JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO.
4, DECEMBER 2001
A Methodology and Model for the Pull-In Parameters
of Electrostatic Actuators
Yael Nemirovsky, Fellow, IEEE, and Ofir Bochobza-Degani, Student Member, IEEE
Abstract—This paper presents a generalized model for the
pull-in phenomenon in electrostatic actuators with a single
input, either charge or voltage. The pull-in phenomenon of a
general electrostatic actuator with a single input is represented
by an algebraic equation referred to as the pull-in equation.
This equation directly yields the pull-in parameters, namely, the
pull-in voltage or pull-in charge and the pull-in displacement.
The model presented here permits the analysis of a wide range of
cases, including nonlinear mechanical effects as well as various
nonlinear, nonideal, and parasitic electrical effects. In some of
the cases, an analytic solution is derived, which provides physical
insight into how the pull-in parameters depend upon the design
and properties of the actuator. The pull-in equation can also
yield rapid numerical solutions, allowing interactive and optimal
design. The model is then utilized to analyze analytically the case
of a Duffing spring, previously analyzed numerically by Hung and
Senturia, and captures the variations of the pull-in parameters
in the continuum between a perfectly linear spring and a cubic
spring. Several other case studies are described and analyzed
using the pull-in equation, including parallel-plate and tilted-
plate (torsion) actuators taking into account the fringing field
capacitance, feedback and parasitic capacitance, trapped charges,
an external force, and large displacements. [655]
Index Terms—Actuators, electrostatic, MEMS, model, pull-in.
I. INTRODUCTION
M EMS operate at several energy domains, including
thermal, mechanical, and electrical. Hence, it is a
challenge to define a unified approach to the modeling of
MEMS [1]. Electrostatic actuators utilize coulombic attraction
between two conductors to induce displacement or to exert
force [2]. Electrostatic actuators are used in a wide range of
important applications, including switched micromirrors for
optical communication [3]–[7] and for digital light processors
[8], [9], radio-frequency microswitches [10], [11], voltage-con-
trolled capacitors [11], and pressure sensors [12], [34], [35].
Electrostatic actuators are attractive because of the high energy
densities and large forces available in microscale devices [13].
Moreover, these actuators are relatively simple to design and
fabricate as well as to integrate in a microsystem. However,
electrostatic actuation may exhibit an inherent instability
situation, known as the pull-in phenomenon [13]–[26]. This
effect may be either derogatory or useful, depending on the
application, and needs to be carefully modeled while designing
an electrostatic actuator.
Manuscript received December 15, 2000; revised March 1, 2001. Subject
Editor H. Fujita.
The authors are with the Kidron Microelectronics Research Center, Elec-
trical Engineering Department, Technion—Israel Institute of Technology, Haifa
32000, Israel (e-mail: nemirov@ee.technion.ac.il; ofird@tx.technion.ac.il).
Publisher Item Identifier S 1057-7157(01)06858-5.
1057–7157/01$10.00 © 2001 IEEE
601
In this paper, a generalized model for the pull-in phenom-
enon in electrostatic actuators with a single input, either voltage
or charge, is presented. Generalized algebraic governing equa-
tions for the pull-in phenomenon are derived and discussed in
Section II. Solving these equations directly yields the pull-in pa-
rameters, namely, the pull-in voltage or pull-in charge and the
pull-In displacement. This approach permits the analysis of a
wide range of case studies, including nonlinear mechanical ef-
fects as well as nonideal and parasitic electrical effects. In Sec-
tion III, the model is first utilized to analyze analytically the case
of a Duffing spring, previously analyzed numerically by Hung
and Senturia [13], and captures the variations of the pull-in pa-
rameters in the continuum between a perfectly linear spring and
a cubic spring. The model is then applied to additional case
studies of interest, which may have various complicating effects,
such as added external force, parasitic and series capacitance,
residual charges, or large-angle and generally shaped plates tor-
sional actuators.
The pull-in effect emerges from the fact that the electro-
static force is an attraction force, as well as from the mixed
dependence of the electrostatic energy upon the electric and
mechanical parameters, namely, the displacement and charge. It
is shown that the pull-in phenomenon is not unique to voltage-
controlled actuators, as seen so far, and may also be exhibited in
charge-controlled actuators. Moreover, similar mixed behavior
of the energy term may be found in certain magnetostatic
actuators, which also exhibit inherent instability [27], [28].
II. THE GENERALIZED ELECTROSTATIC PULL-IN MODEL
A. Definition of the Composite Actuator System
We consider a battery (the source of the electric energy): two
conducting surfaces with a generalized shape forming a capac-
itor with a variable gap and a mechanical part. The latter exerts
a restoring force on the movable electrode of the capacitor and
hence is modeled by a spring, which may be either linear or
nonlinear. The mechanical part may include additional external
forces such as the gravitational force or an external pressure. A
block diagram of the system is shown in Fig. 1. For zero charge
or zero voltage, the initial gap between the two electrodes of the
capacitor is denoted by .
One of the electrodes is assumed to be fixed while the other
one is free to move along a single degree of freedom, denoted in
the most general case by . The latter may represent a linear dis-
placement, denoted by ; an angular displacement, denoted by
; or a more complicated movement along a specific trajectory
(in a general phase plane of the problem), shown schematically
in Fig. 2. In a more general case, both electrodes may be free to
602 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001
Fig. 1. A block diagram of the electrostatic actuator under study, exhibiting the
battery: the electrostatic part with the generalized capacitor and the mechanical
part composed of a suspending element as well as a possible external force,
where Q denotes the applied charge and  the generalized displacement of the
actuator.
Fig. 2. Electrostatic actuators with a single input: (a) displacement type,
where the degree of freedom x denotes the linear displacement of the movable
electrode; (b) torsion type, where the degree of freedom  denotes the titling
angle of the movable plate; and (c) general type, where the degree of freedom
 denotes a displacement along a general trajectory. C (1) and U (1) denote
the capacitance and mechanical energy as a function of the degree of freedom,
respectively.
move, and the separation between the electrodes is defined as
the degree of freedom .
The capacitor formed by the electrodes is assumed as a
linear electrical element, having the charge , proportional to
the voltage , but may be, in general, a nonlinear function of
the generalized displacement . Moreover, when the applied
electrical parameter across the capacitor is the charge, the
actuator is said to be charge-controlled, while when it is the
voltage it is said to be voltage-controlled.
The process under study is considered as static; therefore,
any dissipative processes, such as damping and heat flow, are
ignored. Thus, the electrical energy of the battery is converted
into mechanical and electrical energy, denoted, respectively, by
and , and stored, respectively, in the mechanical part
and the electrical part. denotes the energy of the battery. It
is assumed to be much larger than and and is therefore
practically constant.
B. The Pull-In Phenomenon
The essence of the pull-in phenomenon [1] is due to the mixed
dependence of the electrostatic force upon the electrical, or
, and mechanical, , parameters. Due to the opposite charge
on the two electrodes of the capacitor, an attracting
electrostatic force is formed between the two electrodes. An
equilibrium position of the actuator may be reached via a me-
chanical restoring force. Typical electrostatic actuators exhibit
two possible equilibrium positions, one stable and a second un-
stable, shown schematically in Fig. 3. Fig. 3 exhibits a typical
total energy of an electrostatic actuator as a function of the dis-
placement , at several charges. The minimum points describe
a stable equilibrium, while the maximum points describe an un-
stable equilibrium state, in accordance with the well-known en-
ergy stability criteria [29]. As we increase the charge, the two
states become closer in the -axis and the absolute energy in-
creases. At some critical value of the applied charge, the two
equilibrium states merge into one critically stable state, char-
acterized by an inflection point of the energy, i.e., the first and
second derivatives of the energy with respect to the mechan-
ical displacement become zero. Once this point is reached, the
system collapses since any perturbation will induce an electro-
static attraction force, which is larger than the mechanical force.
This is known as the pull-in phenomenon [1] and is character-
ized by the pull-in parameters denoted by and . Above
the pull-in critical charge, no equilibrium state exists for the
actuator. Even though the above discussion is presented for a
charge-controlled actuator, it is identical for the voltage-con-
trolled actuator using the coenergy representation [1], while the
electrical pull-in parameter is then .
C. The Charge-Controlled Actuator
Considering first the charge-controlled actuator case, and
using the energy representation, which is more adequate in this
case, the total energy stored in the actuator is given by
(1)
where the second term on the right is the electrical energy
stored in the actuator [1], [37]. It is assumed that the internal
energy of the actuator is additive.
As discussed in Section II-B, at the pull-in state, the first and
second derivatives of the total energy with respect to the gener-
alized displacement become zero, explicitly cor-
responding to equilibrium and corresponding to
critical stability. Substituting (1) simultaneously in both condi-
tions and combining the resulting equations by eliminating ,
the following algebraic governing the charge-controlled pull-in
equation is derived:
(2)
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS 603

Fig. 3. Schematic representation of the stable (local minimum), the unstable (local maximum), and critical (pull-in) states. The vertical axis is the total energy
U = U + U = U +( Q=C2 ) and the horizontal axis is the displacement. (a) Q<Q , (b) Q<Q <Q Q Q
, (c) = , and (d) Q>Q .

Moreover, the charge required for pull-in is calculated III. CASE STUDIES
using In what follows, it is shown that while using the pull-in equa-
tions derived in Section II captures several well-known results, it
does so in a way that makes it easy to evaluate the case of pull-in
(3) when both the electrical and mechanical elements are nonlinear
functions of the generalized displacement. To exhibit this, the
case of a Duffing spring, previously analyzed numerically by
Hung and Senturia [13], is discussed. Following this, the pull-in
D. The Voltage-Controlled Actuator equations are applied to a wide variety of cases of parallel plates
Applying a given voltage across the electrodes of the actuator, and torsion actuators considering external forces, fringing fields,
which is known as the voltage-controlled method, provides the parasitic capacitors, residual charges, large-angular displace-
more established control method for electrostatic actuators. The ment, and generally shaped electrodes. Several of these cases are
adequate representation of a voltage-controlled actuator is in reported in the literature and are mentioned here to illustrate the
the coenergy representation [1], where the total coenergy of the usefulness of the pull-in equations. In other cases, the pull-in
actuator is given by equations are used to extend current results or derive new ones.

(4) A. The Parallel-Plate Actuator

where the first term on the right is the electrical coenergy
stored in the actuator [1], [37].
Following the same arguments of Section II-C, at the pull-in
state the first and second derivatives of the total coenergy with
respect to the generalized displacement become zero, explic-
itly, corresponding to equilibrium and
corresponding to critical stability. Substituting (4) si-
multaneously in both conditions and combining the resulting
equations by eliminating , the following algebraic governing
the voltage-controlled pull-in equation is derived:
The parallel-plate actuator, shown in Fig. 2(a), is used in
, many micromachined sensors and devices, including forced re-
balanced accelerometers [31], [32], rate gyroscopes [33], and
force rebalanced microphones (or pressure sensors) [34], [35].
The parallel-plate actuator is governed by one linear mechan-
ical degree of freedom, which is perpendicular to the plane of
the plates.
1) The Spring Stiffening Case: As previously discussed, the
case of the nonlinear mechanical spring stiffening effect, which
results either from large displacements or from residual stress
in the suspensions, on the pull-in parameters is now analyzed.
The actuator displacements are considered large when they are
on the order of the thickness of the suspensions. The mechanical
(5) energy stored in the suspensions is then given by [1], [36]

(7)
Moreover, the voltage required for pull-in is calculated
using where the first term represents the Hooke’s law term and the
second term, characteristic of a Duffing spring, is due to axial
(6) stresses that evolve in the suspensions [1], [36]. The nonlin-
earity is represented by a constant , which can be shown to
604 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001

Fig. 4. The pull-in parameters of the parallel-plates actuator with nonlinear suspensions. (a) The normalized pull-in displacement  =  = and (b) the
normalized pull-in voltage, namely, V divided by the infinite plate approximation pull-in voltage versus  = ( ) , where  is the nonlinearity factor of
the suspensions.

be inversely proportional to the square of the thickness of the where , i.e., it is proportional to the square of
suspensions [1]. This proportionality factor depends on the spe- the ratio between the suspensions thickness and the maximum
cific configuration of the suspensions and the residual stress. gap. The solution of this equation and the pull-in voltage are
Considering the capacitance of the parallel-plate actuator, the given by
most simplified approximation assumes that the side length of
the plates is much larger than the nominal distance between
them as well as their thickness. Thus the fringing field can be
neglected, and the capacitance is given by [37]

(8)

where (10)
area of the plates;
gap at zero voltage;
dielectric constant of the vacuum. These results, illustrated in Fig. 4, exhibit the variations of the
Substituting (7) and (8) into the voltage-controlled pull-in equa- pull-in parameters in the continuum between a perfectly linear
tion (5) yields spring , where , and a perfectly cubic spring
, where . Thus, the effect of spring-stiffening
(9) can be used to extend the travel range of electrostatic actuators,
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS
as discussed by Hung and Senturia [13]. Moreover, Fig. 4 ex-
hibits that for practical design, should be designed at 0.1,
achieving 95% of the maximal possible travel range, with only
two times higher pull-in voltage. For , the gain in the
travel range is small, while the penalty in the increased pull-in
voltage is significant.
Hung and Senturia [13] have previously estimated the effect
of the spring stiffening. They have considered the distributed
problem of a fixed–fixed beam. They showed that the resulting
stiffness was that of a Duffing spring and estimated analytically
the limit where the cubic term dominates, deriving the same
result as for , presented here. They have also demon-
strated this result with numerical finite-difference simulations
on the fixed–fixed beam distributed problem, showing a limit of
, which is slightly above the analytical limit in the
parallel-plate case. Thus the result derived here at the extreme
case is in agreement with Hung and Senturia result.
2) The Effect of a Constant Applied Force: The influence of
a constant applied force on the pull-in parameters of a parallel-
plate actuator is now discussed. The infinite plates approxima-
tion (8) is used for the capacitance of the actuator. Moreover,
the mechanical suspensions are assumed to be linear [36], for
sake of simplicity, and there is additional constant applied force
acting in the direction of . Thus the mechanical energy is
given by
(11)
where is the elastic coefficient of the suspensions. Substi-
tuting (8) and (11) into (5) yields
and therefore
(12)
where is the normalized coordinate . By substituting
(8), (11), and (12) into (6), the pull-in voltage is
(13)
Equations (12) and (13) correspond to the well-known results
obtained by Osterberg and Senturia [1], [14], using . numerical finite difference simulations. The analytic solution
In this case, this approach results in a cubic equation, which can
be solved analytically. The pull-in point is derived by finding
where the stable and unstable solutions become equal. An ex-
perimental verification of the parallel actuator pull-in parame-
ters can be found, for example, in Chan and Dutton [23]. In this
work, the actuator was shown to travel almost up to the theoret-
ical limit. A travel range of was measured before
the actuator collapsed.
The effect of applied force is of great importance in forced
rebalanced accelerometers or forced rebalanced pressure sen-
sors (microphones). It is seen that the pull-in parameters are
strongly affected by the applied force. The pull-in voltage is re-
duced when the applied force is in the direction of the electro-
static force and is increased otherwise.
605
3) The Effect of Finite Plate Dimensions: A model for the
capacitance of a parallel-plate long strip was given by Love and
can be found in [39]. The complete model does not give an an-
alytical expression for the capacitance but provides a method
for calculating it by solving a nonlinear algebraic equation in-
volving elliptic integrals. Nevertheless, an approximate analyt-
ical expression can be derived for cases where the ratio of the
side length of the plates to the gap between them is larger
than one.
This expression can also be used to approximate the effect
of the fringing field capacitance on the total capacitance of a
square parallel-plate actuator
(14)
The first term in (14) expresses the capacitance of the plates
and the second and third terms express the capacitance of the
fringing field. Substituting (14) into (5), assuming linear sus-
pensions, and slightly manipulating the result yields the fol-
lowing equation:
where is the normalized side length . The physical
solution of this equation is
(15)
The approximation used in (15) is for , with less than 1%
deviation from the accurate solution.
This result is within 10% deviation from the result of (12),
which neglects the fringing field capacitance. Fig. 5 exhibits
the normalized [i.e., for the finite plates dimensions
case divided by (13)], showing a deviation within 15% from
(13). To conclude, when we consider the finite dimensions of the
plate, the dynamic range is increased while the pull-in voltage
decreases.
The case of finite dimensions was considered by Osterberg
et al. [16] and Chan et al. [21], for the distributed case of a
fixed–fixed beam. However, the pull-in point was obtained by
[(15)] presented here clearly exhibits the effect of the finite plate
dimensions.
4) The Effect of Residual Charge: Residual charge can accu-
mulate in electrostatic actuators containing electrical insulators
in contact with the conductors. Fig. 6 exhibits a parallel-plate
actuator with a dielectric layer covering the fixed electrode. A
charge is assumed at the interface of the dielectric layer and
air. The total energy of the actuator is given by applying Gauss
law [37]
(16)
where is given in (8) and is
the dielectric layer capacitance. Equation (16) indicates that as
606
Fig. 5. The pull-in parameters of the parallel-plates actuator with finite dimension plates. (a) The normalized pull-in displacement 
normalized pull-in voltage, namely, V divided by the infinite plate approximation pull-in voltage versus the normalized side length !
the side length of the plates.
for the charge source, the capacitors are connected in parallel,
and for the voltage source they are connected in series. Using
the methodology presented here, we obtain
Thus, the residual charge affects the pull-in voltage signifi-
cantly. The series capacitor introduced by the dielectric layer
increases the , while the residual charges, in this case, shift
it to lower values. However, it does not affect the normalized
pull-in coordinate. The latter is only affected by the series
capacitor, provided by the dielectric layer, as discussed below.
This significant effect of residual charges on actuation char-
acteristic was measured by Chan and Dutton [21]. Chan and
Dutton measured the charge buildup in an isolating silicon ni-
tride layer due to fixed–fixed beams, which came into contact
with the nitride. Furthermore, they have measured the magni-
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001
=  = and (b) the
= W= , where W is
(17)
Fig. 6. A parallel-plate actuator with a charged dielectric layer coating the
fixed electrode. A charge Q is assumed at the interface of the dielectric layer
and air. The dielectric layer is assumed having a dielectric constant " and
thickness of x .
tudes of the s in quick succession, i.e., less than 1 min be-
tween measurements, showing progressively lowered s. This
indicates that the charge, which accumulates in the nitride with
each measurement, reduces the pull-in voltage, as predicted by
(17). This effect must be taken into consideration when de-
signing any device that uses pull-in during its operation.
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS
5) The Effect of a Capacitor Feedback: The effect of a ca-
pacitor connected in series with the actuator is obtained from
(17), assuming . Several interesting features can be
concluded.
1) For , i.e., smaller than half the nominal
capacitance of the plates, the pull-in effect is eliminated.
2) For , the pull-in point is in the interval
depending on .
3) When , i.e., and , voltage-controlled case. The generality of this result is
the pull-in parameters of the parallel-plate actuator are
unaffected by the series capacitor.
Thus, by only using a constant capacitor in series with the actu-
ator, the travel range of the actuator can be extended, as reported
by Seeger and Crary [38]. Seeger and Crary explained that the
capacitor in series with the actuator extends the effective elec-
trical gap seen by the source. The movable electrode motion is
now limited by a third of the new effective gap, which may be
larger than the original gap. More recently, Chan and Dutton
realized a parallel actuator design employing a capacitive feed-
back. They have shown that an actuator with a series capacitor
has an extended travel range of , which is almost twice
the conventional range. As in the above model, their simulations
have shown that for perfectly symmetric suspending beams, the
actuator can be made to travel the entire gap with . Nev-
ertheless, due to fabrication issues, asymmetry is introduced in
the geometrical dimensions of the suspending beams, resulting
in pull-in after deflecting close to 60% of the initial gap.
The effect of capacitor feedback can be more generally ex-
plained by considering that the series capacitor shifts the actuator
from the voltage-controlled to the charge-controlled regime.
This is further shown and discussed in Section III-B2 and [30].
Moreover, it can be easily shown that a parasitic capacitor,
which is in parallel with , does not affect the pull-in pa-
rameters. In this case, and the deriva-
tives of are unaffected. However, in the presence of a
series capacitor, a parallel parasitic capacitor can influence the
pull-in parameters. These practical design issues are considered
here from a unified approach.
6) Charge Controlling with a Parasitic Parallel Capac-
itor: We now consider the same parallel-plate actuator with
charge-controlled operation rather than voltage-controlled op-
eration. We consider the thin and infinite plates approximation
and that there is a constant parasitic capacitance connected in
parallel with the actuator plates, . The total capacitance of
the actuator is
(18)
Substituting (18) into (2), assuming linear suspensions, yields
the following solution:
(19)
where is the normalized parasitic ca-
pacitance.
As in Section III-A5 we can conclude the following from
(19).
1) For , i.e., is practically disconnected, the
charge-controlled parallel-plate actuator does not ex-
607
hibits the pull-in phenomenon. Moreover, no pull-in
occurs as long as , i.e., smaller than half the
nominal capacitance of the plates.
2) For , the charge-controlled parallel-plate actuator
also exhibits a pull-in point.
3) The pull-in point of the charge-controlled case is in the
interval , i.e., the travel range of the
charge-controlled parallel-plate actuator is larger than the
discussed elsewhere [30].
The pull-in charge for cases where pull-in occurs is given by
(20)
B. The Torsion Actuator
The torsion actuator, shown in Fig. 2(b), is the building
block of many micromachined devices, usually for optical
uses. Several examples include optical cross-connects [5],
digital light processors [6]–[9], and microscanning confocal
microscope [40]. The torsion actuator is governed by one
angular mechanical degree of freedom about an axis lying in
the plane of the plates.
Several ways were employed for estimating the pull-in pa-
rameters for the torsion actuator. The simplest method is using
the parallel-plate model along with an effective spring coeffi-
cient of the structure for estimating the pull-in voltage [14]. This
method yields rapid calculation but shows errors of up to 20%
compared to finite-element method (FEM) results. Hornbeck
[8], [9] numerically solved the tilt angle, for a given voltage,
where the electrostatic and elastic torque are in equilibrium.
The pull-in voltage is found by increasing the voltage until the
pull-In condition is met. Hornbeck’s method and the FEM ap-
proach require elaborate calculations but are quite accurate.
Pull-in analysis for a torsion actuator with planar square
plates was presented recently by the authors [22]. The tech-
nique used was similar to the roots technique described in
Section III-A1, through the derivation of an th order poly-
nomial approximate equation, which represented the pull-in
displacement. In the following section, a more general analysis
of a torsion actuator is presented, using the pull-in equation.
The analysis starts with the simple planar full-square-plates
torsion actuator shown in Fig. 7(a). Using the pull-in equation,
a nonlinear algebraic equation is derived for the pull-in angle,
replacing the approximate th order equation from [22]. The
equation is derived for the charge-controlled case as well as for
the voltage-controlled case. Following this, the effect of large
angles is analyzed. Then the case of nonplanar plates [41],
[42], which is shown in Fig. 7(b), is considered. Finally, a more
general algebraic equation for the pull-in angle in an integral
form, which is suited for more general shape of the plates [i.e.,
triangular, circular, etc.; see Fig. 7(c)] and location, is derived.
1) A Torsion Actuator with Planar Square Shaped
Plates: We start the analysis with a torsion actuator, where the
configuration and the required parameters for the analysis are
shown in Fig. 7(a). The actuator has two planar square plates,
which extrude a distance in the direction. The total length
of the plates is , and the gap between them at the axis of
608 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001

Fig. 7. Configurations of different torsion actuators. (a) Planar full-square-plate actuator, where a and b are the length and width of the plates, d is the distance
between the plates at the axis of rotation, and  is the tilting angle. (b) Nonplanar square-plate actuator, where  is the angle between the plates at zero voltage.
(c) Generally shaped planar plates actuator, where a is the distance between the axis of rotation to the nearest edge of the plate, a is the distance to the end of
the plate, a is the distance to the end of the proof-mass, and P (x) is the shape function of the plates.

rotation is . Assuming that the plates are large and using the The above derivation is based on the solution of a Laplace equa-
field calculations from [22], the capacitance of the plates at a tion between two semi-infinite tilted plates. Furthermore, in the
tilting angle is given by derivation of the above relation, small angle approximation was
used, i.e., and . We define
and , and thus the small angle ap-
proximation practically means . The capacitance func-
tion becomes

(21) (22)
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS
Using the voltage-controlled pull-in equation (5) and assuming
linear suspensions, i.e., , results in the following
nonlinear algebraic equation:
(23)
Solving this equation by any numerical solver directly yields
that the normalized pull-in angle is given by
(24a)
Using (6) yields the pull-in voltage
(24b)
Substituting (22) in the charge-controlled pull-in equation [(2)],
assuming againlinear suspensions, yields the following equation:
(25)
Solving this equation and substituting into (3) yields the pull-in
angle and charge
(26)
Therefore, assuming small angles, is independent from
similarly to the case of the parallel-plate actuator. Never-
theless, unlike the parallel-plate actuator, the torsion actuator
exhibits pull-in even for charge-controlled operation.
The pull-in parameters of a full plate electrostatic actuator
were experimentally verified in several works. For example,
Buhler et al. [6] have measured a , which
is in good agreement with the above result.
2) The Effect of a Capacitor Feedback: The case of a ca-
pacitor connected in series with the torsion actuator is now
considered in order to examine its effect on the travel range
of the actuator. Assuming that and using
(22), the total actuator capacitance is given by
(27)
Using the voltage-controlled pull-in equation [(5)] and once more
assuming linear suspensions, the following algebraic equation is
derived:
(28)
Fig. 8 exhibits the pull-in parameters derived from the solution
of (28). Fig. 8(a) exhibits the pull-in angle variations as a func-
tion of . For high , i.e., practically short-circuit impedance,
the result of the voltage-controlled case from the (24a) case is
reconstructed. However, for low , the travel range does not in-
crease to the full range, as for the parallel-plate actuator, and
the charge-controlled case from (26) sets the limit. Thus, as the
capacitance of the series capacitor is reduced, the actuator con-
trol is shifted from the voltage-control toward the charge-control
609
regime. A generalization of these results to any electrostatic ac-
tuator is discussed in [30]. Fig. 8(b) exhibits the pull-in voltage,
normalized by (24b), as a function of . The pull-in voltage in-
creases very rapidly as is lowered. Nevertheless, for practical
designs, it is sufficient to use , achieving 95% of the
maximal possible travel range (the charge-controlled case), with
ten times higher pull-in voltage.
3) The Effect of Large Angles: When considering the effect
of large angles, two corrections in the analysis of Section III-B1
should be taken into account. First, in the derivation of , the
full trigonometric relations should be used instead of the small
angle approximation. Second, the suspensions can no longer as-
sumed to be linear. It was shown experimentally in [41] that the
approximation used in Section III-A3 is also valid in this case.
Assuming now that the bottom plate is much larger than the
upper plate (as in Toshiyoshi and Fujita’s [43] configuration),
the capacitance and mechanical energy can be approximated by
(29a)
(29b)
We redefine here .
Substituting (29) into the voltage-controlled pull-in equation
[see (5)], we derive the following:
(30)
Fig. 9(a) exhibits the dependence of the normalized pull-in
angle upon , where , as defined in Section III-A3,
for several . As for the parallel-plate actuator, the nor-
malized pull-in angle is highly dependent upon but is only
slightly changed with . Moreover, Fig. 9(b) clearly ex-
hibits that the pull-In voltage, normalized by the small angle
approximation expression from (24b), is almost independent of
. It is shown here, as for the parallel-plate actuator (see
Section III-A3), that for practical design should be 0.1.
Toshiyoshi and Fujita have fabricated and tested a full plate
torsion actuator with a large actuation angle of 33 [43].
They have measured an increase in the normalized pull-in
angle . This increase is expected and easily
explained using the model discussed above, and its results are
shown in Fig. 9(a).
Nevertheless, it should be noticed that the above approxi-
mation becomes less accurate for very large , since the
fringing field capacitance from the backside of the moving plate
is neglected in the above derivations.
4) Nonplanar Square-Plates Actuator: To achieve large
angle actuation with relatively low operating voltage, a torsion
actuator with nonplanar plates was suggested in [41] and
[42]. The actuator is shown in Fig. 7(b) with all the required
610 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001

Fig. 8. The pull-in parameters of a full-square-plate torsion actuator with a capacitor feedback versus  , the ratio between the series capacitor and the nominal
2 =
capacitance of the actuators. (a) The normalized pull-in angle  = and (b) the pull-in voltage normalized by the small angles approximation pull-in
voltage.

parameters for the analysis. Similarly to Section III-B1, we
derive that the capacitance of the structure is given by
(31)
As in Section III-B1, the plates extrude a distance in the
direction. After carefully observing (31), it can be easily con-
cluded that the results of the parallel-plate actuator are satisfied
also here, due to the similarity of the capacitance function. Thus,
for linear suspension, and
rameters of torsion electrostatic actuators is now derived. This
analysis shows how the pull-in expression exhibits the complete
variation of pull-in with the geometry of the plates. In the deriva-
tions, small angles and negligible fringing field capacitance are
assumed. Fig. 7(c) presents a torsion actuator with the general
symmetric plates’ shape represented by a function , where
is an axis crossing in the middle of the plate and perpendicular
to the axis of rotation . Using again the results from [22], we
derive that the capacitance is given by

(32)

and for nonlinear suspensions, (10) can be used. This results in

a lower pull-in voltage than in the planar plates case.
5) Generally Shaped Planar Plates: Following the deriva-
(33)
tions in Section III-B1, a more general model for the pull-in pa-
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS 611

=(
Fig. 9. The pull-in parameters of a full-square-plate torsion actuator with a large angle approximation versus   ) for several  , where  is the
2 =
nonlinearity factor of the suspensions. (a) The normalized pull-in angle  = and (b) the pull-in voltage normalized by the small angle approximation
pull-in voltage.

where and , and and are defined where we define

in Fig. 7(c). Using the small angle relations ,
, and , where , , , , and are (36)
defined in Fig. 7(c), we derive

and by substituting into (6), we get the pull-in voltage

(34)

(37)
where , , , ,
, and , i.e.,
the normalized plates’ shape [for example, for rectan- where
gular plate for triangular plates and
for circular plates].
Assuming linear suspensions and using (34) and (5), we easily (38)
derive the voltage-controlled pull-in equation for this case
Table I presents and for several different plates’
(35) shape. Fig. 10 presents the equipotential lines in the plane
612 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001

TABLE I
( ) ( )
THE PULL-IN VOLTAGE INTEGRAL f w AND PULL-IN ANGLE INTEGRAL f w FOR THE TORSION ACTUATOR WITH GENERAL PLATES FOR SEVERAL SHAPES,
( ) ( )
WHICH ARE INDICATED IN THE TABLE. Ci w AND Si w ARE THE COSINE INTEGRAL AND SINE INTEGRAL FUNCTIONS, RESPECTIVELY

(a)

(b)
Fig. 10. Equipotential lines of the normalized pull-in angle 2 =  = in the (; ) plane for a torsion actuator with (a) rectangular and (b) triangular
plates, where  = =
a =a and a =a [see Fig. 7(c)].
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS 613

(c)

(d)
2 =  =
Fig. 10. (Continued.) Equipotential lines of the normalized pull-in angle in the (;
) plane for a torsion actuator with (c) reverse triangular
=
and (d) circular plates, where  =
a =a and
a =a [see Fig. 6(c)].
of the normalized pull-in angle for the cases appearing in
Table I. Fig. 10 shows that there is a region where no pull-in
occurs for these actuators and a full swing actuator can be real-
ized. Thus, by changing the bottom electrode location geometry
parameter, the pull-in parameters can be controlled and the phe-
nomenon can be eliminated. Degani et al. examined this issue
in [22], where two torsion actuators were fabricated and mea-
sured. The first actuator location parameters were , due to the optical beams’ angular symmetries.
and have exhibited a measured of 0.385 and
of 11.5 [V], while the calculated parameters are
and [V]. The second actuator location parameters
were , , and have exhibited a measured
of 0.394 and of 19.5 [V], while the calculated parameters
are and of 19.36 [V]. Thus, good correlation
is achieved between the model and experimental results.
It should be noticed that the case studies discussed in Table I
and Fig. 10 are merely typical configurations used in the litera-
ture, and almost any general shape may be modeled as well. Al-
though the rectangular torsion actuator is preferred when com-
paring the various shapes for a given due to its lower opera-
tion voltage, this is not generally the case. As an example, one
may consider a given pixel area. In this case, a higher and
thus higher electrical torque is achieved using the reverse trian-
gular shape. This shape was considered by Hornbeck in Texas
Instruments’ DMD [8], [9]. A reverse triangular shape actuator
with , was fabricated and characterized showing
a , while the model above and Hornbeck calcula-
tions both predict a value of . Other shapes may
be considered due to various requirements. For example, when
considering optical switching, a circular element can be chosen
IV. SUMMARY
In this paper, a model for the pull-in parameters of a general
electrostatic actuator with a single input is derived. Two types
of inputs are presented, voltage or charge, and the difference
between the two types of controlled operations is defined. The
analysis results in two general algebraic equations (one for each
input type), referred to as the pull-in equations, which directly
yield the pull-in parameters. The equations are general and ap-
plicable for any electrostatic actuator, regardless of its shape or
the number of mechanical degrees of freedom, as long as it has
only a single input.
A number of case studies of electrostatic actuators are pre-
sented, exhibiting the straightforward calculations of the pull-in
614 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 10, NO. 4, DECEMBER 2001
parameters using the pull-in equation, even in structures, which
are more complicated than the infinitely large parallel-plate ac-
tuator. Examining the pull-in equation, it is clearly seen that only
two functions are needed in order to calculate the pull-in point:
the mechanical force and the capacitance. These functions can
be expressed analytically or can be calculated numerically. The
case studies presented here are mainly based on analytical solu-
tions, but the methodology can be extended in order to allow fast
and direct calculations of the pull-in parameters in cases where
either the mechanical force or capacitance are calculated numer-
ically [30]. The conclusions of this paper regarding the charge
versus voltage control performance and the capacitive feedback
properties are generalized to any electrostatic actuator and fur-
ther discussed in [30].
ACKNOWLEDGMENT
The illuminating and valuable discussions with Prof. S. D.
Senturia are gratefully acknowledged.
REFERENCES
[1] S. D. Senturia, Microsystems Design. Boston, MA: Kluwer Academic,
2000.
[2] M. Tabib-Azar, Microactuators. Boston, MA: Kluwer Academic,
1998.
[3] C. Marxer, C. Thio, M. A. Gretillat, N. F. de Rooij, R. Battig, O. An-
thamatten, B. Valk, and P. Vogel, “Vertical mirrors fabricated by deep
reactive ion etching for fiber-optic switching applications,” J. Micro-
electromech. Syst., vol. 6, pp. 277–285, Sept. 1997.
[4] W. H. Juan and S. W. Pang, “High-aspect-ratio Si vertical micromirror
arrays for optical switching,” J. Microelectromech. Syst., vol. 7, pp.
207–213, June 1998.
[5] H. Toshiyoshi and H. Fujita, “Electrostatic micro torsion mirrors for an
optical switch matrix,” J. Microelectromech. Syst., vol. 5, pp. 231–237,
Dec. 1996.
[6] J. Buhler, J. Funk, J. G. Korvink, F. P. Steiner, P. M. Sarro, and H. Baltes,
“Electrostatic aluminum micromirrors using double-pass metallization,”
J. Microelectromech. Syst., vol. 6, pp. 126–135, June 1997.
[7] S. W. Chung, J. W. Shin, Y. K. Kim, and B. S. Han, “Design and fab-
rication of micromirror supported by electroplated nickel posts,” Sens.
Actuators A, Phys., vol. 54, pp. 464–467, 1996.
[8] L. J. Hornbeck, “Deformable-mirror spatial light modulators,” SPIE
Critical Review Series, vol. 1150, pp. 86–102, 1989.
[9] , “Spatial light modulator and method,” U.S. Patent 5 061 049, Oct.
1991.
[10] P. M. Zavracky, S. Majumber, and E. McGruer, “Micromechanical
switches fabricated using nickel surface micromachining,” J. Micro-
electromech. Syst., vol. 6, pp. 3–9, Mar. 1997.
[11] C. T. C. Nguyen, L. P. B. Katehi, and G. M. Rebeiz, “Micromachined
devices for wireless communications,” Proc. IEEE, vol. 86, pp.
1756–1768, Aug. 1998.
[12] Q. Zou, Z. Tan, Z. Wang, J. Pang, X. Qain, Q. Zhang, R. Lin, S. Yi, H.
Gong, L. Liu, and Z. Li, “A novel integrated silicon capacitive micro-
phone—Floating electrode ‘electret’ microphone (FEEM),” J. Micro-
electromech. Syst., vol. 7, pp. 224–234, June 1998.
[13] E. S. Hung and S. D. Senturia, “Extending the travel range of
analog-tuned electrostatic actuators,” J. Microelectromech. Syst., vol.
8, pp. 497–505, Dec. 1999.
[14] P. M. Osterberg, “Electrostatically actuated microelectromechanical test
structures for material property measurements,” Ph.D. dissertation, Sept.
1995.
[15] R. K. Gupta, “Electrostatic pull-in test structure design for in-situ
mechanical property measurements of microelectromechanical systems
(MEMS),” Ph.D. dissertation, June 1997.
[16] P. M. Osterberg and S. D. Senturia, “M-TEST: A test chip for MEMS
material property measurement using electrostatically actuated test
structures,” J. Microelectromech. Syst., vol. 6, pp. 107–118, June 1997.
[17] P. Osterberg, H. Yie, X. Cai, J. White, and S. Senturia, “Self-consistent
simulation and modeling of electrostatically deformed diaphragms,” in
Proc. IEEE MEMS ’94, Oiso, Jan. 1994, pp. 28–32.
[18] M. Fischer, M. Giousouf, J. Schaepperle, D. Eichner, M. Weinmann, W.
von Munch, and F. Assmus, “Electrostatically deflectable polysilicon
micromirrors—Dynamic behavior and comparison with results from
FEM modeling with ANSYS,” Sens. Actuators A, Phys., vol. 67, pp.
89–95, 1998.
[19] R. K. Gupta, “Analysis of Analog Devices, Inc., square pixel torsion el-
ement using Rayleigh–Ritz energy method,” MIT Internal Memo., Sept.
1994.
[20] R. Legtenberg, J. Gilbert, S. D. Senturia, and M. Elwenspoek, “Electro-
static curved electrode actuators,” J. Microelectromech. Syst., vol. 6, pp.
257–265, Sept. 1997.
[21] E. K. Chan, K. Garikipati, and R. W. Dutton, “Characterization of
contact electromechanics through capacitance-voltage measurements
and simulations,” J. Microelectromech. Syst., vol. 8, pp. 208–17, June
1999.
[22] O. Degani, E. Socher, A. Lipson, T. Leitner, D. J. Seter, S. Kaldor, and
Y. Nemirovsky, “Pull-in study of an electrostatic torsion microactuator,”
J. Microelectromech. Syst., vol. 7, pp. 373–379, Dec. 1998.
[23] E. K. Chan and R. W. Dutton, “Electrostatic micromechanical actuator
with extended range of travel,” J. Microelectromech. Syst., vol. 9, pp.
321–328, Sept. 2000.
[24] X. T. Wu, R. A. Brown, S. Mathews, and K. R. Farmer, “Extending the
travel range of electrostatic micro-mirrors using insulator coated elec-
trodes,” in Proc. IEEE/LEOS Int. Conf. Opt. MEMS 2000, Kauai, HI,
Aug. 21–24,, 2000, pp. 151–152.
[25] C. Marxer, C. Thio, M. A. Gretillat, N. F. de Rooij, R. Battig, O. An-
thamatten, B. Valk, and P. Vogel, “Vertical mirrors fabricated by deep
reactive ion etching for fiber-optic switching applications,” J. Micro-
electromech. Syst., vol. 6, pp. 277–285, Sept. 1997.
[26] W. H. Juan and S. W. Pang, “High-aspect-ratio Si vertical micromirror
arrays for optical switching,” J. Microelectromech. Syst., vol. 7, pp.
207–213, June 1998.
[27] M. Capanu, J. G. Boyd, and P. J. Heketh, “Design, fabrication and
testing of bistable electromagnetically actuated microvalve,” J. Micro-
electromech. Syst., vol. 9, pp. 181–189, June 2000.
[28] H. Maekoba, P. Helin, G. Reyne, T. Bourouina, and H. Fujita, “Mod-
eling and optimization of bi-stable optical switch,” in Proc. 3rd Conf.
Modeling and Simulation of Microsystems (MSM’2000), San Diego, CA,
Mar. 27–29, 2000, pp. 166–169.
[29] H. B. Callen, Thermodynamics. London, U.K.: Wiley, 1960.
[30] O. Bochobza-Degani, “Investigation of motion sensing and actuation
mechanisms in micro-electro-opto-mechanical systems (MEOMS) and
their applications,” Ph.D. dissertation, Technion—Israel Institute of
Technology.
[31] D. O. King, M. C. L. Ward, K. M. Brunson, and D. J. Hamilton, “Polysil-
icon process development for fully integrated surface-micromachined
accelerometer with CMOS electronics,” Sens. Actuators A, Phys., vol.
68, pp. 238–243, 1998.
[32] B. E. Boser and R. T. Howe, “Surface micromachined accelerom-
eters,” J. Solid-State Circuits, vol. 31, no. 3, pp. 366–375, Mar.
1996.
[33] O. Degani, D. J. Seter, E. Socher, S. Kaldor, and Y. Nemirovsky, “Op-
timal design and noise consideration of micromachined vibrating rate
gyroscope with modulated integrative differential optical sensing,” J.
Microelectromech. Syst., vol. 7, pp. 329–338, Sept. 1998.
[34] B. P. Gogoi and C. H. Mastrangelo, “A low-voltage force rebalanced
pressure sensor with hermetically sealed servo-mechanism,” in Proc.
MEMS’99, Orlando, FL, Jan. 17–21, 1999.
[35] J. W. Kang and K. Simonette, “Surface micromachined multi-layer
moving gate field effect transistor (MOGFET) pressure switch with
integrated vacuum sealed cavity,” in Proc. MEMS’99, Orlando, FL, Jan.
17–21, 1999, pp. 499–504.
[36] S. P. Timoshenko and I. N. Goodier, Theory of Elasticity. New York:
McGraw-Hill, 1970.
[37] R. S. Elliot, Electromagnetics. New York: IEEE Press, 1993.
[38] J. I. Seeger and S. B. Crary, “Stabilization of electrostatically actuated
mechanical devices,” in Proc. Transducers ’97, Chicago, IL, June
16–19, 1997, pp. 1133–1136.
[39] H. B. Palmer, “The capacitance of a parallel-plate capacitor by the
Schwartz–Christoffel transformation,” Trans. AIEE, vol. 56, pp.
363–366, Mar. 1937.
[40] D. L. Dickensheets and R. G. Kino, “Silicon-micromachined scanning
confocal optical microscope,” J. Microelectromech. Syst., vol. 7, pp.
38–47, Mar. 1998.
[41] C. Henri and L. Franck, “Fabrication, simulation and experiment of a ro-
tating electrostatic silicon mirror with large angular deflection,” in Proc.
MEMS’00, Miyazaki, Japan, Jan. 23–27, 2000, pp. 645–650.
NEMIROVSKY AND BOCHOBZA-DEGANI: PULL-IN PARAMETERS OF ELECTROSTATIC ACTUATORS 615

[42] B. Wagner, K. Reimer, A. Maciossek, and U. Hofmann, “Infrared mi- Ofir Bochobza-Degani (S’97) was born in
cromirror array with large pixel size and large deflection angle,” in Proc. Ashkelon, Israel, on April 17, 1974. He received
Transducers ’97, Chicago, IL, June 16–19, 1997, pp. 75–78. the B.Sc. degree in electrical engineering and the
[43] H. Toshiyoshi and H. Fujita, “An electrostatically operated torsion B.A. degree in physics (both summa cum laude) in
mirror for optical switching device,” in Proc. Transducers ’95—Eu- 1996 and the M.Sc. degree in electrical engineering
rosensors IX, Stockholm, Sweden, June 25–29, 1998, pp. 297–300. in 1999, all from the Technion—Israel Institute of
Technology, Haifa, where he is currently pursuing
the Ph.D. degree in electrical engineering.
He investigates motion sensing and actuation
Yael Nemirovsky (SM’84–F’99) received the degree mechanisms in microoptoelectromechanical systems
in chemistry and received the D.Sc. degree in electro- (MOEMS). His research focuses on coupled energy
chemistry from Technion—Israel Institute of Tech- domain modeling as well as noise modeling of MOEMS, including electrostatic
nology, Haifa. actuation, magnetostatic actuation, and optical motion sensing. He is involved
She joined the Department of Electrical Engi- in the development of micromachined inertial (acceleration and rate) sensors
neering, Technion—Israel Institute of Technology, employing integrated optical sensing. His other fields of interest are analog
in 1980. Prior to that, she was a Research Scientist readout and control interfaces, silicon optical benches, and thermal sensors.
specializing in microelectronics with Rafael, a
national R&D organization. For more than 20 years,
she has been active in electrooptical devices in II–VI
compound semiconductors and additional advanced
semiconductor materials as well as infrared focal plane arrays. She has been
involved in growth, processing, device design, and modeling of detectors as
well as VLSI circuits. She has a well-equipped MOCVD laboratory for growth
of heterostructures and extensive facilities for device and interfaces processing
and characterization. She has been a principal investigator in large funded
research programs that ended in prototype infrared detectors and systems
that were transferred to industry. Twice she was Head of the department’s
Microelectronics Research Center. Currently, her research focuses on microop-
toelectromechanical systems (MOEMS), CMOS compatible micromachining,
and microsystems implemented in CMOS technology and integrated with
silicon devices. She has published more than 130 papers in the open literature
and has filed several patents and a large number of classified reports. She has
collaborated with the microelectronics industry as a Consultant in sensors
and VLSI technology and has been quite active in national and international
conferences. She has supervised more than 40 graduate students for the M.Sc
and D.Sc. degrees.
Prof. Nemirovsky is a Fellow of the Institute of Electrical Engineers, U.K.
She was Chairperson of the Israeli Association for Crystal Growth. Currently,
she is Chairperson of the Microelectronics and Photonics Section of URSI.
She received the “Best Teacher” award and an award for “Novel Applied Re-
search” from Technion and The Award for the Security of Israel. She received
The Kidron Foundation award for Innovative Applied Research. She is a Dis-
tinguished Lecturer of the IEEE Electron Device Society.