Stochastic Multi-Scale Reconstruction of 3D Microstructure Consisting of Polycrystalline Grains and Second-Phase Particl
- PDF / 1,708,924 Bytes
- 11 Pages / 593.972 x 792 pts Page_size
- 84 Downloads / 141 Views
INTRODUCTION
ADVANCED experimental techniques such as the electron back-scattered diffraction (EBSD) imaging are routinely used to characterize the microstructure of polycrystalline materials.[1] Due to the nature of this technique, only the polished flat surface of the material can be probed, which results in a 2D micrograph showing a ‘‘slice’’ of the grain morphology, orientation, and spatial distribution. Serial sectioning technique[1] has been used in combination with EBSD to obtain 3D polycrystalline structures. Specifically, after the EBSD micrograph of the top surface is obtained, the specimen is polished such that the top layer is removed and a new surface is imaged. This process is continued until a sufficient number of 2D slices of the material are obtained, which are then stacked to generate a 3D reconstruction. Although widely used, this technique is tedious and very time-consuming. Moreover, it destroys the specimen and is thus not suitable to study timedependent structure evolution under external stimuli. SHAOHUA CHEN and ANTONY KIRUBANANDHAM, Graduate Research Assistants, NIKHILESH CHAWLA, Fulton Professor, and YANG JIAO, Assistant Professor, are with the Materials Science and Engineering, Arizona State University, Tempe, AZ 85287. Contact e-mail: [email protected] Manuscript submitted May 28, 2015. Article published online December 28, 2015 1440—VOLUME 47A, MARCH 2016
Recently, a number of stochastic reconstruction techniques have been developed that allow one to generate ‘‘virtual’’ polycrystalline microstructures from three orthogonal 2D EBSD micrographs associated with the three surfaces of the specimen.[2,3] Specifically, each 3D grain is modeled as an ellipsoid, whose three principal axes are estimated from the 2D micrographs. A packing of ellipsoids is generated using a Monte Carlo simulation, whose semi-axis statistics satisfy those obtained from the 2D images. The ellipsoids are allowed to grow to fill the entire simulation domain, which leads to 3D grain structure. Then the grain orientations are assigned based on the corresponding statistics obtained from the 2D micrographs and a 3D polycrystalline microstructure is reconstructed. More recently, a generalized method[4,5] has been devised that incorporates additional statistics such as orientation correlations directly obtained from 3D EBSD reconstructions. In addition, a new tessellation scheme has been used to better represent grain morphology.[5] These stochastic reconstruction techniques have been proven to be very successful in generated virtual microstructure for large homogeneous polycrystalline materials for subsequent quantitative analysis. An important assumption for the aforementioned stochastic reconstruction techniques is that the 2D micrographs contain a sufficiently large number of grains in order to lead to robust statistics. In addition, the spatial correlations between the grains with different METALLURGICAL AND MATERIALS TRANSACTIONS A
shapes and sizes as well as the distribution of different grains were not explicitly considere
Data Loading...
