Materials Science Research Lecture
Webinar ID: 832 7665 2110
Electron back-scatter diffraction (EBSD) has during the past 25 years become an indispensable characterization tool in both materials science and the earth sciences. While the vendor companies have created new and faster detector systems, the underlying commercial indexing algorithm has not undergone significant updates or improvements after the initial development years. The effectiveness of the Hough-based indexing approach depends on the signal strength of the Kikuchi bands with respect to the background signal, and this is the Achilles Heel of the commercial implementations: if bands cannot be detected, then the indexing algorithm fails. In this presentation, we will begin with a discussion of a forward model for the simulation of EBSD and related diffraction patterns. Then we introduce the basic principles of the dictionary indexing approach, as well as the recently introduced spherical indexing algorithm. We will illustrate the robustness of both approaches with respect to pattern noise and the ability to index overlapping patterns near grain boundaries. Along the way we discuss a series of examples on both materials and geological systems. We conclude with preliminary results of the use of convolutional neural networks to accelerate the dictionary indexing approach.
More about the Speaker:
Professor De Graef received his BS and MS degrees in physics from the University of Antwerp (Belgium) in 1983, and his Ph.D. in physics from the Catholic University of Leuven (Belgium) in 1989, with a thesis on copper-based shape memory alloys. He then spent three and a half years as a post-doctoral researcher in the Materials Department at the University of California at Santa Barbara before joining Carnegie Mellon in 1993 as an assistant professor. He is currently professor and faculty director of the J. Earle and Mary Roberts Materials Characterization Facility. His research interests lie in the area of microstructural characterization of structural intermetallics and magnetic materials. His current focus is on the development of experimental and modeling techniques for the quantitative study of magnetic domain configurations in a variety of materials, including ferromagnetic shape memory alloys, magnetic thin films, and patterned structures. A second major research focus is on the acquisition and representation of the three-dimensional character of microstructures. Work in this area includes development of experimental and numerical techniques to extract quantitative 3-D data from serial sectioning experiments using a focused ion beam. The generation of accurate forward models for many different characterization modalities is also a topic of current interest.