Investigation of carbon-based reductant, low-temperature process for conversion of hematite in red-mud to magnetite

Abstract

The research reported in this Thesis was confined to a specific red-mud originating from Jamaican bauxite with an iron-content, reported as Fe2O3, of ~60 m%. In principle, separation (selective recovery) of this iron from the red-mud would yield a residual product in which the minor (‘high-value’) Al- and Ti-bearing constituents would be concentrated. Thus, the objective of this research was to explore such a strategy to effect this outcome. Conversion of ‘hematite’ to ‘magnetite’ and subsequent magnetic-classification to obtain a high-Fe magnetic fraction and a low-Fe non-magnetic fraction was the strategy selected for this investigation. It was established, on the basis of a cursory equilibrium-thermodynamic analysis, that the desired conversion was a probable outcome at ‘low’ temperatures (300 to 600oC) and a low-cost carbon-based reductant, petroleum-coke, involving C(s)/CO(g),was capable of providing for the range of oxygen partial-pressures necessary for magnetite (Fe3O4) stability. These two process-specifications (low-temperature/low-energy-requirement and low-cost reagent) were considered to be primary requirements for the potential development of an economically-viable value-added conversion-process for red-mud. Reduction experiments were therefore conducted with petroleum-coke as reductant, as well as with a gas-mixture of CO-CO2 containing N2 as diluent. Experiments performed with petroleum coke as a reductant blended with red-mud were unable to achieve the desired objective. In order to elucidate the reaction-mechanism that is likely to facilitate practical reduction- rates in petroleum-coke–red-mud blends, a series of experiments was then performed to observe the reduction-behavior of red-mud when the gaseous oxides CO(g) and CO2(g) were employed directly to provide the requisite oxygen-potential for reduction. Furthermore, N2(g) served as a diluent in order to manage the low partial-pressures of these gases which were necessary, in order that the reaction-rates were not too rapid and could therefore be conveniently measured. ‘Optimal conditions’ for the gas-phase reduction were determined to be: a processing temperature of 540oC ± 10C , partial pressures CO(g) and CO2(g) each of 0.070atm (bar) ± 0.001atm.(bar)/ inert diluent-gas: N2(g), for a conversion-time of 30min. A mathematical-model was developed on the basis of unidirectional-diffusion of CO2(g) within the CO2(g)–CO(g)–N2(g) gas-phase of the porous product-layer – the binary: CO2(g)–CO(g), represent an equimolar, counter-diffusion pair in accordance with the stoichiometry of the conversion reaction: 3 Fe2O3(s) + CO(g)  2 Fe3O4(s) + CO2(g). The model was capable of correlating the time-dependent motion of the (sharp) Fe2O3–Fe3O4 reaction-interface. Experiments conducted to assess magnetic-classification, of the product obtained by gas-phase processing, in each case, did not yield a non-magnetic fraction. Instead, only a small-particle size-range and a larger particle size-range magnetic-fraction were realized. This finding was subsequently attributed to the discovery, via STEM imaging, of nanometer length-scales associated with the nascent crystallites of the entities, which are intrinsic to Bayer-Process precipitate — red-mud. Thus the strategy proposed to effect a classifiable magnetic/non-magnetic product will require additional research to assess the ramifications of this intrinsic characteristic of the precursor red-mud by which the minor-constituents are comingled with the hydrated ferric-oxide(s) at nanometer length-scales.