Authored By:
Burton Carpenter, Mollie Benson, Andrew Mawer
NXP Semiconductors, Inc.
TX, USA
Summary
BGA components used in high reliability automotive microprocessor applications have historically been 0.8mm pitch or larger. Recently, emerging market requirements have pushed BGA pitches down to 0.65mm. However, industry expectations of solder-joint reliability remain constant; customers expect cyclical thermal fatigue lifetimes to be the same as, or in some cases more robust than prior generations of packages. In addition, automotive Tier 1 electronics manufactures often include drop test and other board mounted testing criteria to mitigate the risk of module level failure during SMT, module or final vehicle assembly.
The paper reports the board level reliability of a 0.65 pitch, 10x10mm MAPBGA (molded array plastic BGA) package used as an automotive radar microprocessor. Thermal fatigue life was assessed per IPC-9701A using AATS (air-to-air thermal shock) between -40ºC and +125ºC, while drop testing followed the JEDEC JESD22-B111A at 1500G’s.
Six package configurations were studied: each combination of two package BGA pad surface finishes (Ni/Au and OSP [organic solderability preservative]) and three package solder ball alloys (SnAg, SAC405, and SAC-Bi [a Bi containing SAC derivative]). PCB variables included pad diameter, and the effect of interstitial through vias in the BGA footprint.
The solder-joint lifetimes from the AATS tests were observed to depend on the package variables, with Ni/Au BGA pads performing better than the OSP pads. The PCB variables were not significant. By contrast, the OSP BGA pad finish had better drop test performance.
Conclusions
Thermal cycle fatigue and JEDEC drop test solder-joint reliability were studied on a 0.65mm pitch 141MAPBGA package through in situ monitor of packages mounted on PCBs. The primary conclusions are:
1) The POR package (Ni/Au pad w/ SnAg ball) meets the goal of first failure greater than 1500 cycles using a 0.325mm/0.330mm SRO/Ball Diameter.
2) Packages with BGA pad OSP surface finish were not ready for production, due to solder non-wet, and the inability to meet the temperature cycle requirement.
3) The SAC-Bi alloy was not ready for implementation because the temperature cycle first failure was below target, and the low beta indicate high variability in expected performance.
4) A follow-up study is in progress with improved an improved OSP sphere attach process.
Initially Published in the SMTA Proceedings
|
