Effect of the support on the catalytic properties of vanadyl phosphate in the oxidative dehydrogenation of propane

Article abstract of DOI:10.1016/j.molcata.2010.06.017

Propane oxidative dehydrogenation has been investigated on vanadyl orthophosphate dispersed on different supports (Al₂O₃, TiO₂, SiO₂, ZrO₂, MgO and SnO) in a fixed bed reactor in the temperature range 300-550 °C. Catalysts have been characterized by bulk and surface sensitive techniques, such as X-ray diffraction (XRD), BET analysis, temperature programmed reduction with H ₂ (H₂-TPR) and X-ray photoelectron spectroscopy (XPS). A different dispersion of vanadium species has been found depending on the different nature of the supports. Al₂O₃, ZrO₂ and TiO₂ promote the best dispersion of vanadium and provide the highest catalytic performances, due to the strong interaction of the support with the phosphate phase and the enhanced vanadium reducibility. A different reaction path is activated by catalysts supported on SiO₂, MgO and SnO due to the formation of vanadyl phosphate aggregates and in some cases also to the reducibility of the support which favour propane overoxidation leading to a larger CO₂ production.

Full text of DOI:10.1016/j.molcata.2010.06.017

Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Effect of the support on the catalytic properties of vanadyl phosphate in the
oxidative dehydrogenation of propane
a
b,∗
b
c
c
M.P. Casaletto , G. Landi , L. Lisi , P. Patrono , F. Pinzari
a
Istituto per lo Studio dei Materiali Nanostrutturati, CNR, Via U. La Malfa 153, 90146 Palermo, Italy
Istituto di Ricerche sulla Combustione, CNR, P.le Tecchio 80, 80125 Naples, Italy
Istituto di Metodologie Inorganiche e dei Plasmi, CNR, Monterotondo Scalo, Italy
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
Propane oxidative dehydrogenation has been investigated on vanadyl orthophosphate dispersed on dif-
Received 15 January 2010
Received in revised form 8 June 2010
Accepted 16 June 2010
ferent supports (Al2O3, TiO2, SiO2, ZrO2, MgO and SnO) in a fixed bed reactor in the temperature range
◦
3
00–550 C. Catalysts have been characterized by bulk and surface sensitive techniques, such as X-ray
diffraction (XRD), BET analysis, temperature programmed reduction with H2 (H2-TPR) and X-ray photo-
electron spectroscopy (XPS). A different dispersion of vanadium species has been found depending on
the different nature of the supports. Al2O3, ZrO2 and TiO2 promote the best dispersion of vanadium and
provide the highest catalytic performances, due to the strong interaction of the support with the phos-
phate phase and the enhanced vanadium reducibility. A different reaction path is activated by catalysts
supported on SiO2, MgO and SnO due to the formation of vanadyl phosphate aggregates and in some
cases also to the reducibility of the support which favour propane overoxidation leading to a larger CO2
production.
Available online 23 June 2010
Keywords:
Propane
Oxidative dehydrogenation
Vanadium phosphate
©
2010 Elsevier B.V. All rights reserved.
1
. Introduction
reduced because of the intrinsic exothermicity of the ODH reaction;
ii) there is no thermodynamic limitation due to the formation of
(
During the last years the interest for the selective oxidation
water and (iii) no coke is produced (or if produced is, then, burnt
by oxygen). On the other hand, ODH reaction, like all selective
oxidations, suffers from selectivity problems related to the higher
reactivity of the product with respect to the reactant, leading to
the formation of COx [2,3]. From a general point of view, a selec-
tive oxidation like ODH is promoted by two different factors: (1)
oxygen activation, performed by a redox site, and (2) alkane acti-
vation, performed by an acid site through an acid–base interaction
with the reactant [4]. Even if the contemporary presence of both
types of sites is necessary, Ai underlined that the balance between
redox and acid sites is fundamental in order to avoid non-selective
pathways, leading to overoxidation or cracking products [4].
Two main classes of catalysts have been proposed for the ODH of
propane, both of them containing vanadium as the active phase. The
first is constituted by microporous and mesoporous materials with
structured or extra-lattice vanadium [5–14]. In this case the catalyst
contains a low amount of vanadium, that is well dispersed. The good
“site isolation”, i.e. the formation of isolated active atoms or small
clusters or complexes, is expected to provide an enhanced selec-
tivity [15]. Microporous frameworks could be considered intrinsic
of light hydrocarbons has remarkably grown. As a matter of fact,
selective oxidations could be used in order to produce olefins and
oxygenated products like alcohols, carboxylic acids, anhydrides
through less expensive processes with lower environmental impact
than more traditional processes [1].
Propylene is mainly produced by steam cracking and cat-
alytic dehydrogenation. Non-catalytic steam cracking is widely
used in order to produce light olefins (ethylene and propylene).
Generally by-products, such as methane, diolefins, acetylene, are
co-produced and costly separated from the desired olefins. More-
over, steam cracking needs a large heat supply, due to its high
endothermicity and to the elevated operating temperatures (about
◦
8
50 C). Although more selective, catalytic dehydrogenation is
highly endothermic and is limited by coke production that requires
a regeneration step of the deactivated catalyst.
The oxidative dehydrogenation (ODH) of propane is a promising
alternative route for the production of propylene, due to sev-
eral advantages. In particular, (i) heat supply can be significantly
“single-site” catalysts, due to the possibility to introduce metal ions
into well-defined positions, but, generally, these catalysts do not
show high activity due to the low amount of vanadium sites [16].
The second class of catalysts consists of supported or unsup-
ported vanadia or vanadates (usually magnesium vanadates)
∗ Corresponding author. Tel.: +39 0817682233; fax: +39 0815936936.
E-mail addresses: mariapia.casaletto@ismn.cnr.it (M.P. Casaletto),
landi@irc.cnr.it (G. Landi), lisi@irc.cnr.it (L. Lisi), pasquale.patrono@imip.cnr.it (P.
Patrono), fulvia.pinzari@imip.cnr.it (F. Pinzari).
1
381-1169/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2010.06.017

M.P. Casaletto et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
51
◦
[
17–28]. These catalysts generally show relatively high conversion
digested for 1 h, washed with distilled water, filtered, dried at 90 C
◦
of propane, depending on the amount of the active phase and the
reaction conditions, but propylene yields are limited by a low selec-
tivity to the desired product. In order to obtain better catalytic
performance, modifications of the physico-chemical features of the
V-containing oxide have been induced by changing catalyst and/or
support composition. The use of a TiO2 layer deposited by grafting
overnight and subsequently crushed and calcined at 600 C for 3 h.
Finally, SnO2 was prepared as described in [47]. A 1 M urea
(Sigma ACS 99.0–100.5%) aqueous solution was added to a 0.01 M
SnCl ·5H O (Carlo Erba 98%) aqueous solution and the mixture was
4
2
◦
heated at 80 C under magnetic stirring; the ratio between urea and
tin containing solution volumes was 25. After 6 h the precipitate
was digested for 2 days, washed with distilled water, filtered, dried
[
18], the doping with K and/or P and transition metals [21,24,27]
◦
◦
can be mentioned as attempts to increase catalytic performances,
related to the change of redox and acid properties of the materials.
Bulk vanadyl pyrophosphate (VPO) is industrially used for the
production of maleic anhydride starting from n-butane [29–32]
and has been proposed for propane selective oxidation to acrylic
acid [33–36]. While increasing surface disorder and V /V ratio
are key factors in order to improve maleic anhydride selectiv-
ity [37,38], a higher crystallinity and the absence of V species
lead to better performance of VPO towards acrylic acid formation
at 90 C and calcined at 600 C for 3 h.
Each support was loaded with a different amount of VOP in order
to obtain the deposition of a theoretical monolayer; the proper
amount of VOPO4 was calculated taking into account the support
specific surface area, as reported in Table 1. The so calculated VOP
amounts were dissolved in a little quantity of water, in such a way
that the final volume result 2–3 times the support volume. The sup-
port was added to the solution, then the mixture was heated under
magnetic stirring in order to evaporate the liquid phase, avoiding
solution boiling. At the end the material was crushed, further dried
+5
+4
+5
[
35], suggesting that the effect of catalyst modifications on differ-
◦
◦
ent reactions is not trivial. Recently, bulk VPO, mainly containing
V
at 80 C, and calcined at 550 C in flowing air.
+4
, has been proposed as propane ODH catalyst [39], whereas
All these catalysts are indicated as VOP/M, where M represents
the metal ion of the oxide support.
supported vanadyl orthophosphates (VOP), principally constituted
by V+5, have been proven to be active and selective in ODH of
ethane, titania-supported samples providing the best catalytic per-
formances [40,41]. A deep investigation of these catalysts in ethane
ODH has been carried out [40–46], while no information is avail-
able on the activity in ODH of propane, which is intrinsically more
reactive than ethane and requires lower reaction temperatures.
A strong influence of the support on the properties of the VOP
catalysts have been found by XPS studies, which showed that oxides
In order to study the effect of the VOP loading, the ZrO2 support
was also loaded with a VOP amount corresponding to five layers;
this catalyst was prepared according to the procedure described
above by using a five times larger quantity of VOP. This catalyst is
indicated as VOP/Zr-5L.
X-ray diffraction patterns of calcined catalysts were recorded
by using a Philips PW 1100 diffractometer with Ni-filtered Cu K␣
radiation. Angular measurements (2ꢀ) were accurate to 0.05 . BET
◦
promoting a better VOP dispersion, such as TiO , deeply interact
with VOP favouring the formation of V(IV) and metal–phosphates
compounds [46].
surface areas were measured by N2 adsorption at 77 K with a Carlo
Erba 1900 Sorptomatic.
2
Temperature programmed reduction with hydrogen (H -TPR)
2
The previous results encouraged us to extend our work to the
oxidative dehydrogenation of propane to propylene and, in partic-
ular, to focus the aim of this paper on the investigation of which
properties of the vanadyl orthophosphate/support system are fun-
was carried out by using a Micromeritics TPD/TPR 2900 analyser,
equipped with a thermal conductivity detector and coupled with a
Hiden HPR 20 mass spectrometer, operating with a 2% H /Ar mix-
2
3
−1
◦
◦
−1
ture (25 cm min flux) and a heating rate of 10 C min
630 C. Samples were treated in flowing air at 550 C for 2 h before
analysis.
up to
◦
damental in the selective C H8 activation and which modifications
3
of the active phase are induced by the interaction with the support.
In order to study that, vanadyl orthophosphate has been deposited
onto several supports (TiO , ␥-Al O , ZrO , SiO , MgO, and SnO )
The surface chemical composition of the samples was studied by
−
8
XPS in an ultrahigh vacuum chamber (base pressure ∼10 Torr).
Photoemission spectra were collected by a VG Microtech ESCA 3000
Multilab spectrometer, equipped with a standard Al K␣ excita-
tion source (hꢁ = 1486.6 eV), a nine-channeltrons detection system
and a hemispherical analyser operating at a constant pass energy
of 20 eV. The binding energy (BE) scale was calibrated by mea-
suring C 1s peak (BE = 285.1 eV) from the surface contamination
and the accuracy of the measure was ± 0.1 eV. A non-linear least-
square peak fitting routine was used for the analysis of XPS spectra,
separating elemental species in different oxidation states. Relative
concentrations of chemical elements were calculated by a standard
quantification routine, including Wagner’s energy dependence of
attenuation length [48] and a standard set of VG Escalab sensitivity
factors.
2
2
3
2
2
2
and tested in the propane ODH reaction. The catalytic activity has
been related to the physico-chemical properties of the different
materials using several characterization techniques as XRD, BET,
H -TPR and XPS.
2
2
. Experimental
Vanadyl orthophosphate, VOPO ·2H O (VOP), was prepared as
4
2
described in [41].
␥
-Al O (CK-300 AKZO), anatase TiO2 (Eurotitania), amorphous
2 3
SiO2 (Sigma), ZrO , MgO and SnO2 were used as supports. ZrO2,
2
MgO and SnO were prepared by conventional precipitation meth-
2
ods.
Catalytic activity tests were carried out with the experimen-
tal apparatus described elsewhere [41], equipped with a fixed bed
quartz reactor operating under atmospheric pressure. The feed
composition was 5.2% C H8 and 2.6% O2 in a balance of N . The
In particular, ZrO2 was prepared by slowly adding a 7.6 M
ammonia aqueous solution (Carlo Erba RPE) to an equal volume of
0
.1 M ZrOCl ·8H O (Carlo Erba RPE) aqueous solution pre-heated
2
2
3
2
◦
◦
reaction temperature ranged from 300 to 500 C. The contact time
at 80 C under magnetic stirring for 15 h. The final pH was 10.
At the end the precipitate was digested for 4 h, washed with dis-
−
3
ranged from 0.003 to 0.08 g h N dm . The reaction products were
analysed with a Hewlett Packard series II 5890 gas-chromatograph,
◦
tilled water, filtered and dried at 90 C overnight. Thereafter it was
crushed and calcined at 600 C for 3 h.
◦
equipped with a thermal conductivity detector for O , CO and CO2
2
MgO was prepared in a similar way. A 6.4 M ammonia water
analysis and a flame ionization detector for hydrocarbons anal-
solution was slowly added to a 0.05 M Mg(NO ) ·6H O (Fluka
ysis. The concentrations of O , CO and CO2 were also measured
3
2
2
2
purum ≥99%) aqueous solution under magnetic stirring (without
heating); the ratio between ammonia and magnesium containing
solution volumes was 4.2. After 1 h stirring the precipitate was
on line with a Hartmann & Braun URAS 10 E continuous analyser.
Water produced during reaction was kept apart by a silica gel trap
in order to avoid condensation in the cold part of the experimen-

5
2
M.P. Casaletto et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
Table 1
Nominal VOP content, surface area of catalysts and supports and H2/V ratio evaluated from TPR experiments.
Support surface area (m2
g
−1
)
Catalyst surface area (m2
g
−1
)
H2/V from TPR
Fresh
Catalyst
Nominal VOP content (wt%)
React.
VOP-Ti
VOP-Al
VOP-Si
VOP-Zr
VOP-Mg
VOP-Sn
VOP-Zr-5L
9.6
14.0
14.6
3.6
7.8
2.6
125
190
200
49
105
34
125
183
162
43
56
31
0.71
0.67
0.51
1.07
0.63
1.0
0.63
0.55
–
0.86
–
–
–
18.2
49
27
0.71
tal apparatus. Carbon balance was closed within 3% error in all
experiments. Blank tests performed under the same experimental
conditions excluded the occurrence of homogenous reactions. The
results reported below are only catalytic or catalytically induced.
3
. Results
In Table 1 the nominal VOP content and the surface area of both
supports and catalysts are reported. A general reduction of the orig-
inal surface area of the support occurs upon deposition of VOP,
titania- and alumina-based catalysts showing the less significant
decrease. A further reduction of the surface area was observed in
the VOP-Zr-5L sample due to the great excess with respect to the
VOP monolayer content.
XRD patterns of selected samples are reported in Fig. 1. Among
monolayer catalysts the presence of segregated VOP phase is
detected on SnO supported sample, as shown by the main signal of
2
◦
vanadyl hydrogen phosphate hemihydrate at 2ꢀ = 15.48 [49]. SiO2
and MgO supported samples, whose spectra are not reported, show
a similar behaviour. On the contrary no evidence of VOP signals
has been detected on the other samples, suggesting the absence of
VOP segregation, as indicated by the patterns of TiO - and ZrO -
2
2
supported catalysts reported in Fig. 1. Weak signals of VOP were
also observed for VOP-Zr-5L.
All catalysts can convert propane although with different activ-
ity, while bulk VOP activity towards propane ODH is undetectable
under experimental conditions comparable to those used for sup-
ported catalysts. This behaviour is not surprising taking into
account our previous results on the ethane ODH [43]. Among sup-
ported samples, two main behaviours towards propane oxidation
can be detected under the investigated experimental conditions.
For VOP supported on titania, alumina, zirconia and silica the main
product is propylene in the whole range of investigated tempera-
tures whereas CO2 represents the most abundant product for MgO
and SnO2 supported catalysts. The behaviour of VOP/Mg is not
Fig. 1. XRD patterns of different supported VOP catalysts. Vertical lines indicates
main VOP reflections.
surprising; as a matter of fact, Klisinska et al. [27] reported that
magnesia supported vanadium exhibits higher selectivity to CO2
than propylene. In Fig. 2 the different behaviour of the two classes
of catalysts, represented by VOP/Ti (Fig. 2a) and VOP/Sn (Fig. 2b),
respectively, is compared. For VOP/Ti the decrease of propylene for-
mation is mainly balanced by the oxidation to CO and CO , as also
2
reported for ethane ODH [41]. For VOP/Sn formation of CO slightly
2
decreases with temperature, whereas production of CO is activated
Fig. 2. Propane conversion (᭹) and propylene (ꢀ), CO (ꢂ) and CO2 (♦) selectivity as a function of reaction temperature on VOP/Ti (a) and VOP/Sn (b).

M.P. Casaletto et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
53
Table 2
Values of propane converted per mass and SSA of catalyst and per vanadium atom
◦
(
TOF) and of propylene produced per vanadium atom (TOF2) at 450 C.
Catalyst
Reaction
ratea
Reaction
rateb
TOFc
TOF2d
Active sitese
VOP-Ti
10.3
4.7
2.2
0.2
0.6
0.3
4.6
8.2
2.5
4.5
0.1
0.6
0.9
9.4
1.7
0.6
1.0
0.03
0.1
0.2
0.4
0.94
0.36
0.66
0.02
0.02
0.02
0.26
2.86
2.85
3.35
3.11
5.18
3.12
25.1
VOP-Al
VOP-Zr
VOP-Si
VOP-Mg
VOP-Sn
VOP-Zr-5L
a
−1 −1
6
MolC3H8 g
s
× 10 .
b
−2 −1
4
MolC3H8 m
s
× 10 .
c
−1 −1
−1 −1
2
MolC3H8 molV
MolC3H6 molV
s
× 10 .
d
e
2
s
× 10 .
−2
18
Number m × 10 (on the basis of catalyst surface area).
◦
only at T ≥ 450 C, globally resulting in a decrease of propylene
selectivity with propane conversion. In Table 2 the values of reac-
◦
tion rate evaluated at 450 C and referred to catalyst weight, surface
Fig. 4. Propylene selectivity as a function of propane conversion for VOP/M catalyst
(
area and vanadium content (Turnover frequency – TOF), under the
hypothesis of differential reactor, are reported. VOP/Ti shows the
best performance, whereas VOP/Si, VOP/Mg and VOP/Sn show an
activity at least one order of magnitude lower. Only in terms of spe-
cific surface area VOP/Zr-5 shows an activity comparable to that of
VOP/Ti, but with a vanadium content about double.
The high activity of VOP/Ti is confirmed by comparing the per-
formance of the most active catalysts as a function of contact time
Fig. 3) at a fixed temperature of 450 C. Very low contact times
are required by VOP/Ti in order to convert propane; in particular,
0% propane conversion with 60% propylene selectivity is obtained
at only about 4 ms residence time, calculated at the reaction tem-
perature. This results in a very high TOF, as reported in Table 2.
The difference among catalysts behaviour is not attributable to the
generation of different active sites onto the different supports. As a
matter of fact, for these catalysts the selectivity to propylene seems
basically related to propane conversion independently of the sup-
port (Fig. 4), thus suggesting that the active sites have the same
nature, while the active species of low activity catalysts appears
different, corresponding points in Fig. 4 being far from the line asso-
ciated to the best catalysts. The best active centres are also able to
activate both propane and propylene, and, as a consequence, at high
◦ −3
T = 450 C; W/F = 0.003–0.08 g h N dm ).
be noted, the two plots show a different slope. In particular, in
the multilayer catalyst propylene selectivity extrapolated to zero
conversion approaches 100%, whilst is about 80% for the mono-
layer VOP/Zr catalyst. On the other hand, in this catalyst propylene
selectivity is less affected by the increase of propane conversion,
thus suggesting the presence of sites promoting a different reaction
pathway. It seems that on the monolayer VOP/Zr propylene oxida-
tion to COx is less temperature-sensitive, thanks to the formation of
different active sites, which prevent propylene overoxidation. TOF
values of the two zirconia-based catalysts, reported in Table 2, indi-
cates that some VOP segregation should take place in the VOP/Zr-5L
sample, which markedly reduces the exposure of vanadium species,
as confirmed by XRD analysis.
The relative surface chemical composition of the samples deter-
mined by XPS is reported in Table 3. Elemental concentrations are
expressed as atomic percentage (at.%) and calculated both before
and after ODH test.
VOP/Ti catalyst exhibits the highest surface vanadium concen-
tration (see P/V ratios in Table 3), which could be related to its good
activity in propane ODH. Nevertheless, some surface phosphorous
enrichment occurs upon reaction test.
◦
(
1
contact times, i.e. at higher C H8 conversion, the olefin is further
3
oxidized to COx (Fig. 4).
In Fig. 5 the effect of VOP loading on zirconia support is reported
in terms of propane conversion versus propylene selectivity. As can
Fig. 5. Propylene selectivity as a function of propane conversion evaluated at dif-
−
3
Fig. 3. Propane conversion (full symbols) and propylene selectivity (open symbols)
at 450 C as a function of contact time for VOP/Ti, VOP/Al and VOP/Zr catalysts.
ferent temperatures for VOP/Zr and VOP/Zr-5L catalysts (W/F = 0.02–0.08 g h dm
T = 300–550 C).
;
◦
◦

5
4
M.P. Casaletto et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
Table 3
XPS relative surface chemical composition (expressed in atomic %) of catalysts after
calcination and ODH tests.
Catalyst
V 2p3/2
O 1s
P 2p
Support
ion
P/V
O/P
VOP/Ti
Dried
Calcined at 550
2.7
2.5
2.0
75.5
78.4
71.8
2.7
2.6
3.1
19.2
16.5
23.2
1.0
1.0
1.6
28.0
30.2
23.2
◦
C
◦
◦
◦
Propane ODH at 450
C
C
C
VOP/Al
Dried
Calcined at 550
Propane ODH at 450
1.8
1.3
1.2
61.0
60.8
65.4
2.5
2.1
1.8
34.7
35.8
31.5
1.4
1.6
1.5
24.4
29.0
36.3
◦
C
VOP/Zr
Dried
Calcined at 550
Propane ODH at 450
1.7
1.9
1.5
70.4
67.5
69.2
2.7
2.9
2.4
25.2
27.7
26.9
1.6
1.5
1.6
26.1
23.3
28.8
◦
C
VOP/Zr-5L
Dried
Calcined at 550
5.1
5.2
70.3
69.4
8.3
7.3
16.4
18.1
1.6
1.4
8.5
9.5
◦
C
Also the O/P atomic ratio is reported in Table 3, even if the
investigated samples consisted of vanadyl phosphates supported
on different oxidic matrixes (and not pure compounds) that under-
went different interactions with the support through the oxygen
atom. Since XPS detects only the surface chemical composition of
the samples (not necessarily corresponding to the bulk), the sto-
ichiometric ratio of the pure vanadyl phosphate phase cannot be
found and, as determined by XPS quantitative analysis in Table 3,
an extraordinarily predominance of oxygen atoms is detected (O/P
ratio ∼25–30). Significant is the reduction of the O/P ratio in the
VOP/Zr-5L sample where the amount of vanadyl phosphates on the
surface of ZrO2 exceeded so much the monolayer coverage.
XPS profiles of the V 2p3/2 photoelectron peak in VOP/Ti, VOP/Al
and VOP/Zr samples are reported in Fig. 6(a)–(c), respectively. At
a first visual inspection, a shoulder at high binding energy value is
clearly distinguishable in the V 2p3/2 spectrum of VOP/Ti sample
in Fig. 6a. This shoulder is totally absent in the XPS profile of the
VOP/Zr sample (Fig. 6b).
Fig. 6. XPS profiles of VOP/Ti (a), VOP/Al (b) and VOP/Zr (c) catalysts in the energy
range of V 2p3/2 photoelectron peak.
vanadium phosphate, such as in VOP/Zr catalysts. In this case the
calcination process favours the oxidation of V(IV) oxide species to
V(V) oxide, as expected under oxidizing atmosphere.
Zr 3d peak is located at a quite constant binding energy value
in both VOP/Zr catalysts (BE = 183.1 eV), attributable to Zr4+ ions
as in ZrO2 [48]. This was further confirmed by the curve-fitting of
O 1s peak, which can be deconvoluted into three components at
BE = 530.5, 531.7 and 533.2 eV, assigned to oxide, hydroxide and/or
phosphate species, and to adsorbed water, respectively [50]. The
main component of the O 1s peak (75% peak area) is that related
to oxide species (BE = 530.5 eV), The predominance of the oxide
support peak in the XPS spectrum could be attributed to a bad
VOP surface distribution with a consequent exposure of significant
fraction of the support surface.
The results of the curve-fitting procedure of the V 2p3/2 pho-
toelectron peak in VOP/Ti and VOP/Al samples evidenced the
presence of three components at BE = 518.2, 516.9 and 515.8 eV,
5+
which can be assigned to V phosphate species as detected in
5
+
4+
VOPO , to V oxide species as in V O5 and to V oxide species
4
2
as in V O , respectively [41,50]. In the VOP/Zr sample only the last
2
4
two components are present. As reported in [45], dispersion of VOP
on TiO or Al O results in the formation of a significant amount of
2
2
3
5+
4+
V
and V oxides, since a fraction of phosphorous preferentially
In the dried multilayer VOP/Zr-5L sample the binding energy of
Zr 3d shifts towards a slightly higher value that could hint to an
initial interaction between VOP and ZrO2 finally leading to the for-
mation of Zr(HPO ) species [52]. The main component of O 1s peak
interacts with the support leading to the formation of titanium
or aluminium phosphate, respectively. A fraction of V4 species
is formed even at low temperature in the dried samples. Calcina-
+
4
2
◦
−
tion at 550 C under oxidizing atmosphere surprisingly promotes
was located at BE = 531.7 eV, corresponding to the presence of OH
a further reduction of vanadium, as evidenced by the increase of
the V4+ fraction, as also reported in [46]. This reduction is accom-
panied by the interaction of phosphorous with the support ions,
leading to the formation of a larger fraction of titanium or alu-
minium phosphate, and by the decrease of the amount of vanadium
phosphate (Table 4). In other words, calcination mainly favours
the migration of phosphate ions from vanadium to the support,
generating an additional fraction of V(IV) oxide. The interaction
groups and O–P bonds, as expected due to the greater VOP concen-
tration of this catalyst. Nevertheless, formation of oxides species
prevails on that of phosphates upon calcination, as shown by the
lower binding energy value of Zr 3d peak (as in the monolayer cat-
alyst) and the predominance of the component attributed to the
oxide (BE = 530.5 eV) in the curve-fitting of the O 1s peak (62% peak
area).
Redox properties of these materials are deeply involved in the
reaction mechanism [44,45]. Therefore, a TPR analysis was carried
out in order to explain the large difference of activity observed
for the investigated samples. In Fig. 7 the TPR profiles of VOP/Ti,
VOP/Al and VOP/Zr are compared with that of bulk VOP. Disper-
sion of VOP results into a significant enhancement of reducibility,
which is maximum for TiO2 support. On the other hand, reduction
of phosphorous oxide with both V O5 and TiO2 was also reported
2
by Deo and Wachs [51] who observed the formation of VOPO4
by addition of P O5 to V O5/TiO catalyst and the formation of
2
2
2
vanadium–oxygen–phosphorous bonds by deposition of V O5 to
2
a previously prepared P O5/TiO2 sample. This phenomenon was
2
not observed for supports which do not promote the formation of

M.P. Casaletto et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
55
Table 4
XPS relative surface distribution (expressed as peak area %) of different vanadium species in catalysts after calcination and ODH tests.
Catalyst
V5+ phosphate
V5+ oxide
V4+ oxide
Support ion oxide
Support ion phosphate
VOP/Ti
Dried
Calcined at 550
13.8
13.2
7.6
49.3
42.5
61.4
36.9
44.3
31.0
83.7
65.3
23.8
16.3
34.7
76.2
◦
C
◦
◦
◦
Propane ODH at 450
C
C
C
VOP/Al
Dried
Calcined at 550
Propane ODH at 450
40.0
31.1
20.3
56.0
52.4
56.3
4.0
16.5
23.4
∼100
93.2
19.8
n.da
6.8
80.2
◦
C
VOP/Zr
Dried
Calcined at 550
Propane ODH at 450
n.da
49.5
59.6
44.1
50.5
40.4
55.9
∼100
∼100
∼100
n.d
a
◦
a
a
C
n.d
n.d
n.d
a
a
n.d
a
Not detectable.
of 1 expected for the reduction from V5+ to V3+, are in good agree-
ment with the presence of a fraction of vanadium in 4+ oxidation
state, asfoundbyXPSanalysis. TheH uptake oftheZrO -supported
of VOP/Sn and VOP/Mg catalysts (Fig. 8) mainly takes place dur-
ing the isothermal step of the experiment (630 C), thus suggesting
the presence of hardly reducible bulk-like VOP. Moreover, as sug-
gested by the TPR profile of VOP/Sn, a significant support reduction
◦
2
2
monolayer catalyst can be affected by a slight support reduction,
4+
occurs, not allowing a precise calculation of H /V ratio (Table 1). On
thus leading to H /V higher than 1, even in the presence of a V
2
2
the other hand, this catalyst shows combustion rather than ODH
activity, strongly suggesting that the support contributes to COx
formation. These results confirm that vanadium reducibility is a
key feature for hydrocarbon selective oxidation, while other metal
fraction as revealed by XPS measurements. The exposure to the
reacting environment produces a further reduction, i.e. an increase
of the V4+ fraction, as reported in Table 1, suggesting once more
that vanadium redox cycles are responsible for propane ODH.
Moreover, XPS analysis carried out on the catalysts after the
sites are active in non-selective pathways. The H /V ratio of the
2
◦
best catalysts, except VOP/Zr, lower than the stoichiometric value
catalytic test at 450 C (Table 4) suggests that the ODH reaction
promotes a redistribution of surface species. For both VOP/Ti and
VOP/Al a fraction of vanadium phosphate, initially present in the
fresh sample, is transformed into vanadium oxide and, at the same
time, the fraction of titanium or aluminium phosphate increases,
thus suggesting that the catalytic tests promotes a further migra-
tion of phosphorous from vanadium to the support. Since the
catalytic test was carried out at a lower temperature with respect
to that of calcination, these results suggest that the interaction of
phosphate ion with the support is favoured by a redox cycle of vana-
dium rather than by high temperatures. Due to the low amount
of active phase, the identification of phosphate species, related to
the superficial phosphorus (Table 3), on VOP/Zr surface resulted
difficult.
4. Discussion
The results reported above clearly demonstrate that the activity
of VOP-based catalysts strongly depends on the support. In par-
ticular, the formation of superficial phosphates including metal
phosphates of the support, as detected by XPS measures, seems
the key factor to obtain highly active vanadium species. Indeed,
the samples showing this strong interaction between the active
phase and the support are characterized by very low reduction of
the surface area after VOP deposition, suggesting the formation of
very highly dispersed vanadium-containing species undetectable
by XRD analysis. On the contrary, the samples providing the worst
performances, are those in which most vanadium centres are inac-
tive because of the formation of VOP aggregates, as detected by
XRD, causing significant surface area reduction as well. Due to the
typical redox nature of the investigated reaction, another key point
is the increased reducibility of vanadium centres, as revealed by
TPR measures; the increased reducibility, in fact, is directly related
to the enhanced TOF showed by highly active catalysts. Moreover,
this behaviour appears more related to the high dispersion of the
vanadia species than to the reducibility of the support; as a matter
of fact, the high reducibility of VOP/Sn is not coupled with a high
ODH activity, but more related to a undesired combustion activity.
The behaviour of multilayer VOP/Zr can be explained according to
Fig. 7. TPR profiles of VOP/Ti, VOP/Al and VOP/Zr catalysts compared with that of
bulk VOP.
Fig. 8. TPR profile of VOP/Sn and VOP/Mg catalysts

5
6
M.P. Casaletto et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 50–56
the above considerations; the presence of larger amounts of active
phase leads to a better coverage of the surface, avoiding the propane
reaction with the ZrO2 centres directly, and, as a consequence, the
selectivity towards ODH reaction approaches 100% for zero conver-
sion. On the other hand, the higher reducibility of larger vanadium
cluster obtained on the multilayer catalyst favours the propylene
overoxidation and results in lower catalytic performance at rela-
tively higher propane conversion. Finally, the highest TOF are not
related to the highest support surface area, as revealed by the low
performance of VOP/Si. In conclusion, the high dispersion of the
active phase and, as a consequence, the increased reducibility and
[11] A. Brueckner, P. Rybarczyk, H. Kosslick, G.-U. Wolf, M. Baerns, Stud. Surf. Sci.
Catal. 142B (2002) 1141.
[
12] Y.M. Liu, Y. Cao, K.K. Zhu, S.R. Yan, W.L. Dai, H.Y. He, K.N. Fan, Chem. Commun.
23 (2002) 2832.
[13] Y.M. Liu, Y. Cao, N. Yi, W.L. Feng, W.L. Dai, S.R. Yan, H.Y. He, K.N. Fan, J. Catal.
24 (2004) 417.
2
[
[
14] B. Solsona, J.M. López Nieto, U. Díaz, Micropor. Mesopor. Mater. 94 (2006) 339.
15] J.-C. Volta, Top. Catal. 15 (2–4) (2001) 121.
[16] J.M. Thomas, R. Raja, Top. Catal. 40 (2006) 3.
[17] F. Genser, S. Pietrzyk, Chem. Eng. Sci. 54 (1999) 4315.
[
18] R. Monaci, E. Rombi, V. Solinas, A. Sorrentino, E. Santacesaria, G. Colon, Appl.
Catal. A: Gen. 214 (2001) 203.
[19] B.Y. Jibril, S.M. Al-Zahrani, A.E. Abasaeed, Catal. Lett. 74 (2001) 145.
[20] A. Comite, A. Sorrentino, G. Capannelli, M. Di Serio, R. Tesser, E. Santacesaria, J.
Mol. Catal. A: Chem. 198 (2003) 151.
TOF towards C H8 ODH can be uniquely related to the capabil-
3
[
[
21] G. Garcia Cortez, J.L.G. Fierro, M.A. Ba n˜ ares, Catal. Today 78 (2003) 219.
22] M.J. Holgado, F.M. Labajos, M.J.S. Montero, V. Rives, Mater. Res. Bull. 38 (2003)
1879.
ity of the support to strongly interact with the active phase. As
it can be derived from the above results and considerations, the
active phase obtained on our samples is constituted by small clus-
ters of vanadium oxide; this is the same active phase obtained for
vanadia supported catalysts [53]. With respect to these catalysts,
VOP-based ones show about the same activity towards propane but
an enhanced selectivity towards propylene. This is due to the better
dispersion of vanadium centres induced by the formation of metal
phosphates of the support, leading to more selective active centres.
[
[
23] S. Sugiyama, T. Hashimoto, N. Shigemoto, H. Hayashi, Catal. Lett. 89 (2003) 229.
24] A. Klisi n´ ska, A. Haras, K. Samson, M. Witko, B. Grzybowska, J. Mol. Catal. A:
Chem. 210 (2004) 87.
[25] J.D. Pless, B.B. Barsin, H.-S. Kim, D. Ko, M.T. Smith, R.R. Hammond, P.C. Stair, K.R.
Poeppelmeier, J. Catal. 223 (2004) 419.
[
26] A. Christodoulakis, M. Machli, A.A. Lemonidou, S. Boghosian, J. Catal. 222 (2004)
93.
[27] A. Klisi n´ ska, K. Samson, I. Gressel, B. Grzybowska, Appl. Catal. A: Gen. 309 (2006)
0.
28] G. Martra, F. Arena, S. Coluccia, F. Frusteri, A. Parmaliana, Catal. Today 63 (2–4)
2000) 197.
[29] R.M. Contractor, US Patent 4,668,802 (1987), to DuPont.
2
1
[
(
5
. Conclusions
[
30] G.D. Suciu, G. Stefani, C. Fumagalli, US Patent 4,511,670 (1985), to Lummus
Crest Inc. and Alusuisse Italia SpA.
The study of the oxidative dehydrogenation of propane has
been conducted over vanadyl orthophosphate based catalysts sup-
[
31] H. Taheri, US Patent 5,011,945 (1991), to Amoco Co.
[32] G.K. Kwentus, M. Suda, US Patent 4,501,907 (1985), to Monsanto Co.
[33] G. Landi, L. Lisi, J.-C. Volta, Chem. Commun. 4 (2003) 492.
ported onto TiO , ␥-Al O , ZrO , SiO , MgO and SnO . According
2
2
3
2
2
2
[
[
34] G. Landi, L. Lisi, J.-C. Volta, Catal. Today 91–92C (2004) 275.
35] G. Landi, L. Lisi, J.-C. Volta, J. Mol. Catal. A: Chem. 222 (2004) 175.
to XRD, BET and XPS measurements, the best performing catalysts,
supported on TiO , ZrO and ␥-Al O , have a good active phase dis-
[36] G. Landi, L. Lisi, G. Russo, J. Mol. Catal. A: Chem. 239 (2005) 172.
[37] M. Abon, K.E. Bere, A. Tuel, P. Delichere, J. Catal. 156 (1995) 28.
2
2
2
3
persion and a strong interaction between phosphate and substrate,
evidenced by low reduction in the surface area upon VOP deposition
and the formation of a well dispersed superficial vanadium oxide
layer in addition to vanadium phosphate. On the contrary, VOP seg-
regation is correlated to a low activity and to the formation of CO2
[
[
[
38] K. Aït-Lachgar, A. Tuel, M. Brun, J.M. Herrmann, J.M. Krafft, J.R. Martin, J.C. Volta,
M. Abon, J. Catal. 177 (1998) 224.
39] S. Arias-Pérez, R. García-Alamilla, M.G. Cárdenas-Galindo, B.E. Handy, S. Robles-
Andrade, G. Sandoval-Robles, Ind. Eng. Chem. Res. 48 (2009) 1215.
40] M.P. Casaletto, S. Kaciulis, L. Lisi, G. Mattogno, A. Mezzi, P. Patrono, G. Ruoppolo,
Appl. Catal. A: Gen. 218 (2001) 129.
as the main product. H -TPR and XPS measurements allowed us to
2
[41] M.P. Casaletto, L. Lisi, G. Mattogno, P. Patrono, G. Ruoppolo, G. Russo, Appl. Catal.
A: Gen. 226 (2002) 41.
correlate the catalytic activity to the enhanced vanadium reducibil-
ity, suggesting that an easier reducibility of the active phase is
fundamental for high propylene production.
[
42] P. Ciambelli, L. Lisi, P. Patrono, G. Ruoppolo, G. Russo, Catal. Lett. 82 (3–4) (2002)
43.
2
[
[
43] L. Lisi, P. Patrono, G. Ruoppolo, J. Mol. Catal. A: Chem. 204–205 (2003) 609.
44] M.P. Casaletto, L. Lisi, G. Mattogno, P. Patrono, G. Ruoppolo, Appl. Catal. A: Gen.
2
67 (1–2) (2004) 157.
References
[
[
45] M.P. Casaletto, L. Lisi, G. Mattogno, P. Patrono, F. Pinzari, G. Ruoppolo, Catal.
Today 91–92 (2004) 271.
46] M.P. Casaletto, L. Lisi, G. Mattogno, P. Patrono, G. Ruoppolo, Surf. Interface Anal.
[
[
[
[
[
[
1] F. Cavani, F. Trifirò, Catal. Today 51 (1999) 561.
2] J.M. Thomas, Angew. Chem. Int. Ed. 38 (1999) 3588.
3] J.M. López Nieto, Top. Catal. 41 (1–4) (2006) 3.
4] M. Ai, Kinet. Catal. 44 (2) (2003) 198.
5] G. Centi, F. Trifirò, Appl. Catal. A: Gen. 143 (1996) 3.
6] A. Kubacka, E. Wloch, B. Sulikowski, R.X. Valenzuela, V. CortèsCorberàn, Catal.
Today 61 (2000) 343.
36 (8) (2004) 737.
[
[
[
[
47] K.C. Song, Y. Kang, Mater. Lett. 42 (2000) 283.
48] C.D. Wagner, L.E. Davis, W.M. Riggs, Surf. Interface Anal. 2 (1986) 53.
49] A.A. Rownaghi, Y.H. Taufiq-Yap, F. Rezaei, Chem. Eng. J. 155 (2009) 514.
50] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, in: J. Chastain, R.C. King Jr.
(
Eds.), Handbook of X-Ray Photoelectron Spectroscopy, Phys. Electronics Inc.,
[
[
7] Z.-S. Chao, E. Ruckenstein, Catal. Lett. 94 (2004) 217.
8] B. Solsona, T. Blasco, J.M. López Nieto, M.L. Pe n˜ a, F. Rey, A. Vidal-Moya, J. Catal.
Eden Prairie, USA, 1995.
[
[
51] G. Deo, I.E. Wachs, J. Catal. 146 (1994) 335.
52] L. Lisi, G. Ruoppolo, M.P. Casaletto, P. Galli, M.A. Massucci, P. Patrono, F. Pinzari,
J. Mol. Catal. A: Chem. 232 (1–2) (2005) 127.
2
03 (2001) 443.
[9] J. Santamaria-González, J. Luque-Zambrana, J. Mérida-Robles, J. Maireles-
Torres, E. Rodríguez-Castellón, A. Jiménez-López, Catal. Lett. 68 (2000) 67.
[
53] P. Viparelli, P. Ciambelli, L. Lisi, G. Ruoppolo, G. Russo, J.C. Volta, Appl. Catal. A
[10] E.V. Kondratenko, M. Cherian, M. Baerns, D. Su, R. Schlögl, X. Wang, I.E. Wachs,
184 (1999) 291.
J. Catal. 234 (2005) 131.