499

CRYSTALLOCHE~ISTRY OF V-P-O PHASES AND APPLXCATION TO CATALYSIS

E. BORDES
D~partement de Genie ~himique, Universite de Technologie de CompiPgne, B.P. 233,
60206 CompiPgne (France)

ABSTRACT
Since the crystal structure is the result of all of the interactions between
the atoms, crystal chemistry of catalysts gives information on (i) the kind of
defects, (ii) what can be the atomic arrangement at the surface of the catalyst,
(iii) how the active sites can be displayed. The knowledge of the reactivity of
the catalyst allows one to determine what phases are really present in the stea-
dy state and to examine the influence of the precursors used in the preparation.
These ideas are applied to the case of V-P-O catalysts in the oxidation of bu-
tene and butane to maleic anhydride, in which the same phase (vanadyl pyrophos-
phate) seems to be involved. It is shown how the methods of preparation deter-
mine the kind of redox system which is involved and how the various crystalline
forms of VOP04 account for the different reactivities observed.

INTRODUCTION
The usefulness of the concept of "structure-sensitive reactions" (ref. 1) in
the field of mild oxidation catalysis has now been demonstrated and it
has been experimentally proven (refs, 2-4) that, for a given compound with a de-
finite crystal structure, different kinds of crystallographic faces are cataly-
tically active for different reactions. In the case of supported catalysts, even
the support has been shown to exert an influence on the superficial planes ex-
posed, as the Strong Metal-Support Interaction (SKI) observed for metals (refs.
5, 6). In the case of crystalline V205 supported on Ti02 (anatase or B), this
influence was thought to be exerted by means of coherent interfaces between the
structurally related lattices of the two components (refs. 7, 8) : after indi-
rect methods (refs. 9, lo), this has just been demonstrated directly by HREM
(refs. 11-13). The crystalline state of the catalysts in the mild oxidation of
hydrocarbons or in the oxidative dehydrogenation of oxidized compounds allows
X-ray diffraction and other classical methods of analysis to be used to obtain
information on the nature and texture of the phases involved in catalysis.
Since the structural arrangement is the result of the interactions between
all atoms, is is easily conceivable that the crystal structure, together with
thermodynamics* accounts for the main properties of a solid (ref. 14). This in-
cludes information on (i) the chemical nature of the bonds, (ii) the kind of de-
fects, (iii) what could be the arrangement of the atoms at the surface. The
latter is necessarily related to the bulk structure, particularly in the case of
layered crystalline solids which statistically present their cleavage planes at
the surface.
However, many problems occur when one tries to use crystallographic data in
catalysis because the following must be known : (i) the nature of the precursor
and the methods of preparation and activation, which can deliver unwanted phases
and/or influence the texture of the final catalyst, and (ii) the nature of the
phase(s) realty present at the steady state, particularly when the initial solid
is partly reduced by the reacting mixture (ref. 14). This means that the study
of the reactivity at each step of the elaboration (and if possible of the work)
of the catalyst is not separable from structural studies.
Generally speaking, it can be seen that selectivity in a given reaction is
connected with the presence of one kind of compound (with its own reactivity).
Examples are numerous, such as V205 supported on Ti02 anatase in the o-xylene
oxidation to phthalic anhydride, or V-MO-O (PO/V = l/3) in benzene-maleic anhy-
dride. In each case a specific initial phase (V205 or VI 32Moo S605 respective-
ly) and a given redox system (even if this one is difficult to determine) are
involved. As it appears from the literature, in the case of the V-P-O system va-
nadyl pyrophosphate (VO)~P~O, is involved in the oxidation of both butane and
butene to maleic anhydride, which at first sight seems to be in contradiction
with the preceding remark. Yoreover, according to the P/V atomic ratio, the me-
thod chosen for preparation and the nature of the hydrocarbon different phases
can appear in catalytic experiments, as is often noticed in the literature,
which enhances the difficulties for a better understanding of the system. In our
attempts to clarify the situation, we have chosen to describe first the crystal
structure of the phases more frequently observed. Since the chosen method of
preparation of (VO)2P207 must be appropriate in order to get selectivity from
one or the other C4-hydrocarbon, we intend to show here how the kind of redox
system involved is determined by the preparation and how the various crystalline
forms account for the different reactivities observed. The discussion will deal
with the selective role of the structures during the individual steps of reac-
tion, taking numerous literature observations into account.

CRYSTAL STRUCTURES OF V-P-O PHASES

The known crystal structures of the main phases and precursors of the (V-P-O)
system involved in catalysis can be related to the Re03 framework in which phos-
phate groups are included. The presence of the short V=O bond of the vanadyl
group, in its oxidized (VO)3+ or reduced (VO)*+ form, leads by itself to a cha-
racteristic structural anisotropy which allows (V-P-O) to be compared with the
(V-O) system (refs. 14-16). In most cases the distorted octahedra are linked to-
gether so as to produce chains of alternated short and long vanadium-oxygen
 501

bonds O=V...O=V . ..O. The four equatorial corners are generally shared with
single or double tetrahedra as in ortho (PO4) and pyro (P20,) phosphates respec-
tively, or even with chains of metaphosphate (PO3), according to the valence of
3+
vanadium and the atomic ratio P/V. If we eliminate compounds containing V ,
which are rarely encountered in mild oxidation catalysis, all the structures can
be described by combination of two kinds of elementary unit :
1 - One single octahedron linked to orthophosphate or hydrogenphosphate
groups, each equatorial oxygen of VO6 sharing a corner of one PO4, with vanadyl
chains running in the perpendicular direction.
Infinite (VOPO,)_ layers are present in VOP04.2H20, VOP04.H20, VOPO4.2p20
(refs. 16-18) and in the anhydrous forms aI and aII-VOP04 (refs. 15, 19). In the
hydrates, these layers are linked by water molecules ; one H20 (D20) is coordi-
nated to vanadium and the second is considered as "zeolithic" (see fig. la). For
these and for isostructural aI-VOP04 the space group is P4/nmm (Cih) while it is
P4/n (Cih) for aII -VOP04 (see fig. lb, c) : this brings once more credibility to
the existence of two distinct forms of aI-VOP04, which differ in the relative
position of vanadyl (V=O) as compared to the tetrahedra (ref. 20). In B-VOPU4,
zig-zag chains of octahedra occur in the plane (010) (ref. 21), two oxygens of
the same PO4 being shared with two next V06 linked by O=V...O bonds (see fig. Id).
The structure of vanadyl hydrogenphosphate tetrahydrate VOHP04.4H20 is in
fact close to that of VOP04.2H20 with the same structural unit of alternating
octahedra and tetrahedra, which are arranged as a double chain in one direction
and not as layers (see fig. 2a). Two water molecules are coordinated to vanadium
in transposition to V=O and the remaining two are isolated in the channels
formed by the hydrogen bonding network (ref. 22). The partial dehydration of
VOHP04.4H20 leads to VOHP04.0.5H20 (see fig. 2b), the structure of which will be
described in the next paragraph.
In the case of VO(H2P04)2, the elementary units are diagonally linked in the
(001) plane : chains O-V-O-P-O are diagonally formed, each equatorial oxygen of
the V06 group being shared with one of a tetrahedron 02P(OH)2, and four tetrahe-
dra are hydrogen-bonded by means of the two hydroxyl groups (fig. 3a). In the

perpendicular direction, the O=V...O=V chains are parallel to the square channels
drawn by the hydrogen bonds between the chains of 02P(OH), this alternation en-
suring the stoichiometry P/V=2 (ref. 23). The structure of the vanadyl meta-
phosphate VO(P03)2 is directly related to the preceding (see fig. 3b). The
chains O=V...O=V of distorted octahedra run parallel to the chains of corner-
sharing (-P02-) groups (linear metaphosphate (P03)n), to which they are connec-
ted by means of the equatorial oxygen atoms which do not lie in the same plane
(refs. 24, 27). Short V=O bonds have been observed by UV-visible spectroscopy
(refs. 24, 25), contrarily to the assertions of ref. (26). and positional disor-
der of vanadium inside its polyedron has been evidenced (ref. 27).
502

Fig. 1. Coordination polyhedra around V and P in the crystal structure of :

a) VOP04.2H20 ; b) ~I-VOPO~ ; c) aII-VOP04 ; d) B-VOPO4. Legend : (0) V ;
(0) P ; (0) oxygen ; (@) vanadyl oxygen ; (a) OH2 free or coordinated to V.

a b

Fig. 2. a) Crystal structure of VOHP04.4H 0 ; the double chains (see text) are
perpendicular to the figure ; b) Layers ?010) of VOHP04.0.5H20. Same legend
as above, plus : (0) OH coordinated to P.
Fig. 3. Crystal structure of : a) VO(H2P04) ; b) VO(P0 )2. Vanadyl and me-
taphosphate chains are perpendicular to the F igure. Same ?egend as fig. 2 ; va-
nadium atoms are not shown.

2 - A pair of octahedra linked by an edge in the equatorial plane, or by a

face in the particular case of VOHP04.0.5H20, is connected to tetrahedra.
The framework of VOHP04.0.5H20 is related to that of VOHP04.4H20 if one con-
siders the removal of 3.5 H20 molecules and the connection of the double chains
already described, in such a way that pairs of face-sharing octahedra linked by
corner-sharing tetrahedra are obtained (see fig. 2b). In each pair, the V=O are
in -cis position and the apical oxygen shared between the two vanadium atoms is
from a water molecule. The layers (VOHPO,)_ are held tightly by strong hydrogen
bonds between (HP04) groups and such water molecules of the next layers (refs.
22, 31).
Gorbunova et al. (ref. 28) solved the structure of (V0)2P207 with the resolu-
tion factor R~0.081. The framework consists of sheets made with pairs of edge-
sharing VO6, with trans V=O bonds, equatorially linked to tetrahedra (see fig.
4a). These sheets are stacked upon each other, as to form double columns of dis-
torted VO6 (chains O=V...O=V) parallel to bent pyrophosphate groups (see fig. 4b).
Owing to the low symmetry of the crystal, four crystallographically different
sites of V and P exist ; four values of vanadyl bond ranging from 1.53 to 1.73 \
two kinds of configuration of phosphate and two values of the P-O-P angle, 160'
and 145", are found. Antiphase boundaries and stacking disorder have been alrea-
dy observed by us by electron microscopy (refs. 16, 25). The bad value of the
resolution factor (R =0.198) obtained in an other work performed on (VO)2P207
could be explained to some extent by such defects (ref. 27). Comparable values
of the cell parameters and of the V-O and P-O bonds were found, but only one
kind of P-O-P angle (150") and a different distribution of the pyrophosphate
groups as compared to the sheets. Even if details of the structure are incorrect,
 T a

! (view along j021jI

b
a
Fig. 4. Crystal structure of (VO)2P207 according to ref. (28). a) (109) plane ;
b) (001) plane. \I and P atoms are not shown. Same legend as above.

this determination provides informations on how disordered lattices can be ar-

ranged. As it will be seen further, these discrepancies could be due to diffe-

rent methods of preparation of (VO)2P2~7, which have not been precised in ref.

(281.
The structural arrangement of (~0~2P20, is very particular. As it can be em-
phasized by calculations of the bond strength by Brown's method (ref. 411, the
four crystallographic sites occupied by vanadium correspond to four formal oxi-
dation states ranging from +3.69 to +4.44. In each pair of edge-sharing octahe-
VOf2) are found the couples C3.69, t4.19 and +3.79, t4.44, that
is something like "redox" couples inside the crystalline matrix itself. This

feature is also found in other compounds, such as V6OI3 (which is involved in

o-xylene oxidation catalysts), the structure of which is made with simple zig-
+4.34
zag of corner-shared VO6, with V v and pairs of edge-sharing octahedra, with
,,+4.52 t4.27
and V .
Although single crystals of 6 and y-VOPO4 have not yet been obtained* a
structural model has been proposed (ref. 291, which takes into account the con-
ditions due to stoichiometry and geometrical arrangement of VO6 octahedra and
PO4 tetrahedra, and is based on the structure of VOAs04 found by middlemiss
(ref. 27). In each pair of edge-sharing VO6, the V=O bonds are in
their oxygen uni-coordinated and each opposite oxygen is shared with a tetrahe-
dron uni-coordinated and each opposite oxygen is shared with a tetrahedron of
the upper (or the lower) sheet (see fig. 5). The two forms 6 and y would differ
in the relative organization of the layers. Such a kind of framework allows
stacking disorder which could explain the broadness of some XRO lines as experi-
mentally observed in powder patterns of 6-VOP04 (ref. 30).
At last, the P/V=1 phase of mean stoichiometry VP04 ,5 is considered as an
intergrowth structure made with the two kinds of structural unit, corresponding
 b

fig. 5. Hypothetic fra~work of y-VOP04. a) (010) plane ; b) (100) plane.

Same legend as above.

to, as well Z(a-VOP04) as 2(!3-VOP04), alternating with one (V~)2P2~, (see

fig. 61, as allowed by the coherent character of their interfaces (vide intra)
(ref. 16).

Fig. 6. Hypothetic framework of VP04 ,5 made from (001) a-VOP04 and

(100)(VO)~P20~ slabs,
.

The cell parameters and space groups of the structures and the X-ray diffrac-
tion patterns are gathered in Tables 1 to 5, together with the length of the
short V=O and long V...O bonds and other characteristics such as cleavage planes,
when available.
TABLE 1
Crystal data of vanadium-phosphorus-oxide compounds

Cell parameters V v=o v...o

Space z
Compounds Symmetry point Comments nefs.
group (A) (1) (4)
symm.

VOP04.2H20 tetrag. 2 a =6.202 1.567 2.233 yellow bright platelets 17

':h c4v
P4/nmm c= 7.410 cleavage (001)

VOP04.2D20 tetrag. C3 2 a=6.2154 1.42 2.50 resolution factor 0.186 18

4h C4
c= 7.4029 cleavage (001)
P4/n

a1-VOPO4 tetrag. yellow bright hygroscopic 15, 20

c4v
platelets. Cleavage (001)

"*T-voPo4 tetrag. a=6.014 1.580 2.857 yellow plates. Cleavage 19

C4
c =4.434 (001)

B-VOPO4 orthorh. 4 a=7.770 cS 1.566 2.591 yellow powder. Cleavage 21

4;
b=6.143 (201)
Pnma
c=6.965

6-VOPO4 orthorh. (?) 4 a=6.42 (2) dark-yellow hygroscopic 29

b=6.26 (2) 6#90" powder
c=9.09 (1)

v-VOPO4 monocl. (?) a=9.643 lemon-yellow powder 29

b=15.33 !3=93.04"
c=16.62

VOHP04.4H20 triclin. C: 4 a=6.379 az79.95" blue powder 22

b=8.921 B=76.33”
pl c =13.462 y= 71.03”
507
TABLE 2. X-ray powder pattern of a-VOPO4-type compounds

VOP04.2H20 VOPO4.H20 aI-VOP04 cxII VOPO4

I/I0 dabs I/I, dabs . hk’ dcalc. I/I, dabs 1110 dabs
. hkT %alc. . “’ 'talc. . “’ dcalc.

W 7.45 001 7.410

vs 6.37 001 6.30
vs 4.760 101 4.756 VW 4.43 001 4.43
VW 4.335 I10 4.3&S--- VP 4.40 110 4.39 - - - w 4.37 110 4.38 W 4.25 I10 4.25
W 4.11 001 4.11
vvw 3.703 002 3.705 S 3.57 101 3.569
vvw 3.182 102 3.181 M 3.15 002 3.15
M 3.103 200 3.101-- - w 3.07 200 3.10- - -s 3.100 200 3.105 M 3.07 Ill 3.069
M 3.000 111 2.998 vs 3.01 200 3.007
VW 2.863 201 2.861 2.83 012 2.81
VW 2.833 II2 2.830----vi 2.53 112 2.56
vvw 2.49 201 2.489
VW 2.379 202 2.378 W 2.302 211 2.299
w 2.192 220 2.193- --VW 2.17 220 2.19 --- w 2.193 220 2.192 VVW 2.217 002 2.217
2.105 221 2.103 M 2.10 122 2.08 M 2.16 '220 2.126
M” I.995 301 I.992 W 2.005 300 2.005
M I.963 310 1.961---w I.97 310 1.964---S 1.960 310 1.960 W 1.961 112 I.966
W 1.900 311 I.896 S 1.902 310 1.902
w 1 .a02 302 1.805 1.327 301 1.825
VW I.74 203 I.74 : 1.748 31 I I.748
S I.554 400 1.551- -- VW I.57 400 1.55---s 1.550 400 1.550 M I.562 321 1.561
S 1.519 401 1.518 r! 1.517 222 I.499 S 1.503 400 1.504
M 1.461 330 1.462-- - ---w 1.461 330 1.461 S I.435 103 I.435
'VW 1.417 330 1.417
M I.385 420 I.387 M I.382 41 I 1.385

a = 6.202 i a = 6.21 i a = 6.20 "A a = 6.014 i

c = 7.410 ; c = 6.30 i c = 4.11 i c = 4.434 i

P4/nmm P4/nmm P4/nmm - P4/n

D4;1 - D4;1 - D4; - c4;
(15, 17) (15) (15, I61 (16, 19)

S : strong ; M : medium ; w:weak ; v : very

 509

TABLE 3
X-ray powder pattern of B, Y, 6-VOPO4

8-voPo4 y-VOPO4 6-VOPO4

S5.19 101 5.19 7 6.16 102 6.127 1” 4.54 032 4.54

S4.60 011 4.61 67 4.886 031 4.885 vs 4.03 111 4.021
W 3.96 111 3.963 40 4 346 032 4 '52 Y 3.69 012 3.678
W 3.89 200 3.885 18 4:258 207 4:;63 \'! ?.?I ’ '11
?O? ..L
I' 3.48 002 3.482 rI 3.13 ??r) ?.1'1
vs 3.40 201 3.393 1;: 4.151
?.910 504
221 4.149
3.9!E w 2.96 ??I ?.?6?
FE 3.18 102 3.178 62 3.838 049 3.834 VW 2.73 $13 2.727
vs 3.068 020 3.071 62 3.501 230 3.505 W 2.59 022 2.578
1" 2.974 211 2.970 11 3.442 213 3.449 w,br 2.15 I?4 2.143
K 2.640 121 2.643 45 3.214 223 3.214 \"I 2.03 Xl T.0'4
rrl 2.410 220 2.409 43 3.089 195 3.098 w ‘_ *8”? q’? 1 .I on-l I
r” 2.209 122 2.209 57 0.055 311 ?.15" \” ! .7x ‘1” ‘*-:*
L"
/ 2.172 013 2.: 7'L E
d ' 397 31? ' ?Qp ‘,I l_?f? PJ.Q ? crc

VS 2.093 112 2.092 F i:774 l?fT 21771 \‘J I *

477 33; 1.47"
W 1.993 203 1.993 12 2.726 14T 2.729 1'1 1.427 420 1.4?C
F: 1.961 031 1.964 16 2.698 105 2.697
W 1.947 400 1.942 4 2.613 337 2.618
W 1.701 104 1.700 10 2.487 335 2.484
W 1.638 114 1.638 24 2.448 16i 2.448
ww 1.608 232 1.607 25 2.415 341 2.410
f 1.534
1.539 033
214 1.538 3 2.338 x? 2.333
1.536 9 2.163 146 2.164
S 1.505 133 1.506 5 2.055 018 2.055
VW 1.459 332 1.458 9 2.005 108 2.006
VW 1.444 304 1.445 7 1.991 147 1.992
10 1.961 442 1.960

1.951 405
9' 1.901 272 1.949
1.901
6 1.859 307 1.860

a = 7.770 i a = 9.64(3) i a = 6.42(2)i

b = 6.143 i b = 15.33(5)i I?= 6.26(2)i

c = 6.965 i c = 16.61(8)i c = 9.09(l)ii

16
- Pnma 6 = 93.04" $ = 90"
D2h
(21) (30 1 (30 1

x tentative indexation ; unpublished results

S : strong ; M : medium ; w : weak ; v : very

TABLE 4
X-ray pattern of vanadyl-containing phosphates

"P04.75 “O(PO3)2

d hkl hkl d"c;,c

talc 1'10 d:bs hk' dcalc *'IO dabs *'IO dabs hk' dcalc

6.266 021 W 6.28 102 6.264 S 3.86 020 3.865 S 3.97 011 3.998
5.651 111 W 5.65 Ill 5.652 S 3.66 311 3.667 S 3.88 220 3.880
4.786 002 4.79 200 4.786 VW 3.28 220 3.287 3.470 130 3.470
4.069 131 vt: 4.07 113 4.070 VW 3.21 II? 3.193 1c1 3.217 211 3.231
3.863 200 vs 3.87 020 3.864 M 3.13 102,400 3.124 VW 2.777 301 2.784
3.133 042 VS 3.137 204 3.152 VW 2.98 311 2.993 VW 2.738 040 2.743
3.005 202 M 2.992 220 3.006 S 2.91 112,41_0 2.897 S 2.471 321 2.483
2.904 113 2.906 311 2.903 M 2.81 310,312 2.792 VW" 2.452 420 2.454
2.640 232 ; 2.655 223 2.641 VW 2.63 401 2.626 2.254 141 2.262
2.433 242 M 2.435 224 2.433 W 2.52 222,130 2.524 M 2.034 112 2.03.8
2.393 004 W 2.399 400 2.393 VW 2.39 302 2.392 VW 1.981 202 1.999
2.358 223 W 2.362 322 2.358 VW 2.09 402,600 2.085 M 1.947 051 1.954
2.267 331 W 2.271 133 2.268 M 2.02 331 2.018 M 1.941 431 1.940
2.203 171 W 2.204 117 2.202 M 1.98 132,223 1.987 v: 1.937 440 1.939
2.088 063 M 2.083 306 2.088 VW 1.92 612 1.925 1.832 060 1.829
1.976 224 W 1.981 422 1.976 VW 1.91 140 1.910 VW 1.804 610 1.804
1.931 400 W 1.934 040 1.932 VW 1.89 123 1.885 vvw 1.730 260 1.735
1.837 263 1.840 326 1.837 VW 1.88 232,601 1.880 VW 1.673 142 1.671
1.826 244 vi 1.826 424 1.826 VW 1.78 313,512 1.782 VW 1.632 360 1.636
1.644 442 VW 1.644 244 1.644 VW 1.58 242 1.581 VW 1.583 451 1.581
1.633 01 01 W 1.633 1010 1.633 VW 1.54 54i 1.537 W 1.519 460,152 1.520

a = 7.725 i a = 9.571 i a = 12.82 i; a = 10.973 i

b = 16.576 A b = 7.728 A b = 7.73 A c = 4.293 A
c = 9.573 A c = 16.568 A c = 7.02 A
5 5 11
- Pca2, - Pbc2, 8 = 103" - 14; md
c2v c2v c4v
(23) (27) (16) (241

* possible additive lines : 3.76 (210) VW, 3.50 (220) vvw, 3.29 (221) VW . . . ; ** tentative indexation
 511

TABLE 5
X-ray powder pattern of vanadyl acid phosphate and hydrates

VO(H2P04)2 VOHP04.4H20 VOHP04.0.5H20

M 6.34 110 6.331 S 7.44 011 7.414 3 5.901 110 5.867

S 3.99 002 3.982 S 6.51 002 6.503 100 5.719 001 5.689
s 3.580 121 3.577 VW 5.78 111 5.795 4.818 020 4.796
M 3.374 112 3.371 VW 5.44 012 5.441 4: 4.535 101 4.514
S 3.170 220 3.165 vvw 4.862 OT2 4.858 5 4.099 111 4.084
M 2.980 022 2.975 M 4.192 020 4.162 23 3.684 200 3.708
S 2.835 130 2.831 M 4.022 107 4.009 3.300 121 3.287
M 2.483 230 2.483 VW 3.875 103 3.872 :: 3.1.l6 201 3.106
W 2.373 231 2.371 vvw 3.700 172 3.705 35 2.944 211 2.955
VW 2.310 132 2.307 w 3.365 123 3.378 10 2.796 031 2.787
VW 2.215 123 2.213 vvw 3.260 004 3.252 27 2.663 102 2.656
M 2.098 141 2.095 M 3.185 211 3.187 7 2.615 131 2.609
W 2.005 240 2.002 vvw 2.954 200 2.959 5 2.567 112 2.559
W 1.993 004 1.991 vs 2.867 213 2.870 6 2.454 022 2.447
(UW 1.953 042 1.951 M 2.788 031 2.781 4 2.405 040 2.398
VW 1.945 104 1.944 VW 2.708 203 2.700 6 2.263 301 2.267
W 1.903 142 1.906 VW 2.640 IOT 2.643 5 2.234 231 2.228
VW 1.867 332 1.865 vvw 2.568 015 2.569 2 2.212 311 2.206
W 1.785 124 1.783 vvw 2.450 032 2.451 5 2.131 032 2.125
W 1.758 150 1.756 032 4 2.055 321 2.050
1.688 224 1.685 vvw 2.400 OlF 2.404 6 2.047 222 2.043
M” 1.683 143 1.681 vvw 2.283 2i-? 2.283 1 1.960 330 1.956
W 1.584 440 1.582 VW 2.234 225 2.234 7 1.903 241 1.898
VW 2.164 135 2.166 3 1.864 302 1.866
W 2.055 241 2.055 9 1.854 400 1.854
W 2.011 301 2.009 8 1.836 103 1.837
W 1.967 043 1.967 2 1.808 113 1.804
W 1.918 047 1.918
OB
W 1.811 3T2 1.811
VW 1.752 3i-i 1.752
VW 1.715 326 1.716
315 1.714

a = 8.953 i a = 6.379 i a = 79.95" a = 7.416

c = 7.965 i b = 8.921 i 6 = 76.33" b = 9.592
c =13.462 i y = 71.03" c = 5.689
8
- P4/ncc c; - PI- $3-P
D4h 2h mmn

(16, 281 (22) (31)

S : strong ; M:medium ; w:weak ; v : very

512

PREPARATION AND SOL10 STATE REACTIONS

1. Preparation of the precursors and other phases
Only the main methods used to prepare the precursors of the catalysts and the
results obtained by decomposition will be described here, although other ways,
particularly by solid state reactions, are often possible. Schemes 1 and 2 refer
mainly to the preparation of oxidized (Vt5) and reduced (Vt4) phases respective-

ly.
5+
1.1. Starting from V p recursors. Yellow bright platelets of VOP04/2H20 are
precipitated from an aqueous mixture of V205 and H3P04 (P/V =7.0) heated under
reflux (ref. 32) ; calcination in air or O2 ca 400°C yields pseudomorphic plates
of aI-VOP04 by a topotactic reaction as demonstrated by XRD and electron micro-
scopy (ref. 15). This form is quickly rehydrated if no care is taken and able to
intercalate alcohols or amines (ref. 33), as already noticed by Ladwig (ref. 32).

In air or 02, aI -VOP04 is transformed into aII near 720°C and then to 6-VOP04 ;
on the contrary, aII, which is not sensitive to rehydration, seems to be stabi-
lized by a small amount of V4+, and, according to ref. 19, is stable up to
1140°C (scheme 1).

(15,161 (34)
iX,-WIPCI, > %I *NPVC:-

* See text.

Scheme 1. Numbers in () refer to literature.

If aI-VOP04 is quickly heated in N2 at 76O"C, layered plates of (VO),P207 are

obtained (refs. 16, 25) (scheme 1). The most stable phase 6 can be obtained by
decomposition of aI or aII above 75O"C, by decomposition in air at 500°C of
"NPVC", i.e. (NH4)2(VO)2C204(HP04)2.5H20 (ref. 34) or by calcination of the re-
sidue obtained by drying a solution containing salts of vanadium and phosphorus
with P/V =l.O. The direct decomposition of "NPVC" in N2 yields (VO)2P207 at
550°C. The reduction of a or B-VOP04 under a reactive mixture C4-hydrocarbon/air
(400°C) or in nitrogen (PO2 = 10m5 atm) ca 760°C leads also to the pyrophosphate.
Isothermal equilibrium measurements performed on a or B-VOP04 and (VO)2P207 show
 513

that VP04 75 is formed at the same temperature when pO2 is close to 10" atm
(02'N2 mixture) (see fig. 7). This phase does not exist below 750°C and is not
obtained when (VO),P207 is reoxidized under these conditions (hysteresis) (ref.
16). The reduction of UT leads to samples of (VO),P207 characterized by a laye-
red texture but 6-VOPO4 is generally recovered by reoxidation in 02 and after
several redox cycles ea 75O'C (ref. 29). The form y-VOPO4 can be obtained by de-
composition in O2 of N~4VO(~H)PO4 (ref. 36) (scheme 2) after the formation of
aI-VOP04 in 02 (550°C) ; the formation of alI , promoted by the reducing action
of NH3 diffusing through the solid, is obtained also in N,
‘
at 6OO'C.

c
8
a

zf
--1
reduct. oxid.
tc of of
VOPO, rvop20,

-2 760 p * b *
#.A KA

-3

I:SSZ3homagenei
range
ty

x in VPOx

Fig. 7. Variation of oxygen pressure with composition for VP0 system

(P/V = 1.0). The diagram shows the existence of VP04 76 above 750 C, the range
of non-stoichiometryfor VP04a7S and VPO4.5 and the hysteresis phenomenon when
VP04 5 is reoxidized.
.

1.2. Starting from V4+ precursors. The blue tetra and hemihydrates of vana-
dyl acid phosphate are obtained in non aqueous medium (alcohols) by heating
VOPO4.2~20 or a mixture of V2OS and H3PO4 under reflux (scheme 2) (ref. 35). if5’
can also be reduced in V4+ by HCT and/or oxalic acid (ref. 22), the amount of
water present in the solution determinjng the hydrate obtained. By dehydration
of the hemihydrate, 6 and y-VOPO4 are obtained at 450°C in air and 680°C in O2
respectively, and y-VOPO4 is transformed into 6 at 780°C (ref. 29). The pseudo-
514

morphic layered "y" form of (V@)2P207 is formed directly by a topotactic reac-

tion in the range 500-750°C in nitrogen. When this "y" form is reoxidized in 02,
y-VOPO4 is recovered and several redox cycles can be performed at high tempera-
ture (700°C) before 6 begins to appear (refs. 35, 37).

Scheme 2

Blue acicular crystals of VO(H2P04)2 are precipitated (scheme 2) from a vis-

cous solution of V204 in H3P04 with P/V =lO.O at 180°C (ref. 38), but can also
be found during reduction of V205 in concentrated phosphate solution (wide AU-
pllu). The vanadyl metaphosphate VO(P03)2 is obtained by dehydration of this pre-
cursor above 500°C in air according to the usual mode of decomposition of diacid
phosphates of transition metals, and also by calcination of any mixture of vana-
dium and phosphorus with P/V>1.6 (ref. 25), or by precipitation from the prece-
ding solution heated at 350-400°C (refs. 24, 39). If the temperature is raised
to 45O"C, green trimetaphosphate of vanadium V(P03)3 crystallizes (ref. 24).
When the temperature of the bath is lower than 18O"C, some VOPO4.2H2O can be
formed besides VO(H2P04)2 leading to the parallel formation of VP04 75 whatever
the atmosphere of calcination (ref. 16).

2. Structure and reactivity

The effect of preparation is very important here because the pseudomorphic
relations established between a precursor and the final solid used as catalyst
are responsible for their particular reactivity. On the other hand, the particu-
lar mode of reduction (crystallographic shear planes) and the kinds of defects
known for the Re03-type oxides can be extended to the case of VP0 phases because
of the structural relation already outlined between them (ref. 15). These reduc-
 515

tion mechanisms have been established in the absence of any reactive mixture
(except oxygen) above the solid but, since only the structural features are in-
volved, no reason could be invoked not to use them in catalysis.
2.1. Effects due to the methods of preparation. The topotactic dehydrations
of VOP04.2H20 and VOHP04.0.5H20 lead to layered compounds which mainly present
their cleavage planes at the surface. This feature is not found with methods

which give B-VOPO4.

If the precursor used is VOP04.2H20, the layered forms CST or aII are stabili-
zed as compared to E-VOP04 and their reduction leads to (VO)2P20,. The layered
texture of this sample, as revealed by XRD (enhancement of intensity of the
(200)x lines) and TEM, is not retained when several redox cycles cu 760°C are
performed, since various crystallographic (hkl) planes and not only (200) planes
are seen by TEM at the end of the experiments (refs. 16, 25). Consequently,
"6"-(VO)2P207 is recovered by reduction while y-VOPO4 is never obtained by re-
oxidation in this case. Pseudomorphism was also shown between the planes (010)

of VOHP04.0.5H20, (100)x of the layered form "Y"-(VO),P,O, and (010) of y-VOPO4,

as a consequence of topotaxy (refs. 32, 35). These reactions proceed with rela-
-1
tively low activation energies : 21.4 and 29.7 kcal.mol for the dehydration

and 36 and 45 kcal.mol -' for the reduction of y-VOP04. The comparison between
the reactivity, XRD pattern, IR and UV spectra and external morphology of the

two "forms" 6 and y-(VO)2P207 (ref. 35) does not yet allow the determination of
whether they have distinct (but closely related) structures or if one is only
less disordered than the other. The occurence and displacement in the IR spec-
trum of some bands which reflect the bond lengths would support the first hypo-
thesis perhaps corresponding to the two solutions given for the structure (refs.
27, 28). However, the differences between (VO)2P207 coming from a or 6-VOP04
lead us to think that two kinds of redox systems, "B" and "y", could exist
(refs. 35, 37). Work is still in progress on this subject.
When P/V >l.O, the complexity increases because the effect of preparation may
be preponderant. The clearest point is that, besides a, B-VOPO~ or (VO)2P207
(oxidizing or reducing conditions), VO(P03)2 is easily obtained if 1.6~ P/V ~2.0
in the initial mixture, or by decomposition of VO(H2P04)2. VO(P03)2 has nearly
no influence on the redox process (ref. 16). However in such conditions com-
pounds such as aII-VOP04 and VP04.75
( according to local p02), the formation of
4+
which is favored by the presence of V , may appear during the calcination or
during catalysis.
2.2. Mechanisms of reduction of the different forms of VOP04. As far as the
reduction of a and B-VOP04 is concerned, the crystallographic shear plane (C5)

(*I : notation ref. (28).

516

mechanism found for Re03-type oxides (ref. 40) is the only one able to account
for (i) the same behavior of a and 5, (ii) the formation of pairs of edge-sha-
ring octahedra and of corner-sharing tetrahedra (P207) from single VO5 and PO4
groups respectively, and (iii) the existence of VP04 ,5 obtained equally from a
or 8 under the same conditions (ref. 25). Oxygen vacancies diffuse first, then
aggregate into planar discs and form elements of shear planes, even at the sur-
face. If the vacancies diffuse cooperatively along a given direction, the CS

planes which result are ordered, giving rise to definite structures and not just
extended defects. Considering now the cleavage plane (001) a-VOP04, the coopera-
tive movement of VO5 in the I1101 direction results in their edge-sharing (in
which the V=O bonds are already in osition), while PO4 groups rotate to

fuse as P20, in the upper and lower sheets (see fig. 8a). From B-VOPO~ the same

cooperative movement is considered in the cleavage plane (201) perpendicular to

V=O bonds, along I2111 (see fig. 8b). For the same geometric reasons, slabs of

a b --la R

Fig. 8. Reduction of VOPO4 to (100)(VO)2P207 by CS mechanism. a) (OOl)a-VOP04 :

sharing of VO5 octahedra in the direction Ill01 ; b) (POl)B-VOPO : sharing of
VO5 in the direction I2111. Double circles correspond to V-O-P bon ds to be bro-
ken. Idealized drawings.
 517

(001)a and (201)B match closely with (100) of (VO)2P207 and the crystallographic
misfits are 2 and 3 % respectively for CI and 6. Consequently, not only can
VP04 ,S, which is a definite phase with its own X-ray pattern (ref. 16), be
thought of as an intergrowth of sheets of (201)% or (OOl)~-VOP04 and
(O01)(VO)2P207 with the mean formula 2VOP04,(VO)2P207 (see fig. 6), but any so-
lid mixture obtained at the steady state with stoichiometric composition between

VP06 and VP04_5 can be so pictured. For compositions close to the preceding, ex-
tended defects of the same kind can be envisaged. We have shown earlier by TEM
the validity of these models (ref. 16) used later by others (ref. 42). Conside-
ring the case of isothermal reoxidation of (VO)2P20,, which does not proceed in
the same conditions of ~02 as evidenced by hysteresis (see fig. 7), one can con-
sider that the replenishment of surface vacancies by oxygen is followed by the
diffusion of these species into the bulk, ensuring the disappearance of CS
planes and the restoration of the initial oxidized phase. Such a mechanism,
which is not exactly the reverse of the reduction, accounts for the observed
hysteresis and for the difficulty to come back to a fully stoichiometric VPOS
composition.
Coherent interfaces are therefore possible not only between all these forms,
but also in presence of VO(P03)2. Owing to its great stability, the latter can
4+
only have an influence on the total number ov V ions and on the degree of dis-
order of the other phases by two means : the positional disorder of V in the VO6
of vanadyl chains (ref. 27) which are present and able to match with those of
the other phases, and a disorder in the arrangement of tetrahedra due to the
presence of polyphosphates besides PO4 or P207. Such polyphosphates have been
detected by IR spectroscopy in modified V-P-O catalysts (ref. 43).
In the reduction of y-VOP04 to (VO)2P207, no CS mechanism needs to be consi-
dered because their frameworks are made with the same elementary units (pairs of
octahedra). In this kind of gliding mechanism, only small displacements of the
atoms laying in planes (010)~ are needed : the result of their movement along
(a/6, c/41 is the connection of PO4 and formation of columns of VO6 pairs (refs,
5+
35, 37). It is very easy here to consider the case where some V species are
4+
incorporated in (VO)2P20, (or the reverse, some V in y-VOPO,) by the local
formation of microdomains with the same structural unit, Reoxidation proceeds
probably by the reverse but slower mechanism.

APPLICATION TO CATALYSIS
Because catalysis is essentially a dynamic phenomenon, many difficulties
arise when one tries to determine which phases are actually present, how they
work in the oxidation of C4-hydrocarbons, and what is the influence of crystal
structures. Moreover, some experiments in the literature give contradictory re-
sults, which can often be partly explained by the use of different preparations
518

(refs. 44-46). Summarizing the results found to be in accordance in the litera-

ture (refs. 47-56) and in our own work, selectivity in maleic anhydride would be
connected with the observations listed on Table 6.

TABLE 6
Characteristics of VP0 catalysts according to C4-hydrocarbon oxidation

hydrocarbon butene n-butane

precursors obtained by : scheme 1 scheme 2

steady state V5+ , v4+ mainly V4+

phases found (P/V = 1.0 a or B-VOPC4

to 1.2 1 + (VO)2P207 (VO)2P*O7(x)

(P/V = 1.6) same as above

\ + vo(Po3)2
II 0 1, 11
kind of redox B Y
mechanism of reduction CS plane gliding plane

(x) more or less "amorphous".

According to kinetic studies made with V-P-O catalysts (refs. 47, 54-57), be-
sides the route leading to carbon oxides and water n-butane is directly conver-
ted to maleic anhydride (FCA), which is further partly decomposed, whereas conse-
cutive steps (two or more), at least butene-butadiene and butadiene-F!A (without
decomposition of MA), are found for butene conversion. Sutene and even butadiene
are responsible to a great extent for the formation of by-products (other than
CO, CO2 and H20), the absence of which is certainly significative in the oxida-
tion of butane, At first sight it can be argued that the selectivity in maleic
anhydride from butane is mainly governed by the ability of the catalyst to hin-
der the desorption and the non-selective oxidation of butene or butadiene. Once
adsorbed on the surface, butene or butadiene should find suitable sites to be
oxidatively cyclized by taking three oxygen atoms, this last "step" proceeding
on each kind of catalysts. Another hypothesis, not yet considered because kine-
tics of.butane oxidation were not determined on catalysts suitably prepared for
this reaction, but now supported by the recent work of Pepera et al. (ref. 51),
could be that the two mechanisms are different, the rate-determining step being
the attack of methylene (and not methyl) group. This outlines the necessity for
the organic molecules to keep the symmetry C2v all along the transformation from
butane to maleic anhydride through 2-cis butene (see fig. 9), already invoked in
the case of CoMo04-MoOS/Ti02 (ref. 58). This keeping of symmetry is also neces-
sary in butene oxidation, but numerous experiments have shown that the catalyst
 519

Fi[.=Ckj -fF
- / \ =+ \ +
,-.
ti,C ,1 LHB ",C r-H,
_a
IL
1 -butene Cl trans-butadiene Cp,.,

Fig. 9. Keeping of C2v symmetry in C4-hydrocarbon oxidation

is not able to oxidize selectively 2-cis butene as compared to the other iso-
mers. The challenge is therefore to explain how the specific structural proper-
ties of the solids account for these differences without forgetting that the two
factors, (i) reducing power of the reactive mixture hydrocarbon/air and, (ii)
nature of the solid phases, interact directly. It must be noticed that, since
the preparation and the redox system involved are also related, it is not sur-
prising that catalysts prepared according to scheme 1 are not active in butane
oxidation while catalysts prepared by scheme 2 give mainly carbon oxides in bu-
tene oxidation (refs. 47, 53).

1. Configuration of the surface of the catalysts

At this point it first seems necessary to have an approximate idea of what
the surface of the catalysts could be (provided that they be prepared by the
right method for each reaction) at the steady state, even neglecting the surface
atomic steps and kinks and dislocations which have effectively been observed on
V205 and Moo3 by TEH (refs. 59, 60). For butane oxidation the surface corres-
ponds mainly to the plane (100) of (VO)2P207, which includes local microdomains
of y-VOP04 formed by surface topotactic reoxidation (ref. 61). This disposition,
and the dynamic effect due to the set of reduction-oxidation reactions which oc-
cur in catalysis, account for the observed amorphous surface. The "redox" couple
found in pairs of neighboring vanadium atoms by Brown's method could correspond
to the electronic interactions between the two vanadium atoms as confirmed by
EPR (ref. 51), which favors in turn the positional disorder of V already noticed
(u& &qJ&ul.
Let us assume that, on the surface of (VO)2P207, the trans disposition of V=O
in the pairs of octahedra is retained : on the top of one vanadium is an oxygen
vacancy (Lewis-acid site) which can be replenished by gaseous oxygen, water pro-
duced in the reaction, or OH (Bronsted-acid site). The next V is covered by 0
520

(VO)*P207 ; if-VOPO4

Fig. 10. Model of (100)(VO)2P207 showing the various oxygen sites on the surface.
Same legend as in fig. 2.

of (O=V)*+ (see fig. 10). Five kinds of oxygen sites exist, as confirmed by in-
frared spectra of adsorbed species on (VO)2P207 (ref. 62) ; following increasing
bond strength are found in the series O...(V) (this one being normally abstrac-
ted), (V)2-O-(P), (V)-O-(P), (P)-O-(P) which gives hydrogen-bonded OH groups on
the surface (ref. 62), and O=(V). The bonds (V)*-O-(P) and (V)-O-(P) are also
found on the surface of y-VOPO4, the actual values of bond strength of which are
not known but are certainly higher than for (VO)2P20,. A problem is raised by

the presence of phosphates which are at the same height as the oxygen O=(V) uni-
5+
coordinated to V . If they are abstracted at the surface, there is a local de-
ficit of PO4 which destabilizes y-VOPO4. The excess of phosphate when P/V =1.2
would help to keep this necessary framework, and also contributes to the disor-
der and to the enhancement of activity. The better performance observed in in-
dustrial processes should be interpreted in this sense, because when P/V=1 the
activity drastically decreases with the amount of P, although the y redox is
still present.
In the case of butene oxidation, (VO)2P207 and a or 6-VOPO4 are found, con-
sisting of microdomains more extended than in the above case (mean stoichiometry
VP04 67) and connected by interfaces (refs. 16, 25), the coherence of which is
necessary in order that species such as oxygen, electrons and vacancies be
transferred between two solid phases (refs. 8, 14). Moreover, the activity of
the sites is enhanced as interfacial energy keeps the lattices metastable (refs.
14, 16, 25). This certainly arises when the composition is P/V =1.6 because co-
herent interfaces exist also between VO(P03)2 and the other phases. As a result,
activity increases but selectivity to MA remains approximately constant (refs.
25, 46). It has been demonstrated recently for Bi-Ce-MO-0 systems that in such
cases the free energy of the whole system is minimized at the composition giving
the maximum activity (ref. 63). If the number, and quality of sites of (VO)2P20,
are considered as identical to those of butane catalyst (although the contribu-
 521

tion of the sites of other crystallographic planes should be also taken into
consideration), (Y or 5-VOPO4 certainly present different oxygen and V5+ sites
than y-VOPO4. By adding the fact that butane needs strong-acid sites to be dehy-
drogenated, and butene sites allowing adsorbed r-allylic complex, this would
make the final difference between the two reactions. Conversely, it is also ne-
cessary to examine if analogous sites can be found on each kind of catalyst in
order to account for the similarity of the 'step" butene-EA.

2. Kinetics and mechanism

It has been shown on.VPO catalysts prepared for butene oxidation that the se-
lectivity of the step furan-MA is about 50 % fiefs. 54, 641, and that 2.5-dihy-
drofuran (DHF) is an inte~ediate leading selectively to PA on Mo03/Ti02 cata-
lyst (ref. 64).In the oxidation of butane, the question of butene (or butadiene)
as a necessary intermediate seems, at first sight, not so important because ki-
netic reasons can explain when it is not detected (refs. 47, 51). However, the
comparison with molybdate catalysts in the same reaction (refs. 49, 58, 651, the
kinetics of which are controlled, in the case of CoMoO4, by two consecutive
first order reactions (dehydrogenation of butane and decomposition of I?A) (ref.
66), should be also fruitful. Let us discuss now the mechanism of these reac-
tions in the light of the preceding remarks, as if different steps would actual-
ly occur.
2.1. Butane-maleic anhydride through butene (and/or butadiene~, Pulse reac-

tor experiments have shown that the y form of (VO),P,O, and isostructural B-mo-
lybdates MMo04 (M = Co, Mn, Fe) are selective in the step butane-butadiene even
in the absence of oxygen (refs. 49,57,65), whereas Bi2Mo06 is inactive. A care-
ful examination of their crystal structures provides evidence that the (110)
cleavage plane of 6-M~oO4 is strikingly related to (100) of (VO)2P207 because it
can be described with analogous elementary units, i.e. pairs of edge-sharing
(M06), linked to (Mo04) instead of (PO41 tetrahedra (ref. 14). This also implies
variations of the formal oxidation state of MO in these molybdates analogous to
those found for V in V6013 or (VO)2P207. For inactive Bi2Mo06 only corner-shared
Moo6 are found, SO ability to dehydrogenate butane selectively could be correla-
ted with this kind of bi-dimensional framework exposed at the surface. To be ac-
+3.7_v+4.4 in
tivated, butane needs strong-acid sites (refs. 46, 62) such as V
(VO)2P207, situated near suitably bonded oxygen atoms able to catch the hydrogen
of CH2 groups as recently demonstrated (ref. 51).
The oxidation of n-butane to oxidized C4-compounds and especially BA is scar-
cely observed with transition metal molybdates : only the preceding 6-YMo04 can
yield maleic anhydride, though in small amount because MA is partially converted
in CO2 (ref. 49). Consequently, supplementary conditions that are not found in
molybdates, except in the special case of supported Cot!o04-fJo03/Ti02 (ref. 58),
522

are necessary in order that dehydrogenated butane undergoes oxidative cycliza-

tion on the surface of the catalyst. The fact that the step butadiene-MA depends
on ~02 means that in VP0 catalyst the mean oxidation state of vanadium at least
locally (ref. 67) must be higher than +4. This was observed experimentally by
various authors, who found that only a few subsurface layers were affected by
reoxidation (refs. 47-52, 61, 68). Surface superoxides formed on (VO)2P207 being
recently discarded (ref. 51), we consider that the local and superficial reoxi-
dation of y-(VO)2P207 produces microdomains of y-VOP04, too small to be detected
by physical methods. As just explained IuLcfe dupha), this process is facile and
does not require a large modification of the solid, although it contributes cer-
tainly to the disordering in the layers observed by XRD or RPE (refs. 30, 47,
51).

As far as the adsorption of species is concerned, several hypothesis can be

proposed, which involve mainly the oxygen sites situated on top of vanadium and
phosphorus (see fig. 11).

or

1 bis 3 4

Fig. 11. Possible mechanisms of the formation of MA on an elementary unit of

(VO!gP207 or y-VOP04 (see text). Step 1 bis corresponds to allylic adsorption on
aV center. Hydrogen atoms are not shown ; oxygen not at scale.

+4
1. n-butane is "activated" on a V site, with breaking of the C-H bond of
methylene and of methyl groups, the hydrogen of which is caught by OH-(V=O) of
+4
the V pair and OH-(P) respectively. Adsorbed l-butene or 2-butene can result.
After the desorption of H20 formed, the vacancies (O)-(P) and (O)...(V=O) are
replenished by gliding of the layer below the surface (and eventually by oxygen
of air), thereby producing a microdomain of y-VOP04.
2. Alternatively the preceding mechanism could proceed on both Vt4 centers,
yielding directly adsorbed butadiene trans or a.
 3. The adsorbed -cis butene or -cis butadiene (C2v symmetry) are cyclized by
closing the ring on a O-(P) of y-VOP04, yielding adsorbed DHF or furan respecti-

vely.
4. The formation of maleic anhydride is obtained after the dehydrogenation of
=CH- of furan by acid phosphate which is then able to take oxygen adsorbed on
top of vanadium +5. When MA desorbs, the vacancy due to the loss of water on
phosphate is refilled by means of pyrophosphate formation, vacancies exist on

top of vanadium and (VO),P207 is recovered. If butene were directly adsorbed on

such a surface, it would be mainly broken by taking the loosely bonded oxygen
4+
often adsorbed over an acidic V site ; this is experimentally observed when

the oxidation of butene is attempted with y catalysts (ref. 67).

2.2. Butene-MA. The rate-determining step is the production of an allylic
+5
intermediate by adsorption of butene on V sites followed by abstraction of an
a-hydrogen (refs. 69, 70). Hydrogen of the allylic intermediate hydrogen is lost
faster than oxygen is inserted (ref. 68) if suitable pO2 are used, which deter-
mines the occurrence of oxygenated by-products. This accounts for the necessary
+5
presence of oxygen in the gas phase and V sites in VOP04, which must be nume-
rous enough in the catalyst in order to activate and oxidize butene (refs. 48,
49). As both parameters are related to each other, the partial pressure of oxy-
gen in the mixture butene/air favors the B redox. The greater reducing power of
explains why a catalyst prepared for oxidation of n-butane is deactiva-
n-C4HlO
ted in presence of butene (ref. 47). The consecutive decrease of the surface
area observed in this case would therefore be due to the partial reoxidation of
(VO)2P207 by replenishment of oxygen vacancies and local formation of VOP04. If
we examine the experimental results (ref. 53), the interaction of butene-air
with such a layered y-form results in the formation of aII besides (VO)2P2D7 and
y-VOP04 ; the latter phase does not exist in the spent catalyst. The pseudomor-
phic form aII obtained by this method is induced by layered y-(VO)2P207 with the
resulting catalyst more active than B-VOP04. The surface planes containing the
active sites are indeed more numerous since the layered character is maintained
in the corresponding microdomains (OOl)a,I/(100)(VO)2P207. Why aII instead of CX~
is obtained in such conditions is not yet clear, although its formation was
shown to be promoted by the presence of V 4t (refs. 20, 61).
Consequently, in catalysts suitably prepared for butene oxidation, CL, 6-VOP04
in sufficient amount as to be detected by XRO and 6-(VO)2P207 are thought to be
present at the steady state, and single and double octahedra coexist at the in-
terface of the microdomains. If we argue that cis-2 butene is adsorbed in place
of a vacancy situated on Vt5 (single octahedron), and that a hydrogen of CH3 is
caught by OH-(P) (see fig. ll), then the n-allylic intermediate is further dehy-
drogenated on the same kind of OH to give butadiene. If 1-butene is considered,
methylene and methyl groups could be dehydrogenated in the same way proposed for
524

n-butane, that is by (VO)2P20,. This would justify the observations (ref. 62)
that butene-butadiene conversion proceeds on (VO)2P20,. Butadiene desorbs if it
cannot find an oxygen (V)-O-(P) or O(PO3) of CI, B-VOPO4 in the vicinity, on
which the ring can be closed to give furan or DHF. According to the bond
strength of O...(V=O) adsorbed on (VO)2P207 or on a, B-VOPO4, butadiene would
burn in the first case, or yield crotonaldehyde in the second case. The end re-
sult is the same as for n-butane. It must be recalled at this point that these
kinds of oxygen (V)-O-(P) or O(PO3) are effectively those extracted during a re-
duction by CS mechanism, If the n-allylic intermediate extracts an oxygen

(V)-O-(P) of a, B-VOPO4 instead of being dehydrogenated, it will be preferen-

tially oxidized yielding by-products such as ketones or C3 or C2 acids. On the
other hand, the various crystallographic planes presented on the surface by
these 8 phases (ref. 35) can account by themselves for the obtention of by-pro-
ducts, as an application of the concept of structure sensitive reactions,

CONCLUSION
Although the above picture is somewhat simplified and needs calculations and
more experiments to be checked, it accounts for most results obtained on these
reactions. The selective oxidation of butene should be associated with the pre-
+5
sence of V in single octahedra and with redox of "6" kind implying the reduc-
tion of the solid by CS mechanism, whereas the selective oxidation of butane
+4
should be associated with the presence of V in double octahedra and "y" redox.
As usually encountered in catalysis, when one tries to correlate structural and
catalytic properties, the difficulty is to consider separately working factors
which are intricately mixed up. In our opinion, however, the examination of both
the structure and reactivity of the phases involved in the preparation and in
the catalytic process is necessary. Any study of kinetics, spectroscopy, tempe-
rature-programmed reduction, etc... to be correlated with catalytic performance
should be performed on catalysts prepared in the right conditions, including ac-
tivation. We hope to have contributed to a better understanding of the influence
of these factors in this intricate but thrilling V-P-O system.

ACKNOWLEDGEMENT
Thanks are due to Professor P. Courtine for valuable and enthusiastic dis-
cussions.

REFERENCES

1. M. Boudart, Adv. Catal., 26 (1966) 153.

2 J.C. Volta and B. Moraweck, J. Chem. Sot., Chem. Comm., (1980) 338.
3 J.C. Volta and J.M. Tatibouet, J. Catal., 93 (1985) 467.
4 Y. Ziolkowski, J. Catal., 80 (1983) 263.
5 G.C. Bond and R. Burch, Catalysis, 6 (1984).
 525

Stud. Surf. Sci. Catal., 11 (1984), ref. therein.

A. Vejux and P. Courtine, J. Sol. State Chem., 23 (1978) 93.
8 J. Papachryssanthou, E. Bordes. A. Vejux, P. Courtine, R. Marchand and
M. Tournoux, Catal. Today, 1 (1987) 219-227.
9 Y. Murakami, M. Inomata, K. Mori, T. Ui, K. Suzuki, A. Miyamoto, and T. Hat-
tori, in G. Poncelet, P. Grange and P.A. Jacobs (Eds), Preparation of Cata-
lysts, III, Elsevier, 1983, pp. 531.
10 M. Inomata, K. Mori, A. Miyamoto and Y. Murakami, J. Phys. Chem., 87 (1983)
761.
11 A. Vejux and P. Courtine, J. Sol. State Chem., 63 (1986) 179.
12 P. Courtine and A. Vejux, 9th Europ. Chemistry of Interfaces Conference,
May 19-25, 1987, Zakopane, Pologne.
Z.C. Kang and 9.X. Bao, Appl. Catal., 26 (1986) 251.
:; P. Courtine, ACS Symp. Series, 279 (1985) 37.
15 E. Bordes, P. Courtine and G. Pannetier, Ann. Chim. Paris, 8 (1973) 105.
16 E. Bordes, These d'Etat, Compiegne, France, 1979.
17 H.R. Tietze, Austral. J. Chem., 34 (1981) 2035.
18 N. Tachez, F. Theobald, J. Bernard and A.W. Hewat, Rev. Chim. Mine., 19
(1982) 291.
19 R. Gopal and C. Calvo, J. Sol. State Chem., 5 (1972) 432.
20 M. Tachez, F. Theobald and E. Bordes, J. Sol. State Chem., 40 (1981) 280.
B. Jordan and C. Calvo, Can. J. Chem., 51 (1973) 2621.
;: M.E. Leonowicz, J.W. Johnson, J.F. Brody, H.F. Shannon and J.M. Newsam, J.
Sol. State Chem., 56 (1985) 370 ; J.W. Johnson, D.C. Johnston, A.J. Jacobson
and J.F. Brody, J. Am. Chem. Sot., 106 (1984) 8123.
23 S.A. Linde, Y.U.E. Gorbunova, A.V. Lavrov and V.G. Kuznetsov, Dokl. Akad.
Nauk SSSR, 245 (1979) 584.
24 B.C. Tofield, G.R. Crane, G.A. Pasteur and R.C. Sherwood, J. Chem. Sot.,
Dalt. Trans., (1975) 1806.
25 E. Bordes and P. Courtine, J. Catal., 57 (1979) 236.
26 V.V. Krasnikov and Z.A. Konstant, Russ. J. Inorg. Mater., 15 (1980) 1701.
27 N. Middlemiss, Ph.D. Thesis, McMaster Univ., Hamilton, Canada, 1979.
Yu.E. Gorbunova and S.A. Linde, Dokl. Akad. Nauk SSSP, 245 (1979) 5864.
;: E. Bordes, J.W. Johnson, A. Raminosona and P. Courtine, Mater. Sci. Mono-
graph, 288 (1985), 887.
30 J.W. Johnson, D.C. Johnston and A.J. Jacobson, Proc. 8th Int. Congress on
Catalysis, Berlin, preprint B 5.1 (1984).
31 C.C. Torardi and J.C. Calabrese, Inorg. Chem., 23 (1984) 1308.
32 G. Ladwig, Z. Anorg. Allg. Chem., 338 (1965) 266.
33 J.W. Johnson, A.J. Jacobson, J.F. Brody and S.M. Rich, Inorg. Chem., 21
(1982) 3820.
34 S. Pulvin, M. Ronis and P. Courtine, C. R. Acad. Sci. Paris, 283 C (1976)
585.
35 E. Bordes, J.W. Johnson and P. Courtine, J. Sol. State Chem., 55 (1984) 270.
36 S. Pulvin, E. Bordes, P. Ronis, and P. Courtine, J. Chem. Res., (S) (1981)
29 ; (N) (1981) 362.
37 E. Bordes and P. Courtine, J. Chem. Sot., Chem. Comm., (1985) 294.
38 G. Ladwig, Z. Chem., 8 (1968) 307.
39 A.V. Lavrov. L.S. Guzeeva and P.M. Fedorov. Izv. Akad. Nauk SSSR. Yeoro.
Mat., 10 (1974) 2180.
40 B.G. Hyde and L.A. Bursill, in Le Roy-Eyring and O'Keeffe (Eds), The Chemis-
try of Extended Defects in Non Metallic Solids, North Holland (1970), 347.
41 I.D. Brown and K.K. Wu. Acta Crvst., B 32 (1976) 1957.
42 I. Matsuura, 8th Int. Cong. on Catalysis, Berlin, (1984) Vol. IV, 4.73.
43 M. Brutovsky, S. Gerej and F. Vasilco, Gerejova, Coll. Czech. Chem. Comm.,
47 (1982) 1290.
44 J.S. Buchanan, J. Apostolakis and S. Sundaresan, Appl. Catal., 19 ( 1985) 65.
45 T.P. Moser and G.L. Schrader, J. Catal., 92 (1985) 216.
46 M. Ai, J. Catal., 100 (1986) 336.
47 G. Centi, G. Fornasari, and F. Trifiro, J. Catal., 89 (1984) 44.
48 B.K. Hodnett and 6. Delmon, Ind. Eng. Chem. Fundam., 23 (1984) 465.
526

49 G. Centi and F. Trifiro, Appl. Catal., 12 (1984) 1.

50 B.K. Hodnett, Catal. Rev. Sci. Eng., 27 (1985) 373.
51 M.A. Pepera, J.L. Callahan, M.J. Desmond, E.C. Milberger, P.R. Blum and
N.J. Bremer, J. Am. Chem. Sot., 107 (1985) 4483.
52 Yu.A. Kruchinin, Yu.A. Mischenko, P.P. Nechiporuk and A.I. Gel'bshtein,
Kinet. Catal., 25 (1984) 328.
53 G. Centi, I. Manenti, A. Riva and F. Trifiro, Appl. Catal., 9 (1984) 177.
54 M. Ai and S. Suzuki, Bull. Chem. Sot. Jap., 47 (1974) 3074.
55 R.L. Varma and C.N. Saraf, Ind. Eng. Chem. Prod. Res. Dev., 18 (1979) 7.
56 F. Cavani, G. Centi, I. Manenti, A. Riva and F. Trifiro, ibid. 22 (1983)
565.
57 A. Escardino, C. Sola and F. Ruiz, An. Quim., 69 (1969) 385 & 1157.
58 J.S. Jung, E. Bordes and P. Courtine, in M. Che and G.C. Bond (Eds), Adsorp-
tion and Catalysis on Oxide Surfaces, Stud. Surf. Sci. Catal., 21 (1985).
Elsevier, pp. 345.
59 L. Fiermans, V. Vandenbroucke, R. Vandenberghe, J. Vennik and G. Dalmai, J.
Microsc. Spectrosc. Electron., 4 (1979) 543.
60 J.M. Dominguez-Esquivel, S. Fuentes-Moyado, G. Oiaz-Guerrero and A. Vasquez-
Zavala, Surf. Sci., 175 (1986) L701.
61 A. Raminosona, These, Compiegne, France, 1986.
62 S.J. Puttock, and C.H. Rochester, J. Chem. Sot., Faraday Trans. I, 82 (1986)
2773.
63 J.F. Bradzil, R.G. Teller, R.K. Grasselli and E.K. Kostiner, ACS Symp. Se-
ries, 279 (1985) 57.
64 M. Akimoto and E. Echigoya, 8~11. Chem. Sot. Jap., 48 (1975) 3518.
65 R. Boutry, J.C. Daumas and R. Montarnal, Bull. Sot. Chim. Fr., (1968) 4050.
66 R.A. Agasiev and T.N. Shakhtakhtinskii, Azerb. Khim. Zh., 14 (1969).
67 L. Morselli, F. Trifiro and L. Urban, J. Catal., 75 (1982) 112.
J.S. Buchanan and S. Sundaresan, Appl. Catal., 26 (1986) 211.
:; R.K. Grasselli and J.D. Burrington, Adv. Catal., 30 (1981) 133
70 D.J. Hucknall, "Selective Oxidation of Hydrocarbons", Acad. Press, 1974.