Electro-shape-memory effect in Mn-doped BaTiO 3 single crystals and in situ observation of the reversible domain switching

Materials Science and Engineering A 438 440 (2006) 354 359 Electro-shape-memory effect in Mn-doped BaTiO 3 single crystals and in situ observation of the reversible domain switching L.X. Zhang a,b,c,,x.ren a,b a Multi-Disciplinary Materials Research Center, Xi an Jiaotong University, Xi an 710049, People s Republic of China b Materials Physics Group, National Institute for Materials Science, Tsukuba, 305-0047 Ibaraki, Japan c State Key Laboratory for Mechanical Behavior of Materials, Xi an Jiaotong University, Xi an 710049, People s Republic of China Received 19 April 2005; received in revised form 26 December 2005; accepted 28 February 2006 Abstract Very recently a giant electric-field-induced recoverable shape change effect (electro-shape-memory effect) has been found in aged BaTiO 3 -based ferroelectric crystals. This effect is supposed to come from a reversible domain-switching process with the restoring force provided by a symmetryconforming principle of defects. However, the reversible domain switching process itself behind the giant electro-shape-memory effect is yet to be directly verified. In the present study, we performed in situ domain observation during electric field cycling for aged Mn-doped BaTiO 3 single crystals and simultaneously measured its macroscopic properties. It was found that the aged sample shows a reversible domain switching during electric field cycling; this corresponds very well to the electro-shape-memory effect (with a maximum shape change of 0.4%, which is almost half of the theoretical value). This provides a direct mesoscopic evidence for the defect-mediated reversible domain switching mechanism. Moreover, as all these effects stems from aging in ferroelectric state, we further explained that their microscopic origin is symmetry-conforming property of defects, which is completed during aging by defect migration. 2006 Elsevier B.V. All rights reserved. Keywords: Electro-shape-memory effect; Defect symmetry; Reversible domain switching 1. Introduction Ferroelectric crystals show a spontaneous polarization P S below Curie temperature T c due to a symmetry-lowering phase transition [1]. P S can align along one of the several symmetryallowed orientations, and a region with the same P S is called a domain [2]. When applied electric filed, one domain state can be switched to another; this will generate a large shape change [3]. However, such a shape change is unrecoverable. The reason is that domain switching by electric field is inherently irreversible due to the energetic equivalence of different domain states. If domain switching can be somehow made reversible, a large recoverable electric-field-induced shape change would be expected. Recently we proposed a principle to realize a reversible domain switching in a crystal containing point defects and consequently possible to produce a large recoverable electric- Corresponding author. Tel.: +81 29 8592706; fax: +81 29 8592701. E-mail address: zhang.lixue@nims.go.jp (L.X. Zhang). field-induced shape change [4]. The main idea of this principle is that point defects have a so-far unrecognized statistical symmetry which follows the crystal symmetry when in equilibrium (in short, it is called defect symmetry principle) [4 8]. The symmetry-conforming configuration of point defects cannot have a sudden change when applied electric field, then it acts as an intrinsic restoring force for reversible domain switching [4,8]. Based on this defect-mediated reversible domain-switching mechanism, a giant recoverable electric-field-induced shape change (with a value of 0.75%) has been achieved in aged Fedoped BaTiO 3 single crystals [4] and the same effect with shape change of 0.12% was achieved in aged Mn-doped (Ba,Sr)TiO 3 ceramics [8]. Such an effect is very similar with the rubber-like behavior in physically parallel ferroelastic martensite alloys after aging, which means the memory of original shape during stress cycling and consequently is referred as stress-induced shape memory effect. For analogical reason, we referred the electric-field-induced recoverable shape change as electroshape-memory effect, which means the memory of original shape during electric field cycling. Moreover, it is known that the stress-induced shape memory effect has also been successfully 0921-5093/$ see front matter 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2006.02.127

L.X. Zhang, X. Ren / Materials Science and Engineering A 438 440 (2006) 354 359 355 explained by defect-mediated reversible domain-switching mechanism [5 7] and the mesoscopic reversible domain switching process behind stress-induced shape memory effect when loading and unloading stress had already been observed [7]. However, in ferroelectric system the direct evidence for the existence of the reversible domain switching process behind giant electro-shape-memory effect is yet to be obtained. In the present study, we therefore performed an in situ domain observation during electric field cycling for aged Mn-doped BaTiO 3 single crystals, aiming to verify the reversibility of actual domain-switching process. Moreover, P E hystersis loop and electric-field-induced shape change measurement were also characterized. Thus, a direct link between macroscopic property and the underlying mesoscopic domain switching behavior is unambiguously established. 2. Experiment The Mn-doped BaTiO 3 single crystal sample (denoted as Mn BaTiO 3 ) used in present work was grown by BaCl 2 flux in Ar atmosphere at about 1300 C [9]. The valence state of Mn ion is mostly 3+ under such a growth condition [9]. The concentration of Mn ions was determined by EPMA to be about 0.3 mol% relative to the B-site Ti ions. The grown Mn BaTiO 3 crystal was polished to less than 100 m thick, heated to above T c and kept for 4 5 h. Then it was fast cooled to 80 C and put at this temperature (i.e., in ferroelectric state) for two weeks to establish a fully aged state. In order to verify the domain switching process behind the macroscopic properties, we performed two kinds of experiments on aged samples. Firstly, domain-switching observation and P E hysteresis loop measurement were simultaneously done. The aged sample was painted with air-drying silver electrode on both lateral sides. Then it was placed on a transparent glass plate with the electrodes connecting to the voltage output of a ferroelectric tester (Radiant workstation). During electric field cycling, the domain switching process was observed with an optical microscope and recorded by a digital camera. At the same time P E hysteresis loop was measured by the ferroelectric tester. Then the electric-field-induced shape change of the sample was measured by a photonic sensor during a hysteresis loop measurement [4]. The sample with electrodes on both surfaces was connected to a ferroelectric tester (Radiant workstation). Electric field was thus applied along thickness direction. Above the surface of the sample was the non-contact displacement sensor. During a hysteresis measurement both hysteresis loop and shape change-field curve were obtained. 3. Results 3.1. In situ domain pattern observation during electric field cycling In the following, we show in Fig. 1 the micrographs of domain patterns during electric field cycling and corresponding P E hysteresis loop for aged Mn BaTiO 3 single crystals. Beside each micrograph, the corresponding defect symmetry state and P S state is also illustrated; but this will be discussed later. Here we only show the experimental observation. For the aged multi-domain sample, averaged polarization equals to zero, corresponding to point A in the hysteresis loop curve (see Fig. 1). When electric field is applied, domain switching occurs, leading to a gradual increase of polarization. When the field reaches maximum, an almost single domain configuration is observed, which corresponds to the maximum polarization (point B in P E curve). Interestingly, when electric field decreases to zero, we observed the same multi-domain pattern as the original one (compare the micrograph at point A and C). At the same time, the averaged macroscopic polarization becomes zero (point C in P E curve). Furthermore, the same is true for reverse electric field except that polarization changes to negative (see C D E cycle in Fig. 1). Thus, we observed an interesting reversible domain switching process during electric field cycling in aged sample. It corresponds well to the peculiar double P E hysteresis loop, which is contrasting with the well known normal hysteresis loop without aging. 3.2. Electric-field-induced shape change associated with domain switching process Now we show another important macroscopic consequence accompanying the domain switching process, the electric-fieldinduced shape change. As after treatment for domain observation, the size of specimen is no longer suitable for strain measurement, so we use another sample from the same batch as in situ domain observation experiment. The experimental result is shown in Fig. 2. The shape change versus electric-field curve and the corresponding P E hysteresis loop curve show that with the electric field increasing to maximum, polarization increases to maximum due to domain switching (recall Fig. 1); and simultaneously shape change also increases to the maximum value of 0.4%, which is almost half of the theoretical value. When the field is removed, averaged polarization becomes zero, which is consistent with the restoration of original domain state as shown in Fig. 1. At the same time, shape change also recovers to zero. The same is true for reverse electric field. Thus, we observed an interesting electro-shape-memory effect corresponding to a double hysteresis loop, which also corresponds to the reversible domain switching process shown in Fig. 1. We noticed there is some difference in the coercive field E C between the samples used in the shape change measurement and that in the domain observation. This is attributed to the different Mn concentration of different samples although they are from the same batch of single crystals grown in the same crucible. Another point is that, for domain observation in Fig. 1, we apply a large electric field to observe a fully single domain state; this makes a saturated polarization; while, for strain measurement in Fig. 2, not so large electric field is applied to avoid destroying the aginginduced electro-shape-memory effect; this leads to unsaturated polarization. Up to now, our experiments have correlated the mesoscopic domain switching process with macroscopic properties like hysteresis loop and shape change. Through the experimental results,

356 L.X. Zhang, X. Ren / Materials Science and Engineering A 438 440 (2006) 354 359 Fig. 1. Direct evidence for reversible domain switching process and its relation to double hysteresis loop in aged Mn BaTiO 3 single crystal sample. Beside each micrograph is the schematic configuration of defect symmetry and P S state (solid lines inside the schematic graph A, C, and E shows the 90 ferroelectric domain wall, while dotted lines inside the schematic graph B and D shows the domain wall of different P D configuration formed by defects). See Fig. 3 for the definition of rectangles and arrows. we confirmed the reversibility of domain switching process in aged sample. Moreover, this reversible domain switching is responsible for the macroscopic electro-shape-memory effect. This gives a direct mesoscopic evidence for our reversible domain-switching mechanism mentioned in introduction. On the other hand, we know that reversible domain switching and electro-shape-memory effect all stems from aging in ferroelectric state (without aging, there is no such effect). As aging also causes a symmetry-conforming configuration of the defects (as will be discussed in Fig. 3); this suggests the central role of microscopic defect symmetry in determining mesoscopic domain switching behavior and macroscopic polarization and shape change behavior. This will be discussed in detail in the following section. 4. Discussion 4.1. Defect symmetry principle Here we first give a brief introduction to the defect symmetry principle as it is the basis for understanding the origin of the aging-induced reversible domain switching and correspond-

L.X. Zhang, X. Ren / Materials Science and Engineering A 438 440 (2006) 354 359 357 Fig. 2. Large electro-shape-memory effect and corresponding double P E hysteresis loop in the aged Mn BaTiO 3 sample. The black, dark grey and grey curves show different frequency, 1.0, 0.1, and 0.05 Hz, of electric field cycling. ing electro-shape-memory effect. In the following, we explain this principle in present investigated ferroelectric BaTiO 3 crystal (which has a simple perovskite structure). When Mn 3+ ion is doped into BaTiO 3 crystal, it will occupy Ti 4+ site due to the similarity in ionic radius. To maintain charge neutrality, O 2 vacancies are necessarily produced at O 2 sites [10]. Then we consider the statistical distribution of O 2 vacancies in the host BaTiO 3 crystal. Usually they are thought to be distributed randomly; however, if we are looking at the local (short-range) environment of a given Mn 3+ dopant, the shortrange distribution of O 2 vacancies around it may show certain statistical symmetry. This is what we call defect symmetry, which will conform to crystal symmetry when in equilibrium. As BaTiO 3 has different crystal symmetry at the paraelectric state and ferroelectric state, we will explain this principle in the two states, respectively. Above phase transition point T c, the crystal is in paraelectric state and has a cubic symmetry (see Fig. 3(a)). We now define the conditional probability of finding an O 2 vacancy at one of the six O 2 sites next to a given Mn 3+ dopant as defect probability Pi V (i = 1 6). In the cubic paraelectric structure, the six O 2 sites are equivalent for the Mn 3+ dopant at site-0, thus defect probability is P1 V = PV 2 = PV 3 = PV 4 = PV 5 = PV 6. This means that defects have a cubic symmetry when in equilibrium, which conforms to the cubic crystal symmetry. To simplify the description in the following, we use small and large square represents the cubic defect and crystal symmetry, respectively, as shown next to the structure figure. When cooled down below T c, the crystal changes to the ferroelectric state with a spontaneous polarization P S along 001 directions due to the relative displacement of positive and negative ions (see Fig. 3(b)). Crystal symmetry now is tetragonal. However, point defects cannot migrate during this process because the paraelectric ferroelectric phase transition is diffusionless. Consequently, the defect distribution is the same (cubic symmetry) as in the paraelectric state. The tetragonal crystal symmetry is represented by the large rectangle, while P S is represented by the thick arrow. However, such a state (Fig. 3(b)) is actually not stable. The reason is that in a tetragonal structure, the six neighboring O 2 sites are not equivalent for the Mn 3+ at site-0: site-2, 3, 4, 5 are equivalent, but site-1 is closest and site-6 is farthest in distance. It is natural that the closer site should have larger defect probability due to the Coulomb attractive force between effectively negative Mn 3+ dopant and effectively positive O 2 vacancy. Therefore, the defect probability should change to: P1 V >PV 2 = PV 3 = PV 4 = PV 5 >PV 6, showing tetragonal symmetry when in equilibrium. However, such a change involves the short-range migration of O 2 vacancy (for example, from O 2 site-3 to site-1). Consequently, it requires some time to complete; this is the microscopic origin of the aging of ferroelectric phase (i.e., phenomena of the change in properties with time). After the ferroelectric phase is aged for long time, defect symmetry follows the tetragonal crystal symmetry. The non-centric distribution of charged defects (Mn 3+ dopant and O 2 vacancy) forms defect polarization P D along the direction of P S (Fig. 3(c)). The tetragonal defect symmetry is represented by small rectangle, while P D is represented by thin arrow. 4.2. Relation among microscopic defect symmetry, mesoscopic domain switching behavior, and macroscopic ferroelectric and shape change properties According to the above explanation, we know that aging is the key point to establish a symmetry-conforming defect configuration. According to this principle, for a ferroelectric crystal with multi-domain patterns, after aging at ferroelectric state, defect symmetry follows tetragonal crystal symmetry and defect polarization P D aligns along the direction of P S within each domain [4,8] (see the schematic graph in Fig. 1). When electric field is applied, P S is switched to the direction of electric field and domain switching occurs. This leads to an increase of polarization and shape change. At maximum electric field, a single domain state is obtained. Simultaneously, polarization and shape change reach maximum value. However, defect symmetry and the associated P D cannot be rotated during such a diffusionless domain-switching process. This unswitched defect symmetry and associated P D consequently provide a restoring force favoring a reversible domain switching. Thus, after removing the external field, the switched domain restores the original orientation by defect symmetry and P D. Therefore, the most important consequence of the defect symmetry principle is that defect symmetry and P D can memorize the original crystal domain patterns

358 L.X. Zhang, X. Ren / Materials Science and Engineering A 438 440 (2006) 354 359 Fig. 3. Defect symmetry and related crystal symmetry in perovskite BaTiO 3 structure doped with Mn 3+ ions at Ti 4+ sites (a) equilibrium paraelectric state, (b) immediately after para ferro phase transition at T c, and (c) equilibrium ferroelectric state (after aging at T < T c ). Pi V refers to the conditional probability of finding an O 2 vacancy at O 2 site-i (i = 1 6) next to a given Mn 3+ ion at site-0. Large square/rectangle represents crystal symmetry, while small square/rectangle represents defect symmetry. Thick arrow refers to spontaneous polarization P S, and thin arrow refers to defect polarization P D. T c is the paraelectric ferroelectric phase transition temperature. after aging, and do not change during subsequent domain switching. This assures that P S can be switched back to the original orientation so that defect orientation follows crystal orientation and P D aligns along P S in every domain. As the result, averaged polarization and shape change becomes zero (Fig. 2). Thus, we observed the reversible domain switching during electric field cycling (Fig. 1) and a corresponding double P E hysteresis loop as well as an electro-shape-memory effect (Fig. 2) in aged sample. 5. Conclusion We performed in situ study of domain pattern evolution during electric field cycling for Mn-doped BaTiO 3 single crystals. As a conclusion, we found that aged sample shows a reversible domain-switching process, which corresponds well to the macroscopic double P E hysteresis loop and electro-shapememory effect. This provides a direct mesoscopic evidence for the defect-mediated reversible domain-switching mechanism of the giant electro-shape-memory effect. Moreover, it was explained that the microscopic origin for all these effects is symmetry-conforming property of defects, which is completed during aging by defect migration. Acknowledgements The authors graciously acknowledge the support of the Sakigake-21 of JST and a special fund for Cheungkong professorship, National Science Foundation of China, as well as National Basic Research Program of China under Grant No. 2004CB619303. The authors thank K. Otsuka, W. Chen, K. Nakamura, S. Sarkar, W.H. Wang, and G.L. Fan for helpful discussions. References [1] F. Jona, G. Shirane, Ferroelectric Crystals, Macmillan, New York, 1962. [2] K. Uchino, Ferroelectric Device, Marcel Dekker, New York, 2000.

L.X. Zhang, X. Ren / Materials Science and Engineering A 438 440 (2006) 354 359 359 [3] S.E. Park, T.R. Shrout, J. Appl. Phys. 82 (1997) 1804 1811. [4] X. Ren, Nat. Mater. 3 (2004) 91 94. [5] X. Ren, K. Otsuka, Nature (London) 389 (1997) 579 582. [6] X. Ren, K. Otsuka, Phys. Rev. Lett. 85 (2000) 1016 1019. [7] X. Ren, K. Otsuka, MRS Bull. 27 (2002) 115 120. [8] L.X. Zhang, W. Chen, X. Ren, Appl. Phys. Lett. 85 (2004) 5658 5660. [9] P.V. Lambeck, G.H. Jonker, J. Phy. Chem. Solids 47 (1986) 453 461. [10] D.M. Smyth, The Defect Chemistry of Metal Oxides, Oxford University Press, New York, 2000.