Know What's Happening in Life

A Model for Defect Identification in Materials

MIT researchers have created an AI model that classifies and quantifies up to six point defects in semiconductor materials using non‑invasive neutron‑scattering data. Trained on a database of 2,000 samples covering 56 elements, the model can detect defect concentrations as low as 0.2 percent. The technique offers a universal, quantitative view of defects that traditional methods cannot achieve without damaging the material. The work, published in Matter, signals a shift toward AI‑driven quality control in electronics, solar cells and battery materials.