1

COMPARATIVE STUDY OF PHASE TRANSITIONS IN POLYCRYSTALLINE AND

EPITAXIAL BaTiO3 THIN FILMS BY MEANS OF SPECIFIC HEAT MEASUREMENTS

B.A.Strukov1, S.T.Davitadze1, V.V.Lemanov2, S.G.Shulman2, Y.Uesu3 and S.Asanuma3

1
Lomonosov Moscow State University, Moscow 119992, Russia
e-mail: bstrukov@mail.ru
2
Ioffe Physico-Technical Institution RAS, S.-Petersburg 194021, Russia
3
Waseda University, Tokyo 169-8555, Japan

Abstract The experimental data on the thermal properties of the nanostructured perovskite ferroelectrics
are presented and analysed. The ability of the modified 3ω method for specific heat measurements is dis-
cussed and illustrated by the thermal properties of the thin polycrystalline (BaTiO3 on fused SiO2 , 50-
1100 nm) and epitaxial (BaTiO3 on MgO, 50-500 nm) films. The sharp decrease of the transition tempera-
ture, surplus entropy and spontaneous polarization were obtained for the polycrystalline films. The criti-
cal thickness and crystallite size are estimated for this case from the variation of both crystallite sizes at
constant thickness and film thickness at constant crystallite size. The similarity of specific heat, phase
transition excess entropy evolution in polycrystalline films and nanograined ceramics (the latter was
obtained in [12] ) is revealed. The data on the specific heat of epitaxial films show the extremely diffused
anomaly near 400K, the phase transition has the weak tendency of shifting to higher temperatures with
the decreasing of the film thickness. No anomalies were detected for the thinnest (50 nm) film up to 480K.

In the field of ferroelectricity the problem of finite-size effects received the powerful
practical interest because of the number of applications of the ferroelectric nanostructured
devices [1-3]. It is very important to know the limit of presence of ferroelectric properties
and phase transition when the size of devices is reduced. Thin films are the most interesting
objects in this respect; as well for applications it is important to know the dependence of
physical properties of ferroelectric films and ceramics on the crystallite size. Therefore the
finite-size effects in ferroelectrics have both the practical importance and fundamental scien-
tific sounding and several recent review papers and books on the subject and its development
throw much light on the problem [4-5]. It should be noted that these sources are accented
mainly on consideration of ferroelectric oxides films as elements of memory devices and
therefore the polarization switching processes, coercive electric field and related phenomena
are considered there as subjects of the main interest. In this paper we shall concentrate our
attention mainly on the thermal properties of the nanostructured perovskites directly related
to ferroelectric phase transitions. Perovskite ferroelectric oxides like BaTiO3, PbTiO3,
SrTiO3, (Ba,Sr)TiO3 etc are the most attractive for both applications and as models for finite-
size effect study because of their relatively simple crystal structure and well developed
methods of manufacturing.
 2

Although the switching parameters and dielectric properties of the ferroelectric nanos-
tructures have a prevalence importance at the moment, the critical phenomena and phase
transition singularities seem to have a fundamental sounding for the subsequent understand-
ing of the size-effects nature in these objects. It should be noted that the experimental data
obtained by means of dielectric measurements in films might be essentially distorted by the
depletion layers with a low dielectric constant, arising near electrodes [6,7]. Since we need
an information about volume properties of the nanoparticles or films, the “non-electrical”
data can be useful for the characterization of the phase transitions in these ferroelectric
nanosized objects. We consider specific heat measurements as an important method to obtain
the direct information about transition temperatures and phase transition behavior; moreover
it gives a chance to get indirectly a spontaneous polarization temperature dependence in the
vicinity of the transition point.
The temperature dependence of the specific heat of free-standing (without substrate)
ferroelectric films was analysed in [8-10] within Landau-Ginzburg-Devonshire (LGD) the-
ory phenomenological approach including in the free energy density of the term proportional
to |gradΡ|2 together with a boundary condition involving an extrapolation length δ. The fi-
nite discontinuity in the specific heat was found at the film critical temperature Tc . The
jump of the film specific heat, excess entropy and Tc position were found to decrease to-
gether with the decreasing of a film thickness.
Methodically there are no special problems with the application of the scanning and adia-
batic calorimetry for investigation of the size effects in powder (usually fine particles com-
pacted into pellets) and nanograined ceramics. The rather sharp decrease of the Curie tem-
perature as well as of excess energy of the phase transition were revealed for BaTiO3 (BTO)
and PbTiO3 (PTO) fine particles [11] and BTO ceramics [12-14] for the diminishing size
of particles and crystallites. These effects were considered as an evident manifestation of the
theoretically predicted size effect . The obvious smearing of the phase transition was ob-
served in the both cases; for the ceramics the problems of correlation of dielectric and ther-
mal properties was discussed [12].
The first measurement of the specific heat temperature dependence for the epitaxial BTO
films on SrTiO3 substrate was undertaken by A.Onodera et al. with usage of the ac-method
[15]; the data revealed the small and diffused anomalies for the 6 and 200 nm thick
monocrystalline films. The thermal hysteresis - a visible difference of the specific heat val-
ues on heating and on cooling - was detected. Authors supposed the significant increasing of
 3

the transition temperature in comparison with the bulk samples which was attributed to in-
fluence of the lattice misfit strain between films and substrate.
The next attempts to measure specific heat of the thin ferroelectric films sputtered on the
substrate were undertaken when the 3ω dynamical method was applied to the analysis of
thermal properties of the composite samples dielectric film – massive substrate [16,17]. It
was shown that if the thermal contrast K between film and substrate ( K=cfλf/csλs , c is spe-
cific heat of the film (f) and substrate (s) ) is small enough, the direct measurement of the
film specific heat is possible down to 20 nm film thickness [16]. Therefore BaTiO3 films on
the fused SiO2 and monocrystalline MgO as substrates seem to be convenient match to
measure specific heat of the polycrystalline and epitaxial films (K=0.1 for the BaTiO3 / SiO2
and 0.2 for BaTiO3 / MgO couples ).
BaTiO3 polycrystalline films were grown on the fused quartz by rf magnetron sputtering
at the different substrate temperatures in atmosphere Ar/O2 ≈0.8/0.2 . The rate of sputtering
was about 300 nm/h. AFM analysis of the film surface showed that the size of monocrystal-
line regions depends on the temperature of substrate in time of sputtering. For the tempera-
ture of substrate near 8000C it is to be about 110-150 nm in the interval of film thickness
100-1100 nm and decreases sharply with the further decreasing of a thickness. With decreas-
ing of the substrate temperature the average size of the monocrystalline regions was reduced,
so we had a chance to modify both the film thickness and crystalline size separately. The
columnar-like structure was found for all these films. The epitaxial films were grown on
MgO single crystal substrate (as = 4.213 A) by the pulsed laser deposition method. The sur-
faces of all films were controlled by AFM. In polycrystalline films the mosaic structure was
evident; but in the case of epitaxial films no grains of mosaic were observed revealing the
perfect “epitaxial coupling” between film and substrate. AFM images of 100 nm BTO films
deposited on fused SiO2 and MgO are shown at Fig.1.
The temperature dependence of the specific heat of polycrystalline BaTiO3 films are pre-
sented at Fig. 2a (variation of the thickness for an approximately constant column diameter)
and at Fig. 2b (variation of the column diameter for a constant film thickness). In the both
cases the sharp decrease of the transition temperature and excess entropy of the phase transi-
tion is evident (Fig. 3a, 3b); the phase transition becomes diffused when the film thickness
declined. The temperature dependence of the spontaneous polarization can be taken out
from Fig.3 by the well known LGD theory relations and is presented at Fig. 4a, 4b. It is
clearly seen that spontaneous polarization becomes smaller for the thinner polycrystalline
 4

films; it is important that in these experiments we obtain (although indirectly) the real vol-
ume polarization whereas in the direct measurements of Ps from hysteresis loop or switch-
ing current one can not control the nonswitching regions of the samples. It is known that
LGD theory predicts that the shift of the transition temperature is Tb – Tc = Ad-1 where Tb is
the Curie temperature for a bulk sample, d is the crystal thickness or grain size and A –
combination of LGD free energy coefficients. So we can use the obtained data for evaluation
of critical thickness and critical grain size from the condition Tc = 0. From this condition we
have for films dcrit ≈ 2.5 nm and for grains dcrit ≈ 8 nm. It should be noted that in general the
evolution of the phase transition in polycrystalline thin films, as considered from the ther-
mal properties, looks rather similar to that observed in nanograined ceramics [12]. Evidently
in both cases the observed changes of the transition temperature, excess entropy and sponta-
neous polarization can be attributed to an intrinsic size effect. It was shown in [12] that the
residual stresses resulting from the tetragonal deformation play a minor role on the behavior
of the fine grained ceramics; the critical size of the grains in nanocrystalline ceramics was
estimated as 10-30 nm. In our case of the polycrystalline thin films on the fused quartz, the
additional two-dimensional stresses arising from an interaction between film and substrate
seem also to be almost completely relieved. The observed smearing of phase transition in the
fine particles, nanoceramics and polycrystalline thin films can be described by taking into
account the distribution of the grain sizes which exists in all nanomaterials and results in the
corresponding distribution of transition temperatures [18].

It was revealed that the quite another type of thermal behavior is typical for the epitaxial
films BaTiO3/MgO. The temperature dependence of the specific heat and spontaneous po-
larization of films with the different thickness is presented at Fig. 5,6. It is seen that there is
a weak tendency of increasing of the transition temperature which can be identified as a
maximum of the heavily diffused specific heat anomalous part (see insert at Fig.5). It is also
remarkable that the film with the thickness 50 nm shows no anomalous change up to 480K
(that is an upper temperature limit for our method at present). We consider this result as an
evident competition of the two effects, one is the size-effect theoretically predicted for free-
standing films (declining of Tc with decreasing of the film thickness) and the second is a
mismatch between the MgO lattice parameter as and the cubic cell constant ao of the free-
standing BaTiO3 film, clamping the film in two directions. In fact the lattice mismatch is
very large for BaTiO3/MgO couple: the misfit strain defined as um = (as – ao)/as + (αs –
 5

αo)(T – Tg) (here αs and αo are the thermal expansion coefficients of the substrate and film,
Tg is the crystal growth temperature) is about 5%.

It was shown at [19] that when the substrate lattice parameter is larger than the film lattice
parameter and the film thickness is larger than the critical value hc ≈30-50 nm , the film may
partially relax at the growth temperature by the generation of misfit dislocations. It leads to a
value of the “effective” substrate lattice parameter as eff < as so that the BaTiO3 film on
MgO substrate has in fact a tensile strain of near 0.57% [20]; if one takes into account the
temperature-misfit strain phase diagram for epitaxial (001) BaTiO3 film revealed in [21],
the phase transition is supposed to be at about 600K. Therefore we are forced to assume that
the further release of the tensile mechanical stresses takes place at least for the films thicker
than 50 nm. The Table sums up the data, obtained for polycrystalline (with completely re-
lieved stresses) and epitaxial films; here Tc is the transition temperature of the epitaxial
films, ∆Tc (the second column) – displacement of Tc due to the size effect (in polycrystal-
line films), ∆Tc (the third column) - displacement of Tc due to both size effect and mis-
match effect for the each thickness of epitaxial films. The last column gives the total resid-
ual strain resulting in the observed shift of Tc . The relationship between ∆Tc and um is
taken from [21].

In summary we studied the thermal properties of BaTiO3 polycrystalline and epitaxial

films by means of the modified 3ω method. The clear size effect, similar to one obtained in
the nanostructured ceramics, was detected and the critical film thickness and grain size were
estimated in polycrystalline BaTiO3/SiO2 films. Through the temperature dependence of the
specific heat of epitaxial films BaTiO3/MgO we revealed the misfit strains which were
found to decrease when the film thickness grows up.

This work was partly supported by RFBR grants (projects 05-02-16873 and 06-02-16664)
 6

REFERENCES
1. J.F.Scott, Ferroelectric Review 1, 1-129 (1998).
2. J.F.Scott, Ferroelectric Memories (Springer-Verlag, Berlin, 2000).
3. M.Dawber, K.M.Rabe and J.F.Scott, arxiv: cond-mat/0503372 v1, 15 Mar.2005.
4. Eds C.P.de Araujo, J.Scott and G.W.Taylor, Ferroelectric Thin Films: Synthesis and
Basic Properties (Gordon and Breach Publishers, Amsterdam, 1996).
5. Eds H.Ishiwara, M. Okuyama and Y.Arimoto, Ferroelectric Random Access Memo-
ries (Springer-Verlag, Berlin, 2004).
6. O.G.Vendik and S.P.Zubko, J.Appl.Phys. 88, 5343 (2000).
7. A.Lookman, R.M.Bowman and J.M.Gregg, J.Appl.Phys. 96, 555 (2004).
8. C.Wang and S.Smith, J.Phys.: Cond.Matter 7, 7163 (1995).
9. S.V.Pavlov and O.Yu.Poliakova, Vestnik Mosk. Univ., Ser. Phys. Astronom. №1, 44
(2000).
10. L.-H.Ong, J.Ostman and D.Tilley, Phys.Rev. B, 63, 144109 (2001).
11. W.Zhong et al., J.Phys.: Cond.Matter 5, 2619 (1993).
12. Z.Zhao et al., Phys.Rev. B 70, 024107 (2004).
13. M.H.Frey and D.A.Payne, Phys.Rev. B 54, 3158 (1996).
14. V.Tura et al., Jpn.J.Appl.Phys. 37, 1950 (1998).
15. A.Onodera et.al., J.Europ.Ceramic Soc. 19, 1477 (1999).
16. S.T.Davitadze et al., Solid State Phys., 42, 111 (2000).
17. B.A.Strukov et.al., J.Phys.: Cond.Matter 15, 4331 (2003).
18. M.D.Glinchuk and P.I.Bykov, J.Phys.: Cond.Matter., 16, 6779 (2004).
19. J.S.Speck and W.Pompe, J.Appl.Phys. 76, 466 (1994).
20. F.He and B.O.Wells, arxiv: cond-mat/0511518, 21 Nov. 2005.
21. N.A.Pertsev et al., Phys.Rev.Lett. 80, 1988 (1998).
 7

Captions to figures to the paper of B.A.Strukov et al.

Fig.1. AFM image of BTO films deposited on fused SiO2 (a) and MgO (b).

Fig.2. Temperature dependence of the specific heat for polycrystalline BaTiO3/SiO2 films
with different thickness (a) and different grain size (b).

Fig.3. Temperature and excess entropy of the ferroelectric phase transition in polycrystalline
BaTiO3/SiO2 films with different thickness (a) and different grain size (b).

Fig.4. Temperature dependence of spontaneous polarization in polycrystalline BaTiO3/SiO2

films with different thickness (a) and different grain size (b).

Fig.5. Temperature dependence of the specific heat of BaTiO3/MgO epitaxial films. Inset
– Tc and excess entropy vs film thickness.

Fig.6. Temperature dependence of spontaneous polarization for BaTiO3/MgO epitaxial

films.
 8

Fig.1 (a) (b)

Fig 2 (a) (b)

Fig. 3 (a) (b)

 9

Fig 4. (a) (b)

Fig. 5

Table 1. Estimation of the total residual strain in BaTiO3/Mgo epitaxial films.

h, nm TC (K) ∆ TC (K), (size ∆ TC (K), (misfit Thermal ex- Total residual

effect, poly- strain, epitaxial pansion strain
cryst. films) films) strain
450 395 −1 1 2,8⋅10-5
300 401 −3 9 - 2,2⋅10-4 2,5⋅10-4
100 412 −8 25 7⋅10-4
50 > 470 − 18 > 93 > 2,6⋅10-3