Isotropic Electrically Conductive Adhesives - TheCATweb.cecs.pdx.edu/~jmorris/ece507_microelectronics_seminar/ECE507... · Portland State University ... Percolation limits of different

5/7/2012

1

James E. MorrisDepartment of Electrical & Computer Engineering

Portland State University

P.O.Box 751, Portland OR 97207-0751, USA

[email protected]

[email protected]

IsotropicElectrically Conductive Adhesives

5/7/2012

ICA: Isotropic Conductive Adhesives

Current pathPowder

Flakes (Ag)Polymer

ICA: Flip Chip Interconnect

ICA: SMT Interconnect

Contents•Introduction

•Applications•Processing

•Resistance & measurement•Nanotechnology•Layering effect•Adhesion•Drop test•Galvanic corrosion•Cure models• Miscellaneous

James E. Morris OSU Seminar2

5/7/2012

2

5/7/2012

SMT joint showing Ag flakes

Flip-chip joint showing Ag flakes: Top: Si chip bump Bottom: PWB pad


ICA applications:Solder replacement in SMT &

flip chip

Bosch Diesel Fuel Injection Pump

ICA Smart Cards & RF ID (IZM-Berlin)

5/7/2012James E. Morris OSU Seminar4

5/7/2012

3

ICA Microvias: Ag, Cu, & LMP(Low Melting Point) Alloy

[Das et al (Endicott Interconnect), ECTC’06]

5/7/2012

Failure modes →James E. Morris OSU Seminar5

(Sintered NanoParticle) ICA Die Attach [Bai et al (VaTech), HDP’05]


5/7/2012

4

ACA: Anisotropic Conductive Adhesive (& ACF)IC

BumpAdhesive

Glass substrateElectrode

Conductive particles

Ni particleCoated polymer particleJames E. Morris OSU Seminar7

Technology Drivers

5/7/2012

Environment Environmental:

- No-flux, no-Pb

No-Pb solders -Melting temperatures

-Thermomechanical stress

Silver effects

Economics Manufacturing

Fewer process steps

Available technologies (stencil/screen/dispense)

Technology drivers(c.f. solder)

Thermomechanical stress Low process temperature High compliance

Fine-pitch (area array) interconnect Smaller particles than solder No slumping Inter-metallic diffusion in solders Electromigration


5/7/2012

5

Processinge.g. Flip-Chip Assembly (IZM, Berlin)


Cure Process

5/7/2012

Log Time

Tem

per

atu

re

Tg0

gelTg

Tg∞

Liquid

Sol/Gel Rubber Gel Rubber

Gel Glass

Char

Sol/Gel GlassGelation

Vitrification

Complete CureVitrification

Devitrification

Tc>T∞

Tc<T∞

Sol Glass


5/7/2012

6

Thermal Cure: Optimization

5/7/2012

20

40

60

80

100

120

140

160

0 50 100

Time in min

Te

mp

era

ture

in º

C

Typical specification:

Temperature T and time t

Probably a range of T & t combinations

Possibly ramp rates dT/dt

Pre-cure thermal soak


Cure Effects


5/7/2012

7

5/7/2012

Adhesive &Type

Viscosity(Pa.s)

Potlife Glasstransition

VolumeResistivity

cm

Shear strength(MPa)

Curingtime

(min.)

CuringTemp.

(C)

ICA 1(1 comp.)

65 3 days 80 2.10-4 11 10-15 130-150

ICA 2(2 comp.)

25-35 3-4 days 85 1-4.10-4 10 30 130-140

ICA 3(2 comp.)

50 2 days 50 2-4.10-5 8 90 130-140

ICA 4(1 comp.)

160-200 4 days 80 5.10-3 5 20-30 120-130

ICA 5(1 comp.)

310-350 4 days 80 1.10-3 4 20-30 120-130

ICA 6(2 comp.)

150-200 8-12 hr 75 2-5.10-4 14 10-15 170-180

Compare intrinsically conducting polymers, e.g. Polyaniline: ρ = 105 (intrinsic) to 10-2 .cm (doped)

Compare Ag: ρAg = 1.6 x 10-6 .cm; ρICA = 12.5 to 312.5 x ρAg

Properties Commonly used isotropic conductive adhesives(Jagt, Chapter 11, Conductive adhesives for Electronics Packaging)


Glass Transition

Temperature

5/7/2012

Compare CTEs:Si = 4-5 ppm/°CAg = 20 ppm/°CEpoxy = 54 ppm/°CFR4 = 36 ppm/°C


5/7/2012

8

Low Resistance Measurement

5/7/2012

Wheatstone bridgeJames E. Morris OSU Seminar15

Electrical Test: Four-terminal measurement


Kelvin probe

5/7/2012

9

5/7/2012

Three terminal measurement for pad contact resistance


5/7/2012

(a) Current Crowding(b) EquallyDistributed Current

Current crowding & contact shorting

(a)

(b)

(c)

(d)

4-point measurement using ICA track across proto-board current lines (a) Sense lines short ICA (b) thinned contacts (c) trimmed contacts (d) Surface point contacts

(a)

(b)


5/7/2012

10

ICA Resistance: Percolation

5/7/2012

Polymer Metal

Percolation threshold

Insulating

conducting


Experimental Percolation


5/7/2012

11

Flakes: surface/vol ratio

5/7/2012

Percolation Threshold Comparison (Sancaktar )

Sphere/Powder Flakes

21

5/7/2012

Bi-modal: Flakes + powder

e.g.

10μm diameter flakes

(<1μm thick)

plus

1-3μm diameterpowder

(spheres)


5/7/2012

12

ICA Resistance

5/7/2012

Percolation Particle μm particles 10’s nm e- mfp Negligible size effects

Inter-particle And pad contact

Experimental (Modeling)

ECA ResistanceProtective

of Corrosion

Bulk Resistance

Contact

Percolation PathPossible Penetration of Oxide


Monte Carlo Percolation Models

5/7/2012 (Smilauer)James E. Morris OSU Seminar24

5/7/2012

13

5/7/2012

Size effect in percolation

←p=0.5↓p=0.7


Particle Contact Resistance

Constriction resistance: R = /2a Current crowding Circular contact Diameter 2a

5/7/2012

TEM of contact between nano-particles c. 7nm

diameter


5/7/2012

14

Polymer/Insulating Oxide

5/7/2012

Thermionic emission (with Schottky effect)

Electron tunneling (with image effect)

Poole-Frenkel emission

Fowler-Nordheim tunneling

TCR 0 & field effects for various mechanisms None of these observed


AgO/Ag2O Degenerate Semiconductor

5/7/2012

Intrinsic Extrinsic Degenerate

Silver oxide (degenerate semiconductor) Also 1/a2


5/7/2012

15

Carbon on flake surface appears after cure:

Changes in peak intensities with temperature during cure

Miragliotta, Benson, Phillips, & Emerson: MRS ’01, 2002 Materials

5/7/2012

RHS: 744 cm-1 carboxylate peakLHS: 1600cm-1 C peak

(a) uncured (b) cured

29

TCR Results

5/7/2012

TCRICATCRAg

RICA RAg

Contact R negligible

Resistivities for various temperatures.

(a) Brass stenciled CT-5047-02 thermoset sample (TCR-0.0039/ºC).

(b) (b) CSM-933-65-1 screen printed thermoplastic sample {TCR=0.0038/ºC}


5/7/2012

16

5/7/2012

AC Effects: Sub-Threshold

31

R vs f

L vs 1/f

Frequency Effects: Skin Effect Tunnel Gaps

Low frequency: Solder resistance < ICA resistanceHigh frequency: Solder resistance > ICA resistance

(because areas converge and ρAg << ρsolder

5/7/2012

0.0000001

0.000001

0.00001

0.0001

0.001

0.01

0.1

1

10

100

1000

0 1 2 3 4 5 6

lg(f^0.5)

Imp

ed

an

ce/m

Oh

m

RICA=RDC+R/80 RSolder=RDC+R/80 wL_ICA - ICA I path wL_Solder wL_ICA full area


Skin Effect

SolderICA

ICA

Solder

5/7/2012

17

Nanotechnologies, e.g. Add CNTs

Electrically conductive fillers with CNT

(Courtesy Prof. Zhaonian Cheng, SMIT Center, Shanghai University)


Negligible effect

Nanoparticles in Isotropic Conductive Adhesives (ICAs) (Stefan Kotthaus)

Percolation limits of different ICA filler particles5/7/2012James E. Morris OSU Seminar34

5/7/2012

18

Nanoparticle Sintering

35

[Wong et al, ECTC’06]

Raut, Bhagat, Fichthorn, Nanostruct. Mater. 10(5) 1998, 837-851

Layering Effect

Constable et al (Adhesives’98)

36

Z-axis Size Effect Expt

Resistivity as a function of the thickness.

Resistivity

0.00E+00

1.00E-03

2.00E-03

3.00E-03

4.00E-03

5.00E-03

6.00E-03

7.00E-03

8.00E-03

9.00E-03

0 0.5 1 1.5 2

h/mm

cm

5/7/2012

19

Adhesion

5/7/2012

(Detert, Herzog, Wolter, Zerna (TU-Dresden))

Danfoss


Shear test result with plasma treatment

0

20

40

60

80

100

120

140

Shear strength (%)

M. Tr. Ox. - Ni. Tr. Ox. Tr. Ox. + Hy.Tr.

5/7/2012

Upper two columns show the standard deviation

M. Tr. - Mechanical treatment

Ox. - Ni. Tr. - Oxygen Nitrogen treatment

Ox. Tr. - Oxygen treatment

Ox. + Hy. Tr. - First oxygen treatment then hydrogen treatment

(1230psi)

38

5/7/2012

20

Adhesion: Surface Roughness(Chow, Li, Yuen: 2002 Adv Pkg Materials Sympos)

5/7/201239

Adhesive before and after vacuum process

0

50

100

150

200

250Shear strength

(%)

M. Tr. Ox. Tr. + 2hrvac. afterprinting

2hr. vac.beforeprinting

2hr. vac.after printing

ICA before vacuum process

ICA after vacuum process

Change of surface by:•removal of air and water•contraction of adhesive

100 = (1230 psi)

No Vacuum Vacuum

Average Resistivity cm1022.4 3 cm1092.2 3

Standard Deviation

cm1029.3 3 cm1053.1 3

No

Vacuum Vacuum

Average Pull Test strength

160 N 235 N

Standard Deviation

108 N 56 N

5/7/2012

21

Thermal Cure: Optimization

5/7/2012

20

40

60

80

100

120

140

160

0 50 100

Time in min

Te

mp

era

ture

in º

C

Typical specification:

Temperature T and time t

Probably a range of T & t combinations

Possibly ramp rates dT/dt

Pre-cure thermal soak


Impact Strength/Drop Test (Tong)

Adhesion tests passed

Devices fall off PWB if dropped Large devices fail; small OK

No correlation between drop test results and adhesion strength testing

Complex shear modulus Storage and loss moduli


2

tt

5/7/2012

22

Variation with inertial mass

5/7/2012

Blocks Drops till failure1 More than 120 (no failure)

2 30, 433 17, 214 7, 4,115 4, 46 3

y = 124.7e-0.6556x

R2 = 0 .9712

0

10

20

30

40

50

60

70

0 1 2 3 4 5 6 7

n u mb e r o f we ig h t b lo c ks


Multiple Drop Testing:1. Cumulative damage

2. Pre-heat effects

5/7/2012

Drop test results for PLCC68 Samples cured at 150°C

0

10

20

30

1 2 3 4 5 6

Heig ht in f eet

Variable number oftest survived.

Def inate number oftests survived.

Drop test results for PLCC68 cured at 150°C w ith pre-heating and post-

cooling.

0102030

1 2 3 4 5 5.6

Heig ht in f eet

Variable number oftests survived

Definate number oftests survived


5/7/2012

23

Contact R 85°C/85% RH: Au & Cu contactsGalvanic Corrosion: Dissimilar metals + H2O

5/7/2012

A on Au

A on Cu

C on Au

C on Cu

(Li & Morris)James E. Morris OSU Seminar45

ICA: Lee, Cho, & Morris, Proc. EMAP 2007Ag-ICA 85/85 performance comparison on Cu and immersion-Ag PWB contacts

5/7/2012

A1Cu2

B1Cu1

BiCu2A1Cu1/All Ag

A3Cu2

B3Cu1

B3Cu2A3Cu1/All Ag

185oC cure 185oC cure

170oC cure 170oC cure

B1Cu1

B1Cu2A1Cu2A1Cu1/All Ag

A3Cu1

A4Ag1 B3Cu1

A3Cu2/B3Cu23x Ag

46

5/7/2012

24

5/7/2012

•Contact Resistance Variation

• Ratio Variation of Resistances

47

Creep & Coffin-Manson Lifetimes: Comparisons with

Solder (Rusanen)

5/7/2012

1

10

100

1 000

10 000

0.01 0.10

Non-recoverable strain range

Nf(

50) i

n nu

mbe

r of c

ycle

s

ICA (Rusanen &Lenkkeri 1999)

ICA (Rusanen 2000)

Sn63Pb37 (Doi et al.1996)

Sn63Pb37 (Dudeket al. 1997)

Sn5Pb95 (Darveaux& Banerji 1991)


5/7/2012

25

5/7/201249

0.4 0.6 0.8 1

0

20

40

60

80

[mm]

For

ce [N

]

Multiple plateaus

Primary plateau ~15N (all samples)

Parallel slopes (epoxy shear modulus)

Cure Models

5/7/2012

Reaction rate d/dt = K f(), where K = A exp-(E/RT) K = chemical rate constant, f () reactant concentration. = degree of cure R=Gas constant=8.31J/K.mole=Nk=6.023x1023/mole x 1.38x10-23J/K

n-th order model: f() = (1 - )n

Calculate degree of cure (for constant T, i.e. isothermal cure): 1st order: d/dt = K(1- ), = 1-exp(-Kt) 2nd order: d/dt = K(1- )2, = 1 – [1 + Kt]-1

nth order: d/dt = K(1- )n, = 1 – [1 + (n-1)Kt]-1/(n-1)

Rate proportional to reagent mass available n=1/2 Phase boundary reaction(area) n=1/3 Phase boundary reaction(volume) n=2/3 Nucleation sphere (volume growth)

Auto-catalyzed model: Single-step: d/dt=K f()=K m (1- )n …….. but note d/dt=0 for α=0 Double step (linear combination): d/dt=(K1+K2 m)(1- )n

Note: ONLY model with more than a single activation energy E Modified double step: d/dt=K f()=K(y1+y2 m)(1- )n where y1+y2=1


5/7/2012

26

DSC: Directly determine degree of cure αand rate of cure dα/dt


Cure Modeling:3 Adhesives

5/7/2012

A

B

C

Simple 1st order cure models work well

Manufacturer profiles → incomplete cure

Degree of Cure vs. Time

Resistance vs. degree of cure

52

5/7/2012

27

Particle Swarm Optimisation

53

Fitness

Results – Optimal Coefficient Sets

54

Model E1 A1 n m E2 A2 y1

1st order 37882 1021.64 1 ‐ ‐ ‐ ‐

2nd order 35917 630.83 2 ‐ ‐ ‐ ‐

3rd order 38364 841.82 3 ‐ ‐ ‐ ‐

Single step auto 31017 711.40 1.536 1.118 ‐ ‐ ‐

Double step auto 37039 666.04 0.767 3.958 309930 854.81 ‐

Modified double 33983 1074.40 1.326 1.115 ‐ ‐ 0.0393

5/7/2012

28

Results – Cure variation

55 Cure rate– Emerson and Cuming CE3103 conductive adhesive material

Basic (BADGE) curing reactions

5/7/2012

Etherification significant only at high T Model by 2-step auto-catalytic? For multiple reactions, e.g. 3 above

gcureondaryprimary

diffusiontionetherificaondaryprimaryeffective

TKK

KKKKK

T lowat 11

....11111

sec

sec

I

II

III


5/7/2012

29

Current (to 10A) and surface temperature

5/7/20120

50

100

150

200

250

0 50 100 150 200 250 300

temp/centigrade

resi

sta

nce

/mO

hm

HighCurrent Predicted with TCR

0

50

100

150

200

250

0 20 40 60 80 100 120

current ^2 / I^2

Surface T vs I2

R predicted by TCR

Measured RR vs Surface T

57

Ablebond 84 - 1 LM1

0

50

100

150

200

250

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15I [A]

R [m ]sample #1

sample #2

Electric Field for Fake Alignment

Adhesive “F”

Adhes

ive F

Adhes

ive C

Adh

esive

Bwithout field

during cure

before cure0.0%

50.0%

100.0%

R/R

ma

x

Configuration C

0.8

0.85

0.9

0.95

mO

hm

Electric Field

Adhesive FConfiguration A

before

during

without

Only adhesive F was used for this test.

All measured resistances were in a range of 0.7mOhm to 1.2mOhm

There was no significant drop in volume resistance of the cured samples, whether supplied with an electric field before or during cure, or not.

Previous experiments (different ICA) found significantly lower R.

Adhesive F No significant difference with and without applied field.

58

5/7/2012

30

Imprint Boards


5/7/2012

Summary: ICA Applications; processing

Cure modeling

Electrical conduction Sources of resistance; measurement; layering effects

TCR/frequency data

(Conduction modeling)

Nanoparticles (sintering) & CNTs

Reliability issues Impact resistance

Adhesion

Continuing research Cure modeling

Mechanical cycling

Drop test & crack propagation modeling


Documents

Isotropic Electrically Conductive Adhesives - TheCATweb.cecs.pdx.edu/~jmorris/ece507_microelectronics_seminar/ECE507... · Portland State University ... Percolation limits of different