| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Novel Strategies for Rapid Acquisition and Processing of Large Datasets from Advanced Characterization Techniques
|
| Presentation Title |
Deep Learning-Assisted Study of 3D Damage Evolution in Semiconductor Packages under Thermal Cycling Using X-ray Microcomputed Tomography |
| Author(s) |
Eshan Ganju, Yaw Obeng, William Harris, Nikhilesh Chawla |
| On-Site Speaker (Planned) |
Eshan Ganju |
| Abstract Scope |
The reliability of semiconductor packages under cyclic thermal stresses is critical for ensuring long-term device performance. Within semiconductor packages, Sn-Ag-Cu (SAC) solder interconnects are one of the most widely used Pb-free solder interconnects. In this study, we have used X-ray Microcomputed Tomography data to develop a deep learning-assisted approach for the high throughput 3D characterization of defects in SAC solder interconnects under cyclic thermal stresses. Multiple PCB boards, consisting of grids of SAC solder interconnects, were scanned using XCT, both before and after thermal cycling, to capture the evolution of thermal-cycling induced defects. By leveraging deep neural networks, we have achieved high throughput characterization of defects in the solder interconnects while also revealing how thermal cycling-induced defects interact with preexisting pores within the interconnects. The 3D XCT data was further supplemented by resistivity measurements of the interconnects to understand how thermally induced defects impact device performance. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Characterization, Machine Learning, Other |