ELSEVIER
`
`Senwrs
`
`and Actuators
`
`A 66 ( 1998) 231-236
`
`SE@RS
`ACTIJA~ORS
`A
`
`PHYSICAL
`
`Commercial vision of silicon-based inertial sensors
`
`Cimoo Song *:, Meenam Shinn
`
`Abstract
`
`This paper reviews current technology and market trends in silicon inertial sensors using micromachining technologies. The requirements
`for successful commercialization of the research results will be discussed. Commercial implantation requires involvement at the design,
`process, and manufacturing stages, so that low cost, reliability, and better performance can be achieved. It is also necessary to have a clear
`understanding of the market and IC mentality. The paper also forecasts the potential applications and future market values of silicon-based
`inertial sensors.
`0 1998 Elsevier Science S.A. All rights reserved.
`
`Kry~ords:
`
`Inertial
`
`sensors; Commercialization;
`
`Silicon micromachining;
`
`Accelerometers;
`
`Gyroscopes
`
`1. Introduction
`
`2. Commercial
`
`status
`
`is one of the most important emerging
`Micromachining
`technologies
`for inertial sensors because of the advantages
`that it offers: small size and low cost. Although a variety of
`different materials can be applied for the micromachining
`technology, silicon has been the material of choice in micro-
`machining,
`including
`inertial sensors, due to its many desir-
`able mechanical and electrical properties and its compatibility
`with
`IC fabrication
`technologies. The main benefits
`that
`silicon-based microelectromechanical
`systems
`(MEMS)
`derive from their common base with
`the IC industry are low
`cost, mass-producibility,
`and monolithic
`integration. While
`the fabrication sequences of silicon sensors are similar
`to
`those used in IC processing,
`the mechanical nature of MEMS
`gives rise to additional requirements
`for commercial
`implan-
`tation that do not exist in IC processes, such as the releasing
`of the microstructures,
`handling of these released wafers or
`die, the packaging, and the testing of the devices.
`The R&D on silicon-based
`inertial sensors has advanced
`quite far during the last decade. For transferring
`the research
`results
`to industry, however, a clear understanding of com-
`mercial applications and market trends as well as of the cur-
`rent technology
`is necessary. This paper will
`first briefly
`review
`the present R&D and commercial status so far and
`discuss
`the technological
`requiretnents
`to satisfy
`further
`approach
`to the market.
`It will
`then forecast
`the feasible
`application area and the future market.
`
`l-280-9375;
`+ 82-33
`Tel.:
`author.
`r Corresponding
`6955: E-mail: cmsongesaitgw.sait.samsung.co.kr
`
`Fax:
`
`+ 82-33
`
`l-280-
`
`09241247/98/$19.00
`PllSO924-~2~7(98)OOO~S-X
`
`0
`
`1998 Elsevies Science S.A. All rights
`
`reserved.
`
`Silicon inertial sensors have been developed rapidly during
`the last decade and are considered as the next mass-produced
`mechanical sensors after silicon pressure sensors. Forsensing
`physical quantities such as acceleration or angular rate, a
`mechanical movable component converts the unknown quan-
`tity into a displacement
`that is then detected and converted
`to an electrical signal. The detailed operational principles of
`those inertial sensors will not be covered here and the reader
`is referred
`to the many reference materials available [ 1,2].
`This paper will
`focus on the recent advances there have been
`in R&D of silicon-based
`inertial sensors, especially as it
`relates to commercialization.
`in the
`One of the most successful microaccelerometers
`current market
`is the fully integrated monolithic ADXLSO,
`which was released in 1993 by Analog Devices
`Inc. (ADI)
`[3]. This force-balanced accelerometer consists of a beam-
`mass mechanical structure using surface micromachining, a
`capacitive sensor for the position detection of the mass. the
`signal-processing
`electronics
`for the sensor by means of
`BiCMOS
`IC technology, and an electrostatic actuator to apply
`a feedback
`force to the seismic mass. The device achieves a
`noise floor around 10 mg Hz-
`“a over an input range of k 50g
`or more with a single 5 V power
`supply and with shock
`survival
`in excess of 2000g. Although
`these specifications
`are aimed at applications
`for airbag release control, there are
`potential mass applications
`in a variety of motion-control
`systems, spurring significant
`interest and investment
`from
`industry. They can be marketed at a cost under US$lO apiece.
`
`APPLE 1026
`
`1
`
`

`

`in the F 5Og range are currently being
`Silicon accelerometers
`supplied to vehicles in the USA by ADI, Motorola, and I.C.
`Sensors. In Europe. Bosch is the major player; in Japan, the
`supplier is Nippodenso. With the exception of ADI, all man-
`ufacturers are using multiple-chip configurations.
`Micromachined
`gyroscopes have also received vigorous
`development efforts
`in recent years [4-61 due to their wide
`applications in motion control, including the automotive area.
`Most micromachined gyroscopes are exclusively of vibrating
`type consisting of two resonant modes: lateral and perpen-
`dicular. The main design issue for a vibratory gyroscope
`is
`how to detect the sensing amplitude, which
`is much smaller
`than the vibrating amplitude. For example, since the sensing
`amplitude is roughly at sub-atomic scale for a 10 pm driving
`vibration at lo s - ’ input, output signals are very sensitive to
`various error sources in the system.
`In order to improve
`the
`sensitivity of the devices it is necessary
`to match the reso-
`nance frequencies of lateral (driving) and vertical (sensing)
`modes. to minimize stray capacitance. and to reduce envi-
`ronmental
`vibration effects by keeping high resonance
`frequencies.
`The noise equivalent rate of present gyroscopes announced
`by manufacturers
`including ADI, Motorola, GM, Benz,
`Bosch, Nippondenso. and Sumitomo
`is about 1” s- ’ Hz-
`“’
`with
`the exception of 1” h- ’ Hz-
`“’
`for military application
`by Charles Stark Draper Laboratory
`(CSDL). Out of the
`companies mentioned above, AD1 adopted surface-micro-
`machining
`technology
`for Fabricating devices. It is generally
`known
`that surface micromachining has advantages overbulk
`micromachining:
`the fabrication of free-form, complex, and
`multi-component
`integrated electromechanical structures.
`Recently Samsung has presented a surface-micromachined
`gyroscope
`representing prominent performance with a noise
`equivalent rate of 0. lo s- ’ Hz-
`“’
`[ 71. Fig. 1 shows an SEM
`view of the gyroscope
`fabricated by Samsung. The dimen-
`sions of this microgyroscope
`are 450 Frn wide. 1200 pm
`long, and 7.5 pm thick. The gap between
`the resonator and
`the polysilicon electrode on the substrate
`is designed to be
`
`1.5 pm. Samsung has improved the performance of the device
`in two ways. One is the effective
`reduction of noise level
`using differential driving voltage and electrical
`tuning by
`applying an inter-plate d.c. bias after fabrication. The other
`is to adopt a high-aspect-ratio polysilicon structure with 7.5
`p,rn thickness as shown
`in Fig. 2.
`Fig. 3 demonstrates another force-balancedsurface-micro-
`machined gyroscope presented by Samsung with high angu-
`lar inertia momentum and compact size: 900 p,rn wide, 900
`km long and 7.5 f*m thick [ 81. To increase the linearity and
`dynamic
`range, a force-balancing method was used. This
`device features low fabrication cost since this structure has a
`new force-balancing
`torsional torque mechanism which does
`not require another top electrode layer to reduce the intrinsic
`nonlinearity
`in capacitive-type
`sensors.
`In order to remove
`cross-axis
`sensitivity
`from mechanical coupling, differential
`d.c. bises are applied to the torsional sensing mode and driv-
`ing mode. This enables a resolution of 0.1” s- ’ to be achieved
`at a maximum measurement
`range of’ i 100” s- ’ with 1%
`(FS) nonlinearity.
`
`Fig. 3-. SEM
`pm height.
`
`view of a high-aspect-ratio
`
`polysilicon microstructure
`
`with 7.5
`
`1. SEM
`Fig.
`micromachin&.
`
`view
`
`of a
`
`tunable
`
`vibratory
`
`gyroscope
`
`using
`
`surface
`
`Fig. 3. SEIM view of a force-balanced
`Face micromachining.
`
`dual-axis microgyroscope
`
`using sur-
`
`2
`
`

`

`3.1. Wcfer-level
`
`testing
`
`While functional testing of ICs is generally performed at
`the wafer level, that of MEMS is carried out currently at the
`die level and this is a source of cost increase. For lowering
`testing cost? the performance measurements such as resolu-
`tion, sensitivity, linearity, and drift rate should be carried out
`at wafer level. Then, reliability testing should be done at
`packaged die level. Most difficulties in wafer-level testing of
`micromachined gyroscopes come from the tuning processes
`of two different resonance frequencies resulting from fabri-
`cation errors. The rest of the difficulty is that the device should
`be maintained in a vacuum environment, especially for gyro-
`scopes. Therefore, the testing of inertial sensors requires the
`development of unique test facilities.
`
`Fig. 1. SEIM view of a laterally
`supported
`by LT-bhaped
`spring.
`
`oscillated
`
`and sensed vibratory
`
`gyroscope
`
`3.2. Wafer-level
`
`packaging
`
`vibratory gyroscope
`Fig. 4 also shows a force-balanced
`designed by a new concept and fabricated by surface micro-
`machining
`[9,10]. This gyroscope has a resolution of 0.1”
`s-’ at 2 Hz, a bandwidth of 100 Hz, and a dynamic range of
`90” s-l. Since the direction of the Coriolis motion occurs in
`the wafer direction,
`this device measures the angular rate of
`the vertical direction. For maximizing
`the ratio between
`the
`sensing and mass areas, LT-shaped electrodes as shown
`in
`Fig. 4 are introduced. The specially designed springs serve
`to match the !irst and second modes with
`the mass driving
`and position-sensing modes,
`respectively. The prominent
`shape of the comb-drive helps to improve
`the resolution by
`increasing oscillating displacement. A feature is that the res-
`onance frequencies associated with
`lateral vibration modes
`are not likely
`to be varied by a change in thickness of the
`polysilicon structure, which guarantees the uniform sensitiv-
`ity of products and requires less testing cost for tuning.
`Samsung plans to create a micromachined
`inertial mea-
`surement unit (IMU)
`which consists of three gyroscopes:
`two tunable gyroscopes as shown
`in Fig. 1 for measuring two
`lateral directions and one laterally sensed gyroscope
`as
`viewed
`in Fig. 3 for measuring
`the vertical direction. This
`system enables the three angular rates along three axes to be
`measured. Though the R&D of these micromachined gyro-
`scope has been moved up to the edge of the market, none of
`them has been commercialized
`yet due to the requirements
`that will be discussed
`in the following section.
`
`3. Requirements
`
`for commercialization
`
`of
`for commercialization
`requirement
`important
`An
`is that they should be low cost, offer better perform-
`MEMS
`ance, be reliable, and be easily integrated with electronics
`and provide sufficient
`flexibility
`so that different
`types of
`sensors/actuators
`can be systematized without
`significant
`redesign.
`
`Packaging is one of the most expensive steps in the devel-
`opment of inertial sensors. For example, it costs over 80% of
`the whole price in the case of vacuum packaging using an
`A1203 ceramic case as shown in Fig. 5 [ 71. A reasonable cost
`ratio between die and packaging for production is 1: 1. For a
`gyroscope, the package should provide a vacuum inside with
`no leakage and easy interconnection with electronics by feed-
`through. It should also be free of mechanical stress and should
`prevent external perturbations from reaching the device, pro-
`tect the device from the harsh environment, and be small
`enough. In order to overcome all these problems, the package
`should be designed and fabricated at the same time as device
`fabrication, especially at the wafer level.
`
`The vast majority of companies currently involved in
`MEMS fabrication are performing their proc~essing in facili-
`ties previously fitted-up for IC processing. The choice of bulk
`
`using Al?O,
`gyroscope
`of a vacuum-packaged
`Fig. 5. Photograph
`case. The dimensions
`of rhe package are 8 mmX
`12 mm.
`
`ceramic
`
`3
`
`

`

`231
`
`C. Song, M. Sizinn /Sensors
`
`trnd Amators A hh (1998)
`
`231-236
`
`or surface micromachining and integrated or hybrid devices
`depends on their accessible
`facilities and technical back-
`ground. For example, AD1 and Samsung can take advantage
`of integrated surface-micromachined
`sensors with reasonable
`support from existing
`IC infrastructures. However,
`the most
`important factor is how to minimize the process errors. Thus,
`better equipment such as deep-etching
`facilities and deposi-
`tion furnaces for higher-aspect-ratio
`structures are required
`and this increases production cost. It is necessary to trade off
`between the process errors and process cost. The other factor
`that should be considered in manufacturing
`is the number of
`masks, since fewer masking steps offer a simple and cost-
`effective process.
`
`3.4. Higher performnnce
`
`is an important fac-
`The miniaturized advantage of MEMS
`tor, especially in a battery-operated system which should have
`a power consumption as low as possible. However, any sig-
`nificant size reduction of inertial sensors is accompanied by
`a loss of their performance. The performance of a MEMS
`vibratory gyroscope
`is defined eventually by the minimum
`detectable angular rate. Therefore,
`this problem should be
`solved by optimal design: a selection of low-noise elements
`in the signal-transforming
`chain, providing
`favorable condi-
`tions for
`their maximal
`robustness
`to external noise, and
`increasing the output signal value at any given input angular
`rate. This will be possible by providing effective use of the
`favorable electrical and mechanical properties of silicon as
`design material, as well as providing near-resonant oscilla-
`tions of the proof mass at high values of the Q-factor of
`exciting and sensing oscillation contours. The microstructure
`can also be made as thick and narrow as possible, justifying
`the need for high-aspect-ratio microstructures.
`
`3.5. Reliability
`
`One of the most challenging aspects of commercialization
`is the fabrication of reliable and reproducible devices. For
`example, the drift rate for temperature variation
`in the range
`-40
`to -t 150°C has to be minimized and compensated
`for
`the inertial sensors
`in an automobile. The lifetime of the
`devices can be precisely predictable and the function should
`also be guaranteed for the whole
`lifetime.
`
`4. Application and market
`
`is about 6% per year.
`The overall market growth of MEMS
`The fastest-growing market segments include automotive/
`transportation
`sensors and smart sensors, with growth
`rates
`close to 20% per year [ I 11. Very rapid growth of inertial
`sensors is expected to occur primarily
`in the automotive and
`consumer market, as shown
`in Table 1, Silicon-based sensors
`will occupy more than 90% of all commercial
`inertial systems
`in 2000.
`The future market of inertial sensors will be basically
`spurred
`in ways
`to satisfy all the requirements described
`above. First of all, the automobile market tends to substitute
`the expensive conventional
`‘fully active’
`($5000)
`suspen-
`sion or navigation systems based on total closed-loop control
`with
`low-cost MEMS or ‘semi-active’
`systems
`[ 121. How-
`ever the ‘semi-active’
`systems are still expensive
`($ZClOOj,
`since the unit price of the existing angular rate gyros similar
`to Murata’s piezoelectric vibrating cylinder or Matsushita’s
`tuning fork is around $30-50. Thus, they will only be imple-
`mented in high-end models at this point. This price pressure
`pushes vehicle manufacturers to reduce system cost, making
`systems suitable for less expensive vehicles. The availability
`of a low-cost, MEMS-based angular rate sensor similar to
`those developed by ADI, GM, BEI Electronics, and Samsung
`
`TabIe 1
`Inertial
`sensors market
`
`projections
`
`in the year 2000. The expecred
`
`requirements
`
`and performances
`
`are also shown
`
`iSnmsung
`
`and MRI,
`
`June 1996)
`
`Market
`
`area (requirement)
`
`Applicatioiis
`
`Estimated
`(gyroiaccel.)
`
`units
`
`Gyro performance;
`(OS-‘)
`
`range/res.
`
`Automobile
`environment,
`
`i low-cost,
`lifetime)
`
`reliable, harsh
`
`Consumer/medical
`power.
`small size,
`
`cost. low
`(low
`lifetime)
`
`(small
`IndusWy
`environment)
`
`size, reliable,
`
`harsh
`
`Military
`
`(reliable.
`
`higher
`
`resolution)
`
`Total
`
`’ Varies depending
`
`on situation
`
`airbag
`advanced
`active suspension
`navigation
`ABS/anti-skid
`camcorder
`30 mouse
`VR game/toys
`sports equipment
`roborics
`machine monitoring
`attitude
`controi
`new weapon
`systems
`IMU
`
`316
`214
`l/O
`313
`2/o
`2/o
`212
`212
`616
`d
`
`313
`
`I
`
`200/10
`500.
`100/o.
`5010.5
`5O/O.S
`10012
`10010.1
`50/o. 1
`1010.1
`IO/O.1
`2010.2
`
`Sales
`(SBil.)
`
`0.8
`0.1
`0.02
`0.6
`0.3
`0.3
`0.6
`0.1
`0.3
`0.3
`0.2
`
`3.92
`
`4
`
`

`

`C. Song,
`
`hf. Shim
`
`/Sensors
`
`and Acrmroi-s
`
`A 66
`
`(1998)
`
`231-236
`
`235
`
`in the near future, probably
`is expected to be commercialized
`in a year or two years, and will propel the adoption of these
`enhanced systems
`into less expensive vehicles over a period
`of time.
`that can
`Secondly, only the reliable and robust devices
`survive
`in harsh environments will be adopted to the market.
`Moreover,
`the market also requires an order of magnitude
`better resolution at lower power consumption
`for a very large
`range of applications. Future accelerometers will have a noise
`level as low as 1 ~J-S Hz-“’
`at an input range of 2~. Though
`the necessary accuracy of a gyroscope
`for the recent auto-
`mobile GPS/INS
`is 50-300” h-‘,
`the next-generation de-
`vices for short-term military navigation systems are expected
`to have random drift of about 10” h- ’ or better and the
`acceleration scale factor stability should be in the order of
`0.05%. Several companies including CSDL, ADI,
`I.C. Sen-
`sors, Motorola, and Samsung will be likely
`to compete for
`these opportunities.
`is
`Finally,
`the capability of integration with electronics
`getting more attention in the market since a one-chip system
`is of major
`interest
`for commercial devices, especially
`in
`automobile applications. The role of on-chip electronics
`becomes more important for intelligence and multiplicity:
`the
`combination of a large number of sensors, actuators, and
`electronics
`in one single unit. In addition,
`the demand for
`wireless
`telemetry
`technology
`for communicating with
`the
`outside
`is growing
`fast. Therefore,
`the compatibility of a
`single unit sensor with one-chip systems has to be considered
`from the beginning of the design and fabrication.
`
`5. Conclusions
`
`inertial sensors is increasing,
`The demand for silicon-based
`which promises a new wide market for many areas. In order
`to introduce inertial technology
`into these markets,
`it is nec-
`essary to meet a number of principal requirements: significant
`cost reduction by wafer-level
`testing and packaging, decrease
`of power consumption by miniaturization, enhancement of
`performance and reliability, and on-chip integration for mul-
`tiplicity. Notwithstanding
`the many challenges discussed
`above, a great effort at developing inertial sensors will enable
`these devices to be commercialized, perhaps in as little as two
`years.
`
`Acknowledgements
`
`like to thank all members of the Micro-
`The authors would
`systems Laboratory of Samsung Advanced Institute of Tech-
`nology for their great efforts
`in developing gyroscopes. The
`authors also appreciate Professor Y. Cho of Korea Advanced
`
`Institute of Science and Technology and Professor J. Lee of
`Seoul National University
`for their collaboration.
`
`References
`
`accelerometers,
`
`IEEE
`
`accelerometer
`Way, Northwood,
`
`with single con-
`MA
`02062,
`
`[ 11 B.E. Boser, R.T. Howe, Surface micromachined
`J. Solid-State
`Circuits
`31 (3)
`(1996)
`366-375.
`:-axis
`[2] W.A. Clark, R.T. Howe, R. Horowitz,
`Surface micromachined
`vibratory
`rate gyroscope,
`Tech. Digest, Solid-State
`Sensor and Actu-
`ator Workshop,
`Hiiton Head
`Island
`, SC, USA, 2-6 June, 1996. pp.
`283-289.
`[3] Analog Devices. ADXLSO-monolithic
`ditioning,
`Datahheet,
`I Technology
`USA,
`1993.
`silicon microgyro-
`K. Tanaka, M. Sugimo. Vibrating
`[4] Y. Mochida,
`IEEE Jpn. 116-E
`(7)
`( 1996) 283-288.
`scope, Trans.
`ring gyroscope,
`[ 51 M.W. Putty, K. Najafi, A micromachined
`vibrating
`Tech. Digest, Solid-State
`Sensor and Actuator Workshop,
`Hilton Head
`Island, SC, USA.
`1994, pp. 2 13-220.
`Silicon bulk micromachined
`T.K. Tang, R.C. Gutierrez,
`J.Z. Wilcox,
`vibratory
`gyroscope,
`Tech. Digest, Solid-State
`Sensor and Actuator
`Workshop,
`Hilton Head
`Island, SC, USA, 2-6 June, 1996, pp. 28%
`293.
`[7] Y. Oh, B. Lee, S. Baek, H. Kim, J. Kim, S. Kang, C. Song, A surface-
`micromachined
`tunable
`vibratory
`gyroscope,
`IEEE Proc. MEMS
`‘97,
`Nagoya,
`Japan, 1997, pp. 272-277.
`[8] S. An, Y. Oh, B. Lee, S. Choi, S. Kang, Y. Ko, J. Kim, C. Song, S.
`Lee, Force-balanced
`dual-axis microgyroscope,
`paper presented
`at
`Smart Materials
`Structures
`and ME-MS.
`SPIE, Adelaide,
`Australia,
`lo-13 Dec., 1997.
`and
`oscillated
`[9] K.Y. Park, C.W. Lee, Y.S. Oh, Y.H. Cho, Laterally
`by fish hook
`force-balanced
`micro vibratory
`rate gyroscope
`supporting
`shape springs,
`IEEE Proc. MEMS
`‘97, Nagoya,
`Japan, 1997, pp. 494-
`399.
`oscillatedandforce-balanced
`[ lo] K. Park, C. Lee, Y. Oh, Y. Cho, Laterally
`by fish hook shape springs,
`micro vibratory
`rate gyroscope
`supported
`paper
`presented
`at Smart Materials,
`Structures
`and MEMS,
`SPIE,
`Adelaide.
`Australia,
`lo-13 Dec., 1997.
`[ 111 J. Bryzek,
`Impact of MEMS
`technology
`ators A 56 ( 1996)
`1-9.
`[ 121 R.H. Grace, Automotive
`tems
`(MEMS),
`Conf.
`USA, 7 Oct., 1996.
`
`[6]
`
`on society, Sensors and Actu-
`
`sys-
`of microelectromechanical
`applications
`for Commercialization
`of MEMS,
`Kona, HI,
`
`Biographies
`
`Cirnoo Sorzg received the B.S. and M.S. degrees in mechan-
`ical engineering from Yonsei University, Korea, in I977 and
`1982, respectively, and the Dr.-Ing. degree from the Tech-
`nical University of Berlin, Germany,
`in 1987. From 1977 to
`1982, he specialized in robotics at Korea Institute of Science
`and Technology, Korea. In October 1987, he joined Samsung
`Advanced
`Institute of Technology
`in Korea. Since then he
`has been engaged in the research and development of micro-
`mechatronics
`system
`technologies
`including microsensors
`and microactuators as a program manager.
`
`received the B.S. and MS. degrees in mate-
`Meenanz Shim
`rials science and engineering from Hanyang University,
`
`5
`
`

`

`236
`
`C. Song, hf. Shim
`
`/Srmors
`
`und ActuntorsA
`
`66 (I9981
`
`231-236
`
`Korea. in 1985 and 1988, respectively, and the Ph.D from
`Northwestern
`University
`in 1993. From 1993 to 1994, she
`worked as a post-doctoral
`fellow at Northwestern University
`in the field of thin films. In November 1994, she joined Sam-
`
`in Korea. Since then
`Institute of Technology
`sung Advanced
`she has been engaged in the research and development of
`microelectromechanical
`systems including integratedinertial
`sensors as a senior researcher.
`
`6
`
`