3D Atomic Structure Reconstruction for Missing Wedge Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining three-dimensional surface atomic structures of nanomaterials face challenges due to geometric limitations in electron microscopy, leading to missing wedge defects and inaccurate data, particularly in measuring atomic-scale defects and strain, which affects the precision of nanomaterial properties and mass production.
Innovation Solution
A data generating apparatus and method that uses a deep learning filter to correct and enhance three-dimensional atomic structure data by generating atomic models, simulating tomography across a wider angle range, and reconstructing electron microscope images to overcome missing wedge defects and noise, thereby improving the accuracy and reliability of surface atomic structure determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional electron microscopy methods are used for three-dimensional measurement, then measurement can be performed with current equipment, but measurement precision is insufficient due to missing wedge defects and geometric limitations
Solution Approach 1:
The patent applies preliminary action by generating simulated three-dimensional atomic structure data before actual measurement to train deep learning filters. These pre-generated training datasets, created through molecular dynamics simulations and tomography algorithms, enable the filter to learn and correct missing wedge defects in advance, improving measurement precision without requiring additional experimental equipment.
Solution Approach 2:
The patent introduces a deep learning filter as an intermediary between raw electron microscopy data and final three-dimensional atomic structure determination. This filter, trained on simulated datasets, acts as a mediator that reconstructs missing information in the missing wedge areas, thereby improving both measurement precision and data reliability without direct experimental modification.
2Measurement precision
If deep learning-based neural network models are applied to recover lost data, then image resolution is improved, but the model requires extensive training data which is insufficient in atomic-level three-dimensional structures
Solution Approach 1:
The patent applies copying by creating synthetic copies of atomic structures through molecular dynamics simulations. These simulated atomic models are then used to generate training datasets for deep learning filters, providing sufficient training data without requiring additional experimental measurements. This virtual copying approach solves the data scarcity problem in atomic-level three-dimensional structure analysis.
Solution Approach 2:
The patent employs parameter changes by varying simulation parameters (such as atomic positions, temperatures, and structural configurations) to generate diverse training datasets. By changing these parameters systematically, the patent creates a large volume of varied training data that enables robust deep learning model training, overcoming the limitation of insufficient experimental data.
3Ease of operation
If conventional tomography angle ranges are used (−65° to +65°), then measurement is feasible with current equipment, but missing wedge defects occur that adversely affect surface atomic structure determination
Solution Approach 1:
The patent converts the harmful missing wedge defect into a beneficial training opportunity. By deliberately simulating data with missing wedge artifacts and using it to train deep learning filters, the patent enables the model to learn how to reconstruct and correct these specific defects. This approach transforms the equipment limitation into a solvable pattern recognition problem, improving surface atomic structure determination accuracy.
Data Source
AI summary
A data generating method includes: an atomic model generating step of generating one or more three-dimensional atomic models corresponding to a nanomaterial to be measured; a three-dimensional data generating step of generating three-dimensional atomic level structure volume data corresponding to the nanomaterial to be measured based on the one or more three-dimensional atomic model; a tilt series generating step of generating a tilt series by simulating three-dimensional tomography for a plurality of different angles in a predetermined angle range for at least some of the three-dimensional atomic level structure volume data; and a three-dimensional atomic structure tomogram volume data generating step of generating a three-dimensional atomic structure tomogram volume data set by performing three-dimensional reconstruction on at least some of the three-dimensional atomic level structure volume data based on the tilt series.