Anda di halaman 1dari 10

Available online at www.sciencedirect.

com

Microporous and Mesoporous Materials 111 (2008) 124133


www.elsevier.com/locate/micromeso

The role of Brnsted acidity in the SCR of NO over Fe-MFI catalysts


Michael Schwidder a, M. Santhosh Kumar b, Ursula Bentrup b, Javier Perez-Ramrez c,
Angelika Bruckner b, Wolfgang Grunert a,*
b

a
Laboratory of Industrial Chemistry, Ruhr University Bochum, P.O. Box 102148, D-44780 Bochum, Germany
Leibniz Institute for Catalysis Rostock, Berlin Branch (former Institute of Applied Chemistry Berlin-Adlershof e.V.), Richard-Willstatter-Str. 12,
D-12489 Berlin, Germany
c
Laboratory for Heterogeneous Catalysis, Catalan Institution for Research and Advanced Studies (ICREA), Institute of Chemical Research of
Catalonia (ICIQ), Av. Pasos Catalans 16, 43007 Tarragona, Spain

Received 2 April 2007; received in revised form 8 July 2007; accepted 10 July 2007
Available online 19 July 2007

Abstract
The selective catalytic reduction (SCR) of NO with isobutane and with NH3 was studied over Fe-MFI catalysts which dier strongly
in Brnsted acidity but are similar in Fe content and structure of Fe sites, having shown similar activity in N2O decomposition in related
work. The catalysts were prepared by exchange of Na-ZSM-5 (Si/Al ca. 14) with Fe2+ ions formed in situ by acidic dissolution of Fe
powder and by steam extraction of framework iron from Fe-silicalite or from H-[Fe]-ZSM-5 (Si/Al ca. 30). The characterization of acidic
properties by ammonia TPD and by IR of adsorbed pyridine at dierent temperatures revealed marked dierences in acidity between
exchanged and steam-activated samples, the latter being (almost) void of strong Brnsted sites. The structural similarity of the iron sites
was conrmed by UVVis and EPR spectroscopic results. The weakly acidic samples were inferior both in isobutane-SCR and in ammonia-SCR. With isobutane, dramatic dierences over the whole range of parameters studied imply a vital role of Brnsted acidity in the
reaction mechanism (e.g. in isobutane activation). In NH3-SCR, large reaction rates were achieved with non-acidic catalysts as well, but a
promoting eect of acidity was noted for catalysts that contain the iron in the most favorable site structure (oligomeric Fe oxo clusters).
This suggests that an acid-catalyzed step (e.g. the decomposition of NH4NO2) may be rate-limiting at low temperatures.
2007 Elsevier Inc. All rights reserved.
Keywords: DeNOx; Fe-MFI; Isobutane; Ammonia; UVVis spectroscopy; EPR spectroscopy; Active sites; Acidity; FTIR of pyridine; NH3-TPD

1. Introduction
Fe-ZSM-5 catalysts receive at present much attention
due to their remarkable performance in reactions involving
the activation of nitrogen oxides. In the attempts to elucidate the active Fe sites for the various reactions, the complexity of the Fe site structures coexisting in Fe-ZSM-5 of
dierent preparations has led to many conicting views.
Thus, the selective reduction of NO by isobutane (isobutane-SCR) was suggested to be the catalyzed by binuclear

Corresponding author. Tel.: +49 234 322 2088; fax: +49 234 321 4115.
E-mail address: w.gruenert@techem.ruhr-uni-bochum.de (W. Grunert).
1387-1811/$ - see front matter 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.micromeso.2007.07.019

Fe oxo complexes formed upon washing and calcination


of ZSM-5 loaded with iron by CVD of FeCl3 into the Hform [1,2]. Other authors considered only isolated Fe ions
to support hydrocarbon-SCR over Fe-ZSM-5 [3].
Recently, some of us found that both isolated and oligomeric Fe oxo species can catalyze this reaction (and
NH3-SCR as well) while oligomeric sites attack the reductant at higher temperatures limiting thus the selective temperature window (with isobutane more severely than with
NH3) [4,5]. Both isolated and oligomeric Fe oxo species
have been considered to be active in direct N2O decomposition [6], according to [7,8], however, with a clear preference for oligomeric species. The reduction of NO by CO
or by hydrocarbons proceeds both on isolated and clustered sites [8,9].

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

For several of these reactions, it is obvious that the zeolite Brnsted acidity is not a critical requirement. Thus,
N2O decomposition and reduction by CO were both found
to proceed swiftly over steamed Fe-silicalite which contains
Al in the ppm range at maximum [7,8]. With the SCR of
NO, there is less condence about this question. While
hydrocarbon-SCR is known to require Brnsted acidity
over many catalytic systems (e.g. ZSM-5 modied with
Pd, Ga, In and Ce [1014]), the most plausible reaction
mechanism proposed for isobutane-SCR over Fe-ZSM-5,
which traces NH3-SCR as the nal step of the hydrocarbon-SCR reaction sequence, does not explicitly refer to
acid-catalyzed steps [15]. In NH3-SCR, acidity is generally
considered to be favorable because acidic surfaces can
adsorb NH3 in large quantities and thus provide a reservoir
of the reductant in the vicinity of the reduction site. In the
mechanism suggested by Topsoe et al. for the V2O5/TiO2
catalyst [16], an acidic VOH group is also part of the
active site, but there are also proposals that do not involve
the activation of NH3 by a Brnsted center [17]. In an
investigation of NH3-SCR over VO2+-exchanged ZSM-5,
some of us came to the conclusion that the active site is a
single VO2+ ion, which does not need the cooperation of
Brnsted sites [18]. Very recently, Li et al. [19] identied
an acid-catalyzed step which is probably involved in
NH3-SCR: They found that the decomposition of the likely
reaction intermediate NH4NO2 is accelerated by zeolite
Brnsted sites. This reaction occurs already at rather low
temperatures. Its potential role as rate-limiting step will
depend on the rates of other steps in the reaction sequence.
The assessment of the actual role of acidic sites in SCR
reactions is not straightforward because a reliable conclusion can be only derived if the redox sites in the bifunctional catalysts used for the comparison are identical or
close enough in structure to exclude their involvement in
the catalytic eects observed. The most direct approaches,
which involve poisoning of the acidic sites by bases or by
back-exchange with alkali ions, may also aect the redox
function of the catalyst. In former work, we obtained some
indication that the relevance of Brnsted sites may be different for isobutane-SCR and NH3-SCR from the observation that a catalyst prepared via chemical vapor deposition
(CVD) of FeCl3 onto a high-silica H-ZSM-5 (Si/Al  40)
was active in the latter but very poor in the former reaction
[19]. However, it was also found that this preparation route
results in a multitude of Fe sites (from isolated Fe ions of
dierent coordination via oligomeric sites of dierent
nuclearity to large, disordered aggregates) and that the distribution of these species diers considerably between
materials prepared from parent ZSM-5 samples of dierent
Si/Al ratios [20]. Hence, the dierences in activity may have
originated from dierences in the Fe site structure, which
might be more relevant for isobutane-SCR than for NH3SCR.
In the present paper, we compare Fe-ZSM-5 materials
prepared via dierent routes, which are similar in their iron
content and their catalytic activity in the direct N2O

125

decomposition [21] but should, according to their Si/Al


ratios and the preparation routes employed (exchange of
Fe2+ ions into ZSM-5 vs. steam-activation of isomorphously substituted Fe-MFI) dier markedly in their acidity. UVVis and EPR spectroscopic data from these
catalysts [5,9] show that among them are examples with
close proximity in the structure of the Fe sites. In this
paper, we describe the acidic properties of the catalysts
on the basis of results from temperature-programmed
desorption of ammonia and IR-spectroscopy of adsorbed
pyridine and relate the drastic dierences observed to those
found in the catalytic behavior in the SCR reactions.
2. Experimental
2.1. Materials
We will compare Fe-ZSM-5 catalysts made by exchange
of Na-ZSM-5 (Chemiewerk Bad Kostritz (Germany), Si/
Al  14) with Fe2+ ions produced in situ by acidic dissolution of iron powder (improved liquid ion exchange ILIE,
rst proposed in Ref. [22]), with Fe-MFI (Fe-ZSM-5 and
Fe-silicalite) in which iron initially located in the zeolite
framework was extracted by high-temperature steaming.
For the ion exchange, a slurry of Na-ZSM-5 and Fe powder was allowed to react with dilute HCl (initial pH 1)
under protective atmosphere for several days [5]. The catalysts, which contained 0.21.2 wt% Fe depending on the
amount of Fe powder used in the synthesis, were then
washed, dried and calcined in air at 873 K for 2 h. Along
the manuscript, these samples are labeled as Fe-Z(m . n),
where m . n is the Fe content. Details on the preparation
of steam-activated Fe-MFI zeolites have been described
elsewhere [7,23]. The isomorphously substituted zeolites,
with FeAlSi and FeSi frameworks, were calcined and
activated in steam (300 mbar H2O in 30 ml (STP) min1
of N2) at ambient pressure and 873 K for 5 h, yielding sFe-ZSM-5 (Si/Al = 31 and 0.67 wt% Fe) and s-Fe-silicalite
(Si/Al = 1 and 0.68 wt% Fe). For comparison, acidity and
activity data for a catalyst made by the well-known CVD
technique (CVD of FeCl3 onto H-ZSM-5 [1]) with a defective H-ZSM-5 of low Al content (Si/Al  40; Fe content
2.6 wt%) will be given as well. The structural and catalytic
properties of this material have been reported in Refs.
[20,24] where it was labeled Fe-Z(B) and B(CVD, W1).
The former label will be kept in the present paper. The
information given here about the samples studied is summarized in Table 1.
2.2. Characterization
The temperature-programmed desorption of ammonia
was performed in two of the participating groups with
somewhat dierent procedures. The parent H-ZSM-5 and
Fe-Z(B) were measured with both protocols to allow a
comparison of the results. Procedure A involved a Zeton/
Altamira AMI-1 instrument equipped with a Thermal

126

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

Table 1
Overview over catalyst samples studied
Catalyst

Si/Al

Preparation

Fe
(wt%)

Fe-Z(0.3)
Fe-Z(0.6)
Fe-Z(1.2)
s-Fe-silicalite
s-Fe-ZSM-5
Fe-Z(B)

14
14
14
no Al
31
40a

Improved liquid ion exchange


Improved liquid ion exchange
Improved liquid ion exchange
Steaming of Fe-silicalite
Steaming of H-[Fe, Al]-ZSM-5
Gas-phas exchange of H-ZSM-5
with FeCl3

0.3
0.6
1.2
0.68
0.67
2.6

a
Highly defective material, most Al extra-framework as shown by an IR
study [24].

Conductivity Detector (TCD). After mounting the previously calcined samples, they were treated in owing He
at 373 K, subsequently ammonia (3 vol% in He,
90 ml min1) was adsorbed at 373 K for 30 min. After
physisorbed NH3 had been desorbed at the same temperature into owing He for 90 min, the desorption ramp was
started (5 K min1 between 373 and 873 K, with a 90 min
isothermal period at the nal temperature). The ammonia
desorbed was trapped in dilute 0.05 N H2SO4 and subsequently titrated with 0.05 N NaOH. Procedure B was performed on a Micromeritics Autochem II 2920 equipped
with a TCD as well. The calcined sample (35 mg) was pretreated at 723 K in He (30 ml min1) for 1 h. Afterwards, a
mixture of 5 vol% NH3 in He (40 ml min1) was adsorbed
at 473 K for 15 min. Subsequently a ow of He
(30 ml min1) was passed through the reactor during
30 min to remove ammonia weakly adsorbed on the zeolite.
This procedure was repeated three times. Desorption of
NH3 was monitored in the range of 473873 K at
10 K min1.
The surface acidity of the Fe-zeolites was studied by
FTIR spectroscopy of adsorbed pyridine. Spectra were
recorded using a Bruker IFS 66 spectrometer equipped
with a heatable and evacuable reaction cell with CaF2 windows, which is connected to gas-dosing and evacuation systems. The zeolite powder was pressed into self-supporting
wafers with a diameter of 20 mm and a weight of 50 mg.
Prior to pyridine adsorption, the samples were pretreated
in owing air at 673 K for 1 h followed by cooling to
373 K. Then, pyridine was adsorbed at this temperature
for 1 h by bubbling the Ar ow through a pyridine-containing saturator. Physisorbed pyridine was evacuated during
5 min at 373 K and infrared spectra were recorded at dierent temperatures in the range of 373673 K with 2 cm1
resolution and 100 scans. The relative concentration of
Brnsted and Lewis acid sites was determined from the
area of the absorption bands at 1545 cm1 and
1450 cm1, respectively.
The UVVis and EPR data summarized here were previously published in [5,9] where experimental details can
be found as well. The UVVis-spectra were measured with
a Varian Cary 400 spectrometer equipped with a Harrick
DRS accessory (Harrick), and deconvoluted into sub-

bands using the GRAMS/32 code (Galactic). The X-band


EPR spectra were recorded on a Bruker ELEXSYS
500-10/12 spectrometer (microwave power = 6.3 mW,
modulation frequency = 100 kHz, modulation amplitude = 0.5 mT). The magnetic eld was measured in reference to the DPPH (2,2-diphenyl-1-picrylhydrazyl-hydrate)
standard.
2.3. Catalytic tests
The catalytic studies have been performed as described
in [5]. In brief, the SCR reactions have been measured in
catalytic microow reactors using mixtures of 1000 ppm
NO, 1000 ppm reductant (isobutane or NH3) and 2 vol%
O2 in He, with a gas-hourly space velocity (GHSV) of
30,000 h1 and at temperatures between 823 and 523 K
in the case of isobutane-SCR, and with a GHSV of
750,000 h1 and at temperatures between 873 and 423 K
(temperatures decreasing during the measurement sequence
in both cases) with the NH3 reductant. Prior to the runs,
the catalysts were heated in owing He at the highest reaction temperature for 30 min. The composition of the euent gas was analyzed using calibrated mass spectrometry in
the case of isobutane-SCR and a combination of photometric devices for detection of NO, NO2, and NH3 in the
case of NH3-SCR (Binos 1000, Rosemount, for NO and
NO2, Binos 1, Leybold, for NH3). In both cases, gas-chromatography was additionally used to analyze the quantity
of nitrogen released to trace possible formation of NO2 or
N2O in signicant amounts by setting up a nitrogen balance. This balance typically rendered values between 95
% and 105 %.
3. Results
3.1. Acidic properties
In Fig. 1, NH3 proles recorded during the TPD of
ammonia from H-ZSM-5 and dierent Fe-ZSM-5 samples
are compared. The amounts of NH3 desorbed from the
samples are given in Table 2. The comparison is somewhat
complicated by the parallel application of two dierent
experimental procedures (caused by a limited supply of
some materials). With procedure A (adsorption at 373 K,
Fig. 1a), two peaks arising from weakly and strongly acidic
sites are seen in all runs while the rst one was absent when
procedure B (adsorption at 473 K, Fig. 1b) was applied. In
the latter case, the desorption peaks were somewhat shifted
towards lower temperatures, and the NH3 content in the
euent raised immediately after the start of the temperature ramp, which caused an asymmetry of the desorption
peak of H-ZSM-5 and a second, earlier maximum with
Fe-Z(B) (Fig. 1b). The ammonia desorbed at the start of
the ramp arises obviously from sites that can be populated
at 473 K, because after adsorption at 373 K, the rst
desorption maximum appears at T P 473 K. The downshift of the peak temperature is unexpected and its reason

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133


T,K
400

600

800

isothermal
1000

H-ZSM-5

TC1200D signal, a. u.

Fe-Z(0.6)

Fe-Z(B)

Fe-Z(1.2)

H-ZSM-5

Fe-Z(B)

s-Fe-ZSM-5
s-Fe-silicalite
400

600

800
T,K

Fig. 1. Ammonia TPD from H-ZSM-5 and from Fe-MFI of dierent


preparation. (a) Fe-Z(0.6) and Fe-Z(1.2) prepared by improved liquid ion
exchange, procedure A (see Section 2), (b) Fe-MFI by steam extraction of
framework Fe from Fe-silicalite and H-[Fe]-ZSM-5 (procedure B). For
comparison, the TPD proles obtained with the parent H-ZSM-5 and with
Fe-Z(B) using procedure A and B are given in panels a and b, respectively,
as well.

Table 2
NH3 adsorption capacity of zeolites studied (see Fig. 1)
Catalyst

H-ZSM-5 (Si/Al = 14)


Fe-Z(0.6)
Fe-Z(1.2)
s-Fe-silicalite
s-Fe-ZSM-5
Fe-Z(B)
a

NH3 desorbed (mmol g1) after adsorption at


373 K (Aa, Fig. 1a)

473 K (Ba, Fig. 1b)

1.51
1.43
1.47

0.32

0.75

0.06
0.14
0.21

127

In the samples made via exchange (ILIE; Fig. 1a), the


intensity of the high-temperature signal decreases markedly
with increased Fe content as expected. At the same time, a
shoulder emerges at the high-temperature side of the lowtemperature signal, presumably caused by a weak Lewis
acidity of the Fe ions. The total amount of ammonia desorbed (Table 2) does not decrease signicantly with
increasing Fe content. Apparently, the loss of intensity at
high-temperatures is outweighed by increases due to sites
of lower strength. Fig. 1b and Table 2 conrm that steam
extraction of framework Fe yields materials of inferior
acidity. This applies in particular to s-Fe-silicalite where
aluminium which could give rise to strongly acidic sites,
is completely absent. s-Fe-ZSM-5 exceeds the latter somewhat in number and strength of sites. Fe-Z(B) is a material
with very low acidity as well. The total ammonia amount
desorbed from it is only slightly larger than in the case of
s-Fe-ZSM-5, but it emerges mostly from weak sites.
IR spectra of pyridine adsorbed on H-ZSM-5 and two
Fe-ZSM-5 catalysts made from it via ILIE are given in
Fig. 2. Spectra measured after desorption at 523 K are
given to avoid the distortion of the results by the signal
of hydrogen-bonded pyridine, which is observed in spectra
measured at lower temperatures (Fig. 3, bands at 1445 and
1596 cm1 [2527]). At 523 K, the well-known pyridine
ring-mode bands indicative of Brnsted sites (PyH+
1542 cm1, shoulder at 1635 cm1) and of Lewis sites
(L-Py, 1450 cm1, 16101618 cm1) are well visible. As
expected, the spectrum of H-ZSM-5 is dominated by the
signal of PyH+ although some Lewis sites resulting from
Al3+ are available as well. Upon introduction of iron, the
center of the L-Py band shifts to lower wavenumbers,
and the ratio between Brnsted and Lewis sites (PyH+/
L-Py) changes dramatically. With Fe-Z(0.6), this arises
not so much from a disappearance of Brnsted acidity but
predominantly from a strong enhancement of the L-Py
band. With Fe-Z(1.2), the decrease in Brnsted acidity is
more obvious while the intensity of the L-Py band, which
is now broader due to a shoulder at 1442 cm1, decreases
slightly.

See Section 2.

is unclear, therefore peak temperatures will not be discussed here. According to Table 2, the total NH3 amounts
desorbed after procedures B and A are at a ratio of 1:2 with
H-ZSM-5, and 1:1.5 with Fe-Z(B), the latter reecting the
larger relative weight of weakly bound ammonia on this
material. Thus, apart from the above mentioned discrepancy in temperatures, the data from the two methods are
well consistent.

Fig. 2. FTIR spectra of pyridine adsorbed at 523 K. Background spectra


of the air pre-treated samples were subtracted.

128

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

Fig. 3. FTIR spectra of pyridine adsorbed at dierent temperatures.


Subtraction of background spectra was not applied.

The position of the bands around 1600 cm1 can be considered as a measure of the Lewis acid strength of the surface sites [25]. Bands at higher wavenumbers (1618 cm1)
result from strong Lewis sites whereas bands at lower
wavenumbers (1610 cm1) indicate medium strong sites.
Comparing the ratio of the bands at 1618/1610 cm1 it
can be seen that the intensity of the 1618 cm1 band diminishes with increasing Fe content. This may indicate that
stronger Al3+ Lewis sites are covered by FeOx species,
while Fe3+ Lewis sites of intermediate strength will be
created.

In Fig. 3, the intensity changes of the pyridine bands


during thermal desorption can be seen. The rapid decay
of the 1445 cm1 band on all three samples is obvious,
which is mainly due to the desorption of hydrogen-bonded
pyridine. Furthermore, pyridine adsorbed on Brnsted and
Lewis sites is still observed at 673 K as indicated by the
bands at 1545 cm1 and 1450 cm1, respectively. This also
explains the broad high-temperature peaks in the NH3TPD proles that obviously comprise the desorption of
NH3 from both strong Brnsted as well as Lewis sites.
The trend of band intensities with temperature is summarized in Fig. 4. With the Brnsted sites, the expected
intermediate position of Fe-Z(0.6) between H-ZSM-5 and
Fe-Z(1.2) is seen only after desorption at rather high-temperatures. The excess intensity at low temperature may
indicate additional Brnsted sites of reduced strength
(Fig. 4a). In Fig. 4b, the attention should be focused on
the high-temperature region because at low temperatures
the information about the Lewis sites is obscured by the
band of hydrogen-bonded pyridine. It can be seen that in
the Fe-exchanged samples the Lewis acidic sites are more
abundant than those in H-ZSM-5. The strong decay in
intensity even after desorption of H-bonded pyridine suggests a lower strength of the Fe-derived Lewis sites as indicated by the wavenumber shift discussed above. On the
other hand, even after desorption at 673 K, Fe-Z(0.6) holds
more Lewis-bound pyridine than H-ZSM-5. This may suggest that there is also a minority of strong Fe-derived Lewis
sites. Upon increase of the Fe content to 1.2 wt%, the
intensity of the L-Py signal decreases, obviously caused
by the abundance of more clustered species in Fe-Z(1.2).

Fig. 4. Change of the FTIR band area for pyridium ions (PyH+) at 1545 cm1 (a) and for pyridine coordinated to Lewis-acid sites (L-Py) at 1445 cm1 (b)
with desorption temperature.

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

3.2. Structure of iron species


UVVis and EPR spectra that provide relevant information about the structure of the Fe sites in selected catalysts
are presented in Figs. 5 and 6 (see also [5,9]). The focus will
be on the sample pairs Fe-Z(0.3)/s-Fe-silicalite and FeZ(0.3)/s-Fe-ZSM-5. In the UVVis spectra, all samples
exhibit major bands below 300 nm. While the spectra of
s-Fe-silicalite and of Fe-Z(0.3) are almost completely conned to this region, the remaining spectra extend to higher
wavelengths. It has been reported earlier that signals below
300 nm arise from Fe3+
O charge-transfer bands of isolated ions in tetrahedral and octahedral coordination
(220 nm and 285 nm, (t1 ! t2/t1 ! e transitions unresolved [20,28]). Signals between 300 and 400 nm may be
assigned to oligomeric clusters, while bands at k
> 400 nm indicate the presence of large Fe oxide aggregates
[20]. By deconvoluting the spectra into sub-bands as shown
in Fig. 5 (for numerical results see (see Table 3), it can be
seen that in Fe-Z(0.3) and in s-Fe-silicalite more than
90% of the signal intensity arises from isolated Fe ions.
In the two remaining samples, other species have considerable weight, and in general the degree of iron aggregation
in s-Fe-ZSM-5 is between those in Fe-Z(0.6) and FeZ(1.2). From the viewpoint of UVVis spectroscopy, the
same species occur in all catalysts. The most signicant difference is a predominance of tetrahedrally coordinated isolated Fe species in s-Fe-silicalite, which is not found in its

1.0

counterpart with similar degree of clustering Fe-Z(0.3).


Among the remaining catalysts, there are just graduations
in the degree of clustering, but the spectra are on the whole
very alike.
In Fig. 6a, EPR spectra of the Fe-ZSM-5 catalysts (in
hydrated state) recorded at 77 K and 298 K are compared.
In the spectra, several signals can be discerned, which have
been discussed in our previous papers [9,20]. The signals at
g 0  4.3 and g 0  6, which exhibit Curie-type behavior, are
both attributed to isolated Fe3+ ions, in tetrahedral coordination (intra- or extra-framework location) [2934] and
with higher coordination numbers [35,36], respectively. In
samples Fe-Z(0.3) and s-Fe-silicalite, for which UVVis
data reveal the presence of only a negligible amount of
clusters, there is another signal with Curie-type behavior
at g 0  2, which is assigned to isolated Fe ions in a highly
symmetric environment [5,9,20,31,37]. In s-Fe-ZSM-5,
Fe-Z(0.6), and Fe-Z(1.2), which do contain clusters of different size and concentration, this signal is hardly seen
since it is superimposed by a broader signal at g 0  2, which
does not show Curie-type behavior and is attributed to
clustered species with anti-ferromagnetic interactions
between the Fe ions [5]. Above the Neel temperature, antiferromagnetic ordering collapses, the clustered Fe sites
become paramagnetic and the EPR signal intensity
increases. This is evident by comparing the relative EPR
intensities of Fe-Z(0.3), Fe-Z(0.6) and Fe-Z(1.2) measured
at 77 and 298 K (Fig. 6a) with those measured in air ow at

aggregated

mononuclear

oligomeric

1.0

aggregated
oligomeric

Fe-Z(0.6)
Normalized F(R)

Normalized F(R)

Fe-Z(0.3)

0.5

0.5

Fe-Z(1.2)
Fe-Z(B)

0.0

0.0

1.0

1.0

s-Fe-ZSM-5
Normalized F(R)

Normalized F(R)

s-Fe-silicalite

0.5

0.0
200

300

400

500

Wavelength, nm

600

F(R)

mononuclear

129

700

0.5

0.0
200

300

400

500

600

700

Wavelength, nm

Fig. 5. UVVis spectra of Fe-MFI catalysts used in the present investigation, after calcination in air at 823 K (see also [5,9]).

130

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133


g' = 4.3

g' = 4.3
g' = 6

g' = 6

g' = 2

g' = 2
Fe-Z(0.3)

Fe-Z(B)

Fe-Z(0.6)
77 K
298 K
s-Fe-silicalite
Fe-Z(1.2)
s-Fe-ZSM-5

2000

4000

6000

2000

B, G

4000

6000

B, G
Fe-Z(0.3)
Fe-Z(0.6)
Fe-Z(1.2)

g' = 4.3
g' = 6

2000

4000

6000

B, G

Fig. 6. EPR spectra of Fe-MFI catalysts used in the present investigation (see also [5,9,20]): (a) comparison of spectra recorded at room temperature and
at 77 K, (Fe-Z(B) only at room temperature, not to scale, cf. [20]) and (b) spectra of samples prepared by improved liquid ion exchange measured at 623 K
in air ow.

Table 3
Quantication of the UVVis spectra of Fe-ZSM-5 catalysts in Fig. 5
Catalyst

I1a (area%)

I2b (area%)

I3c (area%)

Fe-Z(0.3)
Fe-Z(0.6)
Fe-Z(1.2)
s-Fe-silicalite
s-Fe-ZSM-5
Fe-Z(B)

95.7
76.2
48.6
93.0
58.7
27

4.3
18.5
23.4
7.0
26.2
38

0
5.3
28.0
0
15.1
35

I1 k < 300 nm, I2 300 < k < 400 nm, I3 k > 400 nm.
a
Isolated Fe3+ in tetrahedral and higher coordination.
b
Oligomeric Fe3
x Oy clusters.
c
Fe2O3 nanoparticles.

623 K (Fig. 6b). The temperature dependence of the g 0  2


signal in samples s-Fe-ZSM-5, Fe-Z(0.6), and Fe-Z(1.2) has
to be dierentiated from the broad g 0  2 signal in s-Fe-silicalite, which decreases in intensity at room temperature,
i.e. exhibits Curie-type behavior. The latter broad feature
arises probably from weak dipolar interactions between
isolated iron sites in s-Fe-silicalite [9]. In summary, it can
be stated in the light of the EPR results that practically
the same iron species occur in all catalysts although some
of them are almost void of FexOy clusters. Again, there
are some special features with s-Fe-silicalite (the weak magnetic interactions between isolated sites, which indicates a
relatively small average distance between very well distributed Fe sites), but beyond that qualitative dierences
between the Fe site structures cannot be discerned.

3.3. DeNOx performance


NO conversions in the SCR with isobutane and with
NH3 are presented in Fig. 7 (data for exchanged samples
cited from [5]). Fe-MFI catalysts prepared by extraction
of Fe from framework positions, which are regarded as
are very promising in N2O abatement reactions [38,39],
are very poor in NO abatement. In isobutane-SCR where
the Fe-ZSM-5 made by ILIE achieve peak conversions
between 70% and 86%, s-Fe-silicalite converts <10% below
700 K, and achieves 20% conversion only above 800 K
(Fig. 7a). The s-Fe-ZSM-5 catalyst exhibits an NO conversion maximum as does the vast majority of SCR catalysts,
but the peak NO conversion is just above 20%. These catalysts behave similar as the Fe-Z(B) sample, which is given
for comparison. As expected, the isobutane conversion
trends dier strongly as well, with steeply increasing curves
found for the exchanged catalysts but gradually increasing
curves for the steamed Fe-MFI samples (Fig. 7c). While
with the former the isobutane conversion is clearly lower
than the NO conversion below the maximum, such high
reductant utilization does not occur with the latter.
In NH3-SCR, the relation between the two catalyst
types is somewhat dierent (Fig. 7b). The steam-activated
iron-containing zeolites are inferior here as well, but the
activities of s-Fe-ZSM-5 and Fe-Z(0.3) are quite similar.
s-Fe-silicalite is the poorest catalyst also for NH3-SCR,
but while it is clearly inferior to s-Fe-ZSM-5 in isobu-

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

a 100

b100
i-C4H10- SCR
NO conversion, %

NH3- SCR

NO conversion, %
0
500

600

700

Fe-Z(0.3)

Fe-Z(0.6)

60
40
20

s-Fe-ZSM-5

s-Fe-silicalite

600

800

Fe-Z(1.2)

Fe-Z(B)

d 100

80

80
NH3 conversion

Isobutane conversion, %

80

0
400

800

c 100

60
40
20
0
500

131

60
40
20

600

T,K

700

800

0
400

600

800
T,K

Fig. 7. Catalytic behavior of FeMFI samples in the selective catalytic reduction of NO with isobutane (a, c 1000 ppm NO, 1000 ppm isobutane, 2 vol%
O2, 30,000 h1) and with NH3 (b, d 1000 ppm NO, 1000 ppm NH3, 2 vol% O2, 750,000 h1).

tane-SCR, both catalysts give identical NO conversions


with ammonia up to 700 K. A remarkable point with
s-Fe-silicalite is its high NH3 oxidation activity. While the
NO and the NH3 conversion curves are practically identical
for all remaining catalysts, unselective oxidation of the
ammonia becomes predominant over s-Fe-silicalite above
823 K. Another dierence to isobutane-SCR is the relatively strong performance of the Fe-Z(B), which, although
not being competitive with the better ILIE catalysts,
achieved sizable NO conversions.
4. Discussion
The relation between acidity and SCR activity will be
discussed on the background of structural information
about the Fe sites present that can be derived from Figs.
5 and 6 (for details see [5,9]). For the samples prepared
via exchange with Fe2+ ions (ILIE), the UVVis spectra
(Fig. 5, Table 1) show that the iron is largely isolated at
low Fe content (>95% at 0.3 wt%). In Fe-Z(0.6), still
75% of the iron is present in isolated species, ca. 20% in
oligomeric clusters, and only a minor amount has further
aggregated into larger (disordered) particles. In Fe-Z(1.2),
the ratio between the Fe atoms in these site types is approximately 50:25:25, i.e. the degree of aggregation is already
signicant here. The Fe speciation in Fe-Z(0.3) is similar
to that in s-Fe-silicalite regarding the high degree of site
isolation (Table 1). There are, however, dierences in the
coordination of iron, which is predominantly tetrahedral

in the latter and both tetrahedral and octahedral in the former (Fig. 5). The analogy in the Fe speciation is closer
between Fe-Z(0.6) and s-Fe-ZSM-5, where similar sites
are present in comparable amounts though with a stronger
aggregation degree in the latter.
The study of the acidity of Fe-MFI SCR catalysts has
conrmed the expected trends, but has also revealed some
interesting details. In the NH3-TPD proles, the introduction of Fe species by ILIE has lead to a decrease of the
high-temperature signal due to zeolite Brnsted sites,
accompanied by the appearance of a shoulder around
523 K, which apparently arises from NH3 bound to Fe ions
(Fig. 1a). It should be noted that the intensity loss of the
high-temperature signal is signicant already at 0.6 wt%
Fe. As expected, the acidity of the samples prepared via
steam extraction of framework Fe or by CVD of FeCl3 into
a high-silica Fe-ZSM-5 is inferior (Fig. 1b). The acidic sites
detected are mostly weak, however, a certain, very low
amount of stronger acidic sites may be discerned in s-FeZSM-5 and in Fe-Z(B).
The pyridine adsorption study (Figs. 24) conrms that
the Fe ions have Lewis-acid character: upon introduction
of 0.6 wt% Fe into H-ZSM-5, the L-Py signal is much
enhanced and shifted from 1455 to 1450 cm1 (Fig. 2). Surprisingly, however, and in disagreement with the NH3 TPD
prole, the intensity of the Brnsted band is almost the
same in Fe-Z(0.6) as in H-ZSM-5 and decreases only at
1.2 wt% Fe. This trend is conrmed in the temperature
dependence of the 1540 cm1 band (Fig. 4a), where the

132

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

initially stronger intensity loss of Fe-Z(0.6) indicates the


presence of a less acidic Brnsted site, which is absent (or
not abundant) in Fe-Z(1.2). The assignment of this site is
not straightforward. We attribute it tentatively to a
Fe3+(OH) group on a site that is abundant in Fe-Z(0.6)
but not in Fe-Z(1.2). According to the structural data cited
here (Figs. 5 and 6, Table 1, for more details see [5,40]), FeZ(1.2) contains both isolated and oligomeric sites in larger
(absolute) abundance than Fe-Z(0.6). It has been found,
however, that some of the isolated sites (those characterized by g values of 4.3 and 6 in EPR) are rare in FeZ(1.2), probably due to their participation in the clustering
that occurs at higher Fe content and higher temperatures
[40] (Fig. 6b). OH groups associated with these Fe species
might be candidates for the Brnsted sites of moderate
acidity that give rise to the extra intensity in Fe-Z(0.6).
Together with the structural data discussed above these
results provide a good basis for the interpretation of the
catalytic data. In isobutane-SCR (Fig. 7a and c), a key
involvement of Brnsted sites is very likely. This is not so
much obvious from the comparison between the non-acidic
s-Fe-silicalite and the acidic Fe-Z(0.3) because the dierence could also originate from the dierence in the coordination of the predominantly isolated Fe sites. However, a
large advantage of the acidic catalyst can be also seen in
the comparison of s-Fe-ZSM-5 and Fe-Z(0.6) which contain Fe sites of similar structure. The poor performance
of Fe-Z(B) supports the conclusion. It appears that the
Brnsted sites are involved already in the activation of
the isobutane because even the isobutane conversion is largely suppressed over the non-acidic samples. Given the
important role of hydrocarbon deposits in isobutaneSCR over Fe-ZSM-5 [41], this is well in accordance with
our earlier observation from an operando-DRIFTS study,
according to which the interaction of isobutane-SCR feed
with H-ZSM-5 leads to very similar hydrocarbon deposits
as the interaction with Fe-ZSM-5 [42]. As these adsorbates
are involved in the reduction of NOx activated on nearby
Fe sites their absence should strongly impede the SCR
reaction. A rate limiting role of acidity at a later stage
(e.g. in NH4NO2 decomposition, which according to
[15,19] is part of the reaction sequence) is less likely because
even if isobutane-SCR proceeds via a concluding NH3SCR step, the rate of the latter is much larger than the
observed rates of the former.
With NH3 as the reductant, an inuence of the Brnsted
sites is less obvious (Fig. 7b and d). Still, the non-acidic
samples s-Fe-silicalite and s-Fe-ZSM-5 exhibit the poorest
performance. However, these two catalysts achieve identical NO conversions over quite a broad temperature range
despite signicant dierences in acidity (Fig. 1b). Moreover, while the NO conversion levels o with s-Fe-silicalite,
s-Fe-ZSM-5 performs quite similar as the strongly acidic
Fe-Z(0.3). The dierence might well be accounted for by
a better accessibility of the Fe species in the latter sample.
Finally, the remarkable performance of the non-acidic FeZ(B) in NH3-SCR (as opposed to isobutane-SCR, cf.

Fig. 7a) shows that the former reaction is possible also


materials decient in Brnsted sites.
However, the very high activity of Fe-Z(0.6) and in particular of Fe-Z(1.2) is not fully compatible with this picture. It has been argued that clustered sites might be
more eective in NH3-SCR than isolated sites as long as
the dispersion remains high [5], but a large dierence in
activity is also seen between the strongly acidic Fe-Z(0.6)
and the weakly acidic s-Fe-ZSM-5 (Figs. 7b and d), which
has a similar Fe content and an only slightly more clustered
Fe phase. This dierence, which is illustrated by a marked
decrease of the temperatures for 50 % conversion (827 K
for s-Fe-ZSM-5, 653 K for Fe-Z(0.6)) is probably too large
to be explained by increased local NH3 concentrations created by neighboring Brnsted sites, rather, it suggests that
an additional reaction channel is opened by the presence of
zeolite protons. This channel might be the acid-catalyzed
decomposition of intermediate NH4NO2 as recently proposed by Li et al. [19]. It is known that this step proceeds
also uncatalyzed, and it might be rapid enough at high temperatures to allow for large reaction rates. At low temperatures, the ammonium nitrite decomposition might become
rate-limiting, and the catalytic eect of Brnsted sites on it
would be felt in the total reaction rate. Therefore, while
considerable NH3-SCR rates can be achieved with nonacidic samples as well, the development of top de-NOx performances may require the combination of oligomeric Fe
oxo clusters with Brnsted sites.
5. Conclusions
In the present study, Fe-MFI catalysts that have a similar structure of Fe sites (as revealed by UVVis and EPR
spectroscopic data) and were found to exhibit comparable
performance in N2O decomposition in companion work
have been compared for their catalytic properties in the
SCR of NO with isobutane and with NH3. The dramatic
dierences observed have been attributed to their strongly
dierent acidity properties, which have been characterized
by TPD of ammonia and by IR of adsorbed pyridine.
Fe-ZSM-5 prepared by exchange of Na-ZSM-5 with Fe2+
ions formed in situ by acidic of Fe powder in acidic solution exhibits high acidity and high SCR activity. Fe-MFI
prepared by steam extraction of framework iron from
Fe-silicalite or from H-[Fe]-ZSM-5 is weakly acidic and
poor in activity. The dierences are dramatic for isobutane-SCR over the whole range of parameters studied,
which indicates a vital role of Brnsted sites in this reaction. In NH3-SCR, large reaction rates can be achieved
with non-acidic catalysts, but acidity denitely increases
the activity provided the catalysts contain the iron in the
most favorable structure (oligomeric Fe oxo clusters). This
suggests that an acid-catalyzed step (e.g. the decomposition
of NH4NO2) may become rate-limiting at low temperatures. The data show that the highest activities in NH3SCR may be achieved with catalysts containing both
oligomeric Fe oxo clusters and Brnsted sites.

M. Schwidder et al. / Microporous and Mesoporous Materials 111 (2008) 124133

References
[1] H.-Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
[2] A.A. Battiston, J.H. Bitter, D.C. Koningsberger, J. Catal. 218 (2003)
163.
[3] Z. Sobalik, A. Vondrova, Z. Tvaruzkova, B. Wichterlova, Catal.
Today 75 (2002) 347.
[4] M. Schwidder, M. Santhosh Kumar, A. Bruckner, W. Grunert,
Chem. Commun. (2005) 805.
[5] M. Schwidder, M. Santhosh Kumar, K.V. Klementiev, A. Bruckner,
W. Grunert, J. Catal. 231 (2005) 314.
[6] G.D. Pirngruber, M. Luechinger, P.K. Roy, A. Cechetto, P.
Smirniotis, J. Catal. 224 (2004) 429.
[7] J. Perez-Ramrez, J. Catal. 227 (2004) 512.
[8] J. Perez-Ramrez, F. Kapteijn, A. Bruckner, J. Catal. 218 (2003) 234.
[9] J. Perez-Ramrez, M. Santhosh Kumar, A. Bruckner, J. Catal. 223
(2004) 13.
[10] C.E. Loughran, D.E. Resasco, Appl. Catal. B 7 (1995) 113.
[11] E. Kikuchi, K. Yogo, Catal. Today 22 (1994) 73.
[12] T. Sowade, C. Schmidt, F.-W. Schutze, H. Berndt, W. Grunert, J.
Catal. 214 (2003) 100.
[13] T. Sowade, C. Schmidt, X. Yu, F.-W. Schutze, H. Berndt, W.
Grunert, J. Catal. 225 (2004) 105.
[14] T. Liese, E. Loer, W. Grunert, J. Catal. 197 (2001) 123.
[15] H.-Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 186
(1999) 91.
[16] N.-Y. Topsoe, J.A. Dumesic, H. Topsoe, J. Catal. 151 (1995) 241.
[17] L. Lietti, G. Ramis, F. Berti, G. Toledo, D. Robba, G. Busca, P.
Forzatti, Catal. Today 42 (1998) 101.
[18] M. Wark, A. Bruckner, T. Liese, W. Grunert, J. Catal. 175 (1998) 48.
[19] M.J. Li, Y. Yeom, E. Weitz, W.M.H. Sachtler, Catal. Lett. 112 (2007)
129.
[20] M. Santhosh Kumar, M. Schwidder, W. Grunert, A. Bruckner, J.
Catal. 227 (2004) 384.
[21] J. Perez-Ramrez, M. Schwidder, M. Santhosh Kumar, A. Bruckner,
W. Grunert, in preparation.

133

[22] R.Q. Long, R.T. Yang, Catal. Lett. 74 (2001) 201.


[23] J. Perez-Ramrez, F. Kapteijn, J.C. Groen, A. Domenech, G. Mul,
J.A. Moulijn, J. Catal. 214 (2003) 33.
[24] F. Heinrich, C. Schmidt, E. Loer, M. Menzel, W. Grunert, J. Catal.
212 (2002) 157.
[25] G. Busca, Phys. Chem. Chem. Phys. 1 (1999) 723.
[26] E.P. Parry, J. Catal. 2 (1963) 371.
[27] R. Buzzoni, S. Bordiga, G. Ricchiardi, C. Lamberti, A. Zecchina, G.
Bellussi, Langmuir 12 (1996) 930.
[28] S. Bordiga, R. Buzzoni, F. Geobaldo, C. Lamberti, E. Giamello, A.
Zecchina, G. Leofanti, G. Petrini, G. Tozzola, G. Vlaic, J. Catal. 158
(1996) 486.
[29] P.N. Joshi, S.V. Awate, V.P. Shiralkar, J. Phys. Chem. 97 (1993)
9749.
[30] T. Inui, H. Nagata, T. Takeguchi, S. Iwamoto, H. Matsuda, M.
Inoue, J. Catal. 139 (1993) 482.
[31] A. Bruckner, U. Lohse, H. Mehner, Micropor. Mesopor. Mater. 20
(1998) 207.
[32] D. Goldfarb, M. Bernardo, K.G. Strohmaier, D.E.W. Vaughan, H.
Thomann, J. Am. Chem. Soc. 116 (1995) 6344.
[33] A.V. Kucherov, A.A. Slinkin, Zeolites 8 (1998) 110.
[34] B. Wichterlova, P. Jiru, React. Kinet. Catal. Lett. 13 (1980) 197.
[35] A.F. Ojo, J. Dwyer, R.V. Parish, Stud. Surf. Sci. Catal. 49 (1989) 227.
[36] P. Wenquin, Q. Shilun, K. Zhiyun, P. Shaoyi, Stud. Surf. Sci. Catal.
49 (1989) 281.
[37] G. Catana, L. Pelgrims, R.A. Schoonheydt, Zeolites 15 (1995) 475.
[38] J. Perez-Ramrez, F. Kapteijn, G. Mul, J.A. Moulijn, Appl. Catal. B
35 (2002) 227.
[39] J. Perez-Ramrez, F. Kapteijn, G. Mul, J.A. Moulijn, Chem.
Commun. (2001) 693.
[40] M. Santhosh Kumar, M. Schwidder, W. Grunert, U. Bentrup, A.
Bruckner, J. Catal. 239 (2006) 173.
[41] H.-Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, Catal. Today 54
(1999) 483.
[42] F. Heinrich, E. Loer, W. Grunert, Stud. Suf. Sci. Catal. 135 (2001)
30.

Anda mungkin juga menyukai