Evaluation of the reduction roast – magnetic separation process to upgrade low-grade ferruginous manganese ore fines

Abstract

Manganese is an important raw material in many industries, such as the steel, non-ferrous and battery industries. Manganese ores are typically mined and are then fed to submerged arc furnaces where a manganese ferroalloy, the most common form in which manganese is used, is produced. For economic ferromanganese production, the ore used should have a manganese grade above 40% and a Mn/Fe ratio greater than 7,5. With the increase demand for manganese in recent years, resources for high grade manganese ores are gradually being depleted. In order to deal with the demand, it has now become imperative to make use of lower grade manganese ores which have historically been dumped due to selective mining of higher-grade ores. Many processes have previously been investigated in order to upgrade these manganese ores, such as physical methods, pyro- and hydrometallurgical methods. One such process that has been investigated has the ability to produce a product with a high manganese grade and Mn/Fe ratio (Elliott & Barati, 2020). This process is the reduction roast – magnetic separation process. In order to upgrade the ore, it is first subject to a carefully controlled reduction process to produce phases with different magnetic susceptibilities, such as MnO and Fe or MnO and Fe3O4. This work focused on the optimization of this process for use with ores obtained from the Nchwaning mine slimes dam. Ore was obtained from the slimes dam, reduced to smaller sample sizes and was then subject to a mineralogical study. XRF and XRD was carried out on the samples, and it indicated that the Mn grade was in the region of 44 wt% Mn, which is typical of a high-grade ore. The Mn/Fe ratio, however, was found to be 3 – much lower than the 7,5 required for economic ferromanganese production. The main manganese minerals were found to be braunite I and braunite II, with a small amount of bixbyite and hausmannite present as well. The main iron mineral was hematite. A significant amount of calcite was also found. An optimization study was carried out in the form of a central composite design, with the variables investigated being temperature and reductant ratio. The conditions were selected based on the thermodynamics of the reduction to obtain the desired phases. The reduction step for the process was carried out in a retort furnace, after which the reduced briquettes were milled and subject to a wet magnetic separation process using a Davis Tube. The resultant magnetic and non-magnetic streams were then analysed with XRF and XRD. The results indicated that the reduction roast – magnetic separation process may not be suitable for this specific ore. The best results showed a manganese grade of the product increase from 44% to 47% and the Mn/Fe ratio increase from 3,00 to 3,72. Despite the poor observed separation, the relevant review metrics; namely the Mn/Fe ratio, the percentage Fe removed and the percentage Mn lost, were modelled and response surfaces were developed. Making use of an optimization technique known as desirability functions, it was concluded that the optimized conditions were a temperature between 850K – 900K and a CO content of the reducing gas between 30 – 35 vol%. In order to determine the reasons for the poor separation, SEM and reduction progress tests were carried out. Solid solution phase formation was evident from the SEM images, with the reduction progress tests indicating that the cause of this solid solution formation was not excessive reduction times or temperatures, but due to the mineralogy of the ore itself. The ore was then subject to a kinetic analysis to understand the reaction mechanisms and rate controlling steps. Two separate kinetic studies were carried out, one with a CO-CO2 gas mixture and another with a H2-H2O gas mixture. The use of hydrogen is growing in popularity due to the formation of water vapour rather than CO2 in the reduction reaction. The different sets of conditions for both kinetic studies were the same as those used for the optimization study. In the case of the CO kinetic study, it was found that the Avrami model fit experimental data the best. This model describes nucleation and growth as the rate limiting step. In the case of the H2 kinetic study, the rate limiting step was found to be activation control. A comparison between H2 and CO reduction indicated that the former occurred at significantly higher rates and reached higher reduction extents than the latter.