Piezoresistive Strain Gauges for use in Wireless

Component Monitoring Systems

Stephen P. Olson1, Dr. James Castracane1, Ryk E. Spoor2

1College
ll off Nanoscale
l Science
i andd Engineering,
i i University
i i at Albany,
lb State University
i i
of New York, 255 Fuller Road, Albany, NY 12203, Email: solson@uamail.albany.edu,
jcastracane@uamail.albany.edu
2International Electronic Machines Corporation (IEM). 60 Fourth Avenue, Albany, NY

12202,, Email: rspoor@iem.net

p @

cnse.albany.edu
 Outline

•Strain monitoring systems for rotating helicopter parts

•Strain gauge requirements – Low power consumption

•Piezoresistive strain gauges

•Strain
S i gauge design
d i andd fabrication
f b i i flowfl

•Test setup

•Strain response simulation

•Strain
Strain gauge measurements - Response and noise

•Conclusions and acknowledgements

cnse.albany.edu
 Component monitoring

Overall Goal:
Reduce unnecessary prescribed maintenance schedules by providing real-time
data on component condition.
• Pitch link controls the angle of the rotor blade as it spins
• One of many parts critical for controlled flight
• Currently these parts are replaced based on usage hours
• Expensive
• Condition is not known while in service

Wireless strain monitoring tags

• Power
P system
t
• Data gathering and signal conditioning
• RF transmitter
• Strain gauge
g g
• Wireless system and properties of the pitch link place
unique requirements on the strain gauge
• Thee focus
ocus oof tthiss pproject
oject iss to design,
des g , fabricate
ab cate and
a d
package strain gauge devices for this system

cnse.albany.edu
 Strain gauge requirements

Company
C Type
T G R ((ohm)
h ) Size
Si
Omega Metal foil 2 120 5x6mm
Soltec Corporation Semiconductor -100 120 11mm dia
Micron Semiconductor 100..155 540..1050 10.6x3

Omega1 Soltec2 Micron3

Requirements for the sensor:
• Multi-axis
M lti i
• Power consumption <100µW, preferably 10µW
• 5 µε resolution, 10 kHz measurement bandwidth
• Small size,, target
g size for entire tag
g is 12.7mm square
q ((sensor,, ppower system,
y , readout
electronics and RF transmitter)
• Low power requires a large gauge factor (sensitivity), large resistance (~100kΩ)
• Operational temperature range -20 to 70 ºC
1. www.omega.com/pptst/Rosettes_Stackedgrid_Strain_SG.html
www.omega.com/pptst/Rosettes Stackedgrid Strain SG.html
2. www.solteccorp.com/products.aspx?catid=48
3. www.microninstruments.com/products/gages/ssgh-halfbackgage.htm

cnse.albany.edu
 Resistive strain gauges

ρl
R= • Conductors change resistance when strained
A
l
ΔR
R = G( ΔL
L ) • Gauge factor is the sensitivity of the gauge
Δl • Resistance of a wire is proportional to its length
A
• Poisson ratio reduces the cross
cross-sectional
sectional area,
area
increasing the resistance
• Piezoresistive materials change resistivity when a
stress is applied, changing the resistance of the
gauge
<100>

G = 1 + 2ν + π l E 110 <110>

~2 >100
Longitudinal
For a strain gauge with a rectangular Transverse
cross section
G Gauge Factor
E Young’s modulus
ν Poisson ratio 10-11 Pa -1
-110
πl Piezoresistive coefficient
P-type Silicon piezoresistance coefficients
Marc J. Madou, Fundamentals of Microfabrication: The
Science of Miniaturization, CRC; 2 edition (March 13, 2002)
cnse.albany.edu
 Noise sources

•Noise
N i will
ill li
limit
it sensor resolution
l ti
•Johnson or thermal noise
• Frequency independent
• Depends on resistance
resistance, temperature and measurement
bandwidth
• kB Boltzmann's constant
• R=100kΩ, T=300K, measurement bandwidth=10kHz
VThermal = k B R T ( f max − f min ) = 2 × 10 −6 VRMS
•1/f noise
• Frequency
q y dependent
p
• Hooge noise model depends on the number of carriers, bias
voltage, and measurement frequency
• N= 9x108, α=1x10-5, Vb =1V, fmax/fmin=10,000

αVb2 ⎛ f max ⎞
V1/ f = ln⎜⎜ ⎟⎟ = 3.2 ×10 −7 V RMS
N ⎝ f min ⎠
Harkey, J.A. Kenny, T.W, Journal of microelectromechanical systems, 9 no. 2 2000

cnse.albany.edu
 Signal to noise

Vout = Vnoise + Vsignal G = 100

Vnoise = 4kB T Δf R
I
R V ↑ Vsignal = I (R + ΔR) T = 300K
ΔR ε = 5 ×10−6
= Gε

NR
Δf = 10kHz
kH

SN
R
Vnoise
I = PR
G=2
Signall
Si Gε
= P
Noise 4kBTΔF
Target Power Power (µW)

• Plot shows the signal to Johnson noise ratio for a 5 microstrain signal
as a function of power
• Large
g ggauge
g factor will be required
q for low ppower operation
p
• Resistance does not affect signal to Johnson noise ratio

cnse.albany.edu
 Silicon rosette sensor

Sensor response to arbitrary stress state 1

ΔRx
= π lσ x + π tσ y + π 12σ z π l = 12 (π 11 + π 12 + π 44 )
Rx π t = 12 (π 11 + π 12 − π 44 )
Ry π s = 12 (π 11 + π 12 )
ΔR y
R45 = π tσ x + π lσ y + π 12σ z π d = (π 11 − π 12 )
Ry
ΔR45
Rx = π s (σ x + σ y ) + π dτ xy + π 12σ z
<110> R45
2
Dopant πl πt πs πd
P 71.8
71 8 -66
66.3
3 2.75
2 75 77
7.7
(001)
N -31.2 -17.6 -24.4 -155.6
All units 10-11 Pa -1
• Sensor will respond to X, Y, Z and in plane shear stress
• Three measurements can resolve X, Y and in plane shear stress if z stress is zero
• Best sensitivity will be achieved using P type x and y sensors and N type 45 degree
sensors
1. D. A. Bittle, J. C. Suhling, R. E. Beaty, R. C. Jaeger, and R.W. Johnson, “Piezoresistive
Piezoresistive stress sensors for structural
analysis of electronic packages”, Journal of electronic packaging, Vol 113(3), 1991, pp. 203-215
2. Simon M. Sze, Semiconductor Sensors, Wiley-Interscience (October 1994)

cnse.albany.edu
 Strain gauge designs

1.5mm
Y Gauge

X Gauge Temperature sensor

Strain gauge variations:

Strain gauge design:
• 2 and 4 contact gauges
• 1.5 mm die
• Single and multi-axis
multi axis configuration
• Sensors formed by boron ion implant
• Multi axis gauges in series
• 0-45-90 rosette configuration
• 4 wire readout
• Integrated Pt temperature sensor

cnse.albany.edu
 Sensing element design

L 500
L=500µm
W=10 µm

on
on concentratio
2 5E 17
2.5E+17
2.0E+17
1.5E+17
• Graph shows piezoresistive factor(P) as a function of 1.0E+17

boron concentration and temperature 5.0E+16

Boro
• Piezoresistive coefficient (πl )= (P) (72x10-11 m2/N) 0 0E 00
0.0E+00
0 0.2 0.4 0.6 0.8 1 1.2

Depth (micron)

• To achieve maximum sensitivity, implant dose • Graph shows boron distribution after
should be below 1018 B/cm3 implant and anneal
2
• Implant parameters, 1.29e13 B/cm chosen based on • Suprem3 simulation software calculates a
Suprem3 calculations to achieve a sheet resistivity of sheet resistivity of 2000 ohms / square
200 Ω/square • 50 square resistor is expected to be 100kΩ

Simon M. Sze, Semiconductor Sensors, Wiley-Interscience (October 1994)

cnse.albany.edu
 Gauge design and fabrication flow

B Doped Si
Oxide
C Si
Al
D Pt

CAD layout showing location E

of the cross sections
Process flow cross section diagrams
Process Flow Steps:
Start with N type (100) wafer (1-10 Ω-cm)
A. Sensors are formed by boron ion implantation
B Contact
B. C t t areas are maded P+ using
i a secondd implant
i l t
C. Wafers are annealed in oxygen to activate the dopant and form 500Å of oxide
D. Oxide is patterned to form contact holes. 1000Å Al is deposited and patterned
E. 1000Å Pt is deposited
p and ppatterned to form the temperature
p sensor
F. Wafers are annealed at 450°C to form the Al –Si contacts (not shown)

cnse.albany.edu
 Fabricated strain gauges
Y Gaauge

T
Temperature sensor O i l image
Optical i off finished
fi i h d 100mm
100 wafer
f
X Gauge

1.5mm die
Optical image showing 3 gauge die with temperature sensor

Strain gauge devices have been fabricated at the

College of Nanoscale Science and Engineering CAD layout of strain gauge die

cnse.albany.edu
 Test setup

Force
5mm

3mm

X-Z Cross section Strain gauge die

Y-Z Cross section

• Sets of four of strain gauges are diced into 3mm test dies
• Test dies are thinned form ~550µm to 100µm
• Bonded to a 15-5PH stainless steel test beam
• Test beam is square in cross section, 5mm on a side
• Wirebond connections are made to a printed circuit board 100 µ
µm

cnse.albany.edu
 Stress simulation

Z Y
Implanted resistor
location

•Strain gauge Die

2 •Adhesive
X s tre s s (M P a )

1.5 •Test beam

1
0.5
0
-0.5
-1500 -1000 -500 0 500 1000 1500 • Solid model includes the test beam,
Die Location (micron) adhesive and strain gauge die
adhesive,
Stress (MPa) • Finite element software Intellisuite is used to simulate the stress
In the direction
of the beam (X
distribution on the surface of the strain gauge
di
direction)
i ) • 50N is applied to the test beam (~10 µε)
• Plot above shows the X direction stress at the surface of the die

cnse.albany.edu
 Gauge factor extraction

Stress at the resistor ΔRx

= π lσ x + π tσ y + π 12σ z
2 Rx
1.5 ΔR = G ΔL( )
MPa)

1
X stress R L
Stress (M

Y stress
0.5
Z stress
0 • Average X stress: 1.44 MPa
• Averageg Y stress: -0.309 MPa
-0.5 • Average Z stress: 1.76 X 10-4MPa
-800 -700 -600 -500 -400 -300 -200 • ΔL/L: 10.2µε
Location (m icron)

• Graphs show the X and Y stress along thee implanted resistor

• Gauge
g factor is calculated byy usingg average
g X and Y direction stress
from the simulation

• Expected gauge factor: 121

cnse.albany.edu
 Strain gauge response

I=10µA Strain g
gauge
g Response
p

↑
139.6
V

esistance (k--ohm)
139.5

139.4

139.3

139.2

139.1
50N 100N 150N
139.0

Re
138.9
0 20 40 60 80 100
• Strain gauge resistance is measured by Tim e (s )
driving 10µA of current and measuring the
voltage
• 14 µW of power is dissipated in the gauge
during the measurement

• Loads of 50, 100, and 150N are applied to the test beam.
• 50N load produces ~10µε in the test beam
• Calculated from the data in the graph, the gauge factor is ~120

cnse.albany.edu
 Noise

Strain gauge Response St i gauge Response

Strain R
139.6
Resistance (k-ohm)

139.5

ohm)
139.350
139.4

139.3 139.345

Resistance (k-o
139 2
139.2
139 340
139.340
139.1

139.0
139.335
138.9 139.330
0 20 40 60 80 100
139.325
Tim e (s)
139.320
139 315
139.315
139.310
30 31 32 33 34 35 36 37 38 39 40

Tim e (s)

Target Measured • Noise in the trace is ~7 ohms RMS, at a gauge

factor of 120, this represents ~0.5 microstrain
Gauge Factor >100 120 RMS.
• Analog
A l tto digital
di it l steps
t in
i data
d t preventt further
f th
Resistance ~100 kΩ 139 kΩ noise analysis
• Measured noise level is suitable for measuring
strain at the 5µε level
Resolution 5µ
µε Acceptable
p

cnse.albany.edu
Summary
• Large gauge factor is required for sensing strain at low powers
• Piezoresistive silicon is a natural choice for its large gauge factor and ability to be batch
fabricated
• Single die, multi-axis strain gauges have been designed and fabricated
• Test data shows that the gauge is suitable for measuring 5 microstrain while dissipating only
14 μW of power

Acknowledgement
We would like to acknowledge Dr.
Dr Hassaram Bakhru for his help with ion
implantation as well as Lawrence Clow and Barry Treloar for their help and useful
discussion.

This work was supported in part by the United States Navy, NAVAIR Contract
N68335-05-C-0216

Thank you for your attention

cnse.albany.edu
 Stress and strain

Strain Stress
σ A
=E Δl
ε
l
Δl
ε= l
F
l σ= Shear stress τ=F/A
A Shear Strain γ= Δl/l
1
Δl
σ Stress
F force
A cross sectional area
ε Strain
E Young’s modulus

• Stress: Normalized force applied to a part

• Strain: Normalized extension of a part
• Stress and strain are related by the Young’s modulus material property
• µstrain = 1 part per million extension

1. Simon M. Sze, Semiconductor Sensors, Wiley-Interscience (October 1994)

cnse.albany.edu