Automated Navigation Framwork for High-Density 3D VNAND using Structural Periodicity-Based Address Mapping and Computer Vision

Thursday, October 8, 2026: 11:20 AM
Ms. Minji Seo , Samsung Electornics, Hwaseong-si, Gyeonggi-do, Korea, Republic of (South)
Mr. Donguk Ko , Samsung Electornics, Hwaseong-si, Gyeonggi-do, Korea, Republic of (South)
Ms. Sooyoun Lim , Samsung Electornics, Hwaseong-si, Gyeonggi-do, Korea, Republic of (South)
Ms. Hyeongki Kim , Samsung Electornics, Hwaseong-si, Gyeonggi-do, Korea, Republic of (South)
Mr. Sangmin Han , Samsung Electornics, Hwaseong-si, Gyeonggi-do, Korea, Republic of (South)
Mr. Sungho Lee , Samsung Electornics, Hwaseong-si, Gyeonggi-do, Korea, Republic of (South)
Mr. Youngjin Cho , Thermo Fisher Scientific, Yongin-si, Gyeonggi-do, Korea, Republic of (South)
Mr. Inchang choi , Thermo Fisher Scientific, Yongin-si, Gyeonggi-do, Korea, Republic of (South)
Dr. Christopher H. Kang , Thermo Fisher Scientific, Yongin-si, Gyeonggi-do, Korea, Republic of (South)

Summary:

Advancements in 3D VNAND technology have increased cell density, making it more difficult to accurately identify failure addresses. Manual counting of thousands of channel holes is now impractical due to increased risk of human error and excessive time demands. To address these challenges, we introduce an automated navigation system that converts logical cell addresses into physical landmarks and integrates with Thermo Fisher Scientific’s iFAST automation platform. The system locates target positions by tracking repeating device structures using computer vision. Testing on Samsung Electronics’ next generation VNAND devices at their developmental stage shows that this method achieves a detection success rate above 97% and reduces analysis time by over 80%. The system provides reliable, continuous automated failure localization and is scalable to other semiconductor devices with repeating structures.