Modeling the Merensky Reef, Bushveld Complex, Republic of South Africa

Abstract

The Merensky pegmatoid (normal reef) in the western Bushveld Complex is commonly characterized as a pyroxene-rich pegmatoidal unit with a base that is enriched in chromite and platinum-group element-bearing sulfides overlying a leuconorite footwall. Models for its formation have ranged from those that view it as entirely a magmatic cumulate succession to those that have suggested that it is a zone of volatile-induced remelting. The consequences of the latter interpretation are investigated using the numerical modeling program IRIDIUM, which links diffusive and advective mass and heat transport with a phase equilibration routine based on the MELTS program. The initial system consists of a simple stratigraphic succession of a partially molten leuconorite overlain by a partially molten pyroxenite, both initially at 1,190°C and 2 kbar. 2 wt% of a volatile fluid composed of 75 mol% H₂O, 20 mol% CO₂ and 5 mol% H₂S is then added to the lower 20 cm of the pyroxenite. The system is then allowed to evolve under conditions of chemical diffusion in the liquid. The addition of the volatile components results in a modest increase in the amount of melt in the pyroxenite. However, chemical diffusion across the leuconorite–pyroxenite boundary leads to more extensive melting at and below the boundary with preferential loss of opx from the underlying leuconorite, preferential re-precipitation of sulfide and chromite and concentration of the PGE at this boundary. These results mimic actual mineral and compositional profiles across the Merensky pegmatoid and illustrate that long-term diffusion process can effectively produce mineralogical and compositional layering not present in the original assemblage.

Appendix

In addition to earlier modifications that added both C and S components (Boudreau and Simon 2007), the version of PELE used here also incorporates a Cr component in orthopyroxene (as Cr-diopside: CaCr₂SiO₆). The values for the thermodynamic constants are somewhat arbitrary and taken as Δ_f H° = −27,90,583 J mol⁻¹, S° = 142.5 J mol⁻¹ K⁻¹, V°₂₉₈ = 6.62 J bar⁻¹, with heat capacity, thermal expansivity and compressibility coefficients the same as for diopside as used in MELTS (Ghiorso et al. 1994). Activity of this component in orthopyroxene is taken to be equal to 0.5 × X _Cr-Diopside. This produces an apparent opx/liquid distribution for Cr equal to about 2 at 1 bar and about 11 at 10 kbar for Bushveld-type liquids, and they agree well with measured opx/liquid distribution coefficients at different pressures of Auwera et al. (2000).

In addition, Cr–H₂O liquid interaction parameters have been changed from 0 (as used in MELTS) to a value of 60,000 J mol⁻¹. This is based on studies such as those of Ford et al. (1972) and Gaetani et al. (1994), that suggest that H₂O expands the phase field of Cr spinel relative to olivine. Although the interaction parameters appear to be a large, this value is only about two times that used by MELTS for liquid component H₂O–MgO and H₂O–FeO interactions. Furthermore, the correction is typically multiplied by the mole fractions of both MgCr₂O₄ and H₂O, both of which are generally low, so the actual change in the free energy of MgCr₂O₄ liquid component is modest. For example, for a MORB liquid containing 0.25 wt% H₂O, the free energy of the liquid component MgCr₂O₄ increases from −21,36,345 to −21,35,554 J mol⁻¹, or a difference of (791 J mol⁻¹.