Microstructures of V-P-O catalysts derived from VOHPO4·0. 5H2O of different crystallite sizes

Article abstract of DOI:10.1016/j.molcata.2004.03.054

The influence of crystallite size of precursor VOHPO₄·0. 5H₂O on the microstructure of a resulting catalyst and selective oxidation of n-butane are investigated. Two kinds of VOHPO₄·0. 5H₂O, small crystallites (av. 1 μm × 110 nm) and large crystallites (av. 10 μm × 415 nm), were prepared by intercalation-exfoliation-reduction of VOPO₄·2H₂O in 2-butanol and the direct reduction of VOPO₄·2H₂O with 2-butanol, respectively. The small VOHPO₄·0.5H ₂O crystallites transformed into single-phase (VO)₂P ₂O₇ of high specific surface area under the reaction conditions of 1.5% n-butane, 17% O₂, and He (balance) at 663 K. In contrast, the catalyst formed from the large VOHPO₄·0.5H ₂O crystallites assumed the form of particles having a double-layered structure, consisting of peripheral (VO)₂P₂O₇ and internal α_II-VOPO₄. As compared with the large (VO)₂P₂O₇, the small (VO)₂P ₂O₇ showed high selectivity to maleic anhydride (～78% at 663 K) and higher catalytic activity. Meanwhile, the catalyst with the double-layered structure exhibited moderate selectivity (～70%). Since (VO)₂P₂O₇ is considered to be a selective phase for the selective oxidation of n-butane, the single-phase of (VO) ₂P₂O₇ derived from the small-sized VOHPO ₄·0.5H₂O crystallites is considered a reason for the high selectivity to maleic anhydride.

Full text of DOI:10.1016/j.molcata.2004.03.054

Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
Microstructures of V-P-O catalysts derived from VOHPO4·0.5H2O
of different crystallite sizes
Yuichi Kamiyaa, Norihito Hiyoshib, Naonori Ryumonb, Toshio Okuharab,∗
a
Japan Science and Technology Corporation, 4-1-8 Honcho, Kawaguchi 332-0012, Japan
b
Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060-0810, Japan
Received 16 December 2003; accepted 3 March 2004
Abstract
The influence of crystallite size of precursor VOHPO4·0.5H2O on the microstructure of a resulting catalyst and selective oxidation of
n-butane are investigated. Two kinds of VOHPO4·0.5H2O, small crystallites (av. 1 ␮m × 110 nm) and large crystallites (av. 10 ␮m × 415 nm),
were prepared by intercalation-exfoliation-reduction of VOPO4·2H2O in 2-butanol and the direct reduction of VOPO4·2H2O with 2-butanol,
respectively. The small VOHPO4·0.5H2O crystallites transformed into single-phase (VO)2P2O7 of high specific surface area under the reaction
conditions of 1.5% n-butane, 17% O2, and He (balance) at 663 K. In contrast, the catalyst formed from the large VOHPO4·0.5H2O crystallites
assumed the form of particles having a double-layered structure, consisting of peripheral (VO)2P2O7 and internal ␣II-VOPO4. As compared
with the large (VO)2P2O7, the small (VO)2P2O7 showed high selectivity to maleic anhydride (∼78% at 663 K) and higher catalytic activity.
Meanwhile, the catalyst with the double-layered structure exhibited moderate selectivity (∼70%). Since (VO)2P2O7 is considered to be
a selective phase for the selective oxidation of n-butane, the single-phase of (VO)2P2O7 derived from the small-sized VOHPO4·0.5H2O
crystallites is considered a reason for the high selectivity to maleic anhydride.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Crystallite size; Microstructure; VOHPO4·0.5H2O; n-Butane oxidation; Maleic anhydride
(n-C4H10/O2/He=1.6/18/80.4, VSHV=1500 h−1) at 673 K.
1. Introduction
Hutchings et al. [6] observed the transformation under the
Vanadium phosphate catalysts have been extensively stud-
ied for selective oxidation of n-butane to maleic anhydride
(MA). Vanadyl pyrophosphate, (VO)2P2O7, is a main ac-
tive component, and is usually prepared from a precursor
VOHPO4·0.5H2O [1–4]. It has been considered that the
transformation from VOHPO4·0.5H2O to (VO)2P2O7 pro-
ceeds through complex processes with dehydration, oxida-
tion, and reduction in the reaction conditions [5–8], whereas
the transformation proceeds topotactically in an inert atmo-
sphere such as He and N2 [9,10].
flow of 1.5% n-butane in air using in situ Raman spec-
troscopy. During the temperature rise from 298 to 667 K,
the intense peaks due to VOHPO4·0.5H2O completely dis-
appeared at 643 K, and weak peaks attributed to (VO)2P2O7
as well as ␣II-, ␥-, and ␦-VOPO4 appeared instead. Then,
treatment at 667 K (final temperature) for 20 h brought
about sharp, intense peaks attributable to (VO)2P2O7.
These observations indicate that the transformation from
VOHPO4·0.5H2O to the resulting catalyst under the reac-
tion gas proceeds through not only a simple dehydration
process, but also complex reactions with oxidation and
reduction of the catalyst.
Kiely et al. [7] revealed that the topotactic dehydration
to (VO)2P2O7 occurs preferentially at the periphery of the
crystallites of VOHPO4·0.5H2O, whereas VOHPO4·0.5H2O
was initially oxidized to ␦-VOPO4 and then reduced to
(VO)2P2O7 in the interior of the crystallites for small size
precursor crystallites (1–2 ␮m long and 0.03–0.1 ␮m thick).
Abon et al. [5] demonstrated that a portion of VOHPO4·
0.5H2O transformed into (VO)2P2O7 through a topotactic
dehydration and was partially oxidized into ␦-VOPO4, fol-
lowed by reduction to (VO)2P2O7 under the reaction gas
∗
Corresponding author. Tel.: +81-11-706-4513;
Fax: +81-11-706-4513.
E-mail address: oku@ees.hokudai.ac.jp (T. Okuhara).
1381-1169/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2004.03.054

104
Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
15 h in a Teflon vessel (45 cm3) equipped with a stainless
steel jacket. The resulting solid was separated by filtration,
washed with acetone, and dried at room temperature.
They further demonstrated that variations in the preparation
procedure of VOHPO4·0.5H2O lead to the catalysts with
very different microstructure; (VO)2P2O7 and some crys-
talline VOPO4 (V5+) phases are contained in various rela-
tive proportions [8]. This implies that the crystallite size of
precursor greatly affects the crystalline phase and structure
of the resulting catalyst.
In the present study, we investigate the influence of crys-
tallite size of VOHPO4·0.5H2O on the microstructure of
the catalyst, by employing two typical kinds of precursors
of small-sized (av. 1 ␮m × 110 nm) and large-sized (av.
10 ␮m × 415 nm) crystallites prepared from VOPO4·2H2O.
Changes in the microstructures of the samples during the
transformation from the precursors to the catalysts are
systematically observed by X-ray diffraction, X-ray photo-
electron spectra, redox titration, thermogravimetric analy-
sis, N2 adsorption–desorption isotherms, scanning electron
microscopy, and transmission electron microscopy. Further-
more, their catalytic performances are discussed in relation
to the microstructure of the catalyst particles.
2.2. Characterization
XRD patterns of solid samples were measured by an XRD
diffractometer (Miniflex, Rigaku) with Cu K␣ radiation. The
adsorption–desorption isotherms of nitrogen were measured
at 77 K by an automatic adsorption apparatus (BELSORP
28SA, BEL Japan Inc.) after the precursors and catalysts
were evacuated at 423 and 523 K, respectively. The sur-
face areas and pore size distributions were calculated by
the BET and DH methods, respectively. Thermogravimetric
(TG) analysis was performed by a TG 8200 (Rigaku) in an
He flow (50 cm3 min−1) at 5 K min−1
.
Scanning electron microscope (SEM) images were taken
by a S-2100 (Hitachi). Transmission electron microscope
(TEM) images were taken with a H-800 (Hitachi) operating
at 200 kV. The average oxidation number of V in the sam-
ple bulk was determined by a redox-titration method using
KMnO4 [12].
2. Experimental section
X-ray photoelectron spectra (XPS) were obtained by a
Shimadzu XPS-7000 with Mg K␣ radiation. All spectra were
referenced to the carbon 1 s peak at a binding energy of
284.5 eV. VOPO4·2H2O and (VO)2P2O7 were used as stan-
dard samples for V5+ and V4+, respectively. The standard
(VO)2P2O7 was prepared by thermal treatment of EP(small)
in a flow reactor at 663 K in He flow (20 cm3 min−1) for
5 h. In order to avoid oxidation of the standard (VO)2P2O7
surface with air, the sample was transferred from the reac-
tor to the sample chamber of the XPS apparatus under an
N2 atmosphere. The oxidation number of the surface was
estimated in the same manner as that reported in [5].
2.1. Materials
2.1.1. VOPO4·2H2O
VOPO4·2H2O serving as a raw material for VOHPO4·
0.5H2O was prepared as follows. A mixture of V2O5 (24 g,
Wako Pure Chem. Ind. Ltd.), aqueous 85% H3PO4 (223 g,
Wako Pure Chem. Ind. Ltd.), and H2O (577 cm3) was re-
fluxed for 24 h. The resulting precipitate was separated
by filtration, washed with acetone, and dried at ambient
temperature and pressure. The precipitant was identified as
VOPO4·2H2O, by XRD and IR.
2.1.2. VOHPO4·0.5H2O
2.3. Dynamic transformation of VOHPO4·0.5H2O to
Two kinds of VOHPO4·0.5H2O, with small and large
crystallites, were prepared from the VOPO4·2H2O. Small
crystallites of VOHPO4·0.5H2O (denoted EP(small)) was
prepared by intercalation-exfoliation-reduction process of
VOPO4·2H2O, which was previously reported [11]. A mix-
ture of VOPO4·2H2O (1.0 g) and 2-butanol (50 cm3, Wako
Pure Chem. Ind. Ltd.) was heated stepwise with stirring at
303, 323, 343, and 363 K for 1 h at each temperature to
yield a homogeneous yellow solution as the product of in-
tercalation of 2-butanol into VOPO4·2H2O layers, followed
by exfoliation [11]. After the addition of a small amount
(5 mg) of VOHPO4·0.5H2O to the homogeneous solution,
the solution was refluxed for 20 h (reduction) to form light
blue precipitates, which were separated by centrifugation,
washed with acetone, and dried at room temperature.
VOHPO4·0.5H2O having large-sized crystallites (de-
noted P(large)) were prepared by direct reduction of
VOPO4·2H2O with 2-butanol. A mixture of VOPO4·2H2O
(2.5 g) and 2-butanol (25 cm3) was heated at 423 K for
catalyst and catalytic oxidation of n-butane
Transformation of VOHPO4·0.5H2O to corresponding
catalyst was performed at 663 K in a flow reactor (Pyrex
tube, 10 mm inside diameter) with a mixture of n-butane
(1.5 vol.%), O2 (17 vol.%), and He (balance) under atmo-
spheric pressure. After the powder of VOHPO4·0.5H2O
(0.68 g and 1.38 g for EP(small) and P(large), respectively)
was placed in the reactor, the reactant gas was fed at a rate of
20 cm3 min−1. Temperature was raised from room tempera-
ture to 663 K at a rate of 5 K min−1 and then maintained at
663 K. In order to observe the structural changes during the
transformation, the samples were prepared while varying
the treatment times (0.1, 0.5, 1, 3, 4, 25, and 300 h, which
represent the time after the temperature reached 663 K).
In parallel with the transformation, the gas at the out-
let of the reactor was analyzed by two on-line gas chro-
matographs. For n-butane and MA, an FID-GC (GC-8A,
Shimadzu) equipped with Porapak QS columns (1 m) was

Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
105
used. A high-speed GC (M200, Aera) equipped with Pora-
pak Q and Molecular Sieves 5A columns was utilized for
the analysis of CO, CO2, and O2.
After 200 h, stationary conversion of n-butane and selec-
tivity to MA for the n-butane oxidation was attained (see
results). In this paper, we call the samples after 300 h cata-
lyst. The catalysts from EP(small) and P(large) are denoted
EC(small) and C(large), respectively.
(Fig. 1c), at least two types of crystallytes were identified;
one was oblong microcrystallite (ca. 50 nm) inside the parti-
cle and other was a distinct shell along the periphery whose
thickness from the outer surface was estimated to be about
60 nm. The TEM image of C(large) revealed more complex
microstructure (Fig. 1d). The particle of C(large) contained
irregular shaped platelets, oblong microcrystallites showing
strong contrast, and white circles in its inside. A distinct
shell (ca. 70 nm thick) similar to that found in EP(small)
was also observed in the periphery of C(large).
Fig. 2 shows SEM images of the precursors and catalysts.
EP(small) was of thin and small leaf-like crystallite. The
lateral lengths were distributed from 0.75 to 1.5 ␮m, with
an average length of 1.0 ␮m (Fig. 2a). In contrast, P(large)
showed a large platelet shape with an average lateral length
of 10 ␮m (Fig. 2b). The shape and size of the particles of the
catalysts were basically retained during the transformation,
although, as described above, the interior of the catalyst
particle changed drastically.
Table 1 summarizes the surface areas, lengths and thick-
nesses of particles, and the oxidation numbers of V in bulk
and surface for the precursors and catalysts. Table 1 also lists
the catalytic activity and selectivity to MA for the selective
3. Results
3.1. Crystalline phases and microstructures of precursors
and catalysts
Fig. 1 shows the bright field TEM images of the precursors
and catalysts. The lateral lengths of EP(small) and P(large)
were 1 and 5 ␮m, respectively. In the TEM images of both
precursors, monolithic particles were observed (Fig. 1a and
b). In contrast, mosaic patterns were observed inside the par-
ticles of both catalysts (Fig. 1c and d), indicating that the
particles of the catalysts were polycrystalline; the patterns
differed drastically depending on the catalyst. For EC(small)
Fig. 1. TEM images of the precursors and catalysts: (a) EP(small); (b) P(large); (c) EC(small) and; (d) C(large) (insets in (c) and (d) are magnified images).

106
Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
Fig. 2. SEM images of the precursors and the catalysts: (a) EP(small); (b) P(large); (c) EC(small); and (d) C(large).
oxidation of n-butane. The lateral lengths were estimated
from the SEM images (Fig. 2). For the precursors, estima-
tion of the thickness from the SEM images was difficult. In
addition, Scherrer’s equation could not be applied for the
precursors, because the line-widths of XRD were too small
(less than 0.05 deg. in 2θ) to allow estimation of the thick-
ness. Thus, the thickness of the crystallites (t) of the pre-
Thickness of crystallites of P(large) was four times that of
EP(small). Eq. (1) was not applied for the catalysts, because
the particle of the catalysts consisted of poly crystals.
The oxidation numbers of V in the bulk and surface, which
were estimated by redox titration and XPS, respectively,
were close to 4.0 for both precursors. For the catalysts, the
bulk oxidation number of V for C(large) greatly exceeded
4.0 (V4.55+), whereas that for EC(small) was close to 4.0.
However, a noteworthy finding is that two catalysts exhibited
approximately equal surface oxidation numbers (V4.23+).
This inhomogeneity of the oxidation state of V between the
bulk and surface for C(large) will be discussed in further
detail later.
cursor was estimated by Eq. (1) using a density (ρ; g cm−3
)
which was calculated from the lattice constant and formula
weight of VOHPO4·0.5H2O, surface area (S; m2 g−1), and
lateral length of the crystallites (L; ␮m).
2000
4
+
= Sρ
(1)
t
L
Table 1
Physical and chemical properties of precursors and catalysts and catalytic results of n-butane oxidation at steady states
Precursor and catalyst
Sa(m2 g−1
)
Lb(×103 nm)
tc(nm)
n of Vn+d
Ratee(10−4 mol g−1 h−1
)
Selec. (Conv.)f (%)
Bulk
Surface
EP(small)g
EC(small)h
P(large)g
8
28
2
10
11
0.75–1.5 (1.0)
0.5–1.5 (0.8)
5.0–11.0 (10.0)
4.0–11.0 (8.0)
4.0–11.0 (8.0)
110
4.01
4.08
3.98
4.55
3.99
4.00
4.23
4.08
4.23
4.31
–
–
17 (0.58)
–
4 (0.39)
5 (0.48)
77.5 (50.9)
–
71.7 (43.2)
71.8 (42.8)
415
–
C(large)h
C(large)-Hei
a
Surface area measured after pretreatment at 423 K (precursor) or 523 K (catalyst) in a vacuum.
Lateral length estimated from SEM. The figures in parenthesis denote average length.
Thickness estimated from the surface area and the length.
b
c
d
e
f
Average oxidation number of vanadium.
The figures in parenthesis denote the specific activity per surface area (10−4 mol m−2 h−1).
The figures in parenthesis denote the conversion at which the selectivity was measured.
Precursor (VOHPO4·0.5H2O).
g
h
i
After transformation under the reaction gas at 663 K for ca. 300 h.
After treatment under He flow at 663 K for 4 h.

Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
107
Fig. 3. Changes in XRD patterns of the samples during the transformation from precursors to catalysts: (A) EP(small)–EC(small) system and (B)
P(large)–C(large) system. (a) 0 h; (b) 0.1 h; (c) 0.5 h; (d) 1 h; (e) 2 h; (f) 25 h and; (g) 300 h. The transformation was performed at 663 K with a mixture
of n-butane 1.5%, O2 17% and He (balance). Symbols ᭹, ᮀ, ꢀ, and ᭛ in the figure denote VOHPO4·0.5H2O, (VO)2P2O7, ␦-VOPO4, and ␣II-VOPO4,
respectively.
3.2. Changes in crystalline phases and structures during
transformation from precursor to catalyst
the treatment conditions for planned times (see experimental
section). In Fig. 4A, the percentage of the residual water in
each sample was calculated from the data of TG analysis
with Eq. (2).
The changes in crystalline phases with time during the
transformation from the precursor to the catalysts were
observed by XRD (Fig. 3). Both EP(small) and P(large)
were confirmed to be well-crystallized VOHPO4·0.5H2O
(Fig. 3A(a) and B(a)). For the EP(small)–EC(small)
system (Fig. 3A), intensities of diffraction lines from
VOHPO4·0.5H2O decreased rapidly with time. These
diffraction lines completely disappeared at 1 h, and weak
diffraction lines from (VO)2P2O7 and ␦-VOPO4 phases
appeared instead (Fig. 3A(d)). The diffraction lines from
(VO)2P2O7 became gradually intense with time, whereas
those from ␦-VOPO4 finally disappeared. Note that after
300 h of the treatment (Fig. 3A(g)), only diffraction lines
from (VO)2P2O7 were observed.
The changes in crystalline phases for the P(large)–C(large)
system (Fig. 3B) were drastically different from those
for the EP(small)–EC(small) system. The diffraction lines
from VOHPO4·0.5H2O also disappeared, but the disappear-
ance took more time for P(large) than for EP(small). At
2 h (Fig. 3B(e)), new diffraction lines due to (VO)2P2O7,
␦-VOPO4, and ␣II-VOPO4 became intense. After that
time, the diffraction lines from ␦-VOPO4 gradually weak-
ened, whereas those from ␣II-VOPO4 became intense with
time. The diffraction lines from (VO)2P2O7 were scarcely
changed after 1 h of treatment. Contrary to the case of
EC(small), ␣II-VOPO4 was a main phase for C(large),
whereas (VO)2P2O7 was a minor phase (Fig. 3B(g)).
Fig. 4 presents the changes in percentage of residual water
in sample (Fig. 4A), bulk oxidation number of V (Fig. 4B),
and surface area (Fig. 4C) as a function of treatment time.
Each sample was prepared by treating of the precursor under
Fig. 4. Change in: (A) percent residual water in the sample; (B) average
oxidation number of V and; (C) surface area as functions of treatment
time. The transformation was performed at 663 K with a mixture of
n-butane 1.5%, O2 17%, and He (balance) ((᭺) EP(small)–EC(small)
system and (᭜) P(large)–C(large) system).

108
Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
TWloss − ((ꢀW/W300 K) × 100)
P
water(%)=100 −
× 100
(2)
Here Pwater is the percentage of residual water in the sam-
TWloss
ple, ꢀW is differential weight of the sample between 300 and
773 K as determined by TG analysis, and TWloss (10.5%) is
a theoretical weight loss for the dehydration calculated from
Eq. (3).
2VOHPO4 · 0.5H2O → (VO)2P2O7 + 2H2O ↑
(3)
Since the reaction of Eq. (3) was completed by 723 K for
both EP(small) and P(large) (Fig. 5), we adopted 773 K as
the temperature of the termination of dehydration. For the
EP(small)-EC(small) system, almost all water was released
from the sample at 1 h, whereas complete dehydration of
the P(large)–C(large) system took 2 h, demonstrating that
the degradation of VOHPO4·0.5H2O for P(large) proceeded
more slowly than that for EP(small). This is consistent with
the results of XRD (Fig. 3). As shown in Fig. 5, the dehy-
dration on EP(small) occurred from lower temperature than
that on P(large).
As shown in Fig. 4B, in both systems the oxidation num-
ber increased rapidly with time at the initial stage. In the
case of the EP(small)–EC(small) system, the oxidation num-
ber decreased with time through a maximum at 1 h (V4.34+),
and finally reached V4.08+ at 300 h. In contrast, P(large) was
more deeply oxidized than EP(small), where the maximum
value was V4.58+ (at 2 h), and subsequently, the bulk oxida-
tion state scarcely changed (V4.55+ at 300 h).
Fig. 6. Changes in mesopore size distributions of the samples as a
function of treatment time. (A) EP(small)–EC(small) system and (B)
P(large)–C(large) system: (a) 0 h; (b) 2 h; (c) 25 h and; (d) 300 h. The
transformation was performed at 663 K with a mixture of n-butane 1.5%,
O2 17%, and He (balance).
isotherms. For the EP(small)–EC(small) system (Fig. 6A),
the samples treated within 2 h did not have mesopores,
whereas mesopores with a peak of about 4 nm in diameter
were formed at 25 h (Fig. 6A(c)). For the P(large)–C(large)
system, no peak appeared in the mesopore region within
25 h, whereas the sample for 300 h; i.e. C(large), had meso-
pores with a peak around 4 nm in diameter (Fig. 6B(d)).
3.3. Selective oxidation of n-butane
Fig. 7 shows the time course changes in the selective ox-
idation of n-butane at 663 K, where the catalyst weights are
0.68 and 1.38 g for EP(small) and P(large), respectively. For
the EP(small)–EC(small) system, the conversion gradually
As shown in Fig. 4C, changes in the surface areas for
EP(small)–EC(small) system were small in the initial few
hours, and after 25 h, surface area increased greatly. The
initial and final surface areas were 8.4 and 28.4 m2 g−1, re-
spectively. In the case of the P(large)–C(large) system, the
surface area did not increase by 25 h, and the final value was
10.3 m2 g−1 (at 300 h).
Fig. 6 shows the changes in mesopore size distributions
of the samples with treatment time, where these distribu-
tions were estimated from the desorption branches of N2
Fig. 7. Time course changes in oxidation of n-butane over the catalysts
formed from (᭺) EP(small)) (W/F = 0.85 kg h mol−1) and (᭜) P(large)
(W/F = 1.7 kg h mol−1). The reaction was performed at 663 K with a
mixture of n-butane 1.5%, O2 17%, and He (balance).
Fig. 5. TG curves of EP(small) and P(large) under He flow at heating
rate of 5 K min−1
.

Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
109
from VOHPO4·0.5H2O (∼1 ␮m × 0.1 ␮m) prepared by re-
duction of VOPO4·2H2O. Although two precursors had dif-
ferent chemical and physical properties such as crystallinity
and morphology, there findings made us suppose that the
crystallite size of the precursor affected the microstructure
of the catalyst.
In the present study, we used two typical precursors,
EP(small) and P(large), having different particle sizes; that
is, 1 ␮m (length) × 110 nm (thickness) and 10 ␮m (length)
× 415 nm (thickness). As shown in Fig. 3, EP(small) trans-
formed into pure (VO)2P2O7 phase (EC(small)), whereas
the catalyst (C(large)) containing mainly ␣II-VOPO4 phase
was formed from P(large). Since the two precursors had
almost the same P/V atomic ratio and oxidation states in
the bulk and surface (Table 1), and were well-crystallized
VOHPO4·0.5H2O (Fig. 3), the difference in crystallite
size between two precursors can be considered the cause
of the difference in the crystalline phases formed in the
catalysts.
As listed in Table 1, the bulk oxidation number of V for
EC(small) was nearly 4.0, and the surface oxidation num-
ber slightly exceeded 4.0+, being V4.2+. Since XPS can
be applied only to surface layers, these results indicate that
only the top few surface layers of EC(small) were oxidized
during the steady-state reaction. The selective oxidation of
n-butane over (VO)2P2O7 has been accepted to proceed via
a redox mechanism involving V4+ and V5+ at a few surface
layers [13–15]; thus, the surface layers at the steady-state
reaction would contain the V5+ species to some extent. The
surface oxidation state of EC(small) was consistent with
this concept. In the case of C(large), the surface oxidation
state was similar to that of EC(small), whereas the catalyst
bulk was deeply oxidized (V4.55+). This result indicates that
the oxidized phases such as ␣II-VOPO4 and ␦-VOPO4 for
C(large) existed mainly in the internal particle. Kiely et al.
[7] demonstrated that the dehydration of VOHPO4·0.5H2O
to (VO)2P2O7 proceeded preferentially at the periphery of
the crystallites in the initial stage, and, in contrast, the inter-
nal crystallite of VOHPO4·0.5H2O tended to transform into
␣II-VOPO4. Our present results revealed that the oxidation
of VOHPO4·0.5H2O to ␣II-VOPO4 was almost completely
suppressed when the small size precursor (1 ␮m long and
110 nm thick) was employed. The reason why the large-sized
catalyst particle was more oxidized than the small catalyst
particle will be discussed in the next section.
Fig. 8. Changes in selectivity to maleic anhydride as a function of con-
version of n-butane: (᭺) EC(small)) and (᭜) C(large). The reaction was
performed at 663 K with a mixture of n-butane 1.5%, O2 17%, and He
(balance). The data were collected after 300 h of reaction.
increased until about 100 h, and then decreased slowly to
a stationary value after 200 h. Selectivity to MA increased
rapidly at initial 20 h, and then increased slowly. In contrast,
for the P(large)–C(large) system, the changes in conversion
were small, whereas the selectivity increased with the lapse
of time, and stable selectivity was obtained after 150 h.
Fig. 8 shows the selectivity to MA plotted against the con-
version, in which flow rates were changed after the station-
ary conversions and selectivities were obtained. The selec-
tivity for EC(small) was higher than that for C(large) over
the entire range of the conversions. As shown in Table 1, the
activity of EC(small) was about four times that of C(large),
whereas the specific activity (per unit surface area) was sim-
ilar. The selectivity around 50% conversion was 78 and 72%
for EC(small) and C(large), respectively.
4. Discussion
4.1. Influence of crystallite size of precursors on
crystalline phases and microstructures of the catalysts
The transformation from VOHPO4·0.5H2O to (VO)2P2O7
proceeds through complex processes with dehydration,
oxidation, and reduction in the reaction conditions [5–8].
Kiely et al. [7] demonstrated that a direct topotactic trans-
formation from VOHPO4·0.5H2O to (VO)2P2O7 occurs at
the periphery of the crystallite, whereas VOHPO4·0.5H2O
initially transforms epitaxially into ␦-VOPO4 in the interior
of the crystallite, followed by reduction of ␦-VOPO4 into
(VO)2P2O7. In addition, they further reported that the cata-
lysts, which were derived from VOHPO4·0.5H2O prepared
by different preparation methods, contained (VO)2P2O7
and some crystalline VOPO4 (V5+) phases in various rel-
ative proportions [8]; VOHPO4·0.5H2O (5 ␮m × 1 ␮m)
prepared in an aqueous medium transformed mainly into
␣II-VOPO4, while single (VO)2P2O7 phase was formed
As shown in Fig. 1c, the internal particle of EC(small) was
made up of oblong microcrystallites (ca. 50 nm). In consid-
eration of the XRD pattern (Fig. 3A(g)) and the bulk oxida-
tion number (Table 1) for EC(small), the crystalline phase of
these oblong microcrystallites was ascribed to (VO)2P2O7.
Meanwhile, the interior of C(large) contained the irregular
shaped platelets and some oblong microcrystallites. In addi-
tion, many white circles were observed in the TEM image
of C(large). In consideration of the bulk oxidation number
and XRD result, about half of V presented as the oxidized
phases, mainly ␣II-VOPO4, and the others as (VO)2P2O7 in

110
Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
Fig. 9. Schematic models of particle structures during the transformation for (A) EP(small) and (B) P(large).
C(large). The difference in the contrast between (VO)2P2O7
and ␣II-VOPO4 in a TEM image is expected to be small,
because the density of (VO)2P2O7 estimated from the lat-
tice constant and molecular weight is almost the same as
that of ␣II-VOPO4. Thus, white circles present in the inter-
nal C(large) are unlikely to be (VO)2P2O7 or ␣II-VOPO4.
Therefore, the observed white circles in C(large) can be as-
sociated with internal voids. Although uncertainty remains
in relation to which of the oblong microcrystallites and the
irregular shaped platelets are (VO)2P2O7 or ␣II-VOPO4 in
C(large), similarity of the oblong shape to that of EC(small)
suggests that the crystalline phase of the oblong microcrys-
tallites is assigned to (VO)2P2O7.
tion, oxidation, and reduction, and the crystallite size of the
precursor obviously dominates these processes.
In the case of the EP(small)–EC(small) system, weak
diffraction lines due to (VO)2P2O7 and ␦-VOPO4 arose at 1 h
of treatment, instead of disappearance of VOHPO4·0.5H2O
(Fig. 3A(d)). Since the diffraction lines were very weak
for the sample at 1 h, amorphous V-P oxides are also in-
cluded. As shown in Fig. 4A and B, the bulk oxidation
number of V increased as residual water decreased in the
initial stage. This relationship suggests that, in the initial
stage, the oxidized phases were formed by direct trans-
formation of VOHPO4·0.5H2O. Abon et al. [5] demon-
strated that a portion of VOHPO4·0.5H2O is oxidized to
␦-VOPO4 under the reaction gas. Kiely et al. [7] speculate
that, in consideration of the structure of VOHPO4·0.5H2O,
water molecules of VOHPO4·0.5H2O are released to the
direction along the layers when the direct dehydration of
VOHPO4·0.5H2O occurs, and further claimed that this de-
hydration is achieved within an extremely short time in the
periphery of the precursor crystallites [7]. Thus, the tem-
porarily existing oxidized amorphous phases in the parti-
cle of EP(small)–EC(small) system are reasonably supposed
to be located inside the particle rather than in surface lay-
ers.
On the basis of these results, schematic models of the
catalyst particles for EC(small) and C(large) are depicted in
Fig. 9.
4.2. Changes in microstructure during transformation
Under flow of helium, the transformation from VOHPO4·
0.5H2O to (VO)2P2O7 at 663 K was completed within 1 h for
both EP(small) and P(large) (data not shown). However, as
shown in Fig. 3, transformation of EP(small) to (VO)2P2O7
in the presence of the reaction gas proceeded slowly, whereas
P(large) transformed into mainly ␣II-VOPO4. These results
indicate that the transformation in the presence of the re-
action gas took place through processes involving dehydra-
As shown in Fig. 4C, the surface area did not change at all
in the initial few hours for the EP(small)–EC(small) system
and no peak was present in the mesopore region (Fig. 6A(a)

Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
111
and A(b)), indicating that the particles did not cleave at this
stage, and the inside of the particle could not come into direct
contact with the surrounding gases. Thus, VOHPO4·0.5H2O,
which exists inside the particle, is considered to be oxidized
by oxygen diffused from the surface.
n-butane over (VO)2P2O7 proceeds via a redox mechanism
at a few surface layers [13–15]. As mentioned in the pre-
vious section, the surface oxidation state of C(large) was
almost the same as that of EC(small), suggesting that the
two catalysts are of similar structure near the surface. This
is a reason for the moderate selectivity of C(large), in spite
of the main phase of ␣II-VOPO4.
After 25 h, since the surface area increased (Fig. 4C)
and a peak at 4 nm in the mesopore region appeared
(Fig. 6A(c)), the particles of EP(small)–EC(small) system
had been cracked. This probably allowed the reduction of
the oxidized phases existing originally inside the particle,
by direct contact with n-butane. As a result, the cata-
lyst particle formed from EP(small) was reduced to about
V4.0+. The mode of the transformation from the small
VOHPO4·0.5H2O crystallites is illustrated in Fig. 9A.
In the case of large crystallites (P(large)), the water
molecules still remained in the sample particle at 1 h
(Fig. 4A), and VOHPO4·0.5H2O was also detected by
XRD (Fig. 3B(d)), indicating that dehydration from the
large-sized VOHPO4·0.5H2O particle was slower than that
from the small-sized particle. The TG analysis in He flow
(Fig. 5) supports this finding.
As mentioned above, the direct dehydration of VOHPO4·
0.5H2O would occur preferentially in the edge of the crys-
tallites. A reasonable assumption is that this dehydration
proceeds to nearly the same depth from the surface inde-
pendent of the crystallite size. The TEM images of the
catalysts (Fig. 1d and e) demonstrate that the thickness of
the shell, which would be formed by direct dehydration,
in C(large) (ca. 70 nm) is similar to that for EP(small)
(ca. 60 nm). Consequently, at the initial stage the par-
ticles of the P(large)–C(large) system contain a greater
amount of VOHPO4·0.5H2O inside the crystallites. Since
VOHPO4·0.5H2O can be oxidized directly to ␦-VOPO4
[5], the large-sized precursor was deeply oxidized, in com-
parison with the small precursor. ␦-VOPO4 is reported
to transform progressively to the more stable ␣II-VOPO4
phase, and ␣II-VOPO4 is not further reduced to (VO)2P2O7
under the reaction conditions [7,16]. Thus, we conclude that
the ␦-VOPO4 phase confined inside the P(large) particle
gradually transformed to stable ␣II-VOPO4, which was no
longer reduced, and this is a reason why the large-sized cat-
alyst particles remained in the high oxidation state (V4.55+).
The process of transformation in the large-sized particle is
depicted in Fig. 9B.
We have further examined the influence of the ␣II-VOPO4
phase, which is present inside the particle of C(large). To
avoid the formation of V5+ phase, precursor P(large) was
dehydrated in He instead of the reactant mixture, and the
resulting sample, C(large)-He, was shown to have a surface
area of 10 m2 g−1 and a crystallite shape similar to that of
C(large) (Table 1). The oxidation number of V of the catalyst
bulk was determined to V3.99+. In addition, the surface oxi-
dation state of C(large)-He was the same as that of C(large).
Accordingly, XRD studies have revealed that C(large)-He
contains a pure phase of (VO)2P2O7. The catalytic activ-
ity of C(large)-He is slightly higher than that of C(large)
(Table 1), and correspondingly, its selectivity is similar to
that of C(large). These results indicate that the influence of
␣II-VOPO4 that is located inside the particles on the cat-
alytic property is not very large, and therefore, the essential
functions of these catalysts are governed by the structure of
the surface layers. Recently Bordes and Duvauchelle [18]
inferred that a mosaic texture was formed during the dehy-
dration process of VOHPO4·0.5H2O to (VO)2P2O7 and pre-
sumed that the intercrystallite boundaries play an important
role in the selective oxidation of n-butane. The influences of
the microstructure of the catalyst crystallites on the activity
and selectivity must be critical, and these remain issues to
be solved.
5. Conclusion
The present study clearly demonstrates that crystallite
size of the precursor VOHPO4·0.5H2O greatly affects crys-
tallite phases and microstructure of the catalyst formed
under the reaction conditions. The small size crystallites
of VOHPO4·0.5H2O (1 ␮m (length) × 110 nm (thickness))
transform into pure (VO)2P2O7. In contrast, the catalyst in
which the inside of the particle is mainly ␣II-VOPO4 is
formed from large-sized crystallites of VOHPO4·0.5H2O
(10 ␮m (length) × 415 nm (thickness)).
4.3. Influence of the microstructure of the catalysts on
selectivity in the selective oxidation of n-butane
The catalyst formed from the small-sized VOHPO4·
0.5H2O crystallites shows high selectivity to maleic anhy-
dride (∼78%), because of the presence of pure (VO)2P2O7
as well as the highly dispersed V5+ species on the surface.
The catalyst obtained from the large-sized VOHPO4·0.5H2O
crystallites exhibits moderately high selectivity (∼70%),
in spite of the main phase of ␣II-VOPO4. The simi-
larity in surface structure independent of the crystallite
size is responsible for the moderately high selectivity for
the catalyst formed from the large-sized crystallites of
VOHPO4·0.5H2O.
Shomoda et al. [17] reported that ␣II-VOPO4 yielded
only 20–30% selectivity towards MA, whereas (VO)2P2O7
yielded about 70% selectivity. In the present study, the
small-sized precursor (EP(small)) transformed into pure
(VO)2PO7 phase, which accounts for the high selectivity
of EC(small). Meanwhile, surprisingly, although the main
phase of C(large) was ␣II-VOPO4, C(large) showed mod-
erate selectivity to MA (∼70%). The selective oxidation of

112
Y. Kamiya et al. / Journal of Molecular Catalysis A: Chemical 220 (2004) 103–112
[6] G.J. Hutchings, A. Desmartin-Chamel, O. Oliver, J.C. Volta, Nature
Acknowledgements
348 (1994) 41.
[7] C.J. Kiely, A. Burrows, G.J. Hutchings, K.E. Bere, J.C. Volta, A.
Tuel, M. Abon, Faraday Discuss. 105 (1996) 103.
[8] C.J. Kiely, A. Burrows, S. Sajip, G.J. Hutchings, M.T. Sananes, A.
Tuel, J.C. Volta, J. Catal. 162 (1996) 31.
[9] J.W. Johnson, D.C. Johnston, A.J. Jacobson, J.F. Brody, J. Am.
Chem. Soc. 106 (1984) 8123.
[10] E. Bordes, P. Courtine, J. Solid State Chem. 55 (1984) 270.
[11] N. Yamamoto, N. Hiyoshi, T. Okuhara, Chem. Mater. 14 (2002)
3882.
This work was partially supported by the New Energy and
Industrial Technology Development Organization (NEDO).
This work was also supported by a Grant-in-Aid for Sci-
entific Research from the Ministry of Education, Science,
Sports, and Culture, Japan.
References
[12] B.K. Hodnett, P. Permanne, B. Delmon, Appl. Catal. 6 (1983) 231.
[13] M. Pepera, J.L. Callahan, M.J. Desmond, E.C. Milberger, P.R. Blum,
N.J. Bremer, J. Am. Chem. Soc. 107 (1985) 4883.
[14] T. Okuhara, M. Misono, Catal. Today 16 (1993) 61.
[15] G. Koyano, T. Okuhara, M. Misono, J. Am. Chem. Soc. 120 (1998)
767.
[16] F.B. Abdelouahab, R. Olier, N. Guilhaume, F. Lefebvre, J.C. Volta,
J. Catal. 134 (1992) 151.
[17] T. Shimoda, T. Okuhara, M. Misono, Bull. Chem. Soc. Jpn. 58
(1985) 2163.
[1] G. Centi, F. Trifirò, J.R. Ebner, V.M. Franchetti, Chem. Rev. 88
(1988) 55.
[2] B.K. Hodnett, Catal. Rev. Eng. 17 (1985) 373.
[3] S. Albonetti, F. Cavani, F. Trifirò, Catal. Rev. Sci. Eng. 38 (1996)
413.
[4] F. Trifirò, F. Cavani, J.A. Horsley, S.R. Vatcha, Selective Partial Ox-
idation of Hydrocarbons and Related Oxidations, Catalytica Studies
Division 4193 So, Mountain View, CA (1994).
[5] M. Abon, K.E. Bere, A. Tuel, P. Delichere, J. Catal. 156 (1995) 28.
[18] N. Duvauchelle, E. Bordes, Catal. Lett. 57 (1999) 81.