Latching Shape Memory Alloy Microactuator

Latching Shape Memory Alloy Microactuator

ENMA490, Fall 2002S. Cabrera, N. Harrison, D. Lunking,

R. Tang, C. Ziegler, T. Valentine

Comments from GWR in yellow boxes.An excellent project and presentation overall.Presentation grade = A.- GWR

Outline

• Background• Problem• Project Development• Design• Evaluation• Applications• Summary/Future Research

ApplicationsDevice and Process Flow

Materials

Problem Statement• Assignment: Develop a design for a microdevice,

including materials choice and process sequence, that capitalizes on the properties of new materials.

• Survey: functional materials and MEMS

• Specific Device Goals:– Actuates– Uses Shape Memory Alloys – Uses power only to switch states

• Concept:– Latching shape-memory-alloy microactuator

NiTi SMA arm

Si island over valve

Project Stimulus

State of the Art: SMA microactuator – Lai et al. “The Characterization of TiNi Shape-Memory

Actuated Microvalves.” Mat. Res. Soc. Symp. Proc. 657, EE8.3.1-EE8.3.6, 2001.

– Uses SMA arms to raise and lower a Si island to seal the valve.

– Uses continuous Joule heating to keep valve open.

Joule heatingTOPVIEW:

SIDEVIEW:

• Martensite-Austenite Transformation

• Twinned domains (symmetric, inter-grown crystals)

Shape Memory Alloys

Austenite

Cooling

Polydomain Martensite

Applied Stress

Single-domain Martensite

Re-heating

Austenite

Applied Stress

Heat SMA2

valve opens

Heat SMA1

valve closes

SMA2 valve stayscools open

SMA1 magnet keepscools valve closed

INITIALDESIGN

Heat SMA1

valve close

s

Heat SMA2

valve open

s

SMA1 magnet keepscools valve closedSMA2 valve

stayscools open

FINALDESIGN

This slide would benefit from labeling the SMA1 and SMA2 for describing the actuation sequence

Cantilever Positions and Forces• Based on beam theory• Non-uniform shape change between SMA

and substrate causes cantilever bending– Thermal expansion causes bulk strain

(2-1)T– Martensite-austenite transformation

creates lattice strain =1-(aaust/amart)– Ω = [(2-1)T] or []

)232(2)()()(6

2221

21212121

22222

22111

21212121

ttttttEEbbtEbtEbttttEEbb

k

2

2kLd 3

3LEIdF

Dr. Wuttig commented that we should be more general in describing the lattice strain portion, not making the specific strains directly coupled to specific lattice constants.

Material PropertiesYoung’s Modulus (GPa)

Thermal Expansion Coefficient (*10-6/K)

Lattice Parameter (nm)

Si 190 2.33 N/A

GaAs 85.5 5.73 N/A

NiTi (martensite) 28-41 11 0.2889 (smallest axis)

NiTi (austenite) 83 6.6 0.3015

http://www.keele.ac.uk/depts/ch/resources/xtal/classes.html, http://cst-www.nrl.navy.mil/ lattice/struk/b2.html

Cantilever Positions and Forces

• Major assumptions:– Can calculate martensiteaustenite strain from

differing lattice constants– Properties change linearly with austenite-martensite

fraction during transformation

• Deflection– Large effect from SMA, negligible effect (orders of

magnitude less) from thermal expansion

Simulation

Simulation – Deflection Results

• 100μm long, 30μm wide, 2.5μm thick substrate, 0.5μm thick SMA• Tip deflection ≈ 39μm, Deflection < ≈ 21°, Tip force ≈ 0.23mN• Heat/cool cantilever 1: F(1) > F(magnet) > F(2)• Heat/cool cantilever 2: F(2) > F(magnet) > F(1)

0.1 0.5 1 5 10 15 20 25 30 35 40 45 50

1050100500100050001.E-06

1.E-05

1.E-04

1.E-03

1.E-02

Tip Deflection Scaling

SMA thickness (um)Length

(um)

Tip

def

lect

ion

(m)

L

0.3L

0.03L

Process Flow (Single Cantilever)-Silicon wafer (green) with silicon dioxide (purple) grown or deposited on front and back surfaces.

-Application of photoresist (orange), followed by exposure and development in UV (exposed areas indicated by green).

-Buffered oxide etch removes exposed oxide layer. Oxide underneath unexposed photoresist remains.

-Removal of photoresist in acetone/methanol is followed by KOH etch to remove exposed silicon until desired cantilever thickness is reached.

-Deposition of NiTi (yellow) via sputtering, followed by 500C anneal under stress to train SMA film.

-Deposition of magnetic material (blue) using a mask via sputtering on bottom of cantilever.

This schematic is more complete than that in the report. It would be helped if the wafer bonding/glueing step could be indicated somehow.

Process Flow (SMA Training)

• Small needles hold down cantilevers during post-deposition anneal

• Training process usually carried out at 500°C for 5 or more minutes

• Thin film will “remember” its trained shape when it transforms to austenite

• Degree of actuation determined by deflection of cantilever during training process

Small green circles indicate needle placement with respect to cantilever

wafer

Side view of needle apparatus

Non-Latching Power Cycle

• Energy use based on time spent in secondary state.– Energy = Power * Time

• Max energy used when 50% of time spent in secondary state.

• Above 50%, other type of actuator more efficient.

Non-latching Duty Cycle

0

20

40

60

80

100

0 10 20 30 40 50 60 70 80 90 100

Time Closed (%)

Cum

ulat

ive

Ener

gy C

onsu

med

(a

rb. u

nits

)

Normally open Normally Closed

Max energy usage

Latching Power Cycle

• Energy use based solely on number of switches.– Energy = Energy per cycle

* frequency of switching * time used

– Least energy used at low power to switch, low frequency of switching

• Low energy to switch, low frequency, latching is more energy efficient.

Latching Duty Cycles

0

5

10

15

20

25

30

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Switches (cycles * 2)

Cum

ulat

ive

Ene

rgy

cons

umed

(a

rb. u

nits

)

Low Power, Low Freq Low Power, High FreqHigh Power, Low Freq High Power, High Freq

The x-axis here is confusing, maybe misleading. If the x-axis is # switching actions, then the steps in energy expended should occur only at the switches. However, if the axis is time, then the faster switch frequency comes out right. More about this in the final report.

Power Considerations• Heat cantilevers to induce shape memory effect

– P = (m•c•T)/t = I2R• m - mass of cantilever, c - specific heat of cantilever, ΔT - difference

between Af and room temperature, t - desired response time

– Power differs slightly for martensite and austenite for constant I because of differing resistivity.

• From simulation:– Required current = 0.27 mA– Required power = 0.097 W

Applications and Requirements

• Electrical Contacts– Sensor– Circuit breaker

• Optical Switching– Telescope mirrors

• Gas/liquid Valves– Drug release system

device

outside world

TI thermal circuit breaker, http://www.ti.com/mc/docs/precprod/docs/tcb.htmSandia pop-up mirror and drive system, http://mems.sandia.gov/scripts/images.asp

Summary

• Final design: dual cantilever system with SMA and magnetic materials to provide latching action

• Power consumption lower than that of a non-latching design when switching occurs infrequently and uses little energy

• Future work:– Research magnetic material, packaging

– Specify application

– Continue analysis and optimization

– Build device

Backup

Shape Memory Effect

Free-energy versus temperature curves for the parent (Gp) and martensite (Gm) structures in a shape memory alloy. From Otsuka (1998), p.23, fig. 1.17.

Martensite-austenite phase transformation in shape memory alloys. From http://www.tiniaerospace.com/sma.html.

Material Choice: NiTi SMA

• Near-equiatomic NiTi most widely used SMA today

Property Value

Transformation temperature -200 to 110 C

Latent heat of transformation 5.78 cal/g

Melting point 1300 C

Specific heat 0.20 cal/g

Young’s modulus 83 GPa austenite; 28 to 41 GPa martensite

Yield strength 195 to 690 MPa austenite; 70 to 140 MPa martensite

Ultimate tensile strength 895 MPa annealed; 1900 MPa work-hardened

% Elongation at failure 25 to 50% annealed; 5 to 10% work-hardened

From http://www.sma-inc.com/NiTiProperties.html

Nickel-Titanium

B2 (cesium chloride) crystal structure. From http://cst-www.nrl.navy.mil/ lattice/struk/b2.html

B19’ crystal structure. From Tang et al., p.3460, fig.5.

Parent β (austenite) phase with B2 structure

Martensite phase with monoclinic B19’ structure