Challenges of Package on Package Devices

Bob Willis
ASKbobwillis.com
Chelmsford, Essex, United Kingdom

David Bernard
Nordson DAGE
Fremont, CA, USA

Ball Grid Array devices, BGAs, are widely used in a vast range of products including consumer, telecommunications and office based systems. As an area array device of solder joints, it provides high packing density with a relatively easy introduction cycle. However, over the last couple of years engineers have started to experiment, and in some cases implement, stacked packages, of the type often called Package on Package, or POP.

In simple terms, POP devices are the stacking of components, one on top of the other, either during the original component manufacture or during printed board assembly. Such packages, allow substantially enhanced functionality but within the same footprint of a single BGA. POP packaging may include direct soldering, wire bonding or conductive adhesives for device to device interconnection.

The main industry problems associated with POP technology are open joints, warping of the two levels of component substrate and, of course, issues with the underlying printed circuit board (PCB). Reworking these stacked components, or just the top mounted part, can be challenging and when you inspect them using 2-D x-ray inspection the data can become difficult to interpret because of the multiple levels of ball interconnection and wire-bonding that may occur within the package.

This paper will outline the process associated with soldering stacked packages using dip flux and dip solder paste that are specifically designed to overcome the incidence of package warp. Based on the process issues involved, inspection results will be presented to better illustrate the challenges in implementing POP, or stacked packages, into production.

Making measurements during the x-ray inspection of POP devices may not only provide a method to quickly confirm the quality of the solder joints within the various package layers but also highlight if there is any warpage that might affect devices where variation in stand-off height can affect performance. A trade off may need to be made in making many measurements over the whole sample against the speed of measurement throughput. Taking additional measurements at the center of the device, as well as at the corner, may also be beneficial.