NANO LETTERS
Site-Specific Fabrication and Epitaxial Conversion of Functional Oxide Nanodisk Arrays
2006 Vol. 6, No. 10 2344-2348
Zixiao Pan, Nasim Alem, Tao Sun, and Vinayak P. Dravid* Department of Materials Science and Engineering, International Institute of Nanotechnology, Northwestern UniVersity, EVanston, Illinois 60208 Received August 13, 2006; Revised Manuscript Received September 1, 2006
ABSTRACT Nanodisk arrays of technologically important magnetic (CoFe2O4) and ferroelectric (BaTiO3) oxides are fabricated on diverse substrates with well-defined size and separation using the soft-eBL approach. We demonstrate that below a certain pattern size, the as-deposited amorphous nanodisks can be readily converted into dense, single-crystal form that exhibit cube-on-cube heteroepitaxy with respect to the underlying single-crystal substrate. Such single-crystal disks show well-defined truncated-pyramid morphology that is consistent with Wulff construction. The mechanism of morphology development with the pattern size change is discussed. Localized characterization of the crystallinity, chemical composition, and magnetic behavior of the CFO nanodisk patterns are carried out using analytical transmission electron microscopy and magnetic force microscopy. Such solution-based epitaxial conversion of patterned arrays of multifunctional oxides has potential for viable cost-effective technological applications.
The family of functional ceramics possesses a broad spectrum of properties in response to electric, magnetic, and stress fields, or combinations of them. Many of these properties show remarkable size dependency1,2 as well as interesting synergistic coupling when materials with different functionalities are positioned in close proximity.3,4 This has become the central driving force for the development of innovative design strategies to (1) manipulate functional ceramic materials at nanometer length scales, (2) control the dimensions and morphology of the nanofeatures, and (3) engineer their internal microstructure. The realization of the above capabilities, while appealing, is equally challenging. It is particularly difficult to fabricate single crystalline or uniformly textured functional ceramic nanopatterns in order to probe their intrinsic or enhanced properties. The main bottleneck for the “top-down” approaches stems from the refractory nature and chemical inertness of ceramics.5-8 Several “bottom-up” approaches have been reported for fabrication of various high-quality ceramic one-dimensional nanostructures using solution-based methods.9-11 However, these methods usually render freestanding nanostructures rather than patterns with well-defined size and separation, and the control of crystallinity and orientation of these nanostructures still remains a challenge compared to substrate-based approaches. The miniaturization trend for functional ceramics calls for development of unique * To whom correspondence may be addressed. Fax: 847-467-6573. E-mail:
[email protected]. 10.1021/nl061905z CCC: $33.50 Published on Web 09/26/2006
© 2006 American Chemical Society
patterning and structural engineering techniques to achieve well-defined dimensions of ceramic nanoelements with controlled internal structure. We have recently introduced a facile patterning technique termed “soft” electron beam lithography (“soft-eBL”) for sitespecific fabrication of solid-state nanostructures, with high versatility for patterning not only two-dimensional planar elements12 but also three-dimensional hierarchical features.13 Herein we present soft-eBL-fabricated nanodisk arrays of technologically important magnetic ceramic CoFe2O4 (CFO) and ferroelectric BaTiO3 (BTO) on diverse substrates with well-controlled size and separation. We further demonstrate that below a certain pattern size, the as-deposited amorphous nanodisks can be converted into dense, single-crystal form that exhibits heteroepitaxy with respect to the underlying single-crystal substrate. The single-crystal CFO and BTO nanodisks present a typical faceted-dome shape (with small variation due to different material/substrate combinations). The mechanism of morphology development with the pattern size change is discussed. Finally, localized characterization of the crystallinity, chemical composition, and magnetic behavior of the CFO patterns has been carried out using analytical transmission electron microscopy and magnetic force microscopy. The single-crystal conversion and shape control strategy can be applied to diverse functional oxidesubstrate systems and therefore contribute to study of the novel properties of materials at different length scales and utilizing miniaturized devices through synergistic coupling.
Figure 1. (a) Schematic illustration of the Soft-eBL procedure. (b) AFM topographic image of a 12 × 8 CFO disk array with 1 µm separation patterned using the soft-eBL. (c) Detailed AFM topographic image and cross sectional profile of the annealed CFO disks with 150 nm diameter and 90 nm height. Scale bar ) 1 µm.
The soft-eBL technique as a combination of electron beam lithography and spinning of liquid precursors for patterning the ceramic nanostructures has been reported in the literature.12 This facile approach allows directed patterning on diverse substrates without any etching process. Figure 1a briefly illustrates the scheme of the soft-eBL process. In this study, CFO was patterned on single crystal (100) SrTiO3 (STO) (with -7.4% lattice mismatch), whereas BTO was patterned on (100) SrRuO3-coated SrTiO3 substrate (SRO/ STO) (with -1.5% lattice mismatch between BTO and SRO). Patterning was also performed on electron transparent amorphous SiNx membrane for direct observation under transmission electron microscopy (TEM). The patterned structures were characterized using scanning electron microscopy (SEM, Quanta 600F, FEI Co.), TEM (JEM-2100F, JEOL, Japan), and atomic force microscopy (AFM, Nanoscope III microsope, Digital Instruments). The as-deposited patterns are in the form of amorphous dry gel containing organic residue. In order to crystallize and fully densify the patterns, all patterned samples were annealed at 1000 °C in air. During annealing, the pattern dimension decreases in both vertical and lateral directions, resulting from both the elimination of organic component and the phase change from an open-structured amorphous state to a dense crystalline structure. Parts b and c of Figure 1 are typical AFM images of soft-eBL-patterned CFO arrays on STO substrate after annealing, illustrating the size uniformity of the CFO patterns. The cross-sectional profile in Figure 1c shows that each disk has uniform diameter of about 150 nm and height of about 90 nm. In the following, samples with different pattern diameters on STO substrate were selected in order to understand the influence of pattern size on the morphology and crystallinity of CFO disks. The height of the final patterns for different samples generally falls in the range of 80-100 nm. Parts a-c of Figure 2 are SEM images of CFO disks with 500, 200, and 100 nm diameters, respectively. In Figure 2a, Nano Lett., Vol. 6, No. 10, 2006
the 500 nm diameter patterns appear to break into several isolated crystallites. This tendency of breaking is significantly diminished for smaller patterns. In fact, two types of morphologies are observed to coexist for pattern diameter between 500 and 200 nm: one showing a polycrystalline CFO network with small pores trapped inside, and the other with dense CFO disks. Patterns with 200 nm and smaller diameters (parts b and c of Figure 2b and 2c) are all developed into fully dense nanodisks. In order to obtain explicit evidence for crystallinity of smaller patterns, tapping mode AFM phase images were taken on the 200 nm diameter disks. AFM phase image can provide superior grain boundary contrast especially when grains are smaller than 20 nm.14 Parts d and f of Figure 2 show the typical phase images of 200 nm diameter patterns, where no grain boundary features are observed. Although the original patterns in the e-beam resist are circular disks, the annealed CFO patterns present a faceted shape which is consistent with the Wulff construction for cubic m3m point group symmetry (truncated octahedron with {100} and {111} facets15), as illustrated in Figure 2g. In fact, the sharp angle between the side/bottom surfaces measured from cross-sectional profiles of Figure 2f is around 56-58°, corresponding well to the angle between cubic {100} and {111} planes (57.4°). It suggests the singlecrystal nature of the nanopatterns and their uniform cubeon-cube epitaxial orientation with respect to the substrate. TEM investigation on a plan view sample (CFO 200 nm diameter disk arrays on STO (100) substrate after annealing) provides unambiguous evidence of the single-crystal nature and detailed information on the orientation of the nanodisks. A well-defined pattern shape can be readily observed from parts a and b of Figure 3. A typical selected area diffraction (SAD) pattern obtained from one such disk (Figure 3c) further reveals the cube-on-cube heteroepitaxy between the single-crystal CFO nanodisks and the STO substrate. This heteroepitaxy implies that one magnetocrystalline easy axis (i.e., 〈100〉) of CFO nanopatterns is aligned normal to the substrate plane, which is highly favorable to harness the magnetization direction of such nanopatterns. Similar epitaxial conversion under appropriate heating condition has also been observed for oxide nanofeatures prepared by pulsed laser deposition.16,17 While current attempts for fabrication of textured or epitaxial ferrite thin films or miniaturized structures are mainly focused on high-energy deposition (e.g., pulsed laser deposition, magnetron sputtering, and molecular beam epitaxy)18 with careful monitoring of the film growth, the approach we demonstrate here can serve as an attractive alternative to fabricate highly textured ferrite nanostructures and heterostructures. The single-crystal conversion and shape control strategy in soft-eBL can be readily applied to other functional oxide/ substrate systems. Shown in Figure 4 as an example, the well-defined faceted shape can also be obtained from BTO patterns on SRO/STO substrate. From the originally defined circular disks defined in e-beam resist, BTO patterns have developed into a perfect truncated pyramid shape after annealing. Unlike CFO on STO, BTO patterns do not contain the {111} small side facets, probably due to its different 2345
Figure 2. (a-c) SEM images of CFO disk arrays with 500, 200, and 100 nm diameters, respectively. Scale bar ) 1 µm. (d-f) AFM phase images of CFO disks with 200 nm diameter and cross sectional profile of Figure 2f. The line contrast on the substrate (indicated by arrows) comes from the terraces on (100) STO due to its surface reconstruction during annealing. (g) The same CFO pattern in Figure 2d with colored facets, illustrating the possible crystallographic orientations that are consistent with the Wulff construction for cubic m3m point group symmetry (inset).
Figure 3. (a, b) TEM plan view images of 200 nm diamter CFO disks on (100) STO substrate. (c) SAD pattern obtained from one CFO disk pattern illustrating the cube-on-cube heteroepitaxy between the pattern and the substrate.
Figure 4. BTO patterns on (100) SRO/STO substrate with truncated-pyramid shape. (a, b) AFM topography and phase images from the same 3 by 3 array. (c) Typical SEM image (inset shows a magnified pattern). White arrows indicate the 〈100〉 directions of the substrate. Scale bar ) 1 µm.
surface energy anisotropy and the smaller lattice mismatch with SRO. A similar faceted shape has been reported by Ramesh et al.19 for CFO nanopillars self-assembled in BaTiO3 matrix during pulsed laser deposition, which is a primary consequence of incomplete wetting of CFO on the 2346
substrate due to its large surface energy anisotropy. This suggests that, by changing the material/substrate combination and therefore the interface energy between them, it is possible to tune the equilibrium shape of the nanopatterns accordingly. This can be readily investigated using soft-eBL, given its Nano Lett., Vol. 6, No. 10, 2006
Figure 5. (a, b) STEM images of CFO nanodisks on SiNx membrane. Inset of Figure 5b shows the diffraction pattern taken from the disk with [001] zone axis, illustrating its single-crystal nature. (c) EDS spectra taken on the disk and on the plain membrane.
Figure 6. (a) Topographic image of a 5 × 5 array of 250 nm diameter CFO disk patterns. (b) MFM phase image before magnetization. (c) MFM phase image after magnetization in a 0.3 T external field. Scale bar ) 1 µm.
high versatility in patterning on a very broad range of substrates. More detailed studies on the morphology development of soft-eBL prepared oxide nanopatterns on diverse substrates will be forthcoming. During this investigation we have also noticed that the pattern morphology appears to diverge at a critical diameter of around 200 nm, i.e., larger patterns tend to break into discontinuous crystallites, whereas smaller patterns develop into dense disks. Considering that the epitaxial nanodisks, to an approximation, resemble strained epitaxial islands in typical semiconductor systems,20-22 the morphology change may be qualitatively explained using the analytical results derived for the latter systems. By assuming the total free energy of the system to consist mainly of elastic strain energy and free surface energy of the islands, studies on the mechanics of epitaxial semiconductor islands20,22 point out the following: (1) The strain energy per unit volume stored in the island is a decreasing function of aspect ratio while the normalized surface energy is an increasing function of aspect ratio. This implies that for CFO patterns with large diameters and therefore small aspect ratio, the strain energy can be so high that the patterns tend to crack to partly release the strain energy at the price of increasing the free surface energy. (2) An equilibrium aspect ratio of the islands at which the total free energy density is a local minimum can be determined with a dimensionless parameter R defined by R∼
πγ Mm2A1/2
(1)
where γ is the free surface energy of the island, M is the Nano Lett., Vol. 6, No. 10, 2006
biaxial elastic modulus of the pattern, m is the lattice mismatch between the island and the substrate, and A is the cross-sectional area of the island. The above argument assumes the islands to be completely coherent with the substrate. Although the contributions to the free energy of a nanopattern from crystallographic shape and facets23,24 and possible dislocations formed in the pattern20,25 are not included, the model can still be used to give approximate estimation on the morphology of CFO patterns. For dense and epitaxial 200 nm diameter CFO patterns on STO (100) substrate, the system parameters have approximate values of γ ) 1 J/m2,26 M ) 100 GPa,27 m ) -0.07, and A ) 15000 nm2 for which R is about 0.1. For this value of R, the minimization of the free energy density in the literature20 implies an equilibrium aspect ratio larger than 0.4 (i.e., pattern height g80 nm), which corresponds to the observed dimension of the CFO patterns with the “critical” diameter of 200 nm. The pattern morphology is also related to diffusion kinetics since nucleation and grain growth are both diffusion-controlled. Although detailed investigation on such correlation can be complicated, one may expect that it is more difficult for large patterns to develop into a single grain than for small ones because a significant amount of mass has to be transferred and rearranged for this purpose. The soft-eBL-prepared nanodisks may develop into singlecrystal patterns even on amorphous substrates (such as SiNx), when the pattern dimensions are small. Although the nucleation energy and strain energy on amorphous substrate differ from that on epitaxial layers, the small pattern size still serves as a favorable factor for diffusion and the strain energy release. CFO nanodisks with 50 nm final diameter 2347
were patterned on an electron transparent SiNx membrane window (SiNx thickness of 50 nm), which serves as a test bed for directly probing the localized structural/chemical information of the nanostructures using TEM without any further sample preparation steps. Parts a and b of Figure 55 show the scanning transmission electron microscopy (STEM) images of 50 nm diameter nanodisk patterns, with the inset diffraction pattern taken from one such disk, indicative of its single-crystal structure. A nanosized electron probe was used for chemical analysis of the nanodisks in this sample. The energy dispersive X-ray spectroscopy (EDS) spectrum obtained from the disk and the plain membrane confirms the chemical constituents of the patterns (Figure 5c). The single-crystal conversion reveals the potential of this approach for structural engineering via nanopatterning, i.e., to obtain polycrystalline or single-crystalline ceramic nanostructures through controlling the pattern size and to achieve specific crystallographic orientation of the nanostructure by choosing proper substrates. The CFO patterns on STO substrate were probed using lift-mode MFM as shown in Figure 6, where an area of a 5 × 5 disk array was scanned before and after magnetization in a 0.3 T out-of-plane static magnetic field (pointing into the substrate). The magnetic force response was collected under phase detection mode (parts b and c of Figure 6) and the topography information was recorded simultaneously (Figure 6a). The phase contrast from the disks before magnetization suggests that most CFO disks have an outof-plane spontaneous magnetization component (coming out normal to the substrate). The reversed contrast after magnetization indicates the magnetic configuration change influenced by the external field and thus identifies the active magnetic nature of the nanopatterns. In summary, we have demonstrated a versatile nanofabrication strategy that allows for conversion of functional oxide nanopatterns into their single crystal form and control over their crystalline orientation and morphology. Through synergistic combination of localized characterization with scanning force microscopy and S/TEM, the structural, chemical, and functional identities have been successfully verified. We believe that such a combined approach is imperative to realize the potential of nanopatterned functional oxides for device-based applications. The flexibility of the approach described here allows patterning diverse oxides into nanodevices in order to explore their behavior at different length scales. In addition, since magnetic oxides can be combined epitaxially into vertical heterostructures with perovskite ferroelectrics4 and high-temperature superconductors,28 this technique also opens up the possibility of fabricating all-oxide heterostructures with novel functionalities. Acknowledgment. This work was supported by the NSFNSEC (Award Number EEC-0118025), NSF-MRSEC (DMR
2348
# 0520513), and the U.S. Department of Energy (DOE-BES). The research work was performed in the EPIC and NIFTI facilities of the NUANCE Center at Northwestern University. The NUANCE Center is supported by NSF-NSEC, NSFMRSEC, the State of Illinois, and Northwestern University. References (1) Buhlmann, S.; Dwir, B.; Baborowski, J.; Muralt, P. Appl. Phys. Lett. 2002, 80, 3195-3197. (2) Alexe, M.; Harnagea, C.; Hesse, D. J. Electroceram. 2004, 12, 6988. (3) Moshnyaga, V.; Damaschke, B.; Shapoval, O.; Belenchuk, A.; Faupel, J.; Lebedev, O. I.; Verbeeck, J.; Van Tendeloo, G.; Mucksch, M.; Tsurkan, V.; Tidecks, R.; Samwer, K. Nat. Mater. 2003, 2, 247252. (4) Zheng, H.; Wang, J.; Lofland, S. E.; Ma, Z.; Mohaddes-Ardabili, L.; Zhao, T.; Salamanca-Riba, L.; Shinde, S. R.; Ogale, S. B.; Bai, F.; Viehland, D.; Jia, Y.; Schlom, D. G.; Wuttig, M.; Roytburd, A.; Ramesh, R. Science 2004, 303, 661-663. (5) Ruzmetov D.; Seo Y.; Belenky L. J.; Kim D-M; Ke X.; Sun H.; Chandrasehar V.; Eom C-B.; Rzchowaki M. S.; X., P. AdV. Mater. 2005, 17, 2869-2872. (6) Hiratani, M.; Okazaki, C.; Hasegawa, H.; Sugii, N.; Tarutani, Y.; Takagi, K. Jpn. J. Appl. Phys. 1 1997, 36, 5219-5220. (7) Cofer, A.; Rajora, P.; DeOrnellas, S.; Keil, D. Integr. Ferroelectr. 1997, 16, 53-61. (8) Clemens, S.; Schnell, T.; Hart, A.; Peter, F.; Waser, R. AdV. Mater. 2005, 17, 1357-1361. (9) Pham-Huu, C.; Keller, N.; Estournes, C.; Ehret, G.; Ledoux, M. J. Chem. Commun. 2002, 1882-1883. (10) Ji, G. B.; Tang, S. L.; Xu, B. L.; Gu, B. X.; Du, Y. W. Chem. Phys. Lett. 2003, 379, 484-489. (11) Zhang, Z. T.; Rondinone, A. J.; Ma, J. X.; Shen, J.; Dai, S. AdV. Mater. 2005, 17, 1419-1425. (12) Donthu, S.; Pan, Z. X.; Myers, B.; Shekhawat, G.; Wu, N. G.; Dravid, V. Nano Lett. 2005, 5, 1710-1715. (13) Pan, Z. X.; Donthu, S. K.; Wu, N. Q.; Li, S. Y.; Dravid, V. P. Small 2006, 2, 274-280. (14) Pang, C. H.; Hing, P.; See, A. J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.sProcess., Meas., Phenom. 2002, 20, 1866-1869. (15) Siem, E. J.; Johnson, E. Philos. Mag. 2005, 85, 1273-1290. (16) Ma, W.; Hesse, D. Appl. Phys. Lett. 2004, 84, 2871-2873. (17) Sanchez, F.; Luders, U.; Herranz, G.; Infante, I. C.; Fontcuberta, J.; Garcia-Cuenca, M. V.; Ferrater, C.; Varela, M. Nanotechnology 2005, 16, S190-S196. (18) Suzuki, Y. Annu. ReV. Mater. Res. 2001, 31, 265-289. (19) Zheng, H.; Zhan, Q.; Zavaliche, F.; Sherburne, M.; Straub, F.; Cruz, M. P.; Chen, L.-Q.; Dahmen, U.; Ramesh, R. Nano Lett. 2006, 6, 1401-1407. (20) Johnson, H. T.; Freund, L. B. J. Appl. Phys. 1997, 81, 6081-6090. (21) Kamins, T. I.; Ohlberg, D. A. A.; Williams, R. S.; Zhang, W.; Chou, S. Y. Appl. Phys. Lett. 1999, 74, 1773-1775. (22) Machtay, N. D.; Kukta, R. V. J. Appl. MechsTrans. ASME 2006, 73, 212-219. (23) Medeiros-Ribeiro, G.; Bratkovski, A. M.; Kamins, T. I.; Ohlberg, D. A. A.; Williams, R. S. Science 1998, 279, 353-355. (24) Jalkanen, J.; Trushin, O.; Granato, E.; Ying, S. C.; Ala-Nissila, T. Phys. ReV. B 2005, 72, Art. No. 084103. (25) Denker, U.; Jesson, D. E. Phys. Status Solidi B 2005, 242, 24552461. (26) Lee, H. Y.; Kim, J. S.; Kang, S. J. L. Interface Sci. 2000, 8, 223229. (27) Krupicka, S.; Novak, P. Oxide Spinels; North-Holland Physics Publishing: Amsterdam, 1982; Vol. 3, p 192. (28) Habermeier, H. U.; Cristiani, G. Physica C 2004, 408-10, 864-865.
NL061905Z
Nano Lett., Vol. 6, No. 10, 2006
