n-Butane oxidation using VO(H2PO4)2 as catalyst derived from an aldehyde/ketone based preparation method

Article abstract of DOI:10.1039/b002553o

A detailed study of n-butane oxidation over vanadium phosphate catalysts derived from in situ activation of VO(H₂PO₄)₂ is described and discussed. New methods of preparation of VO(H₂PO₄)₂ are described using the reaction of V₂O₅ and H₃PO₄ or VOPO₄ · 2H₂O with an aldehyde or ketone as a reducing agent. It is proposed that the enol form of the aldehyde and ketone act as the reducing agent. VO(H₂PO₄)₂ can also be obtained by reduction of VOPO₄ phases with alcohols. The catalysts derived from VO(H₂PO₄)₂ by in situ activation in 1.5% butane in air at 400 °C for 72 h are largely disordered and, by detailed powder X-ray diffraction and laser Raman spectroscopy, are shown to be comprised mainly of disordered material containing some untransformed VO(H₂PO₄)₂, VOPO₄ phases and possibly some VO(PO₃)₂. The catalysts derived from all the forms of VO(H₂PO₄)₂ investigated are found to exhibit maleic anhydride selectivities in the range 20-30% and it is concluded that the low selectivity is due to the presence of VOPO₄ phases in the activated catalysts.

Full text of DOI:10.1039/b002553o

View Article Online / Journal Homepage / Table of Contents for this issue
n-Butane oxidation using VO(H PO ) as catalyst derived from an
4 2
aldehyde/ketone based preparation method
2
Jonathan K. Bartley,a Colin Rhodes,a Christopher J. Kiely,b Albert F. Carleya and
Graham J. Hutchings*b
a Department of Chemistry, Cardi† University, Cardi†, UK CF10 3T B
b Department of Chemistry and Department of Material Science and Engineering,
University of L iverpool, UK L 69 3BX
Received 30th March 2000, Accepted 29th June 2000
First published as an Advance Article on the web 29th September 2000
A detailed study of n-butane oxidation over vanadium phosphate catalysts derived from in situ activation of
VO(H PO ) is described and discussed. New methods of preparation of VO(H PO ) are described using the
2
4 2
2
4 2
reaction of V O and H PO or VOPO É 2H O with an aldehyde or ketone as a reducing agent. It is
2
5
3
4
4
2
proposed that the enol form of the aldehyde and ketone act as the reducing agent. VO(H PO ) can also be
4 2
obtained by reduction of VOPO phases with alcohols. The catalysts derived from VO(H PO ) by in situ
activation in 1.5% butane in air at 400 ¡C for 72 h are largely disordered and, by detailed powder X-ray
di†raction and laser Raman spectroscopy, are shown to be comprised mainly of disordered material
containing some untransformed VO(H PO ) , VOPO phases and possibly some VO(PO ) . The catalysts
derived from all the forms of VO(H PO ) investigated are found to exhibit maleic anhydride selectivities in
the range 20È30% and it is concluded that the low selectivity is due to the presence of VOPO phases in the
2
2
4
4 2
2
4 2
4
3 2
2
4 2
4
activated catalysts.
In this preparation procedure the alcohol is oxidised to form
Introduction
either an aldehyde or a ketone. This has prompted us to con-
sider the role of aldehydes and ketones in the preparation
method, and in this paper we investigate vanadium phosphate
catalysts prepared using aldehydes and ketones in place of
alcohols. Interestingly, we have found that replacement of the
alcohol by an aldehyde or ketone leads to the formation of
VO(H PO ) . This water-soluble vanadium phosphate has
Vanadium phosphate compounds are utilised commercially as
catalysts for the synthesis of maleic anhydride from the partial
oxidation of n-butane. As a result, vanadium phosphates have
been extensively studied,1h4 however, most attention has
focused on the synthesis of suitable catalytic materials. The
current preferred catalyst is derived from in situ activation,5
under reaction conditions, of VOHPO É 0.5H O.6,7 The acti-
vated catalyst comprises a complex mixture of (VO) P O in
combination with a - and d-VOPO
researchers8 favour (VO) P O as the sole active phase, based
2
4 2
been shown to be present in low quantities in many samples
4
2
of VOHPO É 0.5H O4,13 and has been implicated in the poor
2 2
7
4
2
phases.6 Some
catalyst performance of these materials.13 More recently, a
II
4
sample of VO(H PO )
prepared by treatment of
2 2
7
2
4 2
primarily on powder X-ray di†raction data, and have con-
cluded that the presence of other vanadium phosphates is due
to incomplete activation. However, in situ Raman spectros-
copy suggests that speciÐc combinations of V4` and V5`
phases are a necessary requirement for the catalyst to exhibit
high activity and selectivity to maleic anhydride simulta-
neously. Recently, Coulston et al.9 have shown that the pres-
ence of V5` appears to be essential for the initial activation of
n-butane, an observation that is consistent with observations
using a TAP (temporal analysis of products) pulse reactor.10
The debate concerning the importance of speciÐc crystalline
phases of vanadium phosphates is complicated by the obser-
vation of “disordered materialÏ or “partly crystalline
(VO) P O Ï in vanadium phosphate catalysts prepared in
VOPO É 2H O with octan-3-ol14 was found to be highly
4
2
selective when tested in a di†erential reactor.15h17 In this
paper, we investigate the catalytic performance of
VO(H PO ) prepared by a range of methods.
2
4 2
Experimental
Catalyst preparation
Preparation of VOPO Æ 2H O.17 V O (5.0 g Strem.) was
4
2
2 5
reÑuxed with H PO (30 ml, 85% Aldrich) (P/V molar
3
4
ratio \ 15) in water (120 ml) for 24 h. The yellow solid (yield
50% based on V) was recovered by vacuum Ðltration, washed
with cold water (100 ml) and acetone (100 ml) and dried in air
(110 ¡C, 24 h).
2 2
7
organic media.11 These disordered materials are likely to
exhibit V5` and V4` ions in close proximity. In view of the
emphasis on catalysts derived from VOHPO É 0.5H O, few
Preparation of bVOPO .18 NH (VO) PO (1 g, prepared
4
2
4
4
2
4
alternative catalyst precursors have been investigated. To
some extent this is due to the fact that VOHPO É 0.5H O is
easily prepared by a range of methods and readily crystallises
according to the method of Bordes and Courtine19) was oxi-
dised in Ñowing dry air (50 ml min~1). The temperature was
increased from ambient at a rate of 1 ¡C min~1 to 600 ¡C and
then maintained at this temperature for 10 h.
4
2
from solutions containing (VO)2` and H PO in the presence
3
4
of aqueous HCl4,5 or organic reducing agents, typically alco-
hols.7,11,12 A preferred method of preparation involves reÑux-
ing V O in a solution of H PO in an alcohol for ca. 16 h,
Preparation of cVOPO .18 VOHPO É 0.5H O (1 g, pre-
4
4
2
pared according to the method of Kiely et al.6) was heated in
Ñowing dry oxygen (50 ml min~1). The temperature was
2
5
3
4
and this produces very pure samples of VOHPO É 0.5H O.12
4
2
DOI: 10.1039/b002553o
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006
This journal is ( The Owner Societies 2000
4999

View Article Online
increased from ambient at a rate of 2 ¡C min~1 to 680 ¡C and
quartz wool. A thermocouple was located in the centre of the
catalyst bed and temperature control was typically ^1 ¡C.
Carbon mass balances of P95% were typically observed.
VO(H PO ) catalyst precursors were activated in situ (1.5%
maintained at this temperature for 4 h.
Preparation of dVOPO .18 VOHPO É 0.5H O (1 g) was
4
4
2
2
4
heated in Ñowing dry oxygen (50 ml min~1). The temperature
was increased from ambient at a rate of 1 ¡C min~1 to 680 ¡C
and maintained at this temperature for 168 h.
n-butane in air, 1000 h~1) at 400 ¡C for 72 h.
A number of physical techniques were employed to charac-
terise the catalyst microstructure. Powder X-ray di†raction
(XRD) was performed using an Enraf Nonius FR590 X-ray
generator with a Cu-Ka source Ðtted with an Inel CPS 120
hemispherical detector. BET surface area measurements using
nitrogen adsorption were carried out using a Micromeritics
ASAP 2000 instrument. Raman spectra were obtained using a
Renishaw Ramanscope Spectrograph Ðtted with a green Ar`
laser (j \ 514.532 nm). Transmission electron microscopy
(TEM) analysis was carried out as described previously.6
X-ray photoelectron spectroscopy (XPS) was used to deter-
mine the P/V surface atomic ratio. XPS measurements were
made using a VG ESCA3 photoelectron spectrometer employ-
ing an Al-Ka X-ray source operating at 400 W. Samples were
mounted using double-sided adhesive tape. Binding energies
were referenced to the C(1s) peak at 285 eV and P : V ratios
calculated using the relative sensitivity factors due to
Wagner.19
Preparation of ““anhydrousÏÏ VOPO . An unidentiÐed crys-
4
talline V5` vanadium phosphate (designated ““anhydrous
VOPO ÏÏ) was prepared by heating V O (5.0 g, Strem) and
4
2 5
H PO (41.0 g, 100%, Aldrich) (P/V molar ratio \ 15) in an
3
4
autoclave at 150 ¡C for 16 h. The resultant yellow solid was
slurried with acetone, recovered by Ðltration, washed with
acetone (100 ml) and dried in air (110 ¡C, 16 h). The material
was stored under desiccation, on treatment with moist air
VOPO É 2H O formed and it is for this reason that the
4
2
material has been designated ““anhydrous VOPO .ÏÏ
4
Preparation of VO(H PO ) . (a) From V OPO É 2H O.
2
4 2
4
2
VOPO É 2H O (4 g) was reÑuxed with octan-3-ol (50 (mol
4
2
alcohol) (mol VOPO É 2H O)~1, Aldrich) for 16 h. The result-
ant pale blue solid was recovered by vacuum Ðltration,
washed with octan-3-ol (100 ml) and acetone (100 ml) and
dried in air (110 ¡C, 24 h).
4
2
Results and discussion
(b) From ““anhydrous V OPO ÏÏ. Anhydrous VOPO (2 g)
4
4
was reÑuxed with an alcohol (80 ml, Aldrich) for 16 h. The
resultant solid was recovered by vacuum Ðltration, washed
with the alcohol (50 ml) and acetone (50 ml) and dried in air
(110 ¡C, 16 h).
Catalyst precursor characterisation
Aldehydes and ketones as reducing agents. The VPO prep-
aration method6 was investigated using aldehydes and
ketones in place of the alcohol.6,12 Surprisingly, this method
of precursor preparation led exclusively to the formation of
VO(H PO ) and the XRD patterns and laser Raman spectra
(c) Using aldehydes and ketones. (i) VPO method:6 V O
2
5
(1.0 g. Strem), H PO (2.8 g, 85%, Aldrich) (P/V molar
3
4
ratio \ 4.5) and an aldehyde/ketone (40 ml, Aldrich) were
reÑuxed for 16 h. The resultant pale blue solid (yield 85.95%
based on V) was recovered by vacuum Ðltration, washed with
the aldehyde/ketone (50 ml) and acetone (50 ml) and dried
in air (110 ¡C, 24 h). (ii) VPD method:6 VOPO É 2H O
2
4 2
are shown for a range of carbonyl compounds in Fig. 1 and 2.
Previously,14h17 VO(H PO ) had only been formed exclu-
2
4 2
sively in VPD-type preparations6 with VOPO É 2H O using
4
2
octan-3-ol as the reducing agent, and the XRD pattern
obtained for VO(H PO ) from the aldehydes and ketones
4
2
(1.0 g), H PO (0.05 g, 85%, Aldrich) (mole ratio
3
4
2
4 2
VOPO É 2H O : H PO \ 8 : 1) and the aldehyde/ketone (40
ml, Aldrich) were reÑuxed for 16 h. The solid was recovered by
vacuum Ðltration, washed with the aldehyde/ketone (50 ml)
and acetone (50 ml) and dried in air (110 ¡C, 24 h).
was identical with that prepared from VOPO É 2H O with
4
2
3
4
4
2
octan-3-ol14h17 or previously reported by Bordes.2 In addi-
tion, the Raman spectra were very similar to that observed
previously by Guliants et al.20 (1151 cm~1, br, m; 935 cm~1,
vs; 900 cm~1, sh, m; 575 cm~1, m). Scanning electron micros-
copy (SEM) and TEM micrographs for VO(H PO ) pre-
(d) Using excess H PO . (i) VPO method:6 V O (1.0 g,
3
4
2 5
Strem) and H PO (10.0 g, 85%, Aldrich) (P/V molar
3
4
2
4 2
ratio \ 15) and the alcohol (40 ml, Aldrich) were reÑuxed for
16 h. The solid (yield 85.95% based on V) was recovered by
vacuum Ðltration, washed with the alcohol (100 ml) and
acetone (100 ml) and dried in air (110 ¡C, 24 h). (ii) VPD
method:4 VOPO É 2H O (1.0 g), H PO (10.0 g, 85%,
Aldrich) and the alcohol (40 ml, Aldrich) were reÑuxed for 16
h. The solid (yield 40È45% based on V) was recovered by
vacuum Ðltration, washed with the alcohol (50 ml) and
acetone (50 ml) and dried in air (110 ¡C, 24 h).
pared using aldehydes and ketones are contrasted with those
from VO(H PO ) prepared from VOPO É 2H O with octan-
2
4 2
4
2
3-ol and are shown in Fig. 3 and 4. The material prepared
from octan-3-ol has a blocky morphology with regular sided
oblong crystallites with the ratio of the sides being 1 : 1 : 2
regardless of the dimensions of the crystal.17 However, it is
clear that the crystallites are not single crystal faces since the
surfaces are marked with many indentations (Fig. 4(a)). In
contrast, VO(H PO ) prepared using aldehydes and ketones
4
2
3
4
2
4 2
(e) Reduction of V OPO phases. VOPO was reÑuxed in
as reducing agents has needle-shaped crystals which, at higher
magniÐcation, do not reveal the presence of surface indenta-
tions (Fig. 4(b)). Hence, the solvent used to prepare
VO(H PO ) can control the morphology of the crystallites.
4
4
isobutanol (40 ml, Aldrich) for 16 h. The solid was recovered
by vacuum Ðltration, washed with isobutanol (50 ml) and
acetone (50 ml) and dried in air (110 ¡C, 16 h).
2
4 2
All samples were stored in sealed containers under desicca-
tion prior to their investigation as catalysts, for which the
materials were pelleted and sieved to give particles (diameter:
200È300 lm).
This is similar to the e†ects observed for alcohols in the prep-
aration of VOHPO É 0.5H O.14
4
2
Apart from the sample produced using butan-2-one, all
samples of VO(H PO ) prepared using aldehydes and ketones
2
4
were well crystalline and all di†raction and spectral features
could be ascribed to VO(H PO ) . For butan-2-one, two
Catalyst testing and characterisation
2
4 2
The oxidaton of n-butane was carried out using a micro-
reactor using a standard volume of catalyst (0.5 ml). n-Butane
and air were fed to the reactor via calibrated mass Ñow con-
trollers to give a feedstock composition of 1.5% n-butane in
air. The products were fed via heated lines to an on-line gas
chromatograph for product analysis. The reactor comprised a
stainless-steel tube with the catalyst held in place by plugs of
additional weak di†raction lines with d-spacings 6.992 and
6.727 A were observed, but these cannot be indexed with other
vanadium phosphate phases. It is, therefore, considered that
this preparation method represents a novel route to high
quality VO(H PO ) . In addition, it is a general method of
2
4 2
preparation since VO(H PO ) can be produced from the
4 2
reaction of V O and H PO with all aldehydes and ketones
2
3
2
5
4
5000
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006

View Article Online
in the range C ÈC investigated to date and only representa-
4
10
tive data are given in the Ðgures.
The VPD preparation method6 was also investigated with
aldehydes and ketones but, in the absence of an acid,
VOPO É 2H O was not reduced by aldehydes and ketones.
4
2
However, on addition of H PO (VOPO É 2H O : H PO \
3
4
4
2
3
4
1 : 0.1 molar ratio), the reduction reaction was observed and,
again, VO(H PO ) was obtained as the exclusive product for
all aldehydes and ketones examined in the range C ÈC
2
4 2
.
4
12
Representative data of VO(H PO ) prepared for the C
2
4 2
8
series are shown in Fig. 5 and 6 and all the di†raction and
spectral data are consistent with the formation of
VO(H PO ) .
2
4 2
VO(H PO ) prepared using excess H PO . Early reports
4 2
2
3
4
of the preparation of VO(H PO ) by Ladwig21 indicated
2
4 2
that the material was synthesised from the reaction of V O
and excess H PO in aqueous HCl. The reaction of V O
with a stoichiometric quantity of H PO in aqueous HCl
2
2
5
5
3
4
3
4
is well known4 to produce VOHPO É 0.5H O, and
4
2
VO(H PO ) is a known impurity formed by this preparation
4 2
method.13 Although the reaction of V O and H PO in an
2
2
5
3
4
alcohol, VPO preparation method6,12,17 or VOPO É 2H O
4
2
with alcohols, VPD preparation method,6,17 has been well
studied, the addition of excess H PO in this preparation
3
4
route has not been explored. The addition of excess H PO in
the VPO and VPD preparation methods again led exclusively
to the formation of VO(H PO ) . The XRD patterns, Raman
3
4
2
4 2
spectra and TEM analysis showed these materials to be iden-
tical to the material prepared from the reduction of
VOPO É 2H O with octan-3-ol17 (Fig. 4(a)). In the prep-
4
2
aration of VOHPO É 0.5H O from the reduction of
Fig. 1 Powder XRD patterns of VPO precursors prepared using
aldehydes and ketones as reducing agents. The pattern is indexed to
the reÑexions for VO(H PO ) .
4
2
VOPO É 2H O, the nature of the alcohol signiÐcantly a†ected
the morphology of the crystals formed.14 However, the prep-
aration of VO(H PO ) from VOPO É 2H O showed no
e†ect of the alcohol on crystal morphology and all materials
were identical.
4
2
2
4 2
2
4 2
4
2
VO(H PO ) prepared from VOPO . Reduction of the syn-
2
4 2
4
thesised ““anhydrous VOPO ÏÏ with a range of alcohols
4
(C ÈC ) was found to provide an additional preparation
method for the synthesis of VO(H PO ) . The powder XRD
4
8
2
4 2
patterns, as shown in Fig. 7 for hexan-1-,2-,3-ol and octan-1-,
2-,3-ol as representative examples, and no additional vana-
dium phosphates are present in the samples. In addition, it is
apparent that the alcohol has no e†ect on the morphology of
the VO(H PO ) crystallites.
2
4 2
As the ““anhydrous VOPO ÏÏ hydrates to VOPO É 2H O on
exposure to moist air, it was decided to investigate the e†ect
4
4
2
of alcohol reduction on b-, c- and d-VOPO . Reduction of b-
4
and d-VOPO with isobutanol did not result in the formation
4
of VO(H PO ) . Reduction of c-VOPO with isobutanol did
2
4 2
4
form VO(H PO ) but together with a signiÐcant amount of
4 2
VOHPO É 0.5H O (Fig. 8). These experiments show that the
2
4
2
method based on ““anhydrous VOPO ÏÏ provides a facile route
4
to pure VO(H PO ) that is not achievable from the other
2
VOPO phases.
4
4 2
n-Butane oxidation studies and post-reaction characterisation
VO(H PO ) samples prepared according to the range of
2
4 2
methods detailed in this paper, were used as catalyst precur-
sors for the oxidation of n-butane. The catalysts were activat-
ed in situ in the reactor with Ñowing n-butane (1.5%) in air at
385È415 ¡C for 72 h. During this time, the conversion of n-
butane gradually increased, but between 72È300 h reaction
time the catalytic performance was unchanged and the results
for the activated catalysts are given in Table 1, with a fresh
catalyst precursor being used for each condition evaluated.
The catalysts derived from VO(H PO ) gave similar results
Fig. 2 Raman spectra of VPO precursors prepared using aldehydes
and ketones as reducing agents.
2
4 2
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006
5001

View Article Online
Fig. 3 SEM micrographs of VO(H PO ) . (a) VOPO É 2H O and octan-3-ol, (b) VPO method with octanal, (c) VPO method with octan-2-one,
2
4 2
4
2
(d) VPO method with octan-3-one.
and no dependence on the preparation route was apparent.
The surface area of all activated catalysts was found to be in
the range 1È3 m2 g~1 and this is similar to that observed
previously,17 and the speciÐc activity ((mol maleic anhydride)
m~2 h~1) is similar to that reported in the previous study.
The maleic anhydride selectivity observed with these catalysts
is typically 20È30%, even at low conversion levels, and the
temperature) for all the samples, is 30 ^ 5%. This is in con-
trast to our earlier study which indicated that no carbon
oxides were formed with a catalyst derived from VO(H PO )
2
4 2
using a di†erential reactor with a powdered catalyst.16,17
Although it is possible that the di†erent reaction testing con-
ditions may contribute to the signiÐcantly lower selectivity, it
should be noted that the present study is very detailed and
involves the investigation of a wide range of VO(H PO )
primary selectivity S (obtained by extrapolating the maleic
0
2
4 2
anhydride selectivity to zero n-butane conversion, the conver-
samples prepared by a number of di†erent methods. The cata-
sion being varied by changing the Ñow rate at constant
Fig. 4 Bright Ðeld transmission electron micrographs for
VO(H PO ) . (a) VOPO É 2H O and octan-3-ol, (b) VPO method
Fig. 5 Powder XRD patterns of VPD precursors prepared using
octanal, octan-2-one and octan-3-one.
2
4 2
4
2
with octan-2-one.
5002 Phys. Chem. Chem. Phys., 2000, 2, 4999È5006

View Article Online
Fig. 6 The Raman spectra of VPD precursors prepared using alde-
hydes and ketones as reducing agents.
lytic performance is in agreement with a number of previous
studies in which VO(H PO ) has been used as a precursor.
2
4 2
Mount and Ra†elson22 studied VO(H PO ) as a catalyst
4 2
precursor prepared by heating V O and H PO in an auto-
2
2
5
3
3
clave at 150 ¡C. Hannour et al.23 prepared VO(H PO ) by
4 2
heating an aqueous solution of V O , H PO and oxalic acid.
2
2
5
3
4
2
Hutchings and Higgins13 showed that VO(H PO ) could be
4 2
obtained from vanadium phosphate catalysts by water extrac-
tion. In all these cases, the catalyst obtained from in situ acti-
vation of VO(H PO ) gave very low selectivities to maleic
2
4 2
anhydride.
Fig. 7 Powder XRD patterns for VO(H PO ) prepared by alcohol
2
4 2
The catalyst structures were examined following reaction
using powder XRD (Fig. 9) and Raman spectroscopy (Fig. 10).
The XRD patterns show a varying degree of crystallinity.
Most patterns comprise three lines which have been observed
previously16,17 and these have been assigned in those studies
reduction of ““anhydrous VOPO ÏÏ.
4
characteristic of VO(H PO ) .20 Gulients et al. have shown
2
4 2
that the Raman spectra of VO(PO ) has a strong band with a
3 2
to VO(PO ) . However, these peaks cannot be deÐnitively
Raman shift of 957 cm~1. None of the spectra (Fig. 10) show a
3 2
assigned to a speciÐc vanadium phosphate. All three di†rac-
pronounced band at 957 cm~1 although most show a
broadening of the 935 cm~1 band in this spectral region.
However, other bands for VO(H PO ) (575, 900, 1151 cm~1)
tion lines can be indexed to VO(H PO ) which may indicate
2
4 2
that the catalysts are mainly disordered and contain non-
transformed VO(H PO ) in variable amounts. The di†rac-
2
4 2
and VO(PO ) (692, 1065 and 1109 cm~1) are absent from
2
4 2
3 2
tion lines at ca. 23¡ and 29¡ can also be indexed to the [220]
these spectra. There are a number of bands in the 950È1150
and [211] reÑexions of VO(PO ) , which can be formed from
cm~1 region which may be due to VOPO phases5 which are
3 2
4
VO(H PO ) by heat treatment.3,21,22 Raman spectra help
commonly found in catalysts derived from VOHPO É 0.5H O
2
4 2
4
2
clarify the assignment. The Raman spectra of catalysts have a
(V2O stretch: 1091 cm~1, a -VOPO ; 1040 and 1096 cm~1,
II
4
strong broad band with a Raman shift of 935 cm~1, which is
c-VOPO ; 1010 cm~1, d-VOPO ).
4
4
Table 1 n-Butane oxidation over catalysts derived from VO(H PO )
2
4 2
Surface area
n-Butane
Selectivity (%)
Preparation
Method
Temperature
/¡C
Ðnal catalyst
/m2 g~1
conversion
(%)
SpeciÐc activity
10~5 mol m~2 h~1
MA
CO
CO
2
VOPO É 2H /octan-3-ol
385
400
415
385
385
385
385
385
385
385
385
385
385
385
385
385
385
0.9
2.2
2.2
1.2
1.3
1.9
0.9
1.7
1.0
3.6
3.3
1.2
2.7
2.5
1.4
3.1
1.7
18.5
21.8
24.7
16.6
21.1
21.6
16.4
28.5
8.5
35.5
34.2
15.5
29.6
28.7
11.5
25.8
13.2
21
20
18
27
25
25
25
22
28
22
21
24
18
23
27
29
23
50
53
55
42
45
47
43
53
37
57
58
47
55
67
44
58
50
29
27
27
31
30
28
32
25
35
21
21
29
27
10
29
13
27
1.15
1.27
1.34
2.53
2.67
1.92
2.98
2.45
1.60
1.45
1.42
2.09
1.32
1.73
1.47
1.63
1.21
4
2
VOPO É 2H /octan-3-ol
4
2
VOPO É 2H /octan-3-ol
4
2
VPO/octanol
VPO/octan-2-one
VPO/octan-3-one
VPO/butanone
VPO/hexan-2-one
VPO/octanal
““anhydrous VOPO ÏÏ/octan-1-ol
4
““anhydrous VOPO ÏÏ/octan-2-ol
4
““anhydrous VOPO ÏÏ/octan-3-ol
4
““anhydrous VOPO ÏÏ/hexan-1-ol
4
““anhydrous VOPO ÏÏ/hexan-2-ol
4
““anhydrous VOPO ÏÏ/hexan-3-ol
4
VPOX/octan-1-ola
VPDX/octan-1-ola
a The superscript X denotes excess H PO method.
3
4
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006
5003

View Article Online
VO(PO ) . It is possibly the presence of the VOPO phases
3 2
4
that gives rise to the poor maleic anhydride selectivity since it
is known that these materials tend to give very poor selectivity
for n-butane oxidation.18
The catalysts derived from VO(H PO ) were examined
using XPS and the results are summarised in Table 2. The
mean surface P/V ratio was determined to be 2.9 for the cata-
2
4 2
lysts. The charge-corrected V(2p ) binding energy for all
3@2
samples is 518.0 ^ 0.2 eV which is characteristic of V5`,19
whereas the binding energy for V4`, estimated from a plot of
experimental binding energy vs. oxidation state (inset, Fig. 11)
is 516.7 eV. Fig. 11 shows the V(2p)ÈO(1s) region for one of
the catalyst samples with the positions of V5` and V4` indi-
cated. Although the surface comprises vanadium predomi-
nantly in the 5] oxidation state, we cannot completely rule
out a contribution from a small surface concentration of V4`.
The O(1s) spectra exhibit an asymmetry to higher binding
energy which has previously been assigned to the presence of
surface hydroxy groups.17 Deconvolution of instrumental
broadening from the proÐles24 conÐrmed the presence of a
discrete second component, ruling out di†erential charging as
the origin of the apparent asymmetry, and the original data
Fig. 8 Powder XRD pattern of the product of reduction of c-
VOPO with isobutanol, (+) denotes lines attributed to
VOHPO É 0.5H O.
4
4
2
Based on a combination of the powder XRD and Raman
spectroscopy studies, it can be concluded that the activated
catalysts comprise disordered material, together with non-
transformed VO(H PO ) , VOPO phases and possibly
2
4 2
4
Fig. 9 Powder XRD patterns of catalysts after reaction.
Table 2 Photoelectron binding energies and derived atomic ratios for the catalyst samples derived from VO(H PO ) . Binding energies are
2
4 2
charge-corrected by reference to the C 1s peak assumed to have a binding energy of 285.0 eV
Binding energies/eV
Ratios
Catalyst
V(2p
)
O(1s)
P(2p)
OH/O2~
P/V
3@2
VPO : octanan-2-one
VPD : octan-3-ol
VPO : octanol
VPO : octan-3-one
VPO : butanone
VPD:
517.9
517.9
517.9
517.8
518.3
518.2
532.2
532.3
532.3
531.7
532.8
532.6
533.9
534.1
534.1
534.5
534.7
534.5
134.8
134.8
134.9
135.0
135.1
135.3
0.52
0.32
0.41
0.26
0.27
0.21
2.75
2.55
3.05
3.40
2.80
2.95
5004
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006

View Article Online
Fig. 10 Raman spectra of catalysts after reaction.
were, therefore, subjected to a curve-Ðt analysis (Fig. 12).
Unconstrained Ðts to O(1s) spectra from 4 samples identiÐed
two components with binding energies of 532.5 ^ 0.2 eV and
534.4 ^ 0.3 eV, with a constant splitting of 1.85 eV. The ratio
of the intensity of the higher binding energy peak to that of
the lower binding energy peak was determined to be in the
range 0.21È0.52, values which are much lower than that
reported previously (1.45).17 The surface P/V ratio is lower
than that observed in the previous study17 for initial high
maleic anhydride selectivities. Surface enrichment for phos-
phorus is well documented for vanadium phosphates.1 For
Fig. 11 V(2p) ] O(1s) spectral region for the sample corresponding
to the curve-Ðt in Fig. 12. The binding energy scale has been charge-
corrected relative to a C(1s) binding energy of 285.0 eV. The expected
binding energy position for V4` species is estimated from the plot of
Fig. 12 Illustration of the analysis procedure applied to the O(1s)
spectra from the catalysts derived from VO(H PO ) . The raw spec-
2
4 2
trum (a) is subjected to a non-linear background removal procedure
followed by curve-Ðtting (curve (b)), making use of the deconvoluted
proÐle (c) both to conÐrm the existence of 2 spectral components and
to estimate component parameters.
V(2p ) binding energy vs. oxidation state shown inset (data from ref.
3@2
19), and is indicated on the spectrum.
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006
5005

View Article Online
(VO) P O , the surface P/V ratio is typically ca. 2 compared
2 2
with a bulk ratio of 1. In the case of (VO) P O , the excess
ketone and, consequently, it is probable that some
VO(H PO ) will be formed, albeit in trace quantities. Unless
7
2 2
7
2 4 2
this is removed from the precursor, the VO(H PO ) will
surface phosphorus has been proposed to act as a di†usion
barrier that isolates the active sites. For the catalysts prepared
from VO(H PO ) the surface P/V ratio is somewhat higher
2
4 2
transform on activation in butane/air to give a catalyst that is
less selective than the active phases derived from
VOHPO É 0.5H O. This was recognised by Hutchings and
2
4 2
and yet the selectivity is considerably lower. This suggests that
excess surface P does not act as a di†usion barrier. In this
case, it is more likely that the excess phosphorus changes the
surface acidity leading to the preferential adsorption and
further reaction of maleic anhydride.
4
2
Higgins4,13 who demonstrated that a simple water extraction
procedure could be used to remove the water soluble
VO(H PO ) impurity from VOHPO É 0.5H O. The signiÐ-
2
4 2
4
2
cance of the present study is that VO(H PO ) is probably
2
4 2
present in many vanadium phosphate catalyst preparations
and may be responsible for poor catalytic or variable per-
formance.
Comments on the formation of VO(H PO ) using aldehydes
2
4 2
and ketones
Johnson et al.12 showed that, in the preparation of
VOHPO É 0.5H O from the alcohol reduction of V O and
Acknowledgements
We thank the EPSRC for Ðnancial support.
4
2
2 5
H PO , the alcohol (propan-2-ol and butan-2-ol) oxidised to
3
4
the corresponding ketone. Cornaglia et al.25 found that, using
benzyl alcohol and isobutanol in a similar reaction, benz-
aldehyde and benzoic acid were formed as oxidation products.
Use of aldehydes and ketones as reducing agents, in place of
alcohols, would result in di†erent oxidation products, since
both can be oxidised to the corresponding carboxylic acid.
However, the oxidation of a ketone to a carboxylic acid must
involve carbonÈcarbon bond cleavage and it is unlikely that
the reaction conditions used in this study would facilitate such
a reaction. In the present study, no reaction was observed for
a ketone or aldehyde with VOPO É 2H O (i.e. the VPD
References
1
Forum on V anadium Pyrophosphate Catalysts, G. Centi, Catal.
T oday, 1994, 16.
2
3
E. Bordes, Catal. T oday, 1987, 1, 499.
G. Centi, F. TriÐro, G. Busca, J. Ebner and J. Gleaves, Faraday
Discuss., 1989, 87, 215.
4
5
G. J. Hutchings, Appl. Catal., 1992, 72, 1.
G. J. Hutchings, A. Desmartin Chomel, R. Olier and J.C. Volta,
Nature, 1994, 368, 41.
6
C. J. Kiely, A. Burrows, G. J. Hutchings, K. E. Bere, J. C. Volta,
A. Tuel and M. Abon, Faraday Discuss., 1996, 105, 103.
G. Centi, Catal. T oday, 1994, 16, 1.
4
2
method) unless a small amount of H PO was added. ReÑux-
3
4
ing VOPO É 2H O and H PO in water yielded no reaction,
7
8
9
4
2
3
4
which is not unexpected since VOPO É 2H O is prepared in
R. Ebner and M. R. Thompson, Catal. T oday, 1994, 18, 51.
G. W. Coulston, S. R. Bare, H. Kung, K. Birkeland, G. K.
Bethke, R. Harlow, N. Herron and P. L. Lee, Science, 1997, 275,
191.
4
2
aqueous solution with a large excess of H PO and no reac-
3
4
tion is observed. We propose that the involvement of an alde-
hyde or ketone as a reducing agent is via the enol tautomer.
For most aldehydes and ketones only a small proportion of
the enol tautomer is present, but in the presence of a BrÔnsted
acid the equilibrium is more favourable and more enol is
formed. Some carbonyl compounds, e.g. b-diketones, exist
mainly in the enol form. For example, penta-2,4-dione is sta-
bilised in the enol form by forming a conjugated enone and an
intramolecular hydrogen bond between the enol proton and
the remaining carbonyl oxygen.
10 J. T. Gleaves, J. R. Ebner and T. C. Kuechler, Catal. Rev. Sci.
Eng., 1988, 30, 49.
11 M. T. Sananes, A. Tuel, G. J. Hutchings and J. C. Volta, J. Catal.,
1994, 148, 395.
12 J. W. Johnson, D. C. Johnston, A. J. Jacobson and J. F. Brody, J.
Am. Chem. Soc., 1984, 106, 8123.
13 Imperial Chemical Industries, U.K. Pat. 1 601 121, 1981.
14 I. J. Ellison, G. J. Hutchings, M. T. Sananes and J. C. Volta, J.
Chem. Soc., Chem. Commun., 1994, 1093.
15 M. T. Sananes, G. J. Hutchings and J. C. Volta, J. Chem. Soc.,
Chem. Commun., 1995, 243.
16 M. T. Sananes, G. J. Hutchings and J. C. Volta, J. Catal., 1995,
154, 253.
17 M. T. Sananes, I. J. Ellison, S. Sajip, A. Burrows, C. J. Kiely, J. C.
Volta and G. J. Hutchings, J. Chem. Soc., Faraday T rans., 1996,
92, 137.
18 F. Ben Abdelouahab, R. Olier, N. Guilhaume, F. Lefebvre and
J. C. Volta, J. Catal., 1992, 134, 151.
As a demonstration that the ketones and aldehydes operate
as reducing agents in the enol form, VOPO É 2H O was
4
2
19 Practical Surface Analysis, ed. D. Briggs and M. P. Seah, Chi-
chester, 2nd edn., 1990, vol. 1.
reacted with penta-2,4-diene in the absence of added H PO
3
4
and VO(H PO ) was formed as the exclusive product.
The observation that VO(H PO ) can be readily formed
4 2
from V O and H PO when reacted with aldehydes and
20 V. V. Guliants, J. B. Benziger, S. Sundaresan, I. E. Wachs, J. M.
Jehng and J. E. Roberts, Catal. T oday, 1996, 28, 275.
21 G. Ladwig, Z. Chem., 1968, 8, 307.
22 Monsanto, Ger. Pat. 2 727 617, 1977.
23 F. K. Hannour, A. Martin, B. Kubias, B. Lucke, E. Bordes and P.
Courtine, Catal. T oday, 1998, 40, 263.
24 A. F. Carley and R. W. Joyner, J. Electron. Spectrosc., 1979, 16, 1.
25 L. M. Cornaglia, C. A. Sanchez and E. A. Lombardo, Appl. Catal.
A, 1993, 95, 117.
2
4 2
2
2
5
3
4
ketones is of signiÐcance for the preparation of the com-
mercially used hemihydrate precursor, VOHPO É 0.5H O.
4
2
The hemihydrate precursor is typically prepared using V O
2
5
and H PO with an alcohol as a reducing agent. As noted in
an earlier study12 the alcohol is oxidised to an aldehyde and
3
4
5006
Phys. Chem. Chem. Phys., 2000, 2, 4999È5006