SENSOR DEVELOPMENT IN JAPAN

Industry

Much of the work on MEMS-related technology that the JTEC panel saw in Japanese industrial research laboratories focused on sensor applications. This is undoubtedly due to the long history of microsensor technology development, and to the relatively near term nature of commercial sensor products compared to the more speculative nature of many microactuator and micromachine applications. Like most of their U.S. counterparts, MEMS sensor technology developments in Japanese industry concentrate on Si-based technology, including use of integrated electronics.

This section will focus on companies visited by the JTEC panel. This group does not include all sensor research in Japan, but is believed to represent MEMS sensor R&D in Japanese industry. Because of the proprietary nature of most industrial research, the work discussed by the Japanese laboratories was mostly that which had been previously presented or published. There is doubtless a great deal more research underway that is not open for discussion. In some cases, speculation can be made as to present research activities not revealed.

Toyota Central Research and Development Laboratory. This large laboratory has a long history of significant sensor R&D. There are many present and future applications for sensors in automobiles, with requirements for low cost, high volume production, and high reliability. Accordingly, the focus of MEMS research at the Toyota Central R&D Laboratory is almost entirely on silicon microsensors, including an emphasis on monolithic integration of sensors with electronics. This represents a relatively large effort at Toyota, with about ten researchers dedicated to silicon microsensors plus significant support from groups in materials, IC design and processing. A very capable CMOS prototype line is in place at the laboratory, and is used for development of integrated sensors.

Development of physical sensors based on bulk silicon micromachining began many years ago at Toyota; they are reportedly in production. Now the R&D emphasis is on sensors built using surface micromachining of thin film materials such as polysilicon and silicon nitride, along with a major thrust on integration of sensors with CMOS electronics.

Much of the silicon sensor device research at Toyota has been on pressure sensors, critical devices for a variety of control applications in automobiles. This can even involve the very difficult job of sensing pressures within the engine combustion chamber (Morikawa et al. 1993). Toyota's pressure sensor work is generally innovative, and equivalent to similar efforts in the United States. An earlier example of a fully integrated pressure sensor is given in Yamashita et al. (1989). A more recent example of an innovative pressure sensor is the recent surface micromachined microdiaphragm device shown in Figure 3.1 (Shimaoka et al. 1993). This device uses a silicon nitride diaphragm with polysilicon piezoresistors, resulting in a small area diaphragm only 100 mm diameter by 1.6 mm thick. Such small devices can be efficiently combined with electronics for low cost integrated sensors. The ease of integration with electronics is illustrated by an earlier version of the microdiaphragm pressure sensor that was integrated into a 32 x 32 (1K)-element array shown in Figure 3.2 (Sugiyama et al. 1990). This large sensor array is complete with CMOS addressing and readout circuitry, and can be used for tactile sensing.

Another pressure sensor, developed in conjunction with researchers at Toyoda Machine Works and Tohoku University, is shown in Figure 3.3 (Nagata et al. 1992a). This is a bulk micromachined capacitive device, formed by bonding silicon and glass wafers, complete with some CMOS electronic circuitry. It has a 2 mm square diaphragm, much larger than the previous device. The device is built for a low pressure range of 0 to 20 Pa, and can have high overpressure protection. It is combined with a digital electronic chip for compensation and digital output. Using a similar concept and structure, many of the same authors have reported a capacitive accelerometer (Nagata et al. 1992b), another useful automotive sensor for air bag deployment and active suspension control.

Other silicon microsensor work at Toyota has involved uncooled infrared (IR) sensors and arrays utilizing the thermal isolation properties of silicon microstructures. One type detects the temperature rise due to incident IR radiation through a polysilicon pn junction diode (Tanaka et al. 1992), carried out with researchers from Hamamatsu. Another uses a thin film polyvinylidene fluoride (PVDF) pyroelectric (Asahi et al. 1993), as shown in Figure 3.4. IR detection and imaging is an excellent application of the features and advantages of microstructures. These microstructures have high thermal isolation and low heat capacity, and so undergo detectable temperature changes from low levels of incident IR radiation.

In addition to the work on silicon microsensors, there is a similar-sized group working on gas sensors, typically for exhaust emission control. The emphasis is on thin film materials, such as ZnO and SnO. This technology does not use silicon or significant micromachining technology, so is not in the realm of MEMS.


Figure 3.1. Schematic and cross-sectional view of Toyota's surface micromachined piezoresistive microdiaphragm pressure sensor (Shimaoka et al. 1993).


Figure 3.2. Toyota 32 x 32 (1K)-element piezoresistive pressure/tactile sensor array (Sugiyama et al. 1990).


Figure 3.3. Photograph (a) and cross-sectional structure (b) of Toyota capacitive pressure sensor with CMOS electronics (Nagata et al. 1992a).


Figure 3.4. Toyota's pyroelectric IR sensor (Asahi et al. 1993): schematic cross-sectional structure (a) and photograph of the sensor chip before forming an absorbent (b).

Nippondenso Research Laboratories. Since Nippondenso is a member of the Toyota Group of companies and a major automotive component supplier, its focus on sensors using MEMS technology is similar to that of Toyota Central R&D Laboratory. Sensors were stated as Nippondenso's first target for MEMS technology, and the laboratories are extensively developing suitable process technology. Bulk micromachined sensors are in production, and efforts are under way on sensors using poly-Si surface micromachining and wafer bonding. The laboratories' efforts are large, with about thirty people working on micromachining and about ten on the national (MITI) Micromachine Technology Project.

Specific developments shown include an integrated pressure sensor for engine control and an air bag crash sensor/accelerometer. The pressure sensor is a rather standard piezoresistive sensor with a square anisotropically-etched diaphragm but with considerable integrated electronics on a 2.8 mm square chip. The crash sensor is based on a bulk micromachined cantilever beam with piezoresistive sensors, also with extensive integrated electronics. The chip size is 8.3 x 3.6 mm, and is packaged in silicone oil for damping. No detailed information on these devices was available, and no published references have been found. It is most likely that these devices are in or near production. An example of recent work published by Nippondenso is a highly integrated sensor chip with multiple diaphragm pressure sensors and temperature sensors, shown in Figure 3.5 (Fujii, Gotoh, and Kuroyanagi 1992). This device includes use of wafer bonding for cavity and port fabrication, as well as thin film silicon nitride diaphragms similar to the Toyota work described above.


Figure 3.5. Top view of the Nippondenso integrated pressure and temperature sensor chip (Fujii, Gotoh, and Kuroyanagi 1992).

Another interesting device mentioned briefly was a yaw rate sensor (gyro), apparently for automobile steering/braking/suspension control. This sensor was not based on silicon, but was a vibrating metal tuning fork with piezoelectric drive. Similar devices have been demonstrated at several laboratories worldwide, often made from piezoelectric quartz and aimed at low performance inertial systems on missiles, for example, but are relatively expensive. The Nippondenso approach may allow lower-cost fabrication while still providing useful sensitivity for automobile control.

Nippondenso is also working on a variety of wafer bonding technologies, as are U.S. industrial laboratories developing advanced sensors. An earlier paper described an integrated pressure sensor with dielectric isolation, fabricated by silicon wafer bonding (Fujii et al. 1990). Wafer bonding technology is extremely useful for both sensor fabrication and sensor packaging at the wafer level, which is the likely focus of Nippondenso's work.

Nippondenso's work on the MITI micromachine project involves sensors and microactuators, but is said to use materials other than silicon, such as piezoelectrics.

Yokogawa Electric Corporation. In contrast to Toyota and Nippondenso, whose businesses require high volumes of low-cost sensors of moderate performance, Yokogawa's industrial sensing and control business is aimed at high performance sensors that can sell for higher prices in modest volumes. Pressure sensing is the dominant industrial application addressed by MEMS sensors. For Yokogawa's pressure sensing applications, silicon sensor technology is found to be advantageous because of its high performance capabilities.

The recently introduced Yokogawa DPharp industrial pressure transmitter is a high-performance industrial instrument which derives its performance from a unique, state-of-the-art silicon sensor chip. This chip incorporates an electromagnetically driven resonant beam sealed in a vacuum shell, integrated onto a standard silicon pressure sensor diaphragm. It is shown in Figure 3.6 (Ikeda et al. 1988, 1990a, 1990b). Driven into resonance by closed-loop feedback electronic circuitry, the beam is stressed as the diaphragm deflects under pressure, with the resulting change in resonant frequency serving as a highly sensitive measure of pressure. As implemented, there are two such beams on the diaphragm, one in the center and the other at an edge, used differentially. The device offers wide dynamic range, high sensitivity, and high stability, along with a frequency output for easy interface to digital compensation circuitry. The chip fabrication is challenging, comprising four sequential selective epitaxial growth and critical selective etching steps, and the sensor package is extremely complex. It is expensive to produce. It is, nonetheless, an excellent example of the power of silicon microfabrication technology, and represents a major technical achievement.


Figure 3.6. Yokogawa resonant microbeam pressure sensor (Ikeda et al. 1990b): construction of the sensor (a) and cross- sectional SEM photograph of the resonator (b).

The DPharp product is a smart pressure transmitter used to instrument industrial processes. It includes the packaged sensor chip, the associated analog drive/sense circuitry, and digital compensation for linearization, temperature compensation, diagnostics, and communication (Saito et al. 1992). This smart transmitter approach is similar to those offered by Honeywell, Rosemount, and others in the United States. The performance of the Yokogawa transmitter product is excellent, with accuracy of about ▒ 0.01% over a wide temperature and pressure range, equal to the best of such transmitters on the market.

Others. Many other Japanese industrial laboratories appear to be developing MEMS-based sensors, primarily in silicon technology. While Matsushita Research Institute, visited by the JTEC panel, carries out little sensor development, it was stated that the Matsushita Living System Research Laboratory has extensive sensor developments under way. The thrust of that laboratory is consumer products, such as appliances, which are using sensors increasingly. Matsushita has a goal of replacing many conventional sensors with semiconductor-based devices; the major reasons are cost and reliability improvements.

Hitachi representatives have described the development of a closed-loop silicon accelerometer (Tsuchitani et al. 1991). It is intended for automotive systems, and is scheduled for release as a product in 1994. This device, shown in Figure 3.7, requires sophisticated double-sided etching to achieve the desired symmetrical structure that eliminates cross-axis sensitivity (Koide et al. 1992). The 3.2 x 5 mm silicon chip is bonded between glass plates to form a differential capacitor, and mounted with hybrid electronics.

At several of the Japanese companies visited, sensor activities were not discussed in any detail, although they are probably taking place: Seiko makes a number of solid state sensors, using both silicon and quartz piezoelectric technology, but specifics were not described. Omron indicated an emphasis on silicon sensors, including bulk and surface micromachining along with electronic integration, but no specifics were given. Sensor developments also were not covered at Canon or Olympus. At most of these locations, the conversations focused on microactuators and micromachines, rather than sensors.

While not visited by the JTEC panel, a number of other Japanese companies have MEMS sensor technology under development or even in production. Examples include: (1) Fujikura Ltd. has for some years sold piezoresistive silicon pressure sensors as components in the United States and Japan for commercial (including automotive) applications. Recent reports include an integrated pressure sensor with an anisotropically etched diaphragm, bipolar analog signal conditioning electronics, and voltage output (Itoh, Adachi, and Hashimoto 1992). Fujikura is also offering a silicon piezoresistive accelerometer product. (2) Fuji Electric Company has reported a somewhat similar integrated pressure sensor with isotropically etched diaphragm and bipolar electronics, achieving good performance of ▒1% over -50 to +175íC (Kato et al. 1991). These devices seem comparable to integrated pressure sensors recently available in the United States. (3) Nissan Motor Company Central Engineering Laboratories has reported an integrated accelerometer, shown in Figure 3.8, employing a bulk micromachined cantilever, piezoresistive readout, and bipolar electronics for amplification and temperature compensation (Muro et al. 1992).


Figure 3.7. Hitachi closed loop capacitive accelerometer (Koide et al. 1992): schematic (a) and SEM photograph (b).


Figure 3.8. Nissan integrated silicon accelerometer (Muro et al. 1992).

Universities

Unlike Japanese industrial laboratories, most of the Japanese universities visited do not have large MEMS sensor research efforts. Instead, university research is on long-term, innovative approaches to microactuators, micromachines, and microrobots. Carrying out leading edge sensors research at the level of the best United States and European universities requires strong capabilities in materials and processing, including electronic circuit integration. The facilities, technology capabilities, and critical mass needed to do are not present at most Japanese universities. Some niche sensor research is carried out in these and other universities within the technology and facilities available. The major exception is the extensive sensor research program at Tohoku University.

Tohoku University. The research program under Professor Masayoshi Esashi at Tohoku University appears to have by far the largest effort and best facilities directed at sensors R&D at Japanese universities. The facilities are modern and cover a broad range of capabilities, including microfabrication, electronic integration, and devices testing. These facilities are on a par with the best of the U.S. universities engaged in sensor research.

Efforts at Tohoku range from basic material and process development, to development of specific sensors, to specialized sensor packaging. Several sensors were reported as being successfully commercialized in the past, including two ion-sensitive field-effect transistors (ISFETs) for measuring pH and pCO(°2), along with a capacitive pressure sensor. The integration of sensors and electronic circuitry is seen as a dominant trend, and much effort is being devoted to integrated packaging of sensors using glass-silicon bonding (Esashi 1993).

The current research projects on sensors include considerable work on a silicon capacitive accelerometer, including the use of electrostatic force rebalance for expanded dynamic range (Matsumoto and Esashi 1992). The device, shown in Figure 3.9, uses integrated capacitive readout circuitry, and an external phase-locked loop force rebalance circuit. The silicon proof mass is suspended from silicon oxynitride flexures, and the chip is sandwiched between glass covers that include the capacitor electrodes and holes for electrical leadouts, as well as providing overrange protection and damping. The accelerometer operates over the range of a few gravities with good linearity.


Figure 3.9. Tohoku University integrated capacitive accelerometer (Matsumoto and Esashi 1992).

Also developed at Tohoku is the integrated capacitive silicon pressure sensor shown in Figure 3.10 (Matsumoto, Shoji, and Esashi 1990). A Pyrex glass cover is sealed to provide a vacuum reference cavity. The silicon chip includes a bossed sensor diaphragm and CMOS capacitance-to-frequency converter circuitry. The full span was 600 mmHg, with a frequency scale factor of 30 Hz/mmHg. This device technology was probably the pressure sensor described as being commercialized.


Figure 3.10. Tohoku University integrated capacitive pressure sensor (Matsumoto, Shoji, and Esashi 1990).

Another device is a resonant infrared radiation (IR) sensor, which combines bulk silicon micromachining, on-chip NMOS circuitry, and vacuum packaging by silicon-to-glass bonding (Cabuz et al. 1993). The principles of operation involve measuring the change in frequency of a silicon microstructure resonator induced by mechanical stresses due to the heating by incident IR. Frequency sensitivities of about 300 ppm/mW were observed. Also reported were a thermal mass flowmeter and integrated paramagnetic oxygen sensor based on the flow sensor (Esashi 1991). This device combines silicon micromachining and silicon-glass bonding for its fabrication.

Others. The research efforts of Professor M. Kimura at Tohoku-Gakuin University include MEMS-based sensor developments. The Tohoku-Gakuin laboratory facilities are well laid out, using mostly donated equipment, but somewhat limited in capabilities. Professor Kimura works with industry to license his patents and transfer his technology.

One thrust of Professor Kimura's research has been on devices based on thermal isolation bridge microstructures. These have been used for flow and infrared sensors, for example (Kimura 1986). Some gas sensor development by the Ricoh Company is based on this work.

No significant MEMS-based sensor development was discussed at the other Japanese universities JTEC visited. The recent sensor publications from Japanese universities are also limited in number and scope.

Gas, Chemical, and Humidity Sensing

There is excellent work going on at Japanese universities and industry in the areas of microminiature solid state gas, chemical, and humidity sensors, where Japan appears to be doing considerably more research than the United States. For example, the review paper by Yamazoe and Miura at Transducers '93 gives an excellent summary of solid state gas sensor research (1993). These devices do not come under the present definition of MEMS or within the expertise of the panel, and are not included in the present study. However, solid state gas sensor research is an area of strong Japanese capability.

Role of the National Micromachine Technology Project

Sensors listed as required for MITI's micromachine program (Micromachine Center 1993) include microgyroscopes, ultrasonic sensors, and miniature vision and optical sensors. At this stage, the national micromachine program appears to have little effect on the development of sensor technology. MEMS sensor technology is already well advanced, largely based on silicon microfabrication, and enjoying considerable commercial application. The MITI micromachine effort is much more focused on micromachines, microactuators, and so forth. The sensors required seem not yet clearly defined, but may use nonsilicon approaches in many cases. Sensors do not appear to be critical, pacing items in either the national micromachine project or in other micromachine or microrobotic developments.


Published: September 1994; WTEC Hyper-Librarian

MEMS-Based Sensors G. Benjamin Hocker