Linking Microstructure and Materials Properties
Gaining a deep understanding of the link between a material’s properties and its micro- or even nanostructure is essential for developing novel materials. In this context a recent development in the field of microstructure characterization is highly beneficial: the introduction of machine learning (ML) for image segmentation and analysis. In this study it will be shown how robust and stable ML systems are used to characterize inclusions of additive manufactured high-temperature Al alloys. Correlative microscopy approaches between light and electron microscopy provide meaningful data needed to train ML-based segmentation. As a result, inclusions generated during the additive manufacturing process can be quantitatively identified.
Machine learning-based classification can be trained on one dataset, then applied across multiple samples to give repetitive, non-subjective results. The ability to classify based on features other than just local greyscale values, particularly the ability to classify based on textural information, has the potential of being transformative in the ability to extract information from images in materials research.
