Department of Materials Science and Engineering, National Taiwan University, Taipei 106, Taiwan 共Received 31 May 2006; accepted 18 October 2006; published 4 December 2006兲
In this article, the authors demonstrate an imprint method for patterning ferroelectric films. In contrast to conventional nanoimprint lithography, the patterned mold is directly imprinted in a ferroelectric film or a metal/ferroelectric film bilayer structure. In general, direct imprint in a ferroelectric or metal film needs ultrahigh pressure or temperature to form patterns. In this article, the authors improve the direct imprint processes by using a sharp mold and an underlying soft gel film for the reduction of the imprint pressure and temperature. The imprint pressure can be reduced to be compatible with the conventional nanoimprint instrument. The authors also successfully use the metal/ferroelectric bilayer structure to overcome the pattern flattened problem in a gel film. The cover metal layer can also be the upper conductive layer in the ferroelectric application. For direct contact of the metal film with mold, no surfactant should be coated on the surface of mold. It also indicates that no mold-rework processes are necessary for this direct imprint ferroelectric film method. © 2006 American Vacuum Society. 关DOI: 10.1116/1.2395958兴
I. INTRODUCTION
Ferroelectric films have been investigated for applications as sensor, actuator, nonvolatile memory, and optoelectronic devices due to their high piezoelectric and ferroelectric prop-erties. Patterning varied profiles of ferroelectric films with submicrometer scale is important for the microelectrome-chanical systems and optoelectronic applications. Many kinds of ferroelectric films, such as Pb共Zr,Ti兲O3 共PZT兲, are not easy to pattern by conventional semiconductor processes. In general, ferroelectric materials are patterned by lift-off, focused ion beam, and wet-etching or dry-etching processes that are generally complicated and should process with strict conditions.1–3 Recently, Wang et al.4 introduced a two-step etching process, using buffered HF acid in the first step and HCl: H2O in the second step, to etch PZT thin film. How-ever, significant undercutting and brim damage are observed in the achieved PZT pattern. Similar results are also found in the research by Ezhilvalavan and Samper in 2005.5
Nanoimprint lithography共NIL兲, a potential candidate for the next generation lithography technology, has performed rapid, large-area, and low-cost technology for the polymer structures. The standard NIL technique is used in a thermo-plastic resist as shown in Fig. 1共a兲.6
The NIL defines patterns by physical deformation of deformable polymer materials 共resist兲 by adding temperature above their glass transition temperate共Tg兲. After the removal of the mold, the pattern is transferred to underlying substrates by etching processes. Di-rectly patterning underlying materials without etching pro-cesses is desired for rapid propro-cesses. This concept can be
carried out in a silicon substrate by laser-assisted directly imprint.7Using this technique, it needs high power excimer laser 共XeCl, 1.6 J cm−2兲 to melt silicon for embossing and applies external pressure at the same time. However, it is not easy to execute in the common procedure and equipment.
Recently, a process for direct imprint of metal films was reported.8 This experiment needs to be carried out with ul-trahigh pressure共several hundreds of megapascals兲 by using an oil press imprint instrument. But such processes with ul-trahigh pressure are not desirable since they would damage the underlying substrates or devices. Furthermore, the ce-ramic films such as ferroelectric films are harder than metal films and they should apply higher pressure by a direct im-print process. In this article, we demonstrate a direct imim-print method for patterning ferroelectric films with low pressure 共艋20 MPa兲 and low temperature. As shown in Fig. 1共b兲, we improve the direct imprint processes by using a sharp mold and soft gel film for the reduction of the imprint pressure. The imprint pressure of our method is only about 10% as compared with the previous direct imprint method.8,9
The ferroelectric properties are generally caused by ap-plying external electric field. As shown in Fig. 1共c兲, we also demonstrate the direct imprint in a metal/ferroelectric film bilayer structure. The top metal film of the bilayer metal/ ferroelectric structure can be as the upper electrode. Other-wise, the issue during conventional NIL process is that the mold is stuck on the imprinted films, and one needs to coat suitable surfactant on the surface of mold to isolate the sticky film. Therefore, we demonstrate that the metal/ferroelectric film bilayer changes the surface properties between the mold and gel layer and tunes the separate condition between the
imprinted film and mold. We find that no mold-rework pro-cesses are necessary for this bilayer NIL method.
II. EXPERIMENT
We fabricated patterned ferroelectric PZT films by imprint gel films, which were prepared by metal organic decom-position 共MOD兲 solution. The composition of PZT film is Pb1.1共Zr0.52, Ti0.48兲O3 and the 10% excess of Pb is used as the compensation for PZT loss during high temperature annealing. MOD precursor used in our article was a com-mercial solution 共Kojundo, Japan兲 and fabricated by dis-solving stoichiometric amounts of Pb-ethylhexanoate, Ti-isopropoxide, and Zr-n-propoxide in ethyl-hexanoic acid and xylene. The PZT films were deposited on silicon and fused silica substrates by spin coating共500 rpm for 10 s, 2000 rpm for 20 s兲 with the thickness of about 200 nm. After the depo-sition process, the films from MOD-based precursor were dried for 5 min at 50 ° C to remove organic solvent and formed the PZT gel films. Silicon molds used in our experi-ments were fabricated by using electron beam lithography 共Leica, Weprint-200兲 followed by the reactive-ion-etching process. The high-density-plasma reactive-ion-etching 共HDP-RIE兲 system 共Duratek, Multiplex Cluster兲 with induc-tively coupled plasma 共ICP兲 sourced was used to fabricate the hexagonal pyramid molds.10,11The imprint pressure ap-plied was about 10– 20 MPa at room temperature during im-print processes. After imim-print, the patterned gel films were first dried at 120 ° C for 30 min and pyrolyzed at 450 ° C for 30 min in air. Finally, the samples were annealed at 650– 800 ° C for 60 min in oven to obtain the perovskite phase. The structures could be identified by low-angle x-ray analysis 共PANalytical X’Pert Pro, Cu K␣兲. The images and surface profiles of patterned metal films were observed by scanning electron microscope 共JEOL, JSM6500F兲 and atomic field microscope 共NT-MDT, P-47兲, respectively. The hardness was measured by the nanoindenter共CSIRO, UMIS II兲 with a Berkovich-type diamond tip. At each test, the
load-ing speed was adjusted to keep 30 s loadload-ing time, 2 s delay at peak load with 1 mN, and 30 s unloading time. The hard-ness was obtained through dividing the load by the area of the residual indents.
III. RESULTS AND DISCUSSION
Figure 2 shows the x-ray diffraction spectra of PZT films. The PZT films did not form the complete perovskite phase after the pyrolysis process共450 °C兲. When we increase the temperature, the diffraction peaks of perovskite phase of the PZT films appeared. The PZT films begin to be crystallized at 600 ° C. At higher temperature, the PZT films become denser and form the more complete perovskite phase. The better ferroelectric properties can be induced from the perov-skite phase.12,13
In order to reduce the requirement of imprint pressure, the sharp mold is used to increase the tip pressure. Figure 3共a兲 shows the silicon mold used in our experiments that is fab-FIG. 1. Schematic diagrams of imprint technologies:共a兲 conventional nanoimprint lithography,共b兲 direct imprint on a ferroelectric film, and共c兲 direct imprint on a metal/ ferroelectric film bilayer structure.
ricated by using the electron beam lithography followed by HDP-RIE process. Figure 3共b兲 shows the cross section of the sharp mold fabricated by the optimized etching process. The HDP-RIE system with ICP source has a chamber surrounded by rf coils and a rf bias provided for the substrates. For the fabrication of sharp mold, we use the 400 W rf-bias power that would increase ion bombardment on the surface of the mold.
During the imprint process, the imprint pressure is depen-dent on the relative properties between molds and ferroelec-tric thin films. The most important characteristic is the hard-ness of molds and imprinted films. Figure 4 shows a comparison of hardness between molds and the ferroelectric films after different temperature treatments. At room tem-perature 共25 °C兲, the PZT thin film contains much organic solvent and the films formed wet-gel films. When the hard-ness of the wet-gel film with little viscosity was measured, the tip of the nanoindentation system would directly contact to the underlying silicon substrate. As the baking temperature rises to 120 ° C, much solvent of PZT wet-gel films is evapo-rated. The PZT gel film with little solvent would cause the plastic deformation during the tip contact with the PZT film that would decrease the hardness. As temperature gradually increases, the ferroelectric films gradually form ceramic
films with large hardness. We find that the largest hardness difference between the silicon mold and PZT film is achieved after baking at 120 ° C. Therefore, we directly pattern ferro-electric films after baking at 120 ° C for the reduction of imprint pressure.
Generally a metal layer should be coated on a tric film for applying external electric field to get ferroelec-tric properties. Therefore, the direct imprint Au/PZT bilayer structure was also demonstrated in this article. As shown in Fig. 4, the hardness of gold film is lower than a silicon mold and it may be patterned after imprinting processes.
Figure 5 shows the patterned ferroelectric cells obtained by imprinting a single-layer PZT gel film with a hexagonal pyramid silicon mold. As shown in Fig. 5共a兲, the PZT film does not deform markedly the profile by applying the imprint pressure of 14 MPa. Because of the characteristics of a gel film, the imprinted profiles become flattened as the mold leave from the surface of PZT film. As the applied pressure is increased to 20 MPa, the patterns become clear and the depth is only about 60 nm as shown in Fig. 5共b兲. For the flattened effect of a gel film, the imprint depth is much smaller than the step height of the mode.
FIG. 3. SEM images of hexagonal pyramid mold:共a兲 top view and 共b兲 cross section.
FIG. 4. Hardness of different material structures at different temperatures. 共–䊏–兲 PZT/Si, 共–쎲–兲 Au/Si, and 共–䉱–兲 Si.
FIG. 5. SEM images of PZT gel films imprinted by hexagonal pyramid mold under共a兲 14 MPa and 共b兲 20 MPa pressures.
As shown in Fig. 6, the PZT gel film is adhered on the silicon mold after imprint processes. Some of the patterns are destroyed and it is difficult to remove these patterns from the mold especially for the ceramic gel film. Therefore, changing the surface properties between the mold and gel layer, such as adding a surfactant layer, to tune the separate condition is essential.
As shown in Fig. 1共c兲, we deposit a gold film on a spin-coated PZT film to form a Au/PZT bilayer structure. Further-more, we use the same hexagonal pyramid mold to imprint pattern on the bilayer structure with the pressure of
14– 20 MPa at room temperature. Figure 7 shows the scan-ning electron microscope 共SEM兲 images of imprinted Au/ PZT bilayer structure under different imprint pressures. Compared with the single-layer PZT gel film, the bilayer PZT profile is much clearer and deeper. Figures 5 and 7 indicate that the depths are markedly different under the same imprinting pressure. The metal film is used for the upper conductive layer to apply the electric field to induce the ferroelectric properties. Moreover, the metal film helps the PZT gel film to be shaped and also helps prevent the pattern from being flattened during imprint processes. We also find that the metal film can solve the problem of the gel film being stuck on the silicon mold. Figure 8 shows the atomic force microscope 共AFM兲 topography of imprinted Au/PZT bilayer structure under different pressures. As shown in Fig. 8共a兲, we can find that the depth is about 90 nm on the bilayer PZT structure after imprint with 14 MPa pres-sure. As the imprint pressure is increased to 20 MPa, the depth of the dent is increased to 300 nm, which is almost equal to the thickness of bilayer PZT structure.
IV. CONCLUSION
In this article, we demonstrate an imprint method for pat-terning ferroelectric films. In contrast to conventional nanoimprint lithography, the patterned mold is directly im-printed in a ferroelectric films or a metal/ferroelectric film bilayer structure. Compared with the single-layer PZT gel film, the bilayer PZT profile is much clearer and deeper. We also improve the direct imprint processes by using a sharp mold or an underlying soft gel film for the reduction of the FIG. 7. SEM images of Au/PZT structures imprinted by hexagonal pyramid
mold under共a兲 14 MPa and 共b兲 20 MPa pressures.
FIG. 8. AFM topography of imprinted Au/PZT structure with hexagonal pyramid mold under 共a兲 14 MPa and 共b兲 20 MPa pressures.
be the upper conductive layer in the ferroelectric application. For direct contact of the metal film with mold, no surfactant should be coated on the surface of mold. It also indicates that no mold-rework processes are necessary for this direct im-print ferroelectric film method.
ACKNOWLEDGMENTS
The authors are very thankful to the National Science Council, Taiwan, R.O.C. for supporting this study under Project Nos. NSC94-2215-E-002-026 and NSC94-2216-E-002-022.
S. Ezhilvalavan and V. D. Samper, Appl. Phys. Lett. 86, 072901共2005兲. 6S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Science 272, 85共1996兲. 7S. Y. Chou, C. Keimel, and J. Gu, Nature共London兲 417, 835 共2002兲. 8H. Yoshihiko, U. Toshihiko, K. Tomohiro, and M. Takashi, Proc. SPIE
74, 5220共2003兲.
9C. Harnagea, M. Alexe, J. Schilling, J. Choi, R. B. Wehrspohn, D. Hesse, and U. Gosele, Appl. Phys. Lett. 83, 1827共2003兲.
10M. Notomi, T. Tamamura, T. Kawashima, and S. Kawakami, Appl. Phys. Lett. 77, 4256共2000兲.
11H. L. Chen, H. F. Lee, W. C. Chao, C. I. Hsieh, F. H. Ko, and T. C. Chu, J. Vac. Sci. Technol. B 22, 3359共2004兲.
12I. N. Lin, K. C. Hsieh, K. Y. Lee, and N. H. Tai, J. Eur. Ceram. Soc. 24, 975共2004兲.
