Publication

Comparison of methods to determine fibre length distribution: X-ray CT versus standard method

Publication

Outline

D. Salaberger, S. Gleiß, J. Kastner - Comparison of methods to determine fibre length distribution: X-ray CT versus standard method - Int. Conf. Structural Analysis of Advanced Materials, Kos, Greece, 2013

Abstract

For the determination of fibre length distribution of fibre reinforced polymers several methods are available. The standard method is burning or digesting the matrix and imaging the remaining fibres with light optical microscopy (LOM). Alternative methods are 2D image analysis of grinded cuts or 3D X-ray computed tomography (XCT). In this work the fibre length distribution of glass fibres in Polypropylene matrix were determined by XCT and the standard method. The goal was to demonstrate limitations and advantages of both methods. Pyrolysis followed by LOM was performed referring to the ISO standard [1] under consideration of the results of [2]. The standard specifies the number of fibres to be analyzed with at least 400 and it suggests a semi- automatic image analysis. Usually small fibre fragments are not taken into consideration. For the comparison with XCT data also very small fragments have to be analyzed. Therefore the LOM image analysis was performed manually. The XCT analysis was done using the Sub-µm CT device Nanotom and an in- house software development [3]. A reasonable resolution has to be chosen to reach an adequate data quality. The resolution depends mainly on fibre diameter and fibre material. The glass fibres that were analyzed in this study had a mean diameter of 12.5 µm. In previous studies [3] a resolution of 2 µm voxelsize was found optimal for the determination of fibre length distribution. The smallest fibre length that was taken into account was 25 µm for XCT and LOM analyses. For this study Polypropylene filled with 30 % wt. glass fibres produced by injection moulding was analyzed. Samples from multi-purpose test specimens (MPS) were cut out to achieve high resolution. Figure 1 shows slice images on the left, acquired at 2 µm voxelsize and the extracted fibres on the right. The color coding indicates the orientation of each fibre. As a result from CT data analysis the start and endpoint of each fibre is extracted. To be able to evaluate the CT data analysis software and to make a comparison with the standard method, specimens from different locations within the MPS were cut out. These specimens were very small (ca. 0.5 mm edge length) to make sure that the amount of fibres within the specimens is in the range of 1000. This is a number that can be analyzed manually in microscopic images with reasonable effort. A typical fibre length distribution for the investigated material is shown in figure 2. It is obvious that in the class of 25 to 50 µm many fibres or fibre fragments are present. Because of the very small volume that was analyzed, the maximum fibre length is much smaller than in a final part. The comparison of CT and standard method, as shown in figure 2, shows good accordance. Special attention was taken on the small fibre fragments that can act as sources for damage within an injection moulded part. From the comparison a quantitative error that is made by the CT method will be derived. Furthermore notable properties of the fibres that were found by CT like diameter variations will be verified. Acknowledgement: This work is part of the K-Project "Non-Destructive Testing and Tomography" supported by the COMET-Program of the Austrian Research Promotion Agency (FFG) and the Province of Upper Austria (LOÖ).