Composition of the Marginal Rocks and Sills of the Rustenburg Layered Suite, Bushveld Complex, South Africa: Implications for the Formation of the Platinum-Group Element Deposits

The Bushveld Complex contains large ore deposits of platinum-group elements (PGE), V, and Cr. Understanding how these deposits formed is in part dependent on estimates of the compositions of the magmas that filled the Bushveld chamber. Over the past 20 years, estimates for the major oxides and some trace elements in the magmas have been made using the marginal rocks of the intrusion. However, data for most of the trace elements have not been available. This paper presents the results for a full range of trace elements, including the platinum-group elements. The marginal rocks of the Lower and lower Critical zones (B-1. magmas) are tholeiitic Mg-rich basaltic andesites with Mg# 71. It had been suggested that they are boninites but their mantle-normalized incompatible lithophile trace element patterns (spidergrams) resemble those of the upper continental crust and the concentrations of the elements are much higher than those of boninites. The patterns resemble siliceous high magnesium basalts. An unusual feature is that the Pt/Pd ratios are >1.5. The Pt contents of the B-1 rocks (15-25 ppb) are slightly higher than those observed in most primary mantle melts, suggesting that the high Pt/Pd ratio is due to Pt enrichment rather than Pd depletion. The crystallization order and composition of the minerals formed ill equilibrium with the B-1 magma matches that of the Lower and lower Critical zones and thus this magma appears to be representative of the parental magma of these zones. The marginal rocks to the upper Critical zone (B-2) are tholeiitic basalts in terms of major element composition, with Mg# 55. The spidergrams show some similarities with E-MORB; however, the B-2 rocks have strong positive Ba and Pb anomalies and negative P, Ti, Hf, and Zr anomalies, and thus they more closely resemble lower continental crust. The B-2 rocks have lower and more variable Pt + Pd contents than the B-1 magma, suggesting that some of the samples have experienced sulfide saturation, but in common with the B-1. magmas, the Pt/Pd ratios are high, in excess of 1.5. The crystallization order of the Upper Critical zone cannot be modelled by the B-2 magma alone. However, mixtures of B-2 magma and B-1 magma satisfy the crystallization order and mineral composition of the upper Critical zone. The marginal rocks of the Main zone (B-3) are also tholeiitic basalts in terms of major element composition but have a higher Mg# (62) than the B-2 rocks. Trace element patterns in part resemble those of B-2 magmas but are depleted in most incompatible elements with large positive Ba, Pb,. and Eu anomalies and negative Nb, Ta, HI, and Zr anomalies, suggesting the rocks contain a plagioclase component. The PGE contents of the B-3 rocks are lower than those of the B-1. magma and less variable than those of the B-2 magma, but in common with both the other magmas, they have high Pt/Pd. The crystallization order and composition of the minerals in equilibrium with the B-3 magma matches that of the Main zone. Two processes have been suggested to explain the compositions of the Bushveld magmas: mixing of primitive mantle melts with partial melts of continental crust and mixing of primitive mantle melts with melts derived from the subcontinental lithospheric mantle (SCLM). The trace element concentrations of the magmas can be modelled by crustal contamination. This interpretation is supported by oxygen isotopes, initial Sr-87/Sr-86 ratios and epsilon(Nd) of the cumulate rocks. However, the high Pt/Pd ratios of all of the magmas and the overall higher than normal Pt concentrations of the B-1 magma are difficult to explain by mixing of primary mantle melt with crustal components. The SCLM has high Pt/Pd ratios and mixing of primary mantle magma with SCLM-derived magma could account for the high Pt concentrations and high Pt/Pd ratios. This interpretation is supported by recent work on Os isotopes of the Kaapvaal SCLM. It should be kept in mind that the two processes are not necessarily exclusive. A magma with a SCLM component could have been emplaced into the crust and subsequently have been contaminated by partial melts of the crust.