Trends in Electromechanical Transduction I1ene J. Busch-Vishniac Aerospace Mechanical Engineering Department, The Universiy of Texas Austin, R 78712 and and Mechanical Engineering Department, Boston University, 110 Cummington St., Boston, MA 02215 Abstract: Electromechanical transducers are devices which monitor the state of a system or impose a state on a system, and which have one mechanical port and one electrical port. Examples include strain gages and motors. This article examines trends in electromechanical transduction, including the fising pervasiveness of sensors and actuators, the increasing demands placed upon them, the expansion of applications arenas, the push for miniaturization, the growing concern wi[h actuators, and the escalating pace of change. ~TRODUCTION Although most naturally occurring events are primarily either mechanical or chemical in nature, it is common to convert information into an electrical form in order to take advantage of the speed and linearity of electrical systems. The devices that accomplish this energy conversion from one form to another are referred to as transducers. Electromechanical transducers are devices in which one port is associated with electrical energy and one with mechanical energy. Examples include accelerometers, strain gages, loudspeakers, and motors. The largest breakdown generally divides transducers into two There are many ways to categorize transducers. classes which depend on broad function. Those transducers used to monitor the state of a system, ideally without affecting that state, are sensors. Those transducers which impose a state on a system, ideally without regard to the system load, are actuators. The 1970’s and 1980’s brought dramatic changes in electronics and in signal processing techniques, but only As a result, transducers are often the least reliable and most modest changes in electromechanical transducers. expensive elements in measurement and control systems. For this reason, there is a growing emphasis on the field of transduction, and significant changes are beginning to emerge. This article examines the current trends in electromechanical transduction and some of their repercussions. These trends include the rising pervasiveness of transducers, the increasingly stringent demands on transducers, an expansion of applications arenas, a push for miniaturization of sensors and actuators, a growing focus on actuators, and a reduction in the time between discovery of new materials and their introduction into transducers. PERVASWENESS In the last few decades, the decreasing cost of electronics has prompted its adoption into products of all sorts. The growth of electronics in consumer products has been driven by two phenomena: the perception that low technology (nonelectronic) devices are not as good as high technology (electronic) devices, and the push for products with intelligence. Examples of low technology devices whose market is being overtaken by high technology counterparts include office equipment, such as staplers and pencil sharpeners, and kitchen appliances, such as juice squeezers. In many cases this leads to the introduction of electromechanical sensors and actuators. The growing market for intelligent products (i.e. products which include a decision-making process) is prompted by the desires to automate some functions that are performed by people and to add functions that humans cannot perform. Examples of common intelligent products that perfom jobs humans are not able to do are smoke detectors, automobile airbags, and clothing dryers with autodry cycles, In each case sensors aid the decision-making process. Because of the move to high technology devices the growth in sensor and actuator markets has been rapid and is predicted to continue on its current pace through the turn of lhe century. The sensor market alone rose to a $5 billion industry by 1990 [ 1] with projections for a $13 billion worldwide market by the year 2000 (a compound annual growth of 8%). The bulk of this market is divided pretty evenly between the US, Japan, and Europe, with about 14% of the market share going to the rest of the world [ 1]. DEMANDS ARE ON THE RISE In product design it is well accepted that performance demands always increase over time. ~his is sometimes called the “new and improved” syndrome.) In electromechanical transduction, the recent trend in product demands reflects a dichotomy of purpose and is leading to goals that are, for the most part, mutually exclusive. 995 On the one hand, there are proponents of modularity who see sensors and actuators as components of a larger system, and seek to create as few of these components as are necessary to ensure operation of any useful system created by assembling components. This suggests a need for great flexibility in transducer performance - wide ranges of operation, adjustable sensitivities, standardized shapes, etc. On the other hand, the demands on transducers for performance within any particular system in terms of quality, reliability, and cost are becoming progressively more stringent. This has spawned mechatronics, a view that the design process ought to include models of sensors, actuators, and the controlling electronics as well as the main physical plant. The premise of mechatronics is that it is only through consideration of the mechanics and the electronics in an integrated fashion that one can produce an electromechanical product that performs as desired. Taken to its natural limit, this suggests that there is an optimum transducer for a given set of system performance characteristics. In the extreme this leads to custom transducers for each application, the opposite of the few broadapplication transducers used in a modular view. While there is merit in both views of transducers, there are compelling reasons to favor mechatronics. These are well presented by Whitney [2] who argues that the modular approach to electronic system design will never apply to mechanical systems. Among his reasons: mechanical components typically perform multiple functions rather than one, and they perform differently in a system than they do in isolation (which is not the case for electrical components). NEW APPLICATIONS ARENAS A great deal of the expansion of the electromechanical transducer industry has come about because of new applications emerging, especially for sensors. In 1990 approximately half of the sensor market was for automotive applications including position, temperature, emission, pressure, and acceleration sensors [1]. Another third of the market was for industrial applications, and the rest was dominated by biomedical applications. In these statistics, consumer products other than cars accounted for only about one percent of the sensor market. It is clear that the sensor and actuator market is changing its focus both in application and scale. Examples of With medical costs rising rapidly, the impetus for application shifts are in health care and manufacturing. improvements in patient monitoring and early diagnosis prompt more biomedical applications. This is particularly affecting home products and creating more intelligent devices. An example of this trend is the bathroom scale, which is now being outfitted with sensors to aid in determination of balance problems. In manufacturing applications, the most successful industrial control systems have focused on continuous processes that rely on low speed monitoring of chemical characteristics using electrochemical transducers. In recent years attention has shifted toward unit (or discrete) manufacturing processes and operations performed by specific machines. In discrete processes one normally monitors mechanical variables and might need to sense or actuate on short time scales. This shift has been accompanied by a move to faster and more precise industrial sensors. Another industrial change relates to scales of operation, A growing segment of industrial fabrication facilities is concerned with very small dimensions. For instance, the microelectronics industry is constantly pushing toward smaller feature sizes (and is already at about the 0.1 pm level). Small scale features have created a market for sensors with higher sensitivity, and actuators with greater positioning accuracy and resolution. These characteristics normally are achieved by compromising performance in another area, such as range. M~IATURIZATION A clear trend in electromechanical transduction is the push for extreme miniaturization of sensors and actuators. Further, miniaturization is taken to be synonymous with use of microelectronic fabrication techniques for sensors and actuators. The logic driving the production of silicon sensors and actuators deserves careful consideration. Ideally monitoring a system state without affecting it suggests devices that have a small footprint, i.e. that use low amounts of power and are small in size. For sensors then, the push for extreme miniaturization has some compelling logic. A remaining question is whether miniaturization must lead to silicon sensors fabricated using microelectronic approaches. In general, the arguments that have been put forward for solid state sensors focus on three issues: 1). one can achieve better dimensional control, 2). electronics can be integrated wilh the sensors, and 3). costs can be cut through bulk production. However, it is not clear that these issues resolve the miniature sensor question in favor of silicon-based approaches. It is true that VLSI fabrication techniques offer dimensional control at a level (roughly 0.1 pm) that is difficult if not impossible to achieve by conventional methods. However, the range of dimensions offered is small. In conventional processes it is common to have features which are smaller than largest part dimensions by a factor of 105. Using microelectronics approaches, the same ratio is limited to between 103 and 104. 996 circuit~ is twoThe abili(y to integrate electronics with a sensor is another misleading issue. Microelectronic sensors and actuators dimensional, existing on the surface of a silicon substrate. By contrast, electromechanical require a greater dimensionality, and hence different manufacturing techniques have been developed. Unfortunately, the fabrication techniques for transducer structures and for electronics are not wholly compatible. Thus, integrating the two on a single chip requires making compromises in each component that might not be acceptable. Additionally, yields in the two processes are dissimilar, so integration of electronics with the transducer means that a significant amount of functioning electronics are likely to be trashed. While mass production of microelectronic circuits have led to great reductions in their cost, it is not clear that the same savings would be available for sensors. Consider, for instance, one of the most written about solid state The solid state microphone must compete with the miniature electret sensors so far -- the silicon microphone, microphone made by conventional fabrication processes. Estimates of the worldwide annual production of electret microphones are about 800,000,000 devices. Sizes down to a couple of millimeters in diameter and costs as low as $5 including amplification are commercially available. Against this backdrop, solid state microphones generally offer poorer performance in terms of signal to noise ratio and frequency response (see Scheeper et al. [3] for a good review of performance characteristics), and yet it is not clear that they can be made price competitive. Ideal electromechanical actuators seek to impose a state on a system independent of the system load suggesting a device with a large footprint: large size and high power. Thus, miniaturization is not logical unless there are constraints compelling one [o seek small sized actuators. Given this, the case for solid state actuators is made by arguing in favor of large numbers of small actuators put into arrays. Historically, the question of single large actuators versus arrays of smaller actuators has been called the staging problem. The classic example is the problem of the analog watch which could either use a separate motor to turn each of the hands, or a single motor with gearing. In general, which approach will prevail ought to be compelled by the specific application. Of course, even for those cases favoring multiple small actuators, it is not clear that fabrication in silicon presents advantages to miniaturization using conventional approaches. In addition to the problems cited for miniaturized sensors, miniaturized actuators have the additional problem associated with delivery of significant power. Standard options available on macroscopic scales, such as pneumatic and hydraulic power or combustion of a fuel, cease to be logical on very small scales. Also, the use of microelectronic fabrication approaches to produce transducers requires a transition from conventional design essentially independent of fabrication considerations, to design in which fabrication options are the dominant constraints. To date, the limited microelectronic manufacturing options are severely restricting the miniature transducer types. FOCUS ON ACTUATORS The task of imposing a state on a system is generally more difficult than simply monitoring that system state. Hence, it is typically more of a challenge to design and build an actuator than a sensor. As a consequence, the progress in electromechanical sensors has outstripped that in actuators in the last decades. For instance, consider electric motors, a very common actuator for industrial applications. Of the various \ypes of electric motors available, the squirrel cage induction mo[or is the most common found in industry. Because of its dominance in industry one would expect the motivation for improvements in squirrel cage motors to be substantial, However, the current specifications (starting current of 480- 900%, starting torque of 100- 300Y0, efficiency of 80- 92%) match those given in texts in the 1950s (see for instance, Siskind [4]). There is a general recognition that improvements in electromechanical actuators have been few and far between. As demands for enhanced system performance have increased, there has been a growing focus on actuators with aims of speed, efficiency, size, reliability, and cost improvements. Unfortunately, significant improvements have not materialized to date, a fact that is Iikely to fuel an even greater intensity of research on actuators. RAP~ CHANGE One last trend is that the pace of change is dramatically escalating. Consider, for example, the discovery of the piezoelectric effect in quartz. Jacques and Pierre Curie reported this discovery in 1880 [5] but it was not until 1921 that Langevin [6] produced the first patent for a transducer using the piezoelectric effect in quartz. By contrast, let us consider the modern creation of TerfenolR, a rare earth magnetostrictive material. In magnetostrictive materials, the magnetic permeability is a strong function of the mass density, thus coupling mechanical and magnetic fields. The key US patent on TerfenolR was granted to Savage, Clark, and McMasters in 1981 [7]. Excluding the inventors listed in the materials patent, the first US patent to mention Terfenol in a device was filed in 1984 [8], just three years after the materials patent issued. 997 There are a few obvious reasons for the rapid change in the field of transduction: technology itself is improving rapidly, free information is accessible nearly instantaneously via the worldwide web, and the amount of money at stake is enormous. In general, the effects of rapid change on the field of transduction are the same as those seen in all sectors of industry: design times are growing shorter, there is a rise of small and medium-sized companies to address niche needs, and there is a growing variety in the transducers that are commercially available. PREDICTIONS FOR THE FU~RE Based on the current trends in transduction it is possible to make predictions about the future. First, there is no apparent slowing of the demand for intelligent products so there will be a continued expansion of the markets for sensors and particularly for those that are coupled with electronic controls. Second, the current tension between proponents of modularity and those of mechatronics will certainly be resolved, probably with mutually exclusive applications arenas for each. It is easy to imagine that high volume products with noncritical specifications for sensors and actuators (e.g. toys) will move toward standard transducers, sold as OEM (other equipment manufacturers) products which are purchased in bulk. On the other hand, the demands on high end products will continue to rise. As these demands are already “pressing the envelope” of what is feasible, it is quite likely that meeting them will require a move toward more custom devices. Third, the market for sensors and actuators is very likely to shift focus to match the changing world economy. This suggests a growing concentration on biomedical applications and on consumer products other than the automobile. Fourth, eventually there must be a more logical view toward the role of size scales. The current push for miniaturization of sensors will undoubtedly continue but, it will be a small subset of the currently investigated actuator applications which ultimately will be seen to be improvements over alternative, more conventional approaches. Rules for when the staging problem is best solved by one actuator and when it is best solved by many, smal 1act uators w i]1emerge. Fifth, it is clear that progress in electromechanical actuation is stalled and that this will soon bc viewed as a critical situation. This branch of transduction is certainly ripe for major technological advances, although it is not clear at [his point what form they will take, Given that the history of transduction is steeped in exploitation of new materials, one might anticipate advances achieved through collaborative efforts of materials scientists and transducer designers. Finally, given the pervasiveness of information and the market forces pushing for new sensors and actuators, there will be an diminishing delay from the time new technologies and materials are discovered to the production of sensors and actuators incorporating them REFERENCES 1, Ristic, L. (cd), Sensor Technology and Devices, Boston: Artech House, 1994, ch. 1. 2. Whitney, D. E., “Why Mechanical Design Cannot be like VLSI Design,” Res. in Engr. Design 10, 125-138 (1996), 3. Scheeper, P. R., van der Donk, A. G. H., Olthius, W., and Bergveld, P., “Fabrication of Silicon Condenser Microphones Using Single Wafer Technology,” IEEE J. Microeiectromech. Syst. 1, 147-154 (1992). 4. Siskind, C. S., Electrical Machines: Direct and Alternating Current, New York: McGraw-Hill, 1959. 5. Curie, P. J. and Curie, J., “Crystal Physics - Development by Pressure of Polar Electricity in Hemihedral Crystals with inclined Faces,” (in French), Acad. Sci. (Paris) C. R. Hebd. Seances 91, 294 (1880). A translation appeans in Lindsay, R. B. (cd), Acous[ics:Historical and fhilo~ophical Development, Stroudsburg, PA: Dowden Hutchinson Ross, 1973. 6, Langevin, P., “Improvements Relating to the Emission and Reception of Submarine Waves,” British Patent 145,691, Jul. 28, 1921. 7. Savage, H. T., Clark, A. E., and McMasters, 0. D. , “Rare Earth-Iron Magnetostrictive Materials and Devices Using These Materials,” U.S. Patent 4308474, Dec. 29, 1981, 8. Hasselmark, E. D., Waters, J. P., and Wisner, G. R., “Magnetostrictive Actua[or with Feedback Compensation,” U. S. Paten[ 4585978, Apr. 29, 1986. 998