New Reliability Assessment Methods for MEMS

Prof. Mervi Paulasto-Kröckel

Electronics Integration and Reliability

Aalto University

Helsinki School of Economics

Helsinki University of Technology

University of Art and Design

Helsinki

A merger of leading Finnish universities in 2010:

ChemicalTechnology

Electrical Engineering

Economics ScienceArt andDesign

Engineering

Art, Business,Science and Technology

School of…

Electronics Integration and Reliability –Past and Present

Metal – ceramic joining

Electronics assembly technology, soldering

1990 2000 2010

Compatibility of dissimilar materials and interconnect technologies since 20 years !

MEMS 3D integration, Bioelectronics & -sensors

Courtesy of VTI

Outline

• Reliability challenges in MEMS – status of the industry• What kind of changes are needed?• Reliability characterization of gyro

– Development of methods– Results

• TH and shock impact tests• FA

• Reliability characterization of microphone– Reliability assessment methods– Results

• Summary

Characteristics of reliability assessment of MEMS

• Reliability evaluation in functional state– External stimulus (sensors)

– Monitoring of output (actuators)

• Definition of failures– Based on functional characteristics

– More than one criterion for failure

– Requires real-time monitoring system

• Methods for health monitoring are device specific and non-standardized

Note: Standardized methods to produce loading still apply!

• Thermal cycling, mechanical shock, vibration, temperature / humidity (e.g. 85/85), corrosion

MEMSsensor

Test environment

Known stimulus

Control unit

Comparison of output with input

Health monitoring &

system control

MEMSactuator

Detection of output

outp

utou

tput

inpu

tin

put

Reliability challenges in MEMS• Use environment

• Mechanical shock impacts and vibrations (specifically moving parts w and w/o impacting surfaces!)

– Mobile devices vulnerable to high-G shocks

• Rapid changes of temperature

• Moisture (specifically open structure devices!)

• Defects and contaminations from processing

• Typical failure modes• Unwanted interactions at contacting surfaces – friction, adhesion,

stiction and wear• Fracture• Corrosion, delamination

• Package reliability• Maintain hermeticity, package induced stress

Current status in MEMS reliabilityassessment

Design of Experiment

Modeling Reliability test

Failure analysisCompromisestypical !

Trial & ErrorMethod

Functionality test

Observation

MEMS

Microelectronics

• Limited system and package level reliability data from environmental tests available

• Limited physics of failure knowhow

Development by Trial & Error Methods

- Only isolated areas of a system with functional recipes are known

Improvements needed

Design of Experiment

Modeling Reliability test

Failure analysis

Understanding of materials and specifically materials interactions

Methods of reliability evaluation

Methods of reliability simulation

Methods of failure analyses for effective identification of root cause

Development methodology for reliabilityTaSi2 + SiC

TaSi2 + TaC

Ta5Si3 + Ta2C

Ta2Si + Ta2C

Ta3Si + Ta2C

Ta + Ta2C

Ta2C TaC

TaC + C

TaC + SiC

TaSi2 + SiC

TaSi2 + TaC

Ta5Si3 + Ta2C

Ta2Si + Ta2C

Ta3Si + Ta2C

Ta + Ta2C

Ta2C TaC

TaC + C

TaC + SiC

C

Ta Si

x(Si)

x(C)

0 0.2 0.4 0.6 0.8

0.8

1.0

1.0

0

0.2

0.4

0.6

TaC

Ta2C

SiC

Ta3Si Ta2Si Ta5Si3TaSi2

TaC+C+SiC

TaC+SiC+TaSi2

TaSi2+SiC+Si

Ta+Ta2C+ Ta3Si

Ta2C+TaC+TaSi

2Ta2C+TaSi

2+Ta5Si3

C.L.

MEMS Gyroscope reliability

• Device: a 3-axis MEMS Gyroscope– CoC assembly of ASIC and MEMS– Dimensions: 3.1 mm x 4.2 mm x 0.8 mm

• Reliability characterization:– FEM simulations and shock impacts in all

three orthogonal axes and various shock levels 1,500G – 15,000G

– Non-functional and functional tests– Temperature/humidity test 85°C / 90 RH%

100m

m

100mm

1,0 mm thick 8-layer FR4 board for TH

2,0 mm thick single Cu layer alumina board for shock impact

1. Hollow rotating axle – large enough to fit all cables

2. Sample holder jig – Placed in an angle of 54.74°to excite all 3

axes of the gyroscopes

3. Servo motor

4. Clutch coupling

5. Servo drive to control the motor

6. PC software – Control of the angular velocity

– Acquisition of angular velocity data

7. Wireless communication unit on the rotating axle

8. Wireless communication unit of the PC

9. Slip rings – Power to the wireless communication unit

and the gyroscopes

Test methods – 3 axis Gyroscopes

][208)74.54cos(][360..)cos( dpsdpsgeaxlezyx ≈°⇒⋅Ω=Ω=Ω=Ω α

1

2

7

4 3

8

5

9

6

Test methods – healt monitoring

• Failure criterion: predefined change in the – Offset– Sensitivity– Noise

in the output of any of the three axis

• Repeated at different angular velocities: – 0, ± 450, ± 1350, and ± 1800 degrees per second

• The health monitoring procedure was repeated once per hour

Test methods – shock impact

XY

Z+ Z-

Y

Z-

Z+

Shock impact tester(up to 100 000 G)Functional evaluation between shock impacts

velocity

X

• Evaluated parameters– Offset– Sensitivity– Noise

• Four impact orientations: X, Y, Z+, and Z-

Rigid strike surface

Pne

umat

ic

cylin

der

• Health monitoring between the shock impacts

• Devices were not electrically connected during shock impacts

Shock impact results

• Decelerations to produce package failures is about two times that of electrical device failures

• Differences in deceleration tolerance were analyzed statistically

1. Package level failures• Z+ differs statistically significantly from

others• Z- differs from X statistically significantly

2. Electrical device failures• Y differs statistically significantly from Z+

=> Deceleration tolerance has an impact orientation dependency

X Y Z+ Z-v

Y X Z+ Z-

Package failure 8800 10288 14975 8388

Electrical failures 3919 4525 5189 4319

0

2 000

4 000

6 000

8 000

10 000

12 000

14 000

16 000

18 000

De

cele

rati

on

/ [

G]

Impact Orientation

Package failure

Electrical failures

X-o

rient

atio

nY-

orie

ntat

ion

Z-o

rient

atio

nDisplacements Stress distribution Failed device

v

v

v

Shock impact – package failures

Shock impact – package failures

Borosilicate glass– Fracture paths (averages):

• In the borosilicate glass: 70 %• In the silicon: 16 % • Along the fusion bonded interface: 14 %• Along the the anodic bonding interface: 0%

• Y orientation

24 %

74 %

2 %

All orientations

Z-axis failed All axes failed

X and Y axis failed

N = 67

• Transient failures in 22% of the tested gyroscopes X Y Z+ Z-

v

• Electrical failure modes

Shock impact – electrical failures

0 %

20 %

40 %

60 %

80 %

100 %

X Y Z+ Z-

Shock Impact Oroentation

All axes failed Z-axis failed X and Y axis failed

Shock impact – FEM simulationY-

orie

ntat

ion

X-o

rient

atio

nZ

-orie

ntat

ion

Acceleration (to bring moving

structures in contact)

≈≈≈≈ 4 500 G

≈≈≈≈ 1 500 G

(Deformation enlarged 30 times)

≈≈≈≈ 1 800 G

Shock impact – active element failures

• Failure analyses of internal failure sites:– The “cap” of the device is thinned down by DRIE etching– Observation windows are cut in the thinned-down caps by FIB– Observation by the SEM or optical microscopy

• Examples of internal failure modes:– Fractured comb arms(A)

– Fractured comb fingers(B)

– Stuck MEMS elements(C)

– Chipped edges (D)

(A) (B) (C) (D)

Temperature / humidity test (85°C / 90 %RH)

• Failures detected in 13 out of 27 gyroscopes after 180 days of exposure

– 10 failures before 50 days – 11th failure after 148 days

=> At least two different failure mechanisms

• Early failures:– Operation recovered after the devices were

removed from the test environment– Mass decrease of 6% was measured during 7

days at room temperature (devices removed from the substrate by shearing)

=> Failures are most likely due to short circuits by absorbed moisture

– No delamination or voiding of the underfill or the RDL polymer detected

10 1 0001001

5

10

50

90

99

Time / [hours]

Cum

ulat

ive

Fai

lure

Per

cent

age

/ [%

]

Failure mode 1

F=10 / S=0

Failure mode 2

F=3 / S=14

MEMS Microphone reliability

• Device: Multi-chip module composed of– MEMS chip: acoustic sensor

– ASIC

• Reliability characterization:– TH 85°C/85% RH test

– Multigas corrosion

– Shock impact test

Test methods – Microphone

• TH 85 ºC / 85 RH: Loudspeaker outside a test chamber

• Multigas corrosion : Loudspeaker inside the test chamber

Test Chamber

LoudspeakerMicrophones

Heat-resistant

elastic film

11 cm

Loudspeaker

Double chamber configuration

MicrophonesProtected loudspeaker• Toperation ≤ 90°C• “waterproof”• Equipped with internal

• Humidity sensor• Internal microphone

for monitoring

Loudspeaker

Microphones fixed to a jig

Copper coupons for monitoring the atmosphere

Environment chamber Gas control unit

Health monitoring for the microphones

Mixed gas test method – Microphone

2 microphones per board

Power supply

Clock signalgenerator

Audio amplifier

Humidity and temperature sensor

Monitoringprogram

Health monitoring – microphones

Group of ElectronicsIntegration and Reliability,Department of Electronics

Magnitude response

Difference response

Chamber setup

• Microphones in the corrosion chamber for 90 days

• Volume changes 8 times per hour (28.7 litres/minute)

EnvironmentºC % RH

Cl2 (µg/m 3)

H2S(µg/m 3)

NO2(µg/m 3)

SO2(µg/m 3)

Test chamber 30 70 19 262 188 136

Nordic outdoor 5 78 0,9 4,6 28 30

Harsh industry 25 49 15 135 23 550

Response of the microphones

Time / [days]

Mag

nitu

deR

espo

nse

/ [dB

]