May 17 @ 4:00 pm - 5:00 pm
- This event has passed.
Rapid Determination of Grain Size Distributions by EBSD with Quasirandom Sampling
Postdoctoral Fellow, Hopkins Extreme Materials Institute
Johns Hopkins University
Microstructure characterization is an important step in high-throughput screening of new materials and processing routes. Electron backscatter diffraction (EBSD) is a mainstay of microstructure characterization, which is typically used to build a crystallographic orientation map of the grain structure by collecting diffraction patterns from a dense grid of sampling points across an area of interest. Collecting these data can be the rate limiting step in an automated screening system, but throughput can be increased by minimizing the number of points from which data are collected.
Although EBSD can provide a wealth of structural information about polycrystalline materials, in the context of high-throughput screening methods data of lower fidelity may be sufficient for identifying and distinguishing microstructurally interesting regions. For example, instead of using a dense grid of EBSD patterns to obtain a complete grain size distribution, it may be sufficient to only measure some moments of the distribution, such as the average grain size. This opens up the potential to reduce the time needed for characterization by only collecting EBSD patterns from enough points to measure the desired parameters.
In this presentation, I will demonstrate one method for accelerating the characterization of grain size distributions by using quasirandom sampling to reduce the required number of EBSD patterns. I will show work towards quantifying the uncertainty of the resulting grain size distributions and incrementally obtaining EBSD measurements from the area of interest until a specified uncertainty in the grain size distribution is obtained. This approach significantly reduces the number of EBSD patterns that must be collected. A further description of the method for calculating grain boundary lengths from the same data using support vector machine (SVM) clustering will be discussed.
Timothy Long is a postdoctoral fellow with the Hopkins Extreme Materials Institute at Johns Hopkins University. He earned his BS in Materials Science & Engineering in 2013 from the University of Pennsylvania. He then earned his MS (2016) and Ph.D. (2019) from Cornell, where he studied the impact of solute hydrogen on grain-scale plasticity in pure nickel using a combination of high energy diffraction microscopy (HEDM) and modeling. He joined Hopkins in December 2021, working with Todd C. Hufnagel and Tim Mueller. His current research focuses on accelerating EBSD and x-ray diffraction-based materials characterization for high-throughput screening of structural materials.