Selective oxidation of n-butane on a V-P-O catalyst: Improvement of the catalytic performance under fuel-rich conditions by doping

Article abstract of DOI:10.1006/jcat.2000.2903

The effect on the resistance to reduction by n-butane of the vanadium phosphorus oxide (V-P-O) catalyst modified by cobalt and molybdenum doping was studied under fuel-rich conditions at 400°C. Doping with cobalt or molybdenum led to such a promoting effect. Cobalt as dopant had more oxidizing power on the V-P-O system than molybdenum. It favored a higher V⁵⁺/V⁴⁺ ratio under fuel-rich oxidation conditions, limited the carbon deposition, and made possible the production of maleic anhydride (MA) on this modified V-P-O system. The Mo-doped V-P-O catalysts did not give performances as good as those for the Co-doped V-P-O catalyst. It was necessary to dope at a higher percentage, Mo/V = 2.56%, to achieve higher selectivity for MA. The nature of the dopant is more important than its way of introduction into the V-P-O matrix. It opens new routes for adapting V-P-O catalysts for oxidizing-butane to maleic anhydride under fuel-rich conditions and for developing new technologies.

Full text of DOI:10.1006/jcat.2000.2903

Journal of Catalysis 193, 319–329 (2000)
doi:10.1006/jcat.2000.2903, available online at http://www.idealibrary.com on
Selective Oxidation of n-Butane on a V–P–O Catalyst: Improvement of
the Catalytic Performance under Fuel-Rich Conditions by Doping
S. Mota, J. C. Volta, G. Vorbeck,† and J. A. Dalmon
Institut de Recherches sur la Catalyse, 2 Avenue Albert Einstein, 69626 Villeurbanne Ce´dex, France; and †Haldor Topsoe A/S, Nymollevej,
55 P.O. Box, Lyngby 2800, Denmark
Received January 12, 2000; revised April 24, 2000; accepted April 24, 2000
During the activation process under the nC4-air flow, the
VOHPO4, 0.5H2O follows two parallel routes, the dehy-
dration to (VO)2P2O7 and the oxidation to VOPO4 phases
which are progressively reduced (6). This model has been
evidenced by using the complementarity of several tech-
niques which analyze the bulk and the surface properties
of the corresponding V–P–O system during the activation
process. The XPSstudyhasshown that, under fuel-lean con-
ditions, the V5+/V4+ ratio progressively decreases in paral-
lel with the increase of the maleic anhydride yield (6). The
positive role of the V5+ sites was evidenced by compar-
ing the catalytic activity of a standard (VO)2P2O7 catalyst
with that of the same (VO)2P2O7 catalyst slightly reoxidized
(7, 8). From thisstudy, it wasconcluded that, under fuel-lean
conditions, a suitable surface V5+/V4+ ratio estimated to be
0.25 corresponded to the improvement of the MA yield ob-
served with the initial reoxidation of the (VO)2P2O7 cata-
lyst. The selectivity for MA strongly increased at the early
stage of this reoxidation, in parallel with the development
of isolated V5+ sites in strong interaction with (VO)2P2O7. It
was previously postulated that the isolated V5+ sites should
be located in the (100) (VO)2P2O7 face (9). For longer oxi-
dation treatments, the development of amorphous V5+ mi-
crodomains was observed to be detrimental to nC4 selective
oxidation (8). From this study, it was concluded that a suit-
able oxidizing power (redox state) was necessary for the
partial oxidation of n-C4, in agreement with the industrial
DuPont process using the RSR recirculating riser reactor,
where the V–P–O catalyst is submitted to periods of reox-
idation in a regenerator (10, 11) alternating with periods
of interaction with n-butane for production of MA by mild
oxidation.
The oxidation of n-butane to maleic anhydride is possible under
fuel-rich conditions (nC4/O2 = 0.6) at 400 C by modifying the vana-
dium phosphorus oxide (V–P–O) catalysts by doping with Co or Mo;
it is impossible on the undoped V–P–O catalyst, which deactivates
under the same conditions. The doped precursors were prepared by
reduction of the dihydrate VOPO4, 2H2O, and dopants were intro-
duced at different moments of the preparation from cobalt acetyl-
acetonate and ammonium heptamolybdate solutions, respectively.
The Co-doped V–P–O catalyst (Co/V = 0.77%) conserves a higher
capacity for reoxidation of the surface (higher V5+/V4+ ratio) and
maintains a higher surface distribution of the V, P, and O atoms
under fuel-rich conditions: O2 consumption is not total. Maleic an-
hydride is then produced and the catalysts do not deactivate. The
Mo-doped V–P–O catalysts (Mo/V from 0.88% up to 2.56%) do not
give performances as good as those for the Co-doped V–P–O cata-
lyst. It appears that it is necessary to dope at a higher percentage,
Mo/V = 2.56%, in order to achieve higher selectivity for MA. The
corresponding catalyst is less active (5.3% nC4 conversion instead of
25.5%) than the Co-doped V–P–O catalyst. This study demonstrates
that the nature of the dopant is more important than its way of in-
troduction into the V–P–O matrix. It opens new routes for adapting
V–P–O catalysts for oxidizing n-butane to maleic anhydride under
c
fuel-rich conditions and for developing new technologies.
2000
Academic Press
Key Words: n-butane partial oxidation; maleic anhydride; doped
V–P–O catalysts; fuel-rich conditions.
1. INTRODUCTION
The oxidation of n-butane to maleic anhydride (MA) on
vanadium phosphorus oxides has been the subject of many
studies by different groups. Several reviews have been de-
voted to this catalytic system (1–3). Most of the catalytic in-
vestigations and the corresponding physicochemical studies
have been done with a high oxygen/n-butane ratio (1–3%
nC4/air), usually termed fuel-lean conditions. In this case,
a single phase, the vanadyl pyrophosphate (VO)2P2O7 is
considered as being the active phase of the catalyst. Its mi-
crostructure is a key factor in the control of the final cata-
lytic properties by the preparation of its precursor (4, 5).
When considering the V5+/V4+ distribution at the sur-
face of the V–P–O catalyst, it is obvious that the catalytic
properties for mild oxidation of n-butane will depend on
the structural properties of the material, as largely stud-
ied, and also on the redox state of the catalyst imposed
by the nC4/O2 ratio of the gaseous flow, these two aspects
being interconnected. A few publications refer to results
obtained when working with a high nC4/O2 ratio and under
319
0021-9517/00 $35.00
c
Copyright
2000 by Academic Press
All rights of reproduction in any form reserved.

320
MOTA ET AL.
fuel-rich conditions (12–14). Also, some patents claim in- and Mo doping, under fuel-rich conditions. The prepara-
terest in such conditions (15, 16). The industrial interest tion implies a vanadium phosphorus oxide catalyst pre-
in fuel-rich conditions is due to the fact that maleic anhy- pared by the reducing preparation route of VOPO4, 2H2O,
dride could be recovered without condensing water, as is termed in a previous publication the “VPD route” (19).
the case under conventional fuel-lean conditions. But the Indeed, it was observed that among the different routes
fuel-rich conditions require a larger catalyst volume since of preparation, the VPD route gave the (VO)2P2O7 phase
the MA yield in this case would be lower as compared to with the highest surface platelet morphology exposing
under fuel-lean conditions (14). However, the main prob- preferentially the (100) active plane for maleic anhydride
lem is the reduction of the V–P–O catalyst, which affects formation.
its performance. In a recent study (17), we have investi-
gated a V–P–O catalyst under fuel-rich conditions in or-
der to develop a new type of reactor based on a ceramic
2. EXPERIMENTAL
porous tubular membrane. In this reactor, the V–P–O cata-
lyst would be placed as a fixed-bed in the inner volume of
2.1. Catalyst Preparation
the tube and the reactants would be fed separately from
both sides of the membrane. Butane would be introduced
in the inner volume of the tube, and oxygen would be intro-
duced through the membrane which would act as oxygen
distributor along the catalyst bed and thus would maintain
a low O2/C4 ratio. To simulate the situation of the first lay-
ers of the catalyst bed in the membrane reactor (where the
O2/C4 ratio is the lowest), the work was done, using a con-
ventional microreactor, with a O2/C4 ratio of 0.6. For these
conditions, a rapid decrease in maleic anhydride production
was observed with time. It was associated with reduction of
the V–P–O catalyst: the V5+ species were either reduced to
V4+ or masked and/or poisoned by carbonaceous deposits.
The reduction was limited to V4+ since no signal for V3+
species was observed by 31P NMR, spin echo mapping, or
XPS. It was concluded that the V–P–O catalyst cannot be
used in a membrane reactor without cofeeding oxygen with
nC4 at the inlet of the membrane reactor in order to avoid
a total consumption of oxygen and thus to limit total oxi-
dation. In this publication, it was also suggested that these
problems should be overcome by the development of the
V–P–O catalyst which should result in an improvement of
the resistance of the (VO)2P2O7 phase toward reduction
imposed by the high C4 concentration under the fuel-rich
conditions.
Studies under fuel-rich conditions suggested that the MA
yield was improved when the V–P–O catalyst was doped
(14–16). In a previous study, under such conditions, it was
observed that doping with Co resulted in the production
of MA at lower temperature as compared to the undoped
V–P–O catalyst. Selectivity is improved by Co doping (18).
This was associated with this catalyst’s higher susceptibility
to oxidation as evidenced, by Raman spectroscopy, with the
appearance of the VOPO4 structure at lower temperature
during the activation of the corresponding precursor, and
by XPS, with a higher V5+/V4+ ratio on the final catalyst.
It was thus interesting to study the effect of doping by Co
under fuel-rich conditions and to compare it with doping
by Mo.
Two families of doped V–P–O precursors have been pre-
pared using the same route which consists in reduction
of VOPO4, 2H2O by isobutanol and is termed the VPD
route (19). The preparation of VOPO4, 2H2O has been pre-
viously described (20). The resulting VOPO4, 2H2O was
recovered by filtration, washed with water, and identified
by XRD. Then this solid was refluxed with isobutanol for
21 h, and the resulting blue solid was recovered by filtra-
tion and dried in air for 16 h at 110 C. The final solid gives
a VOHPO4, 0.5H2O precursor denoted “VPD” (19).
For the preparation of the Co-doped VPD precursor, the
required massofthe correspondingacetylacetonate salt was
dissolved in isobutanol according to the atomic Co/V stoi-
chiometry, prior to the refluxing of the VOPO4, 2H2O with
the isobutanol solution. The same procedure as for the un-
doped VPD precursor was followed.
For the preparation of the Mo-doped VPD precursor,
a VPD precursor was first recovered after 6 h of reflux-
ing VOPO4, 2H2O with isobutanol. Then the VPD pre-
cursor was impregnated by an aqueous heptamolybdate
solution with the suitable Mo/V ratio. Three MoVPD pre-
cursors, denoted VPDMo1, VPDMo2, and VPDMo3, were
thus prepared with an increasing Mo/V ratio, following
the same procedure. The last MoVPD was prepared dif-
ferently, according to a modified VPD route: in this case,
the heptamolybdate solution was dissolved in isobutanol,
the VOPO4, 2H2O solid was added with the required Mo/V
ratio, and the mixture was refluxed for 16 h. Figure 1 sum-
marizes the preparation of the catalysts.
2.2. Catalyst Activation and Testing
The catalytic experiments were carried out in a fixed-
bed reactor at atmospheric pressure as previously de-
scribed (17). The reactor, a classical tubular Pyrex mi-
croreactor, was equipped with a thermocouple inside the
catalytic bed for continuous temperature control. Sampling
valves were inserted in a hot box heated at 150 C to avoid
any condensation of maleic anhydride. The flow of gas re-
actants was regulated by mass flow controllers. The feed
consisted of a mixture of n-butane, oxygen, and helium.
In this publication, we study the effect on the resistance
to reduction by nC4 of the V–P–O catalyst modified by Co

IMPROVED PERFORMANCE OF n-BUTANE ON V–P–O BY DOPING
321
The 31P nuclear magnetic resonance (NMR) spectra
were recorded with a Brucker DSX400 spectrometer, at
161.9 mHz, equipped with a standard 4-mm probe head.
The 31P spin echo mapping (SEM) spectra were obtained
with a sweep width of 2 mHz, t = 20 ms, and 90 pulse length
of 1.5 ms. The 31P MAS spectra were recorded at 12 kHz
with a P/2 acquisition sequence, a pulse length of 1.5 ms,
and a recycle delay of 60 s.
The XPS analysis was performed in a VG Escalab 200R
machine using Mg K radiation. The electrical charge was
corrected by setting the binding energy (BE) of adventi-
tious carbon (C1s) at 284.5 eV. For quantitative analysis,
we used the integrated area under V2p3/2, O1s, P2p, and
C1s peaks after smoothing and subtraction of a nonlinear
Shirley background. Analysis of the V2p3/2 level allowed
measurement of the changes in the surface oxidation state
of vanadium (V5+, V4+) with a peak decomposition and
curve fitting technique, as already described (6).
The catalysts after activation, as previously described,
were examined by scanning electron microscopy (SEM) us-
ing a Hitachi S800. The atomic composition was measured
by X-ray emission spectroscopy (EDX) with a JEOL 840.
The depth of examination depends on the elements consid-
ered and on the accelerating voltage. It is 0.55 to 1 mm for
V, 1.96 to 3 mm for P, 0.34 to 0.6 for Co, and 0.36 to 0.7 for
Mo with 20 keV of voltage.
FIG. 1. Synthesis of doped and undoped catalysts.
The VPDCo precursor was activated for 1 h at 390 C
under air and then 17 h at 460 C under 1.5% nC4 in air.
The VPDMo precursors were activated for 75 h at 400 C
under 1.5% nC4 in air. The feed consisted of a mixture of
n-butane, oxygen, and helium. The catalytic reaction was
carried out at 400 C under reducing nC4-rich conditions
3. RESULTS
Characterization of the activated catalysts (before testing
under fuel-rich conditions) is given in Table 1.
For comparison, yield to MA, YMA, obtained for n-butane
mild oxidation under fuel-rich conditions (O2/C4 = 0.6) is
given in Fig. 2 for VPD, VPDCo, and VPDMo1 as a func-
tion of the time-on-stream (other VPDMo catalysts behave
similarly to VPDMo1 in Fig. 2, with only the final yield be-
ing slightly affected). In contrast with the VPD catalyst,
it appears that at steady state promoted systems produce
maleic anhydride after 40 h on stream. This demonstrates
the interest of doping V–P–O when catalyzing nC4 oxida-
tion under fuel-rich conditions.
1
(O2/C4 = 0.6) with a constant GHSV of 3000 h and the
corresponding gas composition: nC4/O2/He = 16.6/10/73.4.
Standard fuel-lean conditions, O2/C4 = 12, corresponding
to nC4/O2/He = 0.8/10/92, were also used for the initial ac-
tivating XPS measurements.
Reactants and reaction products were analyzed by on-
line gas chromatography using double FID detection
(HP5890A series II) with HP-PLOT/Al2O3 (C4H10) and
HP-INNOWAX (MA) columns followed by a TCD detec-
tor (Intersmat GC IGC 120 MB) with a CTR1 column (O2,
CO, CO2). A computer was used for integration of the chro-
matographic peaks with HP Chemstation software.
TABLE 1
Some Characteristics of the Catalysts after Activation
under Standard Conditions
2.3. Precursor and Catalyst Characterization
Mass percentage (% )
Atomic (% )
Co/V Mo/V
The chemical analysis was done by a spectrometer of
atomic adsorption (Perkin–Elmer 1100). Solids were dis-
solved in hydrochloric acid. Measurement of the specific
surface area of the catalysts was done by the BET method VPDCo
with volumetric adsorption of N2 at 77.4 K using a home-
made apparatus. X-ray diffraction (XRD) patterns were
collected with a Siemens D500 diffractometer using Cu K
radiation.
Cat.
VPD
V
P
Co
Mo
30.32
28.71
29.39
29.98
29.72
28.97
18.94
17.24
18.09
19.23
19.05
28.45
—
0.27
—
—
—
—
—
0.46
0.67
1.42
0.80
—
0.77
—
—
—
0.83
1.21
2.56
1.45
VPDMo1
VPDMo2
VPDMo3
VPDMo4
—
—
—
—

322
MOTA ET AL.
tion of maleic anhydride (17). Cracking catalytic products
(17% ) are detected due to the reducing feed conditions. The
VPDCo catalyst better oxidizes nC4 due to its higher BET
area. In this case, the conversion of O2 is not total (77.9% ),
which is the main effect of Co doping. Although butenes
are produced with almost the same selectivity (42% ) as
for VPD, the main difference is the production of maleic
anhydride with more CO2 (23.8% ) and less CO (22.4% ).
This is indicative, for the VPDCo catalyst, of the tendency
of nC4 oxidation toward MA. This tendency is also con-
firmed on the VPDMo series for which MA is easily pro-
duced with the positive influence of the addition of molyb-
denum. Note that the effect on the MA selectivity increase
is much more pronounced at almost the same level of dop-
ing for VPDCo (Co/V = 0.77% , SMA = 11.5% ) as compared
to VPDMo1 (Mo/V = 0.83% , SMA = 5.7% ). However, the
nC4 conversion decreases with increased doping by Mo but
produces more selective sites for MA production. This is
also observed from the increase of the MA intrinsic activity
(Table 2).
The surface area of the catalysts was measured after the
test. It is given in Table 2. The VPDCo catalyst shows a high
surface area (28.0 m2/g) as compared to the undoped VPD
catalyst (17.1 m2/g), and this appears to be the consequence
of the doping effect by cobalt at a low level (Co/V = 0.77% ).
Let us recall that, in this case, the doped catalyst was pre-
pared by dissolving the Co dopant into the isobutanol re-
ducing solution as acetylacetonate salt, prior to refluxing
the VOPO4, 2H2O dihydrate (see Fig. 1). The preparation of
the VPDMo series follows a different route which consists
of an impregnation of the VOHPO4, 0.5H2O by a molybde-
num heptamolybdate solution. After the test, the BET area
of the corresponding Mo catalysts is low, particularly at a
low Mo percentage: the BET area is 16.5 m2/g for VPDMo1,
17.5 m2/g for VPDMo2, and 21.9 m2/g for VPDMo3 and
does not differ strongly with the VPD catalyst. The per-
FIG. 2. MA yield for undoped and doped catalysts during the reaction
under fuel-rich conditions: (a) VPD, (b) VPDCo, and (c) VPDMo1.
Steady-state catalytic results (after 45 h) are summa- centage of Mo increases in the VPDMo series, as expected
rized in Table 2. The VPD catalyst converts totally O2 and from the preparation (Table 1). It is important to note that
the main oxidation products are unsaturated C4 (SC4H8
=
the route which consists of a dissolution of Mo heptamolyb-
45.1% ) and CO (SCO = 32.8% ). Maleic anhydride is not date into isobutanol prior to refluxing with VOPO4, 2H2O
produced. CO2 is observed at a low level (5% ), which gives a catalyst with a very low surface area (6.0 m2/g), to-
fits the fact that CO2 is the fingerprint of the overoxida- tally inactive for n-butane oxidation.
TABLE 2
Catalytic Results for n-Butane Mild Oxidation under Fuel-Rich Conditions (O2C4 = 0.6) after 45 h on Stream at 400 C
Catalytic results (% )
IA
SBET
g
2
1
1
Cat.
VPD
C(C4H10
)
C(O2)
SMA
SC4H8
SCO
SCO2
YMA
(10 5 mol MA
m
h
)
(m2
)
15.5
25.5
16
8.2
5.3
100
77.9
96.5
81
0
11.5
5.7
8.3
36.5
45.1
42
47.2
33.7
1.1
32.8
22.4
26.6
30.7
28.8
5
0
0
17.1
28.0
16.5
17.5
21.9
VPDCo
23.8
20.5
27.3
33.8
2.9
0.9
0.7
1.9
4.62
1.28
1.34
3.25
VPDMo1
VPDMo2
VPDMo3
63.2

IMPROVED PERFORMANCE OF n-BUTANE ON V–P–O BY DOPING
323
FIG. 3. XRD spectra of undoped and Co-doped VPD catalysts: (a) VPD activated, (b) VPD after test under fuel-rich conditions, (c) VPD Co
activated, and (d) VPD Co after test under fuel-rich conditions.
The XRD spectra of the activated VPD (a) and equi- ter the catalytic test for the corresponding catalyst. Platelets
librated VPD (b) after test under fuel-rich conditions are appeared more eroded, with a lozenge shape, and curved.
given in Fig. 3, simultaneously with those of the correspond- For the activated VPDMo catalysts, the same type of mor-
ing activated (c) and equilibrated (d) VPDCo catalysts. phology was observed.
All spectra are similar and characteristic of the (VO)2P2O7
Table 3a showsthe EDX resultsobtained on the activated
phase with a broadening of the (200) reflection, typical of VPD, VPDCo, and VPDMo3 catalysts. EDX results are
the well-known VPD morphology which develops platelets given in parentheses for comparison with the XPS results.
with [100] normal (19). The XRD spectra of the VPDMo se- There are almost no differences with the XPS results if we
ries are given in Fig. 4 for the activated catalysts and in Fig. 5 take into account, for EDX, the contribution of C, which
for the VPDMo catalysts after test under reducing condi- cannot be detected by EDX, since C is deposited for the
tions. They are also characteristic of the VPD type morphol- EDX examination but measured by XPS.
ogy. The XRD spectrum of the activated VPDMo4 catalyst
Figure 7 presents the 31P NMR spin echo mapping spec-
(Fig. 4d) can be indexed as a mixture of different VOPO4 tra of the activated VPD, VPDCo, and VPDMo catalysts.
phases ( II-, -, -VOPO4, and VOPO4, 2H2O) with their While the activated VPD catalyst (Fig. 7a) shows a spec-
characteristic lines. The presence of these V5+ phases can trum typical of only the (VO)2P2O7 phase (essentially one
explain the poor performance of the corresponding catalyst signal at 2495 ppm), the activated VPDCo catalyst (Fig. 7b)
(Table 2).
shows, in addition to the signal of (VO)2P2O7, signals near
The activated VPD (a), VPDCo (b), and VPDMo3 (d) 0 ppm, typical of V5+ species, and in the 1000 ppm range,
catalysts were examined by SEM (Fig. 6). Sand-rose typical of V4+–V5+ dimers, which is indicative of a more ox-
platelets were observed for the three catalysts. Their sizes idized situation due to the effect of the Co dopant on vana-
were smaller for the activated VPDCo catalyst, which dium sites. The progressive increase of the oxidation state
should be correlated with the higher BET area observed af- of vanadium for the VPDMo activated catalysts (Figs. 7c,

324
MOTA ET AL.
TABLE 3a
XPS Results on Activated Catalysts (for Activation Conditions, See Paragraph 2.2)
Atomic surface distribution (% )
Doping
V (% )
V5+
Activated cat.
VPD
V
Na
P
O
C
metal
P/V
1.62
V4+
65.8
V5+/V4+
0.53
10.7
(15.7)
11.3
(4.5)
17.3
(15.3)
17.6
65.7
(64.2)
65.7
6.3
—
35.2
31.3
VPDCo
5.3
0.18
1.56
68.7
0.46
(14.5)
10.2
10.1
(2.1)
(15.1)
17.4
16.6
(65.2)
69.0
67.0
VPD Mo1
VPD Mo2
VPD Mo3
3.2
6.0
6.0
0.18
0.28
0.48
1.70
1.64
1.68
72
70
66
28
30
34
0.39
0.42
0.52
9.8
16.5
67.3
(14.3)
(4.5)
(14.9)
(64.9)
Note. EDX results are given in parentheses for comparison with XPS results.
FIG. 4. XRD spectra of fresh doped VPD Mo catalysts (activated): (a) VPD Mo1, (b) VPD Mo2, (c) VPD Mo3, and (d) VPD Mo4.

IMPROVED PERFORMANCE OF n-BUTANE ON V–P–O BY DOPING
325
of the activated VPDMo4 catalyst (Fig. 7f). Figure 8 gives
the 31P NMR spin echo mapping spectra of the VPD (a),
VPDCo (b), and VPDMo (c, d, and e) equilibrated cata-
lystsafter the test under reducingconditions(nC4/O2 = 0.6).
Figure 8 shows that the catalytic reducing treatment has
consumed most of the V5+ species and the V4+–V5+ dimers.
A slight signal near 0 ppm is, however, observed for the
doped catalysts, which indicates that the catalysts have still
conserved an oxidative capacity. This will be confirmed by
the XPS study. The displacement at a higher position of the
signal of vanadium pyrophosphate for the VPDMo series
(2610–2640 ppm) is indicative that this doping procedure
has favored the reorganization of the corresponding mate-
rials during the test under fuel-rich conditions.
Tables 3a and 3b compare the XPS results between
the catalysts activated under fuel-lean conditions (a) (see
Subsection 2.2) and the catalysts after the test under fuel-
rich conditions (b). The atomic surface distribution shows
comparable results for all the activated catalysts (Table 3a).
Note that the percentage of surface oxygen atoms is high
(64–66% ). The doping elements (Co and Mo) are detected
as expected. Note also the usual increased P/V ratio (1.60–
1.70), as generally observed for the V–P–O catalysts work-
ing toward nC4 oxidation under fuel-lean conditions. The
V5+/V4+ ratio for these catalysts as calculated from the de-
convolution of the V2p3/2 levels of V4+ and V5+ observed
at 516.9 and 518.0 eV, respectively (20), shows a value
around 0.4–0.5, usually observed for V–P–O catalysts work-
ing toward nC4 oxidation under fuel-lean conditions. Some
carbon is detected, which can be produced by contamina-
tion from carbonaceous species coming from the nC4/O2
atmosphere of activation of the precursors.Very different
information is observed from the atomic surface distribu-
tion after the test under reducing conditions (O2/C4 = 0.6)
(Table 3b). Results will be discussed with reference to the
same catalysts activated under fuel-lean conditions:
FIG. 5. XRD spectra of doped VPD Mo catalysts after test under
—For VPD, the V5+/V4+ ratio hasalso stronglydecreased
(from 0.53 down to 0.19). A strong increase of C is also ob-
fuel-rich conditions: (a) VPD Mo1, (b) VPD Mo2, and (c) VPD Mo3.
7d, and 7e) with increasing percentage of the Mo dopant is served (from 6.3 up to 55.6% ). V, P, and O have consider-
obvious from the increase of the contributions near 0 and at ably decreased. Note also that the P/V ratio has increased
1000 ppm. The quasi-absence of (VO)2P2O7 and the pres- (from 1.62 to 2.06) This should correspond to a higher acid-
ence of only VOPO4 phases is confirmed from the spectrum ity of the corresponding catalysts.
TABLE 3b
XPS Results on the Catalysts: After Test under Fuel-Rich Conditions (O2C4 = 0.6)
Atomic surface distribution (% )
V (% )
Doping
metal
Equilibrated cat.
V
P
O
C
P/V
V4+
V5+
V5+/V4+
V5+ VTotal
VPD
3.1
9.2
4.9
6.0
9.0
6.4
15.3
10.0
11.2
15.3
34.9
62.9
43.0
47.7
55.8
55.6
12.4
41.9
34.8
19.5
—
0.2
0.15
0.18
0.38
2.06
1.66
2.04
1.86
1.69
84
74
89
86
79
16
26
11
14
21
0.19
0.35
0.12
0.16
0.27
0.496
2.392
0.539
0.840
1.890
VPDCo
VPDMo1
VPDMo2
VPDMo3

326
MOTA ET AL.
FIG. 6. SEM examinations of the activated catalysts.
—For VPDCo, the V5+/V4+ ratio (0.35) is clearly higher and O (43% ). The P/V ratio is still high (2.04). When dop-
than that for VPD (0.19). The masking effect of C is much ing with more Mo, for VPDMo2 and VPDMo3, the atomic
less pronounced, and values not so far from the values ob- surface distribution approaches that of VPDCo, without
served on the corresponding activated VPDCo catalyst are reaching it, however: 9.0% for V, 15.3% for P, 55.8% for O,
observed: 9.2% (b) for V as compared to 11.3% (a), 15.3% and 19.5% for C for VPDMo3.
(b) for P as compared to 17.6% (a), and 62.9% (b) for O as
compared to 65.7% (a). The C percentage (12.4% ) is not
so high as compared to that of VPD (55.6% ). The P/V ratio
(1.66) approaches the values observed for the C4 fuel-lean
conditions of catalysis (1.56).
4. DISCUSSION
In a previous publication (17), we have shown that vana-
dium phosphorus oxide cannot be used for a long time on
—For VPDMo1, the V5+/V4+ ratio is low (0.12). The C stream as a catalyst for n-butane oxidation to maleic an-
(41.9% ) masking effect appears on V (4.9% ), P (10.0% ), hydride (MA) under fuel-rich conditions (C4/O2 = 0.6): the

IMPROVED PERFORMANCE OF n-BUTANE ON V–P–O BY DOPING
327
the production of unsaturated C4 and the total oxidation to
CO rather than to CO2. Carbon then covers the surface of
the catalyst, masks V and P, and limits the accessibility to
oxygen.
The Co-doped VPD catalyst avoids such poor behavior
under fuel-rich conditions. Indeed, the oxidizing power of
cobalt on vanadium is obvious from the 31P NMR spin echo
mapping spectrum of the corresponding activated catalyst,
which shows the increase of the signals associated with V5+
species (Fig. 7b). If most of the bulk V5+ species are con-
sumed under fuel-rich conditions (as observed by NMR,
Fig. 8), the oxidizing capacity of the VPDCo catalyst is ob-
vious from the high atomic surface V5+/V4+ ratio (0.35) as
well from the high atomic surface distribution for V, P, and
O (Table 3b). All XPS analysis information on this cata-
lyst under fuel-rich conditions approaches the XPS analy-
sis information on the same catalyst under fuel-lean con-
ditions: Co retains the active O species for the production
of MA in the vicinity of the V and P active species of the
catalyst.
The catalysts of the VPDMo, series, Mo-doped VPD
catalysts, show a behavior of the same type under fuel-
rich catalytic conditions, but a significant production of MA
FIG. 7. The 31P (SEM) NMR spectra of the activated undoped and
doped VPD catalysts: (a) VPD, (b) VPD Co, (c) VPD Mo1, (d) VPD Mo2,
(e) VPD Mo3, and (f) VPD Mo4.
V–P–O catalyst produces MA for a very short period and
consumes all oxygen in the feed. At this moment CO is
mainly produced with almost no CO2. It was concluded that
CO formation occurs mainly by the oxidation of adsorbed
n-butane whereas CO2 is formed by the overoxidation of
maleic anhydride. In the present publication, we improve
the oxidizing capacity of the V–P–O catalyst, and we show
that it is possible to limit the deactivation under fuel-rich
conditions by modifying the vanadium phosphorus oxide
catalyst by doping with Co or Mo.
The reference VPD type catalyst shows, under fuel-rich
conditions (O2/C4 = 0.6), a total consumption of O2 in the
feed. This is associated with the nonproduction of maleic
anhydride on this catalyst (Fig. 2 and Table 2), as previously
observed on the V–P–O type catalyst (17). The deactivation
of the vanadium phosphorus oxide catalyst is thus observed
under fuel-rich conditions independently of the morphol-
ogy of the (VO)2P2O7 crystals. The V–P–O material does
not suffer from the high reducing conditions of the test: this
corresponds to a decrease of the V5+/V4+ ratio and a P/V
FIG. 8. The 31P NMR (SEM) spectra of undoped and doped catalysts
after test under fuel-rich conditions (O2/C4 = 0.6): (a) VPD, (b) VPD Co,
increase as observed by XPS (Tables 3a and 3b). This favors (c) VPD Mo1, (d) VPD Mo2, and (e) VPD Mo3.

328
MOTA ET AL.
5. CONCLUSIONS
The main objective of this study was to develop a vana-
dium phosphorus oxide catalyst able to work under fuel-
rich conditions for nC4 mild oxidation. This study has
demonstrated that doping with Co or Mo leads to such
a promoting effect. We have observed that the nature of
the doping element is more important than its way of in-
troduction in the V–P–O catalyst. Co as dopant has more
oxidizing power on the V–P–O system than Mo: it favors a
higher V5+/V4+ ratio under fuel-rich oxidation conditions,
limits the C deposition, and makes possible the production
of maleic anhydride on this modified V–P–O system.
This study opens a new route for the development of
a new type of reactor based on a ceramic porous tubu-
lar membrane for which the chemical composition of the
V–P–O catalyst will be adjusted to be adapted to reducing
nC4/O2 catalytic conditions at the inlet of the membrane
reactor. Indeed, besides the choice of a proper membrane,
membrane reactors may also need a specifically designed
catalyst adapted to their catalysis conditions (22). Studies
are underway for optimization of these systems and espe-
cially to define the best method for dopant introduction.
FIG. 9. Correlation between MA intrinsic activity and XPS results.
is observed for doping at the higher level Mo/V = 2.56%
as compared to Co/V = 0.77% . The catalysts are, however,
much less active. The differences observed can be explained
by the different methods of preparation. As a matter of
fact, Co dopant has been dispersed in the V–P–O matrix
at the stage of the preparation of the corresponding pre-
cursor in organic medium. The different method of dop-
ing of the VOHPO4, 0.5H2O precursor by impregnation by
the Mo heptamolybdate solution should explain the dif-
ference observed between the Mo- and Co-doped VPD
catalysts.
ACKNOWLEDGMENTS
This work was conducted in the framework of the Contract Brite
BRPR-CT955-0046. The authors are indebted to the support of the Eu-
ropean Community. We gratefully acknowledge Dr. Khadija Aı¨t Lachgar
for her help in the preparation of one V–P–O precursor, Dr. P. Deliche`re
31
and M. Brun for XPS analysis, M.C. Durupty for P NMR, and V. Martin
for SEM and EDX examination of the catalysts.
In a previous publication (21), we established a corre-
lation for a series of doped and undoped V–P–O catalysts
between the MA intrinsic activity and the contribution of
the V4+–V5+ dimers (measured by 31P NMR by spin echo
mapping in the 200–1500 ppm range) for n-butane oxida-
tion under fuel-lean conditions (nC4/O2/He = 1.5/18.5/80,
T = 430 C, GSHV = 1000 h 1). It was observed by 31P NMR
(compare Figs. 7 and 8) that most of the V5+ species (bulk
as identified by this technique) are consumed under n-
butane fuel-rich oxidation conditions, but that a slight con-
tribution of these species was still present. XPS was used
to detect these species and to make correlations with the
catalytic results. A correlation is observed, Fig. 9, for the
catalysts of the two series and VPD between the maleic
anhydride intrinsic activity, IAMA (see Table 2) and the
surface V5+ species (measured by % V5+ VXPS). It should
thus be concluded that for nC4 oxidation under fuel-rich
conditions, the MA production depends on the superficial
REFERENCES
1. Hodnett, B. K., Catal. Rev. Sci. Eng. 27, 373 (1985).
2. Centi, G., Trifiro, F., Ebner, J. B., and Franchetti, V. M., Chem. Rev.
88, 55 (1988).
3. Cavani, F., and Trifiro, F., Catalysis 11, 246 (1994).
4. Duvauchelle, N., and Bordes, E., Catal. Lett. 57, 81 (1999).
5. Ellison, I. J., Hutchings, G. J., Sanane´s, M. T., and Volta, J. C., J. Chem.
Soc. Chem. Commun., 1093 (1994).
6. Abon, M., Be´re´, K. E., Tuel, A., and Deliche`re, P., J. Catal. 156, 28
(1995).
7. A¨ıt Lachgar, K., Abon, M., and Volta, J. C., J. Catal. 171, 383 (1997).
8. A¨ıt Lachgar, K., Tuel, A., Brun, M., Herrmann, J. C., Kraft, J. M.,
Martin, J. R., Volta, J. C., and Abon, M., J. Catal. 177, 224 (1998).
9. Volta, J. C., Be´re´, K. E., Zhang, Y. J., and Olier, R., in “ACS Symposium
Series, 523” (S. T. Oyama and J. W. Hightower, Eds.), p. 217. Am.
Chem. Soc., Washington, DC, 1993.
10. Contractor, R. M., Bergna, H. E., Chowdry, U., and Sleight, A. W., in
“Fluidization VI” (J. R. Grace, L. W. Shemilf, and M. A. Bergougnou,
Eds.), p. 589. Engineering Foundation, New York, 1989.
density of V5+ sites which should be restricted to the sur- 11. Contractor, R. M., Garnett, D. I., Horowitz, H. S., Bergna, H. E.,
face. It appears clearly that the higher the V5+/V4+ XPS
Patience, G. S., Schwartz, J. T., and Sisler, G. M., in “New Develop-
ments in Selective Oxidation II” (V. Corte´s Corberan and S. Bellon,
Eds.), p. 233, Elsevier, Amsterdam, 1994.
ratio, the lower the carbon deposition. For the unpromoted
VPD, the V5+ species are reduced, leading to dehydrogena-
12. Centi, G., Fornasari, G., and Trifiro, F., J. Catal. 89, 44 (1984).
tion reaction, originating the formation of carbonaceous
species.
13. Centi, G., Fornasari, G., and Trifiro, F., Ind. Eng. Chem. Prod. Res.
Dev. 24, 32 (1985).

IMPROVED PERFORMANCE OF n-BUTANE ON V–P–O BY DOPING
329
14. Hutchings, G. J., Appl. Catal. 72, 1 (1991).
15. Higgins, R., and Hutchings, G. J., U.S. Patent 4222945(1980). [Assigned
to Imperial Chemical Industries]
16. Higgins, R., and Hutchings, G. J., U.S. Patent 4317777(1982). [Assigned
to Imperial Chemical Industries]
17. Mota, S., Abon, M., Volta, J. C., and Dalmon, J. A., J. Catal. 193, 308
(2000).
18. Ben Abdelouahab, F., Olier, R., Ziyad, M., and Volta, J. C., J. Catal.
157, 687 (1995).
19. Kiely, C. J., Burrows, A., Sajip, S., Hutchings, G. J., Sanane´s, M. T.,
Tuel, A., and Volta, J. C., J. Catal. 162, 31 (1996).
20. Sanane´s-Schulz, M. T., Ben Abdelouahab, F., Hutchings, G. J., and
Volta, J. C., J. Catal. 163, 346 (1996).
21. Sanane´s-Schulz, M. T., Tuel, A., Hutchings, G. J., and Volta, J. C.,
J. Catal. 166, 388 (1997).
22. Dalmon, J-A., “Handbook of Heterogeneous Catalysis” (G. Ertl,
H. Kno¨zinger, and J. Weitkamp, Eds.), Chap. 9.3, pp. 1387–1398. VCH,
Weinheim/New York, 1997.