A substrate-versatile catalyst for the selective oxidation of light alkanes I. Reactivity

Article abstract of DOI:10.1016/S0021-9517(03)00008-3

The reactivity of niobium and pyridine-exchanged molybdophosphoric and molybdovanadophosphoric acid catalysts were studied for the selective oxidation of light alkanes. The catalysts were active and selective for the selective oxidation of propane and n-butane over a wide temperature range and under both hydrocarbon-rich and oxygen-rich reaction conditions. For the selective oxidation of n-butane to maleic acid, the presence of vanadium did not significantly influence the reactivity. In contrast, for the selective oxidation of propane to acrylic acid, the presence of vanadium significantly affected the selectivity to the desired product. The selective oxidation of propane also led to the production of maleic acid as a major product under certain reaction conditions. For the selective oxidation of n-butane, the productivity of the catalyst was greater than that of vanadium phosphorous oxide. With propane, the productivity of acrylic acid exceeded that of MoVNbTeO_x. The catalysts are also active for selective oxidation of other substrates, e.g., ethane, isobutane, and toluene.

Full text of DOI:10.1016/S0021-9517(03)00008-3

Journal of Catalysis 218 (2003) 42–53
www.elsevier.com/locate/jcat
A substrate-versatile catalyst for the selective oxidation of light alkanes
I. Reactivity
Joseph H. Holles,a Christopher J. Dillon,a Jay A. Labinger,b and Mark E. Davis a,∗
a
Chemical Engineering, California Institute of Technology, Pasadena, CA 91125, USA
Beckman Institute, California Institute of Technology, Pasadena, CA 91125, USA
b
Received 17 September 2002; revised 11 December 2002; accepted 16 December 2002
Abstract
The reactivity of niobium and pyridine-exchanged molybdophosphoric (NbPMo pyr) and molybdovanadophosphoric (NbPMo Vpyr)
12
11
acid catalysts are investigated for the selective oxidation of light alkanes. Productivities and selectivities are presented for the reactions of
◦
propane to acrylic acid and n-butane to maleic acid. Pretreatment of the material to 420 C in an inert atmosphere is required to form an
active catalyst, and the combination of both niobium and pyridine is necessary for obtaining maximum activity. Vanadium is not essential
for the selective formation of maleic acid from butane, but does appear to affect the selective formation of acrylic acid from propane. Other
metal cations can be used instead of niobium; however, none are as effective as niobium. Comparison of activity results to those from
other molybdate-based structures indicates that high activity requires the use of a reducible molybdenum species. In addition, the central
phosphorus heteroatom of the polyoxometalate is important for high activity. Structures without phosphorus or with other heteroatoms have
lower activity. The catalysts are effective under both hydrocarbon-rich and oxygen-rich conditions. For butane, the productivity to maleic
acid exceeds that of the current industrial standard vanadium phosphorus oxide. With propane, the productivity to acrylic acid is above that of
the MoVNbTe mixed-metal oxide. In addition to efficiently catalyzing the selective oxidation of propane and butane, the catalysts are active
with feeds containing other substrates such as ethane, isobutane, and toluene.
2003 Elsevier Inc. All rights reserved.
Keywords: Heterogeneous catalysis; Selective oxidation; Propane; n-Butane; Polyoxometalate; Keggin unit; Wells–Dawson unit; Strandberg unit; Niobium;
Pyridine
1. Introduction
still the only example of an industrial, selective catalytic
oxidation process that uses an alkane as a feedstock [3].
Other selective oxidation reactions of interest include the
production of acetic acid from ethane and acrylic acid from
propane. In contrast to n-butane, the VPO catalysts are not
effective for these substrates [4]. For ethane, the most com-
mon catalyst system is a MoVNb mixed-metal oxide that
produces mostly ethylene [5]; however, titania-supported
VP oxides have recently been reported to produce signifi-
cant amounts of acetic acid [6]. The most active catalysts
for conversion of propane to acrylic acid are even more
complex mixed-metal oxides such as MoVNbTe [7,8] and
MoVNbSb [9].
Heteropolyacids and their polyoxometalate derivatives
have received extensive study for many different catalytic
applications. These materials are often effective due to their
bifunctional nature as both acid and oxidation catalysts. In-
deed, polyoxometalates, and phosphomolybdates, in partic-
ular, have been extensively investigated for the selective
Catalytic selective oxidation processes are a mainstay of
the modern chemical industry. Today, more than 60% of
the chemicals and intermediates synthesized via catalytic
processes are products of oxidation [1]. One area of selective
oxidation that has received a continuing focus is the selective
oxidation of light alkanes [2]. This is due to the abundance
and low cost of the light alkanes such as ethane, propane,
and butane.
The conversion of n-butane to maleic anhydride over
vanadium phosphorus oxide (VPO) catalysts with molecular
oxygen is a well-established commercial process [1]. This
process typically achieves conversions of around 75% and
selectivities near 67%. Selective oxidation of n-butane is
*
Corresponding author.
E-mail address: mdavis@cheme.caltech.edu (M.E. Davis).
0021-9517/03/$ – see front matter
doi:10.1016/S0021-9517(03)00008-3
2003 Elsevier Inc. All rights reserved.

J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
43
oxidation of light alkanes [10–12]. Most of these studies
have focused on the effect of different exchange cations or
the incorporation of different amounts of vanadium in the
polyoxometalate structure. An advance in this field is the
work of Ueda and Suzuki, who demonstrated that a molyb-
dovanadophosphoric acid (denoted as PMo11V) treated with
pyridine resulted in a active and selective catalyst for the
conversion of propane to acrylic acid [13]. Subsequently, Li
and Ueda found that treating molybdophosphoric acid (de-
noted as PMo12) with pyridine and then activating in nitro-
oxalate (0.064 g) was dissolved in water (20 mL); the solu-
tion was then added to added to PMo12 (0.75 g) dissolved
in water (40 mL) and pyridine-exchanged. For VPMo12pyr,
vanadium pentoxide V2O5 (0.046 g) was dissolved in wa-
ter (40 mL) and then added to oxalic acid in water (0.136 g
in 10 mL). This solution was then added to PMo12 (1.5 g)
dissolved in water (40 mL) and pyridine-exchanged. For
CrPMo12pyr, chromium trioxide CrO3 (0.051 g) was dis-
solved in water (10 mL) and then added to oxalic acid in
water (0.182 g in 10 mL). This solution was then added
to PMo12 (1.5 g) dissolved in water (40 mL) and pyridine-
exchanged.
◦
gen at 420 C improved activity for the selective oxidation
of propane to acrylic acid [14].
Recently, we reported on niobium and pyridine-exchan-
ged molybdophosphoric and molybdovanadophosphoric ac-
id catalysts (NbPMo12pyr and NbPMo11Vpyr) for selective
oxidation of propane and butane [15]. Here, we describe fur-
ther this first example of a substrate versatile catalyst for se-
lective oxidation. Part I reports the reactivity studies in order
to define the components that are necessary for good activ-
ity and selectivity. Part II then provides characterizations of
the better catalytic materials illustrated in Part I.
2.1.2. NbPMo11Vpyr
Molybdovanadophosphoric acid (PMo11V) was prepared
by a known literature method [16]. Sodium hydrogen phos-
phate Na2HPO4 (7.098 g) was dissolved in 100 mL distilled
water at room temperature. Sodium metavanadate◦NaVO3
(6.097 g) was dissolved in 100 mL of water at 80 C. The
two solutions were combined at room temperature and acid-
ified with concentrated sulfuric acid (5 mL) which led to
a dark red solution. Sodium molybdate dihydrate Na2MoO4 ·
2H2O (133.1 g) was dissolved in 200 mL water and then the
two solutions were combined, whereupon the Keggin unit
self-assembled in solution.
2. Experimental
2.1. Catalyst preparation
The sodium form of the Keggin anion was proton-
exchanged by adding concentrated sulfuric acid (85 mL) to
the solution for 90 min (exotherm) with vigorous stirring.
This generated a clear bright red solution that was added
to diethyl ether (200 mL) in a separating funnel and
shaken. Three phases resulted, from which the dark red
layer was decanted and retained. The aqueous layer was
extracted twice with ether, and the ethereal fractions were
combined. The diethyl ether was removed under vacuum
resulting in a thick red syrup that was dissolved in water
and recrystallized. The orange crystals of PMo11V were
dried overnight at 100 ◦C, and then niobium- and pyridine-
exchanged as above.
2.1.1. NbPMo12pyr
Molybdophosphoric acid (PMo12) was purchased from
Aldrich. Niobium pentachloride (1.210 g) was dissolved in
water (10 mL) and the solution basified with ammonium
hydroxide (0.5 mL). The white precipitate was removed by
filtration and dissolved in aqueous oxalic acid (1.008 g in
20 mL water). The niobium oxalate solution was then slowly
added to PMo12 (20 g) dissolved in water (40 mL). The
mixture was stirred and heated to 80 ◦C until the liquid
had evaporated. An aqueous solution of pyridine (1.645 g in
10 mL water) was slowly added to a slurry of the resulting
niobium polyoxometalate (NbPMo12) (5.260 g) in water
(30 mL) to form an immediate precipitate. The mixture was
stirred and evaporated to dryness at 80 ◦C.
Other metal-exchanged phosphomolybdates (MPMo12-
pyr) were prepared similarly. For MoPMo12pyr, ammo-
nium molybdate (NH4)2MoO4 (0.082 g) was dissolved in
water and stirred with Dowex 50 × 8-200 proton form
ion-exchange resin (0.17 g) for 1.5 h. The resin was fil-
tered off and oxalic acid (0.073 g) was added. The molyb-
denum oxalate solution was then added to molybdophos-
phoric acid (1.5 g) and exchanged with pyridine as above.
For ZrPMo12pyr, zirconium sulfate Zr(SO4)2 (0.292 g) was
dissolved in water (20 mL) and basified with ammonium
hydroxide. The white precipitate was removed by filtration
and dissolved in oxalic acid (0.187 g in 20 mL water).
The zirconium oxalate solution was then slowly added to
PMo12 (1.5 g) dissolved in water (40 mL) and evaporated
and pyridine-exchanged as above. For TiPMo12pyr, titanium
2.1.3. NbP2Mo18pyr
Sodium hydrogen phosphate Na2HPO4 · 12H2O (7.14 g)
was dissolved in a mixture of perchloric acid (75 mL) and
◦
water (50 mL). The solution was cooled to 20 C. Then a
solution of sodium molybdate Na2MoO4 · 2H2O (108.27 g)
in water (200 mL) was added dropwise to the above solution.
The resulting solution was clear yellow. The solution was
allowed to evaporate in air and after 2 weeks yellow crystals
of Na6P2Mo18O62 formed [17]. Formation of the Wells–
Dawson unit was confirmed by single-crystal diffraction.
The crystals were filtered off and proton-, niobium-, and
pyridine-exchanged as above.
2.1.4. NbP2Mo5pyr
Sodium molybdate Na2MoO4 · 2H2O (24.682 g) and
sodium dihydrogen phosphate NaH2PO4 · 2H2O (6.369 g)
were added to 10.5 mL of perchloric acid. Distilled water

44
J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
was added to increase the total volume to 50 mL and the
solution was stirred to aid dissolution. The resulting solution
was clear. The solution was allowed to evaporate in air and
after 2 weeks clear crystals of Na6P2Mo5O23 formed [18].
Formation of the Strandberg unit was confirmed by single-
crystal diffraction. The sodium form was then converted to
an exchangeable form by dissolving 4.008 g in 20 mL water
and adding Dowex 50 × 8-200 ion-exchange resin (1.567 g)
and stirring for 90 min to form Na4H2P2Mo5O23. The resin
was filtered off and the sample subsequently niobium- and
pyridine-exchanged as above. Stability of the Strandberg
unit throughout the synthesis procedure was confirmed
by the presence of a 31P NMR peak at approximately
2 ppm [19].
heating to 420 ◦C for 5 h and then held for 6 h at that
temperature before cooling to 380 C in flowing He. This
◦
treatment removes all organic components from the catalyst
precursors [15]. Feed gases included n-butane (99.9%,
Matheson), oxygen (99.5%, Air Liquide), and 5% argon in
helium (99.999%, Air Liquide). Standard flow rates were
4:2:4:5 mL min−1 of n-butane:oxygen:argon/helium:steam.
Water was injected to the reactor feed flow via a syringe
pump to form steam. Total pressure in the reactor was
atmospheric and no pressure drop across the reactor was
observed. Standard reaction temperature was 380 ◦C.
Feed to the reactor was initiated and allowed to come to
steady-state for 1 h. At this point sampling was initiated.
Reported data are the average values for the second hour
on stream. Reaction experiments were typically performed
for 10 h, however, a single run with propane was accom-
plished for 72 h without loss of activity. Reactant and prod-
uct analysis was by GCMS using a HP GCD Plus with a
HP-Plot Q column. Gas analysis was performed online while
oxygenated products were trapped in an ice bath and an-
alyzed offline. Experiments using ethane (99.99%, Math-
eson), propane (99.98%, Matheson), and isobutane (99%,
Matheson) were performed similarly. For the toluene exper-
iment, the helium/oxygen reactant was fed through a bub-
bler containing toluene (99.8%, Aldrich) at room tempera-
ture. Product selectivities are reported in mole percentages.
Surface area measurements (BET) were performed on
a Coulter Omnisorp 100 using nitrogen (99.996%, Air
Liquide) adsorption at 77 K. Surface areas were determined
after pretreating to 420 ◦C.
2.1.5. NbMo8pyr
The procedure of Gili et al. was adapted for pyridine [20].
An aqueous solution of pyridine (1.931 g in 120 mL of
distilled water) was added to an aqueous suspension of
MoO3 (3.213 g of MoO3 in 1800 mL of distilled water).
The mixture was heated under reflux with stirring for
6 h, cooled in an ice bath, and filtered. The solution
was then reduced to 600 mL by a rotary evaporator and
allowed to stand overnight whereupon a white precipitate
(C5H5NH)4(Mo8O26) formed. The crystals were filtered,
washed with water, and dried. Powder diffraction and TGA
curves were identical to those of McCarron et al. [21]. The
sample was niobium-exchanged as above.
2.1.6. PMo11Nbpyr
Molybdophosphoric acid (2.7 mmol; 5.00 g) was dis-
solved in water (27.4 mL). The pH of the solution was then
adjusted to 4.4 using Li2CO3 (0.583 g) to form a lacunary
Keggin ion structure [22]. One molar equivalent of niobium
oxalate was added to the solution. The solution was stirred
for 2 h to form molybdoniobophosphoricacid PMo11NbO40.
Pyridine was added (0.216 g) and a yellow precipitate was
formed, filtered off, and dried.
Elemental analysis was performed by Galbraith Labora-
tories Inc. (Knoxville, TN).
3. Results
3.1. n-Butane reactivity
Previously, the importance of vanadium, niobium, and
pyridine on the n-butane reactivity was determined by in-
vestigating a series of catalysts that were systematically pre-
pared to incorporate all combinations of the three compo-
nents [15]. Results from those and further studies of this
catalyst system are reported in Table 1 (products trapped in
water that presumably convert maleic anhydride to maleic
acid). All results are for 0.2 g catalyst at 380 ◦C under
hydrocarbon-rich conditions (C4/O2 = 2/1). The samples
that contain both niobium and pyridine are the most active
and selective (15% butane conversion is the maximum possi-
ble under these conditions). These materials are also capable
of maintaining high conversion and selectivity even when in-
creasing the total flow rate by a factor of 8 (compare lines 5
and 6). In contrast, the samples that do not contain pyridine
exhibit very low activity. The solids that contain pyridine but
no niobium have intermediate activity, and are not capable
of maintaining high conversion under increased flow rates
2.1.7. (VO)PMo11Nbpyr
From molybdophosphoric acid (5.00 g), molybdoniobo-
phosphoric acid was made in solution as described above,
to which an aqueous solution of vanadyl oxalate (0.213 g)
was added to form a dark green solution. The addition of
pyridine (0.216 g) to the solution caused the precipitation of
a pale green solid that was removed by filtration and allowed
to air dry.
2.2. Reactivity studies
Experiments were performed in a BTRS Jr. reactor
(Autoclave Engineers) with a stainless-steel reactor tube.
A 0.2 g (∼ 0.2 mL) catalyst (35–60 mesh) was mixed
with 1 mL silicon carbide (16 mesh, Abrasives Unlimited)
and held in the reactor with glass wool. Reactant flow
was single pass. The catalyst sample was pretreated by

J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
45
Table 1
Catalyst reactivity using n-butane as substrate
b
c
Catalyst
Flow
Conversion (%)
Selectivity (%)
STY
−1 a
(mL min
)
C
4
O
CO
Ac
AA
MA
x
2
PMo
NbPMo
4:2:4:5
4:2:4:5
4:2:4:5
32:16:32:40
4:2:4:5
32:16:32:40
64:32:64:80
4:2:4:5
4:2:4:5
4:2:4:5
4:2:4:5
32:16:32:40
64:32:64:80
0.2
3
13.5
2.5
15
15
9
0.5
0.4
13.5
15
7
64
13
14
9
25
5
17
31
42
5
16
9
34
7
3
6
3
2
2
1
6
1
1
13
0
4
1
0
70
82
71
71
90
49
50
25
90
76
80
57
0
12
32
89
17
100
100
60
9
0.017
0.097
0.11
0.084
0.84
12
PMo pyr
12
NbPMo pyr
12
3
12
17
19
3
5
5
0.56
−3
−4
PMo
V
1.6 × 10
6.3 × 10
0.11
0.090
0.76
11
NbPMo
V
9
11
PMo Vpyr
90
100
95
82
11
NbPMo Vpyr
2
3
3
11
14
12.3
33
5
0.89
◦
Some of the data in this table have been previously reported [15]. All at 380 C with 0.2 g catalyst.
a
n-Butane:oxygen:helium:water.
b
C , n-butane; CO , carbon oxides; Ac, acetic acid; AA, acrylic acid; MA, maleic acid.
x
4
c
−1
−1
Space-time yield of maleic acid (mmol min (g of catalyst) ).
◦
(compare lines 3 and 4). Surface area measurements for the
series of materials listed in Table 1 are reported in Table 2.
The surface area of the most active catalyst, NbPMo11Vpyr,
is roughly a factor of 2 greater than the acid form of the ma-
terial (PMo11V).
showing that over the temperature range 300 to 420 C the
selectivity to maleic acid remains at 60% or higher.
Table 4 compares the reactivity of n-butane over the nio-
bium/pyridine polyoxometalate catalysts to several of the
catalyst systems reported in the literature, including both
commercial VPO and the best heteropolyacid-based cata-
lysts. The niobium/pyridine catalysts exhibit substantially
higher space-time yields than VPO with similar selectivities.
Even under the more typical oxygen-rich industrial condi-
tions, the new catalyst performs well. When compared to the
heteropolyacid-based catalysts from the literature, the nio-
bium/pyridine system is dramatically more active.
In order to determine whether the Keggin structure
(Fig. 1a) of the niobium/pyridine phosphomolybdate is
essential for the activity, a series of structural variations, all
based on MoO6 octahedral units, were examined. Results
from bulk MoO3, the octamolybdate species (Mo8O26)
(Fig. 1b), the Wells–Dawson (P2Mo18) unit (Fig. 1c), and
the Strandberg (P2Mo5) unit (Fig. 1d) are given in Table 5.
MoO3 alone shows little activity and no selectivity to maleic
acid, while a physical mixture of molybdenum, vanadium,
and niobium oxides produces a selectivity to maleic acid
of 20%. The octamolybdate with niobium and pyridine
produces around half the activity obtained for NbPMo12pyr.
The activity of the Wells–Dawson unit is essentially the
same as the Keggin unit while the Strandberg unit gives no
selectivity to maleic acid at all.
The active catalysts containing both niobium and pyri-
dine perform well under different reaction conditions, as
shown by the data in Table 3. The first three entries demon-
strate the effect of changing the oxygen partial pressure in
a hydrocarbon-richreactant mixture. These studies were per-
formed at low temperatures and low levels of conversion in
order to prevent complete conversion of oxygen. Comparing
entries 1 and 2 shows that doubling the oxygen partial pres-
sure by changing the flow rate from 8 to 16 mL min−1 results
in doubling the conversion of butane and space-time yield of
maleic acid, while reducing the oxygen flow to 2 mL min−1
caused no significant change in butane conversion and some
decrease in space-time yield. Entries 4 and 5 in Table 3 list
the activity of NbPMo11Vpyr under conditions comparable
to those used in industry (VPO catalysts are used in con-
siderably more oxygen-rich reaction mixtures at high butane
conversion levels). Under these conditions, COx and maleic
acid each accounted for around half of the product effluent at
340 and 300 ◦C. The remaining entries in Table 3 provide ad-
ditional reactivity data for NbPMo11Vpyr and NbPMo12pyr
Table 2
Surface area measurements
Instead of ion exchanging niobium into the sample, we
have been able to incorporate it in the Keggin unit as a frame-
work atom. This incorporation of niobium into a phospho-
molybdate heteropolyacid (PMo11Nbpyr) is new and is and
discussed in Part II [23]. The productivity from this sample
(lines 10 and 11) was similar to the niobium-exchangedKeg-
gin analogue or Wells–Dawson unit. A central silicon atom
in place of the phosphorus atom in the Keggin unit results
2
Sample
Surface area (m /g)
PMo
PMo
7.7
5.9
7.0
17
4.8
10
12
V
11
NbPMo
V
11
PMo Vpyr
11
NbPMo Vpyr (as prepared)
11
NbPMo Vpyr
11
◦
All samples, except as noted, activated at 420 C.

46
J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
Table 3
n-Butane reactivity under different reaction conditions
b
Catalyst
Flow
T
Conversion (%)
Selectivity (%)
STY
−1 a
◦
(mL min
)
( C)
C
4
O
CO
Ac
AA
MA
x
2
c
c
c
d
d
d
−2
NbPMo Vpyr
11
16:8:16:20
16:16:16:20
16:2:16:20
1:10:35:5
1:10:35:5
2:10:35:5
32:16:32:40
8:4:8:10
300
300
300
300
340
340
340
340
300
420
1.8
3.7
2.1
27
25
62
33
61
81
90
95
75
46
34
26
33
51
50
54
5
24
26
22
6
6
7
5
4
4
4
11
12
3
2
2
1
1
1
1
1
3
2
3
50
61
50
44
46
41
89
62
60
68
6.4 × 10
NbPMo Vpyr
0.14
11
−2
−3
−2
−2
NbPMo Vpyr
11
NbPMo Vpyr
11
NbPMo Vpyr
11
3.7 × 10
6.4 × 10
1.9 × 10
3.0 × 10
0.76
29
62
63
13.5
14
9
NbPMo Vpyr
11
c
NbPMo pyr
12
c
c
c
NbPMo Vpyr
NbPMo Vpyr
NbPMo Vpyr
11
0.14
0.05
2.37
11
4:2:4:5
256:128:256:80
11
6.9
a
n-Butane:oxygen:helium:water.
Space-time yield of maleic acid (mmol min (g of catalyst) ).
0.2 g catalyst.
0.99 g catalyst.
b
c
d
−1
−1
in a high selectivity to COx and a low space-time yield of
maleic acid.
Table 7. Ammonia and hydrogen pretreatments had little ef-
fect while CO was able to increase the activity by two orders
of magnitude compared to a pyridine-free sample. However,
this is still one order of magnitude lower than the pyridine-
containing sample. Attempts to combine the pyridine and
CO activation were unsuccessful, resulting in a decrease in
activity.
In addition to niobium, other metals can be ion-exchanged
to the PMo12pyr, including Ti, V, Cr, Fe, Cu, Y, Zr, Sb, Ce,
and Bi. Results for the most active of these samples are re-
ported in Table 6. The niobium catalyst is still the most ac-
tive, by a factor of 2 compared to titanium and molybdenum.
Chromium and zirconium also demonstrated some activity.
Although vanadium and niobium are in the same column
in the periodic table, the vanadium catalyst did not perform
nearly as well.
Further attempts to optimize the catalyst activity by
varying the amount of niobium and pyridine present are
shown in Fig. 2. Catalyst samples were prepared with 0.04
to 0.98 niobium atoms per phosphorus atom (or Keggin unit)
with the pyridine balancing the remainder of the charge.
The most active sample contained 0.30 niobium atoms per
Keggin unit, with all samples having similar selectivities.
Instead of using pyridine to activate the sample, other
reducing atmospheres were investigated. Carbon monox-
ide, ammonia, and hydrogen gas were used to pretreat
NbPMo11V. Results from these experiments are shown in
3.2. Propane reactivity
Propane reactivity was determined over a series of cat-
alysts where each of the three different additives was sys-
tematically incorporated. In addition, a material was pre-
pared that incorporated niobium into the structure and was
ion-exchanged with the vanadium species. Results for these
samples are shown in Table 8. As with butane, the cata-
lysts that contain both niobium and pyridine are the most
active [15]. However, in contrast with butane, the pres-
ence of vanadium does strongly affect the reactivity re-
sults primarily by increasing the selectivity to acrylic acid.
The method of introducing vanadium may be important as
well: (VO)PMo11Nbpyr gives somewhat lower selectivity to
Table 4
Comparison of n-butane reactivity to literature data
c
Catalyst
Flow
T
Conversion (%)
Selectivity (%)
STY
Rate
Ref.
−1 a
◦
b
−1 −2
(mL min
)
( C)
C
4
O
CO
MA
(mmol min m )
x
2
−2
NbPMo pyr
12
NbPMo Vpyr
11
VPO
VPOCo
32:16:32:40
1:10:35:5
4:6:26:0
4.2:2.4:18.4:0
6.3:58:218:34
0.6:5.4:24:0
380
340
380
400
360
340
15
62
30
25.5
31.8
90
100
61
–
77.9
–
–
5
50
10
46.2
37
–
90
46
80
11.5
32
0.84
0.019
0.067
0.022
1.3 × 10
2.8 × 10
8.4 × 10
1.9 × 10
3.0 × 10
7.9 × 10
–
This work
This work
[27]
[30]
[11]
−3
−3
−4
2+
−3
−3
BiPMo + 2VO
12
PMo
V
38
–
[10]
10
2
Some of the data in this table have been previously reported [15].
a
n-Butane:oxygen:helium:water.
MA, maleic acid or maleic anhydride.
Space-time yield of maleic acid (mmol min (g of catalyst) ).
b
c
−1
−1

J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
47
(a)
(b)
Keggin unit, (b) Mo O
(c)
6−
octamolybdate, (c) P Mo O
(d)
Wells–Dawson unit, and (d) P Mo O
5 23
3−
4−
6−
Fig. 1. Anion structures: (a) PMo
O
12 40
8
26
2
18 62
2
Strandberg unit. Molybdenum atoms are located near the center of the white octahedra with oxygen atoms at the corners. A phosphorus atom is located
at the center of the shaded tetrahedron with oxygen atoms at the corners.
acrylic acid than NbPMo11Vpyr. Note that these catalysts
produce substantial amounts of maleic acid in addition to
the expected products of acetic and acrylic acid.
most selective, it has a very high space-time yield compared
to the other catalysts.
3.3. Other reactants
Selectivity also depends on flow rate. As shown in
entries 1 and 2 of Table 9, increasing the total flow rate
fourfold causes the selectivity of acrylic acid to increase
from 21 to 49%, while selectivity to maleic acid decreases
from 43 to 15%. Good selectivities and yields are also
obtained at lower temperatures (Table 9).
Table 10 compares the reactivity of propane over the nio-
bium/pyridine polyoxometalate materials to several catalysts
reported in the literature. These include the MoVNb-based
mixed-metal oxides as well as a heteropolyacid-based cata-
lyst. For comparison, reactivity results with the NbPMo11
Vpyr sample under oxygen-rich conditions are also in-
cluded. Since the literature data are at different temperatures,
flow rates, and residence times, direct comparisons are dif-
ficult. However, while the NbPMo11Vpyr catalyst is not the
Reactivity data for ethane, isobutane, and toluene are
shown in Table 11. Comparisons to other catalyst systems
reported in the literature for each of the three substrates
are also included. For ethane, the space-time yield to ethyl-
ene and acetic acid from the NbPMo11Vpyr compare well
to vanadium oxide supported on titania and the MoVNb
mixed-metal oxide catalysts. Other MoVNb-based systems
specifically designed to produce ethylene or acetic acid are
also shown. Similarly, the NbPMo11Vpyr catalyst has a good
space-time yield of methacrylic acid from isobutane. How-
ever, the selectivity is very low for this sample (mostly deep
oxidation products). Finally, space-time yields to benzoic
acid and benzaldehyde from toluene are less than those re-
ported for a AgVCe mixed-metal oxide.
Table 5
Reactivity of solids containing Mo-oxide structural variations
b
Catalyst
Flow
Conversion (%)
Selectivity (%)
STY
−1 a
(mL min
)
C
4
O
CO
MA
x
2
MoO
4:2:4:5
4:2:4:5
4:2:4:5
4:2:4:5
27:16:32:40
4:2:4:5
64:32:64:80
4:2:4:5
64:32:64:80
4:2:4:5
0.7
12
13.5
15
5
15
7.4
15
7.4
15
12.5
2
11
20
95
89
100
49
100
52
100
52
100
83
25
80
95
60
14
50
28
22
19
26
16
12
36
81
69
0
20
82
37
56
71
52
65
52
62
39
0
0
3
c
Phys. Mix.
PMo pyr
0.023
0.097
0.044
0.219
0.084
0.52
0.077
0.52
0.073
0.61
0
12
NbMo O pyr
8
26
NbPMo pyr
12
NbP Mo pyr
2
18
PMo Nbpyr
11
64:32:64:80
4:2:4:5
4:2:4:5
NbP Mo pyr
2
5
NbSiMo pyr
12
18
0.017
◦
All at 380 C with 0.2 g catalyst.
a
n-Butane:oxygen:helium:water.
Space-time yield of maleic acid (mmol min (g of catalyst) ).
Physical mixture of MoO , V O , Nb O , and pyridine with the same stoichiometric ratios as NbPMo Vpyr.
b
c
−1
−1
3
2
5
2
5
11

48
J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
Table 6
Reactivity of MPMo pyr with different exchange cations
12
b
M
Flow
Conversion (%)
Selectivity (%)
STY
−1 a
(mL min
)
C
4
O
CO
MA
x
2
Nb
Ti
Mo
Cr
Zr
V
32:16:32:40
32:16:32:40
32:16:32:40
27:16:32:40
16:8:16:20
8:4:8:10
15
8
7
6
7
100
60
51
43
41
61
22
24
11
15
9
71
69
79
81
83
54
0.67
0.39
0.38
0.23
0.17
0.08
7
22
◦
All at 380 C with 0.2 g catalyst.
a
n-Butane:oxygen:helium:water.
Space-time yield of maleic acid (mmol min (g of catalyst) ).
b
−1
−1
4. Discussion
Fig. 2. Selectivity and space-time yield of NbPMo Vpyr catalysts to
11
maleic acid as a function of niobium loading.
The addition of niobium and pyridine to the phospho-
molybdate Keggin ion leads to a catalyst achieving greatly
enhanced productivity of maleic acid from n-butane. While
PMo11V demonstrates some activity for maleic acid, PMo12
produces no maleic acid. Niobium and pyridine individu-
ally promote the reaction (the individual effect of pyridine
is greater than that of niobium), and the addition of both
niobium and pyridine increases the activity of PMo11V by
almost three orders of magnitude. In addition to increasing
productivity, pyridine and niobium do not appear to have
an adverse effect on selectivity. In contrast, the addition of
vanadium appears to have only a minimal effect on produc-
tivity for the most active catalyst, but does affect selectivity:
NbPMo11Vpyr produces more COx while NbPMo12pyr pro-
duces more acetic and acrylic acid under high flow rate con-
ditions. It is important to note that this dramatic increase in
activity is not simply the result of increased surface area. Al-
though the surface area does increase with the incorporation
of pyridine, it is only by about a factor of 2. This suggests
that the addition of niobium and pyridine to the heteropoly-
acid has a chemical effect on the active state.
most active catalyst. In contrast to the butane results, vana-
dium has a significant effect on the reactivity of propane.
Without vanadium, catalysts containing niobium and pyri-
dine exhibit a high selectivity to complete oxidation prod-
ucts, regardless of whether niobium is synthesized in the
Keggin unit or ion-exchanged. In addition, NbPMo11Vpyr
and (VO)PMo11Nbpyr differ only by about a factor of 2
in acrylic acid productivities, suggesting that the position
where each component is initially located in the precursor
material has some effect (but not a critical one) and that
each species is able to move to its final position during pre-
treatment. X-ray absorption spectroscopy evidence support-
ing this conclusion is provided in the following paper [23].
A similar result has been reported for a n-butane catalyst
system, where exchanged and Keggin substituted vanadium
samples showed very similar activities [11].
Perhaps most surprisingly, substantial oxidation of propa-
ne (a three-carbon reactant) to maleic acid (a four-carbon
product) is observed; maleic acid is the major product under
some reaction conditions. The amount of maleic acid pro-
duced is strongly affected by both the exact catalyst com-
position and the reaction conditions. The presence of vana-
dium shifts the selectivity away from COx to maleic acid,
and increasing the flow rate moves the selectivity toward the
desired acrylic acid product. Formation of maleic acid from
propane, although not previously reported in the literature, is
The multisubstrate usefulness of this catalyst is clearly
demonstrated by the results for the selective oxidation of
propane to acrylic acid. While PMo11V shows no activity,
both NbPMo11V and PMo11Vpyr reveal some activity for
the production of acrylic acid. Again, the individual effect of
pyridine is larger than that of niobium, and as with butane,
the combination of both niobium and pyridine produces the
Table 7
Effects of different pretreatment conditions on reactivity
b
Catalyst
Flow
T
Conversion (%)
Selectivity (%)
STY
−1 a
◦
(pretreatment)
(mL min
)
( C)
C
4
O
CO
Ac
AA
MA
x
2
−4
−3
NbPMo V (He)
11
4:2:4:5
4:2:4:5
4:2:4:5
4:2:4:5
32:16:32:40
27:16:32:40
380
380
380
380
380
380
0.4
11.3
1.4
3.4
14
8
42
14
86
70
9
19
12
2
9
5
4
7
0
3
3
3
25
66
0
15
80
40
6.3 × 10
0.059
0
NbPMo V (CO)
75
36
44
95
84
11
NbPMo V (NH )
11
3
NbPMo V (H )
7.8 × 10
11
2
NbPMo Vpyr
0.76
11
NbPMo Vpyr (CO)
11
12
51
5
0.27
a
n-Butane:oxygen:helium:water.
Space-time yield of maleic acid (mmol min (g of catalyst) ).
b
−1
−1

J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
49
Table 8
Catalyst reactivity using propane as a substrate
a
b
Catalyst
Conversion (%)
Selectivity (%)
STY
=
C
3
O
CO
C
3
Ac
AA
MA
x
2
c
PMo
NbPMo
V
0.4
1.4
3.4
3.4
5.7
10.1
100
95
76
100
55
90
30
22
10
79
65
11
65
45
25
54
44
37
1
1
0
0
0
0
4
16
30
2
3
23
3
0
10
17
2
6
49
3
0
0
0
14
21
15
29
23
22
0
11
c
c
−3
−3
V
1.39 × 10
5.55 × 10
0.008
0.067
0.623
0.012
0.16
0.31
11
PMo Vpyr
11
c
NbPMo pyr
25
12
d
NbPMo pyr
18
21
25
8
12
d
NbPMo Vpyr
11
c
PMo Nbpyr
PMo Nbpyr
(VO)PMo Nbpyr
11
d
8
10
17
28
11
d
18
11
Some of the data in this table have been previously reported [15].
a
b
c
d
=
C , propane; CO , carbon oxides; C , propene; Ac, acetic acid; AA, acrylic acid; MA, maleic acid.
x
3
3
−1
−1
Space-time yield of acrylic acid (mmol min (g of catalyst) ).
Flow rates: 8:4:8:10 mL min of propane:oxygen:helium:water at 380 C.
Flow rates: 32:16:32:40 mL min of propane:oxygen:helium:water at 380 C.
−1
◦
−1
◦
Table 9
Reactivity of NbPMo Vpyr under different conditions for propane
11
b
Flow
T
Conversion (%)
Selectivity (%)
Space-time yield
−1 a
◦
(mL min
)
( C)
C
3
O
CO
Ac
AA
MA
AA
MA
x
2
8:4:8:10
380
380
340
340
25
21
11
15
99
76
63
83
15
11
24
18
19
23
28
32
21
49
17
16
43
15
28
22
0.18
0.62
0.09
0.07
0.27
0.16
0.11
0.07
32:16:32:40
32:16:32:40
16:8:16:20
a
Propane:oxygen:helium:water.
Space-time yield of acrylic and maleic acid (mmol min (g of catalyst) ).
b
−1
−1
Table 10
Comparison of propane reactivity to literature data
b
Catalyst
Flow
(mL min
T
Residence
time (s)
Conv. C
(%)
Sel. AA
(%)
Yield
AA
STY
AA
Ref.
3
−1 a
◦
)
( C)
PMo pyr
12
MoSbVNbO
MoVNbTeO
10:5:25:10
340
400
380
380
380
380
3.6
2.0
1.9
4.0
0.24
0.1
8
31
80
47
10
21
29
30
61
28
29
50
2
9
49
13
3
2.7e−3
4.9e−3
0.018
0.024
0.009
[13]
[9]
[8]
[34]
This work
This work
2:3.2:11.8:28
0.6:2:7.4:8.8
4.5:7.5:3:0
3:8.6:18:21
32:16:32:40
x
x
CsFe(PMo VO
)
40
11
NbPMo Vpyr
11
NbPMo Vpyr
11
0.62
11
Some of the data in this table have been previously reported [15].
a
Propane:oxygen:helium:water.
Space-time yield of acrylic acid (mmol min (g of catalyst) ).
b
−1
−1
not unique to these catalysts. Reacting a similar propane-rich
feed over a VPO catalyst also produces substantial amounts
of maleic acid (data not shown). Additionally, oxidation of
C5 and higher hydrocarbons over VPO catalysts are known
to produce maleic acid as a major product [24,25]. Plausi-
ble reaction mechanisms for the formation of maleic acid
from propane have been previously presented [26]. In that
study, experiments using 13C-labeled propane implicate a
pathway of oxidative dehydrogenationof propane to propene
followed by mostly acid-catalyzed dimerization to C6 hydro-
carbons. The C6 hydrocarbons are then selectively oxidized
to maleic acid.
with previously reported literature results. For butane, com-
parisons are included with VPO and other heteropolyacid-
based-catalysts. NbPMo12pyr has a productivity one order
of magnitude greater than VPO as reported by Ledoux et
al. [27]. Under oxygen-rich conditions the productivity of
NbPMo12pyr is within a factor of 4 of VPO, even though
no attempt has been made to optimize the catalyst un-
der these conditions. Most work on selective oxidation has
been under oxygen-rich conditions; however, investigation
of hydrocarbon-rich conditions has also been reported (see
Hutchings [28] and references therein). In contrast to our
catalyst system that is active under both hydrocarbon-rich
and oxygen-rich conditions, Mota et al. have reported that
VPO deactivates under hydrocarbon-richconditions [29], al-
The space-time yields for the NbPMo12pyr and NbPMo11
Vpyr catalysts for maleic acid and acrylic acid compare well

50
J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
Table 11
Reactivity of other substrates
Ethane
Conversion
b
c
Catalyst
Flow
T
Selectivity
Space-time yield
Ref.
−1 a
◦
=
=
(mL min
)
( C)
C
2
O
C
2
Ac
C
2
Ac
2
−2
−3
−3
−1
−3
−3
NbPMo Vpyr
11
4:2:4:5
380
350
250
350
246
5.7
58
1
32
9.6
–
–
–
–
59
65
18
44
1
13
–
36
2.5
59
3.0 × 10
7.1 × 10
3.5 × 10
3.9 × 10
4.3 × 10
6.5 × 10
This work
[5]
[6]
[37]
Mo
V
Nb
13.3:8.8:125.2:0
141.7:8.3:16.7:0
6:3.6:14.4:0
–
0.73 0.18
0.09
0.09
−3
−2
−2
VO /TiO
6.9 × 10
2.2 × 10
1.2 × 10
x
2
d
Mo
V
Nb
0.73 0.18
Mo V
Nb
7.3:1.5:9.5:0.1
4.7
[38]
1
0.25
O
0.12
e
Pd
0.0005
x
Isobutane
g
c
Catalyst
Flow
T
Conversion
Iso-C
Selectivity
Space-time yield
MAA
Ref.
−1 f
◦
(mL min
)
( C)
O
MAA
MA
MA
4
2
−2
−3
−2
NbPMo Vpyr
28:14:28:35
1:4:15:10
380
300
13
22
98
–
1.7
52
3.9
–
1.3 × 10
1.6 × 10
2.9 × 10
This work
[14]
11
H PMo O
3 12 40
pyr
–
Toluene
i
c
Catalyst
Flow
T
Conversion
Selectivity
Space-time yield
Ref.
−1 h
◦
(mL min
)
( C)
C
O
BzA
BA
BzA
BA
2.2 × 10
7
2
−4
−3
−3
NbPMo pyr
0.16:2:4:5
3:45.7:179.7:0
380
370
26
5
55
–
8
30
29
40
5.6 × 10
4.5 × 10
This work
[39]
12
−3
Ag V Ce
O
6 × 10
1.2
3
0.15 8 + x
a
b
c
d
e
f
Ethane:oxygen:inert:water.
=
C , ethane; C , ethene; Ac, acetic acid.
2
2
−1
−1
Space-time yield (mmol min (g of catalyst) ).
P
= 2.0 MPa.
tot
P
= 1.5 MPa.
tot
Isobutane:oxygen:helium:water.
g
h
i
Iso-C , isobutane; MAA, methacrylic acid; MA, maleic acid.
4
Toluene:oxygen:helium:water.
C , toluene; BzA, benzoic acid; BA, benzaldehyde.
7
though by doping the VPO catalyst with cobalt some activity
could be restored [30]. Addition of niobium to VPO under
conventional oxygen-rich conditions has also been investi-
gated. The niobium allows more rapid activation of the cata-
lysts as well as enhancing n-butane conversion while not af-
fecting the selectivity to maleic anhydride [31,32]. It is also
interesting to note that the NbPMo12pyr catalyst, containing
no vanadium, is extremely active for butane to maleic acid;
previous effective catalysts for this reaction have almost ex-
clusively contained vanadium [1].
Heteropolyacid catalysts have also been examined for the
production of maleic acid from butane. To induce activity
for the reaction, two approaches have been used: (1) incor-
poration of vanadium into the structure [10], and (2) prepa-
ration of vanadyl-exchanged species in combination with
other cations [4,11,33]. In both cases, the presence of vana-
dium has been crucial to obtaining useful activity. The nio-
bium/pyridine polyoxometalate catalysts have activity two
orders of magnitude better than either the structural or ex-
changed vanadium containing heteropolyacid-based cata-
lysts reported in the literature. In addition, vanadium is not
required for this high activity, and when present does not
cause a dramatic difference in the activity.
Similar literature comparisons for propane reactivity over
both heteropolyacid-based catalysts and MoVNb mixed-me-
tal oxides are given in Table 10. The MoVNb mixed-metal
oxide systems of Ushikubo et al. and Takahashi et al. have
received significant attention of late as a large step forward
in the production of acrylic acid from propane [8,9]. The lit-
erature data reported in Table 10 are all under different con-
ditions, and it is therefore difficult to make precise compar-
isons. However, it is clear that the NbPMo11Vpyr catalysts
reported here have exceptional productivity for the conver-
sion of propane to acrylic acid. The selectivity to acrylic
acid for NbPMo11Vpyr is less than MoVNbTeOx, but this
is partly due to the coproduction of maleic acid. Similar to
the butane results, these catalysts are not as effective un-
der oxygen-rich conditions, although the productivity is still
within a factor of 2 of the Mitsubishi MoVNbTeOx catalyst.
When compared to the heteropolyacid-based systems in the
literature [13,34], the NbPMo11Vpyr system is much more
productive for acrylic acid formation.
In contrast to results for butane to maleic acid reported
here, effective formation of acrylic acid from propane
appears to require the presence of vanadium in the catalyst.
The vanadium-free catalyst, while also extremely active,
has selectivity to COx of almost 80%. Similar results were

J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
51
observed for the samples with niobium synthesized in the
Keggin unit and no ion-exchanged vanadium. These trends
are not unprecedented, as a synergistic effect of niobium and
vanadium for the formation of acrylic acid over MoVNbTe
mixed-metal oxides has been previously reported [35,36].
These authors conclude that higher activity can be correlated
with catalysts containing Nb5+ ions, but only if vanadium
is also present in the sample. With mixed-metal oxides,
it was suggested that the role of the niobium ions could
be related to the formation of an active phase with low
crystallinity and higher surface area. For the samples used
in the present study, addition of niobium to a PMo11Vpyr
material actually reduced the surface area of the catalyst.
From the activity results, it is observed that NbPMo12pyr
has good activity, but very low selectivity to the acids.
By comparison, NbPMo11Vpyr has similar activity but
greatly increased selectivity. Also, PMo11Nbpyr exhibits
good activity and poor selectivity while (VO)PMo11Nbpyr
has both good activity and selectivity. Since the PMo11Vpyr
sample does not have nearly the activity of NbPMo11Vpyr,
the presence of niobium is responsible for increased propane
activation. However, only when vanadium is incorporated
into the sample does the acid selectivity improve. Therefore,
both components are critical to a highly active and selective
catalyst. It is unclear whether the niobium and vanadium act
directly as part of the atomic arrangement of the active site or
if they act indirectly by stabilizing the formation of an active
phase. Ai has observed similar effects for n-butane oxidation
of heteropolyacid-based catalysts where multiple species
were required to obtain active and selective catalysts [11].
In addition to butane and propane, the NbPMo11V ma-
terials are capable of catalyzing the selective oxidation of
other substrates. Examples include ethane, isobutane, and
toluene. For ethane, the main products are ethylene and
acetic acid. The observed space-time yield of ethylene for
NbPMo11Vpyr is greater by four times that reported by
Thorsteinson et al. for a MoVNb mixed-metal oxide [5].
Formation of acetic acid is comparable to that of ethylene.
Similar space-time yields of acetic acid at lower tempera-
tures have been reported in the literature, but this may result
from these samples being on high surface area supports [6].
Other data for the MoVNb-based systems at elevated pres-
sures and specifically designed to produce either ethylene
or acetic acid are also shown [37,38]. Comparisons to these
results are even more difficult due to the higher pressures
involved (2.0 and 1.5 MPa, respectively). In contrast to the
propane reaction, no evidence for carbon–carbon coupling
to C3 or higher products is observed.
uct for our system. Toluene was also investigated to exam-
ine selective side-chain oxidation of an aromatic species.
The major product is COx with selectivity to benzoic acid
and benzaldehydebeing 8 and 29%, respectively. Space-time
yields are lower than those reported by Yan and Anders-
son for AgVCe mixed-metal oxides [39]. Attempts to ac-
◦
tivate methane (380 ◦C ꢀ T ꢀ 460 C) produce only small
amounts of complete oxidation products.
The most active catalysts, NbPMo12pyr and NbPMo11
Vpyr, involve three main components: (1) a central oxo-
molybdate cluster, (2) an exchange metal cation, and (3)
an organic base. The role of each in gaining reactivity can
be separately investigated. Catalysts based on metal oxides
of several different types were studied: first, the bulk metal
oxide and a physical mixture of MoO3, V2O5, Nb2O5, and
pyridine; second, the discrete metal oxide cluster octamolyb-
date; third, several different polyoxometalate structures, in-
cluding the Keggin, Wells–Dawson, and Strandberg units
that all contain the MoOx clusters and a heteroatom in differ-
ent geometries. The effect of changing the heteroatom while
leaving the structure the same was also examined by compar-
ing a silicomolybdate to the corresponding phosphomolyb-
date.
Molybdenum trioxide and the physical mixture of metal
oxides both exhibit poor activity for butane oxidation in
comparison to NbPMo12pyr; this strongly suggests the ac-
tive phase is more complex than the combination of bulk
4−
oxide structures. The structures of MoO3 and Mo8O26
(octamolybdate) differ in their arrangement of molybdenum
oxide octahedra; MoO6 are corner-bound through an infinite
4−
network in MoO3 and exist in discrete clusters in Mo8O26
.
In the presence of niobium and pyridine the octamolybdate
exhibits good selectivity; however, the presence of a central
phosphorus heteroatom in the Keggin unit of NbPMo12pyr
enhances catalyst performance substantially. The enhance-
ment in reactivity may be due to the ability of the phospho-
rus atom to stabilize the structure by inhibiting the phase
transition of the material into catalytically inert MoO3. By
TGA-DSC analysis (not shown), NbMo8pyr decomposes to
MoO3 at 290 ◦C, while PMo12 decomposes to MoO3 at
◦
440 C. Interestingly, NbPMo12pyr does not decompose to
MoO3 at all. Clearly, the presence of the phosphorus het-
eroatom helps stabilization of the active structure in the
range of reaction temperatures. A more detailed discussion
of the decomposition characteristics of the catalysts is pre-
sented in Part II [23].
Comparison of reactivity results for NbPMo12pyr and
PMo11Nbpyr indicates that the location of niobium in as-
synthesized samples, either ion-exchanged or incorporated
in the structure, has no significant effect on productivity.
These data suggest that niobium is mobile during the
pretreatment; EXAFS results supporting this conclusion are
presented in the following paper [23].
For isobutane, the activity is very similar to that of butane.
However, the selectivity to COx is 86% while selectivity to
the desired product of methacrylic acid is only around 2%.
The productivity of methacrylic acid is an order of magni-
tude greater than reported by Li a◦nd Ueda; however, they
operated at a temperature of 300 C [14]. A reduction in
reaction temperature may minimize the side reactions of
isobutane to COx and shift selectivity to the desired prod-
The nature of the heteroatom (phosphorus or silicon) in
the heteropolyanion has a significant effect on catalyst per-
formance. The stabilizing effect of phosphorus on the struc-

52
J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
ture during catalyst pretreatment is responsible in part for
the formation of the active phase. By TGA-DSC analysis
(data not shown), silicomolybdic acid and phosphomolybdic
acid differ in their thermal stabilities. Significant decompo-
sition takes place for the SiMo12-based catalyst at pretreat-
ment temperatures and results in a mixture of metal oxides
(this is not the case for PMo12-based catalysts). The activity
and selectivity exhibited by NbSiMo12pyr and by the physi-
cal mixture of mixed oxides for n-butane oxidation are com-
parable (Table 5), supporting this argument.
even though activity can be improved using a gaseous
reductant, it still cannot completely replicate the role of the
pyridine.
In addition to pyridine, other organic bases have been
investigated. Although other bases are capable of activating
the catalyst, none performed better than pyridine. Attempts
to correlate activity with characteristics of the base such as
molecular size and pKa were unsuccessful.
5. Conclusions
The second component of the active catalyst is the
exchanged-metal cation. Elements near niobium in the
periodic table as well as other elements used in mixed-metal
oxides for propane oxidation such as antimony and bismuth
were investigated. Molybdenum and titanium show activity
of about one-half that of niobium. In contrast, vanadium
is almost an order of magnitude less active. In fact, the
sample with only pyridine and no exchange cation is more
active than the vanadium-exchanged sample. The results
of our vanadium-exchanged catalyst and the PMo11Vpyr
catalyst show similar space-time yields of maleic acid.
This demonstrates that the location of vanadium in the
as-synthesized materials has little effect on reactivity and
suggests that vanadium is mobile during activation and
reaction; indeed, regardless of its initial location, once
activated, the vanadium likely plays the same role in both
catalysts. The initial oxidation state of the exchange cation
does not appear important since the three most active
catalysts have 5+, 4+, and 6+ oxidation states (Nb, Ti, and
Mo, respectively). It is also interesting that zirconium gives
better activity than vanadium since zirconium is considered
much less susceptible to oxidation-reduction cycling than
vanadium [40].
An additional aspect of the exchange-metal cation in-
volves the amount of niobium present. Reactivity results
suggest that a ratio of about 0.30 niobium atoms per Keggin
unit is optimal. Since more pyridine is exchangeable with
catalysts that contain less niobium, and since pyridine also
is critical for enhancing activity, the ratio of 0.30 probably
represents a balance between the two effects.
The third component of the active catalyst is the organic
base. As further discussed in Part II [23], the role of the base
is to act as an in situ reductant to reduce both molybdenum
and niobium. An attempt was made to perform the same
role by using the gas phase reducing agents CO, NH3, and
H2. Although CO did increase the activity of pyridine-free
NbPMo11V, it is still one order of magnitude less active
than the sample synthesized with pyridine. The samples
pretreated in H2 and NH3 are even less active. Additionally,
pretreating a pyridine-containing catalyst in CO did not
further increase the activity. Under CO, TGA-DSC data
indicate that the pyridine-free sample first decomposes
to MoO3 at 440 ◦C (which is above our pretreatment
temperature), and then subsequently undergoes reduction.
In contrast, the pyridine-containing sample reduces at lower
temperatures without first going through MoO3. Therefore,
A substrate versatile catalyst for the selective oxidation
of light alkanes has been developed. The catalysts are nio-
bium and pyridine-exchanged salts of molybdophosphoric
acid and molybdovanadophosphoric acid (NbPMo12pyr and
NbPMo11Vpyr). These catalysts are active and selective
for the selective oxidation of propane and n-butane over a
wide temperature range and under both hydrocarbon-rich
and oxygen-rich reaction conditions. The catalysts that con-
tain both niobium and pyridine are very active and selective.
For the selective oxidation of n-butane to maleic acid, the
presence of vanadium does not appear to significantly influ-
ence the reactivity. In contrast, for the selective oxidation of
propane to acrylic acid, the presence of vanadium signifi-
cantly affects the selectivity to the desired product. Surpris-
ingly, the selective oxidation of propane also leads to the
production of maleic acid as a major product under certain
reaction conditions.
Although metal cations other than niobium show some
activity for selective oxidation, none appear better than nio-
bium. The central phosphorus atom of the polyoxometalate
is important. If the phosphorus is not present or replaced by
another heteroatom, productivities are much lower. For both
propane and n-butane, the synthesized location of the nio-
bium (structural or ion-exchanged) does not have a large ef-
fect on reactivity. Similarly for propane, the synthesized lo-
cation of vanadium has minimal effect on activity.
For the selective oxidation of n-butane, the productivity
of the catalyst reported here is greater than that of VPO.
With propane, the productivity of acrylic acid exceeds that
of MoVNbTeOx. The catalysts are also active for selective
oxidation of other substrates such as ethane, isobutane, and
toluene.
Acknowledgments
This work was funded by BP. We thank Larry Henling
and Michael Day (Beckman Institute, California Institute of
Technology) for single-crystal X-ray diffraction.
References
[1] G. Centi, F. Cavani, F. Trifiro, Selective Oxidation by Heterogeneous
Catalysis, Kluwer Academic/Plenum, New York, 2001.

J.H. Holles et al. / Journal of Catalysis 218 (2003) 42–53
53
[2] F. Cavani, F. Trifiro, Catalysis (1994) 247.
[3] B.K. Hodnett, Heterogeneous Catalytic Oxidation, Wiley, Chichester,
2000.
[4] M. Ai, J. Mol. Catal. A: Chem. 114 (1996) 3.
[5] E.M. Thorsteinson, T.P. Wilson, F.G. Young, P.H. Kasai, J. Catal. 52
(1978) 116.
[21] E.M. McCarron III, J.F. Whitney, D.B. Chase, Inorg. Chem. 23 (1984)
3275.
[22] J.S. Min, M. Misono, A. Taguchi, N. Mizuno, Chem. Lett. (2001) 28.
[23] C.J. Dillon, J.H. Holles, R.J. Davis, J.A. Labinger, M.E. Davis, J. Ca-
tal., in press.
[24] F. Cavani, F. Trifiro, Appl. Catal. A 157 (1997) 195.
[25] F. Cavani, A. Colombo, F. Trifiro, G.J. Hutchings, Catal. Lett. 43
(1997) 241.
[6] L. Tessier, E. Bordes, M. Gubelmann-Bonneau, Catal. Today 24
(1995) 335.
[26] C.J. Dillon, J.H. Holles, M.E. Davis, J.A. Labinger, Catal. Today, in
press.
[27] M.J. Ledoux, C. Crouzet, C. Pham-Huu, V. Turines, K. Kourtakis, P.L.
Mills, J.J. Lerou, J. Catal. 203 (2001) 495.
[28] G.J. Hutchings, Appl. Catal. 72 (1991) 1.
[29] S. Mota, M. Abon, J.C. Volta, J.A. Dalmon, J. Catal. 193 (2000) 308.
[30] S. Mota, J.C. Volta, G. Vorbeck, J.A. Dalmon, J. Catal. 193 (2000)
319.
[31] P.G. Pries de Oliveira, J.G. Eon, M. Chavant, A.S. Riché, V. Metin,
S. Caldarelli, J.C. Volta, Catal. Today 57 (2000) 177.
[32] A.M. Duarte de Farias, W.A. de Gonzalez, P.G. Pries de Oloveira, J.-G.
Eon, J.-M. Herrmann, M. Aouine, S. Loridant, J.C. Volta, J. Catal. 208
(2002) 238.
[33] M. Ai, J. Catal. 85 (1984) 324.
[34] N. Mizuno, D. Suh, W. Han, T. Kudo, J. Mol. Catal. A: Chem. 114
(1996) 309.
[35] P. Botella, B. Solsona, A. Martinez-Arias, J.M. Lopez Nito, Catal.
Lett. 74 (2001) 149.
[36] P. Botella, J.M. Lopez Nito, B. Solsona, Catal. Lett. 78 (2002) 383.
[37] K. Ruth, R. Burch, R. Kieffer, J. Catal. 175 (1998) 27.
[38] D. Linke, D. Wolf, M. Baerns, O. Timpe, R. Schlögl, S. Zeyß, U. Din-
gerdissen, J. Catal. 205 (2002) 16.
[7] M.M. Lin, Appl. Catal. A 207 (2001) 1.
[8] T. Ushikubo, H. Nakamura, Y. Koyasu, S. Wajiki, US Patent 5,380,933
Mitsubishi Kasei, 1995.
[9] M. Takahashi, X. Tu, T. Hirose, M. Ishii, US Patent 6,060,422
Toagosei Co. Ltd., 2000.
[10] G. Centi, V. Lena, F. Trifiro, D. Ghoussoub, C.F. Aissi, M. Guelton,
J.P. Bonnelle, J. Chem. Soc. Faraday Trans. 1990 (1990) 2775.
[11] M. Ai, in: 8th International Congress on Catalysis, Vol. 5, Dechema,
Berlin, 1994, p. 475.
[12] B.B. Bardin, R.J. Davis, Appl. Catal. A 185 (1999) 283.
[13] W. Ueda, Y. Suzuki, Chem. Lett. (1995) 541.
[14] W. Li, W. Ueda, in: R.K. Grasselli, S.T. Oyama, A.M. Gaffney, J.E.
Lyons (Eds.), 3rd World Congress on Oxidation Catalysis, Elsevier,
Amsterdam, 1997, p. 433.
[15] M.E. Davis, C.J. Dillon, J.H. Holles, J. Labinger, Angew. Chem. Int.
Ed. 41 (2002) 858.
[16] G.A. Tsigdinos, C.J. Hallada, Inorg. Chem. 7 (1968) 437.
[17] R. Massart, R. Contant, J.-M. Fruchart, J.-P. Ciabrini, M. Fournier,
Inorg. Chem. 16 (1977) 2916.
[18] R. Strandberg, Acta Chem. Scand. 27 (1973) 1004.
[19] L. Pettersson, I. Andersson, L.-O. Ohman, Acta Chem. Scand. A 39
(1985) 53.
[20] P. Gili, P. Martin-Zarza, G. Martin-Reyes, J.M. Arrieta, G. Madariaga,
Polyhedron 11 (1992) 115.
[39] Z.-G. Yan, S.L.T. Andersson, J. Catal. 131 (1991) 350.
[40] C.S.G. Phillips, R.J.P. Williams, Inorganic Chemistry, Oxford Univ.
Press, Oxford, 1966.