Spinel Non-stoichiometry as the Explanation for Ni-, Cu- and PGE-enriched Sulphides in Chromitites

Abstract

High concentrations of PGE are associated with many chromitite layers in layered intrusions. The R-factor argument of Campbell et al. (1983) can explain the difference between the PGE concentrations in chromitite-hosted and Merensky-type PGE ores and the concentrations in magmatic sulphides that are less enriched in PGE. The very high concentrations of Ni and Cu also characteristic of chromititehosted sulphides are not explicable in this way. This has led Gain (1985) and von Gruenewaldt et al. (1986) to suggest that the original mass of these sulphides has been greatly reduced by the removal of Fe and S, with the consequent enrichment of other metals.

We suggest that the mechanism of Fe-loss from sulphide is that of Fe transferring to fill vacancies that exist in chromite crystallizing from basaltic magma.

Thermodynamic data for the reaction

$$4\!/\!3Fe_2O^{(spinel)}_3 + 1\!/\!3FeS = Fe_3O^{(spinel)}_4 + 1\!/\!6S_2$$

((1))

indicate that, on cooling from 1150 to 930°C, if the fS₂ is buffered to that of sulphide cooling as a closed system, Fe-Ni-S sulphide containing 20 mole % NiS in equilibrium with chromite of the composition of the UG-2 can lose Fe to the chromite, using a high proportion of the available vacancies in the chromite at 1150°C. Depending on the mass ratio of the sulphide to chromite, the NiS content of the sulphide can be more than doubled.

Mass balance calculations indicate that if the sulphides of the UG-2 originally had the composition of those of the Merensky Reef (i.e. concentrations of 10·9% Ni, 4·65% Cu), augmenting this to their present concentrations would require the loss of 1578 ppm Fe, which would represent a gain in the mole fraction of Fe in the chromite of 1·8 × 10^-3.

A simple model for spinel, including the various end-members with vacancies (i. e. those involving Al₂O₃ and Fe₂O₃), has been used to calculate the activities of these sesquioxides necessary for the spinel to contain a certain number of vacancies. When this model is used to calculate the vacancies needed to take up the Fe required to explain the enrichment in Cu, Ni and PGE in the Mandaagshoek section of the UG-2 (Gain, 1985), the calculated activities of Al₂O₃ and Fe₂O₃ are of the same order as those measured experimentally for basaltic melts at 1180°C and fO₂ of less than the Ni-NiO buffer. The Mandaagshoek section is one of the most sulphide-rich encountered in the Bushveld chromitite layers, and, therefore, requires more Fe to be removed than most. Thus the model will also account for all of the chromitites containing less sulphide than this area of the UG-2.