Structural and catalytic characterization of solid acids based on zirconia modified by tungsten oxide

Article abstract of DOI:10.1006/jcat.1998.2269

Tungsten oxide species form strong acid sites on ZrO₂ supports and inhibit ZrO₂ crystallite sintering and tetragonal to monoclinic structural transformations. W-L_I X-ray absorption near-edge spectra suggest that the W centers are in a distorted octahedral oxygen environment, even after dehydration at 673 K, in all WO_x-ZrO₂ samples (2-21 wt.% W) oxidized at 1073 K. Maximum o-xylene isomerization turnover rates (per W atom) on WO_x-ZrO₂ solids occur at WO_x surface densities (10 W nm^-2) that exceed the theoretical monolayer capacity of ZrO₂. Similar turnover rates are obtained on WO_x-ZrO₂ samples with similar WO_x surface densities (W nm^-2) over a large range of oxidation temperatures (773-1223 K) and WO_x concentrations (5-21 wt.% W). UV-visible spectra suggest an increase in WO_x domain size with increasing surface density. High isomerization turnover rates appear to require the presence of WO_x domains of intermediate size on ZrO₂ surfaces. WO_x domains of intermediate size appear to provide a compromise between reducibility and accessibility of WO_x centers. These domains are necessary to delocalize a temporary charge imbalance that forms Bronsted acid sites in the presence of H₂ and stabilizes carbocation intermediates. The presence of H₂ during o-xylene isomerization increases turnover rates and prevents rapid deactivation. Slow D₂/o-xylene exchange reactions indicate that H atoms from H₂ are not frequently involved in the activation or desorption of xylenes. H₂ is required, however, in order to reverse the occasional desorption of H atoms during o-xylene isomerization reactions. These desorption processes lead to the destruction of Bronsted acid sites by the formation of strongly adsorbed unsaturated species in the absence of H₂. After promotion with Pt (0.3 wt.%), WO_x-ZrO₂ solids catalyze n-heptane isomerization in the presence of H₂ at 400-500 K with much higher selectivity than sulfated oxides or zeolitic acids at similar turnover rates. On Pt/WO_x-ZrO₂, efficient hydrogen transfer steps prevent extensive cracking of adsorbed carbocations by limiting their surface lifetimes.

Full text of DOI:10.1006/jcat.1998.2269

Journal of Catalysis 181, 57–72 (1999)
Article ID jcat.1998.2269, available online at http://www.idealibrary.com on
Structural and Catalytic Characterization of Solid Acids Based
on Zirconia Modified by Tungsten Oxide
,1
David G. Barton, Stuart L. Soled,† George D. Meitzner,‡ Gustavo A. Fuentes,§ and Enrique Iglesia
Department of Chemical Engineering, University of California at Berkeley, Berkeley, California 94720; †Corporate Research Laboratory,
Exxon Research and Engineering, Route 22 East, Annandale, New Jersey 08801; ‡Edge Analytical, Inc., 2126 Allen Blvd.,
Middleton, Wisconsin 53562; and §Departamento de Ingenieria de Procesos, Universidad Autonoma
Metropolitana-Iztapalapa, A.P. 55-534, 09340 Mexico, D.F., Mexico
Received February 2, 1998; revised August 28, 1998; accepted August 31, 1998
BACKGROUND
Tungsten oxide species form strong acid sites on ZrO2 supports
Current catalytic processes for alkane isomerization and
alkylation require liquid acids (e.g., H2SO4, HF, AlCl3)
or halogen-promoted metal oxides (e.g., Cl–Al2O3), which
pose significant containment, corrosion, and environmen-
tal challenges (1). These problems can be circumvented
by the development of strong solid acid catalysts consist-
ing of metal oxides. The development of solid acids that
are stable, regenerable, and active at the low temperatures
required for favorable isomerization thermodynamics re-
mains a formidable challenge.
Recently, metal oxides promoted by sulfur compounds
have been studied as strong solid acid catalysts (2, 3), espe-
cially SOx- promoted zirconia (4) containing iron or man-
ganese oxides as promoters (5) or noble metals to inhibit
deactivation (6–8). Yet, their low isomerization selectivity
for C7+ alkanes, their poor stability, and their tendency to
form volatile sulfur compounds during catalysis and regen-
eration limit their applicability in isomerization and alky-
lation processes.
Tungsten oxide-based materials comprise another inter-
esting class of solid acids. The small and highly charged
W+6 cations are found in several solid acids: heteropoly-
oxoanion compounds, such as 12-tungstophosphoric acid
(9), oxygen-modified tungsten carbides (10), and WOx dis-
persed on -Al2O3 (11) and ZrO2 (12–19). This paper fo-
cuses on WOx dispersed on ZrO2 (WOx–ZrO2), first re-
ported to be strongly acidic by Hino and Arata (20). These
authors suggested, based on color changes by Hammett in-
dicators, that acid sites stronger than 100% sulfuric acid
and inhibit ZrO2 crystallite sintering and tetragonal to monoclinic
structural transformations. W-LI X-ray absorption near-edge spec-
tra suggest that the W centers are in a distorted octahedral oxygen
environment, even after dehydration at 673 K, in all WOx–ZrO2
samples (2–21 wt.% W) oxidized at 1073 K. Maximum o-xylene iso-
merization turnover rates (per W atom) on WOx–ZrO2 solids occur
at WOx surface densities (10 W nm 2) that exceed the theoretical
monolayer capacity of ZrO2. Similar turnover rates are obtained on
2
WOx–ZrO2 samples with similar WOx surface densities (W nm
)
over a large range of oxidation temperatures (773–1223 K) and
WOx concentrations (5–21 wt.% W). UV–visible spectra suggest an
increase in WOx domain size with increasing surface density. High
isomerization turnover rates appear to require the presence of WOx
domains of intermediate size on ZrO2 surfaces. WOx domains of in-
termediate size appear to provide a compromise between reducibil-
ity and accessibility of WOx centers. These domains are necessary
to delocalize a temporary charge imbalance that forms Brønsted
acid sites in the presence of H2 and stabilizes carbocation interme-
diates. The presence of H2 during o-xylene isomerization increases
turnover rates and prevents rapid deactivation. Slow D2 /o-xylene
exchange reactions indicate that H atoms from H2 are not frequently
involved in the activation or desorption of xylenes. H2 is required,
however, in order to reverse the occasional desorption of H atoms
duringo-xyleneisomerization reactions. Thesedesorption processes
lead to the destruction of Brønsted acid sites by the formation of
strongly adsorbed unsaturated species in the absence of H2. After
promotion with Pt (0.3 wt.%), WOx–ZrO2 solids catalyze n-heptane
isomerization in the presence of H2 at 400–500 K with much higher
selectivity than sulfated oxides or zeolitic acids at similar turnover
rates. On Pt/WOx–ZrO2, efficient hydrogen transfer steps prevent
extensive cracking of adsorbed carbocations by limiting their sur-
c
face lifetimes.
1999 Academic Press
may exist (H0
14.52), but the use of this technique as
a measure of acid strength on solids is generally unreli-
able (21). Butene dimerization studies have confirmed that
acid sites on WOx–ZrO2 have catalytic activity compara-
ble to those on SOx–ZrO2 (14). Hino and Arata (12) re-
ported low-temperature catalytic activity for n-pentane and
n-hexane isomerization, but cracking selectivities were very
high (>50% ) even at low conversion. The addition of a
1 To whom correspondence should be addressed. Fax: (510) 642-4778.
E-mail: iglesia@cchem.berkeley.edu.
57
0021-9517/99 $30.00
c
Copyright
1999 by Academic Press
All rights of reproduction in any form reserved.

58
BARTON ET AL.
metal component (<1% wt. Pt) to WOx–ZrO2 catalysts and (OH)4 2x, x<2) solids with ammonium metatungstate solu-
the presence of H2 resulted in very active and selective cata- tions. High-surface-area ZrOx(OH)4 was prepared by
2x
lysts for n-alkane isomerization (16–18, 22) and cycloalkane hydrolysis of a 0.5 M zirconyl chloride solution (ZrOCl2
ring-opening (15) reactions with excellent stability and high 8H2O, Aldrich, >98 wt.% , Hf <0.5 wt.% , 500 cm3 h 1 feed)
selectivity to isomers.
using the controlled addition of NH4OH (14.8 N) to obtain
Acid-catalyzed alkane isomerization has become in- a constant pH of 10. This pH was chosen based on previous
creasingly important in the synthesis of high-octane blend- studies (25–27), in which the mesoporosity, surface area,
ing components and lubricant basestocks. Currently, iso- and total pore volume of ZrOx(OH)4 2x were shown to in-
merization of heptane and larger alkanes (C7+) is limited crease with increasingprecipitation pH. The precipitate was
by high selectivity to cracking products and by poor yields immediatelyfiltered and washed repeatedlybyredispersion
of high-octane multibranched alkanes containing quater- in a NH4OH solution (pH 10) in order to remove residual
nary carbons. Long carbocation surface residence times chloride. After each wash, a couple ofdropsofa 3M AgNO3
during C7+ alkane isomerization and cracking pathways solution were added to the filtrate liquid until it showed no
that directly form stable tert-butyl surface carbocations visible signsofAgClformation (<10ppm residualCl ). The
favor cracking reactions on strong acids (23). In practice, ZrOx(OH)4
gel was dried at 383 K overnight, crushed,
2x
undesired cracking reactions limit paraffin isomerization and sieved to retain particles with 0.12–0.38 mm diameter.
processes to C4–C6 alkanes. The dried particles were impregnated to the point of in-
Iglesia et al. (24) have shown that high isomerization se- cipient wetness (0.29 cm3 H2O per 1.0 g ZrOx(OH)4
)
2x
lectivity (>70% ) in n-heptane reactions can be obtained with ammonium metatungstate ((NH4)6H2W12O40, Strem
on oxygen-modified tungsten carbide catalysts at high tem- Chemicals, 99.9% ) solutions to give 2–21 wt.% W after ox-
peratures ( 623 K). These catalysts behave as bifunctional idation treatments. The impregnated samples were dried at
metal–acid catalysts with a WC dehydrogenation function 383 K overnight and oxidized in flowing dry air at tempera-
and a WOx acid function, but they require high tempera- tures between 773 and 1273 K. All samples were oxidized in
tures ( 623 K), which favor hydrogenolysis pathways and shallow quartz boats inside a 25-mm-diameter quartz tube
monobranched isoheptanes. In another study, Iglesia et al. with flowing air (Linde Zero Grade, 3.3 cm3 s 1), dried and
(6) reported n-heptane isomerization results at lower tem- purified with Drierite and 13X molecular sieve traps. The
peratures (473 K) on a stronger acid (Pt/SOx–ZrO2). They samples were heated at a rate of 0.167 K s 1 to the final tem-
found predominately isobutane and propane among crack- perature (measured in a quartz thermowell just above the
ing products ( 50% total cracking selectivity at low con- sample bed), held for 3 h, and cooled to room temperature
versions 10% ), consistent with acid-catalyzed cracking in flowing air. After oxidation at 1073 K, a portion of the
of heptyl cations. The addition of hydride transfer cocata- sample containing 12 wt.% W was impregnated to the point
lysts showed that slow hydride transfer to surface carbo- of incipient wetness with a solution of tetra-amine platinum
cations limits isomerization rates and leads to long carbo- hydroxide (Degussa), dried at 383 K overnight, and oxi-
cation surface lifetimes and high cracking selectivity (6). dized for 3 h at 723 K (0.3 wt.% Pt). Nominal tungsten and
These studies along with Hino and Arata’s discovery of the platinum contents (after oxidative treatments) were based
unique properties of WOx–ZrO2 led us to study the syn- on precursor decomposition weight losses obtained from
thesis and characterization of these catalysts promoted by thermogravimetric analysis.
Pt for n-heptane isomerization reactions (Pt/WOx–ZrO2).
This catalyst promoted by Pt couples a strong acid com-
ponent active at low temperatures with clean noble metal
Chemisorption and Physisorption Measurements
particles that increase hydride transfer rates (18). Here, we
examine the structural stabilization of the zirconia support
and the nature ofthe surface WOx groups. We also probe the
effects of H2 on n-heptane and o-xylene isomerization, and
use the turnover rates for the “monofunctional” isomeriza-
tion of o-xylene in order to compare WOx–ZrO2 samples
with a wide range of WOx surface densities.
Chemisorption data were collected on a Quantachrome
1C Autosorb apparatus. Samples were dried in a box fur-
nace at 723 K in air immediately before loading. Pt/WOx–
1
ZrO2 samples were heated in H2 at 0.033 K s to the fi-
nal reduction temperature (either 473 or 823 K) and held
isothermally for 1 h. After evacuation, the samples were
cooled to 313 K, and an 11-point H2 adsorption isotherm
was obtained between 2.7 and 53 kPa. The linear portion
of the isotherm was extrapolated to zero pressure to obtain
the combined (total) uptake. After evacuation at 313 K, the
isotherm wasagain measured;the linear portion wasextrap-
olated to zero pressure to obtain the weakly adsorbed frac-
EXPERIMENTAL
Synthesis of WOx–ZrO2 Catalytic Solids
WOx–ZrO2 samples were prepared by incipient wet- tion of the total initial uptake, and the strong chemisorption
ness impregnation of zirconium oxyhydroxide (ZrOx uptake wasobtained bydifference. After thismeasurement,

ZIRCONIA AND TUNGSTEN OXIDE
59
the catalyst was again reduced at the original reduction tem- Thermal Analysis
perature, and oxygen uptakes were obtained at 313 K. CO
Weight changes during sample oxidation (0.030–0.100 g)
were measured by thermal gravimetry (Mettler TA-2000C).
uptakes at 313 K were measured on a fresh sample pre-
treated as described above. A similar procedure was used
to determine reversible H2 chemisorption uptakes at 523 K
on WOx–ZrO2 samples (without Pt) reduced at 523 K, ex-
cept the uptakes are reported at 45 kPa H2, which results
in a better estimate of the amount of reactive surface H
species present during reaction.
Physical surface areas were obtained by N2 physisorption
using a Quantasorb surface area analyzer (Quantachrome
Corporation) and a standard multipoint BET isotherm (28).
Samples previously oxidized at high temperatures (>473 K)
were degassed in flowing N2 for 1 h at 473 K; other samples
were degassed for 3 h at 383 K before N2 physisorption
measurements.
Weight changes were measured at a heating rate of 0.167 K
1
s
in a 20% O2/He stream. Heat changes as a result of
crystallization of amorphous zirconia were measured with
a differential scanning calorimeter (Dupont 2100, 0.167 K
s
1), using samples (0.025–0.030 g) sealed in aluminum con-
tainers and -Al2O3 samples as reference.
Temperature-programmed reduction studies were car-
ried out in a Quantasorb surface area analyzer (Quan-
tachrome Corporation) modified with electronic mass flow
meters, a programmable furnace, and on-line 13X molecu-
lar sieve traps in order to remove the water formed during
reduction. A 20% H2/Ar mixture (Matheson UHP, certified
mixture) flowing at 1.33 cm3 s 1 was used to reduce samples
(0.025–0.200 g) held within a 4-mm-I.D. quartz u-tube cell
containing a quartz thermowell directly in contact with the
sample bed. Reduction rates were obtained by thermal con-
ductivity measurements of the H2 concentration in the exit
stream as the sample temperature was raised from room
Powder X-Ray Diffraction Measurements
X-ray powder diffraction (XRD) patterns were recorded
using a Siemens D5000 Diffractometer with CuK radi-
ation and a scanning rate of 0.002 s 1. These patterns
were used to quantify the relative amounts and crystal-
lite sizes of each ZrO2 phase (monoclinic and tetragonal)
and to detect the presence of bulk WO3 crystallites. The
1
temperature to 1400 K at 0.167 K s and held for 0.5 h.
The output analog signal was calibrated using the complete
reduction of a standard CuO powder (Aldrich, 99.995% )
in order to determine H2 consumption rates.
¯
ZrO2 diffraction features of interest are the (111) and (111)
reflections of the monoclinic phase and the (111) reflec-
tion of the metastable tetragonal phase. Since these re-
flections overlap even in samples oxidized at high tem-
peratures, they were resolved using a model proposed by
Schreiner and Jenkins (29), which fits both the instrumen-
tal and the wavelength-breadth components to asymmetric
Lorentzians. These pattern-resolved reflections, corrected
X-Ray Absorption Spectroscopy
X-ray absorption spectra were obtained using beamlines
4-1 and 4-3 at the Stanford Synchrotron Radiation Labora-
tory (SSRL). Spectra were measured in transmission mode,
with Ar in the ion chamber detectors. Monochromators on
both beamlines were equipped with Si(220) crystals, de-
tuned to 50% of the maximum inlet signal intensity in or-
der to minimize harmonics. Data points were obtained at
0.25 eV increments near the W-LI and W-LIII edges (12100
for K 1–K
wavelength breadth and for instrumental
2
broadening, were used to calculate the tetragonal volume
percentage (Vt) and the tetragonal and monoclinic crystal-
lite sizes in these crystallographic directions.
and 10207 eV) and at 0.04 A 1 increments in the extended
˚
The integrated areas under the resolved reflections (It,
Im) were used to calculate the intensity ratio (Xt) (Eq. [1])
and then Vt using an empirical nonlinear relationship
(Eq. [2]) obtained by Toraya et al. (30) from physical mix-
tures of monoclinic and tetragonal ZrO2. The Scherrer rela-
tionship (Eq. [3]) (31) was used to calculate crystallite sizes
(Dhkl) using the integral line widths (Bhkl) of the monoclinic
and tetragonal pattern-resolved reflections and a shape fac-
tor equal to one:
fine structure (EXAFS) region. A Hf impurity ( 0.5 wt.% )
in ZrO2 limited the EXAFS to the range between the W-LIII
(10207 eV) and Hf-LIII (10739 eV) edges.
Allcatalyst sampleswere diluted with SiO2 to 2.2wt.% W,
pressed into wafers, crushed, and sieved to retain particles
with 0.18–0.25 mm diameter. Sieved powders were packed
into quartz capillaries (0.9 mm diameter with 0.1 mm
wall thickness) and supported horizontally in a rectangular
X-raybeam path (0.2mm 6.0mm). Four heat cartridgesin
a copper heat sink supplied heat uniformly to the capillary
cell. Samples were dehydrated and oxidized in air flowing
through the capillary(0.167cm3 s 1, Matheson, Ultra Zero).
In processing the spectra, the background was removed
It (111)
Xt =
Vt =
[1]
¯
Im(111) + Im(111) + It (111)
1.311(1 Xt )
[2] by fitting a straight line to the pre-edge region and a third-
order polynomial spline to the EXAFS region. The spectra
1.311 0.311Xt
S
were normalized to the step height, measured at the edge
energy. The edge energy is defined as the energy at the first
Dhkl
=
.
[3]
Bhkl cos

60
BARTON ET AL.
inflection point for W-LIII spectra and as the first inflec- using matrix techniques that account for natural 13C abun-
tion point beyond the pre-edge feature for W-LI spectra. dance and ion fragmentation kinetics (35).
The normalized EXAFS was converted to units of A 1 and
˚
1
˚
then Fourier transformed over the range 2–11.5 A (32).
RESULTS AND DISCUSSION
Small changes in the W-LI near-edge spectra were quanti-
fied by subtracting a bulk WO3 reference spectrum from the ZrO2 Crystal Structure and Crystallite Size
sample spectra, using previously reported procedures (33).
Amorphous ZrOx(OH)4 2x powders treated in air above
723 K convert to crystalline ZrO2 with tetragonal, mono-
clinic, or cubic structures. At temperatures below 1443 K,
Catalytic Measurements
Kinetic studies were carried out in high-pressure flow and monoclinic ZrO2 is the thermodynamically stable crystal
glass recirculating batch reactors described previously (34). structure for crystallites larger than 10 nm (36). Metastable
Catalyst samples were sieved to retain particles with diame- tetragonal ZrO2 crystallites smaller than 10 nm often form
ters between 0.25 and 0.50 mm. These samples were placed after oxidation at temperatures slightly above the ZrO2
along with quartz powder of similar particle size into the crystallization temperature (723 K). The fraction of tetrag-
reactors. Before catalytic tests, samples were pretreated at onal ZrO2 crystallites formed is strongly influenced by the
773 K in flowing air (8.33 cm3 s 1, Matheson, Ultra Zero) nature of the ZrOx(OH)4 2x precursors and by the methods
for 3 h and then reduced at reaction temperature in flowing used in their synthesis (37). After oxidation at higher tem-
hydrogen (8.33 cm3 s 1, Matheson, UHP) for 1 h. Oxygen peratures (>900 K), ZrO2 crystallites sinter and become
and water impurities in H2 were removed using a Pd/Al2O3 predominately monoclinic, regardless of precursor or syn-
catalyst (Johnson–Mathey) and a molecular sieve trap held thetic method. Pure tetragonal ZrO2 crystallites, however,
at ambient temperature. Gas chromatography (Hewlett– may be stabilized by impregnation with metal oxides before
Packard GC/MS 5890/5993, 50-m Ultra-2 column for C7 high-temperature oxidative treatments (38).
separations and 30-m DB-Wax column for xylene separa-
WOx species strongly influence ZrOx(OH)4
crystal-
2x
tions) with flame ionization and mass spectrometric detec- lization into ZrO2. Impregnation of ZrOx(OH)4
with
2x
tion was used to measure product distributions. Reaction ammonium metatungstate solutions stabilizes tetragonal
ratesare reported asW atom turnovers, defined asthe moles ZrO2 crystallites after high-temperature oxidative treat-
of reactant converted per W atom per second. Product se- ments (90% tetragonal for 7.9 wt.% W/ZrO2 vs 3% tetrag-
lectivities are reported on a carbon basis as the percentage onal for pure ZrO2 at 1073 K). This stabilization of tetrago-
of the converted reactant that appears as a given product.
nal ZrO2 crystallites is shown in Fig. 1a, which compares the
n-Heptane isomerization kinetic experiments were car- tetragonal ZrO2 content in samples containing 7.9 wt.% W
ried out between 453 and 513 K at total pressures between with pure ZrO2 samples at various oxidation temperatures.
300 and 2500 kPa in a plug-flow reactor. Electronic mass In addition to stabilizing tetragonal ZrO2 crystallites, WOx
flow controllers were used to control gas flow rates and a liq- surface species inhibit ZrO2 crystallite sintering, leading to
uid syringe pump was used to introduce n-heptane (Fluka, smaller crystallites than in pure ZrO2 (Fig. 1b). In pure
>99.7% ) into a heated mixing zone at 523 K before entering ZrO2, the tetragonal content decreases markedly after oxi-
the reactor.
dation above 800 K. When WOx surface species are present
o-Xylene isomerization kinetic experiments were carried (7.9 wt.% W), this phase transition occurs at higher ox-
out in a glass recirculating batch reactor at 523 K and a total idation temperatures (>1073 K). This phase transition is
pressure of 107 kPa (0.67 kPa o-xylene, 106 kPa H2 or He, not accompanied by a rapid increase in crystallite size; the
0.030 g sample). This reactor has a total volume of 693 cm3, sizes of tetragonal and monoclinic crystallites are similar
a gas recirculation rate of 4.17 cm3 s 1, and it is operated at and much smaller than those in pure ZrO2. This suggests
less than 2% reactant conversion per pass in order to en- that WOx species strongly interacting with ZrO2 crystallites
sure gradientless operation (34). The reactor is evacuated inhibit sintering even at high oxidation temperatures. It ap-
usingmechanicaland diffusion pumpsisolated from the sys- pears that a critical crystallite size is reached ( 20 nm),
tem by liquid nitrogen traps. o-Xylene (NBS, >99.995% ), above which tetragonal crystallites become unstable and
degassed using several freeze–evacuation–thaw cycles, was transform into monoclinic crystallites of identical size.
admitted as a vapor from liquid held at room temperature.
The tetragonal content in ZrO2 also depends on WOx
Hydrogen transfer pathways were studied by H–D iso- concentration at each oxidation temperature. For example,
topic exchange reactions in the glass recirculating batch after oxidative treatments at 1073 K, the tetragonal ZrO2
reactor. Deuterium/n-heptane (343 K, 97 kPa D2, 3.7 kPa percentage increases and the crystallite size decreases with
n-heptane) and deuterium/o-xylene (523 K, 106 kPa D2, increasing WOx concentration (Fig. 2). A minimum ZrO2
0.67 kPa o-xylene) exchange rates were measured during crystallite size (16 nm) is reached by samples containing
isomerization reactions. The deuterium content in reactants 7.9 wt.% W, and crystallites retain this size as WOx con-
and products was calculated from mass spectrometric data centration increases further. At WOx concentrations above

ZIRCONIA AND TUNGSTEN OXIDE
61
FIG. 1. Powder X-ray diffraction results of ZrO2 and WOx–ZrO2 (7.9 wt.% W) samples showing stabilization of (a) the tetragonal ZrO2 phase and
(b) the ZrO2 crystallite size with WOx surface species at various oxidation temperatures.
12 wt.% W, ZrO2 crystallites are purely tetragonal and Surface areas for all samples become quite similar after ox-
monoclinic WO3 crystallites become detectable by X-ray idation above 1173 K, apparently because WOx species are
diffraction. Samples with bulk WO3 crystallites can also be ineffective at stabilizing the ZrO2 surface at these temper-
identified visually by their light yellow color, instead of the atures.
white color of the rest of the samples.
WOx species are not the only reported textural promot-
ers of ZrO2. Several other oxides also inhibit ZrO2 crystal-
lite sintering at high oxidation temperatures (38). Strongly
Catalyst Surface Area Measurements
ZrOx(OH)4 2x powders have a high surface area (289 m2 bound surface oxides inhibit intercrystallite sintering by re-
g
1) after drying at 383 K for 12 h. This surface area ducing the rate of ZrO2 surface diffusion (39). These surface
decreases markedly after oxidative treatments above the oxides may lose their effectiveness at high temperatures be-
ZrO2 crystallization temperature (723 K) (Fig. 3). The sin- cause they diffuse into the bulk, decompose and desorb, or
tering of ZrO2 crystallites is inhibited by impregnating “dewet” and agglomerate to form poorly interacting clus-
metatungstate ions onto ZrOx(OH)4
before oxidation. ters as a separate phase. At high oxidation temperatures
2x
1
The surface area ofthe 7.9wt.% W sample isabout 40m2 g
(>1173 K), surface WOx species agglomerate into mono-
greater than that in ZrO2 samples after oxidation between clinic WO3 crystallites and become less effective sintering
773 and 1173 K (Fig. 3). For example, after oxidation at inhibitors. SOx species also inhibit the loss of ZrO2 surface
1073 K, the surface area of pure ZrO2 is 6 m2 g 1, while the area after oxidative treatments (3). However, these species
1
surface area of a sample containing 7.9 wt.% W is 51 m2 g
.
decompose and desorb as SO3 at relatively low oxidation
FIG. 2. Powder X-ray diffraction results of WOx–ZrO2 samples showing stabilization of (a) the tetragonal ZrO2 phase and (b) the ZrO2 crystallite
size with various WOx concentrations after oxidation at 1073 K.

62
BARTON ET AL.
(0.8 W nm 2) and impregnation with excess metatungstate
ions (>7.9 wt.% W) leads to WOx species that are not ef-
fective at additional stabilization of ZrO2 crystallites. Thus,
it appears that these weakly bound WOx species may form
WOx clusters below WOx surface densities corresponding
to the theoretical monolayer. This idea of a maximum
number of strong binding sites is supported by the results of
Fuentes (42), who found that WOx–ZrO2 samples prepared
by coprecipitation of ZrOx(OH)4
and excess ammo-
2x
nium metatungstate attain maximum W concentration of
8 wt.% W, after washing with a NH4OH solution of pH 10,
a process that removes the weakly bound WOx species.
ZrO2 crystallite phase and size can be controlled by ad-
justing oxidation temperature and WOx loading. These ex-
perimental variables are useful in the design of catalysts
with a given WOx surface density but higher surface areas.
For example, a sample with 21 wt.% W oxidized at 823 K
(140 m2 g 1) has the same nominal WOx surface density
(5.0 W nm 2) as a 7.9 wt.% W sample oxidized at 1073 K
FIG. 3. N2 physisorption BET surface area data of ZrO2 and WOx–
ZrO2 (7.9 wt.% W) samples at various oxidation temperatures.
temperatures ( 873 K), leading to rapid ZrO2 sintering
above 873 K (40).
(50 m2 g 1). We have also detected by X-ray diffraction the
2
incipient formation of WO3 crystallites at about 9 W nm
,
The surface area of WOx–ZrO2 samples also depends on
WOx coverage. At a given oxidation temperature, surface
area increases with WOx concentration (Fig. 4). An asymp-
totic value of surface area is reached near 7.9 wt.% W after
oxidation at 873 or 1073 K. The nominal WOx surface den-
sities on 7.9 wt.% W samples oxidized at 873 and 1073 K are
2.3 and 5.0 W nm 2, respectively. These WOx surface den-
sities are below the theoretical monolayer capacity (7.3 W
atoms nm 2 (41)), yet further increases in WOx concentra-
tion do not lead to higher surface areas. The WOx concen-
tration required for maximum surface area (7.9 wt.% W) is
independent of oxidation temperature and corresponds to
submonolayer coverages. In addition, the maximum tetrag-
onal content was also detected at this WOx concentration.
These data suggest that a fixed number of strong binding
sitesfor metatungstate ionsexist on ZrOx(OH)4 2x powders
irrespective of surface area, WOx concentration, or oxida-
tion temperature. The important effects of WOx surface
density on catalytic activity and W6+ reducibility are dis-
cussed below.
Thermal Analysis
Weight changes during thermal decomposition of
ZrOx(OH)4 2x and WOx–ZrO2 (7.9 wt.% W) samples are
shown in Fig. 5. The ZrOx(OH)4
sample loses 19% of
2x
its weight during decomposition to ZrO2 and heating to
1073 K. This weight loss and the weight loss during ammo-
nium metatungstate decomposition ((NH4)6H2W12O40
WO3; 6.5 wt.% ) were used to calculate nominal WOx
concentrations (wt.% W) for WOx–ZrO2 samples after
→
FIG. 5. Thermogravimetric weight loss data of (A) ZrOx(OH)4
FIG. 4. N2 physisorption BET surface area data of WOx–ZrO2 sam- and (B) WOx–ZrOx(OH)4 2x (7.9 wt.% W) samples heated at 0.167 K s
ples at various WOx concentrations after oxidation at 873 and 1073 K.
in air.
2x
1

ZIRCONIA AND TUNGSTEN OXIDE
63
ZrOx surface diffusion rates are required to reach exposed
nucleation sites. This inhibition of ZrOx surface diffusion
by WOx species also prevents the rapid sintering of ZrO2
crystallites and loss of surface area as reported above.
Thermal analysis results for the WOx–ZrOx(OH)4
2x
samples heated to higher temperature (1273 K) have no
additional features that could arise from decomposition or
loss of WOx species. This thermal stability of WOx species
differs markedly from the poor stability of SOx species sup-
ported on ZrO2. Thermal analysis has detected SOx decom-
position ( 873 K) to volatile SO2 and SO3 products in air,
H2S in H2, and H2SO4 when H2O is present (40). Decom-
position to H2S occurs at even lower temperatures when Pt
crystallites are present. These difficulties limit the applica-
bility of SOx–ZrO2 catalysts, while the inherent stability of
WOx species permits catalyst regeneration and operation
at much higher temperatures.
FIG. 6. Differential scanning calorimetry data showing the influ-
ence of WOx concentration on ZrO2 crystallization exotherms for
(A) ZrOx(OH)4 2x and WOx–ZrOx(OH)4 2x samples with (B) 2.2 wt.% W,
(C) 4.5 wt.% W, (D) 7.9 wt.% W, (E) 16 wt.% W, and (F) 21 wt.% W heated
at 0.167 K s 1 in air.
Temperature-Programmed Reduction
Temperature-programmed reduction (TPR) studies of
supported metal oxides have been used extensively to pro-
vide semi-quantitative descriptions of domain sizes and
monolayer capacities for surface oxides (44). Dispersed
metal oxides tend to reduce at higher temperatures than
bulk crystallites because of strong M–O bonds with the ox-
ide surface and a lack of metal oxide nearest neighbors
to share the reduced charge. TPR methods are often use-
ful indicators of the onset of bulk metal oxide crystallite
formation, caused by a progression from isolated species
to two-dimensional structures and three-dimensional clus-
ters. These methods have been applied to MoO3 on many
supports (45) and to WO3 species on Al2O3 and SiO2 (46).
Temperature-programmed reduction profiles are shown
in Fig. 7 for bulk WO3, WOx–Al2O3, H3PW12O40 on SiO2,
and for a range ofWOx concentrationson ZrO2. Allsamples
oxidation. Elemental analysis (Galbraith Laboratories,
Inc., Knoxville, TN) of a WOx–ZrO2 sample oxidized at
1073 K verified the accuracy of this method (11.9 wt.% W
nominal vs 11.6 wt.% W analyzed) and confirmed that
WOx species were not lost by sublimation during high-
temperature oxidative treatments.
Differential scanning calorimetry (DSC) data (Fig. 6)
contain a single sharp exotherm at 679 K for a pure
ZrOx(OH)4
sample as a result of ZrO2 crystalliza-
2x
tion. ZrOx(OH)4
samples impregnated with ammo-
2x
nium metatungstate solutions have an additional broader
exotherm that increases in relative area and occurs at in-
creasing temperatures with increasing WOx concentration.
The integrated area of these two exotherms remains con-
stant when normalized by the weight of ZrO2 in these
samples ( 120 J g-ZrO2 1), which indicates that both of
these exotherms are due to ZrO2 crystallization. The rela-
tive area of the first exotherm is a measure of the fraction
of ZrOx(OH)4 2x crystallites that contain no WOx surface
species. It is evident that this area is small even for the
WOx–ZrOx(OH)4 2x sample with the lowest WOx concen-
tration (2.2 wt.% W, Fig. 6B), indicating that incipient wet-
ness impregnation leads to WOx species that are very well
dispersed.
The relative area of the second exotherm increases and
occurs at higher temperatures with increasing WOx con-
centration. WOx species on the surface of amorphous zir-
conia crystallites delay the onset of crystallization by bind-
ing to nucleation sites and by limiting the rate of ZrOx
surface diffusion, both of which are required for crystal-
lization (43). As a result, the crystallization temperature
of ZrO2 increases with increasing WOx concentration be-
cause more nucleation sites become blocked and higher
FIG. 7. Temperature-programmed reduction profiles of WOx–ZrO2
samples (0.025–0.200 g) after oxidation at 1073 K: (A) 2.2 wt.% W,
(B) 0.3% Pt–7.9 wt.% W, (C) 9.5 wt.% W, (D) 11.6 wt.% W, and
(E) 21 wt.% W and W standards, (F) 12-tungstophosphoric acid on SiO2
(30 wt.% W), (G) bulk WO3, and (H) 7.9 wt.% W/Al2O3 heated at 0.167 K
s
1 (1.33 cm3 s 1, 20% H2 in Ar).

64
BARTON ET AL.
reduced to W metal at temperatures below 1323 K, ex-
cept WOx–Al2O3. Pure ZrO2 samples showed no signi-
ficant H2 consumption up to 1323 K. Three overlapping
peaks were detected for the reduction of bulk WO3 sam-
ples (Fig. 7G), which occurs via stepwise reduction: WO3 →
WO2.9 →WO2 →W (46).
The reduction profiles for WOx–ZrO2 samples are very
broad (800 K) and have four overlapping peaks (Figs. 7A–
7E) that occur at similar temperaturesat allWOx concentra-
tions. The first three peaks correspond to the same three re-
duction steps found in bulk WO3. A direct comparison with
bulk WO3, however, is impaired by marked differences in
crystallite size between bulk and supported samples. An ad-
ditional reduction step occurs at high temperature (1200 K)
in WOx–ZrO2 samples; it corresponds to the reduction
of WOx species strongly bound to the ZrO2 support
(WO2.9 →W). These WOx species reduce at higher tem-
peratures because they contain unreactive oxygen link-
ages to the ZrO2 support. As WOx concentration increases,
the relative area of this high-temperature reduction peak
decreases, which is consistent with a decreasing fraction
of strong W–O–Zr bonds and increasing fraction of WOx
species in domains that reduce at lower temperatures.
The second reduction step (WO2.9 →WO2) is sensitive
to the size of the WOx domains. At low WOx concentra-
tions, a large fraction of the WOx domains bypass this sec-
ond reduction step and instead reduce directly to W metal
(WO2.9 →W) in the third reduction step; apparently there
are not enough WOx neighbors in these small domains to
form the stable intermediate WO2 species. The relative area
of this second reduction step increases with increasing WOx
concentration and ultimately approaches that for stoichio-
metric reduction to WO2 as in WOx–ZrO2 samples con-
taining primarily bulk WO3 crystallites (21 wt.% W, 1073 K
oxidation). Thus, the increase in the area of this second re-
duction step suggests a growth of WO3-like domains with
W–O–W linkages as WOx concentration increases; these
domains reduce at lower temperatures than isolated WOx
species.
WOx species on ZrO2 reduce at significantly lower tem-
peratures than WOx on Al2O3 at similar surface densi-
ties. For example, reduction of the 2.2 wt.% W on ZrO2
(2.7 W nm 2) sample is complete by 1350 K, whereas re-
duction of the 6 wt.% W on Al2O3 (2.4 W nm 2) sample is
not finished at 1400 K. At low WOx surface densities, the
reduction temperature of these dispersed WOx species is
determined by the strength of the W–O bonds that contain
oxygens shared with the support. The higher reduction tem-
perature of the WOx–Al2O3 sample suggests that the W–O–
Al bonds are stronger than the W–O–Zr bonds. At higher
WOx surface densities, the WOx polyhedra are connected
by W–O–W bonds, and the first two reduction steps of these
polytungstate domains on ZrO2 occur at temperatures sim-
SCHEME 1. Generation of Brønsted acid sites in hydrogen by slight
reduction and delocalization of negative charge in WOx domains.
heteropolyacids (H3PW12O40). This suggests that the onset
of reduction of polytungstate domains is not influenced by
the support when there are enough WOx neighbors to share
the reduced charge and bridging W–O–W bonds that can
be broken.
The addition of Pt (0.3 wt.% ) to WOx–ZrO2 (12.7 wt.%
W) strongly affects the first reduction step (WO3 →WO2.9),
which shifts from 700 to 350 K (Fig. 7B). This suggests that
slight reduction can occur during H2 pretreatment and dur-
ing n-alkane reactions at 440–500 K. The ability of WOx
domains to undergo slight reduction by delocalizing the
net negative charge may be linked with the strong acid-
ity appearing on WOx domains on ZrO2, for which the
domain sizes become large enough to delocalize the elec-
tron charge over several W atoms. These reducible WOx
domains may act as redox sites required for the forma-
tion of H+ species from H2 and alkanes on WOx-based
solid acids (Scheme 1). The accommodation of a proton
by electron transfer and charge delocalization across an ex-
tended W–O network results in electronic structures simi-
lar to those of heteropolyacids (47). This simple mechanism
may also account for the ability of H atoms to desorb car-
bocations as neutral hydrocarbons and to regenerate neu-
tral WO3 domains (Scheme 2) and for the observed effects
of hydrogen on the rate and selectivity of n-heptane and
o-xylene isomerization reactions.
X-Ray Absorption Spectroscopy
W-LI near-edge spectra for Na2WO4 and (NH4)6H2W12O40
standards and for a WOx–ZrO2 sample (11.7 wt.% W)
before and after high-temperature oxidation (1073 K) are
shown in Fig. 8. The pre-edge feature at 10 eV reflects
a 2s → 5d electronic transition that is dipole-forbidden
in centrosymmetric structures, but becomes allowed in
SCHEME 2. Stabilization of carbocations by slight reduction of WOx
ilar to those required for the reduction of WOx domains in domains.

ZIRCONIA AND TUNGSTEN OXIDE
65
FIG. 8. Normalized W-LI near-edge spectra of WOx standards and
WOx–ZrO2 samples: (A) dehydrated Na2WO4, (B) 11.6 wt.% W after
1073K, (C) 11.6wt.% W after 423K, and (D) (NH4)6H2W12O40. Difference
spectra at right, with equivalent vertical scale, calculated by subtracting
the WO3 spectrum from the sample spectrum. Spectra recorded in situ at
298 K after dehydration at 423 K.
FIG. 9. Normalized W-LI near-edge spectra of WOx–ZrO2 samples
(1073 K oxidation): (A) 2.2 wt.% W, (B) 11.6 wt.% W, (C) 0.3 wt.% Pt–
12.7 wt.% W, (D) 21 wt.% W, and (E) bulk WO3. Difference spectra at
right, with equivalent vertical scale, calculated by subtracting the WO3
spectrum from the sample spectra. Spectra recorded in situ at 298 K after
dehydration at 723 K.
tetrahedral or distorted octahedral structures because of
p–d orbital mixing (48). Thus, changes in the intensity of this
pre-edge feature reflect changes in the symmetry of the W
absorber (49). This feature has been used to determine the
distribution of W among octahedral and tetrahedral struc-
tures in Al2O3- and TiO2-supported WOx catalysts (50, 51).
These studies have concluded that W atoms can be 4-, 5-,
or 6-coordinate depending on the W coverage and the state
containing different WOx concentrations and bulk WO3.
W centers in bulk WO3 crystallites reside in relatively low
symmetry positions. For example, in monoclinic WO3 there
are 24 different W–O distances per unit cell, ranging from
˚
1.715 to 2.215 A (52). Because of this structural disorder,
coordination numbers and distances derived from the W
radial structure functions are poorly defined. Figure 10
shows, however, that WOx structures present on the sur-
face of ZrO2 at all WOx concentrations have a static dis-
order similar to that in bulk WO3. W-LI near-edge spectra
and W-LIII radial distribution functions in all WOx–ZrO2
samples resemble those of bulk WO3; therefore, W atoms
reside in structures with distorted octahedral symmetry re-
sembling that in bulk WO3 for all WOx concentrations and
pretreatment conditions. As a result, significant differences
in catalytic activity within these samples cannot be simply
of hydration. The spectrum of a ZrOx(OH)4
sample
2x
impregnated with an aqueous ammonium metatungstate
solution and dried in situ at 423 K (Fig. 8C) differs from
the spectrum of pure ammonium metatungstate (Fig. 8D),
indicating that a reaction occurs between the aqueous
metatungstate ions and the ZrOx(OH)4
surface during
2x
impregnation. After oxidation at 1073 K, which sinters the
ZrO2 crystallites and markedly increases the WOx surface
density, the spectrum changes only slightly (Fig. 8B), sug-
gesting that the local structure and symmetry of the WOx
species does not change significantly after oxidative treat-
ments that lead to catalytically active WOx–ZrO2 samples.
W-LI near-edge spectra for bulk WO3 and WOx–ZrO2
samples oxidized at 1073 K with various WOx concentra-
tions are shown in Fig. 9. All spectra were measured at room
temperature after in situ dehydration in He at 823 K. The
similarity among the three WOx–ZrO2 spectra and the spec-
trum of bulk WO3 is confirmed by subtraction of the bulk
WO3 spectrum from each of the three WOx–ZrO2 spectra
(Fig. 9). These difference spectra differ markedly from the
difference spectrum of Na2WO4 that contains W atoms only
in tetrahedral sites (Fig. 8A). The difference spectra show
that WOx structures have a distorted octahedral symmetry
that is insensitive to WOx concentration in a concentration
range that strongly influences catalytic behavior.
FIG. 10. W-LIII radial structure functions, not phase or amplitude cor-
rected, of WOx–ZrO2 samples after oxidation at 1073 K: (A) 2.2 wt.% W,
(B) 11.6 wt.% W, (C) 0.3 wt.% Pt–12.7 wt.% W, (D) 21 wt.% W, and
The W-LIII fine structure of the samples in Fig. 9 was used
to obtain the radial structure functions shown in Fig. 10.
There are only slight differences among the four samples (E) bulk WO3. Spectra recorded in situ at 298 K after dehydration at 723 K.

66
BARTON ET AL.
explained by differences in the local symmetry of the WOx
octahedra, but may be explained instead by differences in
the size of the WOx extended domain. We discuss how rates
of o-xylene isomerization relate to the WOx domain size de-
termined by UV–visible spectroscopy below.
TABLE 1
Irreversible Chemisorption Uptakea of H2, O2, and CO
on Pt/WOx–ZrO2 (0.3 wt.% Pt, 10 wt.% W)
Reduction
temperature (K)
H/Pt
O/Pt
CO/Pt
Catalytic Isomerization of n-Heptane
473
673
0.63
0.092
1.25
2.04
0.30
0.050
The addition of a metal function (0.3 wt.% Pt) to WOx–
ZrO2 leads to high n-heptane isomerization turnover rates
and selectivities at low temperatures (18). The isomeriza-
tion selectivity at 473 K and 50% n-heptane conversion
is 85% on Pt/WOx–ZrO2. These isomerization selectivi-
ties are much higher than on Pt/SOx–ZrO2 (<50% ) and
on other mesoporous acid catalysts. Here, we expand our
previously reported results (18) in order to contrast these
resultswith the effectsof H2 on o-xylene isomerization cata-
lyzed by WOx–ZrO2 samples without a metal function.
On Pt/WOx–ZrO2, n-heptane isomerization turnover
rates depend weakly on H2 gas phase concentration.
Turnover rates on Pt/WOx–ZrO2 become negative order
in H2 (0 → 0.5) as H2 pressures increase. These data are
consistent with quasi-equilibrated hydrogen transfer steps,
which lead to short surface lifetimes, low steady-state car-
bocation coverages, and desorption of carbocations before
-scission occurs. On Pt/WOx–ZrO2, n-heptane isomeriza-
tion occurs on a surface sparsely populated with carboca-
tions and surface isomerization steps are rate-determining
(Scheme 3). In effect, carbocation formation and desorp-
tion by hydrogen transfer are quasi-equilibrated and car-
bocation coverages are proportional to n-heptane pres-
sure, leading to the observed first-order dependence on
n-heptane concentration. The negative H2 pressure order
arises from competitive binding of H+ and carbocations
at WOx sites, which causes carbocation surface densities
to decrease with increasing H2 pressure. On Pt/SOx–ZrO2,
n-heptane isomerization turnover rates have a positive
H2 order (0.9) and they increase when hydrogen transfer
cocatalysts, such as adamantane, are added to n-heptane re-
actants. Slow hydrogen transfer reactions delay carbocation
desorption and lead to long carbocation surface residence
times and low isomer selectivity (6).
a
Room temperature adsorption; irreversibly chemisorbed species are
defined as those that cannot be removed by evacuation at 313 K.
desorption pathways, which prevent extensive cracking of
adsorbed carbocations. This mechanistic conclusion is con-
firmed by the rapid isotopic exchange between n-C7H16 and
D2 observed on Pt/WOx–ZrO2 during n-heptane isomeriza-
tion (18). H–D equilibrium between n-C7H16 and D2 occurs
much more rapidly than isomerization on Pt/WOx–ZrO2.
These data show that n-heptane adsorbs, exchanges with a
surface hydrogen pool initially consisting of D atoms, and
frequently desorbs before isomerization occurs.
H2 dissociation is required before it can act as a source
of hydrogen atoms. Reduced Pt sites on Pt/WOx–ZrO2 dis-
sociate and store H atoms required for hydrogen transfer
and carbocation desorption steps. As a result, Pt clusters
on Pt/WOx–ZrO2 catalysts chemisorb significant amounts
of H2 at room temperature and catalyze exchange of H
atoms between D2 and “unreacted” n-heptane.
Pt dispersion measurements after H2 reduction at 473 K
using H2, O2, and CO site titrations show that apparent Pt
dispersions measured by hydrogen uptake exceed those ob-
tained from CO chemisorption (Table 1), suggesting that
hydrogen adatoms spillover onto WOx centers during H2
chemisorption at room temperature. O2 uptakes are much
higher than those of H2 or CO, suggesting that WOx species
reduce slightly during H2 treatments at typical isomeriza-
tion reaction temperatures. At higher reduction tempera-
tures, H2 uptakes decrease and O2 uptakes increase; these
phenomena appear to be related to the decoration of Pt
crystallites by reduced WOx, as also proposed to explain
similar effects on TiO2 supported metal particles (53). Thus,
hydrogen atoms formed at H2 dissociation sites may be-
come involved not only in the desorption of adsorbed in-
termediates, but also in the generation and maintenance
of Brønsted acid sites. The formation of Brønsted acid sites
during H2 treatment has been detected by Ebitani et al. (54)
using infrared measurements of the density of Lewis and
Brønsted acid sites in Pt/SOx–ZrO2. In their study, H2 treat-
ments at temperatures typical of catalytic reactions lead to
an increase in the density of Brønsted acid sites and to a
decrease in the density of Lewis acid sites, suggesting that
Lewis acid sites are converted to Brønsted acid sites by
contact with H2. We suggest that a similar mechanism leads
to the formation of Brønsted acid sites on Pt/WOx–ZrO2
Pt/WOx–ZrO2 catalyzes H2 activation and storage of H
atoms and provides rapid carbocation neutralization and
SCHEME 3. n-Heptane isomerization on Pt/WOx–ZrO2.

ZIRCONIA AND TUNGSTEN OXIDE
67
and WOx–ZrO2 samples in the presence of H2 at reaction
temperatures. As a result, attempts to measure the number
and nature of the acid sites on these materials must be done
under reaction conditions.
Catalytic Isomerization of o-Xylene
Acid-catalyzed skeletal isomerization of o-xylene
(Eq. [4]) was used to probe the acid properties of WOx–
ZrO2 samples. Xylene isomerization reactions proceed via
intramolecular 1,2 methyl shifts on Brønsted acid sites (55).
Bimolecular pathways leading to disproportionation side
products (toluene and trimethylbenzenes) can also lead to
isomerization of xylenes (56). In the absence of intrapellet
diffusional restrictions and bimolecular pathways, the rel-
ative selectivity to m- and p-xylenes extrapolated to zero
conversion reflects benzenium ion surface lifetimes. Short
FIG. 12. Influence of hydrogen on o-xylene transformations on a
WOx–ZrO2 (12 wt.% W) sample oxidized at 1073 K (reaction conditions:
surface lifetimes lead to m-xylene, a product formed by sin- 0.67 kPa o-xylene, 106 kPa H2 or He, 523 K). Reactor loaded with 0.0287 g
for (A) and (B) and run in H2. Reactor reloaded with 0.0293 g for (C), (D),
and (E) and run in He. (B) and (D) were run immediately after (A) and
(C) without further pretreatment. (E) was run after oxidative treatment.
See Table 2 for pretreatment details.
gle methyl shifts; longer surface residence times lead to xy-
lene isomer distributions near thermodynamic equilibrium
(57)
under these reaction conditions. Tetramethylbenzene, a dis-
proportionation product, became detectable only at high
[4]
conversions. At low conversions, typical m- to p-xylene
ratios are about 5.6 (thermodynamic equilibrium is 2.3 at
523 K) and disproportionation to isomerization ratios are
0.18. Thishigh m-xylene selectivityreflectsshort benzenium
ion lifetimes. o-Xylene isomerization turnover rates, based
on the total number of W atoms, were used to compare the
relative catalytic activity of the WOx–ZrO2 samples. Nor-
malization of the isomerization rates by the total number of
W atoms is the most conservative estimation of the “true”
acid site turnover rate and is a convenient choice in the
absence of an accurate measurement of the acid site den-
sity in the presence of H2 under reaction conditions. Initial
o-xylene isomerization rates are reported at 5–10% conver-
sion in order to avoid including any initial transients.
o-Xylene reactive turnovers (per W atom) are shown as
a function of contact time in Fig. 12 for a 12 wt.% W sample
oxidized at 1073 K after various H2 and He pretreatments
listed in Table 2. The data in Fig. 12A correspond to the
o-Xylene isomerization was carried out in a recirculating
batch reactor at 523 K (0.67 kPa o-xylene; 106 kPa H2). A
typical plot of selectivity as a function of o-xylene conver-
sion on a 12 wt.% W sample oxidized at 1073 K is shown in
Fig. 11. Aromatic hydrogenation, hydrogenolysis, and xy-
lene oligomerization reaction products were not observed
TABLE 2
Pretreament Conditions and Diluent Gas Composition
for o-Xylene Isomerization Turnover Data in Fig. 12
Figure 12 data
Diluent
Drying
Reduction
A
B
C
D
E
H2
H2
He
He
He
823 K, 2 h
None
823 K, 2 h
None
523 K, 2 h
None
523 K, 2 h
None
FIG. 11. Selectivity as a function of o-xylene conversion on a WOx–
ZrO2 sample (12 wt.% W) oxidized at 1073 K (reaction conditions:
0.67 kPa o-xylene, 106 kPa H2, 0.0287 g catalyst, 523 K).
823 K, 2 h
523 K, 2 h

68
BARTON ET AL.
activity, but again deactivated rapidly. The initial activity in
He appears to be caused by a sacrificial loss of hydrogen
from neighboring adsorbed species, which forms Brønsted
acid sites and allows xylene turnovers to occur, but which
leaves behind strongly adsorbed unsaturated species. The
irreversible buildup of these carbonaceous species leads to
the rapid deactivation during o-xylene conversions in He.
After high-temperature oxidative treatments these samples
completely recovered their initial activity.
The mechanism of H2 promotion in xylene isomerization
reactions was probed by replacing H2 with D2 and moni-
toring the rate of exchange between xylenes and D2 during
o-xylene isomerization. The deuterium contents (Fig. 13)
and the deuterium distributions (Fig. 14) in o-xylene,
m-xylene, and p-xylene are identical, indicating that the ad-
sorption of o-xylene that leads to an isomerization turnover
does not lead to additional deuterium exchange. The rate
of deuterium exchange per W atom (2.3 10 4 s 1) is much
FIG. 13. D2/o-xylene exchange on a WOx–ZrO2 (12 wt.% W) sample
oxidized at 1073 K (reaction conditions: 0.67 kPa o-xylene, 106 kPa D2,
523 K).
experiment in Fig. 11, which was run in H2 at 523 K after lower than o-xylene isomerization turnover rates (1.7
treatment in dry air for 2 h at 823 K and in H2 for 2 h at 10 3 s 1). For example, only 1.2% of the H atoms in xylenes
523 K. Stable isomerization rates were confirmed by flush- have exchanged at 21% o-xylene chemical conversion. The
ing the system with H2 for 1 h at 523 K, refilling the reactor deuterium content in the m-xylene and p-xylene isomers
with o-xylene/H2, and remeasuring o-xylene isomerization increased with contact time and showed a binomial dis-
rates (Fig. 12B). Minimal deactivation was detected on this tribution, suggesting that the surface hydrogen/deuterium
sample after o-xylene reactions to near equilibrium con- species are not in equilibrium with the D2 in the gas phase.
version for 2 h, as shown by a nearly complete recovery of As a result, H atoms formed by H2 dissociative chemisorp-
catalytic activity flushing the system with H2. The influence tion are not frequently involved in the activation or desorp-
of H2 on o-xylene conversions was tested by replacing H2 tion of the xylene reactants. The presence of H2 is required,
with He (Fig. 12C) as an inert diluent. The initial turnover however, in order to reverse the occasional desorption of
rate was lower in He and rapidly decreased as the sam- H atoms as H2, which can occur during o-xylene isomeriza-
ple deactivated. This significant deactivation is shown by tion reactions by the microscopic reverse of the dissociative
the very low rates in Fig. 12D, which was the experiment chemisorption step. These desorption processes lead to the
immediately following flushing the system with He (no ox- disappearance of Brønsted acid sites and to the formation
idation) and refilling with o-xylene/He. After these two ex- of strongly adsorbed unsaturated species when xylene iso-
periments in He, the sample was reoxidized for 2 h at 823 K merization iscarried out in the absence ofH2. H2 can restore
and treated in H2 for 2 h at 523 K and tested again with He as these Brønsted acid sites by rereduction of WOx species or
an inert diluent (Fig. 12E). The sample recovered its initial by formation of acidic HxWO3 species.
FIG. 14. D distribution in o-xylene and in the combination m- and p-xylenes compared to a statistical distribution of 2.4% D atoms.

ZIRCONIA AND TUNGSTEN OXIDE
69
FIG. 16. Influence of oxidation temperature on o-xylene isomeriza-
tion turnover rates on WOx–ZrO2 samples at three WOx concentrations
(5, 12, and 21 wt.% W) (reaction conditions: 0.67 kPa o-xylene, 106 kPa
H2, 523 K).
FIG. 15. o-Xylene isomerization turnover rates (per W atom) as a
function of WOx concentration on WOx–ZrO2 samples oxidized at 1073 K
(reaction conditions: 0.67 kPa o-xylene, 106 kPa H2, 523 K).
o-Xylene isomerization rates on WOx–ZrO2 samples
with various WOx concentrations and oxidation tempera-
tures are shown in Figs. 15 and 16. o-Xylene isomerization
turnover rates for samples oxidized at 1073 K reach a maxi-
mum value at a WOx concentration of 12 wt.% W (Fig. 15),
which is similar to that reported by Hino and Arata [12] and
later by Santiesteban et al. (58) for pentane isomerization
reactions. Isomerization rates were negligible on the sample
with the lowest WOx concentration (2.2 wt.% W). The sam-
ple with the highest WOx concentration tested (21 wt.% W)
gave low but measurable rates, even though it contains pre-
dominately large WO3 crystallites. Clearly, isomerization
rates are not proportional to the number of WOx sites, but
they are strongly affected by the coverage of WOx on ZrO2.
WOx coverage and isomerization rates also depend
strongly on oxidation treatment temperatures. ZrO2 crys-
tallites grow and surface areas decrease with increasing ox-
idation temperature, which increases the WOx surface den-
sity at a given WOx concentration. Three freshly prepared
WOx–ZrO2 samples (5, 12, and 21 wt.% W) that had only
been dried at 423 K were oxidized in the quartz reactor
at increasing temperatures from 800 to 1300 K. o-Xylene
isomerization rates were measured on each sample with-
out removing them from the reactor after each oxidation
treatment. After catalytic measurements, the reactor was
flushed with H2 at 523 K for 1 h, evacuated, and heated
in flowing dry air to the next highest oxidation tempera-
ture for 2 h. Isomerization rates (per W atom) on these
three samples at several oxidation temperatures are shown
in Fig. 16. Each sample reaches maximum isomerization
turnover rates at intermediate oxidation treatment tem-
peratures. The sample with the highest WOx concentration
(21 wt.% W) reaches its maximum isomerization rate
(2.1 10 3 s 1) after oxidation at 973 K. Samples with lower
W content reach maximum rates only after oxidation at
concentrations in WOx–ZrO2 samples decreases the oxi-
dation temperature required to reach maximum catalytic
activity and leads to samples with higher surface areas. It
appears that a critical WOx surface density is required in or-
der to obtain a maximum number of acid sites for o-xylene
isomerization.
Surface areas were measured for these samples at each
oxidation temperature and used to convert WOx concen-
trations to nominal WOx surface densities (W nm 2). When
isomerization ratesfor allsamplesfrom Fig. 16are replotted
(Fig. 17) as a function of this WOx surface density, they
merge into a single curve with a very sharp maximum rate
at about 10 W nm 2. It is apparent that the acid site density
depends only on WOx surface density and that it is indepen-
dent of the oxidation temperature and WOx concentration
FIG. 17. o-Xylene turnover rates replotted from Fig. 16 as a func-
tion of nominal WOx surface density on three WOx–ZrO2 (5, 12, and
21 wt.% W) samples oxidized at various temperatures (800–1300 K). The
higher temperatures (1100–1200 K). Thus, increasing WOx theoretical monolayer capacity for ZrO2 (7 W nm 2) is marked.

70
BARTON ET AL.
used to obtain a particular WOx surface density. As a re-
sult, a WOx–ZrO2 sample with the highest isomerization
1
rate per catalyst weight (1.8 1019
s
g
1) was found on
a sample with the highest WOx concentration (30 wt.% W,
898 K oxidation, 104 m2 g 1) and was limited only by the
solubility of ammonium metatungstate in water.
The critical WOx surface density at which maximum
activity is obtained is higher than the theoretical mono-
layer capacity of ZrO2 (7 W nm 2), calculated by assuming
corner sharing WOx octahedra with bond distances de-
rived from monoclinic WO3; therefore, the most active
WOx–ZrO2 samples must contain a fraction of the WOx
species in extended three-dimensional networks. At high
WOx surface densities (>10 W nm 2), bulk WO3 has been
detected by X-ray diffraction and therefore isomerization
turnover rates normalized to the total number of W
atoms must decrease because a large fraction of W atoms
reside inside the bulk of these crystallites and become
inaccessible to hydrocarbon reactants. At low WOx surface
densities (<5 W nm 2), all the WOx groups are accessible,
and X-ray absorption results indicate that the W atoms
reside in a distorted octahedral symmetry similar to that of
the most active species. Isomerization rates are, however,
very low, suggesting that it is the growth of polytungstate
species at WOx surface densities higher than a monolayer
that is required for the formation of acid sites. This narrow
range of WOx surface densities for which WOx–ZrO2
samples become catalytically active suggests that high
isomerization turnover rates require the presence of WOx
domains of intermediate size, which form selectively as
WOx surface densities reach intermediate values.
Previous analysis of optical absorption edge energies
from UV–vis diffuse reflectance spectra for WOx–ZrO2
sampleswith WOx surface densitybelow10W nm 2 showed
that the number of WOx nearest neighbors and the WOx
domain size increase with increasing WOx surface density
(18). Optical absorption edge energies were determined by
plotting ( h )2 vs h ( , absorbance; h , photon energy),
which assumes that the fundamental optical absorption is
the result of a direct electronic transition. The fundamen-
tal optical absorption energy for bulk WO3 crystallites and
amorphous WOx species, however, has been shown to be
better approximated by plotting ( h )1/2 vs h , which re-
sults in the indirect absorption edge energy (59). Reanalysis
of the absorption spectra based on the indirect transition
formalism gives lower edge energies, but the trends remain
unchanged (Fig. 18). The position of the UV–visible absorp-
tion edge is sensitive to the domain size of transition metal
oxides, which is related to the number of nearest metal ox-
ide neighbors sharing a common oxygen (M–O–M bonds)
(60). An increase in the domain size results in a decrease in
the absorption edge energy. The decrease of the absorption
edge with increasing WOx surface density between 4 and
10 W nm 2 reflects an increase in the WOx domain size. It is
FIG. 18. WOx domains with intermediate size lead to maximum
o-xylene isomerization turnover rates. WOx indirect absorption edge en-
ergies determined from absorption data for WOx–ZrO2 samples in (18)
are plotted over the o-xylene isomerization rate curve from Fig. 17. The
indirect absorption edge of monoclinic WO3 is marked for reference.
evident when the o-xylene isomerization rate curve is plot-
ted over the UV–visible absorption edge energies (Fig. 18)
that increasing WOx domain sizes lead to increasing iso-
merization rates at low surface densities (4–10 W nm 2).
At higher surface densities (>10 W nm 2), bulk WO3 was
detected by a second electronic transition edge and isomer-
ization rates per W atom decrease as a result of a decrease
in WOx dispersion.
The unique acid properties of WOx domains of inter-
mediate size appear to be related to their ability to form
W6 nOx–(n-H+) Brønsted acid centers under reducing con-
ditions at low temperatures (Scheme 1 without Pt). Iso-
lated WOx species that exist at low WOx surface densities
(<5 W nm 2) cannot delocalize the negative charge re-
quired for the formation of Brønsted acid centers, whereas
polytungstate species that form at intermediate WOx sur-
face densities reside in an extended network of WOx species
over which an excess negative charge may be delocalized.
The ability of these WOx species to temporarily reduce
and reversibly adsorb hydrogen was measured by hydro-
gen chemisorption uptakes at 523 K. The reversible H2
uptakes are plotted on the o-xylene isomerization rate
curve (Fig. 19) and show that reversible H2 adsorption oc-
curs at reaction temperatures. Maximum hydrogen uptakes
were recorded on WOx–ZrO2 samples that show maximum
catalytic activity, suggesting that the formation of these
W6 nOx–(n-H+) reduced centers is related to the stabiliza-
tion of carbocationic reaction intermediates. WOx species
present at low surface densities are difficult to reduce and
as a result are not able to form these Brønsted acid cen-
ters required for xylene reactions. Extended WOx domains
that form at higher WOx surface densities can delocalize
the negative charge required in order to stabilize positively

ZIRCONIA AND TUNGSTEN OXIDE
71
tallites and increase the WOx surface density. Surface area
and X-ray diffraction measurements show that dispersed
WOx species inhibit ZrO2 crystallite sintering and stabilize
tetragonal ZrO2 crystallites during high-temperature ox-
idative treatments. UV–visible and X-ray absorption spec-
troscopic studies of precursor and oxidized samples sug-
gest that oxyhydroxide surfaces lead to dispersed WOx
species (<7.9 wt.% W) after initial contact with aqueous
metatungstate solutions and drying at 400 K. X-ray absorp-
tion spectra at the W-LI edge show the predominant pres-
ence of distorted octahedral species similar those in bulk
WO3 at all WOx surface densities. The growth of WOx do-
mains as WOx surface density increases was detected by a
decrease in the absorption edge energies measured by UV–
visible spectroscopy. Temperature-programmed reduction
analysis of WOx–ZrO2 samples show marked changes in
reduction behavior with increasing WOx surface density.
As the WOx surface density increases, WOx species con-
vert from strongly bound WOx species to two- and three-
dimensional structures, which reduce at lower tempera-
tures.
The addition of a metal function (0.3% wt. Pt) increases
the turnover rate (based on W atoms) and selectivity for n-
heptane isomerization. Quasi-equilibrated hydrogen trans-
fer elementary steps during n-heptane isomerization reac-
tions decrease the surface residence time of intermediates
and lead to desorption of isomers before -scission events
occur on the acid sites. H2, O2, and CO chemisorption up-
takes indicate that H atoms spillover onto WOx species
forming W6 nOx–(n-H+) Brønsted acid centers under re-
action conditions. H atoms formed at H2 dissociation sites
become involved not only in the desorption of adsorbed
intermediates, but also in the generation and maintenance
of Brønsted acid sites.
Maximum o-xylene isomerization turnover rates were
found on WOx domains of intermediate size at WOx sur-
face densities (10 W nm 2) greater than monolayer cov-
erage on ZrO2. Turnover rates depend only on the WOx
surface density, and not independently on the oxidation
temperature and WOx concentration used to obtain a spe-
cific density. WOx domains of intermediate size appear to
provide a compromise between reducibility and accessibil-
ity of these WOx domains. At low coverages, turnover rates
increase with increasing WOx surface density, which is re-
lated to the pronounced ability of the polyoxoanion do-
mains to form W6 nOx–(n-H+) Brønsted acid centers. As
WOx surface density is further increased, W-atom turnover
rates decrease because increasing amounts of WOx spe-
cies reside inside WO3 crystallites. H2 is required during
o-xylene isomerization reactions on WOx–ZrO2 to prevent
the occasional desorption of H atoms as H2, which leads to
the destruction of these Brønsted acid sites via the micro-
scopic reverse of the dissociative chemisorption step. These
desorption processes lead to the disappearance of Brønsted
FIG. 19. Reversible H2 uptakes at 523 K and 45 kPa H2 on WOx–ZrO2
samples oxidized at 1073 K, reduced in H2 at 523 K for 1 h, and evacuated
at 523 K, plotted over o-xylene isomerization rates from Fig. 17.
charged species. This delocalization of charge among sev-
eral WOx groups to form W6 nOx–(n-H+) acid centers is
similar to that previouslysuggested for the generation ofthe
strong acid sites in heteropolytungstate clusters (47). In 12-
tungstophosphoric acid, the neutral (WO3)12 shell around
the central PO34 anion delocalizes the negative charge over
the surface of the shell, which is balanced by three acid hy-
drogens with a net positive charge to maintain electroneu-
trality of the cluster.
o-Xylene isomerization turnover rates are strongly influ-
enced by WOx surface density. Two independent param-
eters, WOx concentration and oxidation temperature, are
available in order to control the WOx surface density. Even
though turnover rates depend on WOx surface density, they
are not influenced by the specific oxidation temperature
or WOx concentration required in order to obtain a given
WOx surface density. Maximum turnover rates were ob-
tained on WOx domains of intermediate size (10 W nm 2).
These WOx domains of intermediate size may provide a
compromise between the reducibility and the accessibility
of WOx species in supported domains. H2 is required during
o-xylene isomerization reactionson WOx–ZrO2 for the gen-
eration and maintenance of W6 nOx–(n-H+) Brønsted acid
centers. For this reason, measurements of the Brønsted acid
site density must be carried out under reaction conditions in
order to accurately determine the site density present dur-
ing reaction and to calculate true isomerization turnover
rates.
CONCLUSIONS
WOx–ZrO2 catalysts active for isomerization reactions
at low temperature were prepared by impregnation of
ZrOx(OH)4 2x with a solution of ammonium metatungstate
followed by oxidation treatments that sinter the ZrO2 crys-

72
BARTON ET AL.
acid sites and to the formation of strongly adsorbed unsat- 23. Pines, H., “The Chemistry of Catalytic Hydrocarbon Conversions.”
Academic Press, New York, 1981.
24. Iglesia, E., Baumgartner, J. E., Ribeiro, F. H., and Boudart, M., J. Catal.
131, 523 (1991).
urated species that deactivate the catalysts.
ACKNOWLEDGMENTS
25. Rijnten, H. T., Ph.D. thesis, Delft University of Technology, 1971.
26. Crucean, E., and B., R., Trans. J. Brit. Ceram. Soc. 78, 58 (1979).
27. Gimblett, F. G. R., Rahman, A. A., and Sing, K. S. W., J. Colloid
The authors acknowledge financial support for this work from the
National Science Foundation (CTS-9510575). X-Ray absorption experi-
Interface Sci. 84, 337 (1981).
ments were performed at the Stanford Synchrotron Radiation Laboratory
28. Smith, J. M., “Chemical Engineering Kinetics,” 2nd ed. McGraw–Hill,
(SSRL) and supported, in part, by the U.S. Department of Energy. We also
New York, 1970.
thank Messrs. W. E. Gates and J. E. Baumgartner of Exxon Research and
Engineering Co. for their help with the high-pressure n-heptane isomer-
29. Schreiner, W. N., and Jenkins, R., Adv. X-Ray Anal. 26, 141 (1983).
30. Toraya, H., Yoshimura, M., and Somiya, S., J. Am. Ceram. Soc. 67, C
ization and D /n-heptane exchange experiments.
2
(1984).
31. Klug, H. P., and Alexander, L. E., “X-Ray Diffraction Procedures.”
REFERENCES
Wiley, New York, 1974.
32. Sayers, D. E., Lytle, F. W., and Stern, E. A., Phys. Rev. Lett. 27, 1204
1. Thomas, J. M., Sci. Am. 266, 112 (1992).
(1971).
2. Tanabe, K., Hattori, H., and Yamaguchi, T., Crit. Rev. Surf. Chem. 1, 33. Meitzner, G., and Sinfelt, J. H., Catal. Lett. 30, 1 (1995).
1 (1990).
34. Iglesia, E., Baumgartner, J. E., Price, G. L., Rose, K. D., and Robbins,
J. L., J. Catal. 125, 95 (1990).
35. Price, G. L., and Iglesia, E., Ind. Eng. Chem. 28, 839 (1989).
3. Arata, K., Adv. Catal. 17, 165 (1990).
4. Holm, V. C. F., and Bailey, G. C., U.S. Patent 3,032,599, 1962.
5. Hollstein, E. J., Wei, J. T., and Hsu, C. Y., Sun Refining and Marketing 36. Garvie, R. C., and Goss, M. F., J. Mater. Sci. 21, 1253 (1986).
Co., U.S. Patent 4,918,041, 1990.
37. Norman, C. J., Catal. Today 20, 313 (1994).
6. Iglesia, E., Soled, S. L., and Kramer, G. M., J. Catal. 144, 238 (1993).
7. Ebitani, K., Konishi, J., and Hattori, H., J. Catal. 130, 257 (1991).
8. Hosoi, T., Shimidzu, T., Itoh, S., Baba, S., Takaoka, H., Imai, T.,
38. Yamaguchi, T., Catal. Today 20, 199 (1994).
39. Mercera, P. D. L., Van Ommen, J. G., Doesburg, E. B. M., Burggraaf,
A. J., and Ross, J. R. H., Appl. Catal. 71, 363 (1991).
and Yokoyama, N., Prepr. Am. Chem. Soc. Div. Pet. Chem. 33, 562 40. Srinivasan, R., Keogh, R. A., Milburn, D. R., and Davis, B. H., J. Catal.
(1988).
153, 123 (1995).
9. Misono, M., Catal. Rev. Sci. Eng. 29, 269 (1987).
10. Ribeiro, F. H., Boudart, M., Dalla-Betta, R. A., and Iglesia, E.,
J. Catal. 130, 498 (1991).
41. McCullough, J. D., and Trueblood, K. N., Acta Crystallogr. 12, 507
(1959).
42. Fuentes, G. A., unpublished results.
11. Soled, S. L., McVicker, G. B., Murrell, L. L., Sherman, L. G.,
Dispenziere, N. C., Hsu, S. L., and Waldman, D., J. Catal. 111, 286
(1988).
43. German, R. M., “Sintering Theory and Practice.” Wiley, New York,
1996.
44. Wachs, I. E., “Characterization of Catalytic Materials.” Butterworth–
Heinnemann, Boston, 1992.
45. Regalbuto, J., Catal. Lett. 29, 189 (1994).
46. Vermaire, D. C., and van Berge, P. C., J. Catal. 116, 309 (1989).
47. Pope, M. T., “Heteropoly and Isopoly Oxometallates.” Springer-
Verlag, Berlin, 1983.
48. Shadle, E., Hedman, B., Hodgson, K. O., and Solomon, E. I., Inorg.
Chem. 33, 4235 (1994).
49. Li, P., and Chen, I. W., J. Am. Ceram. Soc. 77, 118 (1994).
50. Horsley, J. A., Wachs, I. E., Brown, J. M., Via, G. H., and Hardcastle,
F. D., J. Phys. Chem. 91, 4014 (1987).
51. Hilbrig, F., Gobel, H. E., Knozinger, H., Schmelz, H., and Lengeler,
B., J. Phys. Chem. 95, 6973 (1991).
12. Arata, K., and Hino, M., in “Proceedings, 9th International Congress
on Catalysis, Calgary, 1988” (M. J. Phillips and M. Ternan, Eds.),
p. 1727. Chem. Institute of Canada, Ottawa, 1988.
13. Hino, M., and Arata, K., Chem. Lett., 971 (1989).
14. Soled, S. L., Dispenziere, N., and Saleh, R., in “Progress in Catalysis”
(K. J. Smith and E. C. Sanford, Eds.), p. 77. Elsevier, Amsterdam, 1992.
15. Chang, C. D., Santiesteban, J. G., and Stern, D. L., Mobil Oil Corp.,
U.S. Patent 5,345,026, 1994.
16. Iglesia, E., Barton, D. G., Soled, S. L., Miseo, S., Baumgartner, J. E.,
and Gates, W. E., in “Proc. 14th North American Meeting Catal. Soc.”
1995.
17. Iglesia, E., Barton, D. G., Soled, S. L., Miseo, S., Baumgartner, J. E., and
Gates, W. E., in Abstracts, 1995 Spring Meeting, American Chemical 52. Loopstra, O., and Boldrini, P., Acta Crystallogr. 21, 158 (1966).
Society.
53. Haller, G. L., and Resasco, D. E., Adv. Catal. 36, 173 (1989).
18. Iglesia, E., Barton, D. G., Soled, S. L., Miseo, S., Baumgartner, J. E.,
54. Ebitani, K., Tsuji, J., Hattori, H., and Kita, K., J. Catal. 135, 609 (1992).
Gates, W. E., Fuentes, G. A., and Meitzner, G. D., Stud. Surf. Sci. Catal. 55. Cortes, A., and Corma, A., J. Catal. 51, 338 (1978).
101, 533 (1996). 56. Morin, S., Gnep, N. S., and Guisnet, M., J. Catal. 159, 296 (1996).
19. Larsen, G., Lotero, E., and Parra, R. D., Stud. Surf. Sci. Catal. 101, 543 57. Taylor, W. J., Wagman, D. D., Williams, M. G., Pitzer, K. S., and Rossini,
(1996). F. D., J. Res. Natl. Bur. Std. 37, 95 (1946).
20. Hino, M., and Arata, K., J. Chem. Soc. Chem. Commun., 1259 58. Santiesteban, J. G., Vartuli, J. C., Han, S., Bastian, R. D., and Chang,
(1987).
C. D., J. Catal. 168, 431 (1997).
21. Farcasiu, D., Ghenciu, A., and Miller, G. J., J. Catal. 134, 118 (1992).
22. Soled, S. L., Miseo, S., Baumgartner, J. E., Gates, W. E., Barton, D. G.,
and Iglesia, E., in “Proc. 13th Int. Conf. Catal.,” p. 17. 1994.
59. Granqvist, C. G., “Handbook of Inorganic Electrochromic Materials.”
Elsevier, Amsterdam, 1995.
60. Weber, R. S., J. Catal. 151, 470 (1995).