Solid State Communications,Vol. 19, pp. 969-973, 1976. Pergamon Press.

Printed in Great Britain

CRYSTAL STRUCTURE OF Ba2 Bia+BiS+O6

D.E. Cox*
Physics Department, Brookhaven National Laboratory, Upton, NY 11973, U.S.A.
and
A.W. Sleight
Central Research and Development Department, E.I. du Pont de Nemours and Company, Wilmington, DE 19898,
U.S.A.
(Received 31 March 1976 by M.F. Collins)

The structure of BaBiOa at 25°C has been solved and refined with the use
of neutron diffraction powder data. The valence situation is found to be
• 3+ .5+
BaBh/2Bh/203 rather than BaBi4+O3. The Bi 3+ and Bi s+ cations take on an
ordered arrangement and the resultant structure has monoclinic symmetry
I2/m (a = 6.181 A, b = 6.136 A, c = 8.670 A,/~ = 90.17°). This is the first
example of an ordered perovskite in which the ordered cations are the
same element, and also represents the first case of Bi 3+ octahedrally co-
ordinated to oxygen. The semiconducting properties of BaBiO3 are readily
understood on the basis of the ordered structure. The structure becomes
rhombohedral at about 130°C, and cubic in the region of 450°C.

1. INTRODUCTION ambiguity. Although they conclude in favor of BaBi4+O3,

we feel that their evidence is not entirely persuasive.
SINCE THE DISCOVERY of superconductivity 1 in the
The symmetry of BaBiO3 has been variously de-
BaPbO3-BaBiO3 system, the question concerning the
scribed as rhombohedral, orthorhombic, monoclinic or
oxidation states in BaBi03 has taken on special import-
triclinic; however, all investigators have agreed that there
ance. Although the oxidation states in BaPbO3 are con-
is no evidence in the X-ray diffraction data to suggest
sistently viewed as Ba2*pb4+O~-, both the formulations
more than one formula unit of BaBiO3 per unit cell.
BaBi4+O3 and BaBl~/2Bl~/203
.3+ .s+ have been proposed 2 - 7
There is thus considerable difficulty rationalizing the
for BaBiO3.
fact that BaBiO3 is semiconducting rather than metallic.
Scholder et al. 2 first prepared BaBiO3, and they
If the situation were BaBi4+O3 with only one Bi4÷ in the
supplied convincing analytical data supporting the
unit cell, metallic properties would certainly be expected.
formula BaBiO3 rather than, for example BaBi3+O2.s. • 3+ .5+
A BaBll/2B11/203 situation in which Bi 3+ and Bi s+ are
They also indicated that the structure of BaBiO3 is
disordered is very unlikely, since ordering would be
closely related to that of perovskite. Scholder et al.
•3+ .S+ favored on lattice energy grounds, and only electron
assumed that the oxidation states are BaBll/2 Bh/203,
transfer with small atomic displacements would be re-
apparently because both Bi 3÷ and Bi s+, but not Bi4+, are
quired for Bi 3÷ and Bi s÷ to order.
known to exist in oxides. A similar assumption has also
A powder neutron diffraction structural analysis of
been made by a number of Russian workers, a - 5
BaBiO3 was therefore undertaken in an attempt to
Nakamura et al. 6 attempted to determine the oxi-
unambiguously resolve the valence situation. Although
dation states in BaBiO3 using ESR, magnetic and electri-
neither neutron no X-ray data are sensitive to the differ-
cal measurements. They found semiconductivity, no
ence between Bi3÷ and Bis÷ per se, the oxygen positions
ESR signal, and a very small, temperature-independent
can be determined much more accurately with neutrons
magnetic susceptibility, and concluded that either the
than with X-rays. Thus, BaBi4+O3 and BaBll/2B11/203"3+
-s+
oxidation states are BaBh/2B11/20
.3+ .s+ a or that there is a
should be readily distinguishable on the basis of whether
"filled Bi4÷ band". Infrared spectroscopy was used by
or not all bismuth cations have identical, or nearly iden-
de Hair and Blasse 7 in an attempt to resolve the valence
tical environments with respect to oxygen.
2. STRUCTURE REFINEMENT
* Work performed under the auspices of the U.S. Energy Powder neutron data were collected at the Brook-
Research and Development Administration. haven HFBR at 25°C from a sintered pellet of material

969
970 CRYSTAL STRUCTURE OF Ba2Bia+Bis+o6
weighing about 10 g, prepared as previously described. 1
Pyrolytic graphite in the (002) setting was used for
monochromator and analyzer. The neutron wavelength
was 2.47 A, higher order components being removed
with a pyrolytic graphite filter. The use of an analyzer
gives substantial improvements in overall resolution and
peak-to-background ratios. Data were obtained at 0.05 °
intervals over an angular range 20 of 25-120 °.
Two sets of peaks were observed; one at positions
corresponding roughly to those expected from a pseudo-
cubic cell with ao = 4.35 A, and a second set which
could be indexed in terms of a doubled cell with edge
2a0 and, h, k and l all odd. This is characteristic of the
face-centered cell adopted by most ordered perovskites
A2BB'06. However, the asymmetry of the peaks showed
quite clearly that the true symmetry was no higher than
orthorhombic, the appropriate cell being body-centered
with edges a ~ b ~ aox/2, c "~ 2So. No peaks character-
istic of a primitive cell of this type, nor impurity peaks,
were observed at the level of the counting statistics, i.e.
about 0.1% of the strongest BaBiOa peak.
Since the peaks overlapped too much for any other
possible systematic extinctions to be determined, the
choice of a space-group posed some problems. The initial
strategy adopted was based upon the method developed
by Glazer s for classifying tilting of rigid octahedra in
perovskites. This shows that lbmm is one logical choice,
with equal tilts about the pseudo-cubic [100] and [010]
axes (a-a-c ° in Glazer's notation). It is to be noted that
this choice of space-group requires the four Bi ions to
occupy equivalent positions.
Glazer's classification can readily be extended to
3rdered perovskites. In this case the parent lattice is
face-centered cubic, space-group Fm3m, with a cell edge
3f 2ao. The octahedra about the B and B' ions are regular
but no longer equivalent, reflecting the different sizes of
the two cations. For the same combination of tilts as
above, once again a body-centered lattice is derived, but
adth monoclinic symmetry, I2/m with a ~ b = aox/~,
c ~ 2So,/3 ~ 90 °. The Bi ions now occupy two sets of
inequivalent positions.
A least-squares refinement of the intensity data was
carried out for each of the above models with the Profile
program written by Rietveld a and modified by Hewat. 1°
Individual isotropic temperature factors were assumed,
and the variable positional oxygen parameters were
given small displacements from their ideal positions in
appropriate directions. The variable Ba parameters were
given arbitrary displacements. In each case refinement
converged smoothly to the same final values irrespective
of the initial choice of Ba parameters. (Table 1, Model I).
Intensity agreement for either model is very satis-
factory, as indicated by the R factors R 1, R2 and R 3.
The first of these corresponds approximately to the
Vol. 19, No. 10
conventional R factor for integrated intensities
100 X [lobs -- Ieale[/Z lobs, while R 2 and R a represent,
respectively, unweighted and weighted values based
upon individual data points as defined by Rietveld. 9
However, although the R factors for the monoclinic
model are lower than those for the orthorhombic, one
cannot necessarily conclude the former is correct. The
crystallographic significance tests usually applied in this
type of situation require the number of independent
integrated intensity measurements to be known. 12 Any
estimate of this quantity in a profile refinement would
be very difficult, and would depend upon such factors
as the resolution, the degree of overlap, and the sym-
metry. However, examination of the profile fit in detail
reveals that most of the discrepancy centers around the
peaks with half-integer indices (indexed in terms of the
pseudo-cubic cell) such as (½ ½ ½). This is demonstrated
in a more quantitative way in Table 2 (Model I), which
lists for both monoclinic and orthorhombic structures
weighted R factors calculated from the individual data
points comprising each of the pseudo-cubic reflections.
Furthermore, this discrepancy remains if anisotropic
oxygen temperature factors or an occupation factor
(corresponding to some degree of disorder between Ba
and Bi(1) ions) are introduced (Models II and III, res-
pectively). The results of the various refinements there-
fore provide strong support for the monoclinic structure
Final confirmation of monoclinic symmetry was
obtained from a heavily over-exposed Guinier X-ray
pattern taken with Cr-K~I radiation, which showed
quite unambiguously three weak but resolved lines at
the pseudo-cubic (111) position. A least-squares refine-
ment of the X-ray data based upon body-centered
indices yielded the lattice parameters a = 6.181(1) A,
b = 6.136(1)A, c = 8.670(1)A,/3-- 90.17(1)A, with
standard errors being given in parentheses. If the neutron
cell edges in Table 1 are scaled by a factor of 0.997
(which reflects a small error in the neutron wavelength),
virtually identical values are obtained, a striking illus-
tration of the high precision attainable by the profile
technique.
The inclusion of anisotropic oxygen temperature
factors yielded an appreciable improvement in the R
factors (1.84, 5.12, and 7.35, respectively, for I2/m
symmetry). Since the anisotropic model involves an
additional eight parameters, it was felt that this improve-
ment was of questionable significance in view of the
amount of data collected, and this model was not con-
sidered further since the changes produced in the
positional parameters were quite small. However, in the
process the temperature factor of Bi(2) attained the
plausible value of 0.3 A 2, an indication that the origin of
the negative value in Table 1 may well lie in the neglect
of anisotropic terms. The introduction of an occupation
Vol. 19, No. 10 CRYSTAL STRUCTURE OF Ba2Bia+Bis+o6 971

Table 1. Results of least-squares refinements fqr BaBiO3 for various models as discussed in text. Scattering ampli-
tudes taken as 0.512, 0.852, and 0.580 x l O-t2 cm for Ba, Bi and 0 respectively.ll Standard errors are given in
parentheses, and refer to the least significant digit(s). U, V, W and Rl, R2, R3 are the half-width parameters and R
factors as defined by Rietveld, ~- f is the fraktion of Ba ions on Bi(1) sites, and the other symbols have their usual
crystallographic significance

Monoclinic I 2[m Orthorhombic Ibmm

Ba 4(i) x 0 z Ba 4(i) x 0 1/4
Bi(1) 2(a) 0 0 0 Bi 4(a) 0 0 0
Bi(2) 2(c) 0 0 1/2 0(1) 4(e) x 0 1/4
O(1) 4(0 x 0 z 0(2) 8(g) 1/4 1/4 z
0(2) 8(j) x y z

Model I Model II Model III

Ba : x 0.5045(12) 0.5047(11) 0.5133(11)
Ba :z 0.2468(19) 0.2471(18) 0.25
O(1):x 0.0604(9) 0.0607(8) 0.0576(10)
0(1 ) : z 0.2608(17) 0.2604(16) 0.25
0(2) :x 0.2613(9) 0.2605(9) 0.25
0(2) :y 0.2552(21) 0.2544(20) 0.25
O(2): z --0.0334(4) --0.0333(4) --0.0358(4)
Ba: B 0.6(1)A 2 1.1(1) A 2 0.5(1) A 2
Bi(1): B 0.8(2) A: 0.6(2) A z J 0.3(1) g 2
8i(2): B -- 0.4(2) A: --0.4(2) g z [ 0.3(1) A:
0(1): B 1.5(3) A 2 1.6(2) A: 1.9(2) g 2
0(2): B 2.0(2) A 2 2.1(2) )k2 1.8(1) M
f 0 0.10(1) 0
a 6.2000(3) )l, 6.2001 (3) A 6.1999(4)
b 6.1561(3) A 6.1561(3)A 6.1557(3)
c 8.6948(4) A 8.6948(4) A 8.6951 (5)
,6 90.15(1) ° 90.15(1) ° 90 °
U 0.84(2) deg2 0.84(2) deg2 0.92(2) deg2
V -0.87(d) deg2 - 0.87(3) deg2 -- 0.96(4) deg2
W 0.33(1) deg2 0.33(1) deg2 0.35(1) deg2
Rl 2.68 2.66 3.40
R2 6.15 6.01 7.30
R3 8.54 8.29 10.85

factor corresponding to disorder between Ba and Bi(1)
ions led to a value of 10% Ba on Bi(1) sites and a small
improvement in R2 and R3. Once again the significance
is questionable, but from size considerations this is a
physically reasonable result, and in fact compounds of
the type Ba2BaWO6 are well known.
Several ref'mements of the neutron data based upon
other body-centered space groups were also tried.
Among these were Immm and some of the sub-groups
of lbmm, namely 12mm, Iam2,1212121, and 12/c. These
all gave poorer agreement than in the case ofl2/m,
particularly with respect to the (½½½) peak, and were
accordingly rejected.
Two phase transitions were found in BaBiO3 by use
of differential scanning calorimetry and high-tempera-
ture X-ray diffraction with a Guinier camera. At about
130°C, BaBiO3 becomes rhombohedral, and this tran-
sition appears to be first order. Near 450°C, BaBiO3 be-
comes cubic via an apparently second-order transition.
It is difficult to determine accurate transition tempera-
tures since there is hysteresis associated with the lower
transition, and the upper transition is very gradual. A
peak in the dielectric constant for BaBiO3 has been
observed4'5 near 400°C. However, the electrical resis-
tivity behavior x does not show any evidence of either
transition. Thus, it appears that BaBiO3 remains as
Ba2Bi3+Bis÷o6 through both transitions and, in fact, all
the way up to at least 700°C. However, a transition to
metallic BaBi4÷O3 at high pressure is probable.
972 CRYSTAL STRUCTURE OF Ba2Bia+Bis*o 6 Vol. 19, No. 10

Table 2. Weighted R factors Rw(hkl) o f the individual data points for several pseudo-cubic reflections from BaBiOa.
Rw(hkl ) is defined as {~ W(yobs -- y~e)2/Z W(yobs)2}v2, where Yobs and Yeale are observed and calculated neutron
counts and w is the weight assigned to each point (i.e. 1/yob~)
Model I Model II Model III
hM
I2/m Ibmm I2/m Ibmm I2/m Ibmm
½½½ 0.22 0.50 0.19 0.43 0,17 0.50
10 0 0.13 0.13 0.11 0.12 0.13 0.13
1 10 0.06 0.07 0.08 0.09 0.06 0.07
½½ 0.07 0.10 0.08 0.11 0.07 0.11
111 0.11 0.09 0.07 0.10 0.10 0.08
200 0.I0 0.11 0.08 0.10 0.10 0.11
~½ 0.12 0.24 0.13 0.19 0.11 0.23
2 10 0.08 0.08 0.08 0.09 0.08 0.08
2 11 0.08 0.10 0.09 0.11 0.07 0.I0
½½,3 ] ] 0.11 0.24 0.08 0.17 0.11 0.24
220 0.06 0.07 0.05 0.05 0.06 0.07
Model I; Isotropic temperature factors.
Model II; Anisotropic oxygen temperature factors.
Model III; Isotropic temperature factors and occupation factor.

3. DISCUSSION distances, (2) to favor a low coordination number, and

With respect to the ideal perovskite structure, the
oxygen parameters correspond to both a tilting of the
octahedra and movement of the oxygen ions away from
Bi(1) towards Bi(2). Thus, the displacements from 0.0
for x of O(1) and for z of 0(2) are indicative of octa-
hedra tilting of roughly 10 ° about the pseudo-cubic
[110] axis. However, the displacements from 0.25 for z
of O(1) and x a n d y of 0(2) represent a movement of
the oxygen ions of about 0.1 A away from Bi(1) towards
Bi(2). Therefore, although Bi(1) and Bi(2) are each
coordinated to six oxygens in a slightly distorted octa-
hedral arrangement, the Bi-O distances are different for
the two bismuth cations, the average being 2.28 A and
2.12 A for Bi(1) and Bi(2), respectively. The latter
agrees very well with that expected for Bi s÷, but a com-
parison of the former with that expected for Bi 3÷ is
difficult because Bi 3. has not previously been found
octahedrally coordinated to oxygen. The average of the
six shortest Bi3÷-O distances is generally in the range
2.35 to 2.4 A. However, these values are taken from
oxides where the coordination polyhedron is very
irregular and the true coordination number is greater
than six; each of these factors is expected to increase the
average of the six shortest Bi3÷-O distances. Covalency
effects may also be important in considering the bonding
of Bi 3÷. The presence of the very electropositive barium
will promote a high degree of covalency for Bi-O bonds.
This covalency will tend (1) to decrease Bia*-O bond
(3) to promote a regular environment. Thus, the bonding
of Bi(1) can be rationalized on the basis that it is Bi3..
Ba2Bia÷BiS+O6 resents not only the first example
of Bi 3÷ octahedrally coordinated to oxygen, but also the
first example of an ordered perovskite structure in which
the ordered cations are the same element. Frequently,
the B cations iri a A~+Ba+Bs+O6 perovskite are not com-
pletely ordered. However, the lack of order is due to
kinetic rather than thermodynamic factors. In
Ba2 Bi3÷BiS÷O6 only electron ordering is required to
attain an ordered structure. However, this electron
ordering must be accompanied by lattice distortions,
which stabilize the disproportion of Bi4÷ into Bi 3÷ and
Bis÷.
Neutron diffraction data have also been collected at
25°C for BaPbO3 and BaPb0.7Bio.303. These have res-
pectively orthorhombic and tetragonal type perovskite
structures in which there is only one set of B cation sites.
The formulation BaPbo.TBlo.303
4+ .4÷ appears appropriate,
since metallic (and superconducting) properties are
observed. 1 Structural studies in the BaPbOa-BaBiO3
system are continuing in an effort to understand the
unusual properties of this system.
Acknowledgements - We are greatly indebted to A.W.
Hewat for providing us with a tape of the modified
PROFILE program. Helpful discussions with J.D. Axe
are also acknowledged.

REFERENCES
1. SLEIGHT A.W., GILLSON J.L. & BIERSTEDT P.E., Solid State Commun. 17, 27 (1975).
Vol. 19, No. 10 CRYSTAL STRUCTURE OF Ba2Bi3÷BiS÷O6 973
2. SCHOLDER R., GANTER K., GLASER H. & MERZ G., Z. Anorg. Allgem. Chem. 319, 375 (1963).
3. SHUVAEVA E.T. & FESENKO E.G., Sov. Phys.-Crystallogr. 14, 926 (1970).
4. FESENKO E.G., SHUVAEVA E.T. & GOUTSOV YU.I., Sov. Phys.-Crystallogr. 17,362 (1972).
5. VENEVTSEV YU.N.,Mat. Res. Bull 6, 1085 (1971).
6. NAKAMURAT.,SABUROK.&SATAT.,J. Phys. Soc. Japan31,1284(1971).
7. DE HAIR J.TH.W. & BLASSE G., Solid State Commun. 12, 727 (1973).
8. GLAZER A.M., Acta Co'st. B28, 3384 (1972); ibid. A31,756 (1975).
9. RIETVELD H.M., J. Appl. Cryst. 2, 65 (1969); Reactor Centrum Nederland Research Report RCN-104
(1969).
I0. HEWATA.W., U.K. Atomic Energy Authority Research Group Report RRL 73/897 (1973) (unpublished).
For an example of the use of the program in another distorted perovskite study see HEWAT A.W., J. Phys. C
C6, 2559 (1973).
11. SHULL C.G., M.I.T. compilation (1972) (unpublished).
12. HAMILTON W.C., Acta CO,st. 18, 502 (1965).