Isotropic Electrically Conductive Adhesives

Transcription

1 Isotropic Electrically Conductive Adhesives James E. Morris Department of Electrical & Computer Engineering Portland State University P.O.Box 751, Portland OR , USA ICA: Flip Chip Interconnect ICA: Isotropic Conductive Adhesives Powder Flakes (Ag) Current path Polymer Contents Introduction Applications Processing Resistance & measurement Nanotechnology Layering effect Adhesion Drop test Galvanic corrosion Cure models Miscellaneous 2 ICA: SMT Interconnect 1

2 ICA applications: Solder replacement in SMT & flip chip SMT joint showing Ag flakes Bosch Diesel Fuel Injection Pump Flip-chip joint showing Ag flakes: Top: Si chip bump Bottom: PWB pad 3 ICA Smart Cards & RF ID (IZM-Berlin) 4 2

3 ICA Microvias: Ag, Cu, & LMP (Low Melting Point) Alloy [Das et al (Endicott Interconnect), ECTC 06] 5 Failure modes (Sintered NanoParticle) ICA Die Attach [Bai et al (VaTech), HDP 05] 6 3

4 ACA: Anisotropic Conductive Adhesive (& ACF) Adhesive IC Bump Glass substrate Electrode Conductive particles 7 Ni particle Coated polymer particle Technology Drivers 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 8 4

5 Processing e.g. Flip-Chip Assembly (IZM, Berlin) 9 Cure Process Char Temperature T g gelt g Sol/Gel Rubber Complete Cure Gelation Vitrification Liquid Gel Rubber Vitrification Devitrification Gel Glass Sol/Gel Glass T c >T T c <T T g0 Sol Glass Log Time 10 5

6 Thermal Cure: Optimization Typical specification: Temperature T and time t Probably a range of T & t combinations Possibly ramp rates dt/dt Tem perature in ºC Pre-cure thermal soak Ti m e i n m i n 11 Cure Effects 12 6

7 Properties Commonly used isotropic conductive adhesives (Jagt, Chapter 11, Conductive adhesives for Electronics Packaging) Adhesive & Type Viscosity (Pa.s) Potlife Glass transition Volume Resistivity cm Shear strength (MPa) Curing time (min.) Curing Temp. ( C) ICA 1 (1 comp.) ICA 2 (2 comp.) ICA 3 (2 comp.) ICA 4 (1 comp.) ICA 5 (1 comp.) ICA 6 (2 comp.) 65 3 days days days days days hr Compare intrinsically conducting polymers, e.g. Polyaniline: ρ = 10 5 (intrinsic) to 10-2.cm (doped) Compare Ag: ρ Ag = 1.6 x 10-6.cm; ρ ICA = 12.5 to x ρ Ag Glass Transition Temperature Compare CTEs: Si = 4-5 ppm/ C Ag = 20 ppm/ C Epoxy = 54 ppm/ C FR4 = 36 ppm/ C 14 7

8 Low Resistance Measurement 15 Wheatstone bridge Electrical Test: Four-terminal measurement Kelvin probe 16 8

9 Three terminal measurement for pad contact resistance 17 Current crowding & contact shorting (a) (a) (b) (c) (b) (d) 18 (a) Current Crowding (b) EquallyDistributed Current 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 9

10 ICA Resistance: Percolation Insulating Percolation threshold conducting 19 Polymer Metal Experimental Percolation 20 10

11 Flakes: surface/vol ratio Percolation Threshold Comparison (Sancaktar ) Sphere/Powder Flakes 21 Bi-modal: Flakes + powder e.g. 10μm diameter flakes (<1μm thick) plus 1-3μm diameter powder (spheres) 22 11

12 ECA Resistance Protective of Corrosion ICA Resistance 23 Percolation Path Possible Penetration of Oxide Contact Bulk Resistance Percolation Particle μm particles 10 s nm e - mfp Negligible size effects Inter-particle And pad contact Experimental (Modeling) Monte Carlo Percolation Models 24 (Smilauer) 12

13 Size effect in percolation p=0.5 p= Particle Contact Resistance Constriction resistance: R = /2a Current crowding Circular contact Diameter 2a TEM of contact between nano-particles c. 7nm diameter 26 13

14 Polymer/Insulating Oxide 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 27 AgO/Ag 2 O Degenerate Semiconductor Intrinsic Extrinsic Degenerate 28 Silver oxide (degenerate semiconductor) Also 1/a 2 14

15 Carbon on flake surface appears after cure: Changes in peak intensities with temperature during cure (a) uncured (b) cured Miragliotta, Benson, Phillips, & Emerson: MRS 01, 2002 Materials LHS: 1600cm-1 C peak RHS: 744 cm-1 carboxylate peak 29 TCR Results TCR ICA TCR Ag R ICA R Ag Contact R negligible Resistivities for various temperatures. (a) Brass stenciled CT thermoset sample (TCR /ºC). (b) (b) CSM screen printed thermoplastic sample {TCR=0.0038/ºC} 30 15

16 Frequency Effects: Tunnel Gaps Skin Effect R vs f AC Effects: Sub-Threshold L vs 1/ f 31 Low frequency: Solder resistance < ICA resistance High frequency: Solder resistance > ICA resistance (because areas converge and ρ Ag << ρ solder 1000 Im p edance/m Ohm Skin Effect ICA Solder Solder ICA lg(f^0.5) RICA=RDC+R/80 RSolder=RDC+R/80 wl_ica - ICA I path wl_solder wl_ica full area 32 16

17 Nanotechnologies, e.g. Add CNTs Electrically conductive fillers with CNT Negligible effect (Courtesy Prof. Zhaonian Cheng, SMIT Center, Shanghai University) 33 Nanoparticles in Isotropic Conductive Adhesives (ICAs) (Stefan Kotthaus) 34 Percolation limits of different ICA filler particles 17

18 Nanoparticle Sintering [Wong et al, ECTC 06] 35 Raut, Bhagat, Fichthorn, Nanostruct. Mater. 10(5) 1998, Layering Effect Z-axis Size Effect Expt Constable et al (Adhesives 98) Resistivity 9.00E E E E-03 cm 5.00E E E E E E h/mm Resistivity as a function of the thickness. 18

19 Danfoss (Detert, Herzog, Wolter, Zerna (TU-Dresden)) Adhesion 37 Shear test result with plasma treatment (1230psi) Shear strength (%) M. Tr. Ox. - Ni. Tr. Ox. Tr. Ox. + Hy. Upper two columns show the standard deviation Tr. M. Tr. - Mechanical treatment Ox. - Ni. Tr. - Oxygen Nitrogen treatment Ox. Tr. - Oxygen treatment 38 Ox. + Hy. Tr. - First oxygen treatment then hydrogen treatment 19

20 Adhesion: Surface Roughness (Chow, Li, Yuen: 2002 Adv Pkg Materials Sympos) 39 Adhesive before and after vacuum process ICA before vacuum process Shear strength (%) = (1230 psi) ICA after vacuum process 0 M. Tr. Ox. Tr. + 2hr vac. after printing 2hr. vac. before printing 2hr. vac. after printing No Vacuum Vacuum Change of surface by: removal of air and water contraction of adhesive Average Resistivity Standard Deviation cm cm cm cm Average Pull Test strength Standard Deviation No Vacuum Vacuum 160 N 235 N 108 N 56 N 20

21 Thermal Cure: Optimization Typical specification: Temperature T and time t Probably a range of T & t combinations Possibly ramp rates dt/dt Tem perature in ºC Pre-cure thermal soak Ti m e i n m i n 41 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 t t

22 Variation with inertial mass Blocks Drops till failure 1 More than 120 (no failure) 2 30, , , 4,11 5 4, y = 124.7e x R 2 = number of weight blocks Drop test results for PLCC68 Samples cured at 150 C Multiple Drop Testing: 1. Cumulative damage 2. Pre-heat effects Height in feet Variable number of test survived. Definate number of tests survived. Drop test results for PLCC68 cured at 150 C with pre-heating and postcooling Height in feet Variable number of tests survived Definate number of tests survived 44 22

23 Contact R 85 C/85% RH: Au & Cu contacts Galvanic Corrosion: Dissimilar metals + H 2 O A on Au C on Au A on Cu C on Cu (Li & Morris) 45 ICA: Lee, Cho, & Morris, Proc. EMAP 2007 Ag-ICA 85/85 performance comparison on Cu and immersion-ag PWB contacts 185 o C cure 185 o C cure A1Cu2 B1Cu1 B1Cu1 BiCu2 A1Cu1/All Ag B1Cu2 A1Cu2 A1Cu1/All Ag 170 o C cure A3Cu2 170 o C cure A3Cu1 B3Cu1 B3Cu2 A3Cu1/All Ag A4Ag1 B3Cu1 A3Cu2/B3Cu2 3x Ag 46 23

24 Contact Resistance Variation Ratio Variation of Resistances 47 Creep & Coffin- Manson Lifetimes: Comparisons with Solder (Rusanen) Nf(50) in number of cycles Non-recoverable strain range ICA (Rusanen & Lenkkeri 1999) ICA (Rusanen 2000) Sn63Pb37 (Doi et al. 1996) Sn63Pb37 (Dudek et al. 1997) Sn5Pb95 (Darveaux & Banerji 1991) 48 24

25 Force [N] [mm] 1 Multiple plateaus Primary plateau ~15N (all samples) Parallel slopes (epoxy shear modulus) 49 Cure Models 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=6.023x10 23 /mole x 1.38x10-23 J/K n-th order model: f( ) = (1 - ) n Calculate degree of cure (for constant T, i.e. isothermal cure): 1 st order: d /dt = K(1- ), = 1-exp(-Kt) 2 nd order: d /dt = K(1- )2, = 1 [1 + Kt] -1 n th 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=(k 1 +K 2 m )(1- ) n Note: ONLY model with more than a single activation energy E Modified double step: d /dt=k f( )=K(y 1 +y 2 m )(1- ) n where y 1 +y 2 =

26 DSC: Directly determine degree of cure α and rate of cure dα/dt 51 A Cure Modeling: 3 Adhesives Resistance vs. degree of cure B 52 Degree of Cure vs. Time C Simple 1st order cure models work well Manufacturer profiles incomplete cure 26

27 Particle Swarm Optimisation Fitness 53 Results Optimal Coefficient Sets Model E 1 A 1 n m E 2 A 2 y 1 1 st order nd order rd order Single step auto Double step auto Modified double

28 Results Cure variation 55 Cure rate Emerson and Cuming CE3103 conductive adhesive material I Basic (BADGE) curing reactions Etherification significant only at high T Model by 2-step auto-catalytic? For multiple reactions, e.g. 3 above II III 56 1 K effective 1 K primary 1 K primary 1 K secondary 1 K secondary K 1 etherification at low T cure 1 K diffusion Tg... 28

29 R [m ] Ablebond 84-1 LM1 sample #1 sample # Surface T vs I I [A] Current (to 10A) and surface temperature resistance/mohm current^2 / I^2 R vs Surface T Measured R R predicted by TCR temp/centigrade HighCurrent Predicted with TCR Electric Field for Fake Alignment Adhesive F 0.95 Adhesive F Configuration A Only adhesive F was used for this test. All measured resistances were in a range of 0.7mOhm to 1.2mOhm mohm Electric Field before during without Adhesive F Configuration C 58 R/Rmax 100.0% 50.0% 0.0% Adhesive F Adhesive C Adhesive B No significant difference with and without applied field. before cure during cure without field 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. 29

30 Imprint Boards 59 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 60 30