Reliability evaluation on low k wafer level packages

Abstract

Wafer Level Package (WLP) technology has seen tremendous advances in recent years and is rapidly being adopted at the 65 nm Low-K silicon node. For a true WLP, the package size is same as the die (silicon) size and the package is usually mounted directly on to the Printed Circuit Board (PCB). Board level reliability (BLR) is a bigger challenge on WLPs than the package level due to a larger CTE mismatch and difference in stiffness between silicon and the PCB. The BLR performance of the devices with Low-K dielectric silicon becomes even more challenging due to their fragile nature and lower mechanical strength. A post fab re-distribution layer (RDL) with polymer stack up provides a stress buffer resulting in an improved board level reliability performance. Drop shock (DS) and temperature cycling test (TCT) are the most commonly run tests in the industry to gauge the BLR performance of WLPs. While a superior drop performance is required for devices targeting mobile handset applications, achieving acceptable TCT performance on WLPs can become challenging at times. BLR performance of WLP is sensitive to design features such as die size, die aspect ratio, ball pattern and ball density etc. In this paper, 65nm WLPs with a post fab Cu RDL have been studied for package and board level reliability. Standard JEDEC conditions are applied during the reliability testing. Here, we present a detailed reliability evaluation on multiple WLP sizes and varying ball patterns. Die size ranging from 10 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> to 25 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> were studied along with variation in design features such as die aspect ratio and the ball density (fully populated and de-populated ball pattern). All test vehicles used the aforementioned 65nm fab node.