Bulk and surface structures of V2O5/ZrO2 catalysts for n-butane oxidative dehydrogenation

Article abstract of DOI:10.1016/j.molcata.2009.05.014

The present investigation focuses on the structural properties and reactivity of zirconia-supported vanadium oxide catalysts, prepared by equilibrium adsorption in basic (pH 10) or in acid (pH 2.7) conditions with vanadium content up to 6 wt.% (pH 10) and

Full text of DOI:10.1016/j.molcata.2009.05.014

Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Bulk and surface structures of V O /ZrO catalysts for n-butane oxidative
2
5
2
dehydrogenation
Delia Gazzoli , Sergio De Rossi , Giovanni Ferraris , Giorgio Mattei , Roberto Spinicci , Mario Valigia
a,∗
b
b
c,1
d
a
Dipartimento di Chimica, Sapienza Università di Roma, P.le A. Moro, 5, I-00185 Rome, Italy
Consiglio Nazionale delle Ricerche, Istituto dei Sistemi Complessi, ISC, UOS “Materiali Inorganici e Catalisi Eterogenea” (MICE), c/o Dipartimento di Chimica,
b
Sapienza Università di Roma, P.le A. Moro, 5, I-00185 Rome, Italy
c
Consiglio Nazionale delle Ricerche, Istituto dei Sistemi Complessi, ISC, Sezione di Montelibretti (Roma) - Via Salaria Km 29.300, C.P. 10, I-00016 Monterotondo St., Italy
Dipartimento di Energetica, Università di Firenze, via S. Marta, 3, I-50139 Florence, Italy
d
a r t i c l e i n f o
a b s t r a c t
Article history:
The present investigation focuses on the structural properties and reactivity of zirconia-supported vana-
dium oxide catalysts, prepared by equilibrium adsorption in basic (pH 10) or in acid (pH 2.7) conditions
with vanadium content up to 6 wt.% (pH 10) and up to 11.6 wt.% (pH 2.7). The samples, heated at 823 K
for 5 h in air, were characterized by X-ray diffraction, Raman spectroscopy and TPR, both as prepared
and after leaching with an ammonia solution to remove species not anchored to the zirconia surface.
Some representative samples were also tested for the n-butane oxidative dehydrogenation (ODH) reac-
tion. Depending on vanadium content, various vanadium species were identified by Raman spectroscopy
that reacted differently on exposure to H2. At similar loading, the fraction of vanadium in a dispersed
state and thus interacting with the support was higher in samples prepared at pH 10 than in those at pH
2.7. Samples prepared at pH 2.7 contained a higher fraction of large polymeric structures in addition to
ZrV2O7 and V2O5.
Received 1 December 2008
Received in revised form 14 May 2009
Accepted 17 May 2009
Available online 27 May 2009
Keywords:
Butane ODH
V2O5/ZrO2
Zirconia
Raman spectroscopy
TPR
In line with literature data for propane ODH on similar catalysts, our catalytic results suggested that
the active sites for the ODH reaction are associated with the V–O–V bonds of the polymeric exposed
structures, whereas the Zr–O–V sites favour alkane combustion.
©
2009 Elsevier B.V. All rights reserved.
1
. Introduction
kinetics, including activation energies and entropies for primary
and secondary combustion [22–24]. Conversely, although ample
information is available on the ODH reaction of n-butane on sev-
eral catalysts [25,26], to our knowledge little is known about
the same reaction on VOx/ZrO2 catalysts. As the reactivity of n-
butane on various vanadium based catalysts is higher than that of
propane [2,27,28], we thought interesting to explore the behaviour
of VOx/ZrO2 catalysts in the ODH reaction of n-butane to butenes.
As a practical aspect, butenes reacting with isobutane yield high
octane number branched C8 alkanes for gasoline reformulation.
The efficiency of supported vanadium oxide catalysts mainly
depends on the dispersion of the active phase, which in turn
depends closely on the supported oxide and the method used for
catalyst preparation. In a previous investigation of the zirconia-
supported vanadium oxide systems, we found that the uptake of
vanadium species on the hydrous support, obtained by equilibrium
adsorption at pH 10 or at pH 2.7, depended on the preparation pH
Vanadium oxides supported on various metal oxides (e.g., MgO,
TiO , Al O , ZrO , SiO ) are an important group of catalysts for
2
2
3
2
2
a variety of reactions, including selective oxidation of hydrocar-
bons [1–6], ammoxidation of aromatics and alkyl-aromatics [7]
and selective catalytic reduction of NO with NH3 [8,9]. Their sur-
face properties have been analyzed in detail and many studies
have investigated the relationship between their surface structure
and catalytic properties [6,10–15]. In particular, ZrO2 supported
vanadium oxide samples have been investigated as catalysts for
the oxidative dehydrogenation (ODH) reaction of alkanes, one of
the most feasible processes to obtain light olefins [16–18]. Numer-
ous studies have focused on propane ODH [19–24]. The vanadium
dispersed species have been extensively characterized by various
techniques and related to the catalytic activity and selectivity in
propane ODH [19,20]. Other studies have reported detailed ODH
[
29]. After heating at 823 K, for catalysts having similar vanadium
loading the fraction of vanadium in a dispersed state and interact-
ing with the support was higher in samples at pH 10 than in those
prepared at pH 2.7. Compared to samples prepared at pH 10, those at
pH 2.7 contained a higher fraction of vanadium as ZrV2O7 and V2O5.
∗ Corresponding author. Tel.: +39 06 49913377; fax: +39 06 490324.
E-mail address: delia.gazzoli@uniroma1.it (D. Gazzoli).
1
Deceased.
1381-1169/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2009.05.014

18
D. Gazzoli et al. / Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
Table 1
V O5 was obtained by decomposing NH VO (Fluka) in air at
2
4
3
Specific surface area, S, and crystalline phases for vanadium-supported zirconia sam-
ples as prepared (heated at 823 K, for 5 h, in air) and after leaching with an ammonia
solution.
8
73 K for 3 h. ZrV O7 was prepared heating a mixture of V O5 and
2
2
ZrO (383) in air at 983 K, following the procedure described by
2
Sanati et al. [34]. Because at 983 K some V O5 becomes volatile,
2
Samplesa
S(m2
g
−1
)
Phasesb
Samplesa
S(m2
g
−1
)
Phasesb
a small excess of V O5, compared with the stoichiometric ratio,
2
ZrO2(823)
ZV0.99,10
ZV2.7,10
ZV3.1,10
ZV3.4,10
ZV4.1,10
ZV4.2,10
ZV4.8,10
ZV5.4,10
ZV6.4,10
49
79
77
105
128
116
96
86
100
114
m, t
m, t
m, t
m, t
m, t
m, t
m, t
m, t
m, t
m, t
was added to the reacting mixture. The pentoxide in excess was
removed by leaching with an ammonia solution [35]. After rins-
ing the solid was re-heated at 773 K 1 h in air. The X-ray diffraction
ZV 1.2,2.7
ZV 1.9,2.7
ZV 2.6,2.7
ZV3.7,2.7
ZV 5.3,2.7
ZV 5.9,2.7
ZV 7.4,2.7
ZV 7.6,2.7
ZV 11.6,2.7
60
77
97
80
42
37
24
38
39
m, t
m, t
m, t
m, t
m, t, Z, V
m, t, Z, V
m, t, Z, V
m, t, Z, V
m, t, Z, V
patterns of the prepared V O5 and ZrV O7 corresponded to those
2
2
reported in the JCPDS diffraction files [36,37]. The crystallite-size, D,
determined from the line broadening of the most intense reflexions
of V O5 and ZrV O7 resulted D
In the most concentrated ZVx,2.7, the crystallite-size of ZrV O7 was
= 75 nm and D
> 200 nm.
2
2
V2O5
ZrV2O7
2
about 100 nm.
ZV2.710L1.4
ZV2.8,10 L1.8
ZV3.4,10L2.5
ZV4.1,10L2.2
ZV4.8,10L2.2
ZV5.4,10L3.0
77
100
128
116
89
m, t
m, t
m, t
m, t
m, t
m, t,
ZV1.2,2.7L1.1
ZV1.9,2.7L1.0
ZV5.3,2.7L2.0
ZV 5.9,2.7L2.3
ZV7.4,2.7L5.8
ZV7.6,2.7L5.7
60
77
45
42
35
45
m, t
m, t
m, t, Z
m, t, Z
m, t, Z
m, t, Z
2.2. Experimental techniques
The vanadium content was determined by atomic absorption
after dissolution of the sample in a concentrated HF solution, and
subsequent dilution [38]. The standard solution (Fluka 1000 ppm)
102
a
For designation of samples, see text.
m = monoclinic ZrO2; t = tetragonal ZrO2, Z = ZrV2O7; V = V2O5.
b
contained ZrO , dissolved in HF, in a concentration similar to that
2
of the samples.
◦
X-ray diffraction (XRD) patterns in the 2ꢀ angular range from 5
◦
to 60 were obtained with a Philips PW 1729 diffractometer using
In both series of samples vanadium invariably tended to condense
into polymeric species resulting in a multilayer material.
In this study, we have examined the two series of samples in
greater detail and tested their catalytic behaviour for the ODH reac-
tion of n-butane. To differentiate the catalytic activity of the various
vanadium species on the zirconia surface or reacted with the sup-
port, the samples were studied as prepared and after leaching with
an ammonia solution. The leaching treatment has already been
applied to vanadium-supported catalysts [30–32].
Cu K␣ (Ni-filtered) radiation.
Surface areas were determined by N2 adsorption at 77 K (BET
method) using a Micromeritics apparatus ASAP 2010.
Raman spectra were collected with a SPECS triple spectrograph,
in macro configuration, equipped with a liquid-nitrogen-cooled
+
CCD detector using the 514.5 nm line of an Ar laser. The laser power
on the sample was always kept under 20 mW. The polarization of
the scattered light was not analyzed. The spectral resolution was
−
1
1
.4 cm . The samples were pressed into 5 mm pellets and ana-
For the study we used several techniques including X-ray diffrac-
tion, Raman spectroscopy, temperature-programmed reduction
determinations and catalytic tests.
lyzed in air or using a in situ cell. Details of the cell setup are given
elsewhere [39]. In the in situ experiment, the sample was heated
in flowing dry oxygen at 673 K, 2 h and the spectra were collected
after cooling to 298 K always in dry oxygen. Raman spectra anal-
ysis included smoothing, base line removal and curve fitting by a
GRAMS/386 software by Galactic.
2. Experimental
2
.1. Sample preparation
Temperature-programmed reduction (TPR) was mea-
sured/analyzed in a flow system using a HWD detector and a
gas mixture containing 10 vol.% of hydrogen in nitrogen 1.8 dm /h,
3
The samples were prepared as previously described [29].
Hydrous zirconium oxide was precipitated by bubbling a stream
of ammonia-saturated N2 into a ZrOCl2 solution for 18 h [33]. After
separation the solid was washed (Cl negative test in the solid), and
heating rate 8 K/min. The gases were dried and purified with
suitable filters and their flows regulated with Brooks electronic
mass flow controllers. For the measurement, 60 mg of sample,
located in a tubular reactor, underwent oxidative pre-treatment
in a flow of oxygen (20 vol.%) in nitrogen at 723 K, 1 h, cooled to
room temperature in a nitrogen flow and then equilibrated in the
reaction mixture.
−
dried at 383 K, 24 h. The obtained hydrous zirconium oxide was
2
−1
microporous, (specific surface area: 330 m g ) and amorphous
to X-ray. It was designated as ZrO (383). The vanadium contain-
2
ing samples were obtained by equilibrium adsorption suspending
ZrO (383) in V(V) aqueous solutions (from NH VO , Fluka) at pH 10
The n-butane ODH reaction was studied at ambient pressure in
a tubular continuous flow microreactor containing 90 mg of pow-
dered sample. The gases were purified and dehydrated by means
of suitable filters and their flows were regulated by means of MKS
electronic mass flow controllers. Before the test, the catalyst was
pre-treated in situ in a flow of oxygen (20 mol%) in helium at
723 K, 1 h, and cooled to reaction temperature in a helium flow.
The experiments were conducted at 603 K with a reactant mix-
ture containing 2.7 mol% n-C H , 4.1 mol% O , balance He, at total
2
4
3
or 2.7 (fixed by NH OH or HNO ). The suspension was shaken for
4
3
72 h at room temperature to equilibrate the system. The solid was
filtered from the liquid fraction, dried at 383 K for 24 h and finally
heated in air at 823 K, 5 h. A fraction of ZrO (383), contacted with
2
water, underwent similar treatment (shaking, filtering, heating at
8
23 K, 5 h) and was indicated as ZrO (823). The samples containing
2
vanadium were designated as ZVx,y, where x stands for the vana-
dium content as wt.% (metal) and y the pH value (Table 1). Some
samples were leached with a hot ammonia solution (about 20 mL,
4
10
2
−
1
flow rate of 73 ml min . In preliminary experiments, some sam-
−
1
2
mol L ) in a closed vessel for 30 min, under stirring to eliminate
ples were tested with different gas feed mixtures (O /n-butane
2
soluble vanadium. The solid was then separated and the leaching
repeated three times. The liquid fractions were collected and used
to determine the soluble vanadium content. The leached samples
were indicated as ZVx,yLz, where x and y as stated, L stands for
leaching and z for the vanadium content remaining in the sample
ratio 1, 2, 4) and at 563 K and 643 K. At constant temperature, the
increase of the reactant O invariably brought about a little increase
2
in the n-butane total conversion and a moderate decrease in the
butene selectivity. At constant gas feed mixture (O /n-butane ratio)
2
the increase of the reaction temperature, apart from the obvious
increase of the total conversion, reduced the butene selectivity.
(
wt.%).

D. Gazzoli et al. / Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
19
Because the results were not unexpected and did not add much
information, they were not reported and for all catalysts we chose
to operate at the condition reported above. Gas analysis was done
with a PerkinElmer Autosystem gas chromatograph, equipped with
Plot-Q capillary columns and HWD detector. The % butane total
conversion, %Ctot, and the % selectivity to butenes, %Sbutenes, were
calculated as:
(
C4H10)in − (C H10)out
4
%
Ctot = 100
(
C H )
4 10 in
(
C H )
4 8
out
%
Sbutenes = 100 (
C4H10)in − (C H10)out
4
The areal rates of total n-C H conversion (rtot) were calcu-
4
10
lated as the number of molecules converted per second and per
2
nm of sample exposed area. All reported catalysis data were col-
lected at the steady state. Owing to the mesoporous character of
the materials [29], we did not expect rate limitation due to mass
transfer within the pores, as they are quite larger than the involved
molecules.
3. Results and discussion
3.1. Catalyst characterization
For all ZVx,y samples the specific surface area values were
higher than that of ZrO (823), increasing for vanadium content
2
up to x ∼ 3 wt.% (Table 1). For higher vanadium content, the spe-
cific surface area values of the ZVx,10 samples remained practically
constant, whereas those of the ZVx,2.7 set markedly decreased.
Leaching with ammonia solution did not affect significantly the
surface area values.
The X-ray diffraction analysis of samples ZVx,10 revealed no lines
other than those of ZrO , both in the monoclinic and tetragonal
2
modification (Fig. 1A). Nor did the ZVx,2.7 samples with x ≤ 3.7 wt.%,
Fig. 1. XRD spectra for ZVx,10 (A) and ZVx,2.7 (B) samples. Section A: (a) ZV2.7,10;
but for samples with higher vanadium concentration the lines of
(
b) ZV4.2,10; (c) ZV6.4,10. Section B: (a) ZrO2(823); (b) ZV2.6,2.7; (c) ZV5.3,2.7; (d)
ZrV O7 and V O5 were also detected (Fig. 1B). The V O5 reflec-
2
2
2
ZV11.6,2.7; (e) ZV11.6,2.7L6. Dotted lines indicate the most intense reflections of
monoclinic, m, and tetragonal, t, zirconia.
tions were no longer present in the patterns of leached samples.
The intensity of the tetragonal peaks increased relative to that of
monoclinic peaks with increasing vanadium loading in both series
of samples (Fig. 1). The comparison of the X-ray patterns of ZV2.7,10
and ZV2.6,2.7 samples (Fig. 1A, curve a and Fig. 1B, curve b reveals
that, despite the almost equal concentration, the fraction of tetrag-
onal zirconia in the ZV2.7,10 is higher than in ZV2.6,2.7.
ilar spectra, characterized by a broad feature centred at about
−
1
885 cm
corresponding to V–O–V vibrations and bands in the
−
1
range 1000–1030 cm depending on V content assigned to V
O
vibrations of VOx species in polyvanadate structures [21,23,42–45].
In the spectra of ZVx,2.7 samples with x ≥ 5.3 wt.% bands at
All these findings are in agreement with the results reported in
a previous paper [29], where we showed that, owing to the pre-
−
1
−1
775 cm and 989 cm (Fig. 2B, curves c–e) due to ZrV O7 (Fig. 2B,
2
vailing microporous texture of ZrO (383) and the influence of pH
curve f) [20], were also clearly detected; their intensities increased
with the vanadium content. The spectrum of the concentrated
ZV7.6,2.7 sample (Fig. 2B, curve e) showed, in addition, sharp bands
2
on the size of vanadium species, after heating at 823 K the two
sets of samples differed in the fraction of the various vanadium
species. The interaction of vanadium species with the zirconia sur-
face affected some solid state processes of the support, including
phase transition and sintering, an effect well documented for vari-
ous systems composed of transition metal oxoanions supported on
zirconia [19,29,40,41].
−
1
−1
at about 142 cm and 998 cm , identifying the presence of V O5
2
(Fig. 2B, curve g) [46]. The Raman assignment is consistent with XRD
results (Fig. 1) indicating that, besides vanadium polymeric species,
the concentrated ZVx,2.7 samples contained ZrV O7 or V O5 or
2
2
both.
Raman spectroscopy has been used to determine the struc-
ture of the vanadium species both in amorphous and crystalline
form. Raman spectra recorded for ZVx,10 (Fig. 2A) and ZVx,2.7
A deeper insight on the molecular structure of the surface
species and on the interaction between vanadium species and the
support is achieved by the analysis of the Raman spectra collected
under dehydrated conditions (“in situ” treatment at 673 K, 2 h in
flowing dry oxygen). It is well documented that hydration affects
vanadium surface species probably via hydrogen bonding result-
ing in a shift to lower frequencies and broadening of the terminal
(
2
Fig. 2B), under ambient conditions, exhibited bands in the range
−
00–700 cm due to ZrO (monoclinic and tetragonal) and a com-
2
1
−
1
posite feature spanning the frequencies range 750–1040 cm due
to vanadium species. At increasing vanadium content the bands of
ZrO2 progressively diminished in intensity and for x > 5 wt.% they
were barely detectable in both series of samples, indicating that the
thickness of the vanadium species on the zirconia surface increased.
All ZVx,10 samples (Fig. 2A) and ZVx, 2.7 with vanadium
content lower than 5.3 wt.% (Fig. 2B, curves a and b) showed sim-
V
O band [47]. Compared to the spectra under ambient condi-
tion, those collected in dry conditions showed a modification in
−
1
−1
the range 1000–1030 cm , but not in the region 750–950 cm .
Raman spectra under dry conditions of some representative ZVx,10
(Fig. 3A and B) and ZVx,2.7 (Fig. 3C and D) samples are presented

2
0
D. Gazzoli et al. / Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
within polyvanadates structures (Fig. 3C, curve a), although the
occurrence of isolated monovanadates species cannot be disre-
garded.
Changes in the surface morphology of zirconia-supported vana-
dium oxide systems with ammonia leaching imaged by atomic force
microscopy (TM-AFM) strengthen the view that large polymeric
vanadium material grows over vanadium species interacting with
zirconia [48]. In samples prepared by ionic exchange between an
aqueous solution of peroxovanadate complexes and hydrous zirco-
nium oxide (V content up to 6.8 wt.%) and heated in air at 823 K for
5
h as for the present systems [48], AFM revealed that, as vanadium
content increased, the surface morphology evolved from a smooth
surface profile towards a granular surface (roundish particles about
9
0–100 nm in diameter and about 20–30 nm in height) with some
larger grains, grown layered, with a lateral size of 200–250 nm and
5–40 nm in height. Leaching produced an overall flattening of the
3
system: the large structures (250–300 nm) were almost completely
removed, the roundish structures were left unchanged in size and
some aggregates became more clearly visible.
TPR analysis of our samples confirmed the presence of var-
ious vanadium species. TPR data may strongly depend on the
experimental conditions and, although the information obtained
is merely comparative [49], this technique may help to charac-
terize the reducibility of different species [50]. ZrO2 showed no
reduction processes in the temperature range investigated (not
shown). The TPR profile of V O5 (not shown) consisted of several
2
peaks in the temperature range 920–1130 K in agreement with lit-
erature data [49–52]. The profile for ZrV O7 displayed a narrow
2
peak at about 813 K and a low-intensity tail in the range 573–773 K
(
Fig. 4B, curve g). We note that a TPR profile of ZrV O7 was already
2
reported by Pieck et al. [51] with three peaks appearing at 700 K,
23 K and 860 K. The two peaks at 700 K and 823 K were proba-
8
bly related to the reduction of vanadium surface species that are
soluble in ammonia solution and thus are absent in TPR profile of
our sample. Three features could be identified from the complex
TPR profiles of the ZVx,10 samples: a main component at about
Fig. 2. Raman spectra of ZVx,10 (A) and ZVx,2.7 (B) samples. Section A: (a) ZV2.7,10;
b) ZV4.8,10; (c) ZV5.4,10. Section B: (a) ZV1.2,2.7; (b) ZV1.9,2.7; (c) ZV5.3,2.7; (d)
ZV7.4,2.7; (e) ZV7.6,2.7; (f) ZrV2O7; (g) V2O5.
(
713 K and H2 uptakes at lower (523–623 K) and higher (773–923 K)
temperatures (Fig. 4A). At increasing vanadium content, the main
peak increased in intensity and became asymmetric toward the
high temperature side. For the ZV5.4,10 sample distinct H2 con-
sumption was detected at 753 K (Fig. 4A, curve f). The reducibility
of the diluted ZVx,2.7 samples (Fig. 4B, curves a and b) almost
matched that of samples prepared at pH 10 confirming the pres-
ence of species of similar typology. At increasing vanadium content
the main hydrogen consumption shifted to the range 770–820 K
(Fig. 4B, curves c–f), partially overlapping the temperature range
together with those of samples after ammonia leaching. Bands at
−1
−1
and 1025 cm were observed in the ZV2.7,10
about 1000 cm
−1
−1
sample (Fig. 3A, curve a) and at about 1017 cm and 1033 cm in
the ZV5.4,10 sample (Fig. 3B, curve a). For the ZVx,2.7 samples, low
−
1
−1
intensity bands at about 1000 cm and 1025 cm were detected
in the spectrum of the diluted ZV1.2,2.7 sample (Fig. 3C, curve
a), whereas in the concentrated ZV7.4,2.7 sample only the band
−1
at about 1033 cm
(Fig. 3D, curve a) was clearly seen, because
was obscured by the intense band
due to ZrV O7 (Fig. 3D, curve a). The shift to
−1
the one at about 1000 cm
at which ZrV O7 reduces (Fig. 4B, curve g) and, the more concen-
2
−
1
at about 985 cm
trated sample showed a distinct peak at about 900 K, due to V O5
2
2
−
1
higher frequencies of the bands in the 1000–1030 cm range and
the concomitant growing in intensity of the broad feature in the
50–950 cm region at increasing vanadium content indicate the
(Fig. 4B, curve f), in agreement with X-ray analysis. We assign the
hydrogen consumption in the range 580–620 K to small polymeric
VOx anchored to the zirconia surface and in the range 680–730 K to
larger polymeric aggregates. The reduction in the range 770–820 K
−1
7
presence of polyvanadate structures that increase in size or number
or both with vanadium loading.
identified ZrV O7.
2
The inspection of the Raman spectra collected on leached
samples (Fig. 3A–D, curves b) specifies that the leaching caused
several effects: removed the bands at ␯ > 1020 cm (V O) (inset
The reducibility of vanadium species interacting with differ-
ent supports has already been investigated [52,53]. Studying the
vanadia/alumina and vanadia/zirconia systems, Roozeboom et al.
[52] observed two reduction peaks, depending on vanadium con-
tent: at low vanadium content (<2.5 wt.%) the peak occurred in
the range 650–700 K, whereas for more concentrated samples
the peak was centred at about 600 K. For vanadia/zirconia sam-
ples, Adamski et al. [53] found a TPR peak at about 800 K at
low vanadium content ascribed to monovanadates; at increasing
concentration the TPR profile first shifted to lower tempera-
tures (formation of small polymers) and then (formation of large
−
1
of Fig. 3A–D), lowered the intensity of the broad feature in the
−
1
region 750–950 cm (Fig. 3A–D) and disclosed the bands at about
−1
775 and 985cm
already at low vanadium content (Fig. 3A–D).
From these findings, more pronounced at increasing vanadium con-
tent, we infer that the leaching removes some large polymeric
vanadates and V O5 when present, leaving on the support surface
2
small sized polymeric structures and ZrV2O7. The band at about
−
1
1
010 cm in the ZV1.2,2.7L1.1 sample, could indicate monovana-
dates species (V O stretching mode) [44], whereas the coexistence
of the broad feature at about 850 cm , suggests V–O–V modes
polymeric species, V O5-like) moved back towards the reduction
2
−1
temperature observed for V O5. These investigators observed that
2

D. Gazzoli et al. / Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
21
Fig. 3. Raman spectra after in situ treatments.
Section A: (a) ZV2.7,10; (b) ZV2.7,10L1.4. Section B: (a) ZV5.4,10; (b) ZV5.4,10L3.0. Section C: (a) ZV1.2,2.7: (b) ZV1.2,2.7L1.1. Section D: (a) ZV7.4,2.7; (b) ZV7.4,2.7L5.8.
as the small polymeric species are reduced faster and at temper-
atures lower than monovanadate structures (T < 800 K) and large
polymeric species, the reducibility sequence proposed by Deo and
case (69%, sample ZV3.45,10). The carbon balance generally closed
at 100 ± 3%.
The total n-butane conversion and the butene selectivity values
are collected in Table 2 together with the apparent surface den-
sity (total number of V atoms in one gram of catalyst/total surface
area of one gram of catalyst) used as a variable for comparison pur-
poses, bearing in mind that it is a functional parameter approaching
the exposed V species for dilute samples only to a first approxima-
Wachs [54]: V–O–V > V–O–(V) > V O was confirmed. We failed
2
to observe an important peak at 800 K in dilute samples, proba-
bly because the monovanadate species, if present, was a minority
species. Therefore, in agreement with the Raman conclusion, the
TPR data confirm the presence of polymeric vanadium species in
our samples. As the concentration increases, the vanadium poly-
meric structures grow, ZrV O7 forms and the reduction shifts to
higher temperature.
2
Table 2
Samples, V surface density (V atoms nm−2) and catalytic properties for n-butane
The leaching treatment, removing large polymeric species
oxidative dehydrogenation at 603 K.
(
V O5-like) and V O5 crystallites, yielded simplified TPR profiles for
2 2
Samplesa
V atoms nm2 % Total n-C4H10 conversion % Selectivity to C4H8
both sets of samples (Fig. 5A–D, curves b). Accordingly, the H2 con-
sumption decreased in the range 600–800 K and for the ZV11.7,2.7Lz
ZV0.99,10
ZV2.7,10
ZV3.1,10
ZV3.4,10
ZV4.1,10
ZV6.4,10
1.5
4.1
3.5
3.2
4.2
6.6
3.7
9.3
11.6
27.0
21.9
42.3
63
25
20
44
41
72
sample the peak at 900 K ascribed to V O5, disappeared (profile not
2
shown). The broad TPR features centred at 640 K for ZVx,10Lz sam-
ples (Fig. 5A and 5B, curves b) and at about 600, 700 and 800 K
for ZVx,2.7Lz samples (Fig. 5C and 5D, curves b), originated from
the reduction of polymeric species strongly held on the zirconia
ZV2.7,10L1.4
ZV4.1,10L2.2
ZV4.8,10L2.2
2.1
2.2
2.9
5.7
5.5
6.3
32
34
30
surface or reacted to form ZrV O7 crystallites or both.
2
ZV 1.9,2.7
ZV 2.6,2.7
ZV3.7,2.7
ZV 5.9,2.7
ZV 7.4,2.7
ZV 7.6,2.7
2.9
3.2
5.5
18.8
36.5
23.6
10.7
10.8
24.0
13.8
8.9
40
31
64
59
60
64
3
.2. Catalytic behaviour for the n-butane oxidative
dehydrogenation
ZrO (823) was inactive. For ZVx,y catalysts, neither CO nor buta-
2
14.9
diene were detected in any experiment. 1-butene was sometimes
observed in traces. The C H8 fraction was almost entirely consti-
tuted by 2-butenes with cis/trans ratio of 0.8–0.85. Thus, hereinafter
ZV 5.9,2.7L2.3
ZV7.4,2.7L5.8
ZV7.6,2.7L5.7
6.5
19.6
15.0
10.8
5.9
11.6
52
39
49
4
“
C H ” or “butenes” will indicate the sum of cis and trans 2-butenes.
4 8
a
The oxygen conversion values did not exceeded 60%, except in one
For designation of samples, see text.

2
2
D. Gazzoli et al. / Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
Fig. 5. TPR profiles of some representative ZVx,10 samples before and after leaching.
Section A: (a) ZV2.7,10; (b) ZV2.7,10L1.4. Section B: (a) ZV5.4,10; (b) ZV5.4,10L3.0.
Section C: (a) ZV1.2,2.7; (b) ZV1.2,2.7L1.1. Section D: (a) ZV7.4,2.7; (b) ZV7.4,2.7L5.8.
For concentrated ZVx,2.7 catalysts (V atoms nm 2 > 7), the lev-
elling off of the areal rate may be explained by considering that the
increasing loading left the surface density of the exposed vanadium
species almost unchanged as the polymeric material increases in
thickness over the zirconia particles. This description is in agree-
ment with high resolution electron microscopy findings of Sanati
et al. [34] who showed that an amorphous structure of polymeric
vanadium material covered the zirconia surface starting from a
sample containing 5.3 wt.% of vanadium (apparent surface den-
−
−
2
sity = 7.94 V atoms nm ).
As for pure compounds, no literature data are available for n-
butane ODH activity of ZrV O7, whereas V O5 reportedly has some
Fig. 4. TPR profile of ZVx,10 (A) and ZVx,2.7(B) samples.
2
2
Section A: (a) ZV0.99,10; (b) ZV2.7,10; (c) ZV3.1,10; (d) ZV3.45; (e) ZV4.25,10; (f)
ZV5.4,10. Section B: (a) ZV1.2,2.7; (b) ZV2.6,2.7; (c) ZV5.9,2.7; (d) ZV7.4,2.7; (e)
ZV7.6,2.7; (f) ZV11.6,2.7; (g) ZrV2O7.
tion. For the ZVx,10 and dilute ZVx,2.7 samples (V atoms nm−2 < 7)
the areal rate of n-C H total conversion, Fig. 6, increased with V
4
10
loading, reaching the maximum value for an apparent surface den-
−2
sity of about 7 V atoms nm . For concentrated ZVx,2.7 catalysts
−2
(
V atoms nm > 7), the areal rate levelled off. Ammonia leaching
reduced the areal rate of n-C H total conversion for both series of
4
10
samples.
In vanadia-supported ZrO2 systems the most active sites for
the ODH of propane were associated with polymeric VOx species
[
19–21]. These sites were reported to increase with loading and a
maximum constant activity was reached, although this occurred
at a vanadium density [20] lower than that we found for n-butane
ODH. The increase in the areal rate of the n-C H total conver-
4
10
sion with the apparent surface density for the ZVx,10 and dilute
−2
ZVx,2.7 samples (up to 7 V atoms nm , x ≤ 3.7 wt.%), Fig. 6, sug-
gests that the polymeric species are the most active sites also for
n-butane ODH. As Raman and TPR determinations showed, these
samples contain an increasing amount of polymeric material; thus
the increasing activity appears correlated with a parallel increase
in exposed polymeric species.
Fig. 6. Areal rate for total n-C4H10 conversion at 603 K vs. apparent V surface density:
(᭹) ZVx,10; (ꢀ) ZVx,10Lz; (ꢀ) ZVx,2.7; (ꢁ) ZVx,2.7Lz; (ꢂ) ZrO2.

D. Gazzoli et al. / Journal of Molecular Catalysis A: Chemical 310 (2009) 17–23
23
activity for the same reaction [4]. In the present study the contri-
bution of V O5 and ZrV O7 to the samples activity are expected
to be low because of their low surface area (as estimated from
crystallite-size).
[7] G. Centi, Appl. Catal. A: Gen. 147 (1996) 267–298.
[
[
8] F. Cavani, E. Foresti, F. Trifirò, G. Busca, J. Catal. 106 (1987) 251–262.
9] P. Forzatti, Appl. Catal. A: Gen. 222 (2001) 221–236.
2
2
[
10] I.E. Wachs, J. Jehng, G. Deo, B.M. Weckhuysen, V.V. Guliants, J.B. Benziger, S.
Sundaresan, J. Catal. 170 (1997) 75–88.
11] J. Haber, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of Heteroge-
neous Catalysis, vol. 5, Wiley VCH, Weinheim, 1997, pp. 2253–2274.
12] M.D. Argyle, K. Chen, A.T. Bell, E. Iglesia, J. Catal. 208 (2002) 139–149.
[13] V. Bràzdová, M.V. Ganduglia-Pirovano, J. Sauer, J. Phys. Chem. B 109 (2005)
23532–23542.
14] M. Olea, I. Sack, V. Balcaen, G.B. Marin, H. Poelman, K. Eufinger, D. Poelman, R.
De Gryse, J.S. Paul, B.F. Sels, P.A. Jacobs, Appl. Catal. A: Gen. 318 (2007) 37–44.
[15] F. Klose, T. Wolff, H. Lorenz, A. Seidel-Morgenstern, Y. Suchorski, M. Piórkowska,
H. Weiss, J. Catal. 247 (2007) 176–193.
[
The decrease in the areal rate of the n-C H total conversion we
4
10
observed for the leached ZVx,10Ly samples (Fig. 6) agrees with the
given picture. In fact the leaching removed the active large poly-
[
meric aggregates and left ZrV O7 and polymeric species strongly
2
[
anchored to the support. Similar behaviour is expected for ZVx,2.7Ly
samples. For concentrated ZVx,2.7Ly samples the apparent sur-
face density remained high owing to the presence of a consistent
[
[
16] C.L. Pieck, M.A. Ba n˜ ares, J.L.G. Fierro, J. Catal. 224 (2004) 1–7.
amount of crystalline ZrV O7.
2
17] H. Poelman, B.F. Sels, M. Olea, K. Eufinger, J.S. Paul, B. Moens, I. Sack, V. Balcaen,
F. Bertinchamps, E.M. Gaigneaux, P.A. Jacobs, G.B. Marinc, D. Poelmana, R. De
Grysea, J. Catal. 245 (2007) 156–172.
Although we did not perform a detailed kinetic study and our
butene selectivity values refer to different n-C H total conver-
4
10
[
[
[
18] M. De, D. Kunzru, Catal. Lett. 96 (2004) 33–42.
sion, we can note that selectivity data (Table 2) for the concentrated
samples from both ZVx,10 and ZVx,2.7 series are higher than those
of the dilute catalysts. Moreover, the leaching decreased both n-
C4H10 total conversion and butene selectivity. In a study of the
propane ODH reaction, Chen et al. [23] found that the Zr–O–V sites,
formed by interaction between vanadium species and support sur-
face, favour alkane combustion. We suggest that the presence of
Zr–O–V sites, in the dilute and in the leached samples of both series,
could also account for the decreased butene selectivity.
19] A. Khodakov, J. Yang, S. Su, A.T. Bell, E. Iglesia, J. Catal. 177 (1998) 343–351.
20] J.L. Male, H.G. Niessen, A.T. Bell, T. Don Tilley, J. Catal. 194 (2000) 431–444.
[21] A. Khodakov, B. Olthof, A.T. Bell, E. Iglesia, J. Catal. 181 (1999) 205–216.
22] K. Chen, A. Khodakov, J. Yang, A.T. Bell, E. Iglesia, J. Catal. 186 (1999) 325–333.
23] K. Chen, A.T. Bell, E. Iglesia, J. Phys. Chem. B 104 (2000) 1292–1299.
24] A. Dinse, B. Frank, C. Hess, D. Habel, R. Schomäcker, J. Mol. Catal. A: Chem. 289
(2008) 28–37.
25] J.M. López Nieto, P. Concepción, A. Dejoz, F. Meloa, H. Knözinger, M.I. Vázquez,
Catal. Today 61 (2000) 361–367.
26] L.M. Madeira, M.F. Portela, Catal. Rev. Sci. Eng. 44 (2002) 247–286.
[
[
[
[
[
[27] M.A. Ba n˜ ares, I.E. Wachs, J. Raman Spectrosc. 33 (2002) 359–380.
[
[
28] O. Ovsitser, E.V. Kondratenko, Catal. Today 142 (2009) 138–142.
29] M. Valigi, D. Gazzoli, G. Ferraris, S. De Rossi, R. Spinicci, J. Mol. Catal. A: Chem.
227 (2005) 59–66.
4
. Conclusions
[
30] J.Ph. Nogier, Catal. Today 20 (1994) 23–34.
[
31] G. Centi, E. Giamello, D. Pinelli, F. Trifirò, J. Catal. 130 (1991) 220–237.
Our Raman and TPR data show that at low vanadium content,
[
[
32] B. Grzybowaska- S´ wierkosz, Appl. Catal. A: Gen. 157 (1997) 263–310.
V < 2.7 wt.%, the structure and spreading of the VOx species do not
depend on the pH (10 or 2.7) adopted for the sample preparation. In
both sets of samples the surface species consist mainly of polyvana-
date entities. For higher loadings, comparing samples with similar
concentration, the fraction of vanadium in a dispersed state and
interacting with the support is higher in samples prepared at pH
33] A. Cimino, S. De Rossi, R. Dragone, G. Ferraris, D. Gazzoli, M. Valigi, J. Colloid.
Interface Sci. 201 (1998) 278–292.
[
[
[
34] M. Sanati, A. Andersson, L.R. Wallenberg, B. Rebenstorf, Appl. Catal. A: Gen. 106
(
1993) 51–72.
35] R.C. Weast, M.J. Astle (Eds.), C. R. C. Handbook of Chemistry and Physics, 63rd
ed., C.R.C. Press, Inc., Boca Raton, Florida, 1982, p. B–162.
36] Powder Diffraction File, JCPDS-International Center for Diffraction Data Swarth-
more, Pennsylvania, 1991. File 41-1426.
1
0 than in those obtained at pH 2.7. In addition, the samples at pH
[37] Powder Diffraction File, JCPDS-International Center for Diffraction Data Swarth-
2
.7 contained ZrV O7 and V O5 along with large polymeric species.
more, Pennsylvania, 1991. File 16-422.
2
2
[
[
38] P.E. Thomas, W.F. Pickering, Talanta 18 (1971) 127–137.
39] I. Pettiti, D. Gazzoli, M. Inversi, M. Valigi, S. De Rossi, G. Ferraris, P. Porta, S.
Colonna, J. Synch. Rad. 6 (1999) 1120–1124.
Leaching with ammonia discloses that small sized polymeric struc-
tures interact with the support and ZrV O7 is barely detectable
2
by Raman spectroscopy at low V content on both sets of samples.
Relying on the literature findings for propane ODH, we propose
that V–O–V bridges could be associated with the active sites for
n-butane ODH, whereas the Zr–O–V sites could favour the alkane
combustion.
[40] K. Chen, S. Xie, E. Iglesia, A.T. Bell, J. Catal. 189 (2000) 421–430.
[41] M. Valigi, D. Gazzoli, I. Pettiti, G. Mattei, S. Colonna, S. De Rossi, G. Ferraris, Appl.
Catal. A: Gen. 231 (2002) 159–172.
[
42] X. Gao, I.E. Wachs, J. Phys. Chem. B. 104 (2000) 1261–1268.
[43] X. Gao, J. Jehng, I.E. Wachs, J. Catal. 209 (2002) 43–50.
[44] A. Christodoulakis, M. Machli, A.A. Lemonidou, S. Boghosian, J. Catal. 222 (2004)
293–306.
[
[
45] D.E. Keller, D.C. Koningberger, B.M. Weckuysen, J. Phys. Chem. B 110 (2006)
14313–14325.
46] C.V. Ramana, R.J. Smith, O.M. Hussain, M. Massot, C.M. Julien, Surf. Interface
Anal. 37 (2005) 406–411.
Acknowledgments
Financial support from Italian MIUR (PRIN, Progetti di ricerca
di rilevante interesse nazionale) is gratefully acknowledged. The
authors thank Mr. Giuseppe Piciacchia for his technical assistance
with the Raman Spectrometer.
[
47] I.E. Wachs, Catal. Today 27 (1996) 437–455.
[
[
[
48] D. Gazzoli, G. Ferraris, S. De Rossi, M. Valigi, L. Ferrari, S. Selci, Appl. Surf. Sci.
255 (2008) 2012–2019.
49] H. Bosch, B.J. Kip, J.G. van Ommen, P.J. Cellings, J. Chem. Soc., Faraday Trans. 1
(
80) (1984) 2479–2488.
50] G.C. Bond, S. Tair, Appl. Catal. A: Gen. 71 (1991) 1–31.
51] J. Pieck, S. del Val, M.L. Granados, M.A. Ba n˜ ares, J.L.G. Fierro, Langmuir 18 (2002)
642–2648.
[
References
2
[
[
[
52] F. Roozeboom, M.C. Mittelmeijer-Hazeleger, J.A. Moulijn, J. Medema, V.H.J. de
Beer, P.J. Gellings, J. Phys. Chem. 84 (1980) 2783–2791.
53] A. Adamski, Z. Sojka, K. Dyrek, M. Che, G. Wendt, S. Albrecht, Langmuir 15 (1999)
[
[
[
[
[
[
1] B.M. Weckhuysen, D.E. Keller, Catal. Today 78 (2003) 25–46.
2] H.H. Kung, Adv. Catal. 40 (1994) 1–35.
3] F. Cavani, F. Trifirò, Catal. Today 51 (1999) 561–580.
4] E.A. Memedov, V. Còrtes Corberàn, Appl. Catal. A: Gen. 127 (1995) 1–40.
5] T. Blasco, J.M.L. Nieto, Appl. Catal. A: Gen. 157 (1997) 117–142.
6] A.A. Lemonidou, L. Nalbandian, I.A. Vasalos, Catal. Today 61 (2000) 333–341.
5733–5741.
54] G. Deo, I.E. Wachs, J. Catal. 129 (1991) 307–312.