A data-constrained computational model for morphology structures

Abstract

A data-constrained computational model has been developed for generation of a microscopic 3D morphology structural map of a material sample, using 3D sets of CT-reconstructed tomograms and a limited number of 2D morphological section maps. The model is defined on a 3D hyper-cubic lattice. Each voxel of the lattice represents an elementary volume of material. It is assumed that the morphological sections are representations of the 3D morphology structure and that the material composition of a given voxel is determined statistically by its neighbouring voxels. Efficient computational algorithms have been developed which incorporate the sum rules of total volume-fractions and the total linear-absorption coefficients on each voxel. The model algorithms have been implemented as a user-friendly MS-Windows software package. The software generated morphology structures agree reasonably welt with the actual structures for a range of simulated testing data sets. Our approach to 3D compositional morphology structures has a number of advantages in materials design and modelling. For example the 3D morphology maps can be used for microscopic chemical processes modelling on a real representation rather than an idealized one.