Oxidation of methanol to formaldehyde over a series of Fe1-xAlx-V-oxide catalysts

Article abstract of DOI:10.1016/j.jcat.2008.06.029

A series of triclinic Fe_1-xAl_xVO₄ phases with 0 ≤ x ≤ 1 were prepared and used in the oxidation of methanol to formaldehyde. The activity measurements revealed that both the activity and especially the selectivity to formaldehyde increased with time of operation for at least 16 h, indicating restructuring of the catalysts. Characterisation of the catalysts with XRD, XANES, and electron microscopy after use in methanol oxidation showed that the stability of the bulk phases improved when Al was substituted for Fe in the structure. XRD and XANES of the used FeVO₄ showed that it partly transformed into a cation-vacant spinel-type Fe_1.5V_1.5O₄ phase, whereas the AlVO₄ phase showed no change in the bulk structure. HRTEM imaging of used catalysts confirmed that structural changes, including in the surface, occurred during catalysis. Quantitative surface analysis by XPS of the catalysts before and after use in methanol oxidation revealed no significant change in the metal composition, in good agreement with the corresponding bulk values, except for a lower Fe value. Steady-state activity data showed a modest increase in specific activity with the Al content, whereas the selectivity to formaldehyde was about 90% for all samples at high methanol conversion. The similar catalytic behaviour of the vanadates irrespective of the differences in the bulk structure indicates that the surface structure differed from the bulk structure. Compared with pure vanadia, the vanadates had lower activity per V atom and slightly greater selectivity to formaldehyde. Consequently, for methanol oxidation, the role of Al and Fe on the catalyst surface can be described as that of a spacer, decreasing the surface concentration of active V sites and the number of less selective V{single bond}O{single bond}V ensembles.

Full text of DOI:10.1016/j.jcat.2008.06.029

Journal of Catalysis 258 (2008) 345–355
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Oxidation of methanol to formaldehyde over a series of Fe1−xAlx-V-oxide catalysts
Robert Häggblad a, Jakob B. Wagner b, Staffan Hansen b, Arne Andersson a,∗
a
Department of Chemical Engineering, Lund University, Chemical Center, P.O. Box 124, SE-221 00 Lund, Sweden
Division of Polymer and Materials Chemistry, Department of Chemistry, Lund University, Chemical Center, PO Box 124, SE-221 00 Lund, Sweden
b
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of triclinic Fe1−xAlxVO4 phases with 0 ꢀ x ꢀ 1 were prepared and used in the oxidation of
methanol to formaldehyde. The activity measurements revealed that both the activity and especially the
selectivity to formaldehyde increased with time of operation for at least 16 h, indicating restructuring
of the catalysts. Characterisation of the catalysts with XRD, XANES, and electron microscopy after use
in methanol oxidation showed that the stability of the bulk phases improved when Al was substituted
for Fe in the structure. XRD and XANES of the used FeVO4 showed that it partly transformed into a
cation-vacant spinel-type Fe1.5 V1.5 O4 phase, whereas the AlVO4 phase showed no change in the bulk
structure. HRTEM imaging of used catalysts confirmed that structural changes, including in the surface,
occurred during catalysis. Quantitative surface analysis by XPS of the catalysts before and after use in
methanol oxidation revealed no significant change in the metal composition, in good agreement with
the corresponding bulk values, except for a lower Fe value. Steady-state activity data showed a modest
increase in specific activity with the Al content, whereas the selectivity to formaldehyde was about 90%
for all samples at high methanol conversion. The similar catalytic behaviour of the vanadates irrespective
of the differences in the bulk structure indicates that the surface structure differed from the bulk
structure. Compared with pure vanadia, the vanadates had lower activity per V atom and slightly greater
selectivity to formaldehyde. Consequently, for methanol oxidation, the role of Al and Fe on the catalyst
surface can be described as that of a spacer, decreasing the surface concentration of active V sites and
the number of less selective V–O–V ensembles.
Received 19 March 2008
Revised 26 June 2008
Accepted 27 June 2008
Available online 30 July 2008
Keywords:
Selective oxidation
Methanol
Formaldehyde
Fe1−xAlxVO4 catalysts
Fe–V–O spinel-type phases
XRD
XANES
XPS
SEM
HRTEM
© 2008 Elsevier Inc. All rights reserved.
1. Introduction
sure drop [7], necessitating replacing the catalyst after 1–2 years
of operation [4]. Consequently, alternative Mo-free catalysts are of
interest [9], but only on the condition that the alternatives are al-
most as selective as the iron molybdate catalyst at high methanol
conversion, because methanol makes the largest contribution to
the production cost.
Today, two competing processes, the silver and oxide processes,
are used for the production of formaldehyde from methanol and
air. The silver process uses a silver catalyst and operates with a
methanol-rich feed (36–40%), whereas the oxide process uses an
iron molybdate catalyst and a methanol-lean (∼8.5%) feed. Which
process is preferred is determined by the operating and capital
costs, such as raw material and energy costs, as well as by plant
size, product end use, and the type of operation [1–3].
Over the last decade, the oxide process has gained market share
due to its higher selectivity to formaldehyde, around 93%. How-
ever, a major drawback of the present MoO3/Fe2(MoO4)3 catalyst
is that it is deactivated as some of the molybdenum reacts with
methanol to form volatile species at the reaction conditions, result-
ing in depletion of molybdenum from the catalyst in the reaction
zone [4–8]. As a consequence, both the activity and the selectiv-
ity to formaldehyde decrease with time, and recondensation of the
sublimed molybdenum in colder regions causes an increased pres-
In the area of alternative catalysts for methanol oxidation, sev-
eral studies have been reported on vanadia-based catalysts includ-
ing pure vanadia, mixed oxides, and supported vanadia [10–13]. In
particular, vanadates with Ni, Fe, Co, Mg, Cr, Mn, Al, Ag, Cu, and
Zn have been found to have selectivities >90% to formaldehyde at
high methanol conversion [9,14–17]. According to Wachs et al. [16],
the vanadium in bulk metal vanadates is not volatile in methanol
oxidation, although XPS analysis before and after use of the sam-
ples in methanol oxidation indicated some structural changes in
the surface and near surface layers during catalysis. Besides the
surface structure, such changes may involve the formation of new
phases; in oxidation catalysis on bulk metal oxides, not only the
surface, but also the catalyst bulk structure may change under the
influence of the catalytic reaction until steady state is reached [18].
Consequently, in the present work, we investigated in more detail
the performance and the stability of a Fe1−xAlxVO4 series of cata-
lysts with x varying from 0 to 1. The catalysts were characterised
Corresponding author.
E-mail address: arne.andersson@chemeng.lth.se (A. Andersson).
*
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.06.029

346
R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
◦
Table 1
5 and 80 degrees 2θ in steps of 0.1 (2.0 s/step). The unit cell pa-
rameters of the five Fe1−xAlxVO4 catalysts were determined both
before and after the catalysts’ use in methanol oxidation. Crystal
structure data were taken from Ref. [20] for FeVO4, from Ref. [21]
for AlVO4, and from Ref. [22] for the spinel-type structure observed
in the used Al-free sample.
Notation, specific surface area and phase composition of prepared catalysts
Catalyst notation
Specific surface area (m2/g)
Phase composition
Fe1Al0
15.2
13.2
22.1
9.3
Triclinic (Fe1Al0)VO4
Fe0.75Al0.25
Fe0.50Al0.50
Fe0.25Al0.75
Fe0Al1
Triclinic (Fe0.75Al0.25)VO4
Triclinic (Fe0.50Al0.50)VO4
Triclinic (Fe0.25Al0.75)VO4
Triclinic (Fe0Al1)VO4
XPS analysis was performed on a PHI 5500 XPS instrument us-
ing monochromatic Al Kα radiation. Powder samples were placed
10.8
on
a conducting and sticky tape. To minimise the effects of
sample charging, the aluminium-containing samples were charge-
neutralised by electrons. PC-ACCESS and MuliPak 6.1A software
were used to evaluate the data, and the quantifications were made
using a Shirley function for the background. The C 1s peak was
used as an energy reference and was set to a binding energy of
285.0 eV. Data used for quantification of the element were col-
lected using a sufficient amount of sample to exclude any signal
from the underlying carbon film. However, due to the charge accu-
mulation observed for the Al-containing preparations, resulting in
peak broadening and asymmetry, the data used to determine the
valence states by peak fitting had to be collected separately using
a small amount of very well-ground sample placed on a conduct-
ing film. In addition, the sample was continuously neutralised with
electrons to compensate for the electrons ejected during the mea-
surements. Because the oxidation number of Al is surely trivalent,
the charge compensation was adjusted to give symmetric Al 2p
and Al 2s peaks, indicating that no charge effects were present. Us-
ing this technique, analysis on the Fe0.50Al0.50 sample was carried
out without peak broadening and asymmetry. For the Fe0Al1 sam-
ple, the charge accumulation problem was too extensive to give
reliable data, however.
by XRD, XANES, XPS, SEM, and HRTEM both before and after use
in methanol oxidation to produce formaldehyde.
2. Experimental
2.1. Catalyst preparation
Fe1−xAlxVO4 catalysts with x = 0, 0.25, 0.50, 0.75, and 1 were
prepared by precipitation from a homogeneous water solution con-
taining dissolved V and Fe and/or Al. The homogeneous solution
was prepared from two separate water solutions, a 0.04 M NH4VO3
(Merck) solution and a 0.5 M solution of Fe(NO3)3·9H2O (Merck)
and/or Al(NO3)3·9H2O (Sigma–Aldrich). The two well-stirred solu-
tions were mixed together and homogenised by lowering the pH
to 1.0 by adding 3 M HNO3. A solid precipitate then was obtained
when the pH was rapidly raised to 4.0 by the addition of 3 M
NH3. Particle coarsening was carried out to stimulate particle re-
◦
covery [19] by heating the turbid solution for 2 h at 50 C under
stirring. The particles were separated by centrifugation (3000 rpm
for 3 min) and then washed three times with water, acetone, and
◦
water. Finally, the samples were dried for 16 h at 80 C and then
◦
calcined for 6 h at 580–620 C depending on the phase purity. Ta-
The XANES measurements were performed at the I811 beam-
line at Maxlab (Lund University) using a Si(111) double-crystal
monochromator and three ionisation chambers. Spectra of the Fe
K- and V K-edges were recorded in transmission mode using Fe
and V metal foils as energy references. The sample was placed in
between the first and the second ionisation chambers, and the ref-
erence was placed before the third chamber. To obtain an optimal
absorption signal, the sample was diluted with boron nitride.
Scanning electron microscopy (SEM) images were acquired with
a Jeol 6700F FEG SEM. The acceleration voltage was set to 5 kV;
the working distance, to 7 mm. For high-resolution transmission
electron microscopy (HRTEM) imaging, the samples were studied
in a Jeol 3000F FEG TEM operated at 300 kV. EELS was performed
using an attached Gatan GIF2002 image filter. The samples were
dry-dispersed on standard Cu grids with holey carbon films for
TEM inspection.
ble 1 summarises the notation, specific surface area, and phase
composition of the catalysts.
2.2. Activity measurements
The prepared catalysts were tested for methanol oxidation to
produce formaldehyde in a stainless steel reactor operating at
isothermal conditions and atmospheric pressure. To obtain isother-
mal conditions, the reactor was embedded in an aluminium block
placed in a tube furnace. For the measurements, the catalyst sam-
ple was ground into fine powder and pressed into tablets, which
were then crushed and sieved to particles of 0.250–0.425 mm
diameter. The reactor was loaded with the desired amount of cat-
alyst diluted three times with quartz particles. The catalyst was
heated up to the reaction temperature in a flow of 10 ml/min
◦
O2 and 84 ml/min N2. When the reaction temperature 350 C was
reached, a gaseous methanol flow of 6 ml/min was added to the
flow of oxygen and nitrogen. The product flow composition was
analysed every 20 min for 1 h and then once an hour for the next
15 h.
3. Results
3.1. Catalytic behaviour
Methanol, formaldehyde (FA), dimethyl ether (DME), methyl for-
mate (MF), dimethoxymethane (DMM), and CO2 were analysed
with an online gas chromatograph equipped with a Haysep C col-
umn and both flame ionisation and thermal conductivity detectors.
CO was analysed online with a Rosemount Binos 100 IR analyser.
The catalytic performance of the pure triclinic Fe1−xAlxVO4
phases listed in Table 1 was evaluated for methanol oxidation.
Fig. 1 shows the methanol conversion and selectivity to formalde-
hyde given as
a function of time on stream for the Fe1Al0,
Fe0.50Al0.50, and Fe0Al1 catalysts. The degree of conversion was
high, with a between-sample variation of 95–98%. After 16 h on
stream, in all three preparations, the difference between the initial
and the final value of the methanol conversion was small; how-
ever, a noticeable increase in the selectivity to formaldehyde with
time could be seen. All three preparations had similar features: a
rapid initial increase in selectivity, followed by an asymptotic fea-
ture reaching a selectivity of 88–91% after 16 h on stream.
2.3. Catalyst characterisation
The specific surface areas of the catalysts were measured with
a Micromeritics Flowsorb 2300 instrument. The single-point BET
method was used with nitrogen adsorption at liquid nitrogen tem-
perature and subsequent desorption at room temperature. All sam-
◦
ples were degassed at 150 C for 24 h before analysis.
Fig. 2 presents activity and selectivity data for the five prepared
Fe1−xAlxVO4 catalysts. All catalysts showed rather similar activ-
ity, with a trend toward activity increasing up to about a factor
Powder X-ray diffraction (XRD) analysis was performed on a
Seifert XRD 3000 TT diffractometer using Ni-filtered Cu Kα radi-
ation and a rotating sample holder. Data were collected between

R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
347
of two when Fe was replaced with Al. The product selectivities
shown in Fig. 2 are for high methanol conversion (95–98%). For
formaldehyde, the selectivity was about 88–91% and the differ-
ences among the samples were small. For the main byproducts
COx and DME, the selectivities were in the ranges 9–12% and 0.3–
1.6%, respectively. No significant differences among the catalysts
were detected. Two possible weak trends were increasing selec-
tivity to DME but decreasing selectivity to COx with increasing Al
content. Compared with the corresponding data shown in Fig. 2
for a commercial-type MoO3/Fe2(MoO4)3 catalyst, the vanadates
were more active but slightly less selective to formaldehyde and
DME, because they produced more carbon oxides. But compared
with the data for pure vanadia, the vanadates were less active
and somewhat more selective for formaldehyde formation. On the
pure vanadia, in contrast to the other samples, a small amount of
methyl formate also was formed.
3.2. X-ray diffraction
The catalyst were characterised by XRD both as synthesised and
after use in methanol oxidation, with the samples designated fresh
and used, respectively. The XRD patterns of the freshly prepared
samples are displayed in Fig. 3. All samples show the characteris-
tic features of a triclinic (P − 1) type of phase, isostructural with
FeVO4 [20,23] and AlVO4 [21,24]. The lattice parameters of the
pure Fe1−xAlxVO4 phases were refined; the values obtained are
given in Table 2. The a-, b-, and c-axes all show a linear depen-
dence with the degree of substitution, confirming the successful
preparation of a solid solution series of Fe1−xAlxVO4 catalysts.
The XRD patterns of the used samples are displayed in Fig. 4.
Comparing this figure with Fig. 3 clearly shows that the used
Fe1Al0 consists of the original triclinic FeVO4 phase to only a slight
extent. Instead, a new diffraction pattern is notable, with broad
and intense peaks at 30.4, 35.5, 43.7, 57.7, and 63.1 degrees 2θ.
As the subtraction diffractogram in Fig. 5 shows, the new phase
formed is a spinel-type phase, similar to Fe3O4 (magnetite) and γ -
Fe2O3 (maghemite). The six strongest diffraction peaks are at 30.2,
35.6, 43.1, 53.5, 57.1, and 62.6 degrees 2θ for Fe3O4 and at 30.3,
35.8, 43.4, 53.9, 57.4 and 63.0 degrees 2θ for γ -Fe2O3. The broad-
ness of the peaks for the used Fe1Al0 indicates that the crystallites
were small in the spinel-type phase formed during catalysis.
For the used Fe0.50Al0.50 and Fe0Al1 catalysts, the diffraction
patterns in Fig. 4 do not reveal the appearance of any new peaks
Fig. 1. Methanol conversion (2) and selectivity to formaldehyde (1) as measured
for the Fe1Al0, Fe0.50Al0.50 and Fe0Al1 catalysts as a function of time on stream at
◦
350 C and high methanol conversion (>95%). The inlet gas composition is 6 vol%
methanol and 10 vol% oxygen in nitrogen.
◦
Fig. 2. Catalytic data as measured after 16 h on stream at 350 C for a series of Fe1−xAlxVO4 catalysts with x = 0, 0.25, 0.50, 0.75 and 1. Corresponding data for a commercial
type MoO3/Fe2(MoO4) catalyst and pure vanadia are included for comparison. For obtaining the specific activities measurements were made at low methanol conversion,
admitting calculation of the specific rates for the defined inlet gas composition. The selectivities to formaldehyde, COx and dimethyl ether (DME) are for high methanol
conversions in the range 95–99%, which is the region of interest in full scale production. Inlet gas composition: 6 vol% methanol and 10 vol% oxygen in nitrogen.

348
R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
Fig. 3. XRD patterns of freshly prepared Fe1−xAlxVO4 catalysts, showing a series of
Fig. 4. XRD patterns for the Fe1Al0, Fe0.50Al0.50 and Fe0Al1 catalysts recorded after
use of the samples for 16 h in methanol oxidation. A comparison with Fig. 3 reveals
that the used Fe1Al0 sample contains only a small amount of the original phase, and
relatively more of the original phase is preserved when increasing the Al content in
the catalyst.
isostructural phases from FeVO4 to AlVO4.
Table 2
Refined unit cell parameters for the triclinic Fe1−xAlxVO4 phases in fresh and used
catalysts
◦
◦
◦
Samplesa
a (Å)
b (Å)
c (Å)
α ( )
β ( )
γ ( )
the positions were 5477.4, 5481.0, and 5481.3 eV, respectively, for
the corresponding used catalysts. Thus, the energy shifts between
the fresh and the used Fe1Al0, Fe0.50Al0.50, and Fe0Al1 were −4.1,
−0.5 and −0.2 eV, respectively, indicating reduction of vanadium
in the bulk. Moreover, the magnitude of the shift shows that the
degree of reduction decreased with increasing Al content in the
catalyst.
The Fe K-edge spectra of fresh and used Fe1Al0 and Fe0.50Al0.50
catalysts are displayed in Fig. 7a. The edge positions for the fresh
Fe1Al0 and Fe0.50Al0.50 samples were 7127.7 and 7127.6 eV, respec-
tively. For the used catalysts, the corresponding positions shifted
toward lower energy and appeared at 7124.9 and 7127.1 eV, re-
spectively. According to the XRD patterns shown in Figs. 4 and 5,
the major phase in the used Fe1Al0 is of spinel-type similar to both
Fe3O4 and γ -Fe2O3. Fig. 7b compares the Fe K-edge spectra for the
used Fe1Al0 catalyst with those for Fe3O4 and γ -Fe2O3. It shows
that the edge position for the used Fe1Al0 was very close to that
for γ -Fe2O3 with Fe3+ only, whereas the edge for Fe3O4 with both
Fe2+ and Fe3+ was further shifted toward lower energy. This find-
ing indicates that Fe can be trivalent in the bulk of the used Fe1Al0
and Fe0.50Al0.50 catalysts. Consequently, the shift in edge position
between the fresh and the used catalysts (Fig. 7a) apparently does
not have to be due to reduction of Fe, but rather to differences in
the coordination of Fe.
Fresh catalysts
Fe1Al0
Fe0.75Al0.25
Fe0.50Al0.50
Fe0.25Al0.75
Fe0Al1
6.708
6.664
6.631
6.576
6.536
8.051
7.985
7.929
7.829
7.758
9.334
9.291
9.259
9.192
9.129
96.7
96.5
96.4
96.3
96.2
106.6
106.7
106.9
107.0
107.3
101.5
101.5
101.4
101.4
101.4
Used catalysts
Fe1Al0
Fe0.75Al0.25
Fe0.50Al0.50
Fe0.25Al0.75
Fe0Al1
Unit cell parameters resemble those of the fresh sampleb
NDc
6.62
7.924
9.234
96.4
106.8
101.4
101.4
NDc
6.537
7.756
9.128
96.2
107.3
a
For notations, see Table 1.
b
c
The sample consisted of mainly a cubic spinel phase with a = 8.33 Å.
Not determined.
compared with the patterns of the corresponding fresh samples
shown in Fig. 3. This finding is confirmed by the subtraction pat-
terns shown in Fig. 5.
3.3. XANES
To determine the valences of the bulk elements, XANES mea-
surements were performed on the fresh and used Fe1Al0,
Fe0.50Al0.50, and Fe0Al1 catalysts. Fig. 6 shows the V K-edge spec-
tra of the catalysts. The main edge position for the fresh Fe1Al0,
Fe0.50Al0.50, and Fe0Al1 was 5481.5 eV for all preparations, whereas

R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
349
Fig. 6. Normalised V K-edge XANES spectra for fresh and used Fe1Al0, Fe0.50Al0.50
and Fe0Al1 catalysts.
Fig. 5. XRD subtraction patterns for the used Fe1Al0, Fe0.50Al0.50 and Fe0Al1 cata-
lysts obtained by subtracting the patterns of the fresh catalysts (Fig. 3) from the
corresponding patterns of the used catalysts (Fig. 5). The diffractograms of the
spinel-type phases Fe3O4 and γ -Fe2O3 are shown for comparison.
3.4. XPS
XPS measurements were performed to determine the elemen-
tal composition and the oxidation states of the elements in the
surface region. The surface compositions of fresh and used Fe1Al0,
Fe0.50Al0.50, and Fe0Al1 are given in Table 3. No significant differ-
ence in the surface compositions can be observed between fresh
and used catalysts. The data show that the concentration of V at
the surface varies from 14.1 to 19.3%, which values are similar to
the value of 16.7% for the bulk. Iron, on the other hand, is signif-
icantly underrepresented at the surface, whereas Al shows similar
values as the bulk.
The Al 2s, Fe 2p3/2, and V 2p3/2 binding energies for fresh
and used catalysts are given in Table 4. The data show that the
binding energy for Al 2s was practically the same for fresh and
used samples, about 119.4 eV. This value is the same as that re-
ported for Al2O3 [25], demonstrating the presence of Al3+ in the
catalysts in all cases. In addition, the measured Fe 2p3/2 binding
energy values were approximately the same for the fresh and used
preparations of Fe1Al0 and Fe0.50Al0.50, varying between 711.3 and
711.6 eV. The values agree with those reported for FeVO4 [26],
Fe2O3 [26,27], and FeOOH [27], verifying that Fe at the surface
of the present catalysts is trivalent. In all cases, a symmetric Fe
2p3/2 peak was observed, revealing no other oxidation state for Fe
Fig. 7. Normalised Fe K-edge XANES spectra for (a) the fresh and the used Fe1Al0
and Fe0.50Al0.50 catalysts, and (b) Fe3O4, γ -Fe2O3 and the used Fe1Al0 catalyst.
ence of more than one oxidation state. Table 5 gives the binding
energy, peak width (FWHM), and relative amounts of the differ-
ent oxidation states of V as determined by resolution of the V
2p3/2 peaks for fresh and used Fe1Al0 and Fe0.50Al0.50 shown in
than Fe3+
.
Opposed to the Al 2s and Fe 2p3/2 peaks, the V 2p3/2 peaks for
the used samples were not always symmetric, indicating the pres-

350
R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
Table 3
Bulk compositions of the catalysts and the surface compositions as determined by XPS
Catalyst
Bulk composition (at%)
XPS fresh catalyst (at%)
XPS used catalyst (at%)
V
Fe
Al
O
V
Fe
Al
O
V
Fe
Al
O
Fe1Al0
Fe0.50Al0.50
Fe0Al1
16.7
16.7
16.7
16.7
8.3
0
0
8.3
16.7
66.7
66.7
66.7
18.6
16.8
14.1
12.2
5.6
–
–
10
16.2
69.2
67.7
69.7
19.3
15.8
15.7
9.5
5.2
–
–
10.2
16.5
71.2
68.8
67.8
Table 4
Catalyst binding energies determined by XPS
Catalyst
Binding energy of fresh catalyst (eV)
Binding energy of used catalyst (eV)
Fe 2p3/2
O 1s
V 2p3/2
Al 2s
Fe 2p3/2
O 1s
V 2p3/2
Al 2s
Fe1Al0
Fe0.50Al0.50
Fe0Al1
711.3
711.6
–
530.4
530.7
530.3
517.2
517.7
517.2
–
711.5
711.6
–
530.6
530.6
530.9
517.5
517.5
517.8
–
119.4
119.6
119.5
119.2
Table 5
Data for vanadium in the catalysts as obtained after resolution of the V 2p3/2 peak into its components by peak fitting
Sample
V5+
V4+
V3+
FWHM
(eV)
BEa
(eV)
V5+/Vtot
(%)
FWHM
(eV)
BEa
(eV)
V4+/Vtot
(%)
FWHM
(eV)
BEa
(eV)
V3+/Vtot
(%)
Fresh catalysts
Fe1Al0
Fe0.50Al0.50
1.12
1.20
517.2
517.7
100.0
100.0
Used catalysts
Fe1Al0
Fe0.50Al0.50
1.15
1.30
517.6
517.5
70.7
90.9
1.20
1.30
516.5
516.3
24.5
9.1
1.20
515.2
4.7
a
BE: binding energy.
Fig. 8. No data are given for higher Al content, because in this
case the analysis was strongly affected by charging effects. For
the fresh samples, the spectra in Fig. 8 show symmetric V 2p3/2
peaks with a binding energy of about 517.4 eV (Table 5), which
can be assigned to V5+ [28,29]. The fact that the peaks are sym-
metric indicates the presence of only one oxidation state. The V
2p3/2 spectra for the samples used in methanol oxidation show
a shoulder from reduced V on the low-energy side of the peak
maximum. The shoulder is more pronounced in the spectrum of
the used Fe1Al0 than in that for the used Fe0.50Al0.50. Peak fit-
ting was performed as shown in Fig. 8. The V 2p3/2 peak of
the used Fe1Al0 can be resolved into the three components, V5+
(517.6 eV), V4+ (516.5 eV), and V3+ (515.2 eV) [28,29], whereas
the spectrum for the used Fe0.50Al0.50 shows only two oxidation
states, V5+ (517.5 eV) and V4+ (516.3 eV). The quantitative data
given in Table 5 indicate more severe reduction of Fe1Al0 (∼70%
V5+) than of Fe0.50Al0.50 (∼90% V5+) after use in methanol oxida-
tion.
3.6. TEM
In general, the samples were very sensitive to the electrons in
the TEM. To avoid beam damage, the electron dose was minimised,
so that the samples were not altered during image acquisition.
Fig. 10a shows a high-resolution image of the fresh Fe1Al0 sample.
The 100-nm particles consisted of FeVO4 single crystals embedded
in a 5- to 10-nm-thick layer showing lattice fringes, indicating that
the layer was crystalline. Because of the low electron dose used, it
was not possible to obtain reliable quantitative analytical data from
the crystalline shell. Qualitatively, however, EELS analyses showed
that both the core crystal and shell structure consisted of iron,
vanadium, and oxygen. After catalytic testing, according to SEM,
the Fe1Al0 sample had the same overall morphology as the fresh
sample (see Figs. 9a and 9b). Fig. 10b presents a TEM image of a
corner of a 100-nm particle, showing that the particle comprised
smaller 5–10 nm crystalline particles and most likely voids, some-
what connected in a crystalline matrix, as indicated by the lattice
fringes. EELS of the V L2,3 and Fe L2,3 ionisation edges revealed a
shift toward lower energy loss compared with the fresh catalyst, in
agreement with the corresponding XANES spectra shown in Figs. 6
and 7.
3.5. SEM
The HRTEM images of the Fe0Al1 samples in Fig. 11 show 10-
to 50-nm crystals in the form of large agglomerates. The smaller
crystal size compared with the iron-containing samples gives the
sample a smoother, less porous appearance. Fig. 11a shows the
crystals of the fresh sample embedded in a thin (1 nm), appar-
ently amorphous layer. After catalytic testing, the sample had a
thicker (2–4 nm), amorphous layer (Fig. 11b). Furthermore, both
the fresh and the used catalysts had 5-nm crystals at the surface
of the larger crystals. It was not possible to determine the exact
composition of the small particles, because of the low EELS signal.
Elemental mapping by means of energy-filtered TEM showed ho-
mogeneous vanadium and iron signals; however, the lack of spatial
resolution because of the low signal intensity made it difficult to
reliably distinguish the 5-nm crystals in the elemental map.
Fig. 9 presents characteristic SEM images of fresh and used
Fe1Al0 samples. The fresh Fe1Al0 sample consisted of approx-
imately 100 nm large and well-separated spherical particles
(Fig. 9a). Although XRD findings demonstrated bulk transformation
of the sample (Figs. 4 and 5), the SEM images of the sample af-
ter catalytic testing revealed the same overall structure of 100-nm
spherical particles (Fig. 9b).
Compared with the Al-free sample, the fresh Fe0.50Al0.50 sample
had slightly larger (∼150 nm) and less distinct particles, whereas
the fresh Fe0Al1 sample consisted of particles agglomerated into a
less porous substance. The corresponding used samples were very
similar to the respective fresh samples, but the used Fe0.50Al0.50
had slightly smaller particles.

R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
351
Fig. 8. XPS spectra of the V 2p region for (a) fresh Fe1Al0, (b) used Fe1Al0, (c) fresh Fe0.50Al0.50, and (d) used Fe0.50Al0.50, showing for the V 2p3/2 peak also the contributions
of different oxidation states as obtained by peak fitting.
Compared with Fe0Al1, the Fe0.50Al0.50 sample showed similar
features on HRTEM, with small (5–10 nm) crystalline particles at
the surface of larger (150 nm) single crystal particles, although the
particles were more distinct and less agglomerated.
The catalytic data in Fig. 1 show changes in the activity and
especially the selectivity to formaldehyde extending over several
hours of operation, clearly indicating that structural transformation
of the catalyst bulk and/or surface occurred in methanol oxidation.
Comparing the XRD patterns for the freshly prepared Fe1Al0 in
Fig. 3 with those for the used catalyst in Figs. 4 and 5 shows that
a partial transformation of the bulk occurred from the triclinic-
type FeVO4 structure in the fresh sample to a spinel-type struc-
ture in the used sample. Although SEM revealed no changes in
either particle size or morphology (Fig. 9), imaging of the cata-
lyst by HRTEM confirmed a major change in the bulk structure
due to its use in methanol oxidation (Fig. 10). Elemental map-
ping in HRTEM showed that the new phase formed, of a spinel
type and similar in structure to Fe3O4 and γ -Fe2O3 (Fig. 5), con-
tained not only iron, but also vanadium. Such types of spinels
have been prepared previously [31–37]. The structural formula for
VxFe3−xO4 spinels with 0 ꢀ x ꢀ 2 has been expressed in more de-
4. Discussion
4.1. Bulk phases
The XRD patterns shown in Fig. 3 and the variations of the unit
cell data given in Table 2 definitely confirm the successful prepa-
ration of a Fe1−xAlxVO4 series of triclinic phases from FeVO4 to
AlVO4. In the structure with V in tetrahedral coordination, the va-
lences of the cations were V5+, Fe3+, and Al3+ [20,21,23,24], in
agreement with our XANES spectra given in Figs. 6 and 7 showing
the V and Fe K-edges for the fresh catalysts. The graph in Fig. 12,
showing the V K-edge positions for a number of compounds with
vanadium in different oxidation state and coordination [30] to-
gether with the data for our fresh and used catalysts, confirms the
presence of V5+ in the fresh catalysts.
tail as (Fe2+Fe13−+α)A(Fe2+ Fe3+ V3x+)BO2− with α = x/2 and tetra-
α
1−α 1−α
4
hedral (A) and octahedral (B) sites [33]. Thermal studies of this
type of spinel phase have demonstrated that oxidation of Fe2+ and
V3+ to form Fe3+ and V5+ via V4+ is possible with preservation

352
R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
Fig. 9. SEM images of (a) fresh Fe1Al0 and (b) used Fe1Al0.
of the basic spinel-type structure [33,34,36,38]. The oxidation of
the cations was accompanied by the appearance of a correspond-
ing number of cation vacancies in the structure to maintain the
electroneutrality.
The V K-edge XANES spectrum for the used Fe1Al0 in Fig. 6
shows a shift of about 4 eV toward lower energy compared with
that for the fresh sample. According to the correlation shown
in Fig. 12, a shift of this magnitude corresponds to an average
valence of +3.9 for V in this sample. Because the used Fe1Al0
catalyst consists of both the triclinic FeVO4 phase and a spinel-
type phase, subtracting the contribution from the triclinic phase
to the XANES spectrum would be possible if the phase compo-
sition of the sample were known. Then, from the edge position
in the resulting spectrum, it would be possible to determine the
oxidation state for the vanadium in the spinel structure. An es-
timate of the content of the triclinic phase in the used Fe1Al0
sample can be obtained by comparing the diffractogram of the
used sample with that of the phase pure fresh sample with re-
gard to some typical XRD peaks from the triclinic phase. Using
the intensities of the two strongest peaks from the triclinic phase
Fig. 10. HRTEM images of the Fe1Al0 catalyst (a) before and (b) after use in
methanol oxidation for 16 h.
5476.2 eV calculated for ESpinel corresponds to an average V va-
lence of +3.5. Using the same approach for the Fe K-edge spectra,
from the estimated edge position for the spinel-type phase, an av-
erage Fe valence of +2.84 is obtained by interpolating between the
edge positions for Fe3O4 and γ -Fe2O3 with an average Fe valence
of +2.67 and +3, respectively.
For a VxFe3−xO4 spinel with x = 1.5, corresponding to the
atomic ratio V/Fe = 1 in Fe1Al0, a stoichiometric phase without
any vacancies can be formulated as V(III)1.5 Fe(II)1.0 Fe(III)0.5O4. Con-
sidering the average valences determined for V and Fe in the used
Fe1Al0 sample, the composition of the oxidised spinel phase in this
sample will be V(III)0.63V(IV)0.63Fe(II)0.20Fe(III)1.06 10.48O4, where 1
denotes a cation vacancy. It has been found that the vacancies can
be located at both octahedral and tetrahedral sites in the spinel-
type structure, whereas the reduced V are in octahedral positions
[34]. The latter finding agrees with the fact that the intensity of
the V K-pre-edge peak is lower in the XANES spectrum of the used
Fe1Al0 than in the spectrum for the fresh sample (Fig. 6), consid-
ering that the intensity of the pre-edge peak for V is expected to
decrease with increasing symmetry from tetrahedral to octahedral
coordination [30].
◦
◦
(25.0 and 27.8 2θ) for the comparison, the calculated intensity
ratios, Iused/Ifresh, indicate that about 20–25% of triclinic FeVO4
was present in the used Fe1Al0 catalyst. From this value, the V
K-edge position of the spinel phase (ESpinel) can be estimated, as-
suming that the measured edge position (EExperimental) is equal to
the sum of the contributions from the two constituent phases ac-
cording to EExperimental = yTriclinic·ETriclinic + (1 − yTriclinic)·ESpinel
,
where yTriclinic is the estimated content of FeVO4 and ETriclinic is
the edge position as measured for the fresh sample consisting of
the pure FeVO4. According to the graph in Fig. 12, the value of

R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
353
Fig. 12. The V K-edge positions for fresh and used Fe1−xAlxVO4 catalysts and ref-
erence compounds with in various type of coordination. Freshly prepared (1)
V
Fe1Al0, (2) Fe0.50Al0.50 and (3) Fe0Al1; and the corresponding used catalysts (4a)
Fe1Al0, (4b) Fe1Al0 after subtraction of the contribution of the triclinic phase, (5)
Fe0.50Al0.50 and (6) Fe0Al1 being submitted to methanol oxidation for 16 hours.
Data for the reference compounds are from Ref. [27] and are marked (A) VO, (B)
KAlV2Si3O10 (OH)2, (C) V2O3, (D) V4O7, (E) V2O4, (F) NH4VO3, (G) CrVO4 and (H)
Pb5(VO4)3Cl, where vanadium is tetrahedrally coordinated in A–E and octrahedrally
in F–H.
from the triclinic phase. Moreover, HRTEM revealed no formation
of new structures, although the SEM images showed that the par-
ticles in the catalyst were slightly smaller after the catalyst was
used in methanol oxidation. That some structural changes in the
catalyst occurred during methanol oxidation is verified by the mi-
nor shift in the position of the V K-edge (Fig. 6), the magnitude
of which, according to the correlation shown in Fig. 12, indicates
an average valence of +4.9 for the V in the used Fe0.50Al0.50 cata-
lyst. In parallel, some shift in the Fe K-edge position also occurred,
as shown in Fig. 7. Together, these findings suggest that a small
amount of a spinel-type V1.5 Fe1.5 O4 was formed during catalysis.
The data for the fresh and the used Fe0Al1 catalyst in Table 2
show the same unit cell dimensions for both samples, indicating
that no bulk transformation of the triclinic AlVO4 phase occurred
from its use methanol oxidation. Further support for the stability
of the AlVO4 bulk in methanol oxidation is provided by the XRD
patterns for the used Fe0Al1 in Figs. 4 and 5, the XANES spectra
in Fig. 6, and the relationship illustrated in Fig. 12, showing exclu-
sively V5+ in the used sample. But the HRTEM images in Fig. 11
reveal some changes at the surfaces, with the thickness of the
apparently amorphous surface layer increasing from 1 nm in the
fresh catalyst up to 2–4 nm in the used catalyst.
Fig. 11. HRTEM images of the Fe0Al1 catalyst (a) before and (b) after use in methanol
oxidation for 16 h. An apparently amorphous surface layer, which has become
thicker after use of the sample in methanol oxidation, is indicated in the figure.
Regarding the used Fe0.50Al0.50 catalyst, the data for the unit
cell axes in Table 2 indicate that the triclinic phase in this sample
is richer in Al compared with the fresh sample. The unit cell has
become slightly smaller, in agreement with Al3+ being a smaller
cation than Fe3+ [39]. From the variations of the unit cell axes
in Table 2, the composition of the triclinic phase in the used
Fe0.50Al0.50 is estimated to be approximately Fe0.48Al0.52VO4, in-
dicating the concurrent formation of either FeVO4, a spinel-type
VxFe3−xO4 phase, or vanadium and iron oxides. From the stoi-
chiometry of the Al-rich triclinic phase, it follows that the total
amount of the minor phase or phases can be only about 3 wt%.
Usually such a small amount cannot be detected on XRD; the XRD
patterns of the used Fe0.50Al0.50 in Figs. 4 and 5 show peaks only
4.2. Surface composition and catalytic performance
Despite the observation that depending on the Al content, the
Fe1−xAlxV-oxide series of catalysts exhibited widely varying de-
grees of bulk transformation after use in methanol oxidation, all of
the samples exhibited very similar activation behaviour during the

354
R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
first 16 h on stream, as shown in Fig. 1. That observation clearly
indicates that the surface structure, which determines the catalytic
properties, changes over time until steady state is reached. In-
deed, Wachs et al. [40] observed similar changes with time of the
catalytic performance in methanol oxidation starting from physi-
cal mixtures of V2O5 and TiO2 or MoO3 and TiO2. Using in situ
Raman characterisation, the investigators showed that reaction-
induced spreading of vanadia and molybdena occurs, forming 2-
dimensional overlayers on the support. Moreover, experimental re-
sults have been reported suggesting that the surface of bulk metal
vanadates [16] and molybdates [41–43] may consist only of sur-
face vanadium oxide and molybdenum oxide species, respectively.
Considering the fact that there is little variation in activity and se-
lectivity between the Fe1−xAlxV-oxide samples (Fig. 2), it can be
proposed that the activation behaviour is due to reaction-induced
surface enrichment of vanadium from the bulk. But the quantita-
tive XPS data in Table 3 give no support for the activation being
primarily related to enrichment of vanadium at the surface. Al-
though the V/Fe ratio for Fe1Al0 increased from 1.5 to 2.0, the
V/(Fe + Al) atomic ratios for Fe0.50Al0.50 and Fe0Al1 were almost
the same before and after the catalysts were used in methanol ox-
idation. For Fe0.50Al0.50, the ratio changed from 1.1 to 1.0 and for
Fe0Al1, the ratio was 0.9 both before and after use; that is, both
samples showed V/(Fe + Al) ratios very close to the value of 1.0
for the bulk. This indicates that the observed activation behaviour
is due to adjustment of the surface structure under the influence
of the reactants. In catalysis, the surface structure is determined
not only by the bulk structure to which it is attached, but also by
the surface interacting with the components being adsorbed from
the gas phase, the reaction conditions, and the associated kinetic
parameters [18]. Moreover, the finding of an activation period of
at least 16 h suggests that the restructuring involves not only the
true surface, but also possibly a few layers beneath. Some sup-
port for this suggestion is given by the HRTEM images of the bulk
stable iron-free Fe0Al1 catalyst in Fig. 11, showing a thicker, appar-
ently amorphous layer in the sample used in methanol oxidation
compared with the fresh sample.
The modest increase with the Al content of the activity per
surface area unit (Fig. 2) definitely was not due to any parallel in-
crease at the surface of the V concentration, which, according to
the XPS results, was higher for Fe1Al0 than for the Al-containing
samples (Table 3). Rather, the variation in activity with the catalyst
composition can be related to differences induced by the bonding
of V to either Fe or Al in the form of V–O–Fe or V–O–Al, respec-
tively. XPS demonstrated a reduction of V in the samples used in
methanol oxidation, as shown in Table 5 and Fig. 8. Moreover, the
degree of reduction diminished with increasing Al content of the
catalyst, in line with the observed reduction of the bulk. There-
fore, the possibility that the reduction detected by XPS may be
from a few layers beneath the surface and that the vanadium in
the true surface layer may be predominately pentavalent irrespec-
tive of the bulk composition of the catalyst cannot be excluded,
considering that the catalyst activities were of similar magnitude
and, moreover, that reoxidation usually is not considered rate-
limiting in methanol oxidation [44,45]. Concerning the differences
between the catalysts with regard to the bulk reduction behaviour,
it should be noted that the reduction of the bulk is an initial tran-
sient process not necessarily indicating any major difference in the
reoxidation rate of the catalyst surfaces under steady-state condi-
tions.
suming a monolayer of metal oxide with approximately 12 μmol
cations/m2 surface area, with half of the cations on the vanadates
being vanadium, the activities per V atom calculated from the data
given in Fig. 2 were 9.9, 3.9, and 2.2 mol methanol/(mol V s) for
pure vanadia, Fe0Al1, and Fe1Al0, respectively. The finding that the
activities per V atom decreased in the same order as the Sander-
son electronegativity values increased, V (1.39) > Al (1.71) > Fe
(2.20), in line with previous findings [46,47], can be attributed to
the electron density on the bridging oxygen decreasing in the or-
der V–O–V > V–O–Al > V–O–Fe. A greater electron density should
facilitate abstraction of the hydrogen atom from the slightly acidic
hydroxyl group of methanol, thereby increasing the surface con-
centration of intermediate methoxy species. But the adsorption of
methanol in the form of a methoxy species is known to be an
easy step [42,46], and the methoxy species can even be mobile
at the surface [49], whereas the abstraction of hydrogen from the
adsorbed methoxy species is rate-limiting [44,50]. In fact, experi-
mental results have indicated that increasing the electron density
at the active site improves the rate constant even more than the
adsorption constant [45].
Although there is general consensus in the literature and spec-
troscopic evidence for the formation of V–OCH3
species [45,46,
48], there is still some disagreement about the details of the final
hydrogen abstraction. Weckhuysen and Keller [48] have proposed
that a high electron density on the vanadium facilitates breaking
of the C–H bond, giving formaldehyde and an intermediate V–H
bond. According to recent theoretical calculations [51,52], hydro-
gen abstraction from the methoxy group may involve H transfer to
a vanadyl oxygen atom; however, the literature gives experimen-
tal evidence that a bridging oxygen rather than a vanadyl oxygen
determines the activity [12,48].
The finding that the selectivity to formaldehyde did not vary
much among the vanadate samples and was around 90% for all
of the samples at high methanol conversion is in agreement with
data previously reported for FeVO4 and AlVO4 [9,16]. Our observa-
tion that pure vanadia gave somewhat lower selectivity (87%) is in
agreement with previous results [53]. Consequently, for methanol
oxidation, the role of Fe and Al in the vanadates can be described
as that of a “spacer,” creating isolation of the V sites and thereby
decreasing the number of V–O–V sites. As the data in Fig. 2 in-
dicate, compared with the V–O–Fe and V–O–Al sites, the V–O–V
sites gave slightly more DME and carbon oxides, respectively. Com-
paring the product distributions obtained in methanol oxidation
on bulk metal vanadates and other metal oxides suggests that the
metal vanadates contain only surface vanadium sites [16]. Concern-
ing our Fe1−xAlxV-oxide series of catalysts, neither the quantitative
XPS data in Table 3 nor the activities expressed per V atom give
support for significantly enriching the surfaces with vanadium. The
selectivity data for the vanadates in Fig. 2 show that the selectivity
to formaldehyde was practically the same irrespective of whether
the second cation was Fe or Al, demonstrating that these elements
are rather inert in the surface structure. Although iron oxide pro-
duces predominately carbon dioxide from methanol, it has very
low activity as expressed per surface area unit [43,54]. Moreover,
our comparative measurements on a low-surface area α-alumina
(Norpro, 5 m2/g, XPS Na/Al ratio = 0.05) showed that it was in-
active for methanol, in contrast to high-surface area γ -alumina
producing DME [12]. The trend shown in Fig. 2 toward increasing
selectivity to DME with increasing Al content of the vanadate agree
with the finding of acidic Al sites producing DME from methanol
present on the surface. However, a DME selectivity of <2% for
Fe0Al1 confirms that the Al on the Fe(Al)–V–O catalysts contributes
little to the selectivity.
For supported oxide systems, Wachs et al. [46,47] reported that
the turnover frequency for methanol oxidation decreased with in-
creasing electronegativity of the metal cation of an oxide support,
suggesting that the electron density on the bridging oxygen in
the V–O–support bond, or, alternatively, the electron density on
the vanadium atom [48], determines the turnover frequency. As-

R. Häggblad et al. / Journal of Catalysis 258 (2008) 345–355
355
5. Conclusion
[5] A.P.V. Soares, M.F. Portela, A. Kiennemann, J.M.M. Millet, React. Kinet. Catal.
Lett. 75 (2002) 13.
[6] A.P.V. Soares, M.F. Portela, A. Kiennemann, L. Hilaire, Chem. Eng. Sci. 58 (2003)
1315.
Our findings indicate that catalysts prepared in the form of
phase pure triclinic Fe1−xAlxVO4 phases with 0 ꢀ x ꢀ 1 were ac-
tive and selective for methanol oxidation to produce formaldehyde.
When submitted to methanol oxidation, all of the triclinic phases
initially showed very similar activation behaviour extending for at
least 16 h of operation. During this period, both the conversion
and especially the selectivity to formaldehyde increased with time
on stream. The extended behaviour indicates restructuring of the
surface and the near-surface region to form a surface structure dif-
ferent from that of the bulk. Supporting information provided by
HRTEM imaging showed growth of an apparently amorphous layer
at the surface of the bulk stable AlVO4.
Elemental analysis by XPS showed no significant difference in
the metal composition between the freshly prepared samples and
the corresponding samples used in methanol oxidation. Moreover,
except for some depletion of Fe, the V/(Fe + Al) atomic ratios de-
termined by XPS agreed quite well with the corresponding bulk
ratios.
Compared with an industrial type MoO3/Fe2(MoO4)3 catalyst,
the vanadates were more active per unit surface area and less
selective to formaldehyde at high methanol conversion (∼90% vs
93%). Compared with pure vanadia, the vanadates were less active
per vanadium atom and per unit surface area, but more selective
to formaldehyde (∼90% vs 87%). Substituting Al for Fe gave slightly
improved activity but had no notable effect on the selectivity to
formaldehyde. The activity data indicate that the electron density
on a bridging V–O–M (M = V, Al, Fe) oxygen determined the ac-
tivity, with increasing activity expressed per V atom decreasing
the electronegativity of the M metal. The improved selectivity to
formaldehyde of the vanadates compared with pure vanadia sug-
gests that the Al and Fe in the structure create isolation of V and
decrease the number of less-selective V–O–V sites.
[7] N. Burriesci, F. Garbassi, M. Petrera, G. Petrini, N. Pernicone, in: B. Delmon, G.F.
Froment (Eds.), Catalyst Deactivation, Elsevier, Amsterdam, 1980, pp. 115–126.
[8] B.I. Popov, V.N. Bibin, G.K. Boreskov, Kinet. Katal. (Eng. Transl.) 17 (1976) 322.
[9] I.E. Wachs, L.E. Briand, US Patent 7 193 117 B2, 2007, to Lehigh University.
[10] P. Forzatti, E. Tronconi, A.S. Elmi, G. Busca, Appl. Catal. A 157 (1997) 387.
[11] T. Kim, I.E. Wachs, J. Catal. 255 (2008) 197.
[12] G. Deo, I.E. Wachs, J. Catal. 146 (1994) 323.
[13] S. Lim, G.L. Haller, Appl. Catal. A 188 (1999) 277.
[14] R. Malin´ ski, React. Kinet. Catal. Lett. 5 (1976) 265.
[15] R. Malin´ ski, M. Akimoto, E. Echigoya, J. Catal. 44 (1976) 101.
[16] L.E. Briand, J.-M. Jehng, L. Cornaglia, A.M. Hirt, I.E. Wachs, Catal. Today 78
(2003) 257.
[17] G.V. Isaguliants, I.P. Belomestnykh, Catal. Today 100 (2005) 441.
[18] A. Bielan´ ski, J. Haber, Oxygen in Catalysis, Dekker, New York, 1991.
[19] S. Denis, E. Baudrin, M. Touboul, J.-M. Tarascon, J. Electrochem. Soc. 144 (1997)
4099.
[20] B. Robertson, E. Kostiner, J. Solid State Chem. 4 (1972) 29.
[21] E. Arisi, S.A. Palomares Sánchez, F. Leccabue, B.E. Watts, G. Bocelli, F. Calderón,
G. Calestani, L. Righi, J. Mater. Sci. 39 (2004) 2107.
[22] B.G. Hyde, S. Andersson, Inorganic Crystal Structures, Wiley, New York, 1988.
[23] S. Denis, R. Dedryvère, E. Baudrin, S. Laruelle, M. Touboul, J. Olivier-Fourcade,
J.C. Jumas, J.M. Tarascon, Chem. Mater. 12 (2000) 3733.
[24] E.J. Baran, I.L. Botto, Monatsh. Chem. 108 (1977) 311.
[25] J. Nilsson, A.R. Landa-Cánovas, S. Hansen, A. Andersson, J. Catal. 186 (1999) 442.
[26] E. Baba Ali, J.C. Bernède, A. Barreau, Mater. Chem. Phys. 63 (2000) 208.
[27] K. Asami, K. Hashimoto, Corros. Sci. 17 (1977) 559.
[28] S.L.T. Andersson, S. Järås, J. Catal. 64 (1980) 51.
[29] L. O’Mahony, T. Curtin, D. Zemlyanov, M. Mihov, B.K. Hodnett, J. Catal. 227
(2004) 270.
[30] J. Wong, F.W. Lytle, R.P. Messmer, D.H. Maylotte, Phys. Rev. B 30 (1984) 5596.
[31] M. Wakihara, Y. Shimizu, T. Katsura, J. Solid State Chem. 3 (1971) 478.
[32] D.B. Rogers, R.J. Arnott, A. Wold, J.B. Goodenough, J. Phys. Chem. Solids 24
(1963) 347.
[33] V. Nivoix, B. Gillot, Solid State Ionics 111 (1998) 17.
[34] B. Gillot, V. Nivoix, Mater. Res. Bull. 34 (1999) 1735.
[35] V. Nivoix, B. Gillot, Chem. Mater. 12 (2000) 2971.
[36] V. Nivoix, B. Gillot, Mater. Chem. Phys. 63 (2000) 24.
[37] B. Gillot, M. Nohair, Phys. Status Solidi A 176 (1999) 1047.
[38] M. Nohair, P. Perriat, B. Domenichini, B. Gillot, Thermochim. Acta 244 (1994)
223.
[39] R.D. Shannon, Acta Crystallogr. Sect. A 32 (1976) 751.
[40] C.-B. Wang, Y. Cai, I.E. Wachs, Langmuir 15 (1999) 1223.
[41] L.E. Briand, A.M. Hirt, I.E. Wachs, J. Catal. 202 (2001) 268.
[42] M. Bowker, R. Holroyd, M. House, R. Bracey, C. Bamroongwongdee, M. Shannon,
A. Carley, Top. Catal. 48 (2008) 158.
[43] M.P. House, A.F. Carley, R. Echeverria-Valda, M. Bowker, J. Phys. Chem. C 112
(2008) 4333.
[44] C.J. Machiels, A.W. Sleight, J. Catal. 76 (1982) 238.
[45] L.J. Burcham, I.E. Wachs, Catal. Today 49 (1999) 467.
[46] L.E. Briand, W.E. Farneth, I.E. Wachs, Catal. Today 62 (2000) 219.
[47] X. Gao, I.E. Wachs, Top. Catal. 18 (2002) 243.
[48] B.M. Weckhuysen, D.E. Keller, Catal. Today 78 (2003) 25.
[49] M. Seman, J.N. Kondo, K. Domen, R. Radhakrishnan, S.T. Oyama, J. Phys. Chem.
B 106 (2002) 12965.
[50] W.L. Holstein, C.J. Machiels, J. Catal. 162 (1996) 118.
[51] J. Döbler, M. Pritzsche, J. Sauer, J. Am. Chem. Soc. 127 (2005) 10861.
[52] A. Goodrow, A.T. Bell, J. Phys. Chem. C 111 (2007) 14753.
[53] F. Roozeboom, P.D. Cordingley, P.J. Gellings, J. Catal. 68 (1981) 464.
[54] G.D. Kolovertnov, G.K. Boreskov, V.A. Dzis’ko, B.I. Popov, D.V. Tarasova, G.G. Bel-
ugina, Kinet. Katal. (Engl. Transl.) 6 (1965) 1052.
XRD and XANES showed that the triclinic FeVO4 phase was un-
stable under the reaction conditions and formed a cationvacant
spinel-type Fe1.5 V1.5 O4 phase. Substituting Al for Fe in the catalyst
gave more stable bulk structures.
Acknowledgments
The Swedish Research Council (VR) is acknowledged for finan-
cial support.
References
[1] G. Reuss, W. Disteldorf, A.O. Gamer, A. Hilt, in: Ullmann’s Encyclopedia of In-
dustrial Chemistry, vol. A11, seventh ed., Wiley–VCH, Weinheim, 2008, pp. 619–
652.
[2] B. Crichton in: Informally Speaking (Newsletter from Perstorp Formox,
http://www.perstorpformox.com), spring/summer 2003, pp. 12–13.
[3] A.R. Chauvel, P.R. Courty, R. Maux, C. Petitpas, Hydrocarbon Process. 52 (9)
(1973) 179.
[4] A. Andersson, M. Hernelind, O. Augustsson, Catal. Today 112 (2006) 40.