Grain reconstruction from boxscan data

A method for reconstructing the 3D shape of a single grain in a polycrystal from far-field diffraction data is presented. The reconstruction is performed using an iterative algorithm, algebraic reconstruction technique (ART), to solve a linear system of equations obtained from data from the Boxscan technique. The Boxscan technique is a 3D X-ray diffractiontype scanning technique, which – in geometric tomography terms – provides 2-dimensional X-rays of the object as data. In this paper, we report on the first experimental 3D grain shape reconstruction based on a beta-titanium sample. By comparing our results to highresolution phase contrast tomography, we find that the average error on the position of the grain boundary in this sample is 2.7 micrometers for a grain radius of 17 micrometers. This result is similar to other current methods.