TEM Sample Preparation Automation: Matching the Level to the Workflow 

In most labs, TEM sample capacity depends on the availability of experienced operators who can reliably navigate to the region of interest, perform lift-out, precisely execute thinning and endpointing, recover from interruptions, and train new users to do the same. 

Automation can reduce that constraint, but not every sample requires the same level of automation. 

For relatively uniform samples, repeatable process execution and controlled thinning may be sufficient. For site-specific semiconductor analysis, the harder problem is knowing exactly where to stop, especially at the most advanced nodes. 

The Thermo Scientific Helios FIB-SEM portfolio reflects these different requirements. Thermo Scientific AutoTEM Software workflows range from guided and automated preparation to ML-based fiducial endpointing and AI-based device endpointing. The right level depends on the sample, the targeting requirement, and where expert intervention is limiting capacity. 

Different samples require different levels of TEM preparation automation

For thin films, epitaxial layers, bulk materials, and other relatively uniform samples, advanced endpointing will not be the primary requirement. If the region of interest is broad, the larger opportunity can be batch automating preparation to improve repeatability and utilization. 

The Helios 5 DualBeam with AutoTEM 5 Software remains relevant for these established, repeatable workflows. Helios PFIBs with AutoTEM 5 Software also have an important role when Ga-free preparation is a priority. 

As TEM preparation moves toward smaller device structures, thinner lamellae, and more demanding site-specific targets, the automation requirement changes. The challenge is no longer just executing the workflow reliably. Automation increasingly needs information about where the lamella is relative to the target. 

That is the distinction between the three AutoTEM 6 Software levels that can be configured with our most advanced Ga-based systems, the Helios 6 HD FIB-SEM and the Helios 6 HX FIB-SEM. 

Helios 6 HD

Read further to learn more about the use-cases and automation levels that best suit a variety of needs and applications. 

AutoTEM 6 Software Essential: Automate preparation and thinning for volume

The Helios 6 HD FIB-SEM with AutoTEM 6 Software Essential is designed for high-volume, automated lift-out and blind thinning. 

This is a strong fit for workflows where structures are sufficiently large or repeating that controlled dose-based thinning can reliably bring the lamella close to the desired final condition. That can include legacy-node and some advanced-node devices where throughput and consistency matter more than sophisticated site-specific endpoint recognition. 

AutoTEM 6 Software coordinates preparation, milling, lift-out, site tracking, grid management, and thinning. High-accuracy stage movement preserves registration during tilts and rotations, while automated preventive maintenance and condition-specific beam alignments support predictable operation across longer runs. 

During pre-thinning, the sample is automatically aligned and oriented before controlled milling of the front and rear faces. The documented workflow uses low-kV steps and/or manual polishing to achieve final thickness. This dose-based approach standardizes the handoff to final polishing and reduces the amount of expert time required at the end of each sample. 

For a lab trying to increase capacity, that’s the key distinction. If thinning consumes operator time but the target does not require active endpoint recognition, the objective is to automate more of the repeatable process, not necessarily to add more targeting intelligence. 

Current Helios 6 HD FIB-SEM productivity measurements show the broader impact of this approach: greater than 15% more samples per hour, a 7% increase in system availability, and up to a 30% increase in productivity through the combined effects of throughput, availability, and automated pre-thinning. 

AutoTEM 6 Software Enhanced: When thinning becomes a targeting problem

For smaller, site-specific device structures, controlled dose alone may not provide enough precision. 

This is typically where experienced FIB users remain most involved. As the lamella approaches the feature of interest, the consequences of a targeting error increase. Maintaining final thickness is only part of the problem; the device also needs to remain correctly centered between both faces. 

AutoTEM 6 Software Enhanced is designed for this more demanding workflow, as it adds ML-based fiducial endpointing to high-volume automated preparation. 

With line indicated termination (LIT), engineered marks establish independent termination coordinates for the lamella faces. AutoTEM Software measures their relationship to the target, then uses SEM observations during thinning to determine when each face should stop. Therefore, the endpoint shifts from a dose-only estimate toward image-based feedback near the final approach. 

That difference is important for advanced-node failure analysis: AutoTEM 6 Software Essential automates thinning, but AutoTEM 6 Software Enhanced adds feedback to determine where thinning should stop. 

In the demonstrated, LIT-assisted workflow, automated pre-thinning achieved less than 15 nm thickness variation, target centering within 10 nm, and greater than 95% process robustness. 

For labs reluctant to trust automation near a small device target, this is an important differentiation. The answer is not necessarily more supervision, but an automation strategy that can guarantee results. 

AutoTEM 6 Software Professional: Extend automation to the device

At the most advanced nodes, AutoTEM 6 Software Professional on the Helios 6 HX FIB-SEM moves beyond fiducial-based endpointing to AI-based device endpointing. This workflow supports not only high-volume capacity but ultimate precision with AI-based device endpointing. 

The Helios 6 HX FIB-SEM combines automated navigation and site positioning with machine learning-assisted endpointing and the ability to train and refine automation models for specific device types. With AutoTEM 6 Software Professional, the system supports fully automated TEM sample preparation of sub-20 nm lamellae with production-grade accuracy. On a selected group of devices, the workflow has demonstrated sub-nanometer targeting accuracy for TEM lamellae below 20 nm. 

The same platform reports greater than 95% robustness for automated weld-to-grid and more than 90% robustness for fully automated thinning to target thickness on selected devices. 

Comparing TEM sample preparation automation levels

To summarize the progression: 

These are workflow-fit examples rather than rigid node boundaries. Target geometry, materials, required lamella thickness, and endpoint requirements all influence the appropriate approach. 

Scaling TEM preparation capacity without scaling operator dependence

Reliable TEM preparation will always depend on process expertise. What changes with automation is where that expertise is required. 

A highly experienced operator does not necessarily need to supervise every navigation step, alignment, lift-out, grid movement, or thinning pass. Depending on the application, automation can carry the sample much further through the workflow and leave only a limited final intervention—or, in the most advanced workflows, automate device-targeted thinning itself. 

That can help laboratories increase sample output while preserving expert capacity for difficult samples, process development, troubleshooting, and training. It can also give new operators a more standardized path to reliable results. 

The practical question is therefore not simply how much automation is available. 

It’s how much information the system needs to prepare your particular samples correctly—and how much of that workflow still needs to depend on your most experienced operator. 

Helios 6 HD with AutoTEM 6 Enhanced prepares high-quality GAA TEM lamellae.
The Helios 6 HD FIB-SEM with AutoTEM 6 Software Enhanced used to deliver high-quality GAA lamellae.

Resources 

Ondřej Bačo

Leave a Reply

Your email address will not be published. Required fields are marked *