Variations in Cr content of magnetite from the upper zone of the Bushveld Complex — evidence for heterogeneity and convection currents in magma chambers

Abstract

Mineral separates of pure magnetite from the upper zone of the Bushveld Complex have been analysed for Cr. Detailed sampling within layers of massive magnetite indicates that the Cr content frequently shows an extremely rapid but regular depletion with height (for example, by a factor of seven over 85 cm of massive magnetite), with sudden reversals of variable magnitude. Rayleigh Law fractionation, from a homogeneous liquid of the volume now seen as cumulate rocks overlying the sampled horizons, cannot satisfy these observed concentration gradients. A diffusion-controlled crystallization model is preferred. In another vertical section of massive magnetite the Cr content remains constant with increasing height, and is interpreted as indicating a steady state situation where the rate of depletion of Cr by magnetite fractionation is balanced by the rate of diffusive addition into the crystallization zone. Reversals in Cr content sometimes occur in the middle of pure magnetite layers and are attributed to convective overturn in the magma chamber. These results provide geochemical support for the model of convection cells and bottom crystallization in large magma reservoirs as proposed by Jackson [20]. Depletion of compatible elements in this bottom layer causes chemical inhomogeneity in the magma. Sections, composed of several layers of magnetite sandwiched between magnetite-bearing gabbro, have also been studied. In one case, a steady decrease in Cr content with increasing height in the separated magnetite was observed; in another, several irregular reversals were found. There appears to be no systematic relation between convection cycles, the depletion in Cr and the formation of pure magnetite layers, indicating that the chemical composition of the magma does not control the production of monomineralic layers. The fluctuating pressure model envisaged by Cameron [16] for the formation of such layers is entirely consistent with these data.