Parametric Analysis of TEBGA Reliability Using a Finite-Volume-Weighted Averaging Technique Hsien-Chie Cheng Computational Solid Mechanics Laboratory National.

Electronic Packaging Lab. of NTHUElectronic Packaging Lab. of NTHU國立清華大學動機系電子構裝實驗室

Electronic Packaging/CAE Labs

National Tsing Hua University







Parametric Analysis of TEBGA Reliability Using a Finite-Volume-Weighted Averaging Technique

Hsien-Chie ChengComputational Solid Mechanics Laboratory

National Center for High-Performance Computing

Kuo-Ning ChiangPower Mechanical Engineering


Chao-Kuang ChenComputational Solid Mechanics Laboratory

National Center for High-Performance Computing










Outline

IntroductionModeling The Finite-Volume-Weighted Averaging TechniqueParametric Design of The TEBGA ReliabilityConclusions










Introduction

BGA: PBGA, CBGA, TBGA, CCGA PBGA over CBGA:

Lower cost of substrate Smaller global CTE mismatch from PCBLower profile

Conventional Over Molded PBGA Packages Experience:Excessive moisture sensitivity of substratePoor thermal performance of over molded compound

TEBGA Provides:The enhanced thermal performance over PBGARelatively cost-effective over CBGA

With Solder Joint Reliability(SJR) in Concern:Similar to PBGA, potential SJR problems still occur in TEBGA.










Introduction (Cont.)

Many Studies on C-up PBGA Reliability:Paydar et. al., 1994; Pan, 1994; Hong, 1997; Jung et. al,1997;Winter and Wallach, 1997; Lau and Pao, 1997, etc.

Few Investigations on C-down TEBGA: Pao et. al, 1998; Lee and Lau, 1998, etc.Most focus on the prediction of SJR.SJR is function of many parameters.Appropriate combinations of these parameters can signigicantly enhance SJR.

Two Techniques for this Purpose:Design of Optimization (DO) or Design of Experiment (Box, 1978)

Require tremendous, time-comsuming trialsParametric FEA (Yeh et. al, 1996)

More cost-effectiveReflect the effect of each parameter on the problem in concern










Introduction (Cont.)

Characterization of Strain/Stress Concentration Field is Critical.Fatigue life Stress/Strain InformationStress/Strain Information Mesh Desnsity

(Due to material and geometry singuarities)

Techniques:Explicit geometry representation (Fillet)

High mesh densityMaterial-nonlinear modeling

Work only for Stress Concentration ProblemAveraging Technique (Clark and Mcgregor, 1993; Akay et.al., 1997)

Considerably conservative result is obtained due to the whole domain area is averaged










Purposes

An Improved Averaging Technique: Finite-Volume-Weighted Averaging Technique

Instead of averaging the response within the whole domain, the structural response is averaged in a finite zoneDetermine the dimension of the finite zone using an engineering emprirical approach

Design Parametric Finite Element AnalysisDesign parameters:

Geomertry/Thickness : Die/PCB/Die Attach/Heat Spreader/BT SubstrateGeometry/Size : DieMaterial/Young’s Modulus : FR-4/BT/Die AttachMaterial/CTE : BT/FR-4










PBGA and Thermal Enhanced BGAPBGA and Thermal Enhanced BGA

Heat spreader Die attachDieBT

Solder MaskSolder BallEncapsulant

Copper Ring

.










256-Pin TEBGA and Modeling

Heat Spreader

PCB

Table: TEBGA Geometry Data

Table: TEBGA Material Property

Co0Co0










Solder Joint Material Property

Temperature-dependent (Plasticity)Time-dependent (Creep)Garfolo Hyperbolic Sine Law

RTH

BA ncp exp)(sinh

etemperatur absolute The T(J/mol/k) 8.31 : constant gas The R

(J/mol) 52961 :energy n Activatio Hstress Equivalent

3 exponent Stress nPa)0.0805(1/M B

c)147.9(1/se Arate strain creep equivalent uniaxial The

:

cp

20 oC

100 oC

0.0100.0080.0060.0040.0020.0000.0

10.0

20.0

30.0

Strain

Stre

ss (

MP

a)










Fatigue Life Prediction

An Empirical Coffin-Manson Relationship

)/1(

22

1c

f

IE

fN

0.5- : exponent ductility fatigue The c0.325 : coeffienct ductility fatigue The

loading thermal of cycle one in range strain inelastic The

failure to cycle mean The

f

IEfN










Improved Volume-Weighted Averaging Technique

Finite-Volume-Weighted Averaging Technique

Rules to Determine the Dimension of the Finite ZoneIn an empirical engineering approximation manner:

Small enough to capture the maximal strain fieldLarge enough to obtain a converging solution as the mesh density increases

n

e

n

ee

ee

dd1 1

~

zone the in strain averaging weighted-volume The element th-e the of (area) volume The

element th-e the of strain The zone the in elements of number total The

~een










Determination of the Finite Zone

# of Elements within the fan shape: 18, 66, 192, 504

Four different radii are defined: 0.01, 0.02, 0.04, 0.06 mm. A net termperature swing = ; plastic behaviors considered.20 zone provide considerable agreement to the proposed criterion.

Co100

m

Fan-shape circular sector 0.12 mm

60050040030020010000.0013

0.0015

0.0017

0.0019

0.0021

0.0023

0.0025

0.0027

0.0029 radius=10 (Micro meter)

radius=20radius=40

radius=60

Mesh Density

Avera

ged E

quiva

lent P

lastic

Stra

in

A CB D

A B

C D










Finite Element Modeling

1/2 package is modeled due to the symmmetric condition2-D plane strain finite element modelElements: 20685 Nodes: 21129The inelastic strain within the characterized finite zone will be averaged using a Volume-Weighted Technique.










Parametric Design of TEBGA Reliability

Design Parameters:Geomertry/Thickness : Die/PCB/Die Attach/Heat

Spreader/BTGeometry/Size : DieMaterial/Young’s Modulus : FR-4/BT/Die AttachMaterial/CTE : BT/FR-4

Thermal Cycling5 min linear temperature loading/unloading20 min low/high termperature dwell periodsTest temperature range: -40 ~ 125 Stress free temperature: 25Assume the is stablized within two cycles

IE

182 1382 1682 2882 3000

25

-40

125

CoCo










Geometry: Die Size

Nonlinear relationship is detected Between:Solder Joint Fatigue Life -vs.- Die Size

The maximum fatigue life appears at die size=7 mmCavity-down TEBGA / / Cavity-up PBGA

Cavity-up PBGA: The larger the die size; the shorter the fatigue life

The location experiencing with the largest CTE mismatch is right beneath the chip

Cavity-down TEBGA: A nonlinear relationship is obtained

A more complicated failure mechanism)

12.010.08.06.04.02.00.02000

2200

2400

2600

2800

3000

3200

Die Size (mm)

Num

ber o

f Cyc

les to

Fail

ure

Die Size










Geometry: Die Size (Cont.)

Two Major Failure Mechanisms:Global Mismatch:

CTE of PCB >> CTE of the PackageThe farther distance from the center of the package, the larger

deformation

Local Mismatch:CTE mismatch between the heat sprader and the die

5432100.0076

0.0086

0.0096

0.0106Die Size=2.Die Size=6.Die Size=10.

Ball Numbering

Max

imum

Ave

rage

Inel

astic

Stra

in

1 2 3 4

Ball Numbering










Geometry:Thickness

0.60.50.40.30.20.12000

2200

2400

2600

2800

3000

3200

Die Thickness (mm)

Num

ber o

f Cyc

les to

Fail

ure

1.51.31.10.90.70.52000

2200

2400

2600

2800

3000

3200

PCB Thickness (mm)N

umbe

r of C

ycle

s to

Failu

re

0.100.080.060.040.020.002000

2200

2400

2600

2800

3000

3200

Die Attach Thickness (mm)

Num

ber o

f Cyc

les to

Fail

ure

0.50.40.30.20.12000

2200

2400

2600

2800

3000

3200

Heat Spreader Thickness (mm)

Num

ber o

f Cyc

les to

Fail

ure

0.50.40.30.20.10.02000

2200

2400

2600

2800

3000

3200

BT Thickness (mm)

Num

ber o

f Cyc

les to

Fail

ure

Die Thickness PCB Thickness Die Attach Thickness

Heat Spreader Thickness BT Thickness










Materials: Young‘s Modulus/CTE

PCB Y’s Modulus BT Y’s Modulus Die Attach Y’s Modulus

BT CTE PCB CTE

26.024.022.020.018.016.014.02000

2200

2400

2600

2800

3000

3200

PCB Young Modulus (GPa)

Num

ber o

f Cyc

les to

Fail

ure

26.024.022.020.018.016.014.02000

2200

2400

2600

2800

3000

3200

BT Young Modulus (GPa)

Num

ber o

f Cyc

les t

o Fa

ilure

3.53.02.52.01.52000

2200

2400

2600

2800

3000

3200

Die Attach Young Modulus (GPa)

Num

ber o

f Cyc

les to

Fail

ure

2e-52e-52e-52e-51e-51e-51e-52000

2200

2400

2600

2800

3000

3200

BT CTE (ppm/Degree in Celcius)

Num

ber o

f Cyc

les to

Fail

ure

2e-52e-52e-52e-52e-51e-51e-51e-50

800

1600

2400

3200

PCB CTE (ppm/Degree in Celcius)

Num

ber o

f Cyc

les to

Fail

ure










ConclusionsIntroduction of a Finite-Volume-Weighted Averaging Technique:

Dimension of the zone determined using an empirical approximation mehtod.The strain/stress concentrtion responses can be characterized, and would no more be mesh-sensitive/dependent.Further experimental verification to really predict fatigue life of Solder Joints.

A parametric study of the reliability of the TEBGA assembly with respect to geometry and material parameters:

The design guideline for obtaining optimal reliability:

Educing a more accurate conclusion requires a sophisticated design optimization technique

MaterialProperty

PCB Y’sModulus

BT Y’sModulus

Die AttachY’s Modulus

BTCTE

PCBCTE

Dimension

Reliability










Conclusions (Cont.)

A Nonlinear Relationship Between SJR and Die Size/PCB CTE.Extensive Investigation on the Major Failure Mechanisms for the case of SJR and Die Size. A different failure mechcanism from that of C-up PBGA.Further investigation:

Effect of other failure mechanisms, such as die cracking on the package reliability.The nonlinear relationship between SJR and PCB CTE.

Parametric Analysis of TEBGA Reliability Using a Finite-Volume-Weighted Averaging Technique Hsien-Chie Cheng Computational Solid Mechanics Laboratory National.

Documents

concern slide

averaged slide

tebga material property

improved averaging technique

pbga reliability

finite zone

pin tebga

ccga pbga