Neural Network Aberration Correction for Electron Microscopes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current aberration measurement methods for electron microscopes are inadequate for achieving optimal imaging effects, especially at the atomic scale, due to reliance on non-crystal areas and instability of the objective lens current, leading to inaccurate aberration calculations and adjustments.
Innovation Solution
A method involving training a neural network using electron microscope simulation images to recognize and correct aberrations in experimental samples, allowing for accurate determination and elimination of aberrations without the need for sample replacement or movement, using a network training unit, image recognition unit, and aberration determination unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If aberration measurement is made using non-crystal areas, then the measurement process can be completed, but the imaging effect at atomic scale cannot be optimized
Solution Approach 1:
The patent creates simulation images that copy the characteristics of actual electron microscope images with known aberration parameters. These simulated images serve as training data for the neural network, allowing the system to learn the relationship between image features and aberration values without requiring physical measurement of non-crystal areas.
Solution Approach 2:
The patent replaces the traditional mechanical/optical measurement system (using non-crystal areas and formula-based calculations) with an artificial intelligence system. The neural network directly analyzes crystal area images to determine aberration parameters, substituting the complex mechanical measurement process with computational analysis.
2Ease of operation
If sample replacement or movement is performed to obtain non-crystal areas, then aberration measurement can be conducted, but time and operational complexity increase
Solution Approach 1:
The system enables the crystal area itself to serve as the measurement target. Instead of requiring external non-crystal areas or sample replacement, the neural network extracts aberration information directly from the crystal structure images, allowing the sample to provide its own measurement data.
Solution Approach 2:
The neural network model serves multiple functions: it not only determines aberration parameters but also identifies crystal structures and analyzes image features. This multi-functional approach eliminates the need for separate measurement procedures and sample handling steps.
3Reliability
If traditional formula-based calculation is used for aberration parameters, then the measurement process is established, but the stability of objective lens current variations cannot be compensated
Solution Approach 1:
The neural network provides a feedback mechanism that continuously learns from simulated images with known aberration values. The system compares predicted aberration parameters with actual values from simulation data, adjusting its internal parameters to minimize errors and compensate for lens current variations.
Solution Approach 2:
The patent changes the approach from fixed formula-based calculations to adaptive parameter learning. The neural network dynamically adjusts its interpretation of image features based on the relationship between image characteristics and aberration parameters learned from simulation data, allowing compensation for varying lens conditions.
Data Source
AI summary
The present disclosure provides a determination method, an elimination method and an apparatus for an electron microscope aberration. The determination method comprises: training a neural network for image recognition using a plurality of electron microscope simulation images to obtain an electron microscope image recognition model; recognizing an electron microscope image of an experimental sample using the electron microscope image recognition model to obtain the electron microscope simulation image corresponding to the electron microscope image of the experimental sample; and obtaining the corresponding set aberration as an imaging aberration of the electron microscope image of the experimental sample according to the electron microscope simulation image corresponding to the electron microscope image of the experimental sample. Through the above solution, an aberration value of an electron microscope can be obtained using a lattice image of the experimental sample, thereby improving an imaging effect of the electron microscope.