Structural changes of surface layer of vanadyl pyrophosphate catalysts by oxidation-reduction and their relationships with selective oxidation of n-butane

Article abstract of DOI:10.1021/ja964437b

The surface structure of vanadyl pyrophosphate ((VO)₂P₂O₇) and its changes upon controlled oxidation and reduction have been investigated comprehensively by means of Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), thermogravimetry (TG), EXAFS, X-ray diffraction (XRD), transmission electron diffraction (TED), and 'micropulse' reaction of n-butane. Oxidation of a well-defined (VO)₂P₂O₇ with O₂ (1 atm) at 733 K formed 'X₁ phase' as a thin surface overlayer on (VO)₂P₂O₇, where X₁ phase is a phase reported previously (Shimoda, T.; Okuhara, T.; Misono, M. Bull. Chem. Sec. Jpn. 1985, 58, 2163-2171) and similar to δ-VOPO₄. By the repeated micropulse reactions of n-butane, the surface X₁ phase was gradually reduced back to (VO)₂P₂O₇, showing that reversible redox reactions between X₁ phase and (VO)₂P₂O₇ occur by the reactions with butane and oxygen. XRD, EXAFS, and TED demonstrated that X₁ phase has a structure analogous to (VO)₂P₂O₇, in both of which V-O-V pair sites exist. The micropulse reaction of n-butane with the surface XI phase produced maleic anhydride with a significantly higher selectivity (~ 40%) than that with β-VOPO₄ (<10%). These results indicate that X₁ phase is the real active phase involved in the redox cycle for the oxidation of n-butane to maleic anhydride(VO)₂P₂O₇.

Full text of DOI:10.1021/ja964437b

J. Am. Chem. Soc. 1998, 120, 767-774
767
Structural Changes of Surface Layer of Vanadyl Pyrophosphate
Catalysts by Oxidation-Reduction and Their Relationships with
Selective Oxidation of n-Butane
Gaku Koyano, Toshio Okuhara,† and Makoto Misono*
Contribution from the Department of Applied Chemistry, Graduate School of Engineering, The UniVersity
of Tokyo, Bunkyo-ku, Tokyo 113, Japan
ReceiVed December 27, 1996
Abstract: The surface structure of vanadyl pyrophosphate ((VO)2P2O7) and its changes upon controlled oxidation
and reduction have been investigated comprehensively by means of Raman spectroscopy, X-ray photoelectron
spectroscopy (XPS), thermogravimetry (TG), EXAFS, X-ray diffraction (XRD), transmission electron diffraction
(TED), and “micropulse” reaction of n-butane. Oxidation of a well-defined (VO)2P2O7 with O2 (1 atm) at
733 K formed “X1 phase” as a thin surface overlayer on (VO)2P2O7, where X1 phase is a phase reported
previously (Shimoda, T.; Okuhara, T.; Misono, M. Bull. Chem. Soc. Jpn. 1985, 58, 2163-2171) and similar
to δ-VOPO4. By the repeated micropulse reactions of n-butane, the surface X1 phase was gradually reduced
back to (VO)2P2O7, showing that reversible redox reactions between X1 phase and (VO)2P2O7 occur by the
reactions with butane and oxygen. XRD, EXAFS, and TED demonstrated that X1 phase has a structure
analogous to (VO)2P2O7, in both of which V-O-V pair sites exist. The micropulse reaction of n-butane with
the surface X1 phase produced maleic anhydride with a significantly higher selectivity (∼40%) than that with
â-VOPO4 (<10%). These results indicate that X1 phase is the real active phase involved in the redox cycle
for the oxidation of n-butane to maleic anhydride over (VO)2P2O7.
Introduction
activation treatment in a flow of reactant is indispensable to
obtain high catalytic performance.11
It is generally accepted that (VO)2P2O7 has a characteristic
function for selective catalytic oxidation of n-butane to maleic
anhydride (abbreviated as MA) and that this is the main
component of industrial catalysts for this reaction.1-3 No other
mixed oxides show performance comparable with (VO)2P2O7.
However, the real active component of this catalyst system has
been controversial for many years. Some researchers have
proposed that the single crystalline phase of (VO)2P2O7 itself
is the active phase for this reaction.4-6 Bordes et al.7 and Volta
et al.8 have indicated that a biphasic catalyst consisting of
(VO)2P2O7 and γ-VOPO4 is active. On the contrary, Yamazoe
et al.9 and Hutchings et al.10 inferred that a phosphorus-rich
phase (P/V ∼2) on the surface of (VO)2P2O7 is important for
this reaction. Besides, it has been claimed that a special
To elucidate the mechanism, the oxidation-reduction pro-
cesses of the surface of (VO)2P2O7 are of great concern, as a
Mars van Krevelen mechanism12 is considered to be involved
in the selective oxidation of n-butane to MA.13 It has frequently
been suggested that the redox between V4+ and V5+ phase
operates in this reaction.8,13-15 On the other hand, Trifiro et
al.1,16 indicated from UV spectroscopy that a redox cycle
between V3+ and V4+ is responsible for the selective oxidation.
Recently, Gai et al.17 presumed that V3+ species, which are
located next to the oxygen defect sites, play important roles in
selective oxidation of n-butane.
As for the behavior of oxygen, Pepera et al.13 reported that
the isotopic oxygen exchange between the surface layer of
(VO)2P2O7 and the gas phase took place rapidly, but that
between the surface and bulk was slow. The authors’ group14
also showed by an isotopic study that the diffusion of oxygen
atoms is limited to a few surface layers of (VO)2P2O7 during
the oxidation of n-butane.
* Author to whom correspondence should be addressed. Tel: +81-3-
3812-2111 (ext 7272). Fax: +81-3-5802-2949. E-mail: tmisono@
hongo.ecc.u-tokyo.ac.jp.
† Present address: Graduate School of Environmental Earth Science,
Hokkaido University, Sapporo 060, Japan. Tel: +81-11-706-5280. Fax:
+81-11-757-5995. E-mail: oku@ees.hokudai.ac.jp.
(1) Centi, G.; Trifiro, F.; Ebner, J. R.; Franchetti, V. M. Chem. ReV.
1988, 88, 55.
(10) (a) Sananes, M. T.; Hutchings, G. J.; Volta, J. C. J. Chem. Soc.,
Chem. Commun. 1995, 243, (b) Sananes, M. T.; Hutchings, G. J.; Volta, J.
C. J. Catal. 1995, 154, 253-260.
(2) Hodnett, B. K. Catal. ReV. - Sci. Eng. 1985, 27, 373.
(3) Centi, G. Catal. Today 1993, 16, 5.
(11) For example: Albonetti, S.; Cavani, F.; Trifiro, F.; Venturoli, P.;
Calestani, G.; Granados, M. L.; Fierro, J. L. G. J. Catal. 1996, 160, 52.
(12) Mars, P.; van Krevelen, D. W. Chem. Eng. Sci. 1954, 3, 41.
(13) Pepera, M.; Callahan, J. L.; Desmond, M. J.; Milberger, E. C.; Blum,
P. R.; Bremer, N. J. J. Am. Chem. Soc. 1985, 107, 4883.
(14) Misono, M.; Miyamoto, K.; Tsuji, K.; Goto, T.; Mizuno, N.;
Okuhara, T. In Studies in Surface Science and Catalysis, Vol. 82, New
DeVelopments in SelectiVe Oxidation; Centi, G.; Trifiro, F., Eds.; Elsevier:
Amsterdam, 1990; p 605.
(4) Shimoda, T.; Okuhara, T.; Misono, M. Bull. Chem. Soc. Jpn. 1985,
58, 2163.
(5) Moser, T. P.; Schrader, G. L. J. Catal. 1985, 92, 216.
(6) Busca, G.; Cavern, F.; Centi, G.; Trifiro, F. J. Catal. 1986, 99, 400.
Yang, T. C.; Rao, K. K.; Der Huang, I. US Patent 4,392,986, 1987.
(7) Bordes, E. Catal. Today, 1987, 1, 499.
(8) (a) Abdelouahab, F. B.; Olier, R.; Guilhaume, N.; Lefebvre, F.; Volta,
J. C. J. Catal. 1992, 134, 151, (b) Hutchings, G. H.; Chomel, A. D.; Olier,
R.; Volta, J. C. Nature 1994, 368, 41.
(9) Morishige, H.; Tamaki, J.; Miura, N.; Yamazoe, N. Chem. Lett. 1990,
1513.
(15) Koyano, G.; Okuhara, T.; Misono, M. Catal. Lett. 1995, 32, 205.
(16) Cavani, F.; Trifiro, F. Chemtech 1994, 24, 18.
(17) Gai, L. P.; Kourtakis, K. Science 1995, 267, 661.
S0002-7863(96)04437-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 01/17/1998

768 J. Am. Chem. Soc., Vol. 120, No. 4, 1998
Koyano et al.
off, washed with acetone, and dried at room temperature. The XRD
Therefore, for the understanding of the unique catalysis of
(VO)2P2O7 in the selective oxidation of n-butane to MA, the
elucidation of the structural changes of the surface of (VO)2P2O7
upon redox treatment is the most critical subject. Besides
(VO)2P2O7, there are several V-P-O phases having P/V ) 1.
Their catalytic features may also be informative. As for V5+
phases, â-VOPO4,18 R-VOPO4,19 X1 phase,4 X2 phase,4 γ-VO-
PO4,20 δ-VOPO4,20 â′ phase,21 and â′′ phase21 have been
reported. X1 and â′′ phases are most probably identical, since
the X-ray diffraction (XRD) patterns agreed.4,21 δ-VOPO4
shows the same Raman spectrum8,15 and a similar XRD
pattern4,20 as the X1 phase; these two phases are considered to
have similar structures. Each phase showed different activity
and selectivity for n-butane oxidation;4,22 e.g., R- and â-VOPO4
were much less selective than X1 phase, γ-VOPO4, and
(VO)2P2O7. Here, it must be reminded that the structure
sometimes changes by the reaction.
Structures of â-VOPO4,18 R-VOPO4,19 and (VO)2P2O723 were
determined by single-crystal X-ray diffraction. VO6 octahedra
are isolated in R- and â-VOPO4, while an edge-sharing structure
of two VO6 units (pair sites) is present in (VO)2P2O7. But the
structures of X1 phase (similar to δ-VOPO4) and γ-VOPO4
remained unsolved. Volta et al.8,24 deduced by Raman spec-
troscopy and solid-state NMR that γ- and δ-VOPO4 do not
contain V-O-V pair sites. On the other hand, Matsuura21
proposed on the basis of the powder XRD pattern that â" phase
contains V-O-V pair sites. The authors25 suggested from the
results of EXAFS that the V-O-V pair sites are present in the
structure of X1 phase.
Raman spectroscopy is proven to be useful for the analysis
of the structural changes of VPO phases with the reaction.8,15,26
Schrader et al. examined the oxidation process of (VO)2P2O7
to â-VOPO4 with 18O2.26 Volta and co-workers8 suggested that
γ-VOPO4 formed on the surface of (VO)2P2O7 is active for the
selective oxidation.
In the present study, the changes of the surface structure of
(VO)2P2O7 upon oxidation and reduction have been investigated
by means of various spectroscopic methods and micropulse
reaction of n-butane and/or oxygen. X1 phase, a plausible V5+
phase involved in the redox process, was characterized by XPS,
EXAFS, and TED. The redox processes during the catalytic
oxidation of n-butane over (VO)2P2O7 are discussed in relation
to these structural changes of the surface of (VO)2P2O7.
pattern of this solid was in agreement with that of VOHPO4‚0.5H2O.28
P-4 was obtained by reduction of VOPO4‚2H2O with 2-butanol under
reflux. VOPO4‚2H2O was first prepared as follows: V2O5 powder (0.15
mol) was added to 250 g of the aqueous solution of 85% H3PO4 (H3-
PO4, 2.2 mol). The resulting yellow solution was refluxed at 348 K
for 24 h. The solid obtained was filtered off and washed with water
for several times. The solid gave the same XRD pattern as VOPO4‚-
2H2O.29 The VOPO4‚2H2O (14 g) obtained was reduced with 2-butanol
(150 mL) at 363 K for 18 h to form P-4.
Vanadyl pyrophosphate catalysts ((VO)2P2O7) were obtained from
P-3 and P-4 as follows. These precursors were treated at 823 K in an
N2 flow for 5 h, and then calcined in a flow of air (90 mL‚min-1) at
733 K for 0.5 h (P-3) and 2 h (P-4). Then these were dispersed into
water. The slurries were stirred for 3 h at room temperature, and the
solids were filtered and washed with water. The obtained solids
(catalysts) are denoted by C-3 and C-4, respectively.
Preparation of Various VOPO4. X1 phase was synthesized from
NH4HVPO6 as described previously.4 An aqueous solution of NH4H2-
PO4 (NH4H2PO4, 0.8 mol; H2O, 600 mL) containing V2O5 powder (0.04
mol) was boiled for 0.5 h and then cooled to room temperature to obtain
the precipitate of NH4HVPO6. After filtration, the solid was treated
in an O2 flow at 823 K for 5 h. The resulting solid was confirmed to
be X1 phase by XRD.4 The surface area was 28 m2‚g-1
.
â-VOPO4 was prepared by calcination of VOHPO4‚0.5H2O, which
was obtained by adding V2O5 powder (0.1 mol) and H3PO4 (0.2 mol)
into an aqueous solution of NH2OH‚HCl (0.2 mol in 200 mL) at 873
K in the O2 flow for 10 h as reported previously.4 The surface area
was 3.2 m2‚g-1. The XRD pattern of the solid was in good agreement
with that in the literature.4
Calcination of (NH4)2[(VO)2C2O4(HPO4)2] in an O2 flow at 873 K
for 10 h produced R-VOPO4, where (NH4)2[(VO)2C2O4(HPO4)2] was
prepared from an aqueous solution of oxalic acid (0.30 mol in 400
mL) by adding V2O5 powder (0.1 mol) and NH4H2PO4 (0.2 mol).4 The
surface area was 5.9 m2‚g-1. The XRD pattern of the solid was in
good agreement with that in the literature.4
Oxidation of (VO)2P2O7 with O2. Oxygen treatment was performed
with O2 (1 atm, 60 mL‚min-1) in the range from 733 to 823 K after
(VO)2P2O7 was treated in a flow of He (1 atm, 60 mL‚min-1) at 773
K for 1 h. The degree of the oxidation state was controlled by the
temperature and time period. Uptakes of oxygen were monitored by
the weight increase, which were measured by a microbalance (Seiko
Instruments, TG/DTA 220), in which the sample (∼6 mg) was set in
a Pt pan. The oxidation state is expressed in the following two ways:
5+
x in V4+
V
2xPO4.5+x ()VPO4.5+x) and the number of V5+ layers
1-2x
(abbreviated as NL), where the latter for NL < 1 is the ratio of the
number of V5+ to the number of vanadium atoms (V4+ + V5+) in the
surface monolayer of (VO)2P2O7. The number of vanadium atoms in
the surface monolayer was estimated to be 8.1 × 10-4 and 1.3 × 10-4
mol‚g-1 for C-3 and C-4, respectively, from the surface area and the
surface concentration of V atoms on the (100) plane.23
Experimental Section
Preparation of (VO)2P2O7. Two kinds of a precursor, vanadium
hydrogen phosphate hemihydrate (VOHPO4‚0.5H2O), P-3 and P-4, were
prepared as reported previously.27 P-3 was obtained from the so-called
organic solvent method as follows. V2O5 (0.08 mol) was added to a
mixture of isobutyl alcohol (90 mL) and benzyl alcohol (60 mL), and
the suspension was then refluxed at 378 K for 3 h. After the suspension
was cooled to room temperature, an aqueous solution of H3PO4 (85%
H3PO4, 0.16 mol) was added to the suspension and was again refluxed
at 378 K for 3 h. The resulting greenish light blue solid was filtered
Characterization of Catalysts. Raman spectra were recorded with
a Laser Raman spectrometer (Jasco Corp., NR-1800) using the 514.5
nm line from an Ar ion laser (NEC GLS3261J). Sample powder was
put on a glass sample holder. The power of the laser was usually set
below 20 mW to avoid the destruction of the samples. Raman
sensitivities for X1 phase and (VO)2P2O7 were determined by macro-
scopic measurement of physical mixtures of (VO)2P2O7 (C-3) and X1
phase having different ratios. It was observed that the sensitivity of
X1 phase was about 10 times higher than that of (VO)2P2O7.
(18) Gopal, R.; Calvo, C. J. Solid State Chem. 1972, 5, 432.
(19) Jordan, B.; Calvo, C. Can. J. Chem. 1972, 51, 2621.
(20) Bordes, E.; Courtine, P. J. Chem. Soc., Chem. Commun. 1985, 294.
(21) Matsuura, I.; Mori, A.; Yamazaki, M. Chem. Lett. 1987, 1897.
(22) Lin, Y. Z.; Forissier, M.; Sneeden, R. P.; Vedrine, J. C.; Volta, J.
C. J. Catal. 1994, 145, 256.
(23) Gorbunova, Y. E.; Linde, S. A. SoV. Phys. Dokl. 1979, 24, 138.
(24) Abdelouahab, F. B.; Volta, J. C.; Olier, R. J. Catal. 1994, 48, 334.
(25) Koyano, G.; Yamaguchi, F.; Okuhara, T.; Misono, M. Shokubai
1995, 37, 80.
X-ray photoelectron spectroscopy (XPS) spectra were recorded with
a JEOL JPS-90-SX spectrometer with Al KR radiation. The sample
was pressed into a self-supporting disk and evacuated in the chamber
at room temperature for 12 h to remove water on the sample. To avoid
the reduction of the sample during the measurement by X-ray, the
(28) Johnson, J. W.; Johnston, D. C.; Jacobson, A. J.; Brody, J. F. J.
Am. Chem Soc. 1984, 106, 8123-8128.
(29) R′kha, C.; Vandenborre, M. T.; Livage, J. J. Solid State Chem. 1986,
63, 202-215.
(26) Moser, T. P.; Schrader, G. L. J. Catal. 1991, 128, 113.
(27) Igarashi, H.; Tsuji, K.; Okuhara, T.; Misono, M. J. Phys. Chem.
1993, 97, 7065.

Oxidation-Reduction of Vanadyl Pyrophosphate Catalysts
J. Am. Chem. Soc., Vol. 120, No. 4, 1998 769
spectra of oxidized (VO)2P2O7 were taken in only one accumulation.
The binding energy was referenced to the O1s binding energy of 532.0
eV.
The shape and size of the particles of (VO)2P2O7 (C-3 and C-4)
were measured by scanning electron microscopy (SEM; JEOL TSM-
T20). The samples were suspended in acetone and dispersed on a
covering glass. After the samples were dried at room temperature, Au
was deposited on them.
X-ray diffraction (XRD) patterns were measured with an X-ray
diffractometer (MAC Science MXP3) using Cu KR radiation (λ )
1.5405). EXAFS was measured at National Laboratory for High Energy
Physics, Photon Factory (BL-7C) at 80 K. The sample (10-45 mg)
was mixed with a cellulose powder as a binder and pressed (100
kgf‚cm-2) into disks (10 mm diameter) in dry air. All data were
analyzed with the same method as described elsewhere.30,31
Transmission electron diffraction (TED) was measured with a JEOL
JEM-4000FX II electron microscope with an acceleration voltage of
400 kV. The bulk X1 phase used for the TED measurement was
prepared from plate-like (VO)2P2O7 (C-4) by the oxidation at 823 K
for 5 h.
Figure 1. Time courses of O2 uptake for (VO)2P2O7 (C-3). Flow rate
of O2: 100 mL‚min-1, 1 atm. The oxidation state, x (in the ordinate),
was calculated from the weight increase by O2 uptake. Oxidation
temperature: (a) 733 K, (b) 753 K, (c) 773 K, and (d) 823 K. Dotted
line corresponds to the monolayer oxidation.
Surface area of the catalyst was measured by the BET method using
N2 with Micromeritics ASAP-2000, after the samples were treated in
a vacuum at 573 K for 1 h.
Micropulse Reaction. A micropulse reaction was performed in the
pulsereactor (Pyrex, 10 mm in diameter) connected with a flow system.
At the outlet of the reactor, a quadrupole mass spectrometer (Anelva
AGA-100; acceleration, 30 eV) was directly connected. After the
catalyst was treated in He (60 mL‚min-1) at 773 K for 1 h, it was
calcined in an O2 flow for 2 h at 733 K. The pulse of n-butane (0.71
mL of 2.0% n-butane in He; n-butane ) 6.0 × 10-7 mol) was injected
into the He flow repeatedly at intervals of about 1 min. The flow rate
was precisely controlled by thermal mass flow controllers (Ueshima-
Brooks 5800). A mass spectrometer was connected to a microcomputer
to collect the data in the range m/z ) 26-60 at every 0.3 s. The
composition of MA at the outlet was estimated using the peak intensity
of m/z 54 after the subtraction of the contributions of fragmentation of
n-butane, butene, and butadiene. The compositions of n-butane,
butenes, butadiene, furan, CO, and CO2 were also calculated from the
relative peak intensities of m/z 28, 39, 41, 43, 44, 54, and 68. Sensitivity
factors of these gases for the mass spectrometric analysis were directly
determined by using each gas.
Catalytic Reaction. Catalytic oxidation of n-butane was carried
out in a conventional flow reactor made of Pyrex tubing (inside diameter
10 mm). A mixture of 1.5% n-butane and 17% O2 (N2 balance) was
passed (20 mL‚min-1) over the catalyst bed (200 mg) at 733 K. Since
nearly stationary state was obtained after 2 h, the conversion and
selectivity were measured at 5 h.
Figure 2. Temperature-programmed desorption (TPD) of O2 monitored
by weight loss with a microbalance (A) and rate of O2 desorption (B).
Prior to the TPD, (VO)2P2O7 (C-3) was oxidized at (a) 733 K, (b) 753
K, (c) 773 K, (d) 823 K, (f) 733 K, and (g) 773 K for 2 h. (e) and (h)
show the results of â-VOPO4. Flow rate of He was 100 mL‚min-1
.
T0.5 shows the temperature at which the amounts of desorbed oxygen
correspond to a half monolayer (NL ) 0.5).
Results
The structures of the two (VO)2P2O7 samples, C-3 and C-4,
were confirmed by XRD to be identical with those reported
previously.27 Surface P/V ratios of C-3 and C-4 were about
1.1, which was estimated from the XPS peak intensity according
to the method described previously by the authors.32 The main
difference between C-3 and C-4 is the shape of the crystallites
and hence the BET surface area. C-3 and C-4 have rose-petal-
like and plate-like hexagon structures, respectively.27 The width
of the hexagons determined by SEM micrograph were 1 µm ×
1 µm and 5 µm × 5 µm for C-3 and C-4, respectively.27 The
thickness of the crystallites estimated from the XRD line width
were 8 and 40 nm for C-3 and C-4, respectively.27 BET surface
areas of C-3 and C-4 were 78 and 13 m2‚g-1, respectively. The
(VO)2P2O7 used for the pulse reaction, Raman spectroscopy,
and XPS was C-3, because of its larger surface area. C-4 was
applied to TED analysis for the reason of crystallinity.
Figure 1 shows the time courses of oxygen uptake measured
by the microbalance for (VO)2P2O7 (C-3) at elevated temper-
atures. Prior to the supply of the O2 flow, the sample was
pretreated in the He flow at 733 K for 1 h. The value of x (x
in VPO4.5+x as defined in the Experimental Section) corre-
sponding to monolayer oxidation is shown by a broken line in
Figure 1. The uptake of O2 was slow at 733 K; the oxidation
was less than the surface monolayer after 2 h (a in Figure 1).
At 753 and 773 K, the degrees of the oxidation after 2 h were
x ) 0.080 (NL ) 1.6) and x ) 0.25 (NL ) 4.9), respectively.
At 823 K, nearly the whole bulk was oxidized to form VOPO4
in 50 min, the oxygen uptake nearly corresponding to V4+
f
V5+
.
Figure 2 provides the temperature-programmed desorption
of O2 from the oxidized (VO)2P2O7 (C-3) and â-VOPO4. The
weight decrease measured by the microbalance and the rate of
the O2 desorption determined by a mass spectrometer are shown
in A and B of Figure 2. As shown in a-d in Figure 2A, oxygen
atoms held by the catalyst were desorbed up to 1030 K. The
(30) Miyanaga, T.; Fujikawa, T.; Matsubayashi, N.; Fukumoto, T.; Yokoi,
K.; Watanabe, I.; Ikeda, S. Bull. Chem. Soc. Jpn. 1989, 62, 1791.
(31) Matsubayashi, N. Thesis, Osaka University, 1986.
(32) Okuhara, T.; Nakama, T.; Misono, M. Chem. Lett. 1990, 1941.

770 J. Am. Chem. Soc., Vol. 120, No. 4, 1998
Koyano et al.
Figure 3. Changes in the XRD pattern upon oxidation. (VO)2P2O7
(C-3) was treated in an O2 flow (60 mL‚min-1, 1 atm) for 2 h at (a)
300 K (x ) 0, NL ) 0), (b) 713 K (x ) 0.011, NL ) 0.21), (c) 733 K
(x ) 0.047, NL ) 0.94), (d) 753 K (x ) 0.080, NL ) 1.6), (e) 773 K
(x ) 0.25, NL ) 4.9), and (f) 823 K (x ) 0.48, NL ) 9.6). XRD
patterns of VOPO4; (g) X1 phase, (h) R-VOPO4, and (i) â-VOPO4.
Figure 4. Changes in the Raman spectra upon oxidation. Oxidation
state of (VO)2P2O7 (C-3) is (a) x ) 0 (NL ) 0, 300 K), (b) x ) 0.047
(NL ) 0.94, 733 K), (c) x ) 0.24 (NL ) 4.9, 773 K), and (d) after the
pulse reaction of n-butane over oxidized (VO)2P2O7. Amount of
(VO)2P2O7 (C-3): 200 mg (surface V ) 1.3 × 10-5 mol). Pulse size
of n-butane ) 6.0 × 10-7 mol. A total of 30 pulses of n-butane was
introduced at 733 K. Raman spectra of VOPO4: (e) X1 phase, (f)
R-VOPO4, and (g) â-VOPO4.
temperatures at which the amounts of desorbed oxygen cor-
respond to a half monolayer (NL ) 0.5) are denoted by T0.5
and are shown in Figure 2A. The values of T0.5 increased as
the oxidation temperature increased and were lower than that
for â-VOPO4 (T0.5 ) 989 K, e in Figure 2A). In accordance
with the results in Figure 2A, the peak temperature of O2
desorption became higher as the oxidation temperature increased
(Figure 2B).
in Figure 4, b. The size of a pulse was 6.0 × 10-7 mol of
n-butane. The amount of V atoms exposed to the surface was
estimated to be 8.1 × 10-4 mol‚g-1 from BET surface area (78
m2‚g-1) and the lattice parameters.23 One pulse thus corre-
sponds to the reduction of 0.052 layer of V5+ to V4+ (5.2% of
V in the monolayer) if 14 electron reduction (corresponding to
n-butane to MA) of the surface occurred. The peaks due to X1
phase disappeared completely after the 30 pulses, and the sole
peak of (VO)2P2O7 recovered. As will be reported in detail in
the forthcoming paper, essentially the same changes in the
Raman spectra as above were observed by using an in situ
Raman cell at 733 K.
Changes of the bulk structure of (VO)2P2O7 (C-3) upon the
controlled oxidation were monitored by XRD. In Figure 3, the
XRD patterns of (VO)2P2O7 and the oxidized (VO)2P2O7 are
shown. (VO)2P2O7 itself exhibited peaks at 2θ ) 18.4, 22.8,
28.4, 29.9, and 33.6° (Figure 3, a) which are in good agreement
with those in the literature.23,27 When (VO)2P2O7 was oxidized
at 713 K for 2 h, the pattern was not changed (Figure 3, b).
Upon further oxidation at 733 K, a new broad peak at 2θ )
21.6° appeared (Figure 3, c). This peak became clearer at 753
K (Figure 3, d), and above 773 K, became more intense than
that at 22.8° (Figure 3, e and f). The XRD pattern after the
oxidation at 823 K (Figure 3, f) was close to that of X1 phase
(Figure 3, g).4 The XRD pattern of X1 phase was clearly
different from those of R- and â-VOPO4 (Figure 3, h and i).
Changes in the Raman spectra of (VO)2P2O7 (C-3) upon the
oxidation to controlled extents are given in Figure 4. One main
peak was observed at 923 cm-1 for (VO)2P2O7. Oxidation of
(VO)2P2O7 at 733 K gave new peaks at 937, 1020, and 1090
cm-1, together with the peak at 923 cm-1 (Figure 4, spectrum
b). When the sample was oxidized at 773 K, the intensities of
the new peaks (937, 1020, and 1090 cm-1) increased (Figure
4, c). These new peaks were not detected for R- and â-VOPO4
(Figure 4, f and g), but are consistent with those of X1 phase
(Figure 4, e).15 Spectrum d in Figure 4 shows the Raman
spectrum obtained after n-butane was pulsed 30 times to the
sample (0.2 g) which had been used for the experiments shown
The oxidation state of V near the surface was measured by
XPS. (VO)2P2O7 exhibited two peaks at 524.8 and 517.7 eV
due to V4+2p1/2 and V4+2p3/2, respectively.15 In the cases of
V
5+ phases such as X1 phase and â-VOPO4, the peaks appeared
at higher binding energies, 526.4 (V5+2p1/2) and 518.9 eV
(V5+2p3/2). For the samples of (VO)2P2O7 oxidized at 733 K
(x ) 0.047, NL ) 0.94) and 753 K (x ) 0.080, NL ) 1.6), the
peaks due to V5+ and V4+ coexisted.15 As summarized in Table
5+
4+
1, the ratios of the 2p1/2 peak intensity, IV /IV , was 0.27 and
0.57 after the oxidation at 733 and 753 K, respectively. The
ratios calculated from two models, bulk and surface, are also
shown in Table 1. These will be discussed in a later section.
In Figure 5, the conversion of n-butane and selectivity to MA
during the pulse reaction of n-butane with the oxidized
(VO)2P2O7 (C-3) is shown as a function of the degree of the
oxidation, x. The data can be compared to the Raman data in
Figure 4. Before the reaction, (VO)2P2O7 was preoxidized by
O2 at 733 K (NL ) 0.94, x ) 0.047). The first pulse (solid
marks in Figure 5) gave the conversion of 80% with the
selectivity of 30%. As the number of pulses increased, the

Oxidation-Reduction of Vanadyl Pyrophosphate Catalysts
J. Am. Chem. Soc., Vol. 120, No. 4, 1998 771
Figure 6. Conversion of n-butane, selectivity to maleic anhydride,
and change of the number of oxidized surface layers (NL) during the
co-pulse reaction of butane-O2. (VO)2P2O7 (C-3) was pretreated in an
He flow at 733 K for 2 h (x ) 0.0, NL ) 0.0): (O) conversion, (2)
selectivity to MA, (0) number of oxidized surface layers; catalyst
weight, 200 mg. Pulse size: n-butane, 2.6 × 10-7 mol; O2 1.3 × 10-6
Figure 5. Conversion of n-butane and selectivity of maleic anhydride
in the pulse reaction as a function of oxidation state of (VO)2P2O7.
(VO)2P2O7 (C-3) was pretreated in an O2 flow at 733 K for 2 h (x )
0.047, NL ) 0.94): (0) conversion, (O) selectivity to MA, catalyst
weight; 200 mg. Pulse size of n-butane ) 6.0 × 10-7 mol. Flow rate
of He: 60 mL‚min-1. Solid marks show the results of the first pulse.
mol. Flow rate of He: 60 mL‚min-1
.
Table 1. XPS Peak Intensity Ratio of V5+ and V4+ (IV /IV
5+
4+
)
5+
4+
peak intensity ratio IV /IV
calculated
catalyst
(VO)2P2O7
fresh
observed bulk model surface model
x
0
0
0
0.0
oxidized at 733 K
oxidized at 753 K
X1 phase
0.27
0.57
0.10
0.19
0.27
0.50
0.047
0.080
0.5
conversion decreased, while the selectivity did not change much
(30-40%). In the region of low conversions, selectivities were
rather scattered, due to the experimental error. When n-butane
was pulsed to fresh (VO)2P2O7 (NL ) 0, x ) 0), the conversion
was low from the first pulse (conversion; 8%, selectivity; 12%)
and further decreased to 0% (fourth pulse).
To estimate the oxidation state of (VO)2P2O7 during the
reaction, a mixture of n-butane and O2 (n-butane; 2.6 × 10-7
mol, O2; 1.3 × 10-6 mol) was pulsed onto the fresh (VO)2P2O7
(C-3). The O2 uptake in the pulse reaction was determined by
using the amount of oxygen atoms at the inlet (O2(inlet)) and
that in the products at the outlet (O2(outlet), CO2, CO, and MA)
as follows: NL ) 4(O2(inlet) - O2(outlet) - 13/2CO2 - 9/2CO
Figure 7. Fourier transform of V K-edge EXAFS of (a) R-VOPO4,
(b) â-VOPO4, (c) (VO)2P2O7 (C-3), and (d) VOHPO4‚0.5H2O.
7
- /2MA)/Vsurf, where the sum of O2 uptake is expressed by
in these ranges of the Fourier transform. It is noted that the
peak positions are similar to those observed for (VO)2P2O7 and
VOHPO4‚0.5H2O.
NL. Vsurf is the number of V atom exposed to the surface,
estimated from the BET surface area and the lattice parameters.23
As shown in Figure 6, NL increased with the pulse number
and became constant, NL ) 0.18, after about 15 pulses.
Figure 7 shows the Fourier transforms of V K-edge EXAFS
of R-, â-VOPO4, (VO)2P2O7 (C-3), and VOHPO4‚0.5H2O (P-
3) measured as standard samples. (VO)2P2O7 has a main peak
around 1.5-2.2 Å and a weak peak at 3.20 Å (Figure 7, c).
VOHPO4‚0.5H2O gives a pattern similar to (VO)2P2O7 (peaks
at 1.5-2.2 and 3.10 Å) (Figure 7, d). Considering the structures
of these compounds,23,28 the peak around 1.5-2.2 Å is assigned
to VdO and V-O distances, and the peak at 3.2 Å corresponds
to V-V and V-P distances.33,34 It should be noted that R- and
â-VOPO4 (Figure 7, a and b) showed only one peak at 1.5-
1.9 Å, which corresponds to the V-O distance.18,19
Transmission electron diffraction (TED) patterns of the bulk
(VO)2P2O7 (C-4) and X1 phase are shown in Figure 9, where
the bulk X1 phase was prepared from C-4 having a plate-like
structure and was determined to have a single phase by XRD.
When the electron beam was incident to (VO)2P2O7 along [100],
which is vertical to the basal plane, spots were obtained as
shown in Figure 9a. The TED image of (VO)2P2O7 was the
same as that already reported,35,36 and the cell parameters were
determined to be b ) 16.53 Å and c ) 9.53 Å by using the
equation d ) Lλ/r, where d is the d spacing, L is the focus
length from the sample to the photo film at the observation point
in the TED, λ is the wavelength of the electron beam, and r is
the distance between observed spots. The values used for the
calculation were as follows, Lλ ) 2.12 mm‚nm and r for
direction b and c were 5.13 and 4.45 mm, respectively. These
cell parameters are close to those obtained from the single-
Figure 8 provides the Fourier transforms of the bulk X1 phase
analyzed by using different Fourier transform regions. Two
peaks around 1.58-2.28 and at 3.15 Å were always observed
(33) Vlaic, G.; Garmassi, F. J. Catal. 1990, 122, 312.
(34) Centi, G.; Trifiro, F.; Busca, G.; Ebner, J.; Gleaves, J. Faraday
Discuss. Chem. Soc. 1989, 87, 215.
(35) (a) Bordes, E.; Courtine, P. J. Catal. 1979, 57, 236. (b) Bordes, E.;
Courtine, P. J. Solid State Chem. 1984, 55, 270.
(36) Torardi, C. C.; Calabrese, J. C. Inorg. Chem. 1984, 23, 1308.

772 J. Am. Chem. Soc., Vol. 120, No. 4, 1998
Koyano et al.
consistent with the above idea. Therefore, the participation of
V3+ in the redox cycle must not be significant, though it has
been suggested to be significant.1,16,17
When the catalyst was oxidized by O2 at 733 K, V4+ atoms
equivalent in quantity to the monolayer were oxidized to V5+
(Figure 1). The location of V5+ can be estimated from the XPS
data as follows. If the whole bulk of (VO)2P2O7 is oxidized
uniformly (“bulk-oxidation”), the ratio, V5+/V4+, at the surface
should be equal to the bulk V5+/V4+ ratio. In this case, the
ratios of the surface V5+/V4+ ()IV /IV ) are calculated to be
0.10 and 0.19 after the oxidation at 733 and 753 K, respectively
(Table 1). If (VO)2P2O7 is oxidized stepwise from the surface
(“surface oxidation”), the V5+/V4+ on the top surface becomes
5+
4+
5+
4+
infinity at the monolayer oxidation, but the IV /IV ratios
become 0.27 and 0.50 at x ) 0.047 (733 K) and x ) 0.080
(753 K), respectively, as estimated by considering the escape
depth of the photoelectron.38 This estimation is based on the
data of Scofield39 (σV ≈ σV ) 6.37 in units of 13 600 ×
5+
4+
10-28 m2) for the photoionization crosssection of V5+ and V4+
,
the mean free paths of the photoelectron (λ)40 λV
,
) 1.53
5+ 5+
V
Figure 8. Fourier transform of V K-edge EXAFS of the bulk X1 phase.
V K-edge EXAFS was transformed at FT region of κ ) (a) 3.15-
13.15, (b) 5.4-13.4, (c) 5.45-15.45, and (d) 7.4-15.4.
5+ 4+
4+ 4+
nm, λV
,
) 1.54 nm, and λV
,
) 1.48 nm, densities of
V
V
V5+ and V4+ (nV ) 1.2 × 10-5 nm-3, nV ) 1.3 × 10-5
5+
4+
nm-3 21,23 and the thickness of an ideal V5+ phase monolayer
)
(t ) 0.41 nm). The XPS data are significantly greater than
those expected from the bulk oxidation model and much closer
to those of the surface oxidation as shown in Table 1. This
result shows that the oxidation occurred at a top few surface
layers of (VO)2P2O7. This conclusion is consistent with our
previous result obtained by using 18O2, that is, the oxygen
diffusion was limited in a few layers from the surface under
the reaction conditions.14
The oxidation of (VO)2P2O7 (Figure 4, b) brought about new
Raman peaks (937, 1020, and 1090 cm-1) which are assigned
to X1 phase (Figure 4, e),8 but not to R- and â-VOPO4. Since
the surface P/V ratio of (VO)2P2O7 was about unity as
determined by XPS, it is reasonable that the oxidation of
(VO)2P2O7 produced X1 phase (bulk P/V ratio is unity). When
(VO)2P2O7 was oxidized at higher temperatures (Figure 3), the
formation of X1 phase was confirmed by XRD, as well. The
Raman peaks of X1 phase, which was formed near the surface
as discussed above, disappeared by the reaction with n-butane,
and a single peak corresponding to (VO)2P2O7 was recovered
(Figure 4). This fact suggests that the surface X1 phase is the
active phase for the oxidation of n-butane and that a redox cycle
between (VO)2P2O7 and X1 phase is involved in the catalytic
oxidation.
Figure 9. Transmission electron diffraction patterns of (a) (VO)2P2O7
(C-4) and (b) X1 phase derived from (VO)2P2O7 (C-4).
crystal XRD.23 The TED pattern of X1 phase is shown in Figure
9b. When the electron beam was incident to the X1 phase along
the [100] direction of the parent (VO)2P2O7, cell parameters of
X1 phase were determined to be a ) 8.22 Å (r ) 5.16 mm)
and b ) 9.40 Å (r ) 4.51 mm). Since these axes are vertical
to each other and both of them possess mirror symmetry, an
orthorhombic unit cell can be suggested for the X1 phase, and
the space group of X1 phase is considered to be C2V.
Structure of Surface-Oxidized Layers. As for the real
active phase for (VO)2P2O7, Volta et al. claimed from Raman
spectroscopy and solid-state NMR8 that the surface layers are
γ- and δ-VOPO4 (this phase is similar to X1 phase) and these
phases have isolated VO6 octahedra. On the other hand,
Matsuura proposed from the powder XRD21 that â′′ phase
(equivalent to X1 phase) contains V-O-V pair sites. The
present authors suggested by using EXAFS that the structure
of X1 phase resembles that of (VO)2P2O7 which contains
V-O-V pair sites.25
Discussion
Changes of the Surface Structure of (VO)2P2O7 upon
Oxidation and Reduction Treatments. The selective oxidation
of n-butane over (VO)2P2O7 has been considered to proceed
via a redox mechanism at the surface layers.8,13-15 Pepera et
al. reported that the number of surface layers involved in the
redox processes during the oxidation of n-butane is about unity.13
The reaction between n-butane and the oxygen atoms of the
catalyst took place only when (VO)2P2O7 was oxidized.13,15 This
was confirmed in the present study, as well. When the V5+
ions on the surface of the oxidized (VO)2P2O7 were reduced
back to V4+ by the reaction with pulsed n-butane, the conversion
of n-butane decreased to almost 0% (Figure 5). This shows
that V5+ is necessary for n-butane oxidation and the redox cycle
between V4+ and V5+ is involved in the catalytic oxidation.
Shimoda et al.4 reported that the valence of V atom after the
steady-state reaction was slightly higher than 4, which is
When (VO)2P2O7 was oxidized to the degree equivalent to
the oxidation of V4+ to V5+ on the surface monolayer, the
(37) (a) Contractor, R. M.; Bergna, H. E.; Horowitz, H. S.; Blackstone,
C. M.; Malone, B.; Torardi, C. C.; Griffiths, B.; Chowdhry, U.; Slight, A.
W. Catal. Today 1987, 1, 49. (b) Contractor, R. M.; Slight, A. W. Catal.
Today 1987, 1, 587.
(38) Inumaru, K.; Okuhara, T.; Misono, M. J. Phys. Chem. 1991, 95,
4826.
(39) Scofield, J. H. J. Electron Spectrosc. Relat. Phenom. 1976, 8, 129.
(40) Penn, D. R. J. Electron Spectrosc. Relat. Phenom. 1976, 9, 29.

Oxidation-Reduction of Vanadyl Pyrophosphate Catalysts
J. Am. Chem. Soc., Vol. 120, No. 4, 1998 773
Figure 10. Proposed structure of X1 phase.
formation of X1 phase was confirmed by Raman spectroscopy
(Figures 3 and 4), while it was hard to detect by XRD. The
average number of oxidized surface layers was below unity,
showing that the small amount of X1 phase is present on the
surface. This result is consistent with the result of XPS that
the oxidation took place only near the surface. X1 phase is the
new phase which was reported by the authors.4 Matsuura et
al.21 reported the same XRD pattern. A similar phase (δ-
VOPO4) showing nearly identical XRD pattern20 and Raman
spectra8 was reported by Bordes et al. and Volta et al., yet the
structure has not been determined.
determined to be 8.21 Å from the XRD pattern in Figure 3.
These values for the size of the V2O6 pair site unit of X1 phase
(a ) 8.22, b ) 9.40) is slightly smaller than that of (VO)2P2O7
(b/2 ) 8.26, c ) 9.56). This result is consistent with the EXAFS
data; the V-V distance was 3.15 Å in X1 phase, and 3.20 Å in
(VO)2P2O7 (Figure 7). This difference is probable if one
considers that the valence of V is 5+ in X1 phase, and as a
consequence, V-O bond length would become shorter.41
Summarizing the above considerations, the structure of X1
phase may be schematically illustrated as in Figure 10. X1 phase
has a structure related to (VO)2P2O7, containing V-O-V pair
sites but not P2O7.
According to the amount of O2 uptake, the fraction of X1
1
phase was only about /10 that of (VO)2P2O7 phase after the
Relation between Catalytic Performance and Structure
of Surface-Oxidized Layers. If the redox mechanism between
(VO)2P2O7 and X1 phase is admitted, the oxidation state of the
catalyst must be equilibrated at the state where the oxidation
rate of (VO)2P2O7 is equal to the reduction rate of X1 phase.
When the mixture of O2 and n-butane was pulsed repeatedly to
the fresh (VO)2P2O7 (C-3), the oxidation state of (VO)2P2O7
gradually increased as the pulse number increased and was
equilibrated at NL ) 0.18 (Figure 6), which corresponds to the
oxidation of C-3 at 733 K (Figure 1a), but the intensity of the
Raman peak of X1 phase was almost comparable to that of
(VO)2P2O7 (Figure 4a,b). This is most likely due to the
difference in the sensitivity that X1 phase is about 10 times more
sensitive than (VO)2P2O7. The different sensitivity observed
may be explained by the difference in the intensities of scattering
caused by the color of each phase and different mode of the
main peak for each phase: ν(PO4)8 for X1 phase and ν(P-O-
P)5 for (VO)2P2O7.
state that about 20% of the surface V4+ is oxidized to V5+
.
The Raman peak at 923 cm-1 for (VO)2P2O7 (Figure 4, a),
has been assigned to P-O-P stretching vibration.5 Volta and
co-workers8 attributed the peak at 935 cm-1 of δ-VOPO4 (ZX1
phase) to PO4 symmetric stretching vibration. The decrease in
the intensity of 923 cm-1 peak (ν(P-O-P) of (VO)2P2O7) upon
the oxidation (Figure 4) shows the structure change of P2O7
groups of (VO)2P2O7. The change in the structure will be
discussed below in more detail.
EXAFS data in Figures 7 and 8 give additional information
about the local structure of X1 phase. The peak at 3.2 is
characteristic of (VO)2P2O7 and VOHPO4‚0.5H2O and assign-
able to the atomic distance of V-V in V-O-V units of edge-
sharing VO6 octahedra. VO6 octahedra are all isolated in R-
and â-VOPO4. The contribution of V-O-P bonds to the peaks
at 3.2 Å may be small, because the V-O-P angle of corner
sharing VO6 and PO4 is more variable than the V-O-V angle
in edge-sharing VO6 dimers, hence decreasing the EXAFS peak
intensity. Furthermore, lighter P atoms have a lower back-
scattering factor than V atoms. If the 3.2 Å peak is due to
V-O-V bonds, the fact that the peak at 3.15 Å was always
observed for X1 phase (Figures 7 and 8), being independent of
the range of the square window, strongly supports that X1 phase
has the V-O-V pair sites in the structure, as in the case of
(VO)2P2O7.
With the reaction between the surface X1 phase and n-butane,
the Raman peaks corresponding to X1 phase (937, 1020, and
1090 cm-1) disappeared (Figure 4), and MA was produced
concurrently with the selectivity of about 40% (Figure 5), while
the selectivity was extremely low for â-VOPO4 (less than
10%).15 The selectivity to MA over (VO)2P2O7 (C-3) under
steady-state catalytic conditions was about 50-60%. Hence
the selectivity obtained by the present pulse reaction on oxidized
(VO)2P2O7 is slightly lower than that under steady-state catalytic
conditions. A difference is also observed for the selectivity with
â-VOPO4 (∼8% (pulse) vs 20% (under steady-state catalytic
conditions). Therefore, the experimental errors in the present
mass analysis due to the low detectability of MA and the very
small amounts of samples could be the major reason for the
differences in the selectivity between the two methods. If these
reasons are taken into account, it may be inferred that the surface
X1 phase is the real active phase for the catalytic formation of
MA over (VO)2P2O7.
Spectroscopic analysis of the X1 phase revealed that X1 phase
contains a structure similar to that of (VO)2P2O7 as illustrated
in Figure 10. This similarity of the structure between (VO)2P2O7
and X1 phase probably makes the redox cycle facile. The
oxygen of the surface X1 phase desorbed at temperatures lower
than that of â-VOPO4, and the desorption temperature from X1
By using a ) 8.22 Å and b ) 9.40 Å obtained from TED in
Figure 9b and assuming that X1 phase is rhombohedral, c is
(41) Hardcastle, F. D.; Wachs, I. E. J. Phys. Chem. 1991, 95, 5031.

774 J. Am. Chem. Soc., Vol. 120, No. 4, 1998
Koyano et al.
phase tended to decrease as the oxidation temperature decreased
(Figure 2), suggesting that X1 phase formed only on the surface
by oxidation at lower temperature is more reactive, and therefore
that the surface redox cycle between (VO)2P2O7 and X1 phase
is more facile than the redox of the bulk.
(VO)2P2O7 is proposed for X1 phase from EXAFS and TED;
both having edge sharing VO6 pairs. Under the steady-state
reaction conditions, about 20% of the surface of (VO)2P2O7 is
oxidized to V5+, crystalline X1 phase detectable by XRD not
being formed. The redox cycle between the surface X1 phase
and (VO)2P2O7 is very probably responsible for the selective
formation of MA from n-butane.
Conclusion
Raman spectroscopy and pulse reaction with n-butane re-
vealed that X1 phase is involved in the redox mechanism on
the surface of (VO)2P2O7. A structure closely related to
JA964437B