MEMS Cell Adhesion Device
Download
Report
Transcript MEMS Cell Adhesion Device
MEMS Cell Adhesion Device
Andrea Ho
Mark Locascio
Owen Loh
Lapo Mori
December 1, 2006
Summary of Fabrication
Based on passive
PDMS pillar arrays
Add 3-axis force
sensitivity on each
pillar
Thin membrane over
pillars
Alignment is critical
Pillars, piezoelectric
elements, electrodes
Use single set of
alignment marks for all
layers
PDMS
Membrane
[Roure, et al. PNAS 2005]
PLA
PDMS
SiO2
PVDF
(Piezoelectric)
Top Electrode
Bottom
Electrodes
Via
Parylene
Ni Traces
(Layer 2)
Si
Ni Traces
(Layer 1)
Fabrication - Alignment Features
1. Si wafer
2. Deposit silicon nitride by LPCVD
3. Spincoat with resist
4. Pattern alignment features in resist
5. Etch silicon nitride using RIE
6. Strip resist in oxygen plasma
Fabrication - Pillar Mold
1. Spincoat with resist
2. Pattern resist by e-beam lithography
3. Etch Si using DRIE
4. Strip resist
5. (Silanize wafer to improve PLA release)
6. Pour PLA
7. Deposit common top electrode by
e-beam evaporation
Fabrication - Piezoelectric Elements
1. Spincoat with PVDF (piezoelectric)
2. Spincoat with resist
3. Pattern using e-beam lithography
4. Etch PVDF using RIE
5. Strip resist
Fabrication - Electrodes
1. Spincoat with PDMS
2. Pattern bottom electrodes and first set of
traces by e-beam lithography and liftoff
3. Deposit SiO2 dielectric layer by PECVD
4. Spincoat with e-beam resist and pattern
by e-beam lithography
5. Etch through SiO2 by RIE
6. Strip resist in acetone
7. Sputter with Ni
8. Spincoat with e-beam resist and pattern
by e-beam lithography
PVDF
Electrodes
9. Etch exposed Ni
10. Strip resist
11. Deposit parylene by CVD
Fabrication - Wafer Bonding
1. Flip over and bond parylene layer to
Si wafer with low heat and pressure
2. Peel off top Si wafer and SU-8 mold
PDMS Membrane
1. Begin with Si wafer
2. Spincoat with photoresist
3. Spincoat with diluted PDMS
4. (Treat in oxygen plasma)
Mold Release
1. Flip over PDMS-coated wafer and
bond to pillars
2. Peel away support wafer
3. (Treat in oxygen plasma)
Parametric Study
Dependence of output
voltage on
pillar geometry
V
64 d33
3 e
material
Diameter
Height
Electrode geometry
material properties
3 D
2
h
6 D s 4 s 2 sin / 2 h L
2
Fx
4
D s D
geometry of electrode and pillar
Parametric Study
Parametric Study
Parametric Study
Response
D 2 μm
s 0.5 m
h 1 μm
L 20 μm
80
Inverse analysis
F V1 V3
z 2 z
V1 V3 V12 2 V1 V2 2 V2 2 2 V2 V3 V32
F
2 V1 V3
V1 V3
arccos
2 V12 2 V1 V2 2 V2 2 2 V2 V3 V32
2 d33 h
z
e s D s
2
2
32 d33 3 D 6 D s 4 s h 2 L h sin / 2
3 e
D s D4
FEM analysis
Model geometry
Mesh
FEM results
It is reasonable to assume constant sz over the piezoelectric material.
Additional results
Resonance frequency
L
m* d ( y )
2
D
0
4
2
dy
33
L D 2
560
d 2d ( y)
3 D 2
k * EI
dy
E
2
dy
128
L
0
L
1
f
2
k*
770
m * 88
ED 2
17.41 Hz
L2
Tip displacement
Fx L3
Fx L3 26
x
0.077 μm
3 E I 3 E D4
Frequency Response
Lumped element model
Long, thin Ni wires in and out of pillar
Electrode of pillar modeled as parallel
resistor & capacitor
Rwire
CPVDF
Rwire
RPVDF
Frequency Response
Circuit element values calculated from
material properties
Rwire
Ni
A
63.1 10
9
m 20 106 m
2.41
15
2
523.6 10 m
C PVDF
7.4 * 0 * 0.5236μm 2
34.3 1018 F
1 μm
R PVDF
(1018 μm)(1 μm)
18
1
.
91
10
2
0.5236μm
Frequency Response
Combine impedances
Take output across ZP
Zw
Zw
ZPR
Zw
ZPC
ZP
Zw
ZEQ
Frequency Response
H ( s)
Bode plot shows
ωC >> any
frequency we will
be sensing
ZP
RP
Z EQ RP 2RW 2RW C P RP s
Thermal Noise
The electrodes and PVDF form an RC
system
As in Senturia, this arrangement will
create thermal noise in the system
Need to ensure RMS thermal noise <<
output voltages
VRMS 4k BTRf
Thermal Noise
Consider noisy resistor to be a
noiseless resistor an a voltage source
H (s)
1
1 RCs
RPVDF
CPVDF
RPVDF
VNOISE
CPVDF
VOUT
Thermal Noise
Calculate noise bandwidth
f
0
1
1
df
0.0038Hz
2
4 RC
1 (2 f RC)
Calculate thermal noise
VRMS 4kBTRf 0.1364mV
This is acceptable, since our outputs will
be hundreds of mV
Actuation
Piezoelectrics allow for both actuation
and sensing
Electromechanical coupling factor k
k2
storedmechanicalenergy
storedelectricalenergy
or k 2
input electricalenergy
input mechanicalenergy
kPVDF ≈ 0.1 to 0.3
Easy to run in reverse to stimulate cell
Actuation
Applied voltages will have to be roughly
10x the voltage out for a corresponding
deflection
storedmechanicalenergy d 2
k
2
input electricalenergy
V
2
This puts it at a reasonable value for
actuation voltage
Actuation would have to be calibrated
experimentally
Sensitivity Analysis
Change in voltage output for a given change in
force: Slope of linear parametric plots
1200
1800
y = 469.71x
y = 24.85x
1600
Output voltage range (mV)
Output voltage range (mV)
1000
800
600
400
200
1400
1200
1000
800
600
400
200
0
0
0
10
20
30
Height of the pillar, L (um)
40
50
0
0.5
1
1.5
2
2.5
3
Distance between electrodes, h (um)
3.5
4
Sensitivity Analysis
Output voltage range (mV)
700
600
500
400
300
200
100
0
0
0.2
0.4
0.6
0.8
Radial width of the electrode, s (um)
1
1.2
Sensitivity Analysis
600
y = 5.5061x
Voltage Output (mV)
500
400
300
200
100
0
0
20
40
60
80
100
Applied Force (nN)
F [nN ]
Resolution where system
noise is the limiting factor
F
V [mV ]
5.5061
0.1364 mV
0.0248 nN
5.5061
120
Sensitivity Analysis
Resolution affected by
fabrication processes
Effect of variation
in pillar diameter
on output voltage
Diameter varies by ~10nm → Output voltage varies ~mV
ΔV = (30mV/μm)(0.06 μm) = 1.8 mV
Sensitivity Analysis
Effect of PVDF layer
uniformity (4% )
At F = 100nN, ΔV[mV]
= 450Δx[μm]
This results in an
output voltage range of
36 mV
ΔF = 36 mV/5.5061 =
6.54 nN
600
y = 5.5061x
500
Voltage Output (mV)
400
300
200
100
0
0
20
40
60
80
Applied Force (nN)
100
120
Sensitivity Analysis
Effect of variation in
pillar height
DRIE allows pillar
height to vary ~μm
At F = 100nN, output
voltage can range over
20 mV
Worst case scenario:
At F=100nN, output voltage varies over a total
range of 20 + 36 + 1.8 mV = 57.8 mV
ΔF = 10.50 nN (~10% error)
Questions