This paper presents an automated focused ion beam (FIB) workflow that revolutionizes FIB-TEM sample preparation for semiconductor analysis, particularly for NAND and DRAM devices. This system integrates with Helios 5 HX and Helios 6 HX platforms to achieve consistent, hands-free sample preparation down to ~10 nm thickness. Our approach transforms traditional manual processes into automated workflows through comprehensive best-known method (BKM) templates, custom recipes, and proprietary code development in AutoTEM and iFast software architectures. The system features advanced machine vision algorithms and custom hardware components that enable precise sample coordination, allowing for both vertical and perpendicular TEM sample preparation from 3D structures at any wafer depth. Implementation results demonstrate significant improvements over manual processes, including thousands of hours in workforce savings, reduced material waste, faster throughput, improved data consistency, and minimized human error in DRAM and NAND structural analysis.

The goal of this project is to develop an automated algorithm of die filter to minimize the number of dies for PFA engineers to review and to confirm before PFA submission. This can reduce the workload of PFA engineers so that the PFA engineers can focus their investigations on few wafers and dies and thus the PFA job submission can be more efficient and accurate. The number, type, and location of dies for PFA sampling can be automatically given by following the rules set by PFA engineers.