Fiber orientation and distribution measurement in a composite polymer

The determination of fiber orientation within composite polymers is a critical parameter for the prediction of a material’s behavior. Over time, different methods have been developed to obtain an estimate of the distribution and orientation of reinforcing fiber materials, which works with a fair level of approximation in global modeling.

However, when a local behavior must be studied or more precision is desired in the mathematical modeling of materials, a far more accurate analysis can be carried out, consisting of the combination of computer-aided tomography (CT) and image processing software.

A challenging sample to image

Benhui Yu et al. described their process very accurately and selected a rather challenging sample to validate their workflow. They combined natural rubber with carbon black, other additives, and short polyester fibers in a mixer at a rolling temperature ranging between 40 and 50 Celsius.

The mixing process resulted in a composite polymer made of material with very comparable density (1.4g/cm3 for the polyester fibers vs ~1g/cm3 for the resulting rubber). This poses a challenge for CT scans, as the difference in density among materials is what provides contrast in datasets resulting from this imaging technique.

When imaging such a sample, the scanner’s sensitivity must be increased to distinguish between the two materials in the results. However, the increase in sensitivity comes at a cost in terms of field of view, so the original 3x3x2.4 mm sample could not be fully scanned. The team, therefore, focused on a sub-volume of 1.05×1.05×1.05 mm.

Converting images into actionable numerical information on composite polymers

Once the scan is completed, imaging reconstruction software needs to be used. This software converts the scan into a 3D model that can be further investigated. The reconstruction algorithms have a strong dependency on the operational parameters, so most CT users tend to rely on the software provided with their hardware for the reconstruction. This is similar to what the team previously did, which involved transferring the data to Thermo Scientific Avizo Image Processing Software for visualization, segmentation, and measurement of the fibers in the sample. The team utilized the software features to remove noise and acquisition artifacts from the dataset, obtaining data that was easier to segment. They then proceeded to identify the polyester fibers using a segmentation algorithm based on image binarization. The result of the algorithm provided the team with a visual representation of the fiber network inside the material.