Charged Particle Microscopy With AI Atom Mapping at Low Dose
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Charged particle microscopy techniques face challenges in obtaining atomic resolution images with sufficient signal-to-noise ratio (SNR) due to limited electron dose budgets, leading to sample damage and poor visibility of atom structures, especially when navigating to find interesting areas.
Innovation Solution
A system and method using a trained machine learning model, such as a convolutional neural network, to predict atom structure probability from low SNR images, enabling live-assisted imaging with reduced charged particle beam dose and fast scanning, and superimposing atom positions on enhanced images for real-time navigation and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high radiation dose is used to obtain atomic resolution images, then image quality and signal-to-noise ratio are improved, but sample damage and alteration increase
Solution Approach 1:
The system uses low-dose imaging by acquiring images with a limited electron dose budget that is lower than conventional doses required for atomic resolution. The machine learning model then processes these low-dose images to reconstruct atomic structures, achieving high measurement precision without the excessive radiation that would cause sample damage.
2Productivity
If fast scanning is used to reduce sample exposure time, then productivity is improved, but image quality and signal-to-noise ratio deteriorate
Solution Approach 1:
The system performs preliminary action by using fast scanning to quickly navigate and identify regions of interest with low-dose images. Once regions are identified, the system can then focus higher doses on specific areas or use machine learning to enhance the fast-scanned images, achieving both high productivity and acceptable image quality for navigation purposes.
3Object-affected harmful factors
If low radiation dose is used to preserve sample integrity, then sample alteration is reduced, but visibility of atom structures and signal-to-noise ratio worsen
Solution Approach 1:
The machine learning model acts as an intermediary between the low-dose images and the final atomic structure visualization. The model processes the noisy low-dose images, enhancing atomic structure visibility and enabling accurate interpretation without requiring high radiation doses that would alter the sample.
4Measurement precision
If navigation is performed to find interesting areas, then analysis accuracy is improved, but time consumption and sample exposure increase
Solution Approach 1:
The system applies partial action by using low-dose fast scanning to navigate and identify regions of interest without performing exhaustive high-dose scanning of the entire sample. This approach achieves sufficient analysis accuracy for navigation while minimizing time consumption and sample exposure, with the understanding that final detailed analysis can be performed on selected regions only.
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
Enables the generation of enhanced images that clearly indicate atomic structure, allowing for accurate sample navigation and reduced sample alteration, while maintaining low radiation exposure.
Implementation Method 1
A system and method using a trained machine learning model, such as a convolutional neural network, to predict atom structure probability from low SNR images
Implementation Method 2
The electrons interact with the sample, resulting in elastically scattered electrons exiting the sample
Data Source
Figure 1
Figure 2
Figure 3A~3C
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.