Implementation of FIB Automation Methodologies for TEM Applications in Memory Devices.
Implementation of FIB Automation Methodologies for TEM Applications in Memory Devices.
Thursday, October 31, 2024
Indigo Ballroom (Hilton San Diego Bayfront)
Summary:
An improved and automated FIB workflow is developed, validated, and implemented into the current FIB-TEM workflow to reduce time-to-data, achieve analysis data consistency, targeting zero human-intervention, covering most semiconductor companies use-cases, especially in NAND and DRAM devices. This research took steps toward closing the gaps and reaching the unmanned lab environment’s final target. All manual use cases were evaluated in detail and converted, reinvented, and optimized into new automation workflows, from ROI definition to the final thinning ~10 nm thickness in Helios 5 HX™ and Helios 6 HX™. All best-known method (BKM) procedures and parameters were coded into FIB process automation templates, recipes, and in-house code development per use case to create the new functions in the automation software architect, also known as AutoTEM™ and iFast™, with in-house hardware parts development to control sample’s coordination system. It provided vertical and perpendicular cut-face-direction TEM samples from 3D structures in wafers at any depth (Z-axis), thanks to the nanometer-scale precision performance of the advanced machine vision (MV) algorithm for various TEM analysis use cases. Thus, the result delivered thousands of hours of workforce savings, reduced resource materials waste, improved the throughput time to consistent data, and reduced human error compared to traditional manual FIB use cases in DRAM and NAND structure TEM analysis.
An improved and automated FIB workflow is developed, validated, and implemented into the current FIB-TEM workflow to reduce time-to-data, achieve analysis data consistency, targeting zero human-intervention, covering most semiconductor companies use-cases, especially in NAND and DRAM devices. This research took steps toward closing the gaps and reaching the unmanned lab environment’s final target. All manual use cases were evaluated in detail and converted, reinvented, and optimized into new automation workflows, from ROI definition to the final thinning ~10 nm thickness in Helios 5 HX™ and Helios 6 HX™. All best-known method (BKM) procedures and parameters were coded into FIB process automation templates, recipes, and in-house code development per use case to create the new functions in the automation software architect, also known as AutoTEM™ and iFast™, with in-house hardware parts development to control sample’s coordination system. It provided vertical and perpendicular cut-face-direction TEM samples from 3D structures in wafers at any depth (Z-axis), thanks to the nanometer-scale precision performance of the advanced machine vision (MV) algorithm for various TEM analysis use cases. Thus, the result delivered thousands of hours of workforce savings, reduced resource materials waste, improved the throughput time to consistent data, and reduced human error compared to traditional manual FIB use cases in DRAM and NAND structure TEM analysis.