Overcoming stereological Bias: A workflow for 3D mineral characterization of particles using X-ray micro-computed tomography

Mineral separation processes operate on properties of individual particle, which can currently be quantified with 2D
characterization techniques, namely 2D automated mineralogy. While 2D automated mineralogy data have driven significant
developments in particle-based separation models, this data inherently correspond to 2D slices of 3D objects, which leads to
stereological bias in the quantification of geometric particle properties. X-ray micro-computed tomography (μCT) is a 3D
imaging technique that can quantify particle geometry. However, μCT only collects limited information regarding material
composition, making mineral identification quantification a challenge. To overcome this challenge, we present a workflow
that utilizes individual particle histograms and corrects image artefacts caused by μCT measurements, such as partial
volume effect. We demonstrate the application of the workflow to perform 3D mineral characterisation of a sulfidic gold ore,
where mineral phases that are commonly mistaken with μCT could be distinguished: pyrite and chalcopyrite, gold, and
galena. Results were verified by comparison with inductively coupled plasma mass spectrometry and 2D automated
mineralogy. As a result, the workflow provides the user with a detailed 3D particle dataset containing the modal mineralogy
and surface compositions, size, and geometrical properties of each particle in a sample – essential data for modelling
mineral separation processes.