Experimental damage analysis and numerical reliability modeling of lead-free ball-grid-array second level interconnects

The reliability of BGA packages is primarily affected by thermo-mechanical deformations in the assembly as a consequence of the CTE mismatch of the materials. Typical failure modes are low-cycle fatigue of second level solder joints or delamination of pads from the PCB. In this work, a fine-pitch BGA package was subjected to a variety of cyclic temperature tests. Micrographs were performed at specific cycle numbers to analyze the local damage. Two different printed circuit boards and two different lead free solder pastes were used in the tests. The ball material was SAC405 alloy and the board metallization was chemical tin. The pad finishes on the package side were varied to investigate their influence. The dominating failure mode observed in the micrographs was crack initiation and propagation in the solder. Further failure modes are discussed and assessed with respect to their relevance for assembly reliability. For all specimen configurations the effective crack length of each solder ball was measured to specify the damage. The crack propagation data were determined by linear regression of the relative damage accumulation versus cycle number. The mean fatigue life was calculated from these data and correlated with local numerical damage parameters obtained by simulations. These relations were compared with fatigue-life models from literature and also used to derive selected parameters of damage laws based on the Coffin-Manson-model. Furthermore, a comparison of the fatigue life for the different test configurations is given. The cyclic accumulated creep strains in the critical solder volume ranged between 0.2 and 1.6% and the crack propagation rate ranged between 0.008 and 0.29 μm per cycle.