Mineralogical Limitations for X-Ray Tomography of Crystalline Cumulate Rocks

Abstract

The use of x-ray computed tomography (XRCT) on igneous rocks enables the visualisation and quantification of the 3D texture of the rock and of the crystal population as opposed to a more traditional 2D vision using thin sections and 3D stereological conversions. Although still in its infancy, the application of XRCT on igneous rocks provides a 3D map of the distribution of each mineral phase, the overall dimensional metrology of crystals (size, area volume, shape, orientation and geometry) and potentially their crystallographic orientation. The precision of crystal size distributions (CSD), which are often used for describing rocks and understanding igneous processes, is enhanced by the use of the 3D analysis of crystal size. XRCT shows promising results when applied to volcanic rocks but only limited applications with intrusive igneous rocks
In this study, we compare the dimensional metrology of crystals in 2D and in 3D of a Tugtutoq peridotite sample using a thin section and 3D tomography data to investigate how different the 3D data is from the 2D to test the utility of using XRCT on a dense cumulate. The tomography data is processed in four steps: i) post-processing which includes filtering noise and correcting artefacts linked to the XRCT acquisition; ii) segmentation of the cumulus phase in the sample (in this case olivine); iii) separation of the segmented olivine into realistic and discrete crystals; and iv) the extraction of the data from the 3D-separated olivine crystals (size, location and distribution in the sample, shape, orientation, …).
The results of the tomography scan of the peridotite sample are close to the thin section data but the error associated with the tomography data is difficult to quantify. That error is likely high due to the low contrast in attenuation coefficients between the crystal populations – linked to the similar densities of the mineral phases in the sample – and the close proximity and extensive contact between the olivine crystals. When applied to cumulate rocks, the method cannot be automated to produce reproducible and objective results, but we suggest the application of this method for processing tomography data with either volcanic rocks or cumulates solidified from a more permeable mush, where there is minimal contact between the crystals of interest and the density contrast between phenocrysts and groundmass is high.