Mosaic Crystals of Vanadyl Pyrophosphate Obtained by Oriented Nucleation and Growth

Article abstract of DOI:10.1006/jssc.1997.7745

A study of the pseudomorphic transformation of the VOHPO₄·0.5H₂O precursor (P/V=1) to (VO)₂P₂O₇catalyst is reported. Attention is paid to the nucleation and growth stages, which are discussed in connection with the concept of topotaxy. The transformation mechanism which accounts for the textural change is established using electron microscopy techniques and X-ray diffraction. The morphology change to a prismatic shape at high temperature, approximately 870°C, is also discussed.

Full text of DOI:10.1006/jssc.1997.7745

JOURNAL OF SOLID STATE CHEMISTRY 137, 311—324 (1998)
ARTICLE NO. SC977745
Mosaic Crystals of Vanadyl Pyrophosphate Obtained
by Oriented Nucleation and Growth
N. Duvauchelle,* E. Kesteman,* F. Oudet,- and E. Bordes*ꢀꢁ
*
De
´
partement de Ge
´
nie Chimique, and -Service d+Analyse Physicochimique, Universite´ de Technologie de Compi e` gne, B.P. 20529, F-60205 Compi e` gne Cedex,
France
Received July 7, 1997; in revised form December 17, 1997; accepted December 19, 1997
By varying the alcohol used as a reflux medium in prep-
A study of the pseudomorphic transformation of the aration of the precursor from V O and H PO , it has
ꢂ ꢅ
ꢆ
ꢄ
VOHPO · 0.5H O precursor (P/V؍1) to (VO) P O catalyst is been shown that primary (plates or thin sheets) and second-
4
2
2
2
7
reported. Attention is paid to the nucleation and growth stages,
which are discussed in connection with the concept of topotaxy.
The transformation mechanism which accounts for the textural
change is established using electron microscopy techniques
and X-ray diffraction. The morphology change to a prismatic
shape at high temperature, approximately 870°C, is also dis-
cussed. ( 1998 Academic Press
ary (piles of plates or ‘‘roses of sand’’) particles may be
obtained for both precursor and catalyst (9, 10). The mor-
phology has also been shown to influence the reactivity of
the particles in an air—butane stream or in noncatalytic
reactions. For a while, two ‘‘forms’’ of vanadyl pyrophos-
phate, c and b (VO) P O , were thought to exist because of
ꢂ ꢂ ꢃ
their differences of reactivity. Indeed, in oxygen, these sam-
ples were reoxidized to c-VOPO and b-VOPO , respec-
ꢄ
ꢄ
tively (11, 12).
INTRODUCTION
Another point is that, in an air—butane stream, the
vanadyl pyrophosphate catalyst is partly oxidized with the
formation of Vꢅ>, or even transformed into several types of
Topotaxy is a very important phenomenon in the field of
reactivity of solids. A topotactic reaction can be defined as
a transformation in which one crystalline phase is converted
into another with a definite and reproducible crystallo-
graphic orientation relationship between the two. Such
a definition is in agreement with the official definition of
topotaxy given by the International Union of Crystallogra-
phy and carries no implication regarding the transforma-
tion mechanism (1). A wide range of reactions can be said
topotactic, from reactions with conservation of the three-
dimensional structure to reconstructive reactions.
When one wants good control of crystal morphology, as
is necessary in the case of structure-sensitive catalytic reac-
tions, finding a precursor which yields the catalyst topotac-
tically provides opportunity. Indeed, the properties of the
pseudomorphic catalyst can be controlled in this way. This
is the case, for example, in the preparation of transition
aluminas from boehmite (2) or of metallic molybdates (3, 4),
VOPO according to the conditions used by the various
ꢄ
authors (13, 14). Nguyen et al. (15) assumed that the exist-
ence of the Vꢅ> species is related to stacking faults, while
Thompson et al. (16) preferred to emphasize the role of
polytypes.
Torardi et al. (17) have recently noted that the final
product is made up of highly oriented vanadyl pyrophos-
phate microcrystals and that the large contraction of the
atomic structure in the direction perpendicular to the basal
plane induces cracks and voids within the crystal. The
present paper aims at describing the dehydration of
VOHPO ) 0.5H O in terms of oriented nucleation and
ꢄ
ꢂ
growth related to the concept of topotaxy, in order to
account for the special texture which characterizes
(
VO) P O catalyst. The morphology change occurring at
ꢂ ꢂ ꢃ
high temperature will also be discussed.
and also for (VO) P O , the well-known catalyst of the
ꢂ ꢂ ꢃ
selective oxidation of n-butane to maleic anhydride (5).
EXPERIMENTAL
Preparation
Indeed, when the precursor, VOHPO ) 0.5H O, is dehy-
drated in an inert atmosphere, pseudomorphic crystals of
VO) P O are obtained because of topotaxy (6, 7, 8).
ꢄ
ꢂ
(
a. VOHPO ) 0.5H O (precursor). When the starting ma-
ꢂ ꢂ ꢃ
ꢄ
ꢂ ꢅ
ꢂ
terials are V O and H PO , two distinct preparation
ꢆ
ꢄ
routes are mainly reported in the literature: preparation in
an aqueous medium with reduction by HCl or other acids,
ꢁ
To whom correspondence should be addressed.
3
11
0
022-4596/98 $25.00
Copyright ( 1998 by Academic Press
All rights of reproduction in any form reserved.

3
12
DUVAUCHELLE ET AL.
and preparation in an organic medium, generally, alcohol a lacey carbon film supported on a copper mesh grid. SEM
18). Another synthetic route consists of the reduction of the analysis was carried out using a JEOL 840C microscope to
oxovanadium phosphate dihydrate, VOPO ) 2H O, in obtain informations on external morphology.
(
ꢄ
ꢂ
alcohol (6, 8, 19).
Potentiometric titration was carried out using the
Well-characterized materials are needed for the present method of Niwa and Murakami (22) to estimate the average
purpose. Among the alcohols chosen, we had already no- oxidation number of vanadium.
ticed (20) that 2-butanol led to the formation of thin plates
The specific surface area was estimated by the one-point
with definite morphology. Therefore, V O (11.8 g) and BET method, using nitrogen gas at 77 K in Quantasorb
ꢂ ꢅ
H PO 85% (14.9 g) were mixed in 2-butanol (250 ml). After equipment.
ꢆ
ꢄ
refluxing for 6 h, the light green suspension was separated
from the organic solution by filtration and washed several
times with the same alcohol (200 ml), and then with acetone
RESULTS
VOHPO ) 0.5H O
(
500 ml), to remove organic residue from the solid. The
resulting light green solid was dried in an oven at 120°C
15 h).
ꢄ
ꢂ
The VOHPO ) 0.5H O phase crystallizes in the ortho-
rhombic system (P ) with a"7.416 A
ꢅꢅꢂ
(23). The XRD pattern of the solid prepared in
-butanol shows that the expected single phase,
VOHPO ) 0.5H O, is well crystallized. No line broadening
ꢄ
ꢂ
(
s s
, b"9.592 A, and
c"5.689 A
s
b. (VO) P O (catalyst). One gram of the precursor was
ꢂ ꢂ ꢃ
2
calcined in an alumina boat placed in a quartz tube and
heated at 5°C/min in a nitrogen flow (1 l h\ꢁ). Various
temperatures and times of calcination were used, yielding
samples called Pyro-420, Pyro-750, and Pyro-870 (Table 1).
ꢄ
ꢂ
of the (001) reflection, which could account for stacking
plates disorder such as when organic molecules are trapped
in between the sheets (24), is observed (Fig. 1). The value of
the internal strain, e, is thus very low (Table 1). The specific
surface area is low and amounts to 4 mꢂ g\ꢁ.
By SEM, the particles of precursor look like very thick
bitrapezoidal platelets (Fig. 2). The TEM observation of
several non-isolated particles confirms this particular mor-
phology (Fig. 3a). The presence of Bragg extinction con-
tours, as seen in Fig. 3a, indicates that the platelets are often
slightly bent, and that their crystalline state is good. The
selected area electron diffraction (SAED) pattern of an iso-
lated particle (Fig. 3b) corresponds to a single crystal lying
on the (001) plane, as confirmed by the sharpness of the
electron diffraction spots. This is also confirmed by the dark
field image of the same particle, which is completely illumin-
ated (Fig. 3b).
c. Partially transformed catalyst. A TGA/DTA appar-
atus was used to control the formation of a material not yet
totally dehydrated (Setaram microbalance TG92, N flow).
ꢂ
A 20 mg of Sample VOHPO ) 0.5H O was heated to 420°C
ꢄ
ꢂ
(
(
heating rate: 5°C/min) and quenched at room temperature
sample called P420-Q). The loss of weight (7.2%), clearly
below the theoretical one (10.46%), confirms the partial
transformation of the sample.
Characterization
The precursor sample and the final products after calcina-
tion were characterized by powder X-ray diffraction (XRD)
and scanning and transmission electron Microscopy (SEM,
TEM).
XRD was performed on powders using an INEL CPS 120
(
curve detector) and CuKa radiation (0.15406 nm). The
ꢁ
Diffrac-At software was used to calculate the full width at
half maximum, b , for each main Bragg peak. The instru-
%
ꢀꢁ
mental broadening b
was determined by a calibration
*
ꢂꢃꢄ
procedure using a well-crystallized V O material. The
ꢂ ꢅ
, was calculated
broadening due to the sample itself, b
ꢃ!ꢅꢁꢆ%
by bꢂ
"bꢂ !bꢂ . The average crystallite size (¸ )
ꢃ!ꢅꢁꢆ%
%ꢀꢁ *ꢂꢃꢄ ꢅ
and internal strain (e) data gathered in Table 1 were evalu-
ated satisfactorily using the Williamson—Hall formula (21):
bꢃ!ꢅꢁꢆ% cos h"0.94 j/L #2e sinh.
ꢅ
TEM observations were made in a JEOL 1200EX micro-
scope, operating at 80 kV to avoid modification of the
sample under the beam. Microdiffraction patterns were also
obtained to identify each sample. Powder samples were
FIG. 1. Powder X-ray diffraction pattern of VOHPO ) 0.5H O pre-
ꢄ
ꢂ
dispersed in CCl by ultrasonic waves and displayed onto pared in 2-butanol.
ꢄ

MOSAIC VANADYL PHOSPHATE CRYSTALS
313
TABLE 1
Characteristics of VOHPO
4
· 0.5H
2
O Prepared in 2-Butanol and of the Three Samples of (VO)
P
2 2
O
7
Prepared at Various Temperatures and Times of Calcination
Temperature of
Time of
Surface area
Oxidation state
Average crystallite
Internal strain@
b (* 10\ꢆ)
calcination (°C)
calcination (h)
Color
(mꢂ/g\ꢁ)
of vanadium?
size@ Lm (A
s
)
Precursor
Pyro-420
Pyro-750
Pyro-870
—
—
48
3
Green
Brown
Grey
4
4
<1000
437
2.35
9.26
6.53
4.32
420
750
870
5.6
6.1
1
4.146
4.012
A
995
24
Blue-grey
<1000
?
@
Potentiometric titration (22).
Calculated by the method of Williamson and Hall (21).
A Titration could not be performed because of non dissolution of the sample.
These results lead us to conclude that the vanadyl hydro- average oxidation state of vanadium decreases, as observed
gen phosphate hemihydrate platelets are monocrystalline by Nguyen et al. (15), from 4.146 to 4.012.
monolith particles, according to the classification proposed
by Figlarz et al. (25).
When compared to the diffraction lines of the VOHPO )
ꢄ
0.5H O starting material, a significant broadening is ob-
ꢂ
served for Pyro-420 and Pyro-750 patterns. A weak
broadening effect is still noted for Pyro-870. The average
crystallite size of Pyro-750 calculated by the method of
(
VO) P O
ꢂ ꢂ ꢃ
s
The XRD pattern of the samples calcined at various Williamson-Hall is 995 A, and the internal strain e is
temperatures, characteristic of the (VO) P O phases, has 6.5;10\ꢆ, clearly below that of Pyro-420 (Table 1). How-
ꢂ ꢂ ꢃ
been indexed in the monoclinic system P2 using the cell ever, let us note that the average line b
cos h"f (2sin h)
ꢁ
ꢃ!ꢅꢁꢆ%
parameters given recently by Nguyen et al.: a"7.7276 A
b"16.5885 A, c"9.5796 A, b"89.5796 (26) (Fig. 4).
The color, the surface area, and the average oxidation
s
,
obtained by the Williamson—Hall method does not fit
s
s
exactly the experimental points all.
SEM pictures show that the morphology of Pyro-420 and
state of vanadium vary with the temperature of dehydration Pyro-750 is different from that of Pyro-870. For Pyro-420
Table 1). Compared with the precursor, a slight increase of and Pyro-750, the pseudomorphism with precursor is evid-
(
the surface areas of Pyro-420 and Pyro-750 is observed. The enced because platelets with the same bitrapezoidal
FIG. 2. SEM image of VOHPO ) 0.5H O prepared in 2-butanol.
ꢄ
ꢂ

3
14
DUVAUCHELLE ET AL.
FIG. 3. (a) Bright field image of several particles of VOHPO )0.5H O; (b) dark field image of a single precursor particle and corresponding SAED
ꢄ
ꢂ
showing the particle lying on the (001) plane.
morphology are observed (Fig. 5a). The surface of each images allow to make some distinctions between them. In
particle is rough and some voids can be seen, especially in the case of Pyro-420, the platelets do not show any Bragg
the case of Pyro-750. For Pyro-870, the shape and size of the extinction contour and seem to be composed of many
precursor are not retained since particles are prismatic and microcrystals (i.e., crystallites) (Fig. 6a). Most of the bright
exhibit a certain degree of sintering between them (Fig. 5b). regions in the TEM image are due to voids, as suggested by
Because of the size and thickness of the prismatic crystals of Torardi et al. (17), that would explain why the surface area
Pyro-870, the TEM study could not be made in the same increases a little as compared with the precursor’s (Table 1).
way as for Pyro-420 and Pyro-750.
SAED shows that the diffraction spots are curved, indicat-
When examining Pyro-420 and Pyro-750, the first obser- ing a small misorientation of these crystallites inside the
vation by TEM is that platelets are displayed on the (100) particle (Fig. 6b). Nevertheless, we can consider this sample
plane as shown by SAED, and exhibit the same character- as made from single crystals lying on the (100) plane. The
s
istic shape as the precursor (Figs. 6, 7). SAED and dark field average crystallite size (437 A) calculated from XRD by

MOSAIC VANADYL PHOSPHATE CRYSTALS
315
FIG. 3—Continued
the Williamson—Hall method, is clearly much smaller than particles (25), i.e., they are built up by a great number of
the observed size of the platelets (several micrometers). The small crystallites slightly misoriented with respect to each
dark field image confirms these results more explicitly (Fig. other.
6
b), because only domains in Bragg position in relation to
The microcrystals of Pyro-750 are bigger and exhibit
the incident electron beam are observed. These images show a more ‘‘chaotic’’ aspect, as for the particle shown Fig. 7a.
a large number of small white spots, the size of which is When looking at the center of this particle, some voids are
about the same order of magnitude than the average crystal- observed, together with a large, individual, crystallite which
lite size calculated from XRD. Consequently, these white seems to stand out of the particle. This is confirmed by the
spots ‘‘materialize’’ the crystallites. Finally, the relatively dark field image which shows two distinct areas (Fig. 7b):
high value (e"9.2;10\ꢆ) of the internal strain can be re- a large dark domain, and a mosaic domain characterized by
lated to the presence of small angle boundaries between the a large number of white spots. These spots are bigger than
crystallites inside the grain. Therefore, we conclude that the those measured on Fig. 6b, confirming thereby the increase
Pyro-420 platelets of (VO) P O are monocrystalline mosaic of the crystallite size (Table 1). The large dark domain,
ꢂ ꢂ ꢃ

3
16
DUVAUCHELLE ET AL.
Whereas each bitrapezoidal platelet of VOHPO )
ꢄ
0
.5H O is a monolithic single crystal, the dehydrated
ꢂ
pseudomorphic platelets (Pyro-420) of vanadyl pyrophos-
phate contain a large number of small, fairly well oriented,
crystallites (mosaic particle). After dehydration at higher
temperature (Pyro-750), the crystallite size Lm increases
while the internal strain decreases, which indicates a tex-
tural change. The prismatic morphology of particles of
vanadyl pyrophosphate is observed when the sample is
calcined at 870°C (Pyro-870).
We propose a mechanism of dehydration based on these
experimental observations, to account for the formation of
monocrystalline mosaic particles and of prismatic particles.
Dehydration Mechanism
FIG. 4. Powder X-ray diffraction pattern of the three samples of
VO) P O . (a) Pyro-420; (b) Pyro-750; (c) Pyro-870.
The mechanism of dehydration is schematized in Figs.
(
ꢂ ꢂ
ꢃ
1
0a—10d to account for the formation of the observed tex-
ture up to 750°C. Steps e and f (Figs. 10e and 10f ) refer to the
which corresponds exactly to the individual crystallite, does formation of prismatic particles at higher temperature.
not present any Bragg orientation in relation to the electron
The dehydration reaction starts by the oriented nuclea-
beam. The fact that the Williamson—Hall relation does not tion of the vanadyl pyrophosphate phase onto the (001)
fit exactly all experimental points means that the sizes of the plane of the precursor (Fig. 10a), as expected from the
crystallites are polydispersed in agreement with the TEM similar atomic arrangement in the (VOHPO ) sheets and
ꢄ ꢇ
observations, and/or that the internal strain e is not fully in the (100) plane of (VO) P O (26, 27). From a thermodyn-
ꢂ ꢂ ꢃ
isotropic. Consequently, the model of mosaic crystal, which amic point of view, this is the most favorable situation
is convenient to describe Pyro-420, is not well adapted for corresponding to the lowest nucleation energy. One can
Pyro-750 because of its heterogeneous texture.
propose that the reaction begins on the precursor platelet
The XRD pattern of the partially transformed solid P420- edges because of the easier escape of the water molecules
Q (Fig. 8) shows that a mixture of VOHPO ) 0.5H O and situated between the layers. However, this proposal is not
ꢄ
ꢂ
(
(
VO) P O is obtained, as expected from the loss of weight supported by our experiments because no interphase
ꢂ ꢂ ꢃ
7.2%). No amorphous phase has been evidenced. The TEM boundary could be evidenced in the partly transformed
micrographs of this mixture show that the texture of the product (P420-Q).
particles is not different from that observed with Pyro-420
During the oriented growth stage, the hemihydrate/pyro-
Fig. 9). However, the SAED of a single platelet indicates the phosphate interphase boundary moves while maintaining
simultaneous presence of the two phases VOHPO ) 0.5H O the structural orientation relationship between the two
(
ꢄ
ꢂ
and (VO) P O , in definite and reproducible crystallo- phases. Due to the low crystallographic misfit (0.13% along
ꢂ ꢂ ꢃ
graphic orientation between each other (Fig. 9):
[010]
// [001]
ꢁꢇꢈꢉ
and 11.8% along [100]
// [010] ),
)
%ꢅ*
)%ꢅ* ꢁꢇꢈꢉ
the existence of a coherent interface between the two phases
makes the (VO) P O nucleation easier during the further
growth. The elastic strains resulting from the mismatch of
the lattices are relaxed by the formation of dislocations,
which move and assemble (dislocation loops) as the reaction
proceeds, to form crystallite boundaries (Figs. 10b, 10c).
These strains limit the pyrophosphate growth in space.
[
001])%ꢅ* // [100]ꢁꢇꢈꢉ
and
ꢂ ꢂ ꢃ
[
100])%ꢅ* // [010]ꢁꢇꢈꢉ
.
No interphase boundaries between these phases could be
highlighted by TEM but the particle P420-Q seems to be
composed of VOHPO ) 0.5H O and (VO) P O small
However, as domains of (VO) P O keep an orientation
ꢂ ꢂ ꢃ
close to the topotactic one, they are weakly misoriented. As
ꢄ
ꢂ
ꢂ ꢂ ꢃ
the interface progresses, some pyrophosphate nuclei in
topotactic relation, but which cannot grow normally be-
cause of the produced strains, are still in formation. For this
reason, each particle, which appears in SAED as a single
microdomains.
DISCUSSION
We can now summarize the essential features characteriz- crystal, is built up of a large number of small, slightly
ing the textural and morphological modifications occurring misoriented, crystallites (Fig. 10d). Both crystallite size and
during dehydration.
internal strain support the model assuming a mosaic texture

MOSAIC VANADYL PHOSPHATE CRYSTALS
317
FIG. 5. SEM image of Pyro-750 (a) and Pyro-870 (b).
for each (VO) P O particle. Indeed, the mosaic texture they occur play also an important role in the textural
ꢂ ꢂ ꢃ
implies the existence of disorganized areas between each modifications of the reaction product. The operating para-
crystallite. These defects, which consist mainly of disloca- meters (temperature, heating rate, and atmosphere) deter-
tions, are responsible for the slight misorientation of these mine the kinetics of the reaction.
crystallites. Therefore, the internal strain e appears as a stat-
istic measure of the inhomogeneous strains due mainly to though topotactic, is a reconstructive reaction: V—O, O—H,
these dislocations.
and P—O bonds are broken and a new VPO network is
It is worth noting that the dehydration mechanism,
According to this reaction mechanism, the formation of reconstructed. Despite the structural similarities between
a mosaic texture is a way to minimize the elastic strains the hydrate and the final oxide phase, the structural ele-
produced, and, as such, is influenced by several factors. ments are not stricto sensu maintained during the course of
Indeed, the textural characteristics of the mother phase the dehydration reaction. In our experiments, the XRD
(
shape, size and texture) and the mismatch between the pattern of the partially dehydrated VOHPO ) 0.5H O
lattices are decisive for the mechanism by which strains are sample P420-Q shows no indication of an amorphous
relaxed. The number of dislocations and the rate at which phase, but exhibits only the lines of the precursor and of the
ꢄ
ꢂ

3
18
DUVAUCHELLE ET AL.
FIG. 6. (a) Bright field image of a particle of Pyro-420; (b) dark field image of the same particle and corresponding SAED showing the particle lying on
the (100) plane.
final (VO) P O solid. However, several reports suggest the Textural and Morphology Evolution
ꢂ ꢂ ꢃ
formation of such an amorphous intermediate (28, 29) which
can be accounted for by our model. Let us consider the stage
When the temperature of calcination is higher than
of the reaction schematized in Figs. 10b and 10c. Particles 420°C, the crystallinity of the material improves, owing to
are composed of many domains of both precursor and the usual mechanism of growth of coherent domains. There-
pyrophosphate, smaller than the initial particle. Line fore, the crystallite size increases while the internal strain
broadening would result from these small domains, as well decreases. During a first period, the evolution of the crystal-
as from the strains produced in this mixed particle, to such linity of (VO) P O leads to a textural change without
ꢂ ꢂ ꢃ
an extent that a disappearance of Bragg peaks could be modification of morphology. The process begins at the
expected (16). These samples being metastable, the ‘‘stabiliz- center of the mosaic particle, inducing the formation of
ation’’ of an X-ray amorphous state is strongly dependent some cracks and holes (Fig. 10e). Then, the central crystal-
on heat and mass transfers.
lites grow before they break away the particle. This stage of

MOSAIC VANADYL PHOSPHATE CRYSTALS
319
FIG. 6—Continued
textural evolution corresponds typically to the Pyro-750 ticles. It is worthwhile to note that the morphology of
sample. Therefore, the mosaic model becomes a less faithful Pyro-870 is quite similar to the equilibrium shape of
representation of the real texture. The previous process (VO) P O crystal obtained by Curie-Wulff construction
ꢂ ꢂ ꢃ
spreads inside all the particle, leading to its total fragmenta- and surface energy considerations (30).
tion in several single crystals. In the same time, the primary
Finally, as far as defects are concerned, the vanadium
particles would sinter during the increase of calcination oxidation state is seen to increase with the increase of
temperature to give the final prismatic morphology internal strain, i.e., with the increase of the calcination
(Fig. 10f). This is in accordance with the TEM picture of temperature (Table 1). Such an observation is not surprising
Fig. 11, which represents a vanadyl pyrophosphate particle if we assume that any Vꢅ> is associated with defects in the
calcined, at 800°C for 24 h under nitrogen. Moreover, such bulk of (VO) P O (intercrystallite boundaries), rather than
ꢂ ꢂ ꢃ
a transformation explains the difference between the sizes of with a Vꢅ>-containing phase, like VOPO . Nguyen et al.
ꢄ
the bitrapezoidal platelets and of the final prismatic par- (16) have the same opinion when they modelize these defects

3
20
DUVAUCHELLE ET AL.
FIG. 7. (a) Bright field image of a particle of Pyro-750; (b) dark field image of the same particle and corresponding SAED showing the particle lying on
the (100) plane.
as bc stacking faults which hinder the dimerization of some oriented, small crystallites, and also why these particles are
(
PO )ꢆ\ groups during dehydration process, and therefore monocrystalline. This process avoids the formation of a ran-
ꢄ
imply the formation of Vꢅ>.
dom texture and is responsible for the pseudomorphism.
The textural modification observed when the temperature
of calcination is very high explains the change of morpho-
logy into prismatic vanadyl pyrophosphate, due to frag-
mentation and sintering of the initial plates.
CONCLUSION
The dehydration of VOHPO ) 0.5H O to (VO) P O is
a kind of oriented nucleation-growth, similar to the dehy- to catalytic performance, especially by the creation of addi-
The formation of such a mosaic texture may be beneficial
ꢄ
ꢂ
ꢂ ꢂ ꢃ
dration of Ni(OH) to NiO studied by Figlarz et al. (31). The tional surface area. Moreover, the intercrystallite bound-
ꢂ
considerations on oriented nucleation and growth stages aries, which are disorganized areas, may play a catalytic role
explain why the dehydration reaction yields (VO) P O in the selective oxidation of n-butane to maleic anhydride
ꢂ ꢂ ꢃ
particles constituted by a great number of slightly mis- (12, 32).

MOSAIC VANADYL PHOSPHATE CRYSTALS
321
FIG. 7—Continued
FIG. 8. Powder X-ray diffraction pattern of partially transformed VOHPO ) 0.5H O (P420-Q), heated at 420°C (TGA/DTA), showing the
ꢄ
ꢂ
coexistence of (VO) P O and VOHPO ) 0.5H O (*) lines.
ꢂ ꢂ
ꢃ
ꢄ
ꢂ

3
22
DUVAUCHELLE ET AL.
FIG. 9. Bright field image and SAED of a partially transformed VOHPO ) 0.5H O (P420-Q) particle, showing the structural orientation relation-
ꢄ
ꢂ
ships between the hemihydrate (H) and (VO) P O (P).
ꢂ ꢂ
ꢃ
FIG. 10. Scheme of the topotactic dehydration showing textural modification (a—d) and morphology change (e—f ). See text for details.

MOSAIC VANADYL PHOSPHATE CRYSTALS
323
FIG. 10—Continued
FIG. 11. Bright field image of a precursor particle calcined at 800°C under nitrogen for 24 h.

3
24
DUVAUCHELLE ET AL.
16. M. R. Thompson, A. C. Hess, J. B. Nicholas, J. C. White, J. Anchell, and
ACKNOWLEDGMENT
J. R. Ebner, in ‘‘New Developments in Selective Oxidation II’’ (V. C.
Corberan and S. V. Bellon, Eds.), p. 167. Elsevier Science, Amsterdam,
Thanks are due to the Institut f u¨ r Angewandte Chemie (Berlin-Adler-
shof ) for potentiometric titrations made within the framework of the
Human Capital and Mobility Program of the European Community
1
994.
1
1
1
7. C. C. Torardi, Z. G. Li, H. S. Horowitz, W. Liang, and M. H. Whangbo,
J. Solid State Chem. 119, 349 (1995).
8. L. M. Cornaglia, C. A. Sanchez, and E. A. Lombardo, Appl. Catal.
A Gen. 95, 11 (1993).
9. I. J. Ellison, G. J. Hutching, M. T. Sananes, and J. C. Volta, J. Chem.
Soc. Chem. Commun., 1093 (1994).
(
Contract ERBCHRXCT930290).
REFERENCES
1
. International Union of Crystallography, Acta Crystallogr. Sect. A 33,
81 (1977).
. B. C. Lippens, Ph.D. Thesis, University of Delft, Netherland (1961).
. P. P. Cord, Ph.D. Thesis, Universite de Paris VI, France (1972).
20. E. Kesteman, Ph.D. Thesis, Universite
´ de Technologie de Compi e` gne,
6
France (1997).
2
3
4
21. G. K. Williamson and W. H. Hall, Acta Metall. 1, 22 (1922).
22. M. Niwa and Y. Murakami, J. Catal. 76, 9 (1982).
´
. O. Lezla, E. Bordes, P. Courtine, and G. Hecquet, J. Catal. 170, 346 23. C. C. Torardi and J. C. Calabrese, Inorg. Chem. 23, 1308 (1984).
(1997).
24. G. Busca, F. Cavani, G. Centi, and F. Trifiro, J. Catal. 99, 400
5
6
. G. Cavani and F. Trifiro, Catalysis 11, 246 (1994).
(1986).
. E. Bordes, P. Courtine, and J. W. Johnson, J. Solid State Chem. 55, 270 25. M. Figlarz, F. Vincent, C. Lecaille, and J. Amiel, Powder ¹echnol. 1, 121
(1984).
(1967).
7
8
. C. C. Torardi and J. C. Calabrese, Inorg. Chem. 23, 1308 (1984).
. J. W. Johnson, D. C. Johnston, A. J. Jacobson, and J. F. Brody, J. Am.
Chem. Soc. 106, 8123 (1984).
26. P. T. Nguyen, R. D. Hoffman, and A. W. Sleight, Mater. Res. Bull. 30,
1055 (1995).
27. E. Bordes, J. W. Johnson, A. Raminosona, and P. Courtine, in Mater-
ials Science Monographs, Vol. 280, p. 887. Elsevier, Amsterdam, 1985.
28. J. C. Vedrine, J. M. M. Millet, and J. C. Volta, Faraday Discuss. Chem.
Soc. 87, 207 (1989).
9
. H. S. Horowitz, C. M. Blackstone, A. W. Sleight, and G. Teufer, Appl.
Catal. 38, 193 (1988).
1
0. E. Kesteman, C. M. Merzouki, B. Taouk, E. Bordes, and R. Contractor,
in ‘‘Preparation of Catalyst, VI’’ (G. Poncelet et al., Eds.), Stud. Surf. 29. P. Amoros, R. Ibanez, A. Beltran, D. Beltran, A. Fuertes, P. Gomes-
Sci. Catal. 91, 707 (1995).
Romero, E. Hernandez, and J. R. Carvajal, Chem. Mater. 3, 407 (1991).
30. J. Ziolkowski, Surf. Sci. 209, 536 (1989).
31. M. Figlarz, B. G e´ rand, A. Delahaye-Vidal, B. Dumont, F. Harb,
A. Coucou, and F. Fievet, Solid State Ionics 43, 143 (1990).
32. N. Duvauchelle, H. Zanthoff, F. Hannour and E. Bordes, in ‘‘Proceed-
1
1
1
1. E. Bordes and P. Courtine, J. Chem. Soc. Chem. Commun., 294 (1985).
2. N. Duvauchelle, P. Courtine, and E. Bordes, to be published.
3. C. J. Kiely, A. Burrows, S. Sajip, G. J. Hutchings, M. T. Sananes,
A. Tuel, and J. C. Volta, J. Catal. 162, 31 (1996).
1
1
4. G. Centi, G. Golinelli, and F. Trifiro, Appl. Catal. 48, 13 (1989).
5. P. T. Nguyen, A. W. Sleight, N. Roberts, and W. W. Warren, J. Solid
State Chem. 122, 259 (1996).
ings. DGMK-Conference ‘C Chemistry—Manufacture and Use of
ꢄ
C
Hydrocarbons’’’ (W. Keim, B. Lu
¨
cke, and J. Weitkamp, Eds.),
ꢄ
p. 181. 1997.