MECS is a revolutionary technology that has the potential to impact many aspects of how we live. The availability of small, inexpensive but high capacity MECS devices will increasingly make the distributed processing of mass and energy a reality. However, to realise this vision, we must be able to economically and reliably fabricate large numbers of MECS devices.
While MECS devices do include geometries with microscale dimensions, the development of the technology has emphasised materials and fabrication techniques that are significantly different from conventional silicon-based micromachining (Paul and Peterson 1999; Martin et al. 1999). Key differences include materials, operating conditions, system sizes and feature sizes. Silicon is not a good material for most MECS devices. Silicon is brittle and has a high thermal conductivity, which can be a problem for many MECS applications (Peterson 1999). Most MECS devices are fabricated out of metals or polymers with some work being done on the fabrication of ceramic MECS devices. One challenge is to develop fabrication approaches that are appropriate for the many different materials and operating conditions encountered in MECS devices.
Many MECS devices operate at high temperatures with chemically reactive products. Often microreactors will operate at 500 to 600°C and in some applications, designers would like to operate the microreactors at temperatures approaching 1000°C. Suitable fabrication techniques exist for materials that can operate at temperatures up to 600°C but above 600°C difficulties in microlamination of refractory metals and ceramics is a significant challenge. MECS technology will require better methods for fabricating high temperature and chemically inert microscale geometries. Regarding system size, MECS applications will often require arrays of microscale geometries that cover between 1 and 400 cm2 and this must be at a low cost. Many conventional microfabrication techniques cannot economically process large surface areas. This requires the development of new approaches or a significant improvement in the capabilities of existing microfabrication techniques.
Much of the promise of MECS depends on the availability of a large number of inexpensive distributed components that can compete with the centralised processing of mass and energy. To be competitive, MECS devices must be extremely inexpensive. This will require fabrication techniques that are appropriate for high volume low cost production with dimensional features across the nano, micro and meso-scales. Traditional prototyping techniques such as diffusion bonding will need to give way to continuous processing techniques such as solder paste bonding and other low cost bonding methods.
The most critical dimension in most MECS devices is the height (or width whichever is smaller) of the microchannel though the integration of nano-scale features within microchannels has begun to challenge this assumption. Microchannel dimensions can range from 25 to 250 ^m high and 25 ^m to tens of mm wide. The sizes of microchannels are determined from a variety of perspectives. From a product design perspective, microchannel heights and lengths impact the residence time of fluid molecules within microchannel reactors or heat exchangers as well as the pressure drop across these devices. In addition, particulate matter within bulk fluids sometimes can present challenges in channel blockage and clogging. For example, consider the separation of hydrogen from gasoline for fuel cell applications. Gasoline contains random particles that can make bulk processing difficult in microchannels with heights below about 100 |im. Further, consider the effect of microchannel size on the overall device shape. MECS devices normally consist of a sequence of unit operations designed either for heat transfer or chemical processing. In order to get bulk fluid flows into the microchannels, a series of headers or plenums are needed each successively decreasing in size. Another way of looking at it is that the headers connect the microchannels to the outside world. Since the major value-added function of MECS devices is carried out in microchannels, generally it is desirable to reduce the volume of header with respect to the volume of microchannel. Due to scaling laws, as the height of microchannels decrease in size, the length of those channels also decrease. Given that the header volume stays relatively the same size, at some point the overall header volume begins to dominate the overall device volume leading to awkward device configurations (e.g. automotive radiators). From a fabrication perspective, as the size of microchannel features decrease, tolerances have more effect on device performance. In other words, a 5 |im error will have more impact on fluid dynamics and heat transfer (due to flow maldistribution) in a 25 |im-high channel than in a 250 |im-high channel. If processing could be performed in a silicon platform, tolerances would permit going well below 5 |im. However, most MECS devices are large compared with IC and MEMS devices and, therefore, bulk material costs are significant. Therefore, monocrystalline silicon is not practical. Further, many MECS devices require thermal or chemical properties not available in silicon architectures. Therefore, many MECS applications need to make use of engineering materials.
The tolerances of many conventional manufacturing processes have been found suitable for microchannels making the economical production of MECS devices more feasible. For example, it has been found that reasonable precision of microchannel heights can be achieved at micro and meso-scale dimensions through the lamination of shim stock. The advantage of this approach is that the precision tolerances are controlled by an economical, high-volume process such as rolling. In this case, the scale of channel height (the critical feature) is dictated by the thickness of the shim stock, which can normally be sourced down to 25-pm thick. Further, early results showed that the warpage produced during the diffusion bonding of copper foils limited microchannel dimensions to no smaller than 80 pm high (Krause et al. 1994) though that constraint has since been lifted.
Current microelectronic integrated circuits (IC) are predominately silicon-based. MECS, on the other hand, require the mechanical and thermal properties provided by other materials. Many thermal applications require low thermal conductivity materials to reduce axial heat transfer (Peterson 1998, 1999). Other requirements for subcomponents could be for highly fatigue resistant material for springs or magnetic steels for generator and motor cores. Many of the prevailing MEMS microfabrication technologies are based on silicon, polymers, or electroplated pure metals (having high thermal conductivity). Adapting these MEMS fabrication techniques for the construction of some MECS devices would be difficult to achieve.
A second requirement for MECS construction is the need for high-aspect-ratio features. Microchannel arrays with >30:1 aspect ratios are commonly needed for heat exchangers and regenerators. Other MECS designs may call for a small gap between adjacent sub-components where the gap is maintained for the entire length of the structure. Other MECS requirements call for heterogeneity in fabrication materials where electrical and magnetic sections may require a metal and dielectric sections may require a polymer or ceramic. Furthermore, electronic circuits may be needed in the overall design of MECS to process information. While MEMS fabrication platforms provide excellent electromechanical integration, these capabilities are limited to planar structures as opposed to three-dimensional, bulk structures.
Finally, MECS must be able to offer all of these capabilities at low cost in order to compete with conventional macro-scale energy and chemical conversion devices. The most notable high-aspect ratio MEMS fabrication technology is LIGA (Becker et al 1986; Ehrfeld et al. 1997). In addition to being primarily a polymer forming method, LIGA is dependent upon highly capital-intensive synchrotron X-ray generation. LIGA and its lower cost derivatives all use lithographic techniques for mold making and electrofoming for mold filling. Weaknesses of this approach include limited material selection, limited geometric complexity (two dimensional structures), and inconsistent pattern-transferring methods (Walsh et al. 1996). Other net-shape microfabrication techniques have been exploited including laser-beam, electron-beam, ion-beam, electrochemical, electrodischarge, and mechanical methods for material removal or deposition. However, all of these approaches are either serial in nature and, therefore, lack the capability of economical mass production, involve single layer thin film forming and, therefore, provide limited aspect ratios, or are incapable of producing embedded microchannel geometry. No well-established micromechanical fabrication method currently exists for addressing MECS device fabrication requirements in a low-cost, high-volume manner.
For over ten years, Oregon State University and the Pacific Northwest National Laboratory (Richland, Washington) have been developing microlamination techniques for producing high aspect ratio MECS devices. The fabrication methods being pursued by these groups rely on the patterning and bonding of thin layers of material (laminae) to make monolithic assemblies with embedded microchannel circuits (Fig. 14.2). Microlamination is not a process rather, it is a processing architecture. All microlamination techniques involve three steps: 1) lamina patterning; 2) lamina-to-lamina registration to form a stack; and 3) bonding of the stack into a monolith.
Typical lamina patterning techniques used for metals include photochemical machining, wire electrodischarge machining, electrochemical machining, punching and laser machining (Wegeng and Drost 1994). Photochemical and electrochemical machining have the potential to produce 'blind' features which can be important for reducing the number of laminae per stack or for creating freestanding features such as posts needed to prevent microchannel warpage. Patterning processes for polymers include hot embossing, laser machining, injection molding and punching. In addition to being computer-controlled and, therefore, not requiring expensive, inflexible photomasks, laser machining in polymers has the added advantage of patterning laminae without leaving a burr. Laser burrs in the laser machining of metals can cause difficulties in bonding laminae. Hot embossing and injection molding both provide the ability to produce blind features. Hot embossing offers the ability to produce thinner laminae, which can be important for reducing device volumes. These molding and forming techniques all require tooling which can be produced using more conventional microfabrication techniques. Variants of these polymer-patterning processes have been studied for patterning green tapes for producing patterned ceramic substrates.
Several registration techniques have been employed in microlamination methods including pin, edge, and self-alignment. The technique used depends on the bonding process chosen. The Thermally-Enhanced Edge Registration (TEER) technique has been demonstrated to be an effective technique for registering laminae during thermal bonding processes (Thomas and Paul 2002, 2003). The TEER technique employs the difference in CTE between the bonding fixture and the laminated material to produce a registration force on the laminae at the bonding temperature. By making the bonding fixture from a material that has a lower CTE than the laminae, a clearance will exist between the fixture and the laminae at room temperature, making the loading of laminae simple compared with other mechanical alignment methods. At the same time, TEER has been shown to achieve a layer-to-layer registration accuracy as small as 2 ^m for small microlaminated structures (Thomas and Paul 2002).
Bonding methods that have been explored for metals include diffusion bonding, diffusion brazing, diffusion soldering and solder paste bonding (Kanlayasiri and Paul 2004; Gabriel et al. 2001; Paul et al. 2004). Advantages of diffusion bonding include the ability to work with a homogeneous material. Diffusion brazing and soldering can result in greatly reduced pressure and temperature conditions yielding more robust production methods (less dimensional error). Solder pastebonding offers the promise of improved economics and the ability to integrate electrical circuits with fluidic circuits. Polymer bonding methods have included solvent welding, thermal bonding, adhesive bonding and ultrasonic welding (Paul et al. 2003). Ultrasonic welding can be easily automated but, along with adhesive bonding, can yield poor channel geometries including particulate matter within microchannels. Solvent welding can leave behind residual solvent which may leach out in subsequent device operation. Thermal bonding can require longer cycle times.
Was this article helpful?
Have you ever been envious of people who seem to have no end of clever ideas, who are able to think quickly in any situation, or who seem to have flawless memories? Could it be that they're just born smarter or quicker than the rest of us? Or are there some secrets that they might know that we don't?