redox state dynamics at the surface of movte(sb)nbo m1 phase in selective oxidation of light alkanes
TRANSCRIPT
ORIGINAL PAPER
Redox State Dynamics at the Surface of MoVTe(Sb)NbO M1Phase in Selective Oxidation of Light Alkanes
Benoit Deniau • Thi Thao Nguyen • Pierre Delichere •
Olga Safonova • Jean-Marc M. Millet
� Springer Science+Business Media New York 2013
Abstract The influence of Te or Sb on the catalytic
properties of the M1 phase of MoVTe(Sb)NbO catalysts in
the oxidation of propane or oxidative dehydrogenation of
ethane has been investigated. The results show that Te and
Sb affect the intrinsic activity of the catalyst for ethane and
propane oxidation in the same way and that the Te-con-
taining M1 phase was significantly more active than the
Sb-containing phase, even when they contained the same
amount of vanadium, which is known to be the active
species for the reactions. This feature has been explained
by a strong modification of the relative number of available
V5? sites, which is related to the fact that Sb tend to sta-
bilize vanadium in active site, in a reduced state. Sb and Te
were also shown to affect the selectivity in the case of
propane oxidation but not in the case of ethane oxidative
dehydrogenation. This demonstrated that they should be
involved directly in the further transformation of the alkene
molecules that are formed as an intermediate and pre-
sumably play a role in the a-hydrogen abstraction from
these molecules, as proposed earlier. Blocking problems
related to the substitution of Te by Sb in view of indus-
trialization are also discussed.
Keywords MoVTeNbO oxide catalysts � Selective
oxidation � Ammoxidation � Alkanes � Ethylene � M1 phase
1 Introduction
Among the best catalysts proposed for light alkane oxi-
dation, the multi-component MoVTe(Sb)NbO catalysts
are very promising [1]. The active phase of these cata-
lysts, called M1, is a solid solution containing the four
metallic cations with V/Mo, Nb/Mo and Te(Sb)/Mo ratios
that can vary, but are most frequently close to 0.3, 0.1
and 0.2, respectively [2–8]. Although the presence of Nb
is not essential, all of the other elements are and key
catalytic functions have been attributed to them. A gen-
eral agreement has been obtained on the fact that V is the
paraffin activator, and that Te or Sb is involved in a-
hydrogen abstraction from the olefin that is formed from
the alkane. In their higher oxidation state, Te and Sb have
been proposed to be the O- or NH-inserting element, in
contrast to Mo, which generally plays this role. As pre-
viously pointed out, all of these elements are strategically
placed adjacent to each other in the active phase of the
catalysts called M1 and comprising the multifunctional
catalytic site assembly [9]. Such site assembly has been
proposed to be associated with a specific surface termi-
nation of the M1 phase corresponding to ab plane
[10, 11].
In the structure of the M1 phase, Te and Sb cations are
located in hexagonal channels formed by MO6 octahedra
(M = Mo, V, Nb) sharing corners. They are formed with
oxygen chains running in the [001] direction, but the
oxygen and Te or Sb sub-stoichiometry in the channels are
not linked to each other [10]. Tellurium has been shown to
be Te?4 on the whole, whereas antimony was shown to be
B. Deniau � T. T. Nguyen � P. Delichere � J.-M. M. Millet (&)
Institut de Recherches sur la Catalyse et l’Environnement de
Lyon, IRCELYON UMR5256, CNRS-Universite Claude
Bernard, Lyon I, 2 Avenue A. Einstein,
69626 Villeurbanne Cedex, France
e-mail: [email protected]
O. Safonova
Institute for Chemical and Bioengineering, ETH Zurich, Zurich,
Switzerland
123
Top Catal
DOI 10.1007/s11244-013-0132-0
present both as Sb5? and Sb3? [12, 13]. The oxidation state
of both elements at the surface of the catalysts has not been
ascertained, since the position of the Te3d3/2 and Sb3d3/2
peaks in the X-ray photoelectrons spectra were most
commonly observed between those of Te4? and Te6?, and
Sb3? and Sb5?; however, most of the time this was closer
to the higher oxidation state cations. Consequently, several
publications have proposed that the oxidation state of Te at
the surface was ?6, while others have suggested ?4
[6, 14–16].
Te-based catalysts have been shown to be more effi-
cient than Sb-based ones, regardless of the paraffin to be
oxidized [6, 17]. However, there is a major drawback of
using tellurium-containing catalysts, which is its volatility
when reduced up to the metallic state, which can always
append in catalytic reaction conditions. This is why it
would be advantageous to substitute Te with Sb or Bi,
both of which are less volatile and more eco-friendly. The
Sb-containing M1 phase can be prepared relatively easily,
as shown earlier, but bismuth-containing M1 phases have
never been obtained despite many attempts [18]. This
may be due to the much larger size of the Bi3? cations,
which cannot accommodate the hexagonal channels of the
structure.
To better understand the complete reaction mechanism
of paraffin oxidation or ammoxidation of the MoV-
Te(Sb)NbO catalysts, it is necessary to determine the exact
respective functionalities of the various elements, including
Te or Sb, but first their catalytic effect has to be better
evidenced. Most commonly, and because of the existence
of a synergetic effect between the M1 active phase and a
second phase called the M2 phase, the studied catalysts
contain this second phase. Furthermore, other phases may
also be present, like Mo5–x(V,Nb)xO14, TeMo5O16 or
VSbO4. The presence of these phases in most of the cata-
lysts of the published studies enables to precisely define the
effect of a given element in the active M1 phase on its
catalytic properties.
In this paper we report the preparation and study of pure
or almost pure M1 phases containing Te or Sb as catalysts.
Depending on the total composition of the catalysts and the
preparation method, Te or Sb can be present in sub-stoi-
chiometry and can exhibit several oxidation states. While
these variations do not alter the crystal structure, they may
affect the activity and selectivity of the catalyst. Thus,
there is a need for a better understanding of the impact of
the presence of these elements, as they will define the
active sites or modify their environment and consequently
their catalytic properties. We have prepared numerous
catalysts but have selected a few that are as pure as possible
in M1 phases, as well as those containing Te or Sb, in order
to study the effect of the nature of the element and its
relative content on the catalytic properties. Furthermore, it
has been shown by Naraschewski et al. [19] that the
initial catalytic rate of propane oxidation is proportional
to the total concentration of vanadium in the structure for
Te-containing M1 phases. Consequently, it was necessary
to compare M1 phases with approximately the same V
content. The catalysts have been studied either in ethane
or propane oxidation to better detail the different poten-
tial roles of Te and Sb and to better understand the
intricacies of the catalytic processes. In the case of eth-
ane, the reaction pathway is simpler with the formation
of only carbon oxides as side products, making it easier
to focus on the effect of the elements on paraffin acti-
vation; in contrast, in the case of propane, the effect of
Te and Sb in the consecutive steps of the reaction, like
the transformation of propene formed intermediately, can
be studied.
Finally in order to try to explain the results obtained, the
M1 phase catalysts have been studied by XANES spec-
troscopy at the V K, Sb and Te L1-edge during reduction
by propane at the catalytic reaction temperature. In a pre-
vious paper on MoVSbNbO catalysts studied by in situ
X-ray absorption spectroscopy, we had shown that the
oxidation state of Sb could change in the reaction condi-
tions and could be involved in the total redox process of the
catalysts, as well as the fact that these changes could be
evidenced by XANES spectroscopy [20]. XPS spectros-
copy has also been used to study the reduction by propane
of a Te containing M1 phase and to confirm the XANES
data obtained.
2 Experimental
2.1 Preparation of the Pure M1 Phase Catalysts
M1 phases containing Te and Sb were prepared using
protocols described elsewhere [21]. Both protocols inclu-
ded the preparation of aqueous slurry containing Mo, V, Sb
or Te precursors in the chosen ratios. In the Sb containing
M1 phase, precursors were dissolved in an aqueous solu-
tion maintained at 80 �C for 2 h. When Sb2O3 was used,
hydrogen peroxide in H2O2/Sb ratio of three was then
added to re-oxidize the V species [12]. Silica Ludox AS40
with an amount corresponding to a Si/Mo = 0.95 was
added to the solution. In parallel, niobic acid was dissolved
at 100 �C in an aqueous solution of oxalic acid and added.
The slurries obtained were dried at 130 �C and heat-treated
in air at 300 �C for 4 h and in N2 at 600 �C for 2 h. The
solids obtained contained the M2 phase, which was
removed by selective dissolution in aqueous hydrogen
peroxide solution at 60 �C. The solids were finally washed
in water, dried at 130 �C and heated in N2 at 600 �C for
2 h.
Top Catal
123
2.2 Characterization of the Catalysts
The metal content of the mixed oxides was determined by
atomic absorption (ICP), and their specific surface areas
were determined using the BET method with nitrogen
adsorption. Powder X-ray diffraction (XRD) patterns were
obtained using a BRUKER D5005 diffractometer with a
Ni-filtered CuKa (0.15418 nm) radiation source.
V K-edge (5.465 keV) and Sb and Te L1-edge (4.9485
and 4.698 keV) XANES spectra were collected at the
BM29 beamline at the European Synchrotron Radiation
Facility (ESRF) in Grenoble, France. Mo spectra were not
recorded since it had been shown that it was not involved in
the redox process or at a level that is difficult to detect [20].
Sb and Te L1-edge XANES spectra were recorded with a
varying energy step of 0.4 eV/20 s in the range
4680–4730 eV and 0.3 eV/20 s in the range 4680–4735.
The V K-edge spectra were collected with a varying
energy step of 0.2 eV/5 s in the range 5420–5560 eV. The
experiments were recorded using a Si(111) double crystal
and in the fluorescence mode using a Camberra Si diode
detector, which was mounted at 90� with respect to the
incident beam. For the in situ spectra, a 20 mL cell
described in a previous work was used [20]. This cell,
which was equipped with a 25 lm Kapton window,
allowed measurement of XAS at the catalytic reaction
temperature under various flowing atmospheres. The study
was performed at atmospheric pressure under a gas flow of
20–30 mL min-1, the composition of which was moni-
tored using Brooks flowmeters. XANES spectra were
recorded at in N2 and O2/N2 (10/90) atmosphere. The
heating from 25 to 380 �C was performed under N2 and the
rate of heating was 5 �C min-1. Spectra were also recorded
in N2 after different pulsing of propane and using a specific
valve allowing 5 mL propane to be sent on the solid to
reduce it. In such cases, we waited 20 min before starting
the recording in order to allow the M1 phase to reach
equilibrium. This length of time was required for the dif-
fusion of oxygen from the bulk to the surface of the M1
crystallites through the hexagonal channels [20].
XANES spectra were normalized and background-cor-
rected using the IFFEFIT program. In order to compare the
different Te and Sb L1-edge XANES spectra, the absorp-
tion background was first determined over the entire range
and the spectra were normalized in the middle of the first
EXAFS oscillation, at ca. 50 eV above the absorption edge.
The relative ratio of SbIII and SbV or of TeII and TeIV in the
compounds could be determined by fitting the L1-edge
XANES spectra considering a combination of two peaks
whose relative areas are proportional to the relative amount
of the two cations [11]. For the V edge spectra, the analysis
was based on the observations of the pre-edge. The position
of the center of mass and the total area of the V pre-edge
peak were calculated using the Peak-Fit program and used
to determine the mean oxidation state of V as previously
described [20]. Ti, V and Sb metal foils were measured for
the different calibrations.
XPS measurements were performed using a Kratos Axis
Ultra DLD spectrometer. The base pressure in the analysis
chamber was better than 5 9 10-8 Pa. XPS spectra of V2p,
Te 3d, Mo 3d, Nb 3d O1 s, and C1 s levels were measured
at 90� (normal angle with respect to the plane of the sur-
face) using a monochromated AlKa X-ray source with a
pass energy of 20 eV and a spot size aperture of
300 9 700 lm. Binding energies were corrected relative to
the carbon 1 s signal at 284.6 eV. The signal intensities
were measured using integrated areas under the detected
peaks. The experimental precision on XPS quantitative
measurements was considered to lead to precision on the
calculated cationic ratio of 0.15.
Reduction of the samples was conducted in an adjacent
chamber at 380 �C. The sample was heated under Ar flow
(50 mL min-1) and two series of 10 pulses of 5 mL of
propane were sent in the chamber. At the end the sample
was re-oxidize at the same temperature by shifting the Ar
flux to an O2 flux for 1 min.
2.3 Testing of the Catalysts
The oxidative dehydrogenation of ethane and the oxidation
of propane were performed in a fixed bed reactor operating
at atmospheric pressure and isothermal conditions using a
glass plugflow micro reactor. The catalytic properties were
determined between 360 and 400 �C with catalyst amounts
varying from 200 to 3000 mg and total flows varying from
6 to 125 mL min-1. The feedstock compositions were
C2H6/O2/N2 ? He = 30/20/50 and C3H8/O2/N2 ? He =
10/5/45/40, respectively. Ethane (99 %) from commer-
cially available compressed gas cylinders was used without
further purification.
After reaction, alkanes, alkenes, CO and CO2 were
analyzed online using a gas chromatograph equipped with
a TCD detector and either Porapak-Q and CPMolsieve 5A
columns or a Carboxen 1010 column. Organic substrates
were condensed during the reaction and analyzed offline
using a gas chromatograph equipped with an FID detector
and CPWax or a Nukol column. Most of the testing was
performed at 385 �C, and only traces of acetone in the case
of propane oxidation and acetic acid in the case of ethane
oxidative dehydrogenation were detected. These products
were not considered for carbon balance and selectivity
calculations. Intrinsic rates of conversion were calculated
at conversion low enough to be in a chemical regime and
compared to each other. No deactivation of the catalysts
was observed during the reaction times.
Top Catal
123
3 Results
3.1 Characterization of the Fresh Catalysts
The XRD patterns of the samples reveal only one phase,
corresponding to the M1 phase. In some cases better fits
were obtained adding a Mo5O14 type phase, which amount
should not overpass 1 or 2 %. To illustrate this the XRD
pattern of one of the samples with its fitting is shown in
Fig. 1. The cell parameters calculated from the powder
patterns were comparable for catalysts of the same type and
to those of the literature (Table 1). The chemical compo-
sitions of the unit cell of the M1 phases have been deter-
mined from chemical analyses. The results obtained were
in relatively good agreement with the desired stoichiome-
try, except that the Te contents of the catalysts were sys-
tematically lower than expected (Table 2). This may be
explained both by the formation in Te containing catalysts,
before the dissolution step, of a slightly higher amount of
M2 phase, which is richer in Te than the M1 phase and
possibly to a small loss of Te by vaporization during the
heat treatments at 600 �C. However it should be noted that
such loss has been evaluated in several cases and never
exceeds 3 %. The specific surface areas of the solids have
been determined using the BET method; they varied
between 6 and 18 m2 g-1 and seemed to slightly increase
with the vanadium concentration of the samples (Table 2).
3.2 Catalytic Activity
3.2.1 Effect of Sb or Te Content
It has been shown previously that the tellurium content of
the M1 phase had no impact on the catalytic properties
when ranging between 1 and 4 tellurium atoms per unit cell
[22, 23]. It is generally not possible to prepare a pure
M1(Te) phase with lower or higher Te content since other
phases are systematically formed at lower contents and all
of the sites in the hexagonal channels are occupied at
higher content; also an excess of Te led to metallic tellu-
rium or the formation of phases richer in Te like M2 [5,
19]. We have thus focused our attention on the effect of Sb
content in the M1 phase to see whether the same results
could be obtained. It was important to check if it was the
same for the Sb containing M1 phase. For that two M1
phases, M1(Sb)-A and M1(Sb)-B with comparable V
contents but different Sb contents were tested as catalysts
in the oxidation of propane (Table 2). The results are
shown in Table 3. It can be seen that the catalytic prop-
erties of the two M1 phases are comparable.
Fig. 1 X-ray diffraction pattern
of M1(Te)-G with its profile
fitting. The 5–14.5� 2h range is
enlarged and a peak indexation
is given
Table 1 Unit cell parameters of the prepared M1 phase catalysts
Catalyst Unit cell parameters Cell volume
(nm3)a (nm) b (nm) c (nm)
M1(Sb)-A 2.1134 (5) 2.6650 (7) 0.4011 (1) 2.2591
M1(Sb)-B 2.1139 (9) 2.669 (1) 0.4012 (2) 2.2635
M1(Sb)-C 2.1133 (3) 2.6633 (4) 0.40150 (5) 2.2598
M1(Te)-D 2.1124 (2) 2.6575 (3) 0.40167 (4) 2.2548
M1(Te)-E 2.1142 (8) 2.6603 (11) 0.4010 (1) 2.2538
M1(Sb)-F 2.1136 (9) 2.6620 (10) 0.4010 (2) 2.2562
M1(Te)-G 2.1118 (5) 2.6602 (7) 0.4008 (1) 2.2516
Table 2 Unit cell content in atoms, specific and surface area (SSA)
of the prepared M1 phases
Catalyst Unit cell composition in atoms SSA (m2 s-1)
Mo V Nb Te Sb
M1(Sb)-A 24.2 7.4 3.8 2.8 19.1 ± 0.2
M1(Sb)-B 28.6 7.6 3.8 4.0 15.4 ± 0.1
M1(Sb)-C 29.2 6.2 4.6 3.3 28.8 ± 0.3
M1(Te)-D 29.8 6.5 3.7 2.3 23.8 ± 0.3
M1(Te)-E 28.4 7.4 4.0 2.6 22.3 ± 0.1
M1(Sb)-F 28.8 7.4 3.8 4.0 19.3 ± 0.2
M1(Te)-G 29.2 6.6 4.2 3.7 24.2 ± 0.1
Top Catal
123
3.2.2 Effects of Te and Sb Nature
In order to show the influence of the nature of Sb or Te
cations, the catalytic properties of M1(Sb)-C and M1(Te)-
D catalysts with again approximately the same V content
were compared with regard to the oxidation of ethane and
propane. For the oxidation of ethane, the results showed
that the presence of Te or Sb has almost no influence on the
selectivity in ethylene (Fig. 2). However a marked effect
on the activity was observed. This clearly appeared when
intrinsic rates of ethane conversion were compared at the
same contact time (Table 4). The apparent activation
energies that were calculated were similar, although
slightly higher for the M1(Te) phase. When the solids were
tested in propane oxidation, the same effect on the rate of
propane conversion was observed (Table 5). Interestingly,
the ratios between the rates measured in both reactions
were comparable. However, a higher selectivity to acrylic
acid on Te containing M1 phase was observed with pro-
pane. This result is consistent with previous studies also
conducted on pure or almost pure M1 phase [24]. We again
observed higher apparent activation energy for the Te
containing phase than for the Sb containing one with regard
to the transformation of propane.
3.2.3 In Situ Characterization by XANES Upon
Reduction by Propane
It is known that Sb or Te can be involved in redox reactions
of the M1 phase [20]. It was therefore interesting to study
the reducibility of these elements in situ by an alkane to see
whether it could explain the catalytic data obtained. We
recorded Te and Sb L1-edge XANES spectra of the M1
phase at the catalytic reaction temperature when propane
was contacted with the phase. Furthermore and because it
is also known that V is the element activating the alkane
molecules, V K-edge XANES spectra were also recorded.
Two M1 phase catalysts, M1(Te)-E and M1(Sb)-F, again
with comparable V contents, were used for this study
(Table 2).
The Te L1-edge and V K-edge XANES spectra of the
M1(Te)-E sample were recorded at 380 �C in N2 and after
10 pulses of propane (5 mL) or after shifting the N2 flux to
and O2/N2 flux (10/90). These spectra are respectively
shown in Figs. 3 and 4 respectively. The V K-edge spec-
trum showed a pre-edge peak, which center of mass cor-
responded to a mean oxidation state of about 4.1, whereas
the Te L1-edge spectrum was typical of Te4? cations. A
slight modification of the V pre-edge was observed under
oxidative atmosphere with a small shoulder appearing
reversibly on the right side, this indicated the partial oxi-
dation of about 3 % of V41 into V51. When the flux is
shifted back to pure N2, the initial V spectrum is observed.
Concomitantly, no change in the Te spectrum could be
Table 3 Catalytic properties of M1(Sb) catalysts in propane oxidation at 385 �C
Catalyst Mass (mg) Conv. (%) Rate (mol s-1 m-2) Selectivity (%)
COx C3H6 AceA ProA. AcrA
M1(Sb)-A 340 26.3 4.8 9 10-8 19 15 6 1 58
M1(Sb)-B 300 25.0 5.4 9 10-8 24 13 7 0 56
AceA acetic acid, ProA. propionic acid, AcrA acrylic acid
50
60
70
80
90
100
0 10 20 30 40 50 60 70
Sélé
ctiv
ité
C2H
4(%
)
Conversion C2H6 (%)
M1(Sb)M1(Te)
Fig. 2 Variation of the selectivity in ethylene as a function of the
conversion on M1(Sb)-C and M1(Te)-D catalysts
Table 4 Intrinsic rate of ethane conversion on M1(Te) and M1(Sb)
catalysts with apparent activation energy (Ea)
Catalyst Ea (kJ mol-1) Rate (mol s-1 m-2)
M1(Sb)-C 68 5.3 9 10-8
M1(Te)-D 76 13.0 9 10-8
Table 5 Intrinsic rate of propane conversion on M1(Te) and M1(Sb)
catalysts with apparent activation energy (Ea)
Catalyst Ea (kJ mol-1) Rate (mol s-1 m-2)
M1(Sb)-C 54 5.5 9 10-8
M1(Te)-D 61 14.5 9 10-8
Top Catal
123
detected. After reduction by propane, no difference in the
position or shape of the pre-peak was observed in the V
spectrum, whereas the Te L1-edge spectrum showed a
shoulder at 4941 eV that was attributed to the presence of
Te2?. The evaluation of the relative area of the new peak
allowed the evaluation of the reduction of Te cations to
5 at.%.
For the M1(Sb)-F sample, the spectra recorded at the
V K-edge and Sb L1-edge are shown in Figs. 5 and 6. The
V K-edge spectrum of the sample was very similar to that
observed for M1(Te)-E and the same mean oxidation state
of 4.1 for the vanadium was calculated. The in situ study of
the oxidation of an Sb-containing M1 phase has previously
been published [20]. It was observed that Sb and V were
a b
Fig. 3 a XANES spectra of M1(Te)-E at Te L1-edge recorded at 380 �C (a) before and (b) after oxidation. b XANES spectra of M1(Te)-E at
V K-edge recorded at 380 �C (a) in N2 and (b) after oxidation
a b
Fig. 4 a XANES spectra of M1(Te)-E at Te L1-edge recorded at 380 �C (a) before and (b) after reduction by propane. b XANES spectra of
M1(Te)-E at V K-edge recorded at 380 �C (a) under N2 and (b) after reduction by propane
Top Catal
123
oxidized in parallel. We have not reproduced similar
experiments and we have focused our attention on the
in situ reduction of the M1 phase by propane.
In this case the reduction was conducted in two steps,
with the first 10 pulses of propane sent to the solid
(reduction 1) and then, after recording of spectra, five more
pulses were sent (reduction 2). We can see that V and Sb
were reduced after the two treatments. Calculation showed
that about 4 % of vanadium and 15 % of Sb were reduced
(1 % and 10 % after the first reduction). It is interesting to
note that Sb contributed for about four times more to the
reduction even though contents were less than half of V.
Such a phenomenon, which was already observed in the
previous study for oxidation of both cations, confirms the
extremely high oxygen mobility in these channels and the
role of the latters as an oxygen reservoir.
3.2.4 In Situ Characterization by XPS Upon Reduction
by Propane
The M1 phase has been characterized by XPS before and
after reduction by propane. Mo, V, Te, Nb and O have been
analyzed. The results obtained are presented on Table 6.
The surface composition of the fresh sample corresponded
to that of the bulk. Contrarily to what was observed before
no surface enrichment in Te was observed and this content
was even lower than the bulk one (3.1 instead of 3.7 Te per
unit cell) [12, 17]. Both V4? and V5? were detected at the
surface leading to as surface mean oxidation state of V
slightly higher than the bulk one (4.3 instead of 4.1). For
Te region only a single Te 3d5/2 peak assigned to Te4? was
observed. The analysis of the literature showed that this
peak has also been in certain cases assigned to Te6? [15,
16]. Such mismatch was related to the fact that the binding
energy of Te 3d levels vary over a large range depending
upon the configuration of the species present (TeO3E or
TeO4E) [25].
After the first series of pulse of propane, a strong
reduction of Te was observed (Fig. 7). In addition to the Te
3d5/2 main peak at 576.7 eV, an additional peak at
574.1 eV was observed. The latter peak has been attributed
to Te2? state [26, 27]. The reduction of tellurium was much
more important that what was observed by XANES (40
instead of 5 %). However the reduction conditions were
different in terms of reaction cell geometry and gas flux. In
the same time, a reduction of Mo and V was detected but to
a much less extend. These results confirmed well those
obtained by XANES spectroscopy. In parallel the relative
amount of surface Te has decreased whereas those of the
other elements remained similar. When the solid was fur-
ther reduced with a second series of 10 pulses of propane,
no further reduction of Te was observed and almost the
same Te4?/Te2? ratio was observed. The surface compo-
sition also remained the same. After the second reduction,
the solid was treated under O2 at 380 �C. The spectrum
recorded after this treatment showed that Te was entirely
re-oxidized to Te4?. In the same time the Te content
increased to reach its starting value. V was also re-oxidized
to reach an oxidation state even slightly higher than the
starting one similarly to the sample studied by XANES
when treated under oxidative atmosphere (Fig. 7).
Fig. 5 XANES spectra of M1(Sb)-F at Sb L1-edge recorded at
380 �C (a) before and (b) after different successive reductions by
propane
Fig. 6 XANES spectra of M1(Sb)-F at V K-edge recorded at 380 �C
(a) before and (b) after different successive reductions by propane
Top Catal
123
4 Discussion
Since the results of the in situ study may help to understand
the catalytic results it appears more fruitful to discuss them
first. We have seen that the oxidation state of V and Te or
Sb in the studied M1 phase corresponded well to those
determined before in various studies using similar tech-
niques. XANES data showed that the mean oxidation state
of V was slightly higher than ?4, and that all of the Te
cations were ?4, whereas Sb cations were both ?3 and ?5,
but mainly ?3 (Sb5?/Sb3? = 30/70). There is a general
agreement that the active species for the light alkane oxi-
dation reaction on these catalysts are vanadium species and
that the V51/V41 redox couple would reasonably be the
first redox couple that would be affected by a reduction by
propane [4]. This was observed for Sb-containing catalysts,
for which we observed that antimony was concomitantly
reduced [20]. However, a different behavior was observed
in the case of Te-containing phase. Tellurium was partially
reduced but vanadium was not. An hypothesis to explain
these results is to consider an oxido-reduction process
between surface Te4? and V4?:
Te4þ þ 2V4þ�Te2þ þ 2V5þ ð1Þ
Furthermore, in the catalytic reaction mechanism pro-
posed by Grasselli et al. [9], the dehydrogenation of pro-
pane was proposed to take place with the transitory
reduction of a V5? cation to V4? and that of a Te4? cation
to Te3?. The charge transfer could then take place between
these cations leading to the stabilization of Te2?cations:
Te4þ þ V4þ� ðTe3þ þ V4þÞ�Te2þ þ V5þ ð2Þ
Surface Te4? can be further reduced to Te0 by undergoing
a second charge transfer or by being reduced by the propene
formed. The same type of transfer does not exist in the Sb-
containing compounds since the Sb5?/Sb3? redox potential
is much lower than the V5?/V4? one (Sb5?/Sb3?:
0.649 eV - V5?/V4?:1.00 eV). This feature is clearly
shown by the stabilization in air and at temperatures as
high as 700 �C of V4? and V3? with Sb5? species in the
VSbO4 mixed oxide [28]. Concerning tellurium, it can be
noted that a similar effect of tellurium had already been
reported in a study of Te-doped VPO used for the oxidative
dehydrogenation of ethane with a positive effect on the re-
oxidation of V4? species. It has to be pointed that tellurium
and vanadium can be reduced independently from one
another and the proposed oxido-reduction may not be so
important in the reaction conditions.
The XPS study confirmed that upon reduction by pro-
pane Te2? was formed whereas the V5? content was
slightly modified. Interestingly this change in oxidation
state was coupled with a decrease in relative content. Both
modifications appeared to be totally reversible upon re-
oxidation by O2. It should be noticed that no significant
change in Mo and V relative contents were observed in
parallel upon oxidation or reduction (Table 6). Two
hypotheses may be considered to explain the decrease of the
Te surface content. Firstly we have seen that upon reduc-
tion, Te was reduced to Te2?, which may itself easily be
further reduced to Te0, which is volatilized. Secondly the
equilibration of charge balance at the surface may take
place not only by reduction of metallic elements but also by
diffusion of Te from the surface into the bulk. In both cases
the replenishment of the surface in Te upon oxidation could
only be explained by diffusion of Te from the bulk to the
surface. It is interesting to note that such diffusion of Te
from the bulk to the surface explained why the bulk Te
content of the M1 phases has, in a certain range of com-
position, a limited impact on the catalytic properties, since
the surface Te content seems to depend mainly from the
Table 6 XPS analyses of the M1(Te) sample before, after reduction
by propane and re-oxidation by O2 at 380 �C
Solid Binding energy (eV) Te/Mo V/Mo
M1(Te)-G before reduction
Mo 3d5/2 Mo6?: 232.9 0.104 0.137
V 2p3/2 V5?: 517.5 (31 %)
Te 3d5/2 V4?: 516.6 (69 %)
Nb 3d5/2 Te4?: 576.7
Nb5?: 207.0
M1(Te)-G after reduction 1
Mo 3d5/2 Mo6?: 232.8 (85 %) 0.037 0.118
Mo5?: 231.6 (15 %)
V 2p3/2 V5?: 517.4 (26 %)
V4?: 516.5 (74 %)
Te 3d5/2 Te4?: 576.7 (62 %)
Te2?: 574.1 (38 %)
Nb 3d5/2 Nb5?: 207.0
M1(Te)-G after reduction 2
Mo 3d5/2 Mo6?: 232.7 (86 %) 0.035 0.131
Mo5?: 231.6 (14 %)
V 2p3/2 V5?: 517.4 (24 %)
V4?: 516.5 (76 %)
Te 3d5/2 Te4?: 576.7 (60 %)
Te2?: 574.1 (40 %)
Nb 3d5/2 Nb5?: 207.0
M1(Te)-G after re-oxidation
Mo 3d5/2 Mo6?: 232.9 0.095 0.127
V 2p3/2 V5?: 517.6 (35 %)
V4?: 516.6 (65 %)
Te 3d5/2 Te4?: 576.7
Nb 3d5/2 Nb5?: 207.0
Top Catal
123
redox atmosphere in equilibrium with the phase. It can be
observed that the surface reduction increased only slightly
after the first 10 pulses. The same behavior was observed by
Kubo et al. [29] in a previously study of the reduction
behavior of MoVTeNbO catalysts by pulse of propane
followed by gas chromatography. It was also shown in the
same study that up to a calculated surface reduction of 33 %
the reduction was reversible. In our case the analysis of the
O1s signal showed that the reduction was about 25 %.
The reduction of antimony by itself has no effect on the
stability of the catalyst. This should not be the case for
tellurium. Indeed, Te2? cations are not stable species and
may easily be further reduced to Te0 and be volatized. If as
shown here, the surface Te content can be replenished, over
long periods of time, this could have an effect on the
selectivity and on the stability of the M1 structure itself.
However it should be recalled that no such loss of selec-
tivity and catalyst deactivation has yet been reported in
catalytic reaction studies performed at the laboratory scale.
This study tend to prove that tellurium may be lost under
strong reducing atmosphere, it also showed that Te is
mobile in the channels of the structure and can easily
replenish the Te surface content. We had already shown
some years ago that Te was mobile at the surface of the
M1(Te) phase when mixtures of M1(Te) and M2(Sb) phases
were studied with the evidence of diffusion of Te at the
surface of M2(Sb) [17]. This confirms the dynamic behavior
of this type of oxidation catalyst and open the door to a
possible additional supra-surface role of tellurium [30].
Before discussing the influence of the nature of Sb or Te
on the catalytic properties, it is important to focus on the
effect of the Sb or Te contents in the M1 phase. The Sb or
Te content apparently has no significant effect for a given
composition range, on either the activity or selectivity of
the phase. The surface Te or Sb content presumably
remained the same whatever the bulk composition was and
the loss of these elements could be compensated by their
migration from the bulk as shown in the XPS study. Other
XPS analyses performed on various catalysts tend to con-
firm this interpretation since they always show a rather
similar content [27]. It can be recalled that, at low content
the M1 phase cannot be prepared alone or in a mixture with
M2, and that other phases like Mo5O14, which cannot be
removed are generally formed. At high content the for-
mation of M2 is preferential and, the Te and Sb content of
the M1 phase cannot be increased since the M2 phase
contains more Te or Sb than the M1 phase.
The results obtained clearly show that the nature of the
Sb or Te element has an important effect on the activity of
the catalyst and this is observed for both ethane and pro-
pane oxidation. Sb and Te also affect the selectivity, but
only in the case of propane oxidation; this second effect
will be discussed later. It has been clearly shown in pre-
vious studies that the oxidative dehydrogenation of the
alkane molecule was the slowest step in the reaction path
[30, 31] and that the alkene was often the only primary
product, while the direct oxidation of propane to carbon
oxides was rarely observed (<1 %). The same conclusion
could be reached for the oxidative dehydrogenation of
ethane, for which the initial selectivity to carbon oxides
was generally lower than 3 %. As stated before, there is a
general agreement on the role of vanadium in this first
reaction step. The alkane activating and methylene
hydrogen abstracting element is V5?, which is able to
Fig. 7 XPS spectra for The M1
phase (M1(Te)-G) of the Te 3d5/2
region before reduction (a) after
reduction with 10 pulses of
propane at 380 �C (b) and after
re-oxidation by O2 at the same
temperature (c)
Top Catal
123
undergo an electron transfer within the vanadyl cation that
imparts a radical character to its oxygen that is capable of
attacking the methylene–H and abstracting it as an H. In
view of this mechanism it was not so easy to understand the
effect of Sb and Te on the reaction rate, which was more
than two times lower for the Sb-containing M1 phase. On
another hand, the apparent activation energy for ethane
oxidative dehydrogenation is lower for the Sb-containing
M1 phase than for the Te-containing phase. If we look to
the variation of Ea with the logarithm of the rate when
moving from Sb- to Te-containing M1 phases we observed
that both increased although Ea increased more gently. This
feature can only be explained by the presence of intrinsi-
cally more active catalytic sites on M1(Sb) than on
M1(Te). The M1(Sb)-C phase has a slightly higher content
in V than the M1(Te)-D phase. Such a difference should
imply a higher rate of conversion of the alkane for the Sb-
containing phase if it was the most determinant parameter,
but this was not the case. One may postulate that the dis-
tribution of vanadium over the different sites and conse-
quently on the active sites is different between the Sb- and
Te-containing M1 phases. However structural determina-
tions made on both types of phases do not show significant
differences in vanadium distribution over the sites that
could account for the difference in activity. Furthermore
both the bulk and surface V content are comparable in both
types of compounds. It can be proposed from the reduction
experiments that the reduction level of V cations in the
M1(Sb) phases in steady state conditions under a reaction
gas mixture is higher than in the M1(Te) phases. Thus, part
of the active sites would be reduced and unavailable for the
reaction. V5?/V4? redox potential lies in between those of
Sb5?/Sb3? and Te4?/Te2? and reduced V species could
rapidly re-oxidized in M1(Te), whereas they are stabilized
in M1(Sb). This leads to a strong difference in number of
active sites in steady state reaction conditions, which
explained the important difference in the initial rate of
alkane conversion, which appeared almost independent of
the type of alkanes considered. Such interpretation stresses
the importance of having the active V5? oxo-species in
close vicinity to Te or Sb sites. One may note that such
requirement is fulfilled if the active site is the M7 site
occupied by V5? as proposed in several studies [4, 32]. It is
also fulfilled if the hypothesis of ensembles of V-Te(Sb)
oxo-species support on the M1 phase is considered [33].
Furthermore, the M1 phase has to be able to undergo
oxidation and reduction to a certain degree without changes
to its structure. However, a structure built from corner-
sharing octahedra should not support a high concentration
of oxygen vacancies. It is important to recall the key-role
played by hexagonal channels as oxygen reservoirs in the
M1 structure and the important role played by Te and Sb
occupying these channels [12, 20]. The surface reduction
level of Te4? cations appeared in this study much higher
than that of possible V5? active sites that are less numer-
ous. This confirms the importance of the former cations in
the re-oxidation of the catalysts in the reaction atmosphere
and in the buffering of oxygen content at the surface of the
catalysts.
Sb and Te also affect the selectivity but only in the case
of propane oxidation. This shows that they may not be
involved directly in the dehydrogenation of the alkanes
molecules but should be involved in the further transfor-
mation of the alkene molecules that are formed.
It was shown a long time ago that elements like Te or Sb
were key to enhancing the selectivity of desired partial
oxidation products [34]. It was proposed that the a-
hydrogen abstracting site of the olefin oxidation or am-
moxidation catalyst is an element possessing a lone pair of
electrons, which imparts a partial radical character to the
oxygen associated with it, thereby causing the abstraction
of the a-H as a H� [4]. In bi-molybdates the a-hydrogen-
abstracting element is Bi and it is reasonable to consider
that Te or Sb play the same role in the M1 phases. Sb and
Te would thus directly play a determinant role in a crucial
step of the reaction and the intrinsic electronic properties of
the two elements could explain the differences in selec-
tivity observed between the two types of M1 phases. The
presence of both types of cation (Sb5? and Sb3?) might
also explain them. Similar effects have been evidenced and
similarly explained in catalysts containing Te that were
used for the oxidation of other alkanes. For example, this
was the case for Keggin-type phosphomolybdic acid and
cesium salts used as catalysts in the oxidation of isobutane
into methacrolein or methacrylic acid, with protons that
were partially substituted by tellurium [35]. In this case,
tellurium induced a strong positive effect on the selectivity
to methacrylic acid and methacrolein when introduced as
counter-cation without any significant effect on the activ-
ity. A similar effect was also observed on Te doped Ni-
MoO4 catalysts [36].
5 Conclusions
This study is part of a general study undertaken to pinpoint
the respective functionalities associated with the various
metals of the active catalytic centers in the M1 phase. In
this study we have tried to determine the impact of Te or
Sb cations on the activity and selectivity either in propane
oxidation or ethane oxidative dehydrogenation and to
understand the redox state dynamics occurring at the sur-
face. We have shown that Te and Sb affect the intrinsic
activity of the catalyst for ethane and propane oxidation in
the same way and that the M1(Te) phases were much more
active than the M1(Sb) ones. These elements should not be
Top Catal
123
directly involved in the first dehydrogenation of the alkane
molecules, which is the rate-limiting step and their effect
on activity has been explained by an indirect effect on the
relative number of V5? active sites in steady state condi-
tions. A study of catalyst reduction at catalytic reaction
temperature by XANES spectroscopy allowed to show that
Te contrarily to Sb participates in the re-oxidation of the
vanadium species by local electron transfer. It leads any-
way to a higher active site density at the surface of the
catalyst than Sb. Such a local effect would be facilitated by
the strategic localization of Te or Sb and V active sites in
adjacent positions to each other in the M1 structure.
However it should be pointed that the observation of this
transfer does not mean that it necessarily takes place in the
reaction conditions.
Sb and Te were shown to affect the selectivity in the
case of propane oxidation but not in the case of ethane
oxidative dehydrogenation. This demonstrates that they
may not be involved directly in the dehydrogenation of the
alkane molecules, but should be involved directly in the
further transformation of the alkene molecules that are
formed. Such involvement has already been assessed in
numerous studies, showing that Te or Sb play the role of a-
hydrogen abstraction from the alkene molecules. The
substitution of Te by Sb that would be beneficial to several
points of view for future industrialization did not lead to
catalysts similarly efficient. Specifically for propane oxi-
dation, the difference in selectivity to acrylic acid (or
acrylonitrile in the case of ammoxidation) should be related
either to the intrinsic property of the Sb or Te element or to
the oxidation state of Sb. Only in the second case could a
solution be envisaged if the oxidation state could be
modified by a judicious surface or bulk doping that should
concomitantly not affect the active V sites. Finally, the
problem of the synergetic effect between M1 and M2,
which is not observed in the case of Sb catalysts remains to
be solved.
Acknowledgments We gratefully acknowledge ESRF for providing
the facilities at ID29 and we thank Dr. Carmelo Prestipino, the local
contact of beamline BM29, for help in preparing of the in situ setup
and XAS spectra recording.
References
1. Ushikubo T, Oshima K, Kayo A, Umezawa T, Kiyono K, Sawaki
I (1992) European Patent No. 529,853
2. De Santo P, Buttrey J, Grasselli RK, Lugmair CG, Volpe AF,
Togy BH, Vogt T (2004) Z Kristallogr 219:152
3. Ueda W, Vitry D, Katou T (2004) Catal Today 96:235
4. Grasselli RK, Buttrey DJ, Burrington JD, Andersson A, Holm-
berg J, Ueda W, Kubo J, Lugmair CG, Volpe AF Jr (2006) Top
Catal 38:7
5. Lopez Nieto JM, Botella P, Concepcion P, Dejoz A, Vazquez MI
(2004) Catal Today 91–92:241
6. Botella P, Dejoz A, Lopez Nieto JM, Concepcion P, Vazquez MI
(2006) Appl Catal A 298:16
7. Baca M, Pigamo A, Dubois JL, Millet JMM (2003) Top Catal
23:39
8. Botella P, Nieto JML, Solsona B, Mifsud A, Marquez F (2002) J
Catal 209:445
9. Grasselli RK, Burrington JD, Buttrey DJ, DeSanto P Jr, Lugmaird
CG, Volpe AF Jr, Weingand T (2003) Top Catal 23:5
10. Guliants VV, Bhandari R, Brongersma HH, Knoester A, Gaffney
AM, Han S (2005) J Phys Chem B 109:10234
11. Guliants VV, Bhandari R, Swaminathan B, Vasudevan VK,
Brongersma HH, Knoester A, Gaffney AM, Han S (2005) J Phys
Chem B 109:24046
12. Baca M, Millet JMM (2005) Appl Catal A 279:67
13. Millet JMM, Baca M, Pigamo A, Vitry D, Ueda W, Dubois JL
(2003) Appl Catal A 244:359
14. Oliver JM, Lopez Nieto JM, Botella P, Mifsud A (2004) Appl
Catal A 257:67
15. Haggblad R, Wagner JB, Deniau B, Millet JMM, Holmberg J,
Grasselli RK, Hansen S, Andersson A (2008) Top Catal 50:52
16. Millet JMM, Roussel H, Pigamo A, Dubois JL, Jumas JC (2001)
Appl Catal A 232:77
17. Baca M, Aouine M, Dubois JL, Millet JMM (2005) J Catal
233:234
18. Nguyen TT, Deniau B, Baca M, Millet JMM (2011) Top Catal
54:650
19. Naraschewski FN, Jentys A, Lercher JA (2011) Top Catal 54:639
20. Safonova O, Deniau B, Millet JMM (2006) J Phys Chem B
110:23962
21. Nguyen TT, Aouine M, Millet JMM (2012) Catal Commun 21:22
22. Naraschewski FN, Kumar CP, Jentys A, Lercher JA (2011) Appl
Catal A 391:63
23. Deniau B (2007) Ph.D. Thesis 315
24. Baca M, Pigamo A, Dubois JL, Millet JMM (2005) Catal Com-
mun 6:215
25. Mekkia A, Khattaka GD, Wengerb LE (2009) J Electron Spec-
trosc Relat Phenom 175:21
26. Zhou WP, Kibler LA, Kolb DM (2002) Electrochim Acta
47:4501
27. Detty MR, Lenhart WC (1989) Organometallics 8:861
28. Birchall T, Sleight AW (1976) Inorg Chem 15:868
29. Kubo J, Watanabe N, Ueda W (2008) Chem Eng Sci 63:1648
30. Schloegl R (2011) Top Catal 54:10
31. Balcells E, Borgmeier F, Grisstede I, Lintz HG, Rosowski F
(2004) Appl Catal A 266:211
32. Oshihara K, Hisano T, Ueda W (2001) Top Catal 15:153
33. Havecker M, Wrabetz S, Krohnert J, Csepei L, Naumann R,
d’Alnoncourt R, Kolen’ko Y, Girgsdies F, Schlogl R, Trunschke
A (2012) J Catal 285:48
34. Grasselli RK, Centi G, Trifiro F (1990) Appl Catal A 57:149
35. Huynh Q, Shuurmann Y, Delichere P, Loridant S, Millet JMM
(2009) J Catal 261:166
36. Kaddouri A, Mazzocchi A, Tempesti E (1999) Appl Catal A
180:271
Top Catal
123