Automated data processing and comprehensive analysis of fracture surfaces of fiber-reinforced unidirectional composites using SEM and confocal laser scanning microscopy

A methodology and algorithms have been developed for the joint automated processing and integrated analysis of scanning electron microscopy (SEM) and confocal laser scanning microscopy (CLSM) data obtained from fracture surfaces of fiber-reinforced composites with parallel fiber alignment. The approach enables extraction of a wide range of quantitative microstructural characteristics that are critical for evaluating the fracture energy components and strength of the composite, understanding the failure mechanisms, and supporting their modeling. Based on the analysis of fracture surface images from both halves of the broken specimen, as well as cross- ectionalmicrographs, the following parameters are determined: contours and count of fibers, shape characteristics of fiber cross-sections (diameters, axis lengths, etc.), distributions of cross-sectional areas, fiber volume fraction, spatial density and heterogeneity of fiber distribution, count of exposed (pulled-out) fibers on the fracture surface, statistical distribution of pull-out lengths (including estimates of mean and variance), segmentation of matrix and delamination boundaries, contours of fiber clusters sharing a common fracture plane, the number of such clusters, and the surface areas of matrix or pull-out fibers. The key idea and novelty of the developed methodology lies in the combined use of two complementary high-resolution microscopy techniques, and in the joint analysis of fracture surfaces from both complementary halves of a specimen failed under bending or tension. This fusion allows for significantly enhanced data extraction compared to the use of individual images, enabling reconstruction of information hidden in deep relief regions of one-half of a specimen using data from the other half. The resulting analysis achieves high accuracy and robustness in contour detection and in the computation of quantitative features. The methodology and its software implementation have been demonstrated using complex fracture surfaces of unidirectional metal matrix composites produced at the Institute of Solid State Physics (RAS) by drawing carbon fiber yarn (1000 filaments, 7 μm diameter) through an aluminum–bismuth melt. The composite wires are 300 μm in diameter and have a fiber volume fraction of 60–65%.