Quantitative image analysis for developing microstructural descriptions of heterogeneous materials

Abstract

A computer-based imaging system for elucidating the microgeometry of porous materials has been developed. The system predicts three-dimensional properties of the porous material by examination of a small number of representative sections of the sample. By generating random three-dimensional particulate structures in the computer and numerically slicing them with parallel test planes, the validity of the methodology for predicting porosity, surface to volume ratio, isotropy, mean particle size, and particle size distribution has been established. Predictions for a variety of randomly generated structures yielded estimates for porosity between 4–6% of actual values and surface to volume ratio between 5–13%. A quantitative threshold of 10% in extent of orientation has been defined to distinguish between isotropic and anisotropic structures. Particle size distributions have been predicted, in good agreement with actual distributions, by observing as few as 600 particle profiles on the section plane. Three-dimensional reconstructions from serial sections have been used to verify the assumptions required for the prediction of particle size distribution. The system developed is quite general and has been applied to a porous membrane device for the controlled release of macromolecules. For the controlled release system and for other problems of conduction in porous media, these quantitative predictions of microgeometry should be useful in the development of predictive models of transport behavior.