Enhancing Reliability of PB-Free Solder Joints
This paper reviews the physical properties of different underfills and how they affect performance through reliability testing.
DOWNLOAD
Authored By:
Brian Toleno, Tom White
Rong Zhang, and Jeff Bowin
Henkel Corporation
Irvine, CA, USA
Summary
Microelectronics devices are moving toward high-density interconnects. The use of flip chips, chip scale packages (CSP's) and ball grid arrays (BGA's) has increased over the past years to meet these needs. It has been found that CSPs and BGAs have particular benefit from the use of underfills to meet assembly reliability requirements. Typically, underfill materials for these devices differ significantly in their formulation to provide different types of protection.
Underfill materials were initially developed to provide coefficient of thermal expansion (CTE) matching for flip chip devices (and now wafer level CSPs). These materials have a high degree of filler and are typically non-reworkable. Underfill systems designed for CSP devices for hand-held electronics tend to be unfilled, reworkable, and reinforce devices for shock and drop.
In this paper the authors compare the reliability of several types of underfill materials across several sizes of area array components: 0.4mm pitch wafer level chip scale packages (WLCSP) and 0.5 mm pitch chip scale packages (CSP), 0.8mm pitch and 1.0mm pitch BGA devices. The reliability of these devices (all Pb-free), are evaluated using industry standard thermal cycling and drop test methodologies. The improvement of the reliability of these devices thorough the use of underfill systems is evaluated. The physical properties of different underfills and how those physical properties affect the performance in the different types of reliability testing and across different packages are also compared.
Conclusions
With the variety of underfill systems available in the market, choosing the correct system for the reliability required is critical. In this study, two materials who have similar physical properties were evaluated. Both have low Tg and are reworkable, but perform very differently in reliability testing.
While underfill A performs poorly in thermal cycling testing, it outperfoms underfill B in drop testing. Ideally, the material of choice would perform the well in both types of testing.
Future work includes more studies that evaluate drop testing on package-level underfill systems and investigating new systems that strive to provide good thermal cycling performance as well as provide drop test enhcancement.
Initially Published in the SMTA Proceedings
|
