I. Introduction
In 1998, Brandenburg and Yeh reported electromigration failure in flip chip eutectic SnPb solder joints[1]. Their research shown that electromigration in flip chip joint would become a serious reliability issue in flip chip technology. Since then, the subject has been included in the international technology roadmap for semiconductors for study. In last ten years, K N Tu et al [2]–[6]. studied electromigration in eutectic SnPb flip chip joint systemically and many interesting findings of electromigration were shown in their works. SnAgCu solder has been the most widely used in electronic industry instead of eutectic SnPb after the transition to lead-free solder in electronics commercial products[7]. While solder joint technology is mature, Pb-free solder technology is not, hence its reliability must be proven. So the reliability of electromigration in SnAgCu solder joint are more noticeable than in SnPb, while the studies of electromigration in SnAgCu solder are less than in SnPb [8]–[10]. The pancake-type void across the entire cathode contact interface to the Si Chip is emphasized as the main failure mode. The pancake-type void is main failure position in electromigration, but the formation of pancake-type void at the interface of Si side can been avoided by using Cu column bumps or increasing the thickness of Cu UBM as advised by K N Tu[11]. Except for the main failure mode, pancake-type void, whether are there others potential failure positions in the pairs of solder joints by electromigration? Our studies will show three potential failure modes of electromigration in the pairs of solder joints. According to IPC standard or U.S. Military Standard (MIL-STD), all theses three kinds of damages would be determined failure in failure-analysis procedures for microcircuits.