Domain structures and piezoelectric properties of Pb„Zr0.2Ti0.8…O3nanocapacitors
Hee Han (한희�,1,2 Yong Jun Park (박용준�,1 Sunggi Baik (백성기�,1,a� Woo Lee (이우�,3
Marin Alexe,4 Dietrich Hesse,4 and Ulrich Gösele4,b�
1Department of Materials Science and Engineering, Pohang University of Science and Technology(POSTECH), Pohang 790-784, Republic of Korea2National Center for Nanomaterials Technology, Pohang University of Science and Technology(POSTECH), Hyoja-dong, Pohang 790-784, Republic of Korea3Korea Research Institute of Standards and Science (KRISS), Yuseong, Daejon 305-340, Republic of Korea4Max Planck Institute of Microstructure Physics, Weinberg 2, Halle D-06120, Germany
�Received 17 May 2010; accepted 3 July 2010; published online 19 August 2010�
Ferroelectric thin films have attracted considerable inter-est due to their potential applications to capacitors with high-dielectric constant, piezoelectric actuators, and nonvolatileferroelectric random access memories.1–3 The electricalproperties of ferroelectrics are determined mostly by domainstructures and their nucleation as well as their dynamics.4,5
For example, in the case of typical perovskite ferroelectrics,highly c-axis oriented ferroelectric thin films could be appli-cable to memory devices, while highly a-axis oriented onesare desirable for capacitor devices. The nucleation of do-mains and their wall motions can affect the coercive fieldduring domain switching, and the mobility of 90° domains isrelated to the extrinsic contribution to physical properties inferroelectric thin films. In this context, considerable researcheffort has been made to systematically investigate the contri-bution of domain structures to the ferroelectric properties bymeans of x-ray diffraction �XRD� �Refs. 6–8� and piezoelec-tric force microscopy �PFM�.9,10 In the case of ferroelectricPb�Zr,Ti�O3 �PZT� or PbTiO3 �PTO�, nonswitchability of90° domains due either to substrate clamping or to domainpinning has been reported by Xu et al. and Kholkin et al.11,12
On the other hand, it was also observed that 90° domains inan epitaxially grown thin PTO film can be switched under anexternal bias, although the fraction of switchable domains israther small.13,14 Nagarajan et al.15 also reported that 180°domains could be pinned by the depolarizing field in spite ofthe high tetragonality of the unit cell. For practical applica-tions of ferroelectric thin films, extensive investigations ofdomain structures and electrical properties in thin films were
carried out. As a consequence, the dependence of the electri-cal properties on the domain structure in ferroelectric thinfilms has been intensively studied both theoretically andexperimentally.16–18 In contrast to extended ferroelectric thinfilms, studies on the impact of the lateral dimension on thedomain structure and the electrical properties of nanostruc-tured ferroelectrics have been somewhat deficient due tosome technical difficulties involved in fabrication and char-acterization of very tiny ferroelectrics.
Recently, an ever-growing interest in practical applica-tions of ferroelectric thin films to data storage devices withultra-high density has spurred systematic investigations onnanometer-sized discrete features under the aspects offabrication,19,20 characterization,21 and theoreticalmodeling.22 For nanostructured ferroelectrics it has been re-ported that strain relaxation can occur to a large extent,23 andpiezoresponse can significantly be enhanced due either to theswitching or the elimination of 90° domains in patternedferroelectric features.24,25 Lee and Baik8 suggested that thefinal domain structures could be engineered simply by pat-terning the ferroelectric thin films, i.e., controlling the degreeof strain relaxation. Therefore, it is crucial not only to fabri-cate structurally well-defined arrays of ferroelectric nano-structures over an extended area but investigate the evolutionof domain structures, the key factors affecting the final do-main structures, and their implications for the electrical prop-erties in the sub-100 nm size regime.
In this paper, we report the effects of the size of epitaxialferroelectrics on the final domain structure and on c-domainvolume fractions both in Pb�Zr0.2Ti0.8�O3 �PZT� nanocapaci-tors and in thin films. Anodic aluminum oxide �AAO� maskassisted-pulsed laser deposition �PLD� was employed for thefabrication of large-area arrays of PZT nanoislands. PLD hasbeen known to be useful for the fabrication of complex func-
tional oxide thin films of high crystalline quality. By extend-ing the PLD approach, we were able to obtain PZT nanoca-pacitors with a uniform piezoresponse over an extendedarea.26 The domain structure and the evolution of strain re-laxation were characterized by reciprocal space mapping�RSM� using high resolution synchrotron XRD. The effect ofa-domains and the strain relaxation on the electrical proper-ties is investigated by PFM.
II. EXPERIMENT
A. Fabrication of ultrathin AAO masks
Ultrathin AAO masks were prepared according to themethod reported previously.26 In brief, a surface finished alu-minum disk �99.999%, Goodfellow� was anodized at 40 Vusing 0.3 M H2C2O4 �1 °C�. First anodization was con-ducted for 12 h. After that, the resulting porous oxide layerwith disordered pores was removed by immersing the sampleinto an acid mixture �1.8 wt % chromic acid and 6 wt %H3PO4� at 43 °C for 12 h to obtain a textured surface on thealuminum surface. For a 400-nm-thick AAO membrane, thesecond anodization was conducted under the same conditionas the first one for 140 s. After the second anodization, a thinlayer of polystyrene �PS� film was spin-coated on the surfaceof the AAO at 3000 rpm for 1 min using an 1.4 wt %PS/CHCl3 solution. Afterwards, the residual aluminum sub-strate was removed by floating the sample onto the surface ofan acid mixture of CuCl2 and HCl, where the aluminummetal was placed to come in contact with the etchant solu-tion. The barrier layer could be selectively removed by plac-ing the PS/AAO composite film on the surface of a 5 wt %H3PO4 solution at 30 °C for 30 min. The resulting PS/AAOcomposite film was transferred onto a desired substrate, andthen immersed into CHCl3 solution to completely remove thePS. The typical dimension of the resulting ultrathin AAOmask is �2 cm2.
B. Fabrication of PZT nanoislands
PLD of PZT material through masks was carried out at650 °C under high vacuum conditions �5�10−6 torr� usinga KrF excimer laser with a wavelength of 248 nm, an energyfluence of 500 mJ cm−2 and a repetition rate of 5–10 Hz.After PZT deposition, extended arrays of PZT nanoislandswere obtained by mechanically lifting off the stencil mask.From scanning electron microscopy �SEM� analyses, the lat-eral dimension of the resulting PZT nanoislands was esti-mated to be approximately 65 nm. In the present nanofabri-cation method, we were able to control the thicknesses �40,30, and 15 nm� of PZT nanoislands by varying the PZTdeposition time. Epitaxial PZT thin films were also fabri-cated by PLD for the comparative investigation of domainstructures. Pt as a top-electrode was deposited at room tem-perature by e-beam evaporation on PZT for the fabrication ofnanocapacitors with a Pt/PZT/Pt stacking configuration �i.e.,metal/ferroelectric/metal; MFM configuration�.
C. Characterization of PZT nanoislands andnanocapacitors
The epitaxial nature of the as-deposited films was con-firmed by XRD with �–2� scan along the substrate normaldirection and by in-plane 360° �-scans. The evolution of thedomain structures in the ferroelectric nanoislands and thecorresponding volume fractions of a-domains and c-domainswere investigated by high resolution synchrotron XRD em-ploying RSM at the 3C2 beam-line of the Pohang LightSource �PLS�. For the two-dimensional �2D� plane scans inreciprocal space, a Huber four-circle diffractometer with ascintillation detector was used. MgO �lattice parameter a=4.213 Å� and SrTiO3 �STO� �a=3.905 Å� were used as aninternal reference for the fully relaxed state of lattice strain.The RSM of the HL-plane was conducted by keeping theMiller index k and varying h by ��h every increment �l inl. In the same way, HK-plane was obtained by varying k by��k every increment �h in h while the l was kept at Braggangle positions of each �001� and �100�. The electrical prop-erties of the nanostructures were investigated by a PFM �Au-toProbe CP Research� with a lock-in amplifier �SR830, Stan-ford Research� and using a Pt-coated cantilever with a springconstant of 2.8 N/m and a resonance frequency of 75 kHz�ATEC-EFM, Nanosensors�. For the piezoresponse imageand for local domain switching, a voltage of 0.3 Vac with245 kHz and �3 Vdc with a probing voltage of 0.5 Vac and30 kHz were applied to the tip, respectively. Conventionaltransmission electron microscopy �TEM� investigations werecarried out using a Philips CM 20 Twin �Philips, Nether-lands� microscope. Some of the thin samples for TEM inves-tigations were prepared by standard methods of mechanicalpolishing and ion milling. The other samples were preparedby focused ion beam �FIB� milling using a gallium ion beamin an FEI Nova 600 NanoLab system.
III. RESULTS AND DISCUSSION
A. PZT nanoislands and nanocapacitors
Figure 1�a� shows an ultrathin AAO mask ��400 nm�prepared by anodization of an aluminum disk, subsequentlyplaced onto a Pt-covered MgO �001� single crystal substrate.By taking advantage of a good thermal stability of the ce-ramic stencil material �i.e., AAO�, PLD of PZT material wascarried out at a rather high growth temperature �e.g.,650 °C�, to obtain arrays of epitaxial, single crystalline PZTnanoislands with a narrow size distribution �Fig. 1�b��. Sub-sequently, 5-nm-thick Pt top electrodes were sputter-deposited at room temperature to realize a novel capacitorstructure with a Pt/PZT/Pt stacking configuration �i.e.,MFM� as shown in Fig. 1�c� �black arrows�. The top metalelectrode provides a homogeneous distribution of the electricfield through the ferroelectric medium and a good contactbetween the tip and the top electrode during PFM operation,enabling a reliable electrical measurement. After that, ex-tended arrays of structurally well-defined MFM nanocapaci-tors were obtained by simply lifting-off the AAO mask. Wecould not observe any structural changes in the stencil mate-rial during the deposition process. The epitaxial nature of aPZT thin film �solid curve� and the nanoislands �dashed
044102-2 Han et al. J. Appl. Phys. 108, 044102 �2010�
Downloaded 23 Aug 2010 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
curve� were confirmed by x-ray �–2� diffraction �Fig. 1�d��and in-plane 360° �-scans �inset in Fig. 1�d��. The in-plane360° �-scans of PZT thin film is not shown here. From the�–2� scan, it is clear that PZT nanoislands grown on Pt/MgO�001� substrates are highly �001�-orientated, and well-developed four peaks with 90º separations in �-scan indicatethe cube-on-cube epitaxial relationship of PZT�100�//Pt�100�and PZT�001�//Pt�001�.
AAO masks of different pore sizes can be convenientlyprepared by anodic oxidation of aluminum under appropriatepotentiostatic conditions using sulfuric, oxalic, or phosphoricacid as electrolyte. In our approach, the lateral dimensions ofthe PZT nanoislands are determined by the pore size of AAOmask �Figs. 2�a�–2�c��. On the other hand, the thicknesses ofnanoislands are determined by PZT deposition time. Figures2�d�–2�f� show representative cross-section TEM images of65-nm-sized PZT nanoislands grown on �d� and �e� Pt and �f�STO with thicknesses of �d� 40 nm, �e� 30 nm, and �f� 15 nm,respectively.
B. Domain structures of PZT nanoislands
In order to investigate the effect of lateral size and thick-ness of epitaxial PZT structures on the evolution of domainstructure, comparative investigations were carried out by per-forming 2D RSMs for arrays of epitaxial PZT nanoislandsand epitaxial PZT thin films with thicknesses of 40 and 15
nm. RSM measurements were performed on HL-plane to ob-tain information on a-domains and c-domains �i.e., �100� and�001� reflections� in the epitaxial PZT films and nanostruc-tures grown on Pt/MgO �001�. Figure 3 shows RSMs of �a�and �b� epitaxially grown PZT film with thicknesses of �a� 40nm and �b� 15 nm, and �c� and �d� arrays of PZT nanoislandswith thicknesses of �c� 40 nm and �d� 15 nm. In the RSMmeasurement, the MgO �a=4.213 Å� was used as a refer-ence for a fully relaxed state of lattice strain and �002� re-flection of MgO corresponds to 2 of q�00l�, and the relativevalues of q�00l� of PZT nanostructures were recorded, i.e.,q�00l�=4.213 /d �d=lattice distance along substrate normaldirection�. The RSMs clearly manifest that the domain struc-ture varies with thickness and dimension of PZT �i.e., dis-crete nanoislands versus continuous film�. In the case of anepitaxial PZT thin film grown on Pt/MgO�001�, the PZT thinfilm is subject to compressive strain during the ferroelectricphase formation, resulting in a major fraction of c-domains�i.e., polarization along the substrate normal� and a minorfraction of fourfold symmetric a-domains �i.e., polarizationalmost parallel to the film plane�. In the latter, the �00l� axesof the a-domains are tilted from the film plane by an angle,��= �90−2 tan−1�a /c��° �0, forming a /c twin boundaries.8
In the 40-nm-thick film, the reflection of a-domain regionbranches off diagonally from the c-domain region to bothsides �Fig. 3�a��, while the corresponding reflection is absentin the case of 15-nm-thick film �Fig. 3�b��. Because thea-domains form to reduce the residual strain in the filmabove a certain critical thickness, it can be said that the strainis accumulated in the case of the 15-nm-thick film wherea-domains do not form. It was observed that the domainstructures of arrays of discrete PZT nanoislands mainly con-sist of a large fraction of c-domains and a residual fraction ofa-domains with different domain configuration as shown inFigs. 3�c� and 3�d�. In the case of 40-nm-thick PZT nanois-lands, the shape of the reflection of the a-domain region inthe RSM is rather flat and extends from side to side �Fig.3�c��. This change in reflection shape of the a-domain region
FIG. 1. �Color online� SEM images of �a� a ultrathin AAO mask and �b�array of 65-nm-sized PZT nanoislands. �c� Cross-section TEM image of ananocapacitor of type Pt-top electrode/PZT nanoisland/Pt-bottom electrode�MFM type�. �d� XRD pattern of a PZT thin film �solid line� and of nanois-lands �dashed line�. In-plane �-scan of PZT nanoislands is shown as an insetin �d�.
FIG. 2. SEM images of PZT nanoislands with size of �a� 350 nm, �n� 65 nm,and �c� 40 nm, and TEM images of 65-nm-sized PZT nanoislands withthickness of �d� 40 nm, �e� 30 nm, and �f� 15 nm, respectively.
044102-3 Han et al. J. Appl. Phys. 108, 044102 �2010�
Downloaded 23 Aug 2010 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
is similar to the case of continuous films under differentstrain states. Our previous works confirmed that when thestrain state of thin films changes from compressive to tensile,the shape of the a-domain reflection also changes from diag-onal to flat shape �Fig. S1�.27–29 This means that the 40-nm-thick PZT nanoislands undergo relaxation of compressivestrain compared to the thin film of the same thickness. Inaddition, the a-domains of PZT nanoislands consist of twotypes of structures as shown in Fig. 4�a� and correspondingrocking curve in Fig. 4�b�: the one is fourfold symmetric�corresponding to four types of tilted a-domains with ap-proximately equal proportion as present in thin films asshown in Fig. 4�c�� and the other is centrosymmetric �corre-sponding to a single well-oriented a-domain�. Unlike thefourfold symmetric a-domains, the centrosymmetrica-domains are nontilted ���=0�, i.e., the �00l� axes of thea-domains are aligned parallel to the plane of the underlyingsubstrate �Fig. 4�d��. It should be noted here that the domainstructure shown in Fig. 4�d� is only a schematic in order tohelp understand the term “nontilted a-domains,” and has noperspective on the real domain structure and the habit planeof the domain boundaries. In the viewpoint of energy mini-mization, ferroelectric domain boundaries should be un-charged. This requirement can be fulfilled only if the com-ponent of the spontaneous polarization �Ps�, that isperpendicular to the domain boundary, equals on both sides
FIG. 3. �Color online� RSMs of the HL-plane near the PZT �001� reflection of �a� 40-nm-thick, �b� 15-nm-thick PZT thin films, and �c� 40-nm-thick �reprintedfrom Ref. 26�, �d� 15-nm-thick PZT nanoislands grown on Pt/MgO. The HL-plane scan involves both PZT �001� and �100� reflections corresponding to c- anda-domains, respectively.
FIG. 4. �Color online� RSM of the HK-plane of the PZT �100� reflection of40-nm-thick PZT nanoislands grown on Pt/MgO �reprinted from Ref. 26��b� shows the rocking curve of the PZT �100� reflections along H-direction.�c� and �d� show schematics of the tilted- and non-tilted a-domain structures.
044102-4 Han et al. J. Appl. Phys. 108, 044102 �2010�
Downloaded 23 Aug 2010 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
of the boundary. In the case of 90° a /c-domain boundaries,this criterion requires that the habit plane of the boundary is�110�. In this respect, the schematic shown in Fig. 4�d� vio-lates the criterion required for the energy minimization; thedomain boundaries are expected to be highly charged. Weassume that such a /c-domain boundaries, if present, mightbe characterized by defective lattices or dislocations, andhave rather broad walls as represented in Fig. 4�d�. In fact,our recent study indicates that charged domain boundaries,which can be metastable due to the stabilization by aliovalentcations �i.e., Ti3+ in PZT� or due to a polarization switchingprocess, are much broader than noncharged boundaries.30 Inthe case of PZT nanoislands grown on STO, unlike the PZTgrown on Pt, the a-domains observed are all centrosymmet-ric �Fig. S2 of Ref. 29�. As the thickness of PZT nanoislandsdecreased, all a-domains disappeared completely. PZTnanoislands grown on the STO undergo stronger compres-sive strain than on Pt, thus it is believed that the strain relax-ation occurs more, resulting in the single nontilteda-domains. The appearance of nontilted a-domains in dis-crete PZT nanoislands can be attributed to the reduced misfitstrain, and this is in good agreement with the recent obser-vations from patterned PTO with an aspect ratio �f =h /d d:lateral dimension, h: height� of almost one, where the com-pressive strain caused by the underlying Pt layer can be sig-nificantly reduced, compared to the case of a continuousferroelectric film.8 In contrast, our previous work also re-vealed that PTO nanoislands fabricated by a chemical routewith low aspect ratio �f =0.15–0.25,� did not contain non-tilted a-domains.31 Therefore, the present observation sug-gests that the presence of non-tilted a-domains is rather re-lated to the high aspect ratio, i.e., a high degree of strainrelaxation. As the thickness of the PZT nanostructures de-creases down to 15 nm, the intensity of both sides of thea-domain reflections decreases but still remains unlike the15-nm-thick film. This observation reveals clearly that thestrain could be relaxed even in the discrete 15-nm-thicknanostructure due to the residual a-domains.
C. Piezoelectric properties of PZT nanocapacitors
The impact of domain structure and strain relaxation inepitaxial PZT nanocapacitors on the electrical properties wasinvestigated by PFM measurements. For accurate electricalmeasurement, ferroelectric nanocapacitors with Pt/PZT/Ptconfiguration �i.e., MFM nanocapacitors� were fabricated.Figure 5 displays �a� topography, �b� piezoresponse image of40-nm-thick MFM nanocapacitors, and �c� and �d� autocor-relation functions of the respective images. Large-area arraysof MFM nanocapacitors showed a uniform piezoresponse.Figure 6 compares local piezoelectric hysteresis loops of adiscrete MFM nanocapacitor measured from the initial twosuccessive hysteresis runs; the curve with solid square marksthe first hysteresis run, and the curve with open square marksthe second hysteresis run. A piezoelectric hysteresis loop of amicrometer-sized capacitor is also presented for direct com-parison �the curve with cross marks�. The micrometer-sizedferroelectric capacitor was fabricated by patterning a Pt-topmetal electrode of 20�20 m2 size onto an epitaxially
grown continuous PZT thin film with the same thickness asthat in the MFM nanocapacitors �40 nm�. Each MFM nano-capacitor exhibited a well-developed ferroelectric hysteresiscurve with a coercive voltage of about 0.8 V and an effectiveremanent piezoelectric coefficient �d33� of �100 pm /V,comparable to the theoretical values. The sharp switchingand the squareness of the hysteresis loops indicate the highquality of the ferroelectric nanostructures, despite the lowvolume of the ferroelectric material.
It should be noted here that the remanent piezoelectriccoefficient of the MFM nanocapacitors decreases from ap-proximately 130 pm/V for the first hysteresis run �solidsquare curve� to approximately 100 pm/V for the secondhysteresis run �open square curve�. The decrease in the rem-anent piezoelectric coefficient in the initial two successive
FIG. 5. �Color online� AFM images of large-area arrays of PZT nanocapaci-tors: �a� topography, �b� piezoresponse and �c� and �d� autocorrelation func-tions of the respective images. The z-ranges of topography and piezore-sponse images are 71 nm and 18.2 V, respectively.
FIG. 6. �Color online� Piezoelectric hysteresis loops of 40-nm-thick PZTnanocapacitors grown on Pt/MgO obtained by two consecutive hysteresisruns: Solid squares mark the first run, and open squares mark the second run�reprinted from Ref. 26�. The hysteresis loop of a 20-m-sized MFM ca-pacitor with a thickness of 40 nm is also shown �cross marks�.
044102-5 Han et al. J. Appl. Phys. 108, 044102 �2010�
Downloaded 23 Aug 2010 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
hysteresis runs can be attributed to the a-domain switching inthe MFM nanocapacitors under an applied electric field. Thepresence of switchable a-domains has already been observedby Nagarajan et al.25 who showed that a significant enhance-ment of the piezoelectric response can take place due to thereduced substrate clamping and the movement of ferroelasticdomain walls, when the continuous ferroelectric thin film ispatterned into micrometer-sized discrete islands. We believethat a-domains in discrete MFM nanocapacitors are alsohighly mobile under an external bias as a result of the re-duced misfit strain, which is supported by our RSM analysesthat evidenced the evolution of a high degree of misfit strainrelaxation in discrete PZT nanoislands. Nagarajan et al.32
also demonstrated that the intrinsic value of the piezore-sponse can be obtained in a patterned thin film due to thestrain relaxation. We observed that the remanent piezoelec-tric coefficient of MFM nanocapacitors is higher than that ofa continuous PZT thin film �see Fig. 6�. About 50% increasein the remanent piezoelectric coefficient can also be attrib-uted to the stress release.
IV. CONCLUSIONS
Large-area arrays of epitaxial 65-nm-sized PZT nano-structures on Pt/MgO �001� were fabricated by AAO mask-assisted PLD. In the case of a 40-nm-thick PZT thin film,only fourfold symmetric a-domains were observed, whilePZT nanoislands with comparable thickness contain bothfourfold symmetric and centrosymmetric a-domains simulta-neously, indicating that the PZT nanoislands are subject tomisfit strain relaxation. In addition, the centrosymmetrica-domains are present only in the nanoislands with low scal-ing ratio, which means that the strain relaxation is related tothe scaling ratio of the nanoislands rather than thickness orlateral dimension itself. According to the PFM analyses, italso turned out that the a-domains in discrete MFM nanoca-pacitors are switchable due to the stress relief, resulting in anincrease in the remanent piezoelectric coefficient. Despite thetiny volume, MFM nanocapacitors also showed a higherremanent piezoelectric coefficient than micrometer-sizedferroelectric capacitors with comparable thickness of PZTdue to a significant strain relaxation.
ACKNOWLEDGMENTS
This work is financially supported by a joint project ofKorea Research Foundation �KRF� and DeutscheForschungsgemeinschaft �DFG� and in part by the Volks-wagen Foundation �Project No. I/80897� and by Korea Re-search Council of Fundamental Science and Technology�KRCF� through the KRISS project. H. Han is grateful forthe award of a fellowship of the German Academic Ex-change Service �DAAD� and the Brain Korea 21 program.
The authors thank Dr. A. Lotnyk for the TEM images. Thispaper is dedicated to the memory of the late Professor UlrichGösele.
1J. F. Scott and C. A. P. de Araujo, Science 246, 1400 �1989�.2Y. Xu, Ferroelectric Materials and Their Applications �Elsevier Science,Amsterdam, 1991�.
3O. Auciello, J. F. Scott, and R. Ramesh, Phys. Today 51�7�, 22 �1998�.4R. Ramesh, T. Sands, and V. G. Keramidas, Appl. Phys. Lett. 63, 731�1993�.
5K. Lee, J. Choi, J. Lee, and S. Baik, J. Appl. Phys. 90, 4095 �2001�.6K. Saito, T. Kurosawa, T. Akai, T. Oikawa, and H. Funakubo, J. Appl.Phys. 93, 545 �2003�.
7D. D. Fong, G. Brian Stephenson, S. K. Streiffer, J. A. Eastman, O.Auciello, P. H. Fuoss, and C. Thompson, Science 304, 1650 �2004�.
8K. Lee and S. Baik, Annu. Rev. Mater. Res. 36, 81 �2006�.9T. Hidaka, T. Maruyama, M. Saitoh, N. Mikoshiba, M. Shimizu, T.Shiosaki, L. A. Wills, R. Hiskes, S. A. Dicarolis, and J. Amano, Appl.Phys. Lett. 68, 2358 �1996�.
10A. Gruverman, O. Auciello, R. Ramesh, and H. Tokumoto, Nanotechnol-ogy 8, A38 �1997�.
11F. Xu, S. Trolier-McKinstry, W. Ren, B. Xu, Z.-L. Xie, and K. J. Hemker,J. Appl. Phys. 89, 1336 �2001�.
12A. L. Kholkin, E. K. Akdogan, A. Safari, P.-F. Chauvy, and N. Setter, J.Appl. Phys. 89, 8066 �2001�.
13F. Xu, F. Chu, and S. Trolier-McKinstry, J. Appl. Phys. 86, 588 �1999�.14K. Lee, Y. K. Kim, S. Baik, J. Kim, and S. Jung, Appl. Phys. Lett. 79,
2444 �2001�.15V. Nagarajan, J. Junquera, J. Q. He, C. L. Jia, R. Waser, K. Lee, Y. K.
Kim, S. Baik, T. Zhao, R. Ramesh, P. Gosez, and K. M. Rabe, J. Appl.Phys. 100, 051609 �2006�.
16V. Nagarajan, I. G. Jenkins, S. P. Alpay, H. Li, S. Aggarwal, L.Salamanca-Riba, A. L. Roytburd, and R. Ramesh, J. Appl. Phys. 86, 595�1999�.
17Y. K. Kim, H. Morioka, R. Ueno, S. Yokoyama, K. Lee, S. Baik, and H.Funakubo, Appl. Phys. Lett. 88, 252904 �2006�.
18R. Ahluwalia and D. J. Srolovitz, Phys. Rev. B 76, 174121 �2007�.19M. Alexe, C. Harnagea, and D. Hesse, J. Electroceram. 12, 69 �2004�.20S. Clemens, T. Schneller, A. Van der Hart, F. Peter, and R. Waser, Adv.
Mater. 17, 1357 �2005�.21M.-W. Chu, I. Szafraniak, R. Scholz, C. Harnagea, D. Hesse, M. Alexe,
and U. Gösele, Nature Mater. 3, 87 �2004�.22I. I. Naumov, L. Bellaiche, and H. Fu, Nature �London� 432, 737 �2004�.23K. Lee, K. Kim, S.-J. Kwon, and S. Baik, Appl. Phys. Lett. 85, 4711
�2004�.24S. Bühlmann, B. Dwir, J. Baborowski, and P. Muralt, Appl. Phys. Lett. 80,
3195 �2002�.25V. Nagarajan, A. Roytburd, A. Stanishevsky, S. Prasertchoung, T. Zhao, L.
Chen, J. Melngilis, O. Auciello, and R. Ramesh, Nature Mater. 2, 43�2003�.
26W. Lee, H. Han, A. Lotnyk, M. A. Schubert, S. Senz, M. Alexe, D. Hesse,S. Baik, and U. Goesele, Nat. Nanotechnol. 3, 402 �2008�.
27Y. K. Kim, K. Lee, and S. Baik, J. Appl. Phys. 95, 236 �2004�.28K. S. Lee and S. Baik, J. Appl. Phys. 85, 1995 �1999�.29See supplementary material at http://dx.doi.org/10.1063/1.3475476 for re-
ciprocal space mapping of HL-plane of PTO films and PZT nanoislands.30C.-L. Jia, S.-B. Mi, K. Urban, I. Vrejoiu, M. Alexe, and D. Hesse, Nature
Mater. 7, 57 �2008�.31H. Han, K. Lee, W. Lee, M. Alexe, D. Hesse, and S. Baik, J. Mater. Sci.
44, 5167 �2009�.32V. Nagarajan, A. Stanishevsky, L. Chen, T. Zhao, B.-T. Liu, J. Melngailis,
A. L. Roytburd, J. Finder, Z. Yu, D. K. Eisenbeiser, and R. Ramesh, Appl.Phys. Lett. 81, 4215 �2002�.
044102-6 Han et al. J. Appl. Phys. 108, 044102 �2010�
Downloaded 23 Aug 2010 to 192.108.69.177. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
