Microstructural Foundations of Magnesium Performance: A Data Mining Approach to High-throughput Electron Microscopy

点击翻译按钮获取中文摘要

In order to meet ambitious targets for reduced energy consumption and emissions in modern vehicles, one of the most widely adopted strategies involves the deployment of lightweight structural materials. At less than a quarter of the density of steel, magnesium is a natural contender to replace legacy materials in automotive and other vehicle components. However, despite its desirable physical properties, magnesium accounts for only a small proportion of the make-up of a typical automobile. The reason, in large part, stems from the difficulty with forming the relatively brittle magnesium alloys into complex shapes required for vehicle components. This award supports research into the science underlying micro-scale behavior of magnesium alloys during forming and other deformation operations. By uncovering the relationships between the metal's microstructure and its response to deformation, modified manufacturing processes and improved alloys can be designed to enable widespread use of lightweight magnesium components in vehicular and other weight-sensitive applications. This will be complemented by STEM-oriented outreach activities and external collaborations.The response of a magnesium component to applied deformation is governed by two atomic-level phenomena: slip and twin activity. Twinning is especially vital to deformation at and near room temperature, due to the difficulty of slip. This project will build upon newly developed microscopy techniques (high-resolution electron backscatter diffraction) to generate snapshots of nano and micro-level structural activity during forming activities. Data mining knowledge extraction techniques will be applied to the resultant huge data sets of twin and slip activity in order to accelerate the discovery of interrelations between microstructure and magnesium deformation mechanics. The data mining will employ a decision-tree type analysis and a neural net approach. The resultant knowledge will be embedded in a meso-scale model of twin / deformation activity as the basis for assessment and design of improved alloys for light-weight structural applications. New insights will emerge into twin development that span various grain-size ranges and critical temperature levels. Key structure parameters will be modeled, including detailed grain-boundary character, accurate local (relative) strain levels, dislocation activity / slip transition temperatures, measures of crystal lattice entropy and other metrics of local heterogeneity. Furthermore, the high-throughput microscopy and data mining advances developed as part of the study will serve as a new framework for accelerated knowledge discovery for constitutive models of deformation in crystalline materials.