Related Experiment Video
Updated: Nov 18, 2025
06:49
In situ Grazing Incidence Small Angle X-ray Scattering on Roll-To-Roll Coating of Organic Solar Cells with Laboratory X-ray Instrumentation
Published on: March 2, 2021
6.5K
Classification of grazing-incidence small-angle X-ray scattering patterns by convolutional neural network
Hiroyuki Ikemoto1, Kazushi Yamamoto2, Hideaki Touyama2
1Department of Physics, University of Toyama, Japan.
Journal of Synchrotron Radiation
|February 10, 2021
Summary
Convolutional neural networks (CNNs) can now classify nanoparticle shapes from grazing-incidence small-angle X-ray scattering (GISAXS) patterns. This AI approach achieves 90% accuracy, efficiently analyzing complex GISAXS data.
Area of Science:
- Materials Science
- Nanotechnology
- Artificial Intelligence
Background:
- Grazing-incidence small-angle X-ray scattering (GISAXS) data contains complex information from nanoscale structure, particle interactions, and layer geometry.
- Manual identification of models from GISAXS patterns is challenging due to numerous superimposed contributions.
Purpose of the Study:
- To apply convolutional neural networks (CNNs) for automated classification of GISAXS patterns.
- To specifically focus on identifying nanoparticle shapes from GISAXS data.
Main Methods:
- Utilized a convolutional neural network (CNN), a type of artificial neural network, to analyze GISAXS patterns.
- Trained the CNN to recognize regularities within GISAXS data corresponding to different nanoparticle shapes.
Main Results:
- The CNN achieved a success rate of approximately 90% in classifying GISAXS patterns based on nanoparticle shape.
- The network effectively identified underlying regularities in the GISAXS data.
Conclusions:
- CNNs provide an efficient method for classifying large volumes of experimental GISAXS patterns.
- This AI-driven approach can accurately categorize patterns based on predefined model shapes and their combinations.
