544 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 9, NO.

4, DECEMBER 2000

An All-Silicon Single-Wafer Micro-g Accelerometer

with a Combined Surface and Bulk Micromachining
Process
Navid Yazdi and Khalil Najafi, Fellow, IEEE

Abstract—This paper reports an all-silicon fully symmetrical The large proof mass of the reported micro-g accelerometers
-axis micro-g accelerometer that is fabricated on a single-silicon is typically formed by wafer bonding and bulk micromachining
wafer using a combined surface and bulk fabrication process. The [3]–[8] or, in some cases, by thick electroplated structures and
microaccelerometer has high device sensitivity, low noise, and
low/controllable damping that are the key factors for attaining g the LIGA process [9]. Multiple wafer bonding is a viable three-
and sub- g resolution in capacitive accelerometers. The microfab- dimensional (3-D) fabrication technology for microsensors and
rication process produces a large proof mass by using the whole microactuators [10]. However, silicon-silicon wafer bonded ac-
wafer thicknes and a large sense capacitance by utilizing a thin celerometers require high temperature processing and could po-
sacrificial layer. The sense/feedback electrodes are formed by a tentially drift. Post-bonding high-temperature treatment and an-
deposited 2–3 m polysilicon film with embedded 25–35 m-thick
vertical stiffeners. These electrodes, while thin, are made very neal has been suggested to relieve the stresses and minimize
stiff by the thick embedded stiffeners so that force rebalancing drift in silicon-silicon bonded structures [11]. Glass and silicon
of the proof mass becomes possible. The polysilicon electrodes wafers can be anodically bonded at lower temperature with a
are patterned to create damping holes. The microaccelerometers much simpler process. However, the difference between tem-
are batch-fabricated, packaged, and tested successfully. A device perature coefficient of expansion of glass and silicon causes
with a 2–mm 1–mm proof mass and a full bridge support has
a measured sensitivity of 2 pF/g. The measured sensitivity of a higher temperature sensitivity in the fabricated structures. In
4-mm 1-mm accelerometer with a cantilever support is 19.4 addition, the wafer bonding process results in thick structural
pF/g. The calculated noise floor of these devices at atmosphere are layers that are desirable for forming the proof mass; neverthe-
0.23 g/ Hz and 0.16 g/ Hz, respectively. [567] less, reducing damping by incorporating damping holes in the
thick bonded structural layers is difficult. LIGA devices are not
I. INTRODUCTION typically inertial grade due to their relatively large temperature
sensitivities and potentially long-term drift caused by creep and
H IGH-precision microaccelerometers are increasingly
needed in numerous applications including self-con-
tained or GPS-augmented navigation and guidance systems,
fatigue in metallic structures. Finally, surface micromachined
accelerometers [12] have the main advantages of being mono-
seismometry for oil-exploration and earthquake prediction, lithically fabricated with the interface circuit and increasing
by using damping holes. However, the proof mass in these de-
microgravity measurements and platform stabilization in space,
vices is small, which results in lower sensitivity and typically
and underwater acoustic measurements. While recent advances
above micro-g mechanical noise floor unless the accelerometer
in fabrication technologies for microaccelerometers have made
is vacuum packaged.
the mass production of these devices with low/medium sen-
This paper presents an all-silicon fully symmetrical -axis
sitivities possible [1], [2], to date, only a few micro-g silicon
micro-g accelerometer that is fabricated on one silicon wafer
accelerometers have been reported [3]–[8]. These devices
using a combined surface and bulk fabrication process [13].
utilize a large proof mass in conjunction with capacitive [3]–[6],
The microaccelerometer has high device sensitivity, low noise,
resonant [7], or tunneling current [8] sensing schemes to achieve
and low/controllable damping, which are the key factors for at-
high sensitivity. Among all these, capacitive accelerometers
taining g and sub- g resolution in capacitive accelerometers.
have become more attractive and promising for high precision
The high device performance is obtained by the novel device
micro-g accelerometers due to their high sensitivity, good dc
structure and fabrication process that enable formation of a large
response and noise performance, low drift, low temperature
proof mass, controllable/small damping, and a narrow uniform
sensitivity, low-power dissipation, large readout bandwidth, and
air gap over a large area. In the following sections, the issues
simple structure.
involved in the development of capacitive micro-g accelerom-
eters are first reviewed. Then, the device structure, design, and
fabrication process are discussed. Finally, fabrication and mea-
Manuscript received April 18, 2000. This work was supported by the
Defense Advanced Research Projects Agency under Contracts JFBI 92-149, surement results are presented.
DABT63-95- C-011, and F30602-98-2-0231. Subject Editor, N. deRooij.
N. Yazdi is with the Department of Electrical Engineering, Arizona State Uni-
versity, Tempe, AZ 85287-5706 USA (e-mail: [email protected]). II. MICRO-G ACCELEROMETER ISSUES
K. Najafi is with the Center for Integrated Microsystems University of
Michigan, Ann Arbor, MI 48109-2122 USA. A precision microaccelerometer is typically operated closed-
Publisher Item Identifier S 1057-7157(00)10871-6. loop to obtain higher bandwidth, full-scale range, and linearity.
1057–7157/00$10.00 © 2000 IEEE
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.
 YAZDI AND NAJAFI: ALL-SILICON SINGLE-WAFER MICRO-g ACCELEROMETER 545

Therefore, the main design consideration for these devices is

meeting the resolution specification. Resolution of a microac-
celerometer is determined by its mechanical noise and readout
electronics noise. The electrical signal-to-noise ratio (SNR) can
be improved by enhancing the sensor and the readout circuit sen-
sitivity and reducing the circuit electrical noise. The mechanical
noise is caused by the Brownian motion of the gas molecules
surrounding the proof mass and the Brownian motion of the
proof mass suspension or anchors. The total noise equivalent
acceleration (TNEA m/s Hz ) is shown to be [14]

TNEA (1)

where Fig. 1. Total noise equivalent acceleration (TNEA) due to Brownian motion at
Boltzmann constant; 2
atmospheric pressure for an accelerometer with 1 mm 2 mm mass/electrode
area and 1.5 m air gap with (a) 4-m damping holes with 9-m pitch, (b) 5-m
temperature in Kelvin; damping holes with 15-m pitch, and (c) no damping holes.
damping factor;
proof mass;
mechanical self-resonance frequency; As can be observed, a sub-micro-g noise floor is achievable at
quality factor. atmosphere by using both thick proof mass and damping holes.
As is evident, mechanical noise can be reduced by increasing Note that by forming the damping holes in the electrode rather
the proof mass size and reducing damping. Device damping is than the proof mass, there is no loss in the mass size. Further-
caused by both structural and viscous damping. However, the more, by using small damping holes, the fringing capacitance
structural component is orders of magnitude smaller than the becomes comparable with the lost capacitance due to reduction
gas viscous damping even at low pressures (a few mTorr) and in surface area caused by the formation of damping holes, and
can be neglected [15]. Accelerometer damping can be decreased thus, there is almost no loss of sense capacitance.
by using damping holes or vacuum packaging. Vacuum pack- The overall sensor SNR can be also improved by increasing
aging of the accelerometer increases the sensor cost and manu- the open-loop sensitivity and enhancing the sense signal at the
facturing complexity. In addition, it may result in long-term drift input of the readout circuit. The static open-loop sensitivity of
due to package leakage. Using damping holes has none of these the accelerometer with capacitive area of and air gap of can
issues. Furthermore, it allows control of the damping factor by be expressed by
changing the hole density and geometry, which effectively sim-
plifies the design of control circuitry for stable closed-loop op- (4)
eration. The damping factor of a perforated plate with equally
spaced damping holes can be expressed by (2) [16] Thus, open-loop sensitivity improves linearly with increasing
proof mass thickness and electrode area. It is also inversely
proportional to the air gap squared and the suspension spring
(2)
constant. In summary, the challenge in developing a capacitive
micro-g accelerometer is to attain thick large mass, soft suspen-
(3)
sion, narrow air gap over a large area, and high-quality factor
simultaneously.
where
number of the damping holes;
viscosity of the surrounding gas; III. DEVICE STRUCTURE
air gap; The microaccelerometer utilizes a novel device structure and
effective plate length and is related to the damping hole fabrication process to address all the micro-g accelerometer de-
size ( ) and pitch ( ) by (3). sign issues. The central idea in this structure is combining sur-
In order to demonstrate the effectiveness of damping holes face and bulk micromachining to fabricate devices with high
in reducing the damping factor and lowering the mechanical sensitivity, low noise floor, and controllable damping on a single
noise floor, an accelerometer TNEA versus the proof mass silicon wafer. Fig. 2 shows the accelerometer structure and its
thickness at atmospheric pressure for three different cases—a) top and cross-sectional views. The main features of the device
4- m damping holes with 9- m pitch; b) 5- m damping holes structure and its fabrication process are 1) use of the entire wafer
with 15- m pitch; c) without any damping holes—is plotted thickness to attain a large proof mass, 2) use of a sacrificial thin
in Fig. 1. The device has a mass/electrode area of 2 mm 1 film to form a uniform and conformal gap over a large area, and
mm, is at atmospheric pressure, and has a 1.5 m capacitive air 3) use of deposited polysilicon and deep trench etching to create
gap between the electrode and the proof mass. Squeeze film mechanically stiff electrodes on both sides of the proof mass cir-
damping and mechanical noise are obtained using (1) and (2). cumventing the need for multiple wafer bonding.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.
 546 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 9, NO. 4, DECEMBER 2000

B. Suspension Beams
The accelerometer structure is entirely symmetrical around
its proof mass center point, and therefore, a low cross-axis sen-
sitivity can be achieved for a cantilever support as well. The
design uses a full-bridge support configuration with eight sup-
port 700 m 40 m 3 m heavily boron-doped Si beams
that yield a spring constant of 65 N/m with an intrinsic stress
of 40 MPa. The beams are slightly misaligned relative to the
wafer flat to allow their undercutting. A more compliant can-
tilever support can be obtained by cutting all the beams on one
side of the proof mass.

C. Air Gap
The accelerometer sensitivity inversely increases with the air
gap squared. The lower limit of the air gap is determined by the
Fig. 2. Micro-g accelerometer structure. (a) Device structure. (b) A-A
cross-sectional view. (c) Top view. device release time and stiction. The air gap is defined by a sac-
rificial silicon oxide (SiO ) layer and is wet etched in the final
The electrodes, while thin, are made stiff in the sense direc- stage of device fabrication. This sacrificial layer is to be etched
tion by embedding thick vertical stiffeners in them. Thus these at the surface and deep in the proof mass trenches to separate
electrodes can be used to electrostatically force-rebalance the the proof mass from the polysilicon electrodes and their vertical
proof mass and provide closed-loop operation. The thick ver- stiffeners. The etch front proceeds in the narrow long channels
tical stiffeners are formed by refilling high aspect-ratio trenches created by the sacrificial layer etch in the trenches and is a non-
with thin polysilicon film deposition. Any damping hole config- linear function of the channel length and height [19]. With a
uration and geometry can be easily formed on the thin polysil- 35- m-deep trench, a conservative air gap size of 1.4 m pro-
icon electrodes to optimize damping factor and capacitance, vides a sufficiently short release time so that the metal and di-
while there is no concern with large hole density and its effect electric silicon nitride layers at the silicon frame are not signif-
on the electrode softening as the plate stiffness is mainly pro- icantly etched. In addition, a gap size of 1.4 m is adequately
vided by the embedded stiffeners. large to avoid severe yield loss due to stiction.
The proof mass and its supporting rim have the whole wafer
thickness and are formed by anisotropic etching of the silicon D. Damping and Noise
wafer (bulk micromachining). When the proof mass and the The damping factor is to be reduced sufficiently to lower the
electrodes touch, electrical isolation is provided by dielectric mechanical noise floor. In order to achieve a few micro–g reso-
layer stand-offs placed on the proof mass. There are eight sus- lution in a 100-Hz bandwidth and satisfy the requirements of the
pension beams, which are symmetric with respect to the proof majority of the applications, a mechanical noise of a few tenth’s
mass centerline and result in low cross-axis sensitivity. These of g/ Hz is needed. The noise and the damping factor are con-
suspension beams are formed using heavily boron-doped strips trolled and reduced by forming damping holes in the electrodes
of the silicon wafer or deposited polysilicon layer. The sense and [16]. It can be shown that for a square hole smaller than 18 m
feedback electrodes are polysilicon plates created on both sides and an air gap of 1.4 m, the fringing capacitance is, in fact, larger
of the proof mass and anchored on an isolation dielectric at the than the lost area capacitance. In addition, in order to avoid plate
frame. The device has low temperature sensitivity as the polysil- softening of the electrodes between the stiffeners, a low hole den-
icon electrodes and silicon frame thermal expansion coefficients sity of 15%-area is selected in the design. Considering all these
match each other closely. Furthermore, the device has a good design constraints, 5- m damping holes with 15- m pitch are in-
long-term stability as it is all-silicon and no wafer bonding is corporated in the electrode plates. The damping factor of the ac-
used in its fabrication process. celerometer with 2 mm 1 mm proof mass is calculated to be
7 e-4 N.m/s, which results in an equivalent noise floor of 0.23
IV. DEVICE DESIGN g/ Hz at atmosphere for the double-side accelerometer. The
A. Proof Mass fringing filed capacitance for each hole is 0.43 fF, whereas the
lost area capacitance is 0.16 fF.
In order to increase the sensitivity and reduce the noise floor,
the proof mass is desired to be as large as possible. Two different
E. Sense and Feedback Electrodes
designs with proof mass of 4 mm 1 mm and 2 mm 1 mm
with a thickness of m for a 4-in Si wafer were selected. The top and bottom electrodes are divided into 5 and 12 elec-
Hence, the proof mass sizes are 2.1 mg and 4.2 mg, respec- trically isolated electrodes for 2 mm 1 mm and 4 mm 1
tively. The proof mass has a rectangular geometry to both keep mm devices, respectively. These electrodes each can be sepa-
the electrode lengths limited and obtain a large mass. The proof rately used for sense, feedback, self-test, or calibration of the
mass is formed by wet anisotropic silicon etching of (100) sil- accelerometer. However, the maximum sensitivity and dynamic
icon wafer, and corner compensation patterns are used to avoid range are obtained when all the electrodes are used for sense and
mask undercutting at the corners [17], [18]. feedback.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.
 YAZDI AND NAJAFI: ALL-SILICON SINGLE-WAFER MICRO-g ACCELEROMETER 547

rounded by wet etching. The dimple height limits the accelerom-

eter open-loop dynamic range, and a high density of dimples
increase the surface adhesion forces when the electrode and the
proof mass touch. The design uses 0.4- m dimples separated 63
m from each other.

V. FABRICATION PROCESS
The fabrication process requires eight masks and is per-
formed symmetrically on both sides of the wafer, as shown in
Fig. 4. The process starts with a shallow (3 m) p boron
diffusion on double-polished p-type Si wafer using
thermal oxide as a mask. Both sides of the wafer are patterned,
and the patterns are aligned to each other. The shallow boron
diffusion is performed at 1150 C for 30 min and defines the
beams, the proof mass, and the supporting rim. Then, 2200
Å of LPCVD nitride is deposited and patterned to form the
polysilicon electrode anchors and isolation dielectric under
the polysilicon electrode dimples. The next masking step is
etching 30- m-deep, 6- m-wide trenches to define the vertical
electrode stiffeners.
The trenches are then refilled completely using 1.4- m
sacrificial LPCVD oxide and 1.8- m LPCVD polysilicon.
The polysilicon is etched back using a blanket RIE etch,
exposing the sacrificial oxide and leaving polysilicon plugs in
the trenches. In this manner, the step height due to the trench
etch is reduced, and sacrificial oxide at the bottom of the
trenches is protected. Next, two patterning steps are performed
Fig. 3. Summary of comparative FEM simulations of electrodes with various
2 2
stiffener geometeries. The electrode is 2000 m 200 m 2 m, the stiffener on the sacrificial oxide. First, it is patterned and partially wet
height is 15 m, the stiffener thickness is 2 m, and the pressure load is 20 Pa. etched to form dimples inside the polysilicon electrodes. These
dimples reduce the contact area and help to reduce stiction.
Second, narrow strips of sacrificial oxide are patterned and
The sidewall capacitance of the stiffeners and proof mass con-
etched away at the anchors. Sacrificial oxide is left in most
tributes to the electrode parasitic capacitance. Thus, the stiff-
areas of the anchors to reduce parasitics and is sealed between
ener design objective is to achieve sufficient rigidity with the underneath nitride and top polysilicon. A 2–3- m LPCVD
minimum sidewall capacitance. In order to determine the most polysilicon is then deposited to connect to the plugs in the
suitable stiffener geometry, stiffness of the electrode in the di- trenches and form the electrodes. The polysilicon electrode is
rection with a uniformly distributed pressure load was simulated doped with phosphorus or boron halfway through deposition.
for a number of different stiffener shapes. The comparative re- The electrodes are sealed with a 4000-Å LPCVD oxide in the
sults of these FEM simulations are shown in Fig. 3. As can be next step. This oxide is patterned to form metal contacts and
observed, the stiffness of the plate with straight stiffener is the openings to bulk Si for the subsequent proof mass release. The
highest, and its sidewall area is the lowest. Hence, it is the most wafer is then patterned for Cr/Au (400-Å/5000-Å) sputtering
suitable choice. In the actual design, each electrode contains two and liftoff. The proof mass is released by an EDP etch at 110 C
vertical stiffeners 3.5 m wide and 30 m high separated from and following sacrificial oxide etch in 1 : 1 HF : DI-water. Then,
each other by 210 m, and three perpendicular stiffeners are the released structures are dried in an oven after a methanol
added to connect these main stiffeners together. The calculated soak or coated with self-assembled monolayer (SAM). SAM
stiffness of each electrode in this case assuming a negligible in- coating makes the polysilicon and silicon surfaces hydrophobic
trinsic stress for the 3- m polysilicon electrode is 510 N/m. and helps substantially with preventing stiction [21].
This stiffness yields a maximum deflection of 16.1 nm under 29
Pa load (2 g acceleration with 2 mm 1 mm proof mass).
VI. FABRICATION AND MEASUREMENT RESULTS

F. Dimples A number of prototype microaccelerometers have been suc-

cessfully fabricated. Fig. 5 shows a SEM view of the device with
The polysilicon electrodes contain dimples underneath to re- 2 mm 1 mm proof mass. A rectangular proof mass has been
duce the contact area between the proof mass and the elec- utilized for this device to achieve a large mass without making
trode and, thus, help with alleviating stiction [20]. These dim- the electrodes too long. The device has five electrically isolated
ples are aligned to the silicon nitride dielectric standoffs on the electrodes on each side, which are anchored at the rim using
proof mass and provide electrical isolation as well. The dimple stiffened polysilicon supports. An enlarged view of a single
size is set by the fabrication process and is 3 m on a side electrode stiffener support is presented in Fig. 6. The electrode
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.
 548 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 9, NO. 4, DECEMBER 2000

Fig. 4. Fabrication process sequence of the symmetrical g accelerometer.

Fig. 7. Close-up SEM view of polysilicon electrode, its vertical embedded

stiffener, and the proof mass. Electrode damping holes and isolating nitride
Fig. 5. SEM view of a high-precision symmetric g accelerometer with 2 mm standoffs can be seen as well.
2 1 mm proof mass.

stiffener, and the silicon proof mass. The isolating nitride stand-
offs can be seen in this SEM as well. A close-up view of the
cross-section of the proof mass and stiffened polysilicon elec-
trode is shown in Fig. 8. As is observed, the air gap conformally
separates the electrode stiffener from the proof mass.
The fabricated accelerometers are tested electrostatically
and mechanically on a turntable. Fig. 9 shows the measured
open-loop response of a microaccelerometer with 2 mm 1
mm proof mass with a full-bridge support. Electrostatic force
is used to generate input acceleration in this measurement.
Small displacement of the electrodes is taken into account by
measuring spring constant of the electrode independently using
the included test structures. The device shows a sensitivity of 2
pF/g, and the calculated sensitivity is 2.2 pF/g, as presented in
Table I. The smaller measured sensitivity is possibly due to the
Fig. 6. Enlarged view of single electrode stiffener support. The electrode slightly thicker suspension beams ( m) and the smaller
damping holes can be seen as well. proof mass. The sensitivity can be increased by cutting half of
the beams on one side and forming a cantilever support. The
damping holes can be also seen in this figure. Fig. 7 shows a turntable test has shown a sensitivity of 19.4 pF/g on each side
close-up view of the polysilicon electrode, its vertical embedded of the 4-mm 1-mm accelerometer with a cantilever support.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.
 YAZDI AND NAJAFI: ALL-SILICON SINGLE-WAFER MICRO-g ACCELEROMETER 549

Fig. 8. Cross-sectional close-up of the proof mass and the stiffened polysilicon
electrode.

Fig. 10. Hybrid packaged accelerometer and the interface chip in a standard
24-pin IC package. Accelerometer is mounted a single edge to reduce the
packaging-induced stresses.

and resting on its dimples. The plate stiffness corresponding to

the measured pull-in voltage is 1135 N/m. This measured value
is higher than 510 N/m calculated stiffness (see Section IV) that
is due to a tensile stress of the boron doped polysilicon plate. ii)
The second pull-in happens at 27.8 V, which is caused by the elec-
trode collapsing on the proof mass in the center between four ad-
jacent dimples. The associated stiffness of the plate segment be-
Fig. 9. Measured open-loop response of microaccelerometer with 2-mm 2 tween the four dimples is 1452 N/m. A summary of the design
1-mm proof mass and full-bridge support.
and measured parameters for the 2-mm 1-mm accelerometer
TABLE I with a full-bridge support is presented in Table I.
SUMMARY OF THE DESIGN AND MEASUREMENT PARAMETERS FOR THE 2-mm
2
1–mm ACCELEROMETER WITH A FULL-BRIDGE SUPPORT
VII. PACKAGING
After completion of the fabrication, the microaccelerometer
is mounted on one edge of its frame on a recessed substrate and
is suspended over the recessed area. In this manner, the package
and mounting-induced stresses will be reduced but not affect
the device performance. The device has bonding pads on both
sides. Each pad at the bottom has a metal overhang over the
frame, which is directly bonded to the pads on the mounting
substrate using ultrasonic wirebonder and provide beam-lead
transfer [22]. Access to the top pads is provided by direct wire-
bonding to them. Fig. 10 shows the hybrid packaged accelerom-
eter, its interface circuit [23], and the reference capacitors in a
standard DIP IC package. This package will be ideally sealed to
protect the sensor and the circuit from environmental humidity,
which may affect the parasitics or potentially cause the sensor
In order to measure the stiffness of the electrodes, the bulk
failure due to stiction.
silicon was partially etched out to form the post-release device
sense capacitors without releasing the proof mass. The sacrificial
VIII. CONCLUSION
oxide layer was etched entirely to release the electrodes. These
electrodes were used to measure the pull-in voltage and obtain In this paper, an all-silicon single-wafer fully symmetrical
their stiffness. As the dc sweep bias is increased, two pull-in volt- capacitive micro-g accelerometer is presented. This microac-
ages are observed: i) The first pull-in occurs with 19.3-V bias celerometer has high device sensitivity, low noise, low/control-
that corresponds to the electrode collapsing on the proof mass lable damping, low cross-axis sensitivity, low temperature sensi-
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.
 550 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 9, NO. 4, DECEMBER 2000

tivity, and potentially good long-term stability. All these features [20] C. H. Mastragelo and C. H. Hsu, “Mechanical stability and adhesion of
are achieved by a novel device structure and a new combined microstructures under capillary forces—Part II: Experiments,” J. Micro-
electromech.Syst., vol. 2, pp. 44–55, Mar. 1993.
surface and bulk microfabrication process that overcome the [21] U. Srinivasan, M. R. Houston, R. T. Howe, and R. Maboudian, “Self-as-
micro-g accelerometer design challenges. Two different micro-g sembeled fluorcarbon films for enhanced stiction reduction,” Tech. Dig.
accelerometer designs with proof mass sizes of 2 mm 1 mm Int. Conf. Solid-State Sensors Actuators, pp. 1399–1402, June 1997.
[22] Q. Bai and K. D. Wise, “A high-yield process for three-dimensional mi-
and 4 mm 1 mm are successfully batch fabricated, packaged, croelectrode arrays,” in Tech. Dig. Solid-State Sensors Actuators Work-
and tested. The device with 2-mm 1-mm proof mass and a shop, Hilton Head Island, SC, June 1996, pp. 262–265.
full-bridge support has a measured sensitivity of 2 pF/g. The [23] N. Yazdi and K. Najafi, “An interface IC for a capacitive silicon g
accelerometer,” in Tech. Dig. IEEE Int. Solid-State Circuits Conf., San
measured sensitivity of the 4-mm 1-mm accelerometer with Francisco, CA, Feb. 1999, pp. 132–133.
a cantilever support is 19.4 pF/g, which yields a differential top
and bottom sensitivity of 38.8 pF/g. The calculated noise floor of
these devices at atmosphere are 0.23 g/ Hz and 0.16 g/ Hz,
Navid Yazdi received the B.S. degree (with honors)
respectively. in 1988 from the University of Tehran, Tehran,
Iran, the M.S. degree in 1993 from the University
of Windsor, Windsor, ON, Canada, and the Ph.D.
REFERENCES degree in 1999 from the University of Michigan,
[1] “ADXL50-monolithic accelerometer with signal conditioning,” Analog Ann Arbor, all in electrical engineering.
Devices, Norwood, MA, 1993. Since November 1998, he has been an Assistant
[2] “MMAS40G10D—Micromachined accelerometer,” Motorola, Professor with the Department of Electrical En-
Phoenix, AZ, 1997. gineering, Arizona State University, Tempe. From
[3] F. Rudolf, A. Jornod, J. Berqvist, and H. Leuthold, “Precision ac- 1988 to 1991, he was an R&D Engineer at University
celerometers with g resolution,” Sensors Actuators, vol. A21–A23, of Tehran, where he worked on hardware/software
pp. 297–302, 1990. development for Computer Numerical Control (CNC) machine tools and large
[4] W. Henrion, L. DiSanza, M. Ip, S. Terry, and H. Jerman, “Wide-dynamic data acquisition systems. From 1991 to 1993, he worked on architectures
range direct digital accelerometer,” in Tech. Dig. Solid-State Sensors Ac- and mixed-signal circuit design for VLSI implementation of artificial neural
tuators Workshop, Hilton Head Island, SC, June 1990, pp. 153–156. networks at the University of Windsor. Since 1993, his research work at the
[5] K. Warren, “Navigation grade silicon accelerometer with sacrifically University of Michigan has been focused on high-performance mixed-signal
etched SIMOX and BESOI structure,” in Tech. Dig. Solid-State Sensors interface circuits for microsensors, high-precision microaccelerometer design
Actuators Workshop, Hilton Head Island, SC, June 1994, pp. 69–72. and fabrication, electromechanical oversampled sigma-delta modulators, and
[6] J. Bernstein, R. Miller, W. Kelley, and P. Ward, “Low-noise MEMS vi- wireless telemetry for microsystems. His research interests include integrated
bration sensor for geophysical applications,” J. Microelectromech. Syst., sensors and actuators, micromachining and microstructure fabrication technolo-
vol. 8, pp. 433–438, Dec. 1999. gies, and mixed analog–digital interface circuits for microelectromechanical
[7] T. V. Roszhart, H. Jerman, J. Drake, and C. de Cotiis, “An inertial-grade systems.
micromachined vibrating beam accelerometer,” in Tech. Dig. Eighth Int.
Conf. Solid-State Sensors Actuators, Stockholm, Sweden, June 1995, pp.
656–658.
[8] C. Liu, A. M. Brazilai, J. K. Reynolds, A. Partridge, T. W. Kenny, J. D. Khalil Najafi (S’84–M’86–SM’97–F’00) was born
Grade, and K. Rockstad, “Characterization of a high-sensitivity micro- in 1958. He received the B.S., M.S., and the Ph.D.
machined tunneling accelerometer with micro-g resolution,” J. Micro- degrees in 1980, 1981, and 1986, respectively, all in
electromech. Syst., vol. 2, pp. 235–244, June 1998. electrical engineering from the Department of Elec-
[9] O. Kromer, O. Fromhein, H. Gemmeke, T. Kuhner, J. Mohr, and M. trical Engineering and Computer Science, University
Strohramann, “High-precision readout circuit for LIGA acceleration of Michigan, Ann Arbor.
sensors,” Sensors Actuators, vol. 46–47, pp. 196–200, 1995. From 1986 to 1988, he was a Research Fellow,
[10] M. A. Schmidt, “Wafer-to-wafer bonding for microstructure formation,” from 1988 to 1990 an Assistant Research Scientist,
Proc. IEEE, vol. 86, pp. 1575–1585, Aug. 1998. from 1990 to 1993 an Assistant Professor, from
[11] M. L. Kniffin and M. Shah, “Packaging for silicon micromachined ac- 1993 to 1998 an Associate Professor, and since
celerometers,” Int. J. Microcircuits Electron. Packag., vol. 19, no. 1, pp. September 1998 a Professor and the Director of
75–86, First quarter 1996. the Solid-State Electronics Laboratory, Department of Electrical Engineering
[12] B. Boser and R. T. Howe, “Surface micromachined accelerometers,” and Computer Science, University of Michigan. His research interests include
IEEE J. Solid-State Circuits, vol. 31, pp. 366–375, Mar. 1996. micromachining technologies, solid-state micromachined sensors, actuators,
[13] N. Yazdi and K. Najafi, “An all-silicon single-wafer fabrication tech- and microelectromechanical systems; analog integrated circuits; implantable
nology for precision microaccelerometers,” in Tech. Dig. Ninth Int. Conf. biomedical microsystems; hermetic micropackaging; and low-power wireless
Solid-State Sensors Actuators , Chicago, IL, June 1997, pp. 1181–1184. sensing/actuating systems. He is an Associate Editor for the Journal of
[14] T. B. Gabrielson, “Mechanical-thermal noise in micromachined Micromechanics and Microengineering of the Institute of Physics Publishing.
acoustic and vibration sensors,” IEEE Trans. Electron Devices, vol. 40, Dr. Najafi was awarded a National Science Foundation Young Investigator
pp. 903–909, May 1993. Award from 1992 to 1997, the Beatrice Winner Award for Editorial Excellence
[15] J. D. Zook et al., “Characteristics of polysilicon resonant microbeams,” at the 1986 International Solid-State Circuits Conference, the Paul Rappaport
Sensors Actuators, pp. 51–59, Oct. 1992. Award for co-authoring the Best Paper published in the IEEE TRANSACTIONS
[16] V. M. McNeil, M. Novack, and M. A. Schmidt, “Design and fabrication ON ELECTRON DEVICES, and the Best Paper Award at ISSCC 1999. In 1994, he
of thin-film microaccelerometer using wafer bonding,” in Proc. IEEE received the University of Michigan’s “Henry Russel Award” for outstanding
Int. Conf. Sensors Actuators, Yokohama, Japan, June 1993, pp. 822–825. achievement and scholarship and was selected “Professor of the Year” in 1993.
[17] B. Puers and W. Sansen, “Compensation structures for convex corner In 1998, he was named the Arthur F. Thurnau Professor for outstanding contri-
micromachining in silicon,” Sensors Actuators, vol. A21–A23, pp. butions to teaching and research and received the College of Engineering’s Re-
1036–1041, 1990. search Excellence Award. He has been active in the field of solid-state sensors
[18] C. Standman, L. Rosengern, H. Elderstig, and Y. Bucklund, “Fabrica- and actuators for more than 18 years and has been involved in several confer-
tion of 45 mirrors together with well-defined V-grooves using wet ences and workshops dealing with solid-state sensors and actuators, including
anisotropic etching of silicon,” J. Microelectromech. Syst., vol. 4, pp. the International Conference on Solid-State Sensors and Actuators, the Hilton
213–219, Dec. 1995. Head Solid-State Sensors and Actuators Workshop, and the IEEE/ASME Micro
[19] D. A. Monk, D. S. Soane, and R. T. Howe, “Sacrificial layer SiO2 Electromechanical Systems (MEMS) Workshop. He is the Editor of Solid-State
wet etching for micromachining applications,” in Tech. Dig. Sixth Int. Sensors for IEEE TRANSACTIONS ON ELECTRON DEVICES, Associate Editor for
Conf. Solid-State Sensors Actuators, San Francisco, CA, June 1991, pp. IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, Associate Editor for IEEE
647–650. JOURNAL OF SOLID-STATE CIRCUITS.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on May 29,2024 at 19:28:12 UTC from IEEE Xplore. Restrictions apply.