Thermolytic molecular precursor routes to Cr/Si/Al/O and Cr/Si/Zr/O catalysts for the oxidative dehydrogenation and dehydrogenation of propane

Article abstract of DOI:10.1016/S0021-9517(03)00141-6

The use of (^tBuO)₃CrOSi(O^tBu)₃ for the generation of high surface area Cr/Si/Al/O and Cr/Si/Zr/O catalytic materials via cothermolyses with [Al(O^tBu)₃]₂ or Zr[OCMe₂Et]₄ in a nonpolar solvent (n-octane) was studied. The materials have high surface areas after calcination at 773 K and did not display phase separation of Cr₂O₃ domains after calcination at 1473 K. The as-synthesized Cr/Si/M materials contained chiefly octahedral Cr³⁺ species with small quantities of tetrahedral Cr⁶⁺ chromates. Only Cr⁶⁺ monochromate centers, with smaller quantities of di- and polychromate species, were observed after calcination at 773 K under O₂. For propane oxidative dehydrogenation, the best results were associated with chromate clusters, while for propane nonoxidative dehydrogenation, well-dispersed (and presumably isolated) chromium species were most effective.

Full text of DOI:10.1016/S0021-9517(03)00141-6

Journal of Catalysis 218 (2003) 123–134
www.elsevier.com/locate/jcat
Thermolytic molecular precursor routes to Cr/Si/Al/O and Cr/Si/Zr/O
catalysts for the oxidative dehydrogenation and dehydrogenation
of propane
Kyle L. Fujdala and T. Don Tilley ∗
Department of Chemistry, University of California, Berkeley, Berkeley, CA 94720-1460, USA
Chemical Sciences Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720-1460, USA
Received 29 October 2002; revised 10 March 2003; accepted 14 March 2003
Abstract
t
t
t
Cothermolyses of ( BuO) CrOSi(O Bu) (1) and either [Al(O Bu) ] (2) or Zr[OCMe Et] (3) as solutions in n-octane resulted in
3
3
3 2
2
4
formation of gels that yielded Cr/Si/Al/O or Cr/Si/Zr/O xerogels upon drying. These materials have high surface areas after calcination at
2
−1
773 K (ranging from 150 to 450 m g ) and do not exhibit phase separation of Cr O domains after calcination at 1473 K (by PXRD). The
2
3
3+
6+
as-synthesized Cr/Si/M materials contain mostly octahedral Cr species with small amounts of tetrahedral Cr chromates (by DRUV–
vis and XANES spectroscopies). Only Cr monochromate centers, with smaller amounts of di- and polychromate species, were detected
6+
after calcination at 773 K under O . These catalysts are active for the oxidative dehydrogenation (ODH) of propane; however, they exhibit
2
poor selectivities for propylene production. In contrast, these catalysts exhibit high selectivities for propylene (> 95%) in the nonoxidative
dehydrogenation (DH) of propane with conversions for the best catalyst above 35% at 723 K.
2003 Elsevier Inc. All rights reserved.
Keywords: Molecular precursor; Cothermolysis; Propane oxidative dehydrogenation; Propane dehydrogenation; Chromium; Oxide catalyst
1. Introduction
be controlled to some extent by careful modification of the
precursor structure, or by the prehydrolysis of one precur-
sor prior to addition of a second, use of this process for
the construction of homogeneous multicomponent oxides
is still rather limited [1–14]. Alternative low-temperature
methods for the synthesis of complex heteroelement oxides
are required for the production of highly homogeneous and
potentially more efficient catalytic materials.
In developing new routes to homogeneous materials,
we have investigated the structure, bonding, and reactivity
of tris(tert-butoxy)siloxy complexes of the form LnM[OSi
(O t Bu)3]x [15–26]. These oxygen-rich compounds serve
as convenient molecular precursors to carbon-free, homo-
geneous M/Si/O materials in low-temperature thermolytic
conversions (typically below 473 K), via the elimination of
isobutylene and water. Solution thermolyses of these precur-
sors in nonpolar media (toluene, n-octane, etc.) give rise to
gels that ultimately form high surface area xerogels upon
conventional drying. This thermolytic molecular precursor
(TMP) route to multicomponent oxides has been utilized
for the production of a variety of homogeneous metal-
An important goal in the synthesis of new heteroge-
neous catalysts is the generation of materials containing
well-dispersed catalytic centers within a support matrix. Ad-
ditionally, efficient catalysts should be highly homogeneous
to ensure uniformity in properties. Significantly, multi-
component oxide materials with isolated (well-dispersed)
catalytic centers are often metastable, requiring use of low-
temperature synthetic procedures that provide a degree of
atomic-level control over the resulting active sites [1–14].
The sol–gel process has been widely employed for the low-
temperature preparation of oxide materials [1–14]. While
this method has proven quite successful for the preparation
of relatively simple oxide materials, it is not generally suit-
able for generating homogeneous binary and higher order
multicomponent oxides due to the different rates of hydrol-
ysis for each sol–gel precursor. Although this problem can
*
Corresponding author.
E-mail address: tdtilley@socrates.berkeley.edu (T.D. Tilley).
0021-9517/03/$ – see front matter
doi:10.1016/S0021-9517(03)00141-6
2003 Elsevier Inc. All rights reserved.

124
K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
[48–51], and TMP-derived V/Zr/O catalysts perform at lev-
els among the very best of these [29,30]. A recent report
claims that Cr/Al2O3 catalysts are moderately active and se-
lective for propane ODH, although their performance was
below that of many vanadium-based catalysts [33].
Given the reported efficiency of Cr-based dehydrogena-
tion catalysts, it was of interest to utilize the TMP cother-
molysis method for the generation of such materials with
high surface areas and highly dispersed active sites. We
previously reported synthesis of the molecular precursor
(tBuO)3CrOSi(O t Bu)3 (1) and demonstrated its efficient
conversion to high surface area (300 m2 g−1) Cr/Si/O ma-
terials via TMP methods [23]. Herein, we describe the use
of 1 for the generation of high surface area Cr/Si/Al/O
and Cr/Si/Zr/O catalytic materials via cothermolyses with
[Al(O t Bu)3]2 (2) or Zr[OCMe2Et]4 (3) in a nonpolar sol-
vent. The efficiency of these materials for the dehydro-
genation of propane was tested under oxidative (ODH) and
non-oxidative (DH) conditions, and the results of these in-
vestigations are presented here.
Scheme 1.
silicon-oxide materials with tailored properties [15–26]. The
TMP method offers several advantages over traditional syn-
thetic procedures for catalyst syntheses. These include use
of well-defined species for control over the stoichiometry,
the preexistence of M–O–E linkages to maximize the ho-
mogeneity, and the use of nonpolar solvents to reduce pore
collapse upon drying. Other groups have also developed
nonaqueous routes to oxide materials, including the conden-
sation of metal alkoxides with metal acetates [27] or metal
halides [28].
We have recently extended the TMP method to the cother-
molysis of precursor complexes [29,30]. This cothermolytic
method has allowed the syntheses of catalytic V/Si/Zr/O
and V/Zr/O materials via thermolyses of solutions contain-
ing OV[OSi(O t Bu)3]3 and Zr[OCMe2Et]4, or OV(O t Bu)3
and Zr[OCMe2Et]4, respectively. By altering the ratio of
precursors, this method allows the synthesis of high surface
area, uniform materials with a wide range of predetermined
compositions [29,30].
2. Experimental details
2.1. General
All synthetic manipulations were performed under an at-
mosphere of N2 using standard Schlenk techniques and/or
a Vacuum Atmospheres drybox, unless noted otherwise.
Solvents were distilled from sodium/benzophenone, potas-
sium/benzophenone, sodium, or calcium hydride, as appro-
priate. Thermal analyses were performed using a TA Instru-
ments SDT 2960 integrated thermogravimetric/differential
thermal analyzer (TGA/DTA) with a heating rate of 10 K
min−1 under N2 or O2, as appropriate. N2 porosimetry
measurements were performed using a Quantachrome Au-
tosorb surface area analyzer with samples heated at 393 K,
under reduced pressure, for a minimum of 4 h immedi-
ately prior to data collection. PXRD data were collected
using a Siemens D5000 diffractometer operating with θ–
2θ geometry at room temperature (Cu-Kα radiation, 1.5 s
collection time, and 0.02◦ step size). DRUV–vis spectra
were acquired using a Perkin-Elmer Lambda-9 spectropho-
tometer equipped with a 60-mm integrating sphere (ratio of
apertures to sphere surface =∼ 8%), a slit width of 4 nm,
and a collection speed of 120 nmmin−1. Samples were
run without dilution in equal quantities (10 mg each) us-
ing BaSO4 as a reference material. XAS measurements were
performed at the Stanford Synchrotron Radiation Laboratory
(SSRL) using beamline 2-3. X-ray detection was performed
in transmission mode using a Si (111) double crystal mono-
chromator and two ion chambers. Data were collected at
298 K on dried samples sealed under an atmosphere of N2
and diluted with BN. The XANES data were analyzed us-
ing WinXas 97 software. TEM images were recorded using
a JEOL 200cx transmission electron microscope operating at
The conversion of propane to propylene is an important
industrial process, and the demand for propylene is rapidly
increasing given its use in the production of propylene oxide,
acrylonitrile, and polypropylene [31–47]. Currently, propy-
lene is primarily produced by the high temperature cracking
of hydrocarbons; however, the controlled dehydrogenation
of propane provides a useful alternative process [31–47]
(Scheme 1). The endothermic dehydrogenation (DH) of
propane, currently employed on an industrial scale, utilizes
Cr-based catalysts with Al2O3 as a support and alkali metal
promoters [32]. Unfortunately, these propane DH reactions
require high temperatures (up to 873 K) which lead to crack-
ing and the deposition of carbon. This reduces the overall
efficiency of the catalysis and requires a catalyst regener-
ation step [32–47]. In studies on various chromium-based
catalysts, DeRossi and co-workers found that Cr/ZrO2 cat-
alysts exhibited increased activity over similarly prepared
Cr/Al2O3 and Cr/SiO2 materials [42–45]. However, it was
also concluded that the practical use of these Cr/ZrO2 cat-
alysts would require an increase in their surface areas and
greater thermal stability under the reaction conditions [42].
The oxidative dehydrogenation (ODH) of propane rep-
resents an alternative synthesis for propylene that can be
performed at lower temperatures and does not suffer from
coking [31,33,38,39,48–51]. However, the tendency of the
propylene product to overoxidize is a significant obstacle for
the commercialization of this process, and therefore the de-
velopment of more efficient catalysts is highly desired. Many
of the best propane ODH catalysts are based on vanadium

K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
125
200 kV. Samples for TEM analysis were prepared by depo-
sition using a pentane suspension of finely ground material
on a “Type A” carbon-coated copper grid obtained from Ted
Pella, Inc. Elemental analyses were performed by Galbraith
Laboratories (Knoxville, TN). (t BuO)3CrOSi(O t Bu)3 (1)
was prepared as previously reported [23], [Al(O t Bu)3]2 (2)
was purchased from Aldrich and purified by sublimation,
and Zr[OCMe2Et]4 (3) was purchased from Gelest and puri-
fied by distillation.
analyzed online using a Hewlett-Packard 6890 gas chro-
matograph equipped with both a capillary column (HP-1)
and a packed column (HAYESEP-Q). C3H6, CO, CO2, and
H2O were detected as the ODH reaction products and CH4,
C2H4, C2H6, and C3H6 were the DH reaction products de-
tected after ∼ 30 min of reaction time. Immediately after
catalyst exposure to propane under DH conditions, some
COx species were detected, presumably due to initial ODH
activity upon reduction of the Cr6+ species [32]. After
∼ 30 min of reaction time no further CO2 was observed
under DH conditions; however, small amounts of CO were
observed throughout (< 1%), presumably as a result of par-
tial oxidation by lattice oxygen. The ODH reactions were
run at 606 and 673 K while the DH reactions were run at
723 or 763 K. Typical propane conversions ranged from ∼ 1
to 10% for the ODH reactions and from ∼ 0.5 to 38% for
the DH reactions. Oxygen conversions in the ODH reactions
ranged from 1 to 15% at 606 K and from 11 to 69% at 673 K.
2.2. Syntheses of Cr/Si/M/O xerogels
A representative nCrSiAl xerogel preparation: A solu-
tion of 1 (0.107 g, 0.2 mmol) and 2 (0.443 g, 1.8 mmol) in
n-octane (5.0 mL) was sealed in a thick-walled Pyrex pyrol-
ysis tube after three freeze-pump-thaw cycles. The tube was
then placed in a preheated oven (453 K) for 48 h. The light
green monolithic gel that resulted was washed with n-octane
(3 mL), air dried for 2 days at 298 K, and dried in vacuo at
398 K for 12 h yielding a xerogel (10CrSiAl). Xerogel cal-
cinations were performed in a Lindberg 1473 K three-zone
tube furnace under flowing O2 (100 mLmin−1) with a heat-
ing rate of 10 Kmin−1 and a 2 h isothermal stage at the final
temperature. For other nCrSiAl samples, amounts of 1 and
2 were varied such that the total concentration of the solu-
tion was maintained at 0.4 M. The nCrSiZr xerogels were
prepared using solutions of 1 and 3 in n-octane (4 M) in an
analogous fashion. For the preparation of pure Al2O3 and
ZrO2 samples, solutions of 2 or 3 in n-octane (0.4 M) were
treated as above except that the solution of 3 was heated at
498 K to facilitate thermolysis.
3. Results and discussion
3.1. Thermolytic molecular precursor route to Cr/Si/M/O
xerogels
Thermolyses of n-octane solutions containing (tBuO)3
CrOSi(O t Bu)3 (1) and either [Al(O t Bu)3]2 (2) or Zr[OCMe2
Et]4 (3) at 453 K resulted in the formation of gelatinous pre-
cipitates that, upon drying in air, formed xerogels [nCrSiAl
and nCrSiZr, where n = mol% of 1 used versus 2 or
3; Eqs. (1) and (2)]. The independent solution thermol-
yses of 2 and 3 (at 453 and 498 K, respectively) gave
gels that, upon drying, provided Al2O3 and ZrO2 xerogels.
Note that the thermal decomposition of 3 occurs at a lower
temperature upon cothermolysis with 1, suggesting that 1
or its decomposition products catalyze the conversion of
3. Similar behavior was observed for cothermolyses of 3
with OV[OSi(O t Bu)3]3 or OV(O t Bu)3 [29,30]. Xerogels
were synthesized with 2.5 (2.5CrSiAl and 2.5CrSiZr), 5
(5CrSiAl and 5CrSiZr), and 10 (10CrSiAl and 10CrSiZr)
mol% of 1 (Table 1).
2.3. Catalytic dehydrogenation of propane
Selectivities and conversions for the propane to propy-
lene reactions were measured using a fixed-bed flow reactor
(quartz) fitted with a medium frit. Quartz chips (0.500 g,
0.246–0.494 mm) were used to disperse each catalyst
within the reactor. The mass of catalyst used varied be-
tween 0.010 and 0.260 g as noted in Tables 3–5. The gas
stream for the ODH reactions consisted of helium (typ-
ically 200 mLmin−1), propane (25 mLmin−1), nitrogen
(2 mLmin−1), and oxygen (9 mLmin−1). Propane and oxy-
gen conversions were varied by changing the total flow rate
(between 40 and 240 mLmin−1). Typical DH reactions used
a gas stream consisting of helium (∼ 32 mLmin−1), propane
(∼ 2 mLmin−1), and nitrogen (10% of the propane flow).
Dilution was achieved by increasing the amount of He and
residence times were varied via changes in the total flow rate.
For ODH reactions, the catalysts were calcined at 773 K
under He/O2 (∼ 10/1; ∼ 200 mLmin−1) for 2 h prior to
reaction. For DH reactions, the catalysts were calcined at
773 K for 2 h under He/O2 (∼ 10/1; ∼ 200 mLmin−1), and
subsequently for 2 h under He (∼ 200 mLmin−1). For ODH
and DH reactions, the gaseous reactants and products were
(tBuO)3CrOSi(OtBu)3 (1) + [Al(OtBu)3]2 (2)
→
453 K, toluene nCrSiAl,
(1)
where n = 2.5, 5, and 10 mol% 1.
(tBuO)3CrOSi(OtBu)3 (1) + Zr[OCMe2Et]4 (3)
→
453 K, toluene nCrSiZr,
(2)
where n = 2.5, 5, and 10 mol% 1.
The gels were dried at room temperature in air to give
xerogels as pale-green powders. The as-synthesized Al2O3
and ZrO2 xerogels from 2 and 3 were colorless pow-
ders. Calcination of the Cr/Si/M/O xerogels under O2
(100 mLmin−1) at 773 K resulted in materials that main-
tained visual homogeneity with colors ranging from light to

126
K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
Table 1
Properties of the xerogels after calcination at 773 K
b
Sample
Cr/Al or Cr/Zr
molar ratio
Cr content
(wt%)
Cr site density
Surface area
Pore volume
(cm g )
Average pore radius
(Å)
−2
2
−1 a
3
−1
(Cr nm
)
(m g
)
2.5CrSiAl
5CrSiAl
10CrSiAl
1:40
1:20
1:10
N/A
1:40
1:20
1:10
N/A
2.4
4.5
8.1
N/A
1.0
2.0
0.6
1.4
2.2
N/A
0.6
1.1
455
1.30
0.66
0.76
1.09
0.12
0.14
0.09
0.21
57.4
34.6
35.4
93.6
13.9
12.9
12.6
47.6
380
430
235
180
220
150
90
Al O
2
3
2.5CrSiZr
5CrSiZr
10CrSiZr
3.8
N/A
2.9
N/A
ZrO
2
a
2
−1
Rounded to the nearest 5 m g to reflect the error in measurement.
b
Calculated from the 2(pore volume)/(area) method, assuming spherical particles.
dark yellow, depending on the Cr content. The Zr-containing
materials were slightly darker in color than the correspond-
ing Al-containing ones. The pure Al2O3 and ZrO2 xerogels
maintained their colorless (white) nature after calcination at
773 K in O2.
3.2. Characterization of the xerogels
Nitrogen porosimetry was used to determine the surface
areas and pore volumes of the nCrSiAl and nCrSiZr xero-
gels. The Brunnaur–Emmett–Teller (BET) method [52] was
used to calculate the surface areas and the Barrett–Joyner–
Halenda (BJH) method was used to obtain the pore size
distributions [53]. Table 1 provides a summary of the nitro-
gen porosimetry data for all of the xerogels after calcination
at 773 K. The presence of 1 in the thermolysis mixture gives
rise to an increase in the surface area of the resulting xero-
gels relative to those of the pure Al2O3 and ZrO2 xerogels
obtained from 2 and 3, respectively. This effect is undoubt-
edly due to the presence of Si (and possibly Cr); however,
the surface areas do not correlate proportionately with the
amount of 1 present [18,21–26]. Also, the total pore volumes
do not vary with the Cr and Si contents. The nCrSiAl mate-
rials have significantly higher surface areas relative to their
nCrSiZr counterparts.
Transmission electron microscopy (TEM) was employed
to determine the morphology of the Cr/Si/M/O samples af-
ter calcination at 773 K under O2. In all cases, the samples
are composed of aggregates of primary particles (from 50 to
> 500 nm in diameter), and no crystalline domains are ap-
parent.
The crystallization behavior of the xerogels was moni-
tored as a function of calcination temperature by powder
X-ray diffraction (PXRD). All of the nCrSiZr xerogels are
amorphous after calcination at 873 K under O2; however,
after calcination at 1073 K crystalline ZrO2 (tetragonal) is
observed, as shown in Fig. 1a for 10CrSiZr. For compari-
son, the as-synthesized (uncalcined) ZrO2 xerogel exhibits
PXRD reflections attributed to both monoclinic and tetrago-
nal ZrO2 crystallites (Fig. 1b).
Fig. 1. Powder X-ray diffraction patterns at various calcination tempera-
tures (under O ) of: (a) 10CrSiZr and (b) ZrO from the thermolysis of
2
2
Zr[OCMe Et] . t, tetragonal ZrO ; m, monoclinic ZrO .
2
4
2
2
that the Cr species are well dispersed. The crystallization of
Cr2O3 has been previously observed at 1173 K (by PXRD)
for Cr/Zr/O samples of similar Cr contents (1–10 wt%)
but prepared by aqueous impregnation methods [54]. The
delay of crystallization to 1173 K was described as aris-
ing from well-dispersed surface Cr species. Therefore, the
lack of Cr2O3 crystallization at calcination temperatures up
to 1473 K in the nCrSiZr samples described here suggests
that the initially formed Cr species are dispersed to an even
greater degree. Similarly, all of the nCrSiAl xerogels are
amorphous in their as-synthesized state, and they remain
No Cr2O3 domains are observed for the nCrSiZr xero-
gels after calcination at 1473 K (by PXRD), which suggests

K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
127
Diffuse-reflectance UV–vis (DRUV–vis) spectroscopy of
the as-synthesized and calcined materials (773 K, O2) was
used to provide insight into the nature of the Cr species in
these materials. Previous DRUV–vis studies by Weckhuy-
sen and co-workers on the nature of Cr centers on various
support materials revealed four types of species [56–60].
Monochromate species exhibit absorption bands at ca. 275
and 370 nm; dichromate and polychromate species (indis-
tinguishable) give rise to bands at ca. 275, 320, and 445 nm;
pseudooctahedral Cr3+ species appear at ca. 295, 465, and
625 nm; and bands for pseudooctahedral Cr2+ species are
at ca. 800 nm [58]. As shown in Fig. 2a, the as-synthesized
(uncalcined) nCrSiAl samples all exhibit DRUV–vis spec-
tra with broad bands (centered at ca. 255, 380, and 435 nm)
characteristic of a ligand-to-metal charge transfer (LMCT)
from O2− to Cr6+ in mono, di-, and polychromates. Signif-
icantly, an intense band at ca. 595 nm suggests that most of
the chromium in these samples exists as octahedral Cr3+, es-
pecially given that d–d transitions are considerably weaker
than LMCT bands [56–60]. After calcination at 773 K un-
der O2, the DRUV–vis spectra of these nCrSiAl materials
exhibit intense bands at ca. 275 and 375 nm with a shoul-
der at ca. 440 nm. This suggests the presence of only
monochromate species, with smaller amounts of di- and
polychromates. Although the presence of small quantities
of Cr3+ species cannot be excluded, the DRUV–vis spec-
tra of the samples after calcination at 773 K do not indicate
the presence of such species (Fig. 2b). Similar behavior has
been observed for related chromium-containing oxide mate-
rials with similar Cr contents [33,56–60].
Fig. 2. Diffuse reflectance UV–vis spectra of 2.5CrSiAl, 5CrSiAl, and
10CrSiAl: (a) as-synthesized and (b) after calcination at 773 K under O
2
for 2 h.
The DRUV–vis spectra of the uncalcined (as-synthesized)
nCrSiZr materials all contain a major band at ca. 270 nm
with a broad shoulder extending out to 500 nm, indica-
tive of various chromate species (Fig. 3a). A small band
at ca. 615 nm, attributed to Cr3+ species, is also observed
for these samples. The DRUV–vis spectra of the calcined
nCrSiZr materials (773 K, O2) exhibit features consistent
with the presence of chromate species, and no Cr3+ species
were detected (Fig. 3b). Qualitatively, the color changes as-
sociated with calcination of the samples are consistent with
the DRUV–vis spectra, in that the green color of the as-
synthesized materials suggests the presence of Cr3+ species,
and the color change to yellow upon calcination at 773 K
is consistent with oxidation to Cr6+ [57]. The initial pres-
ence of significant amounts of Cr3+ species suggests that
the chemistry involved in the transformation to the solid
material primarily involves a Cr4+ to Cr3+ reduction. The
apparently large amounts of monochromate species (vs di-
and polychromates) for the nCrSiAl and nCrSiZr materials
after calcination at 773 K suggest that the samples contain
well-dispersed chromium species. The ability of Al2O3 and
ZrO2 to stabilize dispersed chromate species upon calcina-
tion under O2 at moderate temperatures is well documented
[56–60].
amorphous after calcination at 1473 K for 2 h (i.e., no Cr2O3
or Al2O3 domains are observed). This suggests that the Cr
species are also highly dispersed within these materials. The
high degree of homogeneity in the materials described here
can be attributed, at least in part, to use of the TMP approach
for their syntheses.
Thermogravimetric analyses (TGA) of the as-synthesized
xerogels, under calcination conditions (O2, 10 Kmin−1),
reveal 10 to 25% mass reductions below 873 K attributed
to loss of H2O (physisorbed and chemisorbed) and trace
residual organics. At temperatures greater than ca. 873 K,
no additional mass loss is observed. Differential scan-
ning calorimetry (DSC) studies of the nCrSiZr and pure
ZrO2 samples under calcination conditions (O2, heated at
10 Kmin−1) do not reveal any exothermic transitions (for
crystallization), and this suggests that the crystallization
of ZrO2 is slow under these conditions [18,54,55]. In-
terestingly, the DSC trace for 10CrSiAl reveals a single
small exotherm centered at 1333 K, attributed to mullite
(3Al2O3 ·2SiO2) crystallization from a highly homogeneous
aluminosilicate sample [25]. The apparent mullite domains
were not detected by PXRD, presumably due to their small
size and the relatively small amount of Si present in the
sample. No DSC transitions are observed for 2.5CrSiAl or
5CrSiAl.
To further probe the nature of the chromium centers in the
materials described here, chromium X-ray absorption near-

128
K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
Fig. 5. XANES spectra at the Cr K-edge of various reference materials.
peaks and octahedral species providing very weak peaks,
or none at all [61,62]. XANES analyses of 10CrSiAl after
calcination at 473, 673, and 773 K reveal the presence of in-
creasing amounts of tetrahedral Cr6+ species, as indicated
by the presence of a preedge feature that closely resembles
that for the tetrahedral Cr6+ references (Figs. 4 and 5). No
absorptions can be attributed to the presence of tetrahedral
Cr4+ or distorted octahedral Cr2+ species for the 10CrSiAl
samples. Octahedral Cr3+ species are not readily detected
by XANES measurements (see Cr2O3 in Fig. 5) [61]; how-
ever DRUV–vis spectroscopy revealed the presence of such
components as the dominant species in 10CrSiAl after cal-
cination at 473 K (but not 673 or 773 K). XANES spectra of
10CrSiAl after calcination at 1273 and 1473 K do not ex-
hibit a preedge feature and closely resemble the spectrum
for pure Cr2O3. This suggests that the Cr species in these
samples are predominantly octahedral Cr3+ in small Cr2O3
domains (undetected by PXRD). Chromium XANES spec-
tra of 2.5CrSiAl and 5CrSiAl after calcination at 773 K
are similar to that for 10CrSiAl after calcination at 773 K,
in that they exhibit significant amounts of tetrahedral Cr6+
species (Fig. 4).
Fig. 3. Diffuse reflectance UV–vis spectra of 2.5CrSiZr, 5CrSiZr, and
10CrSiZr: (a) as-synthesized and (b) after calcination at 773 K under O
2
for 2 h.
The first derivative of the XANES spectrum can be used
to identify the position of the main absorption edge, and
the correlation of this edge energy with the oxidation state
of Cr species has been previously reported [61]. It has
been observed that chromium species with higher oxidation
states exhibit absorption edges at higher energy [61]. Simi-
lar analyses of the XANES data reported here have provided
energies for the main absorption edges of the reference ma-
terials and the nCrSiAl samples (Table 2).
Fig. 4. XANES spectra at the Cr K-edge of 10CrSiAl (calcined at 473, 673,
773, 1273, and 1473 K under O ), and 2.5CrSiAl and 5CrSiAl calcined at
773 K under O .
2
2
edge structure (XANES) studies of the nCrSiAl samples
were performed (Fig. 4). XANES spectra were also acquired
for Cr2O3, CrCl2, Cr(O t Bu)4, (t BuO)3CrOSi(O t Bu)3,
CrO3, K2CrO4, and K2Cr2O7 as reference materials (Fig. 5).
The nature of the preedge feature in Cr K-edge XANES
measurements is indicative of specific coordination geome-
tries, with tetrahedral environments giving rise to intense
3.3. Oxidative dehydrogenation of propane
We have previously observed that propane ODH catalysts
prepared by the TMP method exhibit improved performance
relative to catalysts of the same composition but generated
by aqueous impregnation techniques [20,23,29,30,63]. We
therefore investigated the nCrSiAl and nCrSiZr materials

K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
129
Table 2
XANES data for Cr reference materials, 2.5CrSiAl, 5CrSiAl, and
10CrSiAl
Sample/calcination
temperature
Main edge
energy (keV)
Cr oxidation
state
Cr geometry
CrCl
5.9982
6.0033
6.0029
6.0031
6.0071
6.0068
6.0074
6.0042
6.0048
6.0057
6.0038
6.0034
6.0054
6.0058
(II)
Dist. O
2
h
d
d
Cr O
(III)
(IV)
(IV)
(VI)
(VI)
(VI)
O
2
3
h
t
Cr(O Bu)
T
4
d
t
t
( BuO) CrOSi(O Bu)
Dist. T
3
3
CrO
T
T
3
d
d
K CrO
2
4
K Cr O
Dist. T
7
2
2
10CrSiAl/473 K
10CrSiAl/673 K
10CrSiAl/773 K
10CrSiAl/1273 K
10CrSiAl/1473 K
2.5CrSiAl/773 K
5CrSiAl/773 K
Fig. 7. Intrinsic selectivities and activities of the nCrSiAl catalysts plot-
ted versus Cr content for the oxidative dehydrogenation of propane at 606
(circles) and 673 K (squares).
nature and higher thermal stability of the nCrSiAl catalysts
suggest that more Cr is present in the bulk and is not avail-
able for reaction, thus giving rise to lower activities when
compared to their nCrSiZr counterparts. The activity of
10CrSiZr at 673 K is among the highest values reported for
an ODH catalyst, despite its poor selectivity [29,30,50,51].
For all of the nCrSiZr and nCrSiAl catalysts, activities
increase with chromium content (Table 3). For compari-
son, previously reported Cr/Al2O3 catalysts prepared by wet
impregnation (5–20 wt% Cr and 2–10 Cr nm−2) exhibit
good selectivities (from ca. 60–85%) with moderate activ-
ities (from 4.3 to 11.5 mmol propane converted g−1
h
)
−1
cat
at 673 K with propane conversions from 2 to 6% [33].
Previously reported, TMP-derived V/Zr/O catalysts−e1xhibit
activities near 100 mmol propylene produced g−ca1t h with
intrinsic C3H6 selectivities approaching 90% at 673 K [30].
It has been shown that propane reacts with oxygen in
parallel and sequential steps over metal oxide catalysts
(Scheme 1) [50,51]. At low oxygen conversions, pseudo-
first-order rate coefficients for the propane ODH step (k1),
the combustion of propane (k2), and the combustion of
propylene (k3) can be obtained [50,51]. Kinetic analyses of
the propane ODH reactions at 606 K using the nCrSiAl
and nCrSiZr catalysts were performed (Table 4) according
to the procedures outlined by Iglesia, Bell, and co-workers
[50,51]. These studies reveal that the nCrSiZr catalysts ex-
hibit k2/k1 ratios (the rate of propane combustion versus
propane ODH) that decrease with increasing Cr content.
Interestingly, the rates of propane ODH (k1) and propane
combustion (k2) both increase with increasing Cr content;
however, a greater increase in k1 suggests that higher Cr con-
tents are beneficial (Table 4). Additionally, analysis of the
k3/k1 ratios for the nCrSiZr catalysts (the rate of propylene
combustion versus propane ODH) shows that lower values
are obtained for higher Cr contents. These kinetic analyses
show that the rate of propylene combustion (k3) is lower for
higher Cr contents. Among the nCrSiAl catalysts, 5CrSiAl
exhibits the lowest k2/k1 and highest k3/k1 ratios, suggest-
ing that effects other than Cr dispersion (i.e., the nature of
active sites) are also important in these systems [33].
Fig. 6. Intrinsic selectivities and activities of the nCrSiZr catalysts plot-
ted versus Cr content for the oxidative dehydrogenation of propane at 606
(circles) and 673 K (squares).
as catalysts for propane ODH. The ODH reactions were run
in a fixed-bed flow reactor using a large excess of propane
in the feed. The catalysts were calcined at 773 K under an
atmosphere of He/O2 (∼ 10/1) for a minimum of 2 h prior
to reaction. Data corresponding to low propane conversions
were collected to allow for the accurate determination of
intrinsic activities and selectivities by extrapolation to infi-
nite flow rate and 0% conversion, respectively. Bell, Iglesia,
and co-workers have shown that this method for analyz-
ing catalytic data is useful for comparing ODH catalyst
performances [50,51]. Figs. 6 and 7 display plots of the
intrinsic propylene selectivity (%) and the activity for propy-
lene formation (mmol propylene gc−a1t h−1) as a function of
Cr content, for ODH reactions run at 606 and 673 K using
the nCrSiZr (Fig. 6) and nCrSiAl (Fig. 7) catalysts. It is
apparent that the intrinsic selectivity for propylene is low
in all cases (< 30%) and that the ODH activity is signifi-
cantly higher for the nCrSiZr catalysts (Table 3). The high
activity in the nCrSiZr catalysts presumably results from
the crystalline nature of the ZrO2 support, which drives more
Cr species to the surface rendering them accessible for reac-
tion [30]. Despite their higher surface areas, the amorphous

130
K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
Table 3
Summary of the propane ODH reactions using the Cr xerogels
a
b
c
d
e
f
g
Catalyst
Mass (mg)
32
T (K)
Intrinsic activity
Intrinsic selectivity
Intrinsic TOF
Activity
Selectivity
TOF
CO /CO ratio
2
2.5CrSiAl
606
673
606
673
606
673
606
673
606
673
606
673
2.6
12.6
3.2
19.9
5.2
37.5
1.1
7.9
16.2
14.1
20.8
17.7
19.3
19.3
14.5
13.9
19.3
20.5
21.1
25.5
5.7
27.4
3.7
23.0
3.4
2.2
6.6
2.0
6.9
4.4
19.0
0.9
3.9
15.9
11.7
13.6
12.6
17.5
15.8
11.4
9.4
17.4
14.6
19.8
18.3
4.7
14.4
2.3
8.0
2.8
12.3
4.3
19.8
9.1
42.2
14.5
87.9
3.1
2.2
4.0
2.8
4.3
3.0
3.2
2.8
5.5
4.5
7.3
6.1
5CrSiAl
10
15
18
18
10
10CrSiAl
2.5CrSiZr
5CrSiZr
10CrSiZr
24.2
5.6
40.0
10.7
82.5
16.3
209.8
4.1
3.5
31.7
11.9
153.2
16.2
10.6
64.2
−1
a
−1
at 0% conversion.
Millimoles of C H produced g
h
3
6
cat
b
c
Selectivity (%) for C H production at 0% conversion.
3
6
−1
−1
Moles of C H produced mol
h
at 0% conversion.
3
6
Cr
−1
d
e
f
−1
at 0.5% (606 K) or 5% (673 K) C H conversion.
Millimoles of C H produced g
h
3
6
3 8
cat
Selectivity (%) for C H production at 0.5% (606 K) or 5% (673 K) C H conversion.
3
6
3 8
−1
−1
at 0.5% (606 K) or 5% (673 K) C H conversion.
3 6
Moles of C H produced mol
h
3
6
Cr
g
Ratio of CO /CO for the reaction at the lowest flow rate; the ratio was similar at higher total flow.
2
Table 4
in the nCrSiAl and nCrSiZr catalysts may be problematic,
as Si is known to negatively impact the propylene selectiv-
ity for vanadium-based ODH catalysts due to the presence
of Brønsted acid sites [29,30]. All of these results support
the notion that more efficient propane ODH catalysts should
be based on Cr/M/O materials with high Cr loadings and
polymeric CrOx species as the active sites.
Kinetic analyses of propane ODH reactions using nCrSiAl and nCrSiZr
at 606 K
Catalyst
k
k
k
k /k k /k
2 1 3 1
1
2
3
−1 −1
−1 −1
−1 −1
(Lmol
s
) (Lmol
s
) (Lmol s )
2.5CrSiAl
5CrSiAl
10CrSiAl
2.5CrSiZr
5CrSiZr
0.17
0.93
40.2
72.9
22.2
203
110
69.0
5.48 237
3.89 333
4.32 108
5.80 562
4.31 173
0.22
0.21
0.36
0.64
0.92
0.85
0.89
2.09
2.73
3.53
3.4. Dehydrogenation of propane
10CrSiZr
3.85
75.1
The excellent performance of Cr-based oxides as cata-
lysts for the dehydrogenation of propane prompted us to
study the nCrSiAl and nCrSiZr materials in this context
[32–47]. These DH reactions were run in a fixed-bed flow re-
actor with moderate propane concentrations (1–6 mol%) and
propane conversions ranging from ∼ 0.5 to 38%. The cata-
lysts were calcined at 773 K under an atmosphere of He/O2
(∼ 10/1) for 2 h, and subsequently under He for an addi-
tional 2 h prior to reaction. Propane conversions and product
amounts were monitored for an extended period at constant
residence times and reaction temperatures. Table 5 gives the
results of catalytic propane DH reactions at 723 K and low
propane conversions. Overall, the nCrSiZr catalysts outper-
formed their nCrSiAl counterparts in terms of activity, with
all catalysts exhibiting ca. 100% selectivity for propylene
formation. Deactivation of the catalysts, attributed to coking,
was observed with increasing time on-line [32]. However,
heating the catalysts at 773 K (under He/O2 for 2 h and He
for 1 h) after a catalytic run led to full regeneration of the
activity for all catalysts (3 cycles). On a per gram basis, the
activities (mmol C3H6 g−ca1t h−1) of the nCrSiAl catalysts in-
creased with increasing Cr wt%; however 2.5CrSiAl is the
most active when activities are considered on a per mole Cr
The higher propylene selectivities for the recently re-
ported Cr/Al2O3 catalysts prepared by wet impregna-
tion [33], compared to the catalysts described here, appear to
result from the presence of highly active combustion sites in
the nCrSiAl and nCrSiZr catalysts. These sites may result
from the highly dispersed nature of the Cr in these high sur-
face area materials. Indeed, polymeric Cr oxides are thought
to be responsible for the ODH reaction, and the selectiv-
ity for propylene increases with increasing Cr content for
Cr/Al2O3 catalysts [33]. Also, studies of propane ODH re-
actions using Cr/Si/O materials produced from 1 via TMP
methods revealed improved performance for less homoge-
neous materials that contained more polymeric Cr oxide
species [23]. For Cr/Al2O3 catalysts obtained via wet im-
pregnation, it was suggested that a decrease in the k2/k1
ratio is not responsible for the increased propylene selec-
tivity with increasing Cr loading [33]. Similarly for the
catalysts described here, the k2/k1 ratios for the materials
within each nCrSiAl and nCrSiZr set are similar. In con-
trast, the k3/k1 ratios of the catalysts within each nCrSiAl
and nCrSiZr set are quite different, suggesting that the key
difference between catalysts with different Cr loadings is
their activities for propylene oxidation. The 1/1 Si/Cr ratio
basis (mol C3H6 mol Cr−1
h
−1). For the nCrSiZr catalysts,
activities on a per gram or a per mole Cr basis increased with

K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
131
Table 5
Propane DH reactions using the Cr xerogels at 723 K and low C H conversions
3
8
a
b
c
d
Catalyst
Mass
(mg)
mol%
Res. time
(s)
Conversion (%)
Selectivity (%)
C H
Activity
TOF
C H
3
t
1
t
2
t
3
CH
t
t
t
1
t
3
8
3
6
4
1
3
t
1
t
3
t
1
t
3
2.5CrSiAl
5CrSiAl
10CrSiAl
2.5CrSiZr
5CrSiZr
70
50
20
50
50
30
5.5
5.5
5.5
5.1
5.1
5.5
0.67
0.61
0.66
0.69
0.68
0.68
0.59
0.48
0.24
3.02
1.93
0.70
0.38
0.30
0.21
2.75
1.88
0.61
0.32
0.26
0.18
2.70
1.82
0.46
100
100
100
98.5
100
100
100
100
100
99.5
100
100
0
0
0
0.8
0
0
0
0
0.6
0
0.37
0.38
0.53
2.43
1.58
1.02
0.20
0.23
0.35
2.17
1.49
0.66
0.81
0.44
0.35
12.30
4.11
1.40
0.44
0.27
0.25
11.02
3.88
0.90
10CrSiZr
0
0
a
Conversion of C H at t = 35 min, t = 105 min, and t = 170 min (nCrSiAl) or 400 min (nCrSiZr).
3
8
1
2
3
b
c
Selectivity at t = 35 min and t = 170 min (nCrSiAl) or 400 min (nCrSiZr). C H and C H were not observed.
1
3
2
4
2 6
−1
−1
.
Millimoles of C H produced g
h
3
6
cat
−1
d
−1
Moles of C H produced mol
h
at t = 35 min and t = 105 min (nCrSiAl) or 400 min (nCrSiZr).
3
6
1
3
Cr
Table 6
Propane DH reactions using 2.5CrSiZr at 723 or 763 K and high C H conversions
3
8
a
b
c
d
Catalyst
Mass
(mg)
mol%
Res. time
(s)
Conversion (%)
Selectivity (%)
CH
Activity
TOF
C H
3
t
1
t
2
t
3
C H
3
C
2
t
1
t
3
t
1
t
3
8
6
4
t
2
t
3
t
2
t
3
t
2
t
3
2.5CrSiZr
2.5CrSiZr
2.5CrSiZr
2.5CrSiZr
2.5CrSiZr
2.5CrSiZr
50
130
130
130
260
260
5.1
2.9
1.6
2.9
1.9
1.7
0.69
0.9
0.88
1.1
1.02
1
3.02
14.5
20.2
31.9
37.8
33.8
2.75
16.6
20.4
22.3
34.3
31.5
2.70
13.0
16.3
8.40
22.7
20.0
98.5
96.4
95.1
95.2
91.1
91.6
99.4
97.3
96.8
96.1
94.7
95.3
0.8
1.9
2.6
2.4
4.1
3.7
0.6
1.4
1.8
1.8
2.2
1.9
0.0
0.7
0.6
1.1
2.8
3.0
0.0
0.5
0.3
0.3
1.9
2.1
2.4
1.9
1.5
3.2
1.3
1.1
2.2
1.7
1.2
0.9
0.8
0.7
12.3
9.4
7.5
16.4
6.5
5.3
11.0
8.6
6.2
4.5
4.2
3.4
e
f
a
Conversion of C H at t = 35 min, t = 105 min, and t = 400 min.
3
8
1
2
3
b
c
Selectivity at t = 105 min and t = 400 min. C represents the sum of C H and C H (usually 2/1 C H /C H ).
2
3
2
2
4
2
6
2
4
2 6
−1
−1
−1
Millimoles C H produced g
h
at t = 35 min and t = 400 min.
3
6
1
3
cat
−1
d
e
f
Moles of C H produced mol
h
at t = 35 min and t = 400 min.
1
3
6
3
Cr
Reaction at 763 K.
Using regenerated catalyst from the above reaction.
decreasing Cr content. Given that agglomerated species are
more likely to exist at higher Cr loadings, the activity data
for both the nCrSiAl and nCrSiZr catalysts imply that iso-
lated sites are more active for the DH reaction. These results
are consistent with published reports on Cr-based DH cata-
lysts, for which activity levels off or decreases at higher Cr
loadings (4 to 10 wt%) [32]. It has been previously observed
that, in addition to the reversible coking process, slight ir-
reversible deactivation occurs for Cr/Al/O catalysts with
increasing time on-line and through regeneration cycles;
however, consistent performance can be obtained by grad-
ually increasing the reaction temperature [32]. This gradual,
irreversible deactivation of Cr/Al2O3 catalysts has been at-
tributed to the incorporation of Cr3+ species into vacant
Al3+ sites in the Al2O3 support [32]. For Cr/Al2O3 cata-
lysts, it has been shown that coordinatively unsaturated and
isolated Cr3+ species are formed under typical propane DH
conditions (upon reduction of Cr6+ species), and that these
species are active sites [32,45]. Additionally, Cr3+ species in
small amorphous clusters have been proposed as active DH
sites [32].
Thus, we elected to study propane DH reactions with 2.5Cr-
SiZr under conditions that resulted in higher conversions, to
more accurately represent those required for industrial ap-
plication (Table 6, Figs. 8 and 9). Use of larger amounts
of 2.5CrSiZr under various concentrations of propane and
different residence times led to propane conversions that
were greater than 35%, with propylene selectivities exceed-
ing 90%. To monitor catalyst performance over an extended
period, the catalytic runs were executed over the course
of ca. 7 h with GC samples taken approximately every
18 min. Only moderate deactivation due to coke formation
was observed for 2.5CrSiZr at 723 K, with higher propane
conversions giving rise to greater deactivation. Calcination
of the catalysts at 773 K under O2 leads to coke removal
and regeneration of the catalyst. For example, after a 7-h
catalytic run at very high conversions (up to 35%), a 2.5Cr-
SiZr catalyst was treated at 773 K under He/O2 for 3 h
with a subsequent He treatment for 1 h. Exposure of this
regenerated catalyst to propane led to activity and selectiv-
ity values that were nearly identical to those of a previous
run when the (small) differences in the conditions are con-
sidered (Table 6, Figs. 8 and 9). Regeneration times greater
than 3 h may be required to bring the catalyst back to full
At low propane conversions, 2.5CrSiZr is by far the most
active catalyst on a per gram or per mole Cr basis (Table 5).

132
K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
For example, a previously reported Cr/Al2O3 propane DH
catalyst (9% Cr) operated at propane conversions from 26
to 28% with 68 to 76% propylene selectivities at 833 K af-
ter 100 min of reaction time [35]. Other Cr/Al2O3 catalysts
are re−p1orted to have activities of 1.8 mmol C3H6 produced
g−ca1t h at 773 K after 100 min [43], or propane conver-
sions ranging from 25 to 47% with propylene selectivities
from 60 to 90% at 873 K [36,46]. Catalysts based on the
Cr/SiO2 sy−st1em were generally less active (0.6 mmol C3H6
produced gcat h−1 at 773 K after 100 min) [43]. Chromium-
containing Zr/P/O catalysts exhibited activities from 0.4 to
8.4 mmol C3H6 produced g−ca1t h−1 with propylene selectivi-
ties from 79 to 94% at propane conversions from 2 to 18%
at 823 K [38,39]. Finally, Cr/ZrO2 catalysts appear to be the
most active,−w1ith values ranging from 3.6 to 4.8 mmol C3H6
produced gcat h−1 at 723 K [45]. At higher temperatures
(823 K), these catalysts exhibited selectivities for propylene
from 85 to 98% with propane conversions decreasing from
an initial value of 54 to 19% after ca. 100 min [45].
Difficulties in obtaining high surface area and thermally
stable zirconia supports were cited as being a large reason
for limited use of Cr/ZrO2 catalysts for large-scale propane
DH reactions [45]. Use of the TMP-cothermolysis method
has resulted in nCrSiZr materials with high surface areas
after calcination at temperatures above those required for the
propane DH reaction. The excellent selectivity and high ac-
tivity of 2.5CrSiZr as a propane DH catalyst, coupled with
the low temperature required for its use, indicate that this
catalyst is among the best known for this process.
Fig. 8. Propylene selectivity plotted versus time for propane DH using
2.5CrSiZr under various conditions: (a) 50 mg of catalyst, 723 K, 5.1 mol%
C H , and 0.69 s residence time; (b) 130 mg of catalyst, 723 K, 2.9 mol%
3
8
C H , and 0.90 s residence time; (c) 130 mg of catalyst, 723 K, 1.6 mol%
3
8
C H , and 0.88 s residence time; (d) 130 mg of catalyst, 763 K, 2.9 mol%
3
8
C H , and 1.10 s residence time; (e) 260 mg of catalyst, 723 K, 1.9 mol%
3
8
C H , and 1.02 s residence time; (f) 260 mg of catalyst regenerated
3
8
from (e), 723 K, 1.7 mol% C H , and 1.00 s residence time.
3
8
4. Concluding remarks
High surface area, homogeneous xerogels of the types
CrxSixAlyOz (nCrSiAl) and CrxSixZryOz (nCrSiZr) are
easily synthesized via cothermolyses under nonaqueous
conditions using (tBuO)3CrOSi(O t Bu)3 and [Al(O t Bu)3]2
or Zr[OCMe2Et]4. These materials have highly dispersed
chromium centers after calcination at 773 K (by PXRD,
DRUV–vis, TEM, and XANES). The chromium in the
as-synthesized nCrSiAl samples is present as isolated oc-
tahedral Cr3+ species. Calcination of the nCrSiAl samples
at moderate temperatures (673 and 773 K) under O2 results
in high conversion to tetrahedral Cr6+ species and no re-
duced species. Calcination at higher temperatures (1273 or
1473 K) under O2 produces octahedral Cr3+ species, likely
in the form of small (or amorphous) Cr2O3 domains. The
nCrSiZr samples appear to have significant amounts of both
Cr3+ and Cr6+ species in their as-synthesized state; how-
ever, calcination at 773 K leads to formation of only Cr6+
species.
Fig. 9. Propane conversion plotted versus time for propane DH using 2.5Cr-
SiZr under various conditions: (a) 50 mg of catalyst, 723 K, 5.1 mol%
C H , and 0.69 s residence time; (b) 130 mg of catalyst, 723 K, 2.9 mol%
3
8
C H , and 0.90 s residence time; (c) 130 mg of catalyst, 723 K, 1.6 mol%
3
8
C H , and 0.88 s residence time; (d) 130 mg of catalyst, 763 K, 2.9 mol%
3
8
C H , and 1.10 s residence time; (e) 260 mg of catalyst, 723 K, 1.9 mol%
3
8
C H , and 1.02 s residence time; (f) 260 mg of catalyst regenerated
3
8
from (e), 723 K, 1.7 mol% C H , and 1.00 s residence time.
3
8
activity after longer catalytic runs. For the propane DH reac-
tion performed at 763 K with 2.5CrSiZr, initial conversion
was very high with excellent propylene selectivity; however,
the catalyst deactivated at a severe rate to levels below those
observed for the reaction at 723 K (Table 5, Fig. 9). This ap-
pears to reflect a dramatic increase in the rate of coking at
the higher temperature.
Although it is difficult to compare catalysts operating
under different conditions, the excellent performance of
2.5CrSiZr for propane DH at low temperatures suggests that
the TMP cothermolysis method may offer improvements
over the propane DH catalysts studied previously [32–47].
The nCrSiAl and nCrSiZr catalysts are active for
propane ODH, with 10CrSiZr having an activity (based on
propylene production) that is among the highest values re-
ported for this reaction. Unfortunately, the selectivity toward
propylene production is low for all catalysts examined. Ki-

K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
133
netic analyses indicate that the poor selectivities arise from
very high rates for propylene combustion (k3). It appears that
well-dispersed Cr species are not selective for propylene,
and that polymeric chromium oxides are preferred [33]. For
the nCrSiZr catalysts, an increase in the rate of propane
ODH (k1) and a decrease in the rate of propylene com-
bustion (k3) with increasing chromium content indicate that
different sites are at least partially responsible for these reac-
tions. Comparisons to previously reported Al2O3-supported
Cr catalysts synthesized via wet impregnation reveal that the
catalysts prepared here exhibit higher ODH activities, but
lower selectivities for propylene formation [33]. In addition,
comparisons of the Cr-based catalysts reported here with
previously reported V/Zr/O ODH catalysts synthesized via
thermolytic molecular precursor methods reveal that the cat-
alysts have comparable activities, but that selectivities for
propylene are superior for the vanadium-based species [30].
Propane DH reactions under mild conditions (723 K) us-
ing the nCrSiAl and nCrSiZr catalysts are characterized by
high selectivities at low propane conversions. The nCrSiZr
catalysts are generally more active than their nCrSiAl coun-
terparts. Lower Cr contents give rise to higher activities on
a per Cr basis, suggesting that isolated sites are preferred.
Detailed investigations using 2.5CrSiZr led to propane con-
versions greater than 35%, with propylene selectivities ap-
proaching 95%. These values are among the best reported
for propane DH catalysts, particularly at such a low reac-
tion temperature. Deactivation due to coking was observed
over the course of the reactions; however, calcination under
He/O2 at 773 K was effective for regeneration of the cata-
lysts. Therefore, for propane ODH, the best results seem to
be associated with chromate clusters, whereas for propane
DH, well-dispersed (and presumably isolated) chromium
species are most effective.
[5] A.H. Cowley, R.A. Jones, Angew. Chem. Int. Ed. Engl. 28 (1989)
1208.
[6] F. Chaput, A. Lecomte, A. Dauger, J.P. Boilot, Chem. Mater. 1 (1989)
199.
[7] A.W. Apblett, A.C. Warren, A.R. Barron, Chem. Mater. 4 (1992) 167.
[8] C.D. Chandler, C. Roger, M.J. Hampden-Smith, Chem. Rev. 93 (1993)
1205.
[9] A. Stein, S.W. Keller, T.E. Mallouk, Science 259 (1993) 1558.
[10] D.C. Bradley, Polyhedron 13 (1994) 1111.
[11] U. Schubert, N. Hüsing, A. Lorenz, Chem. Mater. 7 (1995) 2010.
[12] D.B. Amabilino, J.F. Stoddart, Chem. Rev. 95 (1995) 2725.
[13] C.L. Bowes, G.A. Ozin, Adv. Mater. 8 (1996) 13.
[14] U. Schubert, J. Chem. Soc., Dalton Trans. (1996) 3343.
[15] K.W. Terry, C.G. Lugmair, P.K. Gantzel, T.D. Tilley, Chem. Mater. 8
(1996) 274.
[16] K. Su, T.D. Tilley, M.J. Sailor, J. Am. Chem. Soc. 118 (1996) 3459.
[17] K. Su, T.D. Tilley, Chem. Mater. 9 (1997) 588.
[18] K.W. Terry, C.G. Lugmair, T.D. Tilley, J. Am. Chem. Soc. 119 (1997)
9745.
[19] K.W. Terry, K. Su, T.D. Tilley, A.L. Rheingold, Polyhedron 17 (1998)
891.
[20] R. Rulkens, J.L. Male, K.W. Terry, B. Olthof, A. Khodakov, A.T. Bell,
E. Iglesia, T.D. Tilley, Chem. Mater. 11 (1999) 2966.
[21] J.W. Kriesel, M.S. Sander, T.D. Tilley, Mater. Chem. 13 (2001) 3554.
[22] J.W. Kriesel, T.D. Tilley, J. Mater. Chem. 11 (2001) 1081.
[23] K.L. Fujdala, T.D. Tilley, Chem. Mater. 13 (2001) 1817.
[24] K.L. Fujdala, T.D. Tilley, J. Am. Chem. Soc. 123 (2001) 10133.
[25] C.G. Lugmair, K.L. Fujdala, T.D. Tilley, Chem. Mater. 14 (2002) 888.
[26] K.L. Fujdala, T.D. Tilley, Chem. Mater. 14 (2002) 1376.
[27] M. Jansen, E. Guenther, Chem. Mater. 7 (1995) 2110.
[28] A. Vioux, Chem. Mater. 9 (1997) 2292.
[29] R. Rulkens, T.D. Tilley, J. Am. Chem. Soc. 120 (1998) 9959.
[30] J.L. Male, H.G. Niessen, A.T. Bell, T.D. Tilley, J. Catal. 194 (2000)
431.
[31] H.H. Kung, Adv. Catal. 40 (1995) 1.
[32] B.M. Weckhuysen, R.A. Schoonheydt, Catal. Today 51 (1999) 223,
and references therein.
[33] M. Cherian, M.S. Rao, W.-T. Yang, J.-M. Jehng, A.M. Hirt, G. Deo,
Appl. Catal. A 233 (2002) 21.
[34] J. Sloczynski, B. Grzybowska, R. Grabowski, A. Kozlowska, K. Wcis-
lo, Phys. Chem. Chem. Phys. 1 (1999) 333.
[35] L.R. Mentasty, O.F. Gorriz, L.E. Cadus, Ind. Eng. Chem. Res. 38
(1999) 396.
[36] O.F. Gorriz, L.E. Cadús, Appl. Catal. A 180 (1999) 247.
[37] F.E. Frey, W.F. Huppke, Ind. Eng. Chem. 25 (1933) 54.
[38] M. Alcántara-Rodríguez, E. Rodríguez-Castellón, A. Jiménez-López,
Langmuir 15 (1999) 1115.
[39] F.J. Pérez-Reina, E. Rodríguez-Castellón, A. Jiménez-López, Lang-
muir 15 (1999) 8421.
[40] W.C. Chang, N. Mimura, M. Saito, I. Takahara, Catal. Today 45 (1998)
55.
[41] V. Indovina, Catal. Today 41 (1998) 95.
[42] S. De Rossi, M.P. Casaletto, G. Ferraris, A. Cimino, G. Minelli, Appl.
Catal. A 167 (1998) 257.
[43] S. De Rossi, G. Ferraris, S. Fremiotti, E. Garrone, G. Ghiotti, M.C.
Campa, V. Indovina, J. Catal. 148 (1994) 36.
Acknowledgments
We are grateful for the support of this work by the
Office of Basic Energy Sciences, Chemical Sciences Divi-
sion, of the US Department of Energy under Contract No.
DE-AC03-76SF00098. We also thank Prof. A. Stacy for the
use of instrumentation (PXRD, DRUV–vis), Dr. J. Male for
assistance with the XANES studies, and the Stanford Syn-
chrotron Radiation Laboratory (SSRL).
[44] S. De Rossi, G. Ferraris, S. Fremiotti, V. Indovina, A. Cimino, Appl.
Catal. A 106 (1993) 125.
References
[45] S. De Rossi, G. Ferraris, S. Fremiotti, A. Cimino, V. Indovina, Appl.
Catal. A 81 (1992) 113.
[46] O.F. Gorriz, V.C. Corberán, J.L.G. Fierro, Ind. Eng. Chem. Res. 31
(1992) 2670.
[47] D. Sanfilippo, F. Buonomo, G. Fusco, M. Lupieri, I. Miracca, Chem.
Eng. Sci. 47 (1992) 2313.
[48] T. Blasco, J.M. López Nieto, Appl. Catal. A 157 (1997) 117.
[49] S. Albonetti, F. Cavani, F. Trifirò, Catal. Rev. Sci. Eng. 38 (1996) 413.
[50] A. Khodakov, B. Olaf, A.T. Bell, E. Iglesia, J. Catal. 181 (1999) 205.
[1] J.-P. Jolivet, Metal Oxide Chemistry and Synthesis. From Solution to
Solid State, Wiley, Chichester, 2000.
[2] C.K. Narula, Ceramic Precursor Technology and Its Applications,
Dekker, New York, 1995.
[3] D.R. Uhlmann, D.R. Ulrich (Eds.), Ultrastructure Processing of Ad-
vanced Materials, Wiley, New York, 1992.
[4] C.J. Brinker, G.W. Scherer, Sol–Gel Science, Academic Press, Boston,
1990.

134
K.L. Fujdala, T.D. Tilley / Journal of Catalysis 218 (2003) 123–134
[51] A. Khodakov, J. Yang, S. Su, E. Iglesia, A.T. Bell, J. Catal. 177 (1998)
[58] B.M. Weckhuysen, A.A. Verberckmoes, A.R. De Baets, R. Schoon-
heydt, J. Catal. 166 (1997) 160.
343.
[52] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309.
[53] E.P. Barrett, L.G. Joyner, P.P. Halenda, J. Am. Chem. Soc. 73 (1951)
373.
[54] J.R. Sohn, S.G. Ryu, Langmuir 9 (1993) 126.
[55] M. Yashima, M. Kakihana, M. Yoshimura, Solid State Ionics 86 (1996)
1131.
[59] B.M. Weckhuysen, A.A. Verberckmoes, A.L. Buttiens, A.R. Schoon-
heydt, J. Phys. Chem. 98 (1994) 579.
[60] B.M. Weckhuysen, L.M. De Ridder, R.A. Schoonheydt, J. Phys.
Chem. 97 (1993) 4756.
[61] P.S. Devi, H.D. Gafney, V. Petricevic, R.R. Alfano, D. He, K.E.
Miyano, Chem. Mater. 12 (2000) 1378.
[56] B.M. Weckhuysen, I.E. Wachs, R.A. Schoonheydt, Chem. Rev. 96
(1996) 3327.
[62] F. Rey, G. Sankar, T. Maschmeyer, J.M. Thomas, R.G. Bell, Top.
Catal. 3 (1996) 121.
[57] B.M. Weckhuysen, R.A. Schoonheydt, Catal. Today 49 (1999) 441.
[63] C. Pak, A.T. Bell, T.D. Tilley, J. Catal. 206 (2002) 49.