1 Vacuum casting, a new answer for manufacturing biomicrosystems M Denoual 1 *, P Mognol 2, and B Lepioufle 1 1 Biomis-SATIE ENS-Cachan antenne de Bretagne, Bruz, France 2 IRCCyN Nantes, France The manuscript was received on 22 July 2004 and was accepted after revision for publication on 20 May 2005. DOI: 10.1243/095440505X32571 Abstract: In this work, the vacuum-casting process is pushed to its dimensional limits thanks to microfabricated original parts. The successful replication of those parts with a slightly adapted vacuum-casting process demonstrates that this technique is suitable for biomicrosystem manufacturing and development. Keywords: vacuum casting, micrometre-scale structures, biomicrosystems 1 INTRODUCTION Microelectromechanical systems (MEMS) or microsystems are considered as a new, rapidly expanding field in research and industry. This field has its origins in microelectronics industry technologies, but for the last few years has been turning towards other materials and other fabrication technologies [1]. The reason for this is that technologies based on the microelectronics industry present two main drawbacks. The first obvious drawback is that microsystems realized through those technologies are limited to silicon-based materials. The standard conditioning of the silicon is also a limitation, 6- to 12-inch wafers with thickness from 500 mm to about 1 mm do not fit all microsystems volume requirements or they result in extra packaging costs. Finally, silicon may not be the right material because of its intrinsic properties. For example, transparency is a key property for bio- or chemical microsystem applications. The second main drawback concerning silicon microfabrication is that the technologies allowing the realization of microsystems such as deep reactive ion etching (RIE) or wafer bonding are still expensive and are not available for mass production. For example, deep etching of silicon is a one-by-one wafer process. The need for non-silicon-based material microsystems has led to the development or adaptation of other technologies. Machining has been pushed to *Corresponding author: Biomis-SATIE, ENS-Cachan antenne de Bretage, Bruz, France. email: denoual@bretagne.ens-cachan.fr the limits (>100 mm). Stereolithography gave birth to microstereolithography [2]. Recently, research and development has turned towards polymer fabrication techniques, guided by the biologists and chemist who demanded transparent, cheap, disposable, and inert microsystems. The adaptation of techniques from the plastics industry resulted in micro hot embossing [3 6] and microinjection moulding. However, these resulting techniques present the same drawbacks as their macro equivalents. A large amount of time and money must be invested before the first part is realized (cost of the tooling) and they only become cost-effective after several tens of thousands of parts have been produced. In addition, the research results presented up to now show some limitations as far as accuracy is concerned [7]. The huge investments required to implement these techniques slow down the development of biochemical microsystems and limit the number of research and development teams working in this field. What is needed is a simple technique that does not require huge investment and that, therefore, makes it possible to develop and manufacture small series of polymer biomicrosystems. This work shows that the vacuum-casting process fits these requirements and meets the need for a low-cost and low-volume polymer microsystem production technique. This paper aims to show the performances that can be achieved with the vacuum-casting technique at the micrometre scale, and which make it suitable for microsystem prototyping and manufacturing. The paper is organized as follows. First, the
2 M Denoual, P Mognol, and B Lepioufle vacuum-casting process with improvements to fit micrometre-scale replication accuracy is presented. Then the performances of the vacuum-casting technique for the replication of micrometre structures, high aspect ratio structures, and submicrometre features are presented and discussed. 2 MATERIALS AND METHODS The original parts used to qualify the vacuum-casting process were obtained through standard microfabrication technologies. Deep silicon etching microfabrication technology was used to manufacture high aspect ratio (up to 1: 50) micrometre-scale structures [8]. In addition, submicrometre-scale structures were obtained through standard microelectronics industry technologies (photolithography). The patterns used for the testing are conventional patterns for microfabrication technique characterization. 6091 resin was used for casting under vacuum in the female mould. First MCP 6091 part B was degassed under vacuum for 10 min. Then, MCP 6091 part A and MCP 6091 part B were mixed in a ratio of 1: 1.8 (A: B) for 1 min. The mixed solution was then cast in the female silicone mould (step 4). After 2 h baking at 70 8C, the resin was hardened and the resin duplicated part was withdrawn from the mould (step 5). 3 RESULTS AND DISCUSSION 3.1 Micrometre structures and high aspect ratio Figure 1 shows the results obtained with the vacuumcasting technology with an anti-adhesive coating and high-quality silicone improvements. The surface quality is comparable with the surface of original part. Moreover, the structures are accurately duplicated. Using this technique, features ranging 2.1 The duplication process The vacuum-casting process consists of the realization of a silicon female mould with an original part followed by resin casting under vacuum to obtain the duplicated parts. The direct application of the standard macro vacuum-casting process leads to poor results at the micro scale. In particular, detailed attention must be paid to the realization of the female mould. Further performances depend on that step. A Teflon-like anti-adhesive layer is preferred to the classical anti-adhesive spray coating on the original part. This Teflon-like layer is deposited in CHF3 plasma by a standard RIE machine. This preparation allows an increased aspect ratio to be achieved at the micrometre scale and prevents the duplication of the anti-adhesive spray droplets that have the same size range (5 100 mm) as the structures to be replicated. Another point where care is needed is the choice of the silicone for the female mould. A high-quality silicone, well known in the microsystem field [9], PDMS (polydimethylsiloxane, Sylgard RTV184 from Dow Corning), was used to make this female mould. This particular choice avoids the surface quality of the duplicated parts being reduced by the sweating that occurs with low- or medium-quality silicones. A Sprue bushing channel and vent were joined to the original part to be duplicated (step 1). A female mould was made of silicone around the original part (step 2). The PDMS prepolymer was first mixed for 2 min at a ratio of 10:1 with the curing agent and then degassed for 15 min before pouring over the original part. The curing time was 1 h at 70 8C. After the bake, the silicone female mould was opened and the original part withdrawn (step 3). The female mould was then ready for casting. MCP Fig. 1 Scanning electron microscopy (SEM) image of duplicated structures. Field of towers (25 mm high and 30 mm in diameter) and set of walls (25 mm high and 12 mm wide). The walls have a slightly negative taper profile due to the microfabrication process used to create them Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture SC03204 # IMechE 2005
Vacuum casting 3 from 10 to 300 mm were successfully replicated. The smallest structures of the original parts were successfully duplicated and the duplicated dimensions were limited only by the dimensions of the original parts. The accuracy is about 1 per cent between the original part and the duplicated part, mainly due to the shrinkage of the resin after baking. The slightly negative taper profiles were also successfully replicated thanks to the elasticity of the mould. Such profiles are not possible with hot embossing or injection moulding since they lead to structural breakdown of the mould or prevent part withdrawal. The aspect ratio of the realized structures varied up to 1:16. Figure 2 shows an example of high aspect ratio structures (towers and wells) successfully replicated. The original in that case part was obtained through deep silicon etching. The visualization of polymer structures using SEM requires that the sample surface is coated with a thin conductive layer (chromium-gold). However a side effect of the high aspect ratio structures is the creation of shadows during the deposition of this layer, which causes Fig. 3 SEM image of resin duplicated patterns: 5-mm diameter and 100-mm high cylinders brightness variations on the images of resin structures compared with the images of silicon structures. Structures with an aspect ratio of 1:20 have also been replicated, as illustrated in Fig. 3. Indeed the replicated cylinders are 5 mm in diameter and 100 mm high. However, the material stiffness seems insufficient to maintain the shape. This indicates that future work is needed on the material choice and processing to improve the aspect ratio. 3.2 Submicrometre structures In order to assess the replication of submicrometre features, the improved vacuum-casting process was used to duplicate oxide layers patterned on silicon. Figure 4 shows an original oxide and silicon part and its resin vacuum-casting replication. The smallest pattern dimensions tested were 0.8 mm and the oxide layer thickness was 300 nm. The results underline that submicrometre structures can be replicated using the improved vacuum-casting process. Features down to 800 nm wide and 300 nm high have been successfully replicated using the improved vacuum-casting process. The limitation here is due to the process environment. The vacuum-casting process is operated in an ambient environment without any particular dust-class control. Therefore some dust is visible on the surface of the duplicated structures. For nanoreplication using the vacuumcasting technique, a clean-room environment is required. Fig. 2 SEM images of high aspect ratio silicon structures: 20-mm diameter and 100-mm high towers 3.3 Polymer technique comparison Hot embossing and injection moulding require huge investments in terms of equipment. Vacuum casting only requires a vacuum chamber and an oven. In addition, the duplicated parts in the case of hot
4 M Denoual, P Mognol, and B Lepioufle Fig. 4 SEM images of a submicrometre oxide dot and its replication in resin below. The dot is 300 nm thick and 2 mm wide embossing and injection moulding are made using a metallic mould. Therefore, moulds with micro-scale definition of structures are required, which means expensive and non-standard fabrication techniques such as LIGA (a German acronym for lithography, electroplating, and moulding) [10 11]. Those moulds become very expensive and may not reach the required dimensions. In contrast, the original part used for the vacuum-casting process can be made out of a wide variety of materials, not only stiff materials, and allows the use of several microfabrication techniques. Processes from the microelectronics industry can be applied to realize silicon microstructure parts whereas silicon is too brittle for microinjection moulding or micro hot embossing. Polymers can also be used to make bulky parts that would be too expensive if made entirely out of silicon. Concerning the duplication performance, improved vacuum casting and micro hot embossing have similar accuracy and aspect ratio and are better techniques than injection moulding when it comes to duplicate microstructures. Table 1 summarizes those characteristics and performances. A cost study has been performed for a specific biochemical microsystem application. The results of the study, illustrated in Fig. 5, stress the cost-effectiveness of this technique: from a few parts to a few thousands parts. For this cost study, the fact that the silicone mould is changed every 40 parts was taken into account. Indeed, the experiments showed decreasing duplication performances after 50 duplicated parts. Therefore, improved vacuum casting meets the need for small series of microstructured systems made of polymers. Because of its high speed to first part and low cost, the adaptation of the vacuum-casting process to microfabrication will lead to a promising technology for MEMS application. This is indeed a suitable technique for rapid prototyping of polymer microsystems dedicated to biology or chemistry when compared with hot embossing or microinjection moulding techniques which require significant investment of time and money. 4 CONCLUSION The performances of the vacuum-casting technique were investigated for the replication of microstructures. According to the experimental results, a conventional vacuum-casting process with minor improvements can be applied to the fabrication of microstructures. High-quality materials are essential to overcome the sweating and adhesion problems found at the micrometre scale. The duplication of classical test structures showed submicrometre Table 1 Comparison between polymer replication techniques Equipment Mould or original part Micro hot embossing Micro-injection moulding Improved vacuum casting Hot embossing machine (US$100 000 300 000) Micro-machined or LIGA-made metal mould Injection moulding machine (US$100 000 300 000) Micro-machined or LIGA-made metal mould Time to first part þþ þ þ þ þ Accuracy þþþ þ þþþ Aspect ratio þþþ þ þþþ Vacuum chamber and oven (US$20 000 30 000) Micro-fabrication technique (micro-stereolithography, micro-machining, etc.), assembly Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture SC03204 # IMechE 2005
Vacuum casting 5 Fig. 5 Part cost versus production volume compared for injection moulding, microfabrication technology, and the vacuum-casting process (down to 300 nm) and high aspect ratio (up to 1:16) performances. The results highlight that vacuum casting fits the requirements for a prototyping technique allowing the development and manufacturing of polymer microsystems. ACKNOWLEDGEMENTS The authors wish to thank Yann Macé for cooperation and discussions during the development of the technique. REFERENCES 1 Gardner, J. W., Varadan, V. K., and Awadelkarim, O. O. Microsensors, MEMS and Smart Devices, 2001 (John Wiley: New York). 2 Varadan, V. K., Jiang, X., and Varadan, V. V. Microstereolithography and other Fabrication Techniques for 3D MEMS, 2001 (John Wiley, New York). 3 Lee, G.-B., et al. Microfabricated plastic chips by hot embossing methods and their applications for DNA separation and detection. Sensors Actuators B, 2001, 75, 142 148. 4 Lin, L., Cheng, T., and Chiu, C. J. Comparative study of hot embossed micro structures fabricated by laboratory and commercial environments. Microsystem Technologies, 1998, 4, 113 116. 5 Shen, X.-J., Pan, L.-W., and Lin, L. Microplastic embossing process experimental and theoretical characterizations. Sensors Actuators A, 2002, 97 98, 428 433. 6 Becker, H. and Heim, U. Hot embossing as a method for the fabrication of high-aspect ratio structures. Sensors Actuators A, 2000, 83, 130 135. 7 Su, Y.-C., Shah, J., and Lin, L. Implementation and analysis of polymeric microstructure replication by micro-injection moulding. J. Micromech. Microengng, 2004, 14, 415 422. 8 Senturia, S. D. Microsystem design, 2000, pp. 67 71 (Kluwer Academic Publishers: Dordrecht). 9 Duffy, D., et al. Rapid prototyping of microfluidic systems in PDMS. Anal. Chem., 1998, 70, 4974 4984. 10 Cheng, Y., et al. Ultra-deep LIGA process. J. Micromech. Microengng, 1999, 9, 58 63. 11 Kim, K., et al. Rapid replication of polymeric and metallic high-aspect ratio microstructures using PDMS and LIGA technology. Microsystem Technologies, 2002, 9, 5 10.