JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006
859
Silicon-Carbide Microfabrication by Silicon Lost Molding for Glass-Press Molds Toru Itoh, Shuji Tanaka, Jing-Feng Li, Ryuzo Watanabe, and Masayoshi Esashi
Abstract—This paper describes two silicon carbide (SiC) microfabrication processes for SiC glass-press molds. One is silicon lost molding combined with SiC chemical-vapor deposition (CVD) and SiC reaction sintering (RS). The other is silicon lost molding combined with SiC CVD and SiC solid-state reaction bonding (SSRB). In both of these processes, an original pattern on a silicon substrate is transferred to a CVD SiC film, and then the film is backed by bulk SiC to obtain rigidity and robustness against pressing force. Finally, the silicon substrate is etched away to release a SiC mold. In the process using SiC CVD and RS, an original pattern on a silicon substrate was transferred to a SiC mold, but the surface , and worse than roughness of the SiC mold was 0.05–0.08 required by the glass-press mold. This was caused by the transformation of amorphous SiC to polycrystalline SiC in RS, which was confirmed by the X-ray diffraction (XRD) data of the CVD SiC film before and after RS. In the process using SiC CVD and SSRB, the surface of the SiC mold was smooth (0.004–0.008 ) without the crystallization of the amorphous CVD SiC film. The SiC mold was pressed to Pyrex glass to demonstrate its high-temperature strength. The Pyrex glass was deformed by the SiC mold at 850 without a void, and no significant deformation of the SiC mold was observed. [1572]
m Ra
m Ra
C
I. INTRODUCTION ILICON CARBIDE (SiC) has heat resistance, chemical inertness, and high hardness, and its microfabrication technology has various potential applications in harsh environments [1]. One of the promising applications is a micromold to press various glasses at higher temperature . Glass press technology is being used to mass produce nonspherical lenses, and it can be also used for microoptics, microreactors, microelectromechanical systems (MEMS), etc., if microstructures can be formed on molds. Especially, microoptics such as light guides, pick-up optics, and microlens arrays are important due to the continuous development and widespread use of optical communication and data storage. Currently, such microoptics are made from plastics by injection molding or press molding at low cost. The optical characteristics of plastics are, however, inferior to glasses. For example, the dispersion and double refraction of the plastics often become problems for optical
S
Manuscript received April 15, 2005; revised August 17, 2005. This work was supported by the Iwate Foundation under Japan Science and Technology Corporation (JST) through the Regional Science Promoter Program. Subject Editor M. Mehregany. T. Itoh, S. Tanaka, and M. Esashi are with the Department of Nanomechanics, Tohoku University, Aoba-ku, Sendai 980-8579, Japan (e-mail:
[email protected]. tohoku.ac.jp). J.-F. Li is with the Department of Material Processing, Tohoku University, Aoba-ku, Sendai 980-8579, Japan. He is also with the Department of Materials Science and Engineering, Tsinghua University, Beijing 100084, China. R. Watanabe is with the Department of Material Processing, Tohoku University, Aoba-ku, Sendai 980-8579, Japan. Digital Object Identifier 10.1109/JMEMS.2006.872231
communication. Thus, the demand for the micromolding of optical glass is increasing. In glass press molding, micropatterns on a mold are transferred to glass around its softening temperature. For example, the press molding of optical glass (BK7) and quartz glass is perand 1400 , respectively, [2]. formed approximately at 700 In such high temperature process, SiC maintains hardness and inertness to some extent, and can be used as the mold material with glassy carbon antisticking coating [2]. Conventional microfabrication techniques, however, have limitations, when applied to SiC due to its hardness and chemical inertness. For example, grinding and electric discharge machining have the limitation of shape resolution by tool size, and suffer from the rapid wear of tools. The etching of SiC is also limited. The reactive ion etching (RIE) [3]–[9] and deep RIE [10], [11] of SiC were studied, but it is generally difficult to mirror-finish the surface. Isotropic and crystal-orientation-dependent wet etching cannot be used for SiC unlike silicon to microfabricate various features. Rajan et al. [12] reported silicon lost molding combined with SiC chemical vapor deposition (CVD) for SiC microfabrication with shape resolution on the order of micrometers. They developed an excellent CVD system, which enables the rapid deposition of low-stressed thick SiC films. However, it is generally difficult to deposit low-stressed SiC films with thickness of a few hundreds micrometers or thicker at a practical deposition rate by CVD. Tanaka et al. [13] developed “microreaction sintering process,” which combines silicon micromachining, reaction sintering (RS) by hot isotropic pressing (HIP) and silicon lost molding. This process allows us to produce millimetersized SiC parts with microstructures, but the surface is too rough (2–4 ) to be used for glass-press molds. In spite of these problems, we thought that the silicon lost molding was useful to fabricate micromolds for glass press, because we can use well-established silicon microfabrication technology to produce original patterns. In this paper, we developed the fabrication processes of SiC micromolds using lithography-based technology and silicon lost molding. II. MICROFABRICATION PROCESSES OF SIC FOR GLASS-PRESS MOLDS To fabricate a glass-press mold using SiC as a material, a high-precision, high-resolution SiC microfabrication process which can mirror-finish the surface is needed. In addition, the mold pattern should be formed on a thick SiC bulk, because the mold must be rigid and robust to stand pressing force. To satisfy the requirements, we have developed two microfabrication processes shown in Fig. 1. One is silicon lost molding combined with SiC CVD and SiC RS [Fig. 1(a)], and the other is silicon lost molding combined with SiC CVD and SiC solid-state reaction bonding (SSRB) [Fig. 1(b)]. In both of these processes,
1057-7157/$20.00 © 2006 IEEE
860
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006
Fig. 2. APCVD system for SiC deposition.
Fig. 1. Microfabrication processes for SiC molds. (a) Silicon lost molding combined with SiC-CVD and RS. (b) Silicon lost molding combined with SiC-CVD and SSRB.
a SiC film is deposited on an original pattern fabricated on a silicon substrate, and then backed with bulk SiC by SiC RS or SiC SSRB. Finally, the silicon substrate is etched away to reveal the CVD SiC surface. Both processes are based on the silicon lost molding, and can produce SiC microstructures on a thick SiC bulk. In this paper, we fabricated SiC micromolds . which have triangular-section gratings with pitch of 5–20 A. Silicon Lost Molding Combined With SiC CVD and RS 1) Experimental Method: A silicon mold, which has tri, is fabricated angular-section gratings with pitch of 5–20 by a standard silicon wet etching process using tetra-methyl ammonium hydroxide (TMAH). A SiC film is deposited using a home-made atmospheric pressure CVD (APCVD) system. Fig. 2 shows the diagram of the APCVD system. It consists of a 100-mm diameter double-walled quartz tube, a SiC-coated graphite suscepter, an inductively heated coil, a water cooling line, and gas lines. The source gas is tetra-methyl silane (TMS) and the carrier gas is hydrogen. In this experiment, a in 300 min. 30- -thick SiC film was deposited at 1100 The TMS flow rate is 0.5 mL/min, and the hydrogen flow rate is 2 L/min. Fig. 3 shows the cross section of the silicon mold covered with a 3- -thick SiC film. The film is amorphous SiC, confirmed by the X-ray diffraction (XRD) data shown in Fig. 4. Fig. 5 illustrates the preparation process of SiC RS for backing the SiC film. This process was done, basically following the method reported in a previous paper [13]. First, the silicon mold is set in a silicon frame and filled with a material powder ( -SiC: 60 wt%, graphite: 30 wt%, phenol resin: 10 wt%) by cold isotropic pressing (CIP). Next, the frame is wrapped in silicon powder and subsequently in boron nitride (BN) powder by CIP. Finally, the sample is encapsulated in
m
Fig. 3. Cross section of the silicon mold covered with a 3- -thick SiC film.
vacuum using a Pyrex glass tube. The BN powder prevents the sample from reacting with the glass tube. SiC RS was performed by HIP at 100 MPa and 1700 for 2 h. Fig. 6 shows the profiles of temperature and pressure during the HIP process. In this step, the silicon powder melts, infiltrates into the material powder and reacts with graphite powder to generate SiC. The original SiC powder is bonded with this newly generated SiC, and dense SiC backing is realized. After the HIP process, the sample is taken out from the capsule. The silicon frame and the silicon mold are etched away to release a SiC mold using the mixture of hydrofluoric acid and nitric acid (1:2). 2) Results: Fig. 7 shows the cross section of a fractured SiC RS sample before silicon etching. The CVD SiC film is successfully backed with reaction-sintered SiC, and the shape of the triangular-section grating is maintained. Fig. 8 shows the surface of the SiC mold after silicon was etched away. The shape of the silicon mold is transferred to the SiC mold. The surface roughness was 0.05–0.08 , which is approximately one fiftieth
ITOH et al.: SILICON-CARBIDE MICROFABRICATION BY SILICON LOST MOLDING FOR GLASS-PRESS MOLDS
861
Fig. 5. Preparation process for SiC RS.
Fig. 4. CVD SiC film data. (a) XRD. (b) XPS.
of that fabricated by our previous process without the CVD film [13], but worse than required by the glass-press mold. The close-up view in Fig. 8 suggests that the rough surface was caused partly by the transformation of amorphous SiC to polycrystalline SiC during the HIP process, and this was confirmed by XRD patterns of the CVD SiC film before and after the HIP shown in Fig. 9. Another possible reasons of the surface roughing is the reaction between the CVD SiC film and the silicon mold during the HIP process. The reaction might result in a silicon-rich SiC surface which could be etched away in wet etching. B. Silicon Lost Molding Combined With SiC CVD and SSRB 1) Experimental Method: First, a silicon mold is fabricated in the foresaid process. Next, a 50- -thick SiC film is deposited on the silicon mold using the APCVD system, and the surface is polished to be mirror-finished using a 1–3 diamond slurry and a Kemet lapping plate, which is rotated at 100 rpm. The SiC film is backed with a 500- -thick SiC ceramic substrate by SSRB [14]. In this paper, the bonding was performed using a 0.5- -thick nickel interface layer at 900 and a pressing pressure of 0.1 MPa for 0.5 h using a simple hot-press system shown in Fig. 10. The nickel interface layer diffuses and reacts with SiC, realizing mechanically and thermally stable bonding [14]. After bonding, the silicon mold is etched away to release a
Fig. 6. Profiles of temperature and pressure during the HIP process.
SiC mold using the mixture of hydrofluoric acid and nitric acid (1:2). 2) Results: Fig. 11 shows the released SiC mold. The CVD film was completely backed with the SiC ceramic substrate by ) comSSRB, and its surface was smooth (0.004–0.008 pared with the CVD SiC film backed by SiC RS. The crystallization of the amorphous SiC does not occur during the bonding process at 900 . The SiC mold was pressed to a glass to confirm its high-temperature strength. Using the hot-press system, the SiC mold was and 1 MPa. In pressed to a 1-mm-thick Pyrex glass at 850 this experiment, the SiC mold was not separated from the Pyrex glass due to the restriction of the experimental setup. Fig. 12 shows the interface between the SiC mold and the pressed glass. The Pyrex glass was deformed by the SiC mold without a void, and no significant deformation of the SiC mold was observed.
862
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006
Fig. 7. Cross section of a fractured SiC RS sample before silicon etching.
Fig. 9. XRD patterns of the CVD SiC film. (a) Before HIP. (b) After HIP.
Fig. 8. Surface of the SiC mold produced by RS.
This result encourages the application of this SiC mold to glass press up to 850 , which is higher than the softening temperature of Corning 7740 Pyrex glass (820 ) and BK7 optical glass (718 ). III. CONCLUSION Two SiC microfabrication processes for SiC glass-press molds were developed. One is silicon lost molding combined with SiC CVD and SiC RS. The other is silicon lost molding combined with SiC CVD and SiC SSRB. In both of these processes, an original pattern on a silicon substrate is transferred to a CVD SiC film, and then the film is backed by bulk SiC to obtain rigidity and robustness against pressing force. Finally, the silicon substrate is etched away to release a SiC mold. In the process using SiC CVD and RS, an amorphous SiC film was deposited on a micromachined silicon mold, and backed with RS SiC by HIP at 100 MPa and 1700 . The surface
Fig. 10. Schematic of the hot-press system.
, and worse roughness of the SiC mold was 0.05–0.08 than required by the glass-press mold. This was caused by the transformation of amorphous SiC to polycrystalline SiC during the HIP process, which was confirmed by the XRD patterns of the CVD SiC film before and after the HIP process. In the process using SiC CVD and SSRB, a CVD SiC film on the micromachined silicon mold was bonded with a SiC ceramic substrate using a nickel interface layer at 900 and a
ITOH et al.: SILICON-CARBIDE MICROFABRICATION BY SILICON LOST MOLDING FOR GLASS-PRESS MOLDS
Fig. 11. Released SiC mold produced by SSRB.
863
[9] F. A. Khan and I. Adesida, “High rate etching of SiC using inductively coupled plasma reactive ion etching in SF -based gas mixture,” Appl. Phys. Lett., vol. 75, pp. 2268–2270, 1999. [10] P. Chabert, “Deep etching of silicon carbide for micromachining applications: Etch rates and etch mechanism,” J. Vac. Sci. Technol. B, vol. 19, pp. 1339–1345, 2001. [11] S. Tanaka, K. Rajanna, T. Abe, and M. Esashi, “Deep reactive ion etching of silicon carbide,” J. Vac. Sci. Technol. B, vol. 19, pp. 2173–2176, 2001. [12] N. Rajan, M. Mehregany, C. A. Zorman, S. Stefanescu, and T. P. Kicher, “Fabrication and testing of micromachined silicon carbide and nickel fuel atomizers for gas turbine engines,” J. Microelectromech. Syst., vol. 8, pp. 251–257, 1999. [13] S. Tanaka, S. Sugimoto, J.-F. Li, R. Watanabe, and M. Esashi, “Silicon carbide micro-reaction-sintering using micromachined silicon molds,” J. Microelectromech. Syst., vol. 10, pp. 55–61, 2001. [14] K. Bhanumurthy and R. Schmid-Fetzer, “Solid-state reaction bonding of silicon carbide (HIP-SiC) below 1000 C,” Mater. Sci. Eng., vol. A220, pp. 35–40, 1996.
Toru Itoh received the B.E. and M.E. degrees from Tohoku University, Sendai, Japan, in 2001 and 2003, in mechatronics and precision engineering, respectively. He has been an engineer with Alps Electric Co., Ltd., since 2003.
Fig. 12. Interface between the SiC mold and the pressed Pyrex glass.
pressing pressure of 0.1 MPa. The surface of the SiC mold was smooth (0.004–0.008 ) without the crystallization of the amorphous CVD SiC film. The SiC mold was pressed to Pyrex glass to demonstrate its high-temperature strength. The Pyrex glass was deformed by the SiC mold at 850 without a void, and no significant deformation of the SiC mold was observed. From this result, we conclude that this process has potential for making a SiC mold for glass press. REFERENCES [1] M. Megregany, C. A. Zorman, N. Rajan, and C. H. Wu, “Silicon carbide MEMS for harsh environments,” in Proc. IEEE, 1998, vol. 86, pp. 1594–1610. [2] H. Maehara and H. Murakoshi, “Quartz glass molding by precision glass molding method,” (in [in Japanese]) Trans. Inst. Elect. Eng. Japan, vol. 122-E, pp. 494–497, 2002. [3] J. Sugiura, W.-J. Lu, K. C. Cadien, and J. Steckl, “Reactive ion etching of SiC thin films using fluorinated gases,” J. Vac. Sci. Technol. B, vol. 4, pp. 349–354, 1986. [4] A. J. Fleischman, C. A. Zorman, and M. Mehregany, “Etching of 3C-SiC using CHF =O and CHF =O =He plasmas at 1.75 Torr,” J. Vac. Sci. Technol. B, vol. 16, pp. 536–539, 1998. [5] P. H. Yih and A. J. Steckl, “Residue-free reactive ion etching of 3C-SiC and 6H-SiC in fluorinated mixture plasmas,” J. Electrochem. Soc., vol. 142, pp. 2853–2860, 1995. [6] G. F. McLane and J. R. Flemish, “High etch rates of SiC in magnetron enhanced SF plasmas,” Appl. Phys. Lett., vol. 68, pp. 3755–3757, 1996. [7] W. Reichert, D. Stefan, E. Obermeier, and W. Wondrak, “Fabrication of smooth -SiC surfaces by reactive ion etching using a graphite electrode,” Mat. Sci. Eng., vol. B46, pp. 190–194, 1997. [8] C. Richter, K. Espertshuber, C. Wagner, M. Eickhoff, and G. Krötz, “Rapid plasma etching of cubic SiC using NF =O gas mixtures,” Mat. Sci. Eng., vol. B46, pp. 160–163, 1997.
Shuji Tanaka received the B.E., M.E., and Dr.E. degrees, all in mechanical engineering, from The University of Tokyo, Tokyo, Japan, in 1994, 1996, and 1999, respectively. From 1996 to 1999, he was a Research Fellow of the Japan Society for the Promotion of Science. He was a Research Associate with the Department of Mechatronics and Precision Engineering, Tohoku University, Sendai, Japan, from 1999 to 2001, and an Assistant Professor from 2001 to 2003. He is currently an Associate Professor with the Department of Nanomechanics, Tohoku University. He is also a Fellow of the Center for Research and Development Strategy, Japan Science and Technology Agency. His research interests include power MEMS, MEMS technology for harsh environments, and microturbo machinery.
Jing-Feng Li, photograph and biography not available at the time of publication.
Ryuzo Watanabe, photograph and biography not available at the time of publication.
Masayoshi Esashi received the B.E. and Dr.E. degrees in electronic engineering from Tohoku University, Sendai, Japan, in 1971 and 1976, respectively. He was a Research Associate with the Department of Electronic Engineering, Tohoku University, from 1976 to 1981, and an Associate Professor from 1981 to 1990. He was a Professor with the Department of Mechatronics and Precision Engineering, Tohoku University, from 1990 to 1998, and moved to New Industry Creation Hatchery Center, Tohoku University, until 2004. He is currently a Professor with the Department of Nanomechanics, Tohoku University. His research interests include MEMS, integrated sensors, micromachining technology, and packaging.
