2/3/2025 | 3:30 PM - 5:30 PM | Regency B

Deep learning-enabled X-ray imaging for rapid assessment of thermal fatigue in SAC solder interconnects

Author(s)

Eshan Ganju | Purdue University
Nikhilesh Chawla | Purdue University

Abstract

 

To ensure electronics perform reliably over time, it's crucial to understand how they behave under cyclic thermal stress. This study focuses on Sn-Ag-Cu (SAC) solder interconnects, a key component in many devices, and how they are affected by temperature fluctuations. Using advanced X-ray Computed Tomography (XCT) and deep learning algorithms, we created a high-speed method to analyze 3D images of these interconnects and identify defects. We examined multiple SAC Ball Grid Arrays (BGA) boards, both before and after exposing them to repeated heating and cooling cycles, to track how damage emerges and evolves over time. Our deep learning model not only allowed us to rapidly analyze the images, but also revealed how heat-induced defects interact with pre-existing flaws in the solder. The imaging data was also supplemented with electrical resistance of the interconnects, linking the observed damage to changes in resistance of the solder.