Lead Free BGA AATC Reliability Round Robin Testing

Lead Free BGA AATC Reliability Round Robin Testing : UIC, Motorola and Rockwell Collins 12/00 automotive 2001

BGA Component Glass Die Encapsulant Layer • • Solder Bumps 1. 27 mm pitch, 256 I/O daisy chain 0. 030” diameter SMD pads 0. 030” diameter spheres (variable alloy) 0. 003” encapsulant 0. 016” FR 4 substrate 0. 040” glass die 2. 95 grams average weight Electroless nickel / immersion gold finish Pb Free/June 5 th 2002 2

Test Board 6 • 0. 062” thick FR 4 • Eight BGAs per board • 0. 020” (2), 0. 022” (4), & 0. 024” (2) NSMD pads • Three surface finishes: Electroless nickel / immersion gold (ENIG) Copper OSP Immersion silver (Imm. Ag) Pb Free/June 5 th 2002 3

Assembly - Stencil 5 mil thick Laser cut Stainless steel Nominal Values Note that square, not circular, apertures were used for this experiment. Pb Free/June 5 th 2002 4

Assembly - Pastes 4 pastes with compositions identical to the solder balls were used: ·Sn/Ag (96. 5, 3. 5); ·Sn/Ag/Cu (95. 5/3. 8/0. 7); ·Sn/Ag/Cu/Sb (96. 2/2. 5/0. 8/0. 5) (Castin), ·Sn/Pb (63/37) All pastes were type III or type IV No Clean formulations supplied by the same vendors as those used in the previous study. Assembly - Bake Out All BGA components were pre-baked at 125 o. C for a minimum of 8 hours prior to reflow. Pb Free/June 5 th 2002 5

Lifetime & Data Analysis Characteristic Lifetime analysis was performed with the use of Winsmith Weibull Software. Characteristic Lifetime is the number of cycles required for 63. 2% of the sample set to fail. · Lifetime was calculated for each metallurgy-pad finish combination. · PCB pad size and printed paste volume effects were not considered. · Failures not associated with in-plane solder fatigue (as determined by cross sectioning) were excluded from lifetime calculations. · Predicted lifetimes were computed using beta (slope) values from the sample set utilizing the same solder alloy and a confidence level of -95% to produce conservative estimates. · Most failures were removed from thermal chambers following the onset of events. Event detection behavior not characterized. Pb Free/June 5 th 2002 6

UIC Sn. Pb Reflow Profile Peak 219 o. C Time Above Liquidus = 44 seconds 2. 67 o. C/sec Cooling Nitrogen Atmosphere (O 2 < 30 ppm) Pb Free/June 5 th 2002 7

UIC Pb-Free Reflow Profile Peak 241 o. C Time Above Liquidus = 45 seconds 2. 55 o. C/sec Cooling Nitrogen Atmosphere (O 2 < 30 ppm) Pb Free/June 5 th 2002 8

UIC Temperature Cycle (Air to Air) 0/100 C AATC 104 maximum -4 minimum 5 minute ramp 5 minute dwell Pb Free/June 5 th 2002 9

UIC Reliability Summary* Solder Alloy Sn. Pb Sn. Ag. Cu. Sb Sn. Ag PCB Finish ENIG Cu OSP # Failures / # Samples 32/32 8/8 23/28 0/8 24/38 13/21 7/32 2/2 First Failure (cycles) 6801 7335 7598 NA 5033 9555 8928 9398 N 63 (cycles) 12715 9850 14600 >16700 15200 16500 24700 NA *Adjusted for non-fatigue failures. Test halted following 16700 cycles. Pb Free/June 5 th 2002 10

UIC Sn. Pb Data 0/100 C AATC Lowest N 63 Cu OSP PCB: First Failure @ 7335 Final Failure @ 11066 N 63 = 9850 OCCURRENCE CDF % ENIG PCB: First Failure @ 6801 Final Failure @ 16698 N 63 = 12715 99 95 90 80 70 60 50 40 30 CMP 205112 20 10 5 2 1 10000 Cycles to Failure CMP 205152 Pb Free/June 5 th 2002 11

UIC Sn. Pb Data Failure Frequency Sn. Pb on ENIG 27 Components Failure Frequency Sn. Pb on Cu OSP 8 Components Pb Free/June 5 th 2002 12

CMP 205152: First UIC Sn. Pb Failure 0/100 C AATC Failure @ 6801 Removed After Cycle 6802 6 Events Detected 4. 42 Ohm Initial Resistance 4. 78 Ohm Final Resistance ENIG PCB Outer Corner of Package Component Side Solder Fatigue Sn. Pb Solder Fatigue Pb Free/June 5 th 2002 13

CMP 205112: Final UIC Sn. Pb Failure 0/100 C AATC Failure @ 16698 Removed After Cycle 16700 2 Events Detected 5. 80 Ohm Initial Resistance 6. 38 Ohm Final Resistance ENIG PCB No Joint Isolated by 4 Point Cleaving in Outer Row Potential Cleaving of Sn. Pb Solder Joint Pb Free/June 5 th 2002 14

UIC Sn. Ag. Cu Data 0/100 C AATC *No Failures on Cu OSP Through 16700 Cycles* 20300 N 63 Prediction OCCURRENCE CDF % ENIG PCB: First Failure @ 2465 Second Failure @ 7598 N 63 = 14600 90 80 70 60 50 40 30 Cu OSP Predicted 20 WB c=-95 10 5 CMP 205048 2 1 10000 Cycles to Failure CMP 205045 (Excluded from Lifetime Analysis: Interfacial Failure) Pb Free/June 5 th 2002 15

UIC Sn. Ag. Cu Data Failure Frequency Sn. Ag. Cu on ENIG 21 Components Pb Free/June 5 th 2002 16

CMP 205045: First UIC Sn. Ag. Cu Failure 0/100 C AATC Failure @ 2465 5. 1 Ohm Initial Resistance Electrically Open at Removal ENIG PCB Outer Corner of Package Non-Solder Fatigue Failure Enlarged Image Interfacial Failure Pb Free/June 5 th 2002 17

CMP 205048: First UIC Sn. Ag. Cu Fatigue Failure 0/100 C AATC Failure @ 7598 Removed After Cycle 7633 5. 10 Ohm Initial Resistance 5. 20 Ohm at Removal ENIG PCB Outer Corner of Package Component Side Solder Fatigue Sn. Ag. Cu Solder Fatigue Pb Free/June 5 th 2002 18

UIC Sn. Ag. Cu Failure Modes CMP 205057 Sn. Ag. Cu on ENIG 11938 CTF PCB Side Fatigue Failure 5. 21 Ohm Initial Resistance 5. 99 Ohm Final Resistance CMP 205052 Sn. Ag. Cu on ENIG 10880 CTF Component Side Fatigue Failure 5. 17 Ohm Initial Resistance 20. 0 Ohm Final Resistance Pb Free/June 5 th 2002 19

UIC Sn. Ag. Cu Failure Modes CMP 205034 Sn. Ag. Cu on ENIG 16700 Cycles to Removal Electrically Good, Non-Failure CMP 205036 Sn. Ag. Cu on ENIG 14115 CTF Electrically Good Joint Cracks Visible in X-ray Pb Free/June 5 th 2002 20

UIC Sn. Ag. Cu. Sb Data 0/100 C AATC ENIG PCB: N 63 = 15200 Glass Delamination (Excluded) Interfacial Failure (Excluded) Large Void(s) (Included) Pb Free/June 5 th 2002 OCCURRENCE CDF % Cu OSP PCB: N 63 = 16500 90 80 70 60 50 40 30 20 10 5 Excluded 2 1 10000 Cycles to Failure 21

UIC Sn. Ag. Cu. Sb Data Failure Frequency Sn. Ag. Cu. Sb on ENIG 19 Components Failure Frequency Sn. Ag. Cu. Sb on Cu OSP 10 Components Pb Free/June 5 th 2002 22

Effects of Delamination is the separation of the underfill material from the substrate and/or the glass die. Visible Delamination Primarily occurs during reflow operations. Delamination affects both in-plane and out-of-plane stresses. Some degree of delamination located near the center of the package is acceptable. Delamination may be visible to the naked eye or may require C-SAM to identify. Pb Free/June 5 th 2002 CMP 205126 Sn. Ag. Cu. Sb on Cu OSP 2137 CTF Effect of Delamination 23

Effects of Delamination : Warpage Room Temperature CMP 205126 Sn. Ag. Cu. Sb on Cu OSP PCB 100 Degrees Celsius Pb Free/June 5 th 2002 24

UIC Sn. Ag. Cu. Sb Failure Modes CMP 205132 Sn. Ag. Cu. Sb on Cu OSP 3990 CTF Interfacial Failure 4. 51 Ohm Initial Resistance 4. 62 Ohm Final Resistance CMP 205013 Sn. Ag. Cu. Sb on ENIG 6105 CTF Solder Fatigue through Large Voids 5. 39 Ohm Initial Resistance 5. 70 Ohm Final Resistance Pb Free/June 5 th 2002 25

UIC Sn. Ag. Cu. Sb Failure Modes CMP 205011 Sn. Ag. Cu. Sb on ENIG 10921 CTF Typical Failure Cracks on Component and PCB Side Numerous Voids Present 5. 93 Ohm Initial Resistance 6. 02 Ohm Final Resistance CMP 205919 Sn. Ag. Cu. Sb on ENIG 16700 Cycles to Removal Electrically Good, Non-Failure Cleaving Observed in Outer Row Pb Free/June 5 th 2002 26

UIC Sn. Ag Data 0/100 C AATC Highest N 63 Cu OSP PCB: First Failure @ 1534 Final Failure @ 10204 6/8 Defective No Prediction Provided OCCURRENCE CDF % ENIG PCB: First Failure @ 8928 7 th Failure @ 15896 24702 N 63 Prediction 90 80 70 60 50 40 30 20 10 5 2 1 10000 Cycles to Failure Pb Free/June 5 th 2002 27

UIC Sn. Ag Data Failure Frequency Sn. Ag on ENIG 7 Components Failure Frequency Sn. Ag on Cu OSP 2 Components Pb Free/June 5 th 2002 28

UIC Sn. Ag Data CMP 205074 Sn. Ag on Cu OSP 1534 CTF Glass Delamination CMP 205069 Sn. Ag on Cu OSP 4575 CTF Interfacial Failure Electrically Open Upon Failure Pb Free/June 5 th 2002 29

UIC Sn. Ag Data CMP 205070 Sn. Ag on Cu OSP 10204 CTF PCB Side Solder Fatigue 5. 10 Ohm Initial Resistance 40. 56 Ohm Final Resistance Cu Thick Cu-Sn Intermetallics Non-reflow Voids Pb Free/June 5 th 2002 30

Motorola Sn. Pb Reflow Profile Peak 221 o. C 1. 3 o. C/sec Cooling TAL = ~115 seconds Nitrogen Atmosphere Pb Free/June 5 th 2002 31

Motorola Pb-Free Reflow Profile Peak 242 o. C 1. 2 o. C/sec Cooling TAL = 46 seconds Nitrogen Atmosphere Pb Free/June 5 th 2002 32

Motorola Temperature Cycle (Air to Air) 0/100 C AATC 107 maximum -7 minimum 10 minute ramp 5 minute dwell Pb Free/June 5 th 2002 33

Motorola Reliability Summary* Solder Alloy Sn. Pb Sn. Ag. Cu. Sb Sn. Ag PCB Finish ENIG Cu OSP # Failures / # Samples 14/14 8/8 14/16 7/8 1/8 4/16 1/8 First Failure (cycles) 5885 7631 7356 8051 7713 11261 7812 13521 N 63 (cycles) 8495 8730 10290 10870 9295 >13707 15318 >13707 *Adjusted for non-fatigue failures. Based on data accumulated through 13707 cycles. Pb Free/June 5 th 2002 34

Motorola Sn. Pb Data 0/100 C AATC Lowest N 63 Cu OSP PCB: First Failure @ 7631 Final Failure @ 9179 N 63 = 8730 OCCURRENCE CDF % ENIG PCB: First Failure @ 2959 Final Failure @ 9914 N 63 = 8495 99 95 90 80 70 60 50 40 30 20 CMP 205656 10 5 2 1 10000 Excluded (Interfacial Failures) Pb Free/June 5 th 2002 Cycles to Failure 35

Motorola Sn. Pb Data Failure Frequency Sn. Pb on ENIG 7 Components Failure Frequency Sn. Pb on Cu OSP 4 Components Pb Free/June 5 th 2002 36

Motorola Sn. Pb Failure Modes CMP 205225 Sn. Pb on ENIG 2959 CTF Interfacial Failure CMP 205219 Sn. Pb on ENIG 3020 CTF Interfacial Failure Pb Free/June 5 th 2002 37

Motorola Sn. Pb Failure Modes CMP 205656 Sn. Pb on ENIG 5885 CTF First Solder Fatigue Failure Pb Free/June 5 th 2002 38

Motorola Sn. Ag. Cu Data 0/100 C AATC Cu OSP PCB: First Failure @ 8051 7 th Failure @ 12200 N 63 = 10870 OCCURRENCE CDF % ENIG PCB: First Failure @ 7356 14 th Failure @ 11188 N 63 = 10290 90 80 70 60 50 40 30 20 CMP 205625 10 CMP 205563 5 2 1 10000 Cycles to Failure Pb Free/June 5 th 2002 39

Motorola Sn. Ag. Cu Data Failure Frequency Sn. Ag. Cu on ENIG 7 Components Pb Free/June 5 th 2002 40

Motorola Sn. Ag. Cu Failure Modes CMP 205563 Sn. Ag. Cu on ENIG 7356 CTF First Sn. Ag. Cu Solder Fatigue Failure Pb Free/June 5 th 2002 41

Motorola Sn. Ag. Cu Failure Modes Non-reflow Voids Copper Silver Needle CMP 205625 First Sn. Ag. Cu on Cu OSP PCB Failure 8051 CTF Pb Free/June 5 th 2002 42

Motorola Sn. Ag. Cu. Sb Data 0/100 C AATC Cu OSP PCB: First Failure @ 11261 14245 N 63 Prediction OCCURRENCE CDF % ENIG PCB: First Failure @ 7713 7 th Failure @ 11062 N 63 = 9295 90 80 70 60 50 40 30 20 WB c=-95 CMP 205455 Cu OSP Predicted 10 5 2 1 10000 Cycles to Failure Pb Free/June 5 th 2002 43

Motorola Sn. Ag. Cu. Sb Failure Modes Cleaving? CMP 205455 Sn. Ag. Cu. Sb on ENIG 7713 CTF First Sn. Ag. Cu. Sb Solder Fatigue Failure Pb Free/June 5 th 2002 44

Motorola Sn. Ag Data 0/100 C AATC Cu OSP PCB: First Failure @ 13521 16000 N 63 Prediction OCCURRENCE CDF % ENIG PCB: First Failure @ 6945 4 th Failure @ 12858 15318 N 63 Prediction 90 80 70 60 50 40 30 ENIG Predicted 20 10 Cu OSP Predicted WB c=-95 CMP 205426 5 2 1 10000 Cycles to Failure Pb Free/June 5 th 2002 45

Motorola Sn. Ag Data Failure Frequency Sn. Ag on ENIG 4 Components Pb Free/June 5 th 2002 46

Motorola Sn. Ag Failure Modes CMP 205426 Sn. Ag on ENIG 6945 CTF First Sn. Ag Solder Fatigue Failure Pb Free/June 5 th 2002 47

Rockwell Data -55 to 125 o. C Air to Air Thermal Cycle All Components were Assembled to Immersion Silver PCBs All Components were Assembled with Sn. Ag. Cu (95. 5/3. 8/0. 7) Solder Paste: Sphere Alloy Composition % on. 022" Pad Sn Pb Ag Cu 63 Sn 37 Pb 65. 270 34. 419 0. 263 0. 049 96. 5 Sn 3. 5 Ag 96. 432 0. 000 3. 519 0. 049 95. 5 Sn 3. 8 Ag. 7 Cu 95. 500 0. 000 3. 800 0. 700 Estimated alloy % based on 90% paste transfer and 50% solidified paste volume. Pb Free/June 5 th 2002 48

Rockwell Reliability Summary* Solder Alloy PCB Finish # Failures / # Samples First Failure (cycles) N 63 (cycles) Sn. Pb Imm. Ag 16/32 413 1990 Sn. Ag. Cu Imm. Ag 13/32 367 2860 Sn. Ag Imm. Ag 9/31 467 2667 *Adjusted for non-fatigue failures. Based on data accumulated through 2000 cycles. ENIG components Pb Free/June 5 th 2002 49

Rockwell Reliability Data -55/125 C AATC Imm. Ag PCB: Sn. Ag N 63 = 2667 Sn. Ag. Cu N 63 = 2860 OCCURRENCE CDF % Sn. Pb N 63 = 1990 ENIG Components on Imm. Ag PCBs 90 80 70 60 50 40 30 20 10 CMP 205205 Excluded 5 2 1 1000 Cycles to Failure Pb Free/June 5 th 2002 50

Rockwell Sn. Pb Failure Modes Copper CMP 205163 Sn. Pb on Imm. Ag 413 CTF Bulk Solder Failure CMP 205254 Sn. Pb on Imm. Ag 426 CTF Bulk Solder Failure Nickel Tin Lead Copper Tin Silver Lead CMP 205239 Sn. Pb on Imm. Ag 1917 CTF Bulk Solder Failure Pb Free/June 5 th 2002 51

Rockwell Sn. Ag Failure Modes CMP 205205 Sn. Ag on Imm. Ag 184 CTF Interfacial Failure Copper Nickel Tin Silver Pb Free/June 5 th 2002 52

Rockwell Sn. Ag Failure Modes CMP 205193 Sn. Ag on Imm. Ag 467 CTF Bulk Solder Failure CMP 205196 Sn. Ag on Imm. Ag 1929 CTF Bulk Solder Failure Pb Free/June 5 th 2002 53

Rockwell Sn. Ag Failure Modes Non-reflow Voids Cu-Sn CMP 205192 Sn. Ag on Imm. Ag 1528 CTF Bulk Solder Failure Pb Free/June 5 th 2002 54

Rockwell Sn. Ag. Cu Failure Modes CMP 205277 Sn. Ag. Cu on Imm. Ag 367 CTF Intermetallic (Cu-Sn) to Bulk Solder Failure Tin Silver Copper Pb Free/June 5 th 2002 55

Rockwell Sn. Ag. Cu Failure Modes Cu-Sn CMP 205278 Sn. Ag. Cu on Imm. Ag 478 CTF Intermetallic (Cu-Sn) to Bulk Solder Failure Pb Free/June 5 th 2002 56

Rockwell Sn. Ag. Cu Failure Modes CMP 205271 Sn. Ag. Cu on Imm. Ag 1528 CTF CMP 205272 Sn. Ag. Cu on Imm. Ag 1528 CTF Pb Free/June 5 th 2002 57

Rockwell Sn. Ag. Cu Failure Modes Silver Needle CMP 205740 Sn. Ag. Cu on Imm. Ag Cu Component 1530 CTF Pb Free/June 5 th 2002 58

Conclusions *Results may be highly influenced by such factors as the reflow profile and test cycle conditions. * UIC and Motorola Sn. Pb: Solder joints consistently fractured through the bulk solder material near the mask defined component attachment pad. A few interfacial failures occurred at the component electroless nickel / solder interface. Minor cleaving was observed on one sample following 16700 AATCs. Cleaving did not lead to failure. Lowest characteristic life among alloys tested. Pb Free/June 5 th 2002 59

Conclusions UIC and Motorola Lead Free: All alloys demonstrated superior fatigue resistance compared to Sn. Pb. Sn. Ag most reliable. Bulk solder fatigue occurred near the component attachment pad and PCB attachment pad. No relation between PCB pad size and crack location was evident. Fatigue cracks are often very fine, with multiple crack fronts influenced by the local microstructure of the solder. A completely cracked joint can be electrically continuous. May explain “glitch”. Pb Free/June 5 th 2002 60

Conclusions UIC and Motorola Lead Free: Several interfacial failures were observed at electroless nickel / solder interfaces. Severe cleaving was observed on electrically good samples. Cleaving may occur before in-plane crack propagation and potentially lead to solder bridging. The CASTIN alloy contained large voids which resulted in early failures. These early failures may occur before Sn. Pb failures. Data indicates higher reliability on copper OSP PCBs. Thick intermetallics, often greater than. 25 mil may form at the copper / solder interface. 5 lead free components on ENIG with atypical event pattern. Evidence of “glitch”? Pb Free/June 5 th 2002 61

Conclusions Rockwell Sn. Pb with Sn. Ag. Cu paste: Solder joints consistently fractured through the bulk solder material near the component or PCB attachment pad. Lowest characteristic life among alloys tested. Rockwell Lead Free: Both alloys demonstrated superior fatigue resistance compared to Sn. Pb. Prediction: Sn. Ag. Cu most reliable. Sn. Ag reliability is comparable. Two failure mechanisms observed: 1. ) Fracture through the bulk solder material near the component or PCB attachment pad. Similar to fine cracks observed in other tests. 2. ) Failure between the solder material and Cu-Sn IMC near the PCB pad. Unique among the other tests. Non-reflow voiding prevalent at failure location. Pb Free/June 5 th 2002 62

Continuing Work Partial repeat of experiment with commercially available components: 1. 27 mm pitch, 256 I/O daisy chain • 0. 0177” diameter SMD pads • 0. 030” diameter spheres (variable alloy) • Sn. Pb Reliability: N 63 = 7700 - 8200 Cycles • Picture by Amkor Technology Completed 1256 0 to 100 o. C cycles without failure. Pb Free/June 5 th 2002 63

Continuing Work Partial repeat of first experiment with NSMD components: Glass Die Encapsulant Layer Solder Bumps AATC Test Matrix Pb Free/June 5 th 2002 64