Thesis (MScEng) –Stellenbosch University, 2010 / ENGLISH ABSTRACT: More mining houses are turning their attention to the processing of ore mined from the UG2 reef. This is mainly due to the depletion of the Merensky reef or the low availability of ore from the Platreef. With the higher UG2 ore concentration in the process feed, companies experience more problems with the processing of the ore due to its high chrome content. Although there are various possible solutions to the problems experienced in the processing of UG2 ore, very little of thise solutions are actually implemented in the industry. When smelting chrome-bearing ore, the chrome can go to any of three phases: matte phase, spinel phase or glass phase. If it reports to the furnace matte phase it can lead to problems in the down stream processing. When chrome forms part of the spinel phase it forms a solid, refractory-like material which, to an extent, is an unwanted material inside the furnace. Lastly the chrome can report to the glass phase (liquid phase) which is the more desireable phase to which chrome should report to since it will have little or no impact on downstream processing. There were four main objectives for this research project namely to conduct a literature review to understand the problem of chrome in the smelting process, to do an experimental simulation of a matte and slag phase in one crucible, to interpret the experimental results and compare the experimental results to thermodynamic predictions obtained using FactSageTM. From literature the following aspects were found to be important: Silica structures Research has shown that silica incorporates different metal ions into its structures. When the metal ions are incorporated into the silica structures they are seen as part of the glass phase. By incorporating cations such as chromium into the silica structure the structure is able to maintain its neutral electrical charge. Basicity of slag Oxides once melted will either donate or accept oxide ions. The group of oxides that are in excess will determine whether that specific slag is acidic or non acidic. This is important to our study since the basicity can determine the stability of the spinel phase. Partial pressures The partial pressure of the system is important since it forms part of the equilibrium constant calculations - meaning that partial pressures determine the stability of certain species. Currently different arguments exist as to how the sulphur pressure inside a molten bath is maintained. What is important, however, is that for the system in this research project a log oxygen partial pressure of -8 and a log sulphur partial pressure of -3.5 was chosen. These values were based on previous research done on PGM and copper smelting processes. Chrome deportment Previous research on chrome deportment shows a relationship between temperature and the amount of chromium dissolving into the glass phase as well as an increased chromium(II)oxide solubility (when compared to chromium(III)oxide) in a silicate melt. Regarding slag chemistry and chrome deportment very little work has been published for the system found in the PGM industry but it is mentioned that by adjusting certain slag additives (alumina, lime and silica) the chrome deportment can be manipulated. The scope of this project was to investigate the effect of chromium, lime and silica on chrome deportment. More specifically, the effect of lime. The reason being that lime was originally added as 10%wt of the feed stream in the processing of Merensky ore to act as a fluxing agent. With the new furnace design and higher power densities the slag are maintained at 200oC to 350oC higher than when Merensky ore was smelted. With these higher slag temperatures fluxing agents will play a smaller role meaning that lime additions become less important. Literature studies also showed that additional lime in a slag system can stabilize the spinel structure (which is an unwanted phase). Removing lime would be advantageous out of a chrome presepective as well as an economic point of view. The research was conducted in three sections namely the determination of the time required for this particular system to reach equilibrium, the investigation of different additives on chrome deportment using a controlled atmosphere and synthetic slags and, lastly, a comparison of the experimental results obtained to thermodynamic predictions. For the equilibration studies reaction time periods of 4,7,11 and 16 hours were used. From the results it was found that the alumina crucible dissolved into the glass phase continually. This indicated that equilibrium was not reached. However, a reaction time, rather than an equilibrium time, was chosen where the species and phases had enough time to react. This was based on literature, on observations of species diffusing between the matte and slag phase as well as on two-point analysis (diffusion gradients). A reaction time of nine hours was chosen With the reaction time fixed, the effect of different slag additives on chrome deportment was investigated. For the addition of chrome it was seen that an 1.5%wt increase in the starting material increased the chrome content of the slag phase by 0.025%wt. For the same increase in chromium in the starting materials the chrome content of the spinel phase increased by 2.1% indicating that chromium has a tendency to report to the spinel phase. The increase in chromium had a minor effect on the chrome content of the matte phase, however, since the chrome content only increased from 0.025%wt to 0.028%wt. Plotting the results showed that increasing the lime in the starting material decreases the chromium content in the glass phase. For the 39%wt silica system the chromium content in the glass phase decreased from 0.75%wt to 0.46%wt for an increase in the lime content from 1.7%wt to 7.3%wt. The same trend was seen for the 33%wt silica system. For the spinel phase an increase of 1.27%wt was seen when the lime content of the starting materials was increased from 1.7%wt to 10.1%wt for a 33%wt silica system. The same increase in lime increased the chrome content of the matte phase from 0.03%wt to 0.06%wt for a 33%wt silica system. Silica also proved to affect chrome deportment. Increasing the silica content of the starting materials from 25%wt to 39%wt increased the amount of chromium in the glass phase from 0%wt to 0.46%wt for a 10%wt lime system. The same effect is seen for a lower lime content except that more chromium were incorporated into the silica structure. A silica increase from 32.4%wt to 39%wt resulted in a chromium decrease from 5.2%wt to 0% in the spinel phase for a 10%wt lime system. The same trend was seen for the 1.5%wt lime system. An increase in the silica levels lead to an increase in the chrome level of the matte phase. When silica is increased from 32%wt to 39%wt the chrome content of the matte phase increased from 0.06%wt to 0.07%wt. The last part of the research project entailed the comparison of the trends observed with the experimental results to trends obtained from thermodynamic predictions. FactSageTM is a program that uses model equations to predict the Gibbs free energies for different phases. The program is therefore also able to predict the amount of different phases present at equilibrium. This is called thermodynamic ”optimization”. In section 6 trends that were observed from FactSageTM results are compared to the trends found in the experimental results. It is important to note that it is only trends that are evaluated and not actual values since FactSageTM calculations are for a system that is at equilibrium and (as explained above) this system is not at complete equilibrium. The comparison however was good. Several trends found in the experimental results were confirmed by the results from FactSageTM. These included the relationships of chromium fed versus chromium spinel, lime fed versus chromium spinel, lime fed versus chromium in glass, silica fed versus chromium in glass, silica fed versus chromium in spinel and silica fed versus chromium in matte. To conclude, slag additions can be used to manipulate chrome deportment to an extent. Secondly, FactSageTM can be used for thermodynamic predictions but a proper understanding as well as some form of validation of the specific system investigated is still needed. Due to time constraints and the difficulty of experimentally simulating this multi-phase system in the lab, only a few parameters were investigated. In order to obtain a more complete understanding of the system the effect of partial pressures and temperature should also be investigated.