The field of MEMS has been recognized internationally only within the last few years, although it is rooted in efforts on sensors and actuators that go back thirty years or more (Wise 1991; Wise and Najafi 1991). The field has been driven by the rapid global progress in the field of microelectronics, where solid-state microprocessors and memory have revolutionized many aspects of instrumentation and control, and have facilitated explosive growth in data processing and communications for more than three decades. Many of the emerging application areas for microelectronics, however, deal with nonelectronic host systems, and thus require that parameters such as pressure or flow be measured and converted to electrical signals that can be processed by computer. After the necessary control decisions are made electronically, the resulting electronic signals can then be fed to actuators to control the parameters of the host system. Figure 1.1 shows the organization of such a control loop, which represents a general microelectromechanical system. It is the promise of MEMS that such systems will eventually find realization in highly integrated, low-cost, and very accurate forms.
By the early 1980s, progress in microelectronics had reduced the cost of a microprocessor to less than that of a typical silicon sensor; today the peripheral functions of sensing and actuation continue to represent the principal bottlenecks in the application of microelectronics to many emerging systems, not only in terms of cost but in terms of reliability and accuracy as well. It is thus evident that continued progress in sensors, actuators, and MEMS is likely to exert considerable leverage on the microelectronics industry beyond the considerable direct markets for these products. Progress will enable the use of microprocessors and memory that otherwise could not be applied. And the direct markets themselves are significant. Indeed, Semiconductor Equipment and Materials International (SEMI) recently noted (SEMICON 1993) that SEMI's mission "now includes not only a commitment to semiconductor technology but to other related industries such as micromachining (MEMS), multichip module, and flat panel display technologies (FPD).... By 1999, it is expected that micromachining and flat panel displays will add some 25 percent more to the market opportunities in dollar terms, giving the total between semiconductors, MEMS and FPD over a $100 billion dollar market potential." SEMI also noted that "MEMS are integrated sensors, actuators, and electronics fabricated with processes similar to those used for ICs [integrated circuits].... By the turn of the century, worldwide sales of MEMS devices could reach $8 billion."
Figure 1.1. Structure of a sensor/actuator control loop typical of evolving microelectromechanical systems. In MEMS, such loops will become highly integrated, reduced in some cases to the level of a single chip.
The role of MEMS in extending electronic signal processing to new types of systems was recently emphasized by K.J. Gabriel (1993) as depicted in Figure 1.2.
Figure 1.2. The role of sensors and actuators in extending integrated microsystems beyond information processing and communication into information gathering and control.
The addition of integrated sensors to microelectronics allows information gathering to occur in highly integrated systems in addition to the more traditional roles of information processing and communication. Beyond that, the emergence of microactuators promises the ability to exert significant measures of control over nonelectronic events at very small sizes. The ability to do sensing and actuation at low cost in distributed systems promises to significantly extend microelectronic applications. However, there is another dimension to MEMS. Microelectronics, most integrated sensors, and many microactuators are based on the ability to batch-fabricate miniature component assemblies. In electronics, this has permitted an explosion of information processing capability that is well known. However, there is a possible parallel path for purely micromechanical parts and subassemblies in which these devices are also batch-fabricated. While many mechanical systems in the macroworld must be large to perform their tasks, there are other applications involving functions at the millimeter or micron levels where perhaps they do not. This parallel path is shown in Figure 1.3. One portion of the current Japanese effort is addressing such micromechanical systems, including micromachines, which represent an extension of both micromechanics and MEMS. Indeed, while in the United States most of the players in the MEMS field originated in the lithography-based world of microelectronics, a larger fraction of the Japanese efforts is derived from mechanical engineering and is based on the miniaturization of more conventional machining processes, including microgrinding and electro-discharge machining (EDM). These nonlithographic processes would presumably be extended to high-volume applications through their use in creating master molds, which could be used for parts replication.
Figure 1.3. Extension of batch-fabrication concepts to microelectronics, MEMS, micromechanics, and micromachines.
Figure 1.4 gives an example of a miniature micromotor and MEMS gear train fabricated at the University of Wisconsin using the LIGA process (Guckel et al. 1991). In LIGA, X-ray lithography is used to form very precise high-aspect ratio patterns in a thick polymer (photoresist), which is then used as a mold to be filled using electroplating. When nickel is used as the plated material, the resulting parts can be driven magnetically. Japanese efforts in LIGA will be discussed in Chapter 2.
For our purposes we will define "MEMS" to mean batch-fabricated miniature devices that convert physical parameters to or from electrical signals and which depend on mechanical structures or parameters in important ways for their operation. Thus, we include batch-fabricated monolithic devices such as accelerometers, pressure sensors, microvalves, and gyroscopes fabricated by micromachining or similar processes. We also include microassembled structures based on batch-fabricated parts, especially when batch assembly operations are used; but we have elected not to focus on individually-fabricated devices which are unlikely to see wide use. It is expected that an interface to electronic signal processing will exist in most MEMS, which implies that they will include sensors, actuators, or (in most cases) both. New materials and processes such as LIGA were an important part of this JTEC study along with testing, packaging, and many issues associated with the design and development infrastructures needed for MEMS. Image sensors, chemical sensors, and purely thermal or magnetic devices, however, are not covered specifically in this report even though they often involve technology and generic microstructures that are similar to those in MEMS and are often lumped under this acronym. In this sense, "MEMS" is a bit of a misnomer and a more general term is needed. Indeed, the most frequent answer to the question of "After pressure sensors and accelerometers, what is the next major sensor based on MEMS that will be mass produced in high volume?" was "chemical sensors for medical applications."
Figure 1.4. Nickel micromotor and gear train formed using the LIGA process at the University of Wisconsin (Guckel et al. 1991). Such structures combine extreme precision with high aspect ratios, can be driven magnetically, and provide one example of MEMS. The rotor diameter here is 150 mm. Magnetic micromotors have been driven at rates exceeding 50,000 rpm.
As noted above, MEMS began as an outgrowth of expanding efforts to realize sensors and actuators using solid-state technology. During the past thirty years, considerable strides have been made in this area (Wise 1991; Wood, Han, and Kruse 1992). Beginning with visible image sensors in the mid-1960s and then pressure sensors in the 1970s, most efforts to realize sensors have drawn extensively from integrated circuit technology and been silicon-based. In the 1980s, accelerometers emerged as additional high-volume product targets, driven primarily by needs in the automotive industry. Both microactuators and MEMS were born during this decade. Today, visible image sensors are approaching the resolution of photographic film and offer the promise of automatic electronic processing of both video and still images. Infrared imaging has similarly resulted in large area arrays, and recently an uncooled array based on micromachining has been demonstrated and shown to produce excellent results in night-vision applications (Wood, Han, and Kruse 1992). Solid-state pressure sensors have been demonstrated over a broad range of applications, from ultrasensitive devices capable of serving as solid-state microphones or capacitive manometers to rugged devices used in electronic transmissions and in the hydraulic control of heavy equipment, spanning at least eight orders of magnitude in pressure. A variety of accelerometers are being merged with on-chip circuitry for high-volume applications (Payne and Dinsmore 1991), and inertial navigation systems based on integrated gyroscopes (Bernstein et al. 1993) are in development in a number of companies. A high-density projection display system based on arrays of electrostatically-driven micromirrors is also in advanced development (Sampsell 1993). Microflowmeters are emerging for industrial process applications, and still other devices are being designed for chemical sensing and for applications in decoding the human genome. Many of these emerging applications are potentially very high in volume and very important to global society.
Table 1.1 lists some of the devices currently in production or in development worldwide. Many of the devices at the top of the columns are currently in production as individual components, although most are evolving upward in sophistication and accuracy. Toward the bottom of the list are more complex systems, including inertial navigation equipment, chromatography systems, mass spectrometers, and chemical sampling/analysis systems. These are in relatively early stages of development, but represent the realization of entire instrumentation systems in miniature, highly-integrated, and potentially low-cost forms. The evolutionary march toward such microinstrumentation systems is expected to continue, both in Japan and in the United States.
While increasingly recognized as important in the implementation of a wide variety of emerging systems, enthusiasm over the realization of sensors, actuators, and MEMS using microelectronic technology must be tempered by the realization that many such devices have been around in various forms for a long time. An example is the gas chromatograph depicted in Figure 1.5. This system integrates a gas sampling/injection system, separation column, temperature control, gas conductivity detection, and associated signal processing on a single chip and was first proposed in 1972 (Wise, Carle, and Donaldson 1972; Terry, Jerman, and Angell 1979). Lack of suitable microvalves was a principal challenge to the realization of this system, but with recent advances in MEMS, such systems may soon be a reality.
Integrated Sensors and MEMS in Production
or Under Development on a Worldwide Basis
Figure 1.5. Organization of a miniature gas chromatography system, an early example of MEMS. All components except the display and main entry valve were proposed as a single chip in 1972.
The commercialization of integrated sensors, actuators, and MEMS has been relatively slow, and in many areas technology push has been considerably stronger than market pull. This has probably been more true in the United States than in Japan and may be due in part to the more extensive system of university research in the United States, which has produced more research prototypes but also been somewhat less dependent on industry. In some cases, the control system hierarchy into which these devices must work must also change considerably in order to accommodate them, and the general lack of synergy between the sensing and control areas globally has undoubtedly retarded progress. Simply replacing earlier devices with solid-state versions has not fulfilled the potential of this area, and going beyond simple component replacement into integrated microsystems requires interdisciplinary cooperation that is difficult to achieve. Has the technology associated with MEMS now matured to the point where high-volume devices can be realized using batch processes with high yields? If not, when will it do so? Can the specialized processes required for the realization of microelectromechanical microstructures really be successfully merged with circuit processes to form microsystems on a chip? Is this necessary, or will hybrid systems do just as well? Where are the markets for these devices that will demand continuous technology improvements similar to what memory has done for microelectronics? What are the principal high-volume markets that will fuel the sensor/actuator/MEMS industry and significantly benefit society? These are some of the questions considered in this study, where the perceptions and approaches found in Japan are contrasted with those in the United States.