Ore deposit studies necessitate modern techniques to understand the processes responsible for forming economically viable mineral resources. In the transition towards a circular economy, mineral resources need to be looked after carefully to ensure a sustainable future. Geological research conventionally relies on macroscopic and microscopic observations of hand specimens and thin/polished sections to characterize the physical and chemical properties of minerals and to reconstruct the mineral paragenesis. Microscopic observations are hereby often limited to traditional two-dimensional (2D) techniques (e.g., optical microscopy). Although these techniques are well-known to characterize the mineral composition of ore deposits, they are not capable of reproducing the real three-dimensional (3D) interior of geological samples.
This thesis addresses the challenges of developing a comprehensive and more standardized methodology to correctly characterize the mineralogy of ore samples in 3D. During this study, well-known techniques like scanning electron microscopy coupled with energy dispersive spectroscopy (SEM-EDS) and state-of-the-art (spectral) μCT were combined to overcome some of the traditional issues in characterizing the different mineralogical phases at the microscopic scale.
As a first step, a methodology has been developed where 2D imagery from SEM-based automated mineralogy (AM) solutions were combined with 3D μCT images to be able to characterize ore samples in 3D. The SEM-based AM mineral phase mapping served as the ground truth data to establish a library of minerals present in the studied ore deposit (i.e., samples from a pegmatite deposit were investigated). The library of minerals was, in turn, used to calculate the expected attenuation values for a given experimental μCT setup. This provided a most correct guide to differentiate and segment different minerals from μCT images, as was easily verified with correlated SEM-based AM images. The segmented mineral phases from μCT allowed us to quantify 3D information on the orientation and the spatial correlation by using statistical descriptors at a low computational expense.
The potential of the combination of SEM-based AM solutions and μCT has subsequently been tested to investigate the formation history of lithium-cesium-tantalum pegmatites, enriched in tin, niobium, tantalum and tungsten, of the Mesoproterozoic Karagwe Ankole belt (KAB) in Central Africa. The textural complex and mineralogical diverse assemblages were now studied in 3D and offered new insights into the paragenetic history of mineralized pegmatites in the KAB. A mineral assemblage of quartz-albite-tourmaline-muscovite revealed a stage of tourmaline crystallization between two stages of albitization. Cornish type cassiterite in greisen pockets confirmed hydrothermal formation conditions, while the spatial association of columbite-tantalite with oriented zircon textures suggested local magmatic-hydrothermal conditions.
The combination of SEM-based AM and μCT allowed us to extend the investigation of ore samples to 3D. However, the absence of direct 3D chemical information in standard μCT imagery limits the differentiation of phases to minerals with distinct differences between their attenuation values. Spectral μCT, as opposed to standard μCT, allows energy-sensitive imaging of attenuation values and the identification of characteristic elemental K-edges in the recorded attenuation profile. Two different spectral laboratory-based μCT setups, sensitive to an atomic number Z higher than 42 (molybdenum), were tested on the potential of non-destructively identifying chemical elements of ore minerals in 3D. Samples from the Lebong Tandai, Tambang Sawah & Cirotan (Indonesia) epithermal gold-bearing breccia, the rare earth element mineralization of Gakara (Burundi) and cassiterite concentrates from quartz veins and pegmatites of the KAB (different locations in Central Africa) were investigated to cover an entire range of chemical elements. K-edge identification of chemical elements within the practical margins of high-Z spectral detectors was shown to be useful to qualitatively differentiate between minerals with otherwise similar attenuation values in standard μCT imagery.
Ultimately, the developed methodologies allowed us to extend and improve the structural, mineralogical and chemical characterization of ore samples to full 3D and get a more thorough understanding of the formation history and geodynamic setting of critical mineral resources.