The design of a new ZrO2-supported V/P/O catalyst for n-butane oxidation to maleic anhydride: The build-up of the active phase during thermal treatment

Article abstract of DOI:10.1016/j.cattod.2010.02.014

This study concerns the generation of the active V/P/O phase in supported catalysts used for n-butane oxidation to maleic anhydride. Catalysts were prepared by impregnating a zirconia support with ammonium vanadate and phosphoric acid, with subsequent thermal treatment under various environments and temperatures. The catalysts were studied by means of in situ Raman spectroscopy and reactivity measurements. It was found that the nature of the active surface is greatly affected both by the composition of the gas phase used for the thermal treatment, and by the P/V ratio used for catalyst preparation. δ-VOPO₄ was initially formed by the high-temperature reaction of vanadium salt with phosphoric acid; however, in the sample prepared with only 10% excess phosphorus (P/V atomic ratio = 1.1), δ-VOPO₄ transformed into either V₂O₅, in a steam-enriched air stream, or α_II- and β-VOPO₄, in a dry air stream. δ-VOPO₄ was more stable in the sample prepared with a large excess of P, i.e., with P/V = 1.3 and 1.5. The latter catalyst showed the best catalytic performance in n-butane oxidation. However, the performance was worse than that of the commercial unsupported vanadyl pyrophosphate catalyst.

Full text of DOI:10.1016/j.cattod.2010.02.014

Catalysis Today 157 (2010) 204–210
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
The design of a new ZrO2-supported V/P/O catalyst for n-butane oxidation to
maleic anhydride: The build-up of the active phase during thermal treatment
A. Caldarellia, F. Cavania,∗,1, F. Folcoa, S. Luciania, C. Cortellib, R. Leanzab
a Dipartimento di Chimica Industriale e dei Materiali, ALMA MATER STUDIORUM, Università di Bologna, Viale Risorgimento 4, 40136 Bologna, Italy
b Polynt SpA, Via E. Fermi 51, 24020 Scanzorosciate (BG), Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 17 March 2010
This study concerns the generation of the active V/P/O phase in supported catalysts used for n-butane
oxidation to maleic anhydride. Catalysts were prepared by impregnating a zirconia support with ammo-
nium vanadate and phosphoric acid, with subsequent thermal treatment under various environments
and temperatures. The catalysts were studied by means of in situ Raman spectroscopy and reactivity
measurements. It was found that the nature of the active surface is greatly affected both by the compo-
sition of the gas phase used for the thermal treatment, and by the P/V ratio used for catalyst preparation.
␦-VOPO4 was initially formed by the high-temperature reaction of vanadium salt with phosphoric acid;
however, in the sample prepared with only 10% excess phosphorus (P/V atomic ratio = 1.1), ␦-VOPO4
transformed into either V2O5, in a steam-enriched air stream, or ␣II- and ␤-VOPO4, in a dry air stream.
␦-VOPO4 was more stable in the sample prepared with a large excess of P, i.e., with P/V = 1.3 and 1.5. The
latter catalyst showed the best catalytic performance in n-butane oxidation. However, the performance
was worse than that of the commercial unsupported vanadyl pyrophosphate catalyst.
Keywords:
Maleic anhydride
n-Butane oxidation
Vanadyl pyrophosphate
Vanadium phosphate
Zirconia
In situ Raman spectroscopy
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
nature of the compounds formed on the surface of VPP, that is still
the main component of the catalyst, is a function of the P/V ratio
Vanadium/phosphorus mixed oxides are catalysts for gas-phase
selective oxidation and ammoxidation of hydrocarbons [1–5]. A
great variety of V/P/O compounds exist that differ in terms of oxi-
dation states, degree of phosphate group condensation, and water
molecule content. It is also known that the transformation of one
V/P/O compound into another can be induced by thermal treat-
ments. The atmosphere at which the treatment is carried out is
a powerful tool that can be used to control such transformations
[6,7]. The most well known V/P/O system is vanadyl pyrophosphate
(VPP), the catalyst used to oxidize n-butane to maleic anhydride [1].
V/P/O compounds have a tendency to undergo changes as a con-
sequence of slight variations in either temperature or composition
of the atmosphere used for thermal treatments. Therefore, in the
past, great attention has been paid to the changes occurring during
the thermal treatment of VOHPO4·0.5H2O (the precursor of VPP)
[8] as well as to possible changes that the VPP may undergo under
reaction conditions. In previous studies, we found that the surface
composition of VPP catalysts is affected by the reaction conditions
[9–11]; during reaction a dynamic equilibrium is established at the
catalyst surface between various V/P/O compounds. However, the
adopted for the preparation of this compound. When an excess of
P is used with respect to the stoichiometric requirement, under
reaction conditions the surface of VPP is oxidized to ␦-VOPO4, and
the redox cycle occurs between V5+ in this latter compound and
V
4+ in VPP; this catalyst gives the optimal performance in n-butane
oxidation to maleic anhydride. Conversely, when a stoichiometric
P/V ratio is used, VPP is oxidized into non-selective ␣I-VOPO4 at
340–400 ◦C which, however, converts into the selective ␦-VOPO4
when temperatures higher than 400 ◦C are used [10,11]. ␦-VOPO4
reduces to VPP during reaction, and the amount of the prevailing
compound in the active phase is a function of both the reaction
conditions and the nature of the rate-determining step [12].
In the present study, we report on the attempt to build-up ␦-
VOPO4 on the surface of a zirconia-supported V/P/O catalyst. The
aim of this approach was twofold: (a) to confirm that ␦-VOPO4 is
the key-compound for the generation of an active and selective cat-
alyst for n-butane oxidation, and that catalyst composition affects
the type of V5+ phosphate that is formed at high temperature, and
(b) to find a method for the preparation of V/P/O-based catalysts
in which the dispersion of the latter onto an inert support may
lead to a material having lower cost compared to the conventional
unsupported VPP. Raman spectra, recorded in situ during the ther-
mal treatment, were used together with reactivity measurements
to characterize the catalyst surface, with the aim of determining
which parameters may influence the generation of the active phase.
∗
Corresponding author.
E-mail addresses: fabrizio.cavani@unibo.it, cavani@ms.fci.unibo.it (F. Cavani).
1
INSTM, Research Unit of Bologna, Italy.
0920-5861/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2010.02.014

A. Caldarelli et al. / Catalysis Today 157 (2010) 204–210
205
Fig. 1. Raman spectra of the sample prepared by impregnation of zirconia with NH4VO3 and H3PO4 (P/V atomic ratio 1.1) recorded in situ during heating from RT to 450 ◦
C
in 14.5% O2 in a N2 flow (figure on the left) and then during the isothermal step at 450 ◦C (figure on the right). Symbols: ␣II = ␣II-VOPO4; ␤ = ␤-VOPO4; ␦ = ␦-VOPO4; • = V2O5;
* = ZrO2.
2. Experimental
200 mm × 4.6 mm 10 ␮ column (eluents formic acid and acetoni-
trile). Maleic acid was determined by titration with NaOH 0.1N,
after subtraction of the contribution deriving from the other acids
formed.
Catalysts were prepared using the wet impregnation technique:
5.00 g of monoclinic ZrO2, with a specific surface area of 14 m2/g,
was dropped into an aqueous solution (volume 70 ml) containing
0.38 g of NH4VO3 and either 0.41 or 0.48 or 0.56 g (P/V atomic ratio
1.10, 1.30 and 1.50, respectively) H3PO4 (85 wt%) (both chemicals
supplied by Sigma–Aldrich). The amounts used made it possible to
obtain the 10 wt% active phase (calculated by hypothesizing the for-
mation of (VO)2P2O7) with respect to ZrO2. Water was evaporated
under slight vacuum at 70 ◦C. The thermal treatment of samples
was carried out either in muffle, or in the cell of the Raman spec-
trometer.
Catalysts were characterized by means of laser Raman spec-
troscopy (Renishaw 1000, laser source Argon ion 514 nm, equipped
with a Leica DMLM microscope and a commercial cell, Linkam
Instruments TS1500 for in situ measurements), and X-ray diffrac-
tion (Philips PW 1710 apparatus, with Cu K␣, ꢀ = 1.5406 Å, as the
radiation source).
Catalyst surface acidity was measured by FT-IR spectroscopy,
by means of a PerkinElmer 1750 Spectrometer. Self-supporting
wafers of pure sample were first evacuated at 450 ◦C in vacuo
(residual p = 10−6 mbar). Then adsorption of pyridine was done at
room temperature, and desorption was carried out by outgassing
the sample at increasing temperatures. The FT-IR spectrum was
recorded after evacuation at each temperature level. The concen-
tration of Lewis and Brønsted acid sites was calculated using the
method proposed by Eines [13] (extinction coefficient for Lewis
sites 2.22, for Brønsted sites 1.67).
Catalytic tests were carried out in a quartz continuous-flow
reactor, loading 0.8 g of catalyst (W/F = 1.3 g s ml−1) and feeding
1.7% n-butane and 17% oxygen (remainder N2). Products were
analyzed on-line by means of a gas chromatograph equipped
with a HP-1 column for the separation of unconverted n-butane,
maleic anhydride and light by-products: formaldehyde, acrylic
acid, acetic acid, crotonaldehyde. A Carbosieve SII column was
used for the analysis of oxygen, CO and CO2. The gaseous efflu-
ent stream was also bubbled for 2 h either in acetone (this fraction
was then analyzed by means of GC) or in water, for the identi-
fication of the heavy compounds (mainly phthalic acid, benzoic
acid, and methyltetrahydrobenzoic acid). The aqueous solution
was analyzed by means of HPLC–MS, using a Lichrosorb RP18
3. Results
3.1. Characterization of samples
Fig. 1 shows the Raman spectra recorded in situ during the
heating of the dried P/V 1.1 sample in 14.8% O2 in N2 flow (fig-
ure on the left) and during the isothermal step at 450 ◦C (figure
on the right). The spectrum of the RT sample shows bands that
are attributable to the zirconia support (at Raman shift 180, 340,
380, 474, 535, 555, 615, 636, 1145 and 1220 cm−1), and bands at
700 cm−1 (very strong), 204, 318 and 976 cm−1; these latter bands
are not attributable to ammonium vanadate or to any other V
compounds. It is worth noting that the solid obtained by reacting
ammonium vanadate with phosphoric acid in aqueous medium,
which yielded NH4(VO2)2PO4 after water evaporation, showed a
Raman spectrum that did not correspond to that shown in Fig. 1.
This suggests that already during the drying step a reaction occurs
between the deposited compounds and zirconia, with the forma-
tion of a mixed V/P/Zr/O precursor spread over the zirconia.
During heating, however, all the bands progressively disap-
peared, and at 300 and 400 ◦C the spectra did not show any
significant bands, not even those attributable to the support. When
the temperature was raised above 400 ◦C, bands attributable to
␦-VOPO4 (1090, 1075, 1020 and 936 cm−1) [14,15] and zirconia
developed, which then became well evident during the isothermal
step at 450 ◦C. However, a further prolongation of the isothermal
step led to the partial transformation of ␦-VOPO4 into ␤- (968
and 892 cm−1) and ␣II-VOPO4 (945 and 933 cm−1). The spectrum
recorded at RT also showed bands attributable to V2O5 (998, 705,
483, 305 and 285 cm−1). A similar behavior was shown when the
thermal treatment was carried out in a wet (3% steam, 14.5 O2,
remainder N2) stream (Fig. 2). Also in this case, after 2 h at 450 ◦C,
bands attributable to ␦-VOPO4 that formed during the heating step,
disappeared; in this case, the compound converted mainly into
V2O5. Therefore, the presence of steam favored the hydrolysis of
VOPO4. In the spectrum recorded at RT, broad bands were visible
at around 870–880 and 810–815 cm−1; these bands are attributable

206
A. Caldarelli et al. / Catalysis Today 157 (2010) 204–210
Fig. 2. Raman spectra of the sample prepared by impregnation of zirconia with
NH4VO3 and H3PO4 (P/V atomic ratio 1.1) recorded in situ in 3% steam and 14.5% O2
(remainder N2) during the isothermal step at 450 ◦C. Symbols as in Fig. 1.
to the stretching vibrations of P(OH)2 [16] and V–O–V in oligomeric
VO units [17], respectively. These bands are even more evident
when the isothermal step is carried out in a flow of 3% steam in
N2, that is, in the absence of oxygen (spectra not reported). Also in
this case, the ␦-VOPO4 formed by reaction between ammonium
vanadate and phosphoric acid decomposed; however, no bands
attributable to bulk V2O5 were visible, neither in the spectrum
recorded in situ at high temperature after 5 h on stream, nor in
the spectrum recorded at RT after the treatment. Therefore, under
these conditions, ␦-VOPO4 evolved into a disordered (amorphous)
layer made of polyphosphoric acid and amorphous VOy.
Thermal treatments were also carried out in tubular muffle,
with different types of flowing stream, and with a final temper-
ature of 450 ◦C. Fig. 3 (top) shows the Raman spectra recorded ex
situ after thermal treatments for sample P/V 1.1. Also in this case,
as shown during the in situ experiments, the nature of the com-
pounds formed was a function of the composition of the gas phase
used during thermal treatment. When the latter was carried out
in a dry air stream, the sample showed bands attributable to var-
ious VOPO4 compounds; the sample was not homogeneous, and
the beam focusing on several particles gave spectra that showed
the presence of different types of vanadium phosphate compounds.
When this treatment was done in a 10% steam air flow, the sam-
ple obtained was homogeneous, and in all cases bulk V2O5 was
the prevailing compound. Finally, a treatment carried out in a wet
and inert (10% steam in He) stream led to a compound character-
ized by a Raman spectrum with a dominant broad band centered
at around 900 cm−1. The latter may derive from the overlap-
ping of bands due to oligomeric V species and polyphosphoric
acid.
Fig. 3. Raman spectra (top) and X-ray diffraction patterns (bottom) recorded ex situ
at RT after thermal treatment of the catalyst precursor (P/V 1.1) at 450 ◦C in different
atmospheres: (i) air, (ii) air + 10% steam, and (iii) He + 10% steam. Symbols as in Fig. 1.
ꢀ: P2O5.
with some uncertainty, to V2O5 and to P2O5; these reflections were
more evident in the sample treated in the 10% steam air flow, which
agrees with Raman spectra. No clear attribution of any VOPO4 com-
pound could be done from X-ray patterns; however, the very weak
reflection at 2ꢁ 29◦, observed in the spectrum of the sample treated
in air, might be tentatively attributed to ␣II-VOPO4.
In samples with a P/V ratio of 1.30 and 1.50 (Fig. 4), the spec-
tra recorded in situ under flow of 3% steam and 14.5% O2 in N2,
showed that, in both cases, the ␦-VOPO4 formed during heating of
the samples was stable and did not decompose into V2O5 nor was it
transformed into other VOPO4 compounds, within the timeframe
of the experiment.
Catalysts were also characterized ex situ after reactivity tests.
Fig. 5 compares the spectra of fresh samples, after calcination in
a 10% steam air flow, and the spectra of the same samples after
reaction in n-butane oxidation (maximum reaction temperature,
450 ◦C). Both samples prepared with P/V 1.1 and 1.3 no longer
contained ␦-VOPO4 after reaction; ␤-VOPO4 and V2O5 were the
only compounds present in these samples. Conversely, the sample
prepared with P/V 1.5 still showed the presence of the desired ␦-
VOPO4, as well as smaller quantities of VOPO4·2H2O and ␣I-VOPO4.
According to both our previous findings and the role of ␦-VOPO4
[6,9–11,18–20], it can be expected that this sample shows the best
catalytic performance.
Fig. 3 (bottom) shows the corresponding X-ray diffraction
patterns. The reflections attributable to monoclinic ZrO2 were dom-
inant, whereas very weak additional reflections are attributable,
Table 1
Results of catalytic experiments carried out at 400 ◦C. All samples are made of 10 wt% V/P/O deposited over ZrO2.
Catalyst
C4 conv. (%)
31
27
7
Maleic anhydride sel. (%)
COx sel. (%)
Lights sel. (%)
Heavies sel. (%)
P/V 1.1
P/V 1.3
P/V 1.5
5
10
18
82
65
72
2
2
4
11
23
6

A. Caldarelli et al. / Catalysis Today 157 (2010) 204–210
207
Fig. 4. Raman spectra of the samples prepared by impregnation of zirconia with NH4VO3 and H3PO4 (figure on the left: P/V = 1.3; figure on the right: P/V = 1.5) recorded in
situ in a 3% steam air flow during the isothermal step at 450 ◦C. Symbols as in Fig. 1.
Fig. 5. Raman spectra recorded ex situ at RT after thermal treatment of the catalyst precursors at 450 ◦C in a 10% steam air flow (figure on the left), and after reactivity tests
(figure on the right).
3.2. Reactivity tests
shown in Fig. 6. The selectivity to Heavies increased when the
reaction temperature was increased, with a concomitant decrease
in the selectivity to carbon oxides. On the contrary, the selectiv-
ity to maleic anhydride was not much affected, and the same was
observed for the selectivity to Lights. This behavior, together with
the nature of heavy by-products, suggests that the formation of
these compounds most likely occurs thanks to the condensation
Table 1 reports the results of the reactivity tests carried out with
samples having P/V atomic ratio of 1.1, 1.3, and 1.5. All the cata-
lysts were firstly calcined in a flow of air with 10% steam (total flow
35 ml/min, ␶ = 0.9 s) for 5 h at 450 ◦C, and then treated at 400 ◦C
under reaction conditions for 100 h. This amount of time is neces-
sary for any type of structural or morphological transformation to
occur in V/P/O-based catalysts, and is also important for developing
an “equilibrated” catalyst with a stable catalytic performance [7].
Catalyst P/V 1.1 was the most active, but it also showed the lowest
selectivity to maleic anhydride. It is worth noting that this lower
selectivity was not simply a consequence of the higher conversion
achieved; in fact, as will be discussed later, also when the compar-
ison is made at similar conversion levels, the selectivity of samples
with a P/V ratio of 1.3 and 1.5 was still higher than that of P/V 1.1.
COx was not the only by-product of the reaction. In fact, other
by-products formed include formaldehyde, acetic acid, acrylic acid,
and crotonaldehyde (all of them grouped into “Lights”); however,
only small quantities of these light by-products were formed (selec-
tivity less than 4%). Heavier by-products (“Heavies”) also formed,
most of which were aromatic compounds; phthalic acid, phthalic
anhydride, benzoic acid, and methyltetrahydrobenzoic acid were
produced in the greatest amounts.
With regards to P/V 1.5, the catalyst that gave the best selectiv-
ity to maleic anhydride, the details of the selectivity to the various
classes of products as a function of the reaction temperature are
Fig. 6. Results of catalytic experiments as a function of temperature with P/V 1.5
catalyst. Symbols: n-butane conversion (♦), selectivity to maleic anhydride (␦), to
COx (ꢀ), to Lights (ꢁ) and to Heavies (ꢂ).

208
A. Caldarelli et al. / Catalysis Today 157 (2010) 204–210
Scheme 1. Surface mechanism suggested for the transformation of n-butane to maleic anhydride, COx and heavy compounds with zirconia-supported V/P/O catalysts.
Table 2
3.3. The role of surface acidity on reactivity behavior
Amount of acid sites in P/V 1.1 and P/V 1.5 samples, as measured from pyridine
adsorption measurements (IR spectra recorded at increasing temperature).
Table 2 compares the amount of acid sites, as inferred from
Catalyst
P/V 1.1
T (◦C)
CB (␮mol/g)
CL (␮mol/g)
the IR spectra recorded after adsorption of pyridine and des-
orption at increasing temperature. Sample P/V 1.5 had a greater
number of acid sites than P/V 1.1, especially of the Br␾nsted-
type; the acidity level, however, was not very different in the
two cases, thus indicating that the nature of the sites was sim-
ilar for the two samples. According to the literature concerning
the acidity of VPP and of V/P/O compounds in general, Br␾nsted
sites are attributable to P–OH groups [21]; Lewis sites can be
tentatively attributed either to coordinatively unsaturated V5+
sites, or to Zr4+ sites (in fact, the acidity characterization of the
zirconia support highlighted the presence of Lewis-type sites,
and no Br␾nsted acid sites whatsoever). It is worth noting that
the surface acidity of VPP is strong enough to catalyze reac-
tions such as aromatic ring alkylation or glycerol dehydration
[22,23].
P–OH sites are generated by the hydrolysis of either V/P/O com-
pounds or polyphosphoric acids. Therefore, the concentration of
sites responsible for heavy compounds formation should be a func-
tion of steam partial pressure. Water is a co-product of the reaction,
and its partial pressure increases in parallel with the conversion
of n-butane, which should enhance the hydrolysis phenomena,
thus contrasting the dehydration reaction that is thermodynam-
ically favored when the reaction temperature is increased. In order
to confirm this hypothesis, we carried out reactivity tests by co-
feeding an additional amount of steam (10% in the inlet stream),
with the P/V 1.1 and 1.5 catalysts. The results of this experiment
25
200
450
28
19
0
23
13
2
P/V 1.5
25
200
450
14
6
0
17
9
1
of unsaturated intermediates. One hypothesis is that the conden-
sation reaction may occur in the presence of acid sites, and that
the latter are generated in situ by hydrolysis of either VOPO4 or
polyphosphoric acid. In fact, the selectivity to Heavies, achieved at
similar levels of n-butane conversion, increased in the following
order:
P/V 1.1 (selectivity 11% at 31% n-butane conversion) < P/V 1.3
(selectivity 23% at 27% n-butane conversion) < P/V 1.5 (selectivity
29% at 25% n-butane conversion).
It is worth noting that under the conditions used, heavy com-
pounds desorbed into the gas phase, and were not retained on
the catalyst surface. In fact, no relevant short-term deactivation
phenomena were observed.
In order to confirm the role of surface acid sites, we carried out
further characterization and reactivity experiments with P/V 1.1
and P/V 1.5 samples.
Table 3
Reactivity of P/V 1.1 and P/V 1.5, both in the absence and in the presence of co-fed steam.
Catalyst
P/V 1.1
Steam in feed (%)
T (◦C)
C4 conv. (%)
Maleic anhydride sel. (%)
COx sel. (%)
Heavies sel. (%)
Maleic acid sel. (%)
Lights sel. (%)
0
10
400
400
31
28
5
3
82
72
11
18
tr
5
2
1
P/V 1.5
0
10
10
440
420
440
25
26
40
15
9
9
53
32
35
29
43
35
tr
14
19
3
2
2

A. Caldarelli et al. / Catalysis Today 157 (2010) 204–210
209
are shown in Table 3. With both samples, the addition of steam
led to an increase in the selectivity to Heavies, with a concomitant
decrease in the selectivity to COx. In the case of P/V 1.5 however,
this variation was greater than with P/V 1.1. Moreover, with both
catalysts, a relevant increase in the selectivity to maleic acid (which
was formed only in trace amounts under conventional reaction con-
ditions), and a decrease in the selectivity to maleic anhydride were
observed.
P is needed not only to develop the desired ␦-VOPO4 during thermal
treatment, but also to hinder its consecutive transformation into
other, undesired VOPO4 compounds or its complete hydrolysis into
V2O5 and polyphosphoric acid [46].
The sample prepared with a P/V ratio of 1.5 gave the highest
selectivity to maleic anhydride, and the lowest selectivity to car-
bon oxides. This confirms that the best performance is achieved
with the catalyst whose active component is made mainly of ␦-
VOPO4. On the other hand, the excess of P loaded is responsible
for the generation of Br␾nsted-type acid sites, which catalyzed the
condensation or addition-type reaction between intermediates to
form heavy compounds. For a n-butane conversion of 25–30%, the
selectivity to these latter products is the highest with the P/V 1.5
catalyst, and is further enhanced with both P/V 1.1 and P/V 1.5 sam-
ples when the partial pressure of steam in the gas phase is very
high.
Let us discuss the reaction mechanism for the formation of
these heavy by-products. Reaction intermediates are butenes, buta-
diene, crotonaldehyde and dihydrofuran [47,48] (Scheme 1). The
nature of the heavy compounds formed suggests that a Diels-
Alder-type addition between butadiene and crotonaldehyde (both
products being obtained from butenes) may lead to the formation of
2-methyltetrahydrobenzaldehyde and, subsequently, of the corre-
sponding acid derivative (one of the heavy products identified). The
latter may then be oxidized to phthalic acid and phthalic anhydride,
whereas benzoic acid likely forms by decarboxylation of phthalic
acid. On the other hand, the olefinic intermediates may either be
oxidized to maleic anhydride, or to carbon oxides. The relative rate
of the three parallel reactions (which have common reaction inter-
mediates) is a function of the reaction temperature, the partial
pressure of steam (an increase of which increases the rate of the
reaction leading to the formation of the heavy compounds), and
catalyst composition.
Diels-Alder reactions are catalyzed by both Br␾nsted and Lewis
acid systems [49]. In general, these reactions are conducted at
room temperature or slightly elevated temperatures, because of
the instability of reactants and products, and because in many
cases reactions are exoenthalpic but reversible, hence the forma-
tion of products may be unfavored at high temperature. In our
case, however, the presence of an oxidizing environment may shift
the equilibrium towards the formation of products, because the
primary addition product – the ring-substituted cyclohexene – is
oxidized to yield the corresponding aromatic compounds. There-
fore, P-OH Brønsted-type acid sites, present in the catalyst having
a large excess of P, may accelerate the reaction between unsatu-
rated intermediates to generate heavy compounds. Acid systems
also catalyze the reaction of butadiene with maleic anhydride [50],
a reaction that also may contribute to phthalic anhydride and
phthalic acid formation [51,52].
4. Discussion
From the literature, it is clear that many authors have tried
to prepare catalysts made of supported VPP [24–37]. This was
achieved either by preparing samples in the presence of precur-
sors for the formation of both the support and the active V/P/O
phase, or by depositing the components, precursors for the active
phase formation, onto the support, e.g., titania, alumina, silica or
Al phosphate [24–33]. With model VO-systems, for catalysts con-
taining approximate monolayer coverage of surface vanadia, the
number of converted n-butane molecules per vanadium atom was
a strong function of the specific oxide support [34]. In most cases,
however, the intrinsic characteristics of the VPP were worsened
due to the chemical interaction that develops between the active
phase and the support. One exception was the catalyst prepared
by Ledoux et al. [35,36], who studied the use of heat-conducting
supports (␤-SiC, Si3N4 and BN) for VPP. This catalyst significantly
increased the maleic anhydride yield with respect to the unsup-
ported one, because of the better control of the catalyst surface
temperature. Moreover, the support apparently did not affect the
chemical properties or the reactivity of the VPP. Outstanding per-
formances were also obtained by Au et al. [37], who used V/P/O
catalysts supported on ZrO2 and H3PO4-treated ZrO2 by means of
precipitation–deposition in an organic medium.
With our samples, the approach used was different to that
reported in the literature. We did not try to develop the VPP pre-
cursor, VOHPO4·0.5H2O (a V4+ compound), by carrying out the
synthesis in the presence of the support; conversely, we decided
to build-up the desired ␦-VOPO4, starting from a V5+ salt and
phosphoric acid. Under reaction conditions, one can expect equilib-
rium to be reached between ␦-VOPO4 and VPP, because structural
similarities between the two compounds facilitate structural inter-
conversion during the redox process.
The role of VOPO4 compounds in the redox behavior of V/P/O-
based catalysts for n-butane oxidation to maleic anhydride is
known since long time [14,15,38–42], but the exact structure
of some of the various VOPO4 polymorphs still is under debate
[43–45]. For example, the structure of ␣I-VOPO4 has not yet
been resolved; this compound, observed in some of our sam-
ples, reversibly transforms into VOPO4·2H2O by hydration [14,38].
Concerning the ␦-VOPO4, this compound was first reported by Bor-
des [38], who proposed that this compound is isostructural with
VOAsO4, featuring pairs of edge-sharing octahedral, like VPP. This
led to the hypothesis that VPP and ␦-VOPO4 may easily intercon-
vert. The structure of ␦-VOPO4 has been recently resolved [45]. The
structure contains infinite chains of trans corner-sharing VO6 octa-
hedra which run parallel to [1 0 0] and [0 1 0] directions. Each VO6
shares single corners with four different phosphate groups, and
each phosphate connects four different chains of VO6. The main
peculiarity of the ␦ structure is that it contains an unusually long
V–O contact (3.1 Å); moreover, it is more strained than other VOPO4
polymorphs.
All in all, the performance of the catalysts prepared is worse than
that of the unsupported VPP catalyst, and of some silica-supported
V/P/O catalysts described in the literature [53,54]. There are various
reasons for this:
(a) The large number of acid sites, arising from the high loading of P,
which in turn is necessary to develop and maintain the ␦-VOPO4
compound. The formation of heavy compounds is favored by
acidity.
(b) In unsupported VPP, the latter and the ␦-VOPO4 are in redox
equilibrium, and they catalyze the transformation of n-butane
to maleic anhydride with high selectivity. The rate-determining
step is V4+ oxidation, and hence VPP is the prevailing compound
in the equilibrated catalyst [12]. A limited amount of ␦-VOPO4
develops at the surface of VPP by V4+ oxidation, thus providing
a moderate concentration of oxidizing V5+ sites. Finally, this
results in mild oxidation, rather than total oxidation. On the
In our previous investigations [10,11], we found that the P/V
ratio is a critical parameter that affects the type of VOPO4 that is
finally obtained; this has also been confirmed with the samples
reported in the present study. In this case, however, a large excess of

210
A. Caldarelli et al. / Catalysis Today 157 (2010) 204–210
other hand, when the catalyst is made of ␦-VOPO4, as in the
case of our supported P/V 1.5 catalyst, the development of a
reduced V4+-containing active phase may be a more difficult
process. Therefore, the supported catalyst is too oxidizing, and
the formation of carbon oxides is kinetically favored over the
formation of maleic anhydride. In other words, the presence of
VPP as the main component of the active phase is essential in
providing a controlled and optimized number of oxidizing sites.
[15] F. Ben Abdelouahab, J.C. Volta, R. Olier, J. Catal. 148 (1994) 334–340.
[16] H. Abderrazak, M. Dachraoui, M.J. Ayora Can˜ada, B. Lendi, Appl. Spectrosc. 54
(2000) 1610–1616.
[17] J.M. Jehng, G. Deo, B.M. Weckhuysen, I.E. Wachs, J. Mol. Catal. A 110 (1996)
41–45.
[18] N. Harrouch Batis, H. Batis, A. Ghorbel, J.C. Vedrine, J.C. Volta, J. Catal. 128 (1991)
248–263.
[19] Y. Zhanglin, M. Forissier, R.P. Sneeden, J.C. Vedrine, J.C. Volta, J. Catal. 145 (1994)
256–266.
[20] C.J. Kiely, S. Sajip, I.J. Ellison, M.T. Sananes, G.J. Hutchings, J.C. Volta, Catal. Lett.
33 (1995) 357–368.
[21] G. Centi, G. Golinelli, F. Trifirò, Appl. Catal. 48 (1989) 13–24.
[22] F. Cavani, F. Trifirò, Appl. Catal. A 157 (1997) 195–221.
[23] F. Wang, J.-L. Dubois, W. Ueda, J. Catal. 268 (2009) 260–267.
[24] G.K. Bethke, D. Wang, J.M.C. Bueno, M.C. Kung, H.H. Kung, Stud. Surf. Sci. Catal.
110 (1997) 453–460.
[25] K.E. Birkeland, S.M. Babitz, G.K. Bethke, H.H. Kung, G.W. Coulston, S.R. Bare, J.
Phys. Chem. B 101 (1997) 6895–6902.
[26] J.M.C. Bueno, G.K. Bethke, M.C. Kung, H.H. Kung, Catal. Today 43 (1998)
101–110.
[27] M. Ruitenbeek, A.J. van Dillen, A. Barbon, E.E. van Faassen, D.C. Koningsberger,
J.W. Geus, Catal. Lett. 55 (1998) 133–139.
[28] R.A. Overbeek, A.R.C.J. Pekelharing, A.J. van Dillen, J.W. Geus, Appl. Catal. A 135
(1996) 231–248.
[29] R.A. Overbeek, P.A. Warringa, M.J.D. Crombag, L.M. Visser, A.J. van Dillen, J.W.
Geus, Appl. Catal. A 135 (1996) 209–230.
[30] X.-K. Li, W.-J. Ji, J. Zhao, Z.-B. Zhang, C.-T. Au, J. Catal. 238 (2006) 232–241.
[31] P.S. Kuo, B.L. Yang, J. Catal. 117 (1989) 301–310.
[32] S. Holmes, L. Sartoni, A. Burrows, V. Martin, G.J. Hutchings, C. Kiely, J.C. Volta,
Stud. Surf. Sci. Catal. 130B (2000) 1709–1714.
[33] F. Cavani, C. Cortelli, A. Frattini, G. Puccinotti, M. Ricotta, F. Rodeghiero, F. Trifirò,
C. Fumagalli, G. Mazzoni, Top. Catal. 38 (2006) 295–301.
[34] I.E. Wachs, J.M. Jehng, G. Deo, B.M. Weckhuysen, V.V. Guliants, J.B. Benzinger,
S. Sundaresan, J. Catal. 170 (1997) 75–88.
[35] M.J. Ledoux, C. Crouzet, C. Pham-Huu, V. Turines, K. Kourtakis, P. Mills, J.J. Lerou,
J. Catal. 203 (2001) 495–508.
[36] M.J. Ledoux, S. Hantzer, C. Pham-Huu, J. Guille, M.P. Desanaux, J. Catal. 114
(1998) 176–185.
[37] R.-M. Feng, X.-J. Yang, W.-J. Ji, Y. Chen, C.-T. Au, J. Catal. 246 (2007) 166–176.
[38] E. Bordes, Catal. Today 1 (1987) 499–526;
5. Conclusions
Supported V/P/O catalysts for n-butane oxidation to maleic
anhydride were prepared by means of the impregnation of zir-
conia with ammonium vanadate and phosphoric acid, and by
thermal treatment under various environments and temperatures.
The nature of the active surface was found to be greatly affected by
both the composition of the gas phase used for the thermal treat-
ment, and the P/V ratio used for catalyst preparation. ␦-VOPO4
was initially formed by high-temperature reactions between the
vanadium salt and phosphoric acid; however, in the sample with
P/V = 1.1, ␦-VOPO4 transformed into either V2O5, in a steam-
enriched air stream, or ␤-VOPO4, in a dry air stream. The ␦-VOPO4
was more stable in the sample prepared with a large excess of
P, i.e., with P/V = 1.3 and 1.5. The latter catalyst showed the best
catalytic performance in n-butane oxidation. However, the selec-
tivity to maleic anhydride was no greater than 20%, mainly because
of the production of heavy compounds. The latter formed by the
acid-catalyzed addition between unsaturated intermediates, and
oxidative dehydrogenation of the addition product to form aro-
matic oxygenated compounds. The number of surface Br␾nsted
sites was highest in the catalyst prepared with P/V 1.5, which
also provided the greatest selectivity to the undesired heavy by-
products.
E. Bordes, Catal. Today 3 (1988) 163–174.
[39] E. Bordes, P. Courtine, J. Chem. Soc., Chem. Commun. (1985) 294–296.
[40] E. Bordes, J.W. Johnson, A. Raminosona, P. Courtine, Mater. Sci. Monogr. 28B
(1985) 877–887.
[41] J.-C. Volta, Catal. Today 32 (1996) 29–36.
[42] C. Doornkamp, M. Clement, X. Gao, G. Deo, I.E. Wachs, V. Ponec, J. Catal. 185
(1999) 415–422.
References
[43] Z.G. Li, R.L. Harlow, N. Herron III, H.S. Horowitz, E.M. McCarron, J. Catal. 171
(1997) 506–508.
[1] N. Ballarini, F. Cavani, C. Cortelli, S. Ligi, F. Pierelli, F. Trifirò, C. Fumagalli, G.
Mazzoni, T. Monti, Top. Catal. 38 (2006) 147–156.
[44] M. Conte, G. Budroni, J.K. Bartley, S.H. Taylor, A.F. Carley, A. Schmidt, D.M.
Murphy, F. Girgsdies, T. Ressler, R. Schlögl, G.J. Hutchings, Science 313 (2006)
1270–1273.
[45] F. Girgsdies, M. Schneider, A. Brückner, T. Ressler, R. Schlögl, Solid State Sci. 11
(2009) 1258–1264.
[2] P. Arpentinier, F. Cavani, F. Trifirò, The Technology of Catalytic Oxidations,
Editions Technip, Paris, 2001.
[3] S. Albonetti, F. Cavani, F. Trifirò, Catal. Rev. Sci. Eng. 38 (1996) 413–438.
[4] J.C. Volta, C. R. Acad. Sci. Paris, Serie IIc, Chim. 3 (2000) 717–723.
[5] E. Bordes, Top. Catal. 15 (2001) 131–137.
[46] Z.Y. Xue, G.L. Schrader, J. Phys. Chem. B 103 (1999) 9459–9467.
[47] U. Rodemerck, B. Kubias, H.-W. Zanthoff, M. Baerns, Appl. Catal. A (1997)
203–216.
[6] N. Duvauchelle, E. Bordes, Catal. Lett. 57 (1999) 81–88.
[7] S. Albonetti, F. Cavani, F. Trifirò, P. Venturoli, G. Calestani, M. Lopez Granados,
J.L.G. Fierro, J. Catal. 160 (1) (1996) 52–64.
[48] G. Centi, F. Trifirò, J.R. Ebner, V. Franchetti, Chem. Rev. 88 (1988) 55–80.
[49] G.J. Meuzelaar, R.A. Sheldon, in: R.A. Sheldon, H. van Bekkum (Eds.), Fine
Chemicals Through Heterogeneous Catalysis, Wiley-VCH, Weinheim, 2001, pp.
284–294.
[8] L. O’Mahony, T. Curtin, J. Henry, D. Zemlyanov, M. Mihov, B.K. Hodnett, Appl.
Catal. A 285 (2005) 36–42.
[9] N. Ballarini, F. Cavani, C. Cortelli, M. Ricotta, F. Rodeghiero, F. Trifirò, C. Fuma-
galli, G. Mazzoni, Catal. Today 117 (2006) 174–179.
[50] S.G. Wilkinson, in: D. Barton, W.-D. Ollis (Eds.), Comprehensive Organic Chem-
istry, vol. 1, Pergamon Press, Oxford, 1979, pp. 640–644.
[51] N. Ballarini, F. Cavani, C. Cortelli, F. Gasparini, A. Mignani, F. Pierelli, F. Trifirò,
C. Fumagalli, G. Mazzoni, Catal. Today 99 (2005) 115–122.
[52] V.A. Zazhigalov, J. Haber, J. Stoch, E.V. Cheburakova, Catal. Commun. 2 (2001)
375–378.
[10] F. Cavani, E. Degli Esposti, S. Luciani, C. Cortelli, R. Leanza, Chem. Eur. J. 16 (2010)
1646–1655.
[11] F. Cavani, D. De Santi, S. Luciani, A. Löfberg, E. Bordes-Richard, C. Cortelli, R.
Leanza, Appl. Catal. A 376 (2010) 66–75.
[12] V.V. Guliants, S.A. Holmes, J.B. Benziger, P. Heaney, D. Yates, I.E. Wachs, J. Mol.
Catal. A 172 (2001) 265–276.
[53] Z.Q. Zhou, H.Y. Xu, W.J. Ji, Y. Chen, Catal. Lett. 96 (2004) 221–226.
[54] C.Y. Xiao, X. Chen, Z.Y. Wang, W.J. Ji, Y. Chen, C.T. Au, Catal. Today 93–95 (2004)
223–228.
[13] C.A. Eines, J. Catal. 141 (1993) 347–354.
[14] F. Ben Abdelouahab, R. Olier, N. Guilhaume, F. Lefebvre, J.C. Volta, J. Catal. 134
(1992) 151–167.