Tungsten oxide monolayer loaded on zirconia: Determination of acidity generated on the monolayer

Article abstract of DOI:10.1021/jp9906381

The benzaldehyde-ammonia titration (BAT) method clarified that the tungsta impregnated on zirconia formed a monolayer and almost completely covered the surface with ca. 4 W nm^-2. Generation of acidity on the monolayer was shown by the water vapor treatment method of ammonia TPD (temperature-programmed desorption). Strength of the acid site was calculated according to the theoretical equation to be ca. 130 kJ mol^-1 in adsorption heat of ammonia, which does not correspond to superacidity. The generated acid site was active for skeletal isomerization of butane.

Full text of DOI:10.1021/jp9906381

7206
J. Phys. Chem. B 1999, 103, 7206-7213
Tungsten Oxide Monolayer Loaded on Zirconia: Determination of Acidity Generated on
the Monolayer
Norihiro Naito, Naonobu Katada,* and Miki Niwa
Department of Materials Science, Faculty of Engineering, Tottori UniVersity, 4-101 Koyama-cho Minami,
Tottori 680-8552, Japan
ReceiVed: February 23, 1999; In Final Form: June 7, 1999
The benzaldehyde-ammonia titration (BAT) method clarified that the tungsta impregnated on zirconia formed
a monolayer and almost completely covered the surface with ca. 4 W nm-2. Generation of acidity on the
monolayer was shown by the water vapor treatment method of ammonia TPD (temperature-programmed
desorption). Strength of the acid site was calculated according to the theoretical equation to be ca. 130 kJ
mol-1 in adsorption heat of ammonia, which does not correspond to superacidity. The generated acid site
was active for skeletal isomerization of butane.
Introduction
species (amount of the produced nitrile) on a binary oxide
sample. The spreading of the acidic promoter loaded on basic
support has been measured on V2O5,20 MoO3,22 SiO2,23-25 and
GeO226 loaded on Al2O3, TiO2, ZrO2, and SnO2. In all of these
cases, it was shown that the acidic oxide formed a monolayer
that almost completely covered the surface upon the strong
interaction between the acidic and basic materials, and that the
formed monolayer was the active species. Recently the validity
of the BAT method was confirmed by the group study carried
out by many researchers using various techniques, i.e., X-ray
photoelectron spectroscopy (XPS), laser Raman spectroscopy
(LRS), electron probe microanalysis (EPMA), and nitrogen
monoxide adsorption. These techniques were applied to the same
MoO3/Al2O3 samples prepared under various conditions, and
they agreed well with the conclusion by the BAT method that
the loaded MoO3 formed a monolayer on Al2O3.27
Subsequently to the discovery of superacidity on a sulfate-
loaded zirconia catalyst,1,2 Hino and Arata found that the
tungsten oxide (tungsta) loaded on zirconia showed a high
catalytic activity for skeletal isomerization of alkanes.3 Later,
the activity was also found on various supports such as tin, iron,
and titanium oxides.4 The early studies proposed the superacidity
on the tungsta-loaded catalyst on the basis of the Hammet
indicator method.3,5 The nature of the tungsta- and sulfate-loaded
zirconia is similar; both catalysts change the color of a base
indicator with pKa < -13, and both catalysts have the activity
for the skeletal isomerization of such an alkane as n-butane.
Studies in the past decade seem to indicate consistently the
presence of the superacidity on the sulfate-loaded zirconia.6-8
However, the surface structure and the acidic property (acid
amount and strength) of the tungsta-loaded zirconia have not
been clarified, although several spectroscopic and catalytic data
have been published.6,8-18 The following fundamental questions
have not been solved for this oxide system:
In the present study, this BAT method was applied to the
tungsta-loaded zirconia in order to clarify the surface structure.
The chemisorption of carbon dioxide has been applied to a
tungsta-titania catalyst.28 The chemisorption of carbon mon-
oxide has been applied to tungsta-alumina catalyst to measure
the coverage.29
(1) Does the tungsta-loaded zirconia indeed have the super-
acidity?
(2) What type of surface/interface species has the acidity?
(3) Does the superacid site catalyze the alkane isomerization?
To overcome the difficulty of analyzing the structure of binary
oxide, it is promising to measure the spreading of loaded oxide.
We have developed a method to distinguish between the surface
areas of two oxides on one binary oxide sample. This is the
benzaldehyde-ammonia titration (BAT) method using the
strong chemisorption of benzaldehyde on basic oxides.19-21
Benzaldehyde is adsorbed on such basic oxides as alumina,
titania, tin oxide, and zirconia in high surface concentration,
while almost no benzaldehyde is adsorbed on such an acidic
oxides as silicon, vanadium, molybdenum, and tungsten oxides.
The amount of the adsorbed benzoate anion can be determined
precisely from the amount of the produced benzonitrile by the
reaction between the adsorbed anion and ammonia. Therefore,
it is possible to estimate the area of exposed fraction of the
surface of a basic oxide from the amount of adsorbed benzoate
On the other hand, we have studied the temperature-
programmed desorption (TPD) method of ammonia to measure
the acidic property of the solid quickly and easily.30 We have
dealt mainly with zeolites in the previous studies. The water
vapor treatment after the adsorption of ammonia removed the
unnecessary peak that was ascribed to the weakly held am-
monia31 at low temperature to simplify the spectrum.32-36 The
ammonia or ammonium cation bonded to the acid site was never
removed because of the strong basicity of ammonia.35 The
desorption of ammonia was found to be controlled by the
equilibrium between the gaseous and adsorbed ammonia, i.e.,
the readsorption freely occurred during the TPD experiments
under the experimental conditions widely utilized.31,37 The
entropy change with respect to the ammonia desorption was
constant on various zeolites,38 and a method to calculate the
adsorption heat of ammonia, namely acid strength, from the
peak intensity, position, and shape has been proposed.39 The
acidic properties of mordenite, ZSM-5,39 â,40 Y,35 and gallo-
* Corresponding author. Phone, +81-857-31-5684; fax, +81-857-31-
5684; E-mail, katada@chem.tottori-u.ac.jp.
10.1021/jp9906381 CCC: $18.00 © 1999 American Chemical Society
Published on Web 08/06/1999

Acidity Generated on Tungsten Oxide Monolayer
J. Phys. Chem. B, Vol. 103, No. 34, 1999 7207
silicate36 have been determined by the improved method of
ammonia TPD.
On zirconia, a broad TPD spectrum is observed by the
conventional method, as shown in the following section. This
observation is inconsistent with the inactivity of zirconia for
acid-catalyzed reactions such as isomerization of alkane. The
activity is created by loading of tungsta. Probably zirconia
adsorbs ammonia via weak interactions with the high concentra-
tion, and the peak showing the acidity on the tungsta-loaded
catalyst may be hindered by the unnecessary peak. In the present
study, we applied the water vapor treatment method of TPD to
the tungsta-zirconia system in order to determine the acidic
property and to relate it to the catalytic activity for the skeletal
isomerization of alkane. The TPD spectra were measured by
varying the W (sample amount)/F (flow rate of carrier gas) ratio
to clarify what controlled the TPD process on these catalysts.
Based on the observed results, theoretical analysis of the TPD
spectrum was first applied to a WO3/ZrO2 system in order to
obtain the acid strength.
Figure 1. BET surface area of WO3/ZrO2 calcined at 573 (O and *)
and 923 K (b).
the W (sample amount)/F (flow rate of carrier gas) ratio was
ca. 12 kg s m-3. To show the influence of the W/F ratio on the
peak temperature, the W/F ratio was varied from 1 to 6000 kg
s m-3 on several samples.
Experimental Section
Catalyst Preparation. To avoid contamination of halogen
and sulfur, zirconium oxynitrate [ZrO(NO3)2] was used as the
precursor of zirconia. It was solved into a nitric acid solution,
and aqueous ammonia was slowly added to precipitate the
zirconium hydroxide until the pH reached ca. 11. The obtained
solid was washed with water and calcined at 573 K. It was put
into an aqueous solution of ammonium tungstate [5(NH4)2O‚
12WO3‚5H2O], followed by drying the water and calcining at
923 K for 4 h under atmospheric conditions.
Catalytic Reaction. The skeletal isomerization of n-butane
into isobutane (2-methylpropane) was carried out by a conven-
tional pulse method. The catalyst (250 mg) was set into a quartz
tube (4 mm i.d.) and pretreated at 873 K for 1 h in 20 cm3
min-1 of flowing helium, which was purified by passing a liquid
nitrogen trap. Then, a pulse of n-butane (1.3 cm3, 5.3 × 10-5
mol) was injected into the helium flow at 623 K, and the
products were analyzed by a gas chromatography (GC) with a
column of VZ-7. Small amount of byproducts, i.e., C1-3
hydrocarbons, were generally detected. The activity was shown
by the yield of major product, namely isobutane.
Structural Characterizations. The total surface area was
determined according to the BET equation from the nitrogen
adsorption experiments at 77 K with p/p0 ) 0.3.
IR Spectroscopy. The IR spectrum was recorded from 950
to 4000 cm-1 on the self-supporting disk 10 mm in diameter,
molded from 10 mg of the sample in an in situ cell with CaF2
windows. After evacuation at 873 K for 1 h followed by cooling
the sample disk for 5 min, the spectrum was recorded. The
spectrum of adsorbed ammonia was collected after the introduc-
tion of ammonia (13.3 kPa) for 30 min, followed by evacuation
for 30 min at 373 K. To show the adsorbed species formed/
diminished by the water vapor treatment, the introduction of
water vapor (ca. 3 kPa) and evacuation were twice repeated
after the adsorption of ammonia, and the spectrum was collected.
On the other hand, vapor of pyridine (ca. 400 Pa, the vapor
pressure at room temperature) was introduced into the cell at
373 K for 30 min, and the spectrum of the adsorbed pyridine
was recorded after evacuation at 573 K for 30 min.
The BAT experiments19 were carried out in a Pyrex reactor
(4 mm i.d.) after the pretreatment at 673 K for 1 h in oxygen
flow. Benzaldehyde was repeatedly injected at 523 K in a flow
of helium, which was purified with a liquid nitrogen trap. The
eluted aldehyde was monitored with an FID (flame ionization
detector), and the injection was done until no further adsorption
of aldehyde was observed. Finally, ammonia was injected at
673 K to desorb the adsorbed material as benzonitrile. The
formed nitrile was analyzed by the FID, and the coverage by
tungsta on zirconia was calculated from the following equation:
Exposure (%) ) 100 × amount of produced nitrile per BET
surface area (BN) on the binary oxide sample (molecules nm-2)/
BN on the unmodified zirconia (2.2 molecules nm-2).
Coverage (%) ) 100 - exposure (%).
Ammonia TPD. The TPD measurements30 were carried out
in a quartz cell 1 cm in diameter. The sample was evacuated at
873 K for 1 h, and ammonia (13.3 kPa) was introduced into
the cell at 373 K. After 30 min, excess ammonia was evacuated
for 30 min. Then, water vapor (ca. 3 kPa, the vapor pressure at
room temperature) was introduced into the cell for 30 min at
373 K, followed by evacuation for 30 min. After the introduction
of water vapor and the evacuation were repeated again, the
adsorbed ammonia was desorbed in helium flow under a reduced
pressure with 10 K min-1 of the heating rate from 373 K, and
the desorbed ammonia was analyzed by a mass spectrometer
(ULVAC, UPM-ST-200P). Although the molecular weight of
ammonia is 17, the fragment with m/e ) 16 was used to quantify
the ammonia, because the fragment 17 is affected by the
desorbed water.30,35 For standard experiments, the sample
amount was 0.1 g, the flow rate of helium was 0.044 mmol
s-1, and the pressure in the sample cell was 13.3 kPa; hence,
Results
Structural Characterizations. Figure 1 shows the change
of BET surface area by the loading of tungsten oxide. Zirconia
had a high surface area, ca. 250 m2 g-1 at 573 K. To show the
effect of the loading of tungsten oxide, the surface areas of the
loaded catalysts were measured after calcination at the same
temperature, 573 K. The loading decreased the specific surface
area shown by the unit of m2 g-cat-1. Since this decrease is
affected by the change of weight of the catalyst by loading, the
-1
surface area shown by the unit of m2 g-ZrO2 is also plotted;
it indicates that the loading of tungsta gradually decreased the
surface area.
The symbol b shows the surface area after calcination at 923
K. The high-temperature calcination seriously decreased the
surface area of zirconia into ca. 30 m2 g-1 by thermal sintering.

7208 J. Phys. Chem. B, Vol. 103, No. 34, 1999
Naito et al.
Figure 3. TPD spectra obtained by conventional (without water vapor
treatment, solid line) and water vapor treatment method (broken line)
over ZrO2 (a) and WO3/ZrO2 (b: 5.2 W nm-2).
Figure 2. Coverage by WO3 on ZrO2 surface measured by BAT
method over WO3/ZrO2 calcined at 573 (O) and 923 K (b).
However, the loaded tungsten oxide prevented the particle from
sintering, and as a result, ca. 100 m2 g-1 of the surface area
was obtained at ca. 20 wt % of WO3.
From the measured surface area, the surface concentration
of tungsten was calculated as follows:
W concentration (W-atom nm-2) ) [WO3 loading (wt %)/
100]/231.8 (formula weight of WO3) × 6.023 × 1023/[BET
surface area (m2 g-1) × 1018].
The WO3 loading, 20%, which provided the maximum
surface area, corresponded to ca. 5 W nm-2 of the surface
concentration of tungsten.
The X-ray diffraction (XRD) showed that the loading of
tungsta prevented the transformation of tetragonal to monoclinic
zirconia (spectrum not shown). As a result, only the XRD pattern
ascribable to tetragonal zirconia was observed on the sample
with 6.4 W nm-2 after calcination at 923 K. On the other hand,
the formation of free WO3 particle was suggested on the sample
with a high concentration of tungsta, 17.3 W nm-2, due to the
diffractions ascribable to the monoclinic phase of WO3. The
suppression of sintering and phase-transformation of zirconia
by the loaded tungsta, and the formation of free WO3 at the
high concentration of tungsta, are in agreement with previous
studies.3,10,17
Figure 2 shows the coverage measured by the BAT method
against the W concentration. On the samples calcined at 573
K, the coverage increased linearly against the W concentration
and arrived at ca. 100% with 10 W nm-2. Calcination at 923 K
made the coverage higher; the coverage arrived at almost 100%
with ca. 4 W nm-2 as shown in Figure 2 (b).
Temperature-Programmed Desorption (TPD). Figure 3
shows the TPD spectra on zirconia and WO3/ZrO2 (ca. 5 W
nm-2) obtained with a constant W/F ratio (ca. 12 kg s m-3) by
the conventional and novel water vapor treatment methods. By
the conventional method, in which the temperature was simply
raised after the adsorption of ammonia, a large and broad
desorption peak of ammonia was observed at 400-600 K on
zirconia. The water vapor treatment after the adsorption of
ammonia almost completely diminished the peak. In place of
the ammonia peak, indeed, the desorption peak of water (m/e
) 18) was observed. The mechanism of replacement is discussed
in the following section.
Figure 4. TPD spectra obtained by water vapor treatment method over
ZrO2 (a) and WO3/ZrO2 [(b) 1.5, (c) 3.1, (d) 5.2, (e) 6.4, (f) 8.2, (g)
17.3 W nm-2).
peak maximum shifted slightly to the higher temperature. As a
result, creation of a certain type of acid site by the loading of
tungsta was clearly indicated. The conventional method showed
the peak intensity on the WO3/ZrO2 (b) about five times higher
than that on zirconia (a); this can be associated with the BET
surface area. On the contrary, the TPD peak after the water vapor
treatment was observed only on the tungsta-loaded catalysts.
The TPD spectra were recorded by the water vapor treatment
method as shown in Figure 4. As shown in Figure 5, the loading
of tungsta created the acid site, and the surface concentration
of the acid site showed the maximum (ca. 1.3 nm-2) at 6.4 W
nm-2 of the tungsten concentration; at the maximum, the number
1
of acid sites was ca. /4-1/5 the number of tungsten atoms.
Further loading decreased the acidity.
On the samples with 5.1 and 5.8 W-atom nm-2, the TPD
spectra were recorded with varying of the W/F ratio from 1 to
6000 kg s m-3. The peak maximum temperature was shifted
with changing the W/F ratio. To interpret this behavior according
to the theory, the relationship between ln 1/Tm - ln A0W/F and
1/Tm, where Tm is peak maximum temperature (K) and A0 is
the amount of desorbed ammonia (mol kg-cat-1), is shown in
Figure 6. An almost linear relationship was obtained, indicating
that the TPD process is controlled by the equilibrium, as
explained in the next section.
On the other hand, on the WO3/ZrO2 catalyst, a peak of
ammonia was observed at 400-700 K by the conventional
method, and the water vapor treatment changed it a little; the

Acidity Generated on Tungsten Oxide Monolayer
J. Phys. Chem. B, Vol. 103, No. 34, 1999 7209
Figure 5. Surface concentration of acid site determined by ammonia
TPD with water vapor treatment method (b) and catalytic activity for
skeletal isomerization of n-butane to isobutane (2-methylpropane) at
623 K (O). The catalytic activity was shown by [yield of iso-
butane (mol) × flow rate of carrier gas (m3 h-1)/surface area of catalyst
(m2)].
Figure 7. IR spectrum of ZrO2 evacuated at 873 K (a), followed by
adsorption of ammonia and evacuation (b) at 373 K and two repetitions
of water vapor treatment and evacuation at 373 K (c).
Figure 6. Plots of ln 1/Tm - ln A0W/F against 1/Tm on WO3/ZrO2 (4,
5.1 and 3, 5.8 W nm-2).
Figure 8. IR spectrum of WO3/ZrO2 (5.2 W nm-2) evacuated at 873
K (a), followed by adsorption of ammonia at and evacuation (b) and
two repetitions of water vapor treatment and evacuation at 373 K (c).
Skeletal Isomerization of Butane. Figure 5 also shows the
catalytic activity for the isomerization of butane. A volcano-
shaped relationship against the loading was obtained. No activity
was observed on zirconia, and the loading of tungsta created
3180 and 1460 cm-1, as shown in Figure 8b. The peaks
generated at 3344 and 1608 cm-1 were also observed on zirconia
at 3338 and 1604 cm-1. A large absorption at 1237 cm-1
appeared in place of the peaks at 1223 (small) and 1174 cm-1
(large) observed on zirconia. The water vapor treatment little
changed the peak shape in the stretching region, 3000-4000
cm-1; made the peak at 1433 cm-1 slightly small; and generated
a small peak at 1262 cm-1, as shown in Figure 8c. Although
these small changes were observed, the spectrum of the
ammonia species adsorbed on the tungsta-loaded zirconia was
approximately maintained.
the activity. The activity showed the maximum at 6.4 W nm-2
,
which agreed with the maximum acidity shown by the TPD,
and further loading diminished the activity.
IR Spectroscopy. The IR spectrum of zirconia evacuated at
873 K (Figure 7 a) showed a peak at 3676 cm-1, assigned to
the hydroxyl group. After the adsorption of ammonia, peaks
were observed at 3338, 1604, 1445, 1223, and 1174 cm-1, as
shown in Figure 7b. As shown below, the peaks at 1604 and
1445 cm-1 are ascribable to the coordinated NH3 molecule and
NH4+ cation, respectively.41 The intensities of these peaks were
comparable. The water vapor treatment strongly suppressed the
peaks at 1604, 1445, and 1174 cm-1, whereas the band at 1223
cm-1 was less affected, as shown in Figure 7c.
In addition to the absorptions by zirconia at 3676 cm-1, the
loading of tungsta generated a new band at 982 cm-1, as shown
in Figure 8a. The ammonia species adsorbed on the tungsta-
loaded sample (5.2 W nm-2) showed new IR peaks at 3262,
After the adsorption of pyridine, the peak at 1447 cm-1
ascribable to the pyridine molecule coordinated on Lewis acid
site8,41 was observed (data not shown). On the other hand, the
WO3/ZrO2 catalyst possessed both Lewis and Brønsted acidity,
as shown by the IR band of pyridinium cation8,41 at 1542 cm-1
.
The spectra were collected on the samples with various tungsta
contents; Lewis acidity was monotonically decreased with

7210 J. Phys. Chem. B, Vol. 103, No. 34, 1999
Naito et al.
is suggested, as proposed by Xie and Tang,42 probably owing
to the strong interaction between the basic support and the acidic
promoter.
The suppression of thermal sintering of zirconia by the loaded
tungsta layer was observed, as shown in Figure 1. On the
zirconia precalcined at 573 K, it is suggested that the loaded
tungsta formed a thermally stable layer like an eggshell to
protect the oxide particle from sintering. Simultaneously, the
phase-transformation of zirconia from tetragonal into monoclinic
was inhibited. The suppression of sintering and phase-
transformation are in agreement with the previous studies.3,10
We also found that the silica monolayer loaded on alumina
prevented the particle from sintering and phase-transformation
at a quite high temperature, 1493 K, and confirmed by means
of the BAT method that the loaded monolayer protected the
particle from sintering like an eggshell.47 It is suggested that
such an acidic oxide promoter as tungsta and silica commonly
suppresses the thermal sintering of such a basic oxide as zirconia
and alumina via the tight bonding between them, based upon
the strong interaction which also induces the formation of an
ultrathin layer as described above.
Figure 9. Intensity of peaks due to pyridinium cation bounded to
Brønsted acid site (b, 1542 cm-1) and pyridine coordinated to Lewis
acid site (O, 1447 cm-1).
loading of tungsta, while Brønsted acidity showed the maximum
With the loading exceeding the monolayer region, the XRD
showed the formation of free WO3 crystal, as reported by other
authors by means of XRD and Raman spectroscopy.3,10,17 This
also supports that the interaction between excess tungsta and
the solid surface becomes weak when the surface has been
covered by the monolayer.
with 6 W-atom nm-2 of the concentration, as shown in Figure
9.
Discussion
Monolayer Structure. As shown in Figure 2, tungsten oxide
almost completely covered the surface with 10 W nm-2 of the
surface concentration at 573 K. This concentration would
correspond to the doubly or further accumulated layer of
tungsten oxide. Calcination at 923 K enhanced the coverage,
Water Vapor Treatment of TPD. On zirconia, ammonia
was strongly adsorbed unless the water vapor treatment was
applied, as shown by the large TPD peak (Figure 3a, solid line),
in which the desorption of ammonia continued up to a relatively
high temperature, ca. 600 K. On the basis of the early studies
of IR spectroscopy carried out on ammine complexes48 and solid
catalysts,41 the IR bands which were observed on zirconia
adsorbing ammonia at 3338, 1604, and 1174 cm-1 (Figure 7b)
were assigned to N-H stretching, asymmetric and symmetric
deformation of H-N-H, respectively, of the NH3 molecule
and the tungsten oxide covered the surface with ca. 4 W nm-2
.
The enhancement of coverage, namely the dispersion of the
tungsta species on the surface, in the temperature region from
573 to 923 K is in agreement with the reported observation by
Raman spectroscopy that the free WO3 phase once formed at
773 K was redispersed into the monolayer by calcination at 1073
K.10 The concentration 4 W nm-2 presumably shows the
monolayer coverage, because the W concentration of tungsta
monolayer loaded on such a basic metal oxide as alumina has
been proposed to be 4 to 7 W nm-2; Xie and Tang estimated
the maximum packing density of planar WO3 species as 0.21
g-WO3/100 m2,42 corresponding to 5.5 W-atom nm-2 of the
surface concentration, whereas Igresia et al. adopted 3.7 nm-2 10
based on the X-ray photoelectron spectroscopy (XPS) studied
+
coordinated to the solid, but not the NH4 cation. Lack of an
NH- anion,48 shown by the absence of absorption at 1510 cm-1
,
is consistent with the TPD spectrum in which the desorbed
species was only NH3 (m/e ) 16 and 17), but not N2 (28) and
H2 (2). The peak intensity at 1445 cm-1 ascribable41 to the
asymmetric deformation of the NH4+ was similar to that of the
NH3 molecule at 1604 cm-1; this indicates the small amount of
+
the NH4 species based on the molar absorption coefficiency,
ca. 1:7 for NH3 and NH4+. Therefore, the major species on
zirconia is concluded to be the NH3 molecule adsorbed on the
Lewis acid site, namely the Zr4+ cation, or hydrogen-bonded.
The former is more reliable, because the desorption temperature
was quite high (up to ca. 600 K, as shown above).
on WO3/Al2O3 or 7 nm-2 based on the crystal structure of
43
WO3.18 Other authors also proposed similar surface concentra-
tions of tungsten atoms on the monolayer loaded on metal
oxides.44-46 In conclusion, after calcination at 923 K, the
monolayer of tungsta completely covered the surface of zirconia.
The full coverage by the monolayer on zirconia supports the
observations by Igresia et al.; they suggested the monolayer
formation based on the change in BET surface area and X-ray
diffraction.17 The IR band at 982 cm-1 on the tungsta monolayer
(Figure 8a) is in agreement with Raman spectroscopy.10,17
These IR peaks on zirconia were almost completely dimin-
ished by the contact with water vapor, as shown in Figure 7c.
The TPD experiments also indicate that most of the adsorbed
species was replaced by the water molecule, resulting in the
removal of the TPD peak, as shown in Figure 3a. It was also
observed on zeolites that the adsorbed ammonia species was
removed by the water vapor treatment after the adsorption of
ammonia, and, as a result, the unnecessary l-peak was dimin-
ished.35,36 The origin of replacement of ammonia by water on
zeolite is suggested to be the polarity of the OH bond in the
water molecule, which is considered to be stronger than that of
NH bond in ammonia based on the higher dipole moment;49
The origin of high coverage is speculated to be the strong
interaction between the support and the promoter. We have
found the similar monolayer coverage on V2O5,20 MoO3,22
SiO2,23,24 and GeO2 loaded on such a basic oxide as Al2O3,
26
TiO2, SnO2, and ZrO2. Moreover, the other authors also
observed the monolayer coverage on WO3/Al2O3 and WO3/TiO2
by means of the adsorption of carbon monoxide29 and dioxide,28
respectively. In these cases, spontaneous monolayer dispersion
+
the ammonia hydrogen-bonded to the NH4 cation50 was
selectively replaced by the water molecule to form a stronger

Acidity Generated on Tungsten Oxide Monolayer
J. Phys. Chem. B, Vol. 103, No. 34, 1999 7211
hydrogen bond.35 However, the present case of the zirconia-
tungsta system seems different, because the adsorbed species
is suggested to be coordinated to the Lewis acid site, as
described above. Another explanation is required for the
adsorption site on zirconia. Probably the hydration of the surface
should completely change the nature of the zirconia surface. It
is suggested that only the dehydrated surface of zirconia has
Lewis acidity, but the hydrated surface, probably filled with
ZrOH groups, has no or quite weak acidity, resulting in the
elimination of ammonia, i.e., removal of the TPD peak.
On the other hand, the large peaks of NH4+ cation adsorbed
on the Brønsted acid site (3180 cm-1 due to the N-H stretching
and 1433 cm-1 due to the H-N-H deformation41) were
observed on the tungsta-loaded zirconia (Figure 8b). This agrees
with the presence of Brønsted acidity shown by the spectrum
of adsorbed pyridine (Figure 9). The peaks ascribed to the NH3
molecule coordinated to the Lewis acid site were also observed;
the asymmetric deformation was observed at 1608 cm-1 and
symmetric deformation was observed at 1237 cm-1. The
wavenumber of the latter was higher than that on zirconia, 1174
cm-1, showing the species coordinated not on Zr4+, but on
another element, namely W; this indicates the Lewis acidity
generated on the loaded tungsta layer. The spectrum of adsorbed
pyridine also shows the presence of both Brønsted and Lewis
acidity on the sample fully covered by the tungsta monolayer.
Both species adsorbed on Brønsted and Lewis acid sites of the
tungsta layer were not diminished by the water vapor treatment,
as shown in Figure 8c. The TPD spectra (Figure 3b) also indicate
that the species adsorbed on the tungsta layer was not removed
by the water vapor treatment.
acid site by a certain type of cluster consisting of multiple W,
Zr, O, and H atoms.
To determine the acid strength from the TPD spectrum, the
TPD measurements were carried out by varying the W/F ratio
on the samples which had almost completely been covered by
the monolayer. Cvetanovic and Amenomiya classified the TPD
experiments into the following three cases: (1) The process is
controlled by the kinetics, namely, the activation energy is high.
(2) The process is controlled by the equilibrium between gaseous
and adsorbed ammonia, namely, readsorption of ammonia freely
occurs. (3) The process is controlled by the slow diffusion.54
From the linear relationship between ln1/Tm - ln A0W/F and
1/Tm shown in Figure 6, the present TPD experiments are
classified into the second case, as observed on various zeo-
lites.30,38 This is the first classification of ammonia TPD on a
nonzeolitic catalyst.
Two parameters, ∆H and ∆S, were calculated based on the
linear relationship shown in Figure 6 according to the derived
equation30
A0W
F
â(1 - θm)2(∆H - RTm)
∆H
RTm
ln Tm - ln
)
+ ln
(1)
∆S
P0 exp
(
)
R
where R is the gas constant (8.314 J K-1 mol-1), â is the heating
rate (K s-1), θm is the coverage of the acid site by ammonia at
the peak maximum, P0 is the pressure at standard conditions
(1.013 × 105 Pa), ∆H is the adsorption heat of ammonia (J
mol-1), and ∆S is the entropy change with respect to the
desorption of ammonia (J K-1 mol-1). The adsorption heat ∆H,
i.e., acid strength, was determined from the slope to be 128 kJ
mol-1 for both samples.
In conclusion, the water vapor treatment selectively removed
the ammonia species adsorbed on zirconia support. Applying
this method to the ammonia TPD clearly demonstrated the
creation of acidity, as shown by the change in the TPD spectrum
by varying the loading amount of tungsta (Figure 4).
On the other hand, the entropy change ∆S was calculated to
be 160-165 J K-1 mol-1 for each experimental run. The entropy
change ∆S must consist of the terms of phase-transformation
Acidic Property. By the measurements of TPD with the
water vapor treatment method, the generation of an acid site is
clearly concerned with the coverage by monolayer, as shown
in Figure 5. The maximum concentration of the acid site was
observed at 6 W nm-2, where the tungsta monolayer almost
completely covered the surface. As shown by the IR study,
Brønsted acidity was generated. The maximum acidity at this
tungsten concentration region is generally in agreement with
the observations on tungsta loaded on alumina51 and other
supports.52 Excess of tungsta decreased the acidity to almost
zero with ca. 18 W nm-2. The diminishing of acidity suggests
that the inactive layer of tungsta further accumulated on the
surface of the active monolayer.
The concentration of the acid site (1.3 nm-2) was almost 1/4-
1/5 of the concentration of the tungsten atom (6 nm-2) on the
monolayer. This suggests the stoichiometric generation of the
acid site by a certain type of cluster consisting of several W,
Zr, O, and H atoms.
We have found similar results on the MoO3/SnO2 system:
The water vapor treatment removed the TPD peak on tin oxide,
while the peak on MoO3/SnO2 was maintained; on the mono-
layer of molybdena, the number of acid sites was almost 1/4 the
number of molybdenum atoms.53 Because tin and molybdenum
have properties similar to zirconium and tungsten, respectively,
these results support the present study.
Igresia et al. proposed distorted octahedral WOx clusters
possessing W-O-W bonds as an active species of WO3/ZrO2
catalyst based on the X-ray absorption and UV-vis spectros-
copy.10,17 This is in good agreement with the generation of an
and mixing30 as ∆S ) ∆Strans + ∆Smix, and ∆Smix ) -R(ln xNH
3
+ xHe/xNH ln xHe), where ∆Strans and ∆Smix are the terms of
3
phase-transformation and mixing, respectively, and xi means the
molar ratio of component i. The term ∆Smix at the peak
maximum is calculated from the experimentally observed
pressures of ammonia and helium to be 40 to 80, in most cases
60-70 J K-1 mol-1 for the experiments shown in Figure 6.
Therefore, the former term is determined to be ca. 90-100 J
K-1 mol-1. This value is almost same as those observed on
zeolites,30 and close to the entropy change with respect to the
vaporization of ammonia (97.2 J K-1 mol-1) and other various
liquids (80-110 J K-1 mol-1). These findings indicate that the
entropy change with respect to the desorption of ammonia from
the WO3/ZrO2 catalyst is mainly determined by the increase of
free volume of molecule,30 and the desorption process on the
WO3/ZrO2 system is controlled by the same rule as that for
zeolites. According to these findings, hereafter we can determine
the acid strength of the WO3/ZrO2 catalyst by one point
experiment of ammonia TPD.
The determined adsorption heat, ca. 130 kJ mol-1, is close
to the strength of the acid site generated by the isomorphous
substitution of aluminum into the silicate framework in the
ZSM-5 zeolite,39 which has never been categorized as a
superacid.
The material with an acid strength stronger than 100%
perchloric or sulfuric acid,55 namely the material whose H0
function is lower than -11.93, is termed superacid.56 Therefore,
the adsorption heat is related to the H0 scale of acid strength in
solution. It is assumed that the solid acid acts as an acid also in

7212 J. Phys. Chem. B, Vol. 103, No. 34, 1999
Naito et al.
an aqueous solution as
where ∆Gf,i0 is the standard Gibbs energy change for formation
of component i.57
WZ + NH3 (aq) a WZ-NH3
(2)
Since reaction (2) can be expressed as -(3)-(6), the Gibbs
0
energy change, ∆G2 , of reaction (2) can be calculated as
where WZ is the tungsta-zirconia catalyst. The equilibrium
constant of the assumed reaction at 298 K is roughly estimated
to be 1.4 × 1016 on the basis of the determined standard enthalpy
change (ca. 130 kJ mol-1) and the entropy change (95 J K-1
mol-1) of the reaction
∆G20 ) -∆G30 - ∆G60 ) -102 + 9.85) -92 kJ mol-1
(8)
Here we assume that the Gibbs energy change of suspension
of WZ into aqueous solution is close to that of WZ-NH3; it
must be balanced.
Therefore, the equilibrium constant K2 of reaction (2) at 298
K is roughly estimated to be
WZ-NH3 a WZ + NH3 (g)
(3)
For such a solid S with H0 < -11.93, the equilibrium constant
of the reaction with ammonia at 298 K in aqueous medium
0
-∆G2
K2 ) exp
) 1.38 × 1016
(7)
S + NH3 (aq) a S-NH3
(4)
RT
is roughly calculated to be > 1.5 × 1021 (see Appendix).
Therefore, it is concluded that the present tungsta-zirconia
catalyst with the apparently lower constant, 1.4 × 1016, is not
a superacid.
2. Calculation of the Equilibrium Constant on a Solid Acid
with H0 < -11.93. It is regarded that the H0 function of solid
acid which can convert the base indicator into the acidic form
is
Catalytic Activity. The catalytic activity for the skeletal
isomerization of n-butane was created by the loading of the
tungsta on zirconia, as shown in Figure 5. The maximum activity
was found at 6 W nm-2, and further accumulation of tungsta
layer diminished the activity. It is therefore considered that the
isomerization proceeded on the Brønsted acid site with 130 kJ
mol-1 in the adsorption heat of ammonia on the tungsta
monolayer. Lack of superacidity on these catalysts suggests that
the superacidity is not required for the skeletal isomerization,
at least not under these conditions.
H0 < pKa
(8)
where pKa is the acid dissociation constant of the conjugate
acid of the used indicator.58 Therefore, in the reaction
S + B a S-B
(9)
where S is a solid superacid with H0 < -11.93 and B is the
base indicator with pKa ) -11.93, the equilibrium constant K9
can be drawn as
Conclusion
[S-B]
[S][B]
K9 )
> 1
(10)
1. Tungsten oxide almost completely covered the surface of
zirconia with 5-6 W nm-2 based upon the strong interaction
between W and Zr.
In an aqueous solution, the equilibrium constant K4 of the
2. The improved ammonia TPD method showed the creation
of a Brønsted acid site with a strength corresponding to ca. 130
kJ mol-1 of the adsorption heat of ammonia on the tungsta
monolayer.
3. The Brønsted acid site on the monolayer was active for
skeletal isomerization of butane.
assumed reaction (4) can be expressed as
[S-NH3] [B][S-NH ]
[S-B]
3
K4 )
)
(11)
[S][NH3] [S-B][NH3] [S][B]
On the other hand, because pKa of the conjugate acid of B is
-11.93, and that of ammonia59 is 9.25, we can obtain equations
Appendix
[H3O+][B]
[BH+]
1. Calculation of the Equilibrium Constant of Reaction
(3) on the Tungsta-Zirconia Catalyst from the Determined
Thermodynamic Parameters. The enthalpy change of reaction
-log
) -11.93
(12)
0
(3), ∆H3 , is assumed to be 130 kJ mol-1, and the entropy
and
0
change ∆S3 is assumed to be 95 J K-1 mol-1. If temperature
dependence of ∆H30 and ∆S30 is ignored, we obtain the standard
[H3O+][NH3]
0
Gibbs energy change, ∆G3 , of reaction (3) at 298 K to be
-log
) 9.25
(13)
+
[NH4 ]
∆G30 ) ∆H30 - T∆S30 ) 102 kJ mol-1
(5)
The difference (12) - (13) shows
With respect to the dissolving of ammonia into an aqueous
solution
+
[B][NH4 ]
-log
) -21.18
(14)
[BH+][NH3]
NH3 (g) + aq a NH3 (aq)
(6)
Therefore, the equilibrium constant K15 of the reaction
BH+ (aq) + NH3 (aq) a B (aq) + NH4+ (aq) (15)
0
the standard Gibbs energy change, ∆G6 , is
∆G60 ) ∆Gf,NH3(aq)0 - ∆Gf,NH3(g)0 ) -26.50 + 16.65 )
-9.85 kJ mol-1 (7)
should be

Acidity Generated on Tungsten Oxide Monolayer
J. Phys. Chem. B, Vol. 103, No. 34, 1999 7213
+
(20) Niwa, M.; Matsuoka, Y.; Murakami, Y. J. Phys. Chem. 1987, 91,
[B][NH4 ]
4519.
(21) Niwa, M.; Suzuki, K.; Kishida, M.; Murakami, Y. Appl. Catal. 1991,
67, 297.
K15
)
) 1021.18 ) 1.514 × 1021 (16)
[BH+][NH3]
(22) Yamada, H.; Niwa, M.; Murakami, Y. Appl. Catal. 1993, 96, 113.
(23) Niwa, M.; Katada, N.; Murakami, Y. J. Phys. Chem. 1990, 94,
It is then assumed that the equilibrium (15) exists on the surface
of solid as well as in aqueous solution. In other words, the
equilibrium constant of the reaction
6441.
(24) Niwa, M.; Katada, N.; Murakami, Y. J. Catal. 1992, 134, 340.
(25) Katada, N.; Niwa, M. AdV. Mater., Chem. Vap. Deposition 1996,
2, 125.
(26) Katada, N.; Niwa, M.; Sano, M. Catal. Lett. 1995, 32, 131.
(27) Okamoto, Y.; Arima, Y.; Nakai, K.; Umeno, S.; Katada, N.;
Yoshida, H.; Tanaka, T.; Yamada, M.; Akai, Y.; Segawa, K.; Nishijima,
A.; Matsumoto, H.; Niwa, M.; Uchijima, T. Appl. Catal., A 1998, 170,
315.
S-B + NH3 (aq) a B (aq) + S-NH3
is assumed to be
(17)
(28) Vaidyanathan, N.; Houalla, M.; Hercules, D. M. Catal. Lett. 1997,
43, 209.
(29) Praff, C.; Zurita, M. J. P.; Scott, C.; Patin˜o, P.; Goldwasser, M.
R.; Goldwasser, J.; Mulcahy, F. M.; Houalla, M.; Hercules, D. M. Catal.
Lett. 1997, 49, 13.
[B][S-NH3]
[S-B][NH3]
) K15 ) 1.514 × 1021
(18)
(30) Niwa, M.; Katada, N. Catal. SurVeys Jpn. 1997, 1, 215.
(31) Niwa, M.; Iwamoto, M.; Segawa, K. Bull. Chem. Soc. Jpn. 1986,
59, 3735.
(32) Chester, A. W.; Higgins, J. B.; Kuehl, G. H.; Schlenker, J. L.;
Woolery, G. L. In Preprint of the 11th Meeting on the Reference Catalyst
of Japan; The Catalysis Society of Japan: Tokyo, 1987; p 22.
(33) Bagnasco, G. J. Catal. 1996, 159, 249.
Substituting (18) into (11) derives
[B][S-NH ]
[S-B]
[S-B]
[S][B]
3
K4 )
) 1.514 × 1021
(19)
(20)
[S-B][NH3] [S][B]
(34) Woolery, G. L.; Kuehl, G. H.; Timken, H. C.; Chester, A. W.
Zeolites 1997, 19, 288.
From equations (10) and (19), we obtain
(35) Igi, H.; Katada, N.; Niwa, M. In Proc. 12th Int. Zeolite Conf. Tracy,
M. M. J., Marcus, B. K., Bisher, M. E., Higgins, J. B., Eds.; Materials
Research Society: Warrendale, 1998, p 2643.
K4 > 1.514 × 1021
(36) Miyamoto, T.; Katada, N.; Kim, J.-H.; Niwa, M. J. Phys. Chem.,
B 1998, 102, 6738.
(37) Sawa, M.; Niwa, M.; Murakami, Y. Zeolites 1990, 10, 307.
(38) Niwa, M.; Katada, N.; Sawa, M.; Murakami, Y. J. Phys. Chem.
1995, 99, 8812.
(39) Katada, N.; Igi, H.; Kim, J.-H.; Niwa, M. J. Phys. Chem., B 1997,
101, 5969.
In summary, the solid superacid whose H0 function is lower
than -11.93 must have an equilibrium constant higher than
1.514 × 1021 for the assumed reaction with ammonia in an
aqueous solution.
(40) Katada, N.; Iijima, S.; Igi, H.; Niwa, M. Stud. Surf. Sci. Catal.
1997, 105, 1227.
(41) Kno¨zinger, H. In Handbook of Heterogeneous Catalysis; Ertl, G.,
Kno¨zinger, H., Weitkamp, J., Eds.; Wiley-VCH: Weinheim, 1997; p 707.
(42) Xie, Y.-C.; Tang, Y.-Q. AdV. Catal. 1997, 37, 1.
(43) Salvatti, L.; Makovski, L. E.; Stencel, J. M.; Brown, F. R.; Hercules,
F. D. J. Phys. Chem. 1981, 85, 3700.
References and Notes
(1) Tanabe, K.; Itoh, M.; Morishige, K.; Hattori, H. In Preparation of
Catalysts; Delmon, B, Jacobs, P. A., Poncelet, G., Eds.; Elsevier: Amster-
dam, 1976; p 65.
(2) Hino, M.; Kobayashi, S.; Arata, K. J. Am. Chem. Soc. 1979, 101,
6439.
(3) Hino, M.; Arata, K. J. Chem. Soc., Chem. Commun. 1987, 1259.
(4) Hino, M.; Arata, K. Bull. Chem. Soc. Jpn. 1994, 672, 1472.
(5) Hino, M.; Arata, K. Chem. Lett. 1989, 971.
(6) Arata, K. AdV. Catal. 1990, 37, 165.
(44) Vuurman, M. A.; Wachs, I. E. J. Phys. Chem. 1992, 96, 5008.
(45) Gru¨nert, W.; Feldhaus, R.; Anders, K.; Shpiro, E. S.; Minachev,
K. M. J. Catal. 1995, 154, 201.
(46) Karakonstantis, L.; Matralis, H.; Kordulis, C.; Lycourghiotis, A.
J. Catal. 1996, 162, 295.
(47) Katada, N.; Ishiguro, H.; Muto, K.; Niwa, M. AdV. Mater., Chem.
Vap. Deposition 1995, 1, 54.
(48) Nakamoto, K. In Infrared and Raman Spectra of Inorganic and
Coordination Compounds, 3rd ed.; John Wiley & Sons: New York, 1962;
p 197.
(49) Lide, D. R. In Handbook of Chemistry and Physics, 73rd ed.; CRC
Press: Boca Raton, Ann Arbor, London and Tokyo, 1992; pp 9-44.
(50) Earl, W. L.; Fritz, P. O.; Gibson, A. A. V.; Lunsford, J. H. J. Phys.
Chem. 1987, 91, 2091.
(7) Lin, C.; Hsu, C. Y. J. Chem. Soc., Chem. Commun. 1992, 1479.
(8) Corma, A. Chem. ReV. 1995, 95, 559.
(9) Vermaire, D. C.; van Berge, P. C. J. Catal. 1989, 116, 309.
(10) Kim, D.-S.; Ostromecki, M.; Wachs, I. E. J. Mol. Catal., A:
Chemical 1996, 106, 93.
(11) Igresia, E.; Barton, D. G.; Soled, S. L.; Miseo, S.; Baumgartner, J.
E.; Gates, W. E.; Fuentes, G. A.; Meitzner, G. D. Stud. Surf. Sci. Catal.
1996, 101, 533.
(12) Zhao, B.; Xu, X.; Gao, J.; Fu, Q.; Tang, Y. J. Raman Spectrosc.
1996, 27, 549.
(13) Santiesteban, J. G.; Vartuli, J. C.; Han, S.; Bastian, R. D.; Chang,
C. D. J. Catal. 1997, 168, 431.
(14) Scheithauer, M.; Grasselli, R.; Kno¨zinger, H. Prepr. Pap.sAm.
Chem. Soc., DiV. Pet. Chem. 1997, 42, 738.
(15) Martin, C.; Martin, I.; Rives, V.; Solana, G.; Loddo, V.; Palmisano,
L.; Sclafani, A. J. Mater. Sci. 1997, 32, 6039.
(16) Boyse, R. A.; Ko, E. I. J. Catal. 1997, 171, 191.
(17) Sohn, J.-R.; Park, M.-Y. Langmuir 1998, 14, 6140.
(18) Barton, D. G.; Soled, S. L.; Meitzner, G. D.; Fuentes, G. A.; Iglesia,
E. J. Catal. 1999, 181, 57.
(19) Niwa, M.; Inagaki, S.; Murakami, Y. J. Phys. Chem. 1985, 89,
3869.
(51) Soled, S. L.; McVicker, G. B.; Murrell, L. L.; Sherman, L. G.;
Dispenziere, Jr., N. C.; Hsu, S. L.; Waldman, D. J. Catal. 1988, 111, 286.
(52) Wachs, I. E. Catal. Today 1996, 27, 437.
(53) Niwa, M.; Igarashi, J. Catal. Today in press.
(54) Cvetanovic, R. J.; Amenomiya, Y. AdV. Catal. 1967, 17, 103.
(55) Hall, N. F.; Conant, J. B. J. Am. Chem. Soc. 1927, 49, 3047.
(56) Gillespie, R. J. Acc. Chem. Res. 1968, 1, 202.
(57) Atkins, P. W. In Physical Chemistry, 3rd ed.; Oxford University
Press: Oxford, 1986; p 818.
(58) Walling, C. J. Am. Chem. Soc. 1950, 72, 1164.
(59) Atkins, P. W. In Physical Chemistry, 3rd ed.; Oxford University
Press: Oxford, 1986; p 826.