Crack Initiation at Underfill/Passivation Interfaces · 2 Problem Statement • Underfill protects...

Post on 13-Apr-2018

239 views 3 download

Transcript of Crack Initiation at Underfill/Passivation Interfaces · 2 Problem Statement • Underfill protects...

Crack Initiation at Underfill/Passivation Interfaces

Raymond A. Pearson

Center for Polymer Science and EngineeringDepartment of Materials Science and Engineering

Lehigh University, Bethlehem, PA 18015rp02@lehigh.edu

October 2002

2

Problem Statement• Underfill protects interconnects from thermal

expansion stresses in flip-chip• Underfill adhesion important to flip chip reliability

Solder Interconnects

UnderfillPassivation

Substrate

Silicon Die

Thermal cycling can result in debonding at UF/PI interface

Baseline 100TC 200TC

300TC 500TC 1500TC

3

• Bulk toughness testing of commercial underfills:

• Dexter underfill resin exhibits higher toughness.

K IC (MPa-m 1/2 ) G IC (J/m 2 )

Dexter FP-4511 2.15 431.4

Zymet X6-82-5 1.47 201.7

UnderfillSEN-3PB Test Method

Background

R. A. Pearson and P. Komnopad, Proc. IMAPS Material, 344 (1999)

4

Interface Tested Failure Mode Gc (dry) J/m2

UF/PI 2525:DEXTER FP 4527ZYMET X6-82-5

Interfacial UF/PICohesive near PI

14.9 ± 1.5 81.4 ± 3.9

• Adhesive strength of commercial underfill resins:

Background

R. A. Pearson and P. Komnopad, Proc. IMAPS Material, 344 (1999)

• Differences in adhesion attribute to chemical interactions.

5

3-LIQUID PROBE METHOD

L

S

V

γsv

γLV

γs L

θ

WA = WLW + WAB = γ1 + γ2 - γ12

WLW = 2 (γ1 γ2 ) 1/2

WAB = 2 (γ1 γ2 ) 1/2 + 2 (γ1 γ2 )1/2

LW LW

+ +- -

SAMPLE BED

VALVE

TWIN SYRINGE PUMPS

Solve nt plusad so rb at e

Solve nt

MICROCALORIMETER

DETECTOR

FLOW MICROCALORIMETRY

-∆HAB = CACB + EAEBBoth methods use probe liquids to characterize interactions between solid surfaces.

Background

6

Surface CA EA CB EB γLW γ+ γ -

“Dexter”

“ Zymet”

PI-2525

BCB

1.70

0.05

0.73

0.40

1.10

0.00

0.61

1.40

1.49

4.60

* * *

4.60

0.22

0.00

* * *

0.44

41.9

46.6

46.8

40.1

0.60

0.11

0.24

0.45

1.25

4.64

5.70

1.63

* * * - no appreciable exotherm detected.Bases: acetone, acetonitrile, ethylacetate, pyridine, triethylamine.Acids:chloroform, iodine, phenol.

Drago-type characterization predicts strong bond between Zymet underfill resin polyimide.

-∆HAB = CACB + EAEB

Background

R. A. Pearson and P. Komnopad, Proc. IMAPS Material, 344 (1999)

7

Interface Tested Failure Mode Gc (dry) J/m2

UF/PI 2525 (control):DEXTER FP 4527

UF/PI 2525 (HPS):DEXTER FP 4527

UF/PI 2525 (UV/ Ozone):DEXTER FP 4527

UF/PI 2525 (Plasma/02):DEXTER FP 4527

Interfacial UF/PI

Interfacial UF/PI

Slightly cohesive

Moderately cohesive

14.9 ± 1.5

18.1 ± 1.0

105.2 ± 7.0

130.9 ± 10.0

Background

R. A. Pearson and P. Komnopad, Proc. IMAPS Material, 344 (1999)

8

A. J. Kinloch, G. K. A Kodokian & J. F. Watts, J. Mater. Sci. Lett., 10 (1991) 815.

Background

9

Underfill side

O2/PlasmaUntreated UV/ Ozone

PI sideBackground - FP 4527 on PI

R. A. Pearson and P. Komnopad, Proc. IMAPS Material, 344 (1999)

10

O

C

O

O

OC

O

C

O

O

C O CH2

CH CH2

O

OCH2

HCH2C

O H

H N

N

H

C2H5

CH3

C

CH3

CH3

O CH2

CH CH2

O

OCH2

HCH2C

OHN N CH2CH2NH2

B. J. McAdams and R. A. Pearson, Proc. Adhesion Soc. Conf. Williamburg, VA, Feb. (2001)

Background - Three Model Underfills

11

Type of Failure Gc (Dry) J/m2Cycloaliphatic/ Anhydride (Unfilled)

Adhesive at UF/PI Interface 24.2 ± 2.5

Cycloaliphatic/ Anhydride (Filled)

Adhesive at UF/PI Interface 13.4 ± 5.3

Bisphenol F/ 2,4-EMI (Unfilled)

Adhesive at UF/PI Interface 27.9 ± 2.8

Bisphenol F/ 2,4-EMI (Filled)

Adhesive at UF/PI Interface 35.2 ± 2.9

Bisphenol A/ AEP (Unfilled)

Adhesive at PI/Al Interface 147.1 ± 10.0

Bisphenol A/ AEP (Filled) Cohesive in PI >200

Adhesion to PI-2555 (Untreated)Underfill

Background

12

Background - Epoxy/PI Strength

C.K. Gurumurthy,” Ph-D Dissertation”, Cornell University (2000)

13

Background

Equilibrium interpenetration zone or width of polymer interface(W)

W = 2b/(6W = 2b/(6χχ))0.50.5

χχ = V/RT(= V/RT(δδaa--δδbb))22

b: statistical segment length

χ: Flory-Huggins intermolecular interaction

V: molar volume

R: gas constant

T: temperature(°K)

δ: solubility parameter

R.P. Wool, Polymer Interfaces-Structure and Strength, Hanser , NY 1995,34

W

W= width of interphase region

14

Background

Polymers can adsorb onto surfaces via dispersive and site specific interactions.

P.G. de Gennes, Soft Interfaces: the 1994 Dirac memorial lecture, Cambridge University Press, 1997

Molecular Interactions

Gc = WA(1 +φ)

Gc ~ 1/χ

15

• To investigate the role of polyimide chemical structure on the strength epoxy-polyimide interfaces.

• To determine the strength of epoxy-polyimide interfaces using a fracture “mechanics-like” approach.

Objectives

16

Materials - Model Polyimides

N N

O

O

O

O

O

n

Poly(pyromellitic dianhydride-oxydianiline)

PMDAPMDA--ODAODA

N

O

O

N

O

O

n

O

5(6)-Amino-1-(4-amino-1-(4-aminophenyl)-1,3,3-trimethylindanbenzophenonetetracarboxylic) dianhydride copolymer

BTDABTDA--DAPIDAPI

17

N

O

O

N

O

O

CF3F3C

O

n

6FDA6FDA--ODAODA

Poly(hexafluoroisopropylidenediphthalic anhydride-oxydianiline)

Materials - Model Polyimides

N

O

O

N

O

On

Poly(3,3’,4,4’-biphenyltetracarboxylic dianhydride-phenylene diamine)

BPDABPDA--PDAPDA

18

Materials - Model Underfill System

H

CH

O OH2C CH2

O O

N

NCH2CH3

H

H3C

Bisphenol F Resin-unfilled

Imidazole Curing Agent(2,4-EMI)

Curing:

60°C- 4 hours

150°C-2 hours

19

Approach - Sample Preparation

SurfacePolishing/Cleaning

SurfacePolishing/Cleaning

SpinCoating/

Sputtering

SpinCoating/

Sputtering

ADCBSample

Assembly

ADCBSample

Assembly

Underfill Flow

And Cure

Underfill Flow

And Cure

20

P

P

a

Approach - ADCB Testing

PC δ=

=

dadC

wPGc 2

2max

C: complianceδ: displacementP: loadw: specimen widtha: crack length

D. Reedy and T. Guess, Sandia National Laboratories

21

P

P

δ

Approach ~ ADCB Testing

2'4/1''1

33'

1

2

])()(1)[1()(6a

hBYhE

PaGc−++= ληρη

Testing ParametersLoading:

Monotonic, UniaxialCrosshead Rate:

1.27 mm/min

0

20

40

60

80

100

120

140

160

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Displacement (mm)

Load

(N)

aP

C →=δ

Bao et al, Int. J. Solids Struct, 29 (1992)

22

3-LIQUID PROBE METHOD

L

S

V

γsv

γLV

γs L

θ

WA = WLW + WAB = γ1 + γ2 - γ12

WLW = 2 (γ1 γ2 ) 1/2

WAB = 2 (γ1 γ2 ) 1/2 + 2 (γ1 γ2 )1/2

LW LW

+ +- -

SAMPLE BED

VALVE

TWIN SYRINGE PUMPS

Solve nt plusad so rb at e

Solve nt

MICROCALORIMETER

DETECTOR

FLOW MICROCALORIMETRY

-∆HAB = CACB + EAEB

Approach -Three Liquid Probe Method

23

Approach - Butt Tensile Joint

• Pull BTJ specimen in screw-driven Instron at 0.2mm/s • Record maximum nominal applied tensile stress• Calculate stress intensity factor associated with the

singular stress field surrounding the interface corner

2hθ

KF =σ h1−λ Ap (ν )

E.D. Reedy and T. Guess, Sandia Nat. Lab.

Fixture LockingHole

Instron GrippingHole

KFC =σ*h1−λ Ap (ν)

24

• Tensile plugs are precision machined to guarantee ends are flat and perpendicular to cylinder axis, and the edges are left sharp.

• Surfaces are polished carefully without breaking corners.

• Surfaces are cleaned, and coated with the appropriate coating, if necessary.

• Samples are locked in a collar fixture, using a removable shim to get the appropriate bond line thickness.

• The edges of the gap are sealed, with small holes in the front and back of the specimen; tubes are pressed up against the holes to form a tight seal.

• The sample is brought to flow temperature, underfill is forced into the gap and cured.

• The sample is removed during cooling and the seal removed.

Approach - Sample Preparation

25

Type of Failure G c (Dry) J/m 2

PMDA-ODA Adhesive at UF/PI Interface 37.16

BTDA-DAPI Adhesive at UF/PI Interface 178.87

6FDA-ODA Adhesive at UF/PI Interface 78.70

BPDA-PDA Adhesive at UF/PI Interface 20.69

Adhesion to UnderfillPolyimides

Results - Interfacial Strength

26

Results - Epoxy on PI contact angles

Wa = γl(1+cosθ)

γBis F/EMI = 45.2 mJ/m2

Polyimide Type θAdv Wa Gc

Bis F/EMI-24 (mJ/m2) (J/m2)

PMDA-ODA 47.9 75.53 37.2

BTDA-DAPI 18.9 87.98 179

6FDA-ODA 30 84.37 78.7BPDA-PDA 19.1 87.94 20.7

Contact Angle Epoxy -> Work of Adhesion

27

Van Oss, Good and Chaudhury

R.J. Good, Journal Adhesion Science and Technology,(6) 12, 1992, 1269-1302

Wa = γl(1+cosθ) = 2(γlLW γs

LW)½ + 2(γl+ γs

-)½ + 2(γl- γs

+)½

Polyimide Type θAdv θAdv θAdv γs LW γs

+ γs - Gc

water Ethylene glycol Diiodomethane (J/m2)

PMDA-ODA 76.6 59.6 49.2 34.7 0.004 12.18 37.16

BTDA-DAPI 71.5 58.7 28.2 44.9 1.44 35.4 178.87

6FDA-ODA 81.4 62.1 39.1 40.1 0.168 8.065 78.7BPDA-PDA 79.2 57.1 19.8 47.84 0.34 7.55 20.69

Results - Probe Molecules on PI

28

• Adsorption of bis F epoxy on PI surface is mostly reversible (~83%)

• Adsorption of EMI is about 25% reversible!

Results - Epoxy and EMI Molecules on PI Surface SA Mass Solvent Probe Conc. Ads Peak Disp kJ/mole

(m^2/g) (g) (mM) (mJ) (uM)

PI 2611 17.4 0.0246 Xylene Bisphenol F 10.4 178.758 1.841 226.98PI 2611 17.4 0.0246 Xylene Bisphenol F 10.4 180.671 1.831 230.54

PI 2611 17.4 0.0246 Xylene Bisphenol F 10.4 149.873 1.868 187.457PI 2611 17.4 0.0246 Xylene Bisphenol F 10.4 149.184 1.84 189.435PI 2611 17.4 0.0246 Xylene Bisphenol F 10.4 149.399 1.842 189.502

PI 2611 17.4 0.025 Xylene EMI 2,4 10.75 84.296 0.813 238.356PI 2611 17.4 0.0245 Xylene EMI 2,4 10.75 80.452 0.802 235.479

PI 2611 17.4 0.025 Xylene EMI 2,4 10.75 28.66 0.59 111.669PI 2611 17.4 0.025 Xylene EMI 2,4 10.75 19.346 0.736 60.426PI 2611 17.4 0.025 Xylene EMI 2,4 10.75 16.597 0.697 54.74

29

ADCB vs. BTJ Results

30

Fracture Surfaces

Completely Cohesive Fracture Predominantly Adhesive

Significant Cohesive Fracture Some Cohesive

Cycloaliphatic/Anhydride

System

Bisphenol F/2,4-EMISystem

6061-T6 Aluminum Surface PI-2555 Polyimide Surface

31

Future Work

AB

C

D

E

FA

B

CA

TOP VIEW

SIDE VIEW

STRESSSINGULARIES

A B C

32

Summary/Conclusions• PI-Epoxy adhesion is sensitive to the structure of the

polyimide.

• Contact angle generated Acid-Base parameters qualitatively predicts the best and the worst adhesion.

• Microflowcalorimetry experiments indicate that EMI adsorption plays an important role in adhesion.

• Crack initiation tests agree well with conventional fracture mechanics tests. More tests are underway!

33

Acknowledgements• HD Microsystems for providing polyimide resins.

• Arch Chemicals for providing polyimide precursors and monomers.

• Zymet Corporation for providing underfill materials.

• Loctite/Dexter Corporation for providing underfill materials.

• SRC and PITA for their financial support.