electrostatic zipping actuators and their application to...
TRANSCRIPT
1
1
Electrostatic Zipping Actuators and Their Application to MEMS
Jian Li
B.S. Mechanical Engineering, Tsinghua University, 1997M.S. Mechanical Engineering, Tsinghua University, 2000
Ph.D CandidateDept. of Mechanical Engineering
Massachusetts Institute of Technology
01/15/2004
2
Introduction
ProblemNeed powerful electrostatic actuators for MEMS
power relay
Huge size using conventional designs
SolutionsZipping actuator with compliant
starting zone
Results
9 mm
~ 3 mm
2
3
Contributions
Design: Starting zone to reduce the pull-in voltageNumerical and analytical modeling: easy and accurateOptimization of the systemSpecified design for implementation in a relayFabrication and measurement of the actuator-relayWet-anisotropically-etched micro relay contact
4
Background: Electrostatic Actuators
Comb DriveForce independent
of stroke: large stroke
Very small force, large deflection
Parallel PlateConfliction
between stroke and force
“Zipper”Large force and
large deflection
High voltage for pull_in
3
5
F-D Curve of Three Electrostatic Actuators
Force-Displacement Character of the three actuatorsWith the same volume and same voltage, the “zipper” actuator achieves a
much larger force than comb drive an capacitor plate actuators
“Zipper” actuator is the only hope to secure milli Newtons of force with a reasonable volume
0
2
4
6
8
10
12
0 20 40 60 80
Displacement (µm)
Forc
e (m
N)
Parallel PlateComb DriveZipping Actutor
6
Zipping Actuators: A Survey
802 ~ 109 * 1RelayLateralThis thesis
100N/A(very small)1 * 0.2Fluid
controlLateralSherman
20N/A(very small)0.5 * 0.2Optical
shutterLateralPerregaux
300.020.8 * 0.5N/ALateralLegtenberg
60.10.8 * 0.8MirrorVerticalDivoux
220N/A (very small)5 * 5ValveVerticalShikida
Stroke (µm)
Force (mN)
Size(mm2)
Applica-tion
Actuation directionAuthor
4
7
Example Load: Double BeamA pre-curved bistable MEMS mechanismPromising as a micro relay structure
Simple structureNo residual stressHigh contact force: good for on-state performance
Need an electrostatic actuator to make the MEMS relay
F
Double beam Double beam working as a Relay
8
Why Is It Hard ? (1)
Stoke: >= 80 µmAsymmetric and high actuation/contact force: ~ 7 mN for the first 10 µm to get a 3mN of contact forceActuation voltage: < 200 V (as small as possible)Scale: 7 × 7 × 1 mm3
Fabrication: Deep Reactive Ion Etching
-4
-2
0
2
4
6
8
0 20 40 60 80 100 120Displacement δ (µm)
Act
uatio
n fo
rce
Fri
ght(
mN
)
~ 3 mN
~ 7.5 mN
5
9
Why Is It Hard ? (2)
Initial Design:DRIE etch throughAttach the cantilever
beam to the double beamApply voltage between
electrode and the cantilever beam.
Problem:Minimum gap due to
DRIE aspect ratio of ~5%Very High pull-in voltage
because of the gap, about 250 V for a 15 µm thick and 4.5 mm long zipping beam.
U
Minimum Gap
Cantilever Beam BeingElectrified
Cantilever Beam Electrode
Double Beam
Double beam actuated by "Zipper"
10
Contributions
Design: Starting zone to reduce the pull-in voltageNumerical and analytical modeling: easy and accurateOptimization of the systemSpecified design for implementation in a relayFabrication and measurement of the actuator-relayWet-anisotropically-etched micro relay contact
6
11
Design: Compliant Starting Zone
To reduce the pull-in voltage:
A compliant starting zoneReciprocity: remove
material not needed and not useful The starting zone
bends up when voltage is applied between electrode and the beamThe cantilever beam
pulls-in and keep zipping at low voltage
U
gmin
ActuatorCantilever Beam
Fixed electrode
Bistable RelayBeam
StartingCantilever a)
b)
12
Contributions
Design: Starting zone to reduce the pull-in voltage
Numerical and analytical modeling: easy and accurateOptimization of the systemSpecified design for implementation in a relayFabrication and measurement of the actuator-relayWet-anisotropically-etched micro relay contact
7
13
Modeling: Overview
( )2'''' 0
2
0
( ) ( )
2 ( ) ( )r
U bE I x y xhy x c x
ε
ε
−=
⎡ ⎤− +⎢ ⎥
⎣ ⎦
Beam bending equation in electric field
Electrode Insulator
h0
h
U
x
y
z
Actuator Cantilever
Starting Cantilever(Component III)
b
L1 L2
L
Component I Component II
WaferNormal
c(x)
14
Numerical Modeling: Before “Pull-in”
Deflection profiles of the cantilever beam and the starting beam at different voltages when the tip is constrained at zero displacement
Numerical MethodModeled as a device
with Multiple-beams.Each beam is bent by
the electrostatic force.The ODEs are solved
simultaneously for each beamDeflection profiles for
each beam are given out at different voltages.Actuation force can be
calculated.Pull-in voltage can be
found
8
15
Numerical Modeling: After “Pull-in”
sL
δ
c(x)
p(x)
Fright
Fleft
sL
δ
x
y
δmax
d
h0
Numerical solutionsEliminate varying B.C by adding parameter
Cantilever Beam Deflection profiles when the tip is constrained at zero displacement(different voltages)
Ls /1−=λ
Model after pull-in
[ ]2
02
0
''''( ) ( )2 ( ) ( ) / r
bUEIy x p xy x c x h
εε
= = −− +
Beam equation: s x L≤ ≤
16
Analytical Modeling: After Pull-in3/ 421/ 4
0
0
~ rright
U hE bFhd
ε ε⎛ ⎞⎜ ⎟⎝ ⎠
0
1
2
3
4
5
6
7
100 120 140 160 180 200 220 240 260
Voltage (V)
Forc
e (m
N)
Device I, Numerical results
Device I, Analytical Results
Device II, Numerical Results
Device II, Analytical results
Approximation equation:
Comparisons between analytical approximations and numerical results:
9
17
Forces: Actuator vs. Switch Beam
Force-Displacement curves of the actuator and the double beam
> 160V is needed to actuate the double beam with ~ 3 mN of contact forceCan we change the
force-displacement curve of the double beam without reducing the contact force?
Force-Displacement curves of the actuator and the double beam
-5
0
5
10
15
20
0 20 40 60 80 100 120
Displacement δ (µm)
Act
uatio
n fo
rce
Fri
ght(
mN
)
Relay with Uniform Thickness, FEA Actuator, U = 100 V
Actuator, U = 120 V Actuator, U = 140 V
Actuator, U = 160 V
18
Contributions
Design: Starting zone to reduce the pull-in voltageNumerical modeling: easier and more accurate
Optimization of the systemSpecified design for implementation in a relayFabrication and measurement of the actuator-relayWet-anisotropically-etched micro relay contact
10
19
Force ratio R = C(τ0,τ2)1
Optimization of The Double Beam
0 2 0 2 0 2( , ) ( ) ( , , , )C I w wδ τ τ δ τ τ= Λ∫Start with constant I(x)
Numerically calculate eigen-functions and eigenvalues [wi , τi]
choose δI = εΛ(τ0, τ2, w0, w2),satisfying thickness constraints.
Small enough R?No
YesEnd
I(x) = I(x) +δI
20
Optimization of The Double Beam (II)
Force-Displacement curves of the actuator and the double beam
-5
0
5
10
15
20
0 20 40 60 80 100 120Displacement δ (µm)
Act
uatio
n fo
rce
Fri
ght(
mN
)
Relay with Uniform Thickness, FEA Optimized Relay, FEAActuator, U = 100 V Actuator, U = 120 VActuator, U = 140 V Actuator, U = 160 V
7.5 mN4.5 mN
Actuation force/Contact force ratio starts from~ 2.5:1.Beam was modulated, force ratio was optimized to ~1.5 :1.~ 120 V is needed to actuate the relay.
11
21
Contributions
Design: Starting zone to reduce the pull-in voltageNumerical modeling: easier and more accurateOptimization of the system
Specified design for implementation in a relayFabrication and measurement of the actuator-relayWet-anisotropically-etched micro relay contact
22
The Relay
UU
UU
U
a)
b)
c)
d)9 mm
~1.5 mm
12
23
Contributions
Design: Starting zone to reduce the pull-in voltageNumerical modeling: easier and more accurateOptimization of the systemSpecified design for implementation in a relay
Fabrication and measurement of the actuator-relayWet-anisotropically-etched micro relay contact
24
Fabrication: Process Flow (1)
Starting substrate: 4''DSP <100> Si Wafer
Spin and pattern 10 µmof photoresist
DRIE etch through
Grow 0.2 µm ofthermal oxide
Strip photoresist inPiranha, RCA clean
v
Silicon Silicon Oxide Photoresist
13
25
Fabrication: Process Flow (2)
Silicon Silicon Oxide Photoresist
Bond shadow waferand device wafer, etchoxide selectively
Spin and pattern 10 µmof photoresist
Starting substrate: 4''DSP <100> Si Wafer
DRIE etch through
Strip photoresist inPiranha
Separate wafers, cleandevice wafer in Piranha
26
Fabrication: Process Flow (3)
Silicon Silicon Oxide Photoresist
Starting substrate: 4''Pyrex 7740 wafer, 500mm thick
Deposite 0.005 mm Ti,0.03 mm Cr
Pattern Cr in CR-7
Etch Pyrex 50 mm in49% HF
Bond device wafer toPyrex wafer at 800 Vand 350 degree C
Titanium Chromium
Sputter Au on relaycontacts
Gold
14
27
Fabrication: Transparency Mask
Device made using Transparency Mask
Device made using transparency mask and 10:1 Stepper
Transparency mask for device wafer:
Much faster and much cheaperLow resolution: 3-4 µm
Electrostatic force is very sensitive to the recess at electrode
Print the mask 10 times larger and use 10:1 stepper to shrink it down
Electrode
Cantilever Beam
Cantilever Beam
Electrode
28
Fabrication Results: Overview
13 7
2 1 11 6
54
12
500 µm
8
10
3 12
9
15
29
Measurement: Pull_in of the Starting Zone
The starting beam bends up and closes the gap between zipper beam and fixed electrode
Pull-in at 75V
140 V. Zipper beam collapse to the fixed electrode, relay is closed
30
Force Measurement
-4
-2
0
2
4
6
8
10
12
0 20 40 60 80 100 120Displacement δ (µm)
Forc
e (m
N)
Relay FEA Relay ExperimentalActuator Numerical Simulation at 140V Actuator Experimental at 140VRelay FEA (Adjusted model)
Flextester
16
31
A Bipolar Drive to Avoid Stiction
74LS74
D
C
Q
Q
74LS74
D
C
Q
Q
+5V
+15V
+15V
+100V
+100V+15V
+100V TopElectrode
BottomElectrode
MiddleElectrode
InputSignal
1k
5k
5k
1k
4k
5k
5k
5k
5k1k
STP6NB50
STP6NB50
STP6NB50
2N3904
2N3904
2N3904
2N3904
InputSignal
TopElectrode
BottomElectrode
MiddleElectrode
t
t
t
t
Change polarity every two actuation cycles
32
Relay Actuated by Zipping Actuator
> 40 millions of life cycles were achieved
Maximum toggling speed: 160 Hz
Shortest excitation pulse: 400 µs
Click for a movie
17
33
Switching Time
Contact bounce time ~ 0.5 ms
Switching time ~ 3ms
34
Contributions
Design: Starting zone to reduce the pull-in voltageNumerical modeling: easier and more accurateOptimization of the systemSpecified design for implementation in a relayFabrication and measurement of the actuator-relay
Wet-anisotropically-etched micro relay contact
18
35
DRIE-etched Relay Contacts
Relay contactsVertical sidewallsSputter Au: low efficiency.Sidewall/surface thickness ratio = 1/9Contact resistance:~1 Ω
36
KOH-etched Relay Contacts: Design
19
37
KOH-etched Relay Contacts: Experiments
38
KOH-etched Relay Contacts: Tests
Sputter Au: high efficiency.Sidewall/surface thickness = 1/1.4 Very low contact resistance: ~ 50 mΩ
20
39
Proposed Full Relay Process Flow
40
Conclusions and Future Work
Conclusions:Design, modeling, optimization, fabrication and measurement of an actuator-relay systemDevelop wet-anisotropically-etched relay contacts
Future work:Combine the actuator and the KOH-etched contactsVertically moving zipping actuators
21
41
Acknowledgements
Advisors: Prof. Alexander Slocum, Prof. Jeffrey Lang, Prof. Michael Brenner (Harvard University)Discussion (PERG): Dr. Jin Qiu, Dr. Joachim SihlerFabrication (at MTL): Kurt Broderick, Dennis Ward, Paul Tierney, Linvhu Hol, Dr. Hongwei Sun …Testing (at ABB Research): Dr. Ralf Struempler, Dr. SamiKotilainen, Dr. Jan-henning Fabian, Eric Wapelhorst, Jeff Renaud
42
22
43
44
Actuator Beam Profile
DRIE Etchdirection
15.24 µm
13.00 µm
15.78 µm
285.8 µm
12.78 µm
23
45
Optimization: Algorithm (1)
0''')'''( =+ wTwIE
( '') '' '' '' ( '') ''i i i i i iE I w T w T w E I wδ δ δ δ+ = − −
( )'' ( '') '' 0i i i iT w E I w w dxδ δ+ =∫
Beam equation:
Fredholm alternative:
( '') ''
''i i
ii i
w E I w dxT
w w dx
δδ = − ∫
∫Cost function variation:
( )( )
2
2
''
'i
ii
I w dxT E
w dx
δδ = ∫
∫
( )( )
2
21
''( ) ( , )
'i
i i ii i i
wdCC T I I T wdT w
δ δ δ∞
=
⎡ ⎤⎢ ⎥= = Λ⎢ ⎥⎣ ⎦∑∫ ∫∫
46
Optimization: Algorithm (2)Beam equation:
Transition state
( )( )
2
2
''
'i
ii
I w dxT E
w dx
δδ = ∫
∫0 0.5 1 1.5 2
-400
-200
0
200
400
600
800
force
displacement
20( ( ) '') '' '' ( / 2)E I w w w f x Lτ δ− + = −
0 ( )t nw w x A w= Γ +
120 ))/(1( −−=Γ ττ
( )*
*
12 2 (1 )
R− Γ∆
= =− ∆ − − Γ
* 1∆ = − Γ
*∆
22 0
22 0
( )2 ( )
τ ττ τ
=−
24
47
Analytical Analysis (1)Beam equation:
[ ]2''''( ) ( )( )
EIy x p xy x β
Γ= = −
+
[ ]1 ( ) ( )
1 ( ) ( )
L
left s
L
right s
F p x L s x dxL s
F p x x s dxL s
⎧ = ⋅ − ⋅ − −⎪⎪ −⇒ ⎨⎪ = ⋅ ⋅ −⎪ −⎩
∫
∫
4 32( ) ( ) ( )
24y x x s A x s
EIβΓ
= − − + −
Near x = s
'''( ) 6leftF EIy s EIA= =
⇒
⇒
sL
δ
c(x)
p(x)
Fright
Fleft
sL
δ
x
y
δmax
d
h0
48
Analytical Analysis (2)
* * 3 * 42( ) ( ) ( )
246left
y x A x s x sEI
F EIA
ββ
Γ ⎫= = ⋅ − − − ⎪⎬⎪= ⎭
[ ]* *
2 2
* 2*
2
1 ( )~ ( ) ~
1 ( )~ ( ) ( )
x
left s
x
right s
x sF L s x dxL s
x sF p x x s dxL s L s
β β
β
⎧ Γ Γ −⋅ ⋅ − −⎪ −⎪
⎨Γ −⎪ ⋅ ⋅ − =⎪ − −⎩
∫
∫
* 3 * 42
*
2
~ ( ) ( )6 24
( )~
left
left
Fx s x s
EI EIx sF
ββ
β
⎫Γ⋅ − − − ⎪
⎪⎬
Γ − ⎪⎪⎭
Assume most of the force comes from s<x<x*
⇒
⇒1~right
EIFL s β
Γ−
3* 1/ 4
* 2
2
( ) ~ ( )
( )~right
EIx s
x sFL s
β
β
⎫− ⎪Γ ⎪
⎬Γ − ⎪
⎪− ⎭
⇒
⇒
25
49
1/ 4
~
1~right
EIL s d
EIFL s
β
β
⎫⎛ ⎞− ⎪⎜ ⎟Γ⎝ ⎠ ⎪⎬
Γ ⎪⎪− ⎭
3/ 41/ 41~ ( )rightF EI
d β⎛ ⎞Γ⎜ ⎟⎝ ⎠
3/ 421/ 40
0
~ rright
U hE bFhd
ε ε⎛ ⎞⎜ ⎟⎝ ⎠
2( ) ( )y x B x s= − ( )2*2
3* 1/ 4
~2
( ) ~ ( )
B x sEI
EIx s
ββ
Γ ⎫− ⎪⎪⎬⎪− ⎪Γ ⎭
Far from x = s
⇒
⇒⇒ ⇒
Analytical Analysis (3)
⇒
2
~
( )
BEI
d B L s
β⎫Γ⎪⎬⎪= − ⎭
⇒
50
Analytical Analysis (4)
0
1
2
3
4
5
6
7
100 120 140 160 180 200 220 240 260
Voltage (V)
Forc
e (m
N)
Device I, Numerical results
Device I, Analytical Results
Device II, Numerical Results
Device II, Analytical results