Selective Catalytic Oxidation (SCO) of ammonia to nitrogen over Fe-exchanged zeolites prepared by sublimation of FeCl3

Article abstract of DOI:10.1016/j.jcat.2004.05.023

The removal of ammonia from waste streams is becoming an important problem. Fe-exchanged zeolites were prepared by subliming FeCl₃ and studied as catalysts for selective catalytic oxidation (SCO) of ammonia to nitrogen in the presence of oxygen. Fe-ZSM-5 prepared by sublimation of FeCl₃ at 700°C exhibited excellent SCO performances at a very high space velocity. Over 99% ammonia conversion and nearly 100% N₂ selectivity were obtained at 400°C. The activity for the SCR of NO with ammonia was also studied on different Fe-zeolites. A higher subliming temperature led to a higher activity for the SCR reaction. This was very similar to the SCO reaction. Fe-ZSM-5 and Fe-MOR showed the highest activities.

Full text of DOI:10.1016/j.jcat.2004.05.023

Journal of Catalysis 226 (2004) 120–128
www.elsevier.com/locate/jcat
Selective catalytic oxidation (SCO) of ammonia to nitrogen over
Fe-exchanged zeolites prepared by sublimation of FeCl3
Gongshin Qi, Joseph E. Gatt, Ralph T. Yang ∗
Department of Chemical Engineering, University of Michigan, Ann Arbor, MI 48109-2136, USA
Received 3 September 2003; revised 18 May 2004; accepted 20 May 2004
Abstract
Fe-exchanged zeolites were prepared by subliming FeCl and studied as catalysts for selective catalytic oxidation (SCO) of ammonia to
3
◦
nitrogen in the presence of oxygen. Fe-ZSM-5 prepared by sublimation of FeCl at 700 C showed excellent SCO performances at a very
3
5
−1
◦
high space velocity (GHSV = 2.3 × 10 h ). Over 99% NH conversion and nearly 100% N selectivity were obtained at 400 C. Among
3
2
◦
Fe-ZSM-5 catalysts prepared by subliming at different temperatures (denoted in parentheses, in C), the catalytic performance decreased
in the sequence of Fe-ZSM-5 (700) > Fe-ZSM-5 (600) > Fe-ZSM-5 (500), Fe-ZSM-5 (400), Fe-ZSM-5 (350). Among different Fe-zeolite
◦
catalysts prepared by subliming at 700 C, the catalytic performance decreased in the sequence of Fe-ZSM-5 > Fe-MOR > Fe-FER >
Fe-Beta > Fe-Y which are similar to the Fe-zeolites prepared by aqueous ion-exchange method. The activity for the SCR (selective catalytic
reduction) of NO with ammonia was also investigated on different Fe-zeolites. The FT-IR results supported the two-step mechanism: NO is
an intermediate for N formation, NH was first oxidized to NO by O , NO then reacts to the unreacted NH to produce N .
2
3
2
3
2
2004 Elsevier Inc. All rights reserved.
Keywords: Selective catalytic oxidation (SCO) of NH ; Selective catalytic reduction (SCR) of NO; Fe-exchanged zeolites; Fe-ZSM-5; Sublimation of FeCl ;
3
3
FT-IR
1. Introduction
the SCR of NO with ammonia in a secondary bed to oxi-
dize the residual ammonia to N2, without introducing other
reactants into the gas mixture. The SCO process can also
be applied to biomass-derived fuels for removing the large
amounts of NH3 impurity [3,4]. Further applications of SCO
are the treatment of reformates for fuel cell systems and the
deodorization of ammonia-containing gases.
Several types of materials have been reported to be ac-
tive for SCO of ammonia to N2, such as Pt, Rh, and Pd
exchanged to ZSM-5 [2]; Ni, Fe, and Mn oxides supported
on γ -Al2O3 [3,4], V2O5/TiO2, CuO/TiO2, and Cu-ZSM-
5 [5]; CuO/Al2O3 [6,7]; Cu–Mn/TiO2 [8]; Fe-exchanged
TiO2-PILC [9]; Fe-exchanged ZSM-5 and other zeolites
[10,11]; Fe2O3–TiO2 [12]; and manganese oxide–silica
aerogels [13]. These catalysts exhibited activities for N2
formation under various conditions. Amblard et al. [3] re-
ported that among the transition-metal oxides supported on
γ -Al2O3, Ni/γ -Al2O3, Mn/γ -Al2O3, and Fe/γ -Al2O3 were
the most active and selective catalysts for the SCO reac-
tion. Wöllner et al. [8] reported that Cu/Mn mixed oxides
were good catalysts for ammonia oxidation to nitrogen and
The removal of ammonia from waste streams is becom-
ing an important problem. Many chemical processes use
reactants containing ammonia or produce ammonia as a
by-product. They are all plagued with the ammonia emis-
sion problem. Selective catalytic oxidation (SCO) of am-
monia to nitrogen is potentially an ideal technology for re-
moving ammonia from oxygen-containing waste gases and
consequently it has become of increasing interesting in re-
cent years [1–11]. Moreover, ammonia is used effectively
in power plants for NOx (x = 1, 2) abatement by selective
catalytic reduction (SCR). In order to control the ammonia
slip, most processes are carried out under conditions such
that NH3/NO < 1, at the expense of decreased NO reduc-
tion efficiencies. For improving the NO reduction efficiency,
the use of stoichiometric or an excess amount of ammo-
nia is desirable. The SCO of ammonia can be applied to
*
Corresponding author.
E-mail address: yang@umich.edu (R.T. Yang).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.05.023
2004 Elsevier Inc. All rights reserved.

G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
121
their activities were much higher than the individual com-
ponent oxides. Cavani and Trifiro [14] studied the SCO of
ammonia over V2O5/TiO2 and found that the vanadium ox-
ide supported on the rutile form of TiO2 was more active and
selective to N2 at low temperatures (i.e., < 300 ◦C). Among
these catalysts, manganese oxides showed interesting ac-
tivity and selectivity at low temperatures, and Fe-ZSM-5
showed high activities for high-temperature SCO of ammo-
nia (i.e., > 300 ◦C). In our previous studies, we found that
iron-exchanged zeolites had high activities for SCR of NO
with ammonia at high temperatures [10,11].
For transition-metal ion-exchanged Y zeolites, pulse re-
action results showed that the activity for ammonia oxidation
decreased in the sequence of Cu-Y > Cr-Y > Ag-Y > Co-
Y > Fe-Y > Ni-Y, Mn-Y, and the main product was nitro-
gen [1]. Gandhi and Shelef [15] found that CuMoO4 was
active for oxidation of NH3 to N2 at 450–550 ◦C, but the ac-
tivity and/or selectivity to N2 was inhibited by water vapor.
More recently, Li and Armor [2] reported that Pt, Rh, and Pd
exchanged to ZSM-5 or supported on Al2O3 showed good
SCO performance in a wet stream. The metal ion-exchanged
ZSM-5 were more active and also less affected by water than
the corresponding Al2O3-supported catalysts. Also, the no-
ble metal catalysts were more active than V2O5/TiO2 and
Co-ZSM-5 [2].
Recently, we studied the SCO reaction on a series of
transition-metal (Cr, Mn, Fe, Co, Ni, Cu, or Pd) ion-
exchanged ZSM-5 [10]. The results showed that the cat-
alytic performance (i.e., NH3 conversion and N2 selectiv-
ity) increased in the trend of Co-ZSM-5 ≈ Ni-ZSM-5 <
Mn-ZSM-5 < H-ZSM-5 < Pd-ZSM-5 < Cr-ZSM-5 < Cu-
ZSM-5 < Fe-ZSM-5 at a high gas hourly space velocity
(GHSV = 2.3 × 105 h−1). Nearly 100% of NH3 conversion
to N2 was obtained at 450 ◦C on the Fe-ZSM-5.
It has been found that Fe-exchanged zeolites have high
activities and selectivities to N2 for the SCO of ammonia at
high temperatures [11]. It was observed that both NH3 con-
version and N2 selectivity increased with iron content (with
iron-exchange level < 100%) at low temperatures for Fe-
ZSM-5 [11]. In order to further increase the activity of SCO,
we need to increase the Fe content. Chen and Sachtler [16]
have developed a method for the preparation of Fe-ZSM-
5 catalysts by subliming FeCl3 into H-ZSM-5. Anhydrous
FeCl3 is sublimated into H-form zeolites at high temper-
atures (e.g., 320 ◦C), where it reacts with zeolite protons
by H+ + FeCl3 = [FeCl2]+ + HCl until all protons are re-
placed by [FeCl2]+. After the reaction, chlorine is removed
by washing the samples with deionized water. This method
is very effective in obtaining high Fe-exchanged Fe-ZSM-5
catalysts in a reproducible manner. More recently, Krishna
et al. [17] studied the Fe-ZSM-5 catalysts prepared by sub-
liming FeCl3 into H-ZSM-5 at different sublimation temper-
atures for the SCR of NO with ammonia.
the effect of the framework types of zeolites which have dif-
ferent pore sizes and different Si/Al ratios, and the effects of
different subliming temperatures on the SCO and NH3-SCR.
2. Experimental
2.1. Preparation of catalysts
The starting materials used for preparation of the catalysts
are as follows. NH4-ZSM-5 (Si/Al ≈ 10) was obtained from
Alsi-Penta Zeolithe Gmbh (Germany). Mordenites (MOR,
Si/Al = 6.4, 10, and 45), NH4-beta (Si/Al = 12.5), and
NH4-ferrierite (FER, Si/Al = 10) were obtained from Ze-
olyst International Company. Y zeolite (Si/Al = 2.4) was
obtained from Strem Chemicals. The non-NH4-form zeo-
lites were first converted to NH4-zeolites by exchanging with
0.5 M NH4Cl solution (four times) at room temperature.
FeCl3 was obtained from Aldrich.
Fe-exchanged catalysts were prepared by chemical vapor
ion exchange. This method was similar to that proposed by
Chen and Sachtler [16]. FeCl3 (97%, Aldrich) was used as
the iron source. During the experiment, 2 g zeolite and 0.5 g
FeCl3, separated by glass wool, were loaded into a quartz re-
actor. Subsequently, the reactor was heated to 350 or 700 ◦C
in flowing He (100 ml/min). FeCl3 was evaporated and then
exchanged with H-zeolite according to the following reac-
tion:
H-zeolite + FeCl3 = [FeCl2]-zeolite + HCl.
(1)
After 1 h, the sample was removed and then washed
with deionized water to eliminate chlorine (detected with
silver nitrate solution). The catalysts were dried at 120 ◦C
overnight, then calcined at 500 ◦C for 6 h in air. Finally, the
obtained samples were ground and sieved to 60–100 mesh.
The Fe and Al contents in the samples were measured by
neutron activation analysis. The iron-exchangelevel was cal-
culated by 3 (number of iron ions)/(number of aluminum
ions) because almost all iron was present as Fe3+. The cata-
lysts are denoted by Fe-zeolite (x), where x is the subliming
temperature in ◦C.
2.2. Catalytic performance measurement
The SCO activity measurement was carried out in a fixed-
bed quartz reactor. The reaction temperature was controlled
by an Omega (CN-2010) programmable temperature con-
troller. One hundred milligrams of catalyst was used in this
work. The reactant gas was obtained by blending different
gas flows. The typical reactant gas composition was as fol-
lows: 1000 ppm NH3, 2% O2, and balance He. The total
flow rate was 500 ml/min (ambient conditions). The pre-
mixed gases (1.0% NH3 in He) were supplied by Matheson.
A magnetic deflection-type mass spectrometer (AERO VAC,
Vacuum Technology Inc.) was used to monitor continu-
ously the effluent gas from the reactor, which contained NH3
In this work, we studied the SCO of ammonia on the Fe-
exchanged zeolites prepared by subliming FeCl3 similar to
Sachtler’s method [16]. For these catalysts, we investigated

122
G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
(m/e = 17 minus the contribution of H2O), H2O (m/e =
18), N2 (m/e = 28), NO (m/e = 30), O2 (m/e = 32) and
N2O (m/e = 44). NO2 (m/e = 46) was not detectable with
this mass spectrometer. The concentrations of the unreacted
NH3 and the NOx formed were also continually monitored
with a chemiluminescent NO/NOx analyzer (Thermo Envi-
ronmental Instruments Inc. Model 42C), in which a high-
temperature converter converted NH3 to NOx by the reaction
NH3 + O2 → NOx + H2O. The NH3 conversion was calcu-
lated by ([NO] + 2[N2] + 2[N2O])/[NH3]0 × 100%, where
[NH3]0 is the initial NH3 concentration. The selectivity is
defined as the percentage conversion of ammonia to N2,
N2O, and NO. The data were collected when the SCO re-
action reached the steady state typically after 20–180 min at
each temperature.
The SCR activity measurement was carried out in a fixed-
bed quartz reactor. The typical reactant gas composition was
as follows: 1000 ppm NO, 1000 ppm NH3, 2% O2, and bal-
ance He. Typically 50 mg sample was used in each run. The
total flow rate was 500 ml/min (under ambient conditions).
Thus, a very high GHSV (gas hourly space velocity) was ob-
tained (4.6 × 105 1/h). The premixed gases (1.01% NO in
He, 1.00% NH3 in He, and 0.99% SO2 in He) were supplied
by Matheson. Water vapor was generated by passing He
through a heated gas-wash bottle containing deionized water.
The tubing of the reactor system was heat-traced to prevent
formation and deposition of ammonium sulfate/bisulfate and
ammonium nitrate. The NO and NO2 concentrations were
continually monitored by the chemiluminescent NO/NOx
analyzer. To avoid errors caused by the oxidation of ammo-
nia in the converter of the NO/NOx analyzer, an ammonia
trap containing phosphoric acid solution was installed be-
fore the sample inlet to the chemiluminescent analyzer. The
products were also analyzed by a gas chromatograph (Shi-
madzu, 8A) at 50 ◦C with a 5A molecular sieve column for
N2 and a Porapak Q column for N2O.
ture in O2/He flow. The reduction of the sample was carried
out from room temperature to 700 ◦C in a flow of 5.32%
H2/N2 (40 ml/min) at 10 ◦C/min. The consumption of H2
was monitored continuously with a thermal conductivity de-
tector. The water produced during reduction was trapped in
a 5 A molecular sieve column.
2.4. FTIR studies
Infrared spectra were recorded on a Nicolet Impact 400
FTIR spectrometer with a TGS detector. A self-supported
wafer of 1.3 cm diameter was prepared by pressing a 15-mg
sample and then loaded into an IR cell with BaF2 windows.
The wafers could be pretreated in situ in the IR cell. The
wafers were first treated at 400 ◦C in a flow of high purity
20% O2/He for 0.5 h and then cooled to room temperature.
At each temperature, the background spectrum was recorded
in flowing He and was subtracted from the sample spectrum
that was obtained at the same temperature. Thus the IR ab-
sorption features that originated from the structure vibrations
of the catalyst were eliminated from the sample spectra. In
the experiment, the IR spectra were recorded by accumulat-
ing 100 scans at a spectra resolution of 4 cm−1
.
3. Results
3.1. Characterization of catalysts
Fig. 1 shows the XRD patterns of different Fe-ZSM-5 cat-
alysts prepared at different temperatures and pure HZSM-5.
The XRD patterns of Fe-ZSM-5 catalysts prepared at differ-
ent temperatures are similar to the patterns of HZSM-5, and
no peak for Fe2O3 species was observed. This indicated that
Fe2O3 was dispersed well on the framework of ZSM-5.
2.3. Catalyst characterization
A Micromeritics ASAP 2010 micropore-size analyzer
was used to measure the N2 adsorption isotherms of the
samples at liquid N2 temperature (−196 ◦C). The specific
surface area was determined from the linear portion of the
BET plot. The pore-size distribution was calculated from the
desorption branch of the N2 adsorption isotherm using the
Barrett–Joyner–Halenda (BJH) formula. Prior to the surface
area and pore-size distribution measurements, the samples
were degassed in vacuo at 400 ◦C for 24 h.
The powder X-ray diffraction (XRD) measurements were
carried out with a Rigaku Rotaflex D/Max-C system with
Cu-Kα (λ = 0.1543 nm) radiation. The samples were loaded
on a sample holder with a depth of 1 mm.
In each H2-TPR (temperature-programmed reduction)
experiment, a 50-mg sample was loaded into a quartz reac-
tor and then pretreated in O2/He (100 ml/min) at 500 ◦C for
0.5 h. The sample was then cooled down to room tempera-
Fig. 1. XRD profiles of Fe-ZSM-5 catalysts prepared by different subliming
temperatures (a) H-ZSM-5, b) Fe-ZSM-5 (350), (c) Fe-ZSM-5 (400), (d)
Fe-ZSM-5 (500), (e) Fe-ZSM-5 (600), (f) Fe-ZSM-5 (700).

G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
123
Table 1
Characterization of the catalysts
2
3
Sample
BET surface area (m /g)
Pore volume (cm /g)
Fe-ZSM-5 (350)
Fe-ZSM-5 (400)
Fe-ZSM-5 (500)
Fe-ZSM-5 (600)
Fe-ZSM-5 (700)
Fe-MOR (700)
Fe-FER (700)
Fe-Beta (700)
Fe-Y (700)
313.02
313.35
300.77
292.13
257.22
366.49
299.90
487.63
145.15
0.073
0.073
0.072
0.081
0.078
0.13
0.084
0.54
0.048
The BET surface areas and pore volumes of the Fe-zeolite
catalysts are summarized in Table 1. After iron exchange
and calcinations at different temperatures for 6 h, the sur-
face areas of Fe-ZSM-5 decreased from 313 to 257 cm2/g.
As the calcination temperature increased, the surface area
decreased. Table 2 shows the Fe contents for Fe-ZSM-5
samples obtained by sublimation of FeCl3 at different tem-
peratures.
Fig. 2 shows the H2-TPR profiles of Fe-ZSM-5 obtained
by subliming at 350 and 700 ◦C. Before the reduction, the
samples were pretreated in O2/He at 400 for 30 min. A peak
at 400 ◦C and a shoulder at 520 ◦C were observed on the
TPR profiles for both catalysts, which can be attributed to the
reduction of iron species at two different sites. Integrations
of area of the TPR peaks yielded an H2/Fe molar ratio of
close to 0.5. This indicates that almost all of the iron was lo-
cated at exchange sites where it could only be reduced from
+3 to +2 under this reduction condition. The result above is
consistent with the literature results [16,17].
Fig. 2. H -TPR profiles of Fe-ZSM-5 catalysts.
2
was detected at any of the temperatures studied, and NO was
detected at lower temperatures for all the catalysts. Among
them, Fe-ZSM-5 sublimed at 700 ◦C showed the highest ac-
tivities for NH3 oxidation to N2 at 350–450 ◦C. N2 selec-
tivity increased with increasing temperature. The conversion
of ammonia for the different Fe-ZSM-5 catalysts decreased
in the sequence of Fe-ZSM-5 (700) > Fe-ZSM-5 (600) >
Fe-ZSM-5 (500), Fe-ZSM-5 (400), Fe-ZSM-5 (350).
3.3. SCO performance on different Fe-zeolites
The catalytic performance of different Fe-zeolites pre-
pared by subliming at 700 ◦C for the SCO of NH3 is summa-
rized in Table 4. Under the conditions of 1000 ppm NH3, 2%
O2, and GHSV = 2.3 ×105 h−1, these catalysts showed var-
ious ammonia conversions at different temperatures. Among
them, Fe-ZSM-5 and Fe-MOR showed the highest activities.
Small amounts of NO were observed. The lowest SCO ac-
tivities were obtained on Fe-Y and Fe-Beta. With increasing
temperature, more NO was produced on Fe-FER and Fe-
MOR, while on Fe-Y and Fe-Beta the NO product decreased
after the temperature was higher than 450 ◦C. These results
suggest that the structure of zeolite has a strong effect on the
SCO of NH3 to N2 for the Fe-exchanged zeolites. The maxi-
3.2. SCO performance on different Fe-ZSM-5 prepared at
different temperatures
The catalytic performance of Fe-ZSM-5 prepared by sub-
liming at different temperatures for the SCO of NH3 is sum-
marized in Table 3. Under the conditions of 1000 ppm NH3,
2% O2, and GHSV = 2.3 × 105 h−1, these catalysts showed
various ammonia conversions at different temperatures. NH3
conversion increased with reaction temperature and all of the
catalysts showed very high NH3 conversions. In all cases,
N2 was the main product for ammonia oxidation, no N2O
Table 2
Catalyst preparation conditions and Fe ion-exchange percentage
a
b
Catalysts
Activation in
He ( C/h)
FeCl sublimation
3
( C/h)
Fe content
(wt%)
Ion exchange
(%)
TOF
◦
◦
3
(×10 /s)
Fe-ZSM-5 (350)
Fe-ZSM-5 (400)
Fe-ZSM-5 (500)
Fe-ZSM-5 (600)
Fe-ZSM-5 (700)
600/1
600/1
600/1
600/1
600/1
350/2
400/2
500/2
600/2
700/2
4.5
5.0
4.7
5.0
3.6
240
267
252
267
192
1.25
1.26
1.26
2.54
3.64
a
Determined from neutron activation analysis.
b
◦
Overall TOF (turnover frequency) is defined as the number of NH molecules converted per Fe per second at 350 C, 1000 ppm NH , 2% O at
3
3
2
5
−1
.
2.3 × 10
h

124
G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
Table 3
Table 4
Catalytic performance of SCO over Fe-ZSM-5 prepared by subliming at
different temperatures
Catalytic performances of SCO over various Fe-exchanged zeolites
a
Catalyst
Temp. NH conv. Selectivity (%)
3
N yield
2
(%)
◦
Catalyst
Temperature NH conversion
3
Selectivity (%)
N O NO
( C)
(%)
N
N O
2
NO
2
◦
( C)
(%)
N
2
2
Fe-FER
(ferrierite)
350
375
400
425
450
18
36
62
84
99
95
92
90
88
85
0
0
0
0
0
5
8
10
12
15
17
33
56
74
84
Fe-ZSM-5 (350)
300
350
375
400
425
450
6
27
46
95
95
99
91
96
97
98
100
100
0
0
0
0
0
0
9
4
3
2
0
0
Fe-ZSM-5
350
375
400
425
450
63
90
100
100
100
96
99
100
100
100
0
0
0
0
0
4
1
0
0
0
60
89
Fe-ZSM-5 (400)
Fe-ZSM-5 (500)
Fe-ZSM-5 (600)
Fe-ZSM-5 (700)
300
350
375
400
425
450
6
29
47
94
96
99
91
96
97
98
100
100
0
0
0
0
0
0
9
4
3
2
0
0
100
100
100
Fe-MOR
(mordenite)
350
375
400
425
450
26
48
74
92
99
88
94
97
99
100
0
0
0
0
0
12
6
3
1
0
23
45
72
91
99
300
350
375
400
425
450
6.4
92
96
98
99
100
100
0
0
0
0
0
0
8
4
2
1
0
0
30
52
94
95
99
Fe-Y
350
375
400
425
450
12
28
50
78
90
89
84
76
69
79
0
0
1
6
0
11
16
23
25
21
10.6
23.5
38
53.8
71.1
300
350
375
400
425
450
10
61
86
99
99
96
99
99
100
100
100
0
0
0
0
0
0
4
1
1
0
0
0
Fe-Beta
350
375
400
425
450
13
26
44
70
89
95
93
88
86
92
0
0
0
0
0
5
7
12
14
8
12.3
24.2
38.7
60.2
81.9
100
300
350
375
400
425
450
15
63
90
100
100
100
96
99
100
100
100
100
0
0
0
0
0
0
4
1
0
0
0
0
Reaction conditions: 0.1 g catalyst, [NH ] = 1000 ppm, [O ] = 2%, He =
3
2
5
−1
.
balance, total flow rate = 500 ml/min and GHSV = 2.3 × 10
h
a
◦
Catalysts prepared by subliming at 700 C.
Reaction conditions: 0.1 g catalyst, [NH ] = 1000 ppm, [O ] = 2%, He =
3
2
5
−1
.
balance, total flow rate = 500 ml/min and GHSV = 2.3 × 10
h
mum N2 yield by the Fe-exchanged zeolites decreased in the
sequence of Fe-ZSM-5 > Fe-MOR > Fe-FER > Fe-Beta >
Fe-Y. The selectivity for N2O was too low to be detected for
all the catalysts.
3.4. Catalytic performance of SCR of NO with ammonia
over Fe-zeolite catalysts
Selective catalytic reduction of NO with ammonia was
also studied on various Fe-based catalysts to demonstrate
the relationship between the SCR reaction and the ammo-
nia oxidation reaction. Fig. 3 shows the NO conversion on
the various Fe-ZSM-5 catalysts prepared by subliming at
different temperatures. All the catalysts showed high activi-
ties and high selectivities to N2. The subliming temperature
affected the activity significantly, particularly at higher tem-
peratures. A higher subliming temperature led to a higher
activity for the SCR reaction. This is very similar to the
SCO reaction. The order of the different activities of the cat-
alysts is Fe-ZSM-5 (700) > Fe-ZSM-5 (600) > Fe-ZSM-5
(500) > Fe-ZSM-5 (400) > Fe-ZSM-5 (350). Fig. 4 shows
Fig. 3. NO conversions over Fe-ZSM-5 catalysts at different temperatures.
Reaction conditions: catalyst 0.05 g, [NO] = [NH ] = 1000 ppm, [O ] =
3
2
h
5
−1
.
2%, He = balance, total flow rate = 500 ml/min, GHSV = 4.6 × 10
the NO conversion on the various Fe-zeolite catalysts pre-
pared by subliming at 700 ◦C. Among them, Fe-ZSM-5 and
Fe-MOR showed the highest activities. The lowest SCR ac-
tivities were obtained on Fe-Y and Fe-Beta. These results
suggest that the structure of zeolites also has a strong effect
on the SCR of NO with NH3 to N2 for the Fe-exchanged

G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
125
Fig. 5. Effect of H O and SO on NH conversion for Fe-ZSM-5 (700).
2
2
3
Fig. 4. NO conversions over Fe-zeolite catalysts at different temperatures.
Reaction conditions: catalyst 0.1 g, [NH ] = 1000 ppm, [O ] = 2%,
3
2
Reaction conditions: catalyst 0.05 g, [NO] = [NH ] = 1000 ppm, [O ] =
3
2
h
He = balance, 500 ppm SO (when used), 2.5% water vapor (when used),
5
−1
.
2
2%, He = balance, total flow rate = 500 ml/min, GHSV = 4.6 × 10
5
−1
.
total flow rate = 500 ml/min, GHSV = 2.3 × 10
h
zeolites. More importantly, a link between the NH3 SCO re-
action and the NO SCR reaction became obvious.
3.5. Effect of H2O and SO2 on NH3 conversion for
Fe-ZSM-5 (700)
Since the waste streams usually contain water vapor and
small amounts of SO2, we further studied the effects of H2O
and SO2 on the catalytic performance of Fe-ZSM-5 (700)
for SCO of ammonia. For Fe-ZSM-5 (700), when 500 ppm
SO2 and/or 2.5% H2O was added to the reactants, NH3 con-
version was decreased only slightly by H2O (Fig. 5). SO2
and H2O + SO2 decreased NH3 conversion at 350–450◦C,
but the decrease was less significant at high temperatures.
N2 was again the dominant product for NH3 oxidation in the
presence of H2O and/or SO2.
◦
3.6. FT-IR study
Fig. 6. IR spectra of Fe-ZSM-5 (700) and Fe-ZSM-5 (350) at 30 C with
empty cell as background. (a) Fe-ZSM-5 (700) and (b) Fe-ZSM-5 (350).
The IR spectra of Fe-ZSM-5 (700) and Fe-ZSM-5 (350)
at room temperature are shown in Fig. 6. The IR spectra of
both samples are almost the same, several bands at 1630,
1870, 1990, and 3740 cm−1 are observed. The weak band
at 3740 cm−1 can be assigned to hydroxyl stretching vi-
bration of the Si-OH group at crystal termination [18]. The
bands at 1990 and 1870 cm−1 are due to the zeolite overtone
bands [19]. The band at 1630 cm−1 and the broad absorption
at 3000–3400 cm−1 are due the adsorbed water. The band at
3610 cm−1 due to OH stretching of the Brønsted acid [19,
20] which is very small and is in accordance with the results
obtained by Chen and Sachtler [16].
The IR spectra of ammonia adsorbed on Fe-ZSM-5
(700) at different temperatures are shown in Fig. 7. Af-
ter the sample was treated in flowing 1000 ppm NH3/He
for 30 min and then purged with He for 30 min at 30 ◦C,
a strong band around 1470 cm−1 and several weak bands
at 1705, 1630, and 1280 cm−1 were observed. The bands at
1705 and 1470 cm−1 are due to the symmetric a+nd asym-
metric bending vibrations, respectively, of NH4 that is
chemisorbed on the Brønsted acid sites, while the bands at
1630 and 1280 cm−1 may be assigned to symmetric and
asymmetric bending vibration of the N–H bonds in NH3
coordinately linked to the Lewis acid sites [21,22]. The
above results indicated that there are many more Brøn-
sted acid sites than Lewis acid sites on the Fe-ZSM-5.
With increasing temperatures, the intensities of the 1630
and 1270 cm−1 bands increased while the intensities of the
1470 and 1705 cm−1 bands decreased. This result indicate+d
that some NH3 species have desorbed and some of NH4
species were transformed to coordinately adsorbed NH3 on
Fe-ZSM-5.

126
G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
Fig. 7. IR spectra of Fe-ZSM-5 (700) treated in flowing 1000 ppm NH /He
3
Fig. 9. IR spectra of Fe-ZSM-5 (700) treated in flow of 1000 ppm NH +2%
3
◦
at 30 C for 30 min and then purged by He at (a) 30, (b) 100, (c) 200, (d)
◦
O /He at (a) 30, (b) 100, (c) 200, (d) 300, and (e) 400 C.
2
◦
300, and (e) 400 C.
with O2/He at 400 ◦C for 30 min followed by cooling to
30 ◦C. Then 1000 ppm NH3 and 2% O2/He were introduced
into the IR cell and IR spectra were recorded as a function
of temperature. As the temperature increased, the intensity
+
of NH4 species decreased and the peak at 1870 cm−1 was
observed. The band at 1870 cm−1 can be attributed to the
adsorbed NO species on Fe2+ sites [20,23,31].
4. Discussion
The results discussed above show that Fe-exchanged zeo-
lites prepared by subliming FeCl3 are highly active for the
SCO of ammonia to nitrogen in the presence of oxygen.
Table 3 shows the ammonia conversions on Fe-ZSM-5 cat-
alysts by subliming at different temperatures in the absence
of water and SO2. All the catalysts showed high activities.
The best catalyst was prepared by subliming at 700 ◦C. More
than 99% NH3 conversion and nearly 100% N2 selectivity
were obtained on Fe-ZSM-5 (700) at 400 ◦C under the con-
dition of GHSV = 2.3 × 105 h−1. A comparison has been
made for the Fe-ZSM-5 catalysts prepared by subliming at
different temperatures, given in Tables 2 and 3. It is clearly
seen from Tables 2 and 3 that the subliming temperature
has a significant effect on the catalytic activities on SCO of
ammonia. Fe-ZSM-5 (700) has the highest TOF, indicating
that this catalyst contains the most active iron species. The
process during the catalyst preparation can be described as
follows [17]. It is known that FeCl3 at 320 ◦C will be present
as a dimmer (Fe2Cl6) [28]. The ion exchange could proceed
according to the following reactions:
Fig. 8. IR spectra of Fe-ZSM-5 (700) treated in flowing 1000 ppm
◦
NO/He + 2% O at 30 C for 30 min and then purged by He at (a) 30,
2
◦
(b) 100, (c) 200, (d) 300, and (e) 400 C.
Fig. 8 shows the IR spectra of Fe-ZSM-5 (700) treated in
flowing NO + O2 and then purged by He at different tem-
peratures. After the Fe-ZSM-5 (700) was heated to 100 ◦C
in He, the bands at 1686 and 1570 cm−1 decreased sharply,
while the band at 1624 cm−1 increased significantly. The
band at 1686 cm−1 was probably due to adsorbed N2O3
species since it is close to the IR band at 1690 cm−1 for
gaseous N2O3 [24]. The bands at 1624 and 1570 cm−1 can
be assigned to adsorbed NO2 and nitrate species, respec-
tively [23,25–27]. This result indicated that N2O3 and ni-
trate adspecies desorbed or transformed to NO2 adspecies.
As the temperature was further increased above 300 ◦C, the
1624 cm−1 band also decreased.
Fe2Cl6 + H-ZSM-5 → Fe2Cl5-ZSM-5 + HCl,
Fe2Cl5-ZSM-5 → Fe-ZSM-5 + FeOx + 5HCl,
(2)
(3)
The IR spectra of the reaction between ammonia and oxy-
gen are shown in Fig. 9. Fe-ZSM-5 (700) was first treated
where the framework and surface hydroxyls also participate
in the reaction.

G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
127
At 700 ◦C, FeCl3 will be present as a monomer (FeCl3),
and the low iron content of the Fe-ZSM-5 catalyst (see Ta-
ble 2) could be due to the rapid ion-exchange process at
700 ◦C:
variable valence of iron cations in the Fe-exchanged zeo-
lites might be beneficial to oxygen adsorption and activation
under the SCO reaction conditions. The oxygen adspecies,
δ−
e.g., O2−, O2 (1 < δ < 2), and O22−, have been observed
on O2-adsorbed Fe2O3 by IR spectroscopy [33]. The high
cation capacities would lead to more iron ions in the Fe-
exchanged zeolites, which provide more active sites for oxy-
gen adsorption and activation. Consequently, high activities
for ammonia oxidation to N2 are expected on the catalysts
with high iron contents. The vapor-phase exchange can pro-
vide the higher exchange level of iron which results in the
high activity of SCO.
FeCl3 + H-ZSM-5 → FeCl2-ZSM-5 + HCl.
(4)
Kaucky et al. [29] and Sobalik et al. [30] have proposed
the presence of α, β, and γ sites in ZSM-5 which are located
at straight channels, at the intersection of the straight and
sinusoidal channels and in a boat-shaped site in sinusoidal
channels, respectively. The dimer (Fe2Cl6) preferentially oc-
cupies the α position along with β and γ positions [17].
Fe-ZSM-5 catalyst prepared by high temperature resulted in
preferential occupation of γ positions. Although the α and β
sites presented in straight channels and intersections are eas-
ily available, these positions were not extensively occupied
when the materials were prepared by sublimation or treated
at 700 ◦C because the FeCl2 is bound weakly at α and β
positions and migrates to the more stable γ positions or mi-
grates out to the crystal surface [17]. Sobalik et al. [30] have
proposed that, in Co-ZSM-5, the α-type Co ions exhibit the
weakest bonding followed by β sites of medium strength of
bonding to framework oxygens. The γ sites showed highest
strength Co ions bound to the framework oxygens, similar to
the above observation.
The catalytic performance, especially N2 selectivity, was
found to decrease in a trend of Fe-ZSM-5 > Fe-MOR >
Fe-FER > Fe-Beta > Fe-Y (Table 4). More than 99% NH3
conversion and nearly 100% N2 selectivity were obtained
on Fe-ZSM-5 at 400 ◦C under the condition of GHSV =
2.3 × 105 h−1. The difference in the SCO performance may
be related to the various pore structure and size and surface
acidity for the Fe-exchanged zeolites. ZSM-5 has a unique
pore structure that consists of two intersecting channel sys-
tems: one straight and the other sinusoidal and perpendicular
to the former. Both channel systems have 10-member ring
elliptical openings (0.52–0.57 nm in diameter). The mor-
denite pore structure consists of elliptical and nonintercon-
nected channels parallel to the c-axis of the orthorhombic
structure. Their openings are limited by 12-member rings
(0.6–0.7 nm). It appears that zeolites with narrow, channel
pore structures favor ammonia oxidation to nitrogen by oxy-
gen [11].
It is also noted that Fe-exchanged zeolites also showed
excellent activities for the selective catalytic reduction of
NO to N2 with ammonia as reductant (Figs. 3 and 4). But
the SCR activity was higher than the SCO activity for the
same catalyst, which was attributed to a higher reactivity
of NH3 with NOx than with O2 [34]. Our results indicated
that almost all of NO was reduced to N2 by ammonia on
Fe-ZSM-5 at 350–450 ◦C under the condition of GHSV =
4.6×105 h−1 (Figs. 3 and 4). This catalysts were much more
active than Fe-Y, Fe-Beta, e-FER which are in line with their
relative SCO performance (e.g., N2 selectivity) obtained in
this work and others [11]. By comparing the SCR activity
with SCO performance for the various Fe-exchanged zeo-
lites, it can be seen that there is a good correlation between
the SCR activity and the N2 selectivity for the SCO reac-
tion; i.e., the higher SCR activity, the higher the N2 selec-
tivity. This conclusion indicates that when NO (the main by-
product) is generated during the SCO reaction, it can be fur-
ther reduced to N2 by unreacted ammonia through the SCR
reaction. Similar results were also obtained on V2O5/TiO2,
which showed both high SCR activity and N2 selectivity of
SCO [1,2,5].
The FT-IR spectra indicate that NH3 were adsorbed on
Fe-ZSM-5 catalysts after the sample was treated with NH3.
NH3 molecules were adsorbed on the Brønsted acid and
Lewis acid sites of catalyst to generate NH4+ ions and coo+r-
dinated NH3, respectively (Fig. 7). In addition to the NH4
ions and coordinated NH3, NH2 species were also detected
in the IR spectra, which indicated that hydrogen abstraction
took place for some ammonia species. The IR spectra of
NO + O2 adsorbed on Fe-ZSM-5 were also studied. NO2,
nitrate, and N2O3 species could be observed. With increas-
ing temperature, the bands due to nitrogen oxides adspecies
decreased (Fig. 8). The NO2 adspecies was dominant at 100–
300 ◦C. The IR spectra in the flow of NH3 and O2 were also
studied (Fig. 9). With increasing temperature, the intensity
of adsorbed ammonia species decreased while the band at
1870 cm−1 increased. From Fig. 9, the band at 1510 cm−1
cannot be detected, which may have resulted from the high
reaction rate between NH2 and NO species.
It is known that ammonia molecules can adsorb on the
Brønsted acid and Lewis acid sites of zeolites to generate,
+
respectively, NH4 ions and coordinated NH3. Our previ-
ous IR spectra showed that the NH4 ions (at 1470 cm−1
)
+
and the coordinated NH3 (at 1280 cm−1) can be detected on
Fe-ZSM-5 catalyst. Our previous XPS and ESR data [32] on
Fe-ZSM-5 indicated that iron cations were present mainly as
Fe3+ ions with tetrahedral coordination, along with a small
amount of Fe2+ and aggregated Fe3+ ions. Also, the Fe3+
ions could be partially reduced to Fe2+ ions by H2 at 300–
600 ◦C (Fig. 2), but the oxidation was reversible when O2
was introduced into the reduced catalyst at 500 ◦C [32]. The
For the SCO reaction, two pathways for oxidation of NH3
to N2 have been proposed in the literature. One is a direct
route by the recombination of 2 NH2 species to NH2–NH2
and then oxidation of NH2–NH2 to N2 [35]. The other is a

128
G. Qi et al. / Journal of Catalysis 226 (2004) 120–128
two-step route that involves oxidation of NH3 to NOx and
then reduction of NOx to N2 by NH3 [1,4]. According to
our catalytic performance and FT-IR results, it seems that
the SCO reaction on the Fe-exchanged zeolites takes place
by the two-step route; i.e., NO is an intermediate for N2
formation. NH3 was first oxidized to NO by O2. This reac-
tion occurs either on the catalyst surface or in the gaseous
phase, or both. Our empty-tube results showed that NH3
conversions were 23–55% at 350–450◦C under the condi-
tion of GHSV = 2.3 ×105 h−1, with NO as the predominant
product. Subsequently, the NO reacts with unreacted NH3 to
produce N2 through the SCR reaction. Therefore, good SCR
catalysts are expected to have high N2 selectivities for the
SCO reaction.
References
[1] N.I. P’chenko, Russian Chem. Rev. 45 (1976) 1119.
[2] Y. Li, J.N. Armor, Appl. Catal. B 13 (1997) 131.
[3] M. Amblard, R. Burch, B.W.L. Southward, Appl. Catal. B 22 (1999)
L159.
[4] M. Amblard, R. Burch, B.W.L. Southward, Catal. Today 59 (2000)
365.
[5] N.N. Sazonova, A.V. Simakov, T.A. Nikoro, G.B. Barannik, V.F.
Lyakhova, V.I. Zheivot, Z.R. Ismagilov, H. Veringa, React. Kinet.
Catal. Lett. 57 (1996) 71.
[6] L. Gang, J. van Grondelle, B.G. Anderson, R.A. van Santen, J. Ca-
tal. 186 (1999) 100.
[7] T. Curtin, F. O’Regan, C. Deconinck, N. Knuttle, B.K. Hodnett, Catal.
Today 55 (2000) 189.
[8] A. Wöllner, F. Lange, H. Schmelz, H. Knözinger, Appl. Catal. A 94
(1993) 181.
[9] R.Q. Long, M.T. Chang, R.T. Yang, Appl. Catal. B 33 (2001) 97.
[10] R.Q. Long, R.T. Yang, Chem. Commun. 5 (2000) 1651.
[11] R.Q. Long, R.T. Yang, J. Catal. 201 (2001) 145.
[12] R.Q. Long, R.T. Yang, J. Catal. 207 (2002) 158.
[13] P. Fabrizioli, T. Burgi, A. Baiker, J. Catal. 207 (2002) 88.
[14] F. Cavani, F. Trifiro, Catal. Today 4 (1989) 253.
[15] H.S. Gandhi, M. Shelef, J. Catal. 40 (1975) 312.
[16] H.Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
[17] K. Krishna, G.B.F. Seijger, C.M. van den Bleek, M. Makkee, Guido
Mul, H.P.A. Calis, Catal. Lett. 86 (2003) 121.
[18] P.A. Jacobs, R. Von Ballmoos, J. Phys. Chem. 86 (1982) 3050.
[19] R. Joyner, M. Stockenhuber, J. Phys. Chem. B 103 (1999) 5963.
[20] L.J. Lobree, I.-C. Hwang, J.A. Reimer, A.T. Bell, J. Catal. 186 (1999)
242.
[21] N.Y. Topsøe, J. Catal. 128 (1991) 499.
[22] W.S. Kijlstra, D.S. Brands, E.K. Poels, A. Bliek, J. Catal. 171 (1997)
208.
5. Conclusions
Based on the above results, it can be concluded that
Fe-exchanged zeolites prepared by vapor-phase exchange
method are highly active for the SCO of ammonia to nitro-
gen. Fe-ZSM-5 and Fe-mordenite (MOR) showed excellent
SCO performances at a very high space velocity (GHSV =
2.3 ×105 h−1). Over 99% NH3 conversion and nearly 100%
N2 selectivity were obtained. Among different subliming
temperature, the catalytic performance decreased in the se-
quence of Fe-ZSM-5 (700) > Fe-ZSM-5 (600) > Fe-ZSM-5
(500), Fe-ZSM-5 (400), Fe-ZSM-5 (350). Among different
Fe-zeolites catalysts by subliming at 700 ◦C, the catalytic
performance decreased in the sequence of Fe-ZSM-5 > Fe-
MOR > Fe-FER > Fe-Beta > Fe-Y. For the Fe-exchanged
zeolites, there existed a good correlation between the N2 se-
lectivity for the SCO reaction and the activity for the SCR
(selective catalytic reduction) of NO with ammonia; i.e., the
higher the SCR activity, the higher the SCO N2 selectivity.
The FT-IR results supported the two-step mechanism: NO is
an intermediate for N2 formation and NH3 was first oxidized
to NO by O2.
[23] H.Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 180 (1998)
171.
[24] J. Laane, J.R. Ohlsen, Prog. Inorg. Chem. 27 (1980) 465.
[25] J. Eng, C.H. Bartholomew, J. Catal. 171 (1997) 27.
[26] Y. Li, J.N. Armor, J. Catal. 150 (1994) 388.
[27] J. Valyon, W.K. Hall, J. Phys. Chem. 97 (1993) 1204.
[28] Q. Zhu, B.L. Mojet, R.A.J. Janssen, E.J.M. Hensen, J. van Grondelle,
P.C.M.M. Magusin, R.A. van Santen, Catal. Lett. 81 (2002) 205.
[29] D. Kaucky, A. Vondroval, J. Dedecek, B. Wichterlova, J. Catal. 194
(2000) 318.
[30] Z. Sobalik, J. Dedecek, D. Kaucky, B. Wichterlova, L. Drozdova, R.
Prins, J. Catal. 194 (2000) 330.
[31] R. Joyner, M. Stockenhuber, J. Phys. Chem. B 103 (1999) 5963.
[32] R.Q. Long, R.T. Yang, J. Catal. 194 (2000) 80.
[33] F. Al-Mashta, N. Sheppard, V. Lorenzelli, G. Busca, J. Chem. Soc.,
Faraday Trans. I 18 (1982) 979.
Acknowledgments
[34] H. Bosh, F. Jassen, Catal. Today 2 (1988) 369.
[35] G. Ramis, L. Yi, G. Busca, Catal. Today 28 (1996) 373.
We are grateful to NSF and EPRI for support.