4
Enabling Manufacturing Technologies

Much of what is called manufacturing science applies to a very wide range of products and size scales. Manufacturability and quality control, which are closely related, vary widely with the type and specifics of a product, whether for the military or for mass markets. Production of microscale surface finishes on a material (for example, a polished metal surface on an aircraft engine turbine blade) is certainly different from making parts for MEMS devices on a similar size scale (for example, a microaccelerometer with submicrometer feature sizes). Similarly, the production of nanocrystalline materials is generally very different from the manufacture of possible molecular electronic devices. In this chapter, the committee considers the general aspects, along with some particular aspects, of manufacturing products whose performance depends on structure, material, or chemistry on the micro- and nanoscales.

The materials, components, subsystems, systems, and platforms used by the military are mainly purchased from industry. Rarely is military hardware manufactured by the Department of Defense. Hence, industry-based manufacture of what the military uses is of central importance. This will be no less true for hardware made by micro- and nanotechnologies than it is for the alloys, antennas, radars, missiles, and airplanes that are now being employed by the military. The role of various communities in acquiring, maintaining, and employing military hardware is summarized in Figure 4-1.

FABRICATION (PATTERNING) APPROACHES

Integrated circuit manufacturing is a top-down process where the starting point is a flat wafer onto which patterns are defined and created by both additive



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies 4 Enabling Manufacturing Technologies Much of what is called manufacturing science applies to a very wide range of products and size scales. Manufacturability and quality control, which are closely related, vary widely with the type and specifics of a product, whether for the military or for mass markets. Production of microscale surface finishes on a material (for example, a polished metal surface on an aircraft engine turbine blade) is certainly different from making parts for MEMS devices on a similar size scale (for example, a microaccelerometer with submicrometer feature sizes). Similarly, the production of nanocrystalline materials is generally very different from the manufacture of possible molecular electronic devices. In this chapter, the committee considers the general aspects, along with some particular aspects, of manufacturing products whose performance depends on structure, material, or chemistry on the micro- and nanoscales. The materials, components, subsystems, systems, and platforms used by the military are mainly purchased from industry. Rarely is military hardware manufactured by the Department of Defense. Hence, industry-based manufacture of what the military uses is of central importance. This will be no less true for hardware made by micro- and nanotechnologies than it is for the alloys, antennas, radars, missiles, and airplanes that are now being employed by the military. The role of various communities in acquiring, maintaining, and employing military hardware is summarized in Figure 4-1. FABRICATION (PATTERNING) APPROACHES Integrated circuit manufacturing is a top-down process where the starting point is a flat wafer onto which patterns are defined and created by both additive

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-1 Communities needed for the production, maintenance, and use of military hardware. (thin-film deposition and growth) and subtractive (etch) processes. This has evolved into an enormously sophisticated enterprise, which is proven for the low-cost, high-yield manufacture of extremely complex (~100 million transistors) and reliable circuits. However, limitations are on the horizon. One is the difficulty of extending today’s optically based lithographic techniques to the nanoscale, which is much smaller than ultraviolet optical wavelengths. Another is the limited number of materials used in ICs. The nanotechnology community is investigating many disparate technologies based on many different materials, but it is far from evident that the different processing requirements of these technologies can be reconciled. Self-assembly is a radically different approach to fabrication at the nanoscale. It takes advantage of molecular and intermolecular forces to define atomic, nanoscale, and macroscale structures. Self-assembly depends on appropriate direction and control being exerted at all stages of the process by preprogramming of the subunits or building blocks such that the required recognition elements for self-assembly are contained in the subunits. Crystal growth is an example of self-assembly with exquisite long-range order. Living species are proof that complex three-dimensional structures with interacting functionality are possible. Integration of the top-down (lithography and pattern transfer) and the bottom-up (self-assembly) approaches offers an attractive approach to bridging the current gaps between these paradigms. The incompatible materials issue may be addressed by individualized optimization of different devices and subsystems, followed by an assembly process akin to the automotive assembly line but at a vastly smaller scale. Here again, top-down (pick-and-place) and bottom-up, self-assembly inspired (DNA-assisted) approaches are among the many being investigated.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies Lithography and Pattern Transfer The microelectronics, computer, and information revolutions can trace their success to several technological roots. The integration of transistors into functional blocks—which are further integrated to form microprocessors, memories, and other ICs—is one of the main reasons for the ever-increasing functionality. Decreasing linewidths, currently 130 nanometers, enable placing more transistors on a chip. Increasing wafer sizes—now up to a 300-millimeter diameter, with projections to 400 millimeters—allow more chips to be produced simultaneously. Mass production of integrated circuits by batch fabrication with high yields has led to declining cost per function and the remarkable reliability of microelectronic devices. Not having to assemble individual parts before packaging chips has also been a major factor in the high yields and low costs of microelectronics. A great triumph of microelectronics has been the high-yield manufacturing of reliable 100-million-part assemblies. With nanotechnology, reliability of even more complex assemblages is possible. There are fundamentally only two kinds of things that have to be done to produce integrated circuits—a pattern must be made and it must be transferred into the work piece by either deposition or removal of materials.1 The pattern definition is done by lithography. Pattern transfer involves any of a number of processes for adding materials to a wafer, such as ion implantation, or removing materials, such as plasma etching. Ancillary techniques are also employed to ensure that the pattern production and transfer techniques work properly. Planarization of a partially processed wafer by chemical-mechanical polishing is an example. Currently, lithographic exposures are done between 24 and 30 times or so during production of a complex IC. Each exposure requires multiple processing steps, including spinning on, prebaking, exposing, postbaking, developing the resist, pattern transfer, and removing the residual resist. Indeed, there may be about 200 processing steps for a modern integrated circuit. Today the mask sets for the most sophisticated circuit can cost about $1 million. The systems that align a given mask level to structures on a wafer and make the exposures cost about $10 million. Aligners are projected to cost close to $30 million in 5 years. Lithography enables production in many technologies besides integrated circuits. Figure 4-2 shows several of the classes of structures and devices that require lithography to produce a pattern on a substrate. The highest resolution lithography is generally required only for cutting-edge integrated circuits. However, techniques developed in the IC industry clearly have applications to many other fields. In the top row of Figure 4-2 is an etched integrated circuit with its copper interconnects, a flexible printed circuit board, and solder bumps. The center row shows a deep-etched silicon structure, a photonic material, and a microfluidic device. In the bottom row are a gene chip, densely packed nano-crystals, and guided cell growth.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-2 Lithography examples. SOURCE: Nagel, D.J. 2002. Technologies for micrometer and nanometer pattern and material transfer. Direct-Write Technologies for Rapid Prototyping Applications, A. Pique and D.B. Chrisey, eds. New York, N.Y.: Academic Press. © 2002, Elsevier Science (USA), reproduced with permission of the publisher. Currently, most aligners use 248-nanometer radiation from KrF lasers to expose photoresist. The switch to ArF lasers, which have a wavelength of 193 nanometers, has already begun for the production of chips with 100-nanometer linewidths. In a few years, the ITRS indicates that it will be necessary either to continue this migration to a 157-nanometer wavelength advanced optical lithography tool or to switch to a next-generation lithography approach. Both alternatives are being intensively investigated and are likely to coexist for some time. The leading contender in the United States for the next lithography technology is extreme ultraviolet (EUV) radiation with a wavelength of 13 nanometers.2 EUV lithography requires the use of plasma sources produced by high-power laser irradiation of atomic xenon, all reflective optics (including the mask), and— as usual for a different wavelength—new photoresists. Numerous technical barriers must be overcome for EUV lithography to be ready for the production of commodity ICs. In Japan, there is still strong interest in x-ray lithography using 1-nanometer wavelength radiation.3 In this case, the “light” source would be either plasmas or high-energy electrons orbiting in an evacuated toroid, so-called synchrotron ra-

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies diation. A fundamental challenge with x-ray lithography is the mask, which must have features the same size as the pattern to be impressed into the photoresist. That is, x-ray lithography is a one-to-one technology and not a projection reduction technique, because optical elements that focus 1-nanometer radiation are not available. Techniques to further extend optics using alternative exposure schemes and various nonlinear processes in the lithography process are also under investigation. It is far from clear which lithography technique will prove to be the industrial workhorse for ICs with linewidths below 100 nanometers. It is possible that electron projection or direct-write technologies will play some role in future chip production. For large-volume commercial applications, economics will be a major driver. Evolution of an incumbent technology, e.g., optical lithography, almost always wins until fundamental limits are reached. For optics the fundamental limit is related to, but is not the same as, the optical wavelength. In a single exposure the limit is on the highest spatial frequency, or how close together two features can be located, not on how small each feature can be. This limit is approximately one-fourth of the wavelength, or 50 nanometers for today’s ArF-laser-based lithographic tools. Immersion techniques allow a reduction by a factor of about 1.5, the refractive index of the immersion fluid. Multiple exposures taking advantage of inherent nonlinearities in the photoresist and subsequent processing stages allow further extensions by factors of one-half, one-third, etc. The manufacturing limits for these optics extensions are associated with process latitude and yield rather than with fundamental physical limitations. It is clear that lithography for the volume production of microelectronics is already a nanotechnology and will become increasingly so in the next decade. Moreover, it will be a long time before any of the methods of modern nanotechnology, such as the growth and use of carbon nanotubes, rival lithography in commercial volume. Beyond the methods that are in use or in contention for mass production of integrated circuits, a number of lithographic methods have been developed in the past decade. Not all of them offer the nanometer resolution that will be needed for IC production in coming years. However, they have been or might prove to be of use for making some MEMS and other structures and devices. These methods are briefly reviewed before discussing pattern transfer methods. A three-stage process, developed in Germany, involves sequential use of lithography, electrodeposition, and molding (Lithographie, Galvanoformung, und Abformung) (LIGA). In contrast to the lithography technologies surveyed so far, which use resists with thickness from nanometers to micrometers, LIGA employs resists with thickness from micrometers to more than millimeters. The steps in the LIGA process are shown schematically in Figure 4-3.4 The next lithography method is both the most recently discovered and the most unconventional compared with commercial techniques. It is called lithographically induced self-construction (LISC),5 and it has a variant termed litho-

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-3 The sequential steps in LIGA. In 1 and 3, el. cond. means electrically conductive. SOURCE: Institut für Mikrotechnik Mainz GmbH. 2001. LIGA Technology. Available online at <http://www.imm-mainz.de/> [August 12, 2002]. graphically induced self-assembly (LISA).6 In both cases, a patterned mask coated with a surfactant is placed in close proximity to a substrate covered with a thin thermoplastic polymer (Figure 4-4). The polymer may be chemically identical to a photoresist—for example, polymethyl methacrylate (PMMA)—but it does not function as a normal photoresist. There is no radiative transfer between the mask and polymer, and the chemistry of the polymer is not modified during pattern transfer. Rather, when the PMMA is heated to 170°C (which is above the softening point), electrostatic forces cause it to move laterally into shapes mediated by the nearby mask. The LISC and LISA techniques are both hybrid contact and

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-4 Schematic of the structures used in LISC. SOURCE: Reprinted with permission from Chou, S.Y., L. Zhuang, and L. Guo. 1999. Lithographically induced self-construction of polymer microstructures for resistless patterning. Applied Physics Letters 75(7): 1004–1006. © 1999, American Institute of Physics. proximity methods. In LISC, the resulting pattern is determined only by the pattern used. In LISA, the pattern that is produced has an overall shape set by the mask but fine details within that shape that arise spontaneously. Embossing is one of the old technologies that has been extended to the micrometer scale in recent years. Hot embossing can be employed to produce micrometer and even nanoscale features and structures. Patterns with structures about 10 μm wide and deep can be embossed into PMMA.7 Recently, a modified wafer bonding system was used to emboss structures as fine as 400 nanometers across an entire 10-cm-diameter wafer.8 A variant of embossing, called nanoimprint lithography (NIL), involves impressing a mold onto the surface of a photoresist-covered substrate.9 In this case, the pattern is transferred to the resist, commonly PMMA, by mechanical rather than chemical action. Subsequent processing of the resist to open the thinned regions to the substrate permits conventional uses of the resist for deposition onto or etching into the substrate. The molds for NIL can be prepared by a wide variety of the normal and developmental lithographic processes. For example, e-beam lithography has been used to make a mold with 10-nm-diameter pillars on a 40-nm pitch that were then imprinted into PMMA.10 A technique called step and flash imprint lithography essentially embosses a layer of liquid on a surface that is then turned into a solid using a photochemical process.11 This technique avoids the elevated temperatures and pressures ordinarily required for embossing. The wafer is first coated with a transfer layer of solid organic material. Then a glass template with the desired pattern is placed near the coated wafer. The template can be micromachined by a variety of meth-

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies ods. A low-viscosity liquid (a photopolymerizable, organosilicon etch-barrier material) is dispensed between the template and transfer layers on the wafer, and the layers are then brought into contact. After UV exposure to solidify the etch barrier and make it adhere to the transfer layer, the template is removed. A plasma etch transfers the pattern from the now solid etch barrier into the transfer layer, and then the etch barrier is removed. This leaves the pattern in the transfer layer on the wafer surface ready for further processing steps. Step and flash imprint lithography has produced 60-nanometer features. Rubber stamps have been used for centuries. In recent years, Whitesides and his group have extended stamping to replicate patterns with features finer than micrometers by the use of polydimethylsiloxane (PDMS) and other elastomers. They term the technique “soft lithography” because of the compliant character of the stamp.12 One of the ordinary lithography methods is used to pattern a thin film on silicon or some other substrate to make a mold. After the surface of the mold has been etched and silanized, the liquid PDMS precursor is cast over the pattern and polymerized by cross-linking. Then, the elastomer is peeled off the mold and placed on a substrate for handling. Wetting of the PDMS stamp with various liquids and suspensions is done before the stamping. In general, the depth of the pattern is in the range of 0.2 to 20 μm, with the maximum limited by the stability of the PDMS structure. The width and spacing of the contact regions are between 0.5 and 200 μm, with the separation limited by the tendency of the region between contacts to bulge toward the substrate being stamped. Soft lithography can be employed in a rolling manner if the PDMS stamp is attached to a cylinder. The methods for pattern production and transfer that are used in the manufacture of microelectronics IC production involve a very limited number of materials and associated processes. The materials used for chips now number only about 10. Copper interconnects, high-dielectric-constant (k) materials for gate insulators and low-k materials for separating interconnect lines are the most recent additions to the list of IC materials. The processes used to make these materials during chip production also number about 10. IC technologies are certainly fundamental to the production of microscale mechanics, optics, and magnetics. However, the number of materials and processes in demonstrated and emerging MEMS and similar technologies far exceed those used for IC production, as indicated schematically in Figure 4-5. One example is the use of piezoelectric materials in MEMS for linking electronic and mechanical behavior. The substrates employed for MEMS devices include semiconductors other than silicon, notably silicon carbide, as well as ceramics, metals, and polymers. There are numerous processes used in the manufacture of MEMS devices that play no role in IC production. The central step in the production of micromechanics on a substrate by surface micromachining is the dissolution of a sacrificial layer of material to release the mechanism. Deep etching of a substrate, called “bulk micromachining,” is another important MEMS pattern transfer tech-

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-5 Integrated circuit production. The number of materials and associated processes for the production of micromechanics, -optics and -magnetics greatly exceeds the number of materials and processes used to manufacture microelectronics. nology that has no role in IC manufacture. Such etching is done by one of two processes. The use of solvents, which attack different planes in silicon at widely varying rates, is called orientation-dependent etching. It now plays a role in the production of some MEMS products. The alternative is deep reactive-ion etching (DRIE). In this case, the plasma ambient over a silicon substrate is alternated every few seconds between etching and passivation (coating) conditions. This process, invented by the Bosch Corporation, can result in deep, narrow trenches in the substrate, the walls of which are scalloped on a very fine scale. Figure 4-6 shows examples, and it also indicates that the rate at which the Bosch process etches into the silicon is geometry dependent.13 The role of pattern transfer for the production of nanoscale structures and devices will probably be similar to that for the production of microscale structures and devices involving mechanics, optics, and magnetics. That is, the transfer processes used in microelectronics manufacture will remain important in many cases. The self-assembly of complex structured materials, as discussed in the next section, is playing an increasingly important role in nanoscale fabrication. New approaches using lithography to direct self-assembly are already emerging in fields as disparate as semiconductor crystal growth (discussed below) and carbon nanotube formation and are likely to play a significant role in the development of nanotechnology. The wider variety of materials that will play a role in nanotechnology will probably involve additional techniques for pattern transfer. This will almost certainly be the case for nanoscale structures that use organic and biomaterials.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-6 Cross-sectional photograph of a silicon wafer processed by deep reactive ion etching. SOURCE: S.D. Senturia. 2000. Microsystem Design, p. 70. Boston, Mass.: Kluwer Academic Publishers. Self-Assembly Self-assembly in nature is older than life itself. Crystalline geological structures spontaneously form highly organized patterns from molten lava. All living organisms, from the simplest single-cell species to humans, depend on some form of molecular self-assembly. Nature performs the most astonishing feats of self-assembly with an artistry and facility that we can only admire and all too often take for granted. Protein folding, nucleic acid assembly and tertiary structures, phospholipid membranes, ribosomes, microtubules, and the nucleocapsides of viruses are but some, representative examples of biological self-assembly in nature that are of critical importance to living organisms.14 Aside from their ability to carry out the functions of life, one of the more remarkable features of self-assembled materials is that their structure may be very complex yet demonstrate long-range order. The power and beauty of spontaneous self-assembly derives from its ability to rapidly, and with seeming ease, generate large, complex, sophisticated “supermolecules,” or ensembles of molecules, from easily available building blocks with high efficiency, generally under mild conditions (at or near room temperature, at atmospheric pressure, in water or other common media). A biological self-assembly,15,16,17 as exemplified by such structures as crystals,18,19 surfactants,20 micelles (nanoscale molecular aggregates),21 colloidal suspensions in confined films or aggregated into fractal structures,22 self-assembled monolayers (SAMs),23 and liquid crystals,24 involves ensembles of molecules with unique properties and function. Because it is possible to manufacture nanostructures with useful properties through self-assembly, interest has recently increased substantially in the study of interactions of molecules that form larger nanostructures of ordered material aggregates.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies Nature’s repertoire of information to guide self-assembly includes hydrogen bonding, π-π stacking, electrostatic and van der Waals interactions, hydrophobic-hydrophilic interaction, dipolar effects, molecular conformations, and phase boundaries, as well as the shape and size of both the final product and the constituent subunits. These effects are commonly referred to as weak interactions in contrast to the strong chemical bonds characteristic of simple molecules,25 which were the basis of the revolution in chemical products of the last 100 years. An important feature of all self-assembly processes is the existence of a kinetically labile, reversible equilibrium between starting materials, intermediates, and products,26 with the final outcome under thermodynamic control (i.e., the end ensemble is the thermodynamically most stable one). As the equilibrium is reversible, the process is self-correcting; an incorrectly formed bond can disassociate and reassociate correctly. As a consequence, self-assembly processes generally engage in self-repair and are self-healing. Self-assembly has the potential for bottom-up fabrication techniques to produce nano- to mesoscale materials and systems, that is, structures and ensembles having spatial dimensions in the range of 1 nm to 1 μm—the size of a large molecule to the size of a living cell. Nano- to microscale systems bridge the molecular and the macroscopic and display unique collective and often nonlinear behavior and properties, different from the bulk characteristics of common substances. Control of the bottom-up products is difficult at present (except for biologically produced and a few select materials) because of the currently limited understanding about how to use these forces. The current state of knowledge about self-assembly and the potential for new products may be comparable to the state of knowledge about the chemical bond at the turn of the previous century. Expanding interest in the fabrication of new materials with these self-assembly methods may give rise to an entirely new discipline. Examples illustrating the potential of self-assembly as a process for materials production include materials such as the shell of an abalone. This composite material is formed by the careful assembly of elongated calcium carbonate crystals in layers, with the long axes of the crystals pointing in perpendicular directions in alternating layers. A protein deposit forms a strong glue to hold these crystals in place, forming the strong final product. A wide variety of useful composite whisker structures are formed by a number of mechanisms such as condensation on specific crystal faces, selected chemical reactivity, and eutectic behavior of alloy mixtures. An example of a fascinating nanostructure formed through the forces involving chemical bonds is that of a carbon nanotube. Selected nanotubes with high conductivities have excited researchers because they might one day be used as conductors and circuits in nanoelectronics (see Box 3-1). Block copolymers form amazingly regular patterns of plastic material with repetition distances on the order of nanometers to micrometers. These plastics interact with light and are used in devices such as optical band gap structures for

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies of a number of separate fabrication sites organized in a fabrication network and made available to the community through a central organization, the MEMS Exchange, was organized. An important and unique element is that designers can have the process sequence for their devices conducted at multiple fabrication sites. This approach affords designers with an enormous range of choices in processing techniques and materials as well as the ability to fully customize the processing sequence. Figure 4-11 shows the rapid growth in user accounts for the MEMS Exchange. Furthermore, over the past 3 years, distributed, multiple-site MEMS device fabrication has been shown not only to be feasible but also to produce advanced and complex prototype devices with distinct advantages over those made at a single foundry. In the last 2 years, nearly 400 process sequence runs were successfully performed in the MEMS Exchange network for designers around the country, each run being different and customized for a particular application. Most of the process sequences delivered by the MEMS Exchange were performed at two sites, with a surprising number of runs having had processing work performed at three or more sites. The designers accessing the MEMS FIGURE 4-11 Cumulative user accounts for the MEMS exchange. Courtesy of the MEMS Exchange.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies Exchange come from a wide range of organizations, including commercial (40 percent), academic (40 percent), and government laboratories (20 percent). Fabrication sites enlisted in the MEMS Exchange include the University of California at Berkeley; Stanford University; Cornell University; the University of Michigan; Case Western Reserve University; Louisiana State University; the University of Illinois; Sony Semiconductor; Integrated Sensing Systems Corporation; Lance Goddard Associates; Microwave Bonding Corporation; Teledyne Electronics Corporation; Advanced MEMS Optical; Tactical Fabs, Inc.; Analog Devices; ASML; Zygo Teraoptics; Intelligent Micropatterning; Axsun Technologies; Aspen Technologies; Fiberlead, Inc.; and American Precision Dicing. Through these 22 fabrication sites that belong to the MEMS exchange, an extremely wide range of MEMS fabrication resources are available to the community, and new process capabilities are being added each week. Effect of Manufacturing Complexity on Commercialization Government-sponsored technology foundry programs (such as MOSIS, the Microelectronics Center of North Carolina, and the MEMS Exchange) have been successful in providing some level of technology access to the broader community, although demand for the services usually far outstrips capacity. However successful these programs may be in providing wide access to a technology, the cost barriers to the adoption of a new technology can only be solved by industry. Thus, government must make sure to involve industry in the new technology as soon as possible. Manufacturing complexity strongly affects manufacturing cost. Thus, for technologies such as MEMS, cost depends on device complexity—in effect, how much integration between IC electronics and micromechanics has been accomplished. The three levels of manufacturing complexity of MEMS are these: all micromechanics and no IC electronics multichip modules or other assembled hybrids integrated micromechanics and IC electronics Commercialization has already occurred for MEMS on all three levels of complexity. The third level is the most complicated and requires the most fabrication sophistication from industry. This level, in turn, can be subdivided into three specialties: preprocessed micromechanics, postprocessed IC electronics integrated micromechanics and IC electronics postprocessed micromechanics, preprocessed IC electronics

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies Case Study: Texas Instruments and the Digital Mirror Device Commercialization does not necessarily happen earliest for the simplest forms of manufactured MEMS (either those containing no IC electronics or hybrids of micromechanics and IC electronics). As an example of how one of the more difficult manufacturing processes (integrated micromechanics and IC electronics) was used for MEMS commercialization, the story of the Texas Instruments (TI) digital mirror device [DMD™] is related below. It shows that even after DMD feasibility had been demonstrated in the late 1980s, the commercialization effort required more than two decades to achieve profitability. This effort took place before the advent of government programs to lower the barriers to commercialization of MEMS. The cost of the DMD and the associated projector, called the digital light processor (DLP), is reported to have been nearly a billion dollars. Other commercialization efforts, most notably the MEMS airbag crash detection accelerometer, manufactured by Analog Devices, Inc. (ADI) had similar difficulties, but commercialization was accomplished at much lower cost and in a shorter time. This was primarily due to the relatively simpler structures ADI had to fabricate but also to the influx of engineers trained in MEMS in government programs. TI was a pioneering company, well in advance of the technology curve, and was forced to solve many newly discovered MEMS problems entirely on its own. For micro- and nanotechnology to be commercialized quickly, it will be necessary for government not only to continue pursuing research in these technologies but also to guarantee wide access to them through programs specially funded for this purpose. The government must ask itself, first, if it really can expect most companies to be persistent in the face of product development adversity and, second, if it can wait so long for the new products to arrive. TI’s DMD, shown in Figures 4-12 and 4-13, consists of an array of a half million or more mirrors that can electrically switch light to a working area. Figure 4-12 shows a cut-away of the DMD structural model. Figure 4-13 shows a photomicrograph of the DMD. The levels labeled on the photomicrograph are identified in Table 4-2 illustrating the complexity of the DMD, with seven functions squeezed into three levels. The original DMD started out as an array of polymer film mirrors originated by a National Security Agency research group whose mission was optical correlation processing. Under DARPA support the mirror array technology was transferred to TI for development. Each mirror had to have a charge placed under the membrane to cause deflection and thus the enable the parallel correlation operations. This called for a large array of semiconductor circuits to enable the deflection. Feasibility was demonstrated, but manufacturing problems—for example, those from dust particles under the membrane causing defective mirror pixels— quickly emerged.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies FIGURE 4-12 Cut-away of the digital mirror device structural model. Reprinted with permission of Michael Mignardi, Texas Instruments. FIGURE 4-13 Photomicrograph of the digital mirror device. Reprinted with permission of Michael Mignardi, Texas Instruments.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies TABLE 4-2 High Complexity of the Digital Mirror Device Label DMD Level Function A Electrode Bias Electrode B Electrode Address Electrode C Hinge/Beam Beam and Hinge Posts D Hinge/Beam Hinges E Mirror Mirror and Mirror Support Post A monolithic mirror approach using an array of aluminum alloy mirrors was developed to overcome the particle problems of the polymer mirrors. The aluminum mirrors worked much better in regard to dust but rapidly failed in operation because of the stress inherent in the mirror layer after plasma etching. Also, the TI printer division interested in the printer application fell on hard times, so the continuation of the program was always in doubt. New DARPA money was supplied, administered by the Air Force. DARPA was then interested in large projection displays. During the 1990s development problems continued. (For instance, in operation, the mirrors would stick permanently to the underlying films; the mirrors did not optimally fill the optical aperture, leading to a checkerboard display image; there was high optical insertion loss from a mismatch in the pixel size and the size of the arc lamp; a very high light flux on the DMD was required; arc lamp longevity was insufficient; and the cost of engineering of a small sophisticated light handling system, including an optically flat window on the package top, was high.) By the late 1980s the DMD was being touted as a MEMS success story, but these manufacturability issues continued to threaten the viability of commercialization. During the 1990s, TI, using its own funds, gradually worked out solutions to the DMD manufacturability problems. Mirror reliability was solved with a hidden mirror hinge made out of an amorphous material. Since there were no grain boundaries in the hinge, there were no sources of the cracks that had led to high failure rates in the metal hinges. Stiction was fixed by fabricating springs and a fluorocarbon coating on the back side of the mirrors. The springs produced a restoring force when the mirror touched the base layer. The hidden-hinge design allowed for a more optically efficient pixel, filling the aperture. Obtaining a robust, very bright light source turned out to be an enormous problem, and the light source is still a major cost component in the DMD system. One of the veteran engineers on the project at Texas Instruments, Michael Mignardi, recently gave an IEEE seminar detailing the scale of TI’s effort to successfully produce the DMD.69 Even after feasibility had been demonstrated in the late 1980s, it took another 10 years to reach profitability, with the effort being spread roughly equally among device improvement, packaging, and testing. Low outgasing adhesives were used to prevent the buildup of a sticky layer inside the package, which would result in stiction problems. A zeolite getter was included

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies in the package to mitigate water-induced stiction in the completed package. These and other innovations led to TI turning its first profit on projectors in 2000. Specifications for a 1.2-million-pixel projector device include mirrors that tilt 10 to 12 degrees to deflect light onto a screen. Only five stuck-on pixels are permitted for a shipped DMD. Clearly, achieving this level of yield and reliability required an enormous effort in developing test equipment and lifetime testing for the semiconductor layer, the mirror level, and the combined levels in action. TI and its technological team deserve a special place in the MEMS Hall of Fame for their perseverance in bringing the projector product to market. The Air Force also deserves credit for early support. Most importantly the development of the successful DMD back-end processing indicates that such processing can be solved by a determined team, even for very complex devices and systems. But it also illustrates the very real difficulties in turning a laboratory demonstration into a reliable product—difficulties that will undoubtedly confront all emerging micro- and nanotechnology applications in as yet unanticipated ways. New technologies are generally thought to become profitable after 10 years. It is generally believed that so far few, if any, MEMS projects have become profitable long-term. All of which is to say that building micro- and nanotechnology materials and systems is very difficult. Continuing to approach these technologies in the cottage industry mode that is currently fashionable will not produce DMD-type successes. There is a need for farsighted management and technical staffs—and for deep pockets. Further, the multidisciplinary research capacity that TI had available does not exist in many of today’s companies in the face of the increased pressure for short-term profitability, and it certainly is not possible in the small-company entrepreneurial environment that is attempting to bring nanotechnology to fruition. FINDINGS AND RECOMMENDATIONS Finding 4-1. Lithography and pattern transfer and self-assembly are key enablers for evolving micro- and nanotechnologies. Recommendation 4-1. The AFRL R&D program will require access to micro- and nanolithography and pattern transfer tools. This should be accomplished using available national facilities or otherwise providing the function internally. Research into new nanolithography and patterning technologies, complementary to the industry push for high-throughput tools, would be a worthwhile investment. The Air Force should not compete with industry efforts, particularly in silicon technology, but should concentrate on developing processes for structures and materials that are outside traditional silicon processing—for example, deep etching for MEMS and integration of new materials with silicon.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies Finding T7. Integration of micro- and nanoscale processes and of different material systems will be broadly important for materials, devices, and packaging. Self-assembly and directed assembly of dissimilar elements will be necessary to maximize the functionality of many micro- and nanoscale structures, devices, and systems. Achievement of high yields and long-term reliability, comparable to those of the current integrated circuit industry, will be a major challenge. Recommendation T7. The Air Force should monitor progress in self- and directed-assembly research and selectively invest its R&D resources. It will be critical for the Air Force to participate in developing manufacturing processes that result in reliable systems in technology areas where the military is the dominant customer—for example, in sensors and propulsion systems. Developments in many of these areas will be driven by the commercial sector. The Air Force must stay aware of advances and apply them to its unique needs. As an example, in sensor applications a wide range of otherwise incompatible materials and fabrication processes is likely to be necessary. Finding 4-2. So much is already known about progress in silicon, with its already highly developed and constantly improving manufacturing processes, that it is unlikely a sui generis technology will spring up sufficiently developed and robust that it will immediately supplant not only the transistor but also all the rest of the integrated circuit. Integrated circuit technology has become extremely sophisticated, and the industry is devoting extensive resources to extending this sophistication in its drive to validate Moore’s law for future IC generations. In contrast, nanotechnology is at a much earlier development stage, concentrating on the behavior of individual devices and circuit components (switches, wires, etc.). It is most likely that these new technologies will first find use as complements to silicon, not as immediate replacements for integrated circuits. Over the longer term, it is not possible to predict the relative roles of integrated circuits and new and evolving nanotechnologies. Recommendation 4-2. The Air Force should emphasize those areas of micro- and nanotechnology for information processing that are potentially integrable with silicon technology and that address Air Force-specific, non-commercial military applications. Finding 4-3. The path from laboratory demonstration to the manufacture of reliable devices and systems is long and arduous, requiring extensive resources and prodigious technology development. This will undoubtedly be as true of today’s emerging technologies as it has been throughout the history

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies of technology. It is worthwhile to consider this lesson when listening to the siren songs appearing daily, particularly in the popular and business press, on the future benefits of nanotechnology. There is undoubtedly an exciting future, and just as undoubtedly, we will find many surprises, both positive and negative, along the way. Recommendation 4-3. Air Force research efforts should be directed not only to the science of micro- and nanotechnology, but also to the development of devices and systems and wide access to the manufacturing technology required to produce them. REFERENCES 1. Nagel, D.J. 2002. Technologies for micrometer and nanometer pattern and material transfer. Pp. 557–701 in Direct-Write Technologies for Rapid Prototyping Applications: Sensors, Electronics, and Integrated Power Sources. A. Pique and D.B. Chrisey, eds. New York, N.Y.: Academic Press. 2. Bjorkholm, J.E. 1998. EUV lithography–the successor to optical lithography? Available online at <http://www.intel.com/technology/itj/q31998/articles/art_4.htm> [July 8, 2002]. 3. Smith, H.I. 2001. Japan could dominate industry with x-ray lithography. Available online at <http://www.e-insite.net/semiconductor/index.asp?layout=article&articleId=CA61809&stt=001> [July 8, 2002]. 4. Institut für Mikrotechnik Mainz. 2001. LIGA Technology. Available online at <http://www.imm-mainz.de/> [July 8, 2002]. 5. Chou, S.Y., L. Zhuang, and L.J. Guo. 1999. Lithographically induced self-construction of polymer microstructures for resistless patterning. Applied Physics Letters 75(7): 1004–1006. 6. Chou, S.Y., and L. Zhuang. 1999. Lithographically induced self-assembly of periodic polymer micropillar arrays. Journal of Vacuum Science Technology B 17(6): 3197–3202. 7. Becker, H., and U. Heim. 1999. Silicon as tool material for polymer hot embossing. Pp. 228– 231 in Proceedings of MEMS ’99: The 12th IEEE International Conference on Micro Electro Mechanical Systems. New York, N.Y.: IEEE. 8. Glinsner, T. 2001. Nanoimprinting can solve pattern-generation problems. R&D Magazine 43(7): 33. 9. Chou, S.Y., P.R. Krause, and P.J. Renstrom. 1995. Imprint of sub-25 nm vias and trenches in polymers. Applied Physics Letters 67(21): 3114–3116. 10. Chou, S.Y. 2001. Nanoimprint lithography. Available online at <http://www.ee.princeton.edu/~chouweb/newproject/page3.html> [July 8, 2002]. 11. Johnson, S.C. 1999. Selective Compliant Orientation Stages for Imprint Lithography. Available online at <http://sfil.org/research/papers/sjthesis.pdf> [July 8, 2002]. 12. Xia, Y., and G.M. Whitesides. 1998. Soft lithography. Annual Review of Materials Science 28: 153–184. 13. Senturia, S.D. 2001. Microsystem Design. Boston, Mass.: Kluwer Academic Publishers. 14. Kauffman, S.A. 1995. At Home in the Universe: The Search for Laws of Self-Organization and Complexity. New York, N.Y.: Oxford University Press. 15. This series, for example, provides a good overview of the topic: Lehn, J.M., J.L. Atwood, J.E.D. Davis, D.D. MacNicol, and F. Vögtle. 1996. Comprehensive Supramolecular Chemistry, Vols. 1–11. New York, N.Y.: Pergamon.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies 16. Lehn, J.M. 1995. Supramolecular Chemistry: Concepts and Perspectives. New York, N.Y.: VCH Press. 17. Balzani, V., and L. DeCola. 1992. Supramolecular Chemistry. Dordrecht, The Netherlands: Kluwer Academic Press. 18. Desiraju., G.R. 1996. The Crystal as Supramolecular Entity: Perspectives in Supramolecular Chemistry, Volume 2. New York, N.Y.: John Wiley & Sons. 19. Desiraju, G.R. 1989. Crystal Engineering: The Design of Organic Solids. New York, N.Y.: Elsevier. 20. Manne, S., and G.G. Warr. 1999. Supramolecular Structure in Confined Geometries. Washington, D.C.: American Chemical Society. 21. Jones, M.N., and D. Chapman. 1995. Micelles, Monolayers and Biomembranes. New York, N.Y.: Wiley-Liss. 22. Wilcoxon, J.P., J.E. Martin, and D.W. Schaefer. 1989. Aggregation in colloidal gold. Physical Review A 39(5): 2675–2688. 23. Kumar, A., N.L. Abbott, E. Kim, H.A. Biebuyck, and G.M. Whitesides. 1995. Patterned self-assembled monolayers and mesoscale phenomena. Accounts of Chemical Research 28(5): 219– 226. 24. De Gennes, P.G., and J. Prost. 1993. The Physics of Liquid Crystals, 2nd Ed. New York, N.Y.: Oxford University Press. 25. Rouvray, D. 2000. Molecular self-assembly. Chemistry in Britain 36(7): 26–29. 26. Huc, I., and J.M. Lehn. 1997. Virtual combinatorial libraries: Dynamic generation of molecular and supramolecular diversity by self-assembly. Proceedings of the National Academy of Sciences 94(6): 2106–2110. 27. Frechet, J.M.J. 2002. Dendrimers and supramolecular chemistry. Proceedings of the National Academy of Sciences 99(8): 4782–4787 28. Whitesides, G.M., and J.C. Love. 2001. The art of building small—Researchers are discovering cheap, efficient ways to make structures only a few billionths of a meter across. Scientific American 285(3): 38–47. 29. Van Hest, J.C.M., and D.A. Tirrell. 2001. Protein-based materials: Toward a new level of structural control. Chemical Communications (19): 1897–1904. 30. Yu, S.J.M., C.M. Soto, and D.A. Tirrell. 2000. Nanometer-scale smectic ordering of genetically engineered rodlike polymers: Synthesis and characterization of monodisperse derivatives of poly (g-benzyl a, L-glutamate). Journal of the American Chemical Society 122(28): 6552– 6559, and references therein. 31. Limberis, L., and R.J. Stewart. 2000. Toward kinesin-powered microdevices. Nanotechnology 11(2): 47–51. 32. Srinivasarao, M. 1999. Nano-optics in the biological world: Beetles, butterflies, birds, and moths. Chemical Reviews 99(7): 1935–1961. 33. Miller, J.S., and A.J. Epstein. 2000. Molecule-based magnets—An overview. MRS Bulletin 25(11): 21–28. 34. Ovcharenko, V.I., and R.Z. Sagdeev. 1999. Molecular ferromagnets. Russian Chemical Reviews 68(5): 345–363. 35. Alivisatos, A.P. 2001. Less is more in medicine—Sophisticated forms of nanotechnology will find some of their first real-world applications in biomedical research, disease diagnosis, and, possibly, therapy. Scientific American 285(3): 67–73. 36. Hooks, D.E., T. Fritz, and M.D. Ward. 2001. Epitaxy and molecular organization on solid substrates. Advanced Materials 13(4): 227–241, and references therein. 37. Choquette, K.D., K.L. Lear, R.P. Schneider, K.M. Geib, J.J. Figiel, and R. Hull. 1995. Fabrication and performance of selectively oxidized vertical-cavity lasers. IEEE Photonics Technology Letters 7(11): 1237–1239.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies 38. Hatzor, A., and P.S. Weiss. 2001. Molecular rulers for scaling down nanostructures. Science 291(5506): 1019–1020. 39. For a contemporaneous snapshot of this field, see, for example, Fan, J.C.C., and J.M. Poate. 1986. Heteroepitaxy on Silicon. Materials Research Society Symposia Proceedings 67. Pittsburgh, Pa.: Materials Research Society. 40. Kapon, E., D.M. Hwang, and R. Bhat. 1989. Stimulated-emission in semiconductor quantum wire heterostructures. Physical Review Letters 63(4): 430–433. 41. Zhou, Y.X., S. Luong, C.P. Hains, and J. Cheng. 1998. Oxide-confined monolithic, multiple-wavelength vertical-cavity surface-emitting laser arrays with a 40-nm wavelength span. IEEE Photonics Technology Letters 10(11): 1527–1529. 42. For a recent survey, see the special section on the growth of heterostructure materials on nanoscale substrates, S.D. Hersee, D.A. Zubia and S.R.J. Brueck, eds. 2002. IEEE Journal of Quantum Electronics 38(8). 43. Cui, Y. and C.M. Lieber. 2001. Functional nanoscale electronic devices assembled using silicon nanowire building blocks. Science: 291(5505): 851–853. 44. Duan, X.F., Y. Huang, Y. Cui, J.F. Wang, and C.M. Lieber. 2001. Indium phosphide nanowires as building blocks for nanoscale electronic and optoelectronic devices. Nature 409(6816): 66– 69. 45. Gudiksen, M.S., L.J. Lauhon, J. Wang, D.C. Smith, and C.M. Lieber. 2002. Growth of nanowire superlattice structures for photonics and electronics. Nature 415(6872): 617–620. 46. International Technology Roadmap for Semiconductors. 2001. Available online at <http://public.itrs.net/> [July 3, 2002]. 47. Moravec, H. 1998. When will computer hardware match the human brain? Available online at <http://www.transhumanist.com/volume1/moravec.htm> [July 8, 2002]. 48. Talghader, J.J., J.K. Tu, and J.S. Smith. 1995. Integration of fluidically self-assembled optoelectronic devices using a silicon-based process. IEEE Photonics Technology Letters 7(11): 1321–1323. 49. Esener, S.C., D. Hartmann, M.J. Heller, and J.M. Cable. 1998. DNA-assisted microassembly: A heterogeneous integration technology for optoelectronics. Paper # CR70-07 in Heterogeneous Integration: Systems on a Chip, Proceedings of SPIE, Volume CR70. A. Husain and M. Fallahi, eds. Bellingham, Wash.: The International Society for Optical Engineering. 50. Hartmann, D.M., M. Heller, S.C. Esener, D. Schwartz, and G. Tu. 2002. Selective DNA attachment of micro- and nanoscale particles to substrates. Journal of Materials Research 17(2): 473– 478. 51. Ozkan, M., C.S. Ozkan, O. Kibar, and S.C. Esener. 2000. Massively parallel low-cost pick-and-place of optoelectronic devices by electrochemical fluidic processing. Optics Letters 25(17): 1285–1287. 52. Ozkan, M., C.S. Ozkan, O. Kibar, M.M. Wang, S. Bhatia, and S.C. Esener. 2001. Heterogeneous integration through electrokinetic movement. IEEE Engineering in Medicine and Biology Magazine 20(6): 144–151. 53. Hartmann, D.M., O. Kibar, and S.C. Esener. 2001. Optimization and theoretical modeling of polymer microlens arrays fabricated using the hydrophobic effect. Applied Optics 40(16): 2736– 2746. 54. Hartmann, D.M., D.J. Reiley, and S.C. Esener. 2001. Microlenses self-aligned to optical fibers fabricated using the hydrophobic effect. IEEE Photonics Technology Letters 13(10): 1088– 1090. 55. Harper, C.A. 2000. Electronic Packaging and Interconnection Handbook, Third Edition. New York, N.Y.: McGraw-Hill. 56. Gilleo, K. 2002. Area Array Packaging Handbook. New York, N.Y.: McGraw-Hill. 57. Kuo, W., W.T.K. Chien, and T. Kim. 1998. Reliability, Yield, and Stress Burn-In: A Unified Approach for Microelectronics Systems Manufacturing and Software Development. Boston, Mass.: Kluwer Academic Publishers.

OCR for page 143
Implications of Emerging Micro- and Nanotechnologies 58. Kim, T., W. Kuo, and W.T.K. Chien. 2000. Burn-in effect on yield. IEEE Transactions on Electronics Packaging Manufacturing 23(4): 293–299. 59. Shideler, J.A, T. Turner, J. Reedholm, and C. Messick. 1995. A systematic approach to wafer level reliability. Solid State Technology 38(3): 47–54. 60. Kim, T., W. Kuo, and W.T.K. Chien. 2000. Burn-in effect on yield. IEEE Transactions on Electronics Packaging Manufacturing 23(4): 293–299. 61. Kim, T. and W. Kuo. 1999. Modeling manufacturing yield and reliability. IEEE Transactions on Semiconductor Manufacturing 12(4): 485–492. 62. Kuper, F., J. van der Pol, E. Ooms, T. Johnson, R Wijburg, W. Koster, and D. Johnston. 1996. Relation between yield and reliability of integrated circuits: Experimental results and application to continuous early failure rate reduction programs. Pp. 17–21 in Proceedings of the 1996 34th Annual IEEE International Reliability Physics Symposium. New York, N.Y.: Institute of Electrical and Electronics Engineers. 63. Van der Pol, J., F. Kuper, and E. Ooms. 1993. Relation between yield and reliability of integrated circuits and application to failure rate assessment and reduction in the one digit fit and ppm reliability era. Microelectronics and Reliability 36(11/12): 1603–1610. 64. Shideler, J.A, T. Turner, J. Reedholm, and C. Messick. 1995. A systematic approach to wafer level reliability. Solid State Technology 38(3): 47–54. 65. Kuo, W., and T. Kim. 1999. An overview of manufacturing yield and reliability modeling for semiconductor products. Proceedings of the IEEE 87(8): 1329–1344. 66. Kim, T., and W. Kuo. 1999. Modeling manufacturing yield and reliability. IEEE Transactions on Semiconductor Manufacturing 12(4): 485–492. 67. Christensen, C.M. 2000. The Innovator’s Dilemma: When New Technologies Cause Great Firms to Fail. New York, N.Y.: HarperBusiness. 68. Personal Communication between Keith Uncapher, Dean Emeritus, University of Southern California, and Al Pisano, Committee on Implications of Emerging Micro and Nano Technologies, March 2001. 69. Mignardi, M. 2001. The Digital Micromirror Device: Its Fabrication and Use. Presentation by Michael Mignardi Texas Instruments, to the IEEE weekly seminar, Southern Methodist University, Dallas, Texas, November 27.