I70 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS. VOL. 1. NO. 4.

DECEMBER 1992

Electrostatic Comb Drive Levitation and

Control Method
William C. Tang, Member, IEEE, Martin G . Lim, and Roger T. Howe, Member, IEEE

Abstmct-This paper presents the theory, slmulatlon results,

and experimental study of the levitating force (normal to the
substr8te) assoc&led with interdigitated capacitor (cleetro-
statk comb) Inter81 actuators. For compliant suspensions, nor-
mal displacements of over 2 pm for 8 comb bias of 30 V are
observed. This phenomenon is due to electrostatic attractkm
induced on top of the suspended structure. By electrically bo-
lating alternating drive-comb Pngers and applying voltages of
equal magnitude and opposite sign, levitation force can be r e
duced by an order of magnitude, while reducing the lateral
drive force by lcss than a factor of 2. The levitation theory in-
corporates elretrostcrtk simulation results, and agrees well with Fig. 1. Layout of I linear lateral resonator driven and sensed with inter-
experimental data. digitated capacitors (electrostatic combs).

I. INTRODUCTION In this paper, the theory of electrostatic levitation is de-

S URFACE-MICROMACHINED plysilicon resona-

tors which are driven by interdigitated capacitors
(electrostatic combs) have several attractive properties
scribed along with the 2-Dsimulation results, which are
then compared with experimental measurements of the re-
duction in levitation effect for the modified comb design
(Fig. 1) [1]-[4]. Vibrational amplitudes of over 10 pm arc with independently biased fingers.
possible with relatively high quality factors at atmo-
spheric pressure, in contrast to structures which move 11. VERTICALLEVITATION THEORY
normal to the surface of the substrate. The comb-drive A. DC Levitation
capacitance is linear with displacement, resulting in an Successful electrostatic actuation of micromechanical
electrostatic drive force which is independent of vibra- structures requires a ground plane under the structure in
tional amplitude. Electrostatic combs have recently been order to shield it from relatively large vertical fields [9],
used for the static actuation of friction test structures [ 5 ] , [IO]. It has been observed that if the underlying nitride
microgrippers [6],and force-balanced accelerometers [7]. and oxide passivation layers are not covered with a
Potential applications of lateral resonators include res- grounded polysilicon shield, the application of a dc bias
onant accelerometers and rate gyroscopes, as well as res- voltage will cause the structures to be stuck down to the
onant microactuators [8]. For efficient mechanical cou- substrate. Furthermore, varying the bias voltage causes
pling between a vibrating pawl and a toothed wheel, it is the structures to behave unpredictably. In previous studies
essential that both structures remain coplanar. However, of the electrostatic-comb drive, a heavily doped polysili-
2-pm-thick plysilicon resonators with compliant folded- con film underlies the resonator and the comb structure.
beam suspensions have been observed to levitate over However, this ground plane contributes to an unbalanced
2 pm when driven by an electrostatic comb biased with a electrostatic field distribution, as shown in Fig. 2 [ll].
dc voltage of 30 V. This effect must be understood in The imbalance in the field distribution results in a net ver-
order to design functioning resonant microactuators, with tical force induced on the movable comb fingers, which
the possibility that levitation by interdigitated combs may levitates the structure away from the substrate. Whether
offer a convenient means for selective pawl engagement. this force causes significant static displacement or excites
Manuxrip received May 20, 1992; revised November 13. 1992. This
a vibrational mode of the structure depends on the com-
work was fupported by the Berkeley Sensor & Actuator Center. an NSFl pliance of the suspension and the quality factor for verti-
IndustrylUnivcrsiiy Cooperative Research Center. Subject Editor, K. Pe- cal displacements.
tersen. Using a 2-Delectrostatic sirnulation program (Maxwell
W. C. Tang is with the Ford Research Laboratory, Ford Motor Com-
pany, Dearborn, MI 48121. Solver [ 111) to simulate the cross section of the comb fin-
M. G. Lim is with the Xerox Palo Alto Resealrh Center, Palo Alto, CA ger biased with a dc voltage, we obtain the induced ver-
94304. tical force per unit length of the movable comb finger at
R . T. Howe is with the Depadment of Electrical Engineering and Com-
puter Sciences. University of California at Berkeley. Berkeley, CA 94720. different levitation positions. This vertical force density,
IEEE Log Number 92066%. Fa [nN-pm-'], is then plotted against levitation, z [pm],

1057-7157/92$03.00 0 1992 IEEE

 TANG et al.: ELECTROSTATIC COMB DRIVE 171

overlap length, then the levitation force is given by F, =

FSAx and the vertical drive capacity, Y~ = yroAx.Equa-
tion (1) can then be rewritten as

This equation implies that since F, a (to- z ) , the levi-

Iv,
- tation force behaves like an electrostatic spring, such that
F, = k t ( q - z), where ke is the electrostatic spring con-
stant,

Fig. 2. Cross section o f the potential contours (dashed lines) and the elec-
tric fields (solid lines) of a comb finger under levitation force induced by Both (2) and (3) will be used extensively in the following
two adjacent electrodes biased at I positive potential.
discussions on vertical transfer function and vertical res-
n 2n
onance.
I
0.15 \ :-4v
1-w
I
!
B. Vertical Transfer Function
The total vertical force acting on the comb fingers in-
cludes the levitation force, FL. and the passive restoring
I 0.10 spring force, Fk, generated by the mechanical suspensions
P of the system, as illustrated in Fig. 4. The vertical dc
0.05 transfer characteristics can be evaluated by solving
i Fnet = Fz - Fk = 0, (4)
f 0.00
L where F,,, is the net force acting on the movable comb
finger, and
-0.05

Fk = kzZ, (5)
-0.10
where k, is the vertical spring constant. Substituting (2)
0 0.5 1 1.6 2
ladtatlo. 11-
and (5) into (4). ,we have
Fig. 3. Simulated levitation force density versus vertical position under
different biasing voltages. Finger dimensions are 2 pm thick X 4 pm wide;
nominal separations from the substrate and the drive electrodes are 2 pm.
Solving for in terms of V, yields
at different dc bias voltages, Vp [VI.Fig.'3 is the simu-
lation results of a 4-pm-wide X 2-pm-thick comb finger zoYr v:
z= (7)
excited by two identically sized electrodes situated 2 pm kzzo + Yzv:'
away from both sides of the finger, and 2 pm above a Figure 5 is a plot of this equation by assuming a value
grounded substrate. There are several important obser- of 86 nN-pm-' for k, [see (27)] and 720 pm for Ax to
vations from this simulation. First, the stable equilibrium yield a value of 1.25 n N . V 2 for rz(18 fingers on each
levitation, ~0 (1.22 pm for this case), is the same for any of the two comb drives and a 20 pm finger overlap). The
nonzero bias voltages. Thus, in the absence of a restoring initial slope of the curve is largely dependent on yrrwhich
spring force, the movable comb fingers will be levitated determines the threshold voltage where levitation reaches
to ~0 upon the application of a dc bias. Second, given z, 90% of the maximum, and the asymptotic value ap-
e
F is proportional to the square of the applied dc bias,
V,. And at any Vp, F,, is roughly proportional to (q - z ) proaches.,z, Therefore, in certain applications where ver-
tical levitation is undesirable, both yZ (which can be in-
as long as z is less than q.Thus terpreted as the sensitivity of levitation to the applied
voltage) and q (the equilibrium levitation in the absence
of returning spring force) should be minimized. The
method to control vertical levitation is discussed later in
this paper.
C. Vertical Resonant Frequency
where the constant of proportionality, ys (1.73 x lo-' In this subsection, we consider the case where the re-
nN-prn-'.V-' for this case), is defined as the vertical sonators are not damped vertically, such as for the case
drive capacity per unit length. If Ax is the comb-finger- of vibrations in vacuum. In the absence of damping, the

___- - ~- -
Rsproduced with psrnission Or copyright m r . Furthsr reproduction p r o h i b i t d .
I72 JOURNAL OF MlCROELECTROMECHANlCAL SYSTEMS. VOL. I. NO. 4. DECEMBER 19'32

Carrying out the indicated operations yields

(to - Zf) 2 (zo - 2,)
F;(VDv 2 ) 72 - v p + 27: -
to to
7: v:
* Vf U&) --zo
Zl(0. (1 1)
Id Substituting this result and (5)for Fk into (8) yields
Ground
Fig. 4. The vertical forces acting on a movable comb finger.

1.2

Since Z, is constant. ( d Z f / d r ) = 0. Further. in the light

1
of (6), the dc terms cancel, yielding

-
10.8

3I 0.6
where

!i 0.4 G = 27, (& -

- ") V, [nN . V-'] (14)
zo
and
0.2
k,,, = k, + k, [nN - pm-'1, (15)
with k, as defined in (3). Equation (13) describes an un-
0 damped oscillator with drive vd(r). The system has a res-
0 10 20 30 40 onant frequency
Applied Voltage M
Fig. 5 . Theoretical levitation (:) versus dc bias: (V,) based on (7). with
0,= k, k,+
(T)112.
7.- = 1.25 nN.V-'. li, = 86 nN-pm'.'. and z, = 1.22 p m .

An important observation is that the resonant frequency

governing equation of motion is a second-order differen- is a strong function of the applied bias, V,. If we define
tial equation given by the mechanical resonant frequency (under zero bias) as
d2r wo, then
Fnc, = M-i-9 (8)
dt
where M is the effective mass of the vibrating structure.
The net vertical force, Facr.which is zero in dc analysis
[eq. (4)], is now nonzero when the bias voltage, V f . is
replaced with a generalized, time-varying drive voltage,
which is plotted in Fig. 6 as a function of V,. In the limit
VD 0):
as V f 03,

(9)
This relationship can be linearized by assuming that vD(t) If we further assume that the input drive vd(t) is sinu-
= V, +
c i d ( r ) and z(r) = 2, + td(t), where V, >> ud(r) soidal, e.g.. U,&) = A sin ( w t ) . then the resulting output
and Z, >> zd(r). It is further assumed that v D ( t ) and z ( t ) is given by
are continuously differentiable. Expanding F, in a Taylor
series about the operating point (V,, 2,) and retaining only zd(r) = B sin (at + $1, (19)
the linear terms yields
where, through standard Laplace transformation analysis,

B = lH(jw)IA, (20)
 TAN0 el al.: ELECTROSTATIC COMB DRIVE I73

3.5

2.5

f a

i.5

0.5
Gi-oulM pcm
0 Fig. 7. Cross seclion of the potential contours (dashed lines) and the elec-
D 10 20 90 40 tric fields (solid lines) rmund a movable comb finger when differential dc
Apmvaly. m bias is applied to the two adjacent electrodes.

-
Pig. 6. Theoretical ftcquency ratio (wl/oo) versus dc bias (V,) based on
(17). with ye 1.25 nN-V-*. k: = 86 nN-pm-’. and ~0 = 1.22 pm.

and
G
H(jo) =
kcq - Mu2*
Note that since no damping is assumed, ( H ( j o ) (is infi-
nite at w = wl and
0 f o r o c w1
m=[ -x f o r o > wI.
(23)

In any real system some damping will exist, such that the
resonant amplitude is finite, and the phase transition is Pig. 8. Cross section of the potential contours (dashed lines) and the elec-
continuous. tric fields (solid lines) around a movable comb finger when differential dc
The previous discussion shows that the comb can be b u s is applied to the two adjacent elestrodcs and the striped ground con-
ductors.
used to control the vertical resonance. In the case where
the vertical mechanical spring constant of the suspension
is very close to the lateral one, i.e., k, = k,. the undesir-
able simultaneous excitation of both vertical and lateral
modes of motion is conveniently avoided, since the dc
bias shifts the vertical resonant frequency.

D. Modijied Comb Design for Levitation Control

In addition to shifting the vertical resonant frequency,
it is desirable to control the dc levitation effect as well.
There are several means to reduce the levitation force. By
eliminating the ground plane and removing the substrate
beneath the structures, the field distribution becomes bal-
anced. Alternatively, a top ground plane suspended above
the comb drive will achieve a balanced vertical force on
the comb. Both of these approaches require much more
complicated fabrication sequences. A simpler solution is
to modify the comb drive itself. Reversing the polarity on 40 Pm
alternating drive fingers results in the field distribution Fig. 9. Scanning electron micrograph (SEM) of the crossover slructure for
electrical isolation of alternating drive electrodes.
shown in Fig. 7. In this case, vertical electric fields ter-
minating on the top surface of the movable comb finger
are eliminated. To further suppress levitation, the ground as illustrated in Fig. 8. The polysilicon layer is used to
plane is modified such that underneath each comb finger form the crossovers to electrically isolate alternating comb
there is a strip of conductor biased at the same potential, fingers (Fig. 9). Simulation shows that the levitation force

Reproduced with pernirtion of copyright omnor. Further reproduction prohibitad.

114 JOURNAL OF MICROBLF13TRC)MECHANlCAL SYSTEMS, VOL. I. NO. 4. DECEMBER 1992

is suppressed by over an order of magnitude compared where k, is the lateral spring constant, L is the length of
with the original biasing scheme. the folded beam (400 pm),Mp is the plate mass, M, is the
This technique results in a stable system. In contrast, it mass of the outer connecting trusses, and M b is beam
is difficult to achieve stability if one attempts to counter- mass. The moment of inertia of the supporting beams with
act the levitation force with an attractive force on the bot- respect to z (I,) is given by [ 131
tom of the movable fingers by driving the underlying elec-
trodes, because of the strong tendency to pull in the fingers
and collapsing them to the substrate.
I: = s x 2 dA = -
h (U
48
Therefore, E can be expressed as
+ b)(a2+ b’). (25)

HI. EXPERIMENTAL
TECHNIQUES 8r2f:L3(Mp + iM, + $Mb)
E= (26)
h(a + @(a2 + b2)
*
Levitation amplitudes are recorded from low-voltage
SEM pictures at various dc biases on the combs. All Using a plysilicon density of 2.3 X lo3 kgsm-’, the
structures are wired together to make possible the mea- value for Young’s modulus is found to be 150 GPa for
surement of a number of structures in a single SEM ses- this process run, which is consistent with earlier results
sion. The angle of tilt and the magnification are fixed for [l], [2]. The vertical spring constant is evaluated as
comparison between different structures. VerticaI dis-
placements are evaluated by accurately measuring the 24E1,
k:=-=
2h3(u2 + 4ab + b2)E
SEM images with a set of standard line widths. Since the L3 3t3(a + b)
planar geometries of the structures are well defined and
the amount of etching undercut of the structural polysili- = 86 nN-pm-’. (27)
con can be measured accurately with optical technique, Finally, with this equation, the levitation force can be
the amount of levitation can be inferred accurately by found as
comparing the dimensions in z with those in x and y. This
technique yields an uncertainty level of around 300 A F, = k,Az. (28)
relative to the planar dimensions, as indicated with the where Az is the vertical displacement.
error bars in Figs. 10, 13, and 14. All the prototype de-
vices are designed with 400-pm-long folded-beam sup- IV. EXPERIMENTAL
RESULTS
pons to provide compliance in both the out-of-plane ( 2 ) A. DC Levitation Results
and the lateral (x) directions. The polysilicon film thick-
ness of the device under test is measured to be 1.94 pm. Levitation is first measured by applying a voltage of 0
Due to difficulties with the plysilicon plasma etching to 25 V to all drive fingers on a prototype with 18 movable
process, the cross section of the suspension is slightly comb fingers and 19 fixed drive fingers, the result of which
trapezoidal, with the width at the top of the beam a = 2.2 is plotted in Fig. 10. Fig. I 1 is an SEM of a comb struc-
pm and the width at the bottom b = 2.8 pm. The finished ture levitated under a 10 V dc bias. Note that the voltage-
comb finger dimensions are 4.2 pm wide on top and 4.8 contrast effect inside the SEM causes the drive fingers at
prn on the bottom, with an overlap length between the a higher potential to appear darkened. The vertical dis-
finger and the driving electrodes of 20 pm. The gap be- placement increases with applied voltage and reaches an
tween the drive electrodes and the comb fingers varies equilibrium near 20 V, where the attractive forces be-
from 1.8 p n near the top surface to 1.2 pm near the bot- tween the displaced interdigitated fingers offset the attrac-
tom. The structures are separated from the substrate by a tive forces induced on the top surfaces of the movable
nominal 2 prn distance, the original thickness of the sac- fingers.
rificial layer. The striped ground planes are similar to The initial negative deflection for a grounded comb,
those illustrated in Fig. 8, and can be biased accordingly. shown in Fig. 10. cannot be attributed to gravity. With
The structures are first resonated laterally to evaluate the test chip at a tilt angle of 7’ from vertical, the com-
Young’s modulus, E, from the measured resonant fre- ponent of the gravitational force normal to the substrate
quencies using Rayleigh’s method [l], [2]. The same is only 0.09 nN. whereas the force required to cause the
value for Young’s modulus is then used for the vertical- initial deflection is 5.9 nN. Nor can there be any built-in
motion calculations based on the assumption that polysil- stress substantial enough to have caused the deflection. A
icon is materially isotropic. The expression for lateral res- structure with one of the two anchors deliberately de-
onant frequency of the test resonator was derived in [3] stroyed and thus free of built-in stress showed even more
and [12], with the following result: zero-bias deflection inside the SEM. Charging effects in
the exposed underlying dielectric films between the inter-
digitated striped ground plane are a likely source of this
offset displacement. With a constant supply of energetic
electrons inside the SEM, it is possible that the trap-charge
density reaches an equilibrium balanced by the relaxation
rate. In order to investigate this hypothesis, we simulated
TANG U 41.: ELECTROSTATIC COMB DRIVE I75

400

300

200

2 Iml
100
f'
Fig. 12. Cross section of the potential contours (hairlines) and the electric
0 fields (solid lines) of a grounded system with an evenly distributed trapped
charge in the silicon nitride layer.

.loot I I I I
charge density should be treated as a lumped parameter
0 5 10 15 20 2s
AppUed voltage M representing the ner fixed-charge effect. If further inves-
Fig. 10. Measured levitation of an 18-finger comb as a result of applying tigation is needed, it is possible to verify this initial ob-
a common voltage to all drive electrodes. servation by optical technique, avoiding the undesirable
charging inside the SEM.
To account for this trap-charge-induced deflection, we
introduce the term F, as the total trap charge force:

F, =
B
-
(z + d)"
where d is the nominal offset of the structure from the
substrate and p is the constant of proportionality, simu-
lated to be 23.6 nN.pm2. Equation (29) is exact for a
layer of trapped charge between a conductor and a ground
plane. To account for the partial shielding from the ground
strips and fringing-field effects, the power of the denom-
inator in (29) may be adjusted. The term F, is now added
to (4) as

Fig. 11.
- 4 pm
SEM of an 18-finger comb levitated under 10 V dc bias. Note.that
F,, = F, - Fk - F,.= 0.
Equations (2) and ( 5 ) are combined with (29) to substitute
the terms in (30). yielding an implicit function as follows:
(30)

the drive fingers, because of the positive bias. appear darkened in the SEM.

the effect of trapped charge in the silicon nitride passi-

vation layer. Figure 12 shows the electric fields and the Fig. 13 is the result of fitting the curve to the data in Fig.
equipotential contours by assuming an evenly distributed 10 by adjusting z,, and yz. The best-fitted values are .Q =
trap-charge density of -4.3 x 10" qecm-' inside the 0.4 pm and z = 1.7 nN*V-2. The equilibrium levitation
0.15-pm-thick nitride layer, with all the comb fingers, (20) of 0.4 Fm is much less than the observed 2 pm levi-
conductors, and the bulk silicon grounded. Simulation re- tation in some structures with soft spring supports and a
sults show that this trap-charge density is sufficient to in- blanket ground plane. The experimental result for yz is
duce the required force on a structure with two 18-finger comparable to the simulated value of 1.25 nN-V-2 for an
combs. The magnitude of this charge density is process ideal comb structure with 2 pm finger gaps formed with
dependent and is not uncommon for LPCVD nitride films vertical sidewalls.
[14]. However, it should be noted that trap charges may The effectiveness of levitation control by alternating the
also be present in the nitride-oxide interface as well as potentials on every drive finger is evaluated next. Fig. 14
the thin native oxide layers on the polysilicon ground is a plot of the measured vertical displacement resulting
strips. Furthermore, surface charges on the nitride layer from holding one set of an alternating drive fingers at + 15
can be either positive or negative, adding complexity to V and varying the other set of electrodes from - 15 V to
the system. Therefore, the simulated value for the trap- +15 V. This structure is the same as that tested in Fig.
 JOURNAL OF MlCROELECTROMECHANlCAL SYSTEMS, VOL. 1. NO. 4, DECEMBER 1992

1
I I I I
- 4pm
Fig. 15. SEM of an 18-fingercomb under f 10 V balanced dc bias on the
5 10 15 20 25 alternating drive fingers, indicating almost no levitation. Note that fingen
Applled voltage M at higher potentials appear darkened due to voltagecontrast effect in the
Fig. 13. M e a s u d and calculated levitation of the IS-finger comb.
SEM .

drive fingers induces a weaker lateral force on the struc-

ture than the unbalanced comb. The balanced comb is
advantageous only if the trade-off between levitation
suppression and loss of lateral drive is favorable. In order
to quantify the comparison of the lateral and vertical force
reductions, we define the lateral drive capacity, yx, of an
electrostatic-comb drive as the lateral force (F,) induced
per square of the applied voltage (Vp):

The value of y, is found to be 0.58 f 0.04 n N . V 2 per

drive finger for all the unbalanced comb designs. Both 7,
and yr for the balanced and unbalanced comb drives are
tabulated in Table I. The reductions in y are defined as
the ratios of the drive capacities of the unbalanced comb
to those of the balanced one.
Although the balanced-comb design does not com-
pletely eliminate the levitation as predicted by idealized
theory, which assumes evenly spaced comb fingers with
Fig. 14. Measured levitation of the 18-finger comb with thc dc bias on onc
- +
set of electrodes varying from 15 V to 15 V, while the bias on the vertical sidewalls, nevertheless a 16:l reduction on the
alternate set is held at + 15 V. vertical drive capacity is achieved while suffering only a
1.6: 1 reduction in induced lateral force.
10. As expected, negative voltages in the range of
- 10 V to - 15 V suppress the lifting behavior. As the C. Vertical Resonant Frequencies and Quality Factors
disparity between the magnitudes of the voltages in- The vertical resonant frequencies are found by stepping
creases, more lifting occurs, with the limiting case of 15 + the output frequency of an HP 4192A LF impedance ana-
V applied to all drive fingers yielding the same vertical lyzer at 0.1 Hz steps at different dc biases from 5.0 V to
displacement as found in Fig. 10. Fig. 15 is the SEM of 15.0 V, and fixing the ac drive amplitude at 50 mV. Fig.
a comb structure under a f 10 V balanced biasing on the 16 is the SEM of a comb structure driven into vertical
alternating drive fingers, indicating almost no levitation. resonance under vacuum torr) inside the SEM
chamber. The vibration amplitude is estimated to be 2 pm
B. Vertical and Lateral Drive Capacities peak-to-peak. The results are plotted in Fig. 17, with the
It is found that, besides suppressing the vertical levi- theoretical curve fitted to the data with zo and yzadjusted
tation, the balanced-biasing approach on the alternating to 0.4 pm and 1.7 nN.V-’, respectively.
 TANG er al.: ELECTROSTATIC COMB DRIVE I77

TABLE I plitude after stopping the drive. The Q for all the struc-
COMPARISON BETWEEN
7: AND 7,
tures tested in vacuum is close to 50 0oO. which is the
7: at z = 0 rx same as the lateral Q in vacuum. However, it is found
TY Pe kO.04 nN-V-' i 0 . M nN.V-' that vertical resonance can be excited over a range of f 10
Unbalanced I .7 0.58
Hz of the resonant frequencies, which, if the system were
Balanced 0. I 1 0.36 linear, would have put the values for Q in the range of
250 to 500. This apparently excessive f3 dB bandwidth
Reduction ratio 16:1 'A:' may be due to the highly nonlinear function of vertical
resonant frequency on the drive voltage. Since the vertical
position is also a function of the applied voltage, the large
vibration amplitude of 2 pm peak-to-peak indicates that
the vertical resonance is in the nonlinear region even at
an ac drive level of 50 mV. Nevertheless, the excellent
fit of the linear theory with the experimental results veri-
fies the usefulness of the frequency-shifting phenomenon
as a way to control vertical resonant frequency.

V. CONCLUSIONS
We have successfully modeled and experimentally in-
vestigated both dc and ac levitation induced by electro-
static-comb drives by direct tests in an SEM,which pro-
vides insights into designing structures for controlled out-
of-plane motions. The best levitation suppression is ob-
tained by alternating the electrodes at every comb finger

- 4 pm
Fig. 16. SEA4 of an 18-finger comb driven into vertical resonance under a
with a striped ground plane underneath the comb struc-
ture. Levitation can be further reduced by designing struc-
tures with vertically stiff suspensions. If controlled levi-
50 mV ac drive on top of a 5 V dc bias. tation is desired, soft suspensions can be used together
with ratioed differential and common mode voltages ap-
plied to the two electrodes. Decoupling of vertical and
lateral resonant, modes can be achieved by simply apply-
3
ing a dc bias to the drive electrodes.
ACKNOWLEDGMENT
The authors wish to thank C. Hsu and the staff at the
Berkeley Microfabrication Laboratory for their assistance
2 in the fabrication process. The discussions on dielectric
trap charges with Prof. J. Chung (Massachusetts Institute
of Technology) and Prof. M.Aslam (Michigan State Uni-
versity) are gratefully acknowledged.
1
REFERENCES
[l] W. C.Tang. T.-C. H.Nguyen. and R. T. Howe. "Laterally driven
polysilicon resonant microstructures," Sensors and Acruators, vol.
20. pp. 25-32. 1989.
[2]W. C. Tang, T.-C.Nguyen. M. W. Judy, and R. T. Howe. "Elcf-
0 6 10 15 20 tmstatictomb drive of lateral polysilicon resonators," in Proc. 5th
h r . Conf. Solid-state Sensors and Actuators (Transducers '89) (Mon-
Applted voltege M IRUX), vol. 2. JUM 25-301989. pp. 328-331.
Fig. 17. Measured and fitted vertical resonant frequencies of the 18-finger [3] W. C. Tang. "Electrostatic comb drive for resonant sensor and ac-
comb as a function of dc bias. The zero-bias resonant frequency, oo, is tuator applications." Ph.D. thesis, Dept. EECS, Univ. California,
extrapolated to be 4.69 kHz. Bc&elcy. Dec. 1990.
[4] R. A. Brennen, A. P. Pisano, and W. C. Tang. "Multiple mode mi-
cromechanical resonators." in Proc. IEEE Micro Elecrro Mech. Syst.
The vertical quality factor, Q,is evaluated using a time- Workshop (Napa Valley, CA), Feb. 1990, pp. 9-14.
[ 5 ] M. G . Lim, J. C. Chang, D. P. Schultz. R. T. Howe. and R. M.
domain method: White, "Polysilicon microstructures to characterize static friction,"
in Proc. IEEE Micro Elccrro Mech. Sysr. Workshop (Napa Valley,
Q = 1.43Atf, (33) CA), Feb. 1990, pp. 82-88.
161 C . 4 . Kim, A. P. Pisano. R. S. Muller, and M.0 . Lim, "Polysilicon
where f,is the resonant frequency and A t is the time for microgripper," in Tech. Dig. I€€& Solid-Stare Sensor and Actualor
the resonance to decay from 90% to 10% of the full am- Workshop (Hilton Head, SC), June 1990. pp. 48-51.

_ _ _ ~ . _ . .. . -----..- -

RsproducSa u l t h ~w-l88lon of m p w l s h t -. Further reproduction p m h i b l t d .

 I78 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 1, NO. 4. DECEMBER 1992

(71 W. Yun. R. T. Howe. and P. R. Gray. “Surfax micromachined. Martin G. Lim received the B.S.M.E. degree in
digitally force-balanced accelerometer with integrated CMOS dctec- 1987 from the Univenity of California at Berke-
tion circuitry.” in Tech. Dig. IEEE Solid-State Sensor and Acruaror ley. Later he joined the Berkeley Sensor and Ac-
Worhhop (Hilton Head, SC), June 1492, pp. 126-131. tuator Center and completed his M.S.in 1990.
[8] A. P. Pisano. *‘Raonant-stmcture micromotom,” in Proc. lEEEMI- He is currently a member of the rescarch staff
cro Elcctro Mcch. Syst. Workshop (Salt Lake City, UT), Feb. 1989. at Xerox Palo Alto Research Center working on
pp. 44-48. acourtic ink printing. His responsibilities include
191 Y.-C. Tai. L.-S. Fan. and R. S. Muller, “IC-processed micro-mo- reseanh and development in micromechanical
tors: design, technology and testing,” in Proc. lEEE Micro Uccfro systems and their fabrication pnxles~es.
Ucch. Syst. Workshop (Salt Lake City. UT). Feb. 1989, pp. 1-6.
[IO] M. Mehregany. P. Nagarknr. S. D. Senturia, and J. H. Lang, “Op-
eration of microfabricated harmonic and ordinary side-drive motors,”
in Proc. IEEE Micro Ekctro Mcch. Sysr. Worhhop (Nap. Valley.
CA), Feb. 1990, pp. 1-8.
[Ill Ansoft Corp.. 4 Station Square. 660 Commerce Court Bldg., Pitts-
burgh, PA, Maxwell Solver, Electrostatic Package.
1121 W. C. Tang, M. 0 . Lim. and R. T. Howe, “Electrostatically bI-
anced comb drive for controlled levitation,” in Tech. Dig. IEEE Solid-
Stare Sensor and Acruntor Workshop (Hilton Head, SC), June 1990.
pp. 23-27.
[I31 J. M. G ~ Rand S. P. Timoshenko. Mechanics of Marcriafs. 2nd ed.
Belmont: Wadsworth, 1984.
1141 S. M. Szc. Physics ofSemiconducror Devices, 2nd ed. New York: Roger T. Howe (S’80-M’84) received the Ph.D.
John Wiley. 1981. degree in electrical engineering from the Univer-
sity of California at Berkeley in 1984.
He was on the faculty of Camegie-Mellon Uni-
Willinm C. Tang (S’86-M’W) received the Ph.D. versity from I984 to 1985 and was an assistant
degree in electrical engineering from the Univer- professor at the Massachusetts Institute of Tech-
sity of California at Berkeley in 1990. nology from 198s to 1987. In 1987. he joined the
He was an associate engineerlscientist at the Depaltment of Electrical Engineering and Com-
General Products Division, 1BM Corp., from 1982 puter Sciences at the University of California at
to 1984. He is currently a senior rrsearch engineer Berkeley, where he is now an associate professor
at the Ford Research Laboratory, Ford Motor and an associate director of the Berkeley Sensor
Company, Dearborn, MI. His primary mponsi- and Actuator Center. His research interests include resonant microsensors
bilities include research in micromachining tech- and microactustors, micromachining technology, and integrated-circuit de-
nology and automotive applications of micmmc- sign.
chanics. Dr. Tang is a member of the Materials Prof. Howe is a member of the Materials Research Society, the Electro-
Research Society- chemical Society. and Sigma Xi.

-.

iproduced with prni=tion of copyright owner. Further reproduction prohibited.