Active sites in vanadia/titania catalysts for selective aerial oxidation of β-picoline to nicotinic acid

Article abstract of DOI:10.1016/j.jcat.2008.07.019

Vanadia/titania catalysts with varying vanadium content were prepared by impregnation using three different titania carrier materials of varying surface area. The structure of active vanadium species for β-picoline oxidation was investigated. Vanadium is mainly in the +5 oxidation state as revealed by electron paramagnetic resonance (EPR) and ⁵¹V magic-angle spinning nuclear magnetic resonance (⁵¹V MAS NMR) spectroscopy techniques. Diffuse reflectance UV-visible (DRUV-vis) spectroscopy and spectral deconvolution enabled identification of at least five different types of vanadium oxide species in these catalysts: monomeric tetrahedral VO^3-₄, polymeric distorted tetrahedral VO^-₃, square pyramidal V₂O₅, octahedral V₂O^2-₆ and V⁴⁺ oxide species. While both VO^3-₄ and VO^-₃ species are active in β-picoline oxidation, the latter having a distorted tetrahedral geometry yielded the desired products-picolinaldehyde and nicotinic acid. High surface area, anatase structure for the support and dispersed, distorted tetrahedral vanadium oxide species are the key parameters determining the activity and selectivity of these oxidation catalysts.

Full text of DOI:10.1016/j.jcat.2008.07.019

Journal of Catalysis 259 (2008) 165–173
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Active sites in vanadia/titania catalysts for selective aerial oxidation of β-picoline
to nicotinic acid
Darbha Srinivas a, Wolfgang F. Hölderich b,∗, Steffen Kujath b, Michael H. Valkenberg b, Thirumalaiswamy Raja a,
Lakshi Saikia a, Ramona Hinze b, Veda Ramaswamy a
a
National Chemical Laboratory, Pune 411 008, India
Department of Chemical Technology and Heterogeneous Catalysis, RWTH, Aachen, Germany
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Vanadia/titania catalysts with varying vanadium content were prepared by impregnation using three
different titania carrier materials of varying surface area. The structure of active vanadium species
for β-picoline oxidation was investigated. Vanadium is mainly in the +5 oxidation state as revealed
by electron paramagnetic resonance (EPR) and 51 V magic-angle spinning nuclear magnetic resonance
Received 17 July 2008
Revised 29 July 2008
Accepted 30 July 2008
Available online 9 September 2008
(
51 V MAS NMR) spectroscopy techniques. Diffuse reflectance UV–visible (DRUV–vis) spectroscopy and
spectral deconvolution enabled identification of at least five different types of vanadium oxide species
Keywords:
Vanadia/titania
−
in these catalysts: monomeric tetrahedral VO34−, polymeric distorted tetrahedral VO3 , square pyramidal
−
V2O5, octahedral V2O2− and V4+ oxide species. While both VO34− and VO3 species are active in
Active vanadium species
Oxidation of β-picoline
Nicotinic acid
Selective aerial oxidation
Spectroscopic investigations
β-picoline oxidation, t6he latter having a distorted tetrahedral geometry yielded the desired products—
picolinaldehyde and nicotinic acid. High surface area, anatase structure for the support and dispersed,
distorted tetrahedral vanadium oxide species are the key parameters determining the activity and
selectivity of these oxidation catalysts.
© 2008 Elsevier Inc. All rights reserved.
1. Introduction
starting materials, calcination temperature, nature of titania sup-
port, additives/impurities and the acid and base properties of the
Supported vanadium oxides have been extensively investi-
gated as interesting catalytic materials [1–11]. Among them, vana-
dia/titania have a high commercial potential as oxidation catalysts
in several industrial processes including oxidation of o-xylene to
phthalic anhydride, butane and pentane to maleic anhydride and
alkyl aromatics to the corresponding aldehydes and acids, oxidative
dehydrogenation of alkanes to olefins and oxidative destruction of
chlorinated hydrocarbons [1–6,12–15]. They are also used in selec-
tive catalytic reduction (SCR) of NOx with NH3 [16]. Their efficient
catalytic activity for the oxidation of β-picoline to nicotinic acid,
a pro-vitamin and an important intermediate in the production of
pharmaceuticals and food additives, with molecular oxygen was re-
ported by several of us [17]. High selectivity and conversion were
achieved. The nature of the TiO2 carrier had a marked influence
on the catalytic activity and selectivity [17]. We report here stud-
ies on the structure of active vanadium species in vanadia/titania
catalysts for this reaction.
support influence the surface vanadium species and catalytic ac-
tivity/selectivity. Despite various characterization studies of the
vanadia–titania catalysts by different techniques, the type and
structure of the surface vanadium species active in oxidation reac-
tions has not been completely understood [18–28]. Different vana-
dium structures viz., isolated vanadium species (in tetrahedral and
octahedral coordination), polymeric-type species (spread over ti-
tanium oxide) and bulk vanadium oxide (either in amorphous or
crystalline form) have been proposed to be present on the surface
in varying amounts. During the oxidation reactions, these struc-
tures can undergo conversion from one to the other. Which of
these species are active in specific oxidation reactions and how
their concentration and reducibility are affected by the reaction
conditions and the presence of hydrocarbon substrates are im-
portant questions. An understanding of the structure, state and
concentration of vanadium species should possibly enable devel-
opment of more efficient selective oxidation catalysts.
In general, structure and physicochemical properties have a
strong effect on the catalytic performance of these catalysts in
oxidation reactions [1–6]. The method of preparation, type of
A heterogeneous composition of vanadium species makes the
analysis more complex and challenging. However, by using a com-
bination of complementary spectroscopic techniques it should be
possible, in principle, to determine the structure of the active vana-
dium species. In view of this, nine different vanadia/titania catalyst
samples have been prepared by impregnation from three differ-
Corresponding author. Fax: +49 (0) 241 8022291.
E-mail address: hoelderich@rwth-aachen.de (W.F. Hölderich).
*
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.07.019

166
D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
ent titania support materials with varying amounts of vanadia.
The catalysts were characterized by X-ray diffraction (XRD), nitro-
gen adsorption/desorption, temperature-programmed desorption of
adsorbed ammonia (NH3-TPD), 51V magic-angle spin nuclear mag-
netic resonance (MAS NMR), diffuse reflectance UV–visible (DRUV–
vis) and electron paramagnetic resonance (EPR) spectroscopy tech-
niques. Different types of surface vanadium species have been
identified and quantified. A correlation between the concentra-
tion of different vanadium species and catalytic activity/selectivity
enabled determination of the active vanadium species in the oxi-
dation of β-picoline to nicotinic acid.
temperature powder XRD experiments. The high temperature XRD
data were recorded with an Anton Paar HTK 1600 attachment. Alu-
mina was used as a standard for calibration. A small amount of
the sample was mounted on a platinum strip, which served as the
sample stage as well as the heating element. A Pt/Rh–13% ther-
mocouple spot-welded to the bottom of the stage was used to
measure the temperature. The HT-XRD patterns were scanned in
◦
◦
the 2θ range of 20–80 with a step size of 0.02 and a scan rate
◦
of 1 /min. The measurements were done in the temperature range
of 298–623 K in static air with a heating rate of 10 K/min and a
soaking time of 10 min. Variable temperature measurements were
done at intervals of 25 K. All the experiments were performed in
static air and the sample stage temperature was calibrated using
thermal expansion of standard phase (α-Al2O3 standard from NIST,
Gaithersburg, USA). Reflections from the Pt-strip sample holder
have been used as an internal standard. This helps in determining
the sample height-displacement error. These sample displacements
at each temperature have been fixed for the refinement of param-
eters in the rutile and anatase titania phases. The XRD profiles
were refined using the X’pert plus refining package provided by
the Philips Co., to obtain the lattice parameters and phase compo-
sition.
Temperature-programmed desorption of ammonia (NH3-TPD)
was used to determine the acidities of the catalysts. For these in-
vestigations, a TPDRO 1100 with a thermal conductivity detector
(TCD) of Thermo Finnigan was used. The evaluation of the data
was carried out with the software—temperature-programmed de-
sorbed oxidation and pulse chemisorption (version 2.3) of Thermo
Electron.
2. Experimental
2.1. Materials
In the preparation of vanadia/titania catalysts, three different
TiO2 anatase materials—A, B, and C were used as supports. The
BET-surface area of A was 300 m2/g, while B and C had surface
areas of around 130 m2/g. The support material was mixed (for
60 min) with oxalic acid and water in a mass ratio of 71:4:25. The
mixtures were extruded (nozzle diameter of the extrusion machine
was 2.0 mm), dried (493 K for 2 h) and then calcined at 773 K for
5 h. The heating rate of the furnace was maintained at 1 K/min.
After grinding and sieving the extrudates, the created granulates
with diameter of 1–2 mm were impregnated with an aqueous so-
lution containing 3 wt% of ammonium metavanadate. After wet
impregnation and evaporation of excess water, the catalysts were
dried and calcined as described above. The impregnation and cal-
cination procedures were repeated up to two or three times to
obtain higher vanadia loadings. The resulting vanadia/titania (VT)
catalysts were designated as VT-X-Y , where X refers to the type
of titania support used in the preparation (A, B or C) and Y indi-
cates the impregnated vanadia content (5, 10 or 15 wt%).
51V Magic-angle spinning nuclear magnetic resonance (MAS-
NMR) studies were carried out on a DSX 500 Bruker spectrometer.
The spectra were recorded and processed with the Bruker software
packages XWIN-NMR and Topspin 1.3.
Diffuse reflectance UV–visible (DRUV–vis) spectra of the de-
hydrated, powdered samples were recorded at 298 K on a Shi-
madzu UV-2500 PC spectrophotometer. The spectral features due
to the support titania were subtracted from the spectra of the
vanadia/titania catalysts by using the support itself as a reference
material. For comparative studies spectra were also recorded us-
ing BaSO4 as a reference material. The spectra consisted of several
overlapping bands corresponding to different types of vanadium
species. They were all deconvoluted and the vanadium species
were quantified using the program Bruker WINEPR.
Electron paramagnetic resonance (EPR) spectra were recorded
on a Bruker EMX spectrometer operating at X-band frequency
(ν = 9.42 GHz) and a 100 kHz field modulation. The spectra were
recorded at 80 K using a Bruker BVTB 3500 variable tempera-
ture controller. The magnetic field was calibrated with a Bruker ER
035 M NMR gaussmeter. Microwave frequency was calibrated with
a frequency counter fitted in the Bruker ER 041 XG-D microwave
bridge unit. Spin Hamiltonian parameters were obtained by sim-
ulating the spectra using the Bruker Simphonia software package.
The samples were taken in specially designed EPR quartz cells hav-
ing a provision for adsorption/desorption studies. Prior to recording
EPR spectra the samples were degassed at 623 K for 4 h.
2.2. Characterization techniques
Inductively coupled plasma (ICP) analysis was carried out on
a Spectroflame ICP-D spectrometer using two monochromators in
the wavelength ranges of 165–460 and 240–790 nm, respectively.
The samples were crushed and dissolved in an aqueous solution
containing 4 vol% conc. H2SO4 and 10 vol% HF. The data were col-
lected and processed with the Spectro Smart Analyzer Software
(version 2.10).
Determination of surface area of the catalyst samples was per-
formed by BET measurement on a Micromeritics ASAP 2000 an-
alyzer. Prior to nitrogen adsorption, the samples were dried in
vacuum at 573 K. The data were recorded and processed using the
Micromeritics software ASAP 2010 (version 4.01).
Powder X-ray diffraction patterns of titania and vanadia/titania
samples were collected on a Philips X’Pert Pro 3040/60 diffrac-
tometer using CuK radiation (λ = 0.1542 nm). Ni-filter and an
α
X’celerator as detector which employs the real-time multiple strip
(RTMS) detection technique were used. The powder diffraction pat-
◦
terns were collected in the 2θ range of 15–95 with a step size
◦
of 0.017 , using a continuous scanning mode and with a scan time
per step of 50 s. The XRD patterns were refined by employing the
Rietveld refinement method [29] to determine the phase compo-
sition using the crystallographic data (both profile and structural
parameters) of anatase and rutile titania. The average crystallite
sizes of titania and vanadia were determined from the broaden-
ing of the XRD peaks corresponding to the (101), (110) and (001)
reflections of anatase/rutile titania and vanadia respectively, using
the Debye–Scherrer equation [30].
2.3. Reaction procedure
The catalyst (8 g) was placed as a fixed-bed in a spiral coil re-
actor made of stainless steel. The inner diameter of the reactor
was 6.0 mm, wall-thickness was 1.0 mm and reactor length was
1 m. The reaction temperature was 553 K and the absolute reaction
pressure was 0.5 bar. β-Picoline was passed through the reactor
with a WHSV of 1 h−1. Air was used as oxidant and the feed was
adjusted to generate an oxygen: β-picoline ratio of 1:1. The prod-
ucts were analyzed by gas chromatography (Siemens R GC 202;
The structural stability of the catalyst samples (VT-A-10, VT-B-
10 and VT-C-10) was determined by subjecting them to in situ high

D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
167
Table 1
Composition, average crystallite size, surface area and acidic properties of vanadia/titania catalysts
Catalysta
V-loading
(wt%)
Average crystallite size (XRD, nm)
NH3-TPD
(mmol/gcat
BET surface area (m2/g)
)
Support
Vanadium oxide
Before extrusion
Before loading vanadium
After loading vanadium
VT-A-5
VT-A-10
VT-A-15
VT-B-5
VT-B-10
VT-B-15
VT-C-5
1.54
3.99
5.90
2.15
4.00
6.52
1.85
4.22
6.02
29
38
38
27
32
36
30
40
40
–
–
31
–
–
45
–
–
129
129
129
130
130
130
300
300
300
63
63
63
64
64
64
65
65
65
38
10
5
41
11
7
29
8
7
0.112
0.039
0.029
0.131
0.042
0.030
0.097
0.035
0.031
VT-C-10
VT-C-15
32
a
Catalyst samples VT-A/B/C-10 and VT-A/B/C-15 were prepared by impregnating twice and thrice with ammonium metavanadate, respectively.
Fig. 1. XRD profiles of vanadia/titania catalysts. Reflections due to anatase and rutile phases are denoted as ‘a’ and ‘r’, respectively. Arrows indicate the reflections of crystalline
V2O5 phase.
a 50 m FS-CW-20M column and a FID detector) and high perfor-
mance liquid chromatography (Merck Hitachi with a UV–Vis L7420
detector and a Superspher 100 RP 18-4μ column; solvent—ethanol;
eluent—50 vol% H2O and 50 vol% of a Na3PO4 buffer solution with
pH 2.9). Operations were done at lower conversions to get a true
correlation between catalytic activity and selectivity and the active
site structure and concentration.
Table 2
Measured and theoretical crystal data of titania in vanadia/titania catalysts
Sample
Phase
Unit cell parameters (Å)
Volume of the
unit cell (Å3)
a
c
ICPDS21-1272a
ICPDS21-1276a
VT-A-10
Anatase
Rutile
Anatase
Rutile
Anatase
Anatase
Rutile
3.785
4.593
3.786
4.581
3.786
3.783
4.586
9.514
2.959
9.512
2.959
9.511
9.505
2.958
136.30(1)
62.42(2)
136.34(3)
62.36(7)
136.32(8)
136.02(6)
62.21(1)
VT-B-10
VT-C-10
3. Results and discussion
3.1. Influence of vanadium loading on support structure—room and
high temperature XRD studies
a
Reference material.
amounts of vanadium (VT-A-15, VT-B-15 and VT-C-15). The crystal-
lite size of anatase in VT-B-10 was 32 nm which was smaller than
that of VT-A-10 (38 nm) and VT-C-10 (40 nm). While the composi-
tions of the anatase and rutile phases in VT-A-10 were 58 and 42%,
respectively, the anatase phase concentration was higher (68%) in
the sample VT-C-10.
High temperature in situ XRD studies revealed that the cata-
lysts are highly stable up to a temperature of 643 K. Crystallite size
and phase composition did not vary appreciably at higher tem-
peratures. The variations in anatase and rutile phase compositions
were only marginal: 57–59% and 41–43% for VT-A-10 and 67–70%
and 30–33% for VT-C-10, respectively.
All the titania supports (A, B and C) used in this study were
in anatase form. Calcination at high temperatures (873 K) influ-
enced only their crystallite size and BET surface area (Table 1) but
not their structural phase. However, the presence of vanadia in-
fluenced the phase of titania which was dependent on the type
of the support (A, B and C) used. The vanadia/titania (VT) cata-
lysts prepared using the type-B titania source had the anatase-type
TiO2 phase whereas those prepared using A and C-type titania ma-
terials contained a mixture of anatase and rutile phases (Fig. 1).
The anatase form was prominent in VT-A-5 and VT-C-5 samples
(vanadium contents in these samples were 1.54 and 1.85 wt%, re-
spectively). On the other hand, VT-A-15 and VT-C-15, containing
about 6 wt% vanadium, had predominantly the rutile phase (Fig. 1).
At low concentrations, vanadium was in a highly dispersed state.
A crystalline V2O5 phase (Fig. 1, corresponding XRD reflection in-
dicated by arrows) was detected in the samples containing higher
Unit cell parameters of titania at 298 K in different samples
compared well with those of the standard values (Table 2). The
lattice had undergone thermal expansion with an increase in tem-
perature. The unit cell volume of anatase increased from 136.34(3)

168
D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
Fig. 2. DRUV–vis spectra of “neat” titania supports and vanadia/titania catalysts recording against BaSO4 as a reference material.
to 137.19(9) Å3 for VT-A-10, 136.32(8) to 136.10(1) Å3 for VT-B-
10 and 136.02(6) to 136.90(8) Å3 for VT-C-10 as the temperature
was increased from 298 to 598 K, respectively. The lattice or vol-
ume thermal expansion coefficient (αv ) values calculated from
the unit cell volume according to a published formula [31] are
The spectra of titania supports (A, B and C) along with those
of the catalyst samples having different loadings of vanadium ox-
ide (5–15%) recorded against BaSO4 as a reference material are
shown in Fig. 2. As seen from the figure, titania shows an intense
CT band in the region 230–400 nm. The supported vanadia shows
weak bands in the region 400–700 nm. Because of some overlap
with the bands of the support titania, those of vanadium oxide
are difficult to distinguish and quantify by using BaSO4 as the ref-
erence material [18–28]. Hence, in further studies, we have used
titania itself as a reference material and avoided spectral compli-
cations arising from the support. The support (A, B or C) used in
the preparation of a particular catalyst was used as the reference
material. As noted from the X-ray studies and textural properties
(Table 1), the vanadia loading effected the phase and surface area
of the support titania. Depending on the type of support (A, B
or C), a part of titania was transformed from the anatase to the
rutile phase. However, both of these structural forms of titania had
similar spectral features. Thus, even if there was some structural
transformation in the support due to the presence of vanadium
that did not affect the spectral pattern. The contribution from the
support was thus cancelled out by using titania itself as a refer-
ence material. Thus, the spectra reported in Fig. 3 are due to the
supported vanadium oxide only and not to the support.
The asymmetric broad absorption in the region 400–700 nm
(Fig. 3) indicates the presence of more than one type of vanadium
oxide species. In order to estimate the type and concentration of
the vanadium species, we have deconvoluted the DRUV–vis spec-
tra. A representative deconvolution plot along with experimental
and simulated curves for VT-A-10 is shown in Fig. 4. The satisfac-
tory fit indicates that the broad, asymmetric UV–vis curve could
be deconvoluted into five bands with band maxima at 397/403 (I),
440 (II), 472 (III), 550 (IV) and 668 (V) nm, respectively. The edge
position, representative of the optical band gap energy, of these
bands was determined in the classical fashion for the allowed
−1
−1
−1
−6
−6
−6
19.11 × 10
K
, 18.06 × 10
K
and 19.88 × 10
K in the
temperature range of 298–623 K for anatase in samples VT-A-10,
VT-B-10 and VT-A-10, respectively. The data obtained agree with
−1
−6
that of the literature (22.0 × 10
K
for anatase titania in the
temperature range 298–723 K) [31]. Similarly, the unit cell volume
of rutile phase increased from 62.36(9) to 62.84(3) Å3 for VT-A-10
and 62.20(6) to 62.67(5) Å3 for VT-C-10. The lattice or volume ther-
mal expansion coefficient (αv ) values calculated from the unit cell
−1
−1
−6
−6
K
volume are 23.38 ×10
K
and 23.19 ×10
in the temper-
ature range 298–623 K for rutile in VT-A-10 and VT-C-10 samples,
respectively.
3.2. Nature and type of vanadium species—diffuse reflectance UV–vis
spectroscopy
DRUV–vis spectroscopy is a powerful tool to elucidate the struc-
ture and types of vanadia species present in vanadia/titania cat-
alysts. The potential of this technique has been used in great
depth by Schoonheydt and co-workers [32] for the study of several
supported vanadium oxide catalysts. Haller and co-workers [33]
reported studies on vanadium-substituted mesoporous molecu-
lar sieves. The local environment of vanadium was investigated
by comparing the results with the spectra of model compounds,
which have a well-defined local symmetry. Among the model
compounds, V2O5 has nearly square pyramidal geometry around
vanadium consisting of four V–O bonds of similar length and a
very short V=O bond. Metavanadates (PbV2O6 and ZnV2O6, for
example) have a structure consisting of strongly distorted octahe-
dral pairs sharing corner oxygen atoms. Monomeric orthovanadates
(Na3VO4 and Mg3(VO4)2, for example) have a structure consist-
ing of isolated tetrahedrally coordinated vanadium ions in a nearly
symmetrical environment. Metavanadates such as NH4VO3, KVO3
and NaVO3 also have a tetrahedral environment, but strongly dis-
torted and sharing two bridging oxygen atoms with other poly-
hedra. DRUV–vis spectra of vanadium oxides are characterized by
ligand-to-metal (O → Vn+) charge transfer (CT) transitions and d–d
transitions, when n = 4 or 3. With increasing coordination number
the band maximum of the CT peak shifts to lower energy. Also
with increasing polymerization the band broadens and shifts to
lower energy.
transitions by finding the energy intercept of the straight line fit-
2
ted through the low-energy rise in the plots of [F(R ) × hν] vs
∝
hν, where F(R ) is the Kubelka–Munk function for the infinitely
∝
thick sample and hν is the energy of the incident photon. The
edge energy is sensitive to the coordination number and the struc-
ture of vanadium species. Monomeric tetrahedral VO34−, polymeric
−
tetrahedral VO3 , square pyramidal V2O5 and distorted octahedral
V2O2− species have UV edge energies (Eg) of 3.21, 3.20, 2.41 and
2.056 eV, respectively. From a comparison of the spectral data of
vanadia/titania catalysts with those of the standard samples (Ta-
ble 3), we conclude that the band I at 397/403 nm is due to
monomeric tetrahedral VO34− or polymeric distorted tetrahedral

D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
169
Fig. 3. DRUV–vis spectra of vanadia/titania catalysts recorded against the support titania (A, B or C) as a reference material.
were significant in the samples containing higher amounts of vana-
dia (Fig. 5).
3.3. Support-vanadia interactions: EPR spectroscopy
The catalyst samples showed weak EPR signals with resolved
eight line hyperfine features especially in the parallel region. The
intensity of the signals increased with increasing vanadium con-
tent and broadened at higher vanadia concentrations due to spin-
spin interactions (Fig. 6). In the preparation of catalysts, vanadium
with +5 oxidation state (ammonium metavanadate) was used as
the precursor. However, the final composition contained both +5
and +4 type vanadium species. In other words, the support influ-
enced and reduced a part of vanadium from +5 to +4 oxidation
state. Quantitative estimation using VO(SO4) as a reference sam-
ple indicated that only about 5% of total vanadium in VT-A-5 was
reduced from the +5 to the +4 state while the majority of the
vanadium was in +5 oxidation state. Only the reduced part of the
vanadium showed EPR signals. Vanadium in the +5 state was de-
tected by NMR spectroscopy. The EPR spectra were simulated using
a rhombic spin Hamiltonian:
Fig. 4. Deconvoluted plots along with the experimental and simulated DRUV–vis
spectra of VT-A-10.
Table 3
DRUV–vis spectral data and band assignments
H = β[gxx Bxx Sxx + gyy B yy S yy + gzz Bzz Szz
]
UV–vis λmax
Band
UV edge
Type of
Type of vanadium species
+ [Axx Sxx Ixx + Ayy S yy Iyy + Azz Szz Izz],
(1)
band
(nm)
width (nm) energy (eV) transition
−
I
II
III
IV
V
397–403 45
2.72
2.42
2.22
1.85
1.51
LMCT
LMCT
LMCT
d–d
Tetrahedral VO34− and VO3
where β is the Bohr magneton; (Bxx, B yy and Bzz) and (gxx, gyy
and gzz) are the magnetic field and g-parameters along x, y and
z directions, respectively; S is the electron spin angular momen-
tum with the components being Sxx, S yy and Szz along the x, y
and z directions, respectively. The spin Hamiltonian parameters are
listed in Table 4. The rhombic spin Hamiltonian parameters indi-
cate a low-symmetry environment around vanadium(IV) in vana-
dia/titania catalysts. The spectra of the VT-C series are broader and
more complicated than those of the VT-A series, indicating higher
amounts of agglomerated vanadium in the former series of cata-
lysts (Fig. 6).
440
472
550
668
60
80
100
125
Square pyramidal V2O5
Distorted octahedral V2O2−
6
Square pyramidal V4+
Square pyramidal V4+
d–d
−
VO3 species. The bands II and III at 440 and 472 nm are due to
V2O5 and V2O2−-type species, respectively. A small part of vana-
6
dium on titania is reduced to +4 state. Bands due to vanadium +4
species (IV and V) appeared at 550 and 668 nm.
A number of conclusions would be drawn from the spectra and
the deconvolution plots. The overall spectral intensity increased
with vanadium content (Fig. 3). Spectral deconvolution indicated
a marginal low energy shift of the band (I) at 397 to 403 nm.
This shift in band position indicates formation of higher amounts
The reducibility of vanadium was investigated by treating the
samples with dry-hydrogen. After reduction at elevated tempera-
tures, the intensity of the EPR signals due to isolated vanadium
species (in VT-A-5) decreased. A broad background signal appeared
indicating formation of clustered and interacting-type V4+/V3+
species (Fig. 7). Also during the oxidation reaction of β-picoline,
larger quantities of V4+ species have formed. The EPR spectrum
of the spent catalyst is shown in Fig. 8. The hyperfine features
were completely lost and only an intense broad signal was ob-
−
of polymeric tetrahedral VO3 at higher vanadium concentrations.
Samples with lower amount of vanadia (VT-A-5, VT-B-5 and VT-C-
5) contained mainly the monomeric VO34− species (397 nm). V2O2−
6
(472 nm; band III) and V4+ (550 and 668; bands IV and V) species

170
D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
Fig. 5. Variation in the areas of UV–vis bands corresponding to different vanadium oxide species as a function of vanadium content in vanadia–titania catalysts: (i) VT-A and
−
(ii) VT-C. Band I (397/403 nm; monomeric tetrahedral VO34−/polymeric distorted tetrahedral VO3 ), band II (440 nm; V2O5), band III (472 nm; V2O2−), band IV (560 nm, V4+
)
6
and band V (668 nm, V4+).
Table 4
EPR spin Hamiltonian parameters of vanadia/titania catalysts
Catalyst
gxx
gyy
gzz
Axx (G)
Ayy (G)
Azz (G)
VT-A-5/10/15
VT-B-5
VT-C-5/10/15
1.980
1.978
1.983
1.975
1.975
1.978
1.928
1.926
1.933
57.0
57.0
58.0
60.0
65.0
60.0
178.0
180.0
177.0
Fig. 7. EPR spectra of vanadia/titania catalyst (VT-A-5) treated with dry-hydrogen at
elevated temperatures.
Fig. 6. X-band EPR spectra of vanadia/titania catalysts as a function of vanadium
concentration.
served corresponding to formation of larger amounts of interacting
V4+ species during the oxidation reactions. Titania is a reducible
oxide. However, such a reduction of titanium from +4 to +3 state
was not observed under our experimental conditions.
3.4. Acidic properties of the catalysts—NH3-TPD
Temperature-programmed desorption of ammonia (NH3-TPD)
was used to investigate changes in the acidic properties of the ma-
terials. As shown in Fig. 9 and Table 1, the amount of highly acidic
sites decreased strongly when the support material was impreg-
nated with NH4VO3 a second or a third time. The amount of acidic
sites did not depend on the catalyst surface area (Table 1). The TPD
results show very strong similarities for the higher loaded materi-
als (VT-X-10 and -15). A real difference can only be found for the
Fig. 8. EPR spectra of fresh and spent catalysts: VT-A-15.

D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
171
complexity of the NMR signal indicates the presence of other
species, but these are too weak to be clearly identified.
The NMR spectra of all catalysts with higher vanadium loadings
(see VT-A-10 and VT-A-15 as examples in Fig. 10) are far less com-
plex. Here, only the peak assigned to bulk V2O5 is visible. MAS
NMR spectroscopy is not able to identify other vanadium oxide
species here. It has to be assumed that vanadium oxide completely
covers the surface in multiple layers. Spectra from VT-B-10 and
VT-C-10 showed the same result and are not presented here. The
intense NMR signals indicate that vanadium in vanadia–titania cat-
alysts is mainly in the +5 oxidation state.
3.6. Catalytic activity—oxidation of β-picoline
The catalytic activity of all previously described catalysts in
the oxidation of β-picoline to nicotinic acid was investigated. By-
products found in this reaction included pyridine-3-carbaldehyde
and maleic acid. The neat supports alone could not catalyze the re-
action. Turnover frequencies (TOF; Table 5) were estimated based
on the total amount of vanadium present in the catalyst as well
as the active vanadium sites (monomeric + polymeric tetrahe-
dral vanadium). The concentration of the active vanadium sites
(Table 5) was determined from the DRUV–vis spectroscopy and
spectral deconvolution. An assumption is made that the ratio of
the areas of the deconvoluted bands corresponds to the ratio of
the concentration of the corresponding vanadium species. The TOF
values estimated based on the active vanadium sites are higher by
two to three times than those estimated based on the total vana-
dium content in the catalyst. As shown in Table 5, experiments
with catalysts VT-A-5, VT-B-5 and VT-C-5 show the highest cat-
alytic activity. At the reaction conditions used here, the conversion
and TOF seem to be proportional to the catalyst’s BET surface area.
With these three catalysts, high amounts (19.0 to 27.4%) of the
side product maleic acid were found as well. This may be due to
the fact that the reaction is highly exothermic, so at higher con-
versions the side reaction is accelerated. The other catalysts lead
to lower conversions (1.2–3.1%), but only the aldehyde and nico-
tinic acid were observed as products. The decomposition product
maleic acid was not found, and the percentage of aldehyde in the
reaction products increased. A comparison of the different sup-
port materials shows a nearly linear dependency of conversion and
TOF on BET surface areas, independently of the support material
used. The acidity of support material or loaded catalyst showed
no correlation to the conversions and selectivities observed. It can
be deduced that the increase of the vanadium loading, at least
when it is as high as in these experiments, has no positive ef-
fect on the conversion. A similar observation can be made for the
selectivity to nicotinic acid. Selectivity decreases with increasing
vanadium content of the catalysts. A direct comparison of the cat-
alysts shows little difference in the product selectivities obtained
using catalysts with high vanadium loadings. Therefore, while the
pure support material is inactive for the reaction, repeated impreg-
nation and calcination cycles of the support material is detrimental
to both conversion and selectivity to the desired products. It may
be noted that the TOF values for oxidation of β-picoline are com-
parable those reported for the selective catalytic reduction (SCR) of
NO by NH3 over these catalysts [23].
Fig. 9. NH3-TPD of vanadia/titania catalysts.
catalysts, which had been impregnated only once. A calculation of
the amount of desorbed NH3 per surface area shows similar val-
ues for VT-B-5 and VT-C-5, while VT-A-5 displays a lower acidity.
A comparison to NH3-TPD measurements of the pure support ma-
terials (not shown here) revealed a strong decrease of highly acidic
sites for all materials. This indicates that surface acidity depended
mostly on the vanadium loading.
3.5. Nature and type of vanadium species—51V MAS NMR studies
3.7. Active vanadium species and structure—function relationships
Catalysts VT-A-5, 10 and 15, as well as VT-B-10 and VT-C-10
were characterized using magic-angle spinning (MAS) NMR tech-
niques. The NMR spectrum of VT-A-5 shows two main peaks. One,
with a δiso of −615 ppm in our spectra, has been attributed to
polycrystalline V2O5 in the literature [2,34,35], while the second
one, at about 640 ppm, can be assigned to strongly bound vana-
dium, which has a distorted octahedral oxygen environment. The
The composition of the anatase phase varies almost linearly
with the BET surface area. Further, the catalytic activity for β-
picoline oxidation correlates with the anatase phase content of the
catalysts (Fig. 11i). Hence, the structure of the support has a sig-
nificant bearing on the catalytic oxidation activity [36]. DRUV–vis
spectra revealed that there exist different types of vanadia species

172
D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
Fig. 10. 51 V MAS NMR of VT-A-5, -10 and -15.
Table 5
Catalytic activity of the different vanadia–titania catalysts in β-picoline oxidation with molecular oxygena
−1
Catalystb
Conversion
of β-picoline
(%)
Product selectivity (%)
Yield of
nicotinic
acid (%)
Turnover frequency (TOF, h
)
Per mol of
vanadiumc
Per mol of (monomeric +
polymeric) T d vanadiumd
Pyridine-3-
carbaldehyde
Maleic
acid
Nicotinic
acid
VT-A-5 (42.1)
VT-A-10 (30.4)
VT-A-15 (30.5)
VT-B-5 (34.3)
VT-B-10 (40.0)
VT-B-15 (34.6)
VT-C-5 (31.9)
VT-C-10 (35.6)
VT-C-15 (27.5)
10.3
1.9
1.2
11.3
3.1
2.0
6.5
2.3
1.9
3.7
0.3
0.1
2.9
0.4
0.2
1.9
0.3
0.2
8.7
0.9
0.3
8.4
1.1
0.5
6.0
0.8
0.6
20.4
59.5
59.4
20.3
67.8
63.1
29.2
57.4
53.3
27.4
0
0
23.5
0
0
19.0
0
0
52.2
40.5
40.6
56.2
32.2
36.9
51.8
42.6
46.7
5.4
0.8
0.5
6.3
0.5
0.7
3.4
1.0
0.9
a
Reaction conditions: Oxygen/β-picoline ratio = 1, WHSV = 1 h−1, pressure = 0.5 bar, temperature = 353 K.
b
c
Values in parentheses indicate percentage of active (monomeric + polymeric T d) vanadium sites estimated from DRUV–vis spectral data.
TOF = moles of β-picoline converted per mole of vanadium per hour.
d
TOF = moles of β-picoline converted per mole of active (monomeric + polymeric T d) vanadium sites.
Fig. 11. Correlation of β-picoline oxidation activity with (i) specific surface area (SBET) and anatase composition, and (ii) concentration of tetrahedral vanadium species.

D. Srinivas et al. / Journal of Catalysis 259 (2008) 165–173
173
on the surface of vanadia/titania catalysts. There is a correlation
References
(Fig. 11ii) between the oxidation activity and the concentration of
(monomeric + polymeric) tetrahedral vanadium oxide species (397
and 403 nm). The latter was estimated based on the areas of the
corresponding UV–vis bands. The higher the concentration of these
tetrahedral vanadium oxide species the higher was the β-picoline
oxidation activity. Catalysts with higher amount of vanadium con-
tained predominant amounts of inactive square pyramidal V2O5
(440 nm) and octahedral V2O2− (472 nm) species (Fig. 5), and as
a consequence catalytic activi6ty was low over these catalysts. As
the concentration of vanadium increased, the band at 397 nm had
shifted to 403 nm. This shift in band position associated with a
shift in the UV edge energy from 2.74 to 2.72 eV indicated the
formation of polymeric type tetrahedral vanadium oxides. While
both these vanadium oxides are active in β-picoline oxidation,
the latter having a distorted tetrahedral geometry is more selec-
tive and yields the desired products—picolinaldehyde and nicotinic
acid. Vanadium in the +5 oxidation state is the active precursor.
Thus, this study reveals that high surface area, anatase phase tita-
nia and dispersed, polymeric, distorted tetrahedral vanadium oxide
species are the determining factors leading to highly efficient and
[1] F. Cavani, C. Cortelli, A. Frattini, B. Panzacchi, V. Ravaglia, F. Trifirò, C. Fumagalli,
R. Leanza, G. Mazzoni, Catal. Today 118 (2006) 298.
[2] U.G. Nielsen, N.-Y. Topsoe, M. Brorson, J. Skibsted, H.J. Jakobsen, J. Am. Chem.
Soc. 126 (2004) 4926.
[3] H.K. Matralis, Ch. Papadopoulou, Ch. Kordulis, A.A. Elguezabal, V.C. Corberan,
Appl. Catal. A Gen. 126 (1995) 365.
[4] B. Grzybowska, J. Słoczyn´ ski, R. Grabowski, K. Samson, I. Gressel, K. Wcisło,
L. Gengembre, Y. Barbaux, Appl. Catal. A Gen. 230 (2002) 1.
[5] A. Bellifa, D. Lahcene, Y.N. Tchenar, A. Choukchou-Braham, R. Bachir, S. Bedrane,
C. Kappenstein, Appl. Catal. A Gen. 305 (2006) 1.
[6] B.M. Reddy, K.N. Rao, G.K. Reddy, P. Bharali, J. Mol. Catal. A Chem. 253 (2006)
44.
[7] A. Klisin´ ska, K. Samson, I. Gressel, B. Grzybowska, Appl. Catal.
(2006) 10.
A Gen. 309
[8] E.V. Kondratenko, M. Cherian, M. Baerns, D. Su, R. Schlögl, X. Wang, I.E. Wachs,
J. Catal. 234 (2005) 131.
[9] M.V. Martínez-Huerta, X. Gao, H. Tian, I.E. Wachs, J.L.G. Fierro, M.A. Bañares,
Catal. Today 118 (2006) 279.
[10] D. Shee, T.V. Malleswara Rao, G. Deo, Catal. Today 118 (2006) 288.
[11] J.L. Lakshmi, N.J. Ihasz, J.M. Miller, J. Mol. Catal. 165 (2001) 199.
[12] G. Deo, I.E. Wachs, J. Haber, Crit. Rev. Surf. Chem. 4 (1994) 141.
[13] B.M. Weckhuysen, D.E. Keller, Catal. Today 78 (2003) 25.
[14] A. Khodakov, J. Yang, S. Su, E. Iglesia, A. Bell, J. Catal. 177 (1998) 343.
[15] T.C. Watling, G. Deo, K. Seshan, I.E. Wachs, J.A. Lercher, Catal. Today 29 (1996)
139.
[16] I.E. Wachs, B.M. Weckhuysen, Appl. Catal. A Gen. 57 (1997) 67.
[17] D. Heinz, W.F. Hoelderich, S. Krill, W. Boeck, K. Huthmacher, J. Catal. 192 (2000)
1.
−
selective vanadia/titania catalysts. The tetrahedral VO3 and VO43−
(397 and 403 nm) are the active species in the selective catalytic
oxidations over vanadia/titania catalysts.
[18] J.P. Balikdjian, A. Davidson, S. Launay, H. Eckert, M. Che, J. Phys. Chem. B 104
(2000) 8931.
4. Conclusions
[19] E.C. Alyea, L. Jhansi Lakshmi, Z. Ju, Langmuir 13 (1997) 5621.
[20] H. Eckert, G. Deo, I.E. Wachs, A.M. Hirt, Colloids Surf. 45 (1990) 347.
[21] J.M. Miller, L. Jhansi Lakshmi, J. Mol. Catal. A Chem. 144 (1999) 451.
[22] L. Jhansi Lakshmi, E.C. Alyea, S.T. Srinivas, P. Kanta Rao, J. Phys. Chem. B 101
(1997) 3324.
[23] I. Giakoumelou, C. Fountzoula, C. Kordulis, S. Boghosian, J. Catal. 239 (2006) 1.
[24] J. Spengler, F. Anderle, E. Bosch, R.K. Grasselli, B. Pillep, P. Behrens, O.B. Lapina,
A.A. Shubin, H.-J. Eberle, H. Knözinger, J. Phys. Chem. B 105 (2001) 10772.
[25] C. Gheorghe, B. Gee, Chem. Mater. 12 (2000) 682.
[26] G. Silversmit, H. Poelman, I. Sack, G. Buyle, G.B. Marin, R. De Gryse, Catal.
Lett. 107 (2006) 61.
Vanadia/titania catalysts with varying vanadium content were
prepared using three different titania carrier materials. The struc-
ture of titania changed from the anatase to the rutile phase during
the preparation of the catalysts. This change in phase composition
depended on the vanadium content and the source and acidity of
the titania material. The active vanadium sites in these catalysts
in β-picoline oxidation were investigated by XRD and DRUV–vis,
EPR and 51V MAS NMR spectroscopy techniques. The catalysts con-
tained at least five different types of vanadium oxide species viz.,
monomeric tetrahedral VO34− (397 nm), polymeric distorted tetra-
[27] A. Brückner, U. Bentrup, J.-B. Stelzer, Z. Anorg. Allg. Chem. 631 (2005) 60.
[28] U. Bentrup, A. Brückner, C. Rudinger, H.-J. Eberle, Appl. Catal. A Gen. 269 (2004)
237.
−
hedral VO3 (403 nm), square pyramidal V2O5 (440 nm), octahedral
[29] H.M. Rietveld, Acta Crystallogr. 22 (1967) 151;
V2O2− (472 nm) and V4+ oxide (550 and 668 nm) species. While
6
H.M. Rietveld, J. Appl. Crystallogr. 2 (1969) 65.
[30] H.P. Klug, L.E. Alexander, X-ray Diffraction Procedures: For Polycrystalline and
Amorphous Materials, Wiley, New York, 1974, 618.
[31] N. Jagtap, M. Bhagwat, P. Awati, V. Ramaswamy, Thermochim. Acta 427 (2005)
37.
[32] G. Catana, R. Ramachandra Rao, B.M. Weckhuysen, P. Van Der Voort, E. Vansant,
R.A. Schoonheydt, J. Phys. Chem. B 102 (1998) 8005.
[33] D. Wei, H. Wang, X. Feng, W.-T. Chueh, P. Ravikovitch, M. Lyubovsky, C. Li,
T. Takeguchi, G.L. Haller, J. Phys. Chem. B 103 (1999) 2113.
[34] O.B. Lapina, A.A. Shubin, A.V. Nosov, E. Bosch, J. Spengler, H. Knötzinger, J. Phys.
Chem. B 103 (1999) 7599.
−
both VO34− and VO3 species (397 and 403 nm) are active in β-pic-
oline oxidation, the latter with a distorted tetrahedral geometry,
yields the desired products (picolinaldehyde and nicotinic acid).
High surface area, anatase structure and dispersed, distorted tetra-
hedral vanadium oxide species lead to more active and selective
oxidation catalysts.
Acknowledgments
[35] V.V. Terskikh, O.B. Lapina, V.M. Bondareva, Phys. Chem. Chem. Phys. 2 (2000)
2441.
[36] D. Habel, J.B. Stelzer, E. Feike, C. Schröder, A. Hösch, C. Hess, A. Knop-Gericke,
J. Caro, H. Schubert, J. Eur. Ceram. Soc. 26 (2006) 3287.
This work was carried out under the DST-DAAD PPP exchange
program. The authors thank DST, New Delhi, India and DAAD, Ger-
many for the financial support.