Reduction of NOx over Fe/ZSM-5 Catalysts: Adsorption Complexes and Their Reactivity toward Hydrocarbons

Article abstract of DOI:10.1006/jcat.1998.2277

Fe/ZSM-5 prepared via sublimation catalyzes the reduction of NO_x to N₂ in the presence of excess O₂ and H₂O. Propane, isobutane, and propene are active reductants; methane is inactive. The NO_x reduction rate is negligible in the absence of O₂; it increases steeply with Po₂ and passes through a maximum. The equilibrium between NO + O₂ and NO₂ is swiftly established over a clean catalyst, but deposits impede this reaction. NO+ O₂ forms chemisorption complexes, NO_y, with y ≥ 2. Upon their reaction with hydrocarbon, a nitrogen-containing deposit is formed on the catalyst. It reacts with NO₂, but not with NO, releasing large quantities of N₂. N₂O seems not to be a precursor of N₂. Isotopic labeling shows that one N atom in every N₂ comes from the deposit; the other comes from NO₂. Less deposit is formed with propane, more with propene. As deposits also block catalyst sites, the rate limiting step in NO_x reduction depends on the nature of the hydrocarbon. Deposits are oxidized by O₂ and volatilized by H₂O at high temperature; these processes contribute to the relative efficiency of different hydrocarbons in NO_x reduction over Fe/ZSM-5.

Full text of DOI:10.1006/jcat.1998.2277

JOURNAL OF CATALYSIS 180, 171–183 (1998)
ARTICLE NO. CA982277
Reduction of NO over Fe/ZSM-5 Catalysts: Adsorption Complexes
x
and Their Reactivity toward Hydrocarbons
Hai-Ying Chen, Timur Voskoboinikov, and Wolfgang M. H. Sachtler
V.N. Ipatieff Laboratory, Center for Catalysis and Surface Science, Department of Chemistry, Northwestern University, Evanston, Illinois 60208
Received March 20, 1998; revised August 31, 1998; accepted September 3, 1998
Little is known about the molecular reaction mechanism
Fe/ZSM-5 prepared via sublimation catalyzes the reduction of of NOx reduction over Fe/ZSM-5. For Co/ZSM-5 Cowan
NOx to N2 in the presence of excess O2 and H2O. Propane, iso- et al. showed that NO reduction with methane displays a
x
butane, and propene are active reductants; methane is inactive. The
NOx reduction rate is negligible in the absence of O2; it increases
large kinetic isotope effect when CH4 is replaced by CD4,
which suggests that H abstraction is rate limiting, leading to
steeply with PO and passes through a maximum. The equilibrium
2
the formation of alkyl radicals (6). Previous work in this lab-
oratory with Cu/ZSM-5 and Co/ZSM-5 had revealed that
exposure of the catalysts to a mixture of NO + O2 gives
rise to the formation of characteristic chemisorption com-
plexes, NOy, which are able to chemically interact with alka-
nes (7). Assuming that propyl radicals are formed from
between NO + O2 and NO2 is swiftly established over a clean cata-
lyst, but deposits impede this reaction. NO + O2 forms chemisorp-
tion complexes, NOy, with y � 2. Upon their reaction with hydro-
carbon, a nitrogen-containing deposit is formed on the catalyst. It
reacts with NO2, but not with NO, releasing large quantities of N2.
N2O seems not to be a precursor of N2. Isotopic labeling shows that
one N atom in every N2 comes from the deposit; the other comes propane, it has been proposed that such radicals react with
from NO2. Less deposit is formed with propane, more with propene. NO to form first nitrosopropane, which could isomerize to
As deposits also block catalyst sites, the rate limiting step in NOx an oxime. Adsorbed oximes have been shown to react with
reduction depends on the nature of the hydrocarbon. Deposits are
oxidized by O2 and volatilized by H2O at high temperature; these
processes contribute to the relative efficiency of different hydrocar-
NO molecules, yielding N2 and N2O (8). When the N atom
14
15
in the oxime was the isotope N, but NO was used in
the feed gas, the product consisted mainly of isotopically
bons in NOx reduction over Fe/ZSM-5.
�c 1998 Academic Press
14 15
14 15
mixed molecules N N and N NO (8). The labeling data
exclude the hypothesis that a nitrogen-free hydrocarbona-
ceous overlayer on the catalyst is the actual reductant for
NOx (9). Ifdepositsare active, theyshould expose nitrogen–
containing groups and N2 would have to be formed by their
interaction with NO2.
Key Words: DeNOx; SCR of NOx; Fe/ZSM-5; adsorbed nitro
groups; adsorbed nitrates; reactive deposits.
1
. INTRODUCTION
Several ZSM-5-supported catalysts are active in the se-
For Cu/ZSM-5 and Co/ZSM-5, a major role of the metal
lective reduction of NOx (any NO + NO2 mixture) in the ions appears to direct NO + O toward forming a metal-
2
presence of excess oxygen. Among this group, certain specific type of adsorption complex, NO . With Cu/ZSM-5,
y
Fe/ZSM-5 catalysts excel by their unique propensity to re- FTIR studies showed bands that were assigned to nitro
tain high activity also in the presence of a large excess of groups, monodentate and bidentate nitrate ions. In con-
water vapor (1–4). This is crucial for the potential applica- trast, with Co/ZSM-5 the reactive NO complex displays IR
y
tion of such catalysts in the emission of internal combustion bands assigned to the nitrito group Co–O–N == O. It appears
engines, in particular under lean burn conditions. In previ- to be able to abstract H atoms even from methane (7).
ous papers we showed that efficient Fe/ZSM-5 catalysts can
In the present work, a similar research strategy has been
be prepared by sublimation of FeCl3 vapor onto HZSM-5, applied to Fe/ZSM-5 catalysts, prepared by our sublima-
followed by hydrolytic removal of chlorine (3, 4). They cata- tion method. First, the role of oxygen has been examined
lyze NOx reduction with iso-butane to N2 in the presence of by comparing temperature-programmed desorption (TPD)
�
1
0% water at 350 C with a yield of 76% . An even higher N2 profiles of Fe/ZSM-5 samples that were exposed to either
yield, near 90% , was obtained with an improved catalyst, NO or a mixture of NO + O . FTIR is used to identify
2
containing a small amount of lanthanum (5). Below the op- adsorption complexes and their reactivity toward hydro-
�
timum temperature of 350 C the NOx to N2 yield over such carbons. In this context a comparison of propane and iso-
catalysts is higher in the presence of water vapor than with butane is relevant because the radical formation concept
a dry feed.
predicts that H abstraction should be easier when a tertiary
171
0021-9517/98 $25.00
Copyright �c 1998 by Academic Press
All rights of reproduction in any form reserved.

1
72
CHEN, VOSKOBOINIKOV, AND SACHTLER
iso-C4 radical is formed. The formation of N2 upon ex- to a temperature-programmed system equipped with a
posure of the NOy complexes to alkanes is monitored by Dycor Quadrupole Gas Analyzer mass spectrometer (TP-
mass spectrometry. There are indications that deposits are MS) (10). Temperature-programmed oxidation (TPO) was
formed in this interaction, and these deposits have again done with a 5% O2/Ar flow of 60 mL/min and a temperature
�
been exposed to NO and NO2. Isotopic labeling is used to ramp of 8 C/min.
identify the processes where N2 and N2O molecules are pro-
duced. Comparison of these results with the overall perfor- 2.4. Adsorption and Desorption Studies
mance in conventional catalytic tests has been used to iden-
All measurements were carried out on a TP-MS system
tify secondary factors affecting measured activities, such as
with a quartz reactor charged with 0.200 g of Fe/ZSM-5 on
site blocking by reaction products or coke.
a porous frit. The samples were first heated in 60 mL/min
� � �
5
% O2/Ar flow from 24 C to 600 C with a 8 C/min ramp,
2
. EXPERIMENTAL
followed by cooling to RT in the same gas flow. The sam-
ples were subsequently purged with a UHP He flow at
2
.1. Catalyst Preparation
6
0 mL/min for 0.5 h. Adsorption was measured with a
Fe/ZSM-5 catalysts were prepared by chemical vapor de- mixture of NO (0.7% ) in He (for “NO only”), or NO
position (3, 4). Na/ZSM-5 (UOP lot# 13923-60, Si/Al = (0.7% ) + O2 (6% ) in He (for “NO + O2”) at a flow rate
4.2, Na/Al = 0.67) was first transformed to H/ZSM-5 by of 60 mL/min. This mixed gas flow was first bypassed to get
ion exchange with a dilute NH4 � NO3 solution, followed by a background signal. The changes of the MS peak intensi-
1
�
�
calcination in an O2 flow at 550 C for 4 h. Anhydrous FeCl3 ties were monitored at an adsorption temperature of 24 C.
Aldrich, 97% ) was sublimed into H/ZSM-5, where it reacts Once equilibrium was established again, the gas flow was
(
with zeolite protons
switched back to UHP He at 60 mL/min and the catalyst
was purged for 1 h. A TPD profile was then registered in
the same flow while the temperature was increased from
+
+
↑
H
+ FeCl3 = [FeCl2] + HCl
[1]
� �
4 to 600 C at 8 C/min.
2
+
until all protons are replaced by [FeCl2] . After washing
with doubly deionized H2O to remove chlorine and drying 2.5. FTIR Measurements
�
in air, the sample was calcined in flowing O2 at 550 C for 2 h.
Spectra were collected on a Nicolet 60SX FTIR spec-
Elemental analysis by inductively coupled plasma atomic
emission shows that the Fe/Al ratio in Fe/ZSM-5 is unity.
Characterization data of such catalysts have been published
elsewhere (4).
trometer equipped with a liquid N2-cooled MCT detector.
An Fe/ZSM-5 sample was pressed into a self-supporting
2
wafer of 8–10 mg/cm and placed into a Pyrex glass cell
sealed with NaCl windows. The sample was calcined in situ
�
to 500 C in an O2/He (5% ) flow of 100 mL/min and then
2
.2. Reaction Studies
�
cooled to 200 C under the same flow. In the following ex-
�
Catalytic tests were carried out in a continuous flow reac- periments the temperature was held at 200 C. For NO + O2
tor (4). A 0.200-g Fe/ZSM-5 sample was charged in a quartz adsorption, a gas mixture of NO (0.5% ) + O2 (3% ) + He at
reactor with a porous frit. The gas flow rate was regulated a total flow rate of 100 mL/min was passed through the
by mass flow controllers, and the total flow rate was main- cell for 30 min, followed by purging with O2 (3% ) + He
tained at 280 mL/min. The inlet feed composition was 0.2% 100 mL/min for 30 min. To probe for reactivity of the ad-
NO, 3% O2, and 0.8% carbon flux (0.2% for i-C4H10, 0.27% sorbed species, a mixture of C3H8 or i-C4H10 (0.2% ) + O2
for C3H8 or C3H6, 0.8% for CH4), with He used as a diluent. (3% ) + He 100 mL/min was introduced into the cell. In situ
�
When desired, 10% H2O was added by a water saturator. To reaction measurements were carried out at 200 C under
avoid condensation of water, the whole line was heated with a flow of the same gas mixture to which 0.5% NO was
�
heating tape. The reaction temperature was increased step- added. All spectra were taken at 200 C, accumulating 50
�
� 1
wise from 200 to 500 C. The catalyst was preconditioned scans with a spectral resolution of 1 cm . The gas inside
at each temperature for 30 min before products were an- the cell immediately after scanning a sample was used as
alyzed by GC-TCD with Alltech 13X molecular sieve and the spectroscopic background. All spectra shown in this
Parapak Q columns.
paper were normalized by subtracting the spectrum of the
calcined sample from the actual spectrum measured under
flow conditions.
2
.3. Temperature-Programmed Oxidation
After testing an Fe/ZSM-5 catalyst for 3 h in the mi-
2
.6. Mass Spectrometry
croflow reactor for SCR of NOx with a given hydrocarbon at
�
3
00 C, the reactor was cooled to RT under a hydrocarbon-
A recirculating manifold equipped with a Dycor Quad-
free gas flow. The reactor was then sealed and attached rupole Gas Analyzer was used to analyze the released gases.

REDUCTION OF NOx OVER Fe/ZSM-5 CATALYSTS
173
A 0.400-g Fe/ZSM-5 catalyst was charged in a Pyrex reac-
tor which was attached to this system. Prior to the exper-
�
iment, the catalyst was calcined to 500 C in an O2 (UHP)
�
flow of 100 mL/min and cooled to 200 C. With the catalyst
held at this temperature, the system was pumped (back-
�
3
ground pressure 10 Torr) for 0.5 h before a dosed gas
mixture was admitted to the sample loop and circulated
over the catalyst. Signal intensities were normalized using
the Ar2 peak (M/e = 20) as an internal standard.
+
3
. RESULTS
3
.1. Effect of O2 Concentration on Catalytic Activity
With Cu/ZSM-5 and other zeolite supported lean burn
deNOx catalysts it is well known that the selectivity to-
ward N2 is enhanced by the presence of O2 in the feed gas
(
11, 12). Previously, we had reported on the SCR activity
of our Fe/ZSM-5 catalysts at constant O2 pressure while
varying the temperature (4). A maximum N2 yield of about
�
7
6% was found near 350 C with a feed of 0.2% NO + 0.2%
�
FIG. 2. Temperature dependence of N2 yield over Fe/ZSM-5 with dif-
ferent hydrocarbons Reaction conditions: 0.200 g Fe/ZSM-5, 0.2% NO,
% O2, 0.2% i-C4H10 (0.8% CH4, 0.27% C3H8 or C3H6), 280 mL/min.
i-C4H10 + 3% O2. In Fig. 1, the N2 yield at 350 C is plotted
against the partial pressure of O2, while the other parame-
3
ters are kept constant. In the absence of O2, NO reduction
�
at 350 C was negligible; it steeply increases with PO , reach-
2
would be 1.2% . Beyond the maximum, the combustion ac-
tivity of the hydrocarbon increases and part of the primary
product CO is oxidized to CO2. These results are similar to
those reported over Cu/ZSM-5 (13–15), but there are also
remarkable differences:
ing an N2 yield of 54% for 0.5% O2. Further increase of the
N2 yield is slower, and a maximum is reached for 2% O2;
this is followed by a slow decrease. The O2 content required
for the stoichiometric reaction
i-C4H10 + NO + 6O2 = 4CO2 + 1/2N2 + 5H2O [2]
(i) more CO is formed over Fe/ZSM-5 than over Cu/
ZSM-5;
(
ii) the decrease in N2 yield for oxygen concentrations
beyond the maximum is steeper over Cu/ZSM-5 than over
Fe/ZSM-5.
3
.2. NOx Reduction with Different Hydrocarbons
The selective catalytic reduction of NOx over Fe/ZSM-5
has been tested with CH4, C3H6, C3H8, and iso-C4H10, al-
ways using the same carbon flux. Figure 2 shows the N2 yield
as a function of reaction temperature. With CH4, the N2
yield is negligible in the whole temperature region. Among
the other hydrocarbons, C3H6 shows the lowest N2 yield,
but the ranking of C3H8 and iso-C4H10 depends on the tem-
perature. Below the temperature of maximum yield, iso-
C4H10, which is the most active reductant over Cu/ZSM-5
(
13), shows a lower N2 yield than C3H8. This unusual rank-
ing at low temperature is due to the formation of deposits,
which were characterized by TPO (not shown). With C3H8
as a reductant, there is virtually no deposit on the catalyst.
With iso-C4H10, and even more so with C3H6, large amounts
of carbonaceous deposits are detected. The CO2 peak areas
FIG. 1. The effect of oxygen concentration on the catalytic activity
�
4
are 1.1 and 3.4 � 10 mol per gram of catalyst for iso-C4H10
of Fe/ZSM-5 Reaction conditions: 0.200 g Fe/ZSM-5, 0.2% NO, 0.2%
�
i-C4H10, 280 mL/min, 350 C.
and C3H6, respectively.

1
74
CHEN, VOSKOBOINIKOV, AND SACHTLER
TABLE 1
N2 Yield (%) over Fe/ZSM-5 at Various Temperatures under Different Conditions
�
�
�
�
�
�
�
�
�
CxHy
NOx
H2O
200 C
250 C
300 C
325 C
350 C
375 C
400 C
450 C
500 C
i-C4H10
NO
NO2
NO
0
0
10%
12.1
22.2
18.1
33.8
40.7
49.7
59.2
74.0
70.4
75.9
80.4
76.6
60.1
68.5
59.0
50.6
60.5
50.5
35.9
43.0
36.3
26.9
31.2
28.1
16.7
C3H8
NO
NO2
NO
0
0
10%
3.9
3.6
2.1
19.9
21.6
6.2
60.3
66.3
33.3
59.1
65.1
43.9
55.5
57.7
44.2
50.4
50.1
42.0
44.4
43.6
37.6
33.3
32.5
29.4
25.8
26.2
22.5
Note. Reaction conditions: catalyst 0.200 g; NOx 0.2% ; O2 3% ; i-C4H10 0.2% (or C3H8 0.27% ); H2O 0% or 10% ; feed rate 280 mL/min.
With C3H8 or iso-C4H10 the catalytic activity of Fe/ZSM-5 to its original value. After purging with He at room temper-
wasalso tested usingNO2 instead ofNO, and in the presence ature and heating, NO is desorbed; the TPD profile is shown
of 10% H2O. The results are compiled in Table 1, where in Fig. 4. Only a low-temperature NO desorption peak with
the N2 yield is used for ranking the catalytic activities. As M/e = 30 was observed. This is mainly due to desorption of
reported previously (4, 5), the presence of H2O increases NO from the zeolite framework (10, 16–18). No desorption
the N2 yield at low temperature when i-C4H10 was used, but peak was observed for M/e = 32 (O2) or M/e = 46 (NO2),
with C3H8 it slightly lowers the N2 yield. With C3H8, almost which indicates absence of dissociation and disproportion-
the same activity is found with NO2 or NO. This is not the ation for this adsorption of “NO only.”
case for iso-C4H10; with that reductant the N2 yield is higher
with NO2.
Different results are observed upon exposing Fe/ZSM-5
to a flow containing 0.7% NO and 6% O2. Figure 5 shows
the NO + O2 adsorption profile. A very low concentra-
tion of NO2 was observed even when the catalyst was by-
3
.3. NO, NO + O2 Adsorption, and TPD
To understand the effect of O2 on the catalytic activity, passed, showing that the metal walls of the ducts catalyze
NO or NO + O2 adsorption and TPD experiments were a detectable reaction even at RT. However, much larger
carried out over the Fe/ZSM-5. Figure 3 shows the NO ad- amounts of NO2 (M/e = 46) were detected when the gas
sorption profile over Fe/ZSM-5. Upon exposing the catalyst mixture flowed over the catalyst. For NO (M/e = 30), sat-
to a He flow containing 0.7% of NO, saturation is reached uration was reached after 30 min exposure time, while the
within � 2 min, after which time the signal intensity returns M/e = 46 peak was still increasing. During purging at room
FIG. 3. Time dependence of M/e = 30 and M/e = 46 signal intensity
FIG. 4. TPD from Fe/ZSM-5 catalyst following “NO only” adsorption
at 24 C.
�
�
upon exposing “NO only” over Fe/ZSM-5 catalyst at 24 C.

REDUCTION OF NOx OVER Fe/ZSM-5 CATALYSTS
175
3
.4. Identification of Adsorption Complexes
and Catalyst Deposits by FTIR
3
.4.1. Adsorption of NO + O2. The adsorption com-
plexes formed upon exposing the catalyst to NO + O2 at
�
2
00 C were examined by FTIR spectroscopy. This temper-
ature was chosen to exclude weak adsorption complexes.
Figures 7a–7c show the spectra for different exposure times
to the NO + O2 flow. Four bands are visible at 2131, 1878,
�
1
� 1
1
625, and 1570 cm . The band at 2131 cm has been as-
+
signed to the N–O stretching vibration of NO2 adsorbed
on the Brønsted acid sites of the zeolite (20–23). Most re-
cently, it was reassigned to NO occupying Brønsted acid
+
sites in the zeolite (24). Its intensity is very low and reaches
saturation quickly, in accordance with our previous finding
that the concentration of Brønsted sites on this catalyst is
�
1
very low (4). The band at 1878 cm has been assigned to
the iron mononitrosyl complex formed by the interaction
of NO with iron (25–27). Since the vibration frequency is
�
1
about the same as for gas phase NO (1880 cm (28)), the
interaction of Fe with this ligand should be very weak. The
FIG. 5. Time dependence of M/e = 30 and M/e = 46 signal intensity
�
1
�
bands at 1625 and 1570 cm cannot be assigned with cer-
tainty at present, because it is impossible to study the entire
upon exposing “NO + O2” over Fe/ZSM-5 catalyst at 24 C.
�
1
spectral region. The frequency of the 1625 cm band is very
near the asymmetric stretching frequency of gaseous NO2
temperature, both peaks decreased much more slowly than
in the case of NO only. During that period, the intensity
ratio of the peaks at M/e = 46 and M/e = 30 increased,
which suggests that NO2 is more strongly adsorbed. Clearly,
Fe/ZSM-5 catalyzes the oxidation of NO to NO2:
�
1
(
1610 cm (28)). We tentatively assign it to a nitro group
�
1
ligated to an iron ion (7, 25, 29). The band at 1570 cm can
be assigned to a nitrate group on an iron site (25, 30). For
the time being, we shall use the term “nitro/nitrate groups”
for both bands. The intensity of the bands at 1878, 1625,
NO + 1/2O2 = NO2.
[3]
�
1
and 1570 cm changed with increasing exposure time; the
The slow return to the baseline suggests formation of N2O3
and/or N2O4, which decompose to NO and NO2 during de-
sorption.
The TPD profile following adsorption of NO + O2 is pre-
sented in Fig. 6. A peak with mass M/e = 30 at low temper-
ature is reminiscent of the peak in Fig. 3, but its height and
area are much larger. This peak is due to the sum of the
parent mass of NO and the M/e = 30 fragment of NO2, as
it is well known that NO2 will decompose to NO inside the
ionization chamber of the mass spectrometer (19). Under
the M/e = 30 peak, small peaks appear for M/e = 32 (O2)
and M/e = 46 (NO2). A second family of TPD peaks for all
�
these masses is registered at � 300 C. However, their shapes
and Tmax values are different from each other. The peak of
�
M/e = 46 (NO2) is rather symmetric with a Tmax = 306 C,
�
while the M/e = 30 peak is tailed with a Tmax = 297 C. Com-
parison with mass spectra of fairly pure NO2 suggests that
the M/e = 30 signal is not entirely due to a fragment of the
M/e = 46 parent, but part of it is the parent mass of NO
molecules desorbed from the catalyst.
Upon exposing the catalyst to NO2, the subsequent TPD
profile is identical to that shown in Fig. 6 after adsorption of
NO + O2. Apparently, the same NOy adsorption complexes
are obtained in both cases.
FIG. 6. TPD from Fe/ZSM-5 catalyst following “NO + O2” adsorp-
�
tion at 24 C.

1
76
CHEN, VOSKOBOINIKOV, AND SACHTLER
�
�
FIG. 7. FTIR spectra taken at 200 C of NOy adsorbed Fe/ZSM-5 sam-
FIG. 8. FTIR spectra taken at 200 C after NO + O2 adsorption and
ple under a flow of 0.5% NO + 3% O2 + He for 5 min (a), 15 min (b),
purge with O2/He (a), subsequently exposed to a flow of 0.2% C3H8 + 3%
O2 + He flow for 5 min (b), 15 min (c), and 30 min (d). The inset depicts a
3
3
0 min (c); followed by purge with 3% O2 + He for 5 min (d), 15 min (e),
�
1
0 min (f).
plot of the band relative intensities at 1625 and 1570 cm as a function of
time.
�
1
band at 1878 cm decreased, while the bands at 1625 and
1
�
1
Figure 9 shows the reactivity of the adsorbed nitro/nitrate
groups toward iso-butane. Clearly, this hydrocarbon reacts
vigorously with these groups. Already after 5 min, both
570 cm increased. This could indicate that mononitro-
syl and nitro/nitrate groups are adsorbed at the same iron
site. When the catalyst is first exposed to the NO + O2 flow,
the NO2 concentration is very low, so that NO will be ad-
sorbed. Once NO is oxidized to the more strongly adsorbed
nitro/nitrate groups, the intensity of the NO band will de-
crease while that of the nitro/nitrate groups will increase.
Figures 7d–7f show the stability of the adsorption bands
upon purging with O2/He at 200 C. In contrast, the bands at
131 and 1878 cm get weaker and vanish after 15 min of
purging. Obviously, NO and NO are weakly adsorbed. The
�
1
bands have decreased dramatically. A band at 1876 cm
�
�
1
2
+
�
1
intensity of the bands at 1625 and 1570 cm also decreases
somewhat, but does not vanish even after 30 min of purging,
indicating that the nitro/nitrate groups are fairly strongly
adsorbed.
3.4.2. Reactivity of the nitro/nitrate groups toward hy-
drocarbons. Whereas Fig. 7 illustrates the thermal sta-
bility of the adsorption complexes, the spectra in Fig. 8
show their chemical reactivity toward propane. Clearly, the
nitro/nitrate groups interact chemically with propane at
�
2
00 C. A plot of the band intensities, which were measured
as peak heights and normalized by their initial intensities,
as a function of time is shown in the same figure. After
3
0 min, the band intensity has dropped to a very low level,
�
1
and new bands appear in the 1700–1300 cm region. In
the C–H stretching region, a weak broad band appears at
�
FIG. 9. FTIR spectra taken at 200 C after NO + O2 adsorption and
purge with O2/He (a), subsequently exposed to a flow of 0.2% i-C4H10 +
�
1
� 1
2
950 cm . Another weak band at 1876 cm might be as-
3
% O2 + He flow for 5 min (b), 15 min (c), 30 min (d), off i-C4H10, with
signed to a mononitrosyl complex (see below).
a flow of 3% O2 + He for 30 min (e).

REDUCTION OF NOx OVER Fe/ZSM-5 CATALYSTS
177
appears immediately when iso-butane is introduced. Its in-
tensity decreases with time. As its frequency is the same as
that of adsorbed NO, it is possible that the reduction of the
nitro/nitrate groups regenerates the mononitrosyl complex.
After more extended exposure to iso-butane, a rather com-
�
1
plicated set of bands appear in the 1700–1300 cm region;
they are indicative for the formation of some polymeric
�
1
deposit on the catalyst. At 2968 cm , a C–H stretching
vibration appears which can be attributed to adsorbed iso-
butane. This band disappears after successive purging with
O2/He (Fig. 9e). In contrast, the bands at low frequencies
are still present basically retaining their intensity. Appar-
ently, the deposit is more stable than the species responsible
�
1
for the 2968 cm band.
Although the deposit is rather stable against O2/He flow
�
at 200 C, it displays a remarkable reactivity toward NO2
(
or NO + O2) as shown by the three spectra in Fig. 10.
Feature (a) was registered after reaction of NOy with iso-
C4H10, followed by purging with He at 200 C. The band at
1
�
�
1
878 cm that might be due to mononitrosyl complex is
still visible, which suggests that it is thermally stable. This
increased thermal stability of the mononitrosyl complex in
comparison to that in Fig. 7 is assumed to be caused by a
FIG. 11. FTIR spectra taken at 200�C under a flow of 0.2% C H +
3
8
0.5% NO + 3% O2 + He flow for 5 min (a), 15 min (b), 30 min (c), sub-
sequently under a flow of 0.5% NO + 3% O2 + He for 30 min (d).
partial reduction of the iron from Fe3 to Fe by the hy-
+
2 +
drocarbon. It is known that this reduction does not change NO + O2 flow. Meanwhile, some carbonate (bands 1437,
�
1
� 1
the position of the 1876 cm band, but NO is ligated 1357 cm ) was formed on the catalyst. In addition to the
2
+
3 +
more strongly to Fe than to Fe (31, 32). Bands in the mononitrosyl complex, other features are observed. As in
700–1300 cm region are due to a polymeric deposit. Re- Fig. 7, they include NO and nitro/nitrate groups.
�
1
+
1
markably, their intensity decreases upon exposure to the
3
.4.3. FTIR spectra under CxHy + NO + O2 flow.
Figure 11 shows the FTIR spectra registered in situ under a
�
flow of C3H8 + NO + O2 at 200 C. The bands at 2131, 1877,
�
1
1
625, and 1570 cm , shown in Fig. 7 under NO + O2, are
� 1
again observed. However, the intensity of the 1877 cm
�
1
band is higher, while that of the 1625 and 1570 cm bands is
lower. Clearly, the nitro/nitrate groups react with propane.
Some new bands appear during reaction in the region of
�
1
1
700–1300 cm , which is due to the formation of carbonate
and carbonaceous deposits. However, their intensity is low.
Figure 12 shows the FTIR spectra under a flow of iso-
�
C4H10 + NO + O2 at 200 C. Replacing propane by iso-
�
1
butane causes rather remarkable changes. The 1876 cm
band, assigned to a mononitrosyl complex, is only visible
in the initial stage of exposing the catalyst to this gas. It
�
1
is difficult to distinguish the bands at 1625 and 1570 cm
,
which are assigned to nitro/nitrate groups, from the compli-
�
1
cated absorption bandsofthe deposit in the 1700–1300cm
region. These increase rapidly with time. Meanwhile, a
�
1
broad band emerges at 3145 cm ; this can be assigned to a
C–H stretching vibration of an unsaturated compound. As-
signment of this band to an O–H stretching and/or an N–H
stretching vibration is also possible (33). Also, the C–H
�
FIG. 10. FTIR spectra taken at 200 C after exposing a NOy adsorbed
�
1
stretching vibration at 2967 cm , assigned to adsorbed
sample to a flow of 0.2% i-C4H10 + 3% O2 + He for 0.5 h, followed by
purge with He for 30 min (a), subsequently exposed to a flow of 0.5%
NO + 3% O2 + He for 5 min (b), 15 min (c), and 30 min (d).
�
1
iso-butane, is visible. In the region of 2300–2100 cm , a
�
1
band at 2255cm appeared immediatelyupon exposingthe

1
78
CHEN, VOSKOBOINIKOV, AND SACHTLER
Mass spectrometry was therefore used to analyze the re-
14
14
leased gas. In order to distinguish N2 and CO, N2O and
CO2, the labeled molecule 15NO was used in this part of the
work. This led to some surprising observations.
First, calcined Fe/ZSM-5 catalysts were exposed to a cir-
culating gas mixture of 10 Torr 15NO + 80 Torr O2 + 10 Torr
Ar for 1 h, followed by evacuation for 0.5 h. These samples,
now loaded with nitro/nitrate groups, were subsequently
�
exposed at 200 C to a circulating mixture of 10 Torr hydro-
carbon, 80 Torr O2, and 10 Torr Ar (used as internal stan-
dard). With one sample the hydrocarbon was propane; with
the other it was iso-butane. With propane, a small amount of
N2 and N2O was released when the hydrocarbon contacted
the NOy-loaded catalyst. Figure 13 shows the relative sig-
nal intensity of 15N2 (M/e = 30) and N2O (M/e = 46) as
a function of time. Clearly, propane reacts with the NOy
species, consistent with the FTIR results shown in Fig. 8.
Initially, the signal intensity increased quite steeply, later
leveling off. This “level-off” was not due to the lack of hy-
drocarbon, as the intensity of signal M/e = 29 (fragment of
C H ) was still rather high and that of M/e = 44 (CO ) was
15
�
FIG. 12. FTIR spectra taken at 200 C under a flow of 0.2% i-
3
8
2
C4H10 + 0.5% NO + 3% O2 + He flow for 5 min (a), 15 min (b), 30 min (c), still increasing (not shown). Upon using iso-C H instead
4
10
subsequently under a flow of 0.5% NO + 3% O2 + He for 30 min (d),
following (c) under a flow of 0.5% NO2 + 3% O2 + He for 30 min (e),
calcined sample under a flow of 0.2% i-C4H10 + 3% O2 + He flow for
15
15
of C3H8, neither the N2 nor the N2O signal intensity
changed much, although the FTIR data in Fig. 9 show that
iso-C4H10 reacts swiftly with the NOy. It is possible that this
quick reaction leads to nitrogen-containing organic com-
3
0 min (f).
catalyst to the gas mixture. Its intensity reached a maximum pounds, which are strongly adsorbed on the catalyst.
and then decreased with exposure time. Another band at
The above samples, which had almost the same history
238 cm continued to grow. Bands in this region can be as- except that one catalyst had been contacted with propane
�
1
2
cribed to an X == Y == Z and/or an X� Y stretching vibration. and the other with iso-butane, were then evacuated for 0.5 h
It has been reported that organic nitriles, cyanates, and/or
isocyanates are potential intermediates of SCR of NO with
hydrocarbons (34–41). It is clear from Fig. 12 that a deposit
is formed on the Fe/ZSM-5 catalyst and that its quantity is
larger upon exposure to iso-C4H10 + NO + O2 than if NO
is not present in the gas (Figs. 12c and 12f). This deposit
appears to block sites for NO adsorption. Thus, reaction
(
3) that is catalyzed by the iron sites will be hindered. This
deposit is not very reactive towards NO + O2 when the rate
of reaction (3) is negligible, as can be seen from Fig. 12d.
After 30 min of flowing NO + O2 over the partially cov-
ered Fe/ZSM-5 catalyst, the band attributed to iso-butane
disappears, while bands attributed to polymeric deposits
decrease only insignificantly. However, this deposit reacts
more quickly with NO2 + O2 flow, as shown in Fig. 12e. Un-
der the same conditions, the intensity of the bands corre-
sponding to the deposits decreases more severely when NO
is replaced by NO2. This result confirms that the catalytic
oxidation of NO to NO2 is suppressed by the deposit.
3
.5. Identification of Reaction Steps by Mass Spectrometry
While the FTIR results show that the nitro/nitrate groups
react with C3 and C4 hydrocarbons, they do not provide in-
formation on the gas phase products of this interaction. over an NOy-adsorbed Fe/ZSM-5 sample.
FIG. 13. Time dependence of the relative signal intensity upon circu-
lating a mixture of 10 Torr C3H8 (or i-C4H10) + 80 Torr O2 + 10 Torr Ar

REDUCTION OF NOx OVER Fe/ZSM-5 CATALYSTS
179
FIG. 14. Time dependence of the relative signal intensity upon circu-
lating a mixture of 10 Torr NO + 80 Torr O2 + 10 Torr Ar over an Fe/ZSM-
FIG. 15. Time dependence of the relative signal intensity: Solid
symbols—upon circulating a mixture of 10 Torr NO + 80 Torr O2 +
10 Torr Ar over an Fe/ZSM-5 sample, on which a CxHyO deposit has
been formed with i-C4H10 + O2; open symbols—upon circulating a mix-
ture of 10 Torr NO + 10 Torr Ar over an Fe/ZSM-5 sample, on which a
nitrogen-containing deposit has been formed with i-C4H10.
5
sample, on which a nitrogen-containing deposit has been formed with
C3H8 or i-C4H10.
and exposed to a circulating mixture of 10 Torr 15NO +
8
0 Torr O2 + 10 Torr Ar. As shown in Fig. 14, this reaction
15
of NOx gave rise to gas evolution: a large amount of N2
15
and some N2O are formed in both cases. The amount of
15
N2 is larger with the sample that was previously exposed
to iso-C4H10. Remember that FTIR had also shown more
deposit formation with iso-C4H10 than with C3H8.
This result triggers the question of whether the carbona-
ceous deposit merely acts as a reductant for impinging NO
or NO2, as has been proposed previously (9). To check this,
we created a deposit by contacting a mixture of 10 Torr
iso-C4H10 + 80 Torr O2 + 10 Torr Ar with a freshly calcined
Fe/ZSM-5 catalyst and exposed it, after evacuation, to the
same recirculating NOx-containing mixture. As shown in
15
Fig. 15 (solid symbols), very little N2 was formed in this
case; however, some 15N2O is formed, the amount of which
is comparable to that in Fig. 14.
It thus seems that for N2 evolution from the reaction
of a deposit with NOx, the deposit must be formed from
adsorbed NOy groups. In order to clarify whether nitrogen-
containing groups of the deposit are directly involved in the
formation of N2, the following experiment using labeled ni-
trogen was carried out. First a deposit was laid down by
14
the procedure mentioned above, using NO for the prepa-
ration of the NOy groups. This deposit-laden sample was
15
then exposed to a circulating mixture of 10 Torr NO +
0 Torr O2 + 10 Torr Ar. The result of this crucial exper-
iment is shown in Fig. 16. Again, N2 and N2O are pro-
FIG. 16. Time dependence of the relative signal intensity upon cir-
8
15
culating a mixture of 10 Torr NO + 80 Torr O2 + 10 Torr Ar over an
14
Fe/ZSM-5 sample, on which a N-containing deposit has been formed
duced, but the isotopically mixed molecule 14N N is the with i-C4H10.
15

1
80
CHEN, VOSKOBOINIKOV, AND SACHTLER
predominant N-containing gas phase product. Small (II) nitrate show bands below 1600 cm 1. Therefore, we pre-
�
amounts of 15N2 and N2O are also observed, but only fer to assign the 1628 cm band to a nitro group, because
in quantities similar to those obtained with the synthetic its frequency is nearly the same as that of gas phase NO2.
nitrogen-free deposit in Figure 15. A very small amount of The high thermal stability of this group on Cu/ZSM-5 may
15
� 1
14
15
isotopically mixed N NO is also measured; its intensity be due to the unique zeolite structure of ZSM-5. Still, we
is even lower than that of 15N2O. The intensive signal at use the cautious term nitro/nitrate in this paper. A band at
14
� 1
M/e = 28 is due to the formation of CO rather than N2. 1605 cm was observed over NO2 adsorbed �-Fe2O3, and
Clearly, oxidation of the nitrogen-containing deposit results it has been assigned to adsorbed nitro group (25). Consider-
�
1
in the formation of CO and CO2, in addition to N2 and H2O. ing these facts, we tentatively assign the band at 1625 cm
These results show that the nitrogen-containing deposit is on (NO + O2) laden Fe/ZSM-5 as a nitro group, while as-
�
1
highly reactive toward NO2 (or NO + O2), while it is much signing the band at 1570 cm as a nitrate group at an iron
�
less reactive toward NO alone. As can be seen from Fig. 15 site. These nitro/nitrate groups are stable at 200 C upon
15
(
open symbols), much less N2 and only negligible amounts purging with He, as shown in Figs. 6 and 7. At higher tem-
of N2O are formed with NO only. This is consistent with perature, they will decompose forming NO, NO2, and O2,
15
the FTIR results.
as shown by the TPD profiles in Fig. 6.
These nitro/nitrate groupsare potentialreaction interme-
diates (7, 48). They show high reactivity toward hydrocar-
bons (see Figs. 8 and 9). In reacting with these groups, iso-
4
. DISCUSSION
Dioxygen has been found to be crucial for the SCR of butane is more active than propane. With iso-butane, the
NO with NH3 or hydrocarbon reductants over various cata- nitro/nitrate groups were virtually consumed within 5 min,
lysts (11). The present work shows that this is also true for while it took about 30 min when propane was used. This
the Fe/ZSM-5 catalyst. Oxygen reacts swiftly with NO to is in agreement with the proposed reaction mechanism for
form NO2, which then interacts with the catalyst to form SCR of NO with hydrocarbons; i.e., the difficult, possibly
chemisorption complexes of the type NOy where y � 2. This rate limiting step in the reduction of NO with alkane is the
is basically the same chemistry as that shown previously rupture of the first C–H bond (6, 49–52). It is obvious that
for other ZSM-5 supported metals (42). In the case of Cu/ this C–H bond fission is much easier with iso-butane than
ZSM-5 there is good evidence that [Cu–O–Cu]2 ions are with propane, since iso-butane, unlike propane, has a ter-
+
the active sites for NO oxidation to NO2 (43). It is possible tiary C–H bond.
that [HO–Fe–O–Fe–OH]2 ions (4, 5) fulfill the same task
+
Over Cu/ZSM-5, Adelman et al. (7) showed that the
in Fe/ZSM-5. An additional beneficial effect of oxygen is nitro/nitrate groups are inactive toward CH4. Over Fe/
in situ removal of excessive carbonaceous deposits (44). ZSM-5, displaying the same bands, the same lack of activity
The present results show that strongly adsorbed NOy toward methane is found in the present work.
complexes are formed on the Fe/ZSM-5 catalyst only if NO2 The reaction between hydrocarbon and nitro/nitrate
is fed or when NO2 is formed in situ from NO + O2. The groups will lead to the formation of organic nitroso and/or
NOy complexes are manifest from IR absorption bands at nitro complexes, as has been proposed before (8, 37, 53,
�
1
1
625 and 1570 cm , which are similar to those observed 54). Organic nitro compounds have very strong IR bands
�
1
� 1 � 1
at about 1550 cm and 1385–1365 cm , while nitroso
by Adelman et al. over Cu/ZSM-5 (1628, 1594, 1572 cm
7). There is some debate on the assignment of these bands. compounds absorb at 1590–1540 cm (33). In the 1700–
Adelman et al. assigned the 1628cm band to a nitro group, 1300 cm region, complicated absorption bands appear in
while bands at 1594 and 1572 cm were assumed to be Figs. 6, 7, and 10. Presumably, they will contain contribu-
caused by nitrate groups. Other researchers observing sim- tions from such organic nitroso and/or nitro compounds.
)
�
1
(
�
1
� 1
�
1
ilar bands over Cu/ZSM-5 (45) assigned them to different
Nitroso compounds can easily undergo spontaneous iso-
kinds of nitrate ions. The main argument for the latter as- merization to oximes. Over Cu/ZSM-5, Beutel et al. (8)
�
1
14
signment is that the stability of the 1628 cm is not eas- found that N-labeled acetone oxime readily reacts with
15
14 15
14 15
ily reconciled with the assumption of a nitro group. How-
NO, forming N N and N NO. They propose that this
ever, assignment to a nitrate ion is also objectionable, since might be a reaction path in SCR of NOx. The results in
�
1
Adelman et al. did not observe a band at 1628 cm after Figs. 8 and 13 confirm that thishypothesisisalso valid for Fe/
impregnation SiO2 or Na/ZSM-5with Cu(NO3)2. The bands ZSM-5. When C3H8 wasused asthe reductant, 2-nitroso-(or
appeared only upon heating the Na/ZSM-5-supported sam- nitro-)propane might be formed, which readily isomerizes
�
ple to 100 C, at which temperature the nitrate started to de- to its oxime. This oxime and/or its hydrolysis compounds
compose. London et al. (46) studied the adsorption of NO2 could further react with NO2, forming N2, N2O, CO, CO2,
on copper oxide. They did not observe this band. Logan and H2O. Therefore, when the NOy-adsorbed sample is first
et al. (47) showed that both the �- and �-forms of copper exposed to the C3H8 + O2 flow, an attenuation of the IR

REDUCTION OF NOx OVER Fe/ZSM-5 CATALYSTS
181
bands corresponding to the nitro/nitrate groups is observed
The amount of N2O released upon circulating NO + O2
see Fig. 8). Simultaneous release of N2 and N2O is found over the deposit-covered sample is nearly identical, re-
see Fig. 13). Later, the amount of NOy decreases, the pro- gardless of whether the deposit is a nitrogen-containing or
(
(
cess leading to N2 formation is suppressed by lack of NO2. nitrogen-free complex, although the amount of N2 formed
Instead nitrogen-containing organic deposits are formed on is markedly different (compare the open plots in Fig. 14
the catalyst (see Fig. 8d). This deposit is stable against O2 at and the solid plots in Fig. 15). Also, in the labeled experi-
�
14 15
15
2
(
00 C, but is active toward NO2 (or NO + O2 when reaction ment, less mixed N NO than isotopically pure N2O is
3) occurs) as shown in Figs. 10 and 14. NO2 reacts quickly observed, in contrast to the predominance of the isotopi-
cally mixed 14N N in Fig. 16. These results indicate that
15
with the deposit forming N2 and some N2O.
Reaction of iso-butane with the nitro/nitrate groups N2O is presumably not an intermediate in the formation of
has to result in different products. While initially tertiary N2.
nitroso and/or nitro compounds will be formed, these
While the nitrogen-containing deposit is crucial for the
molecules cannot isomerize easily because there is no H SCR of NOx, its accumulation may also block the sites for
atom in the �-position to the NO group. Instead of forming NO adsorption, as shown in Fig. 12. Therefore, the overall
an oxime, these molecules apparently polymerize to a SCR activity depends, in a rather precarious way, on the
nitrogen-containing deposit on the catalyst. Very little N2 amount of such deposits in the steady state. Suppression of
or N2O is released in this process (see Figs. 9 and 13). How- NO oxidation to NO2 by site blocking with deposit appears
ever, the most exciting finding of the present research is that to be the easiest rationalization of the nontrivial depen-
this deposit is highly reactive toward NO2. This reaction dence of the SCR rate on the nature of the hydrocarbon,
leads to the formation of gaseous N2. Formation of an active shown in Fig. 2. Also, the temperature dependence of the
nitrogen-containing organic intermediate when C3H6 was relative effectiveness of the reductants between propane
used as a reductant has been reported by Yogo et al. over and iso-butane can be rationalized. At low temperature,
H–Fe–silicalite (55) and by Guyon et al. over Cu/ZSM-5 the N2 yield is lower with iso-C4H10 than with C3H8, be-
(
56). Both works show that N2 is formed upon exposing the cause more deposit is formed with iso-C4H10. At higher
intermediate to NO2. The present work with labeled mo- temperature, when the excess deposit (or its remainder af-
lecules unambiguously shows that 14N N is the predomi- ter N2 evolution) is burnt off, the reactivity sequence of
nant product, indicating that the two nitrogen atoms have these hydrocarbons is reversed and iso-C4H10 is the better
different histories; i.e., one nitrogen atom comes from the reductant. The same reasoning holds true for C3H6, which
deposit, while the other one comes from the gas phase NO2. forms more coke on the catalyst, giving a lower N2 yield as
Details of this reaction are still unclear. The deposit may a result. Propene has been reported to be less effective than
undergo severalreactions, such asoxidation, hydrolysis, and propane also over Cu/ZSM-5 (44, 59).
15
isomerization. Isocyanates and organic nitriles, and possi-
With propane as the reductant, the catalyst surface re-
bly also cyanates, appear as plausible groups from the FTIR mains rather “clean,” as follows from the FTIR spectra
spectra in Fig. 12. These intermediates have been claimed to (Fig. 11) and the TPO result. Mononitrosyl, nitro/nitrate
be active in NO reduction (34–41). One can also speculate groups and carbonates are identified on the catalyst. This
that hydroxylamine is formed upon hydrolysis. One hun- suggests that with propane the rate limiting step may be the
dred years ago, it was shown by Haber that nitrobenzene C–H bond rupture. The equilibrium between NO and NO2
reacts with hydroxylamine to form azo-benzene (57); azo remains established and, consequently, the N2 yield does
and diazo compounds are, of course, easily decomposed, not change upon replacing NO with NO2. With iso-C4H10
releasing N2. Formation of diazonium or diazo compounds as the reductant, the N2 yield is higher with NO2 than with
would also be conceivable if the deposit contains amine NO, because the sites which catalyze the NO oxidation are,
groups, which react with HONO or N2O3 (58). HONO can in part, blocked.
be easily formed by the reaction
Addition of water vapor to the feed is known to sup-
press the formation of carbonaceous deposits (60). CO is
[4] formed in the reaction between the deposit and H2O (4, 5).
The present results show that addition of 10% H2O to
RH + NO2 → R�+ HONO.
This possibility seems to be supported by the fact that only the iso-C4H10 + NO + O2 feed enhances the SCR activ-
NO2, and not NO or O2, reacts quickly with the deposit. ity at low temperature. No such effect was found for a
While this part is still speculative, the present results leave C3H8 + NO + O2 feed, in accordance with the fact that this
little doubt that interaction of NO2 with the N-containing gas leaves the catalyst surface rather “clean.” Instead, H2O
deposit formed from iso-C4H10 and adsorbed nitro/nitrate will compete with NO for the iron sites or interfere with the
complexes is a very efficient step in the reduction of NOx oxidation of NO to form NOy. These actions lead to a slight
to N2 by this hydrocarbon.
decrease in the activity. However, these adverse effects are,

1
82
CHEN, VOSKOBOINIKOV, AND SACHTLER
fortunately, very small for the Fe/ZSM-5 catalyst in contrast 10. Adelman, B. J., Lei, G.-D., and Sachtler, W. M. H., Catal. Lett. 28, 119
1994).
(
to Cu/ZSM-5 and other catalysts (4).
1
1
1
1. Shelef, M., Chem. Rev. 95, 209 (1995).
2. Fritz, A., and Pitchon, V., Appl. Catal. B 13, 1 (1997).
3. Iwamoto, M., and Mizuno, N., Proc. Inst. Mech. Eng. D 207, 23
5
. CONCLUSIONS
(
1992).
Fe/ZSM-5 catalysts prepared via sublimation are active 14. Chajar, I., Primet, M., Praliaud, H., Chevrier, M., Gauthier, G., and
Mathis, F., Catal. Lett. 28, 33 (1994).
for the reduction of NOx to N2 with hydrocarbons such
1
1
1
1
5. Petunchi, J. O., Sill, G., and Hall, W. K., Appl. Catal. B 2, 303 (1993).
6. Schay, Z., and Guczi, L., Catal. Today. 17, 175 (1993).
7. Chang, Y. F., and McCarty, J. G., J. Catal. 165, 1 (1997).
8. Matyshak, V. A., Ilichev, A. N., Ukharsky, A. A., and Korchak, V. N.,
J. Catal. 171, 245 (1997).
as propane, iso-butane, or propene, but not with methane.
They remain active in the presence of a large excess of H2O.
The NOx reduction rate is negligible in the absence of O2;
it increases steeply with increasing O2 content of the feed
and passes through a maximum. The equilibrium between
NO + O2 and NO2 is swiftly established over a clean cata-
lyst, but formation of deposits impedes this reaction. The
formation of N2 appears to be a multistep process. First,
NO + O2 form chemisorption complexesofthe generaltype
NOy, with y � 2. These react with the hydrocarbon and,
dependent on the chemical nature of this hydrocarbon, a
deposit may be formed on the catalyst containing C, O, H,
and N atoms. An intensive formation ofN2 isobserved when
NO2 reacts with such a deposit, in particular if the latter was
created from iso-C4H10. Isotopic labeling shows that one N
atom in every N2 molecule formed in this step comes from
the deposit, while the other comes from NO2. N2O is not
a precursor of N2. The intermediate formation of a diazo
1
9. Corna, A., and Massot, R. (Eds.), “Compilation of Mass Spectroscopy
Data.” Heyden, London, 1966.
2
2
2
0. Chao, C. C., and Lunsford, J. H., J. Am. Chem. Soc. 93, 71 (1971).
1. Teranishi, R., and Decius, J. C., J. Chem. Phys. 22, 896 (1954).
2. Hoost, T. E., Laframboise, K. A., and Otto, K., Catal. Lett. 33, 105
(1995).
3. Szanyi, J., and Paffett, M. T., J. Catal. 164, 232 (1996).
4. Hadjiivanov, K., Saussey, J., Freysz, J. L., and Lavalley, J. C., Catal.
Lett. 52, 103 (1998).
5. Busca, G., and Lorenzelli, V., J. Catal. 72, 303 (1981).
6. Aparicio, L. M., Hall, W. K., Fang, S.-M., Ulla, M. A., Millman, W. S.,
and Dumesic, J. A., J. Catal. 108, 233 (1987).
7. Amiridis, M. D., Puglisi, F., Dumesic, J. A., Millman, W. S., and Topsøe,
N.-Y., J. Catal. 142, 572 (1993).
2
2
2
2
2
2
2
8. Nakamoto, K., “Infrared Spectra of Inorganic and Coordination Com-
pounds” (3rd ed.), p. 109. Wiley, New York, 1978.
9. Valyon, J., and Hall, W. K., J. Phys. Chem. 97, 1204 (1993).
complex is suggested. Deposits can also block catalyst sites, 30. Rochester, C. H., and Topham, S. A., J. Chem. Soc. Faraday Trans. 1
7
5, 872 (1979).
including those needed for the NO oxidation to NO2. The
rate limiting step, therefore, depends on the nature of the
hydrocarbon used and the type and quantity of the deposit.
As deposits are oxidized by O2 and volatilized by H2O at
high temperature, the relative efficiency of different hydro-
carbons in NOx reduction over Fe/ZSM-5 depends on the
temperature and the H2O content of the feed.
3
1. Segawa, K.-I., Chen, Y., Kubsh, J. E., Delgass, W. N., Dumesic, J. A.,
and Hall, W. K., J. Catal. 76, 112 (1982).
2. Guglielminotti, E., J. Phys. Chem. 98, 9033 (1994).
3. Socrates, G., “Infrared Characteristic Group Frequencies” (2nd ed.).
Wiley, New York, 1994.
3
3
3
3
3
4. Bell, V. A., Feeley, J. S., Deeba, M., and Farrauto, R. J., Catal. Lett. 29,
15 (1994).
5. Hayes, N. W., Gr u¨ nert, W., Hutchings, G. J., Joyner, R. W., and Shpiro,
E. S., J. Chem. Soc. Chem. Commun., 531 (1994).
6. Hayes, N. W., Joyner, R. W., and Shpiro, E. S., Appl. Catal. B 8, 343
ACKNOWLEDGMENTS
(
1996).
3
3
7. Yokoyama, C., and Misono, M., J. Catal. 150, 9 (1994).
8. Radtke, F., Koeppel, R. A., and Baiker, A., J. Chem. Soc. Chem. Com-
mun., 427 (1995).
Financial support for this research by an unrestricted grant from the
Ford Motor Corporation is gratefully acknowledged. We thank Dr. Steve
Wilson of UOP, Dr. Arno Tißler ofALSI-Penta Zeolithe GmbH, Dr. Karin
Bartels of Degussa, and Dr. Armin Pfenninger of Uetikon Chemie A.G.
for kindly donating ZSM-5 samples.
3
4
9. Radtke, F., Koeppel, R. A., Minardi, E., and Baiker, A., J. Catal. 167,
127 (1997).
0. Li, C., Bethke, K. A., Kung, H. H., and Kung, M. C., J. Chem. Soc.
Chem. Commun., 813 (1995).
REFERENCES
41. Aylor, A. W., Lobree, L. J., Reimer, J. A., and Bell, A. T., Stud. Surf.
Sci. Catal. 101, 661 (1996).
1
2
3
. Feng, X., and Hall, W. K., Catal. Lett. 41, 45 (1996).
. Feng, X., and Hall, W. K., J. Catal. 166, 368 (1997).
42. Hamada, H., Kintaichi, U., Sasaki, M., and Ito, T., Appl. Catal. 70, L15
(1991).
. Sachtler, W. M. H., and Chen, H.-Y., U.S patent pending, Appl. 43. Lei, G.-D., Adelman, B. J., S a´ rk a´ ny, J., and Sachtler W. M. H., Appl.
No. 60/052,382 (1997).
Catal. B 5, 245 (1995).
4
5
6
. Chen, H.-Y., and Sachtler, W. M. H., Catal. Today 42, 73 (1998).
. Chen, H.-Y., and Sachtler, W. M. H., Catal. Lett. 50, 125 (1998).
. Cowan, A. D., D u¨ mpelmann, R., and Cant, N. W., J. Catal. 151, 356
44. d’Itri, J. L., and Sachtler, W. M. H., Appl. Catal. B 2, L7 (1993).
45. Hadjiivanov, K., Klissurski, D., Ramis, G., and Basca, G., Appl. Catal.
B 7, 251 (1996).
(
1995).
46. London, J. W., and Bell, A. T., J. Catal. 31, 32 (1973).
7
8
9
. Adelman, B. J., Beutel, T., Lei, G.-D., and Sachtler, W. M. H., J. Catal. 47. Logan, N., and Simpson, W. B., Spectrochim. Acta 21, 857 (1965).
1
58, 327 (1996).
. Beutel, T., Adelman, B. J., and Sachtler, W. M. H., Catal. Lett. 37, 125
1996).
. Walker, A. P., Catal. Today 26, 107 (1995).
48. Beutel, T., Adelman, B. J., Lei, G.-D., and Sachtler, W. M. H., Catal.
Lett. 32, 83 (1995).
(
49. Adelman, B. J., Beutel, T., Lei, G.-D., and Sachtler, W. M. H., Appl.
Catal. B 11, L1 (1996).

REDUCTION OF NOx OVER Fe/ZSM-5 CATALYSTS
183
5
5
0. Witzel, F., Sill, G. A., and Hall, W. K., J. Catal. 149, 229 (1994).
1. Lukyanov, D. B., Sill, G., d’Itri, J. L., and Hall, W. K., J. Catal. 153, 265
56. Guyon, M., Chanu, V. Le., Gilot, P., Kessler, H., and Prado, G., Appl.
Catal. B 8, 183 (1996).
(
1995).
57. Haber, F., Z. Elektrochem., 506 (1898).
58. March, J., “Advanced Organic Chemistry: Reactions, Mechanisms,
and Structure” (2nd Ed.), p. 578. McGraw–Hill, New York,
1977.
5
5
5
2. Li, Y., and Armor, J. N., J. Catal. 150, 376 (1994).
3. Li, Y., Slager, T. L., and Armor, J. N., J. Catal. 150, 388 (1994).
4. Lombardo, E. A., Sill, G. A., d’Itri, J. L., and Hall, W. K., J. Catal. 173,
4
40 (1998).
59. Gaudin, C., Duprez, D., Mabilon, G., and Prigent, M., J. Catal. 160, 10
(1996).
5
5. Yogo, K., Ono, T., Ogura, M., and Kikuchi, E., in “Reduction of Nitro-
gen Oxide Emission” (U. S. Ozkan, S. K. Agarwal, and G. Marcelin, 60. Kharas, K. C. C., Robota, H. J., and Liu, D. J., in “AIChE Symposium,”
Eds.), p. 123. ACS, Washington, DC, 1995. Paper No. 240b. Miami Beach, FL, Nov. 1992.