AVS1997 Session MM+VT-WeM: Packaging and Assembly/Materials Issues

Wednesday, October 22, 1997 8:20 AM in Room B1/2

Wednesday Morning

Start	Invited?	Item
8:20 AM	Invited	MM+VT-WeM-1 MEMS Packaging - What have We Learned? J.R. Martin (Analog Devices) Micromachined devices have appeared in scientific publications since the 1970's. These papers continue to proliferate but the growth of high volume MEMS commercial products is considerably slower. This lag between concept demonstration and market penetration is largely due to the belief that MEMS devices can be packaged like other semiconductor devices. However, packaging of MEMS products at high yield and low cost involves unusual challenges. Packaging is the bridge between functional devices and Users. Thus the question becomes: "How do we build bridges that turn academic devices into high volume commercial products?" Two areas are crucial to successful MEMS product packaging: 1. Partitioning, 2. The three "S" factors: Surfaces, Stresses, Surroundings. Product partitioning is central to product/process definition. The "S" factors are fundamental materials science issues that affect device performance. Examples based on surface micromachined accelerometers and bulk micromachined pressure sensors will be used to illustrate why careful consideration of these areas is essential to success.
9:00 AM		MM+VT-WeM-3 Flip Chip System Integration for MEMS K.W. Markus, V. Dhuler, R. Mahadevan, D. Koester (MCNC) Recent advances in MEMS continues to push the need for the integration of systems at both the device and the chip level. Many MEMS processes are not directly (monolithically) integrable with IC’s, and monolithic integration is not always the most manufacturable, economical or capable method. For a number of years MEMS system integration has also been accomplishable through the use of flip chip attachment of the MEMS and IC devices, with both being independently manufactured and joined only when necessary. This method also allows the use of substrate materials other than silicon, such as GaAs, glass, ceramics and metals. Continuing development of flip chip MEMS infrastructure options will soon allow designers the ability to design their system electronics for fabrication through the MOSIS fabrication systems and their MEMS through the MUMPs-TechNet MEMS Infrastructure program at MCNC. The resulting system will be integrated through the flip chip MEMS process available at MCNC.
9:20 AM		MM+VT-WeM-4 Instrumentation and Packaging of a Commercial Pressure Sensor for High-g Impact and Martian Environment J.K. Reynolds (Stanford University); R.C. Blue (Jet Propulsion Lab); D.C. Catling (NASA Ames); N. Maluf (Lucas NovaSensor); T.W. Kenny (Stanford University) NASA has initiated a series of space missions to stimulate the infusion of new technologies into the Space Program, among them Deep Space-2¹. Two DS2 Microprobes will separate from the Mars Surveyor lander prior to atmospheric entry, fall ballistically, and slow to ~ 200 m/s at impact. The probes will communicate to an orbiter the measurements of several meteorological and soil sensors. During impact, the instruments must tolerate ~100,000 g deceleration. For absolute measurement of pressure the sensor must not suffer offset or scale errors due to this deployment. Atmospheric pressure changes as small as a few Pa must also be measured at temperature extremes from 0C to less than -70C. Space-rated pressure sensors meeting the mission requirements are not available since the entire instrument must be delivered for integration on the spacecraft no later than 7/97. The accuracy of commercial micromachined pressure sensors is also lower than required, but selection and individual calibration of devices should overcome this. Their offsets can be minimized and temperature coefficients are very stable. We expect to measure atmospheric pressure on Mars to an accuracy of 1% (~10 Pa) over the 2-week lifetime of the mission. Two elements make up the package: a ceramic substrate for secure mounting and wiring of instrumentation electronics, and a soft die attach (Silicone RTV) for strain relief of the sensor. It is a Lucas NovaSensor piezoresistive SFB device with a monolithic temperature correction resistor. The use of this sensor demonstrates 2 important issues: (1) Micromachined silicon pressure sensors can be packaged to survive the violent deployment and wide temperature range present in space applications. (2) Off-the-shelf MEMS devices not intended for high-performance applications can still be used with careful instrument design and precise calibration. The design, testing, and calibration of the instrument package will be presented. ¹This work is supported by JPL/NASA.
9:40 AM		MM+VT-WeM-5 Electrostatic Self-Assembly Aided by Ultrasonic Vibration K.F. Bohringer, M.B. Cohn, K.Y. Goldberg, R.T. Howe, A.P. Pisano (University of California, Berkeley) The term ``self-assembly'' has been applied to spontaneous ordering processes, such as crystal and polymer growth, that are used to build up large assemblages composed of a large number of identical building blocks. Recently "self-assembly" has been proposed for the manufacture of hybrid circuits and micro electro mechanical systems (MEMS) incorporating large numbers of devices (see e.g. [Cohn,Kim,Pisano'91]¹, [Cohn'92]², or [Hosokawa,Shimoyama,Miura'95]³). In this research, we develop an accurate, reliable method for the parallel positioning, the orienting, and the assembling of microfabricated components. This method uses both electrostatic traps and gravity as motive forces. Frictional forces (necessary to keep the elements in the desired configuration) are overcome during the assembly process by vibratory agitation. In this research, we perform experiments to characterize the dynamic and tribological properties of micro-scale parts when placed on a vibrating substrate and in electrostatic fields. We first demonstrate that ultrasonic vibration can be used to overcome friction and stiction of small particles. In a second set of experiments, we describe how particles are accurately positioned using electrostatic traps. Other experiments performed study the effects of air versus vacuum on the levitated parts, and the behavior and interaction of different kinds of shapes of building blocks for micro assembly. We demonstrate parts sorting by size. ¹M. B. Cohn, C. J. Kim, and Albert P. Pisano. Self-assembling electrical networks as application of micromachining technology, TRANSDUCERS, San Francisco, CA, June 1991. ²M. B. Cohn. Self-Assembly of Microfabricated Devices, United States Patent 7 902 986, Sepember 1992. ³K. Hosokawa, I. Shimoyama, and H. Miura. Dynamics of Self-Assembling Systems --- Analogy with Chemical Kinetics, Artificial Life, pp. 413--427, 1(4), 1995
10:00 AM	Invited	MM+VT-WeM-6 Reliability Testing for MEMS Devices S.B. Brown, C.L. Muhlstein (Failure Analysis Associates) Current MEMS manufacturers naturally wish to optimize the performance of their devices. This may mean pushing the mechanical limits of the device materials. Short-term failure modes can be easily tested by increasing the demands, either electrical or mechanical, placed on the manufactured device. Long term reliability is more problematic given that the demands imposed on devices in proof testing may not reveal long term failure modes. It is therefore important to define the mechanical limits of these microelectromechanical devices if they are to be intelligently designed and applied. Failure modes, including crack initiation and propagation processes, have to be characterized if we expect their application without first requiring years of reliability data. Our group’s work has shown how time-dependent failure can occur in single crystal silicon, with or without cracks. Environmental effects can both exacerbate current failure modes and create new ones. We have demonstrated failure modes associated with silicon MEMS in the presence of moisture that were unexpected by the MEMS community. Therefore we cannot apply our knowledge of “macro” failure processes directly to MEMS; failure testing has to be done on MEMS. The observations of fatigue crack growth in single crystal silicon and time-dependent crack initiation in polysilicon are important because they indicate MEMS failure modes that have not been previously recognized. The effects of microstructure, environment, and fabrication can become significant on the micro scale even when they are irrelevant in larger, macroscale structures. The microstructure of MEMS devices is also different from that of large bodies of the same material. This presentation presents the results from our work and discusses other important failure modes that may be particularly important to MEMS.
10:40 AM		MM+VT-WeM-8 Dynamic Analysis of a Parametrically Actuated MicroElectroMechanical System. K.L. Turner, S. Miller, N.C. MacDonald (Cornell University); S.G. Adams (T.M.S. Technologies, Inc.) Stability and control of MEMS are very important when designing devices for most high-Q applications such as gyros and resonant sensors. In most systems with electrostatic actuation, applying voltage to the device changes the stiffness of the system. The dynamic motion of these and other MEMS systems are described by Mathieu-type equations. Many torsional systems have unique properties such that instabilities can occur at certain predictable frequencies other than resonance. Using time-resolved SEM[1], we observed Mathieu instabilities in the first torsional mode of one such system. This vacuum analysis represents the first experimental observation of eight Mathieu instabilities in a microelectromechanical system. The behavior is visible because of the extrmemly low damping of the system. As damping is increased, the number of observable Mathieu instabilities decreases. Interesting and complex dynamic behavior is clearly illustrated with the torsional system described here, a scanning probe z actuator with integrated tip[2]. Two distinct means of actuation can provide motion to torsional systems of this type. The capacitive actuators can be parallel plate drives, or interdigitated plates which are perpendicular to the substrate[3]. In both mechanisms, force is dependent on the angle of twist of the torsion bar. When excited with a sinusoidal voltage, this force dependence causes a parametric term in the equation of motion. The resulting equation is the well-known Mathieu equation[4]. The interesting stability properties of the Mathieu equation lead to novel device behavior. For capacitive sensing/actuating, the device could be driven at a frequency other than the resonant frequency but which is not a subharmonic, therefore reducing the parasitic signal. We have shown the Mathieu frequencies to be very predictable and easily observable using time-resolved SEM techniques. Correlation between theory and experimental results is within 4%. [1] I. Ogo and N.C. MacDonald, J. Vac. Sci. Technol. B 14, 1630 (1996). [2] S.A. Miller, K.L. Turner, and N.C. MacDonald, "Scaling Torsional Cantilevers for Scanning Probe Arrays: Theory and Experiment," Transducers '97 The Ninth International Conference on Solid State Sensors and Actuators, Chicago, IL, 16-19 June 1997. [3] W.C. Tang, M.G. Lim, and R.T. Howe, J. of Microelectromechanical Systems, 1 170-178 (1992). [4] W.J. Cunningham, Introduction to Nonlinear Analysis, McGraw Hill Book Company, New York, 1958.
11:00 AM	Invited	MM+VT-WeM-9 Thermal Property Measurements for Novel ICs and MEMS K.E. Goodson, M. Asheghi, K. Kurabayashi, M.N. Touzelbaev, Y.S. Ju, B.W. Chui, T.W. Kenny, D.A. Fletcher (Stanford University) The performance of integrated circuits (IC) and MicroElectroMechanical Systems (MEMS) is being improved through the use of novel materials, including organic and/or porous passivation for fast logic and analog circuits, CVD diamond for smart power circuits, and crystalline silicon for cantilevers in MEMS. The thermal conductivities of these novel materials can strongly influence performance and reliability figures of merit, including the time constant and median time to failure. This paper describes thermal-conductivity measurement techniques for novel materials and provides an overview of the resulting data and associated modeling. The experimental techniques are in general novel and make use of unique near- and far-field scanning optical thermometry techniques developed in our laboratory. This paper also describes fabrication challenges associated with realizing the experimental structures, which include doped silicon cantilevers, diamond-coated metallization, and suspended polymer bridges.
11:40 AM		MM+VT-WeM-11 A SiC/Release Layer/SiC/Si Platform for the Formation of Thermally Isolated MEMS Devices H. Busta, R. Amantea, F. Pantuso, L. Goodman, L. White, D. Furst, M. Della Selva, V. Patel, G. Looney (Sarnoff Corporation); I. Jafri, R. Farmer (GT Equipment Technologies, Inc.) Extensive work has been performed in the formation of polysilicon/release layer/silicon MEMS devices for electrostatically driven actuators, micromotors, pressure sensors and others. For thermally isolated devices such as uncooled IR detector arrays, polycrystalline silicon is not suitable. Hydrogenated amorphous SiC, with its low thermal conductivity, offers an alternative materials approach. Anchored SiC cantilever structures of 0.5um thickness and widths ranging from 0.5 to 4um have been processed on silicon dioxide which is deposited on SiC/Si. In some patterns, the cantilevers extend into 50um long SiC plates. The cantilevers are typically 50um long and the structures, when released, reside only 0.5um from the underlying SiC layer. Release techniques, similar to the ones developed for the polysilicon/silicon dioxide/Si MEMS platform have been investigated. These include liquid-to-vapor releases using IPA and methanol at constant pressure and the methanol-to-CO₂ supercritical method. Complete releases, so far, have been obtained with the CO₂ method.