In situ vibrational spectroscopic investigation of surface redox process of vanadyl pyrophosphate

Article abstract of DOI:10.1016/S1381-1169(99)00317-9

(VO)₂P₂O₇ is an active phase for selective catalytic oxidation of n-butane to maleic anhydride (MA). In situ Raman investigation of (VO)₂P₂O₇ and IR observation of low temperature adsorption of CO demonstrated that the surface redox process between (VO)₂P₂O₇ and X₁ phase occurred in a reversible way under the flow reaction conditions on the selective oxidation of n-butane when the butane/oxygen ratio in the feed varied. The state of surface became very close to that of X₁ phase and was distinctly different from that of β-VOPO₄. After the reduction of the oxidized (VO)₂P₂O₇, the surface recovered, indicating that the redox cycle between (VO)₂P₂O₇ and X₁ phase also proceeded reversibly on the surface. (VO)₂P₂O₇ showed a high MA selectivity. However, when butene and butadiene were used as the feeds, the selectivity of (VO)₂P₂O₇ was comparable with or less than that of α- or β-VOPO₄.

Full text of DOI:10.1016/S1381-1169(99)00317-9

Journal of Molecular Catalysis A: Chemical 155 Ž2000. 31–41
www.elsevier.comrlocatermolcata
In situ vibrational spectroscopic investigation of surface redox
process of vanadyl pyrophosphate
)
Gaku Koyano , Takaya Saito, Makoto Misono
Department of Applied Chemistry, Graduate School of Engineering, The UniÕersity of Tokyo, 7-3-1 Hongo, Bunkyo-ku,
Tokyo 113-8656, Japan
Received 28 December 1998; accepted 19 February 1999
Abstract
Redox properties of the surface of vanadyl pyrophosphate ŽŽVO. P O . have been investigated mainly by means of in
2
2
7
situ Raman spectroscopy as well as IR measurement of low temperature adsorption of CO. The results of in situ Raman
spectroscopy were consistent with our previous conclusion wG. Koyano, T. Okuhara, M. Misono, J. Am. Chem. Soc., 120
Ž1998. 767.x that the reversible oxidation and reduction between ŽVO. P O and X1 phase of VOPO4 operates on the
2
2
7
surface during the selective oxidation of n-butane with a relatively high selectivity for maleic anhydride, and showed that the
surface X1 phase formed by oxidation was reversibly reduced to the original ŽVO. P O phase. However, b-VOPO was
2
2
7
4
once formed on the surface, the surface and the selectivity was not recovered by the reduction. It was confirmed by the IR
investigation that the change of the surface were essentially reversible for the oxidation and reduction process between
ŽVO. P O and X1 phase, but the oxidation to b-VOPO changed irreversibly the surface. Furthermore, it was indicated
2
2
7
4
that n-butane preferentially interacts with a Lewis acid site than a Brønsted acid site. q 2000 Elsevier Science B.V. All
rights reserved.
Keywords: Vibrational spectroscopy; Surface redox process; Vanadyl pyrophosphate
4
q
5q
1
. Introduction
that the redox cycle between V
and V
phase operate in this reaction w5–10x. On the
Vanadyl pyrophosphate ŽŽVO.2 P2O7. is an
other hand, Centi et al. proposed on the basis of
active phase for selective catalytic oxidation of
n-butane to maleic anhydride abbreviated as
MA and is the main component of industrial
a UV spectroscopic study w1,11x that redox be-
Ž
3
q
4q
tween V
and V
is responsible for the
.
selective oxidation. Recently, Gai et al. sug-
catalysts for this process w1–3x. A Mars van
3q
gested based on TEM observation that V
Krevelen mechanism w4x is presumed to be in-
species formed together with oxygen defects are
responsible for the selective oxidation of n-
butane w12,13x.
volved in the selective oxidation to produce MA
w5x. It has been claimed by some researchers
As for the role of lattice oxygen, Pepera et al.
1
8
)
w5x carried out an isotopic study using O and
Corresponding author. Tel.: q81-3-5841-7271 ext. 7273; fax:
q81-3-5841-7220; e-mail: gaku@appchem.t.u-tokyo.ac.jp
concluded that the exchange between the sur-
1
381-1169r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S1381-1169Ž99.00317-9

3
2
G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
face layer of
Ž
VO.2 P2O7 and O took place
surface few layers are involved in this reaction,
as indicated by the tracer studies w5,8x, the
2
rapidly, but that between the surface and the
solid bulk was very slow. We also showed that
the diffusion of oxygen atom is limited to only a
few surface layers during the reaction of n-
butane with O w8x. With respect to the active
phase of this catalyst there have been quite a
Žsee references in Ref. w10x.
few studies . Mor-
ishige et al. reported that the surface phospho-
rous rich phase is active for MA production
w14x. Matsuura et al. w15x reported b and b
phases, which have structures close to
surface characterization is important for this
catalytic system. Hence, we investigated in this
study the surface by IR observation of CO
adsorption at low temperature. Recently, the
adsorption of CO has been used to measure the
surface properties of various catalysts w20,21x.
1
8
2
CO is a very weak base and its interaction with
the surface sites is very moderate, especially at
low temperature.
Y
We report here the results obtained by using
Ž
VO.2 P O are highly active. However, the true
in situ Raman investigation of
ŽVO.2 P O and
IR observation of low temperature adsorption of
CO. The results indicate that the surface redox
2
7
2 7
active sites are still controversial w1–3,5–16x.
We have studied the redox process of the
surface and bulk of the
Ž
VO.2 P O and con-
process between
ŽVO.2 P O and X phase ac-
2
7
2 7 1
cluded that local redox cycles on the surface,
very similar to the bulk redox between
tually took place in a reversible way under the
flow reaction conditions of the selective oxida-
tion of n-butane when the butaneroxygen ratio
in the feed was varied.
Ž
VO.2 P O and X phase, are responsible for
2 7 1
the selective formation of maleic anhydride, by
using Raman spectroscopy, XPS, XRD, and
5
q
EXAFS w10x. Here, the X phase is the V
1
2
. Experimental
.1. Preparation of VO P O
7
phase we reported previously w17x, and contains
5O –O–V pair sites as those in VO P O .
V
Ž
.
Ž
.
2
2
7
(
)
2
2
2
This phase shows XRD pattern and Raman
spectrum similar to d-VOPO w18x, and the same
Two kinds of precursor, vanadium hydrogen
phosphate hemihydrate VOHPO P 0.5H O
P-1 and P-3, were prepared as reported previ-
4
Y
XRD pattern to b phase w15x. However, most
Ž
.
,
2
4
of the measurements in our previous studies
w9,10x, i.e., Raman spectroscopy, XRD, XPS,
ously w22x. P-1 was prepared by the so-called
aqueous solvent method and P-3 was obtained
from the so-called organic solvent method. The
solids showed the XRD pattern of VOHPO4 P
and EXAFS, were carried out ex situ, so that the
possible changes of the surface before or during
the measurements could not be excluded. To
solve the problem, in situ measurements have
been attempted in the present study.
0.5H O w23x.
2
Vanadyl pyrophosphate ŽŽVO.2 P2O7.
, C-1
In situ Raman spectroscopy is the one of the
powerful methods to study the VPO system
w6,7,9,10,19x. Schrader’s group examined the
and C-3, were obtained from P-1 and P-3, re-
spectively, by the treatment at 823 K in an N2
flow for 5 h. The C-1 and C-3 obtained were
y
1
oxidation process of
Ž
VO
.
P O to b-VOPO
calcined in the stream of dry air
at 733 K for 2 h C-1 and 0.5 h
Ž
90 ml min
C-3 . Then the
.
2
2
7
4
1
8
with O2 w19x. Abdelouahab et al. w6x con-
Ž
.
Ž
.
5
q
cluded that V
phases, g-VOPO , formed on
catalysts were dispersed in water to form slur-
ries, and the slurries were stirred for 3 h at room
temperature. The catalysts were filtered and
washed with water. The structures of C-1 and
C-3 were confirmed by using XRD w22x. The
4
the surface is active for the selective oxidation.
We applied in situ Raman spectroscopy in the
present study. Above mentioned methods of
characterization, such as Raman spectroscopy
and EXAFS, all provide the information of solid
bulk, not limited to the surface. Since only
BET surface areas of C-1 and C-3 were deter-
2
y1
mined to be 10 and 78 m g , respectively.

G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
33
2
.2. Preparation of VOPO4
X phase was synthesized from NH HVPO
6
using the 514.5 nm Ar laser NEC GLS3261J .
Ž .
The sample powders were placed on a glass
sample holder and pre-treated under the same
1
4
conditions as the steady state reaction
low . The power of laser was usually set below
0 mW to avoid the damage of the samples.
Ž
see be-
as described previously w17x. An aqueous solu-
tion of NH H PO
.
Ž
NH H PO ; 0.8 mol, H O;
4
2
4
4
2
4
2
2
6
00 ml
.
containing V O powder
Ž
0.04 mol
.
2
5
was boiled for 0.5 h and then cooled to room
temperature to obtain precipitate, NH HVPO .
The resulting solid was then treated in an O2
4
6
2.4. IR study of low temperature COrn-butane
adsorption
flow at 823 K for 5 h to form X phase. Bulk
1
1
VOPO4
structure of X phase was confirmed
Ž .
The sample disk
0 mg was set in an in situ cell and evacuated
at 773 K for 1 h. Then, it was cooled to ;90
K, and a predetermined amount of CO 6.82=
1.36=10 mol was
Ž
self support, 20 mmf, ca.
2
by XRD w17x. The BET surface area was 28 m
2
.
y
1
g .
b-VOPO was prepared by the calcination of
VOHPO P0.5H O at 873 K in an O flow for
4
Ž
y
6
y4
4
2
2
10 mol
.
or n-butane
Ž
.
1
0 h w17x. The surface area of b-VOPO was
4
introduced. IR spectra were recorded at 90 K
with a JASCO FT-IR 550 spectrometer. Oxida-
tion treatment of the sample was carried out in
2 1
y
about 3.2 m g . XRD pattern for b-VOPO
4
was consistent with the one reported previously
w17x.
the cell by introducing 200 Torr of O at 733 K
2
for 1 h. Reduction treatment was done in the
presence of 10 Torr of n-butane at 733 K for 1
h. Both treatments were given to the sample
after the adsorbed CO was completely elimi-
nated by the evacuation at 733 K.
2
.3. In situ Raman spectroscopy
In order to observe the changes in the struc-
ture of the catalyst under steady-state reaction
conditions, an in situ Raman cell was con-
structed as illustrated in Fig. 1 w24x. With this
2
.5. Characterization of catalysts
cell, the sample can be treated at 823 K under
the reaction gases. The temperature was cali-
brated by measuring the melting points of zinc
and tin. Raman spectra were recorded with a
X-ray diffraction
Ž
XRD
.
patterns were ob-
MAC Sci-
ls1.5405
A . All the phases were confirmed by XRD.
tained with an X-ray diffractometer
Ž
3
ence MXP
using Cu Ka radiation
. Ž
Laser Raman Spectrometer Jasco, NR-1800
Ž .
˚
.
Fig. 1. Schematic illustration of in situ Raman cell. ŽA. Sample, ŽB. heater coil, ŽC. quartz window, ŽD. cooling water, ŽE. thermocouple,
ŽF. heat insulator, and ŽG. objective lens. Arrows indicate the flow of reactant gas.

3
4
G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
Surface areas of the catalysts were measured by
BET method using N2 adsorption with Mi-
cromeritics ASAP-2000, after the samples were
treated in vacuum at 573 K for 1 h.
2
.6. Catalytic oxidation of n-butane
Catalytic oxidation of n-butane was per-
formed in a flow reactor at 733 K. Prior to the
reaction, 70 mg of catalyst was treated in N
2
3
y1
Ž
60 cm min
reaction was performed with the gas composi-
tion of n-butane 1.5% 17%, O2 content in
parentheses . This is for the ordinary reaction
conditions. Then the partial pressure of n-butane
was decreased to 0.75% 18.5% , and to 0.25%
19.5% , all balanced with N . After this series
of experiments, the catalyst was treated in a
.
at 773 K for 1 h. First, the
Ž
.
Ž
.
Ž
.
2
3
y1
flow of air
temperature. Then, the partial pressure of n-
butane was increased again to 0.25% 19.5%,
O content , 0.75% 18.5% , and 1.5% 17%
Ž
20 cm min
.
for 2 h at the same
Ž
Fig. 2. Changes in Raman spectra of ŽVO. P O ŽC-3. under
2
2
7
.
Ž
.
Ž
.
,
2
a the parent C-3. Partial pressure of
Ž .
steady state reaction.
n-butane was decreased from b 1.5% to c 0.75%, d 0.25%,
and then Že. 0%. Then, it was increased from 0% to Žf. 0.25%, Žg.
all balanced with N . Total flow rate of the gas
Ž .
Ž .
Ž .
2
mixture was varied to maintain the conversion
level of n-butane to be ;50% at each stage. In
each stage of the reaction, the reaction was
continued till a steady state was obtained. Prod-
ucts were analyzed by using two gas chro-
0
.75%, and then Žh. 1.5%. All balanced with air. Ži. X1 phase. All
3
spectra were recorded at 733 K. Flow rate: Žb. and Žh.; 20 cm
y1
3
y1
3
y1
min , Žc. and Žg.; 30 cm min , Žd. and Žf. 40 cm min , Že.;
3
y1
39 cm min
.
matographs; one
Ž
Shimadzu GC, GC-8, TCD
4 mmf, 4.5
4 mmf,
Yanaco
.
showed a single peak at 930 cmy
spectrum a , which is assigned to P–O–P
stretching w25x. At room temperature, the P–O–P
1
equipped with a silica gel column
Ž
Ž
Fig. 2,
mm
1
.
m
for CO , and an MS-13X column
Ž
.
2
.
for CO, O , and N and the other
Ž
2
2
y
1
GC, G-2800, FID
.
equipped with a Porapak QS
for MA and n-butane.
stretching mode appears at 923 cm . With an
increase in the temperature, the peak shifted to a
higher frequency. This is probably because the
angle of the P–O–P is variable depending on
the temperature w25x. After 2 h of the steady
column 4 mmf, 4.5 m
Ž
.
3
. Results
.1. Changes in Raman spectra under the oxi-
state reaction, the spectrum showed the single
ŽVO.2 P2O7 Ž .
peak due to Fig. 2b , and the MA
3
dizing and reducing reaction conditions
selectivity was 52.4%. Upon the decrease of the
n-butane content to 0.75%, there was little
The Raman spectra of C-3 changed as shown
in Fig. 2, when the partial pressure of n-butane
was decreased and subsequently increased. The
selectivities of MA at each stage of partial
pressure are summarized in Table 1. Parent C-3
change observed either in the spectrum
or in the selectivity 48.8% . When the partial
pressure of n-butane was set at 0.25%, the
selectivity decreased to 23.4%, though the spec-
trum did not change much. Considering the fact
Fig. 2c
Ž .
Ž
.

G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
35
Table 1
Changes of the selectivity of maleic anhydride with the change of partial pressure of n-butane over ŽVO. P O ŽC-3.
2
2
7
Phasec
Partial pressure of C H Ž%.
Conversion Ž%.
Selectivity Ž%.
Raman spectrum
4
10
1
0
0
0
0
1
.5a,d
53.8
52.0
56.8
56.0
51.3
45.5
52.4
48.8
23.4
22.8
44.1
61.7
Fig. 2b
Fig. 2c
Fig. 2d
Fig. 2f
Fig. 2g
Fig. 2h
P
P
a,e
.75
.25
.25
.75
a,f
b,f
b,e
P ŽqX.
PqX
P
P
b,d
.5
a
Partial pressure of butane ŽO2 content. was decreased stepwise from 1.5 Ž17.% to 0.75 Ž18.5.% and then 0.25 Ž19.5.% after N2
treatment of the catalyst at 773 K, balanced by N .
2
b
Partial pressure of butane ŽO2 content. was increased stepwise from 0.25 Ž19.5.% to 0.75 Ž18.5.% and 1.5 Ž17.% after treatment under
air at 733 K, balanced by N .
2
c
P: ŽVO. P O , X: X .
2
2
7
1
d
e
f
3
y1
Flow rate: 20 cm min
.
.
.
3
y1
y1
Flow rate: 30 cm min
3
Flow rate: 40 cm min
that new peaks due to X1 phase clearly ap-
peared at 1016 and 1084 cmy1 by the subse-
coincident with the peaks of b-VOPO4
trum i, Fig. 3 . As the partial pressure of n-
Ž
spec-
.
quent oxidation by air
Ž
Fig. 2i
.
, it is probable
butane increased after the oxidation by air, the
y
1
that a very weak band existed at 1016 cm in
spectrum d although the peak was not evident.
Slight differences in the positions and the rela-
tive intensities between these peaks and those
observed at 298 K by ex situ measurements w10x
peaks corresponding to X phase and b-VOPO
1
4
decreased
than
trum i
state conditions still remained low
Ž
Fig. 3g and h
.
, and no peak other
Ž
VO.2 P O was observed
Ž
cf. Fig. 3, spec-
. However, the selectivity at the steady
29.3%
2
7
.
Ž
.
.
3.2. FT-IR obserÕation of surface oxidation and
reduction by using low temperature CO adsorp-
tion
Fig. 4 shows the IR spectra of CO adsorbed
on VPO catalysts at about 90 K. All the spectra
presented are those after the subtraction of the
spectrum of VPO recorded before CO adsorp-
tion. Four peaks at 2141, 2170, 2179 and 2205
y1
cm were observed in the CO stretching vibra-
tion region Fig. 4a . In the n OH region, a
reverse peak was observed at 3661 cm and a
Ž
.
Ž
.
y
1
y1
new band appeared at 3462 cm . As shown in
spectrum b in Fig. 4, the evacuation at 90 K
resulted in the decrease of the bands at 2141
y
1
y1
and 2170 cm . Simultaneously, the 3462 cm
butane was set at 0.25%, the selectivity de-
creased to 20.9%. After 2 h of air treatment,
new three peaks in addition to the peaks due to
band decreased and a very broad peak remained
at a similar position. When the temperature was
y
1
increased in vacuum
at 2179 cm
Ž
ca. 5 K min
decreased gradually and disap-
peared before the temperature reached 190 K
.
, the band
y
1
X phase were observed at 888, 988, and 1065
1
y
1
cm
Fig. 3, e . Those three new peaks were
Ž .

3
6
G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
Fig. 3. Changes in Raman spectra of ŽVO. P O ŽC-1. under steady state reaction. Ža. shows the C-1. Partial pressure of n-butane was
2
2
7
decreased from Žb. 1.5% to Žc. 0.75%, Žd. 0.25%, and then Že. 0%. Then, it was increased from 0% to Žf. 0.25%, Žg. 0.75%, and then Žh.
3
y1
3
1
.5%. All balanced with air. Ži. b-VOPO . All spectra were recorded at 733 K. Flow rate: Žb. and Žh.; 20 cm min , Žc. and Žg.; 30 cm
4
y1
3
y1
3
y1
min , Žd. and Žf. 40 cm min , Že.; 39 cm min
.
spectra c–e . The broad band at around 3460
.
cmy remained. It is remarkable that spectrum
1
Ž
y
1
y1
cm also disappeared. The band at 2205 cm
b was close to the IR spectrum of CO adsorbed
still remained even after the 190 K evacuation.
on X phase spectrum d . Then the sample was
Ž .
1
Effects of preoxidation and prereduction on
IR spectra of CO adsorbed on
ŽVO.2 P O are
2 7
shown in Fig. 5. As shown in spectrum a, four
evacuated at 733 K for 1 h and reduced by
n-butane at 733 K for 1 h, and CO was intro-
duced at 90 K. The recovery of the peaks at
y
1
peaks in the CO stretching region, a reverse
2205 and 2179 cm was noted
trum a and c . CO adsorbed on b-VOPO shows
two peaks at 2179 and 2170 cm
When CO was introduced after the reduction of
Ž
compare spec-
y
1
peak at 3661 cm
tion and a peak at 3462 cm
vibration region were observed. When
VO.2 P O was preoxidized at 733 K for 1 h by
Ž
decrease upon CO adsorp-
.
4
y
1
y1
.
in the OH
spectrum e .
Ž .
y
1
Ž
b-VOPO , the 2205 cm band did not appear
2
7
4
O , IR spectrum of CO subsequently adsorbed
Ž
spectrum f
The effect of adsorption of n-butane was also
examined by using CO adsorption at 90 K Fig.
, and the changes in the peak area are summa-
.
.
2
at 90 K was very different. The bands at 2205,
y
1
2
179, and 3462 cm disappeared
Ž
spectrum b
.
,
Ž
but a small and broad band at around 3462
6
.

G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
37
Table 2
Changes of the selectivity of maleic anhydride with the change of partial pressure of n-butane over ŽVO. P O ŽC-1.
2
2
7
Phasec
Partial pressure of C H Ž%.
Conversion Ž%.
Selectivity Ž%.
Raman spectrum
4
10
1
0
0
0
0
1
.5a,d
72.1
68.7
74.8
71.7
71.3
71.3
46.7
43.8
20.9
15.8
28.0
29.3
Fig. 3b
Fig. 3c
Fig. 3d
Fig. 3f
Fig. 3g
Fig. 3h
P
a,e
.75
.25
.25
.75
PqX
a,f
b,f
b,e
PqXqb
PqXqb
P
P
b,d
.5
a
Partial pressure of butane ŽO2 content. was decreased stepwise from 1.5 Ž17.% to 0.75 Ž18.5.% and then 0.25 Ž19.5.% after N2
treatment of the catalyst at 773 K, balanced by N .
2
b
Partial pressure of butane ŽO2 content. was increased stepwise from 0.25 Ž19.5.% to 0.75 Ž18.5.% and 1.5 Ž17.% after treatment under
air at 733 K, balanced by N .
2
c
P: ŽVO. P O , X: X , b: b-VOPO .
2
2
7
1
4
d
e
f
3
y1
Flow rate: 20 cm min
Flow rate: 30 cm min
Flow rate: 40 cm min
.
.
.
3
y1
y1
3
y
1
rized in Table 3. When n-butane was introduced
on to
VO.2 P O after evacuation at 733 K for
h, no change was observed in the OH stretch-
ing region spectrum a and several peaks due to
CH vibration appeared in 2800–3000 cm .
Since butane molecules was significantly ad-
sorbed on the wall of the cell, it was difficult to
estimate the amount of n-butane adsorbed on
the sample. Spectrum b shows the results of the
Ž
2
7
1
Ž
.
Fig. 5. IR spectra of CO adsorbed at 90 K on pretreated ŽVO. P O
7
2
2
ŽC-3.. Ža. C-3, Žb. C-3 preoxidized at 773 K for 1 h. Žc. Žb. was
evacuated at room temperature and reduced by n-butane at 733 K
Fig. 4. IR spectra of CO adsorbed on ŽVO. P O ŽC-3. at Ža. 90
2
2
7
K. Then, Ža. was evacuated at Žb.90 K, Žc. 150 K, Žd. 170 K, and
Že. 190 K.
for 1 h. Žd. X1 phase, Že. b-VOPO , and Žf. Že. was evacuated at
room temperature reduced by n-butane at 733 K for 1 h.
4

3
8
G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
information about the surface structure by using
low temperature CO adsorption.
As shown in Fig. 4, four peaks in the CO
stretching region 2205, 2179, 2170, and 2141
Ž
y
1
cm
.
Ž
were observed upon CO adsorption on
fresh
VO.2 P O at 90 K. The peak at 3462
2
7
y
1
cm peak can be assigned from its peak posi-
tion w20,21x to acidic OH group interacting with
y
1
CO, and the reverse peak at 3661 cm to the
original OH group of which the peak shifted to
y
1
3
462 cm
upon CO adsorption. The 2141
y
1
cm band is assigned to physisorbed CO in the
case of zeolites and silica w20,21,26x, so that the
same assignment may be applied in the present
y
1
case. Peaks at 2141, 2170, and 3462 cm
disappeared by evacuation at 90 K, and the
y
1
reverse peak at 3661 cm
also disappeared
simultaneously. Therefore, it is concluded that
when CO adsorbs on the acidic OH group hav-
y
1
ing IR peak at 3661 cm its position shifts to
Fig. 6. IR spectra of Ža. n-butane adsorbed on ŽVO. P O ŽC-3.
at 90 K. Žb. CO was introduced after n-butane adsorption at 90 K.
Žc. IR spectra of CO adsorbed on C-3 at 90 K.
2
2
7
y
1
3462 cm and the n CO band appears at 2170
Ž .
cm . Further evacuation by increasing temper-
y
1
ature decreased both the very broad peak at
y
1
y1
3
460 cm and the peak at 2179 cm , keeping
the ratio nearly constant. Thus, the band at 3460
CO introduction after n-butane adsorption. The
bands at the same wave numbers which were
present in the absence of n-butane were ob-
served, but the intensity of the peak at 2205
y1
cm
may be assigned to another acidic OH
group which shows the band of adsorbed CO at
y
1
2
179 cm . The results in the literature that the
CO adsorbed on acidic proton of zeolites was
y
1
cm was smaller.
y1
observed in the region of 2156–2175 cm
w20,21,26x are consistent with the present as-
signments. It is expected that the position of the
CO band absorbed on more acidic OH group
4
. Discussion
4
.1. IR assignment of CO adsorbed and the
surface structure
Table 3
Peak area of CO adsorbed on ŽVO. P O
Recently, we reported that X phase is re-
2
2
7
1
C4 H10 a
CO after C4 H10 b
COc
sponsible for selective MA formation, and pro-
posed its structure by using electron diffraction,
EXAFS, and Raman spectroscopy w9,10x. How-
y1
2205 cm
177q2170 cmy1
–
–
0.72
4.54
1.64
4.51
2
ever, all these methods give the information of
solid bulk and the actual surface structure can-
not be directly evaluated. Even XPS contains
y6
y4
Temperature; 90 K. CO; 6.82=10
mol.
mol, n-butane; 1.36=10
Data are estimated from Fig. 6.
a
Spectrum a.
Spectrum b.
Spectrum c.
˚
the information of the bulk of more than 20 A
b
c
depth. In this study, we obtained additional

G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
39
1
appears at a higher shifts to a lower frequency
so that the acid strength of OH observed at 2179
under O flow, then a and b-VOPO w33x. X
2
4
phase may be transformed to b-VOPO . a-
4
y
1
cm is presumably stronger than OH at 2170
VOPO was not detected in the present study,
4
y
1
cm .
As for the surface OH group, V–OH and
probably because the amount was small.
For C-3, after the X phase formed by air
1
P–OH are possibly present. Both V–OH and
P–OH show OH stretching band at around 3660
cm w27–29x. It is not possible to assign the
treatment was reduced, the MA selectivity was
recovered. However, once b-VOPO was formed
4
y
1
on the surface by decreasing n-butane pressure
Ž .
Fig. 3e , MA selectivity was not fully recov-
ered by the increase in the n-butane pressure
above two bands unambiguously to either V–OH
or P–OH, but it is to be remarked that Spiel-
bauer et al. w27x reported that the CO adsorbed
even after b-VOPO totally disappeared in Ra-
4
y
1
on P–OH shows the n
Ž
CO
.
band at 2177 cm .
man spectrum
demonstrate that the oxidation of
X phase by O and the reduction of X phase
Ž
Table 2
.
. These results clearly
y
1
Since only the peak at 2205 cm
was ob-
Ž
VO.2 P O to
2
7
served above 190 K without any peaks in the
OH region, the band may correspond to CO
adsorbed on Lewis acid sites w30x. If one consid-
1
2
1
to
under the reaction conditions, though the redox
cycle between
VO.2 P O and b-VOPO is not
ŽVO.2 P O by n-butane take place reversibly
2 7
ers that the oxidation state of V in fresh
Ž
2
7
4
Ž
VO.2 P O observed by XPS was close to q4
reversible. The reason for the relatively low MA
selectivity under the condition of 0.25% of n-
butane partial pressure may be due to the con-
secutive oxidation of MA by the excess oxygen
w10x. This is consistent with the fact that the MA
2
7
w9,31x, the band of CO on Lewis acid site
Ž
2205
may be V PPP CO, as in the case of
V O rSiO w30x.
y
1
4q
cm
.
2
5
2
4
.2. Redox cycle producing maleic anhydride
The surface redox mechanism on
selectivity was low in the reaction between
surface X phase and n-butane when the aver-
1
Ž
VO.2 P O
age oxidation state of the surface was high w10x.
The irreversible change of the surface during
2
7
4
q
5
q
is considered to proceed between V and V
w8–10x, and hence V
surface by the oxidation are thought to be im-
5
q
phases formed on the
the redox between
can be a cause of the deactivation.
ŽVO.2 P O and b-VOPO
2 7 4
portant for the MA formation w5,8–10x, while
Raman spectroscopy is a bulk technique so
that the above mentioned observations may not
exclude the possibility that there was a differ-
ence between the surface and bulk and the
surface behaved differently. In this respect the
low temperature adsorption of CO would pro-
vide useful information about the surface. After
3
q
some researchers claim the importance of V
sites on the surface w1,11–13x.
As shown in Fig. 2, X phase was observed
1
after the O treatment. After the reaction in the
2
presence of n-butane, the surface X phase may
1
exist
clearly observed because the amount was small
. On the other
Fig. 2, spectrum b–d , but it was not
Ž .
the oxidation at 733 K
state of surface became very close to that of X1
phase Fig. 5, spectrum d , and was distinctly
different from that of b-VOPO . After the re-
Fig. 5, spectrum b , the
Ž .
Ž
less than 20% of the surface w10x.
hand, X phase was observed in the case of C-1
Ž
.
1
under the reaction conditions in the presence of
4
n-butane
Ž
Fig. 3, spectrum b–d
.
. This is proba-
duction of the oxidized
spectrum c , the surface recovered
spectrum c with spectrum a in Fig. 4
clearly supports that the redox cycle between
ŽVO.2 P2O7 and X1 phase also proceeds re-
Ž
VO.2 P2O7
Ž
Fig. 5,
compare
. This
bly due to the difference in the amount of
oxidized surface layers as indicated by our pre-
vious study w32x. When C-1 was treated under
.
Ž
.
O flow after the reaction, b-VOPO increased.
2
4
This is in agreement with the results that X1
phase was formed first by the oxidation of C-1
versibly on the surface. On the other hand, the
reduction of b-VOPO did not form any de-
4

4
0
G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
tectable amount of Lewis acid site which was
observed with Fig. 5, spectrum e
VO.2 P2O7
5, spectrum f
in n-butane oxidation, Lewis acid site
.
, which showed a low selectivity
4
q
Ž
Ž
.
.
Ž
V
.
Therefore, it supports the conclusion that once
must be important in controlling the selectivity.
b-VOPO4 is formed on the surface of
Ž
VO.2 P O , the surface will not be recovered to
2 7
that of
Ž
VO.2 P O .
5. Conclusion
2
7
We have demonstrated by using techniques
for bulk and surface characterization that the
4
.3. Reaction on the surface
Selective oxidation of n-butane to MA is
reversible redox between
phase produces MA selectively. Redox between
ŽVO.2 P O and b-VOPO is not reversible and
ŽVO.2 P2O7 and X1
considered to proceed consecutively as follows:
2
7
4
butane ™ butene ™ butadiene ™ furan ™ MA
once b-VOPO is formed, the surface and the
4
VO P O
w1x. In the case of butane oxidation, Ž .
2 2 7
MA selectivity irreversibly change. IR observa-
tion of CO adsorbed at low temperature indi-
cated that n-butane preferably adsorbed at Lewis
showed a high MA selectivity, but when butene
and butadiene was used as the feed, the selectiv-
4
q
ity of
Ž
VO.2 P O was comparable with or less
acid site
there.
Ž
probably V
.
, and was activated
2
7
than that of a- or b-VOPO w34x. Thus, the
4
activation of butane, dehydrogenation, is the
important step to control the selectivity. Pepera
et al. w5x concluded from the results of isotope
References
effect by using deuterated butanes that dehydro-
genation at methylene position is the rate-de-
termining step. Busca et al. w35x reported based
w1x G. Centi, F. Trifiro, J.R. Ebner, V.M. Franchetti, Chem. Rev.
8 Ž1988. 55.
w2x B.K. Hodnett, Catal. Rev. Sci. Eng. 27 Ž1985. 373.
w3x G. Centi, Catal. Today 16 Ž1993. 5.
w4x P. Mars, D.W. van Krevelen, Chem. Eng. Sci. 3 Ž1995. 41.
w5x M. Pepera, J.L. Callahan, M.J. Desmond, E.C. Milberger,
P.R. Blum, N.J. Bremer, J. Am. Chem. Soc. 107 Ž1985.
8
on the IR observation of adsorbed acetonitrile
4
q
that V
site is a relatively strong Lewis acid
site and presumed that the Lewis acid site sub-
tracts methylene hydrogen of n-butane, Ai w36x
together with Puttock and Rochester w37x also
reported a correlation between acidity and reac-
tivity.
Fig. 6 shows the effect of butane adsorption
on
ŽVO.2 P2O7 and Table 3 summarizes the
4
883.
w6x F.B. Abdelouahab, R. Olier, N. Guilhaume, F. Lefebvre, J.C.
Volta, J. Catal. 134 Ž1992. 151.
w7x G.H. Hutchings, A.D. Chomel, R. Olier, J.C. Volta, Nature
3
68 Ž1994. 41.
8 M. Misono, K. Miyamoto, K. Tsuji, T. Goto, N. Mizuno, T.
w x
Okuhara, in: G. Centi, F. Trifiro ŽEds.., Stud. Surf. Sci.
Catal., New Developments in Selective Oxidation, Elsevier,
Amsterdam, 1990, p. 605.
change of the peak area accompanied. After the
introduction of n-butane, no change was ob-
served in the OH region, showing that butane
did not interact with Brønsted acid site. Peak
w9x G. Koyano, T. Okuhara, M. Misono, Catal. Lett. 32 Ž1995.
205.
w10x G. Koyano, T. Okuhara, M. Misono, J. Am. Chem. Soc. 120
Ž1998. 767.
y
1
areas of the bands at 2179 and 2170 cm of
VO.2 P O with and without butane were al-
w11x F. Cavani, F. Trifiro, Chemtech 24 Ž1994. 18.
Ž
w
2 P.L. Gai, K. Kourtakis, Science 267 1995 661.
x Ž .
1
2
7
w13x P.L. Gai, Acta Cryst. B 53 Ž1997. 346.
w14x H. Morishige, J. Tamaki, N. Miura, N. Yamazoe, Chem.
Lett. Ž1990. 1513.
most the same. On the other hand, the peak area
of the CO on the Lewis acid site was decreased
after butane adsorption
Ž
spectrum b in Fig. 6
.
.
w15x I. Matsuura, A. Mori, M. Yamazaki, Chem. Lett.
Ž
1987
.
These results indicate that butane preferably
interacts with Lewis acid site rather than with
Brønsted acid site. Since Lewis acid site was
1897.
w16x D. Ye, A. Satsuma, A. Hattori, T. Hattori, Y. Murakami,
Catal. Today 16 Ž1993. 113.
w17x T. Shimoda, T. Okuhara, M. Misono, Bull. Chem. Soc. Jpn.
58 Ž1985. 2163.
not formed by the reduction of b-VOPO4 Fig.
Ž

G. Koyano et al.rJournal of Molecular Catalysis A: Chemical 155 (2000) 31–41
41
w18x E. Bordes, P. Courtine, J. Chem. Soc. Chem. Commun.,
Ž1985. 294.
w19x T.P. Moser, G.L. Schrader, J. Catal. 128 Ž1991. 113.
w20x J.A. Lercher, C. Grundling, G. Eder-Mirth, Catal. Today 27
Ž1996. 353.
w28x H. Kamata, K. Takahashi, C.U.I. Odenbrand, Catal. Lett. 53
Ž1998. 65.
w29x G. Busuca, G. Centi, F. Trifiro, V. Lorenzelli, J. Phys.
Chem. 90 Ž1986. 1337.
w30x A.A. Davydov, A.A. Budneva, N.G. Maksimov, React. Kinet.
Catal. Lett. 20 Ž1982. 93.
w21x H. Kn o¨ zinger, in: R.W. Joyner, R.A. van Santen ŽEds..,
Elementary Reaction Steps in Heterogeneous Catalysis,
Kluwer Academic Publisher, Netherlands, 1993, p. 267.
w22x H. Igarashi, K. Tsuji, T. Okuhara, M. Misono, J. Phys.
Chem. 97 Ž1993. 7065.
w31x P. Delich e` re, K.E. B e´ r e´ , M. Abon, Appl. Catal. A: General
172 Ž1998. 295.
w32x G. Koyano, F. Yamaguchi, T. Okuhara, M. Misono, Catal.
Lett. 41 Ž1996. 149.
w23x J.W. Johnson, D.C. Johnston, A.J. Jacobson, J.F. Brody, J.
w33x G. Koyano, Thesis, The Univ. of Tokyo, 1997.
w34x K. Miyamoto, N. Mizuno, T. Nitadori, T. Okuhara, M.
Misono, Chem. Lett. Ž1988. 303.
Am. Chem Soc. 106 Ž1984. 8123.
w24x T. Saito, G. Koyano, M. Misono, Chem. Lett. Ž1997. 415.
w25x T.P. Moser, G.L. Schrader, J. Catal. 104 Ž1987. 99.
w26x A. Zecchina, S. Bordiga, G. Spoto, D. Scarano, G. Petrini, G.
Leofanti, M. Padovan, C.O. Are a´ n, J. Chem. Soc. Faraday
Trans. 88 Ž1992. 2959.
w27x D. Spielbauer, G.A.H. Mekhemer, T. Riemer, M.I. Zaki, H.
Kn o¨ zinger, J. Phys. Chem. B 101 Ž1997. 4681.
w35x G. Busca, G. Centi, F. Trifiro, V. Lorenzelli, J. Phys. Chem.
90 Ž1986. 1337.
w36x M. Ai, J. Catal. 100 Ž1986. 336.
w37x S.J. Puttock, C.H. Rochester, J. Chem. Soc., Faraday Trans. I
82 Ž1986. 2773.