Maximizing rock-forming minerals differentiation with 1 laboratory spectral X-ray computed tomography

Conventional X-ray micro computed tomography (CT) is widely used for non-invasive 3D material characterization, yet differentiating between the material’s constituents can be challenging. This is notably the case when characterizing rocks from the Earth’s crust. About 95% of them is composed of a mixture of common rock-forming minerals, many of which are indistinguishable with conventional CT. We overcome this limitation by combining laboratory spectral micro computed tomography (sp-CT) with an astute analysis workflow. Where previous studies showed laboratory sp-CT’s potential for identifying heavy elements, we focus on the light-element-bearing minerals that do not feature characteristic absorption edges within the detectable energy range of state-of-the art spectral detectors. For a natural monzo-diorite sample, we show that those minerals appear as separate clusters on a multi-energy histogram derived from a laboratory sp-CT tomogram. Selecting a particular cluster is equivalent to segmenting one constituent in conventional post-processing. The resulting mineral differentiation was corroborated by scanning electron microscopy and Raman spectroscopy. The presented workflow is applicable to any multi-component material. Specifically for the Earth Sciences, it breaks the barrier to analyse a variety of silicate-bearing rocks, their minerals, structures and reactions in three dimensions, non-invasively in the laboratory.