Machine vision characterisation of the 3D microstructure of ceramic matrix composites

A new approach to quantifying the microstructure of continuous fibre-reinforced composites has been presented which reduces the time required to quantify the microstructure and, hence, to better understand the microstructure-sensitive response when compared to prior methods. The technique was demonstrated by characterising the voids and fibre orientation within a continuous SiC fibre-reinforced SiNC ceramic matrix composite as these are known to affect both the oxidation and mechanical behaviour of the material. Microscopy data obtained via automated serial sectioning were analysed using standard digital image correlation algorithms to extract the 3D fibre orientation field, and histogram thresholding to extract the void shape and porosity distribution. Employing orthogonal decomposition, the dimensionality of the void shape data was also reduced, enabling interpretable comparisons between the fibre orientation field and the voids. The approach outlined here is applicable to studying the microstructure-sensitive response and optimization of processing for improved performance.