Package with simulation method to predict underfill flow pattern with different dispensed condition

摘要

Currently electrical product is concerned about high function performance to meet consumer's requirement. Therefore, the package with high function feature is definitely required. The FCBGA (Flip-Chip Ball Grid Array) that provide the solution for low to high I/O, high electrical performance demand in high end memory, microprocessor applications where high frequency, high speed are required. The package offers pin counts in excess of 1800 with higher electrical performance compared to traditional BGA package. To develop the widest range of flip chip interconnect packages to meet consumer's demand is first target. In flip-chip packaging technology, the underfill encapsulation is one of the important processes to obtain a significant improvement in fatigue lifetime for the solder joints between IC chip and substrate. The advanced design of electronic devices aiming at the enhancement of the performance involves the increase of the number of solder bumps, smaller size of the IC chip and smaller gap height between IC chip and substrate. That leads to various problems caused by the flow behavior, such as voids in underfill process. In general, the product process engineer needs to take long time to do short-shot to find out well underfill flow pattern of dispensation parameter that balance flow pattern means underfill melt-front is with small difference gap between center and edge less than 10 bumps. Therefore, we want to use simulation skill to do underfill process prediction and find out good solution to provide well dispense parameter to control void issue and support product process engineer with underfill simulation result to reduce manpower resource. In this paper, we study FCBGA with die size as 22×20 mm and bump counts about more than 7000, and we compare different dispensation of three different I-pass lengths of 30%, 50%, 70% along die edge with different underfill weight (mg). We can observe underfill with different dispensation length and weight performs different underfill melt-front pattern and simulation results also correct with real underfill short-shot measurement.