Flip chip electromigration: Impact of test conditions in product life predictions

Robustness of flip chip joint structure is proven through extensive studies with focus on key failure mechanisms valid for most applications. Solder fatigue is well characterized and widely covered in the literature. Underfilling overcomes the impact of local CTE mismatch between the device and the substrate, providing performance levels well beyond requirements. Over the years, solder joint electromigration has become a key design parameter. Although it is not a concern for most applications, the increasing demand for reduced voltage, and the continued need for increased power on a chip, result in increased cur-rent density per solder joint. In such a case, designers need to take into account maximum allowable current per solder joint for a given operating temperature, IO passivation opening, and a lifetime requirement. The reverse is also true that designers may seek to know minimum passivation opening for a given cur-rent density requirement. This requires better understanding of the numerical models driving product life predictions, as the single test condition, mostly performed as part of a qualification with zero failure requirement, fail to address the need. In this study, fundamentals of flip chip electromigration is reviewed for thin film UBM (Al/NiV/Cu) with SnPb eutectic, SnPbCu, and SnAgCu solder joint structures. Impact of key test conditions (i.e., temperature and current density) on the product life prediction is illustrated through the construction of governing equation, using empirical data collected at different test regimes. Significance of secondary failure mechanisms is also demonstrated in terms of test conditions and substrate pad selection. Obviously, tests conducted at elevated temperatures (150degreesC and above) with high current densities would provide results in short time. Since it involves mix mode failures (UBM consumption and electromigration), it would also yield most conservative predictions for operating conditions. Tests conducted at comparatively lower temperatures (105degreesC - 125degreesC) validate this claim and provide accurate predictions for most field conditions.