A robust mesh size control technology suitable for various empirical equations for predicting solder joint reliability

The solder joint reliability remains one of the most critical reliability issues of ball grid array (BGA) type packaging. The challenge lies in bridging finite element modeling (FEM) results with empirical thermal cycling lifetime, given the inherent nonlinearity of the solder joint material and geometry.This research uses wafer-level packaging (WLP) as a research carrier to stabilize the solder joint lifetime prediction using the empirical equation. Two commonly used empirical equations are investigated: strain-based Coffin-Manson and energy density methods. A data set constituting five WLP thermal cycling experiment results was collected. Chip dimensions ranged from 4x4 to 6x6 mm2, and I/O counts between 100 and 200 solder balls. Each test's mean-time-to-failure (MTTF) lifetimes range from 318 to 1013 cycles. In this research, we developed five distinct finite element models corresponding to these tests, ensuring a geometric and material characteristics representation. The prediction capability is then studied by comparison of the empirical equation prediction with the experiment MTTF lifecycle. Our findings show a consistent theme: with a suitable mesh size control, solder joint lifetime predictions remain stable regardless of the choice of the empirical equation. Furthermore, when evaluating solder joint material properties at strain rates of 10
^-3
, 10
^-4,
and 10
^-5
(1/s), the prediction deviated with an average error margin of approximately 25 cycles or 4.2%. Furthermore, when using the Weibull-based MTTF lifetime metrics, predictions showed robust stability, provided the size of the most critical element remained under stringent control.