Micromachined Systems-on-a-Chip: Technology and Applications

James H. Smith

Intelligent Micromachine Department Sandia National Laboratories

Albuquerque, NM 87185-1080

j .h.smith@ieee.org http://www.mdl.sandia.gov/micromachine

(505) 844-3098, FAX: (505) 844-2991

Mark A. Lemkin Berkeley Sensor and Actuator Center University of California at Berkeley

Berkeley, CA 94720

Abstract Sacrificial polysilicon surface micromachining is

emerging as a technology that enables the mass production of complex microelectromechanical systems by themselves or integrated with microelectronic systems. Early versions of these micromachined systems-on-a-chip have already found application in the commercial world as acceleration sensors for airbag deployment (for example, ADI's ADXLSO). Two technologies described here, enable systems with increasing degrees of complexity to be fabricated. The first is a three-level polysilicon micromachining process which includes a fourth polysilicon electrical interconnect level, while the other is a single-level (+ second electrical interconnect level) polysilicon surface micromachining process integrated with 1.25 micron CMOS. Samples of systems-on-a-chip built in these processes such as combination locks, pop-up mirrors, and multi-axis accelerometers are also given.

Integrated Microelectronic/Micromechanical Systems-on-a-Chip

Recently, a great deal of interest has developed in manufacturing processes that allow the monolithic integration of MicroElectroMechanical Structures (MEMS) with driving, controlling, and signal processing electronics. This integration promises to improve the performance of micromechanical devices as well as reduce the cost of manufacturing, packaging, and instrumenting these devices by combining the micromechanical devices with an electronic sub-system in the same manufacturing and packaging process. For example, Analog Devices has developed and marketed an accelerometer' which

s the viability and commercial potential of . They accomplished this task by interleaving,

combining, and customizing their manufacturing processes which produce the micromechanical devices with the processes that produce the electronics. In another approach, researchers at Berkeley' have developed a modular integrated approach in which the aluminum metallization of CMOS is replaced with tungsten to enable the CMOS to withstand subsequent micromechanical processing.

As recently summarized in a review paper by Howe3, micromechanical structures require long, high- temperature anneals to ensure that the stress in the structural materials of the micromechanical structures has completely relaxed. On the other hand, CMOS technology requires planarity of the substrate to achieve high-resolution in the photolithographic process. If the micromechanical processing is performed first, the substrate planarity is sacrificed. If the CMOS is built first, it (and its metallization) must withstand the high-temperature anneals of the micromechanical processing.

Figure 1. Micromachined resonators (le& next to I 5ir CMOS driving electronics (right) fabricated using the embedded micromechanics integration process.

A unique micromechanics-first approach4 which overcomes the planarity issues of building the MEMS before the CMOS has been developed at Sandia. In this approach, micromechanical devices are fabricated in a trench etched on the surface of the wafer. After these devices are complete, the trench is refilled with oxide, planarized using chemical-mechanical polishing, and sealed with a nitride membrane. The wafer with the embedded micromechanical devices is then processed using conventional CMOS processing. Additional steps are added

DISCLAIMER

This report was prepared as an account of work spo~~-~ored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, make any warranty, express or implied, or assumes any legal liabili- ty or respolm'bility for the accuracy, completeness, or usefulness of any information, appa- ratus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessar- ily state or reflect those of the United States Government or any agency thereof.

Portions of this document may be illegible in electronic image produck h a g s are produced from the best available original document.

at the end of the CMOS process in order to expose and release the embedded micromechanical devices. Completed devices are shown in Figure 1. A cross-section of this technology is shown in Figure 2. This technology has also been named as one of the recipients of the 1996 R&D 100 Award.

A collaboration with designers from the Berkeley Sensor and Actuator Center (BSAC) has recently started. BSAC designs for inertial measurement units (three-axis acceleration’ and three-axis rotation rate6*’) have been built using Sandia’s Modular, Monolithic Micro-Electro- Mechanical Systems (M’EMS) technology. Initial lots of devices have been fabricated and the results from accelerometers built in this technology are reported in the next section.

CMOS Device Area Micromechanical Device Area 4 t 4 c

Figure 2. micromechanics approach to CMOS/MEMS integration.

A schematic cross-section of the embedded

Inertial Sensors One of the principal commercial products utilizing

surface-micromachining is inertial sensors as illustrated by Analog Devices’ ADXL1508 and Motorola’s XMMAS40GWB9. These accelerometers find their primary application as airbag-deployment sensors in automobiles, but are also being used as tilt or shock sensors. (In the following discussion, please note the use of g for gravitational acceleration and the word “gram” for mass). The Motorola device is a +/- 40g (full scale) single-axis accelerometer with a noise floor of 400 mg (400 Hz bandwidth, peak) and has an analog output.’ The Analog Devices’ accelerometer is available as either a single (ADXL150) or dual-axis device (ADXL250), has a +/- 50g full scale output, an analog output, and a noise floor of 10 mg (100 Hz bandwidth, r m ~ ) . ~ Note that the use of peak vs. rms noise specifications along with the difference in bandwidth specifications on each device does not lend itself to a direct comparison between the devices.

The application of these types of accelerometers to inertial measurement units is limited by the need to manually align and assemble them into three-axis systems, the resulting alignment tolerances, their lack of on-chip A/D

conversion circuitry, and their lower limit of sensitivity. In order to overcome some of these limitations, a three-axis, force-balanced accelerometer was designed at U.C. Berkeleys for the integrated MEMS/CMOS technology described earlier in this paper. This three-axis accelerometer system-on-a-chip is shown below in Figure 3.

Figure 3. Three-axis accelerometer micrograph with labeling of functional units as reported by Lemkin et ai.’

The performance of the device is summarized in Table 1 below. Approximately an order of magnitude increase in sensitivity is seen over the commercial devices described previously. The accelerometer chip also includes clock generation circuitry, a digital output, and photolithographic alignment of the sense axes. Thus, this system-on-a-chip is a realization of a full three-axis inertial measurement unit that does not require manual assembly and alignment of sense axes.

Although the bias stability of this accelerometer system has yet to be assessed, these noise numbers indicate that this system can now begin to find application in military systems such as

stability control 0 attitude heading reference

short time of flight navigation and commercial applications such as

automotive control 0 automotive diagnostics

automotive navigation virtual reality environmental sensing.”

Table 1 . Performance of the three-axis accelerometer as reported by Lemkin, et al.5

Noise Floor

Dynamic Range (dB, 100 Hz Bandwidth) Proof Mass

Resonant Frequency (kHz)

(Clg/+W

(pgram)

Power Dissipation

A combined X/Y-axis rate gyro and a Z-axis rate gyro have also been designed by researchers at U.C. Berkeley and are presently being fabricated in this new manufacturing process to yield a full six-axis inertial measurement unit on a single chip. The estimated size for this system is approximately 4 mm by 10 mm.

Multi-Level, Planarized Polysilicon Systems-on-a-Chip

pEngine Micromechanical actuators have not seen the wide-

spread industrial use that micromechanical sensors have achieved. Two principal stumbling blocks to their widespread application have been low torque and difficulty in coupling tools to engines. The MDL has developed devices that overcome these issues. Our three-level polysilicon micromachining process"," enables the fabrication of devices with increased degrees of complexity that greatly enhance the ability to couple tools to engines.

This three-level process includes three movable levels of polysilicon in addition to a stationary level for a total of four levels of polysilicon. These levels are each separated by sacrificial oxide layers. A total of eight mask levels are used in this process. An additional friction- reducing layer of silicon nitride is placed between the layers that form bearing surfaces. The inset (lower right) to Figure 4 iIIustrates a bearing formed between two layers of mechanical poly. The overall photo in Figure 4 shows two comb-drive actuators driving a set of linkages to a set of rotary gears. This engine can be rotated by applying driving forces 90' out of phase with each other to each of the comb- drive actuators. Operation of the small gears (shown in the inset) at rotational speeds in excess of 300,000 revolutions

per minute has been demonstrated. The operational lifetime of these small devices exceed 8x10' revolutions. This smaller gear is shown driving a larger (1.6 mm diameter) gearI3 in Figure 4. This larger gear has been driven as fast as 4800 rpm.

Figure 4. Two sets of linear comb-drive actuators driving the gear shown in the inset. This smaller gear drives a 1.6 mm diameter shutter in the lower left of the photo. Inset (lower right) shows a focused ion-beam cross-sectional image of the small gear.

pTransmission To increase the torque available from a rotary drive, a multi-level microtransmission has been de~eloped. '~ This transmission, shown in Figure 5 , employs sets of small and large gears mounted on the same shaft that interlock with other sets of gears to transfer power while providing torque multiplication and speed reduction. The structure shown in Figure 5 couples the output gear of a microengine similar to the engine shown in Figure 4 to a rack and pinion unit that provides linear motion with high force. pLock

The microengine described above has been used to drive the wheel of a pin-in-maze microcombination 10ck.I~ As shown in Figure 6, a pin suspended on an arm driven by an electrostatic actuator is navigated through a maze cut into a wheel driven by the Sandia microengine. Correct positioning of the pin, allows rotation of the larger wheel. Multiple decision points exist in the path on the wheel. All but one path leads to a series of dead ends. A ratchet and pawl mechanism prevent backwards rotation of the large maze wheel.

Figure 5. An electrostatic microengine output gear coupled to a double-level gear train that drives a rack and pinion slider. This gear train provides a speed-reductiodtorque- multiplication ratio of9.6 to I .

Figure 6. A pin-in-maze microcombination lock. The pin must be electrostatically deflected to traverse the maze cut into the large gear.

pMirror The microengine in combination with the

microtransmission can be used to drive a pop-up mirror out of plane. With the torque increase offered by this setup, the mirror can be deflected without the aid of manual probes.

Figure 7. The microtransmission driving a rack and pinion unit to deflect an out-of-plane mirror. The forces generated by this device are suficient to deflect the mirror without the aid of manualprobes.

Summary

Sandia’s Microelectronics Development Laboratory has developed and is advancing a broad range of sensors and actuators based on polysilicon surface micromachining. A technology where micromachined devices are embedded below the surface of a wafer prior to fabrication of microelectronic devices has also been deveIoped and applied to build a three-axis inertial measurement unit system-on-a-chip. A three-level polysilicon process enables intricate coupling mechanisms that link linear comb-drive actuators to multiple rotating gears. This technology has been applied to microengines, microtransmissions, microlocks, and micromirrors.

Acknowledgments

The process development engineers, operators, and technicians of Sandia’s Microelectronics Development Laboratory should be acknowledged for their contributions to the process development and fabrication of these devices. Programmatic support from Michael Callahan of Weapon System 2010 and Tom Hitchcock of Joint DoD/DOE Munitions Technology is greatly appreciated. Sandia is a multi-program laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04- 94AL85000.

References

1. W. Kuehnel and S . Sherman, "A surface micromachined silicon accelerometer with on-chip detection circuitry," Sensors and Actuators A, vol. 45, no. 1, pp. 7-16, 1994. 2. W. Yun, R. Howe, and P. Gray, "Surface micromachined, digitally force-balanced accelerometer with integrated CMOS detection circuitry," Proc. of the IEEE Solid-state Sensor and Actuator Workshop '92, p. 126, 1992. 3. R. Howe, "Polysilicon integrated microsystems: technologies and applications," Proc. Transducers '95, pp. 43-46 (1 995). 4. J. Smith, S. Montague, J. Sniegowski, J. Murray, and P. Mc Whorter, "Embedded micromechanical devices for the monolithic integration of MEMS with CMOS", Proc. IEDM

5. M. Lemkin, M. Ortiz, N. Wongkomet, B. Boser, and J. Smith, "A 3-axis surface micromachined sigma-delta accelerometer," Proc. ISSCC '97, pp. 202-203, 1997. 6. T. Juneau, et. ai, "Dual axis operation of a micromachined rate gyroscope," to be presented at Transducers '97, Jun. 1997. 7. W. Clark, R. Howe, and R. Horowitz, "Z-axis vibratory rate gyroscope," Micromachining Workshop 111: Technology and Applications, Sept. 1996.Juneau 8. Analog Devices, Datasheet: "ADXL 150/ADXL250 rev.0" Nonvood, MA, 1996. 9. Motorola, Datasheet: "XMMAS40GWB/D7" Phoeniz, AZ, 1996. 10. B. Barbour, J. Elwell, and R. Setterlund, "Inertial instruments: Where to now?" hoc . AIAA Guidance and Control Conf., pp. 566-574, 1992. 11. R. Nasby, J. Sniegowski, J. Smith, S. Montague, C. Barron, W. Eaton, P. McWhorter, D. Hetherington, C. Apblett, and J. Fleming, "Application of chemical- mechanical polishing to planarization of surface- micromachined devices," Proc. Solid-state Sensor and Actuator Workshop, pp. 48-53 (1 996). 12. J. Sniegowski, "Multi-level polysilicon surface micromachining technology: Applications and Issues," to be presented at ASME International Mechanical Engineering Congress and Exposition ( 1 996 13. J. Sniegowski and E. Garcia, "Surface-micromachined geartrains driven by an on-chip electrostatic microengine," IEEE Electron Device Letters, vol. 17, no. 7, p. 366 (1996). 14. J. Sniegowski, S. Miller, G. LaVigne, M. Rodgers, and P. McWhorter, "Monolithic geared-mechanisms driven by a poiysilicon surface-micromachined on-chip electrostatic microengine," Proc. Solid-state Sensor and Actuator Workshop, pp. 178- 1 82 (1 996). 15. J. J. Sniegowski, "Micromachining Technology for Advanced Weapon Systems," Roc. Government

Microcircuit Applications conference (GOMAC '97), Las Vegas, NV, March 10-13, 1997, pp. 36-39.

' 95, pp. 609-612, 1995.