32

Acta Cryst. (1997). B53, 32-43

Octahedral Tilting in Perovskites. I. Geometrical Considerations

PATRICK M . W O O D W A R D t

Department of Chemistry and Center for Advanced Materials Research, Oregon State University, Corvallis, OR
97331-4003, USA. E-mail: pat@pix7a.phy.bnl.gov
(Received 17 April 1996; accepted 28 August 1996)

Abstract (Hwang, Cheong, Radaelli, Marezio & Batlogg, 1995;

The 23 Glazer tilt systems describing octahedral tilting Fontcuberta et al., 1996). The technological importance
in perovskites have been investigated. It is shown that of these compounds and many others has made the
in tilt systems a÷a+a-, a+b+b-, a+a+c-, a+b+c-, a°b+b - perovskite structure one of the most important structures
and a°b+c- it is not possible to link together a three- in all material science.
dimensional network of perfectly rigid octahedra. In The ideal perovskite structure has an AMX3 stoi-
these tilt systems small distortions of the octahedra must chiometry and belongs to the cubic space group Pm3m.
occur. The magnitude of the distortions in the a+a+a- and The A cation is surrounded by 12 X anions in a
a°b+b- tilt systems are estimated. A table of predicted dodecahedral environment, the M cation is octahedrally
space groups for ordered perovskites, A2MM'06, for all coordinated by six X ions, and the X anions are
23 tilt systems is also given. coordinated by two M cations and four A cations. There
are at least two ways of visualizing the structure. The
most common description of the perovskite structure
1. Introduction is a three-dimensional cubic network of corner-sharing
MX6 octahedra. In this description the A cation sits in
The perovskite structure class encompasses a huge
the center of a cube defined by eight corner-sharing
variety of compounds. Most of the metallic ions
octahedral units. An alternative way of visualizing the
in the periodic table can be incorporated into the
perovskite structure is to begin with a cubic close-packed
perovskite structure. Oxides and fluorides comprise the
array of X anions, then replace one out of every four
vast majority of perovskite compounds (Goodenough
X ions with an A cation in an ordered way to obtain a
& Longo, 1970), but the perovskite structure is found
cubic close-packed AX3 array. Finally, all the octahedral
for many combinations of cations and anions. Chlorides
holes in the AX3 lattice that do not border an A cation
(Brynestad, Yakel & Smith, 1966; Horowitz, Amit,
are filled with M cations (25% of the octahedral holes).
Makovsky, Ben Dor & Kalman, 1982), bromides
Even though there are a large number of simple
(Knochenmuss, Reber, Rajasekharan & GEidel, 1986),
perovskites, AMX3, the number of possible compounds
hydrides (Messer, Eastman, Mers & Maeland, 1964),
is greatly expanded when multiple ions are substituted
oxynitrides (Bacher et al., 1988) and sulfides (Clearfield,
for one or more of the original ions. In most cases this
1963; Rodier & Laruelle, 1970; No~l, Padiou & Prigent,
substitution occurs on the cation sites and leads to a large
1975) are all known with the perovskite structure.
class of compounds known either as double or complex
Since the perovskite structure can accommodate
perovskites, AAtMM'X6. When such substitutions are
such a wide variety of ions, the physical properties
made the ions can occupy the original cation site of the
of perovskite compounds are wonderfully diverse.
simple structure in either a random or an ordered fashion.
Perovskite compounds with interesting dielectric,
If an ordered arrangement is adopted the symmetry and,
magnetic, electrical, optical and catalytic properties are
in many cases, the size of the unit cell are changed.
all known and have been widely studied. Technological
Although the ideal perovskite structure is cubic there
applications vary from the dielectric properties of
are several structural deviations from the ideal cubic
BaTiO3 (Shirane, Danner & Pepinski, 1957; Galasso,
structure, both for simple and ordered perovskites. In
1969), which make it an important material in the
fact, there are many more distorted perovskites than
capacitor industry, to the superconductivity displayed
there are cubic perovskites. Even the mineral perovskite
by Ba(Bil_xPbx)O3 (Sleight, Gillson & Bierstedt, 1975)
itself, CaTiO3, is orthorhombic rather than cubic (Sasaki,
and the perovskite-related high-temperature cuprate
Prewitt, Bass & Schulze, 1987). Cubic, tetragonal,
superconductors, to the fascinating magnetoresistive
orthorhombic, rhombohedral, monoclinic and triclinic
and magnetostrictive properties of (Lal_/Cax)MnO3
perovskites are all known. These structural distortions
i" Current address: Department of Physics, Brookhaven National are of interest not only from a crystallographer's point of
Laboratory, Upton, NY 11973, USA. view, but also because they can have important effects

© 1997 International Union of Crystallography Acta Crystallographica Section B

Printed in Great Britain - all rights reserved ISSN 0108-7681 © 1997
 PATRICK M. WOODWARD 33

on the physical properties of perovskite compounds, in Russia around the same time by Aleksandrov (1976).
particularly the electrical and magnetic properties. However, it is the notation developed by Glazer, to
The influence small structural changes have on the describe octahedral tilting distortions, that has become
magnetoresistance properties of manganese- and cobalt- standard notation in the perovskite literature.
based perovskites is an important example of this Somewhat surprisingly, a paper by Leinenweber &
phenomenon (Hwang, Cheong, Radaelli, Marezio & Parise (1995) claimed that tilt systems a÷a÷c- (#5) and
Batlogg, 1995; Fontcuberta et al., 1996; Bricefio, Chang, a÷a÷a- (#7) do not produce structures belonging to space
Sun, Schultz & Xiang, 1995). group P m m n as predicted by Glazer. They contended
Distortions from the ideal perovskite structure can that these two tilt systems actually result in structures
be attributed to one of three mechanisms: distortions of belonging to space group P42/nmc (Leinenweber &
the octahedra, cation displacements within the octahedra Parise, 1995). In addition to this discrepancy, the com-
and tilting of the octahedra. The first two distortion plications that arise when cation ordering and octahedral
mechanisms are driven by electronic instabilities of tilting occur simultaneously have not been addressed
the octahedral metal ion. The Jahn-Teller distortion in in a comprehensive manner. Both of these issues are
KCuF3 (Okazaki & Suemune, 1961) is an example of an discussed in detail in the sections that follow.
electronic instability that leads to octahedral distortions.
The ferroelectric displacement of titanium in BaTiO3
(Shirane, Danner & Pepinski, 1957) is an example of an 2. A brief review of Glazer notation
electronic instability that leads to cation displacements. The notation developed by Glazer over 20 yeai'.s ago
Although these distortion mechanisms are important to describe octahedral tilting distortions in perovskites
in many compounds, they are not discussed further has become the standard by which such distortions are
in this paper. The third and most common distortion described (Glazer, 1972). A familiarity with this notation
mechanism, octahedral tilting, can be realized by tilting is necessary to follow the discussion throughout this
essentially rigid MO6 octahedra while maintaining their paper. As an aid to those unfamiliar with Glazer notation
corner-sharing connectivity. This type of distortion is a brief review is presented below. Further explanation
typically observed when the A cation is too small for the can be found in the original reference (Glazer, 1972).
cubic MO3 corner-sharing octahedral network. In such Using Glazer notation a tilt system is described by
cases it is the lowest energy distortion mode, because the specifying the rotations of the octahedra about each
A----O distances can be shortened while the first coordi- of the three Cartesian axes. The rotations about each
nation sphere about the M cation remains unchanged. axis are described by two parameters. The first of these
(In terms of the MO3 lattice, only the soft M----O---M parameters is a letter specifying the magnitude of the
bond angle is disturbed.) Geometrical considerations rotation about that axis, relative to the magnitude of the
associated with this distortion are the subject of this rotations about the other Cartesian axes. For example,
paper, while the interatomic forces that drive octahedral in the a÷a÷a÷ system the rotation angle is the same
tilting distortions are the subject of the following paper. about each of the three axes; whereas in the a÷a÷c÷ tilt
Over the years many investigators have studied dis- system the rotation angle about the z axis is different
tortions in the perovskite structure. Thomas has devised from the rotation angle about the x and y axes. The
a system for classifying perovskites based on the poly- second parameter is a superscript indicating whether
hedral volumes of the A and M cations, which is par- the rotations in adjacent layers are in the same or
ticularly useful when both cation displacements and opposite directions. A negative superscript indicates that
octahedral tilting occur (Thomas, 1989, 1996; Thomas & the rotations of two neighboring octahedra, along the
Beitollahi, 1994). O'Keeffe & Hyde (1977) structurally tilt axis, are in the opposite directions, while a positive
described the three most common octahedral tilt systems; superscript is used when they tilt in the same direction. A
a÷b-b-, a a a and a÷a÷a÷. Megaw & Darlington (1975) zero superscript is used when no rotations occur about
studied, in great detail, the effect of octahedral tilting an axis.
and octahedral deformation in rhombohedral perovskites The clearest example of how Glazer notation
and Deblieck and coworkers classified possible tilt- describes tilted perovskite structures is a comparison
ing combinations in perovskite-like ABX4 compounds of the two one-tilt systems, a°a°c ÷ and a°a°c -. The
(Deblieck, 1986; Deblieck, Van Tendeloo, Van Landuyt octahedra in both these tilt systems are rotated only
& Amelinckx, 1985). However, almost certainly the about the fourfold axes parallel to the z axis. Fig. 1
most influential work was published in 1972 by Glazer. shows a view looking down the z axis at eight octahedra
This work contained an eloquent description of all in both tilt systems. It is clear in this figure that rotation
possible simple octahedral tilting combinations (Glazer, of one octahedron forces four of the neighboring six
1972). An updated description, including a few minor octahedra to rotate in the opposite sense. This in turn
corrections, can be found in the book 'Space Groups causes all the octahedra in the (001) plane to be rotated
f o r Solid State Scientists', by Burns & Glazer (1990). A with the same magnitude as the first octahedron. The
similar but much less known approach was developed final two neighboring octahedra, connected along the z
34 OCTAHEDRAL TILTING IN PEROVSKITES. I.

axis, are not coupled in the same way and can rotate Table 1. A comparison of the space groups predicted for
independently. If these final two neighboring octahedra each of the 23 simple tilt systems by Glazer (1972) with
rotate in the same direction, as in Fig. 1(b), the a°a°c+ tilt those generated by POTATO
system results. If they rotate in the opposite direction, For tilt systems 4, 5, 6, 7, 17 and 18 the atomic positions calculated by
as in Fig. l(a), the a°a°c - tilt system results. More P O T A T O always indicated slight distortions of the octahedra and,.
complicated structures where independent tilting has to therefore, no space group could be assigned as rigorously correct.
be considered in more than two layers is not described This result is discussed in more detail in the text.
by Glazer notation. Tilt system Tilt system Space group Space group
number symbol (Glazer, 1972) ( P O T A TO)

3. Space-group descriptions of all 23 tilt systems Three-tilt systems

1 a+b+c + I m m m (#71) Immm (#71)
3.1. Simulated structures and their comparison with 2 a+b+b + I m m m (#71) Immm (#71)
Glazer' s space-group assignments 3 a+a+a + Ira3 (#204) Ira3 (#204)
4 a+b+c - Pmmn (#59) Pmmn (#59-2)
Although Glazer completely described the space 5 a+a+c - P m m n (#50) Pmmn (#59-2)
group and symmetry associated with each tilting 6 a+b+b - P m m n (#59) Pmmn (#59-2)
system it is far from trivial in many cases to turn 7 a+a+a - P m m n (#59) Pmmn (#59-2)
8 a+b-c - A 2 1 / m l l (#11) P 2 1 / m (#11-1)
that information into a model structure, complete with
9 a+a-c - A 2 1 / m l l (#11) P 2 t / m (#11-1)
atomic positions. A complete description of the unit cell, 10 a+b-b - Pmnb (#62) Prima (#62)
I1 a+a a P m n b (#62) Prima (#62)
12 a-b-c- Fi (#2) Fi (#2)
aoa%- 13 a-b-b- I2_/a (#15) 12_/a (#15-3)
14 a a a- R3c (#167) R3c (#167-2)
Two-tilt systems
15 a°b+c + I m m m (#71) Immm (#71)
16 a°b+b + 1 4 / m m m (#139) 1 4 / m m m (#139)
17 a°b +c - B m m b (#63) Pmmn (#59-2)
18 a°b+b - B m m b (#63) Pmmn (#59-2)
19 a°b-c - F 2 / m l 1 (#12) 1 2 / m (#12-3)
20 a°b-b - I m c m (#74) Imma (#74)
One-tilt systems
21 a°a°c + C 4 / m m b (#127) P 4 / m b m (#127)
22 a°a°c - F 4 / m m c (#140) 1 4 / m c m (#140)
Zero-tilt systems
23 a°a°a ° P m 3 m (#221) Pm3m (#221)

(a) where all the atomic positions are given, would be very
useful in several situations. Such a model is needed to
(~1)(.10c + compare experimentally determined structures with ideal
structures; as a starting point for Rietveld refinements;
or to systematically study the differences in the ideal
structures associated with each tilt system. With these
applications in mind the FORTRAN program POTATO
/ (Program Originated To Analyze Tilted Octahedra) was
developed (Woodward, 1996). Given the Glazer symbol,
the rotation angle about each of the Cartesian axes and
the M----O bond distances as input, POTATO outputs
a unit-cell description of the compound complete with
atomic positions. The unit cell calculated by POTATO is
always described as centrosymmetric triclinic (Pi) and
doubled along all three axes. The true asymmetric unit
is then found by examination of relationships between
(b) atomic positions or more conveniently with one of the
Fig. 1. A view Iookin~ down the z axis of two adjacent layers of octahe-
automated routines now available [e.g. find symmetry
dra for (a) the a ° a ° c - tilt system and (b) the a ° a % ÷ tilt system. This within the InsightII package (Biosym Technologies,
view is a polyhedral representation of the crystal structure where 1993)].* Using this approach the space group and
each octahedron symbolizes a metal atom M surrounded by six O
atoms. The metal atom is located in the center of the octahedron and * Based on the results contained in this paper the current version of
an O atom is located at each corner. The round circles represent the P O T A T O also gives the true space group and asymmetric unit for each
A ions. tilt system.
 PATRICK M. WOODWARD 35

Table 2. T h r e e d i f f e r e n t d e s c r i p t i o n s o f the a t o m i c p o s i n o n s o f a n a+ a+ a - t i l t e d p e r o v s k i t e (a = 1 0 °)
The first two descriptions were generated in POTATO. In the first structure the distances were constrained to be constant (2.00,~,), while in the
second structure the angles were constrained to remain 90 ° . The third description was generated using the O1, 0 2 and 0 3 positions common to
all three descriptions and the symmetry generators associated with P42/nmc. All other oxygens in the unit cells can be generated from these 12
using the inversion centers on the M ions. The coordinates shown in bold are those where the second and third descriptions differ from the first
description.

Pmmn Pmmn
(POTATO, constant distance) (POTATO, 90 ° angles) P 42/ nmc
x y z x y z x y z
O1 1/4 -0.0404 0.0482 1/4 -0.0404 0.0482 1/4 -0.0404 0.0482
02 0.0482 1/4 -0.0404 0.0482 1/4 -0.0404 0.0482 1/4 -0.0404
03 0.4518 1/4 -0.0404 0.4518 1/4 -0.0404 0.4518 1/4 -0.0404
04 1/4 0.5404 0.0482 1/4 0.54(14 0.0482 1/4 0.5404 0.0482
05 -0.0404 0.0482 1/4 -0.0404 0.0482 0.2513 -0.0404 0.0404 1/4
06 0.5404 0.0482 1/4 0.5404 0.0482 0.2513 0.5404 0.0404 1/4
07 -0.0404 0.4518 1/4 --0.0404 0.4518 0.2513 -0.0404 0.4596 1/4
08 0.5404 0.4518 1/4 0.5404 0.4518 0.2513 0.5404 0.4596 I/4
09 1/4 0.0553 0.4703 1/4 0.0394 0.4667 1/4 0.0482 0.4596
O10 -0.0329 1/4 0.5535 -0.0322 1/4 0.5542 -0.0404 1/4 0.5482
O 11 0.5329 1/4 0.5535 0.5322 1/4 0.5542 0.5404 1/4 0.5482
O12 1/4 0.4447 0.4703 1/4 0.4606 0.4667 1/4 0.4518 0.4596
All distances 2.00 ,~ 1/3 of the distances 1.98 ,~, 1/3 of the distances 1.98 ,~
1/3 of the angles 86.5 ° All angles 90 ° 1/3 of the angles 88.4 °

asymmetric unit were calculated for each of the 23 With regard to those tilt systems where distortions of the
Glazer tilt systems. The results of these calculations octahedra must occur it becomes impossible to uniquely
are compared with Glazer's findings in Table 1. The define a space group. The problem arises because the
complete unit-cell descriptions for each tilt system are symmetry of the unit cell depends on the way in which
given in Appendix B. the octahedra are distorted. Table 2 compares the atomic
One curiosity of this analysis was that for tilt systems positions, bond distances and angles for a hypothetical
a+b+c - (#4), a+a÷c - (#5), a+b+b - (#6), a+a÷a - (#7), compound belonging to the a÷a÷a- tilt system. The first
a°b+c - (#17) and a ° b ÷ b - (#18) slight distortions of the two structures have symmetry described by space group
octahedra were always necessary in order to retain the P m m n , as suggested by Glazer (Glazer, 1972; Bums &
connectivity of the octahedra. Interestingly, the two tilt Glazer 1990). The difference between the two structures
systems, a÷a+c- (#5) and a÷a÷a- (#7), that Leinenweber is that in the first structure all the bond distances have
& Parise (1995) claimed were incorrectly assigned by remained rigid, while in the second structure all the
Glazer are among this group. At first this seemed to bond angles have been constrained to be 90 ° . The third
be a shortcoming of the algorithms used in P O T A T O . structure has symmetry corresponding to space group
However, despite several modifications to P O T A T O this P 4 2 / n m c , as suggested by Leinenweber & Parise (1995).
anomaly persisted. This led to the realization that the It is evident from this table that the atomic positions
octahedral distortions were not caused by an error in are quite similar in all three structures. In Table 3 the
P O T A T O , but rather are an inherent geometric property results of a similar calculation for the a ° b * b - tilt system
of these six tilt systems.* Proof of this claim is given are shown. Here the description in P m m n as output
in Appendix B and demonstrated in Appendix C by cal- from P O T A T O is very close to a unit cell having B m m b
culating the metal-oxygen vectors that define octahedra symmetry, as predicted by Glazer. In this tilt system the
in the a+a+a - tilt system. octahedral distortions are very small. Although it is not
possible to rigorously assign a space group for these
3.2. S p a c e - g r o u p a s s i g n m e n t s f o r tilt s y s t e m s 4, 5, 6, 7, tilt systems, it seems reasonable to assume that actual
17 a n d 18 compounds will crystallize in the more symmetric of the
two available space groups. Based on that assumption
With the exception of the above-mentioned tilt sys- the a÷a+b - and a+a+a - tilt systems are assigned to
tems, P O T A T O was able to generate perfectly linked net- P 4 2 1 n m c and the a°b+ c - and a°b+b - tilt systems assigned
works of rigid octahedra for all tilt systems. Furthermore, to B m m b (standard setting C m c m ) . The a+b÷c - and
for these tilt systems the space groups and lattice con- a÷b+b - tilt systems cannot be tetragonal, because the a
stants predicted by Glazer were found to agree exactly and b lattice constants are different and so they remain
with the unit-cell descriptions generated by P O T A T O . orthorhombic, P m m n , as originally suggested by Glazer
* O'Keeffe & Hyde (1977) had suggested that octahedral distortions (1972). Therefore, in essence none of the assignments
must occur in some tilt systems, but did not identify which tilt systems. in Glazer's work (Bums & Glazer, 1990) are incorrect.
36 OCTAHEDRAL TILTING IN PEROVSKITES. I.

Table 3. Three different descriptions o f the atomic positions o f an a°b+b - tilted perovskite (a = 10 °)
The first two descriptions were generated in POTATO. In the first structure the distances were constrained to be constant (2.00b,), while in the
second structure the angles were constrained to remain at 90°. The third description was generated using the O 1 , 0 2 and 03 positions common to
all three descriptions and the symmetry generators associated with Bmmb (nonstandard setting of space group #63, Cmcm). All other oxygens in
the unit cells can be generated from these 12 using the inversion centers on the M ions. The coordinates shown in bold face type are those where
the second and third description differ from the first description. The Bmmb coordinates have been shifted by ( - 1 / 4 , - 1 / 4 , 0 ) for comparison with
the Pmmn coordinates.
Pmmn Pmmn
(POTATO, constant distance) (POTATO, 90 ° angles) Bmmb
x y z x y z x y z
O1 I/4 -0.0438 0.0438 1/4 -0.0438 0.0438 1/4 -0.0438 0.0438
02 0.0445 1/4 0.0038 0.0445 1/4 0.0038 0.0445 1/4 0.0038
03 0.4555 1/4 0.0038 0.4555 1/4 0.0038 0.4555 1/4 0.0038
04 1/4 0.5438 0.0438 1/4 0.5438 0.0438 1/4 0.5438 0.0438
05 -0.0445 0.0038 1/4 -0.0445 0.0038 1/4 -0.0445 0.0038 1/4
06 0.5445 0.0038 1/4 0.5445 0.0038 1/4 0.5445 0.0038 1/4
07 -0.0445 0.4962 1/4 -0.0445 0.4962 1/4 -0.0445 0.,*962 1/4
08 0.5445 0.4962 1/4 0.5445 0.4962 1/4 0.5445 0.4962 1/4
09 1/4 0.0451 0.4576 1/4 0.0438 0.4576 1/4 0.0438 0.4562
O10 -0.0431 1/4 0.5113 -0.0431 1/4 0.5113 -0.0445 1/4 0.5038
O11 0.5431 1/4 0.5113 0.5431 1/4 0.5113 0.5445 1/4 0.5038
O12 1/4 0.4549 0.4576 1/4 0.4562 0.4576 1/4 0.4562 0.4562
All distances 2.000 A 1/3 of the distances 1.998 A All distances 2.000/k
1/6 of the angles 90.3 ° All angles 90° 1/3 of the angles 90.9 °

However, for tilt systems a+a+b - (#5) and a+a+a - (#7) or translational symmetry. The symmetry of the ordered
the space group P42/nmc suggested by Leinenweber & unit cell must be lower than the disordered or simple unit
Parise (1995) is equally justified and may be favored cell, because the octahedral cation sites are no longer
due to its higher symmetry. equivalent. This destroys any symmetry operations that
relate neighboring octahedral sites. In this example, half

4. Combined effect of cation ordering and

octahedral tilting A

In the cubic perovskite structure 1:1 ordering of the M-

site cations requires the unit cell to be doubled along ) (
all three crystallographic directions, in order to maintain
translational symmetry. The lattice also changes from
simple cubic (space group P m 3 m ) to face-centered cubic
(space group Fm3m). Octahedral tilting can also cause ) (
a doubling of the unit cell and often accompanies - ~ V'-
w
cation ordering. Therefore, it is instructive to consider
the combined effects of cation ordering and octahedral (a) (b)
tilting.
To illustrate the combined effect of cation ordering .9 9
and octahedral tilting consider the tetragonally distorted
a°a°c - tilt system. Upon tilting the structure distorts from
cubic to face-centered tetragonal, space group F4/mmc.
This is a nonstandard setting and the unit cell must
be redefined in order describe the structure in terms
?
of the standard space-group setting, 141mcm. T h e new
unit cell is body-centered and the relationship between (c)
7o 6"* 7o (a)
the lattice constants in the face-centered cell and the
Fig. 2. The relationship between the a and b axes in (a) the simple
new cell is at = arl21/2 and ct = CF. Introducing cation cubic perovskite cell before an octahedral tilting distortion, (b) the
order lowers the symmetry once again to 141m. The face-centered tetragonal cell after an octahedral tilting distortion,
relationship between the a and b axes in the simple (c) the body-centered tetragonai cell and (d) the ordered body-
cubic cell, the face-centered tetragonal cell and the body- centered tetragonal cell. Mirror planes parallel to the c axis cut
through the O atoms and relate the corner M ions (0,0,0) with the
centered tetragonal cell is illustrated in Fig. 2. The 1 1
M ion at the center of the ab face (.~,7,0) in the unit cell shown
introduction of cation ordering will always result in a in (c). Ordering of different cations on the M site destroys those
loss of symmetry, either space-group symmetry elements mirror planes (as well as all the glide planes), as shown in (at).
 PATRICK M. WOODWARD 37

Table 4. Space groups f o r all possible simple tilt systems are generated for each of the 23 tilt systems, distorted
both with and without 1:1 cation ordering octahedra result for several of the tilt systems, including
Tilt system Tilt system Space group Space group
tilt systems 4-7 and 17-18. This is not unexpected
number symbol (no ordering) (1:1 ordering) because they are the same tilt systems where, as shown
Three-tilt systems by geometrical arguments made earlier, the octahedra
1 a+b+ c + I m m m (#71) Pnnn (#48) are forced to distort for simple perovskites. In addition,
2 a+b+b + I m m m (#71) Pnnn (#48) tilt systems 8-11 now also show slight distortions in
3 a+ a+ a + Ira3 (#204) Pn3 (#201) their octahedra as the M---O and M ' - - O bond distances
4 a+b+ c - P m m n (#59-2) P 2 / c (#13)
5 a+ a+ c - P 4 2 / n m c (#137-2) P 4 2 / n (#86)
become increasingly different. It is uncertain at this
6 a+b+b - P m m n (#59-2) P2/c (#13) time whether this is a geometrical limitation of these
7 a+ a+ a - P 4 2 / n m c (#137-2) P4z/_n (#86) tilt systems when the M---O and M - - O bond distances
8 a+b-c - P 2 1 / m (#11-1) PI (#2) differ. More work is necessary to conclusively prove
9 a+a-c - P 2 1 / m (#11-1) Pi (#2)
this point.
10 a+b-b - Pnma (#62) P 2 1 / n (#14-2)
11 a+a a P n m a (#62) P21/n (#14-2)
12 a-b c F1 (#2) FI (#2)
13 a-b-b- I2_/a (#15-3) Fi (#2) 5. Conclusions
14 a a a R 3 c (#167-2) R3 (#148-2) With the aid of POTATO the 23 tilt systems originally
Two-tilt systems described by Glazer have been investigated. It has been
15 a°b+ c + I m m m (#71) Pnnn (#48)
shown that in the tilt systems a+a+a-, a+b÷b-, a+a÷c-,
16 a°b+b + 1 4 / m m m (#139) P 4 2 / n n m (#134)
17 a°b+c - C m c m (#63) C 2 / c (#151) a÷b+c-, a°b÷b- and a°b+c- it is not possible to link
18 a°b+b - C m c m (#63_) C2/_c (#151) together a three-dimensional network of perfectly rigid
19 a°b-c - 1 2 / m (#123) 11 (#2) _ octahedra. In these tilt systems small distortions of the
20 a°b-b - I m m a (#74) 1 2 / m (#123) octahedra must occur to preserve connectivity of the
One-tilt systems octahedra. The distortions are smaller in the a°b+c-
21 a°a°c + P 4 / m b m (#127) P 4 / m n c (#128)
and a°b+b- tilt systems than they are in the other
22 a°a°c 1 4 / m c m (# 140) 14/m (#87)
four tilt systems. These distortions can have important
Zero-tilt systems
23 a°a°a° P m 3 m (#221) F m 3 m (#225)
consequences, particularly on the crystallography. For
example, in tilt systems a+a+c- (#5) and a+a÷a- (#7)
it appears as though P421nmc is a more reasonable
of the symmetry elements are destroyed, including the space-group assignment than Pmmn, based on its higher
mirror plane perpendicular to the fourfold axis and c symmetry. Finally, a table of predicted space groups
glide planes parallel to the fourfold axis, lowering the for compounds that display both cation ordering and
symmetry of the ordered a°a°c - from 14/mcm to 14/m. octahedral tilting has been derived.
In a similar manner the space groups for ordered
perovskites in all 23 of the simple tilt systems can be I would like to thank Dr Arthur Sleight for his support
determined. This analysis was performed using Inter- and many helpful suggestions throughout this project. I
national Tables f o r X-ray Crystallography (1983, Vol. would also like to thank Dr John Evans for a careful
A) and double-checked using POTATO with an ordered and critical reading of the manuscript and countless
arrangement of cations. The results are given in Table 4 discussions over the entire period of this work.
along with the final space-group assignments for simple
tilt systems. For each tilt system the space group describ-
ing the ordered perovskite is a maximal nonisomorphic
subgroup of the space group describing the simple
APPENDIX A
Standard unit-cell descriptions of distorted
perovskite and, of course, all the ordered perovskite
perovskites
space groups are subgroups of Fm3m. Unlike Glazer's
work with the tilt systems of simple perovskites, there A practical difficulty of working with distorted per-
is no precedent in the literature with which to compare ovskites is the widespread use of nonstandard cell set-
the results of Table 4. However, Leinenweber (1996) tings. These types of descriptions are useful when com-
has derived a similar table of predicted space groups paring two different structures, but can be confusing
for ordered perovskites, using a completely different when many different compounds and space groups are
approach, similar to the method used originally by examined. Transforming to a standard setting, which
Glazer, and his results are in complete agreement with is necessary when using many software packages, can
the results of Table 4. sometimes be confusing, time consuming and lead to
POTATO is also able to generate unit-cell descriptions possible errors. Adding to the confusion is the fact that
for ordered perovskites A2MMt06, where the MI---O among the space groups corresponding to the different
distance differs from the M - - O distance. When the tilt systems there are several different choices of origin
two bond distances are nonequivalent and structures and unit-cell size. In the course of this analysis, using
38 OCTAHEDRAL TILTING IN PEROVSKITES. I.

T a b l e 5. A t o m i c p o s i t i o n s a n d unit-cell d e s c r i p t i o n s f o r all the s p a c e g r o u p s g e n e r a t e d by s i m p l e tilting o f the M O 6

octahedra
The number immediately below the space group name is the space group number in International Tables for X-ray Crystallography (1983,
Vol. A). All the space-group descriptions are standard settings. The other number in the first column signifies the tilt systems corresponding to
each space group.
Space Unit-cell
group size Cation positions Anion positions Cornments
Immm a ~- 2ap A cation (81) O,y,z y~--~,z~--¼ Shift by I, I,
(A-71) b ~_ 2ap (2.a) O, O, 0 (8m) x,O,z x~--~,z "~!
c~--2ap (2b) ~i ~,
1 ~.
I
(8n) x,y,O x ~ _ ~ , y _2~
a
Tilt systems (2c) ~, 0
1 , 2 a n d 15 (2d) ~,0,½
M cation
(8k) 1,1,1

Ira3 a ~_ 2ap A cation (24g) O, y,z Y "~ I,~-- I Shift by ~,7,,I

I 1
(A-204) b x 2ap (2a) 0, 0, 0
c~_2ap (6b) 0,½,½
Tilt system M cation
3 (8c) ~,~,I
11

Pmmn a ~_ 2ap A cation (4e) ~,y,z y~-O,z~-O No origin shift

(A-59-2) b _~ 2ap (4e) ~,y,z y~-O,z~-~
(2a) i,i,z z_!
c ~_ 2ae (2a) ~,~z z -~ (4f) x , ! , z y~-O,z~_O Out-of-phase
Tilt systems (2b) i,~ z Z--~ (4f) x,~-,z y~--O,z~-~ tilting about the c
4 and 6 (2b) - ~ - z z - - 4 - (8g) x,y,z axis
M cation x~--O,y~-O,z~_~
(4c) 0 , 0 , 0
(4d) 1,½,½

P42/nmc a ~_ 2ap A cation (8g) ~, y,z y ~-- O,z~-- O No origin shift

(A-137-2) b _~ 2ap (8g) 2,Y,Z y -~- !2 , Z~--
c~2ap (e) x,-x,1 x~_½ Out-of-phase
Tilt systems (4d) z z ~- I tilting about the
5 and 7 M cation c axis.
(8e) 0 , 0 , 0

P21/m a ~-- 21/2% Acation (2e) x, i , z x ~-- O, z ~_ ½ Shift by O, O, ½

(A-11-1) b _~ 2a (2e) x , i , z x~_O,z~_O (2e) x,~ z x ~ - i , z ~ - O
c "~ 21~ap (2e) x,z z x ~ , z ½ (4f) x,y,z In-phase tilting
Tilt systems /3 # 90 ° M cation x _~ ~, y _~ 0, z -- 1 about the b axis
8 and 9 (2b) ½,0,0 (4f) x,y,z
(2c) 0,0,½ x~_1,y~_O,z "-'3

Pnma a ~_ 21/2ap Acation (4c) x , l , z x_~0, z___½ Shift by 0, 0, ½

(A-62) b _~ 2a, (4c) x, ¼,z x _ 0 , z _ 0 (8d) x,y,z
Tilt systems c 21/'2ap M cation x ~ ~, y _~ 0, z -~ In-phase tilting
10and 11 /3#90 ° (4b) ~, 0, 0 about the b axis

Fi a _~ 2ap A cation (8i) x, y, z No origin shift

(A-2) b_~2ap (8i)* x,y,z x_~,y~_0, z~0
Tilt system c ~_ 2ap x ~_ ¼, y ~_ I , z ~- ~ (8i) x, y, z The standard
12 ct#fl#y#90 ° M cation z~--O,y~-1,z~--O setting is P i , add
(4a) 0, 0, 0 (80 x, y, z face-centered
(4b) 0, 0, ½ x _~ 0, y ~ 0, z ~ / generators for F i

12/a a ~_ 2ap Acation (4e) -14,y,0 y_~½ Shift by 0, ½,0

(A-15-3) b 2112ap (4e) ~,y, 0 y--~0 (8f) x,y,z
Tilt system c ~_ 21/2ap M cation x _~ 0, y _~ ~, z ~ ¼
13 /3#90 ° (4b) 0, ½,0

R3c a ~_ 21/2a A cation (6e) x, ½ - x , ¼ x~_~ No origin shift

(A-167-2) b ~ 2 1/2 app (2a) 1 l
Tilt system c ~_ 21/2ap M cation Rhombohedral
14 ot # , 6 # y--~ 60 ° (2b) 0 , 0 , 0 cell
 PATRICK M. WOODWARD 39

Table 5 (cont.)
Space Unit-cell
group size Cation positions Anion positions Comments
14/mmm a ~-- 2ap A cation (8h) x, x, 0 x _ ~- Shift by - ~ 1, ~ ,l ¼
(A-139) b~_2ap (2a) 0 , 0 , 0 (16n) O,y,z y~-~,z~-¼
c "~ 2% (2b) 0, 0, ½ No tilting about the
Tilt system (4c) ½,0, 0 c axis
16 M cation
(8:) ~,-:, ~
Cmcm a ~- 2at, A cation (8e) x, 0, 0 x ~-- ~ Shift by ¼, - 41-,0
(A-63) b~_2at, (4c) 0, y, i y ' ~ 0 (8f) O,y,z y ~ - ! , z ' ~ O
c_~2at, (4c) 0, y,a y-~½ (8g) x,Y, ¼ x !,y_~¼ In-phase tilting
Tilt systems M cation about the c axis
17 and 18 (8d) ¼,¼,0
12/m a "~ 21/2% A cation (4i) x, 0, z x ~ -~, z ~-- ¼ Shift by ~,~,~1
1
,~1
(A-12-3) b~axf2qt,c M(4i)cationX'O'z x ~ - ~ , z ~ - ] (4g)(4h) :,?'y'0y,
0 Y~'iY
- - No tilting about the
Tilt system /~ ~ 90 ° (4e) -~, ¼,¼ b axis
19

Imma a~2a. Acation (4e) 0,~,z z--~ Shift by ¼, ~, ¼

(A-74) b ~ 21]-2a (4e) 0, ¼,z z -~ 3 (8g) x, 0,0 x ~-
c --~ 2 1/2 at,p M cation No tilting about the
Tilt system (4b) =~,
i ~,
1 ~ a axis
20

P4/mbm a ~_ 21/2a_ A cation (2b) 0, 0, ~ ~_ ~ No origin shift

(A-127) b 21/2a~ (2c) 0, ~,½ (4g) x,x + ½,0 x
c ~-- at, M cation In-phase tilting
Tilt system (2a) 0, 0, 0 about the c axis
21

14/mcm a ~_ 21/2al/2p A cation l, (4a) 0, 0, ¼ No origin shift

(A-140) b~2 at, (4b) 0, ¼ (8h) x,x+½,0 x__
c ~_ 2at, M cation Out-of-phase
Tilt system (4c) 0, 0, 0 tilting about
22 the c axis

Pm3m a = b = c = at, A cation (3d) ½,0, 0 No origin shift

(A-221) (lb) ~,~,½
1 t
Tilt system M cation
23 (la) 0 , 0 , 0

Table 6. Atomic positions and unit-cell descriptions for all the space groups generated by tilting of the octahedra in
combination with 1:1 cation ordering
The number immediately below the space group name is the space-group number in International Tables for X-ray Crystallography (1983,
Vol. A). All the space-group descriptions are standard settings. The other number in the first column signifies the tilt systems corresponding to
each space group. Note that the ordered tilt system #20 and simple tilt system # 19 both are assigned to space group 12/m, but the choice of origin
and atomic positions are different for the two structures.
Space Unit-cell
group size Cation positions Anion positions Comments
Pnnn a ~- 2% A cation (8m) x, y, z Shift by ~, ~, ¼
(A-48-1) b"2ap (2a) 0 , 0 , 0 x"O,y~-¼,z~-]
c~-2ap (2b) 0,!,½ (8m) x,y,z =¼
Tilt systems (2c) ½0,½',~'? x~-l'y~-O'z
1, 2 and 15 (2d) : (8m) x,y,z
M and M' cations x "" ~, y "" ¼, z -~ 0
(4e) i'i'i (M)
(4f) ~,~,~ (g')
Pn3 a ~_ 2ap A cation (24h) x, y, z Shift by ~ 1 l
(A-201-1) b~_2ap (2a) 0 , 0 , 0 x~--O,Y~--¼,z~--]
c~_~. (~ o,½,~
40 OCTAHEDRAL TILTING IN PEROVSKITES. I.

Table 6 (cont.)
Space Unit-cell
group size Cation positions Anion positions Comments

Tilt system M and M' cations

3 (4b) i'i'i (M)
(4c) a ~'a (M')

e2/c A cation 6x(4g) x, y, Z No origin shift

(A-13-5) b_~2ap (2e) I z .,~ l x~_¼,Y~_O,z'~O
¢~_2., (2e) ~'Z Z~.}Z x~_O,y~--¼,z~--O Out-of-phase
Tilt systems y~90 ° x~--O,y'~O,z~ tilting about the
4 and 6 (2f) -, z -- - x=O,y=~,zz} c axis
M and M' cations x~i,Y~O'zzi
(2a) 0 , 0 , 0 (M) x ' ~ y~--O,z---
(2b) I,o,i (M)
(2c) 0,0,~ (M')
(2d) ½,0,0 (M')

!i
a ~_2ap A cation (8g) x,y,z No origin shift
P42/n
(A-86) b_~2% x~_¼,y~--O,z~--O
c_~2% (8g) x, y, z Out-of-phase
Tilt systems (4e) -,1,Z Z--- x"~¼,y'~O,z~_½ tilting about the
M and M' (8g) x,y,z, c axis
5 and 7
(4c) 0 , 0 , 0 (M) x~_O,y~--O,z~--¼
(4d) ½,½,½ (M')
P2~/n a ~_ 2~/2a A cation (4e) x,y,z Shift by 0, I, 0
(A-14-2) b 2 1/2 at,p (4e) x, y, Z x~_~,y~_¼,z~--O
c ~_ 2% x"~O,y~_O,z~ (4e) x, y, z In-phase tilting
Tilt systems ~¢90 o M and M' cations x~,y~¼,z=½ about the b axis
10 and 11 (4c) 0,½,0 (M) (4e) x,y,z
(4d) ½,0,0 (M') x=½,y=O,z=¼
A cation 12x(2i) x,y,z No origin shift
pi P i and F i
(A-2) (2i) x,y,z x~_¼,y~.O,z~_O
Tilt systems a ~_ 2ap x~_O,y~-~,z~O The standard
8 and 9 b ~_ 2ap (20 x,y,z x~_O,y~--O,z~--~ setting is P i
Fi c ~_ 2~ x~-l,y=¼,z=l xz~,Y=½ ' z = 0
X ~1

(A-2) (20 x,y,z -,y~_O,x~_ ! For F i use face-

x'~l,y~-l,~-I centered generators
Tilt systems
(20 x,y,z in addition
1, 2 and 13 xx~O,y~---
~i'Y~i' z
li x~_~,y~_~,z=~
(A-2) a ~-- 21/2ap M cation y~o',z For l i use body-
Tilt system b=2a (la) + (le) + (If) + (lg) x~_~_'y~,t z~-.~ centered generators
19 c ~_ 21~ap M' cation x~- y~-- ~-" in addition
~¢9o o (lb) + (lc) + (ld) + (lh)
..~ 2~/2a A cation (6f) x, y,z No origin shift
R~
(A-148-2) a ~.. 21/2aP- (2c) x,x,x x'=- x=l,y=~,z=¼
b _ 1/2 p Rhombohedral
c ~ 2 ap M and M' cations
Tilt system a_~C~Cr=60 ° (la) 0 , 0 , 0 (M) cell
14 (lb) ½,½,½ (M')

a ~_2ap A cation (8m) x,x,z Shift b y - l ~,~,¼

1
P42/nnm
(A-134-1) b ~_2ap (2a) o,o,o x~_~,z'~O
c = 2% (2b) o, o, ½ (16n) x,y,z No tilting about
(4c) I,o,o x=~,y=½,z~l the c axis
Tilt system
16 M and M' cations
(4e) i'i'i (M)
(4f) ~ =,,Z (M')

A cation (8f) x,y,z Shift by I , - ¼,0

C2/c
(A-~5-1) b-~ (4e) 0 , y , i y -~0 x~_~,y~--O,z~--O
c=2a, (4e) 0,y,~ y~-½ (8f) x, y, z In-phase tilting
~¢9o o M and M' cations x'~O,y~,z~-O about the c axis
Tilt systems
17 and 18 (4c) ~.~,o
~ I (M) (sf) x, y,z
(4d) ~,a,~ (M') x~l,y~--~,Z~-- 1
 PATRICK M. W O O D W A R D 41

Table 6 ( c o n t . )
Space Unit-cell
group size Cation positions Anion positions Comments
12/m a ~- 21/2a_ A cation (4i) x,O,z x~--O,z~--¼ No origin shift
(A-12-3) b 21/2a~ (40 x,O,z x~-½,z~-¼ (8j) x,y,z,
c~_Z% M and M' cations x'~,y~_~,z~_O No tilting about
Tilt system /~ g: 90° (2a) 0,0,0 (M) the c axis
20 (2a) ½,½,o (M')
P4/mnc a ~- 21/2a_~, A cation (4e) 0,0, z z~--¼ No origin shift
(A-128) b ~_ 21/2ae (2c) 0, ½, (8h) x,y,O x~--~,y~-~
~_~ 2a~ M and M' cations In-phase tilting
Tilt system (2a) 0,0,0 (g) about the c axis
21 (2b) 0,0,½ (M')
14/m a ~ 21/2a_ A cation (4e) 0,0, z --41 No origin shift
(A-87) b 212a; (rib) 0, ½, (8h) x,y,O x ~ - ~ , y ~ - ¼
c___z% M and M' cations Out-of-phase
Tilt system (2a) 0,0,0 (M) tilting about the c
22 (2b) 0,0,½ (M') axis

Fm3m a : b = c = 2ap A cation (24e) x, 0,0 x~¼ No origin shift

(A-225) (8c) ~,~,~
M and M' cations
Tilt system (4a) 0,0,0 (M)
23 (4b) ½,½,½ (M')

P O T A T O , the standard space-group description of each Vh =Alx+Bly+ Clz,

tilt system was determined. These descriptions are given Vly = A2x + B2y + C2x,
in Table 5 for simple perovskites and in Table 6 for
ordered perovskites. One note of caution: both the a ° b c Vtz = A3x + B3y + C3z.
(#19) tilt system among simple perovskites (Table 5)
While for the second octahedron under consideration
and the a ° b - b - (#20) tilt system for ordered perovskites
the metal--oxygen vectors describing the tilting of this
(Table 6) belong to the 12/m space group, but the atomic
octahedra can be defined as
positions are different for the two tilt systems. Hopefully,
this information will be of practical assistance to others
working with distorted perovskite structures. V2x - a l x + b l y + clz,
V2y = a2x +b2y + c2z,
APPENDIX B V2z = a3x + b3y + c3z.
Vector proof of distorted octahedra in tilt systems
4, 5, 6, 7, 17 and 18 The relationship between these six vectors is shown in
Fig. 3. The other three M - - O vectors associated with
In Glazer's original paper he points out that a mirror
each octahedron can be generated using the inversion
plane exists perpendicular to any axis about which
center at the origin of the octahedron and provide
there is either no tilting or in-phase tilting (Glazer,
no additional information about the orientation of the
1972). These mirror planes are located halfway between
the octahedra they separate (for example, at x = :~ i and octahedra.
The inversion center that must be present at the origin
x = ~ 3 in the doubled unit cell). With this in mind of each octahedron restricts the M-cation positions to
consider any of the + + - or 0+-- tilt systems mentioned lie on the unit-cell axes. The metal centers of these
above. All these tilt systems have mirror planes located two octahedra will be located at (0,0,0) and (0,0,½).
perpendicular to the x and y axes. As a consequence of Couple this with the fact that the axes are orthogonal
having two mutually perpendicular mirror planes these and it is clear that in order for the two octahedra to
tilt systems must have orthogonal axes. Now consider be linked together and the second metal cation to be
the octahedron centered at the origin and the octahedron located directly above the first metal cation on the z axis
located directly above it along the c axis. The orientation the following relations must hold
of each octahedron can be uniquely described by three
metal--oxygen vectors pointing approximately down the
a3 =A3,
x, y and z axes. For the octahedron at the origin these
vectors will be denoted by b3 = B3,
42 OCTAHEDRAL TILTING IN PEROVSKITES. I.

and in order to keep all the M - - O vectors the same The first two equations must be satisfied in order to
length (necessary to maintain perfect octahedra), the keep all the M - - O distances constant (where D is the
third component of vector V2,, is also uniquely deter- M - - O bond distance) and the last three equations must
mined as be satisfied to keep all the O - M - - O bond angles 90 °.
This leaves only four unknowns available to satisfy five
c3= - C3. linearly independent equations. In general, this will not
be possible and one of the equations above will not be
Therefore, in any of the ++- or 0 + - tilt systems an satisfied. The physical consequence of this is that either
arbitrary tilt of the octahedron at the origin fixes the one of the bond angles or bond distances must distort
vector V2z. from its ideal value in a perfect octahedron. To test this
The vectors V2x and Vey are also subject to several assertion calculations were carried out to ensure that
constraints. To begin with, the oxygen ions at the end for an arbitrary tilt of the first octahedron, it was not
of both these vectors must lie on the mirror planes possible to simultaneously satisfy (1)-(5) for the second
discussed previously. This requirement also holds true octahedron. The calculation of V2x, V2y and V2z for tilt
for the octahedron at the origin so that system a + a * a - is given in Appendix C, as an example.

1
al =AI = -a,
2 APPENDIX C
1 Calculation of metal oxygen vectors in the a÷a÷a -
b 2 = B2 = - b ,
2 tilt system
Consider the a+a+a - tilt system with three 10 ° clockwise
where a and b are the lattice constants along the x and y
tilts and a metal--oxygen distance of 2.00 ,~. Performing
axes, respectively. Now the only unknown components
10 ° tilts about each of the three Cartesian axes results
remaining are bl, cl, a2 and c2. In other words, once
in the origin octahedron having an orientation described
the tilting of the first octahedron has been specified, five
by the following vectors
of the nine coefficients describing the orientation of the
second octahedron are determined. However, in order to
Vlx = 1.9397x - 0.3132y + 0.3736z,
keep the second octahedron perfectly rigid the following
equations must be satisfied Vly = 0.3736x + 1.9397y - 0.3132z,
V1z = 0.3133x + 0.3736y + 1.9397z,
D 2 = a l 2 + b l 2 + c l 2, (1)
D 2 = a22 + b22 + c22, (2) where the coefficients represent distances in/~ between
0 = (a 1)(a2) + (b 1)(b2) +(c 1)(c2) (3) the metal ion at the origin and the three oxygen ions
defining the octahedral orientation. Imposing the restric-
0 = (al)(a3) + (bl)(b3) + (cl)(c3)
(4) tions placed on the second octahedron by the mirror
0 = (a2)(a3) + (b2)(b3) + (c2)(c3). (5) planes and the orthogonal axes gives the following
values for the M - - O vectors of the second octahedron

V2x = 1.9397x + b l y + clz,

V2y = a2x + 1.9397y - c2z,
V2z = 0.3133x + 0.3736y - 1.9397z.

V2z By combining (2) and (4) above the vector V2x can be
found so that it is perpendicular to V2z and of length
2.00/1,. When this calculation is c a m e d out a quadratic
equation in c2 is obtained. Since the equation is a
lz
quadratic there are two solutions

Vex = 1.9397x + 0.4299y - 0.2305z

Vlx r y Vex = 1.9397x - 0.313 ly - 0.3736z.

Both of these vectors are 2.00/1, long and perpendicular

to V2z, but the second vector corresponds to in-phase
tilting about the z axis, while the first vector is correct
x
Fig. 3. The metal--oxygen vectors discussed in the text for two for out-of-phase tilting about the z axis. All the tilt
octahedra linked along the z axis of an a ÷ a ÷ a - tilted perovskite. systems considered here have out-of-phase tilting about
 PATRICK M. W O O D W A R D 43

the z axis, so the first of the two vectors above will be Bricefio, G., Chang, H., Sun, X., Schultz, P. G. & Xiang, X.-D.
taken as V2x and the second vector discarded. The final (1995). Science, 270, 273-275.
vector V2y must be perpendicular to both V2x and Vzz. Brynestad, J., Yakel, H. L. & Smith, G. P. (1966). J. Chem.
A vector mutually perpendicular to these two vectors Phys. 45, 4652-4664.
Burns, G. & Glazer, A. M. (1990). Space Groups for Solid
can be generated by taking the cross product V2x x V2z.
State Scientists, 2nd ed., Appendix A9-6. Boston: Academic
The magnitude of this vector is now scaled so that its y
Press.
component is 1.9397, giving the vector Clearfield, A. (1963). Acta Cryst. 16, 135-142.
Deblieck, R. (1986). Acta Cryst. A42, 318-325.
V2y = - 0.3783x + 1.9397y + 0.4347z. Deblieck, R., Van Tendeloo, G., Van Landuyt, J. & Amelinckx,
S. (1985). Acta Cryst. B41, 319-329.
Combining the above calculations the orientation of Fontcuberta, J., Martinez, B., Seffar, A., Pifiol, S., Garcia-
the second octahedron can be described by the following Munoz, J. L. & Obradors, X. (1996). Phys. Rev. Lett. 76,
set of three vectors 1122-I125.
Ga[asso, F. S. (1969). Structure, Properties, and Preparation
V2x = 1.9397x + 0.4299y - 0.2305z, of Perovskite Type Compounds. Oxford: Pergamon Press.
V2y = - 0.3783x + 1.9397y + 0.4347z, Glazer, A. M. (1972). Acta Cryst. B28, 3384-3392.
Goodenough, J. B. & Longo, J. M. (1970). Landolt-Bi~rnstein,
V2z = 0.3133x + 0.3736y - 1.9397z. 4, 126-275.
Horowitz, A., Amit, M., Makovsky, Ben Dor, L. & Kalman,
The way in which these vectors were calculated Z. H. (1982). J. Solid State Chem. 43, 107-125.
guarantees they will satisfy (1) and (3)-(5), but no Hwang, H. Y., Cheong, S.-W., Radaelli, P. G., Marezio, M. &
constraint has been placed on the length of vector V2~,. Batlogg, B. (1995). Phys. Rev. Lett. 75, 914-917.
Calculating the length of V2y gives a value of 2.024 A, Knochenmuss, F., Reber, C., Rajasekharan, M. V. & Giidel,
slightly longer than all the other M - - O vectors. This H. U. (1986). J. Chem. Phys. 85, 4280-4289.
result can be extended to all the ++- and 0 + - tilt systems. Leinenweber, K. (1996). Personal communication.
Therefore, we see that it is mathematically impossible to Leinenweber, K. & Parise, J. (1995). J. Solid State Chem. 114,
achieve linked octahedra in these tilt systems without 277-281.
Megaw, H. D. & Darlington, C. N. W. (1975). Acta Cryst.
distortions of the octahedra. The magnitude of these
A31, 161-173.
distortions will depend on the size of the tilt angles and Messer, C. E., Eastman, J. C., Mers, R. G. & Maeland, J. A.
the length of the M - - O bonds. (1964). Inorg. Chem. 3, 776-778.
Note that when calculating the V2,, vector two so- No~l, H., Padiou, J. & Prigent (1975). C. R. Acad. Sci. Ser.
lutions were found, the second vector was discarded C, 280, 123-126.
because it corresponded to in-phase tilting about all Okazaki, A. & Suemune, Y. (1961). J. Phys. Soc. Jpn, 28,
three axes. However, if the calculations are carried 443-446.
through with this vector the final orientation of the O'Keeffe, M. & Hyde, B. G. (1977). Acta Cryst. B33,
second octahedra satisfies all five equations. This is in 3802-3813.
Rodier, N. & Laruelle, P. (1970). C. R. Acad. Sci. 270,
agreement with the fact that POTATO is able to generate
2127-2130.
structures possessing perfectly rigid octahedra in the +++ Sasaki, S., Prewitt, C. T., Bass, J. D. & Schulze, W. A. (1987).
and 0++ tilt classes. Acta Cryst. C43, 1668-1674.
Shirane, G., Danner, H., Pepinski, R. (1957). Phys. Rev. 105,
856-860.
References Sleight, A. W., Gillson, J. L. & Bierstedt, P. E. (1975). Solid
Aleksandrov, K. S. (1976). Kristallografiya, 21, 249-255. State Commun. 17, 27-28.
Bacher, P., Antoine, P., Marchand, R. L., Haridon, P., Laurent, Thomas, N. W. (1989). Acta Cryst. B45, 337-344.
Y. & Roult, G. (1988). J. Solid State Chem. 77, 67-71. Thomas, N. W. (1996). Acta Cryst. B52, 16-31.
Biosym Technologies (1993). Insight H. Crystal Structure Thomas, N. W. & Beitollahi, A. (1994). Acta Cryst. B50,
Analysis Software. Version 2.3.5. San Diego: Biosym Tech- 1549-560.
nologies. Woodward, P. M. (1997). J. Appl. Cryst. Submitted.