当前位置:首页 >> 能源/化工 >>

Domain Engineering for Enhanced Ferroelectric Properties of Epitaxial


www.advmat.de

COMMUNICATION

Domain Engineering for Enhanced Ferroelectric Properties of Epitaxial (001) BiFeO Thin Films
By Ho Won Jang, Daniel Ortiz, Seung-Hyub Ba

ek, Chad M. Folkman, Rasmi R. Das, Padraic Shafer, Yanbin Chen, Christofer T. Nelson, Xiaoqing Pan, Ramamoorthy Ramesh, and Chang-Beom Eom*
Multiferroic BiFeO3 has attracted great interest due to its promising application to magnetoelectric devices.[1–3] In addition, the high remanent polarization and piezoelectric response of BiFeO3 thin ?lms, which are comparable to those of conventional Ti-rich lead zirconia titanate, suggested BiFeO3 as a strong candidate for lead-free nonvolatile memories.[4] BiFeO3 has a rhombohedral perovskite structure with pseudocubic lattice ? parameters ar ? 3.96 A and ar ? 0.68.[5] Due to this low symmetry, (001)-oriented epitaxial BiFeO3 ?lms possess the rhombohedral distortion along one of the four (111) crystallographic directions of the pseudocubic perovskite unit cell.[6] Thus, eight possible polarization (ferroelectric) variants, which correspond to four structural (ferroelastic) domains, may form in the ?lms, leading to complex domain patterns with both {100} and {101} twin boundaries.[6,7] Such a complex domain structure can deteriorate the ferroelectric response of the system by external electric ?eld, and complicates the examination of the coupling between magnetic and ferroelectric order parameters in BiFeO3.[3] Recently, several studies have shown that high-quality (001)oriented BiFeO3 thin ?lms with two-variant stripe domains can be achieved by using miscut[8] and orthorhombic[9,10] substrates. However, the clear identi?cation of the origin of the stripe domains in BiFeO3 thin ?lms has not yet been reported. Furthermore, correlating the ferroelastic domain structures of BeFiO3 thin ?lms with the ferroelectric properties is a critical, yet missing link between materials properties and device performance. In this communication, we report the origin of the ferroelastic domain variant selection in (001) BiFeO3 ?lms on miscut (001) SrTiO3 substrates with coherent SrRuO3 bottom electrodes, and its effect on the ferroelectric properties of the ?lms. To our best
[*] Prof. C. B. Eom, Dr. H. W. Jang, D. Ortiz, S. H. Baek, C. M. Folkman, R. R. Das Department of Materials Science and Engineering University of Wisconsin Madison, WI 53706 (USA) E-mail: eom@engr.wisc.edu P. Shafer, Prof. R. Ramesh Department of Physics and Department of Materials Science and Engineering University of California Berkeley, CA 94720 (USA) Dr. Y. B. Chen, C. T. Nelson, Prof. X. Q. Pan Department of Materials Science and Engineering University of Michigan Ann Arbor, MI 48109 (USA)

DOI: 10.1002/adma.200800823

knowledge, this is the ?rst report that ferroelectric switching behavior and leakage current in BiFeO3 ?lms are simultaneously improved by domain engineering. For the demonstration of the domain variant selection in BiFeO3 ?lms by substrate miscut, the SrTiO3 substrate was chosen to have either 0.05 or 48 miscut toward the [100] direction, which corresponds to the downhill miscut direction. Since the step width of the 0.058 miscut substrate ($460 nm) is much larger than that of the 48 miscut substrate (6 nm), the effects of the substrate on the strain relaxation and domain structure can be resolved using both substrates (in this letter we call 0.058 miscut exact and 48 miscut miscut, for convenience). Atomic force microscopy (AFM) and reciprocal space mapping (RSM) using high-resolution X-ray diffraction (HRXRD) show that the miscut substrate leads to step?ow growth and two-variant strip domains in the BiFeO3 ?lm. In contrast, the exact substrate causes 3D island-growth and fourvariant domains. Combined with transmission electron microscopy (TEM) and piezoelectric force microscopy (PFM) results, it is suggested that both the preferential distortion of unit cells and the complete step-?ow growth induced by the substrate anisotropy are the origins of the formation of the two-variant stripe domains in (001) BiFeO3 ?lms. Finally, the polarization– electric ?eld (P–E) hysteresis loop and leakage-current measurements allow us to ?nd that two-variant stripe domains provide complete ferroelectric switching in BiFeO3 thin ?lms with low leakage current. The growth mode of the epitaxial BiFeO3 ?lm can be clari?ed by monitoring the change of surface morphology with increasing ?lm thickness. AFM images were obtained before and after the deposition of BiFeO3 ?lms on both exact and miscut (001) SrTiO3 substrates with coherent SrRuO3 bottom electrodes, as shown in Figure 1. The surface of the 100 nm thick SrRuO3 on the exact substrate is atomically smooth with one ? unit cell high (4A) steps (Fig. 1a). The terrace width is $500 nm, which is consistent with the miscut angle of 0.058. After the deposition of BiFeO3 on top of SrRuO3 bottom electrode, the 100 nm thick BiFeO3 ?lm on the exact substrate exhibits many protrusions and holes (Fig. 1b). The propagation of atomic steps is randomly oriented, as indicated by white arrows. With increasing thickness to 400 nm, the ?lm shows a much rougher surface with big islands (Fig. 1c), consistent with the 3D islandformation growth mode.[11] In contrast, completely different surface morphology can be observed in BiFeO3 ?lms on the miscut substrate. The AFM image of the SrRuO3 bottom electrode shows the periodic steps (Fig. 1d), originating from step bunching due to the small terrace width. The actual terrace width (140 nm) is much larger than the

Adv. Mater. 2009, 21, 817–823

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

817

www.advmat.de

BiFeO3. In contrast to the SrRuO3 bottom electrodes, BiFeO3 ?lms show broad peaks with different shapes compared with the substrate. Analysis on the RSM patterns suggested that the BiFeO3 ?lm on exact SrTiO3 has four domains (Fig. 2a), but clearly exhibits only two domains on miscut SrTiO3 (Fig. 2c). The BiFeO3 ?lm on exact SrTiO3 shows the same intensity for each domain, whereas that on miscut SrTiO3 displays two different peak intensities. The RSM pattern around the 013 SrTiO3 re?ection (not shown here) revealed that the peak intensities of both BiFeO3 domains on miscut SrTiO3 are exactly reversed after rotating the ?lm by F ? 1808, indicating the equal amount of both domains in the ?lm. According to Streiffer et al.,[6] there are four structural domains of a rhombohedral phase, r1, r2, r3, and r4. Using these notations, we can identify all domains of both BiFeO3 ?lms, as shown in Figure 2b and d. Note that the BiFeO3 ?lm on miscut (001) SrTiO3 has the rhombohedral distortion only along the [100] direction (distortion angle am ? 0.658), while that on exact SrTiO3 has the distortion along both [100] and [100] directions (ae ? ?0.48). These results provide direct evidence that the miscut substrate signi?cantly affects the domain structure of Figure 1. AFM images of a,d) before and after b,e) 100 nm and c,f) 400 nm thick BiFeO3 ?lms on the BiFeO ?lm. The peak width of BiFeO 3 3 a–c) exact and d–f) miscut SrTiO3, with coherent 100 nm thick SrRuO3 bottom electrodes. The ?lm along the 0k0 direction is much narrower white arrows indicate the directions of step propagation. The line pro?les were obtained across the in the ?lm on miscut SrTiO3, indicating the dotted lines in the corresponding images. The black arrows in f) indicate domain boundaries. improvement of crystalline quality in BiFeO3 ?lms by using that substrate. Details on the crystal symmetry and domain structure of the BiFeO3 ?lms on exact and miscut (001) SrTiO3 substrates will be estimated value based on the miscut angle (6 nm), due to step discussed elsewhere.[10] bunching. After the deposition of BiFeO3, the 100 nm thick BiFeO3 ?lm displays terraces 500 nm wide (Fig. 1e), which means In the cross-sectional views of a (001) rhombohedral ?lm that step bunching also happens during the growth of BiFeO3 on along [100] direction by Streiffer et al.,[6] r1/r2 or r3/r4 pairs form the miscut substrate. The atomic steps propagate along the miscut {100} twin boundaries, and r1/r4 or r2/r3 pairs form {101} twin direction. As the ?lm gets thicker, the ?lm surface gets smoother, boundaries. It suggests that a ?lm with all four r1, r2, r3, and r4 and the ?at terraces fade away with the formation of a number of domains can have both {100} and {101} twin boundaries, but a small steps along the miscut direction (Fig. 1f). Note that the ?lm with only two domains will have one preferred boundary orientation. This is consistent with our experimental results. thicker ?lm still exhibits a surface morphology of the step-?ow growth mode[11] without formation of 3D islands. The step-?ow Figure 3a and c show cross-sectional TEM images of 600 nm thick BiFeO3 ?lms on exact and miscut (001) SrTiO3, growth of BiFeO3 was also reported using orthorhombic DyScO3 respectively. The ?lm on the exact substrate shows irregular substrates.[9] domains with both {100} and {101} twin boundaries. In In order to determine the domain structures and the contrast, the ?lm on the miscut substrate exhibits periodic crystallographic distortion of each domain with respect to the domains with {101} twin boundaries, namely stripe domains. substrate miscut direction, BiFeO3 ?lms were investigated using Corresponding domain con?gurations for both ?lms are the HRXRD RSM technique. Figure 2 shows RSM patterns identi?ed and schematically presented. The width of stripe around SrTiO3 013 re?ections for BiFeO3 ?lms 400 nm thick on domains is determined to be 200$250 nm for 400$600 nm exact and miscut SrTiO3. The single narrow peak for SrRuO3 013p ?lms,[13] which is consistent with the spacing between domain re?ections indicates that the SrRuO3 layers on both substrates are boundaries parallel with the nonmiscut direction seen in the single-domain. Since the crystal structure of SrRuO3 is cubic at AFM image in Figure 1f. It is clear that the BiFeO3 ?lm on the the BiFeO3 growth temperature (690 8C),[12] the surface of the exact substrate has four polarization variants (Fig. 3a), and the SrRuO3 layer is crystallographically identical to that of the BiFeO3 ?lm on the miscut substrate has two polarization variants underlying SrTiO3 substrate. This indicates that the SrRuO3 (Fig. 3c). The in-plane PFM images con?rm four variants in the bottom electrodes have no additional effect on the growth of

COMMUNICATION
818

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2009, 21, 817–823

www.advmat.de

Figure 2. RSM images around 103 SrTiO3 re?ections for BiFeO3 ?lms 400 nm thick on a) exact and c) miscut SrTiO3. Subscript p represents pseudocubic setting. The schematics describe b) four structural domains in the ?lm on the exact substrate and d) two domains in the ?lm on the miscut substrate. The black arrows indicate the polarization direction of each domain in the unit cell and the (010) plane. ae (? 0.48) and am (? 0.658) are the rhombohedral distortion angles.

Figure 3. Cross-sectional TEM images of BiFeO3 ?lms 600 nm thick on a) exact and c) miscut SrTiO3. In-plane PFM images (4 mm ?4 mm) of BiFeO3 ?lms 40 nm thick on b) exact and d) miscut SrTiO3 were obtained with an ac bias of 3 Vpp at 6.39 kHz. The arrows in each PFM image represent directions of in-plane polarization, indicating four variants on exact SrTiO3 and two variants on miscut SrTiO3.

BiFeO3 ?lm on exact SrTiO3 (Fig. 3b) and two variants in the BiFeO3 ?lm on miscut SrTiO3 (Fig. 3d). It was observed that an 800 nm thick BiFeO3 ?lm on miscut substrate still has only two variants. This fact suggested that the elasticstrain energy in BiFeO3 ?lms on miscut substrates are effectively relieved without forming additional domain variants (r2 and r3). There are two competing mechanisms for strain relaxation, namely surface roughening and crystallographic tilt by dislocation multiplication.[14] The AFM images clearly indicate that the BiFeO3 ?lms on exact SrTiO3 exhibit strain relaxation by surface roughening. However, there was no surface roughening in the BiFeO3 ?lms on miscut SrTiO3. Thus, we measured the crystallographic tilt of BiFeO3 ?lms respective to the SrRuO3 bottom electrodes using RSM patterns around 002 SrTiO3 re?ections along two orthogonal directions (miscut and nonmiscut directions), as shown in Figure 4a. As expected, there is no ?lm tilt along the [010] direction (nonmiscut direction) because the miscut angle does not exist for the ?lms to tilt against the substrate. However, the tilt of the ?lms along the [100] direction (miscut direction) is signi?cant. The highly strained 20 and 50 nm ?lms have negative tilt angles, due to the intrinsic tilt mechanism described by Nagai,[15] in which a compressively strained coherent ?lm tilts away from the surface normal due to lattice mismatch on the surface steps of the miscut substrate, as shown in the inset of Figure 4a. With increasing ?lm thickness, the ?lm tilts toward the direction normal to the surface, reducing the angle between this and the [001] direction of the ?lm (Fig. 4b). This positive tilt is evidence of preferential dislocation nucleation, which corresponds to strain relaxation in the ?lm to relieve its total elastic energy.[14] It should be noted that the tilt of the BiFeO3 ?lms is observed along the miscut direction, and not in the nonmiscut direction. The tilting of BiFeO3 ?lms along the [001] direction is attributed to the anisotropic strain relaxation in the ?lms on miscut substrates. Figure 4c and d show the in-plane and out-ofplane lattice parameters of BiFeO3 ?lms on exact and miscut SrTiO3 as a function of ?lm thickness. With increasing ?lm thickness, the in-plane lattice parameters increase and the out-of-lattice parameters decrease, due to the relaxation of biaxial compressive strains. As the ?lm thickness increases, the ?lms on exact SrTiO3 display almost identical variation in inplane lattice parameters along [100] and [010] directions (Fig. 4c). In contrast, the ?lms on

Adv. Mater. 2009, 21, 817–823

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

819

COMMUNICATION

www.advmat.de

described as distortions along both [100] and [100] directions in the (010) plane. As seen in Figure 5a, r1 and r4 are twins with the (101) plane, and r1 and r2 with the (100) plane. As a result, all four r1, r2, r3, and r4 domains are formed in the ?lm {101} and {100} twin boundaries. It is easily speculated that r2 or r3 domains are formed preferentially to the 3D islands, creating {100} boundaries. On the miscut substrate, the small terrace width drives BiFeO3 to adopting a complete step?ow growth mode. This prevents the formation of 3D islands for r2 or r3 domains, because the rhombohedral distortion toward the step edge is not energetically favorable, as shown in Figure 5b. Furthermore, the miscut substrate drives the ?lms to tilting along the [100] direction (Fig. 4a), which corresponds to the preferential rhombohedral distortions to the downhill miscut direction, as shown in Figure 5b. In other words, the miscut substrate removes from the ?lm the degree of freedom of having rhombohedral distortions for r2 and r3. Consequently, r1 and r4 stripe domains with (101) twin boundaries are formed in the ?lm. The effects on the ferroelectric properties of BiFeO3 ?lms can be explored by measuring P–E hysteresis loops. The domain selection by the miscut substrate greatly affects the ferroelectric switching behavior of BiFeO3. The 400 nm thick BiFeO3 ?lm on miscut SrTiO3 exhibits perfect square-like P–E loops (Fig. 6a). The remanent polarization (Pr) value is measured to be 64 ? 2 mC cm?2. Compared Figure 4. a) Tilt angle as a function of ?lm thickness for BiFeO3 ?lms on miscut SrTiO3 substrate with the Pr of bulk single-crystal BiFeO3 with SrRuO3 bottom electrodes. The inset is a schematic for the negative tilting of coherent BiFeO3 ($60 mC cm?2),[16] it is concluded that ?lms on the miscut SrRuO3/SrTiO3 substrate. b) Schematic diagrams for the change of the tilt complete domain switching is obtained from angle for the BiFeO3 ?lms with thickness. Pseudocubic lattice parameters were measured as a the BiFeO3 ?lms with the two-variant stripe function of ?lm thickness for BiFeO3 ?lms grown on c) exact and d) miscut SrTiO3. The dotted line domains. The higher Pr values of the ?lms in d) represents the out-of-plane lattice parameter of the BiFeO3 ?lm exact SrTiO3. than those of BiFeO3 bulk single-crystal can be attributed to the strain-induced polarization rotation.[17] In contrast, the 400 nm thick BiFeO3 ?lm on exact SrTiO3 displays an unclosed and slanted miscut SrTiO3 show faster in-plane strain relaxation along the loops at the lower frequency, and drastically lower Pr values of miscut direction ([100] direction) than along the nonmiscut direction ([010] direction) (Fig. 4d). Thus, the overall strain 43 ? 5 mC cm?2 (Fig. 6b). The slanted loop indicates the relaxation is faster in the ?lms on miscut SrTiO3. This result nonuniformity of the domains, in that each one in the ?lm has a different coercivity. This suggests that some of domains are con?rms the in-plane anisotropy of the miscut substrate, which very hard to switch with applied ?eld, leading to the lower Pr. As has also been observed in the tilting of the crystallographic planes and in the AFM images. seen in TEM and PFM images, more uniform and ordered Based on these experimental results, the origin of ferroelastic ferroelectric (ferroelastic) domains are seen for ?lms on miscut domain variant selection in epitaxial (001) BiFeO3 ?lms by the than on exact substrates, contributing to a square-like loop rather than a slanted P–E loop. To con?rm the signi?cantly different miscut substrate is described as follows. On the exact substrate, switching behaviors between both ?lms, pulsed-polarization the terrace width is too large for BiFeO3 to grow in a step-?ow measurements were carried out as functions of applied ?eld and growth mode. As a result, 3D islands are formed. As the ?lm gets pulse width. Figure 6c shows the switching polarization (DP) as a thicker, the shear strain in the ?lm, induced by the rhombohedral function of electric ?eld. The saturated DP values for both ?lms symmetry of BiFeO3, should be relieved by the rhombohedral are in excellent agreement with the 2Pr value from the P–E loops, distortion of the unit cells. Since the ?lm has biaxially isotropic strain, the rhombohedral distortion occurs randomly, and is clarifying the reduced Pr in the ?lm on exact SrTiO3. From

COMMUNICATION
820

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2009, 21, 817–823

www.advmat.de

Figure 5. Schematic drawings for the initial growth, domain selection, formation of domain boundaries, and ?nal structure for BiFeO3 ?lms on a) exact and b) miscut SrTiO3 substrates with SrRuO3 bottom electrodes. The formation of r2 and r3 domains is prohibited in BiFeO3 ?lms on miscut SrTiO3 substrates, because the rhombohedral distortion for r2 and r3 is mechanically unfavorable.

Figure 6. P–E hysteresis loops of BiFeO3 ?lms 400 nm thick on a) miscut and b) exact SrTiO3 at room temperature. Switching polarization (DP) as a function of c) applied electric ?eld and d) pulse width for BiFeO3 ?lms 400 nm thick on miscut and exact SrTiO3.

starting to increase to saturating, the DP of the ?lm on miscut SrTiO3 shows a narrower and abrupt transition, whereas that of the ?lm on the exact SrTiO3 shows a more gradual transition, consistent with the difference in the shape of P–E loops (square-like vs. slanted). With increasing the pulse width from 1 ms to 1 ms, the DP of the ?lm on miscut SrTiO3 remained constant at 132 mC cm?2 at 200 and 300 kV m?1. However, the ?lm on exact SrTiO3 shows different DP values at 200 and 300 kV m?1, and a stark contrast when DP becomes zero with pulses approaching 1 ms at 200 and 300 kV m?1. This means that the ?lm does not switch at all with long pulse widths. Signi?cant reduction in leakage current density is found in BiFeO3 ?lms on miscut SrTiO3. Figure 7a shows leakage current density as a function of applied electric ?eld for 400 nm thick BiFeO3 ?lms on both exact and miscut SrTiO3. The ?lm on the exact substrate displays a leakage current density around 2 ? 10?3 A cm?2 at 100 kV cm?1, which is comparable to the previously reported values between 5 ? 10?1 and 5 ? 10?3 A cm?2 at 100 kV cm?1.[18–20] Note that the leakage current density is reduced by two orders of magnitude for the ?lm on the miscut substrate. The leakage current density of 3 ? 10?5 A cm?2 at 100 kV cm?1 is the lowest value ever reported for BiFeO3 ?lms.[21] Current–time measurements were carried out to examine current relaxation in both ?lms, as shown in Figure 7b. The leakage current for the ?lm on the miscut substrate stabilizes in 10 ms, whereas the current relaxation for the ?lm on the exact substrate is very slow, and does not stabilize even in 100 ms, implying that the ?lm has a lot of free charge carriers with oxygen vacancies, resulting in high leakage current levels and slow current relaxation times.[22] We believe there are two possible mechanisms for the origin of the higher leakage current in BiFeO3 ?lms on exact SrTiO3, namely i) domain structure and ii) nonstoichiometric point defects. The ?rst mechanism operates through the notion that 1098 domain walls are dominant leakage paths rather than the bulk matrix, as shown in Figure 7c. According to Streiffer et al.,[6] a rotational deformation of opposite sense about the (010) direction in adjacent domains is required to bond the vertical 1098 domains to the substrate, which is not necessary for the 718 domains with (101) twin boundaries to be bonded to the substrate. This additional deformation can cause the width of the 1098 domain walls to be larger and generate charge carriers near the walls. In addition, the 1098

Adv. Mater. 2009, 21, 817–823

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

821

COMMUNICATION

www.advmat.de

BiFeO3 ?lms on exact substrates. The domainwall motions for polarization switching can be delayed and impeded by the larger walls between 1098 domains, resulting in incomplete switching with the applied electric ?led. At very high electric ?elds, nonswitchable domains can be switched, and then complete switching can be obtained. However, at such high ?elds, the leakage current through the vertical 1098 domain walls and/or bulk matrices becomes predominant, and thus complete switching cannot be obtained. For this same reason, the ?lms do not switch with the long pulse widths in Figure 6d. A recent study on PFM of (011) BiFeO3 ?lms showed that 1098 domain switching is less stable and less electrically controllable than 718 domain switching, partially supporting our suggestion.[25] In conclusion, we have demonstrated the selection of domain-structure variants in epitaxial BiFeO3 ?lms, and consequently achieved signi?cant improvement in ferroelectric switching behavior and leakage current by employing miscut in cubic (001) SrTiO3 substrates. BiFeO3 ?lms on miscut SrTiO3 have a step-?ow growth and a preferential rhombohedral distortion toward the miscut direction. This result indicates that both the substrate Figure 7. a) Current–voltage characteristics of BiFeO3 ?lms 400 nm thick on exact and miscut anisotropy and the step-?ow growth of BiFeO3 SrTiO3. The measure time for each point was 100 ms. b) Current–time characteristics of BiFeO3 ?lms 400 nm thick on exact and miscut SrTiO3. c) Schematic drawing for leakage paths in BiFeO3 ?lm are the origins of the two-variant stripe 400 nm thick on exact SrTiO3. The arrows represent leakage current through the ?lm, showing that domains in BiFeO3 ?lms. Square-like P–E loops and very low leakage current densities the vertical 1098 domain walls are dominant leakage paths rather than the bulk matrix. are obtained from these BiFeO3 ?lms. In contrast, BiFeO3 ?lms on exact SrTiO3 exhibit low Pr values and domain wall is parallel to the electric ?eld. Thus, the 1098 domain high leakage currents. From this, we suggest that 1098 domain wall can become a predominant leakage path in the ?lm on exact walls prevent the complete ferroelectric switching of (001) BiFeO3 SrTiO3, leading to the high leakage current. The second mechanism operates through the notion that the domains and act as dominant leakage paths in the four-variant bulk matrix is the dominant leakage path rather than domain BiFeO3 ?lms, whereas the intrinsic ferroelectric properties of walls. Although BiFeO3 ?lms on exact as well as miscut SrTiO3 are BiFeO3 can be observed from two-variant BiFeO3 ?lms with 718 stoichiometric, with a Bi/Fe 1:1 ratio, and have no secondary domain walls.[26] The dependence of ferroelectric properties on phases, there is the possibility of formation of point defects, such domain-wall con?guration opens the exciting opportunity to as oxygen vacancies, in the ?lms. It is generally accepted that the investigate the correlation between domain walls and the miscut substrate leads to the formation of preferential domains antiferromagnetic order in BiFeO3. We believe that such domain engineering can be very useful for growing high-quality BiFeO3 and the stabilization of stoichiometric phases.[23] The longer step width on exact SrTiO3 means longer time for adapted atoms to ?lms on cubic (001) Si substrates for device applications[27] and reach the step edge and crystallize. The relatively volatile bismuth more generally for heterostructures with rhombohedral thin adatoms can evaporate from the ?lm surface during the growth, ?lms, such as Pb(ZrxTi1 ? x)O3, Pb(Mg1/3Nb2/3)O3-PbTiO3, and thus bismuth and oxygen vacancies are formed in the ?lms La1 ? xSrxMnO3, and LaAlO3. after growth. As a result, those nonstoichiometric point defects spread out in the ?lms, leading to the large leakage current. Detailed studies on the high leakage current on exact SrTiO3 are Experimental currently underway. In epitaxial (001) BiFeO3 ?lms, polarization switching occurs Epitaxial (001) BiFeO3 ?lms were grown by off-axis radio-frequency (rf) magnetron sputtering on 0.05 and 48 miscut (001) SrTiO3 substrates[8]. with the formation of both ferroelectric and ferroelastic domain Prior to the deposition of the BiFeO3 ?lms, an epitaxial 100 nm thick walls.[7] According to Wicks et al.,[24] grain boundaries in SrRuO3 bottom electrode was deposited by 908 off-axis rf magnetron Pb(ZrxTi1 ? x)O3 thin ?lms impede domain-wall movement, and sputtering [28]. The thicknesses of the BiFeO3 ?lms were varied from 20 to are able to nucleate domains that are opposite to those generated by 800 nm. The surface morphology and piezoelectric properties of BiFeO3/ an electric ?eld. From this, we believe that 1098 domain walls SrRuO3 heterostructures were investigated using a DI Multimode AFM contribute to the reduction in Pr due to incomplete switching for system [7]. A commercially available high-resolution four-circle X-ray

COMMUNICATION
822

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2009, 21, 817–823

www.advmat.de

diffractometer (D8 Advance, Bruker A?S) was used in HRXRD-RSM measurements. TEM studies were carried out on a Philips CM12 operated at 120 kV with a high-angle (?608) double-tilt holder, and on a JEOL 3011 ultrahigh-resolution TEM operated at 300 kV with a point-to-point resolution of 0.17 nm. Pt top electrodes (100 mm in diameter) were patterned to measure the ferroelectric properties using a Radiant PFH100 ferroelectric measurement system.

Acknowledgements
The authors gratefully acknowledge the ?nancial support of the National Science Foundation through grants ECCS-0708759, 0425914, the Of?ce of Naval Research through grant N00014-05-1-0559, and the Department of Energy under grant DE-FG02-07ER46416. The corresponding author wants to thank Paul Evans for helpful discussions. Received: March 25, 2008 Revised: August 20, 2008

[1] N. A. Spaldin, M. Fiebig, Science 2005, 309, 391. [2] W. Eerenstein, N. D. Mathur, J. F. Scott, Nature 2006, 442, 759. [3] T. Zhao, A. Scholl, F. Zavaliche, K. Lee, M. Barry, A. Doran, M. P. Cruz, Y. H. Chu, C. Ederer, N. A. Spaldin, R. R. Das, D. M. Kim, S. H. Baek, C. B. Eom, R. Ramesh, Nat. Mater. 2006, 5, 823. [4] R. Ramesh, N. A. Spaldin, Nat. Mater. 2007, 6, 21. [5] F. Kubel, H. Schmid, Acta Crystallogr. Sect. B 1990, 46, 698. [6] S. K. Streiffer, C. B. Parker, A. E. Romanov, M. J. Lefevre, L. Zhao, J. S. Speck, W. Pompe, C. M. Foster, G. R. Bai, J. Appl. Phys. 1998, 83, 2742. [7] F. Zavaliche, P. Shafer, R. Ramesh, M. P. Cruz, R. R. Das, D. M. Kim, C. B. Eom, Appl. Phys. Lett. 2005, 87, 252902. [8] R. R. Das, D. M. Kim, S. H. Baek, C. B. Eom, F. Zavaliche, S. Y. Yang, R. Ramesh, Y. B. Chen, X. Q. Pan, X. Ke, M. S. Rzchowski, S. K. Streiffer, Appl. Phys. Lett. 2006, 88, 242904. [9] Y. H. Chu, Q. Zhan, L. W. Martin, M. P. Cruz, P. L. Yang, G. W. Pabst, F. Zavaliche, S. Y. Yang, J. X. Zhang, L. Q. Chen, D. G. Schlom, T. B. Wu, R. Ramesh, Adv. Mater. 2006, 18, 2307.

[10] C. M. Folkman, H. W. Jang, D. Ortiz, S. H. Baek, C. B. Eom, unpublished. [11] M. D. Johnson, C. Orme, A. W. Hunt, D. Graff, J. Sudijono, L. M. Sander, B. G. Orr, Phys. Rev. Lett. 1994, 72, 116. [12] J. P. Maria, H. L. McKinstry, S. Trolier-McKinstry, Appl. Phys. Lett. 2000, 76, 3382. [13] Y. B. Chen, M. B. Katz, X. Q. Pan, R. R. Das, D. M. Kim, S. H. Baek, C. B. Eom, Appl. Phys. Lett. 2007, 90, 072907. [14] P. M. Mooney, Mater. Sci. Eng. R 1996, 17, 105. [15] H. Nagai, J. Appl. Phys. 1972, 43, 4254. [16] D. Lebeugle, D. Colson, A. Forget, M. Viret, P. Bonville, J. F. Marucco, S. Fusil, Phys. Rev. B 2007, 76, 024116. [17] H. W. Jang, S. H. Baek, D. Ortiz, C. M. Folkman, R. R. Das, Y. H. Chu, P. Shafer, J. X. Zhang, S. Choudhury, V. Vaithyanathan, Y. B. Chen, D. A. Felker, M. D. Biegalski, M. S. Rzchowski, X. Q. Pan, D. G. Schlom, L. Q. Chen, R. Ramesh, C. B. Eom, Phys. Rev. Lett. 2008, 101, 107602. [18] X. Qi, J. Dho, R. Tomov, M. G. Blamire, J. L. MacManus-Driscoll, Appl. Phys. Lett. 2005, 86, 062903. [19] Y. Wang, C. W. Nan, Appl. Phys. Lett. 2006, 89, 052903. [20] G. W. Pabst, L. W. Martin, Y. H. Chu, R. Ramesh, Appl. Phys. Lett. 2007, 90, 072902. [21] H. Yang, M. Jain, N. A. Suvorova, H. Zhou, H. M. Luo, D. M. Feldmann, P. C. Dowden, R. F. DePaula, S. R. Foltyn, Q. X. Jia, Appl. Phys. Lett. 2007, 91, 072911. [22] B. Nagaraj, S. Aggarwal, R. Ramesh, J. Appl. Phys. 2001, 90, 375. [23] S. D. Bu, M. K. Lee, C. B. Eom, W. Tian, X. Q. Pan, S. K. Streiffer, J. J. Krajewski, Appl. Phys. Lett. 2001, 79, 3482. [24] S. Wicks, V. Anbusathiah, V. Nagarajan, Nanotechnology 2007, 18, 465502. [25] H. W. Jang, S. H. Baek, D. Ortiz, C. M. Folkman, C. B. Eom, Y. H. Chu, P. Shafer, R. Ramesh, V. Vaithyanathan, D. G. Schlom, Appl. Phys. Lett. 2008, 92, 062910. [26] M. P. Cruz, Y. H. Chu, J. X. Zhang, P. L. Yang, F. Zavaliche, Q. He, P. Shafer, L. Q. Chen, R. Ramesh, Phys. Rev. Lett. 2007, 99, 217601. [27] S. H. Baek, R. R. Das, D. M. Kim, H. W. Jang, C. B. Eom, D. G. Schlom, Y. B. Chen, X. Q. Pan, F. Zavaliche, S. Y. Yang, R. Ramesh, unpublished. [28] C. B. Eom, R. J. Cava, R. M. Fleming, J. M. Phillips, R. B. Vandover, J. H. Marshall, J. W. P. Hsu, J. J. Krajewski, W. F. Peck, Science 1992, 258, 1766.

Adv. Mater. 2009, 21, 817–823

? 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

823

COMMUNICATION


相关文章:
材料物理在计算机上的应用 文献检索
epitaxial perovskite heterostructures 作者 : Mathews,...Enhanced ferroelectric properties of Cr-doped BiFeO...Composite domain walls in a multiferroic perovskite...
Polarization Switching and Light-Enhanced Piezoelectricity in Lead.pdf
Polarization Switching and Light-Enhanced ...ferroelectric domains; therefore, extra ...I. Compositional Engineering of Perovskite Materials...
单畴的多铁性BiFeO3薄膜
Department of Materials Science and Engineering, ...epitaxial BFO films , there are still multi anti...2d, indicating the single ferroelectric P domain....
更多相关标签:
epitaxial | epitaxial growth | epitaxial wafer | ferroelectric | ferroelectricity | enhanced | enhanced steam | enhanced steam怎么用 |