel consumption and
hence lower emissions of CO2. However, a great disad-
vantage with these alternatives is that they operate at high
air-to-fuel ratios that give oxygen excess in the exhaust gas
and thus makes it difficult to reduce the NOx (NO + NO2)
formed during the combustion. Ammonia can act as a re-
ducing agent for NOx in oxygen excess and techniques for
adding ammonia to the exhaust gas in a controlled manner
are currently being developed. The use of ammonia to reduce
NOx is referred to as ammonia selective catalytic reduc-
tion (NH3-SCR) and is one of the most promising methods
for meeting the future legislation demands concerning NOx
emissions from diesel vehicles [1]. The technique has been
commercially used for many years in stationary applications
*
Corresponding author.
E-mail address: mikaela@surfchem.chalmers.se (M. Wallin).
Present address: Volvo Powertrain Corporation, SE-405 08, Göteborg,
1
Sweden.
0021-9517/03/$ – see front matter
doi:10.1016/S0021-9517(03)00148-9
2003 Elsevier Inc. All rights reserved.

M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
355
form [14,18]. In these reports it was observed that the NO
oxidation was suppressed when gaseous NH3 was present
in the zeolite suggesting that oxidation of NO is the rate-
determining step in the SCR reaction over acidic zeolites for
feeds with high NO content [16,19]. It has also been shown
that the SCR reaction with equimolar amounts of NO and
NO2 is much faster than with only NO and more selective
than with only NO2 over acidic zeolites [12].
In the present investigation the selective catalytic reduc-
tion of NOx was studied for different inlet NO:NO2 ratios
over zeolite H–ZSM-5 using transient ammonia supply to
limit the presence of gaseous NH3 in the zeolite. Transient
supply of NH3 is expected to give higher activity for the oxi-
dation of NO to NO2, which in turn should improve the total
NOx reduction activity. Temperature programmed reaction
studies were performed to further investigate the role of the
temperature during adsorption of NH3. In order to reduce
the complexity of the system and to understand the role of
ammonia, a reduced model gas composition was used con-
taining only NO, NO2, NH3, and O2.
of 100 ml/min was used in all experiments, corresponding
to a space velocity of 30,000 h−1. The composition of the
gas at the reactor outlet was monitored continuously by a
quadrupole mass spectrometer (Balzers QMS 200). The ion
currents analysed were m/e = 17 (NH3, H2O), 18 (H2O),
28 (N2, N2O), 30 (NO, N2O, NO2), 40 (Ar), 44 (N2O), and
46 (NO2). These were measured every 22 s and fragmen-
tation patterns, determined experimentally from calibration
gases, were used to quantitatively analyse the data. Some
of the mass spectrometer (MS) signals were too complex to
convert to concentrations due to cross sensitivities for dif-
ferent gases. This concerned especially the nitrogen oxides,
and for these the MS data were only used for qualitative
analysis. Instead the outlet concentrations of nitrogen ox-
ides were quantified using a chemiluminescence instrument
(CLD 700 EL th, TECAN) for NO and NOx detection and
an IR instrument (UNOR 6N, Maihak) for detection of N2O.
The gas flow into these instruments was diluted with air by a
factor of 9.5 to comply with the operating conditions of the
NO/NOx analyser. Prior to each experiment the zeolite was
pretreated with 10% O2 in Ar at 500 ◦C for 20 min.
The effect of transient ammonia supply was investigated
for temperatures between 200 and 500 ◦C in a series of ex-
periments varying the NO:NO2 ratio between 100% NO,
25% NO + 75% NO2, 50% NO + 50% NO2, and 100%
NO2. The experiments were divided in two sections. In the
first section step responses were studied for six different
30-min steps in NOx and NH3 concentration. In the sec-
ond section the pulse responses for six types of ammonia
pulses, varying in duration and concentration of ammonia,
were studied. The experiments were performed at constant
temperature (200–500◦C) and with a constant O2 concen-
tration of 10%. During the step response experiments the
NOx and NH3 concentrations were 1000 ppm whenever in-
troduced. In the subsequent pulse response experiments the
NOx concentration was kept constant at 1000 ppm while the
ammonia was supplied as pulses according to Table 1. The
ammonia concentration was adjusted so that the total amount
of ammonia was in stoichiometric proportion to the amount
of NOx over each cycle.
2. Experimental methods
2.1. Material
A sample of H–ZSM-5 (EZ 476) was used as obtained
from Eka Chemicals AB, Sweden. The SiO2:Al2O3 molar
ratio was determined to be 33.2 and the zeolite contained im-
purities of iron, manganese, and titanium in amounts corre-
sponding to 0.06 wt% Fe2O3, 0.02 wt% MnO, and 0.02 wt%
TiO2 as determined by X-ray fluorescence analysis accord-
ing to the supplier. The specific surface area of the zeolite
was 358 m2/g as determined by N2 adsorption in accordance
with the BET method using a Micromeretics ASAP 2010 in-
strument.
2.2. Catalytic test procedure
The catalytic measurements were performed at atmo-
spheric pressure using 0.10 g of the zeolite powder. The sam-
ple was placed on a small piece of quartz wool supported by
a porous quartz plate in a vertically mounted fixed bed quartz
reactor (i.d. 7 mm). The same catalyst sample was used in all
experiments. The temperature was measured using a thermo-
couple inside the catalyst bed, and measured and controlled
by a thermocouple placed 6 mm after the catalyst bed. The
heating unit consisted of a heating coil (1 mm diameter Kan-
thal wire) in direct contact with the reactor. The reactor and
heating coil were insulated by quartz wool and aluminum
foil. Feed gases were mixed from Ar (99.998%, air liquid),
O2 (99.95%, air liquid), NO (1.02% in Ar, air liquid), NO2
(4865 ppm in Ar, AGA), and NH3 (1.01% in Ar, AGA) and
introduced to the reactor via individual mass flow controllers
(Brooks). The inlet of NO was placed close to the reactor in-
let to avoid oxidation of NO before the catalyst. A total flow
The adsorbed amount of ammonia was quantified at dif-
ferent temperatures and subsequently reacted with NOx and
O2 by either temperature programmed reaction (TPR) ex-
periments or reaction at constant temperature. In the TPR
experiments the zeolite was first saturated with NH3 at 200,
Table 1
Concentration and duration of ammonia pulses
Pulse NH concentration NH pulse length Time without
Times
3
3
type
(ppm)
(min)
NH (min)
3
repeated
5/10
5/5
1/4
5/1
1/1
3000
2000
5000
1200
2000
1500
5
5
1
5
1
1
10
5
4
1
1
3
3
5
5
5
1/0.5
0.5
10

356
M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
250, or 300 ◦C, and thereafter flushed with Ar for 25 min
whereby the temperature was lowered to 50 ◦C. Finally, the
sample was exposed to 1000 ppm NOx (100% NO or 50%
NO + 50% NO2) and 10% O2 and the temperature was
raised from 50 to 600 ◦C at 20 ◦C/min. The NH3 storage
capacity of the zeolite was determined by integrating the am-
monia MS signal. The stored amount of ammonia was com-
pared with the reacted amount of NOx, determined from the
integrated NOx signal from the NOx analyser. In the second
type of preadsorption experiments the sample was first satu-
rated with NH3 at 150, 200, or 300 ◦C, and thereafter flushed
with Ar for 15 min whereby the temperature was adjusted
to 200 ◦C. Finally the sample was exposed to 1000 ppm NO
and 10% O2 at a constant temperature of 200 ◦C.
Some comments regarding the reactor system and the
product gas analysis are necessary for the interpretation of
the results. Most important was the interference from NH3
in the NO/NOx analyser resulting in a reduction of the
NOx signal when high amounts of both NOx and NH3 were
present in the gas feed. This is due to the fact that the SCR re-
action occurs in the part of the instrument where NO2 is con-
verted to NO for NOx analysis. Empty reactor experiments
showed that a mixture of 1000 ppm NH3 and 1000 ppm NO
resulted in a decrease in the NOx signal corresponding to
140 ppm NOx. Experiments with only NH3 also showed that
NH3 gave a contribution to the NOx signal of about 4–8% of
the NH3 concentration due to NH3 oxidation to NOx in the
same part of the NO/NOx instrument. It was further found
that the empty reactor oxidised about 13% NO to NO2 likely
as a result of homogeneous and wall catalysed NO oxidation
in the system. It is not clear exactly where in the system
the oxidation occurred but it is probable that about half the
amount was oxidised before and half after the reactor due to
exposure to approximately equal lengths of room tempered
steel pipes on each side of the reactor.
was added to the gas feed resulting in a slow increase in the
NH3 and H2O outlet concentrations and an immediate de-
crease in the NOx concentration going through a minimum
(for 250 and 300 ◦C) before levelling out at a new steady
state, with lower NO and NOx levels for higher tempera-
tures. The continuous NOx reduction at the end of this step
was 4% at 200 ◦C, 6% at 250 ◦C, and 18% at 300 ◦C af-
ter the NOx signals have been adjusted for the influence of
NH3, as discussed in the experimental section. Considering
that the steady-state NOx signal between 40 and 60 min was
approximately 14% too low, due to the influence of NH3,
there was in fact a minimum in the outlet NOx concentra-
tion also in the 200 ◦C case. Most of the NH3 was adsorbed
in the zeolite during the first 5 min after ammonia was added
to the feed and hence the NOx signal was only marginally in-
fluenced by NH3 during the minimum. Analysis of the MS
signal m/e = 30 confirmed that also the outlet NOx con-
centration at 200 ◦C increased after about 5 min, similar to
the experiments performed at 250 and 300 ◦C. After 30 min
with only NH3 and O2 in the feed, NO was added to the
feed at 90 min in Fig. 1 and the steady-state conversion was
rapidly reached for all three temperatures. When NH3 was
removed from the feed 30 min later (at 120 min in Fig. 1),
there was a rapid decrease in the NO and NOx concentra-
tions at 300 ◦C, a somewhat slower decrease at 250 ◦C, and
markedly slower at 200 ◦C. At 200 ◦C an initial increase of
the NOx level seemed to precede the minimum. However,
considering again that the NOx concentration was about
14% higher just before the NH3 was turned off, it is obvi-
ous that there was no increase in NOx concentration. The
MS signal m/e = 30 also confirmed that the NOx concen-
tration started to decrease at the point where the NH3 supply
was switched off. The removal of NOx proceeded until the
NH3 concentration reached zero, whereupon the outlet NOx
concentration started to increase and finally reached the inlet
NOx concentration. The NOx level at the minimum was 290,
330, and 640 ppm for 300, 250, and 200 ◦C, respectively, and
the minimum was reached 2, 5, and 16 min, respectively, af-
ter the NH3 was switched off. For all three experiments the
water concentration closely resembled the NOx removal pro-
file during this step. The NO concentration levels at the end
of this step (at 150 min) were similar to the corresponding
outlet NO concentrations at 30 min. In the sixth and final
step, NH3 was added to the feed after 30 min with only
NO, which is equivalent to the second step. The outlet NO
and NOx concentration profiles during this step were almost
identical to the corresponding outlet concentration profiles
during the second step.
3. Results
3.1. Step response experiments
Fig. 1 shows the outlet concentrations of NO, NOx, N2O,
NH3, and H2O during step changes in the inlet concentra-
tions of NH3 and NOx (introduced as NO) for three different
temperatures. The experiments started with a 30-min period
where the inlet gas mixture contained NO, O2, and Ar. In
all three experiments the outlet NO concentration was lower
than the total NOx concentration, which is consistent with
NO oxidation over the zeolite. It was also observed that the
NO oxidation was higher for higher temperatures and in all
cases the oxidation decreased slightly with time. After sub-
tracting the amount of NO that was oxidised in the unheated
steel pipes before and after the reactor (i.e., 13%), the NO
oxidation at the end of the 30-min NO step was estimated
to be 16% at 200 ◦C, 23% at 250 ◦C, and 27% at 300 ◦C
of the total NOx concentration. After the first 30 min, NH3
The influence of the NO:NO2 ratio on the SCR reaction
is shown in Fig. 2, where the same experiments as shown in
Fig. 1 were performed at 250 ◦C for three different inlet NOx
compositions (i.e., 75% NO + 25% NO2, 50% NO + 50%
NO2, and 100% NO2). The highest steady-state SCR activity
was observed for equimolar amounts of NO and NO2 in the
feed, where the NOx conversion exceeded 90% (Fig. 2B).
An interesting observation from this experiment was that the

M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
357
Fig. 1. Step responses for outlet concentrations of NO (light grey curve), NO (black curve), N O (dark grey curve), NH (black curve), and H O (grey) after
x
2
3
2
◦
six different steps in inlet NH (dashed line) and NO (dashed line) concentrations. The experiments were performed at 200, 250, and 300 C by exposing
x
3
H–ZSM-5 for 0–1000 ppm NH , 0–1000 ppm NO and 10% O with Ar as balance.
3
2
maximum conversion was not obtained until approximately
10 min after NH3 was added to the feed. During this period
the unreacted NH3 was accumulated in the zeolite, which
was confirmed by MS analysis of the m/e signal 17 and 18
showing no slip of NH3 at all. The SCR reaction continued
for a few minutes after NH3 was removed from the feed
(at 120 min in Fig. 2B), which also proves that NH3 was
accumulated in the zeolite during the SCR reaction. The ex-
periment with only NO2 as the inlet NOx source (Fig. 2C)
showed relatively high conversion of NOx, but accompanied
by a high outlet concentration of N2O. The selectivity to
N2O, calculated as the formed amount of N2O in percent-
age of the reduced amount of NOx, was in this experiment
as high as 36%, compared to less than 5% for the experiment
with equimolar amounts of NO and NO2. An increased ac-
tivity upon removal of NH3 from the feed, similar to the case
with 100% NO shown in Fig. 1, was only seen in the experi-
ment where the inlet NOx consisted of 75% NO + 25% NO2
(Fig. 2A).
3.2. Pulse response experiments
The effect of transient supply of NH3 for NOx reduction
was investigated in the second section of the experiments
discussed above, by following the outlet NOx, NO, N2O,
NH3, and H2O concentrations during pulsing of the NH3
feed. The experiments were performed with constant inlet

358
M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
Fig. 2. Step responses for outlet concentrations of NO (light grey), NO (black), N O (dark grey curve), NH (black curve), and H O (grey) after six different
x
2
3
2
◦
steps in inlet NH (dashed line) and NO (dashed line) concentrations. The experiments were performed at 250 C by exposing H–ZSM-5 for 0–1000 ppm
x
3
NH , 0–1000 ppm NO and 10% O with Ar as balance. The inlet NO consisted of (A) 75% NO + 25% NO , (B) 50% NO + 50% NO , and (C) 100%
x
x
3
2
2
2
NO .
2
NOx concentration for different NO:NO2 ratios and temper-
atures. The pulse responses were investigated for different
pulse lengths and frequencies in the six sets of ammonia
pulse-cycles described in the experimental section. The case
with 100% NO as the inlet NOx source is presented for 200,
250, and 300 ◦C in Fig. 3. As observed in the step response
experiments, it was clear that the SCR activity increased
faster with higher temperature. The average NOx reduction
and selectivity to N2O formation over the different sets of
pulse cycles are summarised in Table 2, where the conver-
sions have been adjusted for the influence of NH3 on the
outlet NOx signal. It was shown in these experiments that
a transient supply of NH3 yields higher total NOx conver-
sion and less formation of N2O, compared to continuous
supply of NH3, for all temperatures and pulse types stud-
ied. The type of pulse that yielded the highest NOx reduction
depended on the temperature. The most pronounced differ-
ences between transient and continuous NH3 supply was ob-
served at 250 ◦C where transient supply of ammonia yielded
a NOx reduction more than five times higher than a continu-
ous NH3 supply for the pulse type 5 min with NH3 followed
by 10 min without NH3. During the longer pulse-cycles at

Fig. 3. Pulse responses for outlet concentrations of NO (light grey curve), NO (black curve), N O (dark grey curve), NH (black curve), and H O (grey) for six sets of ammonia pulses (dashed grey line). The
x
2
3
2
◦
experiments were performed at 200, 250, and 300 C by exposing H–ZSM-5 for 0–5000 ppm NH , 1000 ppm NO, and 10% O with Ar as balance. For pulse characteristics, see Table 1.
3
2

360
M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
Table 2
Table 3
Average NO reduction and N O selectivity (formed N O in percentage
Average NO reduction and N O selectivity (formed N O in percentage of
x
2 2
x
2
2
of the reduced NO ) for different temperatures and type of NH pulses.
the reduced NO ) for different temperatures and type of NH pulses. The
x
x
3
3
The inlet NO feed consisted of 100% NO and the outlet NO signal were
inlet NO feed consisted of 50% NO and 50% NO
x
x
x
2
adjusted for NH influence
3
a
◦
◦
Temperature ( C)
Type of pulse
Temperature ( C)
200 250 300 400 500 200 250 300 400 500
NO reduction (%) Selectivity to N O (%)
a
◦
◦
Temperature ( C)
Type of pulse
Temperature ( C)
200
250
300
200
250
300
x
2
NO reduction (%)
Selectivity to N O (%)
2
5/10
5/5
1/4
5/1
1/1
63 58 51 37 33 2.4 2.9 4.1 5.0 0.6
79 82 76 56 51 2.5 3.0 4.0 3.4 0.3
85 86 77 43 26 2.1 2.0 2.5 3.2 0.5
81 92 96 91 86 2.7 3.5 4.0 2.5 0.2
78 93 97 82 63 3.8 3.5 3.4 2.3 0.2
78 92 98 88 81 3.8 3.9 3.6 2.4 0.2
81 92 98 99 99 3.4 4.7 4.4 3.2 1.0
x
5/10
5/5
1/4
5/1
1/1
16
33
27
30
11
16
13
6.3
28
0.9
0.6
0.5
0.6
0.4
0.4
1.6
0.7
1.1
1.2
3.1
2.2
2.0
9.2
2.9
3.4
1.9
8.8
8.6
10
11
17
7.1
10
10
3.6
41
56
27
30
25
18
1/0.5
Continuous
1/0.5
Continuous
17
a
5/10 means 5 min with NH followed by 10 min without NH .
3
3
a
5/10 means 5 min with NH followed by 10 min without NH .
3
3
Table 4
Average NO reduction and N O selectivity (formed N O in percentage of
250 and 300 ◦C, the same maximum conversion was reached
as when NH3 was removed from the feed in the step response
experiments (at 120 min, Fig. 1). At 200 ◦C, however, the
time without NH3 supply was too short for maximum NOx
conversion to be reached for all pulse cycles. Another inter-
esting observation from the pulse experiments was that the
selectivity to N2O formation was lower for all pulse cycles
compared to continuous NH3 supply.
x
2
2
the reduced NO ) for different temperatures and type of NH pulses. The
x
3
inlet NO feed consisted of 100% NO
x
2
a
◦
◦
Type of pulse
Temperature ( C)
200 250
NO reduction (%)
Temperature ( C)
200 250
Selectivity to N O (%)
x
2
5/10
5/5
1/4
5/1
1/1
63
70
68
76
73
75
77
60
75
74
83
82
84
86
35
32
34
25
29
26
20
38
40
42
38
41
39
37
The effect of increasing the NO2 fraction of the supplied
NOx was investigated in the pulse response experiments pre-
sented in Fig. 4. Increased total NOx reduction was observed
as a result of pulsing the NH3 supply in the case where the
inlet NOx consisted of 75% NO and 25% NO2 (Fig. 4A).
The improvement was, however, not as high as that ob-
served with 100% NO. For equimolar amounts of NO and
NO2 in the feed (Fig. 4B), transient NH3 supply resulted in
a lower total NOx reduction, compared to continuous NH3
supply. The results from experiments performed at five tem-
peratures, 200, 250, 300, 400, and 500 ◦C, with equimolar
amounts of NO and NO2 as the NOx source are summarised
in Table 3, where the average NOx conversion and selectiv-
ity to N2O formation are presented. It is clear that continuous
NH3 supply resulted in the highest NOx conversion at all
temperatures and that increased temperature resulted in in-
creased activity. The corresponding data for the case with
100% NO2 as the NOx source are presented in Fig. 4C and
summarised for 200 and 250 ◦C in Table 4. Here a con-
siderably higher selectivity to N2O formation was observed
compared to the corresponding experiment with equimolar
amounts of NO and NO2. For the case with only NO2 as the
NOx source the continuous NH3 supply showed the highest
NOx reduction activity in combination with the lowest N2O
selectivity compared to all pulse cycles.
1/0.5
Continuous
a
5/10 means 5 min with NH followed by 10 min without NH .
3
3
activity already at 50 ◦C, since the outlet NOx concentra-
tion (top graph) was lower than the inlet NOx concentration
(1000 ppm). Formation of N2 was observed in analysis of
the MS signal m/e = 28, which also verified that the SCR
reaction occurred. The figure also clearly shows that the
NOx reduction activity of the zeolite was initially higher
with higher preadsorption temperature. As the temperature
increased, the NO reduction declined and reached a min-
imum in activity around 130 ◦C, in all three experiments.
Thereafter, the activity increased again and passed through
a maximum. The maximum appeared at the lowest tem-
perature for the experiments with NH3 preadsorbed at 250
and 300 ◦C. The highest maximum in activity was observed
in the experiment with NH3 preadsorbed at 200 ◦C, however,
the SCR activity remained very low after the minimum until
the temperature exceeded 200 ◦C. Above 200 ◦C ammonia
was found to desorb. The amount of desorbed NH3 was high-
est after preadsorption at 200 ◦C. When comparing the NO
and NOx signals in the three experiments in Fig. 5, it was ob-
served that the NO signal was almost equal to the NOx signal
until no further NOx reduction was observed, indicating that
almost all NO2 formed reacted with adsorbed ammonia un-
til all ammonia was consumed. Thereafter, the NO response
followed the same trend in all three experiments.
3.3. Ammonia adsorption studies
The experiments where NH3 preadsorbed at 200, 250,
and 300 ◦C reacted with NO and O2 during a subsequent
temperature ramp are presented in Fig. 5. The experiments
showed that H–ZSM-5 preadsorbed with NH3 showed SCR
The same type of experiments was performed also for
equimolar amounts of NO and NO2 as the NOx source

Fig. 4. Pulse responses for outlet concentrations of NO (light grey), NO (black), N O (dark grey curve), NH (black curve), and H O (grey) for six sets of ammonia pulses (dashed line). The experiments were
x
2
3
2
◦
performed at constant temperature 250 C exposing H–ZSM-5 for 0–5000 ppm NH , 1000 ppm NO, and 10% O with Ar as balance. The inlet consisted of (A) 75% NO + 25% NO , (B) 50% NO + 50% NO ,
3
2
2
2
and (C) 100% NO . For pulse characteristics, see Table 1.
2

362
M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
Fig. 7. Amount of adsorbed NH (black), amount of reduced NO (grey)
x
3
and desorbed NH (white) during the temperature ramp in the experiment
3
in Figs. 5 (left) and 6 (right).
Fig. 5. TPR profiles of NO , NO, and NH after preadsorption of NH at
x
3
3
◦
200, 250, and 300 C for 1000 ppm NO (= 1000 ppm NO) and 10% O ,
x
2
◦
at a heating rate of 20 C/min.
◦
Fig. 8. Reaction of 1000 ppm NO and 10% O at 200 C over H–ZSM-5
2
◦
preadsorbed with NH at 150, 200, and 300 C.
3
the experiment with NH3 preadsorbed at 200 ◦C. This was
not observed when NH3 was preadsorbed at 300 or 250 ◦C.
In Fig. 7, the amounts of preadsorbed NH3 have been
calculated from the experiments in Figs. 5 and 6 and are
compared with the amount of reduced NOx calculated from
the same experiments. The figure shows that when NH3 was
adsorbed at 250 and 300 ◦C, roughly 91% of the amount of
ammonia adsorbed reacted with the NOx. However, when
the NH3 was adsorbed at 200 ◦C the reacted amount of NOx
was approximately 20% less than the amount of adsorbed
NH3. Ammonia desorption peaks starting at 200 ◦C were ob-
served by MS analysis of m/e signals 17 and 18 (Fig. 5), and
the calculated amount of desorbed NH3 was in good agree-
ment with the observed amount of unreacted NH3 (Fig. 7).
To further investigate the influence of adsorption tem-
perature, experiments were performed where NO and O2
reacted at a constant temperature of 200 ◦C, with NH3 pread-
sorbed in the zeolite at three different temperatures: 150,
200, and 300 ◦C (Fig. 8). A higher NOx reduction activity
was observed for the zeolite preadsorbed with NH3 at 300 ◦C
Fig. 6. TPR profiles of NO and NO after preadsorption of NH at 200,
x
3
◦
250, and 300 C for 1000 ppm NO (= 500 ppm NO + 500 ppm NO ) and
x
2
◦
10% O , at a heating rate of 20 C/min.
2
(Fig. 6). In these experiments a high NOx reduction activity
was observed already at the lowest temperature of the ramp,
50 ◦C, which then decreased while the adsorbed ammonia
was consumed. In this case the initial NOx reduction activity
seemed to be less dependent on the NH3 preadsorption tem-
perature. It was observed that the outlet NO concentration
was slightly lower than 50% of the total outlet NOx con-
centration, meaning that oxidation of NO occurred in the
system. When the temperature exceeded 400 ◦C, the outlet
NO:NO2 ratio followed the thermodynamic equilibrium. At
150 ◦C, the NOx concentration levelled out temporarily in

M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
363
compared to the other two experiments, which showed sim-
ilar activities. In the first few minutes the activity slightly
increased in all three experiments and this slow increase in
activity was comparable to the slow increase in NOx reduc-
tion after NH3 was removed in the step response experiment
performed at 200 ◦C (at 30 min in Fig. 1). Additionally the
amount of reacted NO in these experiments was very similar
to the amount of preadsorbed NH3.
in the preadsorption experiments with subsequent NO + O2
exposure at 200 ◦C, when no gaseous NH3 was present in
the feed. However, the increase in NOx reduction was not
as pronounced as in the step response experiments and thus
hindrance of the NO oxidation by physisorbed or weakly
adsorbed NH3 is also likely to be of importance.
In the case with equimolar amounts of NO and NO2, how-
ever, physisorbed NH3 present in the zeolite did not inhibit
the SCR reaction, since a very high steady-state conversion
was observed already at 200 ◦C and the activity decreased as
soon as NH3 was removed from the feed. This supports the
oxidation of NO being the rate-determining step when NO2
is deficient in the gas mixture.
4. Discussion
The efficiency of zeolite H–ZSM-5 to catalyse the
NH3-SCR reaction was found to be strongly dependent on
the relative ratio of NO and NO2 in the feed. The reaction
proceeded with high activity and low selectivity towards
N2O when equimolar amounts of NO and NO2 were used.
When NO was present in excess of NO2 the reaction pro-
ceeded much slower, although with a low selectivity towards
N2O, whereas when NO2 was in excess the activity was still
high, but the selectivity towards N2O formation was unde-
sirably large. This is in agreement with the high formation
of N2O previously reported for high amounts of NO2 over
zeolite-based catalysts [12,20]. It was further found that, for
the cases when NO was in excess, a transient supply of NH3
increased the activity and selectivity of the SCR reaction.
For these cases where NO2 was in excess of NO the activity
was however not improved and the selectivity towards N2O
was increased, with transient supply of ammonia. These re-
sults point towards the importance of the NO:NO2 balance
and confirm that the oxidation of NO to NO2 is the rate deter-
mining step of the NH3-SCR reaction when NO is in excess
of NO2. The improvement by transient supply of NH3 in
these cases suggests that the presence of ammonia in the
zeolite hinders the oxidation of NO to NO2. Blocking the
sites for NO oxidation by ammonia is most likely the ex-
planation for this behaviour and it is plausible to assume
that the NO oxidation in the zeolite occurs at the Brønstedt
acid sites where NH3 adsorbs as NH4+ ions [21,22] and that
NH3 adsorbs more strongly to these sites than NO. Hence,
the NO oxidation declines when NH3 also is present in the
zeolite. Another explanation is that physisorbed or weakly
bound NH3 blocks the oxidation sites. It is however most
Even though experiments showed a very high reduction
of equimolar amounts of NO and NO2 there was a time
delay for reaching maximal NOx reduction (Fig. 2B). One
explanation for this is that the SCR activity increases with a
higher amount of active NH4+ species present in the zeolite.
Anot+her explanation is that rearrangement of the adsorbed
NH4 species into a more active form is rate limiting for
+
the SCR reaction. The adsorbed NH4 ions form either two
or three hydrogen bonds with the zeolite sites (referred to as
2H and 3H, respectively) [23] where the 2H structure is more
stable at lower temperatures and the 3H structure at higher
temperatures. The 3H structure was suggested by Eng et al.
to be the more active structure for NOx reduction [14]. It has
also been argued that the time delay may be due to the forma-
tion of an intermediate NOx–NH4+ complex [14]. However,
after the period without NOx in the feed (Fig. 2) the NOx re-
duction maximum was achieved almost immediately when
NOx was added to the feed. The rate-limiting step should
+
thus be related to the formation of the act+ive NH4 species
and not to the formation of the NOx–NH4 complex.
According to the discussion about different SCR activity
between NH4+ coordinated in 2H and 3H structures, the ad-
sorption temperature for NH3 should influence the activity of
acidic zeolites. A higher initial NOx reduction activity was
indeed observed in the experiments where NH3 was pread-
sorbed at higher temperatures and NOx was added as NO
(see Figs. 5 and 8). However, this effect was not found in the
experiment of equimolar addition of NO and NO2 (Fig. 6),
indicating that the type of coordination is of less importance
or that a restructuring of 3H to 2H occurs upon lowering
the temperature to 50 ◦C. It points to the higher activity ob-
served in Fig. 5 being due to a higher number of available
oxidation sites after preadsorption of NH3 at higher temper-
atures. However, a more favourable NH4+ coordination with
higher adsorption temperature cannot be ruled out as an ex-
planation for the higher SCR reaction at 200 ◦C (Fig. 8).
It was found in the TPR experiments that a considerably
higher amount of NH3 was adsorbed than reacted when NH3
was preadsorbed at 200 ◦C, compared to 250 and 300 ◦C. It
was also observed in the TPR experiment that with ammo-
nia preadsorbed at 200 ◦C, the activity after the oxidation
minimum remained low below 200 ◦C (Fig. 5). These obser-
vations support that the weakly bound NH3 does not react in
+
likely that adsorbed NH4 ions hinder the NO oxidation.
An indication of this is given by the step response study of
NOx reduction at 300 ◦C (Fig. 1) where maximal NOx re-
duction was not obtained until 5 min after NH3 was removed
from the feed. Physisorbed NH3 is not likely to be present
in the zeolite after this period and still the NOx reduction
increased. When no gas phase NH3 is present there will be
sites available for NO oxidation as soon as NH4+ species are
consumed in the SCR reaction. An increased number of sites
available for NO oxidation will give an increased amount of
NO2 and thus a higher activity for the SCR reaction until the
+
amount of NH4 ions becomes limited. There was also an
initial increase in NOx reduction during the first few minutes

364
M. Wallin et al. / Journal of Catalysis 218 (2003) 354–364
the SCR reaction but hinders the NO oxidation. They are
also in agreement with TPD results by Lónyi et al. [24],
whether the high activity was a result of a higher number
of NO oxidation sites available or more favourably bound
NH3 present in the zeolite after preadsorption at high tem-
peratures. A minimum in the activity for NOx reduction was
observed at 130 ◦C independent of the ammonia preadsorp-
tion temperature. The minimum is most likely connected
with the previously observed minimum in NO oxidation ac-
tivity over H–ZSM-5 around this temperature.
+
who showed that NH3 was bound to NH4 species form-
+
ing NH4 (NH3)n after NH3 adsorption in H–ZSM-5 at
150 ◦C but not at 250 ◦C. In the present study it was fur-
ther found that the amounts of reduced NOx and preadsorbed
NH3 were almost equal when NO and O2 reacted at 200 ◦C;
see Fig. 8. This observation suggests that it is possible also
for the weakly bound NH3 to react at 200 ◦C, probably via
rearrangement into more active species.
Acknowledgments
Low temperature NH3-SCR activity of NOx over
H–ZSM-5 has previously been reported to show an opti-
mum around 30–40 ◦C in the temperature range of 0–100◦C
[17]. This is consistent with the minimum in NOx reduction
activity observed at 130 ◦C in the TPR experiments. The
minimum is most likely related to the NO oxidation capac-
ity of the zeolite since this minimum was not observed for
the case with equimolar amounts of NO and NO2. The oxi-
dation of NO over H–ZSM-5 has previously been reported to
proceed via a minimum in the same temperature range [21].
When the oxidation of NO+increases after the minimum, an
increasing amount of NH4 ions is consumed in the SCR
reaction resulting in more available active sites for NO ox-
idation, which further increases the SCR activity until the
number of NH4+ ions becomes limited and the SCR activity
decreases.
This study was performed within the LOTUS-project,
which is financially supported by the European Commission
(Contract GRD1-1999-11135), and partly within the Com-
petence Centre for Catalysis, which is funded by Chalmers
University of Technology, the Swedish Energy Agency,
and the following member companies: AB Volvo, John-
son Matthey-CSD, Saab Automobile AB, Perstorp AB,
Akzo Nobel Catalysts BV, Swedish Space Corporation, and
MTC AB.
References
[1] M. Koebel, M. Elsener, M. Kleemann, Catal. Today 59 (2000) 335.
[2] R.M. Heck, Catal. Today 53 (1999) 519.
[3] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B-Environ. 18
(1998) 1.
5. Concluding remarks
[4] E. Ito, R.J. Hultermans, P.M. Lugt, M.H.W. Burgers, H. van Bekkum,
C.M. van den Bleek, Stud. Surf. Sci. Catal. 96 (1995) 661.
[5] J.A. Sullivan, J. Cunningham, M.A. Morris, K. Keneavey, Appl. Catal.
B-Environ. 7 (1995) 137.
[6] S. Kieger, G. Delahay, B. Coq, B. Neveu, J. Catal. 183 (1999) 267.
[7] R.Q. Long, R.T. Yang, J. Catal. 186 (1999) 254.
[8] B. Coq, M. Mauvezin, G. Delahay, J.B. Butet, S. Kieger, Appl. Catal.
B-Environ. 27 (2000) 193.
[9] B. Ramachandran, R.G. Herman, S. Choi, H.G. Stenger, C.E. Lyman,
J.W. Sale, Catal. Today 55 (2000) 281.
[10] T. Seiyama, T. Arakawa, T. Matsuda, Y. Takita, N. Yamazoe, J. Ca-
tal. 48 (1977) 1.
[11] J.R. Kiovsky, P.B. Koradia, C.T. Lim, Ind. Eng. Chem. Prod. Res.
Dev. 19 (1980) 218.
[12] J.G.M. Brandin, L.A.H. Andersson, C.U.I. Odenbrand, Catal. Today 4
(1989) 187.
[13] E.Y. Choi, I.S. Nam, Y.G. Kim, J. Catal. 161 (1996) 597.
[14] J. Eng, C.H. Bartholomew, J. Catal. 171 (1997) 27.
[15] T. Komatsu, T. Ueda, T. Yashima, J. Chem. Soc.-Faraday Trans. 94
(1998) 949.
[16] S.A. Stevenson, J.C. Vartuli, C.F. Brooks, J. Catal. 190 (2000) 228.
[17] D. Roberge, A. Raj, S. Kaliaguine, D. Trong On, S. Iwamoto, T. Inui,
Appl. Catal. B 10 (1996) L237.
The effects of transient NH3 supply on the selective cat-
alytic reduction of NOx over H–ZSM-5 were investigated
using flow reactor experiments between 200 and 500 ◦C.
With continuous NH3 supply the highest NH3-SCR activ-
ity was observed with equimolar amounts of NO and NO2
in the feed. For inlet NO:NO2 molar ratios > 1, the activ-
ity for NOx reduction transiently increased when NH3 was
removed from the feed, and by pulsing NH3 to the feed the
activity for NOx reduction was enhanced up to five times
compared to continuous supply of NH3. Also the selectivity
towards N2O formation was lower with transient supply of
NH3. For NO:NO2 ratios ꢀ 1, however, the same experiment
showed a decreased NOx reduction activity. The selectivity
towards N2O formation was also low for equimolar amounts
of NO and NO2 and high when the NOx was supplied as
NO2, independent of the pulse type. These results strongly
indicate that the oxidation of NO is the rate-determining step
in the NH3-SCR reaction over H–ZSM-5 for NO:NO2 ra-
tios higher than one. The presence of NH3 in the zeolite was
concluded to suppress the activity for NO oxidation and con-
sequently the SCR activity.
[18] M. Richter, R. Eckelt, B. Parlitz, R. Fricke, Appl. Catal. B 15 (1998)
129.
[19] S.A. Stevenson, J.C. Vartuli, J. Catal. 208 (2002) 100.
[20] G. Delahay, B. Coq, S. Kieger, B. Neveu, Catal. Today 54 (1999) 431.
[21] I. Halasz, A. Brenner, K.Y. Simon Ng, Catal. Lett. 34 (1995) 151.
[22] F. Lónyi, J. Valyon, Micropor. Mesopor. Mater. 47 (2001) 293.
[23] E.H. Teunissen, R.A. Van Santen, A.P.J. Jansen, F.B. Van Duijneveldt,
J. Phys. Chem. 97 (1993) 203.
Temperature programmed reaction experiments with
preadsorbed NH3 showed the highest initial NOx reduc-
tion activity after ammonia adsorption at high temperatures.
From these results it was however not possible to conclude
[24] F. Lónyi, J. Valyon, Thermochim. Acta 373 (2001) 53.