Environmental catalysis on iron oxide-silica aerogels: Selective oxidation of NH3 and reduction of NO by NH3

Article abstract of DOI:10.1006/jcat.2001.3475

The catalytic characteristics of mesoporous iron oxide-silica aerogels prepared by a sol-gel process combined with ensuing supercritical extraction with CO₂ was studied in the selective oxidation (SCO) of NH₃ and the SCR of NO by NH₃. NH₃ was bound to Bronsted and Lewis-type sites, the latter dominant at 300°C. A fraction of low coordinated Fe²⁺ sites were probed by NO adsorption measurements. Lewis-type sites were associated with low-coordinated iron sites. The catalytic activity of the aerogels in SCO correlated with the abundance of more strongly bound NH₃ adsorbed on Lewis sites. High selectivity to nitrogen could be reached up to 500°C, while at higher temperature NO and N₂O formation became significant. The apparent activation energy of N₂ formation was 69-94 kJ/mole, while catalysts with higher selectivity and activity exhibited lower activation energy. In SCR, selectivity to nitrogen was for all aerogels > 98% at < 460°C, and activation energies varied from 38 to 53 kJ/mol. The catalytic activity for SCR did not correlate with the population density of Lewis sites. It was proposed that SCO predominantly occurred on Lewis sites consisting of highly dispersed iron atoms of low coordination, while in SCR these sites did not play an important role.

Full text of DOI:10.1006/jcat.2001.3475

Journal of Catalysis 206, 143–154 (2002)
doi:10.1006/jcat.2001.3475, available online at http://www.idealibrary.com on
Environmental Catalysis on Iron Oxide–Silica Aerogels: Selective
Oxidation of NH and Reduction of NO by NH
3
3
P. Fabrizioli, T. B u¨ rgi, and A. Baiker1
Laboratory of Technical Chemistry, Swiss Federal Institute of Technology, ETH H o¨ nggerberg-HCI, CH-8093 Z u¨ rich, Switzerland
Received September 13, 2001; revised November 15, 2001; accepted November 19, 2001
silicalites (1), which have been reported to possess inter-
The catalytic properties of mesoporous iron oxide–silica aerogels esting catalytic potential in a variety of reactions. Indeed,
3
+
prepared by a sol–gel process combined with ensuing supercriti- the acid properties of the incorporated Fe and the re-
cal extraction with CO2 was investigated in the selective oxidation
dox properties of finely dispersed iron oxide particles, to-
(
SCO) of ammonia and the selective reduction (SCR) of NO by
gether with shape selectivity of the zeolite, are attractive
properties for catalytic application (1). Comparatively lit-
tle attention has been given to supported iron oxide (2–5)
and to iron–silicon oxide aerogels: Wang and Willey (6)
synthesized high-temperature Fe2O3–SiO2 aerogels for the
oxidation of methanol. Willey et al. prepared iron oxide–
chromia–alumina high-temperature aerogels (7), and mag-
ammonia. The main parameters changed in the aerogel prepara-
tion were the type of base used as gelation agent, the iron content,
and the calcination temperature. The aerogels differed significantly
in acidity and iron dispersion. Diffuse reflectance infrared Fourier
transform spectroscopy studies of ammonia adsorption at differ-
ent temperatures revealed that ammonia was bound to Brønsted-
◦
and Lewis-type sites, the latter being dominant at 300 C. A frac-
tion of low coordinated Fe2 sites were probed by NO adsorption nesiumoxide–ironoxidespinelaerogels(8)fortheselective
+
measurements. Lewis-type sites were found to be associated with catalytic reduction of NO by ammonia. No report deals with
low-coordinated iron sites. Catalytic tests were performed in a con- iron oxide–silica aerogels, prepared by low-temperature
◦
tinuous fixed-bed reactor in the temperature 210–550 C range and
extraction with supercritical CO2.
Aerogels can offer high flexibility in the tuning of the
textural properties and chemical composition, as has been
exemplified for the titania–silica-mixed oxides used for the
epoxidation of olefins (9). A problem inherent to the sol–
gel-derived materials is, however, that mixing of the con-
stituents on the molecular scale is often not perfect, and
real “single-site” catalysts are difficult to achieve. Never-
at ambient pressure. The catalytic activity of the aerogels in SCO
correlated with the abundance of more strongly bound ammonia
adsorbed on Lewis sites (low coordinated iron). High selectivity to
nitrogen (97%) could be reached up to 500 C, whereas at higher
temperature the formation of N2O and NO became significant.
The apparent activation energy of N2 formation ranged from 69
to 94 kJ/mol, whereby catalysts with higher selectivity and activity
showed lower activation energy. In SCR, selectivity to nitrogen was
for all aerogels >98% at T < 460 C, and activation energies varied theless comparative studies of titania silicalites and titania–
from 38 to 53 kJ/mol. The catalytic activity for SCR did not cor- silica aerogels proved that the catalytic properties of the
relate with the population density of Lewis sites. We propose that amorphous aerogels are in some cases outperforming those
SCO predominantly occurs on Lewis sites consisting of highly dis-
persed iron atoms of low coordination, whereas in SCR these sites
◦
◦
of the crystalline materials (10). This prompted us to ex-
plore the potential of iron oxide–silica aerogels in cata-
lysis.
In a preceding study (11) we reported on the synthesis
and chemical and structural properties of iron oxide–silica
aerogels. Here we address the catalytic properties of these
materials. Two environmentally relevant reactions are the
focus of our study: the selective catalytic oxidation (SCO)
of ammonia and the selective catalytic reduction (SCR) of
do not play an important role.
ꢀc 2002 Elsevier Science (USA)
Key Words: iron oxide–silica aerogels; selective oxidation of
ammonia; selective reduction of NO; supercritical drying; Brønsted
and Lewis acidity.
1
. INTRODUCTION
Iron–silicon oxide-based materials have received consid- NO by ammonia. The latter is the most frequently used
x
erable attention as potential catalysts for reactions relevant technique to eliminate NO from oxygen-containing waste
x
to environmental and petrochemical catalysis. This is most gases of stationary combustion sources. Although various
prominently demonstrated by the family of crystalline iron catalysts were applied for this reaction (12), most attention
has been given to vanadia- on titania-based catalysts. Com-
1
paratively little is known about the catalytic performance
of iron-containing catalysts, such as iron-exchanged zeolites
To whom correspondence should be addressed. Fax: 41 1 632 11 63.
E-mail: baiker@tech.chem.ethz.ch.
143
0
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.
ꢀ

¨
FABRIZIOLI, BURGI, AND BAIKER
1
44
(
13–17), pillared clays (18–20), mixed oxides (8, 21), and infrared Fourier transform spectroscopy (DRIFTS), and
supported iron oxides (4, 5, 22, 23). electron paramagnetic resonance(EPR) spectroscopy, we
In practice, the SCO to nitrogen succeeds the SCR unit, applied temperature-programmed desorption (TPD) com-
◦
where the gas exits with a temperature of 280–480 C. SCO bined with DRIFTS for the study of ammonia and nitric
of ammonia to nitrogen also finds application in the treat- oxide adsorption.
ment of reformates for fuel-cell systems and in the deodor-
DRIFT spectra were recorded on a Perkin–Elmer 2000
ization of ammonia-containing gas (24, 25). Iron-containing FTIR instrument with a diffuse reflectance cell and a
catalysts applied in SCO include ferrisilicate and Fe- controlled environmental chamber (both Spectra-Tech)
exchanged zeolites (16, 26, 27), supported iron oxide cata- equipped with CaF2 windows. The sample was mounted
lysts (3, 28–30), and pure iron oxides (31–33). As for SCR, on a ceramic frit (Al2O3), which could be heated to ele-
iron oxide–silica aerogels are novel catalyst materials for vated temperature via a PID controller (Tecon 501). Argon
this reaction.
(Pangas, 99.999%) and synthetic air (Pangas, 20% O2 pu-
rity 99.5%, 80% N2 purity 99.995%) was passed through
a cooling trap with an isopropanol–dry ice mixture. The
DRIFT spectra in the reflection mode were transformed to
the Kubelka–Munk function. Then they were normalized
2
. EXPERIMENTAL
2
.1. Preparation and Characterization of Aerogels
The preparation and structural and chemical properties
−1
to the silica overtone vibration at 1848 cm (combination
mode of Si–O vibrations), as proposed by Vansant et al.
(
34). To visualize the species resulting from NH3 or NO
of iron oxide–silica aerogels used in this work were de-
scribed in detail elsewhere (11). In brief, aerogels were
prepared from tetramethoxysilicon(IV) or tetraethoxysili-
con(IV) and Fe(NO3)3 · 9 H2O using a sol–gel process with
different nitrogen-containing bases (N,N-diethylaniline,
trihexylamine, ammonium hydroxide, ammonium carbon-
ate) as gelation agents. The gels were dried by extraction
adsorption, the spectrum of the unloaded catalyst was sub-
stracted.
During TPD of the species adsorbed after reaction and
during temperature-programmed reduction by ammonia,
gas flow composition was analyzed online using a Balzers
quadrupole mass spectrometer QMA 112A.
Four types of measurements were carried out.
◦
of the solvent with supercritical CO2 at 40 C and 22 MPa
and calcined in air at 600 and 900 C, respectively, be-
◦
1. NH3 adsorption and TPD on the calcined catalysts.
fore catalytic tests. The acronyms used for the gels in this
work are the same as those used in a previous study (11).
The first numeral refers to the nominal Fe2O3 content in
weight percent, based on the theoretical system Fe2O3–
SiO2. The subsequent capital letters describe the sol–gel
route applied: first the silicon precursor is given, M indi-
cates that tetramethoxysilicon was used, and E stands for
the tetraethoxysilicon precursor. The “Md” aerogels were
prepared by using a higher amount of alcohol in the hydrol-
ysis as compared to the other aerogels. Then the gelation
agent is represented by two to four letters, “NN” standing
for N,N-diethylaniline; “Tri,” for trihexylamine, “NC,” for
ammonium carbonate, “NH3,” for ammonium hydroxide;
Prior to ammonia adsorption the catalyst was heated in
a synthetic air stream (50 ml min ) to 300 C for 1 h.
The air stream was dried in the cooling trap. Ammonia
−1
◦
(
Pangas, 3600 ppm in argon, dried over a KOH-containing
◦
cartridge) was adsorbed at 50 C. Thereafter the cell was
flushedwithargonfor2.5h. Temperature-programmeddes-
orption of ammonia (NH3-TPD) was performed at a rate
of 5 C min ; spectra were taken by accumulating 50 scans
at a resolution of 4 cm every 50 C.
. NO adsorption and NO-TPD on the reduced catalysts.
◦
−1
−1
◦
2
The aerogels M/NN, M/Tri, M/NC, M/NH3, and E/NH3
were used for NO adsorption studies. The catalysts were
◦
pretreated at 425 C in pure hydrogen for 20 min in the
“
NH3d,” indicating a more diluted ammonia solution. Note
◦
in situ DRIFTS cell. Then cooled down to 50 C, flushed
that the iron content is omitted in the acronym for the gels
with a nominal composition of 10 wt% Fe2O3/90 wt% SiO2.
Aerogels with iron content from 0 to 20 wt% Fe2O3 were
prepared following the synthesis route of Md/NN. Calcina-
with argon, and 2000 ppm NO/ Ar was admitted to the cell
with a continuous flow of 50 ml/min. All gases were dried by
means of a cooling trap (CO2–isopropanol). The pressure
in the cell was set to 2 bar by using an adjustable relief valve
at the outlet of the cell. After 2 h the gas flow was changed
back to argon. The cell was then flushed for 4–5 h before
the steady-state spectra under Ar were recorded.
◦
tion of the aerogel Md/NN in air at 900 C leads to sample
Md/NN900.
2
.2. Characterization
3. TPD and TPR by ammonia. TPD of the adsorbed
In addition to the previously used characterization meth- species after SCO was performed by flushing the reactor
◦
ods (11), which included nitrogen adsorption, X-ray diffrac- with 50 ml/min argon at 50 C for 3 h. Thereafter the cata-
◦
tion (XRD), transmission electron microscopy (TEM), lyst was heated at 10 C/min, and desorbing species were
temperature programmed reduction (TPR), X-ray photo- monitored by mass spectrometry. For comparison, TPD
electron spectroscopy (XPS), UV–vis, diffuse reflectance of ammonia from the calcined and reduced M/NN were

CATALYSIS ON IRON OXIDE–SILICA AEROGELS
145
◦
also measured. The aerogel was pretreated at 550 C in syn- ing that all species, consumed or formed, were accounted
thetic air and in 10% H2/Ar (Pangas), respectively. TPR for.
of the catalysts M/NH3, E/NH3, and Md/NN after use in
◦
SCO and after pretreatment in air at 550 C was investi-
3
. RESULTS
gated in 50 ml/min of 2000 ppm NH3 in a helium (99.999%)
balance (3600 ppm NH3/He, certified ± 2%, Carbagas) at 3.1. Structural and Chemical Properties
◦
◦
a heating rate of 10 C/min in the temperature 50–550 C
range.
4
The structural and chemical properties of the FeOx /SiO2
aerogels have been reported in detail in Ref. (11). Table 1
summarizes the main textural and chemical properties of
the aerogels. Briefly, different aerogels were synthesized by
varying the nitrogen-containing base, the silicon precursor,
and the iron content in a sol–gel process. After calcination
. DRIFTS in situ catalytic studies. To investigate the
role of the acid sites in the SCO of ammonia, in situ
DRIFTS studies with catalysts M/NN and E/NH3 were
performed. The catalysts were pretreated in air at 400 C,
◦
◦
cooled to 250 C, and a background spectrum was taken.
◦
in air at 600 C all aerogels with 10 wt% nominal Fe2O3 were
After adsorption of ammonia (3600 ppm/Ar, 50 ml/min)
for 30 min, the gas flow was changed to argon, and the
catalysts were flushed for 3 h. The oxidation of the ad-
sorbed ammonia was then started by introducing 50 ml/min
.8% O2/Ar into the cell. Spectra were taken in 2-min in-
tervals during the first 10 min. Then every 10 min a spec-
3
amorphous with pore volumes ranging from 0.7 to 2.0 cm
g
−
1
2
−1
and surface areas between 300 and 636 m g . The pore
size ranged from micro- to mesopores, and the maximum of
the pore size distribution was between 20 and 50 nm. The
ironionshadameanoxidationstateofabout3+. Reduction
1
◦
to metallic iron by hydrogen occurred only at 700–1100 C,
trum was taken by accumulating 100 scans at a resolution of
−
1
due to strong interaction with the silica matrix. Electron
spin resonance (ESR) and UV–vis spectroscopy revealed
that different iron sites exist in the aerogels, ranging from
isolated tetrahedrally coordinated iron ions to iron oxide
4
cm
.
2
.3. Catalytic Tests
Catalytic tests for the SCO of NH3 and the SCR of NO clusters, depending on the sol–gel process conditions, the
by NH3 were carried out in a continuous-flow fixed-bed mi- iron content, and the thermal treatment. Tetrahedral co-
croreactor made of a 4-mm-i.d. quartz glass tube. Volumes ordinated iron was detected in all aerogels, except those
3
◦
of 0.126 cm corresponding to 56–90 mg of catalyst (par- calcined at high temperature (900 C) and with high iron
ticle size, 65–120 µm; except for M/NH3, E/NH3, E/NH3d, content (17 wt%, sample 20/Md/NN). For these samples
1
20–300 µm) were used for the measurements. The reaction the presence of some crystalline γ -Fe2O3 or Fe3O4was in-
gas mixture for SCO consisted of 2000 ppm NH3 and 2.0% dicated by XRD.
O2 in a helium balance, whereas for SCR of NO, 900 ppm
NO, 900 ppm NH3, and 1.8% O2 were mixed. The gas mix-
ture was made from single-component gases in a helium
TABLE 1
Structural Properties of Iron Oxide–Silica Mixed Oxide Aerogels a
(
7
99.999%) balance (3600 ppm NH3/He, 2000 ppm NO/He,
% O2/He certified ± 2%, Carbagas). Feed and product
Aerogel SBET (m g ) VPb (cm g ) dmaxc (nm)
−1 −1
2
3
XRDd
concentrations of NH3, N2, NO, N2O, and O2 were quan-
titatively analyzed online using a Balzers quadrupole mass
spectrometer QMA 112A.
M/NN
Md/NN
Md/NN900
310
300
150
610
342
739
335
360
480
636
625
1.10
0.72
0.68
1.80
0.12
2.60
0.72
0.97
1.56
1.73
2.05
40
23
30
45
<2
40
20
45
43
38
50
a
a
γ -Fe2O3 or Fe3O4
Conversion measurements as a function of temperature
5/Md/NN
a
were carried out at a gas-hourly space velocity, GHSV, of 20/Md/NN
γ -Fe2O3 or Fe3O4
−
1
◦
0/Md/NN
M/Tri
M/NC
M/NH3
E/NH3
—
a
a
a
a
2
4,000 h (0 C, 1 atm; standard temperature and pressure
(
(
5
STP)) after pretreatment of the catalysts in synthetic air
Pangas, 20% O2 purity 99.5%, 80% N2 purity 99.995%;
0 ml min ) at 550 C for 30 min. The pressure drop
−
1
◦
over the catalyst bed was below 0.13 bar at room tem- E/NH3d
a
perature. For differential activity measurements the space
a
Ref. (11). Note that samples with no numerals in acronyms contained
0 wt% nominal Fe2O3.
VP designates the BJH cumulative desorption pore volume of pores
−
1
velocity was gradually raised from 17,000 to 50,000 h
STP), and the temperature was adjusted to keep the am-
1
(
b
monia conversion below 20% to approximate differential
reactor conditions. Turnover frequency (TOF) calculations
were referred to the bulk iron content. For every measure-
ment a balance over all nitrogen-containing compounds
was calculated. Even for high conversions of ammonia,
in the maximum diameter range 1.7–300 nm.
c
dmax is the graphically assessed pore size maximum of the pore size
distribution derived from the adsorption branch.
d
XRD patterns observed. “a” stands for amorphous samples; aerogels
Md/NN900 and 20/Md/NN were also predominantly amorphous but con-
tained crystalline domains of γ -Fe2O3 or Fe3O4 not distinguishable by
the deviation in the balance did not exceed ± 3%, indicat- XRD.

¨
FABRIZIOLI, BURGI, AND BAIKER
1
46
3
.2. DRIFTS Studies
3280
Md/NN
M/NC
E/NH3
Ammonia adsorption. Ammonia adsorption was stud-
3282
ied to gain information about the surface acidity of the iron
◦
oxide–silica aerogels. Ammonia adsorption at 50 C, after
◦
dehydration of the samples at 300 C, resulted in strong
−1
3285
ν(NH) vibrations in the 3400- to 2400 cm region (Fig. 1).
No vibrations from the bulk catalyst interfere in this re-
gion. The total amount of adsorbed ammonia decreased
in the order Md/NN > M/NN > M/Tri > M/NC > M/NH3 >
5
0 C
˚
5
0 C
˚
∼
E/NH3d = E/NH3. For comparison a pure silica aerogel was
50˚C
investigated. The ammonium ions observed in the pres-
ence of gaseous ammonia immediately desorbed in pure
argon atmosphere. All mixed oxide aerogels showed Lewis-
bound ammonia (L in Fig. 1), with the stretching vibra-
3
00
˚C
300
˚C
3
00
˚C
3400 3200 3000 2800 2600 3400 3200 3000 2800 2600 3400 3200 3000 2800 2600
-
1
wavenumber / cm
−1
tions in the range 3381–3387 cm (νas(N–H)) and 3276–
287 cm (νs(N–H)) (35–37). The antisymmetric defor-
−1
◦
3
FIG. 2. DRIFTS spectra of ammonia adsorption at 50 and at 300 C
on the dehydrated aerogels Md/NN, M/NC, and E/NH3.
mation vibration was observed at 1609 cm−1. During TPD
the N–H stretching vibrations shifted to lower wavenum-
bers, indicating the presence of sites with different acid-
◦
ity on the catalyst. Strongly bound ammonia at 300 C was
−
1
associated with bands at 3342–3357 cm (νas(NH)) and
250–3274 cm (νs(NH)), whereas at room temperature
the bands were found at 3381–3387 and 3276–3287 cm
−
1
3
−
1
,
respectively (Fig. 2). A correlation of the stretching band
position with the acidity of the metal center was also found
by Nakamoto for amine complexes (38). The vibration at
−
1
3
180 cm was assigned to the overtone of the δas(NH)
of the stronger bound Lewis species, in Fermi resonance
with the nearby stretching vibration νs(NH) (37, 39),
−
1
whereas the band at 3210 cm was attributed to the over-
tone vibration of the weakly bound Lewis species (40),
◦
which is strongly supported by the fact that at 300 C the
latter band almost completely vanished (Fig. 2).
The aerogels possess different Brønsted acid properties.
Brønsted-bound ammonia is characterized by the broad
−
1
band at 3200–3000 cm (B, in Fig. 1). It is mainly com-
posed of the symmetric and antisymmetric stretching vibra-
+
tions of adsorbed NH . A combination band of the sym-
4
+
metric and antisymmetric deformation vibration of NH
species may arise around 3000 cm . The broad vibration
around 2800 cm is likely associated with the overtone
4
−
1
−
1
of δas(NH) (38, 41). The presence of stretching vibrations
due to ammonium ions in the 3400- to 3200-cm region,
◦
−1
FIG. 1. DRIFTS spectra of ammonia adsorption at 50 C on the dehy-
drated aerogels. L and B indicate bands mainly associated with Lewis- and
Brønsted-bound ammonia, respectively. The aerogel denotations are re-
lated to the synthesis conditions and iron content. Top: Aerogels prepared
by N,N-diethylaniline (NN) and a diluted solution of TMOS (Md), with
where Lewis-bound ammonia also appears, cannot be fully
−
1
excluded, as narrow bands above 3300 cm were assigned
+
to the free N–H groups of adsorbed NH on H-ZSM-5, in
4
varying iron content corresponding to 5 wt% (5/Md/NN), 10 wt% (Md/ agreement with ab initio calculations (42–44). However, the
NN), and 20 wt% Fe2O3 (20/Md/NN), respectively. The denotations with-
ν(NH) extinction coefficient of the free N–H is expected
out any starting numeral indicate that 10 wt% Fe2O3 was used. Md/NN900
to be lower than that of hydrogen-bonded N–H (42, 45).
◦
corresponds to sample Md/NN calcined at 900 C. Bottom: Aerogels pre-
Therefore the intense vibrations observed in our study in
pared using the N bases: NN = N,N-diethylaniline, Tri = trihexylamine,
−
1
the region 3400–3250 cm are assigned to Lewis-bound
NC = ammonium carbonate–water solution, NH3 = aqueous ammonia
solution, and TMOS as silicon precursor. For E/NH3 TEOS was used.
ammonia.

CATALYSIS ON IRON OXIDE–SILICA AEROGELS
147
As also reported by Zecchina et al. (42), the decrease in
2+
Fehigh - NO
intensity of the OH stretching vibrations (negative bands)
was correlated with the intensity of the Brønsted bands.
The catalysts M/NN, Md/NN, and M/Tri showed the high-
est decrease in hydroxyls as well as the highest amount of
Brønsted-bound ammonia, followed by M/NC and M/NH3.
E/NH3 and E/NH3d showed nearly no Brønsted sites and
only few hydroxyls were consumed. The acid hydroxyl
groups have already been characterized in the previous
work (11).
.
-2
5 10
2+
Felow - NO
2
0/Md/NN
M/NN
5
/Md/NN
The ratio A = N300/N50 of the amount of ammonia
Md/NN900
◦
◦
adsorbed at 300 C (N300) and 50 C (N50), respectively,
derived from the integrated ν(N–H) band areas, was
taken as a measure of the mean acidity of the cata-
lysts. The values, listed in Table 2, vary from 9 to 18%.
The mean acidity of the catalysts increased in the series
.
-2
3
10
M/NN
M/Tri
∼
E/NH3d < E/NH3 < M/NH3 < M/NC < M/Tri = Md/NN <
M/NC
M/NN.
◦
M/NH3
The stretching vibrations of ammonia adsorbed at 50 C
on the Md/NN aerogels with different iron loading and on
Md/NN900 are depicted in the upper part of Fig. 1. The
amountofBrønsted-boundammoniadecreasesintheorder
E/NH3
1750
1
900
1850
1800
1700
∼
Md/NN > 20/Md/NN > 5/Md/NN = Md/NN900 and corre-
-
1
wavenumber / cm
lates well with the amount of hydroxyl groups on the cata-
lysts (11). The stretching vibrations of Lewis-bound ammo-
FIG. 3. DRIFT spectra of steady-state NO adsorption under argon at
−
1
niaappearedat3372–3350and3286–3280cm . Theratioof
◦
room temperature on the aerogels, reduced at 425 C in hydrogen. Top:
◦
ammoniaadsorbedat300andat50 Cwasagaindetermined spectra of the aerogels with different iron content are shown. Bottom:
as a measure for the mean acidity. The values decrease in spectra of the aerogels prepared by different N-bases are depicted.
the series Md/NN > 20/Md/NN > 5/Md/NN > Md/NN900
(
Table 2).
sites. Figure 3 shows the spectra of NO adsorbed at steady
NO adsorption on the reduced catalysts. The adsorp-
◦
state under argon at 25 C. Two broad bands can be distin-
tion studies of NO on the reduced iron oxide–silica aero-
gels clearly demonstrate the existence of various adsorption
−1
guished at 1830–35 and 1754–58 cm , respectively. Yuen
et al. (46), Segawa et al. (47), and Aparicio et al. (48) studied
2+
the adsorption of NO on silica-supported iron (Fe /SiO2)
and Fe(II)-Y zeolite. The authors could assign NO stretch-
ing bands to NO adsorbed on differently coordinated iron
by comparison with M o¨ ssbauer spectroscopy. The vibra-
TABLE 2
Results from Ammonia and NO Adsorption Studies a
−
1
N50b
N300b
c
2+
2+
tions at 1870–30 cm were assigned to highly coordinated
Catalyst
A = N300/N50
R = Felow /Fehigh
iron with low or intermediate accessibility for NO to adsorb,
−
1
Md/NN
Md/NN900
5
2
M/NN
M/Tri
M/NC
M/NH3
E/NH3
E/NH3d
54
11
13
47
47
38
36
27
10
11
9
0.5
1
6
8
6
5
3
1
0.166
0.045
0.077
0.128
0.176
0.164
0.148
0.116
0.09
—
whereas the vibrations at 1750 and 1767 cm were assigned
0.113
0.121
0.163
0.457
0.35
2+
to a mononitrosyl of Fe in a low-coordinated state. For
iron on amorphous silica (2) the ν(NO) at 1755 cm was
assigned to NO adsorbed on Fe incorporated in the silica.
We therefore assign the band at 1830–35 cm to NO ad-
sorbed on Fe with high coordination, probably incorpo-
rated in small iron oxide clusters, with a broad dispersion of
sites due to differences in coordination and/or accessibility.
The band around 1755 cm is assigned to low-coordinated
Fe , probably due to iron incorporated in the silica ma-
/Md/NN
0/Md/NN
−1
2
+
−1
2
+
0.29
0.272
0.142
—
1
0.09
−
1
a
◦
◦
2+
Ammonia uptake of calcined aerogels at 50 C (N50) and 300 C
2
+
2+
(
N300), respectively. Ratio A = N300/N50, as determined by DRIFTS.
trix. Table 2 lists the ratio, R = Felow / Fehigh , of low-
Relative amount of NO bound to low- and high-coordinated iron,
R = Felow /Fehigh , on reduced aerogels.
2+
to high-coordinated Fe , determined by the integrated
2+
2+
b
area of the corresponding ν(NO) band. The ratio R de-
Area of the ν(NH) vibrations of ammonia in the range 3500–
∼
−
1
creases in the series M/NN > M/Tri > M/NC = M/NH >
2
000 cm
.
3
c
−1
.
Ratio of the area of the ν(NO) band at 1755 and 1830 cm
E/NH3.

¨
FABRIZIOLI, BURGI, AND BAIKER
1
48
NO adsorption on the reduced aerogels with different
iron content and calcination temperature is shown in Fig. 3
top). The highest amount of NO adsorbed was observed
(
on 20/Md/NN, followed by M/NN; very little was adsorbed
on 5/Md/NN and Md/NN900. The ratio R of the integrated
2
+
area of the ν(NO) band on low- and high-coordinated Fe
∼
decreases in the series Md/NN > 20/Md/NN > 5/Md/NN =
Md/NN900 (Table 2).
3
.3. Selective Catalytic Oxidation (SCO) of Ammonia
Catalytic tests. The results of the catalytic tests in the
◦
SCO of ammonia measured in the 210–550 C range are
summarized in Table 3. High selectivity to nitrogen up to
9
8% combined with full conversion of ammonia could be
◦
achieved at 500 C for M/NN, Md/NN, M/Tri, and M/NC.
M/NH3, E/NH3, and E/NH3d afforded a selectivity of 90 to
◦
9
3%N2 at540 C, withN2OandNOasby-products. Activity
decreased in the order Md/NN > M/NN > M/Tri > M/NC >
M/NH3 > E/NH3d > E/NH3, as reflected by both TOF and
◦
rate per surface area at 330 C (Table 3, Fig. 4 bottom). Note
that the reported TOF values are conservative estimates
(
lower limit), since for their calculation the total amount of
iron was considered. The highest activity showed the aero-
gels prepared with N,N-diethylaniline as gelation agents,
−
1
Md/NN and M/NN, with a TOF of 0.16 and 0.14 (ks)
,
respectively. Catalysts M/Tri, M/NC, and M/NH3 were con-
FIG. 4. SCO of ammonia. Arrhenius plot of N2 production rates for
−1
siderably less active (TOF = 0.08 − 0.06 (ks) ). E/NH3 and aerogels prepared by different sol–gel synthesis. Top: aerogels with differ-
∼
−1
ent iron content and calcination temperature. Bottom: aerogels containing
0 wt% nominal Fe2O3 prepared by different N bases.
E/NH3dweretheleastactivecatalysts(TOF = 0.05(ks) ).
1
Reference measurements with the silica component alone
◦
showed some activity, reaching 5% conversion at 500 C
with selectivity to NO of 57%. Among the catalysts
with different iron content the activity decreased in the
order Md/NN > 5/Md/NN ꢁ 20/Md/NN. Note that the
rate per surface area of the 5% Fe2O3-containing aerogel
(
5/Md/NN) is lower than expected from the TOF value be-
cause the surface area of this sample was nearly two times
higher than that of the other aerogels. The activity collapsed
TABLE 3
Catalytic Results of the Iron Oxide–Silica Aerogels in the Selective for the 20/Md/NN catalyst, whereas the selectivity was still
Catalytic Oxidation of Ammonia to N2
◦
very high, amounting to 97% at 500 C (X > 98%). The
◦
Md/NN900 aerogel, calcined at 900 C, was less active than
the corresponding aerogel calcined at 600 C. This can only
a
6
11
TOF × 10
Rate × 10
Selectivity to N2
◦
−1
−2 −1
Eact(N2)
(s
)
(mol N2 m
s
)
(%) (X > 98%,
Catalyst (kJ mol 1) (330 C)
−
◦
(330 C)
◦
500 C)
◦
partly be explained by the decrease in surface area. The rate
per surface area was indeed only 15% lower compared to
that for Md/NN. The Arrhenius plots are shown in Fig. 4.
The activation energies (E(N2)) calculated from the differ-
b
164c
72
68b
58
Md/NN
Md/NN900
80 ± 4
69 ± 5
86 ± 3
90 ± 5
79 ± 5
77 ± 5
76 ± 3
90 ± 5
94 ± 4
94 ± 4
96
93
c
c
d
5
/Md/NN
103
11
96
−
1
ential measurements were in the range 69–94 kJ mol , the
catalysts with the higher selectivity and activity showing
lower activation energies.
2
0/Md/NN
31
137
80
19
55
30
24
97
97
97
97
M/NN
M/Tri
M/NC
M/NH3
E/NH3
E/NH3d
70
The selectivity to nitrogen varied with temperature and
c
c
d
64
22
10
90
93
91
◦
c
c
d
was highest at 450–540 C. The selectivities to the by-
48
c
c
d
products NO and N O, for the catalysts with highest
51
11
2
(
Md/NN) and lowest activity (E/NH3), are depicted in Fig. 5
a
TOF = [N2]/([Fe] · s).
Ninety-five percent confidence limits of Arrhenius-type linear regres-
sion.
(
right). The upper plot is characteristic also for aerogels
b
M/NN, M/Tri, and M/NC. Selectivity to N2O amounted to
◦
c
8–10% and was higher than that to NO up to 330 C. At
Extrapolated values.
d
◦
X > 98% at 540 C.
higher temperature NO selectivity increased to about 5%

CATALYSIS ON IRON OXIDE–SILICA AEROGELS
149
1
00
10
the first 10 min. This effect was not so distinct on E/NH3.
Brønsted-bound ammonia was formed or rearranged due
to the water produced during SCO. After 30 min some
of the Brønsted-bound ammonia started to react. It must
be pointed out that only a slight decrease in these bands
was observed compared to the vibrations due to the Lewis-
bound ammonia.
5
/Md/NN
Md/NN
5
0
NO
N O
5
NO
2
N O
2
0
/Md/NN
0
00
0
10
1
The position of the vibrations of the consumed Lewis-
bound ammonia was nearly constant over the whole re-
20/Md/NN
E/NH3
−
1
−1
action time at 3330 cm (νas(NH)), 3370 cm (νs(NH)),
−
1
and 3172 cm (overtone of the δ(NH)) on M/NN and at
5
0
5
0
−
1
NO
N O
3
340, 3373, and 3174 cm on E/NH3. In both cases the fre-
N O
2
quency of the stretching vibration of the reacted ammonia
was lower than that of the adsorbed species, indicating that
the more acid Lewis species had reacted (cf. ammonia ad-
sorption studies). Brønsted-bound ammonia was consumed
2
NO
50 300 350 400 450 500
0
2
300 350 400 450 500 550
Temperature /
˚C
−
1
−1
at 2600 cm on M/NN and 2730 cm on E/NH3. Also in
FIG. 5. SCO of ammonia. Conversion of ammonia and selectivities thiscasethefrequencyoftheconsumedammoniawaslower
to nitrous and nitric oxide for catalysts 5/Md/NN and 20/Md/NN (left) and
than that of the Brønsted species before reaction (2785 and
Md/NN and E/NH3 during the integral reactivity studies. For comparison,
ammonia conversion with the pure silica aerogel (0/Md/NN) is shown in
section 5/Md/NN.
−1
2810 cm , respectively). Broad bands associated with the
deformation vibrations of these species were observed at
−
1
1700 and 1450 cm
.
TPD and reduction by ammonia after different pretreat-
◦
at the expense of N2O. Between 400 and 480 C both by-
ments of the aerogels. TPD in helium (SCO–TPD) was
performed for all samples after the catalytic test in SCO.
The desorption signals of NH3, N2, NO, and N2O are
products nearly disappeared, and N2 selectivity reached a
maximum. Selectivities to NO and N2O for E/NH3 (Fig. 5,
right), M/NH3, and E/NH3d decreased steadily with in-
◦
creasing temperature. Again with high conversion at 500 C
they nearly disappeared. Comparing the aerogels with dif-
ferent iron contents, a decrease in the formation of by-
products with higher iron loading was observed.
d)
formation
M/NN
b)
a)
Catalytic activity was investigated also for Md/NN af-
ter in situ reduction at 550 C in 5% H2/Ar. The catalyst
◦
a)
showed about half the activity of the calcined one, its selec-
◦
c)
d)
consumption
tivity to N2 increased in the 320–440 C range, and only small
◦
amounts of N2O were produced. Above 450 C NO produc-
◦
tion increased, reaching 14% at 540 C, which is twice the
value found for the calcined catalyst.
In situ DRIFTS measurements during SCO of ammonia.
DRIFTS was used to investigate the role of the acidic sites
in the SCO of ammonia. For this purpose M/NN and E/NH3
E/NH3
a)
were pretreated in air, and then ammonia was adsorbed at
◦
b)
c)
2
50 C. The observed bands corresponded to Lewis- and
Brønsted-bound ammonia. Figure 6 depicts the disappear-
◦
ance of ammonia after admission of 1.8% O2/Ar at 250 C
for catalyst M/NN (top) and E/NH3 (bottom). With both
catalysts admission of oxygen led to an immediate decrease
of the Lewis-bound ammonia, characterized by the vibra-
d)
3
500
3000
2500
2000
1500
−1
−1
tions at 3400–3200 cm (ν(NH)) and 1608 cm (δ(NH)).
Concurrently an increase of the signal in the region 3700–
-
1
wavenumber / cm
−
1
3
500 cm was observed, probably due to hydroxyls hy-
FIG. 6. DRIFTS study of the reaction of preadsorbed ammonia with
oxygen at 250 C. Spectra were recorded after (a) 4, (b) 10, (c) 30, and
d) 60 min. Spectra of adsorbed ammonia in argon at 250 C (0 min) was
drogen bonded to water (34). On M/NN the admission of
◦
oxygen led also to an increase of a broad band at 3000–
◦
(
−1
−1
2
800 cm together with a vibration at 1440 cm during substracted.

¨
FABRIZIOLI, BURGI, AND BAIKER
1
50
◦
at low-temperature (120 C) became prominent, and some
◦
NO and N2 desorbed at 220 C. At high temperature nearly
no oxidation products were observed, but ammonia desorp-
tion was slightlyincreased compared to thecalcined sample.
The reduction of the catalysts by ammonia was investi-
gated using in situ TPR with NH3 for M/NH3, E/NH3, and
Md/NN. Upon heating in 2000 ppm NH3/He, the nitrogen
◦
production started at about 280 C (not shown). The max-
◦
imum N2 production occurred at 460–468 C. Only N2 and
water were produced during reduction, whereas during re-
action with oxygen and also during TPD, products of deeper
oxidation were observed. The amount of nitrogen formed
was higher after use in SCO than after calcination, which
may be explained by the presence of adsorbed intermediate
species from the catalytic reaction.
3.4. Selective Catalytic Reduction of NO with NH3
The Arrhenius plot of the mixed oxide aerogels in SCR
of NO with ammonia is shown in Fig. 8. Additionally the
catalytic performance of the aerogels is summarized in
Table 4. Note that the activity of the aerogels is compared at
◦
atemperaturewhereonlySCRoccurred(210 C). Theactiv-
FIG. 7. TPD of ammonia from differently pretreated catalyst Md/NN.
ity was highest for aerogel E/NH3 with a TOF amounting to
0.12 N2 Fe ks , followed by M/NH3 (0.11 N2 Fe ks ).
M/NC, M/Tri, and M/NN showed lower activity with about
.08 N2 Fe ks . In general the activity was higher than
After in situ calcination (thin line), after use in SCO (thick line), and after
−
1
−1
−1
−1
◦
in situ reduction in 5% H2/He at 550 C (dashed line).
−
1
−1
0
representatively shown for differently pretreated catalyst that in the SCO of ammonia. The selectivity toward ni-
Md/NN in Fig. 7. For all aerogels desorption of ammonia trogen was high over the whole temperature range stud-
◦
◦
showed a maximum at 105–119 C and was observed up to ied, and amounted to 98–99% in the 380–460 C range
◦
5
00 C. Theoverallamount ofdesorbed ammonia decreased (SN2 + S
= 1), that is, when the catalysts reached >98%
N2O
∼
in the series M/NN = Md/NN > M/Tri > M/NC > M/NH3 > conversion of ammonia. The consumption of NO decreased
∼
◦
E/NH3d = E/NH3, and corresponded well to the amount at temperatures higher than 250 C, due to the onset of
of adsorbed ammonia determined by DRIFTS for the cal- ammonia oxidation. The ratio of the conversion of NO
cined samples. N2 and NO were found to desorb at tem- and NH (XNO/XNH3 ) decreased with higher temperature.
3
◦
peratures above 390 C, whereas N2O was not observed. Among the aerogels studied it decreased in the series
◦
The desorption of N2 and NO was maximal at 460–480 C.
Water desorption was prominent in the 170–200 C range
◦
-1.2
for the samples which showed high Brønsted acidity in
DRIFTS (Md/NN, M/NN, M/Tri), whereas on the other
samples it was shifted to lower temperature. The concur-
rent desorption of ammonia and water at higher temper-
ature may therefore be attributed to the desorption of
stronger Brønsted-bound ammonia followed by conden-
sation of the hydroxyl groups. Ammonia desorption can
M/NC
-1.8
-2.4
-3.0
2
10 C
˚
E/NH3
M/NH3
originate from Brønsted- and Lewis-bound ammonia up to
◦
3
00 C, as shown by DRIFTS.
Aerogel Md/NN was also investigated after calcination
in air at 550 C and after reduction in hydrogen at 550 C
Fig. 7). The desorption profiles for ammonia and N2 were
M/NN
M/Tri
◦
◦
(
slightly more intense with the calcined catalyst compared
to that used in SCO. All curves had similar shapes to those
observed after use in SCO. For the reduced catalyst, how-
ever, distinct differences in the desorption behavior were
1
.85 1.90 1.95 2.00 2.05 2.10 2.15 2.20
-
1
1
000 / T K
FIG. 8. Temperature dependence of SCR of NO by ammonia on var-
observed: ammonia desorption decreased, N2O desorption ious aerogels. Arrhenius plot of the N2 production rates.

CATALYSIS ON IRON OXIDE–SILICA AEROGELS
151
TABLE 4
Catalytic Results of Iron Oxide–Silica Aerogels in the Selective
Catalytic Reduction of NO by Ammonia
a
6
a
11
b
TOF × 10
Rate × 10
Selectivity to N2
−
1
−1
−2 −1
Eact (N2) (N2Fe
s
) (mol N2 m
s
)
(%) (X > 98%,
Catalyst (kJ mol 1)
−
(210 C)
◦
(210 C)
◦
380–460 C)
◦
c
M/NN
M/Tri
M/NC
M/NH3
E/NH3
53 ± 2
46 ± 4
47 ± 3
38 ± 1
38 ± 2
80
85
85
106
119
32
32
35
28
23
99
99
98
99
98
a
XNO/XNH = 1.
3
b
c
SN O = 1 − SN .
2
2
Ninety-five percent confidence limits of Arrhenius-type linear regres-
sion.
E/NH3 > M/NH3, M/NC, M/Tri > M/NN, and was therefore
inversely correlated to their activity in the SCO of ammonia
(
Fig. 9). Note, that the influence of the ammonia oxidation
◦
sharply increased above 450 C, as seen by the decrease in
slope of the NO conversion.
4
. DISCUSSION
The following discussion focuses on the correlation be-
tween structural and acidic properties of the catalysts and
their performance in SCO and SCR.
The catalysts showed distinctly different acidic proper-
ties. The DRIFTS results revealed that both the amount of
adsorbed ammonia and the strength of adsorption varied
considerably among the catalysts. The amount of adsorbed
◦
FIG. 10. (a) Correlation between reaction rates of SCO at 330 C (ꢀ),
◦
◦
SCR at 210 C (ꢁ), and ammonia adsorbed at 300 C. (b) Abundance
2+
2+
of low-coordinated iron (R = Felow / Fehigh ). (c) Relative amount of
◦
stronglyboundammonia(A = N300/N50)tototalammoniauptakeat300 C
∼
ammonia increased in the series E/NH3 = E/NH3d < M/ for all aerogels containing 10 wt% Fe O .
2
3
◦
NH3 < M/NC < M/Tri < M/NN < Md/NN. At 300 C mainly
Lewis-bound ammonia was still bound to the surface. The
activity in SCO is depicted as a function of the total amount
ofammoniaadsorbedat300 CinFig. 10. Clearlytheactivity
100
M/NC
E/NH3
◦
9
8
7
6
5
0
0
0
0
of the catalyst positively correlates with the amount of ad-
sorbed ammonia. Figure 10 also demonstrates that the rel-
ative abundance of strongly bound ammonia A = N300/N50
increases with increasing adsorption capacity. Both acid-
ity and adsorption capacity for ammonia may play impor-
tant roles in the oxidation of ammonia. Indeed, for the ox-
idation of ammonia on Md/NN and E/NH3 followed by
DRIFTS (Fig. 6), the reaction was found to proceed mainly
by consumption of the stronger Lewis-bound ammonia.
Lewis-bound ammonia was also suggested to be the ac-
tive species in ammonia oxidation in previous studies for
supported and pure iron oxide catalysts (29, 31) and for
Cu-exchanged zeolites (49). Thebestresults reported to our
knowledge were achieved with ion-exchanged zeolites con-
taining highly dispersed iron (27) or copper ions (49), reach-
M/NH3
M/Tri
M/NN
0
2
50
300
350
400
450
500
550
Temperature / C
˚
FIG. 9. Selective catalytic reduction of NO by ammonia. Ratio of the
NO conversion to the ammonia conversion in the temperature 270–540 C
−
1
◦
−1
◦
◦
ing TOF values of 14 (ks) at 350 C and 4.8 (ks) at 180 C,
respectively.
range.

¨
FABRIZIOLI, BURGI, AND BAIKER
1
52
We propose that the stronger Lewis-bound ammonia is obtained for V2O5/SiO2 (54) and Fe-ZSM-5 (27), whereas
associated with isolated Fe ions, embedded in the silica a lower activation energy was found for Pt catalysts (55).
matrix in tetrahedral coordination, whose relative abun- The activation energy for pure hematite amounted to 95–
dance, R, grows parallel to the relative amount of strongly 100 kJ/mol (31, 32) and was close to that of our less active
bound ammonia, as indicated in Figs. 10b and 10c. These samples. The activation energy was lower for the more ac-
sites are in low coordination in contrast to the iron ions tive catalysts M/NC, M/Tri, M/NN, and Md/NN than for
in agglomerates (46, 48, 50) and could be characterized M/NH3, E/NH3, and E/NH3d (Table 3). The catalysts with
by NO adsorption on the reduced aerogels. The presence low activation energy possess larger acidity than the cata-
of tetrahedrally coordinated Fe3 has been evidenced by lysts with high activation energy (Fig. 10). The acidity and
+
UV–vis and EPR spectroscopy on the calcined samples hence the adsorption strength of the ammonia may play
(
11). The dispersion of the iron ions decreased in the se- important roles in the dehydrogenation step of ammonia.
∼
ries M/NN > M/Tri > M/NC = M/NH3 > E/NH3. Also com- Amide species as intermediate toward N2 production were
paredtotheaerogels5/Md/NN, 20/Md/NN, andMd/NN900, also postulated by Amores et al. (28) for supported transi-
the relative amount of low-coordinated iron was greater on tion metal oxides and by Williamson et al. for Cu-Y (49),
M/NN and Md/NN. Tetrahedral coordination of Fe3 in the whereas dimerization of NH2O species was suggested to
+
silica leads to strong Lewis acid centers, as well as Brønsted occur over molybdena supported on silica (54). On pure ox-
3
+
acid Fe –OH–Si groups, as known from studies on zeolites ides Matyshak and Krylov (32) and Il’chenko and Golodets
51, 52). (31) proposed the reaction of an amide with a nitro/HNO
NO adsorption studies showed that iron oxide nanopar- species to form nitrogen.
ticles were relatively more abundant on E/NH3 than on For SCO the selectivity toward nitrogen increases at
(
M/NN (Table 2, Fig. 3), in agreement with UV–vis and EPR higher temperature (Fig. 5), as also reported by Long
spectroscopy (11). Rethwisch and Dumesic (2) assigned the and Yang (27) for Fe-ZSM-5 and by Amblard et al. (3)
−
1
vibration at 1830 cm to NO bound to iron oxide particles for Ni/Al2O3, Mn/Al2O3, and Fe/Al2O3, on which N2
◦
on amorphous silica. The particle size increased with higher production was maximal at 350–550 C depending on the
calcination temperature (Md/NN900) and with increasing transition metal investigated and decreased at higher tem-
iron content (20/Md/NN), as shown in our previous study perature due to NO production. We observed an increase in
◦
by EPR and XRD (11). On these samples also crystalline NO above 500 C, which was more pronounced on Md/NN
◦
iron oxide particles were detected. The abundance of acidic prereduced at 550 C than on the calcined sample. Reac-
sites and their strength were lower on these catalysts; on tion on the pure SiO2 aerogel, producing mainly NO, has
hematite only weak acid sites were observed, and nearly no to be taken into account at this temperature, but its activity
◦
ammonia adsorption was found at 300 C (not shown). Cor- was too low to fully explain the NO onset on the mixed ox-
respondingly, catalysts with a high abundance of iron oxide ide aerogels. The presence of iron oxide agglomerates did
nanoparticles showed low activity in SCO (Fig. 4). On the not lead to formation of large amounts of by-products, as
other hand, no loss in selectivity toward N2 was observed demonstrated by the high selectivity to nitrogen observed
◦
up to 550 C for the catalysts with high content of iron oxide for Md/NN/900 and 20/Md/NN over the whole temperature
particles. For Md/NN900 the same selectivities as those on range studied. Hence the observed increase in selectivity
◦
Md/NN were observed, whereas on 20/Md/NN, N2 selectiv- toward NO above 500 C is likely not associated with iron
ity was even higher in the low-temperature range (Fig. 5). oxide agglomerates. A possible pathway for N2O forma-
These results are supported by earlier SCO studies on pure tion at low-temperature is the combination of an adsorbed
hematite, which revealed low activity compared to that on dehydrogenated ammonia species with NO (54) or an ad-
other transition metal oxides (31, 32).
sorbed nitro/nitrate species. Alternatively the combination
Iron oxide supported on silica was investigated by Uddin of two adsorbed HNO species was postulated (31). Nearly
et al. (16), but low selectivity toward nitrogen was observed. no N2O was observed during SCR, rendering the internal
On MFI-type ferrisilicalite (16) and Fe/Al2O3 (3), selectiv- SCR mechanism less probable.
◦
ity to N2 > 90% in the 400–500 C range was obtained. Long
We observed NO production also during the TPD of
◦
and Yang (27) investigated SCO on Fe-exchanged zeolites ammonia in argon above 400 C. The catalyst is there-
and reported high activity with selectivity toward nitrogen fore easily reduced above this temperature. On the other
reaching 100% at 450 C. At this temperature a TOF value hand, TPR in ammonia showed only formation of N2. This
◦
−
1
of 22 (ks) was observed for Fe-ZSM-5 with 0.86 wt% Fe. indicates that an increased surface coverage of ammonia
Note that H-ZSM-5 is also active in SCO (53). Increasing suppresses NO formation at high temperature. Differen-
◦
the Si/Al ratio in Fe-ZSM-5 from 10 to 100, 20% loss in ac- tial catalytic measurements in SCO in the 350–400 C range
tivity and up to 50% lower selectivity to N2 were observed. revealed that the activation energy for NO formation was
The activation energy for N2 formation in SCO for the about 120–150 kJ/mol, much higher than that for N2 and
aerogels is in the range 80–90 kJ/mol. Similar values were N2O formation with activation energies between 70 and

CATALYSIS ON IRON OXIDE–SILICA AEROGELS
5. CONCLUSIONS
Iron oxide–silica aerogels with different iron dispersion
153
9
0 kJ/mol. The values for the activation energies explain
the onset in NO production at higher temperature. There-
fore oxidation of ammonia to NO and reaction of NO with
adsorbed ammonia (internal SCR), as suggested by Long
and Yang (27) for Fe-ZSM-5, are likely not the dominant
route for N2 production in SCO on our catalysts. This is
further supported by the different activity in SCO and SCR
for the catalyst series investigated.
3
+
ranging from tetrahedrally coordinated Fe to iron oxide
clusters were synthesized by varying the sol–gel route, the
calcination temperature, and the iron content. The aerogels
showed high selectivity to nitrogen in the SCO of ammo-
◦
nia, reaching 97% at 500 C. Catalytic oxidation of ammo-
nia in the presence and absence of oxygen started at about
The activity of the catalysts in SCR does not correlate
with their activity in SCO (Fig. 10). The reaction rate per
surface area for SCR only increases slightly with increasing
ammonia adsorption capacity. The different influence of
the acidity on the activity indicates that the active sites for
SCR differ from those for SCO. Based on the observation
that the samples containing a substantial amount of iron
oxide clusters showed a consistently higher SCR activity,
we may infer that these species are important for SCR.
Joyner and Stockenhuber (56) reported that on Fe-ZSM-5
iron oxide nanoclusters were more active for SCR with
hydrocarbons than the isolated Fe cations. The activation
energies for SCR lie between 38 and 53 kJ/mol (Table 3)
and are the same for ammonia and NO conversion and
production of nitrogen. Again the more active catalysts
have slightly lower activation energies. All activation
energies are lower than the values reported for H-ZSM-5
◦
270 C. The SCO activity was found to be positively cor-
related to the acidity of the aerogels probed by ammonia
adsorption. The relative abundance of strong acid sites cor-
related to the fraction of low-coordinated iron, probably
incorporated into the silica matrix. Therefore we propose
that the acid sites associated with the low-coordinated iron
are the active sites for SCO. Lewis-bound ammonia is likely
to be the reactive species for this reaction.
The activity for the SCR of NO by ammonia did not cor-
relate to the Lewis acidity of the iron oxide–silica aero-
gels, indicating that these sites do not play a crucial role in
SCR. In general the activity for SCR was higher than that
found for SCO of ammonia. SCO lowered the selectivity
to nitrogen achievable by SCR at temperatures higher than
◦
ca. 250 C.
REFERENCES
(
43, 57) and H-mordenite (58), with activation energies of
6
0–65 kJ/mol.
1
2
. Ratnasamy, P., and Kumar, R., Catal. Today 9(4), 329 (1991).
. Rethwisch, D. G., and Dumesic, J. A., J. Phys. Chem. 90,1625 (1986).
In contrast to vanadia catalysts, where no adsorption of
NO was observed on the oxidized catalyst, both NOx and
NH3 may adsorb on iron-containing catalysts (2, 18, 47, 48,
3. Amblard, M., Burch, R., and Southward, B. W. L., Appl. Catal. B–
Environ. 22, L159 (1999).
4. Janssen, F. J. J. G., and van den Kerkhof, F. M. G., KEMA Scientific
5
0, 56) as well as on molecular sieve-type catalysts. Forma-
Tech. Rep. 3(6), 71 (1985).
tion of NO2 (or an adsorbed NO2-like species) followed by
reduction by ammonia was proposed to be the main route
in the SCR of NO over H-mordenite and H-ZSM-5 (57, 59,
5
6
. Cheng, L. S., Yang, R. T., and Chen, N., J. Catal. 164, 70 (1996).
. Wang, C. T., and Willey, R. J., J. Non–Cryst. Solids 225, 173 (1998).
7. Willey, R. J., Lai, H., and Peri, J. B., J. Catal. 130, 319 (1991).
8
. Willey, R. J., Oliver , S. A., Olivieri, G., and Busca, G., J. Mater. Res.
(6), 1418 (1993).
. Hutter, R., Mallat, T., and Baiker, A., J. Catal. 153, 177 (1995).
6
0). Also for Fe-exchanged-pillared clays and Fe-ZSM-5-
adsorbed NOx species were found to be reactive to NH3
14, 18). Furthermore the presence of adsorbed ammonia
pairs was proposed to increase the catalytic activity in SCR
8
9
(
1
0. Hutter, R., Mallat, T., Dutoit, D., and Baiker, A., Top. Catal. 3, 421
(
1996).
(
18, 59).
11. Fabrizioli, P., Burgener, M., B u¨ rgi, T., van Doorslaer, S., and Baiker,
A., J. Mater. Chem., in press.
The influence of SCO on SCR of NO by ammonia was not
12. Bosch, H., and Janssen, F., Catal. Today 2/4, 369 (1987).
13. Long, R. Q., and Yang, R. T., J. Catal. 194, 80 (2000).
14. Long, R. Q., and Yang, R. T., J. Catal. 198, 20 (2001).
15. Amiridis, M. D., Puglisi, F., Dumesic, J. A., Millman, W. S., and Topsoe,
N.-Y., J. Catal. 142, 572 (1993).
constant over the temperature range analyzed, mainly as
a consequence of the different activation energies of these
processes. Ammonia oxidation during SCR started at about
◦
2
50 C, in agreement with the SCO investigations. The NO
1
1
1
6. Uddin, M., Komatsu, T., and Yashima, T., J. Chem. Soc., Faraday Trans.
91(18), 3275 (1995).
7. Seiyama, T., Arakawa, T., Matsuda, T., Takita, Y., and Yamazoe, N.,
J. Catal. 48, 1 (1977).
conversion steadily decreased relative to the NH3 conver-
sion above this temperature. At 450 C a distinct decrease
in the slope of X(NO)/X(NH3) versus temperature was ob-
served (Fig. 9). At this temperature NO evolution increased
◦
8. Long, R. Q., and Yang, R. T., J. Catal. 190, 22 (2000).
noticeably during ammonia TPD. Hence, the aerogel can be 19. Long, R. Q., and Yang, R. T., Catal. Lett. 59, 39 (1999).
◦
2
2
0. Chen, J. P., Hausladen, M. C., and Yang, R. T., J. Catal. 151, 135 (1995).
1. Kato, A., Matsuda, S., Kamo, T., Nakajima, F., Kuroda, H., and Narita,
T., J. Phys. Chem. 85, 4099 (1981).
2. Chang, T.-H., and Leu, F.-C., Appl. Catal. A Gen. 180, 123 (1999).
3. Naruse, Y., Ogasawara, T., Hata, T., and Kishitaka, H., Ind. Eng. Chem.
Prod. Res. Div. 19, 57 (1980).
quickly reduced above 450 C, which is reflected in the ac-
tivity for SCO. However, TPD of ammonia after different
pretreatments indicates that the catalyst after use in SCO
predominantly maintains the oxidation state of the calcined
aerogel, which was determined to be +3 by TPR (11).
2
2

¨
FABRIZIOLI, BURGI, AND BAIKER
1
54
2
2
4. Biermann, J. J. P., in “KEMA Scientific & Technical Reports” (A. J.
van Loon, Ed.), Vol. 8(5), p. 273. KEMA, Den Haag, 1990.
5. Heck, R. M., Chen, J. M., and Speronello, B. K., Environ. Prog. 13, 45. Pimentel, G. C., and McClellan, A. L., in “The Hydrogen Bond” (L.
44. Teunissen, E. H., van Santen, R. A., Jansen, A. P. J., and van
Duijneveldt, F. B., J. Phys. Chem. 97, 203 (1993).
2
21 (1994).
Pauling, Ed.), p. 221. Freeman, San Francisco, 1960.
46. Yuen, S., Chen, Y., Kubsh, J. E., Dumesic, J. A., Topsoe, N., and Topsoe,
H., J. Phys. Chem. 86, 3022 (1982).
2
2
2
6. Long, R. Q., and Yang, R. T., Chem. Commun. 1651 (2000).
7. Long, R. Q., and Yang, R. T., J. Catal. 201, 145 (2001).
8. Amores, J. M. G., Escribano, V. S., Ramis, G., and Busca, G., Appl. 47. Segawa, K.-I., Chen, Y., Kubsh, J. E., Delgass, W. N., Dumesic, J. A.,
Catal. B–Environ. 13, 45 (1997).
Hall, W. K., J. Catal. 76, 112 (1982).
2
3
9. Ramis, G., Yi, L., Busca, G., Turco, M., Kotour, E., and Willey, R. J.,
J. Catal. 157, 523 (1995).
0. Larrubia, M. A., Ramis, C., and Busca, G., Appl. Catal. B–Environ. 49. Williamson, W. B., Flentge, D. R., and Lunsford, J. H., J. Catal. 37, 258
48. Aparicio, L. A., Hall, W. K., Fang, S.-M., Ulla, M. A., Millman, W. S.,
Dumesic, J. A., J. Catal. 108, 233 (1987).
3
0, 101 (2001).
(1975).
3
3
3
1. Il’chenko, N. I., and Golodets, G. I., J. Catal. 39, 57 (1975).
2. Matyshak, V. A., and Krylov, O. V., Catal. Today 25, 1 (1995).
3. Sil’chenkova, O. N., Matyshak, V. A., and Korchak, V. N., Kinet. Katal.
50. Lobree, L. J., Hwang, I., Reimer, J. A., and Bell, A. T., J. Catal. 186,
242 (1999).
51. Datka, J., and Abramowicz, T., J. Chem. Soc., Faraday Trans. 90(6),
2417 (1994).
4
0, 526 (1999).
3
3
3
4. Vansant, E. F., Van Der Voort, P., and Vrancken, K. C., Stud. Surf. Sci.
Catal. 93, 59 (1995).
5. Davydov, A. A., “Infrared Spectroscopy of Adsorbed species on the
Surface of Transition Metal Oxides.” Wiley, Chichester, 1990.
6. Little, L. H., “Infrared Spectra ofAdsorbed Species,” p. 180. Academic
Press, New York, 1966.
52. Bordiga, S., Buzzoni, R., Geobaldo, F., Lamberti, C., Giamello, E.,
Zecchina, A., Leofanti, G., Petrini, G., Tozzola, G., and Vlaic, G.,
J. Catal. 158, 486 (1996).
53. Long, R. Q., and Yang, R. T., Chem. Commun. 1651 (2000).
54. DeBoer, M., Huisman, H. M., Mos, R. J. M., Leliveld, R. G., vanDillen,
A. J., and Geus, J. W., Catal. Today 17, 189 (1993).
55. Willey, R. J., and Djuhadi, V. S., in “Catalysis 1987” (J. W. Ward, Ed.),
p. 435. Elsevier, Amsterdam, 1988.
3
7. Topsoe, N.-Y., J. Catal. 128, 499 (1991).
3
8. Nakamoto, K., “Infrared and Raman Spectra of Inorganic and Coor-
dination Compounds,” Wiley, New York, 1995.
56. Joyner, R., and Stockenhuber, M., J. Phys. Chem. B 103, 5963 (1999).
57. Stevenson, S. A., Vartuli, J. C., and Brooks, C. F., J. Catal. 190, 228
(2000).
3
9. Busca, G., Centi, G., Marchetti, L., and Trifiro, F., Langmuir 2, 568
(1986).
4
4
4
0. Pittman, R. M., and Bell, A. T., Catal. Lett. 24, 1 (1994).
1. Vedder, W., and Hornig, D. F., J. Chem. Phys. 35 (5), 1560 (1961).
2. Zecchina, A., Marchese, L., Bordiga, S., Paz e` , C., and Gianotti, E.,
J. Phys. Chem. B 101, 10,128 (1997).
58. Andersson, L. A. H., Brandin, J. G. M., and Odenbrand, C. U. I., Catal.
Today 4, 173 (1989).
59. Eng, J., and Bartholomew, H., J. Catal. 171, 14 (1997).
60. Kiovsky, J. R., Koriada, P. B., Lim, C. T., Ind. Eng. Chem. Prod. Res.
Div. 19, 218 (1980).
4
3. Eng, J., and Bartholomew, H., J. Catal. 171, 27 (1997).