Voiding control and reliability of solder joints with …...Voiding Control and Reliability of...

Voiding Control and Reliability of Solder Joints

with Backward Compatibility

Dr. Ning-Cheng Lee

Indium Corporation

1Grace Qin, Lei Nie, Francis Toth Jr, Raiyo Aspandiar, and Ian Williams (Intel ), “ASSESSING THE IMPACT ON TEMPERATURE CYCLING RELIABILITY OF HIGH LEVELS OF VOIDING IN BGA SOLDER JOINTS”, SMTAI 2012, p29-39, Orlando, Florida, October 14-18, 2012

Background

• Pb-components get less readily available due to full conversion to high volume Pb-free components

• For high reliability products, remaining with SnPb assembly result in backward compatibility process

• Reliability and voiding performance more complicate to predict. Control voiding critical in achieving high reliability.

2

Weakest link still hold for backward compatibility system

• The initial failures of both mixed & LF joints occur in the first full row of balls under the die.

• (substrate shrink more than die upon cooling)

3

Richard Coyle, Raiyo Aspandiar, Vasu Vasudevan, Steve Tisdale, Iulia Muntele, Richard Popowich, Peter Read, and Debra Fleming, (Alcatel Lucent, Intel, Sanmina-SCI), “THERMAL FATIGUE RELIABILITY AND MICROSTRUCTURAL CHARACTERIZATION OF A LARGE, HIGH DENSITY BALL GRID ARRAY WITH BACKWARD COMPATIBLE ASSEMBLY”, SMTAI 2012, p886-893, Orlando, Florida, October 14-18, 2012

Half-Pb-mixing

Full-Pb-mixing

Pb-free

TCT: 0/100C, 10 min ramp/10 min dwell

The location of maximum damage accumulation due to inelastic strain energy accumulation are the BGAs under the edge of the die (die shadow BGAs), caused by die warpageECTC2000 Chandran et al (Intel)

Higher Pb mixing higher reliability(1st failure)

4


TCT: 0/100C, 10 min ramp/10 min dwell

Pb dissolution height of bumps increase with increasing heat input

• 3% Pb

5

W. Fox, B. Gumpert, and L. Woody (Lockheed Martin), “DEVELOPMENT OF PROCESSING PARAMETERS FOR SOLDERING LEAD-FREE BALL GRID ARRAYS USING TIN-LEAD SOLDER”, SMTAI 2012, p878-885, Orlando, Florida, October 14-18, 2012

PCB

215C/50S

225C/80S225C/50S

215C/80S

SnPb

SAC

Peak temp Tp has higher impact than TAL on Pbdissolution height

6

Mudasir Ahmad, Kuo-Chuan Liu, Gnyaneshwar Ramakrishna, and Jie Xue (Cisco), ” IMPACT OF BACKWARDS COMPATIBLE ASSEMBLY ON BGA THERMOMECHANICAL RELIABILITY AND MECHANICAL SHOCK, PRE- AND POST-AGING”, SMTAI 2008, p306-321, Orlando, Florida, August 17-21, 2008

Tp205 211 217

TAL123 76

Complete Pb dissolution(at ≥ 217C)

(1) Same Tp(2)1. At same Tp, TAL effect mild

2. When Tp increase, Pb ht increase

Mechanical strength of bumps increase with increasing Pb & heat input

• The shear rate and the pull rate is 250 µm/s

• 3% Pb - the mechanical strength increased with both Tp & TAL (cause more mixing extent)

• 15% Pb - approached the strength of Sn63

• All failed at the IMC layer (interface)-suggesting caused by wetting defect.

• Sn63 strongest joint. (NCL: Due to better wetting/less interface defect, & finer IMC nodule size. Higher Pb wet better-lower surface tension, perhaps finer IMC nodule)

• NCL: Higher heat input may promote better wetting

7


Pb content controlled by stencil thickness

(OSP on PCB)

Surface tension (N/m):Sn63 – 51SAC305 - 57

High Tp cause longer life & less variation (CBGA)

• Fatigue life at ≤ 217°C is lower than that of SnPb assemblies. However, when the peak temperature is >217°C, the characteristic life improves to slightly better than SnPb.

• Slope of the fatigue life curves becomes steeper as Tp increases. (NCL: better wetting. less joint defect)

• The dependence on TAL is weak.

8

Mudasir Ahmad, Kuo-Chuan Liu, GnyaneshwarRamakrishna, and Jie Xue (Cisco), ” IMPACT OF BACKWARDS COMPATIBLE ASSEMBLY ON BGA THERMOMECHANICAL RELIABILITY AND MECHANICAL SHOCK, PRE- AND POST-AGING”, SMTAI 2008, p306-321, Orlando, Florida, August 17-21, 2008

< SnPb >SnPb

5 mil stencil thickness

(Tp ≤ 217C) (Tp > 217C)

Peak Temp vs TAL

• A longer TAL at a lower peak temperature does not equate to a higher peak temperature and shorter TAL.

• It’s more beneficial to raise the peak temperature than lengthen the TAL

9


Aging cause longer ATC life due to Pb diffusion

• Aging for 1000hrs

• Fatigue life increases slightly with aging. May be due to solid state diffusion of the SnPb from the paste into the Pb-free solder joint. (NCL: hence less weak Pb-rich region in joint often associated with voids?)

• (see next slide)

10


11

James Oliver, Margareta Nylén, Olivier Rod, Christofer

Markou, “FATIGUE PROPERTIES OF Sn3.5Ag0.7Cu

SOLDER JOINTS AND EFFECTS OF Pb-

CONTAMINATION”, SMTA International, September 22-

26, 2002, Chicago, IL

Pb concentration at grain boundary associated with void formation at TCT

Shock resistance lower at low Tp & more aging

• Shock induced failures is higher at the lower Tp values (wider variation)(NCL: poorer wetting, uneven Pbdistribution & more defects?)

• The shock test resistance reduces with aging (NCL: due to thicker IMC?)

12 Mudasir Ahmad, Kuo-Chuan Liu, Gnyaneshwar Ramakrishna, and Jie Xue (Cisco), ” IMPACT OF BACKWARDS COMPATIBLE ASSEMBLY ON BGA THERMOMECHANICAL RELIABILITY AND MECHANICAL SHOCK, PRE- AND POST-AGING”, SMTAI 2008, p306-321, Orlando, Florida, August 17-21, 2008

Higher shock resistance

100G = rank 0

Whenever a failure was detected, the testing was stopped at that G level.

Partial mixing cause inconsistency on fatigue cracks at ATC test

• With poorly mixed solder joints (Tp < 217°C), the fatigue cracks can vary significantly depending on the degree of mixing.

• With a fully mixed joint, the fatigue cracksconsistently occur at the top of the joint, near the package/solder interface (more heat history). This could explain the lower slope (higher variation) in the reliability test results on the partially mixed assemblies.

13


Pb impact on crack is indirect• There is minimal interaction

between the Pb phase (white) and the propagating fatigue crack. This provides further evidence that the low Pbcontent from mixed assembly has no appreciable influence on the fatigue cracking of the Pb-free solder.

14


Fatigue life increase with increasing Tp at >217C, correlate to Pb dissolution

• The fatigue life does not begin to increase until 217°C, and after 217°C, its rises continuously until the maximum temperature of 224°C is reached. This indicates that the higher the peak temperature (above 217°C), the higher the fatigue life.

• Fatigue life does not maximize until at least 85% Pb dissolution is achieved. Thus the minimum criterion for acceptable Pbdissolution is at least 85% dissolution

15


217C 85%Life

Desirability

Proposed Reflow Window For Complete Mixing

• The mechanical strength results followed closely to the degree of mixing results, showing similar trends in relation to the Tp and TAL variables. This indicates that degree of mixing and microstructure of the solder joint are good indicatorsof mechanical integrity for these mixed solder joints, and vice versa.

16


relationship between Tp, TAL and Pb content in achieving a strong solder joint

The dashed box in Figure 8 shows the range of reflow profile values used in this study.

Complete mixing

Profile Effect on Reliability Summary

• Higher peak temp (> 217C, up to 227C) cause higher TCT fatigue life and shock resistance (less wetting defect)

• Aging (100C, 150C/1000 hrs) cause higher TCT fatigue life (less Pb-rich phase-voids) but poorer shock resistance (thicker IMC)

• TAL impact not significant

17

Sometimes temp can be higher than target

• Certain components can be 20-30°C higher than target temp

18


SnPbBi paste can be used to bring down the temp for 100% Pb dissolution, but cause voiding

• Introduce Bi into joints reduce ball melting peak (DSC) up to 19C on small balls

• The SnPbBi ternary eutectic 52Bi-30Pb-18Sn (mp 96C) did not form in the samples tested. This is because the percentage of bismuth used in the samples was < 5% (closer to 2-3%).

• During assembly evaluation, 100% Pbdissolution was observed in the samples. However, during assembly, excessive voiding was also observed in the samples. (wide pasty range factor)

19


And, that void does hurt

• The ATC results show that the reliability of the Bismuth system (215C) is comparable to that of the lower peak temp (204C) assembly system. This was likely due to the excessive voiding

• Note:

• 204C Tp exhibit uneven Pb distribution & less wetting & coalescence, more joint defect. Bi improve Pb dissolution by lowering mp.

20


Bi (215C)

204C

224C

Excess Void Cause Sooner 1st Failure

• comparable slopes for the control and the excess void legs indicating similar failure mode.


Failure cycle decrease with increasing maximum void % and total voiding % for (a) control leg, and (b) excess voids leg.


TCT: -40 to 100 C, cycle time 60 min

4017.5

Control XS voids

Crack can happen at both sides


Joints that have voids are reaching the highest level of cracking faster than the non voided joints


At both interfaces the number of highly cracked interfaces in the excess void leg is significantly higher than the control leg

After 2400 cycles, cracks showing > 80% crack area

xs void

Voiding Behavior in Mixed Solder Alloy System

25Yan Liu, Derrick Herron, Joanna Keck, and Ning-Cheng Lee, “VOIDING BEHAVIOR IN MIXED SOLDER ALLOY SYSTEM”, SMTAI, 2012, p870-877, Orlando, Florida, October 14-18, 2012

Introduction

• Transition to Pb-free soldering is incomplete for high reliability or high temperature applications

• For those not fully converted into Pb-free, mixed system is common due to lack of some Pb-containing components

• Mixed system encountered voiding problem, particularly for BGA applications

• Miniaturization aggravate vulnerability of device toward voiding

• Unravel voiding mechanism of mixed system critical for DFR for solder joints


Experimental

• Solder Materials

– Ball Mounting Flux – Indium 446-AL water soluble flux

– Solder Spheres (25 mils diam.)

• Sn62, Sn63, SAC105, SAC305

– Solder Pastes (type 3)

• Sn62, Sn63, SAC105, SAC305


Experimental

• BGA & BGA Assembly

– Substrate

• A 1.6” x 1.6” FR4 coupon with a centered 24 x 24 grid of OSP coated Cu pads used as simulated BGA substrate & for BGA assembly. Pad size was 25 mil with a 40 mil pitch

– BGA ball mounting

• Print flux, place ball, reflow at LT (217C peak) or HT (240C peak), water clean

– BGA assembly

• Print paste, place BGA, reflow at LT or HT under N2


Experimental

• Characterization

– X-ray for voiding

– Cross-section when needed

– DSC for joint melting range

• Print paste dot on ceramic coupon, place ball, reflow with LT or HT under N2, then run DSC up to 300°C.

29

217C240C

LT HTYan Liu, Derrick Herron, Joanna Keck, and Ning-Cheng Lee, “VOIDING BEHAVIOR IN MIXED SOLDER ALLOY SYSTEM”, SMTAI, 2012, p870-877, Orlando, Florida, October 14-18, 2012

DSC Results & Cross-Section Examples

30

Yan Liu, Derrick Herron, Joanna Keck, and Ning-Cheng Lee, “VOIDING BEHAVIOR IN MIXED SOLDER ALLOY SYSTEM”, SMTAI, 2012, p870-877, Orlando, Florida, October 14-18, 2012

Low Surface Tension & Early Bottom Melt Cause Wide Span

• SnPb wider span in voiding than Pb-free

• SnPb lower in surface tension (0.51N/m) than Pb-free (0.57N/m)

– Under N2, no benefit on wetting

– But, less restrictive force on confining entrapped volatiles, thus higher voiding

• Early bottom melt cause low voiding (see next slide)

31

0

2

4

6

8

10

12

Void (%)

SnPb Exhibit A Broader Voiding Span Than Pb-Free

Max Void-LT

Max Void-HT

1. Lower surface tension allow larger void

2. Earlier bottom melt reduce voiding

SnPb Pb-free


Sequence Factor

32

B63P62, Sn62 paste (177.9°C) melt 3.1°C earlier than Sn63 solder ball (180.8°C) – cause ultra-low voiding

B62P63, highest voiding due to reversed sequence

Sn62

Sn63

Sn63

Sn63Sn62

Sn62


Sequence Factor Hold for SnPb

33

R² = 0.7961

0

4

8

12

-4 -2 0 2 4

Void (%)

Onset Gap (Paste - Ball) (°C)

Early Bottom Melt (Paste - Ball) Yield Low Voiding For SnPb System

Max Void-LT

Bottom melt first


Top melt first

Sequence Factor Hold at LT

• Furthermore, also applicable to mixed alloy system with LT reflow

34

0

4

8

12

16

Void (%)

Bottom Melt Earlier Reduce Voiding At Low Reflow Temp

Max Void-LT

Bottom melt earlier than top

Bottom not melt earlier than top

More solid top

B63 very wettable to drained P62

Less solid top

105

305


Solid Top Factor Contribute at LT for Mixed Systems

• At LT, Pb-free solder ball either not fully melted yet or barely completed melting. Solid top would not accommodate formation of voiding.

35

105, less void(solid top)

305, more void


Sequence Factor Fall Apart at HT

• When reflow at HT, early bottom melt (mixed system) more voiding than 0 gap (Pb-free). Why?

36

R² = 0.8223

0

4

8

12

16

-50 -40 -30 -20 -10 0 10

Void (%)

Onset Gap (Paste - Ball) (°C)

Early Bottom Melt Effect On Voiding Outweighed By Entrapped Volatile In Pasty

Joint At High Reflow Temp

Max Void-HT

Pb-free

Mixed

Volatiles entrapped in pasty joint, aggravate voiding


Solid Top Factor No Longer Exist at HT for Mixed System

• At LT, Pb-free solder ball either not fully melted yet or barely completed melting. Solid top would not accommodate formation of voiding.

• At HT, all solid tops are well eliminated during reflow, and voiding can be developed due to entering of volatiles into liquid top.

• Explained voiding HT > LT for mixed

37

HT

LT


Pasty Range Factor Dictate at HT

• Pb-free narrow pasty range, mixed wide pasty range at HT

• Pasty material viscous during heating, retain the volatiles easily and cause more voiding.

38

R² = 0.7123

0

4

8

12

16

0 10 20 30 40 50

Vo

id (

%)

Pasty Range (°C)

Wide Pasty Range Aggravate Voiding At High Reflow

Temp

Max Void %

For wider pasty range

joints, the entrapped

volatiles will have

greater difficulty to

escape due to a higher

viscosity

LF

Mixed


HT always shows a higher voiding than LT

- Partially due to Solid Top Factor at LT

• B105P62 & B105P63 may be affected by solid top factor (with solid top at LT, no solid top at HT)

39


HT always shows a higher voiding than LT

- Partially due to Overheating Factor

• B305P62 & B305P63

– Neither LT nor HT has solid top

– HT (overheating factor) cause more outgassing than LT, hence more voiding than LT.

40


Overheating Factor Dictate For

Mixed System!• The high R-square value indicates that overheating is one

very significant physical parameter in governing voiding

41

R² = 0.9075

0

4

8

12

16

-20 -10 0 10 20 30 40

Void (%)

Overheating (Peak Temp - Joint Liquidus, °C)

Higher Overheating Cause More Voiding For Mixed Alloys

Max Void %LT

HTSolid top


Overheating & Solid Top > Pasty Range Effect

42

• Voiding extent

– Solid top < liquid top

– Overheating extent // voiding extent (LT voiding < HT voiding)

– Pasty range - depends


0

4

8

12

16

0 20 40 60

Vo

id (

%)

Pasty Range (°C)

Wide Pasty Range Aggravate

Voiding At High Reflow Temp

Max

Void %

L

Mix

Paste More Voiding Than Flux

• BGA assembly with solder paste caused much higher voiding than with flux.

• Solder powder wet to the pad concurrently, exclusion of flux from interior becomes fairly troublesome.

• With flux only, the molten ball starts wetting to the pad at the center, followed by progressively pushing the flux outward with advancement of molten solder front.

43

0

2

4

6

8

10

SAC105 SAC305 Sn62 Sn63

Void (%)

Ball/Paste Cause More Voiding Than Ball Alone

Max Void % (B)

Max Void % (B/P)

Paste introduce much more voiding than ball alone

B/P

BYan Liu, Derrick Herron, Joanna Keck, and Ning-Cheng Lee, “VOIDING BEHAVIOR IN MIXED SOLDER ALLOY SYSTEM”, SMTAI, 2012, p870-877, Orlando, Florida, October 14-18, 2012

Point contact expanding

Concurrent multiple point wetting

Conclusion

• Mixed systems less voiding at LT than HT.

• Mixed systems had higher voiding than lead-free systems when reflowed at HT.

• SnPb systems had wider voiding span compared with mixed systems when reflowed at LT under N2.

• Voiding was found to decrease with

– increasing alloy surface tension, (N2 atm)

– earlier melt of bottom solder paste,

– top ball remain solid during reflow,

– reduced joint overheating,

– replacing solder paste with flux alone, and

– narrower pasty range


Additional Approaches to Bring Down Voiding

• Approaches to bring down voiding

– Solder paste formulation less prone to have voiding (better wetting, less volatile at solder molten temp)

– Mechanical shielding fixture, cost through-put

– Longer soaking profile

– Nitrogen reflow atmosphere

– Vacuum or pressurized


Voiding control and reliability of solder joints with …...Voiding Control and Reliability of...

Documents

Transcript of Voiding control and reliability of solder joints with …...Voiding Control and Reliability of...