Geochemistry of ilmenite and applications to ilmenite provenance and Ti-mineral exploration

lmenite is a primary ore feed to the DuPont chloride process for the manufacture of titanium diox-ide pigment. We seek to determine ilmenite geochemistry as a function of original source and as-certain which ilmenites have chemistry most amenable to upgrading via iron-leaching (leucoxenization) and that have favorable, with respect to pigment manufacture, trace element content. Ilmenite geochemistry is a function of its primary origin and of subsequent alteration. Heavy mineral sand deposits may contain ilmenite from one or several dominant source rock(s). In addition, ilmenite may alter in the weathering environment, changing the major element compo-sition (iron-leaching) and certain of the minor element content. A primary objective is to classify ilmenites in relation to origin in order to fingerprint source rocks in provenance determination. The changes in trace element chemistry during alteration are also investigated. A “decision tree” is developed to classify ilmenite types and source rock origins. Ultimately, with those methods and by amassing a database of ilmenite types from worldwide locations, it should prove possible to determine ilmenites (and their sources) more susceptible to iron leaching in the weathering en-vironment. Such ilmenites, enhanced in their titanium content, are the preferred ore targets. The method used is electron microprobe analyses at elevated conditions of voltage, cur-rent, and counting time to enable higher precision. In-situ ilmenites from igneous, metamorphic, and altered rocks (saprolites) are analyzed. Initial foci for sedimentary ilmenites are restricted provenance (approaching monogenetic, examples from Spain, Portugal) and polygenetic prove-nance (Denmark). Altered ilmenites, including the DuPont Florida ore, are used to characterize the chemistry of leucoxenization. Trace element concentration, detectability, and cost considera-tion, led to focus on Mn, Mg, V, Cr, and Nb to develop a classification scheme validated by dis-criminant analysis and “classification-regression trees”. Results show a high degree of success for restricted provenance study and an acceptable one for polygenetic provenance.