Charged Particle Microscopy for Low-Dose Atomic Image Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Charged particle microscopy, particularly scanning transmission electron microscopy (STEM), faces challenges in obtaining atomic resolution images with sufficient resolution for material structure analysis due to limited electron dose budgets and the need for lengthy navigation to find interesting areas, which can result in sample damage.
Innovation Solution
The method involves using low signal-to-noise ratio (SNR) images acquired with charged particle microscopy to generate enhanced images through a trained machine learning model that predicts atom structure probability, allowing for live-assisted navigation and imaging with reduced electron dose and faster scanning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high radiation dose is used during navigation to find interesting areas, then image quality and resolution are improved, but sample damage increases
Solution Approach 1:
The system performs preliminary low-dose imaging to generate predictive maps of atom structure probability before acquiring final high-resolution images. This preliminary action identifies regions of interest and guides subsequent high-dose imaging only where necessary, preventing unnecessary sample damage during navigation and exploration phases.
Solution Approach 2:
The system applies high radiation dose selectively only to specific regions identified as containing interesting features, rather than uniformly across the entire sample. The predictive atom structure maps guide the electron beam to apply high dose only where atom columns are detected, using partial action to minimize overall sample damage while maintaining image quality for critical areas.
2Adaptability or versatility
If lengthy navigation is performed to find interesting areas, then complete exploration of sample is improved, but time consumption increases
Solution Approach 1:
The system introduces predictive atom structure maps as an intermediary representation between raw low-dose images and final high-resolution images. These maps, generated by machine learning models, serve as guides that enable rapid navigation to interesting areas without requiring lengthy manual exploration, thus reducing navigation time while maintaining comprehensive sample exploration.
Solution Approach 2:
The system replaces manual operator navigation with an automated machine learning-based predictive mapping system. The trained models automatically identify regions containing atom columns and generate navigation guides, substituting the mechanical interaction of manual beam scanning with an intelligent automated system that operates faster and more efficiently.
3Object-affected harmful factors
If low electron dose is used during imaging, then sample damage is reduced, but image signal-to-noise ratio deteriorates
Solution Approach 1:
The system creates predictive copies of atom structure information from low-dose images using machine learning models. These predictive atom structure maps serve as enhanced copies that reveal atomic features without requiring high-dose imaging, allowing low-dose acquisition while maintaining the ability to identify and analyze atomic structures through the generated predictive representations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the acquisition of enhanced images indicating atomic structure information with reduced sample damage and faster imaging, allowing for more efficient exploration of sample areas of interest.
Implementation Method 1
The electrons interact with the sample, resulting in elastically scattered electrons exiting the sample
Implementation Method 2
In a transmission imaging mode, the electrons transmitted through the sample are detected and used to form a microscopic image of the sample
Data Source
AI summary
A method of imaging a sample includes acquiring one or more first images of a region of the sample at a first imaging condition with a charged particle microscope system. The one or more first images are applied to an input of a trained machine learning model to obtain a predicted image indicating atom structure probability in the region of the sample. An enhanced image indicating atom locations in the region of the sample based on the atom structure probability in the predicted image is caused to be displayed in response to obtaining the predicted image.