Molecular/electronic structure-surface acidity relationships of model-supported tungsten oxide catalysts

Article abstract of DOI:10.1016/j.jcat.2006.12.018

A series of model-supported WO₃ catalysts were synthesized on preformed Al₂O₃, Nb₂O₅, TiO₂, and ZrO₂ supports by impregnation of aqueous ammonium metatungstate, (NH₄)₁₀W₁₂O₄₁s5H₂O. The molecular and electronic structures of the supported tungsten oxide phases were determined with in situ Raman and UV-vis spectroscopy, respectively. The supported tungsten oxide structures are the same on all oxide supports as a function of tungsten oxide surface density (W/nm²). Below monolayer coverage (<5 W/nm²), both monotungstate and polytungstate surface WO_x species are present under dehydrated conditions and the polytungstate/monotungstate ratio increases with increasing surface coverage. Above monolayer coverage (>5 W/nm²), crystalline WO₃ nanoparticles are present on top of the surface WO_x monolayer. Above ～10 W/nm², bulk-like WO₃ crystallites become dominant. The number of catalytic active sites and surface chemistry of the supported tungsten oxide phases were chemically probed with CH₃OH dehydration to CH₃OCH₃. The specific oxide support was found to significantly affect the relative catalytic acidity of the surface WO_x species (Al₂O₃ ? TiO₂ > Nb₂O₅ > ZrO₂) to that of the supported WO₃ nanoparticles. Consequently, no general relationship exists between the molecular/electronic structures or domain size and the specific catalytic acidity of the supported tungsten oxide phases present in the model-supported WO₃ catalysts.

Full text of DOI:10.1016/j.jcat.2006.12.018

Journal of Catalysis 246 (2007) 370–381
www.elsevier.com/locate/jcat
Molecular/electronic structure–surface acidity relationships
of model-supported tungsten oxide catalysts
a
b
b
a,∗
Taejin Kim , Andrew Burrows , Christopher J. Kiely , Israel E. Wachs
a
Operando Molecular Spectroscopy & Catalysis Laboratory, Chemical Engineering Department, Lehigh University, Bethlehem, PA 18015, USA
b
Center for Advanced Materials and Nanotechnology, Materials Science & Engineering Department, Lehigh University, Bethlehem, PA 18015, USA
Received 30 August 2006; revised 20 October 2006; accepted 30 December 2006
Available online 1 February 2007
Abstract
A series of model-supported WO catalysts were synthesized on preformed Al O , Nb O , TiO , and ZrO supports by impregnation of
3
2
3
2
5
2
2
aqueous ammonium metatungstate, (NH )
determined with in situ Raman and UV–vis spectroscopy, respectively. The supported tungsten oxide structures are the same on all oxide supports
W
O
·5H O. The molecular and electronic structures of the supported tungsten oxide phases were
4
10 12 41
2
2
2
as a function of tungsten oxide surface density (W/nm ). Below monolayer coverage (<5 W/nm ), both monotungstate and polytungstate surface
WOx species are present under dehydrated conditions and the polytungstate/monotungstate ratio increases with increasing surface coverage.
2
2
Above monolayer coverage (>5 W/nm ), crystalline WO nanoparticles are present on top of the surface WOx monolayer. Above ∼10 W/nm ,
3
bulk-like WO crystallites become dominant. The number of catalytic active sites and surface chemistry of the supported tungsten oxide phases
3
were chemically probed with CH OH dehydration to CH OCH . The specific oxide support was found to significantly affect the relative catalytic
3
3
3
acidity of the surface WOx species (Al O ꢀ TiO > Nb O > ZrO ) to that of the supported WO nanoparticles. Consequently, no general
2
3
2
2
5
2
3
relationship exists between the molecular/electronic structures or domain size and the specific catalytic acidity of the supported tungsten oxide
phases present in the model-supported WO catalysts.
3
©
2007 Elsevier Inc. All rights reserved.
Keywords: Spectroscopy; Raman; UV–vis; TPSR; HR-TEM; Catalysts; Supported; WO ; Al O ; ZrO ; TiO ; Nb O ; CH OH; CH OCH ; Dehydration
3
2
3
2
2
2
5
3
3
3
1
. Introduction
the supported WO3/Nb2O5 catalysts have not received as much
attention [14–16].
In many cases, the interaction of a catalytic active compo-
nent, such as tungsten oxide, with an oxide support (Al2O3,
ZrO2, TiO2, and Nb2O5) can dramatically alter the structural
and catalytic properties of supported catalytic active compo-
nent [17]. For most such strongly interacting oxide systems, the
supported catalytic active metal oxide phase is molecularly dis-
persed as a two-dimensional metal oxide overlayer on a high
surface area support oxide [17]. The nature of the surface tung-
sten oxide phases on oxide supports have been characterized
with many different techniques to provide insights into their
molecular and electronic structures as well as provide informa-
tion their corresponding surface chemical properties [17–27]. In
spite of the large body of literature on this subject, the molec-
ular/electronic structure–catalytic activity relationships of sup-
ported tungsten oxide catalysts are still not satisfactorily under-
stood. Consequently, the objective of this paper is to investigate
Supported tungsten oxide catalysts are used in many signif-
icant industrial applications and are known to be efficient solid
acid catalysts [1,2]. Tungsten oxide on alumina catalysts are
used for hydrotreating and for hydrocarbon cracking reactions
3–7], and also exhibit enhanced resistance to reduction [8].
Tungsten oxide supported on titania, in combination with vana-
dia, is effective for both the selective catalytic reduction of
[
NO by NH to N /H O and for alkene isomerization [9,10].
3
2
2
As an alternative to the sulfated zirconia solid acid catalyst,
supported WO3/ZrO2 catalysts also have been reported to be
active for the isomerization of C4–C8 alkanes [11–13]. In con-
trast to the above supported tungsten oxide catalytic systems,
*
Corresponding author. Fax: +1 610 758 6555.
E-mail address: iew0@lehigh.edu (I.E. Wachs).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.12.018

T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
371
the molecular/electronic and the surface reactivity properties of
model-supported tungsten oxide catalysts and to establish the
relationships between these parameters. In the present study
cooling the catalysts back to room temperature in the flow-
ing 10% O /He gas. The spectral acquisition time used was
2
30 scans of 30 s/scan for a total of 15 min/spectrum.
2
both the surface tungsten oxide coverage (W/nm ) and the
specific oxide support, preformed model supports, were var-
ied over a wide range to carefully examine the influence of
these two significant parameters. The tungsten oxide molecu-
lar structures were determined with in situ Raman spectroscopy
and the corresponding electronic structures were determined
with in situ UV–vis diffuse reflectance spectroscopy (DRS).
The acidic nature of the supported tungsten oxide phases were
chemically probed with temperature programmed surface re-
action (TPSR) spectroscopy and steady-state dehydration of
methanol to dimethyl ether (CH3OCH3).
2
.4. UV–vis diffuse reflectance spectroscopy (DRS)
The UV–vis spectra of the dehydrated supported tungsten
oxide catalysts were measured with a Varian Cary 5E UV–
vis–NIR spectrophotometer. The catalysts in powder form were
loaded into an environmental cell (Harrick, HVC-DR2) for ob-
taining their UV–vis spectra. The UV–vis spectra of the dehy-
drated samples were obtained at room temperature after calci-
◦
nation at 400 C in flowing 10% O2/He gas for 1 h and cooling
back to room temperature. A magnesium oxide sample was
used as the standard for obtaining the baseline. Microsoft Ex-
cel software was used to calculate the Kubelka–Munk function
F(R∞) and the edge energy (Eg) from the absorbance data.
Below 300 nm, the absorbance signal was unacceptably noisy
and a filter (Varian, 1.5 ABS) was used to minimize the back-
ground noise. The UV–vis spectra were only collected for the
supported WO3/ZrO2 and WO3/Al2O3 catalysts because the
strong oxide absorbance by the TiO2 and Nb2O5 supports in
the region of interest prevented collection of their supported
tungsten oxide UV–vis spectra. The Al2O3 support possesses
no absorbance and the ZrO2 support exhibits very weak ab-
sorbance in the region of interest. The supported WO3/ZrO2
UV–vis spectra were background subtracted with the ZrO2 sup-
port spectrum to account for zirconia’s weak contribution to the
UV–vis spectrum. In addition, the Eg values for the 2D surface
WOx and the 3D WO3 phases were independently determined
by background subtracting the 2D surface WOx monolayer
UV–vis DRS contribution when monolayer surface coverage
was exceeded.
2
. Experimental
2
.1. Catalyst synthesis
The supported tungsten oxide catalysts were all prepared
by incipient wetness impregnation of aqueous ammonium
metatungstate, (NH4)10W12O41·5H2O (Pfaltz & Bauer, 99.5%
purity), onto different oxide supports: Al2O3 (Harshaw/Engel-
2
2
hard, 178 m /g), TiO2 (Degussa P-25, 51 m /g), Nb2O5
2
(
CBMM AD1927 HY-340, 66 m /g), and ZrO2 (Degussa,
0 m /g). After impregnation, the samples were initially dried
2
6
at room temperature overnight and then calcined in flowing air
◦
for 4 h at 450 C. Several supported WO3/ZrO2 catalysts were
◦
also calcined at elevated temperatures (700–900 C) to examine
the effect of calcination temperature.
2
.2. BET specific surface area measurement
The BET surface areas of the samples were measured by ni-
◦
trogen adsorption–desorption in flowing N2 at −196 C with a
Quantasorb surface area analyzer (Quantachrome Corporation,
Model OS-9). A sample quantity of ∼0.3 g was used for the
2
.5. High resolution-transmission electron microscopy
(HR-TEM)
◦
measurement, and the sample was outgassed at 250 C before
Samples for HR-TEM examination were prepared by dis-
N2 adsorption (Quantachrome Corporation, Model QT-3).
persing the catalyst powder in high purity ethanol, then allow-
ing a drop of the suspension to evaporate on a holey carbon film
supported by a 300 mesh copper TEM grid. HR-TEM images
of the ambient samples were obtained using a JEOL 2200FS
transmission electron microscope, having an accelerating volt-
age of 200 kV, a point-to-point resolution of 0.19 nm and an
information limit of 0.11 nm. Fourier transform (FT) analyses
of lattices fringe periodicities and inter-planar angles were car-
ried out using digital micrograph.
2
.3. Raman spectroscopy
The Raman spectra of the dehydrated supported tungsten ox-
ide catalysts were obtained with either visible (532 nm) or UV
325 nm) excitation (Horiba-Jobin Yvon LabRam-HR). The
(
UV laser excitation was generated from a He–Cd laser (Kim-
mon, Model IK5751 I-G, 30 mW) and the visible excitation
was generated by a Yag double-diode pumped laser (coherent
3
15 m, 20 mW). The scattered photons were directed into a sin-
2
.6. CH3OH temperature-programmed surface reaction
gle monochromator and focused onto a LN2 cooled CCD detec-
tor (JY-CCD3000V). The Raman spectrometer was equipped
with an in situ environmental cell (Linkam T1500 cell) where
both the temperature and the gaseous composition were con-
trollable. The catalysts were maintained in loose powder form,
(
TPSR) spectroscopy
Methanol TPSR spectroscopy was performed on an Al-
tamira temperature programmed system (AMI-200) equipped
with an online quadrupole mass spectrometer (Dycor Dymax-
ion DME200MS). Samples of ∼100 mg were loaded into a
◦
initially dehydrated at 450 C for 1 h in flowing 10% O2/He
Airgas, ultra-high purity and hydrocarbon-free), and the Ra-
man spectra of the dehydrated samples were collected after
(
◦
U-shaped quartz tube and initially calcined at 400 C (Airgas,

372
T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
ultra zero grade air, 30 ml/min) for 40 min to remove any possi-
ble adsorbed organic impurities and to dehydrate the catalysts.
To ensure that the supported tungsten oxide species on the cat-
alysts remain in a fully oxidized state, the pretreated samples
turnover frequency (TOF: number of DME molecules/surface
acid site/s). For the supported tungsten oxide catalysts below
monolayer surface coverage, the number of surface acid sites
was taken as the number of surface WOx species, which as-
sumed 100% dispersion and was independently confirmed with
Raman spectroscopy. Above monolayer surface tungsten oxide
coverage, the number of exposed surface acid sites was deter-
mined by the area under the DME/CH3OH-TPSR curves and
normalized to the value for monolayer coverage.
◦
were first cooled down in flowing air to 110 C and then the
gas stream was switched to an ultra high purity He (Airgas)
◦
flow on further cooling to 100 C. After flushing with flowing
◦
He for another 30 min at 100 C to remove any physically ad-
sorbed oxygen and possible background gases, a CH3OH/He
gas mixture feed (30 ml/min) containing 2000 ppm methanol
was introduced for chemisorption for 30 min. Previous work
3. Results
◦
demonstrated that methanol adsorption at 100 C minimizes
the formation of physically adsorbed methanol on the sam-
ples because physically adsorbed CH3OH desorbs below this
temperature from oxide surfaces [28,29]. After methanol ad-
3.1. BET specific surface area and tungsten oxide surface
density
◦
The BET surface area and tungsten oxide surface density
sorption, the samples were again purged at 100 C with flow-
2
(
W/nm ) for the supported tungsten oxide catalysts were de-
ing He for an additional hour to remove any residual physi-
cally adsorbed methanol. The CH3OH-TPSR experiments were
termined and are listed in Table 1. The BET surface area con-
tinuously decreases with increasing tungsten oxide loading due
to the additional mass introduced by the supported tungsten ox-
◦
performed with a heating rate of 10 C/min in flowing UHP
He and the desorbing molecules were monitored with the on-
line MS. The m/e values used to detect the different desorp-
tion products were CH3OH (m/e = 31), H2CO (m/e = 30),
CH3OCH3 (DME; m/e = 45), (CH3O)2CH2 (DMM; m/e =
2
ide phase. The tungsten oxide surface density (W/nm ) was
calculated by using the initial BET surface area of the oxide
supports because the supports retain their initial BET values for
this rather mild calcination temperature and in the presence of
surface tungsten oxide species that are known to retard sintering
of oxide support.
7
5), H2O (m/e = 18), CO2 (m/e = 44), and CO (m/e = 28).
For those desorbing molecules that gave rise to several frag-
ments in the mass spectrometer, additional m/e values were
also collected to further confirm their identity. The activation
energy for formation of dimethyl ether was calculated using the
Redhead equation for first-order kinetics [30]. Previous studies
demonstrated that the rate-determining step in methanol de-
hydration involves the first-order process that breaks the C–O
3
3
.2. In situ Raman spectroscopy
.2.1. Supported WO3/Al2O3 catalysts
The UV Raman spectra of the dehydrated WO3/Al2O3 sam-
∗
ples as a function of surface tungsten oxide density are pre-
sented in Fig. 1. The UV excitation was used because of sample
fluorescence when excited in the visible region [32]. At low
bond in the surface CH3O intermediate [31]. In addition, the
area under the DME/CH3OH-TPSR curve corresponds to the
number of exposed surface acid sites.
2
surface tungsten oxide density (0.5–4 W/nm ), only the Ra-
man band characteristic of dehydrated surface WOx species is
present. The surface nature of these dehydrated tungsten ox-
ide species is confirmed by their reversible structural transfor-
mations on hydration–dehydration treatments (not shown for
2
.7. Steady-state dehydration of methanol to dimethyl ether
Methanol dehydration was used to chemically probe the cat-
alytic activity and product selectivity of the supported tungsten
oxide catalysts. The dehydration of methanol was examined in
a fixed-bed reactor at 230 C with 30 mg of catalyst. The cat-
−
1
brevity). The Raman band in the 1003–1015 cm region arises
◦
alysts were suspended between two layers of quartz wool in a
vertical glass tube. The volume composition of the gaseous re-
actant feed was CH3OH/O2/He = 6/13/81 (mol%) with a total
flow rate of 100 ml/min from the top to the bottom of the re-
actor. The outlet of the reactor to the gas chromatograph was
◦
heated at ∼130 C in order to avoid product condensation. On-
line analysis of the methanol conversions and reaction products
was performed with an HP-5840A gas chromatograph contain-
ing two packed columns (Porapack R and Carbosieve SII) and
two detectors (TCD and FID). Blank runs without the catalysts
demonstrated negligible methanol conversion in the reactor sys-
tem. The supported tungsten oxide catalysts were pretreated at
◦
3
50 C for 30 min in a stream of O2/He before each run. The
◦
catalytic experiments were run for 4 h at 230 C to obtain the
methanol conversion, selectivity, and activity. The steady-state
methanol dehydration catalytic data were expressed in terms of
Fig. 1. UV Raman spectra of the model-supported WO /Al O catalysts under
3
2 3
dehydrated conditions.

T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
373
Table 1
2
BET surface area and tungsten oxide surface density (W/nm ) of the supported
WO catalysts on the model oxide supports (TiO , Nb O , Al O , and ZrO )
3
2
2
5
2
3
2
Sample (calcination
temperature, C)
Surface density
(W/nm )
BET surface area
(m /g)
◦
2
2
Bulk TiO
0.0
2.7
4.0
4.5
8.0
51.0
48.1
46.4
46.1
43.0
41.4
39.3
2
5.0 wt% WTi (450)
7.3 wt% WTi (450)
8.1 wt% WTi (450)
13.6 wt% WTi (450)
16.4 wt% WTi (450)
19.1 wt% WTi (450)
10.0
12.0
Bulk Nb O
0.0
1.3
4.0
4.5
8.0
66.0
64.0
56.7
55.8
51.3
47.8
43.6
2
5
3.2 wt% WNb (450)
9.2 wt% WNb (450)
10.2 wt% WNb (450)
16.9 wt% WNb (450)
20.3 wt% WNb (450)
23.4 wt% WNb (450)
Fig. 2. Visible Raman spectra of the model-supported WO /Nb O catalysts
3
2 5
under dehydrated conditions.
10.0
12.0
Bulk Al O
0.0
0.5
1.0
4.0
4.5
5.0
6.0
8.0
178.0
166.2
166.0
158.8
145.6
133.0
126.0
125.3
108.2
96.9
2
3
3.2 wt% WAl (450)
6.4 wt% WAl (450)
21.5 wt% WAl (450)
23.5 wt% WAl (450)
25.6 wt% WAl (450)
29.1 wt% WAl (450)
35.4 wt% WAl (450)
40.7 wt% WAl (450)
45.1 wt% WAl (450)
57.8 wt% WAl (450)
10.0
12.0
20.0
75.0
Bulk ZrO
0.0
0.4
1.8
2.3
2.8
3.3
3.9
4.5
4.8
5.9
7.6
10.8
12.0
14.4
28.9
60.0
59.9
59.6
59.4
59.3
59.3
58.7
55.6
55.3
54.5
49.0
45.3
45.1
45.0
32.8
2
1
4
5
6
7
8
9
1
1
1
2
2
2
4
wt% WZr (450)
wt% WZr (450)
wt% WZr (450)
wt% WZr (450)
wt% WZr (450)
.3 wt% WZr (450)
.4 wt% WZr (450)
0 wt% WZr (450)
2 wt% WZr (450)
5 wt% WZr (450)
0 wt% WZr (450)
1.7 wt% WZr (450)
5.0 wt% WZr (450)
0.0 wt% WZr (450)
Fig. 3. Visible Raman spectra of the model-supported WO3/TiO2 catalysts un-
der dehydrated conditions.
3
.2.2. Supported WO3/Nb2O5 catalysts
The visible Raman spectra of the dehydrated supported
2
WO /Nb O catalysts in the 1.3–12 W/nm range are pre-
3
2
5
sented in Fig. 2. The Raman spectra are limited to above
−1
7
00 cm because of the strong Raman vibrations of the Nb O
2
5
−
1
support below 700 cm . Crystalline WO3 NPs are not present
in the 1.3–4 W/nm range and only appear at higher tungsten
2
from the symmetric stretching vibrations of the terminal W=O
bonds of surface WOx species [15,33]. The shift of the surface
W=O band with increasing tungsten oxide surface density is
oxide surface density. Again, note that the supported WO3 NPs
−
1
give rise to a broader Raman band at 805 cm than that for
the corresponding bulk crystalline WO3 at higher tungsten ox-
related to the polymerization of the surface WO species from
x
2
ide surface density (>4 W/nm ). The surface WOx species give
monotungstate to polytungstate species [15]. No Raman bands
−1
−
1
rise to a weak and broad Raman band in the 900–1000 cm re-
gion against the Nb2O5 support background that are further en-
hanced when the Nb2O5 contribution is subtracted (not shown
for brevity).
due to crystalline WO particles at ∼805 and ∼714 cm are
3
2
present for samples with less than 5 W/nm [34]. At tungsten
2
oxide surface densities of 5 W/nm and above, Raman bands
−
1
of crystalline WO3 also appear at ∼805 and ∼714 cm (see
Fig. 1 for the bulk crystalline WO3 Raman spectrum). Note the
Raman bands for the alumina-supported WO3 crystallites are
much broader than that of the crystalline bulk WO3, which re-
flects the nanoparticle (NP) dimension of the supported WO3
crystallites. In addition, the Raman band of the terminal W=O
bond of the surface WOx species further shifts from ∼1015 to
3.2.3. Supported WO3/TiO2 catalysts
The visible Raman spectra of dehydrated supported WO3/
TiO2 catalysts are shown in Fig. 3.
−1
The Raman spectra are limited to above 700 cm because
of the strong vibrations of the TiO2 support at lower values.
−1
−1
∼
1020 cm with increasing surface WOx density.
In fact, the rising background below 750 cm is already a vi-

374
T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
Fig. 4. Visible Raman spectra of the model-supported WO /ZrO catalysts un-
der dehydrated conditions.
Fig. 5. Plots of normalized Raman intensity of the dehydrated surface WO
x
3
2
−
1
−1
species (∼1000–1020 cm ) and crystalline WO particles (∼805 cm ) as a
3
2
function of tungsten oxide surface density (W/nm ) on the model ZrO sup-
2
port.
brational contribution from the TiO2 support. The dehydrated
surface WOx species gives rise to Raman bands in the ∼1000–
the crystalline WO NPs (∼805 cm 1) as a function of tungsten
−
−1
1
016 cm with increasing tungsten oxide surface coverage.
3
oxide surface density and is shown in Fig. 5b. The normalized
This vibrational shift with tungsten oxide surface coverage is
related to the continuous polymerization of the surface mono-
tungstate species to surface polytungstate species [15]. The ad-
2
crystalline WO3 NP Raman intensity is zero until ∼4 W/nm
and then increases linearly with surface W density. Extrapo-
lating the linear portion of the curve back to its intercept with
−
1
ditional Raman band at ∼805 cm at higher surface tungsten
2
2
the x-axis gives a value of ∼4.5 W/nm . This extrapolation
oxide density (>4 W/nm ) reflects the presence of crystalline
method avoids considering the formation of small amounts of
WO3 NPs that typically form as monolayer surface coverage is
approached. Thus, both Raman methods quantitatively demon-
WO3 NPs.
3
.2.4. Supported WO3/ZrO2 catalysts
2
strate that monolayer surface coverage occurs at ∼4.5 W/nm
The visible Raman spectra of the dehydrated supported
for the supported WO3/ZrO2 catalyst system.
WO3/ZrO2 catalysts are presented in Fig. 4. The Raman spec-
2
Comparison of the Raman spectra of the supported WO3/
ZrO2 catalysts with the other supported tungsten oxide catalysts
also provides information about monolayer surface WOx cov-
erage on these other oxide supports. Crystalline WO NPs are
trum for the 0.4 W/nm sample does not give rise to detectable
bands for the surface WOx species against the strong Raman
spectrum of the ZrO2 support. The Raman band of the surface
−
1
WOx species shifts from 1001 to 1015 cm with increasing
surface WOx coverage reflecting the transformation of surface
monotungstate to polytungstate species. Crystalline WO3 NPs
3
2
not present below ∼4 W/nm on any of the supports and the
intensity of the crystalline WO3 NPs increases with increas-
2
2
2
ing W/nm above ∼4 W/nm . Furthermore, the appearance
are present at ∼4 W/nm and higher surface density with Ra-
2
−
1
of crystalline WO NPs at ∼4.5 W/nm was also found to be
man bands at 805 and 714 cm
.
3
independent of catalyst calcination temperature reflecting the
2
universal value of ∼4.5 W/nm for monolayer surface WO
3
.2.5. Monolayer surface WOx coverage on oxide supports
Monolayer surface WOx coverage for the supported WO3/
x
coverage on oxide supports.
ZrO2 catalyst system was determined by using the strong
−1
∼
471 cm band of the ZrO2 support as an internal standard.
3.3. In situ UV–vis diffuse reflectance spectroscopy (DRS)
Two different methods were used to determine monolayer sur-
face WOx coverage. The first method plotted the intensity of
the Raman band of the dehydrated surface WOx species in the
The UV–vis DRS Eg values for the dehydrated supported
WO3/ZrO2 (represented by open symbols) and WO3/Al2O3
(represented by solid symbols) are presented in Fig. 6 as func-
−1
1
000–1020 cm region by normalizing this band to the ZrO2
2
internal standard as a function of tungsten oxide surface den-
sity. The normalized Raman intensity of the dehydrated surface
WOx species as a function of the tungsten oxide surface den-
sity on the ZrO2 support is presented in Fig. 5a. The normalized
Raman intensity for the surface WOx species linearly increases
tions of the tungsten oxide surface density (W/nm ) and cal-
cination temperature. Recall that the UV–vis DRS for the sup-
ported WO3/TiO2 and WO3/Nb2O5 were not collected because
of the strong absorbance by the TiO2 and Nb2O5 supports in the
region of interest. Both supported tungsten oxide catalyst sys-
2
2
2
with W/nm and levels off at ∼4.5 W/nm . Note that no fur-
tems exhibit the same trend with increasing W/nm and essen-
ther increase in the normalized surface WOx Raman intensity
tially fall on the same curve. In addition, the data points for the
supported WO3/ZrO2 calcined at different temperatures also
fall on the same curve. Thus, the UV–vis DRS Eg values are
independent of the specific oxide support and calcination tem-
2
occurs above ∼4.5 W/nm , which reflects saturation of mono-
layer surface coverage of the surface WOx species on ZrO2.
The second method plotted the normalized Raman intensity of

T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
375
Fig. 6. Plot of UV–vis DRS Eg values for the dehydrated model-supported
WO /ZrO and WO /Al O catalysts calcined at different temperatures as a
function of tungsten oxide surface density (W/nm ).
3
2
3
2
3
(a)
2
perature when plotted as a function of tungsten oxide surface
density. The universal Eg vs tungsten oxide surface density plot
exhibits three distinct regions: (1) modest decrease in Eg values
2
between 0 and 4.5 W/nm , (2) somewhat greater decrease in
2
Eg values between 4.5 and 10 W/nm , and (3) relatively con-
2
stant Eg values above 10 W/nm that asymptotically approach
the Eg value of large bulk WO3 crystals. The initial Eg values
at the lowest surface coverage exhibit a value of ∼4.5 eV that
continues to decrease with increasing surface WOx coverage up
2
to monolayer coverage of ∼4.5 W/nm . The initial Eg value
is between the Eg values for isolated (Na2WO4 at ∼5.2 eV)
and extensively polymerized (Na2W2O7 at 4.2 eV) WOx struc-
tures suggesting a mixture of these two surface WOx structures
below monolayer surface coverage. Furthermore, the Eg value
monotonically decreases with increasing surface W density up
to monolayer coverage reflecting that the surface WOx species
are becoming more polymerized as the surface WOx density
increases in the submonolayer region. The Eg value at mono-
layer surface WOx coverage corresponds to that of a linear
(b)
2
WOx polymer present [e.g., (NH4)2W2O7]. Above 4.5 W/nm ,
the Eg value continues to rapidly decrease and Raman simul-
taneously detects the presence of crystalline WO3 NPs in this
tungsten oxide surface density region. The Eg values of the sup-
ported WO3 crystallites, however, are greater than that for large
bulk WO3 crystallites, because they are present as WO3 NPs
2
2
between 4.5 and 10 W/nm . Above 10 W/nm , the Eg values
remain relatively constant with increasing surface W density,
because the WO3 particles are much larger in this tungsten ox-
ide surface density region and exhibit bulk-like characteristics.
In summary, UV–vis DRS in combination with Raman spec-
troscopy is able to discriminate between isolated surface WOx
species, polymeric surface WOx species, WO3 NPs and large
bulk-like WO3 particles.
(
c)
2
Fig. 7. HR-TEM images of (a) 4, (b) 8, and (c) 12 W/nm WO /ZrO catalysts.
3
2
3
.4. High resolution-transmission electron microscopy
2
a surface W/ZrO2 density of 4, 8, and 12 W/nm , respectively.
All of the HR-TEM images obtained clearly display crystalline
ZrO2 lattice fringes, especially at lower magnifications. The ad-
ditional presence of an amorphous WOx overlayer, and very
(HR-TEM)
Representative HR-TEM images of the supported WO3/
ZrO2 catalysts are shown in Figs. 7a–7c for the catalysts with

376
T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
Table 2
◦
DME/CH OH-TPSR Tp temperature ( C) for the supported WO catalysts on
3
3
the model oxide supports (TiO , Nb O , Al O , and ZrO )
2
2
5
2
3
2
Sample (calcination
temperature, C)
Surface density
(W/nm )
DME peak
temperature ( C)
◦
2
◦
Bulk TiO
0.0
4.0
4.5
8.0
10.0
12.0
325
289
289
264
260
259
2
7.3 wt% WTi (450)
8.1 wt% WTi (450)
13.6 wt% WTi (450)
16.4 wt% WTi (450)
19.1 wt% WTi (450)
Bulk Nb O
0.0
4.0
4.5
8.0
10.0
12.0
300
286
286
277
272
271
2
5
9.2 wt% WNb (450)
10.2 wt% WNb (450)
16.9 wt% WNb (450)
20.3 wt% WNb (450)
Fig. 8. CH OH-TPSR spectra for methanol dehydration to CH OCH over the
23.4 wt% WNb (450)
3
3
3
model supported WO /Al O catalysts as a function of the tungsten oxide sur-
face density (W/nm ).
3
2 3
Bulk Al O
196
204
206
208
219
223
225
226
226
2
2 3
2
2
1.5 wt% WAl (450)
3.5 wt% WAl (450)
4.0
4.5
5.0
6.0
8.0
10.0
12.0
20.0
small tungsten oxide NPs that are around 1 nm in size are also
suggested by this sequence of images. The amorphous over-
layer of the surface WOx species appears most prominently at
the edges of the support particles when viewed in profile. In ad-
dition, very small NPs possessing a lateral dimension of around
25.6 wt% WAl (450)
29.1 wt% WAl (450)
35.4 wt% WAl (450)
40.7 wt% WAl (450)
45.1 wt% WAl (450)
57.8 wt% WAl (450)
1
nm can be seen as darker flecks against the ZrO support, and
2
Bulk ZrO
0.0
1.3
2.3
2.8
4.0
4.5
4.8
7.6
374
327
319
305
297
290
286
268
266
262
259
2
appear by virtue of their mass contrast. A general trend of in-
creasing number density of these dark flecks with increasing
surface W/nm density was observed, although it is interesting
3
5
wt% WZr (450)
wt% WZr (450)
2
6 wt% WZr (450)
8.3 wt% WZr (450)
9.4 wt% WZr (450)
10 wt% WZr (450)
15 wt% WZr (450)
20 wt% WZr (450)
to note that their average lateral size did not increase markedly
with W loading. These dark flecks are thought to arise from
the presence of crystalline WO3 nanoparticles and possibly also
from hydrated (WOx)·nH2O clusters. It should, however, be
noted that the WOx overlayers on these ZrO2 support particles
are very sensitive to modification in the electron beam and great
care needs to be taken to minimize the electron dose during HR-
TEM imaging experiments.
10.8
14.4
28.9
25.0 wt% WZr (450)
40.0 wt% WZr (450)
in Table 2. The DME/CH3OH-TPSR spectra exhibit increas-
ing Tp values with increasing tungsten oxide surface density.
The supported surface WOx monolayer on alumina exhibits a
3
.5. CH3OH-TPSR spectroscopy
◦
◦
Tp value of 206 C compared with 196 C for the WO3-free
◦
Methanol was used as a chemical probe molecule because it
Al O support and 254 C for bulk WO . The absence of a
2 3 3
◦
readily discriminates between surface acidic sites (formation of
CH OCH ), basic sites (formation of CO/CO ), and redox sites
DME-TPSR peak at 196 C reveals that exposed alumina sup-
ports sites are not present at monolayer surface WO cover-
3
3
2
x
(
formation of HCHO and CH3OOCH). The supported WO3
age. Increasing the tungsten oxide surface density above mono-
catalysts exclusively yielded CH3OCH3 as the reaction prod-
uct during CH3OH-TPSR demonstrating their surface acidic
nature. The CH3OH-TPSR spectra also allowed determination
of the number of surface acid sites (reflected in the area un-
derneath the DME curve) and the kinetics or strength of the
surface acid sites (reflected by the Tp value). Furthermore, us-
ing the Redhead equation it was possible to directly determine
the activation energy for the surface reaction (Eact) in dimethyl
ether formation and the kinetics for the rate-determining step
layer coverage further increases the Tp temperature from 206
◦
to 226 C indicating that the crystalline WO NPs present in
3
this tungsten oxide surface density region are less active than
the surface WO monolayer on Al O for the acid catalyzed
x
2
3
methanol dehydration reaction. Thus, the surface WO mono-
x
layer on Al O is more reactive than WO NPs and bulk WO .
2
3
3
3
In addition to these reactivity changes with tungsten oxide sur-
face density, the number of catalytic active sites monotonically
decreases above monolayer tungsten oxide surface coverage
2
(
krds), which involves the first-order C–O bond scission of the
with increasing W/nm reflecting the presence of less dispersed
∗
surface methoxy (CH3O ) intermediate [31].
crystalline WO NPs.
3
3
.5.1. Supported WO3/Al2O3 catalysts
3.5.2. Supported WO3/ZrO2 catalysts
The DME/CH3OH-TPSR spectra from the supported WO3/
3 3
The DME/CH OH-TPSR spectra for the supported WO /
Al2O3 catalysts are shown in Fig. 8 and the Tp values are listed
ZrO2 catalysts are presented in Fig. 9 and listed in Table 2.

T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
377
Fig. 9. CH OH-TPSR spectra for methanol dehydration to DME over the model
3
supported WO /ZrO catalysts as a function of tungsten oxide surface density
3
2
2
Fig. 10. First-order kinetic rate constants, krds, for CH3OH dehydration to DME
(W/nm ).
over the model-supported WO catalysts as a function of tungsten oxide surface
3
2
density (W/nm ).
In contrast to the supported WO3/Al2O3 catalyst system that
exhibits an increase in DME/CH3OH-TPSR Tp values with in-
creasing W/nm , the supported WO3/ZrO2 catalysts exhibit
ble 2 and the Tp values decrease with increasing tungsten oxide
2
surface density. The monolayer surface WOx species on TiO2
a decrease in DME/CH3OH-TPSR Tp values with increasing
◦
◦
exhibits a Tp value of 289 C compared with 254 C for bulk
2
W/nm . The surface WOx monolayer on ZrO2 exhibits a Tp
◦
WO3 and 325 C for the WO3-free TiO2 support. The activity of
◦
value of 290 C that is greater than crystalline bulk WO3 at
2
the monolayer surface WOx monolayer on TiO2 is comparable
to that of the surface WOx monolayer on Nb2O5 and ZrO2, but
at higher surface density, the supported WO3/TiO2 catalyst sys-
tem is slightly more active than the supported WO3/Nb2O5 and
WO3/ZrO2 catalyst systems. As for the other supported WO3
catalysts, the number of catalytic active sites decreases with
increasing tungsten oxide surface density above monolayer sur-
face WOx coverage because of the presence of crystalline WO3
NPs.
◦
◦
54 C and less than the WO3-free ZrO2 support at 374 C. At
high tungsten oxide surface density, the Tp value approaches
that of bulk WO3 particles. Thus, the crystalline WO3 NPs
are more active than the surface WOx monolayer on the ZrO2
support and dominate the catalytic properties at high tungsten
oxide surface density. Furthermore, the number of catalytic ac-
tive sites decreases with increasing tungsten oxide surface den-
sity above monolayer surface coverage because of the presence
crystalline WO NPs above monolayer coverage.
3
3
.5.5. Comparison of DME/CH3OH-TPSR results for the
3
.5.3. Supported WO3/Nb2O5 catalysts
The DME/CH3OH-TPSR spectra for the supported WO3/
supported WO3 catalysts
The first-order krds kinetic rate constants for methanol de-
hydration to dimethyl ether over the supported WO3 catalysts
Nb2O5 catalysts follow a similar trend as that for supported
WO3/ZrO2 and are not shown for brevity. The DME/CH3OH-
TPSR Tp values for the supported WO3/Nb2O5 catalysts are
listed in Table 2 and the Tp values decrease with increasing
tungsten oxide surface density. The Tp value for the surface
WO monolayer on Nb O is 286 C, which is higher than that
for bulk WO3 at 254 C and less than that for the WO3-free
Nb2O5 support at 300 C. Increasing the tungsten oxide sur-
face density on Nb2O5 above monolayer surface WOx coverage
further decreases the Tp values from 286 to 271 C, which re-
veals that the WO3 NPs are more active than the surface WOx
species anchored to the Nb2O5 support. As for the supported
WO3/Al2O3 and WO3/ZrO2 catalyst systems, the number of
catalytic active sites decreases with increasing tungsten oxide
surface density above monolayer surface WOx coverage on
Nb2O5.
◦
at 230 C are plotted in Fig. 10 as a function of tungsten ox-
ide surface density. The kinetic krds values were determined
from the DME/CH3OH-TPSR Tp values and application of the
Redhead equation (see Section 2.5 above for details). The sup-
ported WO /TiO , WO /Nb O , and WO /ZrO catalyst sys-
tems follow the exact same krds trend, increasing as a function
of W/nm and approach the activity value of bulk WO3 at high
W/nm values. In contrast, the supported WO3/Al2O3 catalyst
system follows an inverse krds trend, decreasing as a function
of W/nm and does not approach the value of bulk WO at
high W/nm values. Even at the highest tungsten oxide surface
◦
x
2
5
3
2
3
2
5
3
2
◦
◦
2
2
◦
2
3
2
density, the supported WO /Al O catalysts are about an order
of magnitude more active than bulk WO and the other sup-
ported WO catalysts. The initial high k values for supported
WO /Al O below monolayer surface coverage are related to
3
2
3
3
3
rds
3
2
3
exposed alumina sites, but all of the alumina sites are covered
3
.5.4. Supported WO3/TiO2 catalysts
The DME/CH3OH-TPSR spectra for the supported WO3/
with the surface WO monolayer at monolayer surface cover-
x
2
age and higher W/nm . This suggests that the surface WO
x
TiO2 catalysts follow the trends found for the supported
WO3/ZrO2 and WO3/Nb2O5 catalyst systems and are not
shown for brevity. The associated Tp values are listed in Ta-
species coordinated to the Al O surface are highly active acid
2
3
sites and even more active than the supported WO NPs for
3
methanol dehydration.

378
T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
Fig. 11. Number of catalytic active sites for the model-supported WO catalysts
3
as a function of tungsten oxide surface density (W/nm ).
Fig. 13. Catalytic acidity TOR values for CH OH dehydration to DME over the
3
2
model-supported WO catalysts as a function of tungsten oxide surface density
3
2
(
W/nm ).
ence of WO3 NPs above monolayer surface coverage further
alters the TOF values. For the supported WO3/Al2O3 catalyst
system, the TOF value decreases above monolayer surface cov-
erage because of the lower activity of the WO3 NPs relative to
the surface WOx sites anchored to the alumina support. For the
supported WO3/TiO2, WO3/Nb2O5, and WO3/ZrO2 catalyst
systems, the TOF increases above monolayer surface coverage
because of the higher activity of the WO3 NPs compared with
the surface WOx species coordinated to these oxide supports.
The TOF values of the nonalumina-supported tungsten oxide
catalysts asymptotically approach the TOF value of bulk WO3
2
at very high W/nm values. The TOF value of the supported
WO3/Al2O3 catalyst system, however, remains much higher
2
than the TOF for bulk WO3 at high W/nm . The different
trends reflect the contributions of the surface WOx monolayer
relative to WO3 NPs toward methanol dehydration.
Fig. 12. Catalytic acidity TOF values for CH OH dehydration to DME over the
3
model-supported WO catalysts as function of tungsten oxide surface density
3
2
(
W/nm ).
The turnover rates (TOR), number of methanol molecules
dehydrated per all W oxide atoms in the catalyst per second, of
The number of catalytic active sites for the supported WO3
catalysts is plotted as a function of W/nm in Fig. 11. For all
2
◦
2
the supported WO catalysts as a function of W/nm at 230 C
3
are presented in Fig. 13. For the submonolayer region, the TOF
supported WO3 catalyst systems, the number of catalytic ac-
2
and TOR values are identical because the supported tungsten
oxide is 100% dispersed on the supports. Above monolayer
surface coverage, TOF is always greater than TOR because
the supported tungsten oxide phase is not 100% dispersed on
tive sites decreases with W/nm above monolayer coverage
because of the presence of crystalline WO3 NPs on top of the
surface WOx monolayer in this coverage region. The slight dif-
2
ferences in the decrease in Ns with W/nm for the different
the supports. The poorly dispersed crystalline WO NPs are
3
supported WO3 catalyst systems may be related to the different
surface areas and pore structures of the oxide supports.
also present above monolayer surface coverage and decrease
the TOR values because all of the W oxide atoms are used in
normalizing the reaction rate. The maximum TOR values for
3
.6. Steady-state methanol dehydration kinetics
the supported WO catalysts, with the exception of supported
3
2
2
WO3/Al2O3, occurs at ∼8–10 W/nm . At higher W/nm val-
2
The turnover frequency (TOF) values (methanol molecules
ues (>10 W/nm ), the TOR values for all of the supported
dehydrated per exposed tungsten oxide site per second) for the
WO catalyst systems asymptotically approach the same value.
The similar TOR value at high W/nm is a consequence of
3
◦
2
supported WO catalysts at 230 C are presented in Fig. 12 as a
3
2
function of the W/nm . The TOF values were determined from
the predominance of the bulk-like WO crystallites and their
3
knowledge of the number of catalytic active sites obtained from
the CH3OH-TPSR spectra shown in Fig. 11. The relative TOF
values of the surface acidic catalytic active sites at monolayer
surface WOx coverage follow the trend WO3/Al2O3 ꢀ bulk
WO3 > WO3/TiO2 > WO3/Nb2O5 > WO3/ZrO2. The pres-
associated low dispersion at these high tungsten oxide surface
density values.
A more detailed comparison of TOR and TOF values is
shown for the supported WO /ZrO catalyst system in Fig. 14
3
2
(note that the y-scales are different for TOR and TOF). The

T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
379
4
.2. Molecular and electronic structures of the supported
tungsten oxide phases
All of the supported WO3 catalysts exhibit Raman bands
for the dehydrated surface WOx species between ∼1000 and
−
1
1
020 cm as the tungsten oxide surface density is increased
in the submonolayer coverage region (see Figs. 1–4). The cor-
responding UV–vis DRS Eg values for supported WO /Al O
3
2
3
and WO3/ZrO2 catalyst systems decrease simultaneously with
increasing tungsten oxide surface density, reflecting an increase
in extent of polymerization of the surface WO species with
x
increasing surface coverage (see Fig. 6). The simultaneous in-
crease in the Raman vibrational band position of the terminal
W=O bond and the decrease in UV–vis DRS Eg values for
the supported WO /Al O and WO /ZrO catalyst systems re-
Fig. 14. Comparison of catalytic acidity TOR and TOF values for the mod-
el-supported WO /ZrO catalysts as a function of tungsten oxide surface den-
sity (W/nm ).
3
2
3
3
2
3
2
flect the polymerization of the surface WO species with sur-
x
2
face coverage on these model oxide supports. Although the
UV–vis DRS Eg values cannot be obtained for the supported
WO /TiO and WO /Nb O catalyst systems, the Raman shift
3
2
3
2
−
5
TOF and TOR values increase continuously with increasing
W/nm below monolayer coverage. The similar TOF and TOR
1
from ∼1000 to 1016 cm also reflects the polymerization of
the surface WOx species on these model oxide supports, es-
pecially supported WO3/TiO2, with increasing tungsten oxide
surface density. Consequently, the surface WOx species pro-
gressively polymerize with increasing tungsten oxide surface
density in the submonolayer region on these model oxide sup-
ports.
2
trend is related to the 100% dispersion of the supported tungsten
oxide phase in the submonolayer region. The increasing TOF
2
and TOR values with surface tungsten oxide density (W/nm )
below monolayer coverage reflect the greater catalytic acidity
of the surface polytungstate species compared with the sur-
face monotungstate species. The further increase in TOF and
Above monolayer surface coverage of the 2D surface WOx
2
TOR values between 5 and 10 W/nm demonstrates that the
phase, 3D crystalline WO3 NPs are present between ∼5 and
WO3 NPs are more active than the surface WOx species. Above
2
∼
10 W/nm . Raman readily detects the crystalline WO3 phase
2
1
0 W/nm , the low dispersion of the bulk-like WO3 particles
because of the significantly stronger Raman cross-section of
bulk WO crystals than the surface WO species [34]. The
significantly decreases the TOR values as the TOF values con-
tinue to increase. The continued increase of the TOF catalytic
acidity values reflects the higher specific catalytic acidity of the
3
x
HR-TEM images reveal that the supported crystalline WO3 par-
ticles on ZrO2 are on the order of ∼1 nm in size between 5
bulk-like WO particles relative to the WO NPs.
3
3
2
and 12 W/nm . The NP dimension of the supported crystalline
WO3 phase is also reflected in both its broader Raman bands
and higher Eg values relative to large bulk WO3 crystals. At
very high tungsten oxide surface density, the crystalline WO3
NPs are larger and begin to reflect the electronic structural char-
acteristic of large bulk crystalline WO3 particles.
4
. Discussion
4
.1. Tungsten oxide monolayer surface coverage
The Raman spectra of the supported WO3 catalysts quan-
titatively demonstrate that tungsten oxide monolayer surface
4
.3. Number of exposed surface catalytic active acid sites
coverage on the model oxide supports corresponds to ∼4.5
2
2
W/nm . Below ∼4.5 W/nm , surface WO species predom-
x
inate, and only minor amounts of crystalline WO3 NPs are
present as monolayer surface coverage is approached (see
Figs. 1–5). From corresponding X-ray photoelectron spec-
troscopy (XPS) surface analysis of the surface W/Zr ratio as
The number of surface acid sites was quantitatively deter-
mined from the amount of DME formed during the CH3OH-
TPSR experiment (see Fig. 11). Below tungsten oxide mono-
layer surface coverage (<5 W/nm ), the number of surface
2
2
a function of W/nm for the same model-supported WO /ZrO
acid sites increases linearly with the tungsten oxide surface den-
sity. Above monolayer surface coverage, the number of exposed
surface acid sites continuously decreases with tungsten oxide
surface density on all of the supports because of the presence of
crystalline WO3 NPs. The crystalline WO3 NPs reside on top of
the surface WOx monolayer and also cover up some of the sur-
face WOx acid sites. The trend in the quantitative decrease in
the number of surface acid sites varies only slightly with tung-
sten oxide surface density among the different oxide supports.
Furthermore, above tungsten oxide monolayer surface cover-
3
2
catalysts, monolayer surface coverage was determined to oc-
2
cur at ∼4.8 W/nm [35]. The conclusion that the tungsten
oxide monolayer surface coverage on model oxide supports
2
corresponds to ∼4.5–4.8 W/nm is in agreement with almost
all previous monolayer surface coverage determinations that
for model-supported WO catalysts monolayer coverage cor-
3
2
responds to ∼4–5 W/nm and is independent of the specific
oxide support with the exception of the weakly interacting sup-
ported WO /SiO catalyst system [15,17,21,24,25,36–50].
3
2

380
T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
age, the surface acid sites are associated with both the surface
4.6. Steady-state TOR catalytic acidity values
WOx species and the WO3 NPs.
The steady-state TOR catalytic acidity values are some-
times also reported in the catalysis literature because the actual
number of exposed surface tungsten oxide sites is not deter-
mined. As a result, the reaction rate data are normalized by
all of the atoms present in the supported metal oxide phase,
all of the W oxide atoms in supported WO3 catalysts. Conse-
quently, the actual dispersion of the supported WO3 phase is
not taken into consideration. The current study with the model-
supported WO3 catalysts allows a direct comparison between
steady-state TOF and TOR values for the same catalysts and
reaction where the dispersion of the supported WO3 phase is
known. The steady-state TOR and TOF catalytic acidity values
for CH3OH dehydration are identical in the submonolayer re-
gion and at monolayer surface coverage because the supported
WO3 phase is 100% dispersed as surface WOx species on the
oxide support. Above tungsten oxide monolayer surface cover-
age, the TOR and TOF catalytic acidity values begin to deviate
because the presence of 3D crystalline WO3 NPs decreases the
supported WO3 dispersion. As a result, the TOF catalytic acid-
ity values will always be greater than the corresponding TOR
catalytic acidity values because TOR overcounts the number
of actual exposed catalytic acid sites above monolayer surface
coverage. This overcounting of exposed catalytic active sites in
the determination of TOR results in decreasing the TOR cat-
alytic acidity values at high tungsten oxide surface density.
For the supported WO3/Al2O3 catalysts above monolayer
coverage, the TOR catalytic acidity values decrease with in-
4
.4. Relative catalytic acidity of surface WOx species and
crystalline WO3
The relative catalytic acidity of the surface monotungstate
and polytungstate WOx species and the WO3 NPs is dependent
on the specific oxide support and was determined by the kinet-
ics of DME formation during CH3OH-TPSR (see Fig. 10). For
all of the supports investigated, the surface polytungstate WOx
species are more acidic than the surface monotungstate WOx
species. This is reflected in the higher catalytic acid activity of
the monolayer surface coverage catalysts with the correspond-
ing submonolayer surface coverage catalysts. The catalytic acid
activity of the submonolayer-supported WO3/Al2O3 catalyst
could not be determined because the exposed surface Al sites
were very active and overwhelmed the surface reactivity of the
surface monotungstate WOx species at low surface coverage.
For the supported WO3/Al2O3 catalytic system, the sur-
face WOx species at monolayer coverage are more acidic
than the WO3 NPs. The supported WO3 NPs are more acidic
than the surface WOx species for the supported WO3/Nb2O5,
WO3/TiO2, and WO3/ZrO2 catalytic systems, however. This
reveals the important contribution of the specific oxide sup-
port to the acidic character of the surface WOx species. The
catalytic acidity at tungsten oxide monolayer surface cover-
age increases with increasing support cation electronegativity
(
cation electronegativity, Zr < Nb ∼ Ti < Al) [51]. The greater
2
creasing W/nm because of the overcounting of the exposed
the support cation electronegativity, the less electron density re-
sides on the bridging W–O–support bond (less basic and more
acidic). This reflects the ability of the support cation to tune the
acidic character of the surface WOx species relative to the WO3
NPs. With the exception of the supported WO3/Al2O3 catalytic
system, which has an extremely acidic surface WOx monolayer,
the large WO3 particles present on Nb2O5, TiO2, and ZrO2 are
more acidic than the surface WOx species and the WO3 NPs.
surface sites and the lower intrinsic catalytic acidity of the WO3
NPs (see Fig. 13). As mentioned earlier, the TOR catalytic acid-
ity values for the supported WOx species on Al2O3 cannot be
determined below monolayer coverage because of the much
greater catalytic acidity of the exposed surface Al sites in the
submonolayer region.
For the other supported WO3 model catalysts (supported
WO3/Nb2O5, WO3/TiO2, and WO3/ZrO2), the TOR catalytic
2
acidity values increase between ∼5 and 10 W/nm because of
4
.5. Steady-state TOF catalytic acidity values
the much higher intrinsic catalytic acidity of the WO3 NPs rel-
ative to the surface WOx species (see Figs. 13 and 14), while
the number of actual catalytic active sites decreases by only less
than a factor of 2 (see Fig. 12). For tungsten oxide surface den-
For the supported WO3/Al2O3 catalyst system, the TOF cat-
alytic acidity values slightly decreased above tungsten oxide
monolayer surface coverage because of the less acidic WO3
NPs relative to the surface WOx monolayer (TOFPolytungstate
2
sity >10 W/nm , the decrease in the number of actual catalytic
acid sites is greater than the modest increase in catalytic acidity
>
TOFWO ,NPs). However, the TOF catalytic acidity values in-
from the bulk-like WO particles and, thus, results in an over-
3
3
creased continuously with tungsten oxide surface density for
supported WO3/Nb2O5, WO3/TiO2, and WO3/ZrO2, indicating
that TOFMonotungstate < TOFPolytungstate < WO3,NPs < WO3,Bulk.
Thus, the overall TOF catalytic acidity depends on the intrinsic
catalytic acidity of the surface WOx species and the WO3 NPs,
as well as on their relative contributions to the total number of
exposed surface acid sites. Consequently, there is no standard
relationship between the tungsten oxide domain size and the
specific catalytic acidity because of the significant role of the
specific oxide support ligand on the acidity of the surface WOx
species.
all decrease in TOR catalytic acidity values. Consequently, the
initial increase and subsequent decrease of the TOR catalytic
2
acidity values with W/nm above monolayer coverage is just
a consequence of (i) the relative intrinsic catalytic activity of
the WO NPs and the surface WO monolayer, (ii) the partition
3
x
of exposed catalytic acidic sites between the WO NPs and the
3
surface WO species, and (iii) ratio of actual number of exposed
x
catalytic acid sites and the total number of W oxide atoms in
the supported tungsten oxide phase. Thus, the maximum in the
2
TOR catalytic acidity values with W/nm for model-supported
WO3/ZrO2, WO3/Nb2O5, and WO3/TiO2 catalysts is primarily

T. Kim et al. / Journal of Catalysis 246 (2007) 370–381
381
a consequence of overcounting the actual number of catalytic
[9] M. Ai, J. Catal. 49 (1977) 305.
10] S. Morikawa, K. Takahashi, J. Mogi, S. Kurita, Bull. Chem. Soc. Jpn. 55
acidic sites present in supported WO3 catalysts at high W/nm2
[
(1982) 2254.
values.
[
[
11] M. Hino, K. Arata, J. Chem. Soc. Chem. Commun. 18 (1988) 1259.
12] J.G. Santiesteban, J.C. Vartuli, S. Han, R.D. Bastian, C.D. Chang, J.
Catal. 168 (1997) 431.
5
. Conclusion
[13] M. Hino, K. Arata, Chem. Lett. 6 (1989) 971.
The nature of the dehydrated surface tungsten oxide phase
[14] C. Martin, G. Solana, P. Malet, V. Rives, Catal. Today 78 (2003) 365.
[15] D.S. Kim, M. Ostromecki, I.E. Wachs, J. Mol. Catal. A Chem. 106 (1996)
93.
is found to be the same on the different supports examined
2
(
Al2O3, ZrO2, TiO2, and Nb2O5) as a function of W/nm .
2
[16] J.-M. Jehng, A.M. Turek, I.E. Wachs, Appl. Catal. A 83 (1992) 179.
In the submonolayer region (<5 W/nm ), both monotungstate
and polytungstate surface WOx species are present, and the ra-
tio of polytungstate/monotungstate increases with tungsten ox-
[
[
17] I.E. Wachs, Catal. Today 27 (1996) 437.
18] R. Thomas, F.P.J.M. Kerkhof, J.A. Moulijn, J. Medema, V.H.J. de Beer, J.
Catal. 61 (1980) 559.
2
ide surface coverage. Above monolayer coverage (>5 W/nm ),
[19] P. Tiltarelli, A. Iannibello, P. Villa, J. Solid State Chem. 37 (1981) 95.
[
[
[
20] S.S. Chan, I.E. Wachs, L.L. Murrell, L. Wang, W.K. Hall, J. Phys. Chem.
3
D crystalline WO3 NPs are present on top of the 2D surface
2
88 (1984) 5831.
WOx monolayer. Above ∼10 W/nm , bulk-like WO3 large par-
ticles are also present. These different molecular structures of
the supported tungsten oxide phase affect the overall dispersion
of the supported WO3 catalysts. The dispersion is 100% in the
submonolayer region and decreases continuously with increas-
ing tungsten oxide loading above monolayer coverage due to
the presence of the crystalline WO3 particles.
21] L. Salvati, L.E. Makovsky, J.M. Stencel, F.R. Brown, D.M. Hercules,
J. Phys. Chem. 85 (1981) 3700.
22] P. Biloen, G. Pott, J. Catal. 30 (1973) 169.
[23] S. Djerad, L. Tifouti, M. Crocoll, W. Weisweiler, J. Mol. Catal. A Chem.
208 (2004) 257.
[
24] J.A. Horsley, I.E. Wachs, J.M. Brown, G.H. Via, F.D. Hardcastle, J. Phys.
Chem. 91 (1987) 4014.
[25] M.M. Ostromecki, L.J. Burcham, I.E. Wachs, J. Mol. Catal. A Chem. 132
The relative catalytic acidity of the different tungsten oxide
components depends on the specific oxide support. For sup-
ported WO3/Al2O3 catalysts, the surface WOx species are more
active than the crystalline WO3 particles. For the other sup-
ported WO3 catalysts, however, the crystalline WO3 particles
are more active than the surface WOx species. These different
catalytic acidity patterns of the different tungsten oxide struc-
tures reflect the important effect of the specific oxide support
on the relative acidic activity of the surface WOx species to the
WO3 NPs. As the support cation electronegativity increases (Al
(
1998) 59.
[26] L.H. Gielgens, M.G.H. Vankampen, M.M. Broek, R. Vanhardeveld, V.
Ponec, J. Catal. 154 (1995) 201.
27] J. Macht, C.D. Baertsch, M.M. Lozano, S.L. Soled, Y. Wang, E. Iglesia,
[
J. Catal. 227 (2004) 479.
[
[
[
28] T. Feng, J.M. Vohs, J. Catal. 208 (2002) 301.
29] L.E. Briand, W.E. Farneth, I.E. Wachs, Catal. Today 62 (2000) 219.
30] P.A. Redhead, Vacuum 12 (1962) 213.
[31] E.I. Ross, T.J. Kim, I.E. Wachs, in preparation.
[32] L.A. Nafie, in: I.R. Lewis, H.G.M. Edwards (Eds.), Handbook of Raman
Spectroscopy, Dekker, New York, 2001, p. 1.
[
[
[
[
33] M.A. Vuurman, I.E. Wachs, J. Phys. Chem. 96 (1992) 5008.
>
Nb ∼ Ti > Zr), the electron density of the bridging W–O–
34] S.S. Chan, I.E. Wachs, L.L. Murrell, J. Catal. 90 (1984) 150.
35] W.V. Knowles, Ph.D. dissertation, Rice University, Houston, TX (2006).
36] M. Scheithauer, R.K. Grasselli, H. Knozinger, Langmuir 14 (1998) 3019.
[37] Y.S. Chen, I.E. Wachs, J. Catal. 217 (2003) 468.
38] M. Scheithauer, T.K. Cheung, R.E. Jentoft, R.K. Grasselli, B.C. Gates,
support bond deceases and results in a more acidic site. Con-
sequently, there is no general relationship between the tungsten
oxide structure/domain size and the specific catalytic acidity.
[
H. Knozinger, J. Catal. 180 (1998) 1.
Acknowledgment
[
[
39] C.D. Baertsch, S.L. Soled, E. Iglesia, J. Phys. Chem. B 105 (2001) 1320.
40] D.G. Barton, M. Shtein, R.D. Wilson, S.L. Soled, E. Iglesia, J. Phys.
Chem. B 103 (1999) 630.
Support from the U.S. Department of Energy, Division of
Basic Energy Sciences (Grant DE-FG 02-93 ER 14350) is
gratefully acknowledged.
[
41] N. Vaidyanathan, M. Houalla, D.M. Hercules, Catal. Lett. 43 (1997) 209.
[42] N. Vaidyanathan, D.M. Hercules, M. Houalla, Anal. Bioanal. Chem. 373
2002) 547.
43] N. Vaidyanathan, M. Houalla, D.M. Hercules, Surf. Interface Anal. 26
1998) 415.
(
[
[
References
(
44] C. Pfaff, M.J. Perez-Zurita, C. Scott, P. Patino, M.R. Goldwasser, J. Gold-
wasser, F.M. Mulcahy, M. Houalla, D.M. Hercules, Catal. Lett. 49 (1997)
[1] C.L. Thomas, Catalytic Processes and Proven Catalysts, Academic Press,
New York, 1970.
13.
[
2] C.N. Satterfield, Heterogeneous Catalysis in Practice, McGraw–Hill, New
York, 1980.
[
[
45] N. Naito, N. Katada, M. Niwa, J. Phys. Chem. B 103 (1999) 7206.
46] X.F. Yu, N.Z. Wu, H.Z. Huang, Y.C. Xie, Y.Q. Tang, J. Mater. Chem. 11
[3] L.L. Murrell, C.J. Kim, D.C. Grenoble, US Patent 4 233 139 (1980).
[4] D.C. Grenoble, C.J. Kim, L.L. Murrell, US Patent 4 440 872 (1984).
[5] J. Bernholc, J.A. Horsley, L.L. Murrell, L.G. Sherman, S. Soled, J. Phys.
Chem. 91 (1987) 1526.
(2001) 3337.
[
[
47] J. Engweiler, J. Harf, A. Baiker, J. Catal. 159 (1996) 256.
48] G. Bond, S. Flamerz, L. van Wijk, Catal. Today 1 (1987) 229.
[
[
[
6] L.L. Murrell, D.C. Grenoble, C.J. Kim, N.C. Dispenziere, J. Catal. 107
[49] S.S. Chan, I.E. Wachs, L.L. Murrell, N.C. Dispenziere Jr., J. Catal. 92
(1985) 1.
[50] L. Salvati, J.M. Makovsky, J.M. Stencel, F.R. Brown, D.M. Hercules,
J. Phys. Chem. 85 (1981) 3700.
(
1987) 463.
7] C. Wivel, B.S. Clausen, R. Candia, S. Morup, H. Topsoe, J. Catal. 87
1984) 497.
8] I.E. Wachs, C.C. Chersich, J. Hardenbergh, Appl. Catal. 13 (1985) 335.
(
[51] R.T. Sanderson, J. Chem. Educ. 65 (1988) 112.