Abstract: Techniques for training an artificial neural network (ANN) using simulated specimen images are described. Simulated specimen images are generated based on data models. The data models describe characteristics of a crystalline material and characteristics of one or more defect types. The data models do not include any image data. Simulated specimen images are input as training data into a training algorithm to generate an artificial neural network (ANN) for identifying defects in crystalline materials. After the ANN is trained, the ANN analyzes captured specimen images to identify defects shown therein.

Abstract: The invention relates to a method 3D defect characterization of crystalline samples in a scanning type electron microscope. The method comprises Irradiating a sample provided on a stage, selecting one set of crystal lattice planes of the sample and orienting said set to a first Bragg condition with respect to a primary electron beam impinging on said sample, and obtaining Electron Channeling Contrast Image for an area of interest on the sample. The method is characterized by performing, at least once, the steps of orienting said selected set of crystal lattice planes to a further Bragg condition by at least tilting the sample stage with the sample by a user-selected angle about a first tilt axis, and obtaining by Electron Channeling Contrast Image for a further area of interest.