MIT uses AI to detect atomic material defects
Traditional methods often require destructive testing and offer limited insight, while the AI approach reconstructs defect patterns using machine learning.
Researchers at MIT have developed an AI model capable of identifying and quantifying atomic-scale defects in materials without damaging them. The approach aims to improve the design and performance of semiconductors, batteries, and solar cells.
The model analyses data from neutron-scattering experiments and can detect up to six different point defects simultaneously. Trained on 2,000 semiconductor materials, it analyses atomic vibrations to estimate defect types and concentrations that are hard for traditional methods to measure.
Conventional techniques such as X-ray diffraction or electron microscopy typically capture only limited aspects of material defects and often require destructive testing. The AI system uses pattern recognition to build a more complete picture, offering a non-invasive option for manufacturing quality control.
Researchers say the method could eventually be adapted to more widely used tools such as Raman spectroscopy, making industrial adoption more practical. Future work will also extend the model beyond point defects to larger structural features in materials.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!