An electron diffraction and crystal chemical investigation of oxygen/flourine ordering in rutile-type iron oxyflouride, FeOF

Article abstract of DOI:10.1006/jssc.2000.8925

Rutile-type iron oxyfluoride, FeOF, has been synthesized by the reaction of FeF₃ with Fe₂O₃ in a sealed platinum tube at 950°C. The compound was previously believed to have a random distribution of oxygen and fluorine anions surrounding each of the Fe³⁺ cations. In this work, electron diffraction experiments have revealed the presence of a characteristic diffuse intensity distribution in the form of continuous rods of diffuse intensity running along both the [110]* and [110]* directions of reciprocal space through the (h+k+l) = odd parent rutile reflections. Fe³⁺ shifts induced by local O/F ordering are shown to be responsible for the characteristic reciprocal space intensity distribution of this observed diffuse scattering. The continuous 〈110〉* rods of diffuse intensity require the existence of orthogonal {110} planes within the parent rutile structure which exhibit long-range, two-dimensional, oxygen/fluorine ordering, but with no correlation from one {110} plane to the next. A crystal chemical explanation is proposed to support this argument.

Full text of DOI:10.1006/jssc.2000.8925

Journal of Solid State Chemistry 155, 359}365 (2000)
doi:10.1006/jssc.2000.8925, available online at http://www.idealibrary.com on
An Electron Diffraction and Crystal Chemical Investigation
of Oxygen/Fluorine Ordering in Rutile-Type Iron Oxyfluoride, FeOF
Frank J. Brink,*ꢁ- Ray L. Withers,*ꢁꢀ and John G. Thompson*
*Research School of Chemistry, Australian National University, Canberra, ACT 0200, Australia; and -Electron Microscope Unit,
Research School of Biological Sciences, Australian National University, Canberra, ACT 0200, Australia
Received June 26, 2000; accepted August 9, 2000; published online November 29, 2000
The FeOF transition metal oxy#uoride was "rst prepared
Rutile-type iron oxy6uoride, FeOF, has been synthesized by by Hagenmuller et al. (5) and was found to possess a rutile-
the reaction of FeF3 with Fe2O3 in a sealed platinum tube at
9503C. The compound was previously believed to have a random
type structure with cell parameters a"4.647(5) A and
c"3.048(9) A. The possibility of oxygen}#uorine ordering
distribution of oxygen and 6uorine anions surrounding each of
therein was "rst raised by Chappert and Portier (6, 7) as
the Fe3؉ cations. In this work, electron di4raction experiments
a result of their Mossbauer studies that indicated a single
have revealed the presence of a characteristic di4use intensity
unique environment for the Feꢃ> cation. They concluded
distribution in the form of continuous rods of di4use intensity
that there must be some sort of ordered distribution of
running along both the [110]* and [110]* directions of recipro-
oxygen and #uorine ions leading to a superstructure of
cal space through the (h؉k؉l) ؍ odd parent rutile re6ections.
Fe3؉ shifts induced by local O/F ordering are shown to be
responsible for the characteristic reciprocal space intensity dis-
tribution of this observed di4use scattering. The continuous
a rutile-type parent structure that is di$cult to observe with
either X-ray or neutron di!raction. Vlasse et al. (10) fol-
lowed this latter suggestion up with a precise single crystal
11102* rods of di4use intensity require the existence of ortho- X-ray di!raction experiment. No evidence, however, was
gonal {110} planes within the parent rutile structure which
exhibit long-range, two-dimensional, oxygen/6uorine ordering,
but with no correlation from one {110} plane to the next.
A crystal chemical explanation is proposed to support this argu-
ment. ( 2000 Academic Press
found for any such superstructure or for any oxygen}#uor-
ine ordering.
Given the similarity of the atomic scattering factors of
O and F for all three commonly used radiations (X-ray,
neutrons, and electrons) (3) and their weakness relative to
the atomic scattering factor of Fe, it is clear that any di!rac-
tion evidence for O/F ordering is always going to be weak in
intensity relative to the average structure Bragg re#ections
and di$cult to detect. In this study we therefore focus on
electron di!raction which has a proven ability to detect
weak features of reciprocal space often undetected by X-ray
or neutron di!raction.
INTRODUCTION
Oxy#uoride compounds form a natural bridge between
the oxide and #uoride families of compounds and research
into them has been steadily on the increase over many years
(1}4). To date, most oxy#uorides have been reported to
exhibit random anion site disorder. This apparent absence
of oxygen}#uorine ordering is usually attributed to the
rather similar ionic sizes of Oꢂ\ and F\ ions. On the other
hand, the strong electronegativity di!erence between oxy-
gen and #uorine should provide a strong driving force for
ordering, particularly for compounds with simple composi-
tion ratios (2). Perhaps the most simple such formulation
SYNTHESIS AND DATA COLLECTION
The FeOF transition metal oxy#uoride is di$cult to
synthesize in pure single phase form via the action of FeF
ꢃ
on a-Fe O as a result of the unavoidable presence of some
ꢂ ꢃ
water associated with the FeF starting material (5}9). An
ꢃ
additional problem is the volatility of the FeF starting
ꢃ
known corresponds to the Mꢃ>OF (M"Fe, Ti, V, , )
2
material necessitating the use of sealed Pt tubes.
transition metal oxy#uorides (2, 5}9) which crystallize in the
rutile structure type (see Fig. 1), apparently with no oxy-
gen}#uorine ordering.
The FeOF used in this study was prepared by the reaction
of a 1:1.12 molar ratio of a-Fe O (Halewood Chem. Ltd.,
ꢂ ꢃ
99.999%) and FeF (Cerac, 99.5%, nominally anhydrous)
ꢃ
at 9503C in a sealed platinum tube for 24 h. The reactants
ꢀTo whom all correspondence should be addressed.
were manipulated in an argon-"lled dry box to avoid
359
0022-4596/00 $35.00
Copyright ( 2000 by Academic Press
All rights of reproduction in any form reserved.

360
BRINK, WITHERS, AND THOMPSON
ELECTRON DIFFRACTION RESULTS
Initial investigation of reciprocal space revealed a con-
fusing mix of di!raction features in addition to the strong
sharp Bragg re#ections of the underlying rutile-type average
structure (hereafter labeled G) (see Figs. 2}5). In some
EDPs, virtually continuous lines of di!use intensity were
observed, particularly for zone axis orientations relatively
close to [001] (see, for example, Figs. 2a and 2b) while, in
others, apparently sharp superlattice re#ections were visible
(see Figs. 3a and 3b).
Figure 2a, for example, shows typical 10142 and (b)
10132 zone axis microdi!raction patterns, representative of
orientations close to the [001] zone axis orientation. In
addition to the sharp rutile-type parent re#ections, strong
FIG. 1. Schematic [001] projection of the FeOF rutile-type average
structure. The FeA (A"anion) octahedra are shown hatched. The unit
ꢄ
cell is outlined.
contamination with air or water vapor. Initial experiments,
using equimolar amounts (5}9) of the starting materials
always resulted in traces of Fe O in addition to the desired
ꢂ ꢃ
FeOF in the resultant reaction mixture. Subsequent analy-
sis of the FeF starting material by FTIR invariably in-
ꢃ
dicated the presence of a certain amount of water. Some
hydrolysis of the FeF was therefore inevitable during the
ꢃ
synthesis resulting in the loss of #uorine in the form of HF
gas. As dehydration of hydrated FeF and ab initio synthesis
ꢃ
of anhydrous FeF (11) both require exotic apparatus (pure
ꢃ
iron vessels and tubing) and extreme reaction conditions
(HF gas at '8003C) not available to us, commercially
available material had to be used. By adding a relatively
small (0.12 mol) excess of FeF to the initial starting mix-
ꢃ
ture, however, it was found to be possible to account for the
#uorine loss associated with the slightly hydrated FeF
ꢃ
starting material and to consistently produce single phase
FeOF.
In agreement with the previously reported results of
Chappert and Portier (6, 7) and Chamberland et al. (8, 9), we
"nd that FeOF cannot be formed below 9253C and, on
heating from room temperature, begins to unmix back to
Fe O and FeF at temperatures above &6353C.
ꢂ ꢃ
ꢃ
XRD data collected from a Guinier}Hagg camera
(j"1.5406 A), using an internal Si (NBS No. 640)
standard, were used to re"ne the unit cell parameters
(a"4.6431(7) A, c"3.0436(5) A). Samples suitable for
transmission electron microscope (TEM) work were
prepared by the dispersion of "nely ground material onto
a holey carbon "lm. Electron di!raction patterns (EDPs)
were obtained using Philips 430, JEOL 2000FX, and JEOL
100CX TEMs.
FIG. 2. Typical (a) 10142 and (b) 10132 zone axis microdi!raction
EDPs of FeOF. Note the characteristic di!use streaking in both and the
fact that this di!use streaking only runs through rutile parent re#ections for
which (h#k#l) is odd.

RUTILE-TYPE IRON OXYFLUORIDE, FeOF
361
The "nal important feature to notice about Fig. 2 is that
there is clearly an extinction condition operating in that the
di!use streaking does not run through all parent rutile
re#ections, but only through those for which (h#k#l) is
odd (e.g., in Fig. 2a) di!use streaks run through the 041
re#ection but not through the neighboring 141 re#ection
whereas, in Fig. 2b, di!use streaks run through the 131
re#ection but not the neighboring 031 re#ection, etc.). This
characteristic extinction condition can be easily obscured at
more major zone axis orientations such as [001] itself or
11102 and 11202 (cf. Figs. 3 and 4) where stronger multiple
scattering routes connecting the (h#k#l)"odd and
(h#k#l)"even re#ections exist. This essentially body-
centered extinction condition is absolutely characteristic of
the di!use streaking and strongly suggests that atomic dis-
placements of the Fe atoms must be largely responsible
(keep in mind that the Fe atoms occupy the origin and
the body-centered positions in the rutile-type average
structure).
Tilting to zone axis orientations perpendicular to the
[001] orientation, as shown in the close to 11102 and close
to 11202 zone axis orientations of Figs. 3 and 4, results in
FIG. 3. Selected area EDPs taken close to (a) a 11102 zone axis
orientation and (b) a 11202 zone axis orientation. Rutile-type parent
re#ections are labeled. Note the presence of additional satellite re#ections
in both.
(virtually continuous) lines of polarized di!use intensity are
clearly visible running approximately along the [110]* and
[110]* directions of reciprocal space. It is also apparent that
there is a strong angular, or azimuthal, variation of the
intensity of these di!use streaks, indicating that atomic
displacements are making the major contribution toward
the observed intensities. The transverse polarized character
of the di!use streaking (i.e., the fact that the intensity of the
di!use streaking is always strongest when looking out along
directions of reciprocal space perpendicular to the direction
of streaking itself ) implies that the associated atomic shifts
must also be largely transverse polarized; i.e., the atomic
shifts giving rise to the di!use streaking along [110]* must
be largely due to the atomic shifts along [110] and vice
versa.
FIG. 4. EDPs similar to those in Fig. 3, but tilted a little further o! axis
in order to minimize the e!ects of dynamical di!raction. Note the charac-
teristic extinction conditions now clearly visible.

362
BRINK, WITHERS, AND THOMPSON
what appear to be discrete additional &&satellite'' re#ections
occurring at reciprocal space positions G$[m n 0]*, where
G"[hkl]* (h, k, l all integers) and $m#n"1. In Fig. 3a,
for example, G$1/211102* type satellite re#ections corre-
sponding to m"1/2 and n"1/2 are apparent whereas, in
Fig. 3b, G$1/3 12102* type satellite re#ections corre-
sponding to m"2/3 and n"1/3 are clearly visible. Sim-
ilarly, at 13102 zone axis orientations, G$1/4[310]* type
satellite re#ections corresponding to m"3/4 and n"1/4
are visible, etc.
At "rst glance, it appears that these additional satellite
re#ections, e.g., the G$1/3 12102* type satellite re#ections,
apparent on the right-hand side of Fig. 3b, for example, do
not appear to obey any particular extinction condition.
Tilting slightly further away from the exact zone axis ori-
entation to minimize the e!ects of multiple scattering (see
Fig. 4b), however, shows that these satellite re#ections do
indeed obey a characteristic extinction condition (cf.
FIG. 5. An intermediate [135] zone axis EDP. In addition to the
labeled rutile-type parent re#ections, note the presence of di!use dashes of
Figs. 4b, and 3b) and one that is entirely compatible with limited length consistent with an oblique section of a rod in reciprocal
space.
that observed for the 11102* di!use streaking. The same is
also true for the 11102 orientation (cf. Figs. 4a and 3a). The
only way in which the observation of these apparently sharp
satellite re#ections can be reconciled with the more-or-less
continuous di!use streaking apparent in EDPs close to the orthogonal [110] direction. (The equivalent statement
[001] is in terms of essentially continuous rods of di!use must also be true for the symmetry-equivalent one-dimen-
intensity along both the [110]* and [110]* directions of sional rod of di!use intensity running along the orthogonal
reciprocal space running through the (h#k#l)"odd par- [110] direction of reciprocal space).
ent rutile re#ections.
In the case of rutile-type or FeOF, this requires that we
This interpretation may be con"rmed by considering an must look for an explanation of the observed di!use distri-
intermediate orientation such as, for example, the [135] bution in the correlated occupancies and shifts of the atoms
zone axis shown in Fig. 5. Now, in addition to the strong within +110, planes, as shown in Fig. 6a. Consider, for
Bragg re#ections of the underlying rutile-type parent struc- example, the [110]* rods of reciprocal space. The transverse
ture, there exist di!use dashes (or streaks of rather short polarized di!use streaking in EDPs such as those in Fig. 2
duration in reciprocal space). This is exactly as would be requires that the responsible atomic shifts must be perpen-
expected given that we are now intersecting linear rods of dicular to the [110]* rod direction itself, i.e., in the (110)
di!use intensity at an oblique angle. Furthermore, the ori- plane. Indeed it implies that a major component of these
entation, location, and intensity distribution of the di!use shifts must be along the orthogonal [110] direction.
dashes in Fig. 5 are also as would be expected given the
above description.
More speci"cally still, the systematic absence of G$
e[110]* streaks in [110] zone axis EDPs (see Figs. 3a and
4a) requires that the associated displacements must be en-
tirely along the orthogonal [110] direction of real space. In
other words, the rods of di!use intensity along [110]* are
INTERPRETATION
Having established that the reciprocal lattice of FeOF is due to atomic displacements along [110] while the rods of
characterized by continuous, transverse polarized rods of di!use intensity along [110]* are due to atomic displace-
di!use intensity running along both the [110]* and [110]* ments along [110]. (Such basal plane shifts of the Fe atoms
directions of reciprocal space through the (h#k#l) odd are also strongly supported by the average structure re"ne-
re#ections of the rutile-type average structure, the question ment result of Vlasse et al. (10) that the && thermal vibra-
2
now becomes what is the crystal chemical origin of this tion ellipsoid observed for the iron atoms is very much
characteristic di!use distribution?
It is well known that a one-dimensional rod of di!use
#attened perpendicular to the c axis '').
At this point it becomes convenient to consider the ob-
2
intensity running along the [110]* direction of reciprocal served di!use distribution from a modulation wave point of
space implies the existence of (110) planes of atoms whose view (12}14). Each point on the observed di!use distribu-
displacements and occupancies must be correlated within tion is thus described as G$q and the various modulations
the plane but totally uncorrelated from plane to plane along are treated as independent modulation waves. Furthermore,

RUTILE-TYPE IRON OXYFLUORIDE, FeOF
363
Consider, for example, the displacive contribution of the
Fe atoms to the structure factor at the reciprocal space
position (G#q ). Following (12), this is given by
ꢂ
!ꢂ
F
(G#q )J+ f exp+2ni G r ,
'
$*ꢀꢁ
ꢂ
I
I
I
2n(G#q ) e (q )2
[1]
'
ꢂ
I ꢂ
Here r labels the position of the kth independent Fe atom
I
per parent unit cell and e (q ) represents the displacement
I ꢂ
eigenvector (or shift of this atom) associated with the q th
modulation wave-vector.
ꢂ
Now within each (110) plane of atoms (see Figs. 1 and 6)
there exist two independent Fe atoms per parent unit cell,
Fe at (0, 0, 0) and Fe at (1/2, 1/2, 1/2), which will contrib-
ꢀ
ꢂ
ute to the structure factor at (G#q ). The observed extinc-
ꢂ
tion condition, F(G$q , q )"0 unless (h#k#l) is odd,
ꢀ
ꢂ
$%ꢀ ꢂ
requires that e (q )"!e (q ) as can be seen by substi-
$%ꢂ ꢂ
tution into Eq. [1] above, giving
F
(G#q )J2n(G#q ) e
'
ꢂ ꢂ $%ꢀ
$*ꢀꢁ
#2n(G#q ) e exp+ni(h#k#l),
'
ꢂ
$%ꢂ
"[2n(G#q ) e ][1!exp+i n(h#k#l)],
'
ꢂ
$%ꢀ
which"0 unless (h#k#l) is odd as experimentally ob-
served.
While these 11102 displacive shifts of the Fe atoms could
in principle arise without an underlying compositional
cause, it seems far more likely that the Fe shifts are a re-
sponse to oxygen/#uorine ordering within each +110, plane.
FIG. 6. Schematic representation of the 11102 planes of the rutile-type
structure. The "rst diagram (a) shows the honeycomb generated if opposite
corners of each octahedron are constrained to be of opposite type, i.e., if
one is O then the other must be F and vice versa. Note that two such
honeycomb arrays are required to cover all anion sites in this projection.
The two possible fully ordered arrangements are shown in (b) and (c). Note
the relationship between the resulting &&supercells'' in each case.
CRYSTAL CHEMICAL CONSIDERATIONS
The question now becomes what sort of oxygen/#uorine
ordering could be responsible for these 11102 Fe shifts. The
crystal chemical reasons for the Fe shifts can be understood
from a bond length}bond valence analysis (15) of the aver-
age structure of FeOF using the re"ned fractional coordi-
the modulation wave-vectors are broken into two classes, nates of Vlasse et al. (10). The apparent valence (AV) of
those along [110]* and those along [110]*. The former are a particular ion is de"ned as the bond valence sum arising
written in the generic form q "e[110]* and the latter in the from the surrounding ions of opposite charge (15). Such
ꢀ
form q "e[110]*, where it is implicitly understood that apparent valence calculations indicate that oxygen, with
ꢂ
e can run over the range from 0 to 1/2.
a calculated AV of 1.556, is signi"cantly underbonded
As discussed above, the body-centered tetragonal nature in the average structure anion position while #uorine
of the sublattice formed by the Fe atoms in FeOF (see (AV"1.223) is signi"cantly overbonded (see Table 1). Us-
Fig. 1) combined with the observed extinction condition ing the R parameters of Brese and O'Kee!e (15), the ideal
ꢅ
observed for the di!use, i.e., F(G$q , q )"0 unless anion}Feꢃ> separation distance when the anion is Oꢂ\
ꢀ
ꢂ
(h#k#l) is odd, implies not only that shifts on the Fe would be 1.909 A while the ideal anion}Feꢃ> separation
atoms must be primarily responsible for the observed di!use distance when the anion is F\ would be 2.076 A. The
distribution but also that the Fe atom shifts of the two experimental average anion}Feꢃ> separation distance re-
independent Fe atoms per parent unit cell must be anticor- ported by Vlasse et al. (10) is, not surprisingly, halfway in
related, i.e., have the opposite sign.
between at 2.00 A.

364
BRINK, WITHERS, AND THOMPSON
TABLE 1
rutile directions while e is necessarily of opposite sign to
$%ꢂ
, just as required experimentally. There is, theoretically,
a third alternative possibility whereby the two honeycomb
arrays are not correlated. This, however, would give rise to
additional di!use rods at positions of reciprocal space
which are not observed experimentally.
We therefore believe that oxygen and #uorine ions are in
essence completely ordered within any one +110, type plane
with the O/F ordering pattern as shown in Fig. 6c. The
observed 11102* rods of di!use intensity, however, require
that there is no correlation in this ordering pattern from one
such +110, plane to the next. Comparison of Fig. 1 with
Bond Valence Sums (AVs) for the Re5ned Average
Structure of FeOF (10, 15)
e
$%ꢀ
Atom
AV
Expected AV
Fe (assuming anions all F)
Fe (assuming anions all O)
O
F
2.78
3.11
1.56
1.22
3.00
3.00
2.00
1.00
Crystal chemically this implies that Oꢂ\ ions will attract Fig. 6c shows that this should not be unexpected. Indeed
neighboring Feꢃ> cations while F\ ions will repel them. If there would appear to be absolutely no crystal chemical
we consider any particular (110) plane of ions (see Fig. 6a), it reason at all for any such correlation to exist. A possible
seems logical to suggest that opposite corners of each FeA
crystal chemical explanation as to why anions of the oppo-
ꢄ
octahedraon (A for anion) should necessarily be of opposite site type (i.e., O}F, F}O) should necessarily be matched
type, i.e., if one corner is #uorine then the other should be across the short edge of the octahedra can be found in the
oxygen and vice versa (see Fig. 6a). This would then allow 2.609-A separation across the short edge of the octahedra.
the iron atom to shift out of the center of the octahedra Consideration of nonbonded anion}anion interaction ener-
toward the oxygen ions and away from the #uorine ions as gies (16) suggests a minimum for anions of the opposite type
desired. It would also imply, in agreement with the original (i.e., O}F, F}O) at either end of this short 2.609-A separ-
Mossbauer results of Chappert and Portier (6) that each ation distance.
Feꢃ> cation is in a unique environment surrounded always
by 3Oꢂ\ ions and 3F\ ions.
CONCLUSION
The application of such a rule within a +110, plane,
starting at an arbitrary apex of any one octahedron, gener-
The "nal point that has been arrived at is that there is no
ates an ordered honeycomb array of oxygen and #uorine contradiction at all between the original Mossbauer results
ions, with an in-plane superlattice unit cell as shown in (requiring that each Fe atom be in a unique environment)
Fig. 6a. Notice that there are two such honeycomb arrays, and di!raction results, but only if the assumption of three-
at this stage still allowed to be independent of each other. dimensional long-range order is thrown away and the ob-
The requirement that opposite vertices of each octahedron served di!use scattering (in the form of 11102* rods of
should be of opposite type does not, of itself, determine the di!use intensity running through the (h#k#l) odd rutile
phase relationship between these two honeycomb arrays. parent re#ections) taken into account. It is the (completely
Moreover, a single ordered honeycomb array would not of understandable from the crystal chemical point of view) lack
itself force Fe cations to displace solely along 11102 direc- of correlation from one fully O/F ordered +110, plane to the
tions. It is therefore necessary to consider the possible phase next along the orthogonal 11102 direction that is respon-
relationships between the two honeycomb arrays.
sible for the observed di!use distribution and that prevents
Matching anions of the same type (i.e., O}O, F}F) across the condensation of a long-range (O/F) ordered (from the
the short edge of the octahedra leads to the ordering pattern conventional crystallographic point of view) superstructure
shown in Fig. 6b. Note that in this case there is a &&phase'' phase.
reversal of 1803 between the two honeycomb arrays for
Careful consideration shows that there are only four
a parent rutile translation of [110]. Note furthermore that distinct local O F ordering patterns possible for any one
ꢃ ꢃ
such an ordering pattern would constrain half the Fe ca- particular FeO F octahedra that will give rise to a resultant
ꢃ ꢃ
tions to displace along the parent c direction with the other displacement of the Fe atom (away from the octahedral
half constrained to displace along the orthogonal 11102 center) along either the $a or $b directions. The magni-
parent direction (see Fig. 6b). This is not compatible with tude of this shift (essentially away from the local F octahed-
ꢃ
experimental observation as discussed above. The alterna- ral face and toward the opposite O octahedral face) should
ꢃ
tive fully ordered possibility, matching anions of the oppo- be &0.12}0.13 A in order to satisfy local bond valence
site type, (i.e., O}F, F}O) across the short edge of the requirements (see Table 1). By analogy, one might expect
octahedra, gives rise to the ordering pattern shown in similar behavior for all isomorphous Mꢃ> OF (M"Ti, V,
Fig. 6c. This time, the two honeycomb arrays are in &&phase'' Al, ) oxy#uorides (5}9).
2
for a parent rutile translation of [110]. Furthermore, all Fe
Furthermore, it may well be the case that the local crystal
cations are constrained to shift solely along 11102 parent chemistry of a range of other oxy#uoride systems (many of

RUTILE-TYPE IRON OXYFLUORIDE, FeOF
365
which spectroscopic and bond valence considerations sug-
gest should also be locally ordered (3)) cannot be properly
understood until di!use distributions of the type reported in
the present paper are observed and properly taken into
account.
6. J. Chappert and J. Portier, Solid State Commun. 4, 185 (1996).
7. J. Chappert and J. Portier, Solid State Commun. 4, 395 (1966).
8. B. L. Chamberland and A. W. Sleight, Solid State Commun. 5, 765}767
(1967).
9. B. L. Chamberland, A. W. Sleight, and W. H. Cloud, J. Solid State
Chem. 2, 49}54 (1970).
10. M. Vlasse, J. Massies, and G. Demazeau, J. Solid State Chem. 8, 109
(1973).
11. W. Kwasnik, in &&Handbook of Preparative Inorganic Chemistry'' (G.
Brauer, Ed), 2nd ed., Vol. I, p. 226, Academic Press, New York, 1963.
12. R. L. Withers, S. Schmid, and J. G. Thompson, Prog. Solid State Chem.
26, 1 (1998).
REFERENCES
1. J. Ravez, J. Phys. III Fr. 7, 1129}1144 (1997).
2. P. Hagenmuller, in &&Perspectives in Solid State Chemistry'' (K. J. Rao,
Ed), pp. 66}78, Narosa, New Delhi, 1995.
13. T. R. Welberry and R. L. Withers, J. Appl. Crystallogr. 23, 303 (1990).
3. L.-S. Du, F. Wang, and C. P. Grey, J. Solid State Chem. 140, 285}294 14. B. D. Butler, R. L. Withers, and T. R. Welberry, Acta Crystallogr. A 48,
(1998). 737 (1992).
4. T. Vogt, P. M. Woodward, B. A. Hunter, A. K. Prodjosantoso, and 15. N. E. Brese and M. O'Kee!e, Acta Crystallogr. B 47, 192 (1991).
B. J. Kennedy, J. Solid State Chem. 144, 228}231 (1999).
5. P. Hagenmuller, J. Portier, J. Cadiou, and R. DePape, C. R. Acad. Sci.
260, 4768 (1965).
16. M. O'Kee!e and B. G. Hyde, in &&Structure and Bonding in Crystals''
(M. O'Kee!e and A. Navtotsky, Eds.), Vol. I, Chap. 10, pp. 66}78,
Academic Press, New York, 1981.