You are on page 1of 22

Applied Catalysis A: General, 106 (1993) 51-72

Elsevier Science Publishers RV., Amsterdam


APCATA2642
Zirconia-supported vanadium oxide catalysts for
ammoxidation and oxidation of toluene:
Acharacterization and activity study
51
M. Sanati and A. Andersson
Department of Chemical Engineering II (Chemical Technology), University of Lund, Chemical
Center, P.O. Box 124, S-221 00 Lund (Sweden)
L.R. Wallenberg
National Center for HREM, Department of Inorganic Chemistry 2, University of Lund,
Chemical Center, P.O. Box 124, S-221 00 Lund (Sweden)
and
B. Rebenstorf
Department of Inorganic Chemistry 1, University of Lund, Chemical Center, P.O. Box 124,
S-221 00 Lund (Sweden)
(Received 8 March 1993, revised manuscript received 30 August 1993)
Abstract
Aseries of samples of vanadia supported on monoclinic zirconia were prepared with nominal loadings
from a halfup to sixteen theoretical vanadia layers. The samples were characterized with X-ray diffrac-
tion, scanning electron microscopy combined with energy dispersive X-ray analysis, high-resolution
electron microscopy, Raman and diffuse reflectance infrared spectroscopy, and were used in the oxida-
tion and the ammoxidation of toluene. At loadings in the monolayer range, Ramanand infrared bands
from decavanadate-like and dehydrated tetrahedral vanadia species were at ca. 990 and ca. 1025cm"',
respectively. Raman bands at 821 and 880 cm-
t
were present only at the lowest loading and are char-
acteristic of orthovanadate and pyrovanadate species, respectively. X-ray diffraction, Raman and in-
frared spectroscopic results revealed formation of some crystalline V206 and ZrV207 at loadings exceed-
ing a theoretical monolayer. In this case, consideration of Raman intensity variations allowed the
conclusion that additional non-crystalline vanadia must be present. According to high-resolution elec-
tron micrographs, this vanadia consists of an amorphous overlayer, 4-8 atomic layers thick. In toluene
oxidation zirconia-supported vanadia compared with crystalline V
2
0
6
was found less selective for ben-
zaldehyde formation. In toluene ammoxidation, on the other hand, vanadia on zirconia was found to
possess goodactivity and selectivity for benzonitrile formation. Amorphous vanadia was the most active
structure on zirconia, while the selectivities for nitrile and aldehyde formations were almost independent
of the loading for one theoretical layer and above.
Key wards: catalyst characterization (BET / XRD / DRIFTS / FT-Raman spectroscopy / EDX
Correspondence to: Dr. A. Andersson, Department of Chemical Engineering II (Chemical Tech-
nology), University of Lund, Chemical Center, P.O. Box 124, S-221 00 Lund, Sweden. Tel. (+46-
46)108280,fax. (46-46)146030.
0926-860X/93/$06.00 1993 Elsevier Science Publishers B.V. All rights reserved.
52 M. Sanati et aL/ Appl. Catal.A 106(1993) 51-72
/ SEM / HREM); toluene oxidation; toluene ammoxidation; vanadia-zirconia; zirconia
(monoclinic)
INTRODUCTION
Vanadium oxide catalysts are frequently used for the oxidation and the am-
moxidation of aromatic compounds [1-5]. Usually the specific surface area of
vanadium oxides is low [3] and, therefore, the active phase is fixed to a sup-
port. The role of the support is not only to increase the surface area and to give
better mechanical strength, it might also modify the active surface due to sup-
port-active phase interaction [6]. Regarding vanadia, titania is the most stud-
ied support, see for instance refs. 7-15. Especially, the anatase and (B)-phase
polymorphs oftitania are suitable as supports in o-xylene oxidation [8,16] and
toluene ammoxidation [15], respectively, a fact that has been explained as
being due to support-active phase interaction. Other supports for vanadia that
have been well characterized are alumina and silica [17-19].
In comparison with titania, alumina and silica, zirconia has so far received
much less attention as catalyst support. However, in the last few years there
has been increased interest in exploring the catalytic potentials of zirconia
[20], of which there are three stable polymorphs at atmospheric pressure, i.e.,
monoclinic at temperatures below1200C, face-centred tetragonal in the range
1200-2372C, and face-centred cubic at high temperatures. Also, there is evi-
dence for a high-pressure phase of orthorhombic symmetry [21]. In recent
works aiming at preparing a high surface area zirconia that is stable under
thermal and hydrothermal conditions, the textural properties and pore struc-
tures of monoclinic andtetragonal samples, both non-stabilized and stabilized,
have been comprehensively characterized [22-24]. For the CO/H
2
and CO
2
/
H
2
reactions, zirconia has been used as support for various metals [25-29],
and MoO
a
supported on zirconia was found to possess goodcatalytic properties
for the oxidationof methanol [30] and the oxidative dehydrogenation of ethanol
[31] . However, fewstudies have dealt with vanadia/zirconia catalysts for cat-
alytic oxidations. In butene oxidation, furan and buta-1,3-diene were reported
to be formed on zirconia with low vanadia loading, while acetaldehyde and
acetic acid were obtained at high loading [32]. The oxidation activitywas found
to correlate well with the concentration of amorphous vanadia. Also, infrared
spectroscopic studies of the oxidations of toluene, m-xylene, phenol and ben-
zene have been performed [33,34]. Characterizations of vanadia/zirconia and
vanadia/titania catalysts have revealed a similar coverage with vanadia on
both supports [35] and similar extended X-ray absorption fine structure (EX-
AFS) spectral features [36], which were concluded to derive from isolated
vanadate type of complexes. Acomparison of low-temperature oxygen chemi-
sorption (LTOC) results of vanadia supported on zirconia with those of van-
M. Sanati et al. / Appl. Catal: A 106 (1993) 51-72 53
adia supported on alumina, silica and titania, suggested a better dispersion on
the zirconia [37,38]. Niwa et a1. [39], using the benzaldehyde-ammonia titra-
tion and the NO-NH
s
rectangular pulse methods, in their comparison of van-
adia on zirconia (monoclinic), titania, alumina and silica supports concluded
that the exposure of the V
20S
(010) crystal face was the highest on zirconia
and titania, but that the surface of the zirconia support was not covered com-
pletely. Conclusions different from those above were reported by Scharf et al.
[40], who from DRIFT and Raman spectroscopic results suggested that zir-
conia has a lowability to disperse vanadia.
The present study was undertaken to explore the potential for the vanadia/
zirconia catalyst system in the catalytic oxidation of alkyl aromatic com-
pounds. Oxidation and ammoxidation of toluene to benzaldehyde and benzo-
nitrile, respectively, were chosen as model reactions. The monoclinic poly-
morph of zirconia was used as support for vanadia. Catalysts with a wide range
of vanadia loading, from a half to sixteen theoretical monolayers, were pre-
pared and characterized with X-ray diffraction (XRD), diffuse reflectance
Fourier transform infrared spectroscopy (DRIFTS), Raman spectroscopy,
scanning electron microscopy (SEM) combined with energy dispersive X-ray
spectroscopy (EDX), and high-resolution transmission electron microscopy
(HREM).
EXPERIMENTAL
Preparation of support and catalysts
Hydrous zirconia samples were prepared from water solutions saturated at
room temperature with zirconium oxychloride (Janssen Chimica, p.A.), to
which a 25 wt.-% aqueous solution of ammonia was added until the pHwas 10.
The resulting gel was washed repeatedly with portions of water followed by
settling in a centrifuge, until the test with AgNO
s
for chloride ions was nega-
tive. Then the precipitate was dried at 80
0
Cfor 20 h and further calcined in a
flowof air at 550
0
Cfor 4h, producingpredominantly monoclinic zirconia. Two
batches were prepared, givingslightly different specific surface areas (36.2and
23.6 m
2
g-\ respectively).
Catalysts with various loadings of vanadiumoxide on zirconia were prepared
using the impregnation technique. The desired amount of NH
4
VO
s
(Merck,
p.A.) was dissolved in a water solution of oxalic acid (Merck, p.A.) having a
pH < 1. To this solution the support was added, and the excess water was evap-
oratedat 70
0
Cwhile stirring. The resultingsolid was dried at 100
0
Cand finally
calcined in a flowof air at 500
0
Cfor 3 h. After sieving, the fraction of particles
in the range 0.150-0.425 mm was used in the experiments. Six samples with
different vanadia loadings were prepared. The loading, which is also included
in the catalyst notation, and the specific surface area of the samples and the
supports are given in Table 1. Fromthe unit cell dimensions of crystalline V20S
54
TABLE 1
Loading and specific surface area of catalysts
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72
Catalyst Loading Loading Surface area of Surface area of
(mg V
205/m
2
)G (layers)" support (m
2/g)
catalyst (m
2/g)
VZr-0.6 0.6 0.5 36.2 35.3
VZr-1.2 1.2 1.0 36.2 36.0
VZr-2.6 2.6 2.2 36.2 31.1
VZr-6.0 6.0 5.0 23.6 11.3
VZr-9.4 9.4 7.8 23.6 7.1
VZr-19.4 19.4 16.2 23.6 9.0
G Expressed per unit surface area of the fresh support.
b Number of theoretical layers assuming a monolayer corresponds to 1.2 mg V205/m2 surface area
of support.
[41], an average value for loading corresponding to a complete monolayer can
be calculated as 1.2 mg V
205
m-
2
surface area of support, if the (100), (010)
and (001) planes are considered. Thus, from Table 1 it can be concluded that
the vanadia loading varied from! of a theoretical monolayer up to 16 layers.
For comparisons, pure samples of V205and ZrV207were prepared. The pure
V205 sample, having a specific surface area of 5.7 m
2
g-l, was prepared by
decomposing NH
4VOa
(Merck, p.A.) in a flowof air at 450C. Apure ZrV
207
sample with a specific surface area of 0.2 m
2
g-l was obtained from a stoichi-
ometric mixture ofV
205
(Riedel-de Haen, p.A.) and Zr02 (Janssen Chimica,
p.A.) which was heated at 650C for five days with two intermittent grindings
followedby a final calcination at noc for 48h with one intermittent grinding.
The XRD patterns of the two reference products agreed with those in the
JCPDS diffraction file [42] ofV
205
(No. 41-1426) and ZrV
207
(No. 16-422),
respectively.
Characterization of catalysts
X-ray diffraction analysis of the catalysts was carried out on a Philips dif-
fractometer using a PW 1732/10 generator and Cu Ka radiation, and a Varian
AA-1275atomic absorption spectrophotometer was used for determination of
the vanadium content. The specific surface area of the samples was measured
on a gravimetric BET apparatus.
DRIFT spectra were recorded using a Mattson Polaris Fourier transform
infrared spectrometer. The samples were diluted with KBr to give 5-15 wt.-%
of sample. Spectra were recorded without further treatment in dry air. A res-
olution of 4 cm-1 was used, and 1000 scans were collected for each sample.
FT-Raman spectra were recorded on a Bruker IFS 66/FRA 106 instrument
equipped with a low power, diode pumped Nd:YAG laser (1064 nm) and a
M. Sanati et al. / Appl. Caial. A 106 (1993) 51-72 55
liquid nitrogen cooled, highly sensitive germanium diode detector. The laser
power was 100 mW and the resolution was 4 cm-1. For every spectrum, 1000
scans were averaged.
EDX point analysis and mapping of elements were performed with a Link
ANI0oo0 systemattached to a JEM 840Ascanning electron microscope. Sam-
ples of the pure Zr02 support and with vanadia contents of 2.2 and 7.8 theo-
retical monolayers (VZr-2.6 and VZr-9.4, respectively) were investigated. The
samples were gently dispersed without crushing or grinding onto a specimen
stub, clad with a conducting polymer. For elemental mapping of Zr, V, and
background noise, 6-8 particles of each sample were selected for point analysis
and quantification by a ZAF-4 computer program.
High-resolutionelectron microscopy was performed on the pure support, the
VZr-2.6 and the VZr-9.4 samples. A JEM-4000EX transmission microscope
dedicated to high resolution was used for recording of images at 400 kV, and
with a point-to-point resolution better than 0.17 nm. Overviews, final focus
and astigmatismadjustment were performed using image-intensifiedTV cam-
eras and low-dose techniques to ensure minimum beam damage of the speci-
men. Aslowgrowth of amorphous contamination is not uncommon, depending
on the vacuum level in the microscope and on the type of specimen [15]. Great
care was therefore exercised to avoid unnecessary exposure to the electron
beam. The growth rate of contamination was monitored by recording several
micrographs with a time interval of a fewminutes. The growth of contamina-
tion was found to be negligible at the lowcurrent density used for recording of
low-dose images.
The catalytic performance of preparedcatalysts was studied in the oxidation
and the ammoxidation of toluene to benzaldehyde and benzonitrile, respec-
tively. Carbon oxides were formed as the main by-products. An isothermal plug
flowreactor was used in the experiments, andthe catalyst samples were diluted
with quartz of the same particle size. It was verified that the measurements
were not affected by intraparticle and interparticle mass- and heat-transfer
gradients. Reactor and analysis arrangement were the same as described else-
where [43]. Data reported for toluene oxidation and ammoxidation are for
conversions in the range 20-30% and 10-15%, respectively. For the pure Zr02,
the data were measured at 2-4%conversion. The variation of conversion with
amount of catalyst was found to be linear up to this range, admitting the cal-
culation and comparison of specific reaction rates. Considering that two Zr02
preparations with differing specific surface area were used as support (Table
1), a comparison of catalytic performance of the two was investigated for a
vanadia loading of one theoretical monolayer. The activity data obtained
showed no significant difference in reaction rate expressed per unit surface
area and product distribution, why the catalyst preparations can be compared
as one series.
56
RESULTS
X-ray diffraction
M. Sanati et al. / Appl. Catal. A 106(1993) 51-72
The X-ray diffraction patterns of the two zirconia preparations were very
similar except for somewhat broader lines in case ofthe sample with the high-
est specific surface area. Acomparison of the d-values measured with those in
the JCPDS file (nos. 36-420 and 37-1484) [42] showed both samples to be
monoclinic zirconia. Only one weak line at 2.95 Aindicated the presence of
some zirconia with tetragonal structure (JCPDS file no. 17-923) [42]. Adopt-
ing the formula given by Mercera et a1. [22], the volume fraction of the meta-
stable tetragonal phase was estimated to be 6 and 7.5%, respectively, in the
two zirconia samples with specific surface areas of 23.6 and 36.2 m
2
g-l.
Table 2 shows diffraction lines other than those from zirconia of prepared
catalysts and the assignments as obtained from comparison with the reflec-
tions measured for V205 and ZrV207 and corresponding data in the JCPDS
file [42]. The two samples VZr-0.6 and VZr-1.2 with the lowest vanadia load-
ing showed no diffraction lines except for those of the support. In VZr-2.6, the
two strongest reflections of crystalline V205 were observed at lowintensity. In
addition to V205 formation, at high loadings formation of ZrV207 was ob-
served. However, there was a remarkable difference between the ZrV
2
0
7
re-
flections ofVZr-6.0 and VZr-9.4comparedwith those ofVZr-19.4. Two ZrV207
reflections at 3.9 and 3.6 A, which were present for VZr-6.0 and VZr-9.4, were
not in the diffractogram ofVZr-19.4, in spite of these lines being intense both
TABLE 2
X-ray diffraction lines other than those from the support of prepared vanadium oxide catalysts
Catalyst"
VZr-2.6
VZr-6.0
VZr-9.4
VZr-19.4
d/Relative intensity/phase"
4.38/3/V; 3.40/3/V
4.39/20/V+ZrV; 3.92/14/ZrV; 3.58/l8/ZrV; 3.4l/1/V; 2.52/5/ZrV;
2.44/5/ZrV; 2.13/1/ZrV; 1.97/5/ZrV; 1.92/2/V+ZrV
5.79/15/V; 5.06/21/ZrV; 4.38/40/V+ZrV; 4.07/5/V; 3.91/24/ZrV;
3.55/17/ZrV; 3.39/13/V; 3.09/13/ZrV; 2.87/18/V; 2.75/5/V; 2.42/7/ZrV;
2.12/2/ZrV; 1.96/4/ZrV; 1.91/8/V+ ZrV
5.75/17/V; 4.37/33/V+ ZrV; 4.08/14/V; 3.50/2/V; 3.40/44/V; 2.87/34/V;
2.76/16/V; 2.63/30/V+ ZrV; 2.52/5/ZrV; 1.92/13/V; 1.90/3/ZrV;
1.80/35/ZrV; 1.78/50/V; 1.75/5/V; 1.56/2/V+ZrV; l.49/21/V+ZrV
aVZr-0.6 and VZr-1.2 showed no diffraction lines except those from the support.
b The d-spacing is in A. Relative intensity is defined as 100 (I/I ref). where I
ref
is the zirconia line
at 3.16 A. The notations V and ZrV stand for V
2
0
5
and ZrV
2
0
7
respectively.
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72 57
in the reference sample and according to the JCPDS file (No. 16-422) [42].
This observation can be due to a change of ZrV207 morphology or crystallinity
with increase in vanadia loading. Moreover, a decrease of the volume fraction
of the tetragonal zirconia phase was observed with increase in vanadia loading.
No reflection from tetragonal zirconia was observed for five theoretical vana-
dia layers and above.
Raman spectroscopy
Raman spectra of the zirconia support and of prepared catalysts are shown
in Fig. 1 together with the spectra of crystalline V
2
0
S
and ZrV
2
0
7
The spec-
trumof the monoclinic zirconia sample has several strong andsharp bands and
agrees with that observed by other investigators for the same polymorph [22].
The most prominent bands are at 638, 615, 476, 381, 346, 334, 190 and 178
em:", and the very weak features at 147 and 268 cm"" can be from the tetra-
gonal polymorph [44]. Of the reference compounds, V20S has sharp and in-
tense bands at 995,701,284 and 145 cm-
1
, of which the latter band is most
intense. The ZrV207 sample has a band at 990 em-1 with weak shoulders on
both sides and a predominant band at 778 cm-1. For the latter compound,
bands at the same wavenumbers were observed in an earlier investigation [45].
Also, the bands observed for V20S are all in agreement with those reported in
a vibrational study of crystalline V20S [46].
Considering the spectra of the vanadia catalysts, it is seen that the bands of
zirconia decrease in intensity upon deposition of vanadia. Beside the bands
from the support there are a broad band at 987 em-1 and weak features at 880
and 821 cm-
1
in the spectrum for VZr-0.6. For VZr-1.2and VZr-2.6the broad
band is shifted to 1022and 1027cm-1, respectively, and the latter sample gives
additional distinct bands from crystalline V
2
0
S
at 995,701,286 and 145cm-
1

Along with the bands of V20S and zirconia, a strong Raman band at 777cm-
1
is apparent in the spectra of the VZr-6.0 and VZr-9.4 samples. The presence
of this band together with an observed broadeningof the band at 995cm-1 and
the appearance of shoulders at 1024 and 953 cm"' clearly points to the fact
that ZrV207 has formed. Only some weak bands are present in the spectrum of
the VZr-19.4 sample, and they all are from V
2
0
S
'
The Raman technique when applied to supported vanadia catalysts has been
found sensitive to surface structure and coverage of the support [47]. There-
fore, for quantification, the bands at 145 and 778 cm-1 were selected as char-
acteristic for V
2
0
S
and ZrV
2
0
7
, respectively, while the two bands at 615 and
638 em-1 were chosen for the support. For the supported catalysts, the inte-
grated intensities of these bands relative to the intensities of the same bands
in the pure oxides were calculated and are plotted in Fig. 2 as a function of the
vanadium oxide content. A strong decrease in intensity of the zirconia bands
is observed with increase in loading up to a theoretical monolayer. At higher
58
1
0
.
1
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72
i lx
h lx
1200 1000 800 600 400 200
Wavenumber (em-
1
)
Fig. 1. Raman spectra ofBupported catalysts and of reference samples. (a) Zr02; (b) VZr-O.6; (c)
VZr-1.2; (d) VZr-2.6; (e) VZr-6.0; (f) VZr-9.4; (g) VZr-19.4; (h) ZrV
207
; and (i) V
205
Factors
by which the Raman intensity has been expanded are inserted.
loadings, this decrease is more modest. Both V205 and ZrV207 are present at
intermediate loadings. However, at the highest vanadia loading investigated,
only V205 bands were present but with low intensity. The variation of the
difference in balance to 100%, which is also plotted in the figure, indicates at
low loading a predominance at the surface for monolayer species, giving Ra-
man bands in the 1030-820 cm-1 region. At higher loading, beside some V205
and ZrV207 the surface seems covered with an amorphous phase.
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72 59
100


80
iii
c
III

60
c
'" E
'"
40
II:
III


20
a;
II:
o
o 10 20 30 40
Vanadium OXide Content (wt.o V205)
Fig. 2. Relative intensity (compared with pure phases) of some Raman bands characteristic for
Zr02. ZrV
207
and V
205
plotted as a function of the vanadium oxide content of the supported
catalysts. (0) Zr02 bands at 638 and 615 cm-1 integrated within 672-584 em-1; (.) V205band
at 145cm- 1 integratedwithin 162-113 em-1; and (D) ZrV207band at 778cm-1 integratedwithin
724-845 cm"". (e) Difference in balance to 100%.
Infrared spectroscopy
DRIFT spectra of supported catalysts, support, V
2
0
S
and ZrV
2
0
7
are given
in Fig. 3. Two vibrational regions are shown, the 1200-750 cm-
I
region and
the region around 2000 cm-1. The latter region gives information about the
overtones of vanadyl stretching fundamentals around 1000cm-1.
The zirconia support has no band in the two regions. V20S has a strong band
at 1020 cm" ' and a weak band at 982 cm-
I
that are from the symmetric and
asymmetric v(V=O) vibration, respectively, while the band at 845 em>' is a
V-0-Vstretching mode [46] . The overtones of the vanadyl fundamentals are
at 2020 and 1971cm-
1
ZrV
2
0
7
has a broad band centred at 859 cm-
I
Also, a
poorly resolvedbut strong band is at 972 cm-1 surrounded by two weaker bands
at 1023 and 931 cm-
1
Two more bands appear at 1115 and 1064 cm-I, and
they were present neither in the corresponding Raman spectrum, cf. Fig. 1, nor
in any catalyst DRIFT spectrum. It was shown that the appearance of the
latter two bands was related to the particle size of ZrV207' since further grind-
ing of the pure ZrV207 sample down to a particle size around 2/-lm caused 70%
reduction of their intensities, while the intensities of the other bands slightly
increased. At higher wavenumbers, ZrV
2
0
7
has a band at 1979 cm-
1
and a
shoulder is around 1940cm-
1
The spectrum of ZrV
2
0
7
in Fig. 3 was recorded
after grinding.
Considering the spectra of the supported catalysts, VZr-0.6 gives a band at
993 cm" ', and for VZr-1.2 an additional shoulder around 1020cm" ' is appar-
ent. The corresponding first overtones are around 2030 and 1980 cm-1. Pos-
sibly, these bands are from monolayer species interacting with the zirconia.
60
M. Sanati et al. / Appl. Gata/. A 106 (1993) 51-72
I'
I \
/q
20L
h
J ~
"
e (\
.2
/ ~
u
c
"
~
J
d
~
c
~
b
~
a
10.02
lOx
2100 2000 1900
i I \ ~
0\;
h
~
,
g
I ~
~
f
,
/
e?
~
..:
3x
c
,
/
J
~ ~ /
.
a
,
110
4x /
I
, , ,
1100 1000 900 800
WAVENUMBER (cm'
1
)
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72 61
Beside the band at 993 cm-1, two bands at 1020and 845 cm-1 from crystalline
V205 are seen in the spectrum of VZr-2.6. At higher vanadia loadings, both
crystalline V205 and ZrV207 are formed. The spectrum of VZr-6.0 has bands
from V205 at 1020 and 845 cm-1, and from ZrV207 a strong band at 982 cm-
1
and a weak broad band around 936 cm-1 are present. Formation of ZrV207 is
further supported considering the spectral features around 1980 cm-1. In the
spectrum of VZr-9.4, the bands from crystalline V205 are more dominant, but
the band at 982 cm-1 and the shoulder around 1920 cm-1 are from ZrV207'
For the VZr-19.4 sample the spectrum shows in both spectral regions strong
6
Par no. 2
10
27%U
73% 2r
6
Par. no. 1
91% U
U 9% Zr
u
BOX spo a lysis of
par teres
2r
Fig. 4. Secondary electron image of individual grains of the VZr-9.4 catalyst, overlaid with energy
dispersive X-ray maps of zirconiumand vanadium. Amarked difference in composition is obvious
from the spectra shown for two selected particles.
Fig. 3. DRIFT spectra of catalysts and reference compounds. Left: 2150-1850 cm" ': right: 1200-
750 cm" '. (a ) z-o; (b) VZr-0.6; (c) VZr-1.2; (d) VZr-2.6; (e) VZr-6.0; (f) VZr-9.4; (g) VZr-
19.4; (h) ZrV
2
0
7
; and (i ) V
2
0
6
Factors by which the KM-function has been expanded are inserted.
62 M. Sanatietal. I Appl. Catal.A 106(1993)51-72
bands from crystalline V
205
. However, also new bands at 1041 and 1002 cm"
have appeared. These bands were not apparent in the corresponding Raman
spectrum, in which only weak bands from V205 were seen (cf. Fig. 1). This
difference can be due to the Raman technique compared with DRIFT spec-
troscopy is more surface sensitive.
Scanning electron microscopy and energy dispersive X-ray analysis
EDX vanadium maps of the pure Zr02 support did not show any significant
differences from simultaneously recorded maps of background areas without
any characteristic peaks, using windows of the same width. Quantification of
point analyses of random Zr0
2
particles showed the noise level for vanadium
to correspond to less than 1 atom-%. No contaminants were present above that
level.
EDX analyses ofthe VZr-2.6 sample showed the vanadium to be evenly dis-
tributed. The vanadiumatompercentages, V/ (V+Zr) '100, ranged from 16to
34%, to be compared with the nominal value of 11.4%. Since the structures
rich in vanadia form at the surface of the support, the composition measured
will vary with particle size and surface area. The probed depth with 20 keV
primary electrons is approximately 1/lID in this type of material [48). In a few
cases, very high zirconium contents were observed, mostly for grains that were
very smooth in appearance. These probablywere large single crystals of zirconia.
Fig. 5. SEM images of (a) : particle no. 1, and (b) : particle no. 2 from Fig. 4 at higher magnifica -
tion. The vanadium rich particle has a coarser, polycrystalline appearance. The bright spot in (b)
marks the site for spot analysis.
M. Sanati et al. / Appl. Cata!. A 106 (1993) 51-72 63
Mapping of elements in the VZr-9.4 sample, on the other hand, showed a
clear segregation into vanadium- and zirconium-rich areas. Fig. 4 shows the
secondary electron image overlaid with the zirconium map (lighter grey), which
in turn is overlaid with the map of vanadium (white). EDX spectra of the two
different grains marked 1and 2 are inserted in the figure. Quantification showed
in some cases a vanadium content greatly exceeding that of ZrV
207
(67 at.-
%), indicating that these grains possibly were V205' Comparison of the ap-
pearance of grains rich in either zirconium or vanadium, showed the former to
have smooth and clean surfaces, whereas the latter were irregular and deco-
rated with smaller particles, see Fig. 5.
High-resolution electron microscopy
HREM images of the pure support showed the particles, which according to
SEM micrographs were in the 1 JCffi range, to consist of aggregates of a large
number of smaller zirconia particles in the range 30-50 nm. Fig. 6 shows the
50nm
Fig. 6. Transmission electron micrograph of pure zirconia support, showing a grain consisting of
smaller particles with diameters around 30-50 nm. The inset electron diffraction pattern shows
t he particles to be crystalline and with random orientation.
64 M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72
Fig. 7. High-resolution low-dose image of pure zirconia support. The crystal lattices of different
grains have different orientation, but partly sintered together. Note that there is no amorphous
overlayer on the crystal surfaces.
smaller particles to be crystalline and with random orientation, giving ring
patterns in electron diffraction mode (see inset). The graininess of the rings
indicates that the crystallites are at least 20 nm in diameter. Diffraction peaks
derived from line traces of rotated electron diffraction patterns showed good
agreement with experimental and literature X-ray powder patterns. At higher
magnification, Fig. 7, it becomes apparent that the particles to a large extent
are sintered together, making it increasingly difficult to distinguish discrete
particles.
Comparing images of the pure support with those of the VZr-2.6 and VZr-
9.4 samples showed only subtle changes in appearance after deposition of van-
adia. An image of the VZr-9.4 sample is displayed in Fig. 8, and it shows that
the zirconia particles are covered with an amorphous overlayer. This layer is
not due to growth of contamination in the microscope, which is clear compar-
ing with the image of the pure support in Fig. 7 (see also the Experimental
section) . Generally, the surfaces of the crystals showed increased roughness at
the outermost layers.
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72 65
Fig. 8. High-resolution micrograph of the VZr-9.4 catalyst, showing an amorphous surface struc-
ture that is not present on the pure support. The thickness ofthe overlayer (arrowed) is typically
1.6nm.
Activity measurements
The performance of the prepared samples in toluene oxidation is shown in
Fig. 9. Zirconia has low activity and is not selective for partial oxidation, only
for combustion to mainly carbon dioxide. For the vanadia loaded samples, both
total activity expressed per unit surface area and the activity for formation of
benzaldehyde increase with increase in vanadia content up to about eight the-
oretical vanadia layers. With further increase in vanadia content, both activ-
ities decrease. The selectivity for benzaldehyde formation is 7%over VZr-0.6,
and it is around 23%over the other supported samples. For the ZrV207 sample,
the selectivity for aldehyde formation is 15%, while it is much higher, 45% (out
of scale in Fig. 9), over crystalline V20S' The total reaction rates over ZrV207
and V
2
0
S
are lower than for VZr-6.0 and VZr-9.4. Over V
2
0
S
, the rate for al-
dehyde formation is close to the values for VZr-6.0and VZr-9.4, but over ZrV207
it is lower.
Fig. 10 shows the rate per unit surface area and the selectivity for formation
of benzonitrile as a function of vanadia loading. Under the conditions of am-
moxidation, the zirconia support has no activity for formation of benzonitrile
66
12
":"
c
10
E

E
8
"0
E
3- 6
Q)
iii
a:
4
c
2
U
2
'"
Q)
a:
0
0
M. Sanati et al. / Appl. Catol. A 106 (1993) 51-72
25
...

c
20
2
iii
E
0
"
15
u.
Q)
'0
>-
s:

10
Q)
'0
<t
5
.E


u
0
Q)
Qj
5 10 15 20
C/)
Vanadia Loading (No. of layers)
Fig. 9. Total reaction rate (0), rate for formation of benzaldehyde (0) and selectivity for alde-
hyde formation (b,,) plotted as a function of the vanadia loading. Filled and half-filled symbols
showdata for ZrV207and V205,respectively. The reaction temperature was 370C, and the partial
pressures of toluene and oxygen were 0.77 and 11.4 kPa, respectively. For V
2
0
5
, the selectivity for
formation of benzaldehyde was 45% (out of scale).
100
e

c
"
80
2
iii
E
0
60
u,
.S!
.,S
40
Z

.E
.-

.
20
U
Q)
Qj
0
C/)
20 o 5 10 15
Vanadia loading (No. of layers)
Fig; 10. Total reaction rate (0), rate for formation ofbenzonitrile (0) and selectivity for nitrile
formation (b,,) plotted as a function of the vanadia loading. Filled and half-filled symbols show
data for ZrV
2
0
7
and V
2
0
5
, respectively. The reaction temperature was 370C, and the partial
pressures of toluene, oxygen and ammonia were 0.77,11.4 and 2.85 kPa, respectively.
6
c 5
E
'"
'E 4
"0
1-3
Q)
iii
a: 2
c
o
<3 1
'"
Q)
a:
and low activity for carbon dioxide formation. It is seen that both the total
activity and the activity for nitrile formation with increase in vanadia content
pass through maxima at a vanadia loading corresponding to about eight van-
adia layers. The selectivity for nitrile formation increases rapidly with increase
in vanadia loading and attains a value of about 80% for one theoretical layer
and above, which is approximately equal to the values that were obtained for
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72 67
ZrV207and V20S. Both the total rate and the rate for benzonitrile formation
are lower for ZrV207and V20Sthan for most of the supported samples.
DISCUSSION
Dispersion and vanadia species in the monolayer region
Figs. 1 and 2 show a strong decrease with increase in vanadia loading of the
Raman intensity of the support bands. For the VZr-0.6 sample about 30% of
the intensity of the support bands remains, compared with the pure support,
and the corresponding values for samples with loadings of a theoretical mono-
layer and above are less than 5%. These data reveal the dispersion ofvanadia
on the zirconia to be high, a fact that is also supported by the EDX analysis of
the VZr-2.6 sample, showing the vanadium to be evenly distributed. Recently,
for supported vanadia catalysts, we found a correlation between the V/ (sup-
port metal) atom ratio determinedby X-rayphotoelectron spectroscopy (XPS)
andthe intensity of the Raman bands of the supports Ti0
2,
Zr02' Hf0
2,
Nb
20S'
and Ta20s [49], confirmingthat FT-Ramanspectroscopy is a surface sensitive
technique. In the same work [49], for a monoclinic zirconia with low surface
area (4.3 m
2
/ g), we observed almost no further increase in the V/Zr atomratio
determined by XPS when the vanadia loading exceeded half a theoretical
monolayer, indicating a surface coverage with vanadia of about 50%. For the
same samples, the Raman intensity of the support bands compared to those of
the pure support was around 35%. Thus, the corresponding value for the pres-
ent samples 5%) is considerably less and indicates for these almost com-
plete coverage.
The Raman and infrared spectra in Figs. 1 and 3 for low vanadia loading
(VZr-0.6, VZr-1.2 and VZr-2.6) show several bands in the 820-1030 cm-
1
re-
gion that can be attributed dispersed vanadia species. In previous investiga-
tions have bands in the same spectral region been reported for vanadia on
zirconia at ambient conditions, and they have been assigned to different types
of vanadyl and vanadate structures [32,40,50-52]. In viewof the assignments
given by Deo and Wachs [51] for vanadia on various supports and which were
made considering both Raman and SlVNMR data, the Raman bands for VZr-
0.6 at 821 and 880 cm-
1
are assigned to orthovanadate (V0
4
) and pyrovana-
date (V207) species, respectively. The Raman band at 987 em-1 and the cor-
responding infrared bands at 993 cm-
1
(fundamental) and 1980 cm:' (first
overtone) are assigned to octahedrally coordinated vanadium oxide species in
a decavanadate-like (V
lO02S)
environment. The Raman and infrared bands
around 1020-1027 cm-
1
for VZr-1.2 and VZr-2.6, according to Deo and Wachs
[52] are from a dehydrated vanadate species possessing one terminal V=O
bond and three bridging V-O-support bonds. In a study of vanadia on Ti0
2,
Cristiani et a1. [11] assigned Raman and infrared bands at ca. 990 and 1035
68
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72
cm-1 to a mono-oxo species in hydrated and dehydrated form, respectively.
However, considering the high coverage of vanadia on zr0
2
that was obtained
in the present investigation, the band around 990 cm-1 more likely is from a
bidimensional polyvanadate structure of decavanadate type. This conclusion
is supported by calculations [53], showing that complete coverage with mon-
omeric species does not correspond to more than 30% of that expected for a
complete bidimensional polyvanadate layer. In our previous report on vanadia
supported on a zirconia with a low surface area [49], Raman and infrared
bands were both at ca. 1015 em-1. This might be due to another distribution
of Zr02 surface planes, a proposal that can explain the low dispersion of van-
adia on those samples.
Vanadia structures at high loading
At vanadia loadings above the theoretical monolayer region, the XRD (Ta-
ble 2), Raman (Fig. 1) and infrared (Fig. 3) analyses show formation of crys-
talline V205 and ZrV207' According to Fig. 2, the intensities of the Raman
bands of crystalline V205 and ZrV207 in the supported samples are low and do
not balance the corresponding decrease of the support bands. This observation
points to formation of at least one additional structure at high vanadia loading.
High-resolution electron micrographs reveal this structure to be an amorphous
overlayer, see Fig. 8. The formation of amorphous vanadia agrees with the
Raman spectrum for the VZr-19.4 sample in Fig. 1, where no band except very
weak V205 bands is apparent at high magnification. Interestingly, in our pre-
vious microscopy investigation [15], amorphous vanadia was discovered to
form on the surfaces ofthe anatase and rutile forms of Ti0
2
as well.
Formation on supports of some crystalline V205 has frequently been ob-
served [9,15,49,51], especially at high vanadia loading. Concemingthe present
samples, additionally, formation of some crystalline ZrV207is evident for VZr-
6.0 and VZr-9.4. For low loadings, however, there is no indication of ZrV
207
formation, an observation that agrees with data presented in other works where
calcination in air has been performed in the 40D-500C range [40,45,49,51].
Roozeboom et a1. [45], however, observed formation of ZrV207 after calcina-
tion of impregnated samples at 600C, in agreement with the minimum tem-
perature required starting from a mechanical mixture of V205 and Zr02 [35].
The calcination temperature used in the present work, 500C, is lower. It is
also considerably lower than the 710C, that in the present work (see the Ex-
perimental section) was found to be required for preparing ZrV207 from pow-
ders of V
205
and Zr02' Obviously, the formation of ZrV
207
in the catalyst
samples must be due to intimate contact between vanadium and zirconia, re-
sulting from the impregnation with vanadium salt.
There is no univocal evidence for ZrV207 formation on VZr-19.4. Some XRD
lines might be from ZrV
207
(Table 2), but there are no reflections at ca. 3.9
M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72 69
and ca. 3.6A, which typically are intense for crystalline ZrV207 [42]. Further-
more, there are no ZrV207bands in the corresponding Raman spectrum (spec-
trum g, Fig. 1). Possibly, these observations can be explained by formation of
either some poorly crystalline ZrV207 or a new unidentified structure, being
responsible for the infrared bands observed at 1041 and 1002cm-
1
(spectrum
g, Fig. 3).
Catalytic performance
The data in Figs. 9 and 10 show that the Zr02 support has low activity in
toluene oxidation and ammoxidation, respectively, and is selective for com-
bustion. Upon deposition of vanadia on the Zr02 surface, both the total activity
and the selectivity for formation of partial oxidation products increase, show-
ing that vanadia species are active for the formations of benzaldehyde and
benzonitrile. According to the vibrational spectra for VZr-0.6 and VZr-1.2 in
Figs. 1 and 3, the two predominant vanadia species for loadings in the mono-
layer range are an octahedrally coordinated species of decavanadate type and
a tetrahedral mono-oxo species. Although it has been demonstrated that the
former species upon complete dehydration converts to the latter species [51,52],
most likely the octahedral vanadia structure is dominant under reaction con-
ditions since water is formed as a product. For both toluene oxidation and
ammoxidation the total rate for VZr-1.2 expressed per unit surface area of
catalyst is almost twice the rate for VZr-0.6, verifying the dispersion of vanadia
to be high in the whole range up to a theoretical monolayer. Thus, the role of
Zr02 is not limited to that of being a physical support for vanadia, a conclusion
that is supportedby the presented infrared, Raman and EDX results. That the
selectivity for aldehyde formation for VZr-0.6is considerably less than for VZr-
1.2, in spite of the fact that the latter sample is more active per unit surface
area, can be explained by further reaction of benzaldehyde to carbon oxides on
the uncovered support surface. A similar finding has been reported for both
toluene and o-xylene oxidation on vanadia/titania catalysts [54,55]. For tol-
uene ammoxidation the selectivity for nitrile formation is comparatively high
for both VZr-0.6 and VZr-1.2, showing, compared with benzaldehyde, a greater
stability of benzonitrile toward degradation on the bare Zr02 surface.
Assuming the zirconia surface to be almost completely covered at a loading
of one theoretical monolayer, an assumption that is supported by the Raman
data in Fig. 2 and the EDX measurements, allows the conclusion that the sur-
face of multilayer amorphous vanadia with a thickness of 4-8 layers (Fig. 8)
is the most active structure for both toluene oxidation and ammoxidation, see
Figs. 9 and 10. In this regard it is worth noting that also for butene oxidation,
a correlation between activity and the concentration of amorphous vanadia on
Zr02 has been reported [32]. In disagreement with an earlier conclusion [39],
no evidence from activity measurements or microscopy was obtained in the
70 M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72
present work for an oriented growth of V205 on the zr0
2
surface, leading to a
preferential exposure of the (010) V205 surface.
The activity data show the total rate to be higher in oxidation than in am-
moxidation, while a higher rate and selectivity for partial oxidation are ob-
served in ammoxidation. These differences can be explained by considering a
previous analysis of the structure-activity relationships in oxidation of alkyl-
aromatics over metal oxides [56], according to which structurally undercoor-
dinated oxygen participates in degradation and combustion pathways. On
crystalline V205' the (010) plane that is active for o-xyleneoxidation to phthalic
anhydride [57] exposes exclusively oxygens that are highly coordinated [3,5].
The V
2
0
5
(100) and (001) planes, on the other hand, expose strongly under-
coordinated oxygens [3,5] and are active for combustion [57]. Due to the lay-
ered structure of V205 [41], the crystals are anisotropic exposing large (010)
faces and small (100) and (001) faces. Therefore it is not surprising that crys-
talline V
2
0
5
until now is one of the most selective (ca. 45%) catalysts known
for benzaldehyde formation from toluene [15,54]. The much lower selectivity
obtained for the zirconia supported vanadia samples, therefore, possibly is due
to the fact that they expose undercoordinated oxygen species. For the design
of a supported catalyst that is more selective for benzaldehyde formation, a
tuning of the support-active phase interactions is needed to have an optimum
distribution and character of adsorption centres and oxygen species. This can
possibly be achieved using a support with additives. So far titania seems most
promising as support for vanadia in this reaction [15]. The higher rate and
selectivity of vanadia on zirconia for benzonitrile formation in ammoxidation,
compared with those for benzaldehyde formation in oxidation, can be seen as
a consequence of competitive adsorption of oxygen and ammonia, resulting in
a decrease of the concentration of undercoordinated oxygen species active for
combustion [58]. The explanation is supported by the fact that the selectivity
for benzonitrile formation generally is high (ca. 80%) irrespective of support
[15] andwhether the vanadia structure is V205' ZrV207, of decavanadate-type,
or amorphous.
ACKNOWLEDGEMENT
Ms. C. Song is gratefully acknowledged for assisting with EDX measure-
ments. Financial support from the National Board for Industrial and Techni-
cal Development (NUTEK) and the Swedish Research Council for Engineer-
ing Sciences (TFR) is also acknowledged.
M. Sanati et al. / Appl. Catal: A 106 (J993) 51-72
REFERENCES
71
1 A. Andersson and S.L.T. Andersson, in R.K. Grasselli and J.F. Brazdil (Editors), Solid State
Chemistry in Catalysis (ACS Symposium Series, Vol. 279), American Chemical Society,
Washington, D.C., 1985, p. 121.
2 J.K. Dixon andJ.E. Longfield, in P.H. Emmett (Editor), Catalysis, Vol. 7, Reinhold, New
York, 1960, Ch. 3.
3 A. Andersson, J.-O. Bovin and P. Walter, J. Catal., 98 (1986) 204.
4 J.-E. Germain and R. Laugier, Bull. Soc. Chim. Fr., 2 (1972) 541.
5 A. Bielanski and J. Haber, Oxygenin Catalysis, Marcel Dekker, NewYork, 1991, Ch. 8.
6 B. Delmon, J. Mol. Catal., 59 (1990) 179.
7 A. Vejuxand P. Courtine, J. Solid State Chem.,63 (1986) 179.
8 M. Gasior, J. Haber and T. Machej, Appl. Catal., 33 (1987) 1.
9 G. Hausinger, H. Schmelz and H. Knbzinger, Appl, Catal., 39 (1988) 267.
10 G.C. Bond and S. Flamerz, Appl. Catal., 46 (1989) 89.
11 C. Cristiani, P. Forzatti and G. Busca,J. Catal., 116 (1989) 586.
12 H. Eckert and I.E. Wachs, J. Phys. Chem., 93 (1989) 6796.
13 T. Machej, P. Ruiz and B. Delmon, J. Chem. Soc. Faraday Trans., 86 (1990) 731.
14 G. Centi, E. Giamello, D. Pinelli and F. TrifIro, J. Catal., 130 (1991) 220.
15 M. Sanati, L.R. Wallenberg, A. Andersson, S. Jansen and Y. Tu, J. Catal., 132 (1991) 128.
16 V. Nikolov, D. Klissurski and A. Anastasov, Catal. Rev. Sci. Eng., 33 (1991) 319.
17 B. Jonson, B. Rebenstorf, R. Larsson, S.L.T. Andersson and S.T. Lundin, J. Chem. Soc.
Faraday Trans. 1, 82 (1986) 767.
18 B. Jonson, B. Rebenstorf, R. Larsson and S.L.T. Andersson, J. Chem. Soc. Faraday Trans.
1,84 (1988) 1897.
19 B. Jonson, Spectroscopic Studies on Heterogeneous Oxidation Catalysts, Ph.D. Thesis
(LUNKDL/NKOO-1016/001-069/1988), University of Lund, Lund, 1988.
20 R. Burch, Zirconium in Catalysis - Its Uses and Potential, Magnesium Elektron Ltd.,
Twickenham (UK), 1985.
21 Phase Diagrams for Ceramists, 1975 Supplement, The American Ceramic Society, Colum-
bus, Ohio, 1975, p. 77.
22 P.D.L. Mercera, J.G. van Ommen, E.B.M. Doesburg, A.J. Burggraaf and J.R.H. Ross, Appl.
Catal.,57 (1990) 127.
23 P.D.L. Mercera, J.G. van Ommen, E.B.M. Doesburg, A.J. Burggraaf and J.R.H. Ross, Appl.
Catal.,71 (1991) 363.
24 P.D.L. Mercera, J.G. van Ommen, E.B.M. Doesburg, A.J. Burggraaf and J.R.H. Ross, Appl.
Catal., 78 (1991) 79.
25 L. Bruce and J.F. Mathews, Appl. Catal., 4 (1982) 353.
26 T. Iizuka, Y. Tanaka and K. Tanabe, J. Catal., 76 (1982) 1.
27 B. Denise and R.P.A. Sneeden, Appl. Catal., 28 (1986) 235.
28 Y. Amenomiya, Appl. Catal., 30 (1987) 57.
29 G.J.J. Bartley and R. Burch, Appl. Catal., 43 (1988) 141.
30 Y. Matsuoka, M. Niwa and Y. Murakami, J. Phys. Chem., 94 (1990) 1477.
31 T. Ono, H. Kamisuki, H. Hisashi and H. Miyata, J. Catal., 116 (1989) 303.
32 H. Miyata, M. Kohno, T. Ono, T. Ohno and F. Hatayama, J. Chem. Soc. Faraday Trans. 1,
85 (1989) 3663.
33 F. Hatayama, T. Ohno, T. Yoshida, T. Ono and H. Miyata, React. Kinet. Catal. Lett., 44
(1991) 451.
34 F. Hatayama, T. Ohno, T. Muruoka and H. Miyata, React. Kinet. Catal. Lett., 45 (1991)
265.
72 M. Sanati et al. / Appl. Catal. A 106 (1993) 51-72
35 F. Roozeboom, T. Fransen, P. Mars and P.J. Gellings, Z. Anorg.AlIg. Chem.,449 (1979) 25.
36 J. Haber, A. Kozlowska and R. Kozlowski, in M.J. Phillips and M. Ternan (Editors), Proc,
9th Int. Congr. on Catalysis, Calgary, 1988, The Chemical Institute of Canada, Ottawa, 1988,
p.l481.
37 K,V.R. Chary, K, Narsimha, K,S. Rams Reo, B. Rema Raoand P. Kanta Rao,J. Mol. Catal.,
58 (1990) L13.
38 K,V.R. Chary, B. Rama Reo and V.S. Subrahmanyam, Appl. Catal., 74 (1991) 1.
39 M. Niwa, Y. Matsuoka and Y. Murakami, J. Phys. Chem., 93 (1989) 3660.
40 U. Scharf, M. Schraml-Marth, A. Wokaun and A. Baiker, J. Chem. Soc. Faraday Trans., 87
(1991) 3299.
41 H.G. Bachmann, F.R. Ahmed and W.H. Barnes, Z. Kristallogr., 115 (1961) 110.
42 Powder Diffraction File, JCPDS International Centre for Diffraction Data, Swarthmore,
Pennsylvania, 1991.
43 A. Andersson and S. Hansen, J. Catal., 114 (1988) 332.
44 R. Srinivasan, M.B. Harris, S.F. Simpson, R.J. De Angelis and B.H. Davis, J Mater. Res., 3
(1988) 787.
45 F. Roozeboom, M.C. Mittelmeijer-Hazeleger, J.A. Moulijn, J. Medema, V.H.J. de Beer and
P.J. Gellings, J. Phys. Chem., 84 (1980) 2783.
46 L. Abello, E. Husson, Y. Repelin and G. Lucazeau, Spectrochim. Acta, 39A (1983) 641.
47 M. Sanati, A. Andersson and L.R. Wallenberg, in L. Guczi, F. Solymosi and P. Tetany! (Ed-
itors) , New Frontiers in Catalysis, Proc. 10th Int. Congo on Catalysis, Budapest, 19-24 July
1992 (Studies in Surface Science and Catalysis, Vol. 75), Elsevier, Amsterdam, 1993,p. 1755.
48 D.E. Newbury and R.L. Myklebust, Ultramicroscopy, 3 (1978) 391.
49 J. Huuhtanen, M. Sanati, A. Andersson and S.L.T. Andersson, Appl. Catal. A, 97 (1993)
197.
50 H. Miyata, K, Fujii, T. Ono, Y. Kubokawa, T. Ohno andF. Hatayama, J. Chern. Soc. Faraday
Trans. 1,83 (1987) 675.
51 G. Deo and I.E. Wachs, J. Phys. Chem., 95 (1991) 5889.
52 G. Deo and I.E. Wachs,J. Catal., 129 (1991) 307.
53 G.C. Bond and J.C. Vedrine, Catal, Today, to be published.
54 L.R. Wallenberg, M. Sanati and A. Andersson, J. Catal., 126 (1990) 246.
55 I.E. Wachs, R.Y. Saleh, S.S. Chan and C.C. Chersich, Appl. Catal., 15 (1985) 339.
56 A. Andersson, S. Hansen and M. Sanati, in R.K, Grasselli and A.W. Sleight (Editors), Struc-
ture-Activityand Selectivity Relationships in Heterogeneous Catalysis (Studies in Surface
Science and Catalysis, Vol. 67), Elsevier, Amsterdam, 1991, p. 43.
57 M. Gasior and T. Machej,J. Catal.,83 (1983)472.
58 M. Sanati and A. Andersson, Ind. Eng. Chem. Res., 30 (1991) 320.

You might also like