Optimizing Quantification of Microstructural Parameters of Crystal-Glass Assemblages by Electron Backscatter Diffraction

Abstract

We evaluated the use of electron backscatter diffraction (EBSD) for microstructural quantification of silicate crystal- and glass-bearing (hemicrystalline) samples, focusing on the effects of EBSD post-processing (cleaning, grain reconstruction, boundary smoothing, re-gridding). Our case study reanalysed experimental samples of Pontesilli et al. (2019), comprising synthetic trachybasalt melts ± H2O, cooled to 1100°C and held for 30 minutes, producing a phase assemblage comprising dendritic (anhydrous sample) or skeletal (hydrous sample) clinopyroxene (Cpx), titanomagnetite (Tmt), and glass. EBSD indexing quality (Confidence Index and Fit) gradients occur across crystal-glass boundaries, causing a thresholding effect via the cutoff selected to identify glass. Standardising indexing quality to the grain maximum had the greatest impact on results, removing gradients but inflating grains. Phase proportions (Φ) and crystal sizes were influenced mainly by thresholding effects. Number density (NA) and surface area to volume ratio (${\mathrm{S}}_v^{\mathrm{P}}$) were also affected by boundary smoothing, minimum grain size threshold, and step size. Ideally, we recommend analysing differently processed scans simultaneously: non-standardised for Φ, grain sizes and growth rates, and standardised for NA, ${\mathrm{S}}_v^{\mathrm{P}}$, and crystal-crystal misorientations. Comparison with previous backscatter electron (BSE) imaging revealed that EBSD and BSE deliver similar results if spatial resolution is similar and few crystals of the same phase touch. EBSD is recommended if higher resolution is required, BSE contrast is poor, many crystals of one phase touch, and/or crystals are dendritic. Our case study revealed higher ΦCpx and total crystallinity but lower ΦTmt under anhydrous conditions, potentially due to undercooling-enhanced crystallisation of Al- and Ti-enriched Cpx.