piezoMEMS Capacitor Degradation over Time & …...2018/09/14 · Radiant Technologies, Inc....
Transcript of piezoMEMS Capacitor Degradation over Time & …...2018/09/14 · Radiant Technologies, Inc....
Radiant Technologies, Inc.
piezoMEMS Capacitor
Degradation
over
Time & Temperature
Naomi Montross, Gerald Salazar, Bob Howard, Spencer Smith, Scott
Chapman, and Joe Evans
Radiant Technologies, Inc.
International Workshop on Piezoelectric Materials and Applications
Kobe, Japan
September 14, 2018
Radiant Technologies, Inc.
Summary
Why test device properties instead of device
failure over time?
Experiment samples, fixture, and procedures
Results
Conclusions
Radiant Technologies, Inc.
Why Test Properties?
Many non-linear devices depend upon inherent memory for
their function: This includes any piezoelectric device that
requires poling.
Memories accumulated by such a device over its lifetime
cause its performance to change. This is traditionally called
“ageing” but it actually affects all properties.
pMEMS devices might not suffer catastrophic failure but
instead the drift in their properties over time can lead to
system failure.
Radiant Technologies, Inc.
Testing Property Drift
The authors characterized drift in device properties for
packaged thin-PZT-film capacitors using automated testing
techniques.
The capacitors were 250nm-thick, 2% niobium-doped 20/80
PZT with platinum electrodes. (PNZT)
Hundreds of measurements were required to produce
compact representation of property drift.
The following pages describe the fixturing and automated
test procedures used to characterize the properties of the
PZT capacitors.
Radiant Technologies, Inc.
Sample Preparation Top view
COMMON CAP A
CAP B CASE
Cap A Cap B Common
TO-18 Can
Capacitor dice were packaged in
TO-18 transistor packages using
Cr/Au bond pads and gold bond
wires.
Lids were not sealed
hermetically.
The packages were inserted into
a thermal chamber for test.
Radiant Technologies, Inc.
Test Fixture
TC
Thermal
Controller
Signal Multiplexer
USB
I2C
Tester DRIVE
Tester RETURN
PWR
Chamber
Radiant Technologies, Inc.
Test Procedures
→COMBO Test A series of Wake-up stress tests applied
at room temperature immediately after
fabrication.
→Temperature Cycle Measure changes in properties during
two Temperature Cycles from room
temperature to 90C.
→Retention/Imprint Measure changes in Retained State and
Opposite State polarizations with time
at temperature.
Radiant Technologies, Inc.
Test Procedures
→Fatigue Measure polarization loss due to
Voltage Cycling at temperature.
• Switched polarization for memory.
• Monopolar cycling for actuators.
→HALT Measure change in capacitor properties
with Time-At-Temperature under DC
Electrical Bias.
• Typically executed at 300kV/cm for thin
films.
Conducted in this order using 3 packages of two capacitors
each requires 6 days.
Radiant Technologies, Inc.
COMBO Test Definition The COMBO test subjects the virgin sample to multiple cycles
to evaluate film quality.
1. Subject a capacitor to multiple wake-up cycles of
square waves and sine waves.
2. Measure hysteresis after each cycling period.
Radiant Technologies, Inc.
The very first virgin loop after fabrication. It started exactly at zero
polarization as expected indicating random domain orientation as-made.
Total = 1 Hysteresis Cycle
COMBO Test Result Virgin Loop
-50
-40
-30
-20
-10
0
10
20
30
40
50
-10 -5 0 5 10
uC
/cm
2
Voltage
Radiant Technologies, Inc.
These 100 cycles are called Recovery at Radiant. The device shows
improved hysteresis loop after Recovery. This
Total = 101 Hysteresis Cycles
COMBO Test Result 100 Cycles 1Hz Square Wave
-50
-40
-30
-20
-10
0
10
20
30
40
50
-10 -5 0 5 10
uC
/cm
2
Voltage
Green = Virgin
Solid = First 100 Square Cycles
Radiant Technologies, Inc.
The next 100 cycles were Sine Waves. Sine waves always cause Pmax to
decrease. The reason is not known but the amount is affected by process.!
Total = 203 Hysteresis Cycles
COMBO Test Result 100 Cycles 1Hz Sine Wave
-50
-40
-30
-20
-10
0
10
20
30
40
50
-10 -5 0 5 10
uC
/cm
2
Voltage
Black = 100 Original Square Waves
Blue = 100 Sine Waves
Radiant Technologies, Inc.
-50
-40
-30
-20
-10
0
10
20
30
40
50
-10 -5 0 5 10
uC
/cm
2
Voltage
The next 100 cycles are Square Wave. Pmax was restored to near its
original value. This performance meets the process target.
Total = 306 Hysteresis Cycles
COMBO Test Result 100 Cycles 1Hz Square Wave
Black = 100 Original Square Waves
Blue = 100 Sine Waves
Red = 100 Square Waves
Radiant Technologies, Inc.
Temperature Cycles Measure capacitor properties over temperature after Recovery.
1. Program two temperature cycles from room
temperature to 90C and back in 5C steps.
2. Measure Hysteresis, PUND, Leakage, and Small Signal
Capacitance every 5C step after stabilizing at
temperature.
Radiant Technologies, Inc.
Temperature Profile
12 Hrs
0
10
20
30
40
50
60
70
80
90
100
0 10000 20000 30000 40000
T
em
pe
ratu
re (
C)
Elapsed Time(s)
Cycle #1 Cycle #2
Radiant Technologies, Inc.
PUND P*, P^, dP Leakage (A)
Both the PUND and the Leakage profiles on the first temperature cycle
aged quickly. A degradation in polarization is observed.
Total = 144 Measurements over 24 Temperatures
Temperature Cycles Cycle #1
-1210
-1110
-1010
0 20 40 60 80 100
Am
ps
Temperature (C)
0
10
20
30
40
50
60
70
0 20 40 60 80 100
uC
/cm
2
Temperature (C)
P^
P*
dP
Heating
Cooling
Heating
Cooling
Radiant Technologies, Inc.
PUND P*, P^, dP Leakage (A)
The changes in properties around the temperature loop stabilized during the
second cycle.
Total = 288 Measurements over 48 Temperatures
Temperature Cycles Cycle #2
0
10
20
30
40
50
60
70
0 20 40 60 80 100
uC
/cm
2
Temperature (C)
P^
P*
dP
-1210
-1110
-1010
0 20 40 60 80 100
Am
ps
Temperature (C)
Radiant Technologies, Inc.
Retention/Imprint Retention – Switched and unswitched polarization
amplitudes retained since last write event.
Imprint – 1-second Retention Test of the opposite
state measured immediately after the retention
measurement.
1. Write Retained State.
2. Retention Delay
3. Read Retained State.
4. Execute 1-second Retention Test of the Opposite State
Radiant Technologies, Inc.
Why Retention/Imprint? Retention and Imprint would seem to be important only to
memory applications like FRAM.
All piezoelectric properties of ferroelectric materials arise
from the remanent polarization.
Drift in unswitched Remanent Polarization will couple to
piezoelectric actuator performance.
Radiant Technologies, Inc.
Total = 13 Measurements at 85C
Switched Retention Measured change in polarization over time
0
10
20
30
40
50
60
70
100 101 102 103 104 105 106 107 108 109 1010 1011 1012
u
C/c
m2
Time(s)
Radiant Technologies, Inc.
Total = 26 Measurements at 85C
Switched Retention/OS Imprint
0
10
20
30
40
50
60
70
100 101 102 103 104 105 106 107 108 109 1010 1011 1012
u
C/c
m2
Time(s)
Retained Polarization
Imprinted 1s Opposite Retained State
Radiant Technologies, Inc.
Total = 26 Measurements at 85C
UP/DOWN Retention
0
10
20
30
40
50
60
70
100 101 102 103 104 105 106 107 108 109 1010 1011 1012
u
C/c
m2
Time(s)
No retention loss at 85C.
30yr Retained Switched Polarization
Retained Un-switched Polarization
Radiant Technologies, Inc.
Total = 26 Measurements at 85C
UP/DOWN Imprint
0
10
20
30
40
50
60
70
100 101 102 103 104 105 106 107 108 109 1010 1011 1012
u
C/c
m2
Time(s)
30yr
Failure
Memory failure past 30 years
at 85C.
Imprinted DOWN State
Imprinted UP State
Radiant Technologies, Inc.
Bipolar Fatigue vs Temperature Measure the change in capacitor property or properties as a
function of polarization reversals.
1. Execute 2 alternating ±7-volt cycles every millisecond
for 1 second on an imprinted capacitor.
2. Measure PUND at 30C.
3. Double the cycling time and jump back to Step 1.
Repeat the same test on the second imprinted capacitor at 85C
Radiant Technologies, Inc.
Bipolar Fatigue vs Temperature
Fatigue was unaffected by temperature. Platinum-electroded capacitors
reached 50% of starting Remanent Polarization at 1 million cycles.
0
10
20
30
40
50
60
100 101 102 103 104 105 106 107 108
u
C/c
m2
Cumulative Cycles
Total = 30 Measurements at 2 Temperatures
Radiant Technologies, Inc.
HALT HALT – High Accelerated Life Testing
• Measure leakage growth over time with DC Bias at temperature
• Typically the device is run to catastrophic failure.
Non destructive HALT can also track properties over time.
• All properties are measured over time under DC Bias at temperature
• The test can be run to catastrophic failure but it is not necessary.
Radiant Technologies, Inc.
HALT Test Definition
Radiant Technologies, Inc.
HALT Test Definition
The Test Definition has 6 sections:
1. Set Temperature and Configure the
Fixture.
2. Apply DC Bias for increasing time
periods.
3. Measure All Properties
4. Plot All Measurements
5. Apply Branch Condition
1. Shut Down and Exit
Y
e
s
N
o
Y
e
s
No
Radiant Technologies, Inc.
HALT @ 85C & 150C
The Hysteresis Pmax of a monopolar half-loop, as would be applied to a piezoelectric actuator,
shows DC Bias-induced Ageing through the loss of Pmax with time.
Leakage remains below 1A/cm2 after 57 hours at 150C with 300kV/cm DC Bias.
Total = 44 Measurements over 2 Temperatures
0
5
10
15
20
25
100 101 102 103 104 105 106 107 108 109
HALT Hysteresis Parameters
uC
/cm
2
time(s)
150C
85C
-1210
-1110
-1010
-910
100 101 102 103 104 105 106 107 108 109
HALT Leakage
Am
pe
res
time(s)
Monopolar Actuator Half-loop Pmax
150C
85C
Bias = 7.8 volts
Leakage Test = 1 volt
Bias = 7.8 volts
Hysteresis Test = 7.8 volts
Radiant Technologies, Inc.
Conclusion 2/20/80 PNZT integrated thin-film capacitors were subjected to 420
measurements over a period of six days from 30C to 150C.
COMBO Test – Sample exhibited no Pmax decay indicating high quality
film.
TEMPERATURE CYCLING – Rapid ageing caused initial polarization
values to decay 15% during the first temperature cycle while leakage
improved 46%. Properties were stable during second temperature cycle.
RETENTION/IMPRINT– Opposing capacitors tested simultaneously
demonstrated unlimited retention at 85C with imprint failure beyond 30
years.
FATIGUE – Remanent Polarization decayed 50% at 106 reversals at both
30C & 85C .
HALT – Leakage change due to 300kV/cm DC Bias was minimal at both
85C and 150C over 57 hours. However, Actuator-type Pmax decayed
30% at 150C in 57 hours under the same test conditions.