Partial Roasting of a PGM Concentrate

Abstract

PGM concentrates contain more than just the platinum group minerals. They also concentrate three minerals pyrrhotite (Fe1 xS), pentlandite ([Fe,Ni]9S8) and chalcopyrite (CuFeS2). In a sample of Merensky concentrate tested in this study, these minerals account for half the concentrate. During smelting and converting these sulphides collect the PGMs in matte. Smelting produces a matte, the composition of which converting adjusts by oxidizing sulphur to SO2 and iron to Fe2SiO4 (fayalite). The objective of roasting a PGM concentrate would be to alter the composition of a concentrate, at lower temperatures in a kiln before smelting, so that the matte formed contains less iron and sulphur, but, still, collects all of the PGMs, copper and nickel. Roasting tests were conducted in a bench-scale rotary kiln on a sample of Merensky concentrate supplied by Lonmin. The kiln was purged with air to maintain high oxygen partial pressures (~0.21 atm O₂). Test temperatures from 350 °C to 700 °C and residence times from 2 to 30 minutes were explored. The roasted products were smelted in argon at 1500 °C to assess the effect of roasting on matte fall and the deportment of base metals. Roasting for 20 minutes at 550 and 650 °C reduced the sulphur content in the concentrate by 60 and 70 %, respectively from 17.4% to ~5% sulphur. Iron was oxidized to an iron oxide. Fe3O4 (magnetite) predominated at lower temperatures (<500 °C); Fe2O3 (hematite) predominates at higher temperatures. Oxidation in each of these minerals occurred through a number of reactions. Some rates were faster than others. Under the conditions tested all of the iron in pyrrhotite oxidized to iron oxides, and most of the iron in chalcopyrite and pentlandite were oxidized. Copper and nickel remained as sulphides. The smelting of roasted products produced a lower matte fall, the iron in matte was less than 3 %, and matte collected all of the copper, but not all of the nickel. Nickel also partitioned to the slag, where it reported to the spinel phases. The matte formed (from concentrate roasted at 550 °C) in a number of discrete prills (rather than collecting in a single "button"), which might have been a consequence of high slag viscosities. For concentrate roasted at 650 °C discrete alloy prills and not matte prills were formed.