Friday, 4 March 2011

ICAM 2011 presentation on hematite and magnetite discrimination

1) optical microscopy, 2) enhanced BSE image, 3) segmentation, 4) MLA classification

FEI Australia Center of Excellence for Natural Resources is going to present a second talk at the 10th International Congress for Applied Mineralogy (ICAM 2011) in Trondheim, 1-5 August 2011. Here is a preview of our talk on a new MLA approach discriminating hematite from magnetite for iron ore characterisation.

"Advanced discrimination of hematite and magnetite by Automated Mineralogy"
by German Figueroa, Kurt Moeller, Michael Buhot, Gerda Gloy and David Haberlah

As the global growth of steel production and consumption continues to accelerate, innovation in the whole industry from iron ore extraction to processing is needed. By providing quantitative and statistically reliable compositional information, automated mineralogy solutions such as the Mineral Liberation Analyser (MLA) have become important tools for characterising iron ore minerals (i.e. hematite, magnetite, goethite and limonite) and their processing products. Although magnetite (Fe3O4) and hematite (Fe2O3) can be easily distinguished qualitatively using optical microscopy, quantitative characterisation by automated scanning electron microscopy/energy-dispersive x-ray spectroscopy (SEM-EDS) is challenging. Hematite and magnetite are chemically close and display similar backscatter electron (BSE) intensities, making discrimination by energy-dispersive X-ray (EDX) spectra alone difficult.

This study presents an automated mineralogy approach discriminating iron oxides by taking full advantage of the subtle difference in backscatter intensities between hematite and magnetite. The advanced workflow involves three steps: 1) Optimisation of measurement parameters increasing the BSE brightness and contrast. In standard MLA operation mode, the BSE brightness is calibrated so that the mounting media (resin) is kept at backscatter brightness values below 15, and gold at a value of 250, covering all common minerals. The modified settings stretch the BSE range for iron oxides from 115-120 to 195-215, effectively doubling the grey level contrast. Two separate modes emerge representing hematite (~200) and magnetite (~208), which can be separated in the image segmentation stage. 2) EDX spectra acquisition combining two measurement settings. Single EDX spectra are collected from geometric centre points of unsaturated segmented phases corresponding to discrete minerals, including hematite and magnetite. Saturated segments, comprising multiple bright mineral phases, are mapped using a regular grid with further phase discrimination based on EDX spectra. 3) Mineral identification is performed by an advanced classification algorithm combining BSE thresholds and EDX spectra.

The new approach is applied to a synthetic sample including particles displaying complex intergrowth between hematite and magnetite, bright sulphide phases, and common gangue minerals. The automated phase-by-phase approach characterises hematite and magnetite reporting quantitative modal composition, mineral association and locking. The results demonstrate that the advanced approach can successfully discriminate iron oxides into hematite and magnetite while at the same time correctly reporting the modal contributions of other phases. Hematite and magnetite are locked as binary phases reflecting intergrowth, and further occur as ternary phases with quartz and feldspar. Locking of the iron oxides can be slightly overestimated without significantly impacting overall results, due to mineral impurities, defects and boundaries with epoxy showing in the BSE image.

In conclusion, an automated SEM-EDS approach is demonstrated to successfully discriminate and quantify hematite and magnetite by advanced mineral identification based on modified backscatter intensities and EDX spectra matching.

No comments:

Post a Comment