Published by AMSS Press, Wuhan, China. DOI: 10.

1007/s10338-004-0413-z
Acta Mechanica Solida Sinica, Vol. 17, No. 2, June, 2004 ISSN 0894-9166
STATICSTUDYOF CANTILEVERBEAMSTICTION
UNDERELECTROSTATICFORCEINFLUENCE
Zhang Yin
(Department of Mechanical Engineering, University of Connecticut, Storrs, CT 06268-3139, USA)
Zhao Ya-pu
(State Key Laboratory of Nonlinear Mechanics, Institute of Mechanics,
Chinese Academy of Science, Beijing 100080, China)
Received 9 September 2003; revision received 16 April 2004.
ABSTRACT The model and analysis of the cantilever beam adhesion problem under the action
of electrostatic force are given. Owing to the nonlinearity of electrostatic force, the analytical
solution for this kind of problem is not available. In this paper, a systematic method of generating
polynomials which are the exact beamsolutions of the loads with dierent distributions is provided.
The polynomials are used to approximate the beam displacement due to electrostatic force. The
equilibrium equation oers an answer to how the beam deforms but no information about the
unstuck length. The derivative of the functional with respect to the unstuck length oers such
information. But to compute the functional it is necessary to know the beam deformation. So
the problem is iteratively solved until the results are converged. Galerkin and Newton-Raphson
methods are used to solve this nonlinear problem. The eects of dielectric layer thickness and
electrostatic voltage on the cantilever beamstiction are studied. The method provided in this paper
exhibits good convergence. For the adhesion problem of cantilever beam without electrostatic
voltage, the analytical solution is available and is also exactly matched by the computational
results given by the method presented in this paper.
KEY WORDS micro-cantilever beam, adhesion/stiction, electrostatic force, Galerkin method,
constraint, stationary, peel number
I. INTRODUCTION
For capacitor-like microelectromechanical systems (MEMS) structure
[16]
, the voltage between
the structure and substrate causes attractive force. The sources of the voltage can be an articially
mounted device
[2, 79]
or the temporary charge during processing
[10]
or operation
[11, 12]
. At the same
time, stiction, the adhesion of contacting surfaces due to surface force, is one of the most important
reliability problems
[5, 13]
and major failure mechanisms
[13, 14]
of MEMS. Because of the large surface-to-
volume ratio in MEMS devices, adhesion forces become relatively large, much larger than a normally
applied force
[15]
. This fact also causes a major concern about measurement, reduction and control
of friction in the development of micromachines because dry sliding friction occurs in most MEMS
bearings and joints
[16]
. Adhesion can be caused by capillary forces, hydrogen bridging, electrostatic,
van der Waal forces
[14, 17]
, Casimir eect
[18]
, inertial forces
[14]
or by bringing the structure down to the
substrate surface by a probe needle
[5]
or nanoindenter
[19]
. Mastrangelo and Hsu used the energy method
to pioneer the study of the adhesion problem of dierent structures
[20, 21]
. Their model considers the
Vol. 17, No. 2 Zhang Yin et al.: Static Study of Cantilever Beam Stiction 105
nonlinearity due to the large deection of the structure. Jones
[19]
et al. provided a fracture mechanics
approach to the structure adhesion problem. In those researches, adhesion energy/surface force is the
sole source causing structure stiction. The readers are referred to the review by Zhao et al.
[14]
for
further information related to this topic. Yee et al.
[22]
oered a sticking model with the eect of residual
stress gradient and post-release temperature taken into account. Yang
[6]
studied the contact problem
of membrane structure under the action of an electrostatic force. Yang used boundary conditions of the
membrane contact zone to compute the membrane contact. But the adhesion energy/surface force term
does not appear in his governing equations. Chen and Yu
[23]
use the Hankel transform method to solve
the capillary adhesive contact of a punch and piezoelectric half space to show the electric eld inuence
on the adhesion contact. In Chen and Yus paper, the contact piezoelectric substrate is modelled as half
space, thus the emission conditions at innity can be used to assume the solution form of the problem.
And the half space model is also designed to suit their Hankel transform integration domain (from zero
to innity).
In this paper, Mastrangelo and Hsus energy method
[20, 21]
is used to compute the cantilever beam
stiction. Asystematic method of generating polynomials is designed to approximate the beamdeection.
These polynomials are used to compute the beam deection shape in an arbitrarily assumed unstuck
zone. Once the deection shape is found, the beam elastic energy can be calculated. The stiction
equilibrium is found by nding the stationary point of total energy. But arbitrarily assuming beam
unstuck zone may cause the nonexistence of the total energy stationary point. So the basic strategy is to
keep changing the beam unstuck zone and substitute it into governing equations again to compute the
beam deection (potential energy). This iteration continues until the total energy stationary point is
found. Numerically it is that the absolute value of the total energy derivative is less than some tolerance
number.
II. MODEL DEVELOPMENT
In Fig.1(a), a cantilever beam of length L is separated from a substrate by the gap distance g. The
substrate is coatedwith a thin dielectric layer of thickness d
e
to avoidelectric breakdown. Figures 1(b) and
(c) show the arch-shaped and S-shaped stictions. For arch-shaped stiction, the contact area is extremely
small, therefore the measurement largely depends on the contact local properties and statistical error
is relatively much bigger
[5]
. de Boer et al.
[24]
hold that a better way of studying meaningful surface
interaction energy is to investigate S-shaped case only. In this paper, only S-shaped stiction under the
action of electrostatic force case is studied.
Fig. 1. Schematics of beam-dielectric layer-substrate structure and arch-shaped and S-shaped stictions.
2.1. Energy and Governing Equation
In Fig.1(c), the detachment work W
S
is
[14]
W
S
= b(L S)f

W
a
(1)
b is the beam width, S the beam unstuck length, f

the dimensionless roughness function, W

2
w
x
2
_
2
dx (2)
106 ACTA MECHANICA SOLIDA SINICA 2004
EI is the beam bending stiness, E Youngs modulus and I = bt
3
/12 (t the beam thickness) for
rectangular cross-section beam. w(x, S) is the beam displacement.
The beam stretching energy U
S
is
[25]
U
S
=
AE
2S
_
1
2
_
S
0
_
w
x
_
2
dx
_
2
(3)
A is cross-section area, A = bt. And the work done by electrical force f
e
is
W
E
=
_
S
0
_
w
0
f
e
dwdx (4)
Here f
e
is electrical force per unit length and f
e
has the following expression for the substrate coated
with a layer of dielectric material
[3, 6]
.
f
e
=
b
o
2
_
V
o

o
d
e
+(g w)
_
2
(5)

o
is the vacuumpermittivity, the permittivity of the coated dielectric layer material. d
e
is the dielectric
material layer thickness, V
o
the voltage between the substrate and the beam. g is the gap distance between
the beam and dielectric layer. Thus the total energy is
= U
B
+U
S
W
S
W
E
(6)
By using the principle of virtual work (PVW: = 0), the governing equation for the beam in
0 x S domain is derived as
EI

4
w
x
4
=
_
AE
2S
_
S
0
_
w
x
_
2
dx
_

2
w
x
2
+f
e
(7)
And the four boundary conditions are
w|
x=0
= 0,
w
x

x=0
= 0, w|
x=S
= g,
w
x

x=S
= 0 (8)
It is noticed this governing equation is the same as the Abdel-Rahmans governing equation
[2]
for
the case of no axial force and no dielectric layer. There is an implicit assumption in the derivation of
both governing equations, that is, the capacitance of the beam-substrate structure remains unchanged
during beam deection. This fact is reected in the expression of f
e
. Yangs governing equation
[6]
also
implicitly uses this assumption. The experiment conducted by Chan et al.
[3]
does showthat the structure
capacitance changes under dierent beam stiction lengths. But the change is very small and can be
ignored. According to experimental data given by Chan et al.
[3]
for the beam specimen of 300 m 30
m 1.5 m, the change of structure capacitance is only around 0.4 picofarad as the voltage changes
from 27 V to 35 V. (In their experiment, at 27 V, the beam just contacts the substrate and at 35
V, a relatively large portion of the beam is in contact with the substrate.) According to the formula
given by Chan et al. (Eq.(4) in their paper), the original parallel capacitance (when the beam has no
contact with the substrate) is 177.08 picofarad. And as the beam contacts the dielectric layer, residual
charge can accumulate in the dielectric layer and vary in time
[26]
. This fact, as a consequence, will cause
the voltage between the substrate and beam to vary in time. For this static study, the residual charge
varying in time is not considered.
The governing equation (7) can only be used to compute the beam deection w if the unstuck length
S is known. To determine S, the stationary point(s) needs to be computed by letting the functional
satisfy the following equation
[20, 21]
:

S
= 0 (9)
For beam structure (with no electrical force inuence), there is only one stationary point of functional
and there are two stationary points for circular plate structure, one is stable and the other is
Vol. 17, No. 2 Zhang Yin et al.: Static Study of Cantilever Beam Stiction 107
unstable
[20, 21]
. On the other hand, to compute the functional one should know w. So the computation
procedure here is to guess an S rst and use Eq.(7) to compute w. Then one substitutes the expression
of w into to see whether Eq.(9) is satised. For numerical computation, Eq.(9) is changed to

< (10)
here is a tolerance number.
If Eq.(10) is not satised, change S and compute w again by using Eq.(7). The iteration continues
until Eq.(10) is satised. Once Eq.(10) is satised, S is found. To illustrate this procedure of nding
S, a simple example with neither electrical force nor nonlinearity due to axial stretching is given. For
such a case, the governing Eq.(7) changes to
EI

4
w
x
4
= 0, 0 x S (11)
The boundary conditions are the same as Eq.(8). The solution of w is
w =
2g
S
3
x
3
+
3g
S
2
x
2
(12)
S is an unknown constant. Now = U
B
W
S
= 6EIg
2
/S
3
b(L S)f

W
a
(U
S
= W
E
= 0).
To determine S, Eq.(9) is used and /S = 18EIg
2
/S
4
+ bf

W
a
= 0. From this equation, S is
determined as S = [18EIg
2
/(bf

W
a
)]
1/4
= [3Et
3
g
2
/(2f

W
a
)]
1/4
. This S value is the long slender beam
equilibrium unstuck length with peel number N
p
= 1 case
[14, 20, 21]
. For N
p
> 1, the beam elastic energy
is greater than the adhesion work, and the beam cannot adhere to the substrate. For N
p
1, the beam
will adhere to the substrate
[14]
. Clearly for this simple case, only one iteration is needed to solve the
problem and obtain the analytical solution.
2.2. Dimensionless Equations
To non-dimensionalize the equations, the following quantities are introduced
[2]
W =
w
g
, =
x
L
, R =
S
L
(13)
The governing equation(7) now changes as
W

=
_

1
R
_
R
0
(W

)
2
d
_
W

+

2
V
o
2
(W )
2
(14)
here
1
= 6(g/t)
2
,
2
= 6
o
L
4
/(Eg
3
t
3
) and = 1 + d
e
/(g
r
).
r
is relative permittivity dened as

r
= /
o
. Here the operator ( )

= /.
The boundary conditions now change to
W|
=0
= 0,
W

=0
= 0, W|
=R
= 1,
W

=R
= 0 (15)
The third boundary condition should be noticed. Because of it, it has become very dicult (if not
impossible) to use the eigenvalue problem method to get the mode shapes to approximate the beam
displacement.
The functional now becomes
(, R) =
EIg
2
2L
3
_
_
_
_
R
0
(W

)
2
d +

1
2R
_
_
R
0
(W

)
2
d
_
2

3
(1 R) 2
2
V
o
2
_
R
0
_
1

1
W
_
d
_
(16)
here
3
= 2bL
4
f

W
a
/(EIg
2
) = 24L
4
f

W
a
/(Et
3
g
2
). For /S = 1/L (/R), the expression is
extremely lengthy and not given here. The general rule of how to take derivative (/R) of functional
_
b(R)
a(R)
f(, R)d is given in Appendix.
108 ACTA MECHANICA SOLIDA SINICA 2004
III. GALERKIN METHOD AND CONSTRAINT
The electrical force f
e
(, W) = b
o
V
o
2
/{2g
2
[W() ]
2
} is unknown except at the boundaries
( = 0, R). Let f
e1
= f
e
(0, W) = f
e
(0, 0) = b
o
V
o
2
/[2g
2

2
] and f
e2
= f
e
(R, W) = f
e
(R, 1) =
b
o
V
o
2
/[2g
2
(1 )
2
]. To use the Galerkin method, the polynomials (
i
s) are used to approximate
W
W =
N

i=1
a
i

i
(, R) (17)
a
i
s are unknown constants to be determined. For each
i
, the same boundary conditions for W are
satised

i
|
=0
= 0,

i

=0
= 0, |
=R
= 1,

=R
= 0 (18)
And each
i
has such an expression

i
(, R) = P
i
(, R) +C
i1
(R)
3
+C
i2
(R)
2
+C
i3
(R) +C
i4
(R) (19)
C
i1
, C
i2
, C
i3
and C
i4
are unknown constants to be determined by boundary conditions. P
1
(, R) = 0
and for i 2, P
i
(, R) = K
i1
(R)
i+3
+ K
i2
(R)
4
(K
i1
, K
i1
are constants). And P
i
(i 2) are also
required to satisfy the following conditions at = 0, R
EI

4
P
i
x
4

x=0
=
EI
L
4
P
i

|
=0
= f
e1
, EI

4
P
i
x
4

x=S
=
EI
L
4
P
i

|
=R
= f
e2
(20)
In other words, P
i
s (i 2) are required to match the electrical force distributions at the boundaries.
By using Eqs.(18 20), the polynomials (
i
s) can be determined as

i
(, R) =
_

2
R
3

3
+
3
R
2

2
(i = 1)
f
i
R
i1

i+3
+f
4

_
2
R
3
+ (i + 1)f
i
R + 2fR
_

3
+
_
if
i
R
2
+fR
2
+
3
R
2
_

2
(i 2)
(21)
here f
i
=
2
V
o
2
(2 1)/[(1 )
2

2
(i + 3)(i + 2)(i + 1)i] and f =
2
V
o
2
/(24
2
).
A natural question to ask is mathematically and physically what exactly are these polynomials?
1
actually is the solution of the linear dierential equation
1

= 0 with the boundary conditions of

=
i1
(i+3)(i+
2)(i +1)if
i
/R
i1
+24f with the boundary conditions of Eqs.(18).
i1
(i +3)(i +2)(i +1)if
i
/R
i1
+24f
is the distributed load. P
i
(, R) is the particular solution of this distributed load. C
i1
(R)
3
+C
i2
(R)
2
+
C
i3
(R) + C
i4
(R) is the homogeneous part solution. Also remember Eq.(20),
i
s (i 2) are designed
not only to satisfy the boundary conditions but also to match electrical force f
e
at the boundaries. So

1
is the solution of no load case,
2
the solution of linearly distributed load case,
3
the solution of
quadratically distributed load case,
4
the solution of cubicly distributed load case and so on and so
forth.
Although each
i
is designed to satisfy the boundary conditions of Eqs.(18), the boundary conditions
for W (Eq.(15)) can be satised if and only if the constraint

N
i=1
a
i
= 1 is satised. Now the governing
Eq.(14) will change to
W

1
R
_
R
0
(W

)
2
d
_
W

2
V
o
2
(W )
2
+(
N

i=1
a
i
1) = 0

N
i=1
a
i
1 = 0
(22)
is a Lagrangian multiplier. One substitutes W =

N
i=1
a
i

i
(, R), multiplies
i
and integrates from
0 to R for the rst equation in Eq.(22). Equation (22) becomes a set of the following N + 1 equations
_
R
0

1
_
_
_
N

i=1
a
i

d
_
_

1
R
_
R
0
(
N

i=1
a
i

)
2
d
_
_
N

i=1
a
i

Vol. 17, No. 2 Zhang Yin et al.: Static Study of Cantilever Beam Stiction 109

2
V
o
2
(

N
i=1
a
i

i
)
2
+(
N

i=1
a
i
1)
_
d = 0
_
R
0

2
_
_
_
N

i=1
a
i

d
_
_

1
R
_
R
0
(
N

i=1
a
i

)
2
d
_
_
N

i=1
a
i

2
V
o
2
(

N
i=1
a
i

i
)
2
+(
N

i=1
a
i
1)
_
d = 0 (23)


_
R
0

N
_
_
_
N

i=1
a
i

d
_
_

1
R
_
R
0
(
N

i=1
a
i

)
2
d
_
_
N

i=1
a
i

2
V
o
2
(

N
i=1
a
i

i
)
2
+(
N

i=1
a
i
1)
_
d = 0
N

i=1
a
i
1 = 0
For the N + 1 unknowns, a
i
(i = 1, 2, . . . , N) and are solved by the Newton-Raphson method.
Here the dimensionless unstuck length R is assumed before using Eq.(23) to solve a
i
and . Whether
the assumed R is the true equilibrium unstuck length depends on whether |/R| < is satised.
Because it uses the stationary point(s) of total energy to nd equilibrium, unstable equilibrium
(local maximum point of total energy) may be encountered. It is necessary to detect the sign change of
/R around the stationary point in the program to tell whether the stationary point is stable or not.
According to Mastrangelo and Hsu
[21]
, there is only one and stable equilibrium for the beam stiction
of the electrical force free case. The computation in this paper also nds only one (stable) equilibrium
in the case of electrical inuence.
IV. RESULTS AND DISCUSSION
The dimensionless unstuck length R
o
under no electrical force inuence is introduced. R
o
has the
analytical solution
R
o
=
S
o
L
=
_
3Et
3
g
2
/(2f

W
a
)

1/4
L
=
_
36

3
_
1/4
(24)
Fig. 2. Beam deection under dierent voltages at the orig-
inal stiction Ro = 1/2 and = 1.05.
Fig. 3. Beam deection under dierent s at the original
stiction Ro = 1/2 and
2
Vo
2
= 50.
110 ACTA MECHANICA SOLIDA SINICA 2004
For simplicity and in order to compare with the analytical solution R
o
, whose governing equation
does not account for the stretching nonlinearity at all,
1
in all the cases computed here are taken as
0. The polynomials (
i
, i = 1, 2, . . . , N) used in the Galerkin method exhibit good convergence on the
computation of R. The results given here are all computed by 6 polynomials (N = 6).
Figure 2 shows the voltage inuence on beam stiction. is 1.05 and
3
is taken as 576, thus
R
o
= 0.5, half of the beam originally adheres to the substrate under no electrical force inuence. When

2
V
o
2
= 50, the unstuck length shrinks to 0.36101. is another parameter determining the electrical
force distribution. depends on the dielectric material permittivity and the layer thickness. Smaller
causes larger electrical distribution, which causes larger stiction. In Fig.3,
2
V
o
2
is xed at 50. The
unstuck length R = 0.36101 for = 1.05, R = 0.39576 for = 1.08 and R = 0.41107 for = 1.10. The
original unstuck length R
o
also has its inuence on the beams further stiction on the substrate. In Fig.4,

3
is taken as 2916 (R
o
= 1/3), thus two thirds of the beam is stuck to the substrate. is also taken as
1.05. As
2
V
o
2
changes from 0 to 50, the unstuck length R shrinks from 1/3 to 0.3. So compared with
the results shown in Fig.2, the change is relatively much smaller. This is physically understandable. As
R
o
shrinks, the beam span in governing equation also shrinks. Thus the whole system becomes stier
and at the same time, the electrical force acting on the beam unstuck part is reduced. Figure 5 shows
the inuence on the beam stiction and
2
V
o
2
is also taken as 50 and R
o
= 1/3. For = 1.05 case,
R = 0.3, for = 1.08 case, R = 0.311864 and for = 1.10 case, R = 0.316384. So again compared
with the results in Fig.3, the change is relatively smaller too.
Fig. 4. Beam deection under dierent voltages at the orig-
inal stiction Ro = 1/3 and = 1.05.
Fig. 5. Beam deection under dierent s at the original
stiction Ro = 1/3 and
2
Vo
2
= 50.
Fig. 6. Beam unstuck length R under dierent s and
2
Vo
2
at the original stiction Ro = 1/2.
Fig. 7. Beam unstuck length R under dierent s and
2
Vo
2
at the original stiction Ro = 1/3.
Vol. 17, No. 2 Zhang Yin et al.: Static Study of Cantilever Beam Stiction 111
Figure 6 shows the combining eects of two parameters
2
V
o
2
and inuence beam stiction for the
case of R
o
= 1/2. Figure 7 shows the same for R
o
= 1/3 case.
V. CONCLUSION AND REMARKS
A systematic method of generating polynomials is given. The polynomials are the solutions of (linear)
beam deection governing equation under dierent distributed loads. At the same time, each polynomial
matches the electrical force distribution at the two end boundaries and satises the boundary conditions
of the S-shaped beam. The dielectric layer thickness, permittivity and the voltage between the beam and
substrate eects on the beam stiction are studied. As the voltage reduces to zero, the results converge
to the analytical solution of no electrical force. All the results are the cases of peel number N
P
= 1 case.
This work could oer a more accurate way of predicting beam stiction under electrostatic inuence in
real world. However, as pointed out by van Spengen, Puers and De Wolf
[13]
, the surface interaction
energy should be considered as a distribution. They drew the conclusion by referring to the experiments
done by de Boer et al.
[24]
that a number of beams, which are expected to be exactly the same, give widely
diering values for the surface interaction energy under identical conditions. Our surface interaction
energy (W
S
) is treated as a constant for each specic adhered beam though one can arbitrarily change
its values. The model presented in this paper may be enhanced by introducing random variables with
certain distributions for (the part of) the beam contact length, displacement etc.
APPENDIX
If a function f(, R) and f/R are both continuous in [A, B; C, D] and both [a(R)]/R and
[b(R)]/R exist when C R D (A a(R) B and A b(R) B), then
[27]

R
_
_
b(R)
a(R)
f(, R)d
_
=
_
b(R)
a(R)
[f(, R)]
R
d +f(b(R), R)
[b(R)]
R
f(a(R), R)
[a(R)]
R
References
[1] Osterberg,P.M. and Senturia,S.D., M-Test: A rest chip for MEMS material property measurement using
electrostatically actuated test structures, Journal of Microelectromechanical Systems, Vol.6, 1997, 107-118.
[2] Abdel-Rahman,E.M., Younis,M.I. and Nayfeh,A.H., Characterization of the mechanical behavior of an
electrically actuated microbeam, J. Micromech. Microeng., Vol.12, 2002, 759-766.
[3] Chan,E.K., Carikipati,K. and Dutton,R.W., Characterization of contact electromechanics through
capacitance-voltage measurement and simulation, Journal of Microelectromechanical Systems, Vol.8, 1999,
208-217.
[4] Adams,S.G., Bertsch,F.M., Shaw,K.A., Hartwell,P.G., Moon,F.C. and MacDonald,N.C., Capacitance based
tunable resonators, J. Micromech. Microeng., Vol.8, 1997, 15-23.
[5] van Spengen,W.M., Puers,R. and De Wolf,I., A physical model to predict stiction in MEMS, J. Micromech.
Microeng., Vol.12, 2002, 702-713.
[6] Yang,F., Electromechanical instability of microscale structures, Journal of Applied Physics, Vol.92, 2002,
2789-2794.
[7] Hung,E.S. and Senturia,S.D., Extending the travel range of analog-tuned Electrostatic Actuators, Journal
of Microelectromechanical Systems, Vol.8, 1999, 497-505.
[8] Tilmans,H.A. and Legtenberg,R., Electrostatically driven vacuum-encapsulated polysilicon resonators
part II: theory and performance, Sensors Actuators A, Vol.45, 1994, 67-84.
[9] Toshiyoshi,H., Piyawattanametha,W., Chan,C. and Wu,M.C., Linearization of electrostatically actuated
surface micromachined 2-D optical scanner, Journal of Microelectromechanical Systems, Vol.10, 2001, 205-
214.
[10] Alley,R.L., Cuan,G.J., Howe,R.T. and Komovopoulus,K., The eect of release-etch processing on surface
microstructure stiction, Proc. IEEE Solid-State Sensors and Actuators Workshop, Hilton Head Island, SC,
1992, 202-207.
[11] Bhushan,B., Tribology and Mechanics of Magnetic Storage Devices, New York: Springer, 1990.
[12] Anderson,K.M. and Colgate,J.F., A model of attachment/detachment cycle of electrostatic microactuators,
Micromech. Sensors.Actuators and Systems, DSC, Vol.32, 1991, 255-268.
[13] van Spengen,W.M., Puers,R. and De Wolf,I., The prediction of stiction failures in MEMS, IEEE Transac-
tions on Devices and Materials Reliability, Vol.3, 2003, 167-172.
112 ACTA MECHANICA SOLIDA SINICA 2004
[14] Zhao,Y.P., Wang,L.S. and Yu,T.X., Mechanics of adhesion in MEMSa review, J. Adhesion Sci. Technol,
Vol.17, 2003, 519-546.
[15] Yeh,R., Kruglick,E.J.J. and Pister,K.S.J., Microelectromechanical Component for Articulated Microrobots,
Proc. 8th Int. Conf. on Solid-State Sensors and Actuators (Transducers 95) and Eurosensors IX, Stockholm,
Sweden, 1995, 346-349.
[16] Tas,N.R., Gui,C. and Elwenspoek,M., Static friction in elastic adhesion contacts in MEMS, J. Adhesion
Sci. Technol, Vol.17, 2003, 547-561.
[17] Tas,N.R., Sonnenberg,T., Jansen,H., Legtenberg,R. and Elwenspoek,M., Stiction in surface micromaching,
J. Micromech. Microeng., Vol.6, 1996, 385-397.
[18] Serry,F.M., Walliser,M. and Maclay,G.J., The Role of the Casimir Eect in the Static Deection and Stiction
of Membrane Strips in Microelectromechanical Systems (MEMS), Journal of Applied Physics, Vol.84, 1998,
2501-2506.
[19] Jones,E.E., Begley,M.R. and Murphy,K.D., Adhesion of micro-cantilevers subjected to mechanical point
loading: modeling and experiments, Journal of the Mechanics and Physics of Solids, Vol.51, 2003, 1601-1622.
[20] Mastrangelo,C.H. and Hsu,C.H., Mechanical stability and adhesion of microstructures under capillary
forcespart I: basic theory, Journal of Microelectromechanical Systems, Vol.2, 1993, 33-43.
[21] Mastrangelo,C.H. and Hsu,C.H., Mechanical stability and adhesion of microstructures under capillary
forcespart II: experiment, Journal of Microelectromechanical Systems, Vol.2, 1993, 44-55.
[22] Yee,Y., Park,M. and Chun,K., A sticking model of suspended polysilicon microstructure including residual
stress gradient and postrelease temperature, Journal of Microelectromechanical Systems, Vol.7, 1998, 339-
344.
[23] Chen,Z.R. and Yu,S.W., Capillary adhesive contact between a spherical rigid punch and a piezoelectric
half space, Journal of Applied Physics, Vol.94, 2003, 6899-6907.
[24] de Boer,M.P., Knapp,J.A., Mayer,T.M. and Michalske,T.A., The role of interfacial properties on MEMS
performance and reliability, Proc. SPIE., Vol.3825, 1999, 2.
[25] McDonald,Jr.P.H., Nonlinear Dynamic Coupling in a Beam Vibration, Journal of Applied Mechanics,
Vol.22, 1955, 573-578.
[26] Bochobza-Degani,O., Socher,E. and Nemirovsky,Y., On the eect of residual charges on the pull-in param-
eters of electrostatic actuators, Sensors and Actuators A, Vol.97-98, 2002, 563-568.
[27] Chen,C.Z. et al., Mathematical Analysis, Volume 2, 2nd ed., Beijing: Gaodeng Jiaoyu Press, 1983(in Chi-
nese).