Vanadium phosphorus oxide catalyst modified by niobium doping for mild oxidation of n-butane to maleic anhydride

Article abstract of DOI:10.1006/jcat.2002.3562

The interest of doping (VO)₂P₂O₇ with Nb⁵⁺ ions was studied for n-butane oxidation to maleic anhydride. Niobium was solubilized as niobium (V) ethoxide into isobutanol and used as a reducing reagent to prepare the VOHPO₄·0.5H₂O precursor. Under n-butane fuel-lean oxidation conditions, the VNbPO precursor was activated at 20°C lower than for undoped vanadium phosphorus oxide (VPO) due to a modification of its morphology. The activated VNbPO catalyst was more disorganized than the activated VPO. It comprised more defects and concentrates niobium at the surface. NB acted an n-type dopant for the p-type (VO)₂P₂O₇ semi-conductor. VNbPO catalyst was more oxidized, particularly at the surface. Doping with niobium created defects, responsible for C-H n-butane activation, which were associated with Lewis acid sites of low acidity. This was the reason for the enhancement of the n-butane conversion. Selectivity to maleic anhydride was not modified by Nb-doping while CO₂/CO formation was increased due to the higher surface V⁵⁺/V⁴⁺ ratio.

Full text of DOI:10.1006/jcat.2002.3562

Journal of Catalysis 208, 238–246 (2002)
doi:10.1006/jcat.2002.3562, available online at http://www.idealibrary.com on
Vanadium Phosphorus Oxide Catalyst Modified by Niobium Doping
for Mild Oxidation of n-Butane to Maleic Anhydride
∗,
∗
Andre´a M. Duarte de Farias, † Wilma de A. Gonzalez, Paulo G. Pries de Oliveira,† Jean-Guillaume Eon,‡
,1
Jean-Marie Herrmann,§ Mimoun Aouine,¶ Ste´phane Loridant,¶ and Jean-Claude Volta¶
∗
Instituto Militar de Engenharia, Pc¸a Gen. Tiburcio, 80, Rio de Janeiro, RJ, Brazil; †Instituto Nacional de Tecnologia, Laboratorio de Catalise, Av.
Venezuela, 82, Rio de Janeiro, RJ, Brazil; ‡Universidade Federal do Rio de Janeiro, Inst de Quimica, Bl. A, Cidade Universitaria, Rio de Janeiro,
RJ, Brazil; §Ecole Centrale de Lyon, Laboratoire de Photocatalyse, Catalyse et Environnement, IFOS, CNRS, BP 163, 69131, Ecully Cedex,
France; and ¶Institut de Recherches sur la Catalyse, 2 Avenue Albert Einstein, 69626, Villeurbanne Ce´dex, France
Received December 3, 2001; revised February 14, 2002; accepted February 14, 2002
tivity to mild oxidation (5). The effect of Fe doping has
been compared to that of Co in fuel-lean oxidation of n-
butane (6, 7): both dopants improve the selectivity to MA,
The interest of doping (VO)2P2O7 with Nb5+ ions has been stud-
ied for n-butane oxidation to maleic anhydride. Nb was solubilized
as niobium (V) ethoxide into isobutanol and used as a reducing but Co decreases n-butane conversion whereas Fe increases
reagent to prepare the VOHPO4 · 0.5H2O precursor. Under n-
it. These results were explained by a different dispersion of
butanefuel-leanoxidationconditions, theVNbPOprecursorwasac-
the VOPO4/(VO)2P2O7 phases during the period of acti-
tivated at a temperature 20◦C lower than for undoped VPO because
of a modification of its morphology. Electron microscopy showed
vation of the precursor. By electron microscopy and 31P
NMR by spin-echo mapping, it was observed that Co de-
that the activated VNbPO catalyst is more disorganized than the
lays the transformation of the precursor, increasing the dis-
activated VPO. It contains more defects and concentrates Nb at the
order of the VPO material at lower temperature by stabi-
surface. Nb acts as an n-type dopant for the p-type (VO)2P2O7 semi-
lizing an amorphous VPO phase for which Co was acting
as the promoter (8). Co was also observed to be an efficient
dopant for the VPO catalyst in fuel-rich n-butane oxidation
conditions, making possible the mild oxidation of n-butane
conductor, as observed in electrical conductivity measurements. 31
P
NMR by spin-echo mapping and XPS spectroscopy provide evi-
dence that the VNbPO catalyst is more oxidized, particularly at the
surface. Doping with Nb creates defects, responsible for C–H n-
butane activation, which have been observed to be associated with to MA; this was observed to be impossible without this
Lewis acid sites of low acidity. This is the reason for the enhance-
ment of the n-butane conversion. Selectivity to maleic anhydride is
not modified by Nb-doping while CO2/CO formation is increased
dopant (9).
Many other dopants of VPO have been considered to
improve the catalytic performance in fuel-lean oxidation
conditions (4). There is not a general explanation concern-
ing their specific role and the examination of the published
results shows that they should affect differently the nature
and the distribution of the nanocrystalline oxidized VOPO4
phases which are generated during the activation period of
the catalyst and affect the properties of the equilibrated
catalyst (10, 11). Among these dopants, Nb has not yet
been seriously considered, although it is mentioned in a
patent assigned to ICI for mild oxidation of n-butane to
MA (12). Apart from this, a promotional catalytic effect
of niobium phosphate for vanadyl pyrophosphate was ob-
servedformildoxidationofn-butane:theauthorsexplained
this result by considering the development of strong Lewis
acid sites induced by Nb and associated to the formation
of a (V1−x Nbx O)2(P2O7)1−x (PO4)2x solid solution (13). We
showed that this interaction in close proximity to V should
occur for the VOHPO4 · 0.5H2O precursor, which was mod-
ified by amorphous niobium phosphate: a new signal was
detected by 31P NMR (MAS and spin-echo mapping) on
the corresponding VPO precursor (14).
due to the higher surface V5+/V4+ ratio. ꢀ 2002 Elsevier Science (USA)
c
Key Words: n-butane mild oxidation; maleic anhydride; vana-
dium phosphorus oxide; niobium doping.
INTRODUCTION
Vanadium phosphorus oxide catalysts (VPO) are known
to catalyze the mild oxidation of n-butane to maleic an-
hydride (MA) (1–3). Catalytic performance may be im-
proved by adding specific doping agents to the VPO com-
position, but their effect on conversion and selectivity may
be different. The nature, the location, and the role of these
dopants have been previously reviewed in the open liter-
ature (4). For example, it was observed that the presence
of Co changes the surface properties by modifying the re-
dox properties of the VPO catalyst, which results in an
easier desorption of maleic anhydride and a higher selec-
1 To whom correspondence should be addressed.
238
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

SELECTIVE OXIDATION OF n-BUTANE ON MODIFIED (VO)2P2O7
239
In the present publication, the transformations induced by the laser beam. XPS examinations of the VPO cata-
by Nb doping of the VPO precursor have been investigated, lysts were performed with an ESCALAB 200R spectrom-
as well as the effect on the catalytic properties and on the eter using Al Kα radiation. The binding energy of ad-
physicochemical characteristics of the corresponding cata- ventitious carbon was taken as a reference at 284.5 eV.
lyst for n-butane oxidation to maleic anhydride. This work Examinations were focused on the binding energies of
is in the continuation study demonstrating the interest of P 2p, C 1s, V 2p3/2 (V5+ and V4+), Nb 3d, and O 1s. Electri-
associating Nb with V in mixed oxides catalysts for mild cal conductivity measurements were performed in a pre-
oxidation of light hydrocarbons (15, 16).
viously decribed cell (18), especially designed to study
electronic interactions between powdered samples under
various atmospheres. VPO catalysts were heated in air up
to 400◦C at a heating rate of 5◦C/min. Conductivity mea-
surements were then followed at 400◦C as a function of
time. Acidity measurements on VPO catalysts were per-
formed by thermodesorption of pyridine, used as a basic
probe. Catalysts were compressed to obtain disks of 25 mg.
After outgassing at 350◦C, pyridine was adsorbed at room
temperature. The catalysts were further outgassed at 25,
100, 150, 200, 250, 300, and 350◦C. The acid properties were
then compared by measurement of the intensity of the char-
acteristic bands at 1540 cm−1 (Brønsted acid site) and at
1450 cm−1 (Lewis acid site) as a function of the tempera-
ture of outgassing.
Catalytic tests were conducted on 1 ml of precursor at
400◦C and atmospheric pressure in a differential flow Pyrex
microreactor. The VPO and VNbPO precursors were acti-
vated on line for 80 h under fuel-lean oxidation conditions
(n-C4H10/O2/He = 1.6/18.0/80.4) at 2000 h−1 GHSV from
25◦C up to 400◦C at a heating rate of 0.5◦C/min. Catalytic
performances (n-butane conversion and MA, CO, CO2 se-
lectivities) were compared at 400◦C as a function of time
of testing of the VPO and VNbPO catalysts. Reactants and
reaction products were analyzed online by gas chromato-
graphy. O2, CO, CO2, H2O, and He were detected by
thermal catharometry after separation on a 4 m × 1/8
in. 80- to 100-mesh Carbosieve B column. Hydrocarbons
were detected by flame ionization after separation on a
4 m × 1/8 in. 80- to 100-mesh Durapak column. Maleic an-
hydride, aldehydes, and acids were detected by flame ion-
ization after separation on a 2 m × 1/8 in. 80- to 100-mesh,
10% AT 1200-imp. 1% H3PO4/Chromosorb WAW column.
EXPERIMENTAL
A reference VOHPO4 · 0.5H2O precursor (designated
VPO) was prepared according to the organic route by
refluxing a suspension of V2O5 (11.5 g) into isobutanol
(250 ml) in the presence of 85% H3PO4 with a P/V ratio =
1 : 1 for 16 h (17). A VPO precursor modified by niobium
(designated VNbPO) was prepared according to the same
procedure by solubilizing niobium(V) ethoxide into isobu-
tanol (Nb/V = 0.01). All the operations of refluxing were
similar to those used for reference VOHPO4 · 0.5H2O pre-
cursor. The two resulting blue solids were recovered by
filtration, washed with fresh isobutanol and ethanol, and
finally dried in air for 16 h at 110◦C. Chemical analysis was
performed by atomic emission spectroscopy with an ICP
spectro-flame(inductingcouplingplasma)fromSPECTRO
after dissolution of the solids in hydrochloric acid.
The BET specific surface areas of the solids were con-
ventionally measured by N2 adsorption at 77.4 K. The de-
composition of the two precursors under air was followed
in a SETARAM TGA–DTA 92 thermobalance from room
temperature up to 700◦C at a heating rate of 5◦C/min.
Activation under reaction conditions over 80 h was con-
ducted in order to obtain activated solids which were
then studied by XRD, SEM, TEM, NMR, Raman spec-
troscopy, electrical conductivity, and acidity measurements.
X-ray diffraction (XRD) patterns were collected with a
SIEMENS D500 diffractometer using Cu Kα radiation.
For transmission electron microscopy (TEM), the catalysts
were sonically dispersed in ethanol and deposited on a cop-
per grid coated with holey carbon before examination in a
JEOL 2010 microscope with a high-resolution pole piece.
Microanalyses were performed with a link Isis system. A
HITACHIS800microscopewasusedforSEMexamination
of the VPO precursors. Samples for SEM examination were
RESULTS AND DISCUSSION
The BET areas of the VNbPO and VPO precursors were
deposited on a metallic grid and then sputtered with gold. both observed to be 7 m2/g. Activated catalysts showed
The 31P NMR spectra were recorded with a BRUCKER surface areas of 17 and 12 m2/g for VNbPO and VPO, re-
DSX400 spectrometer, at 161.9 MHz, equipped with a stan- spectively. Figure 1 shows the XRD patterns of the VPO
dard 4-mm probe head. The 31P spin-echo mapping (SEM) and VNbPO precursors. Both are characteristic of the
spectra were obtained with a sweep width of 2 mHz, t = VOHPO4 · 0.5H2O precursor phase only (17). The mod-
20 ms, and 90◦ pulse length of 1.5 ms. Raman spectra were ification of the relative intensity of the (001) and (220) lines
recorded with a DILOR XY spectrometer. The excitation for the VNbPO precursor as compared to the VPO pre-
line at 514.53 nm of an argon–krypton laser was used with cursor is indicative of a difference of morphology of the
a power of 0.5–1 mW on the sample. It has been checked VOHPO4 ·0.5H2O platelets induced by the presence of Nb,
that such a power is low enough to avoid heating effects which was confirmed by SEM examination (Fig. 2). In the

240
DUARTE DE FARIAS ET AL.
Figure 4 shows the variation of the n-butane conversion
of the two catalysts with time when activated under the n-
C4/O2/He flow. While conversion is similar in the first 20 h,
the VNbPO catalyst shows significantly higher conversion
in the next 60 h as compared to the VPO catalyst. After
80 h of activation, the n-C4 conversion is 75% for VNbPO
as compared to only 58% for VPO. Maleic anhydride was
observed to be the only mild oxidation product for both
VNbPO and VPO. Almost no acetic nor acrylic acids (less
than 1%) were observed in both cases. Only CO and CO2
were detected as total oxidation products. Figure 5 shows
that almost no modification of the MA selectivity is ob-
served between the two catalysts. However, Table 1 shows
that the ratio CO2/CO is higher for the VNbPO catalyst as
compared to the VPO one.
FIG. 1. XRD patterns of the VNbPO and VPO precursors.
Figure 6 shows the XRD pattern of the two catalysts af-
ter being activated. Both are very similar and are charac-
teristic of a disorganized (VO)2P2O7 phase with the (200),
(024), and (032) lines at 23.00, 28.45, and 29.95◦, respec-
tively. In the case of VNbPO, the (200) reflection is broad-
ened as compared to the (200) line for VPO. This result
may be interpreted as due to a decrease in the size of the
(VO)2P2O7 crystallites in the [100] direction, in accordance
with the modification of the (VO)2P2O7 crystal morphol-
ogy observed by SEM. Another possibility is peak overlap
due to the presence of some residual VOPO4 phases: on
reference phases (20), the intense (111) and (012) lines of
δ-VOPO4 are observed at 22.08 and 24.17◦, whereas the in-
tense(004), (221), and(040)linesof γ -VOPO4 areobserved
at 21.37, 22.66, and 23.16◦, respectively (20). Moreover, the
background in the XRD pattern of VNbPO appears to be
slightly higher than in the case of VPO.
Figure 7 shows the Raman spectra of the two catalysts.
For both samples, the background is probably due to elec-
tronic defects. Its curved aspect may correspond to the pres-
ence of an amorphous phase. Bands observed at 924, 1125,
and 1181 cm−1 are typical of the (VO)2P2O7 phase (20).
The lower relative intensity of the 924-cm−1 band from
VNbPO is linked to a lower crystallinity (21). In the case
of VPO, the higher crystallinity may be due to a polariza-
tion effect appearing with the particular growth of crystal-
lites. The spectrum of VNbPO does not show evidence for
case of VNbPO, the platelets of the precursor are smaller
and do not agglomerate into sand roses, as seen for VPO.
ChemicalanalysisconfirmedthepresenceofNbwiththeex-
pected composition (Nb/V = 0.0102). Nb is well dispersed
throughout the bulk of VOHPO4 · 0.5H2O, as observed by
EDX in a TEM.
It is known (19) that the decomposition under air of
VOHPO4 · 0.5H2O to (VO)2P2O7 occurs in two steps.
The vanadium hemihydrate retains its water of hydration
up to 360◦C. The departure of this water molecule re-
sults in vacancy formation at the apex of the vanadium–
oxygen octahedra. At this stage, an electronic rearrange-
ment proceeds, which corresponds to the formation of
VOHPO4 layers (Eq. [1]). These layers subsequently join,
expelling the constitutional water molecules to form one
P2O7 group from two HPO4 groups (Eq. [2]). (VO)2P2O7
is then further oxidized to VOPO4 (step [3]). All these
transformation steps were identified by XRD examina-
tion of the samples (19). The process of dehydration–
oxidation under air may be summarized by the correspond-
ing scheme:
VOHPO4 · 0.5H2O → VOHPO4 + 0.5H2O,
2VOHPO4 → (VO)2P2O7 + H2O,
(VO)2P2O7 + 1/2O2 → 2VOPO4.
[1]
[2]
[3]
The decomposition under air of the VNbPO precursor is bands due to NbOPO4 although the Raman response of
compared to that of VPO in Fig. 3. While the departure this phase is much higher than (VO)2P2O7. Indeed, we de-
of water of hydration (endothermic process) occurs at the tected the presence of niobium phosphate by micro-Raman
same temperature (360◦C), the second departure of consti-
tutional water molecules from the VNbPO precursor un-
TABLE 1
der air is lower by 21◦C (414◦C for VNbPO as compared
to 435◦C for VPO). Differences are also observed for the
Catalytic Performance of VNbPO and VPO Catalysts Measured
at 400◦C after 80 h Activation under n-C4/O2/He = 1.6/18.0/80.4 at
exothermic oxidation stage to VOPO4(510◦C for VNbPO
as compared to 555◦C for VPO). This is another indica-
tion of the differences between the VNbPO and VPO mor-
phologies. Note that in the case of Co doping of VPO, a
contrary result was observed with the dehydration step de-
layed instead of accelerated (19).
GHSV 2000 h−1
Catalyst
Conv C4 (%)
SMA (%)
SCO (%)
SCO (%)
2
VNbPO
VPO
75.0
58.0
70.1
70.0
18.9
20.6
11.0
9.4

SELECTIVE OXIDATION OF n-BUTANE ON MODIFIED (VO)2P2O7
241
FIG. 2. Scanning electron micrographs of VNbPO and VPO precursors.
spectroscopy on samples synthesized by a different route, P atoms in (VO)2P2O7 (22). Indeed V4+ in the (VO)2P2O7
even at levels as low as 0.25% Nb/V. Finally, bands observed structure gives rise to a large signal around 2400–2700 ppm,
at about 1040 and 1075 cm−1 may be indicative of the pres- depending on temperature of calcination. When cooling
ence of some VOPO4 phases (20).
(VO)2P2O7 at −123◦C, this large signal has been observed
31P NMR by spin-echo mapping gives information about to be split into four distinct signals, at 3200, 3650, 4200,
the different oxidation states of vanadium surrounding the and 4850 ppm. These signals have been attributed to the

242
DUARTE DE FARIAS ET AL.
FIG. 3. DTA profiles of the decomposition of the VNbPO and VPO
precursors under air.
FIG. 6. XRD patterns of the activated VNbPO and VPO catalysts.
existence in V atoms in the unit (VO)2P2O7 cell of four
distinct oxidation states, at 4.44+, 4.19+, 3.79+, and 3.69+,
respectively (22), in accordance with previous calculations
of the individual bond strengths (23). We used these re-
sults to compare the oxidation states of V in the unit cell
of (VO)2P2O7 for the VNbPO sample with the undoped
VPO catalyst from the spectra recorded at room temper-
ature (Fig. 8). The two spectra are typical of a disorga-
nized (VO)2P2O7 catalyst with a complex signal around
2600 ppm. This complex signal, which is associated with
the unit cell of (VO)2P2O7, shows, for VNbPO, a global
displacement to the right (contributions at 2740, 2176, and
1916 ppm), indicative of a higher oxidation state of V as
compared to VPO. Signals at about 10 and 1140 ppm show
FIG. 4. Conversion of n-butane measured at 400◦C under n-
C4/O2/He = 1.6/18.0/80.4 at GHSV 2000 h−1 for the VNbPO and VPO
catalysts as a function of time.
FIG. 5. Selectivity to maleic anhydride measured at 400◦C under n-
C4/O2/He = 1.6/18.0/80.4 at GHSV 2000 h−1 for the VNbPO and VPO
catalysts as a function of time.
FIG. 7. Raman spectra of the activated VNbPO and VPO catalysts.

SELECTIVE OXIDATION OF n-BUTANE ON MODIFIED (VO)2P2O7
243
catalysts. The relative distribution of surface O, V, C, and
P atoms is similar for both catalysts, except for O, which
is higher for VNbPO (66.9% as compared to 64.3% for
VPO). The C concentration is lower for VNbPO (4.1%) as
compared to VPO (7.3%), which may be indicative of less
carbonaceous residues for the more oxidizing VNbPO cata-
lyst. These observations favor a higher average oxidation
state of the VNbPO catalyst as compared to the VPO one,
as is experimentally observed by the higher percentage of
surface V5+(49%) in VNbPO, as compared to surface V5+
(29%)inVPO. ThedifferenceobservedforthesuperficialC
deposit may also be explained by the difference of morphol-
ogy of the two catalysts. The P/V ratio is also slightly lower
for VNbPO. It is important to notice a higher Nb/V surface
ratio (Nb/V = 0.032) for the VNbPO catalyst as compared
to the bulk composition (Nb/V = 0.010), indicating a Nb
surface enrichment for this catalyst.
TEM examination showed that both catalysts have
platelet morphologies and exhibit the characteristic
surface-normal electron diffraction pattern of the [100] di-
rection of (VO)2P2O7 (Fig. 9). However the diffraction
spots were less intense for VNbPO, which may be indicative
of a lower crystallinity. The analysis of the images revealed
a strong difference: while the periphery of the VPO crys-
tals appeared quite regular in profile when viewed along
the [100] direction, the periphery of the VNbPO crystals
showed the existence of many kinks and steps in the same
[100] direction. This was indicative of the existence of a
high density of defects. EDX analysis of the platelet identi-
fied the presence of Nb in the whole crystal of the VNbPO
catalyst.
FIG. 8. 31P NMR spectra by spin-echo mapping of the VNbPO and
VPO activated catalysts.
the presence of V5+ entities attributed to either isolated
V5+ species or V5+ microdomains and V4+–V5+ dimers,
respectively. The spectrum of VNbPO shows a higher con-
tribution of signals associated with V5+ sites as compared
to the spectrum of VPO. Finally, the comparison of the two
31P NMR spectra by spin-echo mapping shows that VNbPO
is globaly more oxidized than VPO.
XPS experiments have been performed on the activated
catalysts to obtain information on the surface distribution
of atoms. For the VNbPO catalyst, niobium was observed as
Nb5+ with a binding energy of 208 eV. Previous studies have
shown that the binding energies at the V 2p3/2 level for V4+
and V5+ are observed at 516.9 and 518.0 eV, respectively
(24). The percentage of surface V4+ and V5+ of the two
catalysts (as determined by XPS) has been calculated using
a deconvolution method (25). Table 2 gathers together the
XPS results obtained from the activated VPO and VNbPO
Doping of (VO)2P2O7 by Nb, i.e., the dissolution of Nb5+
into the lattice of vanadyl pyrophosphate, was evidenced,
directly, by electrical conductivity measurements. Indeed,
Fig. 10 shows the evolution of σ (in ꢀ−1 · cm −1) with time
under air for the two catalysts. The electrical conductivity σ
of VNbPO is half an order of magnitude lower than that of
VPO. (VO)2P2O7 is known to be a p-type semiconductor
(26), which suggests an n-doping character of Nb. The same
observation was found when comparing the same materials
calcined at 750◦C under N2. According to the valence induc-
tion law, described in Ref. (18), n-type doping by Nb may
be explained by a substitution to V4+ by Nb5+ in (VO)2+
according to the scheme
TABLE 2
Superficial Atomic Percentage and Atomic Ratios
as Measured by XPS
[(VO)2P2O7] + [Nb5+] → V1−x Nbx O2P2O7+x
,
[4]
VPO (%)
VNbPO (%)
O 1s
V 2p3/2
64.3
10.9
66.9
11.2
which is made possible because of the similar ionic radii of
the corresponding cations: Nb5+ = 6.4 × 10−2 nm, V4+
=
29 (V5+)–71 (V4+
)
49 (V5+)–51 (V4+
4.1
)
5.8 × 10−2 nm.
C 1s
Nb 3d
7.3
—
For (VO)2P2O7, a V4+ phase which is known to be
a p-type semiconductor, the majority charge carriers
are positive holes p+ (i.e., electron vacancies), the ex-
istence of which is linked to the V5+/V4+ redox couple
0.36
Nb/V = 0.032
P 2p
17.5
P/V = 1.60
17.4
P/V = 1.55

244
DUARTE DE FARIAS ET AL.
FIG. 9. TEM micrographs and electron diffraction patterns from the activated VNbPO and VPO catalysts.

SELECTIVE OXIDATION OF n-BUTANE ON MODIFIED (VO)2P2O7
245
involved in Eqs. [7] and [8]:
ꢀ
ꢁ0
1
2
O2S− + (VO
)
+
O2 ↔ 2(O−S )+.
[10]
2+
2−
The overall redox process can be related to a Mars–van
Krevelen mechanism. These results suggest that the initial
step of n-butane activation on the V–P–O catalyst is the
attack by O−S originating from the p-type semiconductor
character of (VO)2P2O7. The doping of the V–P–O catalyst
by Nb5+ gives rise to a decrease of σ (Fig. 10) correspond-
ing to a substitution to V4+ in (VO)2+ by Nb5+ (Eq. [4]).
The decrease in V4+ in (VO)2+ sites implies the creation of
additional or excess anionic oxygen (O2S−)0. (See Eq. [4].)
This “constitutional” presence of excess an−ionic oxygen in-
creases the concentration in oxidizing (OS )+ species, ac-
cording to Eq. [6]. Since these species are responsible for
the n-butane activation (Eqs. [8] and [9]), one observes a
significantly higher conversion of n-butane as a function of
the time of activation (Fig. 4).
FIG. 10. Electrical conductivity measurements at 400◦C under air for
the activated VNbPO and VPO catalysts.
(associated with residual V5+ in (VO)2P2O7) according to
the equilibrium:
By electron microscopy, the VNbPO catalyst showed a
higher density of structural defects (Fig. 9). It may be in-
ferred that it will be connected to an increase in the number
of p+ and of O− species. The first step of the mechanism will
be favored by an increase in the n-butane conversion, which
is effectively observed. This result is confirmed by the in-
crease in the V5+ surface density, as observed by XPS, since
this goes in parallel with the density of p+ entities (Eq. [5]).
For the VNbPO catalyst, however, the subsequent steps
of the reaction mechanism will not be modified, as may be
deduced from the similar selectivity profile when compared
to VPO.
V5+ ↔ V4+ + p+.
[5]
The temporal increase of the electrical conductivity, ob-
served for both catalysts, is indicative of gradual oxidation
of the solids. Indeed, the p+ entities are strong oxidizing
agents, as observed in photocatalysis (27).
The initial activation step of the n-butane molecule, i.e., a
C–H bond cleavage from a methylene group (28), involves
a hole in the valence band of O2− anions. At the surface
(S), this hole can be filled by an electron hopping from a
neighboring anion according to the scheme:
ꢀ
ꢁ0
On (VO)2P2O7, Lewis acidity has been explained by
unsaturated cations or oxygen vacancies, associated with
structural defects (29), while Brønsted acidity correlates
with acid hydroxyl groups connected to P or V atoms
(30, 31). By selective poisoning of acidic Brønsted sites,
it was demonstrated that Brønsted acid sites (V–OH or
P–OH) do not participate to the initial n-butane activa-
tion, since conversion was not inhibited in this case (32).
Lewis acid sites associated with structural defects should
be considered as the activating sites for n-C4. Figure 11
shows the variation of the area of the characteristic pyridine
bands (33) of Lewis sites at 1450 cm−1 and Brønsted sites at
1540 cm−1 as a function of the temperature of thermodes-
orption. It appears that there are globally fewer acid sites
on VNbPO as compared to VPO. This figure provides in-
teresting information on the influence of Nb doping on
the relative distribution of these two types of acid sites:
on both catalysts there are more Lewis sites as compared
to Brønsted sites, which is generally observed on vanadium
phosphorus oxide catalysts. However, there are many more
Lewis sites of lower acidity (sites desorbing pyridine at a
O2S− + p+ ↔ (O−S )+.
[6]
(O2S−)0 is a −neutral entity with respect to the bulk solid,
whereas (OS )+ is positively charged and constitutes the
chemical support of the hole p+. Therefore, O−S species are
the oxidative ones for n-butane activation:
ꢀ
ꢁ0
C4H10 + (O−S )+ ↔ C4H9 + H+ + O2S−
.
[7]
·
·
The butyl radical C4H9 subsequently reacts with a second
O−S species to generate butene as the first reaction interme-
diate:
ꢀ
ꢁ0
C4H9 + (O−S )+ ↔ C4H8 + H+ + O2S−
The two protons H+ formed in Eqs. [7] and [8] are elimi-
nated as water by reaction with one surface oxygen anion
and by the formation of a doubly ionized anionic vacancy
.
[8]
·
2+
2−
(VO
)
:
ꢀ
ꢁ0
2H+ + O2S− ↔ H2O + (VO
)
.
[9]
2+
2−
2−
The anionic vacancy (VO
)
2+ will be subsequently filled by temperature lower than 250◦C) on VNbPO as compared to
dioxygen from the gas phase with the participation of one VPO (sites desorbing pyridine at a temperature lower than
surface oxygen anion to regenerate the two holes (O−S )+ 100◦C). This result confirms the fact that Lewis acid sites

246
DUARTE DE FARIAS ET AL.
REFERENCES
1. Hodnett, B. K., Catal. Rev.–Sci. Eng. 27, 373 (1985).
2. Centi, G., Trifiro, F., Ebner, J. R., and Franchetti, V. M., Chem. Rev.
88, 55 (1988).
3. Hodnett, B. K., Catal. Today 1, 477 (1987).
4. Hutchings, G. J., Appl. Catal. 72, 1, 1991.
5. Zazhigalov, V. A., Haber, J., Stoch, J., Pyatnitzkaya, A. I., Komashko,
G. A., and Belousov, V. M., Appl. Catal. A 96, 135 (1993).
6. Ben Abdelouahab, F., Olier, R., Ziyad, M., and Volta, J. C., J. Catal.
157, 687 (1995).
7. Sanane`s-Schulz, M. T., Ben Abdelouahab, F., Hutchings, G. J., and
Volta, J. C., J. Catal. 163, 346 (1996).
8. Sajip, S., Bartley, J. K., Burrows, A., Rhodes, C., Volta, J. C.,
Kiely, C. J., and Hutchings, G. J., Phys. Chem. Chem. Phys. 3, 2143
(2001).
9. Mota, S., Volta, J. C., Vorbeck, G., and Dalmon, J. A., J. Catal. 193, 319
(2000).
10. Hutchings, G. J., Desmartin-Chomel, A., Olier, R., and Volta, J. C.,
Nature 368, 41 (1994).
11. Guliants, V. V., Holmes, S. A., Benzinger, J. B., Heaney, P., Yates, D.,
and Wachs, I. E., J. Mol. Catal. A 172, 265 (2001).
12. Higgins, R., andHutchings, G. J., U. S. Patent4,418,003(1983), assigned
to Imperial Chemical Industries.
13. Matsuura, I., Ishimura, T., Hayakawa, S., and Kimura, N., Catal. Today
28, 133 (1996).
14. de Oliveira, P., Eon, J. G., Chavant, M., Riche´, A. S., Martin, V.,
Caldarelli, S., and Volta, J. C., Catal. Today 57, 177 (2000).
15. Thorsteinson, E. M., Wilson, T. P., Young, F. G., and Kasai, P. H.,
J. Catal. 52, 116 (1978).
16. Ciambelli, P., Lisi, L., Russo, G., Viparelli, P., and Volta, J. C., Appl.
Catal. A 184, 291 (1999).
17. Johnson, J. W., Johnston, D. C., Jacobson, A. J., and Brody, J. F., J. Am.
Chem. Soc. 106, 8123 (1984).
18. Herrmann, J. M., in “Catalyst Characterization, Physical Techniques
for Solid Materials” (B. Imelik and J. C. Ve´drine, Eds.), p. 559. Plenum
Press, New York, 1994.
FIG. 11. Relative variation of the area of the Lewis (1450 cm−1 and
Brønsted (1540 cm−1) pyridine bands as a function of the temperature of
desorption for VNbPO and VPO catalysts.
of lower acidity are associated with the structural defects.
Indeed by Nb doping, as observed by electron microscopy,
defect density is strongly increased and this enhances the
number of weaker Lewis acid sites. We agree on the impor-
tantroleofLewisacidsites, but, unlikepreviouslypublished
results (13), we disagree on the strength of their acidity.
Our studies (14; this publication) lead us to propose that
Nb is located in close proximity to vanadium in the vana-
dium phosphorus oxide structure, near to a positive hole,
which will facilitate the C–H bond breaking by surface O−
species. It should be noted that the V–O–Nb moiety has
recently been proposed to be a paraffin activator (34).
19. Martin, V., Millet, J. M. M., and Volta, J. C., J. Therm. Anal. 53, 111
(1998).
CONCLUSIONS
20. Ben Abdelouahab, F., Olier, R., Guilhaume, N., Lefebvre, F., and
Volta, J. C., J. Catal. 134, 151 (1992).
21. Sanane`s, M. T., Tuel, A., Hutchings, G. J., and Volta, J. C., J. Catal. 148,
395 (1994).
22. Tuel, A., Canesson, L., and Volta, J. C., Colloids Surf. A 158, 97
(1999).
23. Brown, I. D., and Wu, K. K., Acta Crystallogr. Sect. B 32, 1957
(1976).
24. Garbassi, F., Bart, J. C., Tassinari, R., Vlaic, G., and Lagarde, P.,
J. Catal. 98, 317 (1986).
25. Abon, M., Bere, K. E., Tuel, A., and Delichere, P., J. Catal. 156, 25
(1995).
26. Herrmann, J. M., Vernoux, P., Bere, K. E., and Abon, M., J. Catal. 160,
106 (1997).
27. Herrmann, J. M., Catal. Today 53, 115 (1999).
28. Pepera, M. A., Callahan, J. L., Desmond, M. J., Milburger,
E. C., Blum, P. R., and Bremer, N. J., J. Am. Chem. Soc. 107, 4883
(1985).
This study has demonstrated that Nb may be used as an
effective dopant to increase the conversion of n-butane for
mild oxidation to maleic anhydride. The main interest of
Nb stems from the increase in the number of defects at the
periphery of (VO)2P2O7 platelets, which increases the num-
ber of O-vacancies and facilitates the initial C–H activation
of n-butane. These sites correspond to unsaturated cations
which act as Lewis acid sites of low acidity. However, doping
(VO)2P2O7 by niobium does not improve the selectivity to
maleic anhydride. It was observed that its presence changes
the CO2/CO product distribution in agreement with the
higher V surface oxidation state of the VNbPO catalyst as
observed by XPS.
ACKNOWLEDGMENTS
29. Busca, G., Centi, G., and Trifiro, F., Appl. Catal. 25, 265 (1986).
30. Centi, G., Trifiro, G., Busca, G., Ebner, J. R., and Gleaves, J., Faraday
Discuss. Chem. Soc. 87, 215 (1989).
31. Cornaglia, L. M., Lombardo, E. A., Anderson, J. A., and Garcia Fierro,
J. L., Appl. Catal. A 100, 37 (1993).
32. Centi, G., Colinelli, G., and Trifiro, F., Appl. Catal. 48, 13
(1989).
33. Kno¨zinger, H., Adv. Catal. 184, 25 (1976).
34. Grasselli, R. K., Top. Catal. 15(2–4), 93 (2001).
The authors are indebted to Vincent Martin for the SEM experiments,
to Miche`le Brun for the XPS measurements, and to Jean Billy for the
acidity measurements. They thank CNPq (Conselho Nacional de De-
senvolvimento e Pesquisa—Brazil) and CNRS (Centre National de la
Recherche Scientifique—France) for the financial support of this work
within the framework of a programme PICS (Programme International
de Coope´ration Scientifique).