X-ray diffraction (XRD) as a fast industrial analysis method for heavy mineral sands in process control and automation-Rietveld refinement and data clustering

摘要

X-Ray diffraction (XRD) provides useful information about the composition of an ore sample in terms of quantification of crystalline phases and amorphous content. The use of new, fast detection systems in XRD creates the opportunity to use this technique in modern process control. In the analysis of heavy mineral sands, XRD can identify the main mineral phases, such as ilmenite FeTiO_3, rutile TiO_2, Zircon ZrSiO_4 and quartz SiO_2, and any other minor components present, such as anatase TiO_2, magnetite Fe_3O_4, haematite Fe_2O_3 or monazite (Ce, REE)PO_4. Quantitative analysis is possible by various classical methods such as straight line or polynomial calibration with standards, but modern quantification analysis techniques such as Rietveld analysis (Bish and Howard, 1988) are attractive alternatives, as they do not require any standards or monitors. These methods offer impressive accuracy and speed of analysis. The Rietveld method compares calculated versus experimentally derived X-ray powder diffraction patterns for the sum of all crystalline phases. Another analysis technique offering great benefit to mining industries is cluster analysis. Enormous amounts of XRD measurement data are generated during the process control and material evaluation. New ways of handling such vast amounts of data are required to produce meaningful information for the end users. Cluster analysis greatly simplifies the analysis of a large amount of data from different processes or different raw materials, and automatically sorts closely related scans of an experiment into clusters and marks the most representative scan of each cluster as well as outlying patterns. The use of cluster analysis to evaluate the XRD data allows fast and reliable tracking of the process. It is the most economical procedure to have automatic data evaluation without involving any dedicated personnel in the process. Details of the techniques used, sample optimization methodologies, results, data precision and limitations will be discussed. The approach has potential as a relatively inexpensive, reliable tool, useful in the characterization of heavy minerals sand materials.