A stepwise Study on the characterisation and processing of South African Platinum Group Tailings

Thesis / Dissertation


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title
The endurance of the mining industry has led to the near-depletion of some of the most processible ore types. This has resulted in a unique challenge that necessitates unique innovation for the industry. Firstly, existing technologies are increasingly geared towards improved efficiency in processing lower grade ores. Secondly, ores that were previously processed with older, and in some cases, inefficient technologies, have emerged as potentially viable solutions that would help maintain concentrator capacities. On the other hand, as the industry has endured, so has its waste accumulation. Tailings dumps continue to grow, and continue to pose various environmental issues. But although they are a waste product, tailings also have several merits to them. They are already mined and readily available, and reprocessing them immediately addresses two conundrums i.e. how can the industry source alternative ores, and how can it deal with the steady accumulation of waste? It is expected that as a result of their initial processing from so-called fresh ores, and their stay in their respective dumps, the tailings will be altered or tarnished. The surface properties of their compositional minerals will be oxidised and layered with other compounds that might hinder their interaction with flotation reagents such as collectors and hence, hinder their flotation response, presenting a challenge for their proposed reprocessing. More obviously, as they are a waste product, their grades will be far lower than the fresh ores that produced them. Numerous studies aim to elucidate the viability of the metallurgical reprocessing of tailings. However, this flurry of innovation thus far extends to the vast stretches of the Witwatersrand dumps, and thus, to gold. The issue with processing other tailings types then becomes threefold. Is there enough value in the dumps to justify tailings beneficiation? What beneficiation method would be suitable? Would the method yield economically viable results? This study acquired three bulks of PGM tailings to investigate these questions. The first bulk was Merensky, the second was UG2, and the third was a deslimed version of UG2 in which a portion of denser minerals were separated out. For the sake of convenience, the bulks are referred to by their shorthands throughout the thesis. Merensky became MER, the normal or raw UG2 became UG2R, and the deslimed UG2 became UG2D. The study conducted the investigations along two lines: first, by characterising the tailings, and secondly, by floating them. The first characterisation step was a full elemental and mineralogical analysis that quantified the amount of valuables as well as gangue minerals; this was done via XRD and QEMSCAN. The second step was to determine the degree of oxidation by using qualitative, in-situ experimental methods; these are EDTA extraction and oxygen reactivity. It was hypothesised that EDTA extraction would measure the concentration of secondary and oxidised materials on the mineral surfaces; and oxygen reactivity would measure the demand of oxygen in each tailings, and therefore the tailings' capacity to react with species in the pulp phase. The qualitative methods have only ever been used for fresh ores, and have shown to be reliable in predicting and/or explaining the flotation behaviour of those ores. They have never been used to predict an/or explain the flotation behaviour of an ore material that has already been processed, and is therefore very low grade, and oxidised. If they are as viable for tailings as they are for fresh ores, they would determine different EDTA extraction indices and oxygen demand constants. QEMSCAN and XRD provided the different concentrations of the different minerals across the tailings, and showed that some minerals were present in one tailings but not the other. For instance, MER contained the highest fraction of chalcopyrite, as well as the highest fraction of sulphides. UG2R and UG2D each had more than ten times the fraction of chromite seen in MER. MER, on the other hand, had more pyroxenes, plagioclase and amphibole. It was expected that these differences in composition would be the cause of the different extraction indices and oxygen demand constants. The robustness of both methods thus had to be tested. This was done by altering each of the tailings and testing whether the extraction indices and oxygen demand constants would change. The surface and composition alteration was induced with ultrasonication and desliming. The study thus answered the question: can the EDTA extraction and oxygen reactivity methods detect when a change has occurred on the mineral surfaces of the same tailings? The tailings were then floated, and it was here that the question of economic viability was assessed. For the concentrators from which MER, UG2R and UG2D were collected, a reprocessing venture would be deemed economically viable if 30% of the copper present in the tailings was recovered. The flotation performance was thus analysed with copper recovery as the primary positive indicator. Nickel behaviour was also tracked in case of any supporting elucidations, and also because pentlandite is the primary PGE-carrier for Merensky and UG2. The tailings were floated with DOW200 as the frother, SIBX as the collector, and CMC as the depressant. The results showed that the presence of the depressant resulted in very low solid quantities being recovered to the concentrate. In fact, less than 2% of each tailings was recovered, and less than 10% of the present copper (and therefore chalcopyrite) was extracted. When the depressant was removed from the reagent scheme, recovery of the solids improved to 10%, but the copper yield was still below 30%. So, the collector dosage was increased under the fundamental assumption that the hydrophobicity of the valuables would be improved, and in this way, the efficiency of separating the hydrophilic gangue from the valuables would also improve. The plan worked, and UG2R finally achieved the industrial objective of recovering 30% of the present copper. While MER and UG2D failed to do the same, their performance was also at its best under these conditions, with each tailings yielding roughly 23% copper recovery. In an effort to improve floatation, the tailings were cleaned via ultrasonication and desliming. These cleaning methods both had a detrimental effect on copper recovery. However, nickel (and therefore pentlandite) behaviour improved, showing that while the methods were disadvantageous to one mineral, they were favourable to another, and they might be useful for a study that uses a different metal as its positive performance indicator. The study also showed that MER, UG2R and UG2D have different copper EDTA extraction indices. UG2D had the highest index, followed by UG2R, and then MER. The copper minerals associated with UG2D can therefore be concluded to be more oxidised than those associated with UG2R and MER. Moreover, UG2D was the least reactive to oxygen, having an oxygen consumption rate constant of 0.113 min-1 when compared to 0.198 and 0.152 min-1 for UG2R and MER, respectively. Ultrasonication and desliming decreased each of these constants, indicating that when cleaned with the chosen methods, the mineral surfaces became less reactive to oxygen. And so, it was concluded that of the investigated tailings, UG2D was more oxidised than the other two, reacted the least with oxygen, and yielded the lowest copper recoveries. When UG2R achieved the highest reactivity with oxygen, it also yielded the lowest copper extraction index and the highest copper recovery. Overall, nickel behaved contrary to copper.