The effects of steaming and sodium exchange on the selective catalytic reduction of NO and NO2 by NH3 over HZSM-5

Article abstract of DOI:10.1006/jcat.2002.3544

The SCR of NO and NO₂ by NH₃ over a series of HZSM-5 catalysts, which were modified by mild steaming and by sodium exchange, was studied. Mild steaming of a HZSM-5 catalyst increased its activity for the reduction of NO by NH₃. A slight decrease (from 52 to 47 kJ/mole) in apparent activation energy was observed. Mild steaming increased the rate constants for SCR and NH₃ oxidation by about a factor of two but did not change the ammonia adsorption constant. The activity of a series of ZSM-5 catalysts containing varying amounts of sodium for NO reduction was weakly dependent on the Na level. The extent of sodium exchange strongly influenced the activity of the catalysts for NO₂ reduction; although the fully exchanged sample was still active for this reaction, it was ≤ 20 times less active than catalysts containing less sodium. The framework and extra-framework acid sites were suggested to be active for the reduction of NO₂.

Full text of DOI:10.1006/jcat.2002.3544

Journal of Catalysis 208, 106–113 (2002)
doi:10.1006/jcat.2002.3544, available online at http://www.idealibrary.com on
The Effects of Steaming and Sodium Exchange on the Selective
Catalytic Reduction of NO and NO2 by NH3 over HZSM-5
Scott A. Stevenson,1 Jim C. Vartuli, and Sanjay B. Sharma2
ExxonMobil Research & Engineering Company, Corporate Strategic Research, Active Materials Laboratory/Materials Synthesis Section,
1545 Route 22 East, Annandale, New Jersey 08801-3059
Received August 31, 2001; revised January 25, 2002; accepted January 25, 2002
Although a variety of materials show some catalytic activity
for this reaction, catalysts based on mixtures of vanadia and
titania are currently used in most commercial SCR units.
There has been significant interest, however, in developing
zeolite-based catalysts. Zeolites offer a number of advan-
tages over vanadia/titania catalysts: they are active over a
wider temperature range, they are more resistant to ther-
mal excursions, and the spent catalyst can present less of a
disposal problem.
We have studied the selective catalytic reduction of NO and NO2
by NH3 over a series of HZSM-5 catalysts that have been modified
by mild steaming and by sodium exchange. We find that the activity
of an HZSM-5 sample for the reduction of NO is increased follow-
ing a moderate steaming. Kinetic data collected for this sample can
be accurately fit using the rate expression developed for unsteamed
catalysts. Mild steaming increases the rate constants for SCR and
ammonia oxidation by approximately a factor of two but requires no
change in the ammonia adsorption constant, suggesting that steam-
ing increases the rate by increasing the number of active sites. We
also observe that the rate of NO reduction is not strongly affected by
the level of sodium exchange. Given that mild steaming is known
to increase the concentration of extra-framework alumina while
sodium exchange is known to primarily affect the acidity of the
framework Brønsted alumina, these observations strongly support
our earlier conclusion that the active site for the selective catalytic
reduction of NO by ammonia over HZSM-5 is highly acidic extra-
framework alumina. We also find that while the fully exchanged
sample is active for the reduction of NO2 by NH3, the rate is strongly
dependent on the number of unexchanged framework Brønsted
sites, suggesting that both framework and extra-framework alu-
c
Although the literature on the SCR reaction over metal-
exchanged zeolite catalysts is relatively large, there have
been far fewer studies on the mechanism or kinetics of this
reaction over zeolites in the hydrogen form. In a previous
publication (2) we have summarized the relevant literature
on the kinetics of NOx reduction by NH3 over H-form ze-
olites (3–12). Literature reports of the kinetics and mech-
anism of the SCR reaction over zeolites have focused on
H–mordenite (3–7, 11, 12) and H–ZSM-5 (8–12) catalysts.
Studies have observed this reaction to be first order in NO
(7, 9) or nearly so (11) and first order in oxygen (3, 11) or
nearly so (9), but negative order in ammonia (9, 11). Mea-
sured activation energies have generally been observed to
be near 60 kJ/mol (4, 7, 11), although lower values have also
been reported (9). Interestingly, the reaction is observed to
be more rapid when mixtures of NO and NO2 are fed than
when either is fed separately (3–5). The oxidation of NO
to either NO+ or NO2 has been suggested to be rate de-
termining (4, 5, 9, 11, 12), although not all authors are in
agreement with this conclusion (10). The disproportiona-
tion of NO to give NO2 and N2O has also been proposed as
the rate-determining step (8). Both Lewis (4) and Brønsted
acid sites (12) have been suggested to be the active site for
this reaction.
ꢀ 2002 Elsevier Science (USA)
mina are active for NO2 reduction.
Key Words: selective catalytic reduction; HZSM-5; zeolite; NO;
NO2; NOx; acid site modification.
INTRODUCTION
The selective catalytic reduction (SCR) of nitric oxide by
ammonia is the most widely used process for the removal
of NOx from flue gas (e.g., 1). The generally accepted stoi-
chiometry for this reaction involves the combination of
equimolar amounts of NO and NH3 in the presence of oxy-
gen to produce nitrogen and water.
In two previous publications (2, 13), we have reported
the results of our own kinetic studies of the reduction of
NO and NO2 by NH3 over an HZSM-5 catalyst. We col-
lected kinetic data over a range of temperatures and inlet
NO, NH3, and O2 concentrations and developed a model
that accurately describes the experimental results; a similar,
but much less extensive, set of data was collected for NO2
reduction. Based on these results, we reached two main
4NO + 4NH3 + O2 → 4N2 + 6H2O
1 To whom correspondence should be addressed. Current address:
SABIC Technology Center, 1600 Industrial Boulevard, Sugar Land, TX
77478. E-mail: scott.stevenson@sabicusa.com.
2 Current address: UOP Llc., 25 East Algonquin, P.O. Box 5017, Des
Plaines, IL 60017.
106
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

REDUCTION OF NO AND NO2 OVER MODIFIED HZSM-5 CATALYSTS
107
conclusions: the rate-determining step in the reduction of sized hydrothermally at approximately 100◦C using an n-
NO by NH3 over HZSM-5 is the oxidation of NO, and the propylamine template. The Si/Al molar ratio in the synthe-
active site for this reaction is highly acidic extra-framework sis mixture was 27 : 1; elemental analysis indicated that the
alumina.
actual Si/Al ratio in the final catalyst was roughly 22 : 1 and
Our conclusion that NO oxidation was rate determining that the catalyst contained approximately 700 ppm sodium.
was based on the following observations:
The material used as the parent for the sodium exchange
studies was very similar; chemical analysis suggested it to
have a silicon/aluminum ratio of 22 : 1 and, as received, ap-
proximately 1500 ppm sodium. The average crystal size of
both samples, as measured by TEM, was approximately
0.50µm. Sampleswerecarefullypretreated(2)toensurere-
producibility from run to run; we have previously observed
that differences in pretreatment can lead to measurable dif-
ferences in activity.
• The SCR reaction was first order in NO and O2 but
negative order in NH3, suggesting that NO and O2 partic-
ipated in the series of steps up to and including the rate-
determining step but that NH3 did not.
• Only small amounts of NO2 were observed in the gas
exiting the reactor even though the same HZSM-5 catalyst
converted significant amounts of NO to NO2 in the absence
of NH3.
For the steaming experiments, 0.46 g of catalyst was
loaded into a reactor, as described below. Kinetic experi-
ments using various amounts of NO, ammonia, and oxygen
were performed; the kinetics and kinetic constants were
found to be essentially the same as those we reported pre-
viously using a different sample of the same catalyst (2).
This sample, while still loaded in the open glass reactor
tube, was allowed to adsorb moisture from the air for ap-
proximately 6 weeks. The tube was then remounted in the
kinetics unit and 2074 ppm NH3 in He was flowed over the
catalyst at a rate of 50 cm3/min and a temperature of 50◦C
for 18 min. Based on ammonia measurements at the reactor
exit we calculate that approximately 75 µmol of ammonia
was adsorbed. The flow was then stopped, and the sam-
ple was heated under static conditions to 400◦C at a rate of
20◦C/min and held at this temperature for 130 min. Because
there was no gas flow during this heating, the sample was
exposed to both ammonia and water vapor, due to the des-
orption of adsorbed ammonia and water. The sample was
then treated in a flowing stream of 2% oxygen in helium for
45 min at 500◦C to remove all ammonia, after which it was
cooled to room temperature in flowing helium to be used
for the kinetic studies described below.
The sodium-exchanged samples were prepared by the
following procedure: a known amount of sodium nitrate
was dissolved in distilled water and mixed with 3–4 g of
HZSM-5. Approximately 10 ml of solution was used for
each 1 g of catalyst. The solution and the catalyst were
stirred for 30 min; in some cases the solution and catalyst
were allowed to remain in contact unstirred for an addi-
tional 18 h. Following this, the liquid was decanted, the
sample was washed with 20 ml of water, and the mixture
was filtered. The resulting material was dried in air for 2 h
at 120◦C and calcined in dry air for 2 h at 400◦C. In this
manner samples determined by chemical analysis to con-
tain 0.158, 0.367, and 1.27 mol of Na/mol of Al were pre-
pared. They were compared with two additional samples:
one in which a 2 M solution of ammonium nitrate was used
in place of the sodium nitrate solution in the procedure de-
• The kinetics of NO oxidation were similar to the kinet-
ics of NO reduction and the rate constants were approxi-
mately equal.
• The reduction of NO2 was observed to be two to three
orders of magnitude faster than the reduction of NO under
the same conditions.
• The kinetics of NO2 reduction were found to be sig-
nificantly different from the kinetics of NO reduction, sug-
gesting that the primary barrier to NO conversion had been
removed by feeding NO2.
Our conclusion that the active site for NO oxidation was
highly acidic extra-framework alumina was based on less-
direct evidence. We observed that strong ammonia adsorp-
tion inhibited the rate of NO reduction, suggesting that am-
monia was decreasing the rate by adsorbing on the active
site and that the active site was therefore an acid site. Our
kinetic modeling indicated that the coverage of ammonia
on this site was very high, considerably higher than could
be accounted for by the known strength of ammonia ad-
sorption on framework zeolite alumina.
In order to further test our conclusions concerning the ac-
tive site in the selective catalytic reduction of NO by NH3
over zeolites, we tested the activity of catalysts modified by
two methods known to alter the relative amounts of frame-
work and extra-framework acid sites: mild steaming and
sodium exchange. Mild steaming is known to increase the
amount of highly acidic extra-framework alumina and, if
our hypothesis concerning the active site is correct, might
be expected to increase the rate of NO reduction without
changing its kinetics. Sodium exchange is known to pri-
marily affect the framework Brønsted sites and might be
thought to provide some insight into the importance of
these sites in the SCR reaction. The results of these studies
are reported in this work.
EXPERIMENTAL
The parent catalyst used for the steaming study was a scribed above, giving a sample containing only 0.029 mol of
Mobil commercial preparation of HZSM-5. It was synthe- Na/mol of Al, and a second sample that was merely washed

108
STEVENSON, VARTULI, AND SHARMA
TABLE 1
Analysis Results for Exchanged Samples
Al
Si
Na
Na/Al
Acidic Al contenta
Sample
Treatment
(wt%)
(wt%)
(ppm)
(mol/mol)
(µmol/g)
EX-3
EX-2
EX-4
EX-5
EX-8
2.10 M NaNO3, 18 h
0.19 M NaNO3, 18 h
0.03 M NaNO3, 30 min
Distilled water, 30 min
2.03 M NH4NO3, 18 h
1.76
1.79
1.78
1.80
1.65
39.0
40.0
39.9
39.6
40.9
19,100
5600
2390
1580
403
1.274
0.367
0.158
0.103
0.029
0
412
548
584
632
a Acidic Al content calculated as total Al/g × (1-Na/Al). Al content normalized to constant total Al
concentration in all samples.
with distilled water, dried, and calcined, giving a material 1000 sccm. Standard inlet conditions were 500 ppm NO or
containing 0.103 mol of Na/mol of Al. Some properties of NO2, 500 ppm NH3, and 1% O2; variations of these concen-
these samples are summarized in Table 1.
trations were used to study the reaction kinetics. The cata-
One FeZSM-5 catalyst was used to compare the behavior lyst activity was measured at temperatures between 300 and
of a metal-exchanged zeolite catalyst to that of the HZSM-5 500◦C.
catalysts that were the focus of this work. The parent mate-
Effluent gas samples were recorded every 5 min. The
rial used for this catalyst was the same as that used for the catalyst was left on stream at each set of conditions for
steaming and sodium-exchange studies. It was prepared by 90 min when the total flow rate was 1000 sccm and 3 h when
the incipient wetness impregnation of the HZSM-5 powder the flow rate was 500 sccm; this was a sufficient length of
using an aqueous solution of ferrous acetate. The resulting time for the catalyst to reach steady state in all cases. An
catalyst slurry was thoroughly mixed at room temperature MTI 200 Gas Chromatograph was used to determine O2,
and dried in air overnight at 100◦C in a rotary evaporator. N2, and N2O concentrations; calibrations were made using
Theresultingimpregnatedanddriedsolidwasthencalcined cylinders of known concentration supplied by Matheson.
in air at 600◦C for 10 h to produce the final catalyst. The Two Siemens NDIR analyzers were used to measure NO
final iron loading was determined to be 2.5% by elemental and NH3 levels. NO2 concentrations were estimated from
analysis.
Kinetic data were collected using the same reactor sys-
an overall mass balance.
When NO2 was fed, some measurable conversion of NO2
tem described in our earlier studies (2, 13). A plug flow re- and NH3 to N2 was observed even when the gas flow by-
actor operated in a downflow configuration was used; a by- passed the catalyst. We believe that this conversion is due
pass loop allowed the measurement of feed concentrations. to the homogeneous reaction of NO2 and ammonia in the
Catalyst was loaded into a 3.49 mm inner diameter quartz heated lines leading to the ammonia analyzer. Simple by-
tube reactor and held in place with quartz wool. For the pass tests showed that this background conversion is ap-
NO reduction experiments, approximately 0.46 g of cata- proximately first order in NO and half order in NH3; these
lyst, sized to 20/40 mesh, was loaded and a total flow rate results were used to correct the amount of N2 formed over
of 500 cm3/min was used, giving a space velocity of approx- thecatalystduringkinetictesting. Thiscorrectionwassmall,
imately 36,000 h−1. Due to the significantly higher rate of typically only 1–10% of the N2 formation rate over the cata-
NO2 conversion, the space velocity for experiments using lyst. The HiSil used to dilute the zeolite bed was also tested
NO2 was increased by using less catalyst and, in some cases, for NO, NO2, and NH3 conversion activity; its contribution
a higher gas flow rate. Some 0.019–0.038 g of 20/40 mesh to the observed conversion was found to be negligible.
catalyst was mixed with approximately 0.10 g of HiSil, also
When NO was fed, conversions typically varied from 5
sized to 20/40 mesh, giving a total bed volume of approxi- to 25%. However, when NO2 was fed conversions were
mately 0.4 cm3 and a space velocity on zeolite of 630,000– often much higher, ranging from 25% to more than 80% in
1,750,000 h−1
.
some experiments. In both cases, the measured rates were
Oxygen, nitric oxide or nitrogen dioxide, and ammonia corrected for nondifferential conversions using the kinetic
in balance helium, supplied from gas cylinders, were me- equations developed in our previous work (2, 13). These
tered using mass flow controllers and mixed with helium corrections were small and, we believe, accurate for the
to give the desired inlet concentrations. Trace amounts (2– NO reduction experiments. For the runs where NO2 was
15 ppm) of nitrogen were present in the feed mixture. The fed and the conversion was high, however, the rates must be
NO2 contained a small amount of NO (NO2/NO ratio of considered somewhat approximate. In addition, for catalyst
approximately 63.) The total flow was set to either 500 or EX-2 the measured rate of NO2 reduction at 350 and 400◦C

REDUCTION OF NO AND NO2 OVER MODIFIED HZSM-5 CATALYSTS
109
increased significantly as the space velocity was increased, the rate equation developed in our earlier work (2):
suggesting that the reaction is being limited by bulk mass
kSCR[NO][O2]
1 + Ka[NH3]
kox Ka[NH3]
1 + Ka[NH3]
transfer. For this sample it is likely that the true kinetic
rates, i.e., the rates measured under conditions where mass
transfer was not important, would be somewhat higher than
we observed.
rN
=
+
+ kO[NO].
2
The rate constants for the SCR reaction, kscr, and for am-
monia oxidation, kox, as well as the ammonia adsorption
constant, Ka, were estimated by fitting the data for the
steamed catalyst; kO was assumed to be the same as for
the unsteamed catalyst (2), since data were collected only
at one oxygen concentration. The excellent agreement of
the fits (Fig. 2, solid lines) to the data show that the model
RESULTS AND DISCUSSION
Effect of Steaming on SCR Activity
Figure 1 shows the effect that the mild steaming treat- accurately describes the observed rate data.
ment described above had on the activity of the HZSM-5
The values of the constants used to obtain the fits in
sample; rates were measured with inlet concentrations of Fig. 2 are presented in Table 2, where they are compared to
500 ppm NO, 500 ppm NH3, and 1% O2 at temperatures those reported previously for the unsteamed catalyst (2).
ranging from 300 to 500◦C. At every temperature the activ- From this table we see that the SCR and ammonia oxida-
ity of the catalyst was increased by steaming; the rate en- tion rate constants, kscr and kox, have nearly doubled, while
hancement varied from approximately 2.25 times at 300◦C the ammonia adsorption constant, Ka, remains almost un-
to 1.6 times at 500◦C. The measured apparent activation en- changed. Error analysis indicates that the increase in the
ergy of the catalyst decreased slightly after steaming, from SCR rate constant is significant at the 95% level; the in-
52 ( 2) kJ/mol to 47 ( 4) kJ/mol (95% confidence limits.) crease in the ammonia oxidation rate constant is significant
Figure 2 shows the effect of the steaming treatment on the only at the 70% level. The observation that the same kinetic
kinetics of NO reduction. The rate was observed to be first model fits the data very well suggests that steaming does not
order in NO concentration and negative order in ammonia change the mechanism of the reactions, but rather increases
both before and after steaming. The kinetic data were fit to the number of active sites. Because the rate constants for
FIG. 1. The effect of steaming on NO reduction activity. Inlet conditions: 500 ppm NO; 500 ppm NH3; 1% O2; GHSV = 36,000 h−1
.

110
STEVENSON, VARTULI, AND SHARMA
FIG. 2. The kinetics of NO reduction over steamed HZSM-5. Experimental data (points) fit using kinetic model (solid lines). Solid circles show
effects of varying NO (NH3 held constant at 500 ppm) and open squares show effects of varying NH3 (NO held constant at 500 ppm). Inlet O2 = 1%;
GHSV = 36,000 h−1
.
ammonia oxidation and the SCR reaction increase by ap- Effect of Sodium Exchange on NO Reduction
proximately the same amount, it would also seem reason-
Table 3 shows the effect of sodium exchange on the rate
able to suggest that both occur on the same active site.
of NO reduction. At 350◦C there is no effect of sodium
It is well-known from the literature (e.g., 15–20) that mild
exchange; the rate of N2 formation over the completely ex-
steaming increases the number of strongly acidic extra-
changed sample is the same as the rate over the low-sodium
framework alumina sites in a variety of zeolites, includ-
sample prepared by exchange with ammonium nitrate. At
ing HZSM-5. The observation that steaming also increases
400◦C the activity of all the samples is the same within ex-
the number of active sites for the SCR reaction suggests
perimental error with the exception of the fully exchanged
that extra-framework alumina is in fact the active site for
sample, which is about 25% less active. Only at 450◦C do we
this reaction. This suggestion is in agreement with our ear-
see a consistent trend of declining activity with increasing
lier work (2), where we proposed that the active site for
the SCR reaction over HZSM-5 was extra-framework alu-
mina based on the extent of ammonia inhibition of the
reaction.
TABLE 3
Effect of Na Exchange Level on Activity for NO Reductiona
SCR rate (mol/g/s × 108)
TABLE 2
b
Free Al content
Ea
Sample
(µmol/g)
350◦C
400◦C
450◦C
(kJ/mol)
Changes in Kinetic Fitting Parameters due to Steaminga
EX-3
EX-2
EX-4
EX-5
EX-8
0
412
548
584
632
2.81
2.87
2.90
2.63
2.76
3.98
5.39
5.28
4.91
5.41
4.65
9.03
9.35
9.35
10.5
19 ( 6)
43 ( 6)
44 ( 6)
47 ( 6)
50 ( 6)
Constant
Units
Unsteamed
Steamed
% change
kscr
Ka
kox
kO
cm6/mol/g/s × 10−7
cm3/mol × 10−8
mol/g/s × 109
cm3/g/s
9.85
5.36
8.11
0.558
19.5 (13.2–42)
4.75 (2.5–14.1)
14.1 (2.8–23.7)
0.558b
98%
−11%
74%
—
a Inlet conditions: 500 ppm NH3 and NO; 1% O2. GHSV =
a T = 400◦C; 95% confidence intervals (14) shown in parentheses.
b This parameter was not varied in the data fitting.
36,000 h−1
.
b 95% confidence intervals shown in parentheses.

REDUCTION OF NO AND NO2 OVER MODIFIED HZSM-5 CATALYSTS
TABLE 4
111
sodium level. Even at this temperature, however, the rate
is not proportional to the amount of unexchanged alumina;
sample EX-2 has 35% less unexchanged aluminum than
sample EX-8, but its NO reduction activity is only 15%
less. Likewise, sample EX-3, which has little or no unex-
changed aluminum, still retains 45% of the activity of the
low-sodium samples.
It is known (e.g., 21) that while the framework Brønsted
alumina sites are easily exchanged with sodium, the extra-
framework alumina is not. If framework alumina was the
active site for the reduction of NO, we would expect to see a
linear relationship between sodium content and SCR acti-
vity. Instead, we see almost no effect of sodium level at
lower temperatures and only a weak effect at higher tem-
peratures. We view this data as very strong supporting ev-
Effect of Na Exchange Level on Activity for NO2 Reductiona
N2 + N2O formation rate
Free
Al content
(µmol/g)
(mol/g/s ×106)
GHSV
(h−1
Sample
)
300◦C
350◦C
400◦C
EX-3
EX-2
EX-8
0
412
632
1,750,000
1,250,000
850,000
1.20
4.15b
5.45
1.02
13.8
—
0.88
19.4
—
a Inlet conditions: 500 ppm NH3 and NO2; 1% O2.
b GHSV = 630,000 h−1
.
Effect of Sodium Exchange on NO2 Reduction
Table 4 shows the effect of sodium exchange on the rate
idence that extra-framework alumina is the active site for of NO2 conversion to N2 and N2O. As observed in our ear-
the reduction of NO by NH3 over HZSM-5.
lier work on NO2 reduction over ZSM-5 (13), the rate of
It is not clear why the relative activity of the sodium- NO2 conversion is several hundred times faster than the
exchangedsampleschangesasthetemperatureisincreased, rate of NO reduction. Somewhat to our surprise, the rate of
with the samples containing the most sodium showing NO2 reduction is strongly affected by the level of sodium
somewhat less activity than the low-sodium samples at the exchange and hence the concentration of free framework
higher temperatures. This effect also manifests itself in the Brønsted sites, unlike the rate of NO reduction, which was
apparent activation energies for the five samples; Table 3 not greatly affected by the sodium level. The data in Table 4
shows that the measured activation energies decline as the show that at 300◦C there is an almost linear relationship
sodium content increases, although the relationship is not between the calculated amount of free aluminum and the
linear. Several explanations are possible for this behav- NO2 conversion rate. At 400◦C sample EX-2 was more than
ior. It is possible that at reaction temperatures of 400◦C 20 times more active than sample EX-3, which was fully ex-
and above the sodium moves in the zeolite to some extent changed. This suggests that the reduction of NO2 occurs
and blocks a certain number of the extra-framework acid readily on the framework acid sites in HZSM-5. From the
sites. In agreement with this hypothesis, the activity of sam- fact that the fully exchanged sample retains a measurable
ple EX-3, which was essentially fully exchanged, decreased level of activity as well as the observation that the effect
with time on stream to a much greater extent than any other of sodium exchange increases with temperature, it can be
sample; at 400◦C its activity declined 16% across the course concluded that both the extra-framework and the frame-
of 13 h, while no other sample showed more than a 7% de- work acid sites can catalyze NO2 reduction. However, in
activation under similar conditions. A second possibility is the low-sodium sample the relative number of framework
that the sodium may partially block the zeolite channels, sites is so much larger relative to the concentration of extra-
hindering access of the reactants. If so, such an effect would framework aluminum that the activity of the former domi-
be expected to be more pronounced at higher temperature, nates. At 350 and 400◦C sample EX-2 showed evidence of
where the rate is higher, and for samples containing higher bulk mass transfer limitations while sample EX-3 did not,
sodium levels. Such an effect might be especially severe suggesting that the difference between the low-sodium and
for the fully exchanged sample, since it contains approxi- high-sodium samples is in fact even greater than measured.
mately 25% excess sodium. Finally, the possibility must be It is also worth noting that even in the fully exchanged sam-
considered that the framework Brønsted sites take part in ple the rate of NO2 reduction was 20–35 times faster than
the reaction to some extent at the higher temperatures. If the rate of NO reduction.
so, it would be expected that the activity of all the sam-
The conclusion that both framework and extra-
ples would be the same at low temperature while the highly framework aluminum species can reduce NO2 is in agree-
exchanged samples would not show as high an activity at ment with our previously reported data on the effect of
450◦C. However, we do not believe that this hypothesis is inlet NH3 concentration on the rate of NO2 reduction (13).
consistent with the strong inhibition of the rate due to am- Adsorption constants estimated from this data were of the
monia that we observe at 450◦C (2). In any case, it is clear same order of magnitude as those given in the literature for
from the data that the framework aluminum contributes lit- ammonia adsorption on framework aluminum in HZSM-5
tle to the rate at 400◦C and below, where the sodium level (22).
is observed to have little or no effect on the NO reduction
rate.
Although the strong correlation between sodium level
and activity for NO2 reduction suggests that both

112
STEVENSON, VARTULI, AND SHARMA
framework and extra-framework aluminum are active for ing to speculate that the iron-exchanged catalyst is so active
this reaction, it does not necessarily follow that framework for NO reduction because the iron cations catalyze the oxi-
aluminum participates in the reduction of NO via the oxi- dation of NO much more efficiently than extra-framework
dation of NO to NO2 on an extra-framework site followed aluminum can.
by the desorption of NO2 and its subsequent reduction on
a framework aluminum site. Because of the high activity
of the extra-framework sites for NO2 reduction, it is possi-
ble that little or no NO2 desorbs from the extra-framework
SUMMARY
We have previously proposed (2, 13) that highly acidic
sites before it is converted to N2, especially at lower temper- extra-framework aluminum is the active site for the re-
atures, where NO oxidation and desorption are slower. At duction of nitric oxide by ammonia on HZSM-5. In or-
400◦C and above, however, where small amounts of NO2 der to provide more direct evidence for this hypothesis,
are observed in the effluent during NO reduction, it is prob- we have treated samples by mild steaming and by sodium
able that some of the NO2 is indeed reduced to N2 over the exchange, two procedures that are known to change the rel-
framework Brønsted sites. This may explain the decrease ative amounts of framework and extra-framework acidity
in NO reduction rate observed at 400◦C and above for the of ZSM-5.
fully exchanged sample; in this sample, there are no frame-
We find that a mild steaming of an HZSM-5 catalyst in-
work acid sites available to reduce NO2 and the rate may, creases its activity for the reduction of NO by NH3. A slight
as a consequence, be decreased.
decrease in apparent activation energy is observed, from
52 to 47 kJ/mol. The kinetic data collected for the steamed
catalyst can be fit using the kinetic equation previously de-
veloped for unsteamed samples. Mild steaming increases
the rate constants for SCR and ammonia oxidation by
approximately a factor of two but does not change the
ammonia adsorption constant, suggesting that steaming
increases the rate by increasing the number of active sites.
Because mild steaming is known to increase the amount
of highly acidic extra-framework aluminum, these results
provide support for our hypothesis that such sites are the
active sites for NO reduction in H-form zeolites.
By exchange with sodium nitrate and ammonium nitrate,
we have prepared a series of ZSM-5 catalysts containing
varying amounts of sodium. It is known that framework
Brønsted acid sites are easily exchanged with sodium while
extra-framework acid sites are not. We find that the activity
of these samples for NO reduction is only weakly depen-
dent on the sodium level, suggesting that the framework
Brønsted acid sites play little or no role in this reaction and
indicating that extra-framework species are most likely the
active site. The extent of sodium exchange does, however,
strongly affect the activity of the catalysts for NO2 reduc-
tion; although the fully exchanged sample is still active for
this reaction, it is observed to be as much as 20 times less
active than catalysts containing less sodium. We would sug-
gest that both framework and extra-framework acid sites
are active for the reduction of NO2. Since NO oxidation is
the rate-determining step for NO reduction, the framework
aluminum may play some role in converting NO2 after it is
formed at the extra-framework sites, especially at higher
temperatures, where NO2 is more likely to desorb from
these sites and readsorb on the framework acid sites.
Comparison of HZSM-5 and FeZSM-5 Catalysts
We have also made limited measurements of the effects
of modifying our HZSM-5 catalysts by impregnation of the
parent material with ferric acetate to a final Fe concentra-
tion of approximately 2.5 wt%. A large number of reports
attest to the fact that the addition of iron (e.g., 5, 23–25)
and other transition metals (e.g., 5, 26–32) to zeolites signifi-
cantly increases their activity for NO reduction. The results
of our measurements are shown in Table 5, which shows the
activity of the HZSM-5 and FeZSM-5 catalysts for the re-
duction of NO and NO2 by NH3. As we previously reported
(13), under these conditions the rate of reduction of NO2
by HZSM-5 is almost three orders of magnitude faster than
the rate of NO reduction. In contrast, the FeZSM-5 cata-
lyst shows only a fivefold increase in activity when NO is
replaced with NO2 in the feed. Although the FeZSM-5 cata-
lyst is two orders of magnitude more active than the HZSM-
5 catalyst for NO reduction, it is actually less active for
NO2 conversion. Given that we have shown considerable
evidence that the rate-determining step for the reduction of
NOoverHZSM-5istheoxidationofNOtoNO2, itistempt-
TABLE 5
Comparison of HZSM-5 and FeZSM-5 Catalysts
for NO and NO2 Reduction
Rate of reduction
(mol/g/s ×108)
Sample
NO
NO2
HZSM-5
FeZSM-5
2.3
240
2000
1250
Note. T = 350◦C, 500 ppm inlet NH3, 500 ppm in-
let NO or NO2, 1% inlet O2. GHSV for HZSM-5,
36,000 h−1 (NO) and 1,750,000 h−1 (NO2); GHSV for
FeZSM-5, 415,000 h−1 (NO) and 835,000 h−1 (NO2).
ACKNOWLEDGMENTS
We acknowledge Carlton Brooks, Dave Shihabi, Al Lang, Bill Bundens,
and Pat Cunningham for their helpful assistance.

REDUCTION OF NO AND NO2 OVER MODIFIED HZSM-5 CATALYSTS
113
REFERENCES
18. Haag, W. O., in “Zeolites and Related Materials: State of the Art
1994,” Studies in Surface Science and Catalysis (J. Weitkamp, H. G.
Karge, and W. Ho¨lderich, Eds.), Vol. 84, p. 1375. Elsevier, Amsterdam,
1994.
19. van Bokhoven, J. A., Tromp, M., Koningsberger, D. C., Miller, J. T.,
Pieterse, J. A. Z., Lercher, J. A., Williams, B. A., and Kung, H. H.,
J. Catal. 202, 129 (2001), doi:10.1006/jcat. 2001. 3265.
20. Lukyanov, D. B., Zeolites 14, 233 (1994).
1. Bosch, H., and Janssen, F., Catal. Today 2, 1 (1987).
2. Stevenson, S. A., Vartuli, J. C., and Brooks, C. F., J. Catal. 190, 228
(2000), doi:10.1006/jcat. 1999. 2747.
3. Hirsch, P. M., Environ. Prog. 1, 24 (1982).
4. Andersson, L. A. H., Brandin, J. G. M., and Odenbrand, C. U. I., Catal.
Today 4, 173 (1989).
5. Brandin, J. G. M., Andersson, L. A. H., and Odenbrand, C. U. I., Catal. 21. Lercher, J. A., and Rumplmayr, G., Z. Phys. Chem. 146, 113 (1985).
Today 4, 187 (1989).
6. Odenbrand, C. U. I., Andersson, L. A. H., Brandin, J. G. M., and Ja¨ras,
S., Catal. Today 4, 155 (1989).
7. Medros, F. G., Eldridge, J. W., and Kittrell, J. R., Ind. Eng. Chem. Res.
28, 1171 (1989).
8. Mastikhin, M. N., and Filimonova, S. V., J. Chem. Soc., Faraday Trans.
88, 1473 (1992).
9. Moon, I. S., Namba, S., and Yashima, T., Sekiyu Gakkaishi 36, 339
(1993).
10. Halasz, I., Brenner, A., and Ng, K. Y. S., Catal. Lett. 34, 151
(1995).
11. Eng, J., and Bartholomew, C. H., J. Catal. 171, 14 (1997).
12. Eng, J., and Bartholomew, C. H., J. Catal. 171, 27 (1997).
13. Stevenson, S. A., and Vartuli, J. C. 208, 100 (2002).
14. Box, G. E. P., Hunter, W. G., and Hunter, J. S., “Statistics for Experi-
menters.” Wiley, New York, 1978.
15. Lago, R. M., Haag, W. O., Mikovsky, R. J., Olson, D. H., Hellring, S. D.,
Schmitt, K. D., and Kerr, G. T., “7th International Zeolite Conference,
Tokyo, 1986.” Kodansha, Tokyo.
22. Sharma, S. B., Meyers, B. L., Chen, D. T., Miller, J., and Dumesic,
J. A., Appl. Catal. A 102, 253 (1993).
23. Uddin, M. A., Komatsu, T., and Tashima, T., J. Chem. Soc., Faraday
Trans. 91, 3275 (1995).
24. Amiridis, M. D., Puglisi, F., Dumesic, J. A., Millman, W. S., and Topsøe,
N. Y., J. Catal. 142, 572 (1993).
25. Byrne, J. W., Chen, J. M., and Speronello, B. K., Catal. Today 13, 33
(1992).
26. Medros, F. G., Eldridge, J. W., and Kittrell, J. R., Ind. Eng. Chem. Res.
28, 1171 (1989).
27. Kiovsky, J. R., Koradia, P. B., and Lim, C. T., Ind. Eng. Res. Prod. Res.
Dev. 19, 218 (1980).
28. Ito, E., Hultermans, R. J., Lugt, P. M., Burgers, M. H. W., Rigutto,
M. S., van Bekkum, H., and van den Bleek, C. M., Appl. Catal. B 4, 95
(1994).
29. Komatsu, T., Nunokawa, M., Moon, I. S., Takahara, T., Namba, S., and
Yashima, T., J. Catal. 148, 427 (1994).
30. Komatsu, T., Ogawa, T., and Yashima, T., J. Phys. Chem. 99, 13053
(1995).
16. Zholobenko, V. L., Kustov, L. M., Kazansky, V. B., Loeffler, E., Lohser,
U., Peuker, C., and Oehlmann, G., Zeolites 10, 304 (1990).
17. Campbell, S. M., Bibby, D. M., Coddington, J. M., Howe, R. F., and
Meinhold, R. H., J. Catal. 161, 338 (1996).
31. Nam, I.-S., Eldridge, J. W., and Kittrell, J. R., in “Catalysis 1987” (J. W.
Ward, Ed.), p. 589. Elsevier, Amsterdam, 1988.
32. Ham, S.-W., Choi, H., Nam, I.-S., and Kim, Y. G., Catal. Today 11, 611
(1992).