Novel microstructures and reactivity for n-butane oxidation: Advances and challenges in vapor phase alkane oxidation catalysis

Article abstract of DOI:10.1016/j.molcata.2004.03.053

The development of novel and improved micro structures was presented for vanadium phosphorus oxide catalysts for the oxidation of n-butane to maleic anhydride (MA) by grafting alkoxides of bismuth and molybdenum on to pre-formed vanadium phosphorus oxide precursor, VO(HPO₄)·0.5 H₂O. The bismuth/molybdenum grafted catalysts showed no change in selectivity to MA at 40% conversion, when these catalysts were evaluated in an oxygen rich, 2% butane/air mixture. A general decrease in MA selectivity was observed with increasing levels of bismuth and molybdenum for this catalyst both in 2% butane/air and 9% butane/10% O₂. The compositional gradient profile and the importance of the formation of a surface region depleted of some bismuth was further supported by the observation that catalysts prepared by co-precipitating the promoter cations with the vanadium phosphate precursor during precursor synthesis did not show a similar enrichment of molybdenum on the catalyst surface. Those systems exhibited smaller performance improvements, as a function of promoter level, than was observed for the supported catalyst systems. For those systems, the largest activity gains required higher levels of promoter cations and were accompanied by some loss in selectivity to MA.

Full text of DOI:10.1016/j.molcata.2004.03.053

Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
Novel microstructures and reactivity for n-butane oxidation: advances
and challenges in vapor phase alkane oxidation catalysis
Kostantinos Kourtakis, Pratibha L. Gai∗,1
Central Research and Development Laboratories, Experimental Station, DuPont Science and Engineering, P.O. Box 80356,
Wilmington, DE 19880-0356, USA
Received 20 January 2004; accepted 3 March 2004
Abstract
The development of novel and improved microstructures is reported for vanadium phosphorus oxide (VPO) catalysts for the oxidation
of n-butane to maleic anhydride (MA) by grafting alkoxides of bismuth and molybdenum on to pre-formed vanadium phosphorus oxide
precursor, VO(HPO4)·0.5 H2O. Depth profiling shows that the surface is molybdenum-rich, following a calcination and activation procedure
which converts the catalyst precursor to the final, active phase(s). This novel composition profile in the catalyst may be responsible for the
reactivity enhancements observed for this catalyst system. Catalysts prepared using bismuth or molybdenum containing precursors individually
do not show a performance enhancement.
© 2004 Elsevier B.V. All rights reserved.
Keywords: n-Alkane oxidation; Bi; Mo; V; P; O; Novel microstructures; Grafted catalysts
1. Introduction
(ETEM) studies of dynamic butane oxidation over VPO
at operating temperatures (∼360–390 ◦C) have shown the
direct evidence of catalyst surface structure modifications
leading to glide shear defects and active sites [2]. The stud-
ies have shown that anion vacancies are formed following
catalyst oxygen loss in butane oxidation and lie in the
basal plane between corner-sharing phosphate tetrahedra
and vanadyl octahedra. The glide shear defect regions cor-
respond to a slightly anion-deficient defect pyrophosphate
phase, with pyrophosphate groups sharing corners similar
to metaphosphate (PO3)n groups. The local glide defects
regions, which maintain the anion vacancies (V3+ phases),
do not change the overall VPO structure under the catalyst
operating conditions and as such are not visible in X-ray
diffraction (XRD). In situ ETEM and electron diffraction
have thus played a key role in unraveling the catalyst defect
structures in butane catalysis.
Several international groups, including ourselves, are
working to improve the performance of the catalysts by
adding various promoters. A number of cations (e.g. Fe, Cr,
Ce, Li, Zn) have been added [6–14] and the literature con-
nected to the promotion of VPO catalysts has been reviewed
recently by Hutchings and coworkers [11,14]. However, the
surface and bulk microstructural changes of the catalyst due
The selective oxidation of n-butane to maleic anhydride
(MA) over vanadium phosphorus oxides (VPO) is an im-
portant commercial process [1–12]. Maleic anhydride (MA)
is used as a raw material for a variety of products ranging
from tetrahydrfurans (THF) to agricultural chemicals. The
most successful industrial process to manufacture this com-
pound is the vapor phase oxidation of n-butane to MA over
active vanadyl pyrophosphate catalysts. In the selective ox-
idation of the alkane to MA, the most active and selective
phase is believed to be vanadyl pyrophosphate, (VO)2P2O7
catalysts (hereafter referred to as VPO), with active centers
located at the exposed (0 1 0) basal planes [1]. There are
also some claims that the active phase is a combination of
VPO and VOPO4 phases, or a P-rich phase, or a mixture
of VPO with unknown or amorphous phases [1,10]. Direct
in situ environmental transmission electron microscopy
∗
Corresponding author. Tel.: +1-302-695-9203;
fax: +1-302-695-1664.
E-mail address: pratibha.l.gai@usa.dupont.com (P.L. Gai).
1
(PL Gai also at: Department of Materials Science, University of
Delaware, Newark DE, USA).
1381-1169/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2004.03.053

94
K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
to promoters, and their effects on the catalytic reactivity are
not well understood. It is crucial, therefore, to elucidate the
interplay between the promoted structures and reactivity
to obtain a deeper understanding of the role of promoted
microstructures in alkane oxidation catalysis.
catalyst system. VPO is orthorhombic with a = 1.659 nm,
b = 0.776 nm and c = 0.9588 nm [15]. VPO is a layered
structure and the basal (0 1 0) plane, considered to be the
active plane, consists of edge sharing pairs of distorted
vanadium octahedra shared at corners by the phosphate
tetrahedral groups (Fig. 1).
Here we describe microstructures and chemical composi-
tions of promoted VPO catalysts for butane oxidation cataly-
sis. In this study, we investigate alternate synthetic pathways
for promoting vanadium phosphates, in order to improve
the catalysts further. To understand the observed catalytic
improvements, we also investigate the important relation-
ships between the catalyst microstructure and reactor per-
formance for a number of promoted vanadyl pyrophosphate
catalysts. Our goal is to obtain insights into the performance
improvements observed for these new catalyst systems to
guide further design efforts. To understand the complex
interactions of VPO with the cation promoters and their
effect on reactivity our studies have focused on the bismuth
and molybdenum promoted vanadyl pyrophosphates using
advanced electron microscopy, X-ray methods and parallel
reaction chemistry, which are described in the following
sections.
In addition to Bi and Mo, we have prepared other novel
compositions using promoter cations. Our approach has
been: (1) to substitute a lower valent cation such as Fe3+
or, Cr3+ for V4+, or (2) to replace V4+ with an equimolar
combination of cations of different valencies such as Fe3+
and Sb5+ or bismuth and molybdenum. We have also pre-
pared cerium oxide additions to the VPO catalyst to explore
the properties of the promoter as an oxygen reservoir in
the alkane oxidation process. In this paper, we discuss our
novel finding in the bismuth–molybdenum promoted VPO
2. Experimental procedures
2.1. Catalyst synthesis and characterization
A variation of an organic method was employed to syn-
thesize the “base” catalyst precursors [6,16]. Vanadium pen-
toxide and anhydrous phosphoric acid, prepared by mixing
P2O5 with aqueous phosphoric acid, are refluxed in 11:1
isobutyl alcohol/benzyl alcohol solvent mixtures under an
inert atmosphere. Enough anhydrous phosphoric acid to sat-
isfy a 1.1/1, P/V atomic ratio is slowly added to the alcohol
mixture over a period of 2 h. The solution is allowed to re-
flux an additional 15 h. Soluble pentavalent vandyl species
are reduced predominantly by benzyl alcohol to precipitate
the blue tetravalent precursor, which in the case of pure VPO
catalysts, crystallizes as VO(HPO4)·0.5 H2O. Excess phos-
phoric acid is washed from the blue filter cake with fresh
isobutyl alcohol. The precursors were subsequently dried
under vacuum at 150 ◦C for 18 h.
To create the novel grafted promoted vanadium phos-
phate catalyst, a mixture of bismuth and molybdenum alkox-
ide was used with vanadium hydrogen phosphate hemihy-
drate precursor. The alkoxides were prepared using bismuth
and molybdenum chloride; when reacted with ethanol, they
form a soluble mixed alkoxide complex. The bismuth alkox-
ide is insoluble in the absence of the molybdenum species,
suggesting that a soluble alcohol complex has formed con-
taining both bismuth and molybdenum. The mixed alkox-
ide, dissolved in alcohol, can be reactively grafted onto the
pre-formed vanadium phosphate precursor. The catalysts in
these experiments contain only 1 mol% bismuth and molyb-
denum supported on VOHPO4·0.5 H2O. This process is a
grafting procedure, since it is likely that the alkoxides of bis-
muth and molybdenum are reacting with the hemihydrate of
the precursor or the hydroxyl groups associated with the hy-
drogen phosphate, thereby forming covalent linkages of the
alkoxy species with the vanadium hydrogen phosphate. This
precursor complex is then heated to transform the precursor
to the final, promoted catalyst. Calcinations were performed
in a small, 3.5 cm fluidized bed on 60 g catalyst samples.
Samples were heated in air at 390 ◦C for 1 h, followed by
460 ◦C for 18 h in 1.5% butane/air. The relatively high tem-
perature of 460 ◦C was used for these calcination to for the
rapid equilibration of the catalysts.
Powder X-ray diffraction was carried out on the samples.
Vanadium oxidation state measurements, as determined by
redox titrimetry show small changes in vanadium oxidation
state after activation. Likewise, large changes in N2/BET
Fig. 1. 3D structural model of VPO catalyst along [0 1 0] with edge-shared
vanadyl octahedra interconnected by phosphate tetrahedra.

K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
95
surface area are not observed for these systems, suggest-
ing that the increased n-butane oxidation activity is due to
modification of active sites or an increase in the site density
(sites/m2) of the catalyst surface.
croscopy, catalysts were deposited on 3 mm carbon-filmed
copper or titanium microscope grids.
Both as-prepared powders, and cross-sectioned samples
were used for compositional depth profile in the samples.
First, we carried out analyses from the surfaces of the
as-prepared powders, by performing experiments at differ-
ent electron accelerating voltages in the LVSEM, from the
same regions to understand the differences in their surface
and bulk composition. The data were confirmed by analyses
on cross-sectioned samples; the catalyst particles were cross
sectioned to have access to the surface and to the interior
(core). EDX compositional analyses, from the surface of
cross-sectioned catalyst particles were then repeated at high
resolution in the STEM using electron nanoprobes. The
electron nanoprobe was moved gradually from the surface to
the core of the particle. Analysis spectra were recorded from
several dozen particles. The spectra were calibrated using
well-characterized Bi2MoO6 standards [9] and quantified.
The analyses provide high spatial resolution and increased
chemical sensitivity, uniquely, from localized regions and
the extent of penetration of the promoters into the bulk of
the catalyst and the microstructure across the catalyst.
The microstructures were correlated with parallel reac-
tion chemistry in a microreactor. Using these techniques,
we elucidate the structure and chemistry of modified cata-
lysts, the nature of catalyst–adsorbate interactions and their
role in catalytic selectivity. The synthetic techniques de-
scribed here have also been applied to robotic synthetic
methods and combinatorial tools, in which a number of
promoted vanadium phosphorus oxides can be prepared in
a parallel manner to cover a broad, promoter composition
[22].
To directly probe the catalyst microstructure, morphol-
ogy and composition, novel electron microscopy instrumen-
tation and techniques, developed by ourselves, have been
used. These include, the pioneering development of in situ
atomic resolution- environmental transmission electron mi-
croscopy (in situ ETEM) by Gai and co-workers to probe
gas–catalyst reactions directly at the atomic scale to provide
the dynamic atomic structural information and chemical evo-
lution in catalysts under working conditions [17,18,20,21].
A modified CM30 (S)TEM was used for our in situ ETEM
studies of catalysts under controlled reaction conditions to
yield a deeper understanding of active sites, transient states
and reaction mechanisms which could be directly related to
structure-activity relationships in practical reactors.
In the in situ ETEM studies, appropriate gas compositions
are used using a conventional reactor-type gas-manifold sys-
tem, which enables the inlet of flowing gases into the in situ
ETEM microreactor. A mass spectrometer is added for gas
analysis. Compositions of gases (e.g. 1.5% butane/air) in an
in situ ETEM microreactor are comparable to those in a re-
actor. A sample stage with a furnace allows samples to be
heated in the in situ ETEM. Lower gas pressures are used
for high resolution imaging. Higher gas pressures are pos-
sible with some loss of resolution due to multiple scatter-
ing of electrons through gas layers. Experiments on a large
number of catalysts [9] have shown that the gas layer in
immediate contact with the catalyst surface is the most crit-
ical to the catalytic reaction compared to the total amount
of gas present. We have used careful conditions in electron
microscopy to ensure that the material is not damaged in
the electron beam [9]. These include the use of very low
dose electron beam currents for imaging materials. The im-
age signal is then amplified by a low light television camera
connected to a video monitor, or recorded on a charge cou-
ple device (CCD) camera. Calibration ‘blank’ experiments
(without the electron beam) are carried out with the beam
switched on for a few seconds only, to record the image and
checked with in situ data. Under these careful conditions, no
invasive beam damage is observed. Parallel chemical reac-
tion experiments are performed in a microreactor on larger
amounts of samples.
2.2. Microreactor evaluations
A fixed bed microreactor was used for catalytic evalua-
tions. The microreactor consisted of a 40.64 cm by 0.64 cm
stainless steel tube enclosed by an aluminum sheath (3.12 cm
thick, to assist in minimizing thermal gradients) which itself
was enclosed in a tube furnace. The reactor was mounted
vertically with the direction of the gas flow from top to bot-
tom. Pressure gauges were mounted near both the entrance
and exit of the reactor to monitor pressure changes. A bypass
valve installed near the reactor entrance allowed the enter-
ing gas stream to pass through either the reactor or to bypass
the reactor and pass directly to the sampling valve system,
allowing analysis of the reactor feed gases prior to reac-
tion. Also, a T-connection at the top of the reactor allowed
a thermocouple to be mounted such that it monitored the
temperature at the entrance to the packed catalyst bed. Com-
mercially available mass flowmeters (Tylan Model FC-260)
of ranges 0–100 and 0–10 cc/min respectively, were used at
the reactor feed. The heated exit gas stream (200 ◦C) passed
through a heated sampling valve allowing the selection of a
gas sample (250 ␮l) of the stream for analysis by gas-liquid
chromatography (GLC), using commercially available in-
In addition high resolution TEM, scanning TEM (STEM)
with electron diffraction and electron stimulated energy
dispersive X-ray spectroscopy (EDX) for chemical analy-
sis were carried out using a high resolution CM 200 field
emission (FE)–(S)TEM system operating at 200 keV, using
low electron beam currents. Surface topography of cata-
lysts was investigated using a high resolution low voltage
SEM (LVSEM) using an advanced Hitachi SP5000 [19].
The (S)TEM and the SEM systems were fitted with en-
ergy dispersive X-ray spectrometers (EDX) and were used
for composition analyses of the catalysts. For electron mi-

96
K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
Fig. 2. XRD patterns (following calcination and activation in 1.5% butane/air at 460 ◦C), for 3 catalysts: (a) Calsicat vanadium hydrogen phosphate
precursor with 1 mol% alkoxide solution derived from bismuth trichloride and molybdenum pentachloride; (b) VPO catalyst with ethanol alone; (c)
control VPO catalyst. (Peaks, e.g. 020 and 402 are at 2θ (degrees) of ∼ 23 and 28.2, respectively).
struments (Gow-Mac Series 740P FID GLC). The GLC was
used to analyze for butane and for reaction products, e.g.
maleic anhydride, acetic acid, and acrylic acid.
The pseudo first order rate constant k, given for the dis-
appearance of butane, is obtained by fitting the reactor data
to a classical first order rate expression:
tors including the dependence of the reaction rate on oxy-
gen concentration (which remains relatively constant under
our conditions) and the concentration of catalyst active sites
(also assumed constant).
3. Results and discussion
d[butane]
= −k[butane]
dt
Powder X-ray diffraction data of Bi,Mo promoted cat-
alysts, catalysts with ethanol alone and control catalysts
are shown are shown in Fig. 2a–c, respectively. These
diffractograms show single phase crystalline vanadium py-
rophosphate. Surface area and redox titrimetry data are
listed in Table 1. Table 2 shows data for higher butane
concentrations.
As described in subsequent sections, improved reactor
performance is observed for the grafted promoted VPO sys-
tems. Detailed analysis is also described of the catalyst mi-
crostructure using high resolution depth profiling, and this
unusual catalyst microstructure which is evolved may be
related to improved reactor performance. The results have
d(x0 − x)
= −k(x0 − x)
dt
where x0 is the initial concentration of butane, x the portion
of butane reacted.
Integrating this expression gives the concentration of bu-
tane exiting the reactor as a function of contact time t, in the
reactor:
[exit butane] = x0 − x = x0 exp (−kt)
In addition to describing the reaction rate of butane with
a catalyst, the rate constant k, includes several other fac-

K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
97
Table 1
Fixed Bed Microreactor Data in 2% butane/air, 380 ◦C, of grafted catalyst systems compared to control catalysts
Catalyst composition
k (s−1
)
Selectivity at 40%
conversion
Vox
N2/BET SA(m2/g)
34
(A) 1 mol% Bi,Mo VO(HPO4)·0.5 H2O
(B) 1 mol% Bi,Mo VO(HPO4)·0.5 H2O (duplicate)
1.02
0.88
0.56
86
83
88
4.07/4.06
(C) EtOH on VO(HPO4)·0.5 H2O (control),
4.07/4.05
4.00/4.01
26
30
precursor before activation
(D) HCl + EtOH on VO(HPO4)·0.5 H2O (control),
0.49
0.56
83
86
precursor before activation
(E) VO(HPO4)·0.5 H2O (control), precursor before
activation
been compared with data from the co-precipitated promoter
catalyst.
parison is especially important since some HCl is released
during the synthesis of the bismuth/molybdenum alkoxide
complex, and the grafting experiments are performed in
ethanol. For these treated catalysts (D) and (E), the perfor-
mance was not improved compared to the control catalyst,
(E). N2/BET surface area measurements show nearly neg-
ligible changes in surface area, indicating that the perfor-
mance improvements of the grafted systems are most likely
related to an increase in the density of active sites on the
catalyst.
The bismuth/molybdenum grafted catalysts show no
change in selectivity to maleic anhydride at 40% conver-
sion, when these catalysts are evaluated in an oxygen rich,
2% butane/air mixture. Aqueous based precursor systems,
derived from dissolving bismuth nitrate and ammonium
molybdate in water followed by a similar impregnation
procedure onto pre-formed vanadium phosphate precursor,
do not improve the catalyst.
3.1. Microreactor evaluations in 2% butane/air
3.1.1. 1 mol% Bi, 1 mol% Mo supported on pre-formed
VPO precursor and comparisons
The catalyst systems described earlier (i.e. catalyst A)
in Table 1 was prepared by reactive grafting of the mixed
alkoxides of bismuth on molybdenum on pre-formed vana-
dium hydrogen phosphate precursor followed by calcina-
tion and activation at 460 ◦C in 1.5% butane/air. In Table 1,
fixed bed microreactor data obtained using a 2% butane/air
feed mixture are shown. The catalyst activity of the cata-
lysts grafted with 1 mol% of bismuth and molybdenum, af-
ter the calcination and activation heat treatments in 1.5%
butane/air, is shown for catalysts (A) and (B). Remarkably,
we find that the pseudo first order rate constant is approx-
imately 60–80% greater than that of the untreated, control
catalyst (catalyst E). Catalysts (C) and (D) are additional cat-
alysts which were prepared to determine the effects of sol-
vent (ethanol, for catalyst C) and aqueous hydrochloric acid
(catalyst D), on the performance of the pre-formed precur-
sor which is subsequently calcined and activated. This com-
A series of catalysts were prepared to determine the effect
of Bi/Mo ratio and loading on these catalyst systems. As
before, each catalyst system is subjected to a calcination
and activation heating procedure in 1.5% butane/air prior to
reactor evaluations. In this series of experiments, reactor data
in 2% butane/air and 9% butane/10% O2 are tabulated. For
Table 2
Fixed bed microreactor reactor, 2% butane/air, 9% butane/air at 380 ◦C
Catalyst
2% Butane/air
9% Butane, 10%O2
k (s−1
)
Selectivity at 40%
conversion
Yield
7.2
Selectivity
(1) Control catalyst (VOHPO4·0.5 H2O, subsequent to calcination/activation to
form the (VO)2P2O7 crystalline phase))
0.50
80
80
(2) 5BiMo (5 mol% Bi, 5 mol% Mo on VPO precursor (on VOHPO4·0.5 H2O))
(3) 1BiMo (1 mol% Bi, 1 mol% Mo on VPO precursor (on VOHPO4·0.5 H2O))
(4) Higher Bi (5.537 mol% Bi 0.35 mol% Mo on VPO precursor (on
VOHPO4·0.5 H2O))
0.89
1.12
1.84
66
87
62
6.2
10
12.8
60
70
75
(5) Higher Bi (9.788 mol% Bi, 0.492 mol% Mo on VPO precursor (on
VOHPO4·0.5 H2O)); highest Bi/Mo ratio possible using anhydrous chloride
alkoxides in EtOH
1.22
0.92
71
63
7.6
7.4
71
65
(6) Higher Mo (5 mol% Bi, 10 mol% Mo on VPO precursor (on VOHPO4·0.5
H2O))
(7) Mo only (1 mol% Mo on VPO precursor (on VOHPO4·0.5 H2O))
(8) 5 mol% Mo on VPO precursor (on VOHPO4·0.5 H2O)
0.91
1.22
90
71
9.2
7.6
75
60

98
K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
the more reducing, “oxygen lean” gas feed stoichiometries
(9% butane/10% O2), the yield and selectivity to maleic
anhydride at that condition is tabulated.
In depth profiling experiments, a comparison was made
between a catalyst formed by the grafting techniques de-
scribed in the preceding paragraphs and those formed by
co-precipitating bismuth and molybdenum precursors with
the vanadium hydrogen phosphate precursors during the or-
ganic reflux. As shown in Table 1, performance improve-
ments have been observed for the grafted systems in 2%
butane/air.
An analysis of these results shows very interesting and
remarkable trends which relate to the performance of these
catalysts in hydrocarbon lean conditions (2% butane/air) and
the hydrocarbon rich, less oxidizing (9% butane, 10% O2)
conditions. Catalysts grafted with molybdenum alone at 1
and 5 mol% concentrations (catalysts (7) and (8)) show im-
provements in reactivity compared to the control catalyst
(1) in 2% butane/air. The addition of bismuth to the 1 mol%
composition, as shown in catalyst (3), further improves cat-
alyst performance in 2% butane/air (two-fold increase in ac-
tivity). Bismuth rich compositions, as exemplified in catalyst
(4), show even larger (four-fold) improvements in reactivity,
but at the expense of selectivity at 40% conversion in 2%
butane/air. Perhaps most interesting however, is the perfor-
mance in 9% butane/10% O2. Under reducing conditions, a
significant increase in activity is observed and is accompa-
nied by only a small decrease in selectivity. This increase in
activity, at high selectivities, in the more reducing environ-
ments of 9% butane/air is greater than that observed for the
equimolar Bi,Mo composition in catalyst (3) or catalyst (2).
Other trends in this reactor data indicate that increasing
the catalyst loading from 1 mol% Bi, 1 mol% Mo (catalyst
(3)) to 5 mol% Bi, 5 mol% Mo (catalyst (2)) is not benefi-
cial. A general decrease in maleic anhydride selectivity is
observed with increasing levels of bismuth and molybdenum
for this catalyst both in 2% butane/air and 9% butane/10%
O2.
To summarize, microreactor data taken over 20 h of ex-
posure to the reaction stream suggests that in the more
oxidizing environments (2% butane/air), an equimolar com-
bination of bismuth and molybdenum is preferred. In a hy-
drocarbon rich environment, however (9% butane/10% O2),
a composition rich in bismuth, in our case with a Bi/Mo
ratio of roughly 16, is preferred (catalyst (4) in Table 2). In
either case, bismuth is a significantly beneficial co-catalyst,
and together with molybdenum, greatly enhances catalyst
activity while maintaining a high selectivity to maleic anhy-
dride. Bismuth may function to stabilize molybdenum (or
vanadium) in a more stable and selective oxidation state,
and this role may become more important under the more
reducing conditions of the 9% butane/10% O2 reactant
stream.
The synthesized catalysts were examined in both plan
view and in cross-section. The chemical analyses have
shown the presence of a primary phase, (V0.9 Bi0.05 Mo0.05
O)2 P2O7. The presence of this phase has been demonstrated
by ETEM structural analyses at the atomic scale and by EDX
spectroscopic composition analyses on the nanometer scale,
calibrated with Bi2MoO6 standards (gamma Bi2MoO6).
The phase represents a new composition isostructural with
vanadyl pyrophosphate. Figs. 3–5 show microstructural
morphology, ordered crystal lattice and chemical composi-
tion, respectively. Microstructural morphology and micro-
crystallinity of the grafted (1 mol% Bi,Mo)VPO catalyst are
shown in Figs. 3 and 4, respectively. The overlapping mi-
crocrystals, all of them showing lattice fringes, are clearly
seen Fig. 4. The presence of microcrystals is confirmed
by the selected area electron diffraction, which is inset in
Fig. 4. Overlapping crystals can create the so-called Moire’
fringes [23]. Fig. 5 shows an EDX chemical spectrum from
a general area of the catalyst, with the smaller Bi and Mo
promoter peaks in VPO. They are magnified in the inset.
The 1 mol% (Bi,Mo)VPO catalysts (catalyst A in Table 1)
generally contain variable size (i.e. diameter, or extent
across) ranges of plate-like catalyst crystals: many small
catalyst particles in the size range 50−150 nm and larger
particles upto 0.5−0.7 ␮m (Fig. 3). Cross-sectional sam-
ples and tilting experiments of crystals in the LVSEM were
employed to establish crystal (plate) thicknesses. The com-
positions of the surface and the interiors of the catalyst
3.2. Microchemical investigations: depth profiling
3.2.1. Comparison between grafted and co-precipitated
Bi,Mo-promoted VPO catalysts
Important relationships between the catalyst microstruc-
ture and reactor performance are being investigated for pro-
moted vanadyl pyrophosphate catalysts. An understanding
of these performance improvements observed for these new
catalyst systems will help guide further catalyst design ef-
forts.
Fig. 3. Microstructural morphology of our novel, grafted 1 mol% (Bi,Mo)
promoted-VPO catalysts using LVSEM. Catalysts contain crystals of vari-
able sizes, from 50 to 150 nm upto 0.5 ␮m.

K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
99
Fig. 4. Atomic resolution environmntal transmission electron microscopy (ETEM) image at room temperature (RT), showing a well ordered grafted
1 mol% (Bi,Mo) promoted-VPO. Crystalline nature of the 1 mol% (Bi,Mo)-promoted catalyst: The microcrystals are in the active (0 1 0) orientation and
are aligned randomly with respect to each other. The associated electron diffraction pattern (inset) confirms microcrystallinity.
particles were probed with a view to establishing the ex-
tent of penetration or depth of Mo and Bi species into the
interior of the catalyst particles and concentrations of the
promoters in the VPO structure were estimated.
crystal, from the surface to the center (interior) of the parti-
cles. Statistical errors for each data point in the plot are as
follows: in the depth, 0.4 nm and in the Mo/Bi ratios of
about 0.15. Plot A, shows the composition depth profile
for the grafted catalyst 1 mol% (Bi,Mo)VPO and plot B, is
that for co-precipitated (Bi0.05 Mo0.05)VPO. Fig. 7 shows
the compositional depth profile of Mo/V, confirming the sur-
face enrichment of Mo.
The precursor alkoxides, originally on the surface of the
vanadium phosphate precursor, decompose during the calci-
nation and activation procedure and diffuse from the surface
to the catalyst core. Essentially, no secondary phases are ob-
served in the promoted catalyst phase, either by high reso-
lution electron microscopy, or by powder X-ray diffraction.
However, detailed compositional depth profiling and micro-
chemical analysis reveal the final catalyst composition is not
completely homogeneous in bismuth and molybdenum. In-
deed, these studies show that the catalyst surface layers are
molybdenum rich and the core of the catalyst particles are
enriched in bismuth, in the 1 mol% (Bi,Mo)VPO catalyst.
Fig. 6 shows compositional depth profile of Mo/Bi, plotted
(with a scaling factor) as a function of depth in the catalyst
These changes in the depth profile distribution of catalyst
composition are not accompanied by changes in surface
area, as measured by N2/BET. The presence of the Mo-rich
phase may, therefore be significant and responsible for the
observed enhanced catalytic activity, by altering the nature
or introducing new active sites on the catalyst surface. As
described previously in [2], anion vacancies and glide de-
fects are introduced in VPO following catalyst oxygen loss
in butane oxidation, leading to V3+ phases. We believe that
4−␦
in the proposed global stoichiometric compound, (V0.9

100
K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
Fig. 5. Compositional spectrum of 1 mol% (Bi,Mo)VPO, using electron
stimulated, energy dispersive X-ray spectroscopy (EDX) in the electron
microscope. High precision compositional analyses at the nanoscale (using
electron nanoprobes) from several dozen crystals away from bars of the
grid supporting the sample and various catalyst regions have shown Bi
and Mo to be in solid solution with VPO, forming the promoted VPO
catalysts.
Fig. 7. Compositional depth profile of Mo/V of the grafted 1 mol%
(Bi,Mo)VPO is shown in plot A, confirming surface enrichment of Mo.
Data from the co-precipitated catalyst (bulk) are shown in B.
Bi0.05 Mo0.05 O)2P2O7−␦/2, anion vacancies resulting from
the substitution of Bi3+ in V4+ are countered by the higher
valent cation Mo6+, with the net effect of optimizing the
anion defect concentration and glide defect concentration
and maintaining the surface vanadium oxidation state. From
the compositional profile we have observed, the Bi–Mo
ratio varies from the surface into the bulk. In view of the
large ionic size differences between vanadium and bismuth,
it is unlikely that bismuth will be located on the vanadium
Fig. 6. Compositional depth profile of Mo/Bi ratios (scaled) plotted as a function of depth in the catalyst from the surface to the interior (core) of
cross-sectioned samples. Plot A is for grafted 1 mol% (Bi,Mo)VPO catalyst and plot B is for co-precipitated catalyst prepared by co-precipitation of
promoter cations with the precursor. The grafted catalysts show surface enrichment of Mo and the core enrichment of Bi. Statistical variations are
shown. The co-precipitated catalyst (B) does not show this behavior. We have estimated the promoter concentrations from the surface and the bulk of the
catalyst using standards. It may be noted that the data are derived uniquely, from localized regions of the sample which contains small catalyst particles.
This information is difficult to obtain from other molecular diffraction methods (especially when particle sizes are quite small), as they tend to average
information from a large area of sample.

K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
101
sites in the vanadyl pyrophosphate structure, except in com-
positional regions largely depleted of bismuth, which are
near the surface of the catalyst. The compositional gradient
profile and the importance of the formation of a surface
region depleted of some bismuth is further supported by
the observation that catalysts prepared by co-precipitating
the promoter cations with the vanadium phosphate pre-
cursor during precursor synthesis do not show a similar
enrichment of Mo on the catalyst surface. Those systems
exhibit smaller performance improvements, as a function of
promoter level, than is observed for the supported catalyst
systems. For those systems, the largest activity gains require
higher levels of promoter cations and are accompanied by
some loss in selectivity to maleic anhydride.
Co-precipitated catalyst systems are described here for
comparison with the supported promoter synthetic method
described earlier. Our investigations of the systems have also
shown a mixture of some minor phases, namely, VPO con-
taining only Mo, Bi–VPO compounds and undoped VPO
crystals. Mo–VPO crystals exhibited defective structures.
BiPO4, Bi5PO10 type and Mo-oxide phases have so far not
been detected in our study. In general, the co-precipitated
systems exhibit smaller performance improvements, as a
function of promoter level, than is observed for the grafted
catalyst systems. For those systems, the largest activity gains
require higher levels of promoter cations and are accompa-
nied by some loss in selectivity to maleic anhydride. Mor-
phologies of the surface and in the interior of the particles
were also examined in the cross-sectioned samples. This
catalyst composition is different from the calcined and ac-
tivated (Bi,Mo)-promoted VPO catalysts prepared by the
co-precipitated method which did not show compositional
depth profile, with Mo/Bi ratio of approximately 1 (Fig. 6,
plot B).
in the area can be made when the nature of the surface and
surface species of VPO catalysts with several other promot-
ers is elucidated under in situ conditions. By understanding
the nature of active species, more active catalysts (e.g. those
having higher surface areas, or placed on high surface area
supports) and perhaps more stable catalysts could be devel-
oped. Steps towards this goal have been realized by in situ
microscopy studies, as described in this paper, and in situ
EXAFS studies. For VPO chemistry, a second major chal-
lenge is to understand the role of the nature of the equili-
brated catalyst and the role played by activation conditions
on generating the final, equilibrated catalyst, and the effect
of catalyst compositional changes on the catalyst activation.
Large, parallel microreactor systems are probably best suited
to investigating this very large parameter space. Here too,
in situ studies can have an important role to play to advance
the field. In this paper, we have discussed the nature of an-
ion vacancies, especially near the catalyst surface, and the
marked effect on performance which was observed. These
vacancy structures may also be induced and perhaps sta-
bilized by investigating catalyst/support interactions. These
approaches may be crucial to creating even more active cat-
alysts, with higher productivities for this chemistry and for
other selective oxidations. Finally, we speculate that the sta-
bilization of furan intermediate in the butane oxidation and
its subsequent hydrogenation to THF (or THF in a single
step) by overcoming the thermodynamically favorable MA
production, can have a huge economic impact.
Acknowledgements
Pratibha L. Gai thanks Prof. G.J. Hutchings for the kind
invitation to prepare the manuscript. We thank L.G. Hanna,
F. Gooding and L. Wang of DuPont (CRD) for technical
support.
4. Conclusions
In summary, our novel findings indicate elevated levels of
Mo in the catalyst surface layers, and we believe that this
improved microstructure is related to the enhanced catalytic
performance. These surface compositional findings of the
promoted catalyst and the synthetic procedures are novel and
provide important insights into the surface microstructure
and the unusual composition of the catalyst.
References
[1] G. Centi, Catal. Today 16 (1993) 1.
[2] P.L. Gai, K. Kourtakis, Science 267 (1995) 661.
[3] G.J. Hutchings, et al., Catal. Today 33 (1997) 161.
[4] E. Bordes, Catal. Today 16 (1993) 27.
[5] P.L. Gai, Acta. Cryst. B53 (1997) 346.
[6] P.L. Gai, K. Kourtakis, R. Coulson, Jacobsen, J. Phys. Chem. 101
(1997) 9916.
[7] P.L. Gai, Topics Catal. 8 (1999) 91.
[8] P.L. Gai, Topics Catal. 21 (2002) 161.
[9] P.L. Gai, E.D. Boyes, Electron Microscopy in Heterogeneous Catal-
ysis, Institute of Physics, UK, 2003.
5. Future challenges
The oxidation of n-butane to maleic anhydride is a remark-
ably selective fourteen electron oxidation. To date, vanadium
phosphorus oxide based catalysts are still the most active and
selective system for this chemistry. It is likely however, that
other, highly active and selective catalysts exist with perhaps
different properties (stability, lattice oxygen capacity). How-
ever, to date none have been discovered. Significant progress
[10] M.T. Sananes, G.J. Hutchings, J. Volta, J. Catal. 154 (1995) 253.
[11] G.J. Hutchings, Appl. Catal. 72 (1991) 1.
[12] J. Haber, V. Zazhigalov, J. Stoch, L. Bogutskaya, I. Batcherikova,
Catal. Today 33 (1997) 39.
[13] S. Irusta, A. Boix, B. Pierini, C. Caspani, J. Petunchi, J. Catal. 187
(1999) 298.
[14] G.J. Hutchings, R. Higgins, J. Catal. 162 (1996) 153.
[15] Y. Gorbonova, Linde, Dockl. USSR, 245 (1979) 584.

102
K. Kourtakis, P.L. Gai / Journal of Molecular Catalysis A: Chemical 220 (2004) 93–102
[16] K. Kourtakis, et al., US 4,442,226.
[17] E.D. Boyes, P.L. Gai, Ultramicroscopy 67 (1997) 219.
[18] P.L. Gai, Adv. Mater. 10 (1998) 1251.
[19] E.D. Boyes, Adv. Mater. 10 (1998) 1277.
[20] J. Haggin, C&E News, Am. Chem. Soc. 73 (1995) 39.
[21] M. Jacoby, C&E News, Am. Chem. Soc. 80 (2002) 26.
[22] K. Kourtakis, P.L. Gai, Vanadium-phosphorus oxide catalysts with
promoter reagents, WO 2001 052983 20010726.
[23] P.B. Hirsch, et al., Electron Microscopy of Thin Crystals, Krieger,
NY, 1977.