Professional Documents
Culture Documents
School of Mechanical Engineering, Faculty of Engineering, Tel Aviv University, Ramat Aviv 69978, Tel Aviv, Israel
School of Applied and Engineering Physics and Cornell Nanoscale Facility, Cornell University, Ithaca, NY, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 16 June 2011
Received in revised form
30 June 2011
Accepted 1 July 2011
Available online 23 September 2011
In this work we report on a nite element modeling and design methodology, fabrication and
characterization of a large-displacement low voltage multistable micro actuator with an integrated
electrostatic comb drive transducer. The compliant suspension of the device incorporates multiple
serially connected bistable arch-shaped beams and exhibits controllable sequential snap-through
buckling under an increasing actuation force. The device can be considered therefore as an example of a
compliant multistep structure. The device is also distinguished by its ability to remain in several
different stable congurations at the same actuation voltage while the forcedisplacement characteristic of the suspension can be tailored by changing the geometry parameters of the exures. A model
built using the shallow arch approximation along with a nonlinear nite element analysis were used in
order to study the inuence of the suspension architecture on the stability limits of the structure and
for evaluation of design parameters of the actuator. Bistable and multistable devices were fabricated by
a Deep Reactive Ion Etching (DRIE) based process using silicon-on-insulator (SOI) wafers. Experimental
results, which are consistent with the model predictions, demonstrate that the compliant multistep
devices exhibit improved lateral stability and consequently larger stable displacements compared to
the conventional comb drive actuators. Stable displacements up to 80 mm at a voltage of 30 V were
registered in the experiments while three snap-through and snap-back events took place during
loading and unloading, respectively. Our computational and experimental results show that the
suggested device has clear functional advantages and can be efciently used in applications including
switches, threshold inertial sensors, variable optical attenuators as well as in micro-and nanomechanical logical elements.
& 2011 Elsevier B.V. All rights reserved.
Keywords:
Large-displacement actuator
Finite element analysis
Multistability
Multistep
Snap-through buckling
Pull-in
Comb drive
1. Introduction
Bistability and multistability, namely, the existence of two or
several different stable congurations at the same loading, is an
intrinsic feature of many mechanical structures. This behavior
typically originates from the geometric nonlinearity of the structure resulting in a non-monotonous stiffnessdisplacement characteristic. One of the most common examples is a exible arch
loaded by a transverse force [13], Fig. 1(a). This structure is
bistable in the interval of the force between the snap-back
(release) and snap-through values (see Fig. 1(b)). The analysis of
structures liable to snap-through buckling, mainly arches, frames,
cylindrical panels and spherical caps, is a well-established topic in
structural mechanics [15].
Corresponding author.
E-mail address: vadis@eng.tau.ac.il (S. Krylov).
0168-874X/$ - see front matter & 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.nel.2011.08.021
59
Fig. 1. An arch loaded by a transverse force in a pre-buckling and post-buckling conguration (a) and schematics of a corresponding limit point buckling curve (b).
Operational principle of the device-schematics of a device incorporating multiple serially connected bistable beams (c) and a generic limit point buckling curve (d).
2. Computational model
2.1. Curved beam
In order to provide an insight into the inuence of different
parameters on the stability properties of a curved beam and
choose the design parameters, the most suitable for the control of
the multistable behavior, we rst consider a model of a single
initially curved beam, Fig. 2.
We consider a exible, initially curved, prismatic micro beam
of length L, of a rectangular cross-section of area Ab d and
second moment of the area Iyy bd3 =12. The initial shape of the
beam is described by the function z0 x hc0 x (for convenience
it is considered positive in the negative direction of the z-axis,
Fig. 2) where h is the initial elevation of the central point of the
beam about its ends and c0 x is a non-dimensional function such
60
u
2 !2 2
2
1 1 L 2 1u
L
L
2x
t
1
c0 x
1
2 8 h
2
2h
h
L
3
0
00
000 T
approach is implemented (see [49] for the case of congurationdependent electrostatic force). The force F is considered as an
unknown parameter while the midpoint deection of the beam is
prescribed, i.e., wL=2 wM .
In addition, the stability of the beam was analyzed using the
nite element method by means of the commercially available
software. The planar straight beam element with an extensible
axis, three nodal degrees of freedom (two translations and one
rotation) and Hermitian polynomials as interpolation functions
was used. The element could also account for the shear deformation of the beam. Note however that the inuence of the shear
deformation on the behavior of very slender beams considered in
this work is minimal. To enforce clamped boundary conditions,
the translation in the x and z directions as well as the rotation of
the end nodes of the nite element model were precluded. The
calculations were performed using the large deection analysis.
The unstable branches of the limit point buckling curves were
described using force control combined with the arc-length
continuation method (e.g. see [50]) implemented in the commercially available software. The parameters of the arc-length procedure were chosen by trial and error in such a way that the
entire buckling curve was obtained. A total number of 200 force
increments in the nonlinear solution was used. The mesh was
rened until convergence. The results presented hereafter correspond to the convergent solution and to the beam subdivided into
80 elements. Hereafter in this section the width and the thickness
of the beam used in calculations were b30 mm and d3 mm,
respectively, Youngs modulus was E 169 GPa.
The results of calculations are shown in Fig. 3. Comparison
between the shallow beam model, Eq. (2) and the nite element
solution is shown in Fig. 3(a) for three different elevations. Excellent
agreement between the two models is observed. For h10 mm the
relative error in the snap-through value of the force was 0.38%. We
attribute the certain discrepancy mainly to the approximate character of the shallow beam model, which disregards the nonlinear
curvature of the beam. A graphical representation of the curved
beam during the loading is shown in Fig. 3(b).
In accordance with Fig. 3(b), each of the elements of the
multistable suspension should exhibit bistable behavior while
the value of the critical force corresponding to the snap-through
instability should be different for each of the beams. Generally
speaking, for the prescribed initial shape and material of the
clamped arch, the forcedisplacement characteristic of the beam
can be controlled by three parametersthe initial elevation, the
thickness and the length of the beam. It is well known that
the beam described by Eq. (2) is bistable when the ratio between
the initial elevation of the beam and its thickness is higher than a
certain value. In accordance with [2], in the case of a beam with a
rectangular cross-section, the snap-through takes place when the
Fig. 3. (a) Limit point buckling curves of an arch-shaped beam for different initial midpoint elevations h (numbers, in mm). The length of the beam is L 1000 mm. Solid line
corresponds to the shallow beam model, Eq. (2); markers represent the numerical nite element solution. (b) Snapshots of the beam at different actuation forces for
h 6 mm.
p
ratio m d=h 3 o0:42, which corresponds to h 4.13 mm in the
case of 3 mm thick beam (for the case of an initially sinusoidal
arch the value of mo0.4 was obtained in [51] using the two-term
modal expansion solution). Since the bistability criterion is
independent of the length and is very sensitive to the width and
thickness, the buckling behavior of the arch can be controlled by
choosing appropriate values of h (see Fig. 3(a)) and/or d. However,
for microstructures, both h and especially d can be very uncertain
due to low tolerances of micromachining. The structures considered in this work are fabricated from single crystal silicon using
deep reactive ion etching (DRIE). In the framework of this process,
thin (typically a few micrometers) beams are surrounded by large
open area and may suffer from signicant over etch. As a result,
the actual thickness of the beams is usually smaller than the
nominal value. Although corrections (bias) of the nominal dimensions could be made at the design stage in order to account for
possible over etch, in view of sensitivity of the buckling force to
the thickness, the uncertainty originates from the lack of repeatability and uniformity of the process is still high. Uncertainty in
the initial elevation is related to small residual stress or stress
gradients appearing in the bonded silicon-on-insulator (SOI)
wafers as well as to possible variations in temperature (see [21]
and references therein). In view of the aforementioned, in this
work we keep the thickness and the elevation of each of the
beams forming the suspension to be constant and use the
length of the beam to tailor the forcedeection characteristic
of the beam.
2.2. Multistable suspensiona chain of curved beams
The solution for a single beam was used as a building block for
the description of the multistable suspension incorporating
multiple serially connected bistable elements. Hereafter we refer
a suspension as multistable or multistep if the corresponding
limit point buckling curve contains several different (not adjacent) stable branches. A nite element solution was obtained
similarly to the case of the single beam, namely using a nonlinear
large deection analysis combined with the arc-length procedure.
To enforce boundary conditions, translation in the x-direction as
well as the rotation of the end nodes were precluded whereas the
translations in the z-direction were released. The beams were
connected by rigid links in such a way that the compliance of the
system was associated solely to the compliance of the beams.
The solution based on the shallow arch model was obtained
using the following procedure. First, the dependence between the
force and the midpoint deection was obtained separately for
each of the beams distinguished by different length. The displacement control procedure was used and the prescribed increments
of the midpoint displacements are taken to be identical for each of
61
the beams. However, since the beams are of dissimilar length, the
values of the forces corresponding to these displacements are
different for each beam. On the other hand, within the suspension
the beams are connected serially and, from the equilibrium
considerations, the force acting on each of the bistable elements
of the chain is the same. To overcome this difculty, the table
look up approach was used and the forcedisplacements characteristic for each of the beams was approximated using a
polynomial t F i Pni wi
M . Here n is the order of the polynomial
(in most cases seventh order polynomial was used) and
wi
M , i 1::N is the midpoint deection of the ith beam. Next, in
the framework of the displacement control approach, the displacement wA of the end point of the suspension (hereafter
referred to as an actuator displacement) was prescribed, and
the midpoint displacements wi
M , i 1::N of each of the beams
along with the force were found as the solutions of the system of
N 1 nonlinear algebraic equations
Pni wi
M F 0
N
X
wi
M wA 0
i1
The limit point buckling curves of the separate beams are shown
in Fig. 4(a), the limit point buckling curve of the suspension
assembled from four beams is shown in Fig. 4(b). The length of
the beams is L 700 mm, 800 mm, 900 mm and 1000 mm, the initial
elevation of all the beams is h8 mm and the width of the beams is
d 2.7 mm. One observes that, as a result of the serial connection of
the beams, the snap-through values of the chain are identical to
critical values of the individual beams whereas the corresponding
displacements are larger in the chain. Fig. 4(b) illustrates also the
robustness of the suspension to the uncertainty in the beams width.
One observes that while the critical values of the forces are strongly
affected by the beams width, the bucking behavior is qualitatively
preserved and the sequential snap-through can be achieved in the
suspensions with uncertain geometric parameters. Note that the
device can be viewed as multistable in a sense that it may have
several overlapping or non-overlapping bistability regions. A comparison between the results obtained using the shallow beam model
and the nite element analysis revealed very good agreement
between the two. In the considered example, the error in the
snap-through value of the force corresponding to the highest limit
point of the multistable chain was 0.4%.
One of the central advantages of the suspension conguration
considered in the present work is the ability to control the force
displacement curve in a very large range by choosing the appropriate values of the beams parameters. Examples of the limit
point buckling curves corresponding to different geometrical
parameters of the beams are presented in Fig. 5 where the limit
Fig. 4. (a) Limit point buckling curves of an arch-shaped beam for different length L(numbers, in mm) and the midpoint elevation h 8 mm. (b) The limit point buckling
curve of the multistable suspension assembled from four bistable beams. Nominal width of the beam is d 3 mm; dashed and dotted lines correspond to the width of the
beam of d 3.3 mm and d 2.7 mm, respectively.
62
Fig. 5. (a) Limit point buckling curves of the multistep suspension assembled from four bistable beams-nite element solution. The lengths of the beams is L 750 mm,
820 mm, 900 mm, 1000 mm and the midpoint elevation is h 5 mm. Inset illustrates the suspension stiffness increase at larger forces. (b) Limit point buckling curves of the
multistable suspension. The length of the beams is L 850 mm, 900 mm, 950 mm, 1000 mm and the midpoint elevation is h 15 mm. Dashed lines illustrate the boundaries of
the bistability region in terms of the force.
ne0 bw0 wA V 2
2
2g0 uA
ne0 bw0 wA V 2
ne0 bV 2
ne0 bV 2
2g0 uA 2
5
s
kx
w0 2 w0
g0
2kz
2g0
2
Fig. 7. The limit point buckling curve (solid line) of the multistable suspension
assembled from bistable beams of four different length L 700 mm, 800 mm,
900 mm, 1000 mm, the midpoint elevation of h 8 mm and the width of d 3 mm
and corresponding tangent stiffness kTz (dashed line).
63
Fig. 9. (a) Design #1bistable device containing a pair of curved beams connected in parallel. The length and the initial elevation of the beams are L 1160 mm, h 25 mm.
(b) Finite element analysis results of the lateral displacements (in mm) of the structure under a lateral force of 100 mN applied to the shutter.
64
Fig. 10. Limit point buckling curve of the bistable device shown in Fig. 9(a)
(design #1)nite element analysis result.
3. Designed congurations
Multistable devices of four different congurations incorporating one or two (connected in parallel) chains of three or four
serially connected curved beams and actuated electrostatically by
a comb drive transducer were designed. In addition, a simple
bistable device incorporating a pair of curved beams connected in
parallel was designed as well. In all cases, the nominal width and
the thickness of the beams were b30 mm and d3 mm, respectively. In all multistable designs, the curved beams of different
lengths were attached by their ends to a relatively stiff frame
realized as a truss-like structure. The midpoints of the shortest
beams in each of two chains of bistable elements were connected
to a central beam (hereafter referred as a shutter) with the
movable part of the electrostatic transducer attached to it. The
midpoints of the longest beams were anchored to the substrate.
For each design, the shallow beam model was used for the
preliminary evaluation of the design and operational parameters.
Next, the nite element method was used at the stage of the
detailed analysis and design to obtain the forcedeection characteristics and estimate the stability of the device. Note that in all
cases two-dimensional (planar) nite element models were used
and all the components of the structures were modeled using
Fig. 12. Limit point buckling curve of the multistable device incorporating three
serially connected bistable beams and supported by a tilted folded exure
(Fig. 11(a), design #2)nite element analysis result.
Fig. 11. (a) Design #2device incorporating three serially connected bistable beams and supported by a tilted folded exure. The lengths of the curved beams are
1500 mm, 1300 mm, 1100 mm, the initial elevation of all the beams is h 19 mm. The length and thickness of the beams of the folded suspension are 1100 mm and 3.5 mm,
respectively, the inclination is 3.51. (b) Finite element analysis results of the lateral displacements (in mm) of the structure under two lateral forces of 100 mN each applied
to the shutter.
65
Fig. 14. Limit point buckling curve of the multistable device incorporating two
chains of three serially connected double beams (Fig. 13(a), design #3)nite
element analysis result.
Fig. 13. (a) Design #3device with two chains of three double bistable beams. The lengths of the curved beams are 1100 mm, 1000 mm, 900 mm, the initial elevation of all
the beams is h 13 mm. (b) Finite element analysis results of the lateral displacements (in mm) of the structure under the lateral force of 100 mN applied to the shutter.
66
Fig. 15. (a) Design #4device with curved beams connected by frames. The lengths of the curved beams are 1250 mm, 1150 mm, 1050 mm, the initial elevation of the
beams is h 15 mm. (b) Finite element analysis results of the lateral displacements (in mm) of the structure under three lateral forces of 100 mN each applied to the
outer frame.
Fig. 16. Limit point buckling curve of the multistable device with curved beams
connected by frames (Fig. 15(a), design #4)nite element analysis result.
4. Experiment
Using the detailed nite element analysis designs #4 and #5
were found to exhibit the largest stable displacement combined
with a relatively small footprint and the possibility to integrate
relatively large number of electrodes. Fabrication and characterization efforts were focused mainly on these devices. Design #1
(bistable device) was fabricated as well due to its simplicity,
robustness and convenience of operation and for comparison with
the model results.
The rst step in the fabrication of MEMS devices is preparation
of the detailed layout of the structure. For the sake of compatibility with the mask making tools, the layout should be accomplished using dedicated layout software originally developed for
the needs of the semiconductor industry. On the other hand,
mechanical design is typically carried out using three-dimensional CAD tools allowing easy visualization and parameterization
of complex geometries. In addition, the geometries created by the
mechanical CAD tools can be conveniently imported into nite
element software and then meshed and analyzed. In this work,
the three-dimensional geometry of the devices was rst built
using mechanical CAD tools (SolidWorks [57]), see Fig. 8, and
nite element analysis of the devices was performed, as was
described previously in Section 3, using the imported geometry.
Then, the mechanical geometry was converted into the GDSII
format compatible with the standard layout tools [57] (see
Fig. 19). Note that mask generation was simplied due to the
single layer architecture of the SOI devices.
The devices were fabricated from highly doped single crystal Si
using silicon-on-insulator (SOI) wafers with (1 0 0) surface orientation as a starting material and etched using a deep reactive ion
etching (DRIE) based process. The patterning of the photoresist
spun on top of the 30 mm thick device layer of the SOI wafer was
67
Fig. 17. (a) Design #5device with two chains of four curved beams connected by frames. The lengths of the beams are 1000 mm, 900 mm, 800 mm, 700 mm, the midpoint
elevation of all the beams is h 11 mm. (b) Finite element analysis results of the lateral displacements (in mm) of the structure under a lateral force of 100 mN applied to the
shutter.
Fig. 18. Limit point buckling curve of the multistable device with two chains of
four curved beams connected by frames (Fig. 17(a), design #5)nite element
analysis result.
and of the frames simplies the release and decreases the area
prone to stiction.
The structures were mounted on a wafer prober Karl Suss
PSM6 located on an anti-vibration table (Kinetic systems, vibroplane), and were operated at room temperature and under
ambient air conditions. The actuation voltage provided by a
voltage source was applied to the unmovable electrodes of the
comb drive transducer while the movable parts of the device and
the substrate were connected to ground. The in-plane motion was
captured by a CCD camera mounted on an optical microscope
Mitutoyo FS70 (0/100, switchable microscope with long working
distance objectives). The displacements of the actuator were
measured using an analysis of captured images, each corresponding to specic values of the actuation voltage.
Preliminary experimental results demonstrating the feasibility
of the suggested approach are shown in Figs. 21 and 22.
Fig. 21(a) shows a bistable device suspended using a pair of
curved beams connected in parallel by a shutter (Design #1).
Corresponding experimental and calculated limit point buckling
curves are shown in Fig. 21(b). Note that actual geometry
parameters of the device (mainly the thickness of the beams
and the distance between the electrodes), which were measured
by high magnication optical microscope, were used in the
model. The device is bistable in the interval of the voltages
between 14 and 30 V. It should be noted that all the designs
incorporate relatively large (a few mm in size) connecting frames
with limited out of plane/tilting stiffness. For these reasons the
devices of designs #4 and #5 were found to be prone to stiction to
the substrate. In this perspective, device #4 demonstrated better
functionality when compared with design #5. Optical microscope
68
Fig. 21. (a) An optical microscope micrograph of the bistable device (Design #1). The length of the beam is 1160 mm, initial elevation of the midpoint is 25 mm. The comb
drive transducer contains 180 electrodes with a gap of 3.5 mm. Blue arrow illustrates the actuation direction. (b) Experimental limit point buckling curve (markers) and
nite element model results (dashed line) of the device. Arrows illustrate the direction of loading/unloading (for interpretation of the references to color in this gure
legend, the reader is referred to the web version of this article).
Fig. 22. (a) An optical microscope micrograph of the multistable device (design #4). Blue arrow illustrates the actuation direction (b) Experimental limit point buckling
curves (markers) and nite element model results (solid line) of the device incorporating two chains of three serially connected bistable beams connected by frames
(design #4). The lengths of the beams are 1250 mm, 1150 mm, 1050 mm and the initial elevation of the midpoint is 14 mm. The device is actuated by 330 combs (for
interpretation of the references to color in this gure legend, the reader is referred to the web version of this article).
between the experimental and the model results to the differences in the device geometry, mainly in the thickness of
the beams, which is highly uncertain due to low fabrication
tolerances of micromachining. The device exhibits stable total
displacement of 80 mm at the relatively low actuation voltage of
33 V.
5. Conclusions
Fig. 23. Snap shots of the suspension (design #4) at different actuation voltages:
(1) 16 V, (2) 22 V, (3) 25 V, (4) 32 V.
We presented a design approach of a multistable long displacement micro actuator. The device incorporates serially connected
bistable elements realized as shallow curved beams. One of the
advantages of the suggested design approach is that a desired limit
point buckling curve can be achieved by changing geometrical
parameters of each of the bistable elements. We found that in view
of high uncertainty in the device geometry due to low tolerances of
micromachining the length of the beams is the most suitable for
the tailoring of the device forcedisplacement characteristics.
Several design congurations were considered and the feasibility
of the suggested approach was demonstrated using the detailed
nite element model. We show that low lateral compliance of the
device actuated by a comb drive transducer and containing several
serially connected bistable beams may make the device to be prone
to the lateral pull-in instability. Direct numerical evaluation,
by means of nite element modeling, of the axial and lateral
Acknowledgments
This work was supported by the Israel Science Foundation
(Grant no. 1426/08) and the National Science Foundation (Grant
ECS-0335765). A preliminary version of this work initially appeared
in [45].
References
[1] S. Timoshenko, Theory of Elastic Stability, 2nd ed., McGraw-Hill, New York,
1961.
[2] C.L. Dym, Stability Theory and its Applications to Structural Mechanics,
Noordhoff Pub., Groningen, 1974.
[3] G.J. Simitses, D.H. Hodges, Fundamentals of Structural Stability, Boston
Elsevier/Butterworth-Heinemann, Amsterdam, 2006.
[4] J. Singer, J. Arbocz, T. Weller, Buckling Experiments: Experimental Methods in
Buckling of Thin-walled Structures, Wiley, Chichester, New York, 1998.
[5] A.P. Seyranian, I. Elishakoff (Eds.), Modern Problems of Structural Stability,
Springer, Vienna, New York, 2004.
[6] T. Gomm, In-plane linear displacement bistable microrelay, J. Micromech.
Microeng. 12 (2002) 257264.
[7] R.A.M. Receveur, C.R. Marxer, R. Woering, V.C.M.H. Larik, N.-F. de Rooij,
Laterally moving bistable MEMS DC switch for biomedical applications, J.
Microelectromech. Syst. 14 (2005) 10891098.
[8] J. Ko, Y. Kim, B. Kwak, Parametric study and optimization of a micro-optical
switch with a laterally driven electromagnetic microactuator, J. Micromech.
Microeng. 12 (2002) 939947.
[9] R.R.A. Syms, H. Zou, J. Stagg, Multistate latching MEMS variable optical
attenuator, IEEE Photonics Technol. Lett. 16 (2004) 191193.
[10] J. Zhao, J. Jia, H. Wang, W. Li, A novel threshold accelerometer with
postbuckling structures for airbag restraint systems, IEEE Sensors J. 7
(2007) 11021109.
[11] R. Vitushinsky, S. Schmitz, A. Ludwig, Bistable thin-lm shape memory
actuators for applications in tactile displays, J. Microelectromech. Syst. 18
(2009) 186194.
[12] B. Charlot, W. Sun, K. Yamashita, H. Fujita, H. Toshiyoshi, Bistable nanowire
for micromechanical memory, J. Micromech. Microeng. 18 (2008) 045005.
[13] B. Halg, On a micro-electro-mechanical nonvolatile memory cell, IEEE Trans.
Electron Dev. 37 (1990) 22302236.
[14] J. Rubin, S. Tiwari, An electronic nonvolatile memory device based on
electrostatic deection of a bistable mechanical beam, in: Proceedings of
the Technical Digest of MRS Fall Meeting, Boston, 27 November1 December
2006 (oral presentation) O310.
[15] Y. Shim, J. Lee, Modeling and experimental characterization of the chevrontype bi-stable microactuator, J. Micromech. Microeng. 13 (2003) 948954.
[16] j. Tsay, S. Liang-Qing, S. Cheng-Kuo, Design of a linear micro-feeding system
featuring bistable mechanisms, J. Micromech. Microeng. 15 (2004) 6370.
[17] D.A. Wang, H.T. Pham, Y.H. Hsieh, Dynamical switching of an electromagnetically driven compliant bistable mechanism, Sensors Actuators A: Phys. 149
(2009) 143151.
[18] S.M. Baker, L.L. Howell, On-chip actuation of an in-plane compliant bistable
micromechanism, J. Microelectromech. Syst. 11 (2002) 566573.
[19] N.D. Masters, L.L. Howell, A self-retracting fully compliant bistable micromechanism, J. Microelectromech. Syst. 12 (2003) 273280.
[20] J. Casals - Terre, A. Fargas - Marques, A.M. Shkel, Snap-action bistable
micromechanisms actuated by nonlinear resonance, J. Microelectromech.
Syst. 17 (2008) 10821093.
[21] Y. Gerson, S. Krylov, B.R. Ilic, Electrothermal bistability tuning in a large
displacement micro actuator, J. Micromech. Microeng. 20 (2010) 112001.
[22] J. Lang, A. Slocum, A curved-beam bistable mechanism, J. Microelectromech.
Syst. 13 (2004) 137146.
[23] J. Qiu, A bulk-micromachined bistable relay with U-shaped thermal actuators, J. Microelectromech. Syst. 14 (2005) 10991109.
[24] Y. Backlund, A lateral symmetrically bistable buckled beam, J. Micromech.
Microeng. 8 (1998) 2932.
[25] M.T.A. Saif, On a tunable bistable MEMS-theory and experiment, J. Microelectromech. Syst. 9 (2000) 157170.
69