Ferroelectric domain evolution with temperature in BaTiO3 film on (001)SrTiO3 substrateGuoliang Yuan, Jiangpeng Chen, Hui Xia, Junming Liu, Jiang Yin et al. Citation: Appl. Phys. Lett. 103, 062903 (2013); doi: 10.1063/1.4818138 View online: http://dx.doi.org/10.1063/1.4818138 View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v103/i6 Published by the AIP Publishing LLC. Additional information on Appl. Phys. Lett.Journal Homepage: http://apl.aip.org/ Journal Information: http://apl.aip.org/about/about_the_journal Top downloads: http://apl.aip.org/features/most_downloaded Information for Authors: http://apl.aip.org/authors

Ferroelectric domain evolution with temperature in BaTiO3 film on (001)SrTiO3 substrate

Guoliang Yuan,1,a) Jiangpeng Chen,1 Hui Xia,1 Junming Liu,2 Jiang Yin,2 and Zhiguo Liu2,b)

1School of Materials Science and Engineering, Nanjing University of Science and Technology,Nanjing 210094, People’s Republic of China2National Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093,People’s Republic of China

(Received 20 April 2013; accepted 26 July 2013; published online 7 August 2013)

The evolution of ferroelectric domains in BaTiO3 film on (001) SrTiO3 substrate is studied at

20–120 �C. The dominant domains orient in (001) plane due to strong internal strain when

BaTiO3 island grains are embedded in the film, most of which is of layer-by-layer growth.

These domains are commonly larger than grains at 20 �C; however, they are close to grain size

at 60–80 �C, which largely enhances the coercive field. The in-plane domains are very stable at

20 �C; however, they can transform to out-of-plane domains at 40–80 �C under electric field,

which is likely the cause of enhanced piezoelectric response. VC 2013 AIP Publishing LLC.

[http://dx.doi.org/10.1063/1.4818138]

Domain evolution in ferroelectric materials is of great im-

portance for both fundamental physics and applications. Many

aspects of this topic are not yet fully understood. Domain

nucleation, growth, and switching depend significantly on

grain size, crystal orientation, thermal/epitaxial strain, and so

on. It is important to clarify the coupling between them using

direct imaging techniques.1–4 For example, mesoscale flux-

closure domain formation in BiFeO3 and BaTiO3 single crys-

tal or epitaxial thin films has been investigated using scanning

transmission electron microscopy (TEM) and piezoelectric

force microscopy (PFM).5–9 Compared with other ferroelec-

trics,10 BaTiO3 has a Curie temperature of �120 �C, which is

convenient for domain evolution study.4,11,12 Hofer and

McGilly have reported the annihilation of stripe domains

across the ferroelectric-paraelectric (FE-PE) phase transition

in BaTiO3 single crystal.12,13 Lu et al. reported switching of

polarization direction in BaTiO3 epitaxial thin film using

strain gradient with the tip of PFM).14

It is important to study the dependence of in-plane (IP)

domain evolution on temperature and strain in BaTiO3

film.12,13 If domains are larger than crystal grains at room

temperature, their size will be close to the grain at a certain

high temperature in ferroelectric films. It is interesting to

study the ferroelectric and piezoelectric properties of this

mono-domain. Besides, the strain can change electrical

polarization through piezoelectricity, and recently the strain

gradient is highlighted because it may switch polarization

through flexoelectricity.5 In a word, epitaxial/thermal strain

and its gradient are important to clarify and control the vari-

ous aspects in ferroelectric domain growth, such as domain

nucleation, orientation, size, and boundaries.

In this paper, we report that the strain induce mainly IP

a-domains in the BaTiO3 films grown on (001) SrTiO3

substrates. At elevated temperature, electric field can trigger

transformation from a-domain to out-of-plane (OP)

c-domain, leading to enhanced piezoelectric responses.

BaTiO3 films and La0.67Sr0.33MnO3 (LSMO, 50 nm)

buffer layers are grown on commercial (001) SrTiO3

substrates at 710 �C and 13 Pa oxygen pressure by pulsed laser

deposition (PLD). Laser energy per pulse is modulated in

order to adjust the growth modes of BaTiO3 films. Besides,

the target’s rough surface is helpful to get the island growth of

BaTiO3 film. There are three samples, i.e.,1-BaTiO3 (250 mJ,

300 nm thickness) with layer-by-layer growth, 2-BaTiO3

(200 mJ, 200 nm thickness) with the coexistence of island

and layer-by-layer growths, and 3-BaTiO3 (70 mJ, 60 nm

thickness) with island growth in most area. The as-grown

samples are cooled to 300 �C in 1000 Pa oxygen pressure with

5 �C/min, and then they are fast cooled to �30 �C within

2 min. The crystal structure and strain of BaTiO3 films are

characterized using X-ray diffraction (XRD, Bruker D8

advanced). PFM measurements are conducted using an Atomic

Force Microscope (AFM, Bruker multimode 8) with an open

loop controller, a conductive AFM tip (MESP-RC, Co/Cr coat-

ing, 35 nm tip Radius), and contract mode. An ac probing volt-

age (f¼ 41 kHz, 1–2 V sample bias) is used here, and the lower

one is used at higher temperature to insure that it is below the

coercive voltage of the BaTiO3 films. A heater is used to heat

the sample from 20 �C to 120 �C with accuracy of 0.1 �C.

Only diffraction peaks from (001), (002), and (003)

crystal planes of SrTiO3 substrate and BaTiO3 films can be

observed as shown in Fig. 1. There is no impurity observed

within the limit of XRD. The diffraction angles of the (003)

plane of 1-, 2- and 3-BaTiO3 films are 70.70�, 70.12� and

69.96�, respectively, indicating small compressive strain for

1-BaTiO3, mild strain for 2-BaTiO3, and large compressive

strain for 3-BaTiO3. It is noted that the strain and the thermal

expansion coefficient of BaTiO3 film on SrTiO3 substrate

change little at 20–120 �C, and thus the domain evolution

with temperature mainly originated from other factors, such

as the polarization change.

Figure 2 shows the topographies and PFM phases of the

three samples. Atomically flat surface with <1 nm roughness

is observed for 1-BaTiO3 (Fig. 2(a)). The c domains with

upward polarization are dominant as revealed by the OP

a)yuanguoliang@mail.njust.edu.cnb)liuzg@nju.edu.cn

0003-6951/2013/103(6)/062903/5/$30.00 VC 2013 AIP Publishing LLC103, 062903-1

APPLIED PHYSICS LETTERS 103, 062903 (2013)

(Fig. 2(b)) and IP PFM phases (Fig. 2(c)). By comparison, 2-

BaTiO3 shows a roughness of �10 nm in the 5� 5 lm2 area

(i.e., the 1st area) in Fig. 2(d), which suggests island growth

mode. Usually, the layer-by-layer growth leads to atomic

smooth surface, while the island growth generates rough sur-

face. When the deposition condition is properly controlled, a

transient between the two growth modes can be achieved,

and they can even coexist in a narrow window. Here, these

two growth modes coexist and their ratio can be adjusted

through the subtle change of film growth rate and the surface

roughness of the target. The phase angle is close to 0� in

Fig. 2(e), which is indicative of no OP domains. Indeed,

domain structures are observed in the IP PFM phase

(Fig. 2(f)). Most of the domains are different from the corre-

sponding topographic features. It suggests that the strain

between grains pin most domains in the (001) plane in this

sample. This is different from the common domain configu-

ration of BaTiO3 single crystal, i.e., a–c stripe zigzag domain

structure.6,13 The roughness of 3-BaTiO3 increases to

�15 nm as shown in Fig. 2(g). Both OP domains (Fig. 2(h))

and IP domains (Fig. 2(i)) are clearly identified. Most

domains are larger than the grain size in the corresponding

topography image. Clearly, the strain between neighboring

grains plays an important role in determining the domain

structure.4 In the following discussion, we focus on 2-

BaTiO3. Its OP PFM phases (Fig. 3), IP PFM phases (Fig. 4),

and phase/amplitude versus dc voltage loops (Fig. 5) of the

1st and/or its nearby areas are collected in the same heating

and cooling processes at 20–120 �C, respectively.

The evolution of OP phase on temperature was shown in

Fig. 3. A special feature labeled with dashed red circle is

used to identify the same location. At each temperature, the

3� 3 lm2 and 1� 1 lm2 regions are polarized with �8 V

and 8 V sample biases upon LSMO bottom electrode, respec-

tively, and then the PFM phases are scanned with an ac prob-

ing voltage immediately. At 20 �C, the OP phase does not

show obvious change between the two polarized regions in

Fig. 3(a). This suggests that (1) 8V is not large enough to

induce a- to c-domain transition or (2) the switched domain

quickly relaxes back to a-domain. The following PFM phase

vs. tip bias loops confirms that (2) is the case. With tempera-

ture increasing from 40 �C to 60 �C, the color contrast

between the two polarized regions becomes more and more

clear (Figs. 3(b)–3(d)), which suggests that the domains with

upward or downward polarization orientations begin to stabi-

lize. Furthermore, the phase degree of the as-grown film out-

side the 3� 3 lm2 region is close to that of the 8V polarized

region at 60 �C in Fig. 3(d). This proves that some IP

domains have switched to upward orientation at �60 �Ceven without external electric field, which is consistent with

the upward self polarization in many as-grown BaTiO3 films

on (001) SrTiO3 substrate.14 The color contrast of two

FIG. 2. The surface images, OP phase

and IP phase of (a)–(c) 1-BaTiO3,

(d)–(f) 2-BaTiO3, and (g)–(i) 3-

BaTiO3 films, respectively, where

insets show the layer-by-layer growth,

the island growth, and the strains

between neighboring island grains

guided by red dashed circles.

FIG. 1. The XRD patterns of 1-, 2-, and 3-BaTiO3 samples, where the inten-

sities of 2- and 3-BaTiO3 samples were amplified 5 and 25 times, respec-

tively, for clear exhibition.

062903-2 Yuan et al. Appl. Phys. Lett. 103, 062903 (2013)

polarized regions becomes weak at 80 �C (Fig. 3(e)) and then

nearly disappears at 120 �C (Fig. 3(f)) due to the

ferroelectric-paraelectric phase transition of BaTiO3.11,12

Afterward, the sample was cooled down to 60 �C and then

20 �C for further measurements. Fig. 3(g) proves that most

domains can be switched at 60 �C, and Fig. 3(h) suggests

that some domains can be switched at 20 �C in the 1st area

by 68 V sample biases. The OP phase behaviors after the

sample cooling down (Fig. 3(h)) are different from those

before its heating process (Fig. 3(a)), which is due to the

multiple switching processes or the influence of ac and dc

voltages on the later domain growth. To avoid this influence,

a nearby 5� 5 lm2 area, which was not measured during

heating and cooling processes, was measured after the sam-

ple was cooled down to 20 �C, and its OP phase in Fig. 3(i)

is similar to that in Fig. 3(a).

FIG. 3. The OP PFM phase of the 1st

5� 5 lm2 area in 2-BaTiO3 film at (a)

20 �C, (b) 40 �C, (c) 50 �C, (d) 60 �C,

(e) 80 �C, and (f) 120 �C during heat-

ing process and at (g) 60 �C and (h)

20 �C during cooling process; (i) the

OP PFM phase of a nearby area at

20 �C after cooling process, where the

68 V polarized regions are marked

with dashed blue/red squres,

respectively.

FIG. 4. The morphology surface of a

2� 2 lm2 area in 2-BaTiO3 film at (a)

20 �C; its corresponding IP PFM phase

at (b) 20 �C, (c) 40 �C, (d) 60 �C, (e)

80 �C, and (f) 120 �C during heating

process and at (g) 60 �C and (h) 20 �Cduring cooling process; (i) the IP PFM

phase of another nearby area at 20 �Cafter cooling process.

062903-3 Yuan et al. Appl. Phys. Lett. 103, 062903 (2013)

Most IP domains of 2-BaTiO3 are far larger than the cor-

responding grains at 20 �C; however, the structure of a domain

per grain is common at 60 �C. They gradually become smaller

with increasing temperature up to its Curie temperature. Note

that surface screening charges are not needed to neutralize

bound charges for IP domains unlike that of OP domains, thus

their influence on domain size can be ignored.3 Most grains

have diameters of �180 nm in a 2� 2 lm2 area that is near the

1st area (Fig. 4(a)), while the maximum length of the ferroelec-

tric domains is �1.2 lm at 20 �C (Fig. 4(b)). There are two dif-

ferent colors, i.e., milk white and dark yellow, suggesting at

least two types of IP domains.13 As temperature increases, the

topography features of the sample are maintained. However,

the contrast between the two types of domains becomes

weaker and the big domains become smaller at 40 �C (Fig.

4(c)). The temperature and strain may be not homogenous;

thus their gradient can decompose a big domain to several

smaller ones with temperature increasing. By comparison,

McGilly et al. reported that the IP domains of BaTiO3 single

crystal was relatively stable below TC; however, they abruptly

annihilated in a narrow temperature window (<1 �C) near

TC.13 At 60 �C, lots of IP domains are close to the size of corre-

sponding grain (Fig. 4(d)), and the structure of one domain per

grain can largely enhance the coercive field of domain. The

domains continually decrease its size with temperature increas-

ing (Fig. 4(e)) and finally disappear at 120 �C (Fig. 4(f)). As

discussed above, the ac probing voltage may influence the do-

main growth to some extent.15 Even so, the IP domains of the

1st area are in the great majority when samples were cooled

down to 60 �C (Fig. 4(g)) and 20 �C (Fig. 4(h)), respectively.

Most importantly, in-plane domains still overwhelm the most

area that was not measured during heating and cooling proc-

esses, and one example is shown in Fig. 4(i).

The PFM phase/amplitude versus tip bias loops also con-

firm that the in-plane domains are less stable with temperature

increasing. At each temperature, phase/amplitude loops of

BaTiO3 were repeatedly measured at more than five positions

but only the most typical loops were shown in Fig. 5 for con-

venient observation. It is noted that the ac sample bias and dc

tip bias were applied simultaneously during measurements

with this method; the electrostatic interaction between the

sample and the tip may contribute a linear slope of the PFM

amplitude as function of DC bias. The room-temperature

loops of 1- and 3-BaTiO3 are nearly symmetric in Figs. 5(a)

and 5(b), while those of 2-BaTiO3 are asymmetric in Fig.

5(c). The coercive voltages, VC, are �3.1 V and �1.2 V at

20 �C in Fig. 5(c). Domains with downward polarization exist

under <�3.1 V tip bias, and they can switch to an upward

state when the tip bias is above �1.1 V. However, soon after

the tip bias is removed, strain pushes the domain to be IP

again since the upward-polarization domain is just an interme-

diate state at 20 �C. By comparison, the PFM phase versus tip

bias loop measured at 40 �C shows VC values of �1.1 V and

0.4 V in Fig. 5(d). The smaller VC shift suggests that IP

a-domains are easier to transform to OP c-domains at 40 �C,

mainly due to the smaller polarization at higher temperature.

Accompanying the a- to c-domain transformation induced

by electric field, the converse piezoelectric responses of

2-BaTiO3 sample shows a dramatic increase as temperature

increases. The amplitude is due to the piezoelectric d33 coeffi-

cient of OP domains and the d31 coefficient of IP domains.

The d33 value (i.e., about 150 pm/V) is far larger than d31 (i.e.,

about �80 pm/V) in BaTiO3 crystal.16 The amplitude at 20 �Cmainly involves the as-grown IP domains, thus the depend-

ence of amplitude on tip bias is very weak in Fig. 5(c). The IP

domains become smaller, and hence, it is much easier for

them to become OP domains under tip bias at 40 �C, which

can contribute to larger amplitude through their d33 coefficient

in Fig. 5(d). Most of IP domains can switch to OP domains

under enough tip bias and then they contribute a large increase

of amplitude at 60 �C in Fig. 5(e) and 80 �C in Fig. 5(f),

respectively. The piezoelectric enhancement may originate

from a-domains which can switch to c-domains under outer

electric field at an elevated temperature. However, the ampli-

tude of 2-BaTiO3 becomes very small at 120 �C (Fig. 5(g))

because the FE-PE phase transition occurs and the values of

both d33 and d31 are nearly zero in the paraelectric phase. As

comparison, the amplitude of epitaxial PbZr0.8Ti0.2O3 (PZT,

200 nm thickness) film is shown in Fig. 5(h), which suggests

that the amplitude of 2-BaTiO3 at 60–80 �C is �70% ampli-

tude of the PZT film.

In conclusion, the strain between island grains and other

BaTiO3 films can pin most domains in (001) plane. The sizes

of most IP domains are much larger than island grains at

20 �C; however, they become smaller with temperature

increasing. Piezoelectric response and coercive voltage

increase largely when mono-domain appears in most regions

at 60–80 �C. The IP domains are very stable at 20 �C; how-

ever, they are less stable and can switch to the OP when the

ferroelectric polarization becomes smaller at a higher tem-

perature. This finding is important for the understanding of

FIG. 5. The OP phase and amplitude versus tip bias loops of (a) the 1-

BaTiO3 at 20 �C and (b) the 3-BaTiO3 at 20 �C; those of the 2-BaTiO3 film

at (c) 20 �C, (d) 40 �C, (e) 60 �C, (f) 80 �C, and (g) 120 �C during heating

process and those of (h) the PZT film at 20 �C.

062903-4 Yuan et al. Appl. Phys. Lett. 103, 062903 (2013)

domain evolution in ferroelectric films and the engineering

of high performance piezoelectric devices.

This work was supported by the National Natural

Science Foundation of China (11134004), the National Key

Project for Basic Research of China (2012CB619406 and

2009CB623303), Natural Science Foundation of Jiangsu

Province (SBK201123822), and the Fundamental Research

Funds for the Central Universities (30920130111018).

1B.-K. Lai, I. Ponomareva, I. A. Kornev, L. Bellaiche, and G. J. Salamo,

Phys. Rev. B 75, 085412 (2007).2X. Y. Liu, Y. M. Liu, S. Takekawa, K. Kitamura, F. S. Ohuchi, and J. Y.

Li, J. Appl. Phys. 106, 124106 (2009).3D. Y. He, L. J. Qiao, A. A. Volinsky, Y. Bai, and L. Q. Guo, Phys. Rev. B

84, 024101 (2011).4J. Paul, T. Nishimatsu, Y. Kawazoe, and U. V. Waghmare, Phys. Rev.

Lett. 99, 077601 (2007).5G. Catalan, A. Lubk, A. H. G. Vlooswijk, E. Snoeck, C. Magen, A.

Janssens, G. Rispens, G. Rijnders, D. H. A. Blank, and B. Noheda, Nature

Mater. 10, 963–967 (2011).

6R. G. P. McQuaid, L. J. McGilly, P. Sharma, A. Gruverman, and J. M.

Gregg, Nat. Commun. 2, 404 (2011).7L. J. McGilly, A. Schilling, and J. M. Gregg, Nano Lett. 10, 4200

(2010).8N. Balke, B. Winchester, W. Ren, Y. H. Chu, A. N. Morozovska, E. A.

Eliseev, M. Huijben, R. K. Vasudevan, P. Maksymovych, J. Britson, S.

Jesse, I. Kornev, R. Ramesh, L. Bellaiche, L. Q. Chen, and S. V. Kalinin,

Nat. Phys. 8, 81 (2012).9C. L. Jia, K. W. Urban, M. Alexe, D. Hesse, and I. Vrejoiu, Science 331,

1420 (2011).10K. S. Wong, J. Y. Dai, X. Y. Zhao, and H. S. Luo, Appl. Phys. Lett. 90,

162907 (2007).11S. V. Kalinin and D. A. Bonnell, Appl. Phys. Lett. 78, 1116 (2001).12A. Hofer, M. Fechner, K. Duncker, M. Holzer, I. Mertig, and W. Widdra,

Phys. Rev. Lett. 108, 087602 (2012).13L. J. McGilly, T. L. Burnett, A. Schilling, M. G. Cain, and J. M. Gregg,

Phys. Rev. B 85, 054113 (2012).14H. Lu, C.-W. Bark, D. Esque de los Ojos, J. Alcala, C. B. Eom, G.

Catalan, and A. Gruverman, Science 336, 59 (2012).15V. Ya. Shur, V. A. Shikhova, A. V. Ievlev, P. S. Zelenovskiy, M. M.

Neradovskiy, D. V. Pelegov, and L. I. Ivleva, J. App. Phys. 112, 064117

(2012).16R. Tazaki, D. Fu, M. Itoh, M. Daimon, and S.-Y. Koshihara, J. Phys.:

Condens. Matter 21, 215903 (2009).

062903-5 Yuan et al. Appl. Phys. Lett. 103, 062903 (2013)