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: 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: A substrate is alignable for ion beam milling or other inspection or processing by obtaining an electron channeling pattern (ECP) or other electron beam backscatter pattern from the substrate based on electron beam backscatter from the substrate. The ECP is a function of substrate crystal orientation and tilt angles associated with ECP pattern values at or near a maximum, minimum, or midpoint are used to determine substrate tilt. Such tilt is then compensated or eliminated using a tilt stage coupled the substrate, or by adjusting an ion beam axis. In typical examples, circuit substrate “chunks” are aligned for ion beam milling to reveal circuit features for evaluation of circuit processing.