Possible reasons for the superior performance of zeolite-based catalysts in the reduction of nitrogen oxides

Article abstract of DOI:10.1016/j.jcat.2005.06.022

Literature data and results obtained in this laboratory are used to compare catalysts supported on either zeolites or amorphous oxides and used in NO _x reduction to N₂. The data show that the catalytic activity of the zeolite-based materials is higher if alkanes, alkenes, or organic oxygenates are used as reductants, but with ammonia the performance of both groups of catalysts is comparable. These findings are rationalized in the context of the present knowledge of the mechanism of NO_x reduction. Transition metals form oxo-ions on zeolites, but solid solutions on alumina. On zeolites, optimal Coulomb stabilization can be achieved if heterolytic dissociation of molecules transforms multipositive cations to monopositive cations. N₂O₄ dissociates to form NO++NO3- on BaNa/Y. In contrast, no NO⁺ ions are detected on BaO/γ-Al ₂O₃. Another important parameter is the high heat of adsorption of small molecules in the narrow zeolite pores. Because oxygenates are superior to alkanes in competing against water for active sites, NO _x reduction with alkanes is favored by zeolites with low Al/Si ratios, but NO_x reduction with acetaldehyde is more efficient on a faujasite with a high Al/Si ratio. Over these catalysts, water vapor actually enhances NO_x reduction by preventing formation of crotonaldehyde, which would poison catalyst sites.

Full text of DOI:10.1016/j.jcat.2005.06.022

Journal of Catalysis 235 (2005) 201–208
www.elsevier.com/locate/jcat
Possible reasons for the superior performance of zeolite-based catalysts in
the reduction of nitrogen oxides
Meijun Li, Younghoon Yeom, Eric Weitz, Wolfgang M.H. Sachtler ∗
Department of Chemistry and Institute for Environmental Catalysis, Northwestern University, Evanston, IL 60208, USA
Received 21 April 2005; revised 21 June 2005; accepted 23 June 2005
Abstract
Literature data and results obtained in this laboratory are used to compare catalysts supported on either zeolites or amorphous oxides
and used in NO reduction to N . The data show that the catalytic activity of the zeolite-based materials is higher if alkanes, alkenes, or
x
2
organic oxygenates are used as reductants, but with ammonia the performance of both groups of catalysts is comparable. These findings are
rationalized in the context of the present knowledge of the mechanism of NO reduction. Transition metals form oxo-ions on zeolites, but
x
solid solutions on alumina. On zeolites, optimal Coulomb stabilization can be achieved if heterolytic dissociation of molecules transforms
+
−
+
multipositive cations to monopositive cations. N O dissociates to form NO + NO on BaNa/Y. In contrast, no NO ions are detected
2
4
3
on BaO/γ -Al O . Another important parameter is the high heat of adsorption of small molecules in the narrow zeolite pores. Because
2
3
oxygenates are superior to alkanes in competing against water for active sites, NO reduction with alkanes is favored by zeolites with low
x
Al/Si ratios, but NO reduction with acetaldehyde is more efficient on a faujasite with a high Al/Si ratio. Over these catalysts, water vapor
x
actually enhances NO reduction by preventing formation of crotonaldehyde, which would poison catalyst sites.
x
2005 Published by Elsevier Inc.
Keywords: De-NO catalysis; Zeolites; Competitive adsorption; Charge dissipation
x
1. Introduction
of benzene to phenol [4]. In this case, the propensity of ze-
olites to favor formation of oxo-complexes, such as ferryl
groups, appears to be crucial [5]. Silicalite, a material with
the same structure as MFI zeolites, has been identified as an
optimal support for the formation of hydrogen peroxide over
Ti ions and for the synthesis of epoxides [6].
In the field of selective catalytic reduction (SCR) of ni-
trogen oxides to nitrogen, zeolite-based materials such as
Cu/MFI [7,8], Fe/MFI [9–11], Co/MFI [12,13], and Ba/Y
[14,15] are known to display both high activity and selectiv-
ity. However, because their thermal stability is inferior to γ -
Al2O3, TiO2, or ZrO2 supports, their efficacy in “on board”
applications for abating toxic exhaust emissions from cars
and diesel trucks is uncertain, although high performance
under laboratory conditions has been well established. Thus
catalysts have been tested that use the same metal ions on
nonzeolitic oxide supports, but these catalysts were often
found to be less active for NOx reduction. Bethke et al. [16]
tested Al2O3 supported V, Cr, Mn, Fe, Co, Ni, Cu, and Zn
The introduction of zeolites in catalytic cracking and re-
forming has had a major impact on the petroleum indus-
try [1]. One motivation for replacing traditional alumina and
silica–alumina catalysts by zeolites in this field is their su-
perior acidity in the H form. Zeolites have also been used
as catalysts for other processes. A prominent example is
H-ZSM-5, a zeolite of MFI structure and an excellent cat-
alyst for the conversion of methanol to high-octane gasoline
in Mobil’s MTG process. For this application, the geome-
try of the MFI structure is crucial [2]. The same holds for
the selectivity of acid zeolites for direct xylene isomerization
toward the para-isomer [3]. More recently, superior zeolite-
based catalysts have been reported for the one-step oxidation
*
Corresponding author.
E-mail address: wmhs@northwestern.edu (W.M.H. Sachtler).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.06.022
2005 Published by Elsevier Inc.

202
M. Li et al. / Journal of Catalysis 235 (2005) 201–208
catalysts for NOx reduction with ethylene, and compared
them with catalysts that have the same metal ions in the cav-
ities of an MFI zeolite. These authors report that the temper-
ature at which the catalysts achieve their maximum conver-
sion to N2 is significantly higher for the γ -Al2O3-supported
catalysts than for the MFI-supported systems. This finding
implies superior activity for the zeolite-based systems. Only
zinc catalysts reach their activity maximum at almost the
same temperature (near 600 ◦C) for either support. Because
it is well known that transition metal ions form aluminates or
solid solutions with γ -Al2O3 at high temperature [17], these
authors wondered whether the superior performance of the
zeolite-supported catalysts could be due to a better mainte-
nance of the original dispersion of the active ions.
Because SCR catalysts are usually tested in flow reactors
under conditions where the temperature is increased in an
atmosphere with a large excess of O2 over NOx, it is useful
to remember that the shapes of plots of the N2 yield versus
temperature reflect a competition between two reactions. At
low temperature, the reductant reacts mainly with NO2 or
N2O3 [18] in a sequence of steps leading to the production
of N2. At high temperature, however, the thermodynamically
preferred combustion of organic reductants with O2 prevails.
Likewise, if ammonia is used as the reductant, then its oxi-
dation to NO becomes significant at high temperature. For
all of these reductants, the N2 yield will ultimately decrease
with temperature. Thus, one can argue that the activity of de-
NOx catalysts for the desired reaction can be defined by their
“light off” temperature and the early part of the ascending
branch of the N2 yield-versus-T plot, whereas the yield max-
imum reflects the point at which the undesired combustion
process starts to win out over the desired reaction because of
limited selectivity for the latter process.
literature data were measured at the same space velocity
and/or with the same feed composition; we use such data
only if differences between catalysts with different supports
are large.
2. Experimental
Here we describe the procedure for catalysts prepared in
our laboratory. For catalysts used by other authors, refer to
their papers.
Briefly, 5.5 wt% BaNa/Y was obtained by a three-fold ion
exchange of Na/Y (Si/Al = 2.5, Aldrich) with an aqueous
0.1 mol Ba(NO3)2 solution. For the preparation of 2 wt%
Ag/Y, an AgNO3 solution was used at ambient temperature.
After every exchange, the slurry was stirred for 48 h before
vacuum-filtered, washed thoroughly with double-deionized
H2O, and dried in air. The alumina-supported catalysts
4 wt% Ag/γ -Al2O3, 5.5 wt% Ba/γ -Al2O3, and 5.5 wt%
Ba/TiO2 were prepared by wet impregnation.
We recorded in situ Fourier transformed infrared spectra
in transmission mode with a Bio-Rad Excalibur FTS-3000
infrared spectrometer equipped with a MCT detector. Unless
otherwise stated, we obtained each spectrum by averaging
70 scans at a resolution of 2 cm−1. The “homemade” in-
frared cell, described in more detail in [15], is based on the
design in [19] and comprises a stainless-steel cube with two
CaF2 windows that can be pumped differentially. A fresh
catalyst sample was used for each new set of experiments.
We carried out catalytic tests in a fixed-bed microflow
reactor, described in more detail in [20]. Briefly, 0.2 g of
the powdered catalyst was packed into the quartz reactor.
The composition of the feed gas was regulated by mass-flow
controllers (URS-100, UNIT Instruments). H2O was intro-
duced into the reaction system by bubbling He through a
H2O saturator. The composition of the effluent was analyzed
“on-line” with a gas chromatograph (GC) equipped with a
thermal conductivity detector (TCD) (5A column for N2 and
CO, Porapak Q for CO2). A cold trap was positioned be-
tween the reactor and the GC to protect the column from
H2O.
If different resistance against the interaction of the ac-
tive ion with the support were the only cause of the dif-
ferent effects of zeolite and non-zeolite supports, then one
would expect to find similar performance with either sup-
port at low temperature. In the present work we tested this
hypothesis by comparing the N2 yield over different cata-
lysts, not only at their yield maximum, but also in the low-
temperature regime, where other causes, such as differences
in physisorption, the presence or absence of an electric field
in the cavities of the support, and/or different energies of
charge separation might affect the impact of the support on
catalytic performance.
3. Results
We compared literature data and results measured in
our laboratory for NOx reduction to N2 with the following
reductants: propane, isobutane, propene, ethanol, acetalde-
hyde, and ammonia. The zeolite-based catalysts are Cu/MFI,
Cu/Mor, Fe/MFI, Co/MFI, Ag/Y, and BaY (or BaNa/Y).
We compare the performance over these catalysts with the
performance over Cu/TiO2, Cu/ZrO2, Co/γ -Al2O3, Ag/γ -
Al2O3, Ba/γ -Al2O3, and Ba/TiO2. We compare the temper-
atures at which the maximum N2 yield and a lower yield
(20% in most cases) are achieved with catalysts with the
same active element supported on different carriers. Not all
Fig. 1a shows data for the reduction of NO2 with
acetaldehyde over BaNa/Y as a function of the reaction tem-
perature with a feed containing 2% water. The maximum N2
yield is obtained at 200 ◦C. Kung et al. [21] reported that
over TiO2, this maximum is obtained at T > 300◦C. Fig. 1b
shows our results for the same reaction over BaO/γ -Al2O3
and BaO/TiO2. In agreement with the results of Kung [21],
the temperature of maximum N2 yield is >300 ◦C for both
catalysts. The temperature at which a 20% N2 yield is regis-
tered is 95 ◦C for BaNa/Y, but a much higher temperature is

M. Li et al. / Journal of Catalysis 235 (2005) 201–208
203
Fig. 2. NO reduction with ethanol in a wet feed as a function of reaction
2
Fig. 1. NO reduction with acetaldehyde in a wet feed as a function of
2
temperature over (a) Ag/γ -Al O and (b) AgY. Ethanol, 1000 ppm; NO ,
2
3
2
reaction temperature over (a) Ba,Na/Y, (b) BaO/γ -Al O and BaO/TiO .
500 ppm; O , 7.0%; H O, 2.0%; total flow rate: 200 ml/min, 0.2 g catalyst.
2
3
2
2
2
AA, 500 ppm; NO , 500 ppm; O , 7.0%; H O, 2.0%; total flow rate,
2
2
2
200 ml/min; 0.2 g catalyst.
Table 3
NO + C H ; Co sites
x
3
6
Table 1
Catalyst
T at 20% yield
( C)
T at max. yield
( C)
Y
Ref.
max
NO + CH CHO; Ba sites
x
3
◦
◦
(%)
Catalyst
T at 20% yield
( C)
T at max. yield
( C)
Y
Ref.
max
a
◦
◦
Co/MFI
310
400
350
460
83
78
[22]
[22]
(%)
b
Co/Al O
2
3
Ba,Na/Y
BaO/γ -Al O
BaO/TiO
95
250
135
200
315
400
90
51
45
This work
This work
This work
a
0.095% NO, 0.1% C H , 5% O and 1.7% H O; total flow rate,
3
6
2
2
2
3
100 ml/min; 0.5 g 1.24 wt% Co/MFI.
2
b
0.095% NO, 0.1% C H , 5% O and 1.7% H O; total flow rate,
3
6
2
2
100 ml/min; 0.5 g 2 wt% Co/Al O .
2
3
Table 2
NO + C H OH; Ag sites
x
2
5
Ag/Y; the relevant parameters are given in Table 2. Clearly,
a much higher temperature is required to attain both the
20% yield point and the maximum N2 yield over Ag/γ -
Al2O3 than over Ag/Y. The required temperature for the 20%
yield point for Ag/γ -Al2O3 is 110 ◦C higher than that for
Ag/Y. All of these criteria indicate a superior activity for the
zeolite-supported catalysts.
For the SCR of NOx with propene, we again compare
(in Table 3) the temperature at which the catalysts reach
the 20% yield point [22]. This temperature is 90 ◦C lower
Catalyst
T at 20% yield
( C)
T at max. yield
( C)
Y
Ref.
max
◦
◦
(%)
Ag/Y
70
290
350
83
100
This work
This work
Ag/γ -Al O
180
2
3
required for BaO on γ -Al2O3 or TiO2; the relevant parame-
ters are given in Table 1.
Fig. 2 shows NO2 reduction with ethanol in a wet feed,
as a function of reaction temperature, over Ag/γ -Al2O3 and

204
M. Li et al. / Journal of Catalysis 235 (2005) 201–208
Table 4
NO + C H ; sites Fe
x
2
4
Catalyst
T at 20% yield
◦
T at max. yield
( C)
Y
Ref.
max
◦
( C)
(%)
a
Fe-MOR-97
250
400
250
400
20
21
[27]
[27]
b
FeO/Al O
2
3
a
−3
−3
NO, 1000 ppm; C H , 250 ppm; O , 2%; and W/F, 0.2 g s cm
.
.
2
4
2
b
NO, 1000 ppm; C H , 250 ppm; O , 2%, and W/F, 0.2 g s cm
2
4
2
Table 5
NO + NH , Cu sites
x
3
Catalyst
T at max. yield
( C)
Y
T at 50% yield
( C)
Ref.
max
◦
◦
(%)
a
Cu/MOR
240
300
220
100
35
95
180
–
180
[29]
[30]
[30]
b
CuO/ZrO
2
b
CuO/TiO
a
2
NO, 500 ppm; NH , 500 ppm; O , 5%; GHSV, 100,000/h; 2.3 wt%
3
2
Cu/MOR.
b
NO, 35 ml/min; NH , 35 ml/min; air, 1350 ml/min; GHSV, 20,000/h.
3
over Co/MFI than over Co/γ -Al2O3. Though the maximum
yields are comparable, the zeolite-based catalyst achieves
this yield at a temperature that is also 90 ◦C lower than for
the Co/γ -Al2O3 catalyst.
Supported iron catalysts have been used in the SCR of
NOx. High yields are achieved with isobutane over Fe/MFI
[23–26]. As is evident from Table 4, with ethylene a much
higher temperature is required over γ -Al2O3-supported Fe
catalysts than over Fe/Mor for both the 20% yield point and
the maximum yield [27].
Although there are data for SCR with propane over
Fe/MFI [11], Co/γ -Al2O3 [22], and Cu/MFI [28] catalysts,
comparisons are problematic because different space veloci-
ties have been used. Nonetheless, the data show that much
higher temperatures for both the maximum yield and the
20% yield are required for Co/γ -Al2O3 than for Fe/MFI or
Cu/MFI.
Reduction of NOx with ammonia has been studied by nu-
merous authors [29,30]. Table 5 presents literature data ob-
tained with Cu supported on mordenite, titania, and zirconia.
The highest yield, of 95%, is found for CuO/TiO2 at 220 ◦C
and Cu/MOR at 240 ◦C; the 50% yields are registered at ap-
proximately the same temperature over both catalysts. It is
not clear, and not within the scope of this paper, to con-
sider, why the TiO2 and ZrO2 supports behave differently.
Sun et al. reported that over Fe/MFI ammonia, conversion
to N2 exceeds 70% at 275 ◦C and reaches 90% at 300 ◦C.
Clearly, very high N2 yields are possible with ammonia over
both zeolite-supported and TiO2-supported transition metal
catalysts.
The totality of these results suggests that for NOx reduc-
tion with organic reductants, the zeolite-supported catalysts
are more active than catalysts using the same ions on other
supports. But when ammonia is used as the reductant, cat-
alytic performance is comparable for zeolitic and nonzeolitic
supports.
Fig. 3. FTIR spectra after exposure of Na-Y (a) and BaNa/Y (b) to 1 Torr
◦
of NO . Each sample was activated at 430 C for 3 h in vacuum before
2
the NO was admitted to the cell. For both top to bottom (as indicated by
2
◦
the arrows) panels, the spectra were taken at 50, 100, 150, and 200 C,
and after subsequent evacuation. The temperature was increased at a rate of
◦
2.5 C/min.
To rationalize these findings, it is useful to consider
the adsorption complexes that have been identified through
IR spectroscopy. Fig. 3a shows a series of FTIR spectra
recorded after exposing Na/Y to NO2. Absorption bands at
2011 and 2160 cm−1 are due to NO+ ions formed by the
reaction
−
N2O4 (ꢀ 2NO2) → NO+ads + NO3ads
.
Details of this reaction have been reported elsewhere [15,31].
The nitrate counter ion for NO+ is initially observed as a
strong absorption band at 1407 cm−1 (see the top trace). In-
creasing the temperature causes this band to split into two
bands, at 1370 and 1415 cm−1. Fig. 3b shows a series of
FTIR spectra recorded by exposing BaNa/Y to NO2. The ab-
sorption bands at 2034 and 2176 cm−1 are due to the NO+
ions, that is, the co-product of dissociative chemisorption of
N2O4. These bands are shifted to higher energy on BaNa/Y
than on Na/Y (∼23 and ∼16 cm−1, respectively). This in-
dicates that the oxygen anions of the zeolite framework in
Ba,Na/Y donate less electron density to NO+ than those in
Na/Y. This result is ascribed to the lower negative charge of
the O2− ions in Ba,Na/Y in comparison to Na/Y, because
Ba2+ ions more strongly attract electrons from their oxygen
neighbors than Na+ ions do. The stretching frequency of

M. Li et al. / Journal of Catalysis 235 (2005) 201–208
205
Fig. 5. FTIR spectra after exposure of BaO/γ -Al O to 1 Torr of NO (a)
2
3
2
15
◦
and NO (b) at 50 C.
2
(see the top trace in (a)). Based on literature data, these bands
are assigned to bridging bidentate nitrates [32–34]. There
are more bands on BaO/γ -Al2O3 than on γ -Al2O3 (see (b)),
because in the former system two solid phases, BaO and γ -
Al2O3, coexist. The features at 1303, 1451, and 1481 cm−1
are due to ionic nitrates formed on BaO. The bands at 1261
and 1632 cm−1 are assigned to bridging bidentate nitrate
[ν(N=O) and νasym(NO2)] on BaO/γ -Al2O3. The band at
1566 cm−1 is ascribed to bridged chelating bidentate nitrate
on BaO. A broad absorption band at 1940 cm−1 on γ -Al2O3
and at 1955 cm−1 on BaO/γ -Al2O3 gradually decreases in
intensity with upon increasing temperature and can hardly be
seen at 200 ◦C. Ba/γ -Al2O3 was exposed to 15NO2 to check
whether the 1955 cm−1 band shifted. Indeed, it did shift
(by ∼41 cm−1) to a lower frequency with 15NO2 (Fig. 5).
Fig. 4. FTIR spectra after exposure of γ -Al O (a) and BaO/γ -Al O (b)
2
3
2 3
◦
to 1 Torr of NO . Each sample was activated at 430 C for 3 h in vacuum
2
before the NO was admitted to the cell. For both top to bottom (indicated
2
◦
by the arrows) panels, the spectra were taken at 50, 100, 150, and 200 C,
and after subsequent evacuation. The temperature was increased at a rate of
◦
2.5 C/min.
NO+ varies with the nature of the cation because it depend∗s
on the degree of electron transfer from the anion to the π
orbital of NO+. Nitrate ions on BaNa/Y are characterized
by strong absorption bands at 1366 and 1414 cm−1 (see the
top trace). For both catalysts, the bands centered at 1616 and
1766 cm−1 are due to gas phase NO2 and the ν1 + ν2 com-
bination band of NO3−, respectively. The thermal stability
of NO+ is different for Na/Y and BaNa/Y. Raising the tem-
perature from 56 to 100 ◦C causes the intensity of the ba+nd
for NO+ads on BaNa/Y to decrease by 7%, and that of NOads
on Na/Y to decrease by 84%. At 200 ◦C, no NOa+ds band
can be detected on Na/Y, but on BaNa/Y the NO+ band re-
mains clearly detectable (although not on the absorbance
scale used +in this figure). We conclude that the concentra-
tion of NOads is higher at 200 ◦C on BaNa/Y than on Na/Y.
When the temperature is raised from 50 to 200 ◦C, the in-
tensity of the nitrate bands on BaNa/Y decreases by ∼84%
and that for the bands on Na/Y decreases by ∼96%. This in-
dicates that the thermal stability of nitrates is higher in the
presence of exchanged Ba2+ ions.
δ+
This band is probably due to the interaction of NO2 and
Al3+. A similar band has been observed at ∼1970 cm−1
when γ -Al2O3 was exposed to NO + O2 [32]. This band
δ+
has been assigned to NO2 interacting with Al3+ [32]. No
NO+ bands are observed between 2000 and 2400 cm−1 with
the sensitivity scale used here. If any NO+ ions were formed
on this catalyst, their concentrations were orders of magni-
tude lower than on the zeolite-based catalysts BaNa/Y and
Na/Y. This indicates that dissociative chemisorption of N2O4
to NO+ +NO3− is much less on alumina than on the zeolite-
based catalysts.
An absorption band at 2253 cm−1 is observed with
BaO/γ -Al2O3 but not with γ -Al2O3. This band becomes
stronger with increasing temperature (Fig. 4a) and does not
shift when BaO/γ -Al2O3 is exposed to 15NO2. The as-
signment of this band is beyond the scope of this paper.
If the extinction coefficients for adsorbed nitrates in the
1100–1700 cm−1 region are similar for BaO/γ -Al2O3 and
γ -Al2O3, then it follows from the integrated absorbances
at 50 ◦C that the concentration of nitrate is roughly 2.5 or-
Fig. 4 compares γ -Al2O3 and BaO/γ -Al2O3. Panels (a)
and (b) show the FTIR spectra after exposure to NO2. Strong
bands appear immediately at 1250, 1591, and 1625 cm−1

206
M. Li et al. / Journal of Catalysis 235 (2005) 201–208
ders of magnitude higher on BaO/γ -Al2O3 than on Al2O3.
Raising the temperature from 50 to 200 ◦C decreases the in-
tegrated absorbance by ∼56% for both solids; however, the
overall intensity of adsorbed nitrate in this region at 200 ◦C
is still ∼2.5 times higher on BaO/γ -Al2O3.
acid. The acetate ions exchange their carboxylate group with
NO2, yielding the acid-form of nitromethane. This interme-
diate is subsequently dehydrated (probably through a path-
way involving formo-hydroxamic acid) to isocyanic acid,
which then reacts with water
For BaNa/Y, the diss−ociation of N2O4 is an important
source of adsorbed NO3 ions. But for the alumina-based
catalysts, this route is apparently unavailable; as previously
proposed, a major path is
CH3NO2 → HNCO + H2O → CO2 + NH3.
(4)
Considering the reaction mechanism along with data on the
efficiency of different catalytic systems for reducing NOx
to N2, it appears that zeolite’s special role is manifested in
the reduction of NOx to NH3 and that zeolite has less ef-
fect on the subsequent steps. To rationalize these findings,
we consider three characteristic propensities that distinguish
zeolites from other solid oxides:
BaO + 3NO2 → Ba(NO3)2 + NO.
(1)
Despres et al. [35] studied the forward and reverse reactions
in Eq. (1) on BaO/TiO2 using TPD. They observed that NO2
consumption and NO production in the forward reaction oc-
cur in a 3:1 ratio. They also observed the reverse process,
that is, reduction of the nitrate with NO. As mentioned in
Ref. [32], formation of nitrate from NO2 is also possible
through BaO2.
1. Selective oxidation is preferred on isolated oxo-ions,
such as [Cu–O–Cu]2+ or [FeO]+, in zeolites, but non-
selective combustion is preferred on oxide particles or
their solid solutions. This aspect is crucial for the selec-
tivity of catalysts containing transition metal ions. But
since this paper focuses on the effects of the support on
the activity of the catalysts, with special emphasis on
catalysts that do not contain transition metal ions, this
aspect is not pursued further here.
2. Dissociative chemisorption of molecules such as H2O
(to H+ + OH−) or N2O4 (to NO+ + NO3−) is energet-
ically preferred on zeolites with multipositive cations in
their cavities.
4. Discussion
The catalytic data show that with alkanes, alkenes, or
oxygenates as reductants, the NOx reduction activity of
zeolite-supported catalysts is superior to that of catalysts us-
ing conventional nonzeolitic oxides as the support. Both the
maximum yield and the yield measured at low conversion
support this conclusion.
Before discussing possible reasons for this difference in
performance, we briefly recapitulate our present understand-
ing of the reaction mechanism for NOx reduction. Isotopic
labeling clearly shows that in each N2 molecule formed by
SCR, one of the N atoms stems from a precursor in which
it had the formal oxidation state of −3, as in NH3, and the
other N atom had an oxidation number of +3, as in N2O3
[18,36–38]. N2 formation thus occurs in a process for which
the decomposition of ammonium nitrite is the prototype,
3. The heat of physisorption of many molecules is higher
in the nanopores of a zeolite than on amorphous oxides.
We now discuss points 2 and 3 in more detail. For point 2,
a characteristic of zeolite structure is that the negative charge
is effectively localized on Al-centered oxygen tetrahedra in
the rigid lattice, but the positive-charge carriers are ions lo-
cated in the cavities. Only the positive-charge carriers are
mobile. For mono-positive ions, such as Na+, it is obvious
that Coulomb interactions will be optimum when the posi-
tive ions are positioned near the negative-charge carriers. But
for multipositive ions, such as Ba2+ and Ga3+, a compro-
mise position must be found for the positive-charge carrier,
because it cannot simultaneously be a close neighbor of each
of the two or three immobile negative charges that it com-
pensates. In this situation the Coulomb energy of the system
will be lowered by adsorbates that dissipate the (n+) charge
of the positive ion into n(1+) charges that can seek positions
near an Al-centered tetrahedron of the zeolite.
NH4NO2 = N2 + 2H2O.
(2)
As was found with aqueous solutions of NH4NO2 by Millon
in 1847 [39] and with NH4NO2 deposited on quartz by Li
et al. [40], this decomposition is rapid near 100 ◦C and does
not require a catalyst. For the SCR of NOx with ammonia,
the highest yield of N2 is always obtained if NO and NO2
are present in a 1/1 ratio. A catalyst can facilitate formation
of this ratio by increasing the rate of either the forward or
the backward reaction,
Examples of such charge dissipation are hydrolysis reac-
tions, such as
2NO + O2 ꢀ 2NO2.
(3)
For reductants other than ammonia or an amine, the reduc-
tant must first reduce part of the NOx to NH3, as was first
suggested by Poignant et al. [41]. This ammonia will then
react with the remaining NOx [15]. Conversion of part of the
NOx to ammonia has been studied in considerable detail for
BaNa/Y catalysts using acetaldehyde as the reductant [15].
In brief, the acetaldehyde is first oxidized by NO2 to acetic
Ga3+ + H2O = [GaO]+ + 2H+.
(5)
In the case of BaNa/Y, the results of this paper show that
charge dissipation is achieved by the dissociative chemisorp-
tion of N2O4,
N2O4 + Ba2+ = [Ba–NO3]+ + [NO]+,
(6)

M. Li et al. / Journal of Catalysis 235 (2005) 201–208
207
which effectively transforms a 2+ charge carrier into two
mobile 1+ charged carriers. The IR results in this paper
demonstrate that the [NO]+ ions formed in this process on
BaNa/Y exhibit a fairly strong IR band, but no such band
is detectable with BaO/γ -Al2O3 or γ -Al2O3. Because the
[NO]+ ion is known to form nitrous acid (HONO) on reac-
tion with water, its presence provides a path for the forma-
tion of ammonia nitrite. This path may be less important in
the presence of a high partial pressure of ammonia, because
reduction of nitrate ions to nitrite ions provides another ef-
ficient route toward ammonia nitrite. However, for catalytic
reactions where ammonia must be formed by NOx reacting
with an organic molecule, nitrite formation through hydra-
tion of the [NO]+ ion is likely to become dominant, because
the concentration of ammonia will be low in the steady state.
In terms of point 3, the third characteristic difference
between zeolite and nonzeolite supports is the heat of ph-
ysisorption in the pores of the support. Data obtained by
calorimetry or derived from isosteric equilibria have been
reported by various authors, including Breck [42] and Stach
et al. [43]. Using silicalite analogs of MFI and FAU lattices,
Eder and Lercher [44] showed that the heat of adsorption
of n-hexane depends strongly on the pore diameter, being
>70 kJ/mol on MFI but only 45 kJ/mol on FAU. When
discussing the consequences of adsorption on the effective-
ness of SCR catalysts, one must consider that actual de-NOx
processes always occur in the presence of a partial pressure
of water exceeding that of the organic reductant or ammo-
nia. Adsorption of the reductant thus must compete with
adsorption of water on the same sites. Work focusing on
this competitive adsorption has been published by Hunger
et al. [45], who used temperature-programmed desorption
to determine the activation energy, Edes, which will be very
similar to the enthalpy of adsorption. These authors showed
that methanol displaces adsorbed water from the strongest
adsorption sites of Na/MFI. The Edes of methanol is 75–
100 kJ/mol. This finding explains why oxygenates such as
acetaldehyde or ethanol can achieve NOx reduction in the
presence of water vapor at a lower temperature than alka-
nes, which also require that a transition metal oxo-complex
be partially oxidized.
We conclude that enhanced adsorption of the reductant and
formation of NO+ ions are likely causes of the superior ac-
tivity of zeolite-based de-NOx catalysts. Because adsorption
of the reducing agent occurs in competition with water va-
por, one can understand that for nonpolar hydrocarbons, the
“hydrophobic” zeolites [46] with low Al/Si ratio are op-
timum, but with polar molecules, such as acetaldehyde or
ethanol, high N2 yields are obtained at lower temperatures
over “hydrophilic” zeolites with high Al/Si ratios.
Acknowledgments
This research was supported by the Chemical Sciences,
Geosciences and Biosciences Division, Office of Basic En-
ergy Sciences, Office of Science, US Department of Energy
(Grant No. DE-FG02-03ER15457). The authors thank the
donors of the American Chemical Society Petroleum Re-
search Fund for their partial support of this work and Pro-
fessor R. Snurr for his helpful suggestions.
References
[1] H. Heinemann, in: G. Ertl, H. Knözinger, H. Weitkamp (Eds.), Hand-
book of Heterogeneous Catalysis, vol. 1, Chemie, Weinheim, Ger-
many, 1997, pp. 35–48.
[2] S.L. Meisel, J.P. Mc Cullough, C.H. Lechthaler, P.B. Weia, Chem.
Tech. 6 (1976) 86.
[3] S.M. Csicsery, Zeolites 4 (1984) 202–213.
[4] G.I. Panov, A.S. Kharitonov, V.I. Sobolev, Appl. Catal. A 98 (1993) 1.
[5] K.A. Dubkov, N.S. Ovanesyan, A.A. Shteinman, E.V. Starokon, G.I.
Panov, J. Catal. 207 (2002) 341.
[6] B. Notari, Adv. Catal. 41 (1996) 253.
[7] M. Iwamoto, H. Yhiro, Y. Yu-U, S. Shundo, N. Mizuno, Sholubai 32
(1990) 430.
[8] J. d’ Itri, W.M.H. Sachtler, Catal. Lett. 15 (1992) 289.
[9] X. Feng, K. Hall, Catal. 166 (1997) 368.
[10] H.-Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
[11] H.-Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, Catal. Today 54
(1999) 483.
[12] Y. Li, J.N. Armor, Appl. Catal. B 1 (1992) L31.
[13] X. Wang, H.-Y. Chen, W.M.H. Sachtler, Appl. Catal. B 19 (2001) 47.
[14] G. Panov, R.G. Tonkyn, M.L. Balmer, C.H.F. Peden, SAE 2001 01-
3513.
[15] Y.H. Yeom, B. Wen, W.M.H. Sachtler, E. Weitz, J. Phys. Chem. B 108
(2004) 5386.
[16] K.A. Bethke, M.C. Kung, B. Yang, M. Shah, D. Alt, C. Li, H.H. Kung,
Catal. Today 26 (1995) 169.
[17] P.H. Bolt, Transition metal aluminate formation in alumina supported
model catalysts, thesis, University of Utrecht, The Netherlands, 1994.
[18] Q. Sun, Zh.-X. Gao, H.-Y. Chen, W.M.H. Sachtler, J. Catal. 201 (2001)
89.
[19] P. Basu, T.H. Ballinger, J.T. Yates, Rev. Sci. Instrum. 59 (1988) 1321.
[20] B. Wen, Y.H. Yeom, E. Weitz, W.M.H. Sachtler, Appl. Catal. B 48
(2004) 125.
[21] A.P. Kozlova, H.Y. Law, M.C. Kung, H.H. Kung, Catal. Lett. 95 (2004)
211.
[22] J.Y. Yan, M.C. Kung, W.M.H. Sachtler, H.H. Kung, J. Catal. 172
(1997) 178.
[23] H.-Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
[24] T.V. Voskoboinikov, H.-Y. Chen, W.M.H. Sachtler, Appl. Catal. B 19
(1998) 275.
The concept of competitive adsorption of reductant and
water is illustrated by the well-known negative effect of wa-
ter vapor on the NOx reduction rate over such catalysts as
Cu/MFI and Co/MFI. It is, therefore, quite remarkable that
water vapor has a much less inhibiting effect when an oxy-
genate is used as the reductant. In the case of NOx reduction
with acetaldehyde over BaNa/Y, our data show higher NOx
conversion in the presence of water than in its absence [20].
This remarkable result has been rationalized by considering
that the presence of water will shift the equilibrium for aldol
condensation (7) to the left, thus minimizing the formation
of crotonaldehyde, which is a potential precursor of coke that
deactivates catalyst sites:
2CH3CHO = CH3–CH=CH–CHO + H2O.
(7)

208
M. Li et al. / Journal of Catalysis 235 (2005) 201–208
[25] H.-Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 186 (1999)
91.
[36] H.-Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 180 (1998)
171.
[26] H.-Y. Chen, X. Wang, W.M.H. Sachtler, Phys. Chem., Chem. Phys. 2
(2000) 3083.
[37] H.-Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 186 (1999)
91.
[27] S. Sato, H. Hirabayashi, H. Yahiro, N. Mizuno, M. Iwamoto, Catal.
Lett. 12 (1992) 193.
[38] H.-Y. Chen, Q. Sun, B. Wen, Y.-H. Yeom, E. Weitz, W.M.H. Sachtler,
Catal. Today 96 (2004) 1.
[28] R. Gopalakrishnan, P.R. Stafford, J.E. Davidson, W.C. Hecker, C.H.
Bartholomew, Appl. Catal. B 2 (1993) 165.
[29] S.W. Ham, H. Chio, I.S. Nam, Y.G. Kim, Catal. Lett. 42 (1996) 35.
[30] T. Iizuka, H. Ikeda, J. Chem. Soc., Faraday Trans. 1 82 (1986) 61.
[31] Y.H. Yeom, J. Henao, M.-J. Li, W.M.H. Sachtler, E. Weitz, J. Ca-
tal. 231 (2005) 181.
[32] F. Prinetto, G. Ghiotti, I. Nova, L. Lietti, E. Tronconi, P. Forzatti, J.
Phys. Chem. B. 105 (2001) 12732.
[33] F. Prinetto, G. Ghiotti, I. Nova, L. Castoldi, L. Lietti, E. Tronconi, P.
Forzatti, Phys. Chem. Chem. Phys. 5 (2003) 4428.
[34] J. Szányi, J.H. Kwak, D.H. Kim, S.D. Burton, C.H.F. Peden, J. Phys.
Chem. B. 109 (2005) 27.
[39] E. Millon, Ann. Chim. Phys. 19 (1847) 255.
[40] M.-J. Li, J. Henao, Y.-H. Yeom, E. Weitz, W.M.H. Sachtler, Catal.
Lett. 98 (2004) 5.
[41] F. Poignant, J. Saussey, J.C. Lavalley, G. Mabilon, Chem. Comm.
(1995) 89.
[42] D.W. Breck, Zeolite Molecular Sieves, Structure, Chemistry and Use,
Wiley, New York, 1974, p. 654.
[43] H. Stach, U. Lohse, H. Thamm, W. Schirmer, Zeolites 74 (1986)
74.
[44] F. Eder, J.A. Lercher, Zeolites 18 (1997) 7.
[45] B. Hunger, S. Matysik, M. Heuchel, W.-D. Einicke, Langmuir 13
(1997) 6249, J. Thermal Anal. Calorim. 64 (2001) 1183.
[46] D.H. Olson, W.O. Haag, W.S. Borghard, Microporous Mesoporous
Mater. 35–36 (2000) 435.
[35] J. Despres, M. Koebel, O. Krocher, M. Elsener, A. Wokaun, Appl.
Catal. B. 43 (2003) 389.