Professional Documents
Culture Documents
Doctoral Thesis by S Chaterjee
Doctoral Thesis by S Chaterjee
SAIKAT CHATERJEE
A thesis submitted in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
(Engineering)
under the supervision of
to the
JADAVPUR UNIVERSITY
KOLKATA-700032, INDIA
INDEX NO. 22/08/Engg.
1. Title of Thesis:
Modelling and Simulation of Static and Dynamic Behaviour of Cantilever-Based
Electrostatically Actuated MEMS Devices
2. Name, Designation & Institution of the Supervisor:
Dr. Goutam Pohit
Professor, Department of Mechanical Engineering,
Jadavpur University, Kolkata-700032.
3. List of Publications (Referred Journals):
i. Chaterjee, S. and Pohit, G., 2010, Squeeze-film damping characteristics of
cantilever microresonators for higher modes of flexural vibration, International
Journal of Engineering Science and Technology 2(4), 187-199, Multicraft.
ii. Chaterjee, S. and Pohit, G., 2009, A large deflection model for the pull-in analysis
of electrostatically actuated microcantilever beams, Journal of Sound and
Vibration 322, 969-986, Elsevier.
iii. Chaterjee, S. and Pohit, G., Squeeze-film damping characteristics of cantilever
microresonators under large electrostatic loading, Journal of Mechanics of
Advanced Materials and Structures, Taylor & Francis. (in press)
iv. Chaterjee,
S.
and
Pohit,
G.,
Dynamics
of
electrostatically
actuated
4.
List of Patents:
5.
-- i --
-- ii --
AND
DYNAMIC
BEHAVIOUR
OF
CANTILEVER-BASED
-- iii --
The author, Sri Saikat Chaterjee was born in Kolkata in 1975. He completed his
graduation (B.E.) in Mechanical Engineering from Indira Gandhi Institute of Technology,
Sarang, Odisha in 1997. He pursued post-graduation (M.E.) course in Mechanical
Engineering (Machine Design) from Bengal Engineering College (presently BESU),
Kolkata, and successfully completed the same in 2003. He is engaged in academic
activities since the last 12 years. He is presently a faculty member in the Department of
Mechanical Engineering of C. V. Raman College of Engineering, Bhubaneswar, Odisha.
-- v --
Acknowledgement
Being a QIP research scholar, I got the rare opportunity to interact with some of
the best minds of the Department of Mechanical Engineering, Jadavpur University. But
perhaps it was destiny to have met the person who enthused and enthralled me with his
humble yet inspiring ideas. With successful transformation of those ideas into this volume
of work, I take this opportunity to acknowledge my sincere appreciation for my
honourable guide Prof. Goutam Pohit, Professor, Department of Mechanical Engineering,
Jadavpur University. In spite of his towering intellect, I always felt at ease during the
several hours of discussion held in course of preparation of this thesis. I learnt and
enriched myself from his methodical guidance and will cherish every moment of his wise
company.
I sincerely acknowledge the authority of C V Raman College of Engineering for
showing confidence in me and sponsoring me for the doctoral work.
I would like to express my gratitude to all the members of the Machine Element
Laboratory, research and post graduate scholars who have been with me throughout the
completion of this thesis. Their valuable suggestions and input have certainly helped to
improve the overall content of the thesis. Their very presence and the lighter moments
that we shared definitely brought solace in the days of hardship.
I am also grateful to all my well-wishers and family members whose
overwhelming support and belief in me helped me to curve my way out to the timely
completion of the thesis.
I would like to express my sincere indebtedness to the staff and faculty members
of the Department of Mechanical Engineering, Jadavpur University.
Saikat Chaterjee
-- vii --
To,
Maa, Mithu Maasi and Mama
Contents
Page No.
iii
Acknowledgement
vii
Contents
xi
List of Notations
xv
List of Figures
xix
List of Tables
xxiii
Abstract
xxv
01-31
Chapter 1: Introduction
1.1 Background and Motivation
1.2 Literature Review
1.2.1 Cantilever Beam Theories
1.2.2 Squeeze-film Damping
1.2.3 Static and Dynamic Behaviour of Electrostatic MEMS
1.2.3.1 Pure DC Voltage Actuation
1.2.3.2 Combined DC and AC Voltage Actuation
1.2.4 Summary of the Literature Review
1.3 Description of Thesis Problems
1.4 Layout of the Thesis
1.5 Contributions of the Thesis.
Chapter 2: Static and Transient Behaviour under a DC Electrostatic
Force
2.1 Introduction
2.2 Model Description
2.3 Boundary Value Problem
2.4 Galerkin Formulation
2.5 Results and Discussion
-- xi --
33-55
57-86
3.1 Introduction
3.2 Mathematical Formulation
3.2.1 Compressible Flow Model
3.2.2 Static Deflection
3.2.3 Small Amplitude Motion
3.2.4 Compressibility Ratio
3.2.5 Incompressible Flow Model
3.3 Finite Element Model
3.4 Results and Discussion
3.4.1 Validation Study
3.4.2 Effects of DC Bias Voltage
3.4.3 Effects of Variations in Ambient Pressure
3.4.4 Comparison Between Incompressible and Compressible Flow
Models
3.5 Summary
Chapter 4: Nonlinear Response under Large DC and AC Excitation
4.1 Introduction
4.2 Mathematical Formulation
4.2.1 Reduced Order Model
4.3
4.4
Summary
-- xii --
87-107
109-114
5.1 Conclusions
5.1.1 Static and Transient Behaviour under Pure DC Actuation
5.1.2 Small Amplitude Dynamics under Combined Actuation
5.1.3 Nonlinear Dynamics under Combined Actuation
5.2 Recommendations for Future Work
115-131
References
-- xiii --
List of Notations
1
Variational operator
Permittivity constant
Damping ratio
xx
Axial strain
Non-dimensional damping parameter
eff
Coefficient of viscosity
Effective coefficient of viscosity
Poissons ratio
Excitation frequency
res
Resonance frequency
Density of the beam material
Squeeze-number
Cut-off squeeze-number
Pressure function
Compressibility ratio
ai
-- xv --
beff
cl
d0
Plate modulus
fd
Fd
Fs
g
Nonlinear curvature
Air-gap spacing
ka
ke
kl
Kn
Kn0
l
leff
m, M
ml
n, N
Knudsen number
Linearized Knudsen number
Length of the beam
Effective length used for non-trivial pressure boundary conditions
Number of terms retained in the expression for pressure function
Lumped mass of the single DOF system
Number of terms retained in the reduced order model
pa
Ambient pressure
p0
P*
Quality factor
R
t
-- xvi --
Ue
Us
Small AC voltage
v0
Driving voltage
VAC
VDC
w
w0
wd
wmax
zm
zmR , zmI
-- xvii --
List of Figures
Page No.
Figure 1.1
Figure 1.2
03
05
Figure 2.2
35
42
Figure 2.3
42
Figure 2.4
Figure 2.5
44
Figure 2.6
Figure 2.7
46
47
Figure 2.9
47
48
-- xix --
49
Figure 2.10
1 = 2.1e 5 .
50
Figure 2.11
51
Figure 2.12
51
Figure 2.13
2
Phase plot at representative values of 2VDC
in the vicinity
53
53
cantilever microbeam.
Figure 3.2
Figure 3.3
68
Figure 3.4
54
71
Figure 3.5
71
75
76
-- xx --
Figure 3.7
Figure 3.8
77
Figure 3.9
77
Figure 3.10
78
Figure 3.11
80
Figure 3.12
81
Figure 3.13
83
Figure 4.1
85
92
93
Figure 4.3
94
Figure 4.4
94
Figure 4.5
hysteresis.
Figure 4.6
-- xxi --
96
Figure 4.7
Figure 4.8
Figure 4.9
98
Figure 4.11
99
Figure 4.12
97
Figure 4.10
96
99
Figure 4.13
100
Figure 4.14
Figure 4.15
101
102
103
Figure 4.17
104
Figure 4.18
104
-- xxii --
106
List of Tables
Page No.
Table 2.1
Table 2.2
44
Table 3.1
52
Table 3.2
72
Table 3.3
73
The percentage error in damping ratio of the present semianalytical model ( ) with reference to the FE model ( FE )
compared at different compressibility ratios in the first
flexural mode for zero ( VDC = 0 V ) DC bias voltage.
Table 3.4
73
-- xxiii --
74
Abstract
The outstanding advantages and rapidly growing potential applications of Micro-ElectroMechanical Systems (MEMS) have generated a tremendous impetus to illuminate and
illustrate microscale phenomena, and, to aid in the design of MEMS devices. Amongst
the several MEMS devices which have already reached the market or are yet to pass the
functionality tests, electrostatically actuated devices are the forerunner and are most
extensively used. The thorough understanding, prediction, and control of static and
dynamic behaviour of such devices at the microscale are critical issues. Experimental
investigation at the microlevel may not be always possible as it involves sophisticated
instrumentation and novel techniques in all the three stages including observation,
measurement and testing. The development and prototype of MEMS devices rely heavily
on trial and error and consume considerable time and expense. In this respect,
formulation of realistic models and subsequent simulation study emerges as a domain of
research which truly complements the complexities of experimental characterization.
Many approaches have been proposed to make realistic models of microsystems
and simulate their behaviour. Modelling of MEMS, however, pose several challenges due
to the coupling amongst the several physical domains such as structural mechanics,
electrostatics, and fluid mechanics. Moreover, the electrostatic domain is inherently
nonlinear, and, the structural and fluidic nonlinearities need to be further considered to
develop a model as realistic as possible. Starting from simplified lumped models to
detailed three-dimensional models, various techniques for extracting mathematical
models have been devised to analyse MEMS. The computational expenses associated
with simulation of full three-dimensional models have triggered a recent shift in the
construction of reduced-order distributed models for MEMS. Fundamentally, all
numerical methods for solving the field equations rely on the process of discretization
through which the degrees of freedom of the field are reduced from an infinite number to
a finite number. The reduced-order model (ROM) approach incorporates relatively small
number of basis functions in the process of discretization and effectively balances the
need for a realistic model against the numerical efficiency.
The present thesis investigates the static and dynamic behaviour of
electrostatically actuated cantilever beams in two important MEMS-based devices. The
first is resistive microswitches, in which a cantilever microbeam is actuated by DC
-- xxv --
electrostatic forces. Static behaviour and transient dynamics are analysed for deformable
cantilever electrodes separated from the ground electrode by large gaps, with special
emphasis on the nonlinear effects due to geometry, electrostatic forces, and inertial terms.
The non-linear curvature and the von Krmn non-linearity in the axial straindisplacement relationship are incorporated to account for the geometric non-linearity of
the microcantilever associated with large deflection. To realize a compatible model,
higher order electrostatic terms associated with large gap separations are also considered.
The results obtained with numerical simulation of the ROM are validated with the
experimental and numerical results available in open literature. The static and transient
results are shown to be much improved when higher order correction terms are taken into
account. The large deflection model to treat the large gap separation problem is derived
and simulated for the first time and the results presented may serve as benchmarks for
further study. The second category of device studied is resonant microsensors, in which
the microcantilever is deflected by a bias DC voltage and an AC harmonic voltage forces
it to vibrate in the resonant mode. The response characteristics of small and large
amplitude motions are investigated for small air-gaps in presence of squeeze-film
damping accounting for rarefied flow conditions. The results for small amplitude
dynamics generated with a semi-analytical method are validated with the analytical,
numerical and experimental results available in open literature. The dynamic problem is
simulated with the coupled field finite element analysis software ANSYS and results are
compared. The applicability of the present model for the range of flow compressibility is
studied. A comparison of the squeeze-film characteristics obtained with compressible and
incompressible flow models is also presented. The large amplitude dynamic behaviour is
studied in the primary and superharmonic frequency range. The distributed model is
formulated accounting for the nonlinearities of the system, arising out of electrostatic
forces and the squeeze-film damping terms. Several nonlinear phenomena are illustrated
using frequency-response curves, time-histories, and phase portraits as effective analysis
tools. The comprehensive approach and the results presented in this dissertation will help
the designers to realize better performance of both pull-in and non-pull-in microscale
devices.
-- xxvi --
Chapter 1
INTRODUCTION
Outline of the chapter: 1.1 Background and Motivation, 1.2 Literature Review, 1.2.1
Cantilever Beam Theories, 1.2.2 Squeeze-film Damping, 1.2.3 Static and Dynamic
Behaviour of Electrostatic MEMS, 1.2.3.1 Pure DC Voltage Actuation, 1.2.3.2 Combined
DC and AC Voltage Actuation, 1.2.4 Summary of the Literature Review 1.3 Description of
thesis problems, 1.4 Layout of the thesis, 1.5 Contributions of the thesis
1.1
Chapter 1
pressure sensors, accelerometers used in automotive air-bags, digital mirror displays
(DMD), fine-positioning actuators for high-performance disk drives, and inkjet printerhead nozzles are a few MEMS products that have been commercially successful.
However, market analyses portray MEMS as an Enabling Technology with huge
potential
of
application
in
sectors
like
IT/peripherals,
medical/biochemical,
Introduction
Chapter 1
The actuation method offers high sensitivity, high energy density, fast response, and high
precision control of microscale motion [Price et al., 1989; Hung and Senturia, 1999b].
However, high voltage is needed to generate electrostatic forces to be used as the driving
force. Electrostatic sensing devices like capacitive accelerometers [Sasayama et al., 1996]
and capacitive sensors [Bao et al., 2000] are widely used. In MEMS, electrostatic forces
are often used to actuate microstructures including microswitches [Lee et al., 2004],
microgrippers [Millet et al., 2004], microrelays [Schiele et al., 1998], and electrostatic
motors [Mehregany et al., 1990; Ishihara et al., 1996]. These MEMS devices use either
pure direct current (DC) voltages or a combination of DC and alternating current (AC)
voltages. The present work focuses on electrostatically actuated cantilever microbeams
which form the heart of MEMS devices like resistive microswitches and resonant sensors.
In microswitch, a cantilever beam with contact tips at the free end is actuated by DC
voltages and the resulting electrostatic forces deflect the beam to close an external electric
circuit allowing current to flow. On the other hand, in resonant sensors, the
microcantilever is deflected by a bias DC voltage and an AC harmonic voltage forces it to
vibrate in the resonant mode. Resonant sensors rely upon correlating the shift in the
resonant frequency of the vibrating beam with the quantity to be sensed [Beeby et al.,
2004; Ziegler, 2004].
Even though electrostatically induced forces seem to be rather weak, because of
the large surface area to volume ratio of MEMS, surface effects such as electrostatics
dominate volume effects such as gravity. Electrostatic phenomena arise from the forces
that stationary electric charges exert on each other, and such forces are described by
Coulomb's law. As shown in Figure 1.2, a typical electrically actuated MEMS device
consists of a deformable electrode (cantilever beam) made of a conductive material
suspended above a rigid conductive electrode with a dielectric medium, generally air,
between them. An electric potential difference applied between the two electrodes
induces electrostatic charges on the surface of the conductors as shown in Figure 1.2(a).
The distribution of electrostatic charges on the surface of the conductors depends on the
relative position of the two conductors. These electrostatic charges give rise to
electrostatic pressure, also known as Coulomb pressure, which acts normal to the surface
of a perfect conductor and deflects the deformable conductor towards the rigid one. As
the movable electrode deforms, the charges redistribute (Figure 1.2(b)) on the surface of
the conductors and, consequently, the Coulomb pressure and the gap-separation between
Introduction
the two electrodes also change. Thus, there exists the structure-electrostatic coupling
effect which needs to be taken into account while designing such microstructures.
As shown in Figure 1.2(b), the electrostatic load, the mechanical restoring force and the
damping force together govern the dynamic behaviour of this system. For small initial
gap-separation to beam-length ratios, the surface of the deformable conductors can be
assumed to be flat and locally parallel to the lower rigid electrode and parallel plate
(parallel capacitor) approximation is used to determine the electrostatic pressure [Pelesko
and Bernstein, 2003]. However, more accurate estimates of the electrostatic pressure can
be developed by considering the slope and curvature of the deformable electrode [Pelesko
and Driscoll, 2005; Krylov and Seretensky, 2006]. The present work studies the case of
large gap-length ratios incorporating the corrections to the parallel plate capacitance
model and establishes the ratio limiting the validity of the parallel plate approximation.
Electrostatic phenomenon introduces a new instability concept called Pull-in.
The applied electrostatic voltage has an upper limit beyond which electrostatic force is
not balanced by the counter-balancing forces (Figure 1.2(b)) in the deformable conductor
that eventually snaps and touches the lower rigid electrode, and the MEMS collapses.
This phenomenon, called pull-in instability, has been observed experimentally
[Nathanson et al., 1967; Taylor, 1968]. The critical displacement and the critical voltage,
associated with this instability, are called pull-in displacement and pull-in voltage,
respectively. Their accurate evaluation is crucial in the design of both pull-in and non5
Chapter 1
pull-in MEMS devices. In particular, the designer avoids this instability in case of
microresonators in order to achieve stable motions; however, in switching applications
the designer exploits this effect to optimize the performance of the device. In the present
dissertation, pull-in effects have been investigated under pure DC as well as for combined
DC and AC loads.
Damping in MEMS strongly affects their performance, design and control.
Damping influences the behaviour of MEMS in various ways, depending on their design
criteria and operating conditions. For a resonant sensor, damping should be minimised to
enhance quality factor to achieve high resolution. On the other hand, for a
microaccelerometer, damping should be so designed that the overall damping of the
device is about the critical value to prevent large resonance responses which might result
in a mechanical or electrical failure. The common energy loss mechanisms [Tilmans et
al., 1992] include losses to surrounding fluid due to acoustic radiation and viscous
friction, losses due to microslips at support mounts (anchor/clamping losses), and
intrinsic losses of the material. Acoustic radiation results from sound waves generated by
the motion of the mechanical elements in direction normal to their surfaces. Acoustic
radiation is insignificant unless acoustic wavelength is equal to or less than a typical
dimension of the device such as the air-gap [Veijola and Lehtovuori, 2009].
Anchor/clamping losses refer to energy losses to substrate when vibrating elements are
directly attached to stationary substrate. They can be alleviated by mechanically isolating
the vibrating elements from the substrate. Intrinsic damping depends primarily on the
material and geometric properties of the structure. In single-crystal materials such as
silicon, intrinsic losses are very low and their contribution to the total damping is almost
negligible. Hence, among the various damping sources, viscous damping is the most
significant source of energy loss in MEMS.
The motion of small parts in a MEMS device can be significantly affected by the
surrounding air. Damping force of the surrounding air being a surface effect increases
significantly as the micromachined structures decrease in size. Drag force damping of air
is significant when the moving structure is not in the close proximity of any surrounding
objects [Nouira et al., 2007]. Typical MEMS devices employ parallel plate capacitors, as
shown in Figure 1.2, in which air fills the narrow gap between the two parallel conductors
and the deformable conductor moves normally against the stationary conductor. Under
such conditions, the damping due to squeeze-film action of the air between the conductors
is pronounced. As the cantilever beam moves in the trapped air, the resistive force is
6
Introduction
caused by the damping pressure in the air-gap. The electrostatic pressure and the damping
pressure acting on the beam together cause the beam to deform to a state where they are
balanced by the inertial stiffness and the inertial forces at that time instant (Figure 1.2(b)).
Thus, the dynamics of the electrostatically actuated cantilever moving in trapped air is
influenced by the fluid-structure-electrostatic coupling effect. The damping pressure
consists of two main components: the component to cause the viscous flow of air when
the air is squeezed out of (or sucked into) the gap region and that to cause the
compression/expansion of the air-film. The viscous flow dissipates the movement and the
compression/expansion of the air-film causes a spring force. The combined effect is
called the squeeze-film damping. Therefore, viscous damping in many MEMS devices
corresponds to squeeze-film damping. The present work takes into account the squeezefilm damping effects while studying the dynamics of microcantilevers excited by
small/large amplitude AC harmonic voltages.
The thorough understanding, prediction, and control of static and dynamic
behaviour at the microscale are critical issues. Experimental investigation may not be
always possible as it involves sophisticated instrumentation and novel techniques in all
the three stages including observation, measurement and testing. The development and
prototype of MEMS devices rely heavily on trial and error and consume considerable
time and expense. An interface, which connects design and fabrication, should be made
to provide MEMS designers with convenient modelling and simulation tools. Modelling
of MEMS devices is challenging because of the types of forcing and nonlinearities
encountered in MEMS. In contrast to conventional macrostructures, the surface effects
dominate over volume or bulk effects as the feature size of mechanical elements is scaled
down. Most MEMS devices with electrostatic actuation involve some form of structural
or system nonlinearity: intrinsic nonlinearities due to electrostatic forces, structural
nonlinearities due to large deflections compared to device dimensions, and nonlinear
squeeze-film damping. Furthermore, MEMS technology rests on multidisciplinary
foundations. In the present dissertation, several physical domains such as structural
mechanics, electrostatics, and fluid mechanics have been treated with due emphasis on
the coupling amongst the different fields.
Many approaches have been proposed to make realistic models of microsystems
and simulate their behaviour. Starting from simplified lumped models to detailed threedimensional models, various techniques for extracting mathematical models have been
devised to analyse MEMS. The lumped model has the advantage of simplicity and full
7
Chapter 1
analytical description but has the disadvantage of deviating far from the reality. Full
three-dimensional models are closer to the reality but their simulation involves massive
numerical calculations. In order to alleviate the computational expenses associated with
the three-dimensional analysis, considerable efforts have been devoted to the
development of reduced-order distributed models [Batra et al., 2007] for MEMS.
Approximations can be made in any of the three fields of the coupled field MEMS
analysis: structural, electrical and fluidic. In the present work, deformable electrode is
modelled as a cantilever microbeam (one-dimensional), the electric field is modelled
incorporating suitable approximations depending upon the gap-length ratios, and the
three-dimensional Navier-Stokes equation governing fluid flow is reduced to the
Reynolds equation to model the fluid field. Reduced-order modelling is a way to balance
the need for a realistic model against the numerical efficiency necessary to make the
model of practical use in MEMS design.
The various fundamental equations of continuum physics are differential
equations governing the different fields. Very often these equations are difficult or
impossible to solve analytically with the current mathematical techniques. Under such
circumstances, computational methods are resorted to obtain approximate solutions to the
equations describing the realistic models. Fundamentally, all numerical methods for
solving field equations rely on the process of discretization through which the degrees of
freedom of the field are reduced from an infinite number to a finite number. The reduced
problem of finite number of degrees of freedom is then solved utilizing the numerical
schemes. In this connection, many numerical methods have been developed over the
years which find application in simulation of MEMS. Amongst these the noteworthy are
the finite-element methods (FEM), the boundary-element methods (BEM), the finitedifference methods (FDM), and, the various gridless and meshless methods. These
methods may be adequate for simulation of static behaviour of MEMS devices but are
generally inadequate for dynamic simulations because they require time integration of
large number of second-order ordinary-differential equations (ODEs). The computational
expenses associated with the above mentioned methods have triggered a recent shift in
the construction of reduced-order models (ROM) for MEMS applications. The ROM
generated through the Galerkin procedure of discretization [Nayfeh et al., 2005] uses
relatively small number of basis functions globally defined (instead of defining a basis
function for each element as in FEM) over the entire domain of the problem. The present
dissertation employs a ROM to simulate the static and dynamic behaviour of electrostatic
8
Introduction
MEMS using the linear undamped mode shapes of the undeflected microcantilever as a
basis set in the Galerkin procedure.
1.2
Literature Review
Several studies have investigated the behaviour of electrically actuated
1.2.1
the Euler-Bernoulli beam theory, and (c) the Timoshenko beam theory. The theory of
elasticity approach has a major drawback that only a few problems can be solved exactly
[Cowper, 1968], and hence it is not very attractive. The Euler-Bernoulli beam theory
[Shames and Dym, 1985] assumes that plane cross sections, normal to the neutral axis
before deformation, continue to remain plane and normal to the neutral axis and do not
undergo any strain in their planes (i.e., their shape remains intact). In other words, the
warping and transverse shear-deformation effects and transverse normal strains are
considered to be negligible and hence ignored. Incorporating the effect of transverse shear
deformation into the Euler-Bernoulli beam model gives us the Timoshenko beam theory
[Shames and Dym, 1985]. In this theory, to simplify the derivation of the equations of
motion, the shear strain is assumed to be uniform over a given cross section. In turn, a
shear correction factor is introduced to account for this simplification and its value
depends on the shape of the cross section [Cowper, 1968]. In the presence of transverse
shear, the rotation of the cross section is due to both bending and transverse (or out-of-
Chapter 1
plane) shear deformation. Few contributions have considered the effects of shear
deformation and rotary inertia. Abramovich (1992) investigated the influence of
compressive axial loads on the linear frequencies of Timoshenko beams having various
boundary conditions. This formulation, however, neglected the joint action of rotary
inertia and shear deformation. Foda (1999) studied the nonlinear vibrations of a
Timoshenko beam using perturbation method. While, both shear deformation and rotary
inertia were considered, it was restricted to the pinned boundary condition. Rao et al.
(2006) studied the large amplitude vibrations of hinged-hinged and clamped-clamped
Timoshenko beams with central point concentrated masses using the coupled
displacement field method. Recently, Ramezani et al. (2006) and Mestrom et al. (2010)
investigated the effects of shear deformation and rotary inertia on nonlinear vibration of
doubly clamped microbeams. They demonstrated that the effects of rotary inertia and
shear deformation can be neglected for slender microbeams. Furthermore, Crespo da
Silva (1988,1991) and Patel (1997) showed that beams with one end free to move behave
essentially as if they were inextensional when the slenderness ratio is large. In fact in the
absence of large axial forces, slender cantilever microbeams can be approximated [Batra
et al., 2006; Mahmoodi and Jalili, 2007] as inextensional members.
A linear beam model would suffice when dealing with small deformations. But
when the deformations are moderately large, for accurate modelling, several
nonlinearities also need to be included. The relevant structural nonlinearities can be
classified as [Evan-Iwanowski, 1976; Nayfeh and Mook, 1979; Moon, 1987] geometric
nonlinearity, inertia nonlinearity, nonlinear boundary conditions, and material
nonlinearity. Under large rigid-body rotations, structures like cantilever beams undergo
large deformations but small strains so that the material behaviour can be assumed to be
linear. Geometric nonlinearity exists in systems undergoing large deformations or
deflections. This nonlinearity arises from the potential energy of the system. In structures,
large deformations usually result in nonlinear strain-displacement and curvaturedisplacement relations. Several authors [Wagner, 1965; Hodges, 1984; Hsieh et al., 1994;
Nayfeh et al., 1995] accounted for geometric nonlinearity by incorporating, with varying
levels of complexity and sophistication, the nonlinear curvature and the von Krmn
nonlinearity in the axial strain-displacement relationship. Inertia nonlinearity arises from
the kinetic energy of the system. It is associated with the motion of an inextensional
cantilever beam. Crespo da Silva and Glynn (1978a,b) systematically derived the
nonlinear equations of motion and boundary conditions governing the flexural-flexural10
Introduction
torsional motion of isotropic, inextensional beams. They included nonlinearities due to
inertia and geometry up to order three and showed that the nonlinearities arising from the
curvature are of the same order of magnitude as those due to inertia. Pai and Nayfeh
(1990) and Anderson et al. (1996) investigated the nonlinear motions of cantilever beams
and observed that, for the first mode, the geometric nonlinearity, which is of the
hardening type, is dominant; whereas for the second and higher modes, the inertia
nonlinearity, which is of the softening type, becomes dominant. The structural geometric
nonlinearity of microbeams associated with large deflection is quite considerable
[Hassanpour et al., 2007; Mahmoodi and Jalili, 2007; Batra et al., 2008] because of their
low mass and high flexibility.
1.2.2
Squeeze-film Damping
Prior to the advent of MEMS, squeeze-film effect relating to air-film lubrication
had already found application in air-bearing and levitation systems. Detailed analysis and
experiments [Langlois, 1962; Griffin et al., 1966] revealed that squeeze-films have a
significant effect on bearing stability. Langlois (1962) employed a linearized Reynolds
lubrication equation [Hamrock, 1994] over a two-dimensional domain that represents the
compressed volume to capture the characteristics of the squeeze-film. Griffin et al. (1966)
treated the compressible squeeze-film problem for a strip plate. They obtained the
approximate expression for the cut-off frequency at which viscous damping force and
elastic force of squeeze-film air damping are equal. In electrostatic MEMS, modelling of
the squeeze-film damped system involves coupling of electrostatic, structural, and fluid
domains. When the microstructure is assumed to be rigid, the Reynolds equation is
decoupled and, on linearization, simpler expressions for the damping coefficient can be
obtained. Blech (1983) solved the linearized compressible Reynolds equation and
evaluated the ratio between the contributions of the elastic and damping effects for trivial
boundary conditions of rigid rectangular plates. This ratio is referred to as the squeeze
number in literature. He evaluated the dependence of cut-off squeeze number
(corresponding to cut-off frequency) on the aspect ratio of rectangular plate. Sadd and
Stiffer (1975) discussed the oscillating amplitude effect of squeeze-film damping for
plates of typical shapes at low squeeze numbers. They concluded that the elastic and
viscous damping effects can be represented by linear spring-damper approximation when
the oscillatory amplitude is not very large. Starr (1990) assumed incompressible flow and
linearized the Reynolds equation to model a capacitive parallel plate accelerometer. He
11
Chapter 1
also used the finite-element package ANSYS to calculate the pressure distribution
underneath a plate with a hole. Pan et al. (1998) presented dynamic angular response of a
MEMS torsion mirror by analytically solving the linearized compressible Reynolds
equation with the surface of the mirror assumed to be rigid. The results were validated
experimentally as well as compared with the numerical results obtained by solving the
nonlinear Reynolds equation using an implicit finite difference scheme. An analytical
method based on Greens function solution to the linearized Reynolds equation was
proposed by Darling et al. (1998) for nontrivial pressure boundary conditions. With this
relatively flexible method, they treated the rigid plate squeeze-film problems with
boundary conditions for one edge closed, two edges closed, etc. Trivial pressure boundary
condition, i.e., the gas at the borders is at ambient pressure, can be assumed while
modelling squeeze-film damping if the plate dimensions are much larger than the gap
dimensions. The most common method to include the border effects is to extend the
surface dimensions such that the coefficient of damping force for the enlarged plate with
trivial boundary conditions has the same value as the coefficient of damping force of the
real device size with the border effects. A series of two-dimensional and threedimensional FEM simulations were performed by Veijola et al. (2005) to extract the
approximate expressions for the elongations. The surface elongation model of Veijola et
al. (2005) was validated experimentally as well as numerically by Pandey et al. (2007)
while studying a fairly complex torsional MEMS structure with large air-gap and
complicated boundary conditions.
The air in the gap can be treated as a continuum as long as the gap distance is
much larger than mean free path of the gas molecules. For low pressure and/or a very
small gap, collisions among the gas molecules are reduced and the flow is said to be
rarefied. There have been two basic approaches so far in considering the damping in
rarefied air: the effective coefficient of viscosity and the free molecular model. The first
approach suggests the extension of the viscous flow model to non-continuum regimes.
The validity [Li and Hughes, 2000] of the continuum model depends on the nondimensional parameter Knudsen number which is defined as the ratio between the
molecular mean free path and the variable gap distance. Continuum flow is assumed as
long as the Knudsen number is less than a value equal to 0.01 . For rarefied gas,
Burgdorfer (1959) used a traditional approach of modifying the Reynolds equation
applying an effective viscosity to analyse the gas lubricated bearings. However, the
approach is limited to the slip-flow regime (Knudsen number less than unity). Christian
12
Introduction
(1966) proposed that in low vacuum the viscous flow model is no more valid and a free
molecular model has to be considered. He obtained the resistive damping force on an
oscillating plate from the momentum transfer rate between the plate and the molecules.
Newell (1968) used cantilever resonators to study the effect of damping in different
pressure regimes. In the regime of high pressure, viscous damping was found to be
dominant. Legtenberg and Tilmans (1994) calculated the quality factors of a clampedclamped microresonator based on the kinetic theory [Gupta, 1990] of gases in the
molecular regime. The theoretical values were found to overestimate the experimental
results. Veijola et al. (1995) investigated the behaviour of a capacitive accelerometer with
different intrinsic gas pressures. The dynamic behaviour was studied using a spring-massdamper model. They presented a simple approximate equation of pressure dependent
viscosity that is valid for both viscous and molecular damping regions. The coefficient of
effective viscosity becomes dependent on pressure via Knudsen number, and its use
extends the validity of the Reynolds equation to the non-continuum regimes. Bao et al.
(2002) proposed an improved model based on energy transfer phenomenon to deal with
squeeze film air damping in low vacuum. However, the results obtained were found to
deviate from the experimental results. Minikes et al. (2005) modified the model of Bao et
al. (2002) to account for the fact that their plate rotated about a pivot rather than
translated up and down as assumed in Baos model. They showed that the modified Baos
model agreed with their measured data. A systematic review of squeeze-film damping
models in the free molecular regime and a comparison of the measured damping ratios
with predictions from the various published models have been presented by Sumali
(2007). He concluded that the measured damping ratios are close to predictions from
models that are based on the Reynolds equation and take into account the inertia of the
gas.
For completeness certain related peripheral literature are presented in this
subsection. The deep reactive ion etching (DRIE) or induction coupled plasma (ICP)
technologies developed in recent years produces devices with thick, perforated plate. The
adequate model for the squeeze-film air damping of the perforated plates becomes
desirable. Squeeze-film damping of thick hole-plates were treated by Bao et al. (2003a,b)
and Kwok et al. (2005) utilizing the Reynolds equation for damping pressure of holeplates. Several FEM based simulations of dynamic behaviour of oscillating perforated
microplates under the effect of squeeze-film damping were developed and compared by
some authors [Som and Pasquale, 2008; Pasquale and Veijola, 2008]. At low pressures
13
Chapter 1
or in ultra-thin films, the molecular interactions with the surfaces also affect the effective
viscosity. The complex interactions at the air-gap surface are characterized through an
experimental parameter, the accommodation coefficient. Veijola et al. (1998) derived a
simple approximate equation for the effective viscosity based on the solution of linearized
Boltzman equation taking into account the molecular interactions. The accommodation
coefficient for the single-crystal silicon surface was experimentally extracted and
agreeably compared with the reported values. The coupling effects of surface roughness
(characterized by Peklenik number) and gas rarefaction for ultra-thin films has been
analyzed by Li and Weng (1997) and Chang et al. (2002) using the so-called modified
molecular gas film lubrication (MMGL) equation. For small gaps and low operating
frequencies, the squeeze-film damping effects in MEMS devices can be modelled using
the Reynolds equation. For large gaps and high operating frequencies, inertia forces need
to be considered, pressure boundary conditions are nontrivial, and fluid conditions are no
more isothermal. Under such conditions several authors [Veijola, 2004; Maali et al.,
2005; De and Aluru, 2006] have resorted to the simplified as well as full versions of the
Navier-Stokes equations. In the commercial FEM based software CoventorWare, a hybrid
approach called Navier-Stokes-Reynolds (NSR) is used to extract the damping
characteristics. An overview of modelling and simulation of squeeze-film damping
effects in MEMS for rigid plate problems has been presented by Bao and Yang (2007).
1.2.3
14
Introduction
1.2.3.1. Pure DC Voltage Actuation
When DC voltage is applied gradually, evaluation of the static deflection and the
static pull-in conditions are of great interest. Experimental investigations have
encouraged several modelling approaches to estimate the static pull-in parameters. From
simple lumped models, on one hand, to powerful 3-D numerical simulations based on
FEM/BEM schemes, on the other hand, have been established to elucidate the pull-in
behaviour. Osterberg et al. (1994) proposed two models, namely, a one-dimensional (1D)
lumped parallel plate spring model and a 3D simulation model to predict the static pull-in
behaviour of cantilever structures. Zavracky et al. (1997) fabricated and experimented
with cantilever based micromachined switches. They calculated the static deflections for
different DC bias voltages by iteratively solving the fourth order differential equation
describing the small deflection model. They showed that two solutions exist at each DC
voltage, and that both solutions coalesce at pull-in voltage. They also inferred that one of
them is unstable since it gives a maximum deflection beyond the size of air-gap. The
static pull-in parameters were also evaluated using a lumped model. They concluded that
the lumped model is inaccurate since it predicts the pull-in voltage at a maximum
deflection near one-third the air-gap, while the numerical solution predicts a higher value
of the pull-in air-gap. Choi and Lovell (1997) analysed doubly-clamped beams using
closed-form solutions as well as numerical integration. They concluded that the inclusion
of mid-plane stretching effects due to large deflection of the beam resulted in decrease in
the static displacements under electrostatic loading. Computer aided design systems
MEMCAD (currently known as CoventorWare) was developed [Senturia et al., 1992] to
simulate electrostatic MEMS. The mechanical analysis was performed using the
commercial FEM software ABAQUS, while the electrostatic force was computed using
boundary element method (BEM) based program FASTCAP. MEMCAD starts by
computing the electrostatic force on the undeformed geometry which is then used to
deform the microbeam. Then the electrostatic force on the deformed geometry is
recomputed and used to re-deform the structure, and so forth. This required re-meshing
and re-computation of interpolation functions during each iteration. An automated
procedure to generate macromodel from a 3-D FEM simulation was presented by Gabbay
et al. (2000). This procedure is not suitable for problems involving large displacements as
the mid-plane stretching effect for the clamped-clamped beams was neglected. Finite
cloud method (FCM), a meshless technique, was used by Li and Aluru (2001) to simulate
linear and non-linear static behaviour in electrostatic MEMS. Geometric nonlinearity was
15
Chapter 1
considered while using a total Lagrangian description to develop the governing equations
for large deformation analysis. Effect of different geometric variables on the static pull-in
voltage was presented for both clamped-clamped and cantilever beams. They concluded
that for clamped-clamped beams, linear theories hold comparatively over a small
deflection regime than for cantilever beams.
A closed form expression for the static pull-in voltage has been derived by
Pamidighantam et al. (2002) starting from the known expression of a lumped model.
Though nonlinear stiffening effects due to large deflection has been considered for
clamped-clamped beam, the effective stiffness for cantilever case is valid only for small
deflections. The expression of effective stiffness used in the model is derived assuming a
uniformly distributed electrostatic load. Moreover, the beta correction coefficient used by
the authors is not easily determined directly. Abdel-Rahman et al. (2002) emphasized the
efficacy of accounting for the mid-plane stretching of the beam in order to treat large
deflections of clamped-clamped beams. The stable equilibria were obtained by
numerically solving the boundary-value problem utilizing a shooting method. The model
successfully highlights the importance of including the deflection distribution, geometric
nonlinearities and mid-plane stretching in the analysis to avoid underestimation of the
stability limits. They also concluded that both axial force and mid-plane stretching have a
stiffening effect on the microbeam while the electrostatic forces introduced a softening
effect. The approach of Chowdhury et al. (2005) is based on a linearized uniform
approximate model of the non-linear electrostatic pressure and the load deflection model
of a cantilever beam under uniform pressure incorporating fringing field corrections.
They introduced a compensation factor to compensate for the errors arising out of noninclusion of the geometric nonlinearity and the linearization of the electrostatic forces.
Pull-in voltage results obtained through experimental measurements by Hu et al.
(2004) were compared to the analytical model based on small deflection and linearized
electrostatic forces. An energy approach was used to derive the system governing
equations and the Galerkin method of decomposition was used to obtain the ROM. Zhang
and Zhao (2006) used a ROM to evaluate the static pull-in parameters of electrostatically
actuated cantilever, clamped-clamped beams and simply-supported plate under variable
axial load. A common formulation accounting for mid-plane stretching was used to treat
large deflections for clamped-clamped as well as cantilever beams. The mode shapes used
in the Galerkin method of discretization to formulate the ROM were computed separately
for each axial load. In low range of axial load, the pull-in voltage results obtained with the
16
Introduction
ROM retaining single mode and five modes, respectively, in the discretization procedure
were found to agree well. Chao et al. (2006) developed a ROM to predict the static pull-in
parameters for a square microplate. Retaining only the first mode in the Galerkin method
of discretization, the obtained numerical results were found to agree well with the FEM
and experimental results. The fifth-order Taylor series representation of electrostatic force
generated pull-in results close enough to those obtained with the full-order representation.
Generalized differential quadrature method (GDQM) has been used by Sadeghian et al.
(2007) to study the pull-in behaviour of both clamped-clamped and cantilever based
MEMS switches. Various effects like intrinsic residual stress, axial stress due to beam
stretching, fringing field, stress gradient and trapezoidal cross-section were considered.
They concluded that for clamped-clamped beams the pull-in voltage is most affected by
the presence of residual stress. Recently, Ahmad and Pratap (2010) analysed the static
deflection and pull-in of thin clamped circular plates neglecting the stretching effects due
to large deflection. Simple analytical expressions for static deflection were obtained using
Galerkin-weighted residual technique, and the analytical results were found to compare
well with the ANSYS simulations except for higher DC voltages.
Legtenberg et al. (1997) proposed the use of curved electrodes to determine the
characteristics of large-displacement actuators. A 3-D coupled electromechanical model
was presented to understand the interaction between the deflection of the actuator and the
electrostatic force. The simulated results were compared with the experimental data in
this paper. Kuang and Chen (2005) studied the nonlinear electrostatic behaviour for
curved electrodes and shaped cantilevers using DQM. Their model included nonlinear
curvature for the inextensible cantilever but incorporated the parallel plate approximation
for the electrostatic domain. The electrostatic pull-in results for various actuator
structures, such as cantilever beams, fixed-fixed beams, and clamped diaphragms, have
been adopted to determine the Youngs modulus and the residual stress in the microactuators [Osterberg and Senturia, 1997]. Pull-in voltages and capacitance-voltage
measurements along with 2-D simulations of electromechanical systems have also been
made to explicate the material properties of electromechanical structures [Chan et al.,
1999].
With the DC voltage applied as non-smooth function of time as in case of
applications like microswitches or micromirror positioning, transient behaviour needs to
be studied. Pull-in under dynamic conditions is referred to as the dynamic pull-in and is
affected by inertia and damping effects. Rocha et al. (2004) identified a metastable region
17
Chapter 1
just beyond the pull-in displacement for over-damped electrostatic devices. They used a
lumped parameter model of the MEMS device incorporating a gap dependent damping
coefficient representing the linearized squeeze-film damping. Pull-in time measurements
in presence of external acceleration and for step voltage amplitudes close to the static
pull-in voltage were compared with the simulation results. Lumped parameter models
were used by Nielson and Barbastathis (2006) for analytical and numerical analyses of
both parallel plate and torsional actuators. For the ideal case of no damping, significant
decrease in the pull-in voltage was obtained for an applied step voltage. Static and
dynamic pull-in conditions were analytically examined by Fargas-Marques et al. (2007)
using energy analysis of a lumped parameter model. Under dynamic conditions, reduction
in stable voltages was obtained whereas the travel range was much extended. Leus and
Elata (2008) analysed the undamped response of parallel plate actuators driven by step
voltages using an energy method. From the experimentally validated results, they
concluded that the cycle time of periodic response and the switching time of non-periodic
response are linearly related to the applied voltage on a semilog and log-log scale
respectively. Liu et al. (2008) presented identification of system parameters of a
capacitive dual-backplate microphone. The lumped model involved linear and cubic
stiffness, linear viscous damping, and nonlinear electrostatic force with up to third-order
terms retained in a Taylors series expansion. The stiffness parameters were extracted
from finite element analysis using the commercial package CoventorWare. An
approximate solution of the nonlinear dynamic equation was obtained using a multiple
scales method. The solution was subsequently used to obtain the system parameters by
nonlinear curve-fitting of the measured transient data. Kim et al. (2010) experimentally
evaluated the performance of a resistive microswitch consisting of moving membranes
and mechanical springs. The pull-in actuation voltage was shown to be much reduced by
modifying the fabrication process in order to reduce the thickness of mechanical spring
only, while maintaining the switch membrane thickness. The deviations from the
numerically simulated results were attributed to fabrication errors.
For flexible structures vibrating even in the first mode of vibration, the elasticity
equation that gives the dynamic displacement under electrostatic load is coupled to the
Reynolds equation. Pull-in time of a pressure sensor modelled as a clamped-clamped
microbeam was studied by Hung and Senturia (1999a) using coupled Euler-beam and
nonlinear Reynolds equations. They generated a macromodel by discretizing the two
equations using the Galerkin method. They used few runs of a slow finite difference
18
Introduction
method to generate the global basis functions required in the Galerkin formulation. The
results were validated with experiments and finite difference simulations. Behaviour of
cantilever microswitches under step voltages, including contact bounce, was analyzed by
McCarthy et al. (2002) using finite difference method and was agreeably compared with
the experimental results. The model used Euler-Bernoulli beam theory for cantilever
beam, included a simple spring model of the contact tips, and used nonlinear
compressible Reynolds equation to represent squeeze-film damping. They also presented
the simplified case of a uniform cross-section switch neglecting the flow compressibility
effects. The microswitch modeled as a clamped-camped microbeam actuated by step
input was studied by Xie et al. (2003) using invariant manifold approach. The model
accounts for geometric nonlinearity associated with large deflection. The undamped
behaviour obtained numerically is favourably compared, except near pull-in, with those
obtained from the finite difference method. Full-Lagrangian based relaxation and Newton
schemes for dynamic analysis of electrostatic MEMS was presented by De and Aluru
(2004). The highlight of the method is elimination of surface re-meshing and recomputation of interpolation functions before each electrostatic analysis thereby making
the method very efficient and fast. For highly nonlinear structure-electrostatic coupling,
Newton scheme was shown to over perform relaxation method. Rochus et al. (2005) used
a FEM approach to simulate the dynamic behaviour of a clamped-clamped beam type
resonator under a suddenly applied voltage. It was shown that the system may become
unstable before the static pull-in voltage due to the dynamical effects. Pursula et al.
(2006) presented a FEM based sequential method to solve the coupled equations
governing the transient response of a MEMS accelerometer consisting of a proof mass
suspended at the ends of two cantilevers. The structural, electrostatic, and damping
models were implemented as separate modules in open source finite element software
Elmer. The differences between the model and experiment were obtained larger with fast
actuation speeds than with slower ones.
A ROM was developed by Younis et al. (2003) to investigate both static and
dynamic pull-in behaviour of a clamped-clamped beam based microdevice. The
macromodel uses few linear-undamped mode shapes of a microbeam in its straight
position as basis functions in a Galerkin procedure. The model accounts for the mid-plane
stretching and the electrostatic force is represented exactly in the discretization procedure.
They observed that numerical results obtained using even number of modes did not
converge. Using five modes for discretization, pull-in time of a pressure sensor was
19
Chapter 1
validated with the experimental results. An elastic cantilever coupled to a plate at the free
end was studied by Krylov and Maimon (2004) to investigate the static and transient
behaviour in the presence of squeeze-film damping. The boundary value problem was
formulated taking into account the rotational inertia of the plate attached to the free end of
the microcantilever. The nonlinear incompressible Reynolds equation, assuming non-slipflow conditions, was used to model squeeze-film damping. A ROM, with three modes
used in the Galerkin discretization procedure, predicted the transient dynamics revealing
good agreement with the experimental data. They observed that when small vibrations of
the beam around its underformed state are considered, the contribution of the squeezefilm damping is small and the model of linear damping is adequate. Near the instability
point, the squeeze-film damping dominates and the system is overdamped. Further they
concluded that a high level of damping leads to an increase in the voltage required to
cause dynamic instability while the region of stable deflection shrinks. The Lyapunov
exponents had been used by Krylov (2007) to indicate dynamic pull-in instability of
double clamped microbeam actuated by a suddenly applied voltage subjected to nonlinear squeeze-film damping. Static analysis results obtained with the ROM were
compared with the numeric solution of the boundary value problem. The number of
modes to be used in the discretization procedure was shown to be dependent on the
nonlinear parameter representing geometric nonlinearity. The transient behaviour
obtained with the reduced order and finite difference models were in good agreement for
an undamped beam. Lin et al. (2007) focused on the construction of a ROM based on a
modal projection method to deal with electrostatically actuated rectangular microplates.
During the construction of the ROM, the energy functions are calculated by several runs
of finite element analyses and a subsequent nonlinear function fitting scheme. Though the
model can take into account the effect of residual stress, this method is limited to energy
conservative systems.
Introduction
Tilmans and Legtenberg (1994) solved the small deflection electromechanical
problem by applying the energy minimum principle. The derived equation expressed the
resonant frequency as a function of AC excitation amplitude. Comparison of the
calculated results with the experimental results revealed a good qualitative agreement
only. Turner and Andrews (1995) modelled a microbeam as a spring-mass system with a
cubic nonlinearity representing the mid-plane stretching. They studied the effects of
electrostatic force and mid-plane stretching separately. They concluded that an increase in
resonant frequency is caused by mid-plane stretching whereas the resonant frequency
decreases due to the electrostatic effect. Ayela and Fournier (1998) experimentally
studied the response of microbeams of variable geometry to a general electric load
composed of DC and small AC components. For the operating conditions used, the
experimental results revealed softening behaviour for some of the devices, while others
exhibited hardening behaviour. They commented that predicting the nonlinear behaviour
of these devices analytically is quite impossible. Frequency shifting due to
electromechanical coupling was analysed by Brusa et al. (2004) using lumped parameter
approach as well as FEM simulations neglecting damping. Errors introduced due to
linearization of electrostatic forces were found to be fairly high at higher voltages. They
further suggested the possibility of approximating the characteristic frequency versus
voltage by second order curves.
The dynamic behaviour of a torsion micromirror driven by a DC bias voltage and
a small AC voltage as a function of air pressure was studied by Fischer et al. (1998)
experimentally and the results were found to be approximated well by the squeeze-film
theory. Zhang et al. (2004) analysed the quality factors of a doubly clamped beam
microresonator. The compressible Reynolds equation linearized about the undeflected
position of the beam was coupled to the Euler-Bernoulli beam equation to study the
characteristics of the resonator actuated by small amplitude AC excitations. The squeezefilm damping effects were characterized by two dimensionless parameters completely
determined by the physical properties of the beam and gas. Nayfeh and Younis (2004)
studied the effects of squeeze-film damping on the resonant frequency and quality factor
of a microplate resonator actuated by small DC and AC excitation. They used the 2-D
compressible Reynolds equation, linearized around the undeflected position of the plate,
coupled with the small displacement 2-D plate equation to predict the quality factors
under wide range of gas pressures. They used the model of Veijola et al. (1995) to
calculate the effective coefficient of viscosity to account for the rarefied gas effects.
21
Chapter 1
Singular-perturbation method was used to derive an analytical expression for the pressure
distribution in terms of the structural mode shapes, and the resulting linear damped
eigenvalue problem is solved using FEM. The simulated quality factor values agreed well
with the experimental results. They added that the singular-perturbation method is not
applicable to cantilever microbeams. Gaillard et al. (2006) presented an alternative
measurement technique for cantilever based MEMS with fully electrical actuation and
detection. They observed that the resonance frequency of the cantilevers for small
amplitude oscillations can be tuned by adjusting the DC and harmonic AC amplitudes
without appreciably affecting the quality factor under ambient operating conditions.
Pandey and Pratap (2007) obtained the analytical and numerical values of
damping ratios for a MEMS cantilever resonator, actuated by very small DC and AC
voltages, under the effect of squeeze-film damping and compared them favourably with
the experimental results. The linearized compressible Reynolds equation coupled with the
elastic beam equation was solved using Greens function approach. However, they
neglected the static deflection due to DC bias voltage. Subsequently, they solved the
coupled structural-fluid problem for several flexural modes of vibration using the modal
projection technique available in the commercial FEM package ANSYS. They concluded
that quality factors can go up appreciably for higher modes of vibration of the resonator.
Li et al. (2007) presented a model for calculating squeeze-film damping of electrically
actuated microbeam under static bias deflection. The model does not consider the
dependency of the effective viscosity on the gap height which is no longer uniform for a
statically deflected beam. Moreover, the ANSYS / FLOTRAN model is performed on the
steady-state deflection calculated separately utilizing a numerical method. Using the
model, they obtained the quality factor from the decay rate of the step responses and
verified the results with that of McCarthy et al. (2002). The problem of large DC bias and
small AC voltage actuation of microplates under different gas pressures was studied by
Younis and Nayfeh (2007). The static bias deflection for the microplate was obtained
using the beam model and a perturbation method was used to obtain the pressure
distribution. The coupled dynamic equation was solved using a finite element method and
the quality factors were compared with those obtained using the coupled field analysis
software ANSYS. Poor agreement in the quality factors was observed at small values of
DC voltages. They indicated that the use of a single value for the quality factor in
modelling electrostatically actuated MEMS devices for different DC loadings might lead
to erroneous results. Chakraborty et al. (2010) experimentally analysed the resonant
22
Introduction
frequency and quality factor as a function of dimensions of microcantilevers for small
sinusoidal AC excitation. The results were compared with that obtained through simple
analytical models and CoventorWare simulations. While the frequency results are in
reasonable compliance, mismatch was observed in the quality factor values.
Li et al. (2008) studied the static and dynamic behaviour of microplates subjected
to non-classical boundary conditions and unsymmetrical electrostatic forces due to
multiple actuators. Non-classical boundary conditions are modelled using rotational and
translational springs at the edges of the microplate. An energy method using boundary
characteristic orthogonal polynomials was applied to formulate the equations of motion of
the microsystems. Based on this method, influence functions were built and least squares
method was used to optimize the desired deflection under electrostatic forces. The
Rayleigh Ritz method was used to obtain the response frequencies. The results under
different sets of bias voltages were verified against ANSYS simulation based on a ROM
method. Ballestra et al. (2008) studied the influence of residual stress gradient on
frequency shift and pull-in for curled gold microcantilevers actuated by DC and small AC
voltages. Residual stress gradient was determined by measurement of deflection profiles
through a non-contact interferometric profilometry system and subsequently calculating
the slope and curvature of the cantilever. Frequency shift variations with DC voltage were
experimentally obtained and compared with the results of implemented ANSYS models.
The effect of initial upward curvature and hence the residual stress gradient was
accounted for in ANSYS by building a model with increasing initial gap. The stress
gradient was found to introduce a hardening effect in contrast to the softening effect of
the electrostatic forces. Batra et al. (2008) studied the small amplitude vibrations of
undamped clamped-clamped narrow microbeams predeformed by a DC bias voltage. The
elctromechanical model accounts for axial strain associated with large deflection and the
electrostatic fringing field due to both the finite width and the finite thickness of the
microbeam. The lowest frequency is extracted with a simple and computationally
efficient one degree-of-freedom model obtained by approximating the deflection field
with the static deflection of a clamped-clamped microbeam loaded by a uniformly
distributed force. The models predictions are in good agreement with those from threedimensional finite-element ANSYS simulations based on a ROM method. They
concluded that for relatively large gaps both the axial strain and the fringing field effects
are important and neglecting one of them may result in improper estimates of both the
pull-in parameters and the resonance frequencies.
23
Chapter 1
Characterization of complex nonlinear dynamic behaviour of MEMS actuated by
large DC and AC excitations in the presence of nonlinear squeeze-film damping is quite
involving. Various nonlinear phenomena like superharmonic and subharmonic
resonances, jump phenomenon and frequency hysteresis, period-doubling bifurcations,
chaos, and dynamic pull-in instabilities have been reported. An early experiment with
electrostatic cantilever microresonators was carried out by Jin and Wang (1998). They
observed primary as well as second order superharmonic resonances and found the
superharmonic resonance peak lower than that of primary resonance. They further
concluded that the superharmonic resonance amplitude is in direct proportion to the
square of the AC voltage and has no relationships with the DC bias when the bias is
small. Wang et al. (1998) presented numerical simulations with experimental validation
of the period-doubling route to chaos for an electrostatic tuning actuator. The lumped
model incorporated both linear and cubic stiffness, and linear damping. By proper control
of the actuation voltage, the electrical force generated from the left and right comb drives
of the actuator was shown to bring the system to chaotic steady state. Using energy
increment approach, Luo and Wang (2004) investigated the chaotic frequency band for an
undamped lumped model of the resonant gate transistor developed by Nathanson et al.
(1967). They further concluded that the lower-order resonant motion can be easily sensed
as the lower-order resonant motion has a higher energy than the higher-order resonant
motions. Rocha et al. (2006) simulated and experimented with parallel plate actuators to
demonstrate that through proper choice of amplitude and frequency of the driving
sinusoidal AC voltage, stable displacements can be obtained beyond static pull-in
limitations. The phase portraits generated using the lumped model were analysed to
underline the importance of initial conditions and the external drive as influential aspects
of dynamic pull-in. They further concluded that damping enhances the limits of stable
zone.
Mestrom et al. (2008) compared the experimental observations with the
numerically simulated results of linearly damped lumped parameter model of a doubly
clamped beam microresonator. In the model, mechanical nonlinearity was represented by
fourth-order spring stiffness while a third-order Taylor series approximated the
electrostatic nonlinearity. The frequency response curves revealed jump and hysteresis
near the fundamental frequency with a noteworthy influence of AC amplitude on the size
of the frequency hysteresis interval. Subharmonic resonance of order one-half preceded
by period doubling bifurcation of period two was also observed. Mestrom et al. (2010)
24
Introduction
further extended this study using a ROM derived from the distributed model. The
distributed model coupled in the mechanical, electrical, and thermal domains considered
shear deformation and rotary inertia effects for thick beams. Alsaleem et al (2009)
numerically simulated a lumped parameter model of a microcantilever attached to a proof
mass at its end using both finite difference method (FDM) and long-time integration
approach. Simulations and experiments conducted under low pressure revealed large
primary and subharmonic deflections with softening behaviour while superharmonic
resonance exhibited almost linear behaviour with the smallest response amplitude. Longtime integrations for several initial conditions were carried out to obtain basins of
attraction of stable solutions near primary resonance. Alsaleem et al. (2010) further
conducted a series of experiments on several microstructures to demonstrate nonlinear
phenomena like fractal escape, secondary resonances, and dynamic pull-in. They
concluded that sweeping the AC frequency or the AC amplitude generates almost the
same results. Experimentally obtained pull-in bands were compared with the inevitable
escape bands determined through numerical simulation of a simple lumped model.
Nayfeh and Younis (2005) studied secondary resonances of a clamped-clamped
beam microresonator using a combination of a shooting technique and a ROM obtained
through Galerkin decomposition procedure. The distributed model incorporated axial
force, mid-plane stretching and considered linear viscous damping. They observed that
under dynamic loading conditions, the pull-in instability occurs at a much lower voltage
than under static loading conditions. Though pull-in instability for second order
superharmonic resonance was obtained at certain AC amplitude, subharmonic resonance
of order one-half once activated exhibited pull-in regardless of the magnitude of the AC
forcing. Najar et al. (2006) simulated the dynamic behaviour of doubly clamped
microactuators with the cross-section of the beam varying along the length of the beam.
The electrostatic nonlinearity was represented in two separate forms: firstly, by a thirdorder Taylor series expansion, and secondly in its exact form. A combined DQM-FDM
discretization technique was used to generate frequency-response curves for various
microstructure geometries and different voltages. They concluded that a Taylor-series
approximation of the electrostatic force breaks down for large motions and produces
qualitatively erroneous frequency-response curves. They further concluded that the
microactuator response is much more sensitive to variations in the gap size than the
variations in the microbeam cross-section dimensions. Effects of quadratic and cubic
nonlinearities due to nonlinear electrostatic forces and nonlinear damping were
25
Chapter 1
investigated by Liqin et al. (2006) for a clamped-clamped microswitch. A ROM was
derived from the distributed model retaining two eigenmodes in the Galerkin method of
discretization and expanding the nonlinear terms into a Taylor series. For large AC
amplitudes and low frequencies, appearance of period doubling bifurcation and inverse
period doubling bifurcation was reasoned to lead the system to an unsteady state and even
to chaos. De and Aluru (2006) investigated the effects of electrostatic, mechanical, and
fluidic nonlinearities on electrostatically actuated microstructures, namely, a fixed-fixed
microbeam and a MEMS torsion mirror. Full-Lagrangian based relaxation and Newton
schemes were presented to simulate period doubling route to chaos under superharmonic
excitations. A mass-spring-damper model retaining only electrostatic nonlinearity was
numerically simulated to indicate that the nonlinear electrostatic force is the underlying
physical force responsible for the observed dynamic properties. The mechanical and
fluidic nonlinearities present in the system were shown to have a stabilizing effect on the
microsystem.
The pull-in dynamics near the primary resonance of a doubly clamped microbeam
resonator for variations in initial conditions were studied by Nayfeh et al. (2007). The
system was shown to jump to pull-in at bifurcation points irrespective of the initial
conditions. Alternatively, under certain operating conditions, the transient dynamics of
the system determined whether the system goes to pull-in or settles to a stable orbit. They
have also demonstrated that the effective nonlinearity of the resonator changes from
hardening to softening type with increase in the level of voltage actuation. However, they
neglected the fluidic nonlinearities. Park et al. (2008) used FEM to model a doubly
clamped beam microresonator under linearized incompressible squeeze-film damping
conditions. Period-doubling bifurcations leading eventually to chaotic oscillations were
simulated by increasing the AC amplitude while holding the DC voltage just below the
pull-in value. They observed that under a control perturbation, chaos can be converted
into periodic motion and at the same time energy enhancement of the MEMS resonator
can be achieved. Ouakad and Younis (2010) investigated, using a perturbation technique
and the long-time integration of a ROM, the non-linear resonance frequency of a linearly
damped clamped-clamped arch as a function of the AC harmonic amplitude, the DC load,
and the arch initial rise for small values of AC and DC loads. Parallel plate approximation
of electrostatic force was followed assuming a shallow arch. Eigenmodes of the straight
beam and the exact mode shapes of the unactuated arch used separately as the basis
functions in the ROM generated agreeable results. Dynamic snap-through motion for a
26
Introduction
certain band of frequency near primary resonance was detected with long-time integration
approach. The work was further extended [Younis et al., 2010] to simulate results near
primary, subharmonic, and resonances near higher order harmonics.
Liu et al. (2004) studied the lumped model of a MEMS cantilever system under
the effect of nonlinear electrostatic actuation and a separate nonlinear capacitive sensor.
Utilizing Poincare maps, they have shown bistability and Hopf bifurcation for closed loop
controlled cantilever, and, observed period doubling route to chaos for both open and
closed loop control. Zhao et al. (2005) presented the transient as well as harmonic
response of an electrostatically actuated torsional micromirror considering the coupled
effect of torsion and bending. The dynamic behaviour of the lumped model incorporating
electrostatic nonlinearity was investigated using the Runge-Kutta method. The transient
results obtained with the lumped model compared well with the ANSYS simulations well
below the pull-in voltage. The period and amplitude of the responses were found to
increase with increase in applied voltage. Superharmonic resonances up to fourth order
were numerically simulated. Krylov et al. (2005) derived a linear parametric equation to
study the small amplitude dynamics around the equilibrium position of a microbeam
electrostatically actuated by two symmetrically located electrodes. Parametric
stabilization of microstructures was observed under symmetric actuation by DC and AC
voltages with the DC voltage exceeding the pull-in value. Harish et al. (2009) used a
lumped model of a micro-ring resonator to characterize parametric resonance of first
order and parametric amplification under combined harmonic and parametric excitations.
The stability boundary corresponding to the principal parametric resonance has been
determined experimentally from the upward frequency sweep and has been shown
to compare well with the theoretically predicted boundary. Jump features were observed
at high values of parametric excitation. Zhu et al. (2009) considered a large array of
interacting microcantilevers to study the effects of elastic nonlinearity and nonlinear
coupling between the adjacent microbeams. Analysis of the linearly damped lumped
parameter model of the array using harmonic balance method in conjunction with Newton
iteration method suggested the existence of high-order parametric subharmonic
resonance. Haghighi and Markazi (2010) analysed the lumped model of a
micromechanical resonator actuated from both sides to evaluate the chaotic dynamics.
Results of the numerically simulated problem were compared with the Melnikov analysis
of weakly nonlinear problem derived for small AC amplitudes. A control strategy to
27
Chapter 1
eliminate chaotic motion and take the system response to a stable orbit was evaluated
through numerical simulations.
1.2.4
research which needs to be further explored are identified and described as follows:
28
Introduction
order forms. An improved distributed model incorporating the nonlinear terms in
the exact forms needs to be formulated to further evaluate the nonlinear dynamics
of microbeams actuated by large electrostatic loads.
1.3
microbeams is carried out in the present thesis work. Static behaviour and transient
dynamics are analysed for pure DC voltage actuations. Behavioural analysis of
deformable cantilever electrodes separated from the ground electrode by large gaps is
taken up with special emphasis on the nonlinear effects due to geometry, electrostatic
forces, and inertial terms. The non-linear curvature and the von Krmn non-linearity in
the axial strain-displacement relationship are incorporated to account for the geometric
non-linearity of the microcantilever associated with large deflection. Higher order
correction terms are applied to the parallel plate formulation to model the electrostatic
domain for large gap separations. Analysis for large gap separation problem neglects any
form of damping in the system.
The present work also focuses on the linear and nonlinear dynamics studied under
combined DC and AC voltage actuations. The entire analysis in this part of work is
restricted to the range of initial gap-length ratios, as concluded from the large-gap
analysis discussed above, which validates the application of parallel plate nonlinear
electrostatic model in conjunction with the small deflection theory for the structural
model. Small amplitude motions under the influence of large DC load coupled with small
AC harmonic voltage are treated with due consideration of the static deflection due to the
large DC bias. The small amplitude linear dynamic behaviour is analysed in the presence
of linearized squeeze-film damping accounting for flow rarefaction effects. The large
amplitude dynamic behaviour of the microcantilever under the combined effect of DC
bias and harmonic AC excitation is studied next in the primary and superharmonic
frequency range. The model is formulated accounting for the nonlinearities of the system,
arising out of electrostatic forces and the squeeze-film damping terms. Full-order
representations of the nonlinear terms are used in the present work.
The entire investigation in the present thesis is based on numerical simulations of
reduced-order models (ROM). The set of governing equations of motion of the system is
obtained by the application of Hamiltons principle to the energy functions, derived
29
Chapter 1
conforming to the microsystem under consideration. For the case of damped systems, the
equation of motion is coupled to the Reynolds equation to capture the characteristics of
squeeze-film damping. From the boundary value problem, a ROM is developed using the
linear undamped mode shapes of the undeflected microcantilever as a basis set in the
Galerkin formulation. The ROM with the correct number of modes is then used to
numerically simulate the static and large amplitude dynamic behaviour. For cantilever
microbeams actuated by large DC bias and small AC excitation, small amplitude motions
about the statically deflected position of the beam is analysed using an easy to implement
semi-analytical approach.
1.4
30
Introduction
flow compressibility is studied. A comparison of the squeeze-film characteristics obtained
with compressible and incompressible flow models is also presented.
Nonlinear dynamic behaviour of a cantilever microbeam actuated by a
combination of DC and AC harmonic loading in presence of squeeze-film damping is
presented in chapter four. From the coupled field governing equations, a ROM is
developed and numerically simulated to study the dynamics for the fundamental flexural
mode of vibration under primary and superharmonic AC excitations. Beginning with a
static analysis, several nonlinear phenomena are illustrated using frequency-response
curves, time-histories, and phase portraits as effective analysis tools.
Finally in chapter five, conclusions from the present study and recommendations
for future work are included.
1.5
31
Chapter 2
STATIC AND TRANSIENT BEHAVIOUR UNDER A DC
ELECTROSTATIC FORCE
Outline of the chapter: 2.1 Introduction, 2.2 Model Description, 2.3 Boundary Value
Problem, 2.4 Galerkin Formulation, 2.5 Results and Discussion, 2.5.1 Static Analysis,
2.5.2 Simulation of Static Behaviour, 2.5.3 Simulation of Transient Dynamics, 2.6
Summary
2.1
Introduction
Typical MEMS structures consist of thin beams with cross-sections in the order of
microns and lengths in the order of ten to hundreds of microns. Efficient electrostatic
actuation at reasonable actuation voltages is achieved for the size of the devices such that
the electrostatic gap to beam length is typically of the order of 102 103 . Further
reduction in size of the devices, and, low actuation voltages operations have been made
possible by the development of new materials and advancement in fabrication
technologies. In such devices, the length of the electrode is typically of the order of
several micrometers with the gap-length ratio not necessarily small and can be typically
of the order of 101 102 [Teva et al., 2004; Decuzzi et al., 2007] or even larger
[Sazonova et al., 2004]. For small gap-length ratios, parallel plate approximation of the
electrostatic forces and small deflection assumption for deriving the mechanical model
are justified. Parallel plate electrostatic model assumes the deformable electrode to be flat
and locally parallel to the rigid ground electrode. However, slope and curvature of the
deformable electrode need to be considered while modelling the electrostatic forces for
large electrostatic gaps. Moreover, the structural geometric nonlinearity due to large
deformation effect is quite considerable in microstructures because of their low mass and
high flexibility. Thus, for relatively larger gaps, a model involving geometric nonlinearity
and a more accurate approximation of the electrostatic forces needs to be derived.
The overall literature review related to this field is already presented in the
preceding chapter. Although several studies have investigated the behaviour of
microbeams under a DC electrostatic force, few models have accounted for geometric
33
Chapter 2
nonlinearity associated with large deflection. Geometric nonlinearity due to mid-plane
stretching effect was incorporated in the models [Choi and Lovell, 1997; Li and Aluru,
2001; Abdel-Rahman et al., 2002; Xie et al., 2003; Krylov, 2007] to demonstrate that
linear theories hold over a small regime and grossly underestimate the stability limits.
However, literature investigating the effects of nonlinear curvature and inertia
nonlinearities on the response of electrically actuated cantilever microbeams separated by
large gaps is scarce. Hirai et al. (2000), and, Kuang and Chen (2005) developed models
of the shaped cantilever beam and the curved electrode for static analysis of electrostatic
actuators. The formulations incorporate nonlinear curvature for the inextensible cantilever
to model geometric nonlinearity effects but considered parallel plate approximation for
the electrostatic domain. Ding et al. (2007) studied the nonlinear static response of
inextensible cantilever force-sensing elements as a combined result of large deflection
and non-vertical loading. They considered the large nonlinear bending effect. Mahmoodi
and Jalili (2007) studied the nonlinear flexural vibrations of piezoelectrically driven
microcantilever beams. The model included cubic nonlinear inertia and stiffness terms
along with quadratic and cubic nonlinear terms due to piezoelectric effect.
In this chapter, the static and dynamic pull-in behaviours of undamped wide
microcantilevers separated by relatively larger gaps and acted upon by DC electrostatic
forces are studied. The boundary value problem accounting for the nonlinearities of the
system due to electric forces, the geometry of the deflected beam, and the inertial terms is
derived by applying Hamiltons principle. The compatibility of the electrostatic model,
with the large deflection theory being used for the structural model, has been taken care
off by incorporating higher order correction of the electrostatic forces [Krylov and
Seretensky, 2006]. From the boundary value problem (BVP), a reduced order model
(ROM) is developed utilizing the linear undamped mode shapes of the undeflected
microcantilever as a basis set in the Galerkin method of discretization. The derived ROM
is then used to study the effect of nonlinearities on the static response and the transient
dynamics. The effect of inclusion of higher order electrostatic terms on the static and
dynamic pull-in parameters is analysed. Behaviour is studied as a function of lumped
non-dimensional design parameters representing, respectively, the strength of
nonlinearities and the strength of electrostatic forces. For large initial gap to beam length
ratios, the large deflection model is observed to be more effective in and around pull-in
zone.
34
2.2
Model Description
Figure 2.1 illustrates a perfect conductor clamped at one end through dielectric support
and suspended over a ground plane. The upper electrode is acted upon by attractive
electrostatic force which is non-uniformly distributed along the length due to the
redistribution of the charges as the beam deflects. Even at small voltages, the tip
deflection will be comparable to the air-gap and will be large enough in comparison to the
electrode thickness. The deformable electrodes with thickness much smaller than the
characteristic in-plane dimension can be treated as 2-D plate like bodies [Batra et al.,
2007]. In this work, the electrostatic force is assumed to be uniform across the width and
the deformable conductor is assumed to undergo cylindrical bending deformations and is
treated as a wide beam with the effective Youngs modulus equal to the plate modulus
[Osterberg and Senturia, 1997]. Further, the Euler-Bernoulli beam theory is used to model
the microbeam, and accordingly the effects of warping and shear deformation are
neglected. Under large rigid-body rotations, structures like cantilever beams undergo
large deformations but small strains, and the beam is modelled by incorporating the von
Krmn nonlinearity in the expression for the axial strain and consideration of nonlinear
curvature.
Fringing fields emanating from the lateral and the top surfaces of the deformable
electrode need to be considered while modelling the electrostatic field by accounting for
finite width and finite thickness of the beam. For wide beams with width to air-gap ratio
35
Chapter 2
greater than 1.5 [Rabaey, 1996] and width-thickness ratio greater than 5 [Batra et al.,
2006], the fringing fields are neglected. For flexible structures, the parallel plate
capacitance is usually justified by the smallness of the gap-length ratio typically of the
order of 102 103 [Pelesko and Bernstein, 2003]. The present work formulates the
electrostatic model for relatively larger gaps by incorporating the second order
corrections [Krylov and Seretensky, 2006] which depends on the slope and curvature of
the deformable beam.
2.3
thickness h separated from the ground plane by an initial gap of d0 . When subjected to a
driving DC voltage VDC , the beam undergoes a transverse deflection w ( x, t ) and an axial
extension u ( x, t ) , which are dependent on the position x along the beam length and time
t . Geometric nonlinearity arises from two distinct mechanisms: (1) Green strain relation
that quantifies the extensional deformation of the centroidal plane of the beam, and, (2)
the nonlinear curvature of the deflection curve. The axial strain xx [Hsieh et al., 1994]
associated with the material located at the neutral axis is given by
1
2
2
ds dx u ( x, t ) w( x, t ) 2
= 1 +
xx =
+
1
x x
dx
(2.1)
u w
1 + +
=1
x x
(2.2)
Following the formulation derived by Hodges (1984) and as shown in Figure 2.1, the
exact expression for the nonlinear curvature k can be given by
2
2
u w u w
1+ 2 2 .
x x
x x
k=
=
3
s
u 2 w 2 2
1 + +
x x
(2.3)
(2.4)
36
1l
E.I .k 2 .dx
20
(2.5)
T=
1 l w
1 l u
A dx + A dx
2 0 t
2 0 t
(2.6)
1+
.
. ( d 0 w ) dx
U e = 0bVDC
2
3 d 0 2 x
d w)
0 ( 0
(2.7)
where permittivity constant for free space, 0 = 8.854 1012 , Fm 1. It may be noted that
Eq. (2.7) incorporates the corrections [Krylov and Seretensky, 2006] up to the second
order which depends on the slope and curvature of the deformable beam.
t2
governing equations as well as the boundary conditions can be derived. Using Eqs. (2.4),
(2.5), (2.6) and (2.7), and retaining all nonlinearities up to O ( 3 ) , where x l = O (1) ,
2
2
w
+ EI
2
2
t
x
2
2
w u
2 w 2 w u 2u w
+
+
2
E
I
2
2
x 2
x x x
x 2 x x 2 x
1 0bVDC
2 d w 2
0
( d0 ) 2
1 l 2
3d 0
and,
37
w 2
2 w
+ 2 ( d0 w ) 2
x
x
Chapter 2
2 2
2
2
u
2 w w
w
A
+ EI
+ EI x 2 = 0
2
2
t
x x 2 x
x
(2.8)
2
4
2
w
w
w w w
w x x 2 w w w 2
+ EI
+ EI
+ A
dx.dx
.
2 +
2
4
2
x x x x x
x x l 0 x t x x t
t
x
( ( ))
( )
0
1 0bVDC
1 l 2
=
2 d w 2
3d 0
0
2
w 2
2 w
+
2
d
w
(
)
x 2
x
(2.9a)
w(0, t ) =
w
=0 ,
x x = 0
(2.9b)
2w
3 w
=
=0
x 2 x =l x 3 x =l
Using the non-dimensional variables w = w d 0 , x = x l , t = t s , s = l 2 ( A
EI ) the
xx
+ w + 1 w ( ww ) + 1 w w
w + ( w ) dx.dx
w
10
iv
2
2VDC
(1 w )
1
2
1 3 2 w 2 ww + ( w )
w (0, t ) = w(0, t ) = 0 ,
w(1, t ) = w(1, t ) = 0.
(2.10a)
(2.10b)
38
actual values. The influences become more prominent as the ratio increases and the
results are dictated by the relative values of 1 and the electrostatic force represented by
2
2VDC
. The terms on the right hand side of the equation represent the nonlinear
electrostatic terms. The second order correction of the electrostatic forces which is
represented by the second term on the right hand side of the equation depends on the
slope and curvature of the deformable beam. The influence of the second order correction
is further dependent on the ratio 1 .
2.4
Galerkin Formulation
Modal decomposition [Younis et al., 2003] is performed in this section to
w( x, t ) = ai (t )i ( x)
(2.11)
i =1
where N represents the number of modes retained in the solution. i ( x) is the i th linear
undamped mode shape of the undeflected microcantilever obtained from the following
linear undamped eigenvalue problem of a straight beam
iiv = ( si )2 i
(2.12a)
(2.12b)
61.70 for the third mode, etc. The above equation is derived from the undamped free
vibration problem expressed by Eq. (2.10a) with all the nonlinear terms in the equation
1
39
Chapter 2
Multiplying Eq. (2.10a) by (1 w) 2 , substituting Eqs. (2.11) and (2.12a) into the
resulting equation, multiplying by n ( x) , and integrating the outcome over x = 0 to 1 , the
coupled nonlinear ODEs of the system can be derived as
N
i =1 j =1
+ ai a j ak ( si ) ni jk dx 2 ai a j ( si ) ni j dx
2
i =1 j =1 k =1
i =1 j =1
1
1
N N N N N
N N N
+1 ai a j ak ni jkdx + ai a j ak a p aq ni jk pq dx
i =1 j =1 k =1 p =1 q =1
0
0
i =1 j =1 k =1
N
i =1 j =1 k =1
2 ai a j ak a p ni jk p dx + 4 ai a j ak ni jkdx
i =1 j =1 k =1 p =1
N
+4 ai a j ak a p aq ni jk pq dx
i =1 j =1 k =1 p =1 q =1
N
i =1 j =1 k =1
8 ai a j ak a p ni jk p dx + ai a j ak ( si ) 2 ni jk dx
i =1 j =1 k =1 p =1
+ ai a j ak a p aq ( si ) 2 ni jk pq dx
i =1 j =1 k =1 p =1 q =1
N
i =1 j =1 k =1
2 ai a j ak a p ( si ) 2 ni jk p dx + ai a j ak n ijk dx
i =1 j =1 k =1 p =1
+ ai a j ak a p aq n pq ijk dx
i =1 j =1 k =1 p =1 q =1
N
i =1 j =1 k =1
+ ai a j ak a p aq n pq ijk dx
i =1 j =1 k =1 p =1 q =1
1
N N N N
2 ai a j ak a p n p ijk dx
i =1 j =1 k =1 p =1
0
1
N N
1
1 N 1
2
2VDC
n dx 2 ai nidx 2 ai a j ni j dx
3 i =1 0
i =1 j =1
0
0
1
N N
+ ai a j ni j dx
i =1 j =1
0
= 0,
n = 1, 2, ...., N
(2.13)
40
xx
It should be noted that the above method of treating the electrostatic force term ensures its
exact representation. In the process, the mass matrix of Eq. (2.13) no longer remains
diagonal as obtained in the literature [Krylov and Maimon, 2004; Krylov, 2007].
2.5
BVP. The model is then validated for the static pull-in voltage and the static pull-in
deflection. The convergence of the model obtained through Galerkin decomposition
method is studied next by comparing the results with the static solution of the BVP. The
ROM with the correct number of modes is then used to study the dynamic behaviour
under a suddenly applied DC voltage. The transient results are validated with the
numerical results available in the open literature. Finally, influences of nonlinearities
arising out of electrostatic forces, geometry, and the inertial terms are investigated.
2.5.1
Static Analysis
The static deflection equation is obtained by setting all time derivatives in Eq.
(2.10) to zero. The resulting two-point BVP is numerically solved for w ( x) using the
Matlab boundary value problem solver [Shampine et al.]. The numerical procedure
implements a collocation method for the solution of two-point BVP. The first step is to
express the governing equation as a system of first order differential equations
(collocates). An initial guess is supplied for each variable used to define the first order
differential equations. The guess for an initial mesh is then used by the finite difference
code to obtain accurate numerical solution within an absolute tolerance of 10-6.
Figure 2.2 reveals the existence of two solutions for each value of the gradually
applied DC voltage. The static pull-in condition is predicted as both the solutions coalesce
at a certain voltage. It has been shown by earlier authors [Abdel-Rahman et al., 2002;
Krylov, 2007] that the upper-branch solution is unstable whereas the lower branch
corresponds to the stable equilibrium configurations. Hereafter, in this paper, the lower
branch will suffice for the static deflection of the microcantilever. To validate the
proposed model the pull-in results are compared with the results obtained by Hu et al.
(2004). Figure 2.3 clearly shows that the pull-in results of the present work are in better
41
Chapter 2
Figure 2.2: Variation of the non-dimensional tip deflection wmax with voltage VDC for the
beam properties E = 155.8 GPa , b = 5000 m , h = 57 m , l = 20000 m ,
d 0 = 92 m , = 0.06 as used by Hu et al. (2004).
42
voltage (VDC ) SPI is 66.78 V using the proposed model compared to 68.5 V obtained
experimentally by Hu et al. (2004). The pull-in voltage predicted by Sadeghian et al.
(2007) for the same system configuration was 66.4 V , only after due consideration of
fringing field and stress gradient effects which have been neglected in this study.
To study the influence of the nonlinear curvature, the microbeam static deflection
equation was solved for a range of electrostatic forces ranging from zero to the forcing
level where structural instability (pull-in) develops. This procedure was repeated for
various values of 1 which represents the strength of geometric nonlinearity. Figure 2.4
shows the variation of the non-dimensional tip deflection wmax with the electrostatic force
2
for various values of 1 . At low levels of the electrostatic force, the
represented by 2VDC
2
curve is linear but nonlinearity starts showing its effects as 2VDC
increases to higher ( > 1 )
2
values. As 1 increases, the value of 2VDC
at which pull-in develops also increases. This
is in support of the fact that geometric nonlinearity has got a stiffening effect on the
system. However, the effect of geometric nonlinearity is predicted to be significant for 1
values 0.09 and above. It has been predicted in the earlier works [Nathanson et al., 1967;
Chowdhury et al., 2005; Fargas-Marques et al., 2007] that the pull-in instability develops
at the normalized maximum deflection ( wmax ) value of 0.33 , but the results in Figures
2.2, 2.3 and 2.4 show that the non-dimensional static pull-in deflection lies in the range of
0.45 to 0.5 . These results indicate that the linear theories grossly underestimate the
stability limits of the microcantilever. The static pull-in parameters are also compared
with the results obtained by neglecting the second term on the right hand side of Eq.
(2.10a), and thus assuming parallel plate capacitance (PPC) model. As shown in Table
2.1, for values of 1 equal to 0.09 and above, significant improvement is observed in the
2
static pull-in electrostatic strength ( 2VDC
)
SPI
( wmax )SPI
parameters when second order correction (SOC) of electrostatic forces is taken into
account over parallel plate capacitance approximation. The parallel plate capacitance
approximation that neglects the higher order terms is found to underestimate the pull-in
parameters.
43
Chapter 2
2
Figure 2.4: Variations of the non-dimensional tip deflection wmax with 2VDC
for various
values of 1 .
Table 2.1
Comparison of the static pull-in parameters obtained using the parallel plate capacitance
(PPC) model and the second order correction (SOC) of the electrostatic forces.
( V )
( wmax )SPI
2
2 DC SPI
SOC
PPC
SOC
PPC
2.1e-5
1.681
1.681
0.4462
0.4455
0.04
1.706
1.693
0.4530
0.4585
0.09
1.739
1.708
0.4603
0.4584
0.16
1.788
1.728
0.4760
0.4710
0.25
1.854
1.751
0.4894
0.4672
0.36
1.943
1.783
0.5008
0.4728
44
2.5.2
all time derivatives in Eq. (2.13) are set equal to zero, a set of nonlinear algebraic
equations is obtained. First the set of nonlinear algebraic equations is numerically solved
for a particular number of modes. This is also known as reduced order model (ROM)
analysis. The microcantilever static deflection is thereafter obtained from Eq. (2.11) for
any DC voltage VDC . Figure 2.5 compares the static solution obtained by directly solving
the BVP with the static deflection results of the ROM simulations. The number of modes
is varied from one to five. The design specifications used are same as in Figure 2.2. It is
observed (Figure 2.5b) that the deflection curve obtained through ROM simulation shifts
on either side of the BVP solution, depending on the number of modes considered.
Results obtained retaining five modes in the ROM simulation matches well with the BVP
solution. The results obtained by Younis et al. (2003) shows poor convergence for even
number of modes while others [Krylov and Maimon, 2004; Krylov, 2007] published the
results only for odd number of modes without providing any reasons for non-inclusion of
results for even modes . In the present study, numerically converged results were obtained
for both odd and even number of modes. Using different number of modes in the
discretization procedure, the predicted values of the (VDC ) SPI have been plotted in Figure
2.6. It is further noted from Figure 2.6 that the results exhibit good convergence when
five or more number of modes is taken in the analysis irrespective of the number being
even or odd. Convergence is also exhibited with five modes for higher values of 1 . It is
therefore recommended that at least five modes should be taken to obtain converged
results.
2.5.3
45
Chapter 2
microcantilever is plotted as shown in Figure 2.7. In the same plot, numerical results
obtained by De and Aluru (2004) are also shown. It has been observed that both the
(a)
(b)
Figure 2.5: (a) Predictions of the normalized tip deflection by the ROM of different
orders and (b) enlarged plot of the predictions near the pull-in.
46
Figure 2.6: Comparison of the static pull-in voltages (VDC ) SPI predicted by the ROM of
different orders.
Figure 2.7: Non-pull-in response at a suddenly applied voltage of 0.5 V for the beam
properties E = 169 GPa , b = 10 m , h = 0.5 m , d 0 = 0.7 m , l = 80 m , = 0.3 ,
Chapter 2
results are in excellent agreement. The resonant frequency at 0.5 V is found from the
time period as 108.5 kHz compared to 109 kHz as obtained in the reference [De and
Aluru, 2004]. It has been further shown in Figure 2.8, the dynamic pull-in characteristics
of the cantilever beam observed under pure DC actuation at dynamic pull-in voltage
(VDC ) DPI of 2.12 V in the reference [De and Aluru, 2004] match well with the pull-in
characteristics at (VDC ) DPI of 2.25 V obtained with the present numerical scheme. A small
deviation in the response characteristics is observed (Figure 2.8) after the beam reaches
the dynamic pull-in deflection (about 0.44 m ) due to the extreme sensitivity of the
response to the actuation voltage at pull-in.
Figure 2.8: Comparison of the pull-in responses under suddenly applied DC voltage
actuation.
For a particular value of 1 and varying excitation voltages, the tip deflection
time history, as obtained in Figure 2.9, predicts periodic motion below a certain critical
value of the voltage known as the dynamic pull-in voltage (VDC ) DPI . With the time period
increasing with voltage, a definite softening effect of the electrostatic forces and the
inertial nonlinearity is concluded. For a certain voltage above the critical value, the
periodic motion gives away to a divergent motion and the beam abruptly collapses onto
48
the electrode. The quantitative estimation of the dynamic pull-in parameters can be made
from the corresponding phase plots as shown in Figure 2.10. The dynamic pull-in
condition is marked by the separatrix on the phase plane. The dynamic pull-in of the
microbeam occurs at an applied voltage (VDC ) DPI of 60.79 V , which is well below the
static pull-in voltage (VDC ) SPI of 66.78 V . The dynamic pull-in deflection ( 0.6 ~ 0.7 ),
however, increases thus increasing the travel range of the beam under applied voltages
varying as non-smooth function of time. In order to highlight the effect of higher mode,
the dynamic pull-in condition for the single-mode model (dark continuous line) is also
included in Figures 2.9 and 2.10 which clearly show that single-mode model predictions
vary significantly from the five-mode model results, for example, the dynamic pull-in of
the microbeam is obtained at an applied voltage of 60.40 V with single-mode analysis
whereas the phenomenon is not observed for five-mode analysis carried out at the same
( 60.40 V ) voltage. Ripples can be observed in the response curves for the five-mode
analysis due to the excitation of higher modes.
Figure 2.9: Deflection time history under various suddenly applied DC voltages VDC for
1 = 2.1e 5 .
49
Chapter 2
Figure 2.10: Phase plot for the five-mode ROM of the microcantilever excited by various
suddenly applied DC voltages VDC for 1 = 2.1e 5 .
The effect of geometric nonlinearity on the transient behaviour has been studied in
detail. The dynamic pull-in behaviour of beams for various values of 1 has been
simulated. For brevity, simulations for few values of 1 have been presented in Figures
2.11 and 2.12. Figure 2.11 reveals that the electrostatic force for the dynamic pull-in,
2
) DPI , increases significantly with 1 . However, as predicted in the
represented by ( 2VDC
case of static analysis, the effect of geometric nonlinearity is found to be significant for
1 values 0.09 and above. From the corresponding phase plot, as shown in Figure 2.12,
it is observed that before the beam deflection reaches the value of dynamic pull-in
deflection ( 0.6 ~ 0.75 ), the tip velocity at a certain tip displacement is higher for a higher
value of 1 , and, is lower for a higher value of 1 as the beam deflects beyond and
approaches the ground electrode. This may be attributed to the fact that for beams with
2
) DPI values are higher resulting in higher beam velocities and
higher values of 1 , ( 2VDC
larger softening effects whereas the stiffening effects of geometric nonlinearity are larger
in the large deflection range. Thus, the overall effect, in the entire deflection range, due to
the nonlinearities may be stiffening/softening depending on the relative strengths of the
50
51
Chapter 2
nonlinearities. Moreover, these variations in the tip velocity affect the total time (pull-in
time) the beam takes to collapse (represented by non-dimensional tip deflection of unity)
onto the electrode as shown in Figure 2.11. To highlight the importance of inclusion of
higher order corrections in the electrostatic model, a comparison of the dynamic pull-in
2
electrostatic strength ( 2VDC
) DPI obtained with the second order correction (SOC) model
and the parallel plate capacitance (PPC) model is shown in Table 2.2. As observed in the
static analysis, significant improvement in the dynamic pull-in parameter is obtained with
the second order correction model for 1 values equal to 0.09 and above.
Table 2.2
2
) DPI obtained using the
Comparison of the dynamic pull-in electrostatic strength ( 2VDC
parallel plate capacitance (PPC) model and the second order correction (SOC) of the
electrostatic forces.
1
2
( 2VDC
) DPI
SOC
PPC
2.1e-5
1.393
1.393
0.04
1.414
1.404
0.09
1.441
1.416
0.16
1.480
1.434
0.25
1.534
1.457
0.36
1.606
1.485
52
2
Figure 2.14: Phase plot at representative values of 2VDC
in the vicinity and well below
53
Chapter 2
shown in Figures 2.13 and 2.14, the curves overlay each other. In the vicinity of pull-in
2
= 1.35, 1.62 ), the overall effects due to the nonlinearities are such that the
( 2VDC
Figure 2.13. This increasing trend can be further explained with the help of Figure 2.14
which predicts lower tip velocities, almost in the entire deflection range, for beams with
higher values of 1 .
2
Figure 2.15: Variations of the non-dimensional resonant frequency with 2VDC
for
various values of 1 . SOC: second order correction; PPC: parallel plate capacitance.
inertial terms are further depicted in the figure. As the applied voltage increases, the
structure is softened and the resonance reduces. Moreover, for periodic motions well
2
bellow pull-in ( 2VDC
= 0.8 ), varying 1 results in no significant variation in resonant
54
frequency, whereas, in the vicinity of pull-in the resonant frequency of stable periodic
motion increases significantly with increase in 1 . For comparison, resonant frequency
plots for various values of 1 have also been obtained with the parallel plate capacitance
approximation and the plot for 1 value of 0.36 has been shown in Figure 2.15. The
parallel plate capacitance model predictions of resonant frequencies for higher values of
2
2VDC
( > 0.8 ) are found to be significantly underestimated.
2.6
Summary
The chapter presents the static and transient behaviour of an undamped
microcantilever, with relatively large initial gap to beam-length ratios, under pure DC
voltage electrostatic actuation. A comprehensive large deflection electromechanical
model is formulated accounting for the nonlinearities of the system. Since the gap is
relatively large, the electrostatic model is formulated incorporating correction up to
second order. The related boundary value problem is described, and a reduced order
model is formulated using Galerkin method. An investigation is carried out in detail to
ascertain number of modes to be employed for the reduced order model analysis to arrive
at results within admissible error. The mathematical model is successfully validated with
experimental, analytical, and numerical results available in open literature. Conclusive
results are presented on the impact of varying strengths of nonlinearities on the static and
dynamic behaviour in the vicinity and well below pull-in. Efficacy of the large deflection
model, for behavioral simulations of electrostatic MEMS, with respect to gap-length ratio
of the devices is demonstrated.
55
Chapter 3
SMALL AMPLITUDE MOTION UNDER A COMBINED DC
AND AC VOLTAGE ACTUATION
3.1
Introduction
Microstructures actuated at resonance condition, referred to as microresonators,
are an integral part of MEMS devices like resonant sensors. In case of an electrically
actuated resonant sensor, the microcantilever is deflected by a bias DC voltage and
subjected to vibration in the resonant mode by an AC harmonic voltage. In case of small
DC voltage the static bias deflection may be neglected but should be accounted for
voltage large enough to cause non-uniform gap spacing between the cantilever electrode
and the ground plane [Batra et al., 2008]. Moreover, large DC voltages cause accountable
softening of the system [Nathanson et al. 1967]. In practice, the electromechanical system
is complicated by the effects influenced by the fluid surrounding the system. The
damping of electrically actuated microstructures vibrating in presence of fluid trapped in
the narrow gap is dominated by squeeze-film damping [Nouira et al., 2007]. When the
microcantilever vibrates, there is appreciable resistance to motion due to squeeze-film
damping which may be of the same order of magnitude as the electric and mechanical
forces. Thus, the dynamics of the electrostatically actuated cantilever moving in trapped
fluid is influenced by the structure-electrostatic-fluid coupling effect. It is, therefore,
essential to effectively model the electro-mechanical coupling of the system and the
damping effects to correctly predict the most important design characteristics of the
resonator, namely, the resonant frequency and the quality factor.
A detail account of studies investigating the dynamic behaviour of microbeams
driven by a DC bias and small harmonic voltage has been already presented. It may be
57
Chapter 3
noted from the review that majority of them [Zhang et al., 2004; Nayfeh and Younis,
2004; Pandey and Pratap, 2007; Chakraborty et al., 2010] neglects the static bias
deflection due to the DC voltage. Frequency shift variations with DC voltage were
investigated by some authors [Li et al., 2008; Ballestra et al., 2008; Batra et al., 2008]
while taking into account the electrostatic softening effects. However, their models
neglect the influences of damping. Younis and Nayfeh (2007) considered the static
deflection while dealing with the problem of large DC bias voltage actuation of damped
microplates under different gas pressures. They used a singular perturbation method to
obtain the pressure distribution in terms of the structural mode shapes. However, the
perturbation approach does not apply to cantilevers because of their relatively small
fundamental natural frequencies. While accounting for the static bias deflection, few
models [Li et al., 2007; Younis and Nayfeh, 2007] have also attempted to incorporate the
squeeze-film damping effects including the rarefied flow conditions. But the effective
viscosity approach followed to account for the rarefied gas effects neglects the
dependency of the effective viscosity on the gap height which is no longer uniform for a
beam statically deflected by a DC bias.
This chapter deals with small amplitude motions under the influence of large DC
load coupled with small AC harmonic voltage, treated with due consideration of the static
deflection due to the large DC bias. The small amplitude linear dynamic behaviour of the
resonator operating in different ambient pressure conditions is analysed in the presence of
linearized squeeze-film damping accounting for flow rarefaction effects. The entire
analysis in this part of work is restricted to the range of initial gap-length ratios, as
concluded from the large-gap analysis discussed in the preceding chapter, which validates
the application of parallel plate nonlinear electrostatic model in conjunction with the
small deflection theory for the structural model. The dynamic problem coupled in the
electrostatic, structural, and fluid domains is formulated by coupling the Euler-beam
equation, incorporating the electrostatic force terms and the squeeze-film damping terms,
with the linearized compressible Reynolds equation. The present model takes into account
the dependency of the effective viscosity on variable gap spacing through the variable
Knudsen number. The results generated with a semi-analytical method for the first three
flexural modes of vibration are validated with the analytical, numerical and experimental
results available in open literature. The dynamic problem is simulated with the coupled
field finite element analysis (FEA) software ANSYS and results are compared. The
effects of variations in DC bias voltage, ambient pressure, and flow compressibility on
58
3.2
Mathematical Formulation
Figure 3.1 (reproduced here from Chapter 2 for convenience) shows a damped cantilever
beam of length l , width b , thickness h separated from the ground plane by a small
initial air-gap of d 0 . When subjected to a driving voltage V (t ) comprising of a DC bias
voltage VDC and a small AC component v(t ) with VDC >> v(t ) , the beam undergoes small
amplitude motions around the statically deflected position. These small oscillations
redistribute pressure in the air trapped in the non-uniform gap spacing causing a damping
effect. Let w( x, t ) denote the transverse displacement of the beam being dependent on the
position x along the beam length and time t . Neglecting the nonlinear curvature and the
inertia nonlinearity terms, retaining only the parallel plate electrostatic force terms, and,
adding term representing the squeeze-film damping effect, the governing equation of
motion for small initial air-gap to length ratio can be obtained from Eq. (2.9a) as
2w
4 w 1 0bV 2
EI
+
=
f d ( x, t )
t 2
x 4 2 ( d 0 w ) 2
(3.1a)
w(0, t ) =
w
=0 ,
x x = 0
(3.1b)
2w
3 w
=
=0
x 2 x =l x 3 x =l
59
Chapter 3
The first term on the right hand side of Eq. (3.1a) represents the excitation force per unit
length while the second term represents the force acting on the beam owing to the
pressure of the air-film squeezed between the beam and the ground plane.
The one-dimensional force
integrating the two-dimensional pressure distribution along the width of the beam, and, is
given by
b
f d ( x, t ) = [ p ( x, y, t ) pa ] dy
(3.2)
where p ( x, y, t ) is the absolute pressure in the gap and pa is the ambient pressure.
3.2.1
gaps and the motion of the beam being restricted to normal approach, the pressure
p ( x, y, t ) is governed by the nonlinear compressible Reynolds equation [Hamrock, 1994]
a g 3 p a g 3 p
+
= ( a g )
x 12 x y 12 y t
(3.3)
is the dynamic viscosity of air under standard temperature and pressure. For the ideal
gas under isothermal conditions, p. a 1 = C where C is a constant, and, considering
rarefied flow, Eq. (3.3) can be reduced to
(d 0 w)3 p (d 0 w)3 p
p
w
p +
p = (d 0 w) p
x 12 eff
x y 12 eff
y
t
t
(3.4a)
(3.4b)
mean free path 0 at standard temperature and pressure p0 = 1.013 105 Pa conditions is
about 65 10 9 m . The present model takes into account the dependency of the effective
viscosity on variable gap spacing through the variable Knudsen number Kn( x, t ) .
60
x = x l , y = y b , w = w d0 ,
1
2w 4w
2V 2
( p 1)dy
+
=
P
t 2 x 4 (1 w )2
0
(3.5a)
w(0, t ) = ( w x ) x =0 = 0 ,
( w x )
2
x =1
= ( 3 w x 3 )
x =1
(3.5b)
=0
x eff
x
y eff
y d 0 pa s
t
t
(3.6a)
(3.6b)
where 2 = (1 2 ) ( 0bl 4 EId 0 3 ) , P = l 4bpa EId 0 , = l b . It may be noted that the overscores have been removed for convenience.
The response of the microcantilever w( x, t ) under an excitation voltage comprising of a
large DC voltage and a small AC component can be expressed as
w( x, t ) = w0 ( x) + wd ( x, t )
(3.7)
where w0 ( x) represents the static deflection under the DC bias voltage VDC , and,
wd ( x, t ) represents the small amplitude vibration, around the statically deflected position,
61
Chapter 3
3.2.2
Static Deflection
Substituting Eq. (3.7) in Eq. (3.5a) and dropping the dynamic terms, the equation
(3.8a)
w0 (0) = ( dw0 dx ) x =0 = ( d 2 w0 dx 2 )
x =1
= ( d 3 w0 dx 3 )
x =1
=0
(3.8b)
w0 ( x) = aii ( x)
(3.9)
i =1
where N represents the number of modes retained in the solution; ai are the unknown
coefficients to be determined. i ( x) is the i th linear undamped mode shape of the
undeflected microcantilever.
Multiplying Eq. (3.8a) by (1 w0 ) 2 , and proceeding as described in Section 2.4, a set of
coupled nonlinear algebraic equations can be derived as
N
i =1 j =1 k =1
2
an ( sn ) 2 2 ai a j ( si )2 ni j dx + ai a j ak ( si )2 ni jk dx 2VDC
n dx
i =1 j =1
= 0,
n = 1, 2, ....., N
(3.10)
Prediction of the static deflected shape for different bias voltages is very crucial to
estimate the characteristics of the resonator in the entire working range of voltage
polarization. ROM analysis performed in Section 2.5.2 has shown that at least five modes
( N = 5 ) are required in Eqs. (3.9) and (3.10) to get the correct deflected shape for beam
based MEMS devices even for small air-gap to length ratios.
3.2.3
is also small and the absolute pressure underneath the vibrating beam can be written as
p ( x , y , t ) = 1 + P * ( x, y , t )
(3.11)
where P* ( x, y, t ) << 1 . Substituting Eq. (3.7) and Eq. (3.11) in Eq. (3.5), expanding the
62
wd (0, t ) = ( wd x ) x = 0 = ( 2 wd x 2 )
x =1
= ( 3 wd x 3 )
(3.12a)
x =1
=0
(3.12b)
(1 w0 )
*
2 *
2 P*
P*
2 P dw0
3 P
2
3
1
1
1
w
w
w
(
)
(
)
(
)
0
0
0
s t
x 2
x dx
y 2
0.159
*
Kn0
wd
3 P ( 9.638 )(1.159 ) Kn0
+ (1 w0 )
1.159
s t
x
x
1 + 9.638 Kn0
=0
(3.13a)
(3.13b)
(3.14)
The present study deals with small amplitude motions around the statically deflected
position, under very small AC excitations. With these prevailing conditions, it is assumed
that there is no nonlinear resonance and the spatial shape of the deflection wd at i th
resonance mode can be expressed as a function of the linear undamped mode shape i ( x)
modified by the complex amplitude A of small vibration. Thus, the steady-state solution
of Eq. (3.12) and Eq. (3.13) can be expressed, respectively, by
wd ( x, t ) = Ai ( x )e j ( s )t
(3.15)
P* ( x, y, t ) = ( x, y )e j ( s ) t
(3.16)
and
The spatial distribution of pressure ( x, y ) for compressible flow conditions has been
obtained by various researchers [Langlois, 1962; Blech, 1983; Darling et al., 1998]
employing rigid plates but few analytical models [McCarthy et al., 2002; Pandey and
Pratap, 2007; Li et al., 2007] are present for flexible microstructures. McCarthy et al.
(2002) assumed ( x, y ) to be separable being the product of a parabolic function along
the beam width and an unknown function along the beam length which is determined
during solution. Pandey and Pratap (2007) expanded it using Greens function approach
63
Chapter 3
of assumed double cosine series. Li et al. (2007) represented the pressure distribution as a
product of an assumed parabolic function along beam width and a cosine series along
beam length. In the present work, following Li et al. (2007), the spatial distribution of
pressure ( x, y ) satisfying the boundary conditions (Eq. (3.13b)) is given as
M
( x, y ) = zm cos (2m 1)
m =1
x ( y y 2 )
2
(3.17)
zm
AU
( x ) = i ( x)
(3.18)
m =1
where
U m ( x) = um ( x) cos (2m 1)
2
x + um ( x) sin (2m 1)
2
(3.19)
um ( x ) = (1 6 ) (2m 1) (1 w0 )3 + 2 2 (1 w0 )3 + j (1 w0 )
2
6
(2m 1)
um ( x ) =
(1 w0 ) 2
dw0 1
( 9.638)(1.159 ) Kn0
+ (2m 1) (1 w0 )3
dx 6
2
1 + 9.638 Kn01.159
j
0.159
Kn0
x
(3.20)
(3.21)
(3.22)
0
1
x dx
64
(3.23)
0
1
x dx
(3.24)
Z( m) = zm = zmR + jzmI
(3.25)
where zmR and zmI are the real and imaginary parts of z m , respectively. Using Eqs. (3.16)
and (3.17), the distributed force acting on the beam owing to the time varying pressure
distribution underneath it can be rewritten as
1
(3.26)
where
Fs ( x) =
P M zmR
cos (2m 1)
6 m =1 A
2
(3.27)
Fd ( x) =
P M zmI
cos (2m 1)
6 m =1 A
2
(3.28)
The spring effect of the trapped air is represented by Fs ( x) while Fd ( x) stands for the
damping effect.
Substituting Eqs. (3.15), (3.16), and (3.26) into Eq. (3.12a), multiplying throughout by
A=
2 2VDC v0
0
i ( x)
(1 w0 ( x) )
dx
2
2
{(si ) ke + ka } 1 2 (s )(2sik) + k + j (s )2 c k + k
i
i
e
a
i
i
e
a
(3.29)
where
1
2
ke = 2 2VDC
i 2 ( x)
(1 w0 ( x) )
(3.30)
dx
represents the softening effect due to electrostatic coupling. The squeeze-film stiffness k a
and the damping coefficient c are given by
1
ka = Fs ( x )i ( x)dx =
0
c=
P M zmR
H ( m)
6 m=1 A
(3.31)
i 1
i P M zmI
(
)
(
)
=
F
x
x
dx
H ( m)
i
0 d
6 m=1 A
65
(3.32)
Chapter 3
Considering a lumped spring-mass-damper system under a harmonic excitation, the
equation of motion can be expressed as
ml
d 2r
dr
+ cl
+ kl r = Fe jt
2
dt
dt
(3.33)
where ml , k l , and cl are the lumped mass, spring constant and damping constant
respectively. The steady-state solution is given by
m
c
R = ( F kl ) 1 l 2 + j l
kl
kl
jt ,
r = Re
(3.34)
The damping ratio , quality factor Q and the resonance frequency res are given by
cl
2 kl
kl
,
ml
Q = 1 2 ,
res = 1 2 2
kl
ml
(3.35)
In open literature, several definitions exist relating the quality factor Q and the damping
ratio . Pandey and Pratap (2007) used the relation between Q and expressed through
Eq. (3.35) to calculate the damping ratio from the experimentally obtained quality factor.
Since the results of the present work have been validated with the results obtained by the
same authors, the relation between Q and expressed through Eq. (3.35) has been
uniformly used for the entire analysis.
By comparing Eqs. (3.29) and (3.34), an equivalent mass-spring-damper system may be
obtained by making the corresponding terms equal. Subsequently following the
definitions given by Eq. (3.35) leads to the expressions of the quality factor and the
resonance frequency of the microcantilever resonator under squeeze-film damping as
Q = ( si )
res =
3.2.4
1
s
( si )2 ke + ka
( si )2 ke + ka
)c
(3.36)
c2
2( si )2
(3.37)
Compressibility Ratio
To examine whether the damping effect or the spring effect of the air-film in the
gap is dominant, a non-dimensional parameter called the squeeze-number is defined in
the literature [Langlois, 1962; Blech, 1983] as
= 12 b 2 pa d 0 2
(3.38)
For rigid plates undergoing small sinusoidal motion, Blech (1983) found the elastic force
66
c = 2 1 +
1
2
(3.39)
For a given set of operating conditions, the flow compressibility can be assessed by
comparing the values of and c . For << c , the spring effects are negligible and the
flow can be assumed to be incompressible. However, the present model of a flexible
microbeam statically deflected by a large DC bias voltage and undergoing flexural
vibrations does not facilitate such a comparison. Alternatively, the relative dominance of
the spring and damping effects in the present case can be obtained by directly calculating
the ratio (hereafter called the compressibility ratio) of the elastic force and the viscous
damping force due to the squeeze-film. The expression for can be given by
1
= Fs ( x)dx
0
F ( x)dx
(3.40)
For << 1 , the elastic forces are negligible and the flow can be assumed to be
incompressible whereas for finite values of compressibility effects cannot be ignored.
For > 1 , spring effects dominate over the damping effects of the squeeze-film.
3.2.5
The present section assumes an incompressible and non-inertial flow in the airgap. As the flow is incompressible, the squeeze-film spring forces will be absent and
there will be only dissipation of the movement. Proceeding as in Section 3.2.1, the
nonlinear incompressible Reynolds equation can be expressed in the non-dimensional
form as
12l 2 w
1
1
3 p
3 p
(1 w ) + 2 (1 w ) = 2
x eff
x
y eff
y d 0 pa s t
(3.41a)
(3.41b)
As explained in Section 3.2.3, Eq. (3.41a) can be linearized and expressed as Eq. (3.13a)
with the exclusion of the term (1 w0 )
modified by excluding the term
P*
. For incompressible flow, Eq. (3.20) can be
s t
1
j (1 w0 ) . Repeating the rest of the procedure as
6
67
Chapter 3
described in Section 3.2.3, the quality factor Q and the resonance frequency res can be
expressed by Eqs. (3.36) and (3.37) respectively.
3.3
projection technique [Mehner et al., 2001] available in the commercial FEA software
ANSYS. Three coupled field domain solvers: electrostatic, structural, and fluid have been
utilized. The structural domain has been modelled with SOLID 45 elements used for 3-D
modelling of solid structures, and, capable of modelling stress stiffening associated with
static deflection due to large DC bias voltages. Direct coupling of electrostatic and
structural domain has been achieved using TRANS 126 elements. TRANS 126 elements
are lumped electromechanical transducers preferable for highly nonlinear coupled field
interaction but valid for small air-gap to length ratio. The model uses FLUID 136
elements to account for the squeeze-film damping effect which is based on linearized
compressible Reynolds equation modified for the rarefied gas effect and is thus applicable
for small pressure changes in the small air-gap.
Figure 3.2: An enlarged sectional view of the coupled field model simulated with
ANSYS.
68
eff being dependent on the variable Knudsen number Kn is assigned to each fluid
element with the updated local gap separation. A modal analysis is performed next to
compute the natural frequencies and the mode shapes (eigenvectors) for the flexural
modes of vibration. For a film of fluid adjacent to the structure, there is a true dependency
between the structural velocity and the fluid pressure. For a beam undergoing flexible
body dynamics, the displacement and velocity vary along the structure. Thus, the air-film
underneath the beam is subjected to varying velocity profile. In case of modal projection
method, the harmonic analysis on the fluid elements is performed with the nodal
velocities equal to the eigenvector of the appropriate mode obtained through the modal
analysis performed in the previous step. The resulting pressure distribution and hence the
modal forces are computed to obtain the modal squeeze stiffness and damping
parameters. In ANSYS simulations, the real component of the modal forces contributes to
damping and the imaginary part contributes to stiffness. Subsequently, the modal
damping ratio is obtained corresponding to the flexural modes.
3.4
previous sections, namely, semi-analytical method and finite element method. The semianalytical method uses the compressible flow model to generate the results presented in
this section. A comparison of the squeeze-film characteristics obtained with the
incompressible and compressible flow models has also been presented in Section 3.4.4.
First the convergence criterion is checked for both the semi-analytical and finite element
methods. Subsequently, the converged results are validated with the analytical, numerical
69
Chapter 3
and experimental results for zero DC bias voltage as available in open literature. The
design properties used in the present analysis are l = 150 m and 350 m , b = 22 m ,
h = 4 m , d 0 = 1.4 m , = 2330 kg / m3 , E = 160 GPa , pa = 1.013e5 Pa , = 1.8e 5 Ns / m 2
as used for the experimental study by Pandey and Pratap (2007). The model is also
validated for wide beams and high squeeze numbers. The applicability of the model with
regard to the compressible effects of the squeeze-film is explained through the
compressibility ratio in the subsequent section. For large DC bias voltages, the results
obtained with the present model are validated with the ANSYS results. The effects of
large DC bias voltages on resonance frequency and quality factor are studied next for the
first three flexural modes of vibration of the resonator. Variations in ambient pressure in
presence of DC bias voltages are shown to affect the resonance frequency and the quality
factor of the microresonator.
The unknown coefficients ai present in Eq. (3.9) are evaluated by numerically
solving the set of nonlinear algebraic equations obtained retaining five ( N = 5 ) modes in
Eq. (3.10). The static deflection for a DC bias voltage is then expressed as a function of
position x using Eq. (3.9). The spring softening effect k e , squeeze-film stiffness k a , and
the damping coefficient c are obtained using Eqs. (3.22)-(3.24) and (3.30)-(3.32) where
the integrations are obtained numerically. Thereafter, using Eqs. (3.35)-(3.37), the quality
factor Q , the damping ratio , and the resonance frequency res are calculated for the
flexural modes of vibration. The undamped natural frequency is calculated from Eq.
(3.37) replacing terms k a and c by zero.
To ensure convergence, damping ratio is evaluated for different values of M
used in Eq. (3.17). As shown in Figure 3.3, results exhibit good convergence when 15 or
more number of pressure terms are used. In this study, M is taken as 25 which exhibit
good convergence even at large DC voltages. The results obtained through simulations
carried out in ANSYS are also checked for convergence by computing the damping ratio
while varying the number of fluid elements. It is evident from Figure 3.4 that excellent
convergence is observed if the fluid domain is meshed with more than 7000 elements.
Therefore in the present study, more than 7000 fluid elements are used to ensure
convergence even at large electrostatic loads. Number of electromechanical TRANS 126
elements used in each simulation equals the number of surface nodes present in the
meshed fluid domain and has been checked to ensure convergence of the static deflection.
70
Figure 3.3: Convergence of the damping ratio with the number of pressure terms M
for the 350 m long resonator.
Figure 3.4: Convergence of the computed damping ratio with the number of fluid
elements used in simulations with ANSYS for the 350 m long resonator.
71
Chapter 3
3.4.1
Validation Study
Table 3.1
Comparison of the damping ratio results for zero ( VDC = 0 V ) DC bias voltage.
Length
l
( m )
350
150
Flexural
mode
Experimental
[Ref.]
FE model
[Ref.]
FE model
[Present
Work]
Analytical
[Ref.]
Present
model
1st
0.415 0.002
0.45
0.4475
0.422
0.4484
2nd
0.066 0.002
0.072
0.0710
0.067
0.0713
3rd
0.027 0.002
0.025
0.0244
0.024
0.0247
1st
0.071 0.002
0.074
0.0735
0.070
0.0747
In order to further validate the model for wide cantilever beams and high squeeze
numbers, damping ratio results have been obtained for a different set of beams with
design properties b = 80 m ; l = 100 m , 88.89 m and 80 m ; h = 4 m ; d 0 = 1.4 m ;
72
Width
b
( m )
80
Length
l
( m )
Squeeze
number
Analytical
[Ref.]
Present
model
FE model
[Present
Work]
100
18.37
0.105
0.1056
0.1019
88.89
23.25
0.067
0.0676
0.0639
80
28.71
0.044
0.0447
0.0418
Table 3.3
The percentage error in damping ratio of the present semi-analytical model ( ) with
reference to the FE model ( FE ) compared at different compressibility ratios in the
first flexural mode for zero ( VDC = 0 V ) DC bias voltage.
Compressibility
ratio
FE
100
FE
0.11
9.91
0.0093
0.20
150
0.63
10.08
0.0488
1.63
80
100
18.37
16.19
1.0727
3.63
80
88.89
23.25
17.86
1.2575
5.79
80
80
28.71
19.74
1.4308
6.94
Width
b
( m )
Length
l
( m )
Squeeze
number
22
350
22
Cut-off squeeze
number
73
(%)
Chapter 3
In order to estimate the range of compressibility effect of squeeze-film over which
the present model is valid, the magnitudes of compressibility ratio (as mentioned in
Section 3.2.4) are evaluated for different microbeam widths and lengths as presented in
Table 3.3. In addition, percentage error in damping ratio of the present semi-analytical
model ( ) with reference to the FE model ( FE ) is also mentioned in the table. The
relatively small (less than 5 % ) percentage error indicates the applicability of the semianalytical model for values of ( 1 ) at which the compressibility effects are
appreciable. To further illustrate the effectiveness of , the values of cut-off squeeze
number c (Eq. (3.39)) obtained following the rigid-body formulation [Blech, 1983] are
also presented in Table 3.3. It is evident from Table 3.3 that when << c , is very
small ( << 1 ) as the elastic forces are negligible and the flow can be assumed to be
incompressible. When is comparable with c , is high enough so that the
compressibility effects cannot be ignored. Thus, for statically deflected flexible structures
undergoing small amplitude motion, compressibility ratio provides an accurate
quantitative estimation of the relative strengths of the elastic and damping forces of the
squeeze-film.
Table 3.4
Comparison of the damping ratio results in the first flexural mode for trivial and
nontrivial pressure boundary conditions under zero ( VDC = 0 V ) DC bias voltage.
Length
l
( m )
Experimental
[Ref.]
trivial
nontrivial
350
0.415 0.002
0.4484
0.5403
150
0.071 0.002
0.0747
0.0907
nontrivial have been calculated using surface dimensions leff and beff , as described earlier
74
Figure 3.5: Variations of undamped natural frequency with DC voltage VDC for two
different lengths of the cantilever microresonator computed with ANSYS (o) and the
present semi-analytical model (+).
To validate the present model for large electrostatic loads, the semi-analytical
results for undamped natural frequency and quality factor are compared with the ANSYS
results. It is to be highlighted that the results of the present model are in excellent
agreement with the ANSYS results even when the electrostatic load is large. For the first
flexural mode of vibration of the beam of two different lengths, Figure 3.5 shows the
exclusive effect of electromechanical coupling on the undamped natural frequency found
to be monotonically decreasing with DC voltage VDC up to pull-in instability. The
undamped frequencies are obtained in ANSYS by carrying out modal analysis of the
three-field coupled model, with the fluid elements not affecting the results. The
monotonically decreasing pattern corroborates the softening effect of electrostatic forces
75
Chapter 3
as observed in earlier works [Batra et al., 2008; Younis and Nayfeh, 2007]. With increase
in static bias voltage the gap width decreases and the effect of squeeze-film damping
increases thus decreases the quality factor of the resonator as can be seen in Figure 3.6.
The mechanism behind this increase in damping has been further discussed in detail in the
following section. It is noteworthy that the results obtained with the semi-analytical
model compares very well with the ANSYS results.
Figure 3.6: Variations of quality factor Q with DC voltage VDC for two different lengths
of the cantilever microresonator computed with ANSYS (o) and the present semianalytical model (+).
3.4.2
Figure 3.7: Variations of quality factor Q with DC voltage VDC for the first three flexural
modes of vibration.
77
Chapter 3
the quality factor decreases monotonically with DC voltages. However, the percentage
decrease in quality factor Q
( (Q Q ) 100 / Q )
ref
ref
voltage V ( VDC Vref ) is observed (Figure 3.8) to be higher for the first mode of
vibration in comparison to the higher modes. As shown in Figure 3.8, the quality factor
decreases sharply by 75 % approximately for the first mode as pull-in is approached.
Figure 3.9: Variations of resonance frequency with DC voltage VDC for the first three
flexural modes of vibration.
Figure 3.9 shows the variations of the resonance frequency for the first three
flexural modes of vibration of the resonator. For the first mode of vibration of the
resonator operating under ambient ( pa = 1.013e5 Pa ) conditions, the resonant peak
disappears at bias voltages ( VDC 6 V ) much lower than the static pull-in ( VDC = 7.704 V )
voltage due to increase in damping thus limiting the working range of voltage
polarizations for the resonator. Though the resonance frequency for the first mode
decreases monotonically with DC bias, the resonance frequency for the higher modes is
more or less uniform for the entire range of DC polarization.
78
79
Chapter 3
Figure 3.10: Dynamic pressure distribution (in Pa) on the bottom surface of the resonator
actuated by (a) VDC = 0 V and (b) VDC = 7.5 V for (I) First mode, (II) Second mode, and
(III) Third mode of vibration.
80
The effects of DC voltage with varying air-gap pressure on quality factor and
resonance frequency are investigated next. Pressure dependency of the damping has been
analyzed with the model by considering the pressure dependency of the mean free path
81
Chapter 3
Figure 3.11: Variations of quality factor Q with DC voltage VDC under various ambient
air pressures for (I) First mode, (II) Second mode, and (III) Third mode of vibration.
82
83
Chapter 3
Figure 3.12: Variations of resonance frequency with DC voltage VDC under various
ambient air pressures for (I) First mode, (II) Second mode, and (III) Third mode of
vibration.
Figure 3.12 shows variation of the resonance frequency with DC voltage VDC for
the three modes of vibration of the resonator for various values of ambient pressure pa .
As shown in the figure 3.12(I), at a certain DC voltage the undamped and damped
frequencies have no appreciable difference at low pressures. Thus, it can be concluded
that the squeeze-film damping at low pressures has negligible effect on first mode
resonance frequency for the entire range of DC polarization. For the higher modes of
vibration, resonance frequency is observed (Figures 3.12(II) and 3.12(III)) to be nearly
invariant with pressure variations for the entire range of DC polarization.
3.4.4
84
Figure 3.13: Variations of quality factor Q with DC voltage VDC for the different flow
models for ambient pressures (a) pa = 1.013e5 Pa , and (b) pa = 1e3 Pa .
85
Chapter 3
As shown in Figure 3.13(a), no appreciable difference exists in the quality factor
values obtained with the compressible and incompressible flow models. The
compressibility ratio values ( 0.009 for the first mode, 0.06 for the second mode
and, 0.15 for the third mode) computed for case (a) explains the negligible
compressible effects. However, Figure 3.13(b) shows appreciable difference in the quality
factor values obtained with the two flow models for the third mode ( 0.34 ) of
vibration. Thus, it is clear that the difference between the incompressible and
compressible flow models used in predicting the squeeze-film characteristics becomes
more pronounced as the compressibility ratio increases. Moreover, as shown in Figure
3.13(b), quality factors obtained with the incompressible flow model is comparatively
lower for the entire range of DC polarization as squeeze-film stiffening effects are absent
for incompressible flow and the damping is overestimated.
3.5
Summary
This chapter presents the small amplitude linear dynamic behaviour of cantilever
86
Chapter 4
NONLINEAR RESPONSE UNDER LARGE DC AND AC
EXCITATION
4.1
Introduction
The central component in many MEMS devices is the mechanical resonator
87
Chapter 4
excitation. Thus, the investigation of large amplitude dynamic behaviour requires
effective modelling involving the physical effects such as mechanical motion, damping
and electrostatic actuation.
There is a large body of work devoted to the dynamics of resonators undergoing
periodic excitation. An overall review of the research work performed by several authors
mainly for excitation frequencies close to one of the system resonances has been already
presented. This interest is explained by applications where the resonant amplification is
beneficial [Rocha et al., 2006; De and Aluru, 2006; Park et al., 2008; Vyas et al., 2008]
or, in contrast, should be avoided [Luo and Wang, 2004; Liqin et al., 2006; Nayfeh et al.,
2007; Alsaleem et al., 2010]. Theoretical investigations utilizing models of varying
complexity have revealed a multitude of nonlinear phenomena including resonance-shifts,
secondary resonances, jumps and hysteresis, period-doubling, chaos, and dynamic pull-in
instabilities. Majority of these models are based on simplified lumped-parameter systems
[Wang et al., 1998; Luo and Wang, 2004; Rocha et al., 2006; Mestrom et al., 2008, 2010;
Alsaleem et al., 2009, 2010]; considered linear [Wang et al., 1998; Nayfeh and Younis,
2005; Nayfeh et al., 2007; Park et al., 2008; Mestrom et al., 2008; Alsaleem et al., 2010]
or no [Luo and Wang, 2004; Mestrom et al., 2010] damping; incorporated truncated
expansion series for the nonlinear electrostatic [Liqin et al., 2006; Mestrom et al., 2008]
and squeeze-film damping [Liqin et al., 2006; Alsaleem et al., 2009] terms. However, for
accurate prediction of the nonlinear dynamics of the microbeams actuated by large
electrostatic load, a nonlinear spatial distributed parameter model that accounts for
forcing and system nonlinearities needs to be developed and studied.
The present chapter focuses on the large amplitude dynamic behaviour of a
cantilever microbeam actuated by a combination of DC and harmonic AC loading in
presence of ambient air. A distributed parameter model is derived accounting for the
nonlinearities of the system due to nonlinear electrostatic forces and nonlinear squeezefilm damping. The present work emphasizes on the full-order representations of the
nonlinear terms unlike the truncated expansion series. Structural nonlinearities due to
geometry and inertia are neglected limiting the validity of the model to small initial gaplength ratios as inferred from Chapter 2. The model further assumes inertia-less,
incompressible flow conditions, and, accounts for the rarefaction effects limited to the
slip-flow regime. From the comprehensive model, a ROM is developed and numerically
simulated to study the large amplitude dynamic behaviour of the microcantilever for the
fundamental flexural mode of vibration under primary and superharmonic AC excitations.
88
Nonlinear Response
Beginning with a static analysis, several nonlinear phenomena are illustrated using
frequency-response curves, time-histories, and phase portraits as effective analysis tools.
Response sensitivity to initial conditions is investigated near bifurcation points. A
comparison of dynamic characteristics simulated near primary and superharmonic
resonances for various combinations of DC and AC loading is presented.
4.2
Mathematical Formulation
Proceeding as described in Section 3.2, the equation of motion can be expressed
by Eq. (3.1a) with the boundary conditions expressed by Eq. (3.1b). In this case, the
driving voltage V (t ) comprises of a DC bias voltage VDC and a harmonic AC component
VAC cos (t ) . It may be mentioned that the presently applied AC component is not
f d ( x, t ) due to squeeze-film damping is obtained as earlier using Eq. (3.2). The equation
of motion is coupled to the nonlinear Reynolds equation expressed by Eq. (3.3) with the
trivial pressure boundary conditions given by Eq. (3.4b).
For very small air-gap, air in the gap is rarefied even under standard temperature and
pressure conditions. In case of many microsystems, the rarefaction effects can be
accounted for by assuming slip-flow boundary conditions for the air-flow. Assuming the
flow to be incompressible, and, considering an effective coefficient of viscosity
eff [Burgdorfer, 1959] to take into account the slip-flow ( Kn < 1 ) conditions, Eq. (3.3)
can be reduced to
(d 0 w)3 p (d 0 w)3 p
w
+
= 12
x eff
x y eff
y
t
where
variable
effective
viscosity
eff ( x, t ) = (1 + 6 Kn)
(4.1)
is
calculated
using
Burgdorfers model modified to take into account the dependency of the effective
viscosity on variable gap spacing through the variable Knudsen number Kn( x, t ) .
Using the non-dimensional variables as defined in Section 3.2.1, the equation of motion
can be non-dimensionalized as expressed by Eq. (3.5a) with the non-dimensional form of
the boundary conditions given by Eq. (3.5b). Further, the non-dimensional form of the
nonlinear incompressible Reynolds equation, given by Eq. (4.1), can be expressed as
2
2
3
3
d 0 (1 w) p d 0 (1 w) p 12 w
+
=
l x
b y eff y pa s t
x
eff
89
(4.2a)
Chapter 4
with the pressure boundary conditions, Eq. (3.4b), rewritten in the non-dimensional form
as
(4.2b)
It may be noted that the over-scores have been removed for convenience.
Taking into consideration the boundary conditions, Eq. (4.2b), absolute pressure
underneath the vibrating beam can be written as
p ( x , y , t ) = 1 + P * ( x, y , t )
(4.3)
( x) along the beam length. It needs to be mentioned that for the beam undergoing large
amplitude dynamics, the dynamic pressure variation P* ( x, y , t ) is not necessarily small as
assumed in the preceding chapter.
Substituting Eq. (4.3) in Eq. (4.2a) and integrating the resulting equation across the width
of the beam leads to
2
2
w
1
1 1
d0 1
2
3
(1
)
(
)
(1
)
w
w
l 2
x x 6eff
x 2 6 x eff
eff
j ( s )t
3
e
(1 w)
x
d 2
(1 w)3 e j ( s ) t
0
b
eff
12 w
pa s t
(4.4)
As observed from the analysis carried out in Chapter 2, for air-gap to length ratio d 0 l
less than 0.3 , the ( d 0 l ) 2 terms in the expression for the electrostatic force and structural
nonlinearities can be safely neglected. In the present model, the equation of motion, Eq.
(3.1a), neglects the (d 0 l )2 terms and thus assumes parallel plate nonlinear electrostatic
model in conjunction with the small deflection theory for the structural model. For
compatibility of the fluid model with the electrostatic and the structural model, the
(d 0 l )2 terms in Eq. (4.4) are neglected [Krylov and Maimon, 2004; Park et al., 2008] to
obtain the pressure function ( x) as
( x) =
6 eff b 2
1 w j ( s ) t
e
d pa s (1 w)3 t
(4.5)
2
0
90
Nonlinear Response
Using Eqs. (4.3) and (4.5), Eq. (3.5a) can be rewritten as
2V 2
2w 4w
w
+ 4 =
2
2
2
t
x
(1 w) (1 w + 6 )(1 w) t
where = ( l 2
(4.6)
on the right hand side of Eq. (4.6) denotes the damping term and is proportional to the
beam velocity with the coefficient being a nonlinear function of the beam displacement.
The distributed [Nayfeh and Younis, 2005; Nayfeh et al., 2007; Park et al., 2008] and
lumped [Mestrom et al., 2008; Alsaleem et al., 2010] parameter models derived earlier
assumed a coefficient corresponding to linear viscous damping.
4.2.1
N N
d 2 an
dan
d 2 ai 1
2
2
(1
6
)(
s
V
a
(3
12
)
a j ni j dx
+
+
+
+
n
2
2
n
dt 2
dt
i =1 j =1 dt
0
1
d 2 ai
(3 + 12 ) ai a j ( si ) ni j dx + (3 + 6 ) 2 a j ak ni jk dx
i =1 j =1
i =1 j =1 k =1 dt
0
0
N
i =1 j =1 k =1
1
d 2 ai
a
a
a
j k p ni jk p dx
2
p =1 dt
0
N
+ (3 + 6 ) ai a j ak ( si ) 2 ni jk dx
i =1 j =1 k =1
ai a j ak a p ( si ) 2 ni jk p dx (1 + 6 ) 2V 2 n dx
i =1 j =1 k =1 p =1
=0,
4.3
n = 1, 2, ......., N
(4.7)
carried out by numerically solving the non-explicit set of nonlinear ODEs obtained
retaining five ( N = 5) modes in Eq. (4.7). The time step used in the simulations presented
here is 0.03 s which ensures a large number ( 95 ) of simulation points per time period
even for the highest simulated frequency ( 350 kHz ). The design parameters used in the
present analysis are l = 90 m , b =10 m ,
91
h = 2 m ,
d 0 = 1.4 m ,
E = 160 GPa ,
Chapter 4
Firstly, a static analysis is carried out to better understand pull-in instability under
dynamic excitation as discussed in the subsequent parts of this section. The static
deflections are obtained by numerically solving the set of nonlinear algebraic equations
obtained from Eq. (4.7) while considering ai (t ) being independent of time, VAC equal to
92
Nonlinear Response
zero, and all the time derivatives set equal to zero. Figure 4.1 reveals the existence of two
equilibrium solutions w ( x = 1) for each value of the gradually applied DC voltage VDC .
As discussed earlier in Section 2.5.1, the lower stable branch and the upper branch of
unstable fixed points (saddles) coalesce at certain voltage ( 71V ) known as static pull-in
voltage.
4.3.1
Figure 4.2: Frequency-response curves for various AC amplitudes below dynamic pull-in.
93
Chapter 4
94
Nonlinear Response
in Figure 4.2, for VAC equal to 1.0 V , a forward frequency-sweep past the bifurcation
point at frequency 258.16 kHz results in a jump up of the system from a smaller orbit
corresponding to the lower stable branch to a larger orbit on the upper stable branch.
Figure 4.3 shows such a time-history evolution for an excitation frequency ( 258.17 kHz )
past the bifurcation point. Consequently, upon reversal of the frequency-sweep, the
system settles down (Figures 4.2 and 4.4) on the lower stable branch at the same critical
bifurcation value of 258.16 kHz , with no hysteresis.
appreciable
peak
is
observed
95
at
VAC
equal
to
1.5 V
Chapter 4
96
Nonlinear Response
Figure 4.8: Comparison of phase plots for (a) primary resonance and (b) superharmonic
resonance of order two.
though the superharmonic resonant amplitude is much smaller than the amplitude near
primary resonance. Figure 4.8 compares the phase-plane curves for primary and
superharmonic resonances. Two cycles are observed for superharmonic resonance of
97
Chapter 4
order two. As shown in Figure 4.8, the geometrical asymmetry of the orbits confirms the
nonlinear state of the system [Nayfeh and Balachandran, 1995].
4.3.2
Figure 4.9: Frequency-response curve showing the onset of dynamic pull-in instability.
98
Nonlinear Response
displacement w = 0.733 , the motion diverges and the cantilever microbeam collapses
onto the ground electrode (pull-in).
Chapter 4
Under identical excitation conditions, a variation in initial condition close to the
orbit of bifurcation point D is next shown to bring a qualitative change in the system
response. For a small variation in single initial condition da1 dt from a value of
( 0.0010) to ( 0.0006) , the time-history (Figure 4.12) as well as the phase plot (Figure
4.13) depicts a stable and periodic motion. It is interesting to note from the phase plot that
the initial condition triggers a transient response that approaches very close to the saddle
followed by a retreat to settle onto a stable orbit. Further to comment on the possible
existence of such a stable solution, the size of its basin of attraction [Thompson and
Stewart, 2001] needs to be estimated by long-time integration of Eq. (4.7) for various
combinations of initial velocity and displacement conditions. The present work employs a
five-mode ROM which corresponds to a ten-dimensional phase space. Consequently,
generating the basins of attraction is a nontrivial task for the present study unlike the
work of Alsaleem et al. (2009, 2010) which deals with a two-dimensional phase space.
However, given the large size of the orbit (Figures 4.11 and 4.13) and the proximity to the
saddle, the present analysis predicts that there exists a fair chance for the microcantilever
to experience dynamic pull-in near bifurcation point D.
Figure 4.12: Deflection time-history showing the impact of variation in initial condition
close to the orbit of bifurcation point D of Figure 4.9.
100
Nonlinear Response
Figure 4.13: Phase plot corresponding to the time-history of Figure 4.12. represents the
location of the saddle.
4.3.3
amplitudes higher than the dynamic pull-in voltage ( 1.86 V ) are shown in Figure 4.14.
For AC amplitude of 2.0 V , the frequency-response curve is similar to that for 1.86 V
with a jump of the system to a larger stable orbit on the upper branch in course of forward
frequency-sweep and a jump to pull-in near primary resonance in course of the reverse
sweep. On further increasing the AC amplitude, a conspicuous change is observed near
primary resonance as shown in Figure 4.14. For an AC amplitude of 5.0 V , the figure
shows a frequency band ( 220.08 kHz ~ 291.74 kHz ) where no stable solution of the
system exist as the system escapes to dynamic pull-in. A forward frequency sweep past
bifurcation point E ( 220.08 kHz ) and a reverse sweep past bifurcation point F
( 291.74 kHz ) result in dynamic pull-in instability. On increasing the AC amplitude to
6.0 V , as shown in Figure 4.14, the frequency interval further opens up around the
primary resonance and the locations of the bifurcation points shift to higher values of the
maximum displacement wmax ( x = 1) of the tip of the microcantilever.
101
Chapter 4
Figure 4.14: Frequency-response curves for various AC amplitudes above dynamic pullin.
Figure 4.15 compares the phase plots for an excitation frequency ( 220.09 kHz )
past the bifurcation point E (Figure 4.14) for two sets of initial conditions close to the
orbit of bifurcation point E. The initial condition da1 dt is set equal to ( 0.0021 ) in
Figure 4.15a and ( 0.0027 ) in Figure 4.15b while the rest of the initial conditions are kept
identical. As shown in Figure 4.15a, the transient response oscillates for a while as it
approaches the saddle ( w = 0.733 ) and ultimately escapes to pull-in. On the other hand,
Figure 4.15b shows that the transient response settles onto a stable orbit quite away from
the saddle. Thus, the frequency bands (EF and GH) of no stable solution shown in Figure
4.14 are not necessarily the inevitable escape bands [Thompson and Stewart, 2001] since
in these frequency intervals the system may acquire a stable state (for example in Figure
4.15b) under specific initial conditions. Dynamic pull-in scenario observed near
superharmonic resonance ( 120.60 kHz ) of order two in course of the reverse sweep is
shown in Figure 4.14 for an AC amplitude of 6.0 V , and the corresponding phase plot is
shown in Figure 4.16. Existence of superharmonic resonances of higher order is revealed
by Figure 4.14, and the corresponding phase plots are shown in Figure 4.17.
102
Nonlinear Response
Figure 4.15: Comparison of phase plots for two sets of initial condition da1 dt equal to
(a) 0.0021 and (b) 0.0027 , close to the orbit of bifurcation point E of Figure 4.14.
103
Chapter 4
104
Nonlinear Response
Figure 4.17: Phase plots for superharmonic resonances of order (a) three and (b) four.
4.3.4
frequency with the amplitude of AC excitation VAC , is compared for primary (Figure
4.18a) excitation as well as superharmonic (Figure 4.18b) excitation of order two. Figure
4.18 shows the variation of resonance frequency with AC amplitude till the onset of
dynamic pull-in for two values of DC bias. At higher DC bias, as shown in both the
figures, the softening effect of AC loading is more pronounced as indicated by the rapid
decrease of resonance frequency with increase in AC amplitude. Moreover, the onset of
dynamic pull-in is observed at larger AC amplitude when the DC bias voltage is small.
This can be explained as at lower DC bias, the saddle is located at higher w (equal to
0.830 for VDC equal to 50 V as obtained from Figure 4.1) and the system requires a
comparatively larger AC amplitude so that under the combined effect of DC and AC
loading the transient response approaches the saddle and ultimately escapes to pull-in.
Consequently, the dynamic displacement range of the system before pull-in increases
when the system is actuated by a comparatively small DC bias and large AC excitation.
Comparison of Figures 4.18a and 4.18b reveals that under superharmonic excitation the
resonance frequency at a particular DC bias varies almost linearly for small AC
105
Chapter 4
amplitudes indicating a comparatively weak softening effect. Further, at a particular DC
bias, the amplitude of AC excitation corresponding to dynamic pull-in is larger for
superharmonic excitation than it is for primary excitation.
106
Nonlinear Response
4.4
Summary
The chapter presents the nonlinear dynamic behaviour of a cantilever microbeam
107
Chapter 5
CONCLUDING REMARKS AND RECOMMENDATIONS
Outline of the chapter: 5.1 Conclusions, 5.1.1 Static and Transient Behaviour under
Pure DC Actuation, 5.1.2 Small Amplitude Dynamics under Combined Actuation, 5.1.3
Nonlinear Dynamics under Combined Actuation, 5.2 Recommendations for Future Work
5.1
Conclusions
The present dissertation focuses on the static and dynamic behavioural analysis of
electrostatically actuated cantilever microbeams. Simulation studies have been carried out
to predict the behaviour under actuations by pure DC voltage, and, combined DC and
harmonic AC voltage. Device models of deformable cantilever electrodes separated from
the ground electrode by small and large gaps have been considered. Special emphasis has
been laid on modelling of the electrostatic domain, for large gap separations, with
inclusion of the higher order correction terms. The structure-electrostatic model described
for small gap separation problems has been further coupled with the fluid domain to
account for the squeeze-film damping influences.
The set of governing equations of motion of the system has been obtained by the
application of Hamiltons principle to the energy functions. For the case of damped
systems, the equation of motion is coupled to the two dimensional Reynolds equation to
capture the characteristics of squeeze-film damping. Reduced-order model (ROM)
developed from the related boundary value problem have been effectively used in the
present thesis to numerically simulate static behaviour, as well as, local and global
dynamics. The ROM uses few linear undamped mode shapes of a straight cantilever beam
as basis functions in a Galerkin procedure, and hence reduces the complexity of the
simulation. Moreover, the ROM is found to be effective in predicting many possible
nonlinear phenomena in electrostatically actuated MEMS devices.
5.1.1
to beam-length ratios, actuated by DC electrostatic force has been studied with special
109
Chapter 5
emphasis on the nonlinear effects due to geometry, electric forces, and inertial terms. The
deflections near pull-in being large, model derived on the basis of large deflection
assumption effectively predicts the behaviour. In case there is large gap between
deformable conductor and ground plane, it is essential to consider higher order
corrections of electrostatic forces during the formulation of the model. In the present
work, it has been shown that results are much improved when second order correction
terms are taken into account during static and transient analysis. The numerical results of
the static analysis are validated with experimental and analytical results available in open
literature. According to the results of static analysis, where the voltages are applied
gradually, a linear model can appreciably predict the static behaviour for small
electrostatic forces as the deflections are small. For higher strengths of electrostatic forces
close to pull-in, coupled effects of structural nonlinearity, and nonlinear electrostatic
forces with higher order correction terms cause deviation from the linearized results while
the effects are significant for initial gap to beam-length ratios of 0.3 and above.
Consideration of nonlinearities gives a better estimation of the stability limits which can
be advantageously used for design of non-pull-in devices.
The numerical results of the dynamic analysis are validated with other numerical
results available in open literature. According to the results obtained, nonlinear
electrostatic forces and the inertial effects are observed to cause softening of the
microstructure whereas geometric nonlinearity has got a stiffening effect on the
microstructure. The overall effect, in the entire deflection range, due to the nonlinearities
may be stiffening/softening depending on the relative strengths of the nonlinearities. At
and in the vicinity of pull-in, geometric nonlinearity has got significant effect on the
response characteristics for systems with initial gap to beam-length ratios of 0.3 and
above. For actuations at applied voltages well below the dynamic pull-in voltage,
geometric nonlinearity does not play any significant role. Thus, for applications in and
around the pull-in zone, the large deflection model needs to be considered for effective
design of microcantilever based microsystems.
5.1.2
Conclusions
has been used to study the squeeze-film damping characteristics of cantilever
microresonators for the first three flexural modes of vibration in the rarefied and noninertial flow regime. Elastic beam equation for small air-gap coupled with linearized
compressible Reynolds equation has been used to model the small amplitude motions.
The pressure function assumed as a product of a parabolic function along the beam width
and a cosine series along the beam length gives converged results for computationally
reasonable number of pressure terms. Rarefaction effects modelled utilizing the effective
coefficient of viscosity approach accounts for the dependency of the effective viscosity
on the gap height through the use of a variable Knudsen number. The model is validated
by comparing the results with the analytical, numerical, and experimental results
available in open literature as well as through ANSYS simulations. The results show that
the quality factor is higher for higher mode of vibration for the entire range of DC
polarization. With variations in DC bias voltage, quality factor and resonance frequency
are most affected in the first mode and are found to decrease with an increase in DC
voltage. For flexible microbeams, DC bias voltage in conjunction with mode shape has a
profound effect on the back pressure distribution which ultimately affects the squeezefilm characteristics.
The present study also focuses on a range of ambient pressures limited to the
viscous flow regime considering rarefaction effects. Under rarefied flow conditions, the
results show that at a certain DC voltage the quality factor increases with decrease in
ambient pressure. The amount of increase in quality factor is more or less uniform for the
entire range of DC polarization. On the other hand, for low ambient pressures, there is not
much change in resonance frequency to be considered as a response characteristic of the
resonator (e.g., pressure sensor) at the first harmonic. For higher modes, the resonance
frequency is found to be invariant with pressure as well as DC voltage variations. Further,
a comparison of quality factor values computed with compressible and incompressible
flow models shows that the incompressible flow model overestimates the damping and
the effect is more pronounced for higher compressibility ratio.
5.1.3
excitation has been carried out to provide further insight into the dynamic phenomena
exhibiting rich nonlinearities. The large amplitude dynamic behaviour of cantilever
microbeams under the combined effect of DC bias and harmonic AC excitation is studied
111
Chapter 5
in presence of squeeze-film damping. A distributed parameter model for small gap
separation is derived accounting for the nonlinearities of the system due to nonlinear
electrostatic forces and nonlinear squeeze-film damping. From the coupled field
governing equations, a ROM is developed and numerically simulated to study the
dynamics for the fundamental flexural mode of vibration under primary and
superharmonic AC excitations.
For large DC bias, the onset of local bifurcation phenomena such as resonance
shift, and jump in the response are observed for comparatively small AC amplitudes. On
further increasing the AC amplitude, the system response jumps to pull-in, indicating the
onset of a global bifurcation phenomenon, typically in the course of reverse frequencysweep. The simulations reveal that dynamic pull-in is marked by the approach of the
system transient response towards the unstable fixed point saddle, corresponding to the
DC bias, and subsequent divergence of the motion. However, the system response is
sensitive to initial conditions, and consequently, the excitation that triggers dynamic pullin may lead to a stable periodic motion for specific initial conditions. At higher electric
loads, the present study shows frequency bands where no stable solution of the system
exist. Response sensitivity to initial conditions, however, reveals that these are not
necessarily the frequency bands of inevitable escape to dynamic pull-in.
The present work also focuses on the effect of DC bias on the dynamic behaviour
under harmonic AC excitations. The nonlinear effects of AC excitation are observed to be
more pronounced for comparatively large DC bias with the response behaviour, under
certain loading combination, being comparatively more nonlinear near primary resonance
than it is near superharmonic resonance. Moreover, the stable dynamic displacement
range is found to increase when the system is actuated by a comparatively small DC bias
and large AC excitation which can be put to advantageous use in non-pull-in devices.
Alternatively, a comparatively large DC bias combined with a small AC excitation may
be effectively used for pull-in devices.
5.2
and successful in large scale, major issues concerning successful static and dynamic
design of microstructures needs to be tackled. The present thesis work has been envisaged
to formulate realistic models of such microstructure based systems and simulate their
112
Conclusions
behaviour. Varying the complexity of models with assumptions made in the involved
physical domains of the coupled field MEMS analysis, further opens up different domains
of research. Following is the list of recommendations and notes for future work:
The structural domain is modelled, in the present work, as one dimensional beam
and the electrostatic force is assumed to be uniform across the width of the beam.
Modelling of the coupled field problem for two dimensional bodies like plates,
diaphragms etc. separated by large gaps, and, subsequent behavioural analysis can
be a potential area of research.
In the present dissertation, analysis for large gap separation problem neglects any
form of damping in the system. For narrow gaps, viscous damping corresponds to
squeeze-film damping which is modelled using the Reynolds equation. However,
for large gaps, inertia forces need to be considered which calls for modelling the
fluid domain using the Navier-Stokes equations. The dynamic problem coupled in
the electrostatic, structural, and fluid domains for large gap separations remains to
be a potential area and can be taken up as an extension of the present thesis work.
Nonlinear dynamic behaviour in the subharmonic frequency range has not been
evaluated in the present work. The present model restricts the operating
frequencies to superharmonic and primary resonances so as to avoid frequencies
high enough to nullify the incompressible flow assumptions. An improved
nonlinear model incorporating the compressible effects of the air-film needs to be
formulated to carry out further investigations.
In case of pull-in devices like microswitches, attention has to be paid to model the
interaction between the microcantilever tip and the ground electrode as the
microbeam makes contact with the ground substrate at pull-in. The present work
can be extended by incorporating a suitable contact model in the present
formulation.
The present work can be extended to investigate the nonlinear dynamic behaviour
of a large array of interacting microcantilevers. The present model incorporating
113
Chapter 5
elastic nonlinearity of individual cantilever oscillators may be modified to
accommodate the effects of nonlinear coupling between adjacent oscillators.
An investigation needs to be carried out into the effects neglected while modelling
the behaviour of cantilever microbeams. These include the fringing effects of the
electrical field, and stress gradients which in released cantilevers cause them to
curl out of plane.
114
References
A
Abdel-Rahman, E.M., Younis, M.I. and Nayfeh, A.H., 2002, Characterization of the
mechanical behaviour of
of
115
B
Ballestra, A., Som, A. and Pavanello, R., 2008, Experimental-numerical comparison of
the cantilever MEMS frequency shift in presence of a residual stress gradient, Sensors
8, 767-783.
Bao, M. and Yang, H., 2007, Squeeze-film air damping in MEMS, Sensors and Actuators
A 136, 327.
Bao, M., Yang, H., Sun, Y. and French, P., 2003a, Modified Reynolds equation and
analytical analysis of squeeze-film air damping of perforated structures, Journal of
Micromechanics and Microengineering 13, 795800.
Bao, M., Yang, H., Sun, Y. and Wang, Y., 2003b, Squeeze-film air damping of thick
hole-plate, Sensors and Actuators A 108, 212217.
Bao, M., Yang, H., Yin, H. and Shen, S.Q., 2000, Effects of electrostatic forces generated
by the driving signal on capacitive sensing devices, Sensors and Actuators A 24, 213
219.
Bao, M., Yang, H., Yin, H. and Sun, Y., 2002, Energy transfer model for squeeze-film air
damping in low vacuum, Journal of Micromechanics and Microengineering 12, 341346.
Batra, R.C., Porfiri, M. and Spinello, D., 2006, Electromechanical Model of Electrically
Actuated Narrow Microbeams, Journal of Microelectromechanical Systems 15(5),
1175-1189.
Batra, R.C., Porfiri, M. and Spinello, D., 2007, Review of modelling electrostatically
actuated microelectromechanical systems, Smart Materials and Structures 16, R23
R31.
Batra, R.C., Porfiri, M. and Spinello, D., 2008, Vibrations of narrow microbeams
predeformed by an electric field, Journal of Sound and Vibration 309, 600-612.
Beeby, S., Ensell, G., Kraft, M. and White, N., 2004, MEMS mechanical sensors, Artech
House Inc., Norwood, MA.
Blech, J.J., 1983, On isothermal squeeze-films, Journal of Lubrication Technology A 105,
615-620.
Brusa, E., De Bona, F., Antonio, G. and Som, A., 2004, Modelling and prediction of the
dynamic behaviour of microbeams under electrostatic load, Analog Integrated
Circuits and Signal Processing 40, 155-164.
116
Burgdorfer, A., 1959, The influence of the molecular mean free path on the performance
of hydrodynamic gas lubricated bearings, Journal of Basic Engineering, Trans. ASME
81, 94-99.
C
Chakraborty, S., Swamy, K.B.M., Sen, S. and Bhattacharyya, T.K., 2010, An
experimental analysis of electrostatically vibrated array of polysilicon cantilevers,
Microsyst Technol 16, 2131-2145.
Chan, E.K., Garikipati, K. and Dutton, R.W., 1999, Characterization of contact
electromechanics through capacitance-voltage measurements and simulations, Journal
of Microelectromechanical Systems 8(2), 208-217.
Chang, K.M., Lee, S.C. and Li, S.H., 2002, Squeeze-film damping effect on a MEMS
torsion mirror, Journal of Micromechanics and Microengineering 12, 556-561.
Chao, C-P.P., Chiu, C-W. and Tsai, C-Y., 2006, A novel method to predict the pull-in
voltage in a closed form for micro-plates actuated by a distributed electrostatic force,
Journal of Micromechanics and Microengineering 16, 986998.
Chen Kuo-Shen and Ou Kuang-Shun, 2007, Development and verification of 2D dynamic
electromechanical coupling solver for micro-electrostatic-actuator applications,
Sensors Actuators A 136, 403-411.
Choi, B. and Lovell, E.G., 1997, Improved analysis of microbeams under mechanical and
electrostatic loads, Journal of Micromechanics and Microengineering 7, 24-29.
Chowdhury, S., Ahmadi, M. and Miller, W.C., 2005, A closed-form model for the pull-in
voltage of electrostatically actuated cantilever beams, Journal of Micromechanics and
Microengineering 15, 756763.
Christian, R.G., 1966, The theory of oscillating-vane vacuum gauges, Vacuum 16, 175178.
CoventorWare Documentation, 2003, Version 2003.1, Coventor Inc.
Cowper, G.R., 1968, On the accuracy of Timoshenkos beam theory, Journal of the
Engineering Mechanics Division, 94, 14471453.
Crespo da Silva, M.R.M. and Glynn, C.C., 1978a, Nonlinear flexural-flexural-torsional
dynamics of inextensional beams. I:
Mechanics 6, 437-448.
117
D
Darling, R.B., Hivick, C. and Xu, J., 1998, Compact analytical modelling of squeeze-film
damping with arbitrary venting conditions using a Greens function approach, Sensors
and Actuators A 70, 32-41.
De, S.K. and Aluru, N.R., 2004, Full-Lagrangian schemes for dynamic analysis of
electrostatic MEMS, Journal of Microelectromechanical Systems 13, 737-758.
De, S.K. and Aluru, N.R., 2006, Complex nonlinear oscillations in electrostatically
actuated microstructures, Journal of Microelectromechanical Systems 15, 355-369.
De, S.K. and Aluru, N.R., 2006, Coupling of hierarchical fluid models with electrostatic
and mechanical models for the dynamic analysis of MEMS, Journal of
Micromechanics and Microengineering 16, 17051719.
Decuzzi, P., Granaldi, A. and Pascazio, G., 2007, Dynamic response of microcantileverbased sensors in a fluidic chamber, Journal of Applied Physics 101, 024303.
Dequesnes, M., Tang, Z. and Aluru, N.R., 2004, Static and dynamic analysis of carbon
nanotube-based switches, Journal of Engineering Materials and Technology 126,
230237.
Ding, W., Guo, Z. and Ruoff, R.S., 2007, Effect of cantilever nonlinearity in nanoscale
tensile testing, Journal of Applied Physics 101, 034316.
E
Elmer: Finite Element Software for Multiphysical Problems, <http://www.csc.fi/elmer>.
Evan-Iwanowski, R.M., 1976, Resonance oscillations in mechanical systems, Elsevier,
NY.
118
F
Fargas-Marques, A., Casals-Terr, J. and Shkel, A.M., 2007, Resonant pull-in condition
in parallel-plate electrostatic actuators, Journal of Microelectromechanical Systems
16, 10441053.
Fischer, M., Giousouf, M., Schaepperle, J., Eichner, D., Weinmann, M., von Miinch, W.
and Assmus, F., 1998, Electrostatically deflectable polysilicon micromirrors- dynamic
behaviour and comparison with the results from FEM modelling with ANSYS,
Sensors and Actuators A 67, 89-95.
Foda, M.A., 1999, Influence of shear deformation and rotary inertia on nonlinear free
vibration of a beam with pinned ends, Computers and Structures 71, 663670.
Fricke, J. and Obermaiar, C., 1993, Cantilever beam accelerometer based on surface
micromachining technology, Journal of Micromechanics and Microengineering 3,
100102.
G
Gabbay, L.D., Mehner, J.E. and Senturia, S.D., 2000, Computer-aided generation of
nonlinear reduced-order dynamic macromodels I: Non-stress-stiffened case, Journal
of Microelectromechanical Systems 9, 262269.
Gaillard, J., Skove, M.J., Ciocan, R. and Rao, A.M., 2006, Electrical detection of
oscillations in microcantilevers and nanocantilevers, Review of Scientific Instruments
77, 073907.
Gorthi S., Mohanty A. and Chatterjee A., 2006, Cantilever beam electrostatic MEMS
actuators beyond pull-in, Journal of Micromechanics and Microengineering 16, 18001810.
Granaldi, A. and Decuzzi, P., 2006, The dynamic response of resistive microswitches:
switching time and bouncing, Journal of Micromechanics and Microengineering 16,
1108-1115.
Griffin, W.S., Richarson, H.H. and Yamanami, S., 1966, A study of squeeze-film
damping, ASME Journal of Basic Engineering, 451-456.
Gupta, M.C., 1990, Statistical Thermodynamics, Wiley, New Delhi.
119
H
Haghighi, H.S. and Amir Markazi, H.D., 2010, Chaos prediction and control in MEMS
resonators, Commun Nonlinear Sci Numer Simulat 15, 3091-3099.
Hamrock, B.J., 1994, Fundamentals of Fluid Film Lubrication, McGraw-Hill, NY.
Harish, K.M., Gallacher, B.J., Burdess, J.S. and Neasham, J.A., 2009, Experimental
investigation of parametric and externally forced motion in resonant MEMS sensors,
Journal of Micromechanics and Microengineering 19, 015021.
Hassanpour, P.A., Cleghorn, W.L., Esmailzadeh, E. and Mills, J.K., 2007, Vibration
analysis of micro-machined beam-type resonators, Journal of Sound and Vibration
308, 287-301.
Hirai, Y., Marushima, Y., Soda, S., Jin, D., Kawata, H., Inoue, K. and Tanaka, Y., 2000,
Electrostatic actuator with novel shaped cantilever, Proceedings of the International
Symposium on Micromechanics and Human Science, 223-227.
Hodges, D.H., 1984, Proper definition of curvature in nonlinear beam kinematics,
American Institute of Aeronautics and Astronautics Journal 22, 1825-1827.
Hsieh, S.R., Shaw, S.W. and Pierre, C., 1994, Normal modes for large amplitude
vibration of a cantilever beam, International Journal of Solids and Structures 31,
1981-2014.
Hu, Y.C., Chang, C.M. and Huang, S.C., 2004, Some design considerations on the
electrostatically actuated microstructures, Sensors Actuators A 112, 155-161.
Huang, J.M., Liew, K.M., Wong, C.H., Rajendran, S., Tan, M.J. and Liu, A.Q., 2001,
Mechanical design and optimization of capacitive micromachined switch, Sensors
Actuators A 93, 273285.
Hung, E.S. and Senturia, S.D., 1999a, Generating efficient dynamical models for
microelectromechanical systems from a few finite-element simulations runs, Journal
of Microelectromechanical Systems 8, 280289.
Hung, E.S. and Senturia, S.D., 1999b, Extending the travel range of analog-tuned
electrostatic actuators, Journal of Microelectromechanical Systems 8(4), 497-505.
I
Ishihara, H., Arai, F. and Fukuda, T., 1996, Micromechatronics and microactuators,
IEEE/ASME Transactions on Mechatronics 1(1), 68-79.
120
J
Jin, Z. and Wang, Y., 1998, Electrostatic resonator with second superharmonic resonance,
Sensors and Actuators A 64, 273-279.
K
Keskar, G., Elliott, B., Gaillard, J., Skove, M.J. and Rao, A.M., 2008, Using electric
actuation and detection of oscillations in microcantilevers for pressure measurements,
Sensors and Actuators A 147, 203-209.
Kim, Jong-Man., Lee, S., Park, Jae-Hyoung., Baek, Chang-Wook., Kwon, Y. and Kim,
Yong-Kweon., 2010, Electrostatically driven low-voltage micromechanical RF
switches using robust single-crystal silicon actuators, Journal of Micromechanics and
Microengineering 20, 095007.
Kovacs, G.T.A., 1998, Micromachined transducers sourcebook, McGraw-Hill, New
York.
Krylov, S. and Maimon, R., 2004, Pull-in dynamics of an elastic beam actuated by
continuously distributed electrostatic force, Journal of Vibration and Acoustics 126,
332-342.
Krylov, S. and Seretensky, S., 2006, Higher order correction of electrostatic pressure and
its influence on the pull-in behaviour of microstructures, Journal of Micromechanics
and Microengineering, 16, 13821396.
Krylov, S., 2007, Lyapunov exponents as a criterion for the dynamic pull-in instability of
electrostatically actuated microstructures, International Journal of Non-Linear
Mechanics 42, 626642.
Krylov, S., Harari, I. and Cohen, Y., 2005, Stabilization of electrostatically actuated
microstructures using parametric excitation, Journal of Micromechanics and
Microengineering 15, 1188-1204.
Kuang, J.H. and Chen, C.J., 2005, The nonlinear electrostatic behaviour for shaped
electrode actuators, International Journal of Mechanical Sciences 47, 11721190.
Kwok, P., Weinberg, M. and Breuer, K., 2005, Fluid effects in vibrating micromachined
structures, Journal of Microelectromechanical Systems 14, 770781.
121
L
Langlois, W.E., 1962, Isothermal squeeze films, Quarterly of Applied Mathematics XX2, 131-150.
Lavrik, N.V., Sepaniak, M.J. and Datskos, P.G., 2004, Cantilever transducers as a
platform for chemical and biological sensors, Review of Scientific Instruments 75(7),
2229-2253.
Lee, S., Ramadoss, R., Buck, M., Bright, V.M., Gupta, K.C. and Lee, Y.C., 2004,
Reliability
testing
of flexible
printed
circuit-based
RF
MEMS
capacitive
122
Liqin,
L.,
Gang,
T.Y.
and
Zhiqiang,
W.,
2006,
Nonlinear
dynamics
of
M
Maali, A., Hurth, C., Boisgard, R., Cohen-Bouhacina, T. and Aime, J.P., 2005,
Hydrodynamics of oscillating atomic force microscopy cantilevers in viscous fluids,
Journal of Applied Physics 97, 074907.
Madou, M.J., 2002, Fundamentals of microfabrication, CRC Press, Boca Raton, FL.
Mahmoodi, S.N. and Jalili, N., 2007, Non-linear vibrations and frequency response
analysis
of
piezoelectrically
driven
123
Mestrom, R.M.C., Fey, R.H.B., van Beek, J.T.M., Phan, K.L. and Nijmeijer, H., 2008,
Modelling the dynamics of a MEMS resonator: Simulations and experiments, Sensors
and Actuators A 142, 306-315.
Millet, O., Bernardoni, P., Regnier, S., Bidaud, P., Tsitsiris, E., Collard, D. and
Buchaillot, L., 2004, Electrostatic actuated micro gripper using an amplification
mechanism, Sensors and Actuators A 114, 371-378.
Minikes, A., Bucher, I. and Avivi, G., 2005, Damping of a micro-resonator torsion mirror
in rarefied gas ambient, Journal of Micromechanics and Microengineering 15 17621769.
Moon, F.C., 1987, Chaotic vibrations- An introduction for applied scientists and
engineers, Wiley, NY.
N
Najar, F., Choura, S., Abdel-Rahman, E.M., El-Borgi, S. and Nayfeh, A., 2006, Dynamic
analysis of variable-geometry electrostatic microactuators, Journal of Micromechanics
and Microengineering 16, 2449-2457.
Naniwa I., Makamura S., Saegusa S. and Sato K., 1999, Low voltage driven piggy-back
actuator of hard disk drives, Proceedings of IEEE Micro Electro Mechanical Systems,
Orlando, USA, 49-52.
Nathanson, H.C., Newell, W.E., Wickstrom, R.A. and Davis, J.R., 1967, The resonant
gate transistor, IEEE Transactions of Electron Devices ED-14(3), 117133.
Nayfeh, A.H. and Balachandran, B., 1995, Applied Nonlinear Dynamics, Wiley, NY.
Nayfeh, A.H. and Mook, D.T., 1979, Nonlinear oscillations, Wiley, NY.
Nayfeh, A.H. and Younis, M.I., 2004, A new approach of the modelling and simulation of
flexible microstructures under the effect of squeeze-film damping, Journal of
Micromechanics and Microengineering 14, 170181.
Nayfeh, A.H. and Younis, M.I., 2005, Dynamics of MEMS resonators under
superharmonic and subharmonic excitations, Journal of Micromechanics and
Microengineering 15, 1840-1847.
Nayfeh, A.H., Chin, C. and Nayfeh, S.A., 1995, Nonlinear normal modes of a cantilever
beam, Journal of Vibration and Acoustics 117, 477-481.
Nayfeh, A.H., Younis, M.I. and Abdel-rahman, E.M., 2005, Reduced-order models for
MEMS applications, Nonlinear Dynamics 41, 211236.
124
Nayfeh, A.H., Younis, M.I. and Abdel-Rahman, E.M., 2007, Dynamic pull-in
phenomenon in MEMS resonators, Nonlinear Dynamics 48, 153-163.
Newell, W.E., 1968, Miniaturization of tuning forks, Science 161, 13201326.
Nguyen, C.T.-C., October 2004, Vibrating RF MEMS for next generation wireless
applications, Proceedings of the IEEE Custom Integrated Circuits Conference,
Orlando, Piscataway, FL 36.
Nielson, G.N. and Barbastathis, G., 2006, Dynamic pull-in of parallel-plate and torsional
electrostatic MEMS actuators, Journal of Microelectromechanical Systems 15, 811821.
Nouira, H., Foltte, E., Hirsinger, L. and Ballandras, S., 2007, Investigation of the effects
of air on the dynamic behaviour of a small cantilever beam, Journal of Sound and
Vibration 305, 243260.
Nugaeva, N., Gfeller, K.Y., Backmann, N., Lang, H.P., Dggelin, M. and Hegner, M.,
2005, Micromechanical cantilever array sensors for selective fungal immobilization
and fast growth detection, Biosensors and Bioelectronics 21(6), 849-856.
O
Oden, P.I., Chen, G.Y., Steele, R.A., Warmack, R.J. and Thundat, T., 1996, Viscous drag
measurements utilizing microfabricated cantilevers, Applied Physics Letters 68, 14651469.
Osterberg, P.M. and Senturia, S.D., 1997, M-TEST: a test chip for MEMS material
property measurement using electrostatically actuated test structures, Journal of
Microelectromechanical Systems 6, 107-118.
Osterberg, P.M., Yie, H., Cai, X., White, J. and Senturia, S., 1994, Self-consistent
simulation and modelling of electrostatically deformed diaphragms. Proceedings of
the IEEE Conference on Micro Electro Mechanical Systems, Osio, Japan, 2832.
Ouakad, H.M. and Younis, M.I., 2010, The dynamic behaviour of MEMS arch resonators
actuated electrically, International Journal of Non-Linear Mechanics 45, 704-713.
P
Pai, P.F. and Nayfeh, A.H., 1990, Non-linear non-planar oscillations of a cantilever beam
under lateral base excitations, International Journal of Non-Linear Mechanics, 25,
455-474.
125
Pamidighantam, S., Puers, R., Baert, K. and Tilmans, H.A.C., 2002, Pull-in voltage
analysis of electrostatically actuated beam structures with fixed-fixed and fixed-free
end conditions, Journal of Micromechanics and Microengineering 12, 458464.
Pan, F., Kubby, J., Peeters, E., Tran, A.T. and Mukherjee, S., 1998, Squeeze-film
damping effect on the dynamic response of a MEMS torsion mirror, Journal of
Micromechanics and Microengineering 8, 200-208.
Pandey, A.K. and Pratap, R., 2007, Effect of flexural modes on squeeze film damping in
MEMS cantilever resonators, Journal of Micromechanics and Microengineering 17,
24752484.
Pandey, A.K., Pratap, R. and Chau, F.S., 2007, Influence of boundary conditions on the
dynamic
characteristics
of
squeeze-films
in
MEMS
devices,
Journal
of
J.A.
and
Bernstein,
D.H.,
2003,
Modelling
MEMS
and
NEMS,
126
Q
--------R
Rabaey, J.M., 1996, Digital Integrated Circuits, Englewood Cliffs, Prentice-Hall, NJ,
438445.
Ramezani, A., Alasty, A. and Akbari, J., 2006, Effects of rotary inertia and shear
deformation on nonlinear free vibration of microbeams, Journal of Vibration and
Acoustics 128, 611-615.
Rao, G.V., Meera Saheb, K. and Janardhan, G.R., 2006, Fundamental frequency for large
amplitude vibrations of uniform Timoshenko beams with central point concentrated
mass using coupled displacement field method, Journal of Sound and Vibration 298,
221232.
Rocha, L.A., Cretu, E. and Wolffenbuttel, R.F., 2004, Behavioural analysis of the pull-in
dynamic transition, Journal of Micromechanics and Microengineering 14, S37-S42.
Rocha, L.A., Cretu, E. and Wolffenbuttel, R.F., 2006, Using dynamic voltage drive in a
parallel-plate electrostatic actuator for full-gap travel range and positioning, Journal
of Microelectromechanical Systems 15(1), 69-83.
Rochus, V., Rixen, D.J. and Golinval, J.C., 2005, Electrostatic coupling of MEMS
structures: transient simulations and dynamic pull-in. Nonlinear Analysis 63, e1619e1633.
S
Sadd, M. and Stiffer, A., 1975, Squeeze-film dampers: amplitude effects at low squeeze
numbers, Journal of Engineering for Industry, Transactions of the ASME 97, 13661370.
Sadeghian, H., Rezazadeh, G. and Osterberg, P.M., 2007, Application of the generalized
differential quadrature method to the study of pull-in phenomena of MEMS switches,
Journal of Microelectromechanical Systems 16, 13341340.
Sandberg, R., Svendsen, W., Molhave, K. and Boisen, A., 2005, Temperature and
pressure dependence of resonance on multi-layer microcantilevers, Journal of
Micromechanics and Microengineering 15, 1454-1458.
127
Sasayama, T., Suzuki, S., Tsuchitani, S., Koide, A., Suzuki, M., Ichikawa, N. and
Nakazawa, T., 1996, Highly reliable silicon micromachined physical sensors in mass
production, Sensors and Actuators A 54, 714-717.
Sazonova, V., Yaish, Y., Ustunel, H., Roundy, D., Arias, T.A. and McEuen, P.L., 2004, A
tunable carbon nanotube electromechanical oscillator, Nature 431, 284-287.
Schiele, I., Huber, J., Hillerich, B. and Kozlowski, F., 1998, Surface-micromachined
electrostatic microrelay, Sensors and Actuators A 66, 345-354.
Senturia, S.D., 2001, Microsystem Design, Norwell, Kluwer, MA.
Senturia, S.D., Harris, R.M., Johnson, B.P., Kim, S., Nabors, K., Shulman, M.A. and
White, J.K., 1992, A computer-aided design system for Microelectromechanical
systems (MEMCAD), Journal of Microelectromechanical Systems 1, 313.
Shames, I.H. and Dym, C.L., 1985, Energy and finite element methods in structural
mechanics, McGraw-Hill, NY.
Shampine, L.F., Reichelt, M.W. and Kierzenka, J., Solving boundary value problems for
ordinary differential equations in MATLAB with bvp4:<http://www.mathworks.com>
Som, A. and Pasquale, G.D., 2008, Numerical and experimental comparison of MEMS
suspended plates dynamic behaviour under squeeze-film damping effect, Analog
Integr Circ Sig Process 57(3), 213-224.
Starr, J.B., 1990, Squeeze-film damping in solid-state accelerometers, Proceedings of
IEEE Solid-State Sensor and Actuator Workshop, 44-47.
Sumali, H., 2007, Squeeze-film damping in the free molecular regime: model validation
and measurement on a MEMS, Journal of Micromechanics and Microengineering 17,
2231-2240.
T
Taylor, G.I., 1968, The coalescence of closely spaced drops when they are at different
electric potentials, Proceedings of Royal Society A 306, 423-434.
Teva, J., Abadal, G., Davis, Z.J., Verd, J., Borris, X., Boisen, A., Prez-Murano, F. and
Barniol, N., 2004, On the electromechanical modeling of a resonating nanocantilever-based transducer, Ultramicroscopy 100, 225-232.
Teva, J., Abadal, G., Torres, F., Verd, J., Prez-Murano, F. and Barniol, N., 2006, A
femtogram resolution mass sensor platform, based on SOI electrostatically driven
128
U
--------V
Veijola, T. and Lehtovuori, A., 2009, Numerical and analytical modelling of trapped gas
in micromechanical squeeze-film dampers, Journal of Sound and Vibration 319, 606621.
Veijola, T. and Pasquale, G.D., 2008, Comparative numerical study of FEM methods
solving gas damping in perforated MEMS devices, Microfluid Nanofluid 5,517528.
Veijola, T., 2004, Compact models for squeezed-film dampers with inertial and rarefied
gas effects, Journal of Micromechanics and Microengineering 14, 1109-1118.
Veijola, T., Kuisma, H. and Lahdenper, J., 1998, The influence of gas-surface
interaction on gas-film damping in a silicon accelerometer, Sensors and Actuators A
66, 83-92.
Veijola, T., Kuisma, H., Lahdenper, J. and Ryhnen, T., 1995, Equivalent-circuit model
of the squeezed gas film in a silicon accelerometer, Sensors Actuators A 48, 239-248.
Veijola, T., Pursula, A. and Raback, P., 2005, Extending the validity of squeezed-film
dampers models with elongations of surface dimensions, Journal of Micromechanics
and Microengineering 15, 1624-1636.
129
Vyas, A., Peroulis, D. and Bajaj, A.K., 2008, Dynamics of a nonlinear microresonator
based on resonantly interacting flexural-torsional modes, Nonlinear Dynamics 54, 3152.
W
Wagner, H., 1965, Large-amplitude free vibrations of a beam, Journal of Applied
Mechanics 32, 887-892.
Wang, Y.C., Adams, S.G., Thorp, J.S., MacDonald, N.C., Hartwell, P. and Bertsch, F.,
1998, Chaos in MEMS, parameter estimation and its potential application, IEEE
Transactions on Circuits and Systems-I: Fundamental Theory and Applications 45,
1013-1020.
Wilfinger, R.J., Bardell, P.H. and Chhabra, D.S., 1968, The resonistor a frequency
selective device utilizing the mechanical resonance of a silicon substrate, IBM J. 12,
113-118.
X
Xie, W.C., Lee, H.P. and Lim, S.P., 2003, Nonlinear dynamic analysis of MEMS
switches by nonlinear modal analysis, Nonlinear Dynamics 31, 243256.
Y
Younis, M.I. and Nayfeh, A.H., 2007, Simulation of squeeze-film damping of microplates
actuated by large electrostatic load, ASME Journal of Computational and Nonlinear
Dynamics 2, 101112.
Younis, M.I., Abdel-Rahman, E.M. and Nayfeh, A.H., 2003, A reduced-order model for
electrically actuated microbeam-based MEMS, Journal of Microelectromechanical
Systems 12, 672680.
Younis, M.I., Ouakad, H.M., Alsaleem, F.M., Miles R. and Cui, W., 2010, Nonlinear
dynamics of MEMS arches under harmonic electrostatic actuation, Journal of
Microelectromechanical Systems 19(3), 647-656.
130
Z
Zavracky, P.M., Majumder, S. and McGruer, N.E., 1997, Micromechanical switches
fabricated using nickel surface micromachining, Journal of Microelectromechanical
Systems 6(1), 3-9.
Zhang, C., Xu, G. and Jiang, Q., 2004, Characterization of the squeeze-film damping
effect on the quality factor of a microbeam resonator, Journal of Micromechanics and
Microengineering 14, 1302-1306.
Zhang, Y. and Zhao, Ya-pu, 2006, Numerical and analytical study on the pull-in
instability of micro-structure under electrostatic loading, Sensors and Actuators A
127, 366-380.
Zhao, J.P., Chen, H.L., Huang, J.M. and Liu, A.Q., 2005, A study of dynamic
characteristics and simulation of MEMS torsional micromirrors, Sensors and
Actuators A 120, 199-210.
Zhu, J., Ru, C.Q. and Mioduchowski, A., 2009, High-order subharmonic parametric
resonance of multiple nonlinearly coupled micromechanical nonlinear oscillators,
Acta Mech 212(1-2), 69-81.
Ziegler, C., 2004, Cantilever-based biosensors, Anal Bioanal Chem 379, 946-959.
131