2010 SMTA International Conference and Exhibition, October 24-28, 2010, Orlando, FL

Thermal Cycling Reliability of Package on Packages Assembled by One-Pass and Two-Pass Techniques

Nicholas Williard, Preeti Chauhan, Vikram Srinivas, Michael Osterman, and Michael Pecht
CALCE Electronic Products and Systems Center
University of Maryland
College Park, MD, USA

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


Package on Package (PoP) assemblies find use in applications that require high performance with increased memory density. PoP technology has gained wide acceptance in cell phones and other mobile applications. These packages can be assembled using a one-pass or two-pass assembly processes. In the one-pass technique, the processor is first positioned on the board, the memory is positioned on top of the processor and the board is 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 in reflow operation. These joined devices are then mounted on a printed circuit board in the board assembly reflow process.

To understand the reliability impact of one-pass and two-pass PoP assembly, an experimental test was conducted. Test vehicles consisted of PoP packages assembled using the one-pass technique and the two-pass technique with different solder combinations. SAC105 (98.5%Sn + 1.0%Ag + 0.5Cu) were used in the top ball grid array (BGAs) package while SAC125Ni (98.25%Sn + 1.2%Ag + 0.5%Cu+0.05%Ni) and SAC305 (96.5%Sn + 3.0%Ag + 0.5%Cu) solder spheres were used in the bottom BGA packages. Sixteen test specimens of each type were thermally cycled from -55°C to 125°C as per IPC SM 785 standard to examine thermo-mechanical fatigue durability as a function of assembly technique and solder combination. In-situ resistance monitoring was carried out during thermal cycling to determine the cycles to failure. Dye and pry analysis was performed to determine failure sites. Cycles to failure data was plotted as Weibull 2-parameter distributions and Kruskal–Wallis analysis was performed to evaluate the statistical relevance.

Complete article available to CALCE consortium members

[Home Page] [Articles Page]
Copyright © 2010 by CALCE and the University of Maryland, All Rights Reserved