AGING INDUCED VOID FORMATION AND ITS IMPACT ON THE MECHANICAL INTEGRITY OF SOLDER TO BARE Cu OR ENIG BGA SURFACE FINISHES
Authors: B. Vaccaro, J. Osenbach, S. Gasper-Gerlach, G. Lib Company: Agere Systems Date Published: 9/24/2006
Abstract: The mechanical performance of Pb-Sn solder on both ENIG and bare Cu eutectic SOP BGA pad surface finishes as a function of exposure time at 150C was evaluated. In addition, the impact of aging on the solder joint microstructures was investigated. High purity, 99.99%, fully annealed Cu pins were reflow soldered to a subset of the BGA balls on FCBGA substrates that were manufactured with either ENIG or bare Cu-SOP surface finish. The Cu pins were attached to assess the influence of PCB Cu pad metallization on the mechanical stability of the solder joint and also provided a surface for mechanical clamping to enable pull testing. Pull tests were performed at strain rates corresponding to load speeds between 0.0005mm/s and 5.0mm/s. Both the strength and failure mode were recorded. Ball shear strength data was also collected on balls with and without Cu pins on the same samples at a strain rate corresponding to a shear tool speed of 0.4mm/sec. The ball shear and pin pull tests were done prior to aging and after 10, 20, 40, and 80 days of aging at 150C. Scanning electron microscopy and EDS microanalysis were completed on cross-sections taken from the devices at each test point to verify the failure modes and investigate the microstructure for each type of assembly. For all samples, the shear and pull strengths decreased with aging time. Brittle solder/pad IMC failures were first observed after 10 days of aging for ENIG samples with Cu pins. No such failures were observed in ENIG joints not exposed to Cu. In contrast, brittle failures at the substrate solder/pad IMC interface were first observed after 80 days of aging for bare Cu-SOP samples both with and without Cu pins attached. No occurrences of brittle failure were observed at the Cu pin/solder interface. Brittle solder/pad IMC failures for bare Cu-SOP samples occurred as a result of micro-void formation within the IMC layer that formed near the substrate Cu pad. Significant differences in the microstructure and void density at the solder/IMC/Cu interfaces on the BGA pad side of the joint as compared to the Cu pin side of the joint were observed for the bare Cu- SOP samples. Both shear and pin pull test methods revealed the brittle solder/pad IMC failure mode at the same aging times for both pad finishes when Cu pins were attached to simulate PCB assembly. The results indicate that high speed testing may not be required to observe brittle solder joint fracture induced by long term high temperature aging.