Issue link: https://iconnect007.uberflip.com/i/253291
40 SMT Magazine • February 2014 TeSTING INTerMeTaLLIC FraGILITy ON eNIG uPON aDDITION OF LIMITLeSS Cu continues FEATUrE for the test board were through the (Cu,Ni)6Sn5 intermetallic at the pad surface. Further research needs to be conducted of this test method in order to develop a pro- cess that produces variability with solder al- loy and ball attach process as seen in the CBP testing. In CBP testing peak load to failure distri- butions showed a marked difference between the various lead-free morphologies formed during ball attached and subsequent reflow with copper powder. All but one lead-free case produced a large standard deviation in pull testing. Only the SnAg Long ball attach sample resulted in a narrow peak load distri - bution similar to the SnPb cases as shown in Figure 19. This suggests that the SnAg Long sample has more consistent intermetallic properties from ball to ball. The fracture sur- faces were compared and the SnAg long ball attach process seemed to have slightly more intermetallic remaining at the pad surface than the other lead-free samples as shown in Figure 20. Although brittle failures modes are not favorable in electronics packaging, hav- ing consistent result does improve an engi- neer's ability to predict characteristic life of the device in mechanical accelerated life test- ing. The SnPb samples, by comparison, also failed in a narrow distribution of peak load to failure. This has driven the average up and ar- guably reduces the possibility of infant mor- talities in mechanical stress conditions. Fail- ure modes were similar for the control sample however the Intel 845 device exhibited a dif- ferent fracture characteristic with more inter- metallic remaining at the electroless Ni pad surface following ball removal as shown in Figure 21. Conclusions Intermetallic morphology affects the peak load to failure distribution in CBP testing and a possible narrow distribution can be created for SnAg using a long ball attach profile. This con- dition is favorable for reliability predictions and mitigation of infant mortalities for lead-free product. Figure 19: CbP load to failure distribution for vari- ous morphologies. Figure 21: SEM SnPb fracture morphology comparison (A) SnPb Short (b) SnPb long and (C) Intel 845. Figure 20: Cross-sectional SEM comparison of CbP fracture surfaces (A) SnAg Short (b) SAC304 Short (C) SnAg long (D) SAC304 long.