Journal of Alloys and Compounds 372 (2004) 40–48

Formability of ABO3 perovskites

Chonghe Li a , Kitty Chi Kwan Soh b , Ping Wu a,∗
a Institute of High Performance Computing, 1 Science Park Road, 01-01 The Capricorn, Singapore 117528, Singapore
b Department of Materials, National University of Singapore, Lower Kent Ridge Road, Singapore 119260, Singapore

Received 7 July 2003; received in revised form 6 October 2003; accepted 6 October 2003

Abstract

Regularities governing perovskite formability are investigated, by using empirical structure map methods and a total of 197 binary oxide
systems. Each of the 197 systems belongs to one of the three groups (A2 O–B2 O5 , AO–BO2 and A2 O3 –B2 O3 ), and among them √ only 121
systems have a perovskite compound. It is found that the octahedral factor, rB /rO , is as important as the tolerance factor, (rA +rO )/ 2(rB +rO ),
with regards to the formability of perovskite. On the structure map constructed by these two parameters, sample points representing systems of
forming (perovskite) and non-forming, are distributed in distinctively different regions. Prediction criterions for the formabilities of perovskites
are obtained, by which only four systems are misclassified. The developed models can be used to search for new perovskites by screening
all possible elemental combinations in any of the above three groups. It may assist in the design of new substrate or buffer materials for
semiconductors where perovskite can play an important role.
© 2003 Elsevier B.V. All rights reserved.

Keywords: Perovskite; Oxide; Tolerance factor; Octahedron; Formability

1. Introduction try interests of perovskites are: colossal magnetoresistance

Perovskite is one of the most frequently encountered
structures in solid-state inorganic chemistry, and it accom-
modates most of the metallic ions in the periodic table with
a significant number of different anions. An ideal perovskite
structure has an ABX3 stoichiometry and a cubic crystal
structure, which is composed of a three-dimensional frame-
work of corner-sharing BX6 octahedron. The A-site cation
fills the 12 coordinate cavities formed by the BX3 network
and is surrounded by 12 equidistant anions [1], as seen
in Fig. 1. Although majority of the perovskite compounds
are oxides or fluorides, other forms like heavier halides,
sulphides, hydrides, cyanides, oxyfluorides and oxynitrides
are also reported [2,3].
Perovskite and perovskite-related materials are important
crystal structure due to their diverse physical/chemistry
properties [2,4,5] over a wide temperature range. For ex-
ample, perovskite ceramics with ferroelectric and/or piezo-
electric properties, such as BaTiO3 and Pb(ZrTi)O3 play a
dominant role in the electroceramics industry. Other indus-
∗ Corresponding author. Tel.: +65-6419-1212; fax: +65-6778-0522.
E-mail address: wuping@ihpc.a-star.edu.sg (P. Wu).
((SrCaLa)MnO3 ), PTC thermistor (BaTiO3 ), electroopti-
cal modulator ((PbLa)(ZrTi)O3 , BaTiO3 ), optical switch
(LiNbO3 ), battery Material ((Li0.5−x Lax TiO3 )), etc.
It is of interest to find out regularities governing the for-
mation of perovskite-type compounds and use it to fur-
ther guide the exploration of new materials. In early 1920s,
Goldschmidt√ [6] has proposed a “tolerance factor” (t =
rA + rO / 2(rB + rO ), where rA , rB and rO are the ionic
radii of A, B and O, respectively) to study the stability of
perovskites. Geometrically, for an ideal perovskite, the ra-
tio of D(A–O), the bond length of√A–O bond, to D(B–O),
the bond length of B–O bond, is 2 : 1. Thus, if the bond
length is roughly assumed to be the sum of two ionic radii,
the t value of an ideal perovskite should be equal to 1.0.
However, Goldschmidt found that, as an experimental fact, t
values of most cubic perovskites are in the range of 0.8–0.9,
and distorted perovskites occur in somewhere wider range
of t.
Goldschmidt’s tolerance factor t has been widely ac-
cepted as a criterion for the formation of the perovskite
structure, a number of researchers have used it to discuss
the perovskite stability, including oxides [7–11], fluorides
[12–14], chlorides [15]. Up to now, almost all known per-
ovskite compounds have t values in the range of 0.75–1.00.

0925-8388/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2003.10.017
 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48
Fig. 1. Ideal perovskite structure illustrated for ABO3 . Note the corner-
shared octahedra, extending in three dimensions.
However, it seems that t = 0.75–1.00 is not a sufficient
condition for the formation of the perovskite structure, as
indicated later, for some systems whose t are even within the
most favourable range (0.8–0.9), no perovskite structure is
stable.
Muller and Roy [1] proposed to plot “structural map”,
which took the ionic radii of A and B as coordinates to study
the distribution of different structures for many ternary
structural families, including A2 BX4 , ABX4 and ABX3
systems [1]. Furthermore, the schematic distribution map
of different crystal structure for A1+ B5+ O3 , A2+ B4+ O3 ,
and A3+ B3+ O3 systems separately, were given by the same
method [1,16]. However, the criterion for perovskite formal-
ity was not discussed, possibly due to the lack of accurate
data of crystal structure of some ABO3 compounds at that
time.
In China, a group of scientists [17] use a pattern
recognition—atomic parameter method to study the regular-
ities of formation of perovskite, including simple (ABO3 -
type compound) and complex oxides (A(B1 B2 )O3 -type
compounds). Their model contains seven atomic parame-
ters, including the radius of ion A and ion B, the electroneg-
ativities of ion A and ion B, and the d electron number of
ion B. The underline physics is not quite clear and a special
software package is needed when predicting new compound
formation using their model.
In this paper, 197 ABO3 -type compounds (of which, 121
are perovskite, and the rests are non-perovskite) are collected
from Phase Diagram for Ceramics (I–XII) [18]. First of all,
more recently available systems are used to test Muller and
Roy’s results [1]. And a structural map spanned by ionic
radius of cations A and B is obtained combining all three
41
groups (A1+ B5+ O3 , A2+ B4+ O3 , and A3+ B3+ O3 ) in one
map, which further indicated that the different charges of the
systems have little effect on the perovskite formability rule.
The tolerance factor t and a newly defined octahedral factor
(the radii ratio of the small cation B over the anion O) are
further applied to construct a new and effective structural
map leading to new criteria of the formability of perovskite
compounds.
2. Structural map method
The most important atomic parameter that dominates the
crystal structure of ionic compound is the ionic radius, which
may be explained by a thermodynamic analysis. By the Sec-
ond Law of thermodynamics, the Gibbs free energy of a
solid is:
G = H − TS (1)
G = U + PV + 1
2 hν − TS (2)
Thus, Gibbs free energy = lattice energy + terms which are
very small at low P and T.
Thus, at low temperatures and pressures, G ≈ U, i.e.
the lattice energy determines which structure is stable. The
lattice energy U for ionic compounds can be expressed as
[1]:
−Mz2 e2
U= (3)
rc + r a
where M is the Madelung constant, which is a constant for
certain structure, z the valence of ions, rc and ra the radii of
anion and cation. So it can be seen that the crucial impor-
tance of ionic radius is rooted directly in the fact that the
lattice energy determines the Gibbs free energy and that in
turn the lattice energy bears a simple inverse relationship to
the interatomic separation (sum of cation and anion radius).
In 1959, Mooser and Perason [19] applied a two-dimension
graphic to study the stability of different compounds, the
two factors they used were the difference of electroneg-
ativity between the cation and the anion and the average
principle quantum number. They succeeded to discriminate
the crystal structures of AB-type compounds, of AX2 -type
halides, and the metallic or non-metallic ternary fluorides
ABX. Similar methods were called the structural map tech-
nology, and more parameters were used to draw the graphic.
In this paper, the formability of perovskite ABO3 will be
studied by a similar method.
3. Data collection
A total of 197 binary oxide systems are collected from
Phase Diagram for Ceramics [18], which can be divided
into three groups: A2 O–B2 O5 system (corresponding to the
42 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48

A+ B5+ O3 -type perovskites); AO-BO2 system (correspond- Table 1 (Continued )

ing to A2+ B4+ O3 -type perovskites) and A2 O3 –B2 O3 system No. System Formability rA rB Tolerance rB /rO
(corresponding to A3+ B3+ O3 -type perovskites). Of which, (A) (A)
121 systems (denoted ‘yes’) are found to have perovskite 14 K2 O–P2 O5 No 1.38 0.38 1.14 0.30
compounds and 76 systems (denoted ‘no’) have no per- 15 Li2 O–P2 O5 No 0.76 0.38 0.87 0.30
ovskite compounds. 16 Tl2 O–Sb2 O5 No 1.50 0.60 1.05 0.48
17 Na2 O–Sb2 O5 No 1.02 0.60 0.87 0.48
According to Lufaso and Woodward [5], an ideal cu-
18 Li2 O–Sb2 O5 No 0.76 0.60 0.77 0.48
bic perovskite can transform into other crystal structure 19 K2 O–Sb2 O5 No 1.38 0.60 1.00 0.48
through tilting the octahedral BO6 , these resulted structures 20 Na2 O–Bi2 O5 No 1.02 0.76 0.80 0.60
have close relationship to the ideal perovskite. In general, 21 Li2 O–Bi2 O5 No 0.76 0.76 0.71 0.60
all compounds with these 15 kinds of structures (one ideal 22 Ag2 O–Bi2 O5 No 1.15 0.76 0.84 0.60
23 K2 O–Nb2 O5 Yes 1.38 0.64 0.98 0.51
structure plus 14 structures resulted by the tilting) belong
24 K2 O–Ta2 O5 Yes 1.38 0.64 0.98 0.51
to ‘perovskite’ catalog. It is well known that [1,2] crystal 25 Na2 O–Nb2 O5 Yes 1.02 0.64 0.85 0.51
structures can change with variation of temperature and/or 26 Na2 O–Ta2 O5 Yes 1.02 0.64 0.85 0.51
pressure. Compounds with other structure can, therefore, 27 K2 O–U2 O5 Yes 1.38 0.76 0.92 0.60
transform into perovskite, and vice versa at different tem- 28 Na2 O–U2 O5 Yes 1.02 0.76 0.80 0.60
29 Na2 O–W2 O5 Yes 1.02 0.62 0.86 0.49
perature and/or pressure. So the criterion for classifying a
30 Na2 O–V2 O5 Yes 1.02 0.54 0.90 0.43
forming system (denoted ‘yes’ in Table 1) is that a com- 31 Ag2 O–V2 O5 Yes 1.15 0.54 0.95 0.43
pound of perovskite (any of the 15 crystal structures) is 32 Ag2 O–Ta2 O5 Yes 1.15 0.64 0.90 0.51
stable at room temperature and one atmosphere. Systems 33 Ag2 O–Nb2 O5 Yes 1.15 0.64 0.90 0.51
denoted ‘no’ represent at least one of the following three 34 Cs2 O–I2 O5 Yes 1.67 0.95 0.94 0.75
35 Tl2 O–I2 O5 Yes 1.50 0.95 0.88 0.75
conditions: (1) there is no new ternary compound; (2) there
36 Rb2 O–I2 O5 Yes 1.52 0.95 0.89 0.75
are new ternary compounds but there is no oxide of the 37 K2 O–I2 O5 Yes 1.38 0.95 0.84 0.75
chemical formula ABO3 ; and (3) there is at least one ABO3 38 Ag2 O–Sb2 O5 Yes 1.15 0.60 0.92 0.48
compound but it is not a perovskite structure. 39 K2 O–Bi2 O5 Yes 1.38 0.76 0.92 0.60
One hundred ninety-seven binary oxide systems with their (b) Formability of perovskites in AO–BO2 system
perovskite formability, ionic radius of constituent ions A 1 BaO–GeO2 No 1.35 0.53 1.03 0.42
and B, tolerance factor and the octahedral factor (the ra- 2 CaO–SiO2 No 1.00 0.40 0.96 0.32
tio of radii of the small cation B over the radii of anion 3 CoO–TiO2 No 0.65 0.61 0.72 0.48
4 MgO–GeO2 No 0.72 0.53 0.78 0.42
O) are listed in Table 1, part (a) (for A2 O–B2 O5 system), 5 MgO–TiO2 No 0.72 0.61 0.75 0.48
part (b) (for AO–BO2 system) and part (c) (A2 O3 –B2 O3 6 MnO–TiO2 No 0.83 0.61 0.79 0.48
system). All perovskite structures in the tables are con- 7 ZnO–TiO2 No 0.74 0.61 0.76 0.48
firmed by four references [1–3,20]. In addition, the ionic 8 FeO–TiO2 No 0.61 0.61 0.71 0.48
radii used here are the values for six coordination num- 9 NiO–TiO2 No 0.69 0.61 0.74 0.48
10 FeO–SiO2 No 0.61 0.40 0.80 0.32
ber (Shannon’s value), most of them cited from Handbook 11 ZnO–SiO2 No 0.74 0.40 0.85 0.32
of Chemistry and Physics [21], some values which are not 12 MnO–GeO2 No 0.83 0.53 0.83 0.42
available in this handbook are complemented from The el- 13 MgO–SnO2 No 0.72 0.69 0.72 0.55
ements [22] and Perovskites and high TC Superconductor 14 SrO–SiO2 No 1.18 0.40 1.04 0.32
[2]. 15 EuO–SiO2 No 1.17 0.40 1.04 0.32
16 SmO–SiO2 No 1.19 0.40 1.04 0.32
17 BaO–SiO2 No 1.35 0.40 1.11 0.32
18 CoO–SiO2 No 0.65 0.40 0.81 0.32
Table 1 19 SnO–PbO2 No 0.93 0.78 0.76 0.62
20 NiO–SiO2 No 0.69 0.40 0.83 0.32
No. System Formability rA rB Tolerance rB /rO
21 PbO–SiO2 No 1.19 0.40 1.04 0.32
(A) (A)
22 MgO–SiO2 No 0.72 0.40 0.84 0.32
(a) Formability of perovskites in A2 O–B2 O5 system 23 SrO–GeO2 No 1.18 0.53 0.96 0.42
1 Li2 O–Nb2 O5 No 0.76 0.64 0.75 0.51 24 PbO–GeO2 No 1.19 0.53 0.97 0.42
2 Li2 O–Ta2 O5 No 0.76 0.64 0.75 0.51 25 CdO–GeO2 No 1.10 0.53 0.93 0.42
3 Li2 O–V2 O5 No 0.76 0.54 0.79 0.43 26 CaO–GeO2 No 1.00 0.53 0.89 0.42
4 Rb2 O–Nb2 O5 No 1.52 0.64 1.03 0.51 27 BaO–MnO2 No 1.35 0.53 1.03 0.42
5 Li2 O–As2 O5 No 0.76 0.46 0.83 0.37 28 CoO–MnO2 No 0.65 0.53 0.75 0.42
6 Rb2 O–Ta2 O5 No 1.52 0.64 1.03 0.51 29 NiO–MnO2 No 0.69 0.53 0.77 0.42
7 Cs2 O–Nb2 O5 No 1.67 0.64 1.09 0.51 30 SrO–MnO2 No 1.18 0.53 0.96 0.42
8 K2 O–As2 O5 No 1.38 0.46 1.09 0.37 31 CaO–VO2 Yes 1.00 0.58 0.87 0.46
9 K2 O–V2 O5 No 1.38 0.54 1.04 0.43 32 SrO–HfO2 Yes 1.18 0.71 0.88 0.56
10 Na2 O–As2 O5 No 1.02 0.46 0.94 0.37 33 CaO–MoO2 Yes 1.00 0.65 0.84 0.52
11 Na2 O–P2 O5 No 1.02 0.38 0.98 0.30 34 CaO–RuO2 Yes 1.00 0.62 0.85 0.49
12 Cs2 O–V2 O5 No 1.67 0.54 1.15 0.43 35 CaO–SnO2 Yes 1.00 0.69 0.82 0.55
13 Cu2 O–P2 O5 No 0.77 0.38 0.88 0.30
 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48 43

Table 1 (Continued ) Table 1 (Continued )

No. System Formability rA rB Tolerance rB /rO No. System Formability rA rB Tolerance rB /rO
(A) (A) (A) (A)
36 SrO–CeO2 Yes 1.18 0.87 0.81 0.69 33 Gd2 O3 –Fe2 O3 Yes 0.94 0.55 0.86 0.44
37 SrO–RuO2 Yes 1.18 0.62 0.92 0.49 34 Ho2 O3 –Fe2 O3 Yes 0.89 0.55 0.84 0.44
38 SrO–SnO2 Yes 1.18 0.69 0.88 0.55 35 La2 O3 –Mn2 O3 Yes 1.03 0.58 0.88 0.46
39 BaO–CeO2 Yes 1.35 0.87 0.87 0.69 36 La2 O3 –Ti2 O3 Yes 1.03 0.67 0.84 0.53
40 SrO–ZrO2 Yes 1.18 0.72 0.87 0.57 37 Nd2 O3 –Fe2 O3 Yes 0.98 0.55 0.88 0.44
41 SrO–PbO2 Yes 1.18 0.78 0.85 0.62 38 Nd2 O3 –Ni2 O3 Yes 0.98 0.56 0.87 0.44
42 SrO–TiO2 Yes 1.18 0.61 0.92 0.48 39 Nd2 O3 –Ti2 O3 Yes 0.98 0.67 0.82 0.53
43 BaO–NbO2 Yes 1.35 0.68 0.95 0.54 40 Nd2 O3 –V2 O3 Yes 0.98 0.64 0.83 0.51
44 BaO–SnO2 Yes 1.35 0.69 0.95 0.55 41 Pr2 O3 –Fe2 O3 Yes 0.99 0.55 0.88 0.44
45 BaO–ZrO2 Yes 1.35 0.72 0.93 0.57 42 Pr2 O3 –Mn2 O3 Yes 0.99 0.58 0.86 0.46
46 SrO–MoO2 Yes 1.18 0.65 0.90 0.52 43 Pr2 O3 –Ni2 O3 Yes 0.99 0.56 0.87 0.44
47 SrO–VO2 Yes 1.18 0.58 0.94 0.46 44 Sm2 O3 –Fe2 O3 Yes 0.96 0.55 0.87 0.44
48 CaO–UO2 Yes 1.00 0.89 0.74 0.71 45 Sm2 O3 –Ni2 O3 Yes 0.96 0.56 0.86 0.44
49 BaO–ThO2 Yes 1.35 0.94 0.84 0.75 46 Sm2 O3 –Ti2 O3 Yes 0.96 0.67 0.81 0.53
50 BaO–UO2 Yes 1.35 0.89 0.86 0.71 47 Tb2 O3 –Fe2 O3 Yes 0.92 0.55 0.85 0.44
51 CaO–HfO2 Yes 1.00 0.71 0.81 0.56 48 Tb2 O3 –V2 O3 Yes 0.92 0.64 0.81 0.51
52 PbO–CeO2 Yes 1.19 0.87 0.81 0.69 49 Tm2 O3 –Fe2 O3 Yes 0.88 0.55 0.84 0.44
53 BaO–HfO2 Yes 1.35 0.71 0.94 0.56 50 Tm2 O3 –V2 O3 Yes 0.88 0.64 0.80 0.51
54 CaO–TiO2 Yes 1.00 0.61 0.85 0.48 51 Yb2 O3 –Fe2 O3 Yes 0.86 0.55 0.83 0.44
55 CaO–ZrO2 Yes 1.00 0.72 0.81 0.57 52 Yb2 O3 –V2 O3 Yes 0.86 0.64 0.79 0.51
56 BaO–TiO2 Yes 1.35 0.61 0.99 0.48 53 Y2 O3 –Cr2 O3 Yes 0.9 0.62 0.81 0.49
57 PbO–TiO2 Yes 1.19 0.61 0.93 0.48 54 Y2 O3 –Fe2 O3 Yes 0.9 0.55 0.84 0.44
58 PbO–ZrO2 Yes 1.19 0.72 0.87 0.57 55 Lu2 O3 –Al2 O3 Yes 0.86 0.54 0.83 0.43
59 BaO–MoO2 Yes 1.35 0.68 0.95 0.54 56 Y2 O3 –Ti2 O3 Yes 0.9 0.67 0.79 0.53
60 BaO–TbO2 Yes 1.35 0.76 0.91 0.60 57 Gd2 O3 –Ti2 O3 Yes 0.94 0.67 0.81 0.53
61 BaO–PrO2 Yes 1.35 0.85 0.87 0.67 58 Yb2 O3 –Al2 O3 Yes 0.86 0.54 0.83 0.43
62 BaO–PbO2 Yes 1.35 0.78 0.90 0.62 59 La2 O3 –Al2 O3 Yes 1.03 0.54 0.90 0.43
63 CaO–PbO2 Yes 1.00 0.78 0.78 0.62 60 Nd2 O3 –Al2 O3 Yes 0.98 0.54 0.88 0.43
64 CaO–MnO2 Yes 1.00 0.53 0.89 0.42 61 Bi2 O3 –Fe2 O3 Yes 1.03 0.55 0.89 0.44
62 La2 O3 –Co2 O3 Yes 1.03 0.55 0.89 0.44
(c) Formability of perovskites in A2 O3 –B2 O3 system
63 La2 O3 –Cu2 O3 Yes 1.03 0.73 0.81 0.58
1 La2 O3 –B2 O3 No 1.03 0.23 1.09 0.18
64 La2 O3 –Ni2 O3 Yes 1.03 0.56 0.89 0.44
2 Sc2 O3 –B2 O3 No 0.75 0.23 0.95 0.18
65 Dy2 O3 –Fe2 O3 Yes 0.91 0.55 0.85 0.44
3 Sm2 O3 –B2 O3 No 0.96 0.23 1.05 0.18
66 Ce2 O3 –Al2 O3 Yes 1.01 0.54 0.89 0.43
4 Al2 O3 –B2 O3 No 0.54 0.23 0.85 0.18
67 Ce2 O3 –Cr2 O3 Yes 1.01 0.62 0.85 0.49
5 Ga2 O3 –Al2 O3 No 0.62 0.54 0.74 0.43
68 Eu2 O3 –Cr2 O3 Yes 0.95 0.62 0.83 0.49
6 Eu2 O3 –Ln2 O3 No 0.95 0.8 0.76 0.63
69 Gd2 O3 –Cr2 O3 Yes 0.94 0.62 0.83 0.49
7 Dy2 O3 –B2 O3 No 0.91 0.23 1.03 0.18
70 Lu2 O3 –Fe2 O3 Yes 0.86 0.55 0.83 0.44
8 Er2 O3 –B2 O3 No 0.89 0.23 1.02 0.18
71 Nd2 O3 –Cr2 O3 Yes 0.98 0.62 0.84 0.49
9 Eu2 O3 –B2 O3 No 0.95 0.23 1.05 0.18
72 Sm2 O3 –V2 O3 Yes 0.96 0.64 0.83 0.51
10 Gd2 O3 –B2 O3 No 0.94 0.23 1.04 0.18
73 Bi2 O3 –Al2 O3 Yes 1.03 0.54 0.90 0.43
11 Ho2 O3 –B2 O3 No 0.89 0.23 1.02 0.18
74 Dy2 O3 –Cr2 O3 Yes 0.91 0.62 0.82 0.49
12 Y2 O3 –B2 O3 No 0.9 0.23 1.03 0.18
75 Er2 O3 –Cr2 O3 Yes 0.89 0.62 0.81 0.49
13 Yb2 O3 –B2 O3 No 0.86 0.23 1.01 0.18
76 Yb2 O3 –Cr2 O3 Yes 0.86 0.62 0.80 0.49
14 Bi2 O3 –Sm2 O3 No 1.03 0.96 0.73 0.76
77 Ho2 O3 –Cr2 O3 Yes 0.89 0.62 0.81 0.49
15 V2 O3 –Cr2 O3 No 0.64 0.62 0.71 0.49
78 Pr2 O3 –Al2 O3 Yes 0.99 0.54 0.88 0.43
16 V2 O3 –Al2 O3 No 0.64 0.54 0.75 0.43
79 Ho2 O3 –Cr2 O3 Yes 0.89 0.62 0.81 0.49
17 As2 O3 –B2 O3 No 0.58 0.23 0.87 0.18
80 Tm2 O3 –Al2 O3 Yes 0.88 0.54 0.84 0.43
18 Gd2 O3 –Y2 O3 No 0.94 0.9 0.72 0.71
81 La2 O3 –Y2 O3 Yes 1.03 0.9 0.75 0.71
19 Sm2 O3 –Y2 O3 No 0.96 0.9 0.73 0.71
82 Tm2 O3 –Cr2 O3 Yes 0.88 0.62 0.80 0.49
20 Tm2 O3 –B2 O3 No 0.88 0.23 1.02 0.18
83 Lu2 O3 –Cr2 O3 Yes 0.86 0.62 0.80 0.49
21 In2 O3 –Cr2 O3 No 0.8 0.62 0.77 0.49
84 Gd2 O3 –Al2 O3 Yes 0.94 0.54 0.86 0.43
22 In2 O3 –Fe2 O3 No 0.8 0.55 0.80 0.44
85 Sm2 O3 –Al2 O3 Yes 0.96 0.54 0.87 0.43
23 Sc2 O3 –Al2 O3 No 0.75 0.54 0.79 0.43
86 La2 O3 –Ga2 O3 Yes 1.03 0.62 0.86 0.49
24 Sc2 O3 –Cr2 O3 No 0.75 0.62 0.76 0.49
87 Eu2 O3 –Al2 O3 Yes 0.95 0.54 0.87 0.43
25 La2 O3 –Cr2 O3 Yes 1.03 0.62 0.86 0.49
88 Er2 O3 –Al2 O3 Yes 0.89 0.54 0.84 0.43
26 La2 O3 –Fe2 O3 Yes 1.03 0.55 0.89 0.44
89 Y2 O3 –Al2 O3 Yes 0.9 0.54 0.85 0.43
27 La2 O3 –V2 O3 Yes 1.03 0.64 0.85 0.51
90 Dy2 O3 –Al2 O3 Yes 0.91 0.54 0.85 0.43
28 Sm2 O3 –Cr2 O3 Yes 0.96 0.62 0.83 0.49
91 Nd2 O3 –Ga2 O3 Yes 0.98 0.62 0.84 0.49
29 Ce2 O3 –V2 O3 Yes 1.01 0.64 0.84 0.51
92 Gd2 O3 –Ga2 O3 Yes 0.94 0.62 0.83 0.49
30 Er2 O3 –Fe2 O3 Yes 0.89 0.55 0.84 0.44
93 Eu2 O3 –Ga2 O3 Yes 0.95 0.62 0.83 0.49
31 Er2 O3 –V2 O3 Yes 0.89 0.64 0.80 0.51
94 Ho2 O3 –Al2 O3 Yes 0.89 0.54 0.84 0.43
32 Eu2 O3 –Fe2 O3 Yes 0.95 0.55 0.86 0.44
44 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48

Fig. 2. (a) Perovskite formability in A+ B5+ O3 oxide systems. (b) Perovskite formability in A2+ B4+ O3 oxide systems. (c) Perovskite formability in
A3+ B3+ O3 oxide systems.
 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48
Fig. 2. (Continued ).
4. r A –r B structural map for perovskite formability
According to Muller and Roy’s proposal [1], the rA –rB
structural map is a powerful tool in studying the ternary
oxide stability. First of all, we use more data which is cur-
rently available to further test Muller and Roy’s regularities
of perovskite formability [1]. To simplify the problem, these
197 binary oxide systems are divided into three groups,
A2 O–B2 O5 group, AO–BO2 group and A2 O3 –B2 O3 group,
based on the valence of ions A and B, and are investigated
by the structural map respectively as seen in Fig. 2a–c. It
is shown that points representing perovskites and those of
non-perovskites are located in distinctively different zones.
Encouraged by the above successes, three groups are com-
bined into one as shown in Fig. 3. It is astonishing that even
in this mixed group, the separation of compounds is reserved
well. This may imply that the stability of perovskite ABO3
is mainly determined by the radii of cation A and B, and
the valence of ions A and B have less effect on the forma-
bility of perovskites. All systems in Fig. 3 has the common
anion O, effect of anion size on the perovskite stability may
be overlooked. On the other hand, the tolerance factor is a
widely used parameter in perovskite study, which takes the
anion radii into consideration. We further try to build up a
structural map containing the tolerance factor that may lead
45
to simple and more effective new criteria for the perovskite
formability.
5. Tolerance factor and its limit
Goldschmidt’s tolerance factor t has been widely accepted
as a criterion for the formation of the perovskite structure
[1,2,6]. It can be defined as:
r A + rO
t=√
2(rB + rO )
And it is known that almost all perovskites have a t value
ranging from 0.75 to 1.00. However, it seems that t =
0.75–1.00 is a necessary but not a sufficient condition for
the formation of the perovskite structure. Even in the range
of t = 0.8–0.9 which is the most favourable value for per-
ovskites, there exist many systems that do not have any
perovskite structure [17], see Cu2 O–P2 O5 , Li2 O–As2 O5 ,
MnO–GeO2 , Al2 O3 –B2 O3 in Table 1.
Additional parameters may, therefore, be necessary for the
prediction of perovskite formation. It is well known that [4]
the octahedron BO6 is the basic mosaic or unit for perovskite
structure. If one cation and six anions form an octahedron,
the ratio of their ionic radius (rB /rO ) is reported within a
46 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48

Fig. 3. Classification of perovskite compounds in 197 binary oxide systems.

limited range [23,24], this factor is defined as the octahedral one can never expect a stable perovskite even this system
factor. So it is natural to try to construct a structural map by has a very favourable tolerance factor, i.e., Cu2 O–P2 O5 sys-
the tolerance factor and the octahedral factor for perovskite tem (radius of P5+ is 0.38 Å), Li2 O–As2 O5 (radius of As5+
formability. is 0.46 Å) and MgO–SiO2 (radius of Si4+ is 0.40 Å). This
As shown in Fig. 4, all perovskites and non-perovskites may be explained as follows: in perovskites, the octahedron
are located in two different regions, and a clear border be- BO6 is the basic unit, if ion B is too small, this unit may
tween two kinds of compounds is identified. The criterion of become unstable. So does the perovskite. This can also be
perovskite formability is, then, expressed by the following better explained by the ratio of cation and anion size, rc /ra ,
equations: in the octahedron. The reported lowest limit of rc /ra for oc-
rB tahedron [23,24] is 0.414. In our structural map spanned
= 0.425 (4) by the tolerance factor and the octahedral factor (seen in
rO
Fig. 4), the lowest limit of the octahedral factor (rB /rO ) for
rA + rO
√ =1 (5) perovskite formation is 0.425, these two values agree well
2(rB + rO ) with each other.
−4.317(rA + rO ) rB Although the criterion mentioned-above cannot correctly
√ + 3.912 = (6) classify the CaO–MnO2 system, it is noticed in the Fig. 4
2(rB + rO ) r O
that this system locates at the boundary zone. Furthermore,
For all 197 systems, only one system, MnO–GeO2 , some points near the boundary between perovskite and
is wrongly discriminated as a “perovskite”, which is a non-perovskite in Fig. 4, i.e. Eu2 O3 –In2 O3 , MnO–TiO2 ,
“non-perovskite”. MnO–GeO2 , SrO–GeO2 , PbO–GeO2 , CdO–GeO2 , and
CaO–GeO2 system, can undergo perovskite–non-perovskite
structure transformation (all of them [1,25,26] can form
6. Discussion perovskites under higher pressures). Also it is observed that
some antimodides and bismuthades do not obey this regu-
As noticed in Fig. 3 and Table 1, there exists a lowest limit larity, seen in Figs. 2a, 3 and 4. Such as in the Na2 O–Sb2 O5 ,
of radii of cation B, 0.53 Å. If ion B is smaller than 0.53 Å, Na2 O–Bi2 O5 and Ag2 O–Bi2 O5 systems, where favourite
 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48
Fig. 4. Classification of perovskite compounds (octahedral factor vs. tolerance factor).
values of the tolerance factor and the octahedral factor are
observed, they do not form perovskite but ilmenite [1].
It is not clear why our model does not work for systems
containing Sb2 O5 , and Bi2 O5 .
It can also be seen from Figs. 3 and 4 that, whether per-
ovskites can be formed in a binary oxide system is almost
decided only by the size of two cations, and their valence
have few effect on this formability. This result is consistent
with the previous report [27].
Goldschmidt’s tolerance factor t has been widely accepted
as a criterion for the formation of the perovskite structure,
and it is known that almost all perovskite compounds have
t values ranging from 0.75 to 1.00. However, the tolerance
factor t is a necessary but not a sufficient condition for the
formation of the perovskite structure. On the other hand, oc-
tahedron BO6 is the basic unit of perovskites, so the octa-
hedral factor rB /rO that related to the stability of octahedron
BO6 is another governing parameter for the formability of
perovskites.
Although Muller and Roy’s rA –rB structural map has
proven useful for gross structural separation, the regions
bordering different structural types are not well defined
[28]. In order to solve this problem, in 1990s, Giaquinta
and Loye [28] proposed a new structural map for predom-
inantly A2 O3 –B2 O3 group, which relies on the combina-
tion of ionic radii and bond ionicities. With their model,
47
they successfully predicted two compound’s crystal struc-
ture. Comparing to Giaquinta and Loye work, our model
has the following features. (1) Two parameters that are
used in our model (tolerance factor and octahedral factor)
are physically meaningful. (2) Unlike predominately the
A2 O3 –B2 O3 group, our model works well for all three
groups (A2 O3 –B2 O3 , A2 O–B2 O5 and AO–BO2 group). (3)
For perovskite formability, our model gives a simple and ef-
fective prediction criterion, which may be easily adopted in
the design of advanced materials with perovskite structure.
7. Conclusions
In this research, 197 binary oxide systems, including
A2 O–B2 O5 , AO–BO2 and A2 O3 –B2 O3 groups, are col-
lected, of which, only 121 systems can form perovskites.
Using these samples, regularities of formability of per-
ovskites are investigated by two structural maps. One map
is drawn by the ionic radii of cation A and √ B, another is
spanned by the tolerance factor, (rA + rO )/ 2(rB + rO ),
and the octahedral factor, rB /rO . Through this study, the
following conclusions are made:
1. In the rA –rB structural map, the systems with per-
ovskite and those without perovskite are distributed
48 C.H. Li et al. / Journal of Alloys and Compounds 372 (2004) 40–48

in distinctively different regions, there exists a clear [4] Z.L. Wang, Z.C. Kang, Functional and Smart Materials, Plenum
boundary between these two kinds of samples. Press, New York, 1998.
[5] M.W. Lufaso, P.M. Woodward, Acta Cryst. B57 (2001) 726.
2. The tolerance factor is a necessary but not sufficient con- [6] V.M. Goldschmidt, Skrifer Norske Videnskaps-Akad. Oslo, I.
dition for perovskite formability, another necessary pa- Mat.-Nat. Kl. 8 (1926).
rameter is the octahedral factor. Using these two factors [7] E.A. Wood, Acta Cryst. 4 (1951) 353.
the perovskites formability can be reliably predicted. [8] M.L. Keith, R. Roy, Am. Miner. 39 (1954) 1.
3. In the structural map consisting of the tolerance factor, [9] J.M. Moreau, Mater. Res. Bull. 3 (1968) 427.
√ [10] R.D. Shannon, Inorg. Chem. 6 (1967) 1474.
(rA + rO )/ 2(rB + rO ), and the octahedral factor, rB /rO , [11] H. Muller-Buschbaum, C. Teske, Z. Anorg, Allg. Chem. 369 (1969)
the points for perovskites and those for non-perovskites 249.
are located in different zones via a clear boundary defined [12] D. Babel, Structure and Bonding, vol. 3, Springer, Berlin, 1967.
by: [13] D. Babel, Z. Inorg. Allg. Chem. 369 (1969) 117.
[14] A. Vedrine, J.P. Besse, G. Baud, M. Capestan, Rev. Chim. Miner. 7
rB
= 0.425 (4) (1970) 593.
rO [15] H.F. McMurdie, J. de Groot, M. Morris, H.E. Swanson, J. Res. Natl.
Bur. Std. 73A (1969) 621.
rA + rO
√ =1 (5) [16] A.S. Bhalla, R. Guo, R. Roy, The Perovskite Structure—A Review
2(rB + rO ) of its Role in Ceramic Science and Technology, 2002, p. 25.
[17] C.Z. Ye, J. Yang, L.X. Yao, N.Y. Chen, Chin. Sci. Bull. 47 (6) (2002)
−4.317(rA + rO ) rB 458.
√ + 3.912 = (6)
2(rB + rO ) rO [18] M.L. Ernest, et al., Phase Diagram for Ceramists, vols. I–XII,
The American Ceramic Society, 735 Ceramic Palace, Ohio 43081,
These equations form the criterions for perovskite forma- 1958–1994.
bility. Among 197 samples, only one system is mispre- [19] E. Mooser, W.B. Pearson, Acta Cryst. 12 (1959) 1015.
[20] Inorganic Crystal Structure Database (ICSD), CD-ROM, Version
dicted. This simple model may be applied to design new 2002, FIZ Karlsruhe, Fachinformationszentrum Karlsruhe, 76344
substrate or buffer materials in compound semiconductor Eggenstein-Leopoldshafen, Germany.
epitaxy. [21] D.R. Lide, Handbook of Chemistry and Physics, Electronic Resource,
CRC Press, Boca Raton, FL, 1999.
[22] J. Emsley, The Elements, third ed., Clarendon Press, Oxford, 1998.
[23] R.C. Buchanan, Materials Crystal Chemistry, Marcel Dekker, New
References York, 1997.
[24] G.S. Roher, Structure and Bonding in Crystalline Materials, Cam-
[1] O. Muller, R. Roy, The Major Ternary Structural Families, Springer- bridge University Press, New York, 2001.
Verlag, New York, 1974. [25] N.L. Ross, B. Reynard, F. Guyot, Zei. Fuer Krist. 204 (1993) 43.
[2] F.S. Galasso, Perovskites and High Tc Superconductors, Gordon and [26] J. Ko, C.T. Prewitt, Phys. Chem. Miner. 15 (1988) 355.
Breach, New York, 1990. [27] R. Roy, J. Am. Ceram. Soc. 37 (1954) 581.
[3] http://www.chemistry.ohio-state.edu/∼mlufaso/spuds/index.html. [28] D.M. Giaquinta, H.C. Loye, Chem. Mater. 6 (1994) 365.