- Email: [email protected]
Kinetics of nanodomain growth in ferroelectric artificial superlattices
Available online at www.sciencedirect.com
Scripta Materialia 69 (2013) 501–504 www.elsevier.com/locate/scriptamat
Kinetics of nanodomain growth in ferroelectric artificial superlattices Taekjib Choi,a,b Bae Ho Park,c Hyunjung Shind and Jaichan Leea,⇑ a
School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon 440-746, Republic of Korea b Hybrid Materials Research Center and Department of Nanotechnology and Advanced Materials Engineering, Sejong University, Seoul 143-747, Republic of Korea c Department of Physics, Konkuk University, Seoul, Republic of Korea d Department of Energy Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea Received 21 March 2013; revised 21 May 2013; accepted 26 May 2013 Available online 4 June 2013
We report the kinetics of nanoscale domain growth in ferroelectric PbZrO3/PbTiO3 artificial superlattices. Ferroelectric superlattices behave as single ferroelectric materials with only 180° domain structure (monodomain). Domain size increases linearly with the pulse voltage, and is linearly dependent on the logarithmic value of the pulse duration. It was found that ferroelectric superlattices have a relatively high activation field for domain growth and enhanced domain stability, resulting in a long-term retention behavior of more than one month. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Ferroelectric materials; Domain kinetics; Piezoresponse force microscopy; Artificial superlattices
Significant interest in the artificial superlattice of perovskite oxide has developed over the last decade as superlattices have the potential to create novel materials with enhanced physical properties or new functionality [1–6]. The formation of an artificial superlattice for ferroelectricity offers opportunities to manipulate ferroelectric properties and expand the functionality of these materials [4–6]. Recent experimental and theoretical studies have revealed that ferroelectric superlattices have intriguing structural and electronic properties that are closely related to their domain structures [6–10]. The domain morphologies of ferroelectric superlattices have features that depend on the stacking periods [8,9]. For example, 180° periodic nanostripe domain structures were observed at relatively large periods, while monodomain-like structures appeared at ultrashort-stacking periods. Such domain morphologies are critical to domain engineering for practical applications. In an earlier study, we reported that the ultrashort-period ferroelectric superlattice was more advantageous than nanodomain engineering due to its prototype
⇑ Corresponding
author. Tel.: +82 31 290 7397; fax: +82 31 290 7410; e-mail: [email protected]
supercell with only two polarities along up and down directions and the resulting domain structures were comprised of only 180° domain boundaries [10]. Even though the kinetics of domain wall motion (domain growth) in nanodomain engineering of ferroelectrics has practical implications in scanning probe microscopy (SPM)-based applications, such as high-density data storage, few experimental studies of the domain dynamics of superlattices have been performed. In terms of electronic device miniaturization and the development of nanotechnology, nanoscale characterization of local ferroelectric properties has been a key issue. Various characterizations of ferroelectric materials at the nanoscale, such as visualization and manipulation of the domain structure, domain reversal and domain growth, have been provided by piezoresponse force microscopy (PFM) [11,12]. Recently, in situ transmission electron microscopy and time-resolved X-ray microdiffraction studies have provided direct and accurate information on the dynamics of ferroelectric switching, which complements PFM measurements [13,14]. Following nanodomain engineering via superlattices, i.e. the formation of nanoscale domains, the kinetics of the domain growth in nanoscale superlattices have been
1359-6462/$ - see front matter Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.scriptamat.2013.05.037
502
T. Choi et al. / Scripta Materialia 69 (2013) 501–504
investigated with an ultrashort stacking period, e.g. a two-unit-cell PbZrO3/two-unit-cell PbTiO3 (PZO2/ PTO2), by PFM. Nanoscale domain formation as a function of the applied pulse voltage and time for the domain growth (domain wall motion) was specifically investigated. The domain wall motion is reported in terms of the electric field dependence of the domain wall velocity and inhomogeneous electric field distribution in the sample beneath the PFM tip. The domain wall velocity of the ferroelectric superlattice is proportional to exp ( d/E) (Merz’s law), where d is the activation field and E is the applied electric field [15]. The superlattices showed a relatively high activation field for domain growth and enhanced domain stability, resulting in longterm retention greater than one month (i.e. 42 days). Artificial PZO/PTO superlattices were fabricated on a (La0.5,Sr0.5)CoO3(LSCO)/MgO (100) substrate by multi-target pulsed laser deposition (PLD, k = 248 nm, KrF excimer laser). Epitaxial PZO and PTO layers with the same number of unit cells were alternately deposited at 500 °C with a 100 mtorr oxygen ambient after epitaxial growth of LSCO to form a bottom electrode on the MgO substrate. The stacking period was varied from one unit cell of the PTO layer and one unit cell of the PZO layer (i.e. PZO1/PTO1) to PZO100/PTO100. In our previous work, maximum polarization of the superlattice was obtained at a stacking period of PZO2/PTO2. Additionally, the formation of the superlattice structure was also confirmed by X-ray diffraction analysis with the appearance of satellite peaks near the main diffraction peak [16]. Only a single peak of the (0 0 1) domain structure was observed in reciprocal space mapping (data not shown), indicating that PZO/PTO superlattices have a perfect single c-axis domain and tetragonal structure. For the microscopic aspects of domain growth and switching of 50 nm thick PZO2/PTO2 superlattices, the PFM domain imaging and study of domain growth were performed with a PFM system consisting of a commercial scanning probe microscope (Seiko, SPA 300), a low-pass filter and a lock-in amplifier (Stanford Research System, SR830). A key parameter for estimating and manipulating ferroelectric domains is the film surface roughness. Figure 1a shows the topography of the 50 nm thick PZO2/PTO2 superlattice. The root-mean-square (rms) roughness over a 4 lm 4 lm surface was 0.4 nm, indicating that the films were atomically smooth. A highly smooth and uniform surface is advantageous for effective writing or imaging of ferroelectric domains in SPM-based applications. The ferroelectric domain structure and switching behavior in artificial PZO/PTO superlattices were elucidated by forming the polarized domains using PFM. Figure 1b shows the piezoresponse images of the polarized domain patterns written by alternately applying +10 V and 10 V DC to the LSCO bottom electrode over four successive scans. In sequence, the written areas were 4, 3, 2 and 1 lm2. The tip was electrically grounded while scanning to enable writing. In Figure 1b, the dark region represents the positively polarized area (positive domain with the polarization vector oriented upward) written by applying +10 V, while the negative domain with polarization oriented toward the bottom electrode written by 10 V appears as a bright region. As
Figure 1. (a) Topographic image of a 50 nm thick PZO2/PTO2 superlattice showing 0.42 nm rms roughness over a 4 lm 4 lm area. Dust particles are at the left of the image (dotted circle). (b) The piezoresponse and (c) amplitude images of the patterned domain with various polarized areas. (d) Cross-sectional piezoresponse profile of the polarized domain across the A–B line in (b).
observed in the well-defined domain image, each region produces a uniform piezoresponse, indicating a homogeneous polarization state. The image contrast was also clearly observed, implying ferroelectric polarization reversal. Figure 1d shows a cross-section profile of the piezoresponse domain image along the A–B line, as indicated in Figure 1b. The piezoresponse signal of the positive domain (polarized by +10 V) was equal to that of the negative domain ( 10 V), suggesting a single polarization with opposite polarity. The abrupt transition from a positive domain to a negative domain was also observed in Figure 1d. Furthermore, piezoresponse images revealed the absence of the a-domain and 90° domain boundary. Note that the ultrashort stacking-period superlattices have only a single c-axis orientation, e.g. a monodomain-like domain, even throughout the 200 nm sample thickness. Based on these results, PZO/PTO superlattices with ultrashort stacking periods (i.e. PZO1/PTO1 and PZO2/PTO2) may act as a single-component system with a single ferroelectric domain structure, even though the superlattice consists of alternating ferroelectric and antiferroelectric layers. Recent first-principles studies on ultrashort-period KTaO3/KNbO3 showed that the polarization in the B-site modulation superlattice is more sensitive to the variations in composition and atomic configuration than that in the A-site modulation superlattice, such as the BaTiO3/SrTiO3 superlattice, since the B-site modulation breaks the translational symmetry of the oxygen octahedron [17]. Sepliarsky et al. also predicted the anisotropy in correlation of each PZ (polarization along the modulation direction) and PX (the in-plane component of the polarization) through the B-site modulation superlattice via molecular-dynamics simulation and revealed that the superlattice acts as a single artificial ferroelectric structure at short modulation lengths [18]. It is necessary to consider the amplitude images in detail in order to understand the polarized domain struc-
T. Choi et al. / Scripta Materialia 69 (2013) 501–504
503
ture. Since the tip vibration or piezoresponse signal is an integration of the domain displacement along the thickness, the amplitude of the domain patterns depends on the portion of opposite domains along the thickness direction. Figure 1c illustrates the amplitude image of the domain patterns shown in Figure 1b. The amplitude reaches its minimum at the boundary, appearing as a dark line in Figure 1c when the two opposite domains contribute equally to the tip vibration signal, whereas it reaches a maximum when one of the opposite domains fully penetrates the thickness direction. Moreover, the domain boundaries obtained from both amplitude and phase are the same scale. Therefore, the PZO2/PTO2 superlattice has 180° domain boundaries extended to the bottom electrode and exhibits domain switching between the two opposite polarities. The piezoresponse image of nanosized domains on the 50 nm thick PZO2/PTO2 superlattice is shown in the inset of Figure 2a. The domains were formed by pulse voltages of 6 V (bottom row) and 8 V (upper row) and pulse durations of 10, 20, 100, and 500 ms (from left to right column) on the reverse poled area with a +10 V bias. The superlattice exhibited well-defined circular domains. The size of the polarized domains depended strongly on various pulse parameters. The domain radius was estimated by averaging the full width at half maximum value of the line profile across the vertical and horizontal axes. Figure 2a shows the dependence of the domain radius on the pulse voltage for 10 and 20 ms, respectively. The domain radius increases linearly with increasing pulse amplitude for various durations. To elucidate the change of domain size with pulse voltage, inhomogeneous distribution of the tip-generated field within the sample was considered using the sphere model [19]. We assumed that below the sphere tip (radius, R, 30 nm), the local electric field inside the film, E, was VR/d, where V and d (50 nm) are the applied voltage and film thickness, respectively. As the pulse voltage increases, the electric field concentrates beneath the tip and spreads out inside the film along the lateral direction, resulting in an increase in domain size. When the domain switching under the tip is considered, the switching begins with nucleation of a
new domain beneath the tip if the pulse voltage (or tip-generated field) exceeds a threshold value (or coercive field), where the electric field has an intensity maximum. As a result, under an electric field greater than the threshold value, immediate nucleation and fast forward growth through the entire thickness occurs as the superlattices have a domain structure with 180° boundaries penetrating the entire film. Therefore, the domain size is primarily determined by the electric field (or applied voltage). Subsequently, the newly formed domain then expands by a sidewise motion of the domain walls if additional time of the applied pulse is allowed. As a result, the domain growth and size is also kinetically governed by the sidewise motion of the domain wall with applied pulse duration in respect to inhomogeneous field distribution inside the film. A detailed study on domain wall motion was performed by measuring the domain size as a function of the pulse duration. Figure 2b shows the dependence of the domain radius on the pulse duration and the PFM phase image of the three bit arrays written by the 10 V pulse with various durations. The domain radius is linearly proportional to the logarithmic value of the pulse duration. Specifically, the domain size did not saturate with pulse amplitude and duration. This dependence of domain size on the pulse parameter is different from the theory of the equilibrium domain formation reported by Molotskii [20]. This behavior has been observed by Rodriguez et al. in experiments on domain growth kinetics in LiNbO3 single crystals that are also known to have only a 180° domain structure [21]. It was suggested that the domain size is kinetically limited to the pulse parameter and the domain kinetics was described as an activation process via nucleation at the existing 180° domain wall. Based on the time dependence of the domain radius (Fig. 2b), the domain wall velocity, v(r) = dr/dt, was extracted for the variation of domain radius of two subsequent pulse durations. Figure 3a shows the domain wall velocity as a function of the domain radius and inverse applied electric field, 1/E(r), to the domain wall (inset of Fig. 3a). The corresponding E(r) = VR/rd, where r = [r(t1) + r(t2)]/2, was calculated using a model from Tybell et al. [22]
Figure 2. (a) Dependence of the domain radius on the pulse voltage for different pulse widths. The inset is a piezoresponse image of nanoscale domains written by applying various pulse parameters, i.e. 6 and 8 V with pulse durations of 10, 20, 100 and 500 ms. (b) Dependence of the domain radius on the pulse time at a pulse voltage of 10 V. The domain size with pulse time was estimated from the PFM phase image (inset) where the same three bit arrays were written at 10 V with various pulse times on an area prepolarized at +10 V.
Figure 3. (a) Domain wall velocity vs. domain radius calculated using data from Figure 2b. The inset is the domain wall velocity as a function of the inverse applied electric field. The trends follow an activated process (Merz’s law) with an activation field of 2.7 MV cm 1. (b) The retention behavior of the 50 nm thick PZO2/PTO2 superlattice. PFM image of domain patterns on the superlattice polarized at ±10 V: (upper) immediately after poling; and (down) after 42 days.
504
T. Choi et al. / Scripta Materialia 69 (2013) 501–504
The data fits well with the exponential field dependence of the domain wall velocity as v exp [ d/E(r)], where the activation energy (d) was 2.7 MV cm 1. Therefore, it is suggested that the domain wall motion in the superlattice is an activation process following Merz’s law. As previously mentioned, the domain morphologies in ferroelectric superlattices are strongly governed by stacking periods, which could affect the dynamic characteristics of the domain growth. Lisenkov et al. predicted the dynamics of domain evolution from nanostripes to monodomains in ferroelectric BaTiO3/SrTiO3 superlattices [8]. For a nanostripe domain structure in ferroelectric superlattices with relatively large periods, the domain wall motion deviated from Merz’s law. In contrast, our ultrashort stacking-period PZO/PTO superlattice typically obeyed Merz’s law, which is attributed to monodomain-like structures [10]. Moreover, the activation energy of superlattices is greater than those of films, ferroelectrics, e.g. for Pb(Zr0.2Ti0.8)O3 1.0 MV cm 1 from the experimental results of Tybell et al. [22] and for PbTiO3, 0.8 MV cm 1 from the theoretical results of Shin et al. [23]. Since the symmetry breaking of the ferroelectric double-well potential via an electric field on the ferroelectric domain wall leads to domain wall motion, the high activation energy implies a high energy barrier in the double-well potential of ferroelectric superlattices. The larger energy barrier for domain wall motion can be associated with unique structural properties of ferroelectric superlattices involving epitaxially coherent strain and interactions of ferroelectric polarizations within artificial supercells. A first-principles study by Beckman et al. revealed that the barrier to domain wall motion in the PZO/PTO superlattice is greater than those in PbTiO3 and PZT solid solutions [24]. The TiO2 layer contributes more to the net polarization than the ZrO2 layer due to the small spacing of TiO2 planes, i.e. the Ti–O bond is shorter, whereas in the PZT solid solution, Ti–O and Zr–O bonds contribute equally to polarizations. The larger energy barrier to the domain wall motion can lead to strong stability of the ferroelectric domains in superlattice thin films. Furthermore, the domain stability is especially important for the retention characteristics of the superlattice. Figure 3b illustrates a PFM image of domain patterns on the superlattice at an interval (i.e. 42 days) after poling, which clearly indicates that significant retention loss does not occur. The long-term retention behavior can be explained as the highly anisotropic supercell (like tetragonal structure) superlattices have 180° domain switching that penetrates the entire thickness and has strong domain stability. In summary, the domain structure and growth behavior of the PZO/PTO superlattice at the nanoscale have been investigated by PFM. The superlattice showed a single domain structure throughout the entire superlattice thickness, even though the superlattice consists of alternating ferroelectric and antiferroelectric layers. In the superlattice system, the domain size is kinetically limited to pulse amplitude and duration, as found in various other ferroelectrics. From these results, the key parameters are identified by controlling the domain size
with pulse voltage and time for high-density data-storage applications. The domain growth in superlattices is an activation process following Merz’s law with relatively large activation field. The larger activation barrier for domain wall motion in the superlattice can lead to higher domain stability than found in solid solutions. This work was supported by the Basic Research Program (2009-0092809 and 2011-0025607) through the National Research Foundation of Korea. [1] A. Ohtomo, D.A. Muller, J.L. Grazul, H.Y. Hwang, Nature 419 (2002) 378–380. [2] K. Ueda, H. Tabata, T. Kawai, Science 280 (1998) 1064– 1066. [3] L. Kim, D. Jung, J. Kim, Y.S. Kim, J. Lee, Appl. Phys. Lett. 82 (2003) 2118–2120. [4] H.N. Lee, H.M. Christen, M.F. Chisholom, C.M. Rouleau, D.H. Lowndes, Nature 433 (2005) 395–399. [5] T. Choi, J. Lee, J. Kor. Phys. Soc. 46 (2005) 116–119. [6] E. Bousquet, M. Dawber, N. Stucki, C. Lichtensteiger, P. Hermet, S. Garigilio, J.-M. Triscone, P. Ghosez, Nature 452 (2008) 732–736. [7] K. Rogdakis, J.W. Seo, Z. Viskadourakis, Y. Wang, L.F.N. Ah Qune, E. Choi, J.D. Burton, E.Y. Tsymbal, J. Lee, C. Panagopoulos, Nat. Commun. 3 (2012) 2061. [8] S. Lisenkov, I. Ponomareva, L. Bellaiche, Phys. Rev. B 79 (2009) 024101. [9] J.Y. Jo, P. Chen, R.J. Sichel, S.J. Callori, J. Sinsheimer, E.M. Dufresne, M. Dawber, P.G. Evans, Phys. Rev. Lett. 107 (2011) 055501. [10] T. Choi, B.H. Park, H. Shin, J. Lee, Appl. Phys. Lett. 100 (2012) 222906. [11] V. Nagarajan, A. Roytburd, A. Stanishevsky, S. Prasertchoung, T. Zhao, L. Chen, J. Melngailis, O. Auciello, R. Ramesh, Nat. Mater. 2 (2002) 43–47. [12] A. Gruverman, A. Kholkin, Rep. Prog. Phys. 69 (2006) 2443–2474. [13] H. Chang, S.V. Kalinin, S. Yang, P. Yu, S. Bhattacharya, P.P. Wu, N. Balke, S. Jesse, L.Q. Chen, R. Ramesh, S.J. Pennycook, A.Y. Borisevich, J. Appl. Phys. 110 (2011) 052014. [14] J.Y. Jo, P. Chen, R.J. Sichel, S.L. Callori, J. Sinsheimer, E.M. Dufresne, M. Dawber, P.G. Evans, Phys. Rev. Lett. 107 (2011) 055501. [15] W.J. Merz, Phys. Rev. 95 (1954) 690–698. [16] T. Choi, J. Lee, Thin Solid Films 475 (2005) 283–286. [17] S. Hao, G. Zhou, X. Wang, J. Wu, W. Duan, B.-L. Gu, Appl. Phys. Lett. 86 (2005) 232903. [18] M. Sepliarsky, S.R. Phillpot, D. Wolf, M.G. Stachiotti, R.L. Migoni, Phys. Rev. B 64 (2001) 060101. [19] J. Woo, S. Hong, N. Setter, H. Shin, J.-U. Jeon, Y.E. Pak, K. No, J. Vac. Sci. Technol. B 19 (2001) 818–824. [20] M. Molotskii, J. Appl. Phys. 97 (2005) 014109. [21] B.J. Rodriguez, R.J. Nemanich, A. Kingon, A. Gruverman, S.V. Kalinin, K. Terabe, X.Y. Liu, K. Kitamura, Appl. Phys. Lett. 86 (2005) 012906. [22] T. Tybell, P. Paruch, T. Giamarchi, J.-M. Triscone, Phys. Rev. Lett. 89 (2002) 097601. [23] Y.-H. Shin, I. Grinberg, I. Chen, A.M. Rappe, Nature 449 (2007) 881–884. [24] S.P. Beckman, X. Wang, K.M. Rabe, D. Vanderbilt, Phys. Rev. B 79 (2009) 144124.
Scripta Materialia 69 (2013) 501–504 www.elsevier.com/locate/scriptamat
Kinetics of nanodomain growth in ferroelectric artificial superlattices Taekjib Choi,a,b Bae Ho Park,c Hyunjung Shind and Jaichan Leea,⇑ a
School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon 440-746, Republic of Korea b Hybrid Materials Research Center and Department of Nanotechnology and Advanced Materials Engineering, Sejong University, Seoul 143-747, Republic of Korea c Department of Physics, Konkuk University, Seoul, Republic of Korea d Department of Energy Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea Received 21 March 2013; revised 21 May 2013; accepted 26 May 2013 Available online 4 June 2013
We report the kinetics of nanoscale domain growth in ferroelectric PbZrO3/PbTiO3 artificial superlattices. Ferroelectric superlattices behave as single ferroelectric materials with only 180° domain structure (monodomain). Domain size increases linearly with the pulse voltage, and is linearly dependent on the logarithmic value of the pulse duration. It was found that ferroelectric superlattices have a relatively high activation field for domain growth and enhanced domain stability, resulting in a long-term retention behavior of more than one month. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Ferroelectric materials; Domain kinetics; Piezoresponse force microscopy; Artificial superlattices
Significant interest in the artificial superlattice of perovskite oxide has developed over the last decade as superlattices have the potential to create novel materials with enhanced physical properties or new functionality [1–6]. The formation of an artificial superlattice for ferroelectricity offers opportunities to manipulate ferroelectric properties and expand the functionality of these materials [4–6]. Recent experimental and theoretical studies have revealed that ferroelectric superlattices have intriguing structural and electronic properties that are closely related to their domain structures [6–10]. The domain morphologies of ferroelectric superlattices have features that depend on the stacking periods [8,9]. For example, 180° periodic nanostripe domain structures were observed at relatively large periods, while monodomain-like structures appeared at ultrashort-stacking periods. Such domain morphologies are critical to domain engineering for practical applications. In an earlier study, we reported that the ultrashort-period ferroelectric superlattice was more advantageous than nanodomain engineering due to its prototype
⇑ Corresponding
author. Tel.: +82 31 290 7397; fax: +82 31 290 7410; e-mail: [email protected]
supercell with only two polarities along up and down directions and the resulting domain structures were comprised of only 180° domain boundaries [10]. Even though the kinetics of domain wall motion (domain growth) in nanodomain engineering of ferroelectrics has practical implications in scanning probe microscopy (SPM)-based applications, such as high-density data storage, few experimental studies of the domain dynamics of superlattices have been performed. In terms of electronic device miniaturization and the development of nanotechnology, nanoscale characterization of local ferroelectric properties has been a key issue. Various characterizations of ferroelectric materials at the nanoscale, such as visualization and manipulation of the domain structure, domain reversal and domain growth, have been provided by piezoresponse force microscopy (PFM) [11,12]. Recently, in situ transmission electron microscopy and time-resolved X-ray microdiffraction studies have provided direct and accurate information on the dynamics of ferroelectric switching, which complements PFM measurements [13,14]. Following nanodomain engineering via superlattices, i.e. the formation of nanoscale domains, the kinetics of the domain growth in nanoscale superlattices have been
1359-6462/$ - see front matter Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.scriptamat.2013.05.037
502
T. Choi et al. / Scripta Materialia 69 (2013) 501–504
investigated with an ultrashort stacking period, e.g. a two-unit-cell PbZrO3/two-unit-cell PbTiO3 (PZO2/ PTO2), by PFM. Nanoscale domain formation as a function of the applied pulse voltage and time for the domain growth (domain wall motion) was specifically investigated. The domain wall motion is reported in terms of the electric field dependence of the domain wall velocity and inhomogeneous electric field distribution in the sample beneath the PFM tip. The domain wall velocity of the ferroelectric superlattice is proportional to exp ( d/E) (Merz’s law), where d is the activation field and E is the applied electric field [15]. The superlattices showed a relatively high activation field for domain growth and enhanced domain stability, resulting in longterm retention greater than one month (i.e. 42 days). Artificial PZO/PTO superlattices were fabricated on a (La0.5,Sr0.5)CoO3(LSCO)/MgO (100) substrate by multi-target pulsed laser deposition (PLD, k = 248 nm, KrF excimer laser). Epitaxial PZO and PTO layers with the same number of unit cells were alternately deposited at 500 °C with a 100 mtorr oxygen ambient after epitaxial growth of LSCO to form a bottom electrode on the MgO substrate. The stacking period was varied from one unit cell of the PTO layer and one unit cell of the PZO layer (i.e. PZO1/PTO1) to PZO100/PTO100. In our previous work, maximum polarization of the superlattice was obtained at a stacking period of PZO2/PTO2. Additionally, the formation of the superlattice structure was also confirmed by X-ray diffraction analysis with the appearance of satellite peaks near the main diffraction peak [16]. Only a single peak of the (0 0 1) domain structure was observed in reciprocal space mapping (data not shown), indicating that PZO/PTO superlattices have a perfect single c-axis domain and tetragonal structure. For the microscopic aspects of domain growth and switching of 50 nm thick PZO2/PTO2 superlattices, the PFM domain imaging and study of domain growth were performed with a PFM system consisting of a commercial scanning probe microscope (Seiko, SPA 300), a low-pass filter and a lock-in amplifier (Stanford Research System, SR830). A key parameter for estimating and manipulating ferroelectric domains is the film surface roughness. Figure 1a shows the topography of the 50 nm thick PZO2/PTO2 superlattice. The root-mean-square (rms) roughness over a 4 lm 4 lm surface was 0.4 nm, indicating that the films were atomically smooth. A highly smooth and uniform surface is advantageous for effective writing or imaging of ferroelectric domains in SPM-based applications. The ferroelectric domain structure and switching behavior in artificial PZO/PTO superlattices were elucidated by forming the polarized domains using PFM. Figure 1b shows the piezoresponse images of the polarized domain patterns written by alternately applying +10 V and 10 V DC to the LSCO bottom electrode over four successive scans. In sequence, the written areas were 4, 3, 2 and 1 lm2. The tip was electrically grounded while scanning to enable writing. In Figure 1b, the dark region represents the positively polarized area (positive domain with the polarization vector oriented upward) written by applying +10 V, while the negative domain with polarization oriented toward the bottom electrode written by 10 V appears as a bright region. As
Figure 1. (a) Topographic image of a 50 nm thick PZO2/PTO2 superlattice showing 0.42 nm rms roughness over a 4 lm 4 lm area. Dust particles are at the left of the image (dotted circle). (b) The piezoresponse and (c) amplitude images of the patterned domain with various polarized areas. (d) Cross-sectional piezoresponse profile of the polarized domain across the A–B line in (b).
observed in the well-defined domain image, each region produces a uniform piezoresponse, indicating a homogeneous polarization state. The image contrast was also clearly observed, implying ferroelectric polarization reversal. Figure 1d shows a cross-section profile of the piezoresponse domain image along the A–B line, as indicated in Figure 1b. The piezoresponse signal of the positive domain (polarized by +10 V) was equal to that of the negative domain ( 10 V), suggesting a single polarization with opposite polarity. The abrupt transition from a positive domain to a negative domain was also observed in Figure 1d. Furthermore, piezoresponse images revealed the absence of the a-domain and 90° domain boundary. Note that the ultrashort stacking-period superlattices have only a single c-axis orientation, e.g. a monodomain-like domain, even throughout the 200 nm sample thickness. Based on these results, PZO/PTO superlattices with ultrashort stacking periods (i.e. PZO1/PTO1 and PZO2/PTO2) may act as a single-component system with a single ferroelectric domain structure, even though the superlattice consists of alternating ferroelectric and antiferroelectric layers. Recent first-principles studies on ultrashort-period KTaO3/KNbO3 showed that the polarization in the B-site modulation superlattice is more sensitive to the variations in composition and atomic configuration than that in the A-site modulation superlattice, such as the BaTiO3/SrTiO3 superlattice, since the B-site modulation breaks the translational symmetry of the oxygen octahedron [17]. Sepliarsky et al. also predicted the anisotropy in correlation of each PZ (polarization along the modulation direction) and PX (the in-plane component of the polarization) through the B-site modulation superlattice via molecular-dynamics simulation and revealed that the superlattice acts as a single artificial ferroelectric structure at short modulation lengths [18]. It is necessary to consider the amplitude images in detail in order to understand the polarized domain struc-
T. Choi et al. / Scripta Materialia 69 (2013) 501–504
503
ture. Since the tip vibration or piezoresponse signal is an integration of the domain displacement along the thickness, the amplitude of the domain patterns depends on the portion of opposite domains along the thickness direction. Figure 1c illustrates the amplitude image of the domain patterns shown in Figure 1b. The amplitude reaches its minimum at the boundary, appearing as a dark line in Figure 1c when the two opposite domains contribute equally to the tip vibration signal, whereas it reaches a maximum when one of the opposite domains fully penetrates the thickness direction. Moreover, the domain boundaries obtained from both amplitude and phase are the same scale. Therefore, the PZO2/PTO2 superlattice has 180° domain boundaries extended to the bottom electrode and exhibits domain switching between the two opposite polarities. The piezoresponse image of nanosized domains on the 50 nm thick PZO2/PTO2 superlattice is shown in the inset of Figure 2a. The domains were formed by pulse voltages of 6 V (bottom row) and 8 V (upper row) and pulse durations of 10, 20, 100, and 500 ms (from left to right column) on the reverse poled area with a +10 V bias. The superlattice exhibited well-defined circular domains. The size of the polarized domains depended strongly on various pulse parameters. The domain radius was estimated by averaging the full width at half maximum value of the line profile across the vertical and horizontal axes. Figure 2a shows the dependence of the domain radius on the pulse voltage for 10 and 20 ms, respectively. The domain radius increases linearly with increasing pulse amplitude for various durations. To elucidate the change of domain size with pulse voltage, inhomogeneous distribution of the tip-generated field within the sample was considered using the sphere model [19]. We assumed that below the sphere tip (radius, R, 30 nm), the local electric field inside the film, E, was VR/d, where V and d (50 nm) are the applied voltage and film thickness, respectively. As the pulse voltage increases, the electric field concentrates beneath the tip and spreads out inside the film along the lateral direction, resulting in an increase in domain size. When the domain switching under the tip is considered, the switching begins with nucleation of a
new domain beneath the tip if the pulse voltage (or tip-generated field) exceeds a threshold value (or coercive field), where the electric field has an intensity maximum. As a result, under an electric field greater than the threshold value, immediate nucleation and fast forward growth through the entire thickness occurs as the superlattices have a domain structure with 180° boundaries penetrating the entire film. Therefore, the domain size is primarily determined by the electric field (or applied voltage). Subsequently, the newly formed domain then expands by a sidewise motion of the domain walls if additional time of the applied pulse is allowed. As a result, the domain growth and size is also kinetically governed by the sidewise motion of the domain wall with applied pulse duration in respect to inhomogeneous field distribution inside the film. A detailed study on domain wall motion was performed by measuring the domain size as a function of the pulse duration. Figure 2b shows the dependence of the domain radius on the pulse duration and the PFM phase image of the three bit arrays written by the 10 V pulse with various durations. The domain radius is linearly proportional to the logarithmic value of the pulse duration. Specifically, the domain size did not saturate with pulse amplitude and duration. This dependence of domain size on the pulse parameter is different from the theory of the equilibrium domain formation reported by Molotskii [20]. This behavior has been observed by Rodriguez et al. in experiments on domain growth kinetics in LiNbO3 single crystals that are also known to have only a 180° domain structure [21]. It was suggested that the domain size is kinetically limited to the pulse parameter and the domain kinetics was described as an activation process via nucleation at the existing 180° domain wall. Based on the time dependence of the domain radius (Fig. 2b), the domain wall velocity, v(r) = dr/dt, was extracted for the variation of domain radius of two subsequent pulse durations. Figure 3a shows the domain wall velocity as a function of the domain radius and inverse applied electric field, 1/E(r), to the domain wall (inset of Fig. 3a). The corresponding E(r) = VR/rd, where r = [r(t1) + r(t2)]/2, was calculated using a model from Tybell et al. [22]
Figure 2. (a) Dependence of the domain radius on the pulse voltage for different pulse widths. The inset is a piezoresponse image of nanoscale domains written by applying various pulse parameters, i.e. 6 and 8 V with pulse durations of 10, 20, 100 and 500 ms. (b) Dependence of the domain radius on the pulse time at a pulse voltage of 10 V. The domain size with pulse time was estimated from the PFM phase image (inset) where the same three bit arrays were written at 10 V with various pulse times on an area prepolarized at +10 V.
Figure 3. (a) Domain wall velocity vs. domain radius calculated using data from Figure 2b. The inset is the domain wall velocity as a function of the inverse applied electric field. The trends follow an activated process (Merz’s law) with an activation field of 2.7 MV cm 1. (b) The retention behavior of the 50 nm thick PZO2/PTO2 superlattice. PFM image of domain patterns on the superlattice polarized at ±10 V: (upper) immediately after poling; and (down) after 42 days.
504
T. Choi et al. / Scripta Materialia 69 (2013) 501–504
The data fits well with the exponential field dependence of the domain wall velocity as v exp [ d/E(r)], where the activation energy (d) was 2.7 MV cm 1. Therefore, it is suggested that the domain wall motion in the superlattice is an activation process following Merz’s law. As previously mentioned, the domain morphologies in ferroelectric superlattices are strongly governed by stacking periods, which could affect the dynamic characteristics of the domain growth. Lisenkov et al. predicted the dynamics of domain evolution from nanostripes to monodomains in ferroelectric BaTiO3/SrTiO3 superlattices [8]. For a nanostripe domain structure in ferroelectric superlattices with relatively large periods, the domain wall motion deviated from Merz’s law. In contrast, our ultrashort stacking-period PZO/PTO superlattice typically obeyed Merz’s law, which is attributed to monodomain-like structures [10]. Moreover, the activation energy of superlattices is greater than those of films, ferroelectrics, e.g. for Pb(Zr0.2Ti0.8)O3 1.0 MV cm 1 from the experimental results of Tybell et al. [22] and for PbTiO3, 0.8 MV cm 1 from the theoretical results of Shin et al. [23]. Since the symmetry breaking of the ferroelectric double-well potential via an electric field on the ferroelectric domain wall leads to domain wall motion, the high activation energy implies a high energy barrier in the double-well potential of ferroelectric superlattices. The larger energy barrier for domain wall motion can be associated with unique structural properties of ferroelectric superlattices involving epitaxially coherent strain and interactions of ferroelectric polarizations within artificial supercells. A first-principles study by Beckman et al. revealed that the barrier to domain wall motion in the PZO/PTO superlattice is greater than those in PbTiO3 and PZT solid solutions [24]. The TiO2 layer contributes more to the net polarization than the ZrO2 layer due to the small spacing of TiO2 planes, i.e. the Ti–O bond is shorter, whereas in the PZT solid solution, Ti–O and Zr–O bonds contribute equally to polarizations. The larger energy barrier to the domain wall motion can lead to strong stability of the ferroelectric domains in superlattice thin films. Furthermore, the domain stability is especially important for the retention characteristics of the superlattice. Figure 3b illustrates a PFM image of domain patterns on the superlattice at an interval (i.e. 42 days) after poling, which clearly indicates that significant retention loss does not occur. The long-term retention behavior can be explained as the highly anisotropic supercell (like tetragonal structure) superlattices have 180° domain switching that penetrates the entire thickness and has strong domain stability. In summary, the domain structure and growth behavior of the PZO/PTO superlattice at the nanoscale have been investigated by PFM. The superlattice showed a single domain structure throughout the entire superlattice thickness, even though the superlattice consists of alternating ferroelectric and antiferroelectric layers. In the superlattice system, the domain size is kinetically limited to pulse amplitude and duration, as found in various other ferroelectrics. From these results, the key parameters are identified by controlling the domain size
with pulse voltage and time for high-density data-storage applications. The domain growth in superlattices is an activation process following Merz’s law with relatively large activation field. The larger activation barrier for domain wall motion in the superlattice can lead to higher domain stability than found in solid solutions. This work was supported by the Basic Research Program (2009-0092809 and 2011-0025607) through the National Research Foundation of Korea. [1] A. Ohtomo, D.A. Muller, J.L. Grazul, H.Y. Hwang, Nature 419 (2002) 378–380. [2] K. Ueda, H. Tabata, T. Kawai, Science 280 (1998) 1064– 1066. [3] L. Kim, D. Jung, J. Kim, Y.S. Kim, J. Lee, Appl. Phys. Lett. 82 (2003) 2118–2120. [4] H.N. Lee, H.M. Christen, M.F. Chisholom, C.M. Rouleau, D.H. Lowndes, Nature 433 (2005) 395–399. [5] T. Choi, J. Lee, J. Kor. Phys. Soc. 46 (2005) 116–119. [6] E. Bousquet, M. Dawber, N. Stucki, C. Lichtensteiger, P. Hermet, S. Garigilio, J.-M. Triscone, P. Ghosez, Nature 452 (2008) 732–736. [7] K. Rogdakis, J.W. Seo, Z. Viskadourakis, Y. Wang, L.F.N. Ah Qune, E. Choi, J.D. Burton, E.Y. Tsymbal, J. Lee, C. Panagopoulos, Nat. Commun. 3 (2012) 2061. [8] S. Lisenkov, I. Ponomareva, L. Bellaiche, Phys. Rev. B 79 (2009) 024101. [9] J.Y. Jo, P. Chen, R.J. Sichel, S.J. Callori, J. Sinsheimer, E.M. Dufresne, M. Dawber, P.G. Evans, Phys. Rev. Lett. 107 (2011) 055501. [10] T. Choi, B.H. Park, H. Shin, J. Lee, Appl. Phys. Lett. 100 (2012) 222906. [11] V. Nagarajan, A. Roytburd, A. Stanishevsky, S. Prasertchoung, T. Zhao, L. Chen, J. Melngailis, O. Auciello, R. Ramesh, Nat. Mater. 2 (2002) 43–47. [12] A. Gruverman, A. Kholkin, Rep. Prog. Phys. 69 (2006) 2443–2474. [13] H. Chang, S.V. Kalinin, S. Yang, P. Yu, S. Bhattacharya, P.P. Wu, N. Balke, S. Jesse, L.Q. Chen, R. Ramesh, S.J. Pennycook, A.Y. Borisevich, J. Appl. Phys. 110 (2011) 052014. [14] J.Y. Jo, P. Chen, R.J. Sichel, S.L. Callori, J. Sinsheimer, E.M. Dufresne, M. Dawber, P.G. Evans, Phys. Rev. Lett. 107 (2011) 055501. [15] W.J. Merz, Phys. Rev. 95 (1954) 690–698. [16] T. Choi, J. Lee, Thin Solid Films 475 (2005) 283–286. [17] S. Hao, G. Zhou, X. Wang, J. Wu, W. Duan, B.-L. Gu, Appl. Phys. Lett. 86 (2005) 232903. [18] M. Sepliarsky, S.R. Phillpot, D. Wolf, M.G. Stachiotti, R.L. Migoni, Phys. Rev. B 64 (2001) 060101. [19] J. Woo, S. Hong, N. Setter, H. Shin, J.-U. Jeon, Y.E. Pak, K. No, J. Vac. Sci. Technol. B 19 (2001) 818–824. [20] M. Molotskii, J. Appl. Phys. 97 (2005) 014109. [21] B.J. Rodriguez, R.J. Nemanich, A. Kingon, A. Gruverman, S.V. Kalinin, K. Terabe, X.Y. Liu, K. Kitamura, Appl. Phys. Lett. 86 (2005) 012906. [22] T. Tybell, P. Paruch, T. Giamarchi, J.-M. Triscone, Phys. Rev. Lett. 89 (2002) 097601. [23] Y.-H. Shin, I. Grinberg, I. Chen, A.M. Rappe, Nature 449 (2007) 881–884. [24] S.P. Beckman, X. Wang, K.M. Rabe, D. Vanderbilt, Phys. Rev. B 79 (2009) 144124.