Evaluation of POP Assembly Under Load

Vikram Srinivas, Michael Osterman
Center for Advance Life Cycle Engineering
Department of Mechanical Engineering, University of Maryland USA

Robert Farrell
Benchmark Electronics, Inc., Nashua, NH USA

Package on Packages (PoP) find use in applications that require high performance with increased memory density. One of the greatest benefits of PoP technology is the elimination of the expensive and challenging task of routing high-speed memory lines from under the processor chip out to memory chip in
separate packages. Instead, the memory sits on top of the processor and the connections are automatically made during assembly. For this reason PoP technology has gained wide acceptance in cell phones and other mobile applications. PoP technology can be assembled using one-pass and two-pass assembly processes. In the one-pass technique the processor is first mounted to the board, the memory is mounted to the processor and the finished board is then run through the reflow oven in a single pass. The two-pass technique has an intermediate step in which the memory is first mounted onto the processor. Then, these two parts are placed in a carrier tray and reflowed. These joined devices are then mounted on the circuit board and the finished board is reflowed a second time.

The two-pass technique has a distinct advantage in that the PoPs can be checked for defects before final assembly using a non-destructive test (such as X-Ray) and hence one would expect higher yield. For this study, identical test vehicles were assembled with eight PoP packages assembled with SAC105 and SAC125 solder for the bottom BGA and top BGA respectively. One-pass technique and two-pass technique were used to assemble two test vehicles each. These test vehicles were evaluated under mechanical torsion loading to establish if method of assembly used has any impact on the mechanical fatigue durability. This was followed by failure analysis to determine failure sites. Time to failure data was plotted as Weibull 2-parameter distributions and ANOVA analysis was performed. No statistically significant difference was found in the reliability of the packages assembled using the two different techniques

Mechanical torsion loading was used to compare the different assembly techniques for package on package technology and no statistically significant difference was observed. The bottom nets were always found to fail first in PoP assemblies. No statistically significant difference in durability was observed for PoP technology with increased Ag content in the solder interconnects of bottom package under mechanical torsion. No difference in trend of failure distribution was observed between solder types or assembly techniques for package on packages. Pad cratering was identified to be a potential reliability concern. Use of under fill is recommended to redistribute stress uniformly and avoid trace failures.