Interaction between sulfated zirconia and alkanes: Prerequisites for active sites - Formation and stability of reaction intermediates

Article abstract of DOI:10.1016/j.jcat.2004.11.045

Two sulfated zirconia catalysts were prepared via sulfation and calcination of zirconium hydroxide at 873 K; the zirconium hydroxide had been aged at room temperature for 1 h (SZ-1) or aged at 373 K for 24 h (SZ-2). SZ-1 was active for n-butane isomerization at 373 K; SZ-2 reached a similar performance level only at 473 K. Both materials contained about 9 wt% sulfate and were tetragonal. Because of a lower BET surface area (105 vs. 148 m²/g) SZ-1 featured a higher sulfate density, and XRD and EXAFS analysis showed larger (ca. 10 nm) and more well ordered crystals than for SZ-2. n-Butane TPD on SZ-1 showed a butene desorption peak at low temperature, whereas no obvious butene desorption was observed with SZ-2, suggesting that SZ-1 has a higher oxidizing power at low temperature than SZ-2. The number of sites capable of dehydrogenation are less than 5 μmol/g, because the differential heats of n-butane adsorption as measured by microcalorimetry were 45-60 kJ/mol for higher coverages, indicating weak and reversible interaction. TAP experiments describe the adsorption and desorption behavior of n-butane at different activity states and are the basis for a simple adsorption model. Reactant pulses and purge experiments show that the active species, presumably formed in an oxidative dehydrogenation step, are stable at the surface under reaction conditions.

Full text of DOI:10.1016/j.jcat.2004.11.045

Journal of Catalysis 230 (2005) 214–225
www.elsevier.com/locate/jcat
Interaction between sulfated zirconia and alkanes: prerequisites for active
sites—formation and stability of reaction intermediates
Xuebing Li, Katsutoshi Nagaoka, Laurent J. Simon, Johannes A. Lercher
Technische Universität München, Department of Chemistry, Lichtenbergstraße 4, 85747 Garching, Germany
Sabine Wrabetz, Friederike C. Jentoft
Department of Inorganic Chemistry, Fritz-Haber-Institut der Max-Planck Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany
∗
Cornelia Breitkopf, Silke Matysik, Helmut Papp
Universität Leipzig, Institute of Technical Chemistry, Linnéstr. 3, 04103 Leipzig, Germany
Received 22 September 2004; revised 24 November 2004; accepted 26 November 2004
Available online 13 January 2005
Abstract
Two sulfated zirconia catalysts were prepared via sulfation and calcination of zirconium hydroxide at 873 K; the zirconium hydroxide had
been aged at room temperature for 1 h (SZ-1) or aged at 373 K for 24 h (SZ-2). SZ-1 was active for n-butane isomerization at 373 K; SZ-2
reached a similar performance level only at 473 K. Both materials contained about 9 wt% sulfate and were tetragonal. Because of a lower BET
2
surface area (105 vs. 148 m /g) SZ-1 featured a higher sulfate density, and XRD and EXAFS analysis showed larger (ca. 10 nm) and more
well ordered crystals than for SZ-2. n-Butane TPD on SZ-1 showed a butene desorption peak at low temperature, whereas no obvious butene
desorption was observed with SZ-2, suggesting that SZ-1 has a higher oxidizing power at low temperature than SZ-2. The number of sites
capable of dehydrogenation are less than 5 µmol/g, because the differential heats of n-butane adsorption as measured by microcalorimetry
were 45–60 kJ/mol for higher coverages, indicating weak and reversible interaction. TAP experiments describe the adsorption and desorption
behavior of n-butane at different activity states and are the basis for a simple adsorption model. Reactant pulses and purge experiments show
that the active species, presumably formed in an oxidative dehydrogenation step, are stable at the surface under reaction conditions.

2004 Elsevier Inc. All rights reserved.
Keywords: Sulfated zirconia; Alkanes; Isomerization; Active sites; Reaction intermediates; Transient measurements
1
. Introduction
sulfation [5–7], the sulfation agent [8,9], or the calcination
conditions [10–14].
It was claimed that an active sulfated zirconia can only
be prepared from amorphous zirconia hydroxide [15], ob-
tained from hydrolysis of zirconium salts, that is, ZrOCl2,
ZrCl4, or ZrO(NO3)2. Sulfation of crystalline zirconia was
claimed to result in inactive sulfated zirconia. Recently, how-
ever, the preparation of active sulfated zirconia by sulfation
of crystalline zirconia has also been reported [16,17]. The
calcination of sulfated amorphous zirconium hydroxide in
the temperature range 723–923 K always yielded predomi-
nantly the metastable tetragonal zirconia phase, which was
Sulfated zirconia, capable of converting alkanes to their
isomers at low temperatures, attracted significant interest in
recent years [1–4]. Even though numerous investigations fo-
cused on the preparation of sulfated zirconia, the essential
procedure is still unclear, since many factors significantly af-
fect the resultant sulfated zirconia, such as the precursor for
*
Corresponding author.
E-mail address: papp@sonne.tachemie.uni-leipzig.de (H. Papp).
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2004.11.045

X. Li et al. / Journal of Catalysis 230 (2005) 214–225
215
believed to be the catalytically active phase [18–20]. In spite
of this, monoclinic sulfated zirconia was also recently shown
to be active for n−butane isomerization [21,22], but the ac-
tivity was lower than that of tetragonal zirconia.
tered with a 0.45-µm filter. The sulfate content was measured
by ion chromatography (Metrohm; 690 ion chromatograph
equipped with an IC anion column). The BET surface area
of sulfated zirconia was determined with a PMI automated
BET-sorptometer at 77.3 K with nitrogen as the analysis gas.
X-ray diffractometry (XRD) of the catalysts was performed
in transmission geometry on a STOE STADI-P X-ray dif-
fractometer equipped with a primary monochromator and
a curved position-sensitive detector with an internal resolu-
When Arata et al. [2,14] discovered the property of sul-
fated zirconia for butane isomerization at room temperature,
it was categorized as a superacid, since only superacids are
claimed to activate light alkanes at such a temperature. How-
ever, experiments using NMR [23,24], UV–vis [25], or IR
spectroscopy [26,27] revealed that the acidity of sulfated
zirconia is not higher than that of sulfuric acid. Alterna-
tively, alkane isomerization may be related to the generation
of olefins via oxidative processes [28–30]. In this context it
should be mentioned that several papers have demonstrated
the promoting effect of butenes for n-butane isomerization,
via formation of carbenium ions on the catalyst surface
◦
tion of 0.01 , with Cu-Kα radiation (λ = 1.542 Å). Samples
were mixed 1:1 with corundum, which was used as an in-
ternal standard. Nonsulfated samples were analyzed with a
Philips X’Pert-1, also with Cu-Kα radiation. Scanning elec-
tron microscopy (SEM) was conducted on a Hitachi S-4000
microscope equipped with a cold field emission gun at an ac-
celeration voltage of 5 kV. X-ray absorption spectra of the Zr
K-edge (17.998 keV) were recorded at beamline X1 at the
Hamburger Synchrotron Radiation Laboratory (HASYLAB)
with a Si(311) double crystal monochromator. The storage
ring was operated at 4.45 GeV with injection currents of
150 mA. X-ray absorption fine structure (XAFS) analysis
was performed with the software package WinXAS v2 [39].
To estimate the concentration of Brønsted and Lewis acid
sites, IR spectra for pyridine adsorption tests were obtained
at 373 K in vacuum with a Perkin–Elmer 2000 spectropho-
[
31–33].
In general, n-butane skeletal isomerization on sulfated
zirconia or other sulfated metal oxide shows an induction
period at low temperature, which is assumed to be due to
the formation of the active species [34]. After the activity
reaches its maximum, the catalyst deactivates, while coke is
being formed on the surface [35]. The rate constant of de-
activation decreased dramatically when alkenes in the feed
were eliminated via a suitable trap [36,37].
Here we use two sulfated zirconia materials of very dif-
ferent activities, but both tetragonal and equal in sulfate con-
tent, to discriminate on a finer scale between relevant and
irrelevant properties. In addition to extensive structural and
site characterization, we investigate the capacity of the ma-
terials to generate butenes and the dynamics of the catalytic
behavior with variations in reaction temperature and feed
composition on a minute to hour scale.
−1
tometer at 4 cm resolution. Samples were pressed into
wafers, heated to 673 K at 10 K/min, kept at 673 K for
−6
2 h (residual pressure of 10 hPa), and exposed to pyri-
dine (0.1 hPa) at 373 K for 60 min, followed by evacuation
at this temperature for 30 min. The number of sites available
for alkane adsorption was determined from measurement of
adsorption isotherms of propane, n-butane, and isobutane
(all Messer–Griesheim, 99.5%) at 313 K after outgassing at
7
23 K in vacuum. The differential heats of adsorption were
determined with a SETARAM MS-70 Calvet calorimeter.
The setup is described in detail in [40]. The error bars given
for the data result from the uncertainty in the heat generated
by valve opening, which contributes to the heat signal for
each dosing step.
2
. Experimental
2
.1. Catalyst preparation
Zirconium hydroxide powders were prepared by precip-
itation of ZrO(NO3)2 with NH4OH in water at pH 8.4 with
stirring. The resulting precipitate was aged in solution at
room temperature for 1 h followed by filtering, washing, and
drying at 373–393 K, denoted as Z-1. Alternatively, the pre-
cipitate was aged at 373 K for 24 h and processed as above;
this sample is denoted Z-2. We obtained sulfated zirconias
2.3. Temperature-programmed desorption of n-butane
The temperature-programmed desorption of n-butane
was performed with a vacuum system. The sulfated zirco-
nia sample was activated under vacuum at 673 K for 2 h and
cooled to 323 K for n-butane adsorption. n-Butane (2 hPa)
(Messer–Griesheim; 99.5%, purified with a H-Y zeolite trap
for alkene removal) was allowed to equilibrate with the sam-
ple for 5 min. The sample was heated to 873 K at 10 K/min.
The desorbing species were analyzed with a mass spectrom-
eter (QME 200, Pfeiffer vacuum).
(
(
SZ-1 and SZ-2) by sulfating the powders prepared above
Z-1 and Z-2) with (NH4)2SO4 (20 mol%) followed by cal-
cination at 873 K for 3 h in flowing air.
2
.2. Catalyst characterization
The sulfate contents (SO42 wt%) of the catalysts were
−
2.4. Isomerization of n-butane
determined by ion chromatography (IC) as described in the
literature [38]. For this, 0.02 g of sulfated zirconia was sus-
pended in a 0.1 N solution of NaOH, and the solution was fil-
n-Butane isomerization was carried out in a quartz micro-
tube reactor (8 mm i.d.) under atmospheric pressure. Sul-

216
X. Li et al. / Journal of Catalysis 230 (2005) 214–225
fated zirconia pellets (0.2 g, 315–710 µm) were loaded into
the reactor and activated in situ at 673 K for 2 h in flowing He
Samples were activated before use in TAP at 573 K for
2 h in vacuum in the microreactor and then cooled down
at 10 K/min to the reaction temperature. The fresh sam-
ples were exposed to single pulses of n-butane (Messer–
Griesheim; 99.5%) first. We conducted the flow experiments
by closing the slide valve of the TAP reactor and expos-
ing the microreactor to a constant flow of 2 ml n-butane
(Messer–Griesheim; 99.95%) in 18 ml helium. The reaction
products were analyzed by on-line GC (Shimadzu).
n-Butane was monitored for all possible fragments known
for C4 alkanes. Only their main mass fragments (m/e = 27,
29, 41, 43, 58) and their ratios are discussed in detail here.
Modeling of a simple first-order adsorption rate constant
was performed with modeling software by Schuurman and
Gleaves [41], which assumes a first-order Langmuir adsorp-
tion for the reactant in a one-zone model.
(10 ml/min). The catalyst was cooled to the desired reaction
temperature, and the reactant mixture (5% n-butane in He,
total flow of 20 ml/min) was passed through the catalyst bed.
We purified n-butane (Messer–Griesheim; 99.5%) of alkene
impurities by flushing it through a trap containing H-Y zeo-
lite. Alkenes were not detected in the reactant mixture after
purification. The reaction products were analyzed with an
on-line HP 5890 gas chromatograph (GC) equipped with a
capillary column (Plot Al2O3, 50 m × 0.32 mm × 0.52 mm)
connected to a flame ionization detector (FID). The rate of
isobutane formation was taken as a measure of catalytic ac-
tivity.
The influence of temperature on the catalytic behavior
of both catalysts was studied. After reaction at 373 K over
6
0 h, when SZ-1 was deactivated completely and SZ-2 had
not shown any activity, 20 ml/min of He flow was passed
through the catalyst bed for 30 min. The reaction was then
restarted after the temperature had been increased to 473 K.
Pulsing during the induction period of the reaction on
SZ-1 at 373 K was performed by alternately changing the
reactant (5% n-butane) and He flow every minute. A stop-
restart experiment was conducted on the SZ-1 catalyst at
3. Results
3.1. Catalyst characterization
BET surface areas, pore sizes, particle sizes, sulfate con-
tents, and acid concentrations of different samples are com-
piled in Table 1. The zirconium hydroxide samples without
thermal treatment exhibited the highest BET surface areas:
uncalcined Z-1, 288 m /g; uncalcined Z-2, 342 m /g. Cal-
cination at 873 K for 3 h converted zirconium hydroxide to
zirconia and caused a significant BET surface area decrease:
calcined Z-1, 49 m /g; calcined Z-2, 76 m /g. Calcination
at 873 K for 3 h of zirconium hydroxide doped with sulfate
also resulted in a loss of BET surface area. However, the cal-
cined sulfate-doped samples had higher BET surface areas
3
73 K during the deactivation phase. At the end of the in-
duction period, the catalyst bed was purged with He for a
certain time, and then the reaction was restarted. This proce-
dure was repeated several times.
2
2
2
2
2
.5. Temporal analysis of products switch experiments with
n-butane
We probed surface sites and their reactivity in a TAP-
2
II reactor [41] by pulsing n-butane over the fresh catalysts
and over catalysts that had been used in isomerization in a
flow experiment at 423 K. Pulse responses were analyzed
with a quadrupole mass spectrometer (QMS) from HIDEN,
where single masses (m/e) of interest can be followed. The
reactor was filled with a corundum/catalyst mixture analo-
gous to one used in atmospheric flow experiments. For SZ-1,
than the calcined undoped samples: SZ-1, 105 m /g; SZ-2,
2
148 m /g (twice that of calcined Z-1 and Z-2, respectively).
The pore and particle size distributions of the calcined sam-
ples indicate that the calcined sulfate-doped and undoped
zirconia samples were all mesoporous materials with irreg-
ular pores. The difference in the morphologies of SZ-1 and
SZ-2 becomes evident from scanning electron microscopy
(see Fig. 1). The SZ-1 particles were typically 5–20 µm in
size; the surfaces exhibit smooth and stepped regions, and
often smaller particles were found to be adhering. SZ-2 con-
sisted predominantly of particles 1–5 µm in diameter, which
occasionally enveloped larger particles.
Fig. 2a shows the XRD patterns for SZ-1, SZ-2, Z-1, and
Z-2, which were all calcined at 873 K for 3 h in air. The cal-
cined Z-1 was monoclinic and calcined Z-2 was tetragonal,
which indicates that Z-1 was more sensitive to the thermal
treatment than Z-2. Both calcined sulfated materials exhib-
ited only reflections of the tetragonal phase. Normalization
to internal standard corundum demonstrates that the reflec-
tions of SZ-2 are lower in intensity and broader than those
of SZ-1 (Fig. 2b). The sizes of the crystalline domains as ob-
tained via the Scherrer equation were 9.9 nm for SZ-1 and
7.5 nm for SZ-2.
3
00 mg of catalyst was mixed with 700 mg of corundum,
whereas for SZ-2, which is more voluminous, 258 mg of
catalyst material was mixed with 604 mg of corundum. The
particle diameter for all materials was 0.1–0.315 mm. The
temperature was controlled by a thermocouple positioned in
the center of the bed.
Neon was used as the reference gas. Knudsen diffusion
in the reactor during pulse experiments was ensured by con-
trol of the pulse intensity of the pulses. The shape of the
response curves has to be independent from the pulse size.
Pulse sizes were then kept constant. Reactant mixtures were
used at a constant pulse width of 130 µs and mixed in the
blending tank with 200 hPa inert gas and 200 hPa reactant
gas. A single experiment consisted of several cycles of puls-
ing with signal averaging (in general 10 pulses; for m/e of
lower intensity up to 50) to improve the signal-to-noise ratio.

X. Li et al. / Journal of Catalysis 230 (2005) 214–225
217
Table 1
Preparation and properties of raw materials
Z-1 uncalcined
SZ-1 uncalcined
Z-2 uncalcined
SZ-2 uncalcined
Precursor
Aging
Drying
Precipitation from ZrO(NO ) with NH OH until pH 8.4
1 h at room temperature
After filtering and washing dried at 373–393 K
3
2
4
24 h at 373 K
–
Sulfation
–
(NH ) SO
(NH ) SO
4 2 4
4
2
4
2
13.4
n.d.
0 mol%
20 mol%
n.d.
n.d.
2
−
SO
(wt%)
–
288
–
342
4
2
BET surface area (m /g)
Z-1 calcined
SZ-1 calcined
Z-2 calcined
SZ-2 calcined
Calcination
BET surface area (m /g)
Pore size (nm)
Crystallite size (nm)
Flowing air, 873 K, 3 h
2
49
8.0
12.6
–
105
3.5
9.9
9.0
2.5
76
7.0
11.0
–
148
3.7
7.5
8.9
3.0
2
−
SO
SO
(wt%)
H O leached (wt%)
4
4
2−
–
–
2
Brønsted acid sites
(
mmol/g)
n.d.
n.d.
0.069
0.66
n.d.
n.d.
0.047
0.32
2
(µmol/m )
n-Butane adsorption
(mmol/g) at 1 kPa
0.062
0.51
0.056
0.38
2
(µmol/m ) at 1 kPa
n.d.: not determined.
(a)
(b)
Fig. 1. SEM images of (a) SZ-1 and (b) SZ-2.
These findings are corroborated by Zr K edge XAFS (see
Fig. 3). The radial distribution functions, consistent with
tetragonal or cubic zirconia, showed a lower intensity in the
second and higher maxima for SZ-2 than for SZ-1. These
higher shells are dominated by Zr–Zr contributions, and the
decreased intensity may indicate a reduction in the Zr–Zr
coordination due to smaller crystal domains. In addition, the
reduced intensity may indicate an increase in the disorder
of the structure, caused by strain, a small amount of a less
ordered crystalline, or amorphous ZrO2 phase.
3.2. Surface sites
The sulfate content of the uncalcined SZ-1 was 13.4%,
whereas calcination at 873 K for 3 h in air resulted in a
sulfate content of 9.0 wt%, which indicates that the high-
temperature treatment eliminated approximately 30% of the
sulfate groups. The sulfate contents of the resulting two cal-
cined sulfated zirconia samples, SZ-1 and SZ-2, were both
9% in SO42 wt%. This is higher than a hypothetical mono-
−
2
layer, if 0.31 nm is assumed to be the surface area occupied

218
X. Li et al. / Journal of Catalysis 230 (2005) 214–225
Fig. 4. Differential heats of adsorption of n-butane on SZ-1 and SZ-2 as a
function of coverage. Activation at 723 K in vacuum, adsorption at 313 K.
(a)
The Brønsted acid site concentration (determined by ad-
sorption of pyridine) of calcined SZ-1 was 0.069 mmol/g,
which is higher than the 0.047 mmol/g observed with cal-
cined SZ-2. Thus, SZ-1 provides 1.5 times as many Brønsted
sites as SZ-2. This ratio was also confirmed by impedance
spectroscopy, with which the relative concentrations of pro-
tonated pyridine molecules were also explored [43].
Alkane sorption sites on both materials were probed by
adsorption of propane and butanes. Adsorption isotherms
could be adequately described by a first-order Langmuir
model for low pressures. The curvature at higher pressures
could be reproduced with a higher order Langmuir model,
but a slightly better fit was obtained with a Freundlich
model. The differential heats decline with coverage, which
is considered in the Freundlich model, but the decline is very
gentle, so that the Langmuir model, which assumes uniform
heats, is still a reasonable match. The Freundlich isotherms
were used to estimate the number of sites covered at the typ-
ical alkane partial pressure of 5 kPa. Typical C4 coverages
at 5 kPa were between 85 and 160 µmol g . The difference
in the concentration of sites between the two catalysts, how-
ever, was not larger than differences arising through varia-
tion of the activation temperature. The heats of adsorption
decreased slightly with coverage. For all three alkanes on
(
b)
Fig. 2. XRD patterns of (a) calcined Z-1, Z-2, SZ-1, and SZ-2 and (b) cal-
cined SZ-1 and SZ-2 measured with internal standard. M, monoclinic
phase; T, tetragonal phase; *, internal standard corundum.
−1
−
1
both catalysts, differential heats of about 60 kJ mol were
−1
determined at a coverage of ca. 5 µmol g . The heats slowly
−1
−1
declined to about 45 kJ mol at a coverage of 30 µmol g .
The presence of a small amount of sites (i.e., < 5 µmol g 1)
with a different heat of adsorption cannot be excluded;
however, data scatter considerably at coverages below 2–
−
−1
5
µmol g . For the majority of sites, there were no signif-
icant differences between SZ-1 and SZ-2 in the differential
heats (see, for example, n-butane adsorption, Fig. 4).
Fig. 3. Uncorrected radial distribution function for Zr K-edge of SZ-1,
SZ-2, and an yttria-stabilized reference material.
3
.3. n-Butane temperature-programmed desorption
by a SO42 group based on its kinetic diameter [42]. The
−
Figs. 5a and b shows the TPD after n-butane was ad-
water-soluble sulfate fractions were 2.5 and 3.0 wt% for SZ-
sorbed at 323 K on sulfated zirconia samples SZ-1 and
1
and SZ-2, respectively.
SZ-2. Butene (m/e = 56) was the most abundant desorbing

X. Li et al. / Journal of Catalysis 230 (2005) 214–225
219
Fig. 6. n-Butane reaction rate versus time on stream at 373 K on (F) SZ-1
and (2) SZ-2.
Fig. 7. Arrhenius profile of n-butane isomerization on SZ-1 at temperature
range of 333–393 K.
Fig. 5. n-Butane TPD profiles (a) SZ-1 and (b) SZ-2, (Q) butene
m/e = 56), (F) butane (m/e = 58).
(
action is primarily isobutane with 95–96% selectivity. The
main by-products are propane and pentanes (n-pentane and
iso-pentane with a ratio of 1:4).
Fig. 7 shows the Arrhenius correlation of n-butane iso-
merization on SZ-1 at the reaction temperature from 333 to
393 K, where the maximum reaction rates were used as in-
trinsic activity. The apparent activation energy for n-butane
conversion on SZ-1 of 41.5 kJ/mol is in agreement with re-
sults reported previously [28,34,44].
Because SZ-2 showed no catalytic activity at 373 K for
n-butane isomerization, we switched to He after 40 h and
purged for 25 h; then we increased the temperature to 473
K and tested the catalytic performance again (see Fig. 8).
Even though SZ-2 is almost inactive at 373 K, it showed ac-
tivity for n-butane isomerization at 473 K, and the rate was
on the same order as that of SZ-1 at 373 K. The selectivity
for isobutane was 96–97% with SZ-2 at 473 K, and the ma-
jor by-products were propane and pentane. SZ-2 deactivated
only slowly.
species, with an intense peak at 430 K on SZ-1. However, no
obvious butene desorption peak was observed in n-butane
TPD on SZ-2 at this temperature. Much lower intensities
of butene and butane were observed in the gas phase up to
7
3
3
00 K for both catalysts, SZ-1 and SZ-2.
.4. Catalytic experiments at atmospheric pressure
.4.1. Activity of n-butane isomerization on SZ-1 and SZ-2
Fig. 6 shows the n-butane isomerization activity versus
time on stream (TOS) of SZ-1 and SZ-2 at 373 K. SZ-1
showed a much higher catalytic activity than SZ-2, which
was almost inactive at 373 K under the conditions used. An
induction period characterized by a steady increase in the
catalytic activity was observed during n-butane reaction on
SZ-1, followed by deactivation. The maximum conversion
for SZ-1 at 373 K was 0.9%. The product of n-butane re-

220
X. Li et al. / Journal of Catalysis 230 (2005) 214–225
Fig. 8. n-Butane reaction rate on SZ-2 versus time on stream at (F) 373 K
followed by reaction at (2) 473 K.
Fig. 10. n-Butane reaction rate on SZ-1 at 373 K (Q) continuous reaction
and (F) stop-restart reaction during deactivation period.
2
0, 60, 300 min of He purging, the catalyst showed high ac-
tivity for n-butane isomerization without signs of a period of
increasing activity. The average activity was nearly identical
to that observed during the continuous-flow mode opera-
tion. It is obvious that the active species, the carbenium ion
(
alkoxy group), is stable at 373 K on the catalyst surface and
is removed by He purge. However, the initial activity after
the restart was always higher than that at steady state, indi-
cating higher concentration of active species on the surface,
which decreases to steady-state values after a short time.
3
.5. TAP experiments on catalysts during continuous-flow
operation
Fig. 9. n-Butane reaction rate on SZ-1 at 373 K (2) continuous reaction
and (Q) pulse reaction during induction period (alternative pure helium and
reactant per 1 min).
TAP experiments were conducted intermittently with
flow experiments to investigate the adsorption and desorp-
tion during various stages of the life cycle of sulfated zirco-
nia.
3
.4.2. Pulsed reaction mode during the induction period
A pulse method was applied in this study to explore the
The TAP intensity profiles recorded at 423 K for frag-
ment 43 (main fragment of n- and isobutane) are presented
in Fig. 11. All measured data represent the overall response,
which is determined by adsorption, desorption, and reac-
tion [41]. As a single mass unit that can be used to follow the
formation of products does not exist, fragmentation ratios
are compared. Differences in the fragmentation patterns are
thought to represent different surface processes and indicate
indirectly the formation of products. Side product detection
is also difficult because of the low intensities and the super-
position of single mass fragments.
Relative to the inert materials corundum and unsulfated
zirconia, the profile for the fresh SZ-1 was lower in inten-
sity (not shown here). This consistently points to a strong
interaction of n-butane and/or isobutane with SZ-1. The res-
idence time (and hence the strength of interaction) increased
in the presence of surface sulfates, as indicated by a com-
catalytic behavior during the induction period. He and re-
actant were alternatively replaced every minute and passed
into the reactor for the first 60 min during n-butane reaction
at 373 K on SZ-1 (see Fig. 9). The overall reaction behav-
ior was identical to the reaction with continuous flow, which
indicates that the 1-min He purge did not affect the forma-
tion of active species. After the pulse reaction, a continuous
flow of reactant was used to test the catalytic behavior in the
deactivation period. It was also identical to the conventional
continuous-flow reaction.
3
.4.3. Stopping and restarting the reaction
We also investigated the stability of the active species on
the catalytic surface by stopping and restarting the reaction
during the deactivation phase. Fig. 10 shows that even after

X. Li et al. / Journal of Catalysis 230 (2005) 214–225
221
Table 2
Fragmentation pattern for SZ-1 and SZ-2 for fresh and deactivated catalysts
(ratios of different mass fragments)
43/20 58/20 43/58 27/20 29/20 41/20 42/20
SZ-1, fresh
SZ-1, 40 min 2.55
SZ-2, fresh 1.18
1.73
0.18
0.31
0.13
0.15
9.31
8.09
9.12
9.20
1.82
2.07
1.13
1.39
1.96
2.27
1.24
1.43
0.33
1.15
0.58
0.69
0.43
0.48
0.28
0.29
SZ-2, 40 min 1.40
Table 3
Comparison of residence times [s] for SZ-1 and SZ-2 for fresh and 40 min
used catalysts (m/e = 43); over corundum τ = 0.08 s, T = 423 K
SZ-1
SZ-2
Fresh
0 min catalysis
0.52
0.46
0.84
0.83
(a)
4
In the TAP experiment, three trends were observed with
increasing deactivation of SZ-1: (i) the intensity of the peaks
slowly recovered compared with the inert material and in-
creased (Fig. 11a), (ii) the residence times decreased, and
(
iii) the ratio of the m/e 43 to 58 fragments decreased (Ta-
ble 3).
It is now interesting to compare the observed fragmenta-
tion ratios for SZ-2 with the same experimental setup. The
measured response curves for SZ-2 (inactive under the con-
ditions chosen) are shown in Fig. 11b. The displayed m/e
4
3 responses were reduced in their intensities relative to an
inert material, too (inert material not shown). n-Butane is
also strongly adsorbed to the surface of the material. The
shape of the curves was broadened compared to SZ-1, and
the residence time for the alkane was much higher than that
observed for SZ-1. Evaluation of the online GC data revealed
that no significant conversion of n-butane to isobutane was
observed. This is also displayed by the constant fragmenta-
tion ratios of SZ-2 for the start reference values and the 40-
min fragmentation pattern. With time on stream the intensity
of the main fragments also increased, but their relations re-
mained constant. Only adsorption and desorption determine
the observed fragmentation ratios.
Modeling of the first step of the n-butane interaction with
the catalyst surface was done with a simple one-zone model
assuming first-order adsorption of the reactant. Comparison
of the modeled adsorption rate constants for the fresh and
used SZ-1 showed a significant decrease. As all parame-
ters in the modeling are the same, the only variable that
causes this decrease is the number of available adsorption
sites on the catalyst surface. Thus it can be concluded that
after 40 min of treatment of SZ-1 under flow conditions,
nearly half of the available adsorption places are blocked or
deactivated. It has to be noticed that in this simple model
no term for reaction to the product is included. For SZ-2 the
trend was opposite to that observed for SZ-1. After 40 min
of treatment the adsorption rate constant was increased by a
factor of 2.5, which can only be interpreted by assuming that
adsorption site centers have been created.
(
b)
Fig. 11. TAP response pulses for n-butane (m/e = 43) at 423 K for two
different reactivity states of the catalysts (a) SZ-1, (b) SZ-2.
parison of residence times between sulfated and nonsulfated
samples.
After pulsing in vacuum, the operation mode was switched
to flow conditions at atmospheric pressure for distinct time
intervals. SZ-1 converted initially about 15% n-butane to
isobutane and traces of propane and pentanes. With time
on stream, SZ-1 deactivated to about 2% conversion after
1
20 min. After 40 min, the flow was stopped, and it was
switched back to vacuum and n-butane pulses were admit-
ted again. The residence times and fragmentation patterns
discussed here are taken from pulsing after 40 min of treat-
ment in n-butane flow, when SZ-1 is still active.
Table 2 summarizes all calculated fragmentation ratios.
The comparison of the fragmentation ratios for SZ-1 (start
values as a reference) and SZ-1 (40 min) clearly shows that,
consistent with GC analysis, the conversion from n-butane
to isobutane can be followed by changes in the main frag-
ments 43, 27, 29, 41, and 58 here. All measured intensities
vary in their intensity/fragmentation ratios. Because of pos-
sible irreversible adsorption on the surface during pulsing
and flow, the mass balance is affected and may influence the
measured ratios as well.

222
X. Li et al. / Journal of Catalysis 230 (2005) 214–225
4
. Discussion
general, a higher sulfate density on the surface may lead
to more condensed structures, such as pyrosulfates. Hence,
more pyrosulfate configurations may exist on the surface of
SZ-1 than on the surface of SZ-2, which would explain the
higher activity of SZ-1.
4
.1. Prerequisites for active sites
It has been discussed for some time that an active sul-
fated zirconia catalyst must consist of the tetragonal phase,
which is metastable below 1300 K. Only a few authors have
claimed a monoclinic or a mixture of monoclinic and tetrag-
onal sulfated zirconia to be active for n-butane isomeriza-
tion [21,22]. In this study, however, two tetragonal sulfated
zirconia samples prepared from zirconium hydroxides with
different aging conditions exhibited very different catalytic
activities for the n-butane reaction at low temperature. The
tetragonal phase can thus be seen as a necessary but not
sufficient requirement for good catalytic activity, which is
consistent with the findings by Yang et al. [45].
The results confirm that highly active catalysts can be ob-
tained with tetragonal materials; however, not all tetragonal
materials are necessarily highly active. It is also required that
the exposed facet planes be of the right type and right qual-
ity to allow the formation of certain sulfate structures. It is
interesting in this respect that excellent promoters such as
manganese or iron are incorporated into the tetragonal phase,
causing a constriction of the lattice [47].
4.2. Dehydrogenation capacity of sulfated zirconia
Recent investigations by Li et al. [30] suggest that the
isomerization reaction is initiated via oxidative dehydro-
The initial step of butane skeletal isomerization has been
proposed to be the formation of a surface carbenium ion, but
the mechanism of the formation of such an active species
is still under debate. On a superacid catalyst, carbenium ions
can be formed directly by the protonation of alkanes through
Brønsted superacid sites and subsequent decomposition of
the alkanium ions under cleavage of H2 or by hydride ab-
straction from alkanes through Lewis superacid sites. How-
ever, the superacidity of sulfated zirconia has been in doubt
recently, and the formation of carbenium ions was ascribed
to the protonation of alkene impurities; the contribution of
this pathway has been confirmed by the dramatic decrease in
the isomerization activity after the removal of trace amounts
of butenes in the feed.
In a preceding study, the dehydrogenation capacity of sul-
fated zirconia was identified during n-butane isomerization
at 373 K by the detection of butene (with TPD), water (by
in situ IR spectroscopy), and SO2 [30]. The in situ genera-
tion of butenes has been shown to be the initial step for the
n-butane isomerization reaction. Such a step would be non-
catalytic if no pathways for reoxidation of the sulfur species
existed and irreversible if sulfur were lost. However, once
an activate intermediate has been formed, the isomeriza-
tion proceeds in cycles, including rearrangement and hydride
transfer with generation of new intermediates. Given a high
number of cycles per generated butene, a pseudo-steady-
state behavior can be observed.
The results for n-butane TPD on SZ-1 indicate that it is
able to oxidatively convert n-butane to butene at low tem-
peratures. However, for the catalyst SZ-2, which is inactive
for n-butane isomerization at low temperatures, a discrete
butene desorption peak was not observed. Thus we conclude
that the dehydrogenation capacity of SZ-1, which leads to
the in situ formation of butenes, is responsible for isomer-
ization activity at low temperatures.
Increasing the reaction temperature to 473 K can effec-
tively overcome the catalytic inactivity of SZ-2, which then
shows an activity comparable to that of SZ-1 at 373 K. The
activity is related to the oxidative dehydrogenation on SZ-
2 at 473 K, either as less active sites become active or as
genation of n-butane by a pyrosulfate group (S2O72 ). Ac-
cording to DFT calculations by Hofmann and Sauer [46],
pyrosulfate is a possible and stable structure at the tetrago-
nal (101) face. Following this hypothesis, two explanations
emerge for the lower reactivity of SZ-2 in comparison with
SZ-1, both based on the question of which circumstances
will favor the formation of pyrosulfate.
−
SZ-1 and SZ-2 differ in the size of the crystalline domains
(
XRD), and, moreover, the short-range order (< 8 Å) of their
structures is also decidedly different (Zr K-edge XAFS).
Such variations in the bulk structure may lead to variations
in the surface structure; more explicitly, the poorly crys-
tallized, maybe highly defective tetragonal phase of SZ-2
may not be a suitable substrate for pyrosulfate. The pyro-
sulfate configuration on the zirconia surface is tetradentate;
that is, the three oxygen atoms of the pyrosulfate are con-
nected to three surface zirconium atoms, and a fourth oxy-
gen is hydrogen-bridged to a water molecule adsorbed on a
fourth zirconium atom. While the fourth link can be con-
sidered to be flexible, the other three bonds require three
zirconium atoms arranged in an isosceles triangle. The lower
intensity of the second shell in the EXAFS analysis of SZ-2
indicates that there are fewer zirconium atoms with the next
zirconium neighbor at a distance corresponding to those of
the tetragonal phase, implying that the described triangular
configuration on the tetragonal (101) face is less abundant.
The pyrosulfate, which is already strained on a perfect (101)
face, may not be able to accommodate aberrations in the dis-
tances, whereas, for example, a bidentate monosulfate struc-
ture can.
The two samples SZ-1 and SZ-2 exhibit the same sul-
fate content, but different surface areas, resulting in different
sulfate densities on the surface. The sulfate content is higher
than required for a hypothetical monolayer coverage, that is,
5
.2 and 7.2 wt% for SZ-1 and SZ-2, respectively, if we as-
2−
2
sume that the SO4 group occupies an area of 0.31 nm cor-
responding to its kinetic diameter [42]. However, the mono-
layer coverage depends on the assumed sulfate structure. In

X. Li et al. / Journal of Catalysis 230 (2005) 214–225
223
−
surface sulfate structures rearrange with temperature. In this
context it should be emphasized that the heterogeneity of
sulfated zirconia could lead to a wide spectrum of active sites
with only a few very active ones being responsible for the re-
action at low temperatures.
If the reaction is limited in the generation of alkenes,
we conclude that at 373 K SZ-2 is not producing a suffi-
cient quantity of butene to start the reaction. Thus, the dif-
ference in the generation of alkenes is responsible for the
low activity of SZ-2. At 473 K the production of alkenes
is sufficient for both materials, and, hence, catalytic activity
is found on both. However, the rates of reaction still dif-
fer for the two materials by one order of magnitude (SZ-1,
n-butane [50]. The range of 45–60 kJ mol 1 for the adsorp-
tion of propane and isobutane on sulfated zirconia indicates
sites that interact more strongly with alkanes than those of
the zeolite HZSM-5. Data for n-butane and isobutane ad-
sorption on sulfated zirconia presented by González et al.
[51] show a more rapid decline of the heat with coverage to
−1
−1
values below 40 kJ mol at 30 µmol g coverage. The sur-
faces of SZ-1 and SZ-2 are obviously more homogeneous
than those of the catalysts in [51].
Heats of adsorption from calorimetry data can be com-
pared with those accessible through TAP data and can be
used to validate the transferability of TAP results, which are
obtained in the Knudsen regime (base pressure < 10 hPa),
to pressures in the hPa range. The heats of adsorption of n-
butane that result from van’t Hoff plots of the equilibrium
constants given by the ratios of the rate constants for ad-
sorption and desorption were similar for the two catalysts,
namely, about 52–53 kJ mol . The TAP response curve is
an average over all interactions of the molecules in the pulse
with the surface, and the obtained heat of adsorption should
correspond to an average value. The average heat estimated
−9
−2
−2
r = 20 × 10 µmol/(g s); SZ-2, r = 1 × 10 µmol/(g s)).
In this context it should be emphasized that the concentra-
tion of Brønsted acid sites is 50% higher on SZ-1 than on
SZ-2. This suggests that the dramatic differences between
the two samples are the result of a number of subtle effects
rather than one effect.
−1
4
.3. Adsorption sites and the nature of interaction with
−1
alkanes
from the calorimetric results is about 50 kJ mol , indicating
perfect agreement between the two techniques.
The difference in catalytic activity between the two sam-
The heats of adsorption are relatively small, and adsorp-
tion at 313 K is reversible within the detection limits of the
calorimeter (< 10%). The values obtained for the heats in-
dicate a weak type of interaction for the majority of the
alkane molecules, such as hydrogen bonding via induced po-
larization of the alkane. It has been reported that n-alkanes
interact with the OH groups of pure [52] and sulfated zirco-
nia [53]. The bands of sulfate were claimed not to be shifted
appreciably after n-butane adsorption [54]; after adsorption
of n-pentane the S=O band at 1416 cm was decreased
without the formation of additional bands. Thus the role of
sulfate in the adsorption process is not clear, but it appears
that alkanes adsorb on OH-groups.
ples is not reflected in the number of sites available for
alkane adsorption or in the differential heats. The projected
coverage under our reaction conditions (5 kPa) for n-butane
−
1
is 85–160 µmol g . The sulfate content of both materials is
−1
9
wt% or about 938 µmol g . If a site consisted of a sin-
gle sulfate group, only about 10–15% of all sulfate groups
would participate under our reaction conditions; for a pyro-
sulfate group, about 20–30% of the sulfur would be involved
in adsorption of the reactant. This confirms that not all sul-
fates are active and that the catalyst can be destroyed by the
removal of 40% of the sulfate [48], and it may explain why it
has been so difficult to identify any relevant sulfate structure
spectroscopically.
−1
Our results do not reveal an obvious difference in the
quality of the sites for reactant adsorption that would al-
low an explanation of the order-of-magnitude difference in
catalytic activities. The small heats of adsorption are con-
sistent with the observations made during n-butane TPD,
which did not show much desorption of butane because it
is mostly weakly and thus reversibly adsorbed and is re-
moved during evacuation before the TPD. However, the
presence of a small number of sites with a different heat of
adsorption cannot be excluded from calorimetry, and these
might be the sites that dehydrogenate butane during TPD to
give butene(s). The calorimetry results limit the number of
The differential heats of adsorption for all three alka-
nes on the two catalysts did not deviate significantly from
−
1
each other. Typically, about 60 kJ mol evolved at low
−1
coverages, and beyond 10 µmol g the heats dropped to
−1
4
2
(
(
5–50 kJ mol . These values correspond to only about
–3 times the heat liberated upon condensation of propane
− −1
1
18.8 kJ mol ), n-butane (22.4 kJ mol ), and isobutane
−1
21.3 kJ mol ) [49] and indicate a weak interaction.
The similarity of the heats of adsorption for C3 and C4
alkanes is explainable by an interaction with the surface that
does not encompass the entire molecule. An adsorption with
the n-alkane flat on the surface appears unlikely because of
the corrugated nature of the sulfated zirconia surface [46].
The pocket-like interstitial space between the surface sulfate
groups may accommodate only part of an alkane in an end-
on-like fashion.
−1
sites capable of butane dehydrogenation to 5 µmol g or
less.
4.4. Carbenium ions and/or alkoxy groups on the surface
at low temperature
For the adsorption of n-alkanes on the zeolite HSZM-5,
the heat of adsorption increases with the chain length; 32 and
An induction period was always observed during the n-
butane isomerization reaction on sulfated zirconia at low
−1
4
8 kJ mol were reported for the adsorption of propane and

224
X. Li et al. / Journal of Catalysis 230 (2005) 214–225
temperatures. It was concluded that during this period
alkoxy groups (carbenium ions) formed by protonation of
butene generated by oxidative conversion accumulated on
the catalytic surface (see also Ref. [30]). On sulfated zirco-
nia SZ-1, pulse and continuous-flow n-butane reaction at low
temperature showed an identical induction period. The num-
ber of molecules streaming through the catalyst bed in the
pulse experiment is only half of that in the continuous-flow
experiment, indicating that the accumulation of a species by
adsorption from the gas phase is not rate determining, as
shown previously by Cheung et al. [55]. Rather, changes to
the surface or slow conversion of already adsorbed species
must be relevant. Furthermore, purging with helium did not
affect the transition of the surface to a more active state.
We speculate that some of the adsorbed n-butane molecules
were retained on the surface because they have already been
transformed and no longer interact merely by the weak in-
teraction seen in the calorimetry and TAP experiments; that
is, they could be surface carbenium ions.
The stop-restart method was applied in the deactivation
period of n-butane isomerization. It is very clear that the
active species are stable on the catalyst surface during He
purge, even for 5 h. Thus, when the reaction was restarted,
the reaction rate in the first minute is much higher compared
with that of the continuous reaction, because a large frac-
tion has been converted to tert-butyl carbenium ions, and
these undergo preferred hydride transfer in this initial stage
of the reaction. This is also why the catalytic activity, in
the first minute after the reaction was restarted after 20 and
5. Conclusions
Two sulfated zirconia samples with different catalytic ac-
tivities were fully characterized and investigated with re-
spect to n-butane skeletal isomerization. The sulfated zirco-
nia prepared from zirconium hydroxide aged in a solution
at room temperature for 1 h showed much higher activity
than that prepared from zirconium hydroxide aged at 373 K
for 24 h. The stoichiometric oxidation (of butane to butene)
of these two samples is concluded to be the critical dif-
ference determining the variations in catalytic activity. The
SZ-2 sample exhibits catalytic activity when the reaction
temperature is raised to 473 K, which indicates that the dehy-
drogenation property of sulfated zirconia was dramatically
promoted by the higher temperature.
Since the two materials are tetragonal and have the same
sulfate content, the difference in reactivity is speculated to
originate from the difference in sulfate density on the surface
(consequence of the BET surface areas), from the difference
in the size and the defect structure of the zirconia crystals
(revealed by XRD and EXAFS), or from a combination of
these properties. The number of sites for butane oxidative
dehydrogenation must be very small (less than 5 µmol/g),
because microcalorimetry experiments prove the interaction
with C3 and C4 alkanes to be weak for the majority of sites.
Heats of adsorption for n-butane determined by calorimetry
were about 45–60 kJ/mol, consistent with estimations from
TAP data.
By analysis of the catalytic behavior of the surface ac-
tive species, the surface alkoxy group/surface carbenium ion,
with pulse and stop-restart methods on SZ-1 at 373 K, it was
clearly shown that the carbenium ion is very stable under the
conditions used, and the deactivation of sulfated zirconia is
due to the decay of carbenium ion species.
6
0 min of He purge, is higher than the maximum activity of
the continuous-flow reaction. During He purge, the surface
sec-butyl carbenium ion undergoes skeletal isomerization to
its isomer, the tert-butyl carbenium ion, in the absence of
n-butane. This suggests also that the skeletal isomerization
and not the hydride transfer step is the rate-determining step.
After the reaction was restarted following He purge for 5 h,
the reaction showed an induction period, although the active
species is still on the surface, indicating that the purge car-
ried out over a long time can remove a very small fraction of
the surface alkoxy groups/carbenium ions, probably via the
decomposition of carbenium ion to its corresponding olefin
and a proton.
Acknowledgments
The Deutsche Forschungsgemeinschaft is gratefully ac-
knowledged for supporting this project in the framework of
the priority program no. 1091, “Bridges between real and
ideal systems in heterogeneous catalysis.” The authors thank
E. Kitzelmann, R.E. Jentoft, T. Ressler, and the HASYLAB
staff at DESY for the XAS and XRD measurements; G.
Weinberg for the SEM measurements; and R. Schlögl, A.
Hofmann, and J. Sauer for fruitful discussions. MEL Chem-
icals kindly provided the Y-stabilized ZrO2.
In the deactivation period, without any butane flow, the
deactivation rate of the stop-restart reaction was nearly iden-
tical to the continuous-flow reaction (at least for 20 and
6
0-min purges), which implies that the deactivation is due
to the decay of the active species. The most plausible reac-
tion is the reaction between olefins desorbing and surface
carbenium ions forming larger surface intermediates, which
hamper the accessibility to the active sites. The larger sur-
face species also undergo slower hydride transfer. The rapid
deactivation limits the use of bare sulfated zirconia in in-
dustry applications. Thus, usually a noble metal function is
additionally introduced, and the reaction is carried out un-
der hydrogen to control the olefin concentration in the reac-
tor.
References
[
[
[
[
[
1] X.M. Song, A. Sayari, Catal. Rev. Sci. Eng. 38 (1996) 329.
2] K. Arata, Adv. Catal. 37 (1990) 165.
3] G.D. Yadav, J.J. Nair, Micropor. Mesopor. Mater. 33 (1999) 1.
4] V. Adeeva, H. Liu, B. Xu, W.M.H. Sachtler, Top. Catal. 6 (1998) 61.
5] T. Yamaguchi, K. Tanabe, Y.C. Kung, Mater. Chem. Phys. 16 (1986)
67.

X. Li et al. / Journal of Catalysis 230 (2005) 214–225
225
[
6] C. Sarzanini, G. Sacchero, F. Pinna, M. Signoretto, G. Cerrato, C.
Morterra, J. Mater. Chem. 5 (1995) 481.
[33] J.E. Tabora, R.J. Davis, J. Catal. 162 (1996) 125.
[34] A. Sayari, Y. Yang, X. Song, J. Catal. 167 (1997) 346.
[
[
[
7] C. Guo, S. Yao, J. Cao, Z. Qian, Appl. Catal. A Gen. 107 (1994) 229.
8] J.R. Sohn, H.W. Kim, J. Mol. Catal. 52 (1989) 361.
9] J.M. Parera, Catal. Today 15 (1992) 481.
[35] R. Ahmad, J. Melsheimer, F.C. Jentoft, R. Schlögl, J. Catal. 218 (2003)
365.
[36] K.B. Fogash, R.B. Larson, M.R. Gonzalez, J.M. Kobe, J.A. Dumesic,
J. Catal. 163 (1996) 138.
[
10] M.-T. Tran, N.S. Gnep, G. Szabo, M. Guisnet, Appl. Catal. A Gen. 171
1998) 207.
11] D. F a˘ rca s¸ iu, J.Q. Li, Appl. Catal. A Gen. 175 (1998) 1.
(
[37] H. Liu, V. Adeeva, G.D. Lei, W.M.H. Sachtler, J. Mol. Catal. A 100
[
[
(1995) 35.
12] C. Morterra, G. Cerrato, C. Emanuel, V. Bolis, J. Catal. 142 (1993)
[
38] P. Canton, R. Olindo, F. Pinna, G. Strukul, P. Riello, M. Meneghetti,
G. Cerrato, C. Morterra, A. Benedetti, Chem. Mater. 13 (2001) 1634.
39] T. Ressler, J. Synchrotron Rad. 5 (1998) 118.
349.
[
[
13] R.A. Comelli, C.R. Vera, J.M. Parera, J. Catal. 151 (1995) 96.
14] A. Hahn, T. Ressler, R.E. Jentoft, F.C. Jentoft, Chem. Commun. (2001)
[
[
40] L.C. Josefowicz, H.G. Karge, E.N. Coker, J. Phys. Chem. 98 (1994)
537.
8
053.
41] J.T. Gleaves, G.S. Yablonski, Y. Schuurman, Appl. Catal. A: Gen. 160
1997) 55.
42] N. Katada, J. Endo, K. Notsu, N. Yasunobu, N. Naito, M. Niwa,
J. Phys. Chem. B 104 (2000) 10321.
[
[
15] M. Hino, S. Kobayashi, K. Arata, J. Am. Chem. Soc. 101 (1979) 6439.
16] C. Morterra, G. Cerrato, G. Meligrana, M. Signoretto, F. Pinna, G.
Strukul, Catal. Lett. 73 (2001) 113.
[
[
(
[
[
[
[
17] C.R. Vera, J.M. Parera, J. Catal. 165 (1997) 254.
18] D. F a˘ rca s¸ iu, J.Q. Li, S. Cameron, Appl. Catal. A Gen. 154 (1997) 173.
19] D. F a˘ rca s¸ iu, J.Q. Li, Appl. Catal. A Gen. 128 (1995) 97.
20] F.R. Chen, G. Coudurier, J. Joly, J.C. Vedrine, J. Catal. 143 (1993)
[43] S. Matysik, C. Breitkopf, H. Papp, submitted.
[44] C.-Y. Hsu, C.R. Heimbuch, C.T. Armes, B.C. Gates, J. Chem. Soc.
Chem. Commun. 1645 (1992).
616.
[45] X. Yang, R.E. Jentoft, F.C. Jentoft, Catal. Lett., in press.
[46] A. Hofmann, J. Sauer, J. Phys. Chem. B 108 (2004) 14652.
[47] F.C. Jentoft, A. Hahn, J. Kröhnert, G. Lorenz, R.E. Jentoft, T. Ressler,
U. Wild, R. Schlögl, J. Catal. 224 (2004) 124.
[
[
21] W. Stichert, F. Schüth, J. Catal. 174 (1998) 242.
22] W. Stichert, F. Schüth, S. Kuba, H. Knözinger, J. Catal. 198 (2001)
277.
[
23] J.B. Nicholas, J.F. Haw, L.W. Beck, T.R. Krawietz, D.B. Ferguson,
J. Am. Chem. Soc. 117 (1994) 12350.
24] J. Zhang, J.B. Nicholas, J.F. Haw, Angew. Chem., Int. Ed. Engl. 39
[
[
48] X. Li, K. Nagaoka, J.A. Lercher, J. Catal. (2004), in press.
49] D’Ans Lax, Taschenbuch für Chemiker und Physiker (Handbook for
Chemists and Physicists), Organic Compounds, vol. II, fourth ed.,
Springer, Berlin, 1983.
[
(2000) 3302.
[
[
[
25] B.S. Umansky, K. Hall, J. Catal. 124 (1990) 97.
26] F. Babou, G. Coudurier, J.-C. Vedrine, J. Catal. 152 (1995) 341.
27] L.M. Kustov, V.B. Kazansky, F. Figueras, D. Tichit, J. Catal. 150
[50] F. Eder, M. Stockenhuber, J.A. Lercher, J. Phys. Chem. B 101 (1997)
5414.
[51] M.R. González, K.B. Fogash, J.M. Kobe, J.A. Dumesic, Catal. To-
day 33 (1997) 303.
(1994) 143.
[
[
[
28] K.T. Wan, C.B. Khouw, M.E. Davis, J. Catal. 158 (1996) 311.
29] D. F a˘ rca s¸ iu, A. Ghenciu, J.Q. Li, J. Catal. 158 (1996) 116.
30] X. Li, K. Nagaoka, L.J. Simon, A. Hofmann, J. Sauer, J.A. Lercher,
J. Chem. Phys. B (2004), submitted for publication.
31] J.E. Tabora, R.J. Davis, J. Am. Chem. Soc. 118 (1996) 12240.
32] S. Hammache, J.G. Goodwin Jr., J. Catal. 211 (2002) 316.
[
[
52] B. Klose, Thesis (Diplomarbeit), TU München 2001.
53] E.A. Paukshtis, N.S. Kotsarenko, V.P. Shmachkova, Catal. Lett. 69
(2000) 189.
[
[
54] F. Babou, G. Coudurier, J.C. Vedrine, J. Catal. 152 (1995) 341.
55] T.-K. Cheung, J.L. d’Itri, B.C. Gates, J. Catal. 151 (1995) 464.
[
[