R 180 Philips Res. Rep.

7, 21-27, 1952 ,

THE SPECIFIC HEAT OF BARIUM TITANATE

BETWEEN 100 oK AND 410 -x
by J. VOLGER 536.631: 548.0.536

Summary
The specific heat of polycrystalline BaTi03 has been measured in
a Nernst calorimeter. The lattice contribution cannot be deseribed
accurately with a single Debye function. In three transition regions
the anomalous peaks in the specific heat have been determined.
Rsum
On a mesur la chaleur spcifique de BaTi03 polycristalline dans un
calorimtre de Nernst. La contribution du rseau ne s'exprime pas
exactement par une function de Debye. On a dtermin les anomalies
de la chaleur spcifique dans les trois rgions de transition.
Zusammenfassung
Die spezifische Wrme von polykristallinischem BaTi03 wurde mit
einem Kalorimeter nach Nernst gemessen. Es zeigt sich, daB der
Beitrag des Gitters nicht mit einer einzigen Debye-Funktien he-
schrieben werden kann. Die Anomalien der spezifischen Wrme in
den drei bergnngsgebietcn wurden bestimmt.

1. Introduetion
From X-ray and dielectric measurements it is well known that barium
titanate exhibits polymorphism 1), 2) with transition temperatures of
about 120C, 5 C and -70C, where maxima of the permittivity occur.
Coming from high temperatures the crystal shows cubic, tetragonal, ortho-
rhombic and rhombohedral symmetry. Anomalies of the specific heat have
been observed in the transition regions at 5 C and 120C 3), 4), 5), 6).
The purpose of this paper is to offer some new caloric measurements
on polycrystalline BaTi03' The specific heat va temperature curve of the
compound between 100 oK and 410 oK will be given, with special attention
to the behaviour in the three transition regions.

2. Methods
A vacuum calorimeter of the classical Nernst type was used. The appa-
ratus is shown in fig. 1. Heat exchange between the sample and its sur-
roundings could be reduced by adjusting the temperature of a thick copp~r
shielding. The temperatures of sample and shielding were determined
from voltage measurements of silver-constantan thermo-elements on a
Diesselhorst compensator. Heat was supplied by passing electric current
through heating coils. For low-temperature measurements the whole set
was immersed in a bath of liquid nitrogen, thereby providing for an initial
cooling of shielding and sample.
22 J. VOLGER

Two methods have been followed.

(A) In the first place absolute measurements have heen carried out.
During an exactly known period (e.g. 2 minutes) heat was supplied to the
sample, the total amount of this heat being calculated from the time, current
and voltage, whilst the resultant increase of temperature (about 1 or 2
degrees) could be determined from the increase of thermo-e.m.f., taking
into account the small drift of temperature before and after the heating
period. From these data Cp can be calculated.
(B) Secondly, relative measurements have been taken, in particular
in the transition ranges. During a long period (e.g. an hour) the reciprocal
drift of temperature (expressed in seconds per 10 [LVincrease of thermo-
e.m.f.) was determined while supplying a small and practically constant
amount of heat, which in any case varied only very slowly. In other ex-
periments a slow cooling was provided for by making the shielding much

vacuum pump

68537

Fig. 1. The Nernst calorimeter. The sample is supported with a nylon thread. The heating
coil is made of 01 mm thick nickel-chromium wire. At A and B wires for current and
voltage measurements are connected. The temperature of the sample is read with the help
of a thermo-element (C). The shielding (S) has also a heating coil (D); its temperature
is read from the thermo-e.m.f. at E. The whole system is mounted in an iron jar (P) which
is made vacuum-tight with a lead ring (l). The vacuum is provided by a mercury
diffusion pump.
 SPECIFIC HEAT OF BARIUM TITANATE 23

colder than the sample. The reciprocal drift is proportional to Cp of the sub-
stance, the proportionality factor being"~ smooth function of time (tempera-
ture) during the experiment. This factor can be determined if Cp is known.
By extending such a heating or cooling experiment sufficiently far beyond
and in advance of the transition region, to temperatures where absolute
measurements of Cp had been carried out, and then interpolating the
calculated factors, Cp could be calculated also in the transition regions.
With the second method more reliable values of the anomalies of Cp can
he obtained than with the first one.

3. Sample
Barium titanate in the form of a sintered polycrystalline product was
prepared by Dr G. H. Jonker of this laboratory. Barium carbonate and
titanium dioxide were mixed, milled and prefired at 1000 oe. Afterwards
the titanate obtained was milled again, mixed with a suitable organic
hinder and shapedinto a cylindricalrod which was definitely fired at 1360 oe.
The rod had a weight of about 54 grams and a diameter 2r = 15 mm.
An important question is whether the heat applied is distributed fast
enough in the sample. This, however, was certainly the case. Putting the
diffusitivity of the material at about 5.10-3 cm2 sec, by a calculation based
upon the model of a cylinder in which initiallythe temperatureis everywhere
the same whilst from a given moment onwards the surface temperature
assumes' another, constant value, it is found that within a minute the dis-
tribution has practically finished 7). This is confirmed experimentally.
With method (B) one degree of temperature variation took about 2 minutes.

4. Results
The results of the absolute measurements (method (A)) are given
in table 1.
These values have been plotted in fig. 2, where it is seen that they follow
a smooth curve. The dashed parts of the curve indicate the three transition
regions (I, II and Ill) where humps in the specific heat occur.
The anomalies of Cp in these regions were determined by method (B).
As an example, the observations obtained in the lowest transition region
are reproduced in fig. 3. Fig. 4 shows the anomaly at 118 oe as calculated
from such observations. The values marked x were measured directly
(method (A)).
It appeared that the temperatures where maxima of Cp are found de-
pend" on whether the sample is heated or cooled: the substance shows
thermal hysteresis at the transitions. Both kinds of critical points (Tt and
Tt) were well reproducible. The anomalous heat has been integrated
with the help of a planimeter. This integral, the heat of transition L1E,
 24 J. VOLGER
.....
~~~~r--------'--------'--------'--------'--------'--------'-~-----'-4
~
ia!
Is,
-8
~50~------~--------+--------+--------+--------+--------4--------+~
:;
~
..._

vv--
~~----/~------+------+------+------4------~--~-4
I 11

5<~O------~~~------~I~~----~200~----~25kO~-----~~------~35kO
_TroK)
,
Fig. 2. Cp of BaTiOa. The dashed parts of the curve indicate the transition regions J, II i
and Ill.

is given in table Ir. In this table some other characteristics of the substance
investigated are collected.
. The quantity LIT is the width of a peak, taken as the difference between
the temperatures at which the anomalous specific heat has half its top value.
25

I:>.t (sec)

t
~ .:
\

-3000 -2500 -2000

~E(pV)
I I I I
190 200 210 220 TroK)
___
68539

Fig. 3. Example of observations made with the slow heating method (method (B. Transi-
tion I.
 SPECIFIC HEAT OF BARIUM TITANATE 25

TABLE I

Cp of BaTi03 (polycrystalline) as a function of temperature

Cp (Joule degree-1 Cp(Joule degree "

T(OK) T(OK)
Mole-1) Mole-1)

99 575 295 1015

106 64 302 102
112 71 311 1025
117 71 3265 103
1215 755 338 1035
129 79 346 1065
1375 83 349 1045
147 86 358 107
1535 865 365 1105
1635 915 373 112'5
176 955 380 117
248 98 387 1215
254 99 398 1215
261 99 406 1235
273 1005 407 1255

LlT was the same whether the sample was cooled or heated, with exception
of transition III where in a cooling experiment a somewhat broader peak
was found (marked (*) in the table) than in a heating experiment. In
view of the very-slow drift of temperature in these experiments it seems
very plausible to assume that LlT is an intrinsic property of the substance
under investigation.

TABLE n

Characteristics of critical points of BaTi03

(For explanation of symbols see text)

Tt (OK) T-j,(OK) LIE (Joule Mole-1) LlT(OK)

Transition
heating cooling heat of transition half-width

I 207 193 60 3
n 287 280 65 3
In 392 390'5 195 1;2*
 26 J. VOLGER

5. Remarks
Specific heat of the lattice. The curve ill fig. 2 shows a saturation of about
12 R at room temperature, which is 3 R less than the Dulong and Petit
value. At higher temperatures, however, Cp increases again and reaches
15 R. Obviously a single Debye function is not sufficient to describe the
whole curve. Probably a sum of a Debye function

12 RfD (~D)
and Einstein functions would fit better. e
D will be of the order of 400 oK.
For an exact determination of the characteristic temperatures in the func-
tions, measurements ought to be carried out at lower, respectively higher
temperatures than those taken here.
The Cp- Cv correction. From thermodynamical reasoning it follows that
av2
Cp- Cv = TV. p'
300~------~-------r------~--------~-------r------~--------~--~

i; 25m-------_4--------4-------_4--------~~----_4--------4_--------~~
~
10.0
0.0

i0.0200~------~--------~------~--------~~----~--------~------_+--_1
:;
~

x
x
..........
-!----x
x

roo~------~------_+--------~------~------_+------~~------4_--_1

50~------~----~~----~------~~----~------~------~--_;
I

~7~O~----~3~~----~3~8~O------3~8~5~----~3~W~----~395~-----4~O
I
-T(
Fig. 4. Cp calculated from the reciprocal drift of temperature by slow heating. Transition Ill.
 SPECIFIC HEAT OF BARIUM TITANATE 27

where V is the molar volume, av the coefficient of volume expansion and

fJ the compressibility of the substance. From Mason's 8) data fJ is readily
found to be

As the molar volume is 42 cm" and alin about 1.10-3 degree"! 6),9) one gets
at room temperature

which is very small in comparison with the measured value of Cp' At higher
tempera~ures the correction will be somewhat larger, but as no exact values
of fJ are available in this higher range no calculation has been made.
The anomalies. The value LIE at the transitions andn In
are in accord-
ance with those given by Blattner, Knzig and Merz 5) (71 respectively
196 Joule Mole-I). By heating their sample these authors found in region n
a critical temperature of about 5C, which value, however, agrees with T-j,
rather than with Tt of table Ir. They do not find peaks as sharp as those
shown in fig. 3 or fig. 4, but this may be accounted for by the fact that their
sample was a powder. Harwood et al. found the Curie point at Tt = 398 oK
which is 6 degrees higher than that given by Blattner et al. 5), and that
given in this paper. There exists also a considerable difference with the
results in the papers cited at 3) arid 6).
Mr W. Bol of this laboratory carded out some electrical measurements
at 5 kc/so When heating the sample he found a maximum of e at 393 "K
and of tan 0 at 388 "K. Cooling it he found a maximum of e at 3915 "K
and of tan 0 at 3865 "K,

Acknowledgement. The author is very grateful to Mr J. H. Aalberts

for his valuable help during the measurements and to Dr G. H. Jonker
and Dr J. H. van Santen for their interest in this investigation.

Eindhoven, June 1951

REFERENCES
1) H. F. Kay, H. J. Wellard and P. Vousden, Nature 163, 636-637, 1949;
H. F. Kay and P. Vousden, Phil. Mag. 40, 1019-1040, 1949.
2) W. Merz, Phys. Rev. 75, 687, 1949.
3) B. Vul, J. Phys. U.S.S.R. 10, 95-106, 1946.
4) M. G. Harwood, P. Popper and D. F. Rushman, Nature 160, 58-59, 1947.
5) H. Blattner, W. Knzig and W. Merz, Helv. Phys. Acta 22, 35-65, 1949.
G) S. Sawada and G. Shirane, J. Phys. Soc. Japan 4,52-56,1949.
7) See e.g.: H. S. Carsla w and J. C. Jaeger, Conduction of heat in-solids, Clarendon "
Press, Oxford, 1947, p. 175. .
8) W. P. Mason, Phys. Rev. 74, 1134-1148, 1948.
9) H. D. Megaw, Experientia 2, 183-184, 1946.