Abstract
Microwave-induced plasma (MIP) decapsulation of microchips traditionally requires constant manual monitoring and verification to ensure an optimized process, a time-consuming task for researchers. This paper explores the feasibility of utilising an AI-driven approach to automate both the operation and verification of the MIP process. By leveraging machine learning trained on a dedicated image database, the AI successfully managed the etching cycles, automatically halting the process upon full die exposure. Furthermore, the system demonstrated the potential to identify the decapsulation states of “foreign” microchip not included in the original training set. These results suggest a significant opportunity to streamline industrial workflows through AI and expand automated failure analysis to more advanced microchip packaging.
S.Y. Sim, Nanyang Technological University
Q. Liu, Nanyang Technological University
Z. Qiu, Nanyang Technological University
Prof. C.L. Gan, Nanyang Technological University
FIGURES
Visual highlights from the research and analysis presented in this publication.
Fig. 5 – Screenshots of STM chip sample gradually decapsulated from (A) to (E), exposed its die shown within the red rectangle via MIP process automatically through AI interaction script.
