Mat. Res. Bull. Vol. 6, pp. 209-218, 1971.

in the United States.

Pergamon Press, Inc.

Printed

PREPARATION AND PROPERTIES OF

(Ba,Sr)TiO 3 SINGLE CRYSTALS
K. Bethe and F. Welz
Philips Forschungslaboratorium Hamburg GmbH, Germany

(Received February 12, 1971; Communicated by G. H. Jonker)

ABSTRACT
High quality solid solution single crystals of
(BavSrl_x)TiO~ (BST) with any value of x can be pulled
from a melt cSntaining about 36 mol-% BaO/SrO and
64 mol-% TiO 2. The low frequency dielectric behaviour
resembles that of the respective ceramics. The character of phase transition changes from Ist to 2nd order
when the SrTi05-content exceeds around 50 mol-%. Accordingly the dielectric nonlinearity constant of the
free crystal reverses its sign at that composition.
Growth of Single-Crystalline BST
It is well known that pure BaTiO 3 (BT) single crystals can
be grown from a nearly eutectic melt in the system BaO/Ti02 (I).
The suitability of this method for the growth of solid
solution single crystals of (Ba,Sr)TiO 3 (BST) was checked by
the authors using DTA and direct optical observation of the
melt. These preliminary experiments revealed that the region
of the phase diagram which was used for crystallization was
only gradually affected by a replacement of Ba by Sr. Particularly the peritectic temperature (1322 C) was changed by less
than 30 C for BT-contents between 100 and 35 mol-%. That means
the desired mixed crystals of (Ba,Sr)TiO 3 could be grown nearly
the same way as pure BaTiO 3 from this chemically clean system
without foreign ions.
The apparatus: The crystals were grown by a modified
Czochralski technique using a hydraulic precision pulling

209

210

(Ba, Sr)TiO3 SINGLE CRYSTALS

Vol. 6, No. 4

machine provided with an illumination/observation system*.

The resistively heated furnace was temperature controlled to
a stability better than 0.1 C. The 200 ml platinum crucible
(Platin S, Heraeus) containing the melt was surrounded by a
thick alumina shield for temperature homogenization. A seedless platinum finger was used as crystallization site. The
temperature of this double-walled platinum finger could be
lowered by a controlled and programmable stream of pressurized
air in its interior. The difference in temperature between the
tip of this air-cooled pulling finger and the measuring thermocouple immersed in the melt was determined by an optical pyrometer. It amounted to 20 C at a maximum air-stream of 6000 1/h.
The vertical position of the crucible was so adjusted that there
was a nearly linear increase of temperature from bottom to top
of 4 C. This difference was necessary to prevent the growth of
dendrite-like crystals on the surface.
The crystal growth process: Figure I shows the temperature
progr~mm at the crystallization site, which remained nearly unchanged for all compositions of the (Ba,Sr)Ti03 crystals. At
the soak temperature of 1600 C all mixtures formed a melt of
rather low viscosity which was well stirred by a platinum sheet
mounted instead of t h e pulling finger and rotated at 60 rpm.
Soak

Nucleation
Annealing,Reox~dahon

1600

.c

~n 0 2

/ ,oF,oo ;
1

/
~oc /

I
~ t i m e
W i t h d r a w a l of c r y s t a l
f r o m melt

lO h
Pt - f i n g e r :
c o o l i n g air

FIG. I
Program of a crystal growth run for (Ba,Sr)TiO 3

*500 W high-pressure mercury lamp/stereoscopic microscope with

green-bluish filter.

Vol. 6, No. 4

(Ba, Sr)TiO 3 SINGLE CRYSTALS

211

At a constant furnace temperature of 1500 C the seedless

platinum finger tip which touched the melt surface was cooled
down at a rate of I /h by a programmed increase of the airstream through the platinum finger. Immediately after the first
appearence of crystallites the cooling air-stream was held
constant at that value to keep the number of nuclei as low as
possible. The pulling speed during the crystallization period
was as low as 0.1 mm/h. Rotation rates were 12 rpm. The comparatively slow cooling to room temperature was found necessary
to prevent inhomogeneous strain in the che~ry-sized crystals.
These crystals (approx.
BaTiO:
20 grms) consisted of 2 to 6
oo,
. . . . . . . . .
single crystals, from which
single crystalline cubes of
4 to 8 mm could be cut. They
showed slightly yellow color,
which disappeared after annealing of the single crystals at 600 C in air.
The Ba/Sr-ratio of the
obtained crystals was markedly different from that of
the melt, the crystal always
containing less Ba than the

60

c.)~

10

10 20 30 40 50 60 70 80 90 100"/.
Melt with 28 mob-*/. TiO2-excess

BaTiO3

~ t Sr Ti 03

melt.

FIG. 2
Substitution ratio Ba/Sr
in melt and crystal
Homogeneity

of the Prepared Solid Solution Single Crystals

The most stringent proof of homogeneity can be deduced

from the authors'
microwave measurements on single crystalline
BST-spheres (0.5 to 6 mm in diameter) (2): The minimum line
width of these temperature-tuned spherical dielectric resonators
amounted 0.3 to 0.5 C (BST 27 to BST 47). This bandwidth gives
an upper limit for the local distribution of Curie temperatures
within the spheres. With the difference in Curie temperatures
of pure BT (+ 104 C) and pure ST (- 247 C) the maximum equi-

212

(Ba, Sr)TiO 3 SINGLE CRYSTALS

Vol. 6, No. 4

valent spread in Ba/Sr-ratio becomes

0.~o

However,

..0.5
351

= 0.9...I .4 10 -3

the actual spread should be at least a factor of 5

smaller, because the measured linewidth is mostly due to dielectric losses in the perfect

(homogeneous)

crystal (compare

theory of dielectric microwave losses and bandwidth of pure ST

in (2)).
Taking into account the geometrical resolution of this
quasioptic microwave measurement we can conclude,

that our

crystals contain no regions larger than about 0.3 mm with a

spread in the Ba/Sr-ratio larger than 3 " 10 -4
Further proofs of the macroscopic homogeneity of the BSTcrystals are:
I) Same measured Curie temperature

for all samples taken from

different parts of a crystal (electrodeless microwave measurements, accuracy better than I C).
2) Abrupt disappearence

of the optical birefringence at the

Curie point all over a 5 mm platelet

These microscopic

observations

(accuracy better than 2 C).

of solid solution crystals

crossing their Curie point revealed no striations.

The occurence

of such striations is intimately connected with thermal fluctuations within the melt which could not be detected during
our growth process

(aT < 0 .05 C , measured 3 mm from the

growing crystal).
The microscopic homogeneity was checked by X-ray diffraction: The 200-reflex near 2@ = 46 showed no extra broadening
(b2 in Table I). Thus the upper limit for inhomogeneities
smaller than about 5 ~m is set to 1.5% spread in Ba/Sr-ratio.
Dielectric Measurements
From the obtained single crystals,
strain,

flaws or inclusions

regions free of visible

(> 5 ~m) were selected.

After an

X-ray orientation, circular cylinders were cut out of these by

ultrasonic drilling. Faces and wall of the cylinders were ground
and polished by hand using different diamond pastes down to 1 ~m.
Finally these cylinders measured 3 mm in diameter, 2.5 mm in
height, the axis being parallel to the (lO0)-direction within

Vol.6, No.4

(Ba,Sr)TiO3SINGLECRYSTALS

213

0.5 . After an etching procedure in HF/HNO 3 both faces were

provided with 1 ~m vacuum-deposited gold electrodes on a 300
chromium layer. Painted-on silver electrodes were found to be
quite insufficient,

especially in the case of crystals with

high Sr-contents.
Most low-frequency dielectric measurements were made with
a Wayne Kerr precision bridge between 30 Hz and 20 kHz. Probing
field strength always was less than 10 V/cm. The parallel
biasing field usually was a pulsed field with a repetition
frequency around 10 Hz as indicated in Figure 3. The voltage
integral over one period was adjusted to zero the prevent
space charge near the electrodes within the crystal. During
these measurements the samples were kept in dry pressurized
helium.

In the middle of Fig. 3, as an example of (Ba,Sr)Ti0 3

solid solution single crystals, a crystal containing 47 mol-%

BT is s h o ~ .
clear separation of

One notes a high and sharp maximum of E r and a

ST

the ferro~o

phase tran-

the second

BT

2o

electric
sition and

BST-47

\~

!i!~x.....~ i

~0 ~

(orthorhombic )
phase transition. All
mixed crys-

//~,/

100

tals of
(Ba,Sr)TiO 3
grown show
very distinct

200

300

K
400
~T

FIG. 3
Dielectric constant vs temperature of SrTiO 3,
Bao.A~Sr 0 =~Ti0~ and BaTiO~ (Parameter: biasing
~
" ~ voltage in kVlcm).

ferroelectric phase transitions

(see also Fig. 4 and Table I );

this statement is at variance with some former publications,

but in full accord with results on ceramic samples.
Dielectric Nonlinearity
For a stress-free ferroelectric crystal the dielectric

214

(Ba, Sr)TiO 3 SINGLE CRYSTALS

Vol. 6, No. 4

response on a biasing electric field can be deduced from

Devonshire's thermodynamic theory:
E

4BP 3 + ...

(1)

+ 12BP 2 + ...

(2)

= 2AP +

1/c = 2A

Taking into account only a first order nonlinearity constant

N(w 12 Boo3), we get for the field dependence above the
phase transition
I

~~ 1

. p2

'~(p) C(o) +Co~

1

C(o) C(s)
C(o )3 C(Z )3
-

"

= 7

tO kVlcm
20kVIcm
m 30 kV/cm
o 52kVIcm

?
10
Z

40

BST- 30

C~

o
0

I|i

I
I

30

KI

I
I

20

"( 2 C(E) +3 c() )

(3)

where (o) is the dielectric

constant at zero biasing field
and C(E ) the dielectric
constant at the field strength
E. Equation (3) fits the
experimental values of all
BST-crystals with an accuracy
of 5% up to 60 kV/cm biasing
field for temperatures at
least several degrees above
the field-dependent (I) phase
transition. The striking

tO

\ , , ~ a . 8c

-,~o

-1()0
~T

oC

-50

FIG. 4
Complex permittivity and nonlinearity constant N of a single crystal (Ban.3Srn 7) Ti03
at 1.6 kHz.
( ~ = Y~ V/cm).

deviation from Equation (3)

observed near Tph (Fig. 4) is
caused by a rounding of the
s-peak under biasing field
(see Fig. 3). The falling off
at low s-values may result
from eL, the non-ferroelectric
portion of c. Thus we conclude
that the measured nonlinearity
constant N corresponds to the

Vol. 6, No. 4

(Ba, Sr)TiO 3 SINGLE CRYSTALS

quadratic coefficient in
Devonshire's theory and
that higher order terms
of the power series
expansion can be neglected.

215

5O
~
~-~ 4 0 [ - \

= Ceromics
a Single cryslols

The character of the

observed response of
mechanically free BST-crystals
on a polarizing field is
qulte different in crystals
of different composition:
Above Tph in pure BT (and

'1

- ~

BST-87) the permittivity

increases upon application
of an electric field (negative

- SrTiO3'KTa03

1.Order 2.0rcler

N) (see also (3)), whereas

mixed crystals with only 47%
BT and less always show a
marked decrease
of ~r
with increasing biasing field
(positive N). According to the
theory a negative nonlinearity
constant N is connected

FIG. 5
Difference of ferroelectric
phase transition temperature
(Tnh) and extrapolated Curie
~temperature (Tc) as a
function of substitution
ratio for (Ba,Sr)TiO~ and

K(Ta,Nb)O 3.

with a first order ferroelectric phase transition,

positive N indicates a second order transition.

whereas a

Order of Phase Transition

Pure BT shows a typical first order phase transition
characterized by:
a) thermal hysteresis,
b) discontinous drop of Er just below Tph,
c) finite slope of Sr-VS.-T-curve just above Tph,
d) negative nonlinearity constant N.
However, the phase transition of BST-47, BST-30 and
pure ST are of the second order type:
a) no thermal hysteresis,
b) finite slope of the ~r-VS.-T-curve just below Tph,

216

(Ba, Sr)TiO 3 SINGLE CRYSTALS

Vol. 6, No. 4

c) very steep curve just above Tph,

d) positive N.
(The characteristics b and c can be observed in thoroughly
prepared BST-ceramics too.)
The gradual change of the character of the Sr-VS.-Tcurves in the neighbourhood of the phase transition can be
seen from Fig. 5: The difference of the ferroeleetric phase
transition temperature Tph and the extrapolated Curie-temperature T c decreases with increasing ST-contents and
zero at about 50% ST.

Thus we conclude:

Within the

(Ba,Sr)TiO 3 the ferroelectric phase transition

becomes
system

changes from

first to second order, if the ST-contents exceeds about 50%.

Such a change of phase transition order has been reported before
by Triebwasser for the system K(Ta,Nb)O 3 (5). It may also be
predicted from Devonshire's theory (6).
Values taken from
Triebwasser, together with some of the authors' measurements,
are shown in Figure 5, too.

Table of Measured Data

Some physical data of the investigated solid solution
crystals of (Ba,Sr)Ti03 are compiled in Table I: The molar BTcontents (x) was calculated from the lattice cell dimensions
(a,c) assuming a linear change. These values were confirmed by
the extrapolated Curie temperatues (Tc). Dc-resistivity was
measured with a field strength of 100 V/cm at 24 C (120 C).
The very high values of resistivity indicate the absence of
foreign ions. The occurrence of thermal hysteresis at the
phase transition (Tph) is indicated by (H). The values of the
Curie constant C and the extrapolated Curie-temperature T c are
calculated from microwave measurements thus avoiding electrode
effects. This Curie constant shows a monotonous decrease with
increasing Sr-content just as ceramics do. The maximum of Sr
at Tph and the width bo. 9 of this peak (at ~r = ~r " 0.9) are
hard to determine exactly because both are highly influenced by
internal strain (electrodes, quick temperature cycling, residual space charge). Therefore these values c a n n o t be taken
as a proof of inhomogeneity. In the last column, the dielectric
nonlinearity constant N reverses its sign in this system, be-

Vol. 6, No. 4

(Ba, Sr)TiO 3 SINGLE CRYSTALS

217

TABLE 1
List of Data of (Ba,Sr)TiO 3 Single Crystals
xl%

a,cl~ b2el!Q/ncm TclC Tph/C C.10-5/C r-10-3

.10 -12

bo.9/C N-101~(I~) 2

(BaxSr1_x)TiO 3
BT
BST 87

=98

3.992
4.036

--

4.8

+104

+119(H)

87

3.981
4.004

<0.1

3.9

+60

+67(H)

1.26

I0
(0.1)

-80

-80

1.6

8.6

7-0.32

~9.5

<5

=-0.4

0.885

~I0

<4

4.2

BST 47

47.5

3.935

0.05

BST 34

34.2

3.928

0.06

40

-126

-127

0.813

~12

74

=6

BST 30

30

3.925

0.05

-141

-141

0.81

~11

<4

7.5

BST 27

27.6

3.922

0.05

20

-151

-150

0.795

~16

--

=9

0.05

I0
(0.01)

-247

0.80

25.5

--

12.9

ST

3.905

ginning with small negative values at pure BT, crossing the

Curie-Critical-Point somewhere near x = 50% and then going to
increasing positive values for high ST-contents.
Acknowledgement
The authors express their thanks to Dr. W. Tolksdorf and
his coworkers for many fruitful discussions and their advice
and help concerning the chemical problems. We are furthermore
indepted to Dr. P. Holst for carrying out the X-ray examinations.
References
I.

A. von Hippel and co-workers, Technical Report 178, Lab.

Insulation Res., MIT (AD 404 776) (1963).

2.

K. Bethe, Philips Res. Repts. Suppl., No. 2 (1970).

3.

S. Triebwasser,

4.

G. Rupprecht et al., First Quarterly Progress Rept., Rept.

No. I, Contract No DA 36-039-SC-87369 (1960).

5.

S. Triebwasser,

6.

W. Heywang and R. Schofer,

Phys. Rev. 118, 100 (1960).

Phys. Rev. 114, 63 (1959).

Z. angew. Phys. 20, 10 (1965).