Mineral quantification by laser-induced breakdown spectroscopy: towards on-stream mineral characterization.

2019

Francis Vanier, National Research Council Canada; Elton Soares de Lima Filho, National Research Council Canada; Josette El Haddad, National Research Council Canada; Aïssa Harhira, National Research Council Canada; Christian Padioleau, National Research Council Canada; Mohamad Sabsabi, National Research Council Canada; Greg Wilkie, CRC ORE; Alain Blouin, National Research Council Canada

Real-time mineral quantification of coarse rock streams could greatly enhance the decision-making processes and the mining operation efficiency. In the laboratory, automated mineralogy instruments based on electron microscopes and energy dispersive x-ray spectroscopy (EDS-SEM) provide reliable information on the mineral abundance and texture of prepared rocks. However, such technique cannot be applied in the field, especially on coarse rock streams. Laser-induced breakdown spectroscopy (LIBS) has been used for elemental analysis in many geology and mining applications such as rock characterisation on conveyor belts, but LIBS was not used for true mineral identification and quantification. Here, we present a novel method for mineral identification and quantification using LIBS, even when several mineral phases of similar elemental composition are present. This technique could be scalable to perform automated mineralogy measurements in coarse rock streams. A set of rock tiles from mining operations in Australia was characterised using (EDS-SEM) and the resulting quantitative mineral analysis (QMA) data were used to guide and validate the results obtained by LIBS. The use of a tailored multivariate analysis technique applied to the LIBS data allowed the identification, quantification and imaging of minerals on rock tiles, even in the presence of mixed mineral phases within the laser spot area. Mineral abundance and imaging were obtained with success for the mineral phases selected in the present work, which includes bornite, chalcopyrite, pyrite, molybdenite, quartz, chlorite, K-feldspar, albite, fluorite and calcite. The method presented a mineral quantification root mean square error below 10 % for the listed minerals. In addition, mineral quantification by point-counting using single laser shots per LIBS measurement is demonstrated, achieving absolute errors below 3.5 % for major minerals and below 1 % for minor minerals. This work was jointly funded by the National Research Council Canada and the Collaborative Research Centre for Optimising Resource Extraction (CRC ORE) in Australia.

Keywords: Mineral phase identification; LIBS; multivariate analysis; quantitative mineral analysis;