Nickel in komatiites

Abstract

THE principle of partitioning of trace elements between magma and crystalline phases is an extremely useful tool in deciphering the origin and evolution of magmas. With our present knowledge of partition coefficients calculations usually assume them to be constant. The nickel contents of suites of komatiitic lavas from South Africa, the evolution of which is controlled by olivine fractionation, suggest that the partition coefficient for nickel between olivine and liquid is not constant, but increases from <3 for liquids with >20% MgO and >1,000 p.p.m. Ni to 11 for liquids with 10% MgO and 200 p.p.m. Ni. There are two fundamentally different views on the origin of basic magmas. O'Hara¹ proposed that the mantle melts to produce a picritic liquid which undergoes extensive olivine fractionation prior to eruption. The alternative model, summarised by Ringwood², assumes that basic magma is a direct product of melting and undergoes no fractionation. Ringwood's rejection of the fractionation hypothesis is based largely on calculations of the Ni content of magmas². All his calculations assume a constant partition coefficient for Ni between olivine and liquid in the range 12–15. Several recent investigations concluded that K_D is not constant but that it decreases (1) with increasing temperature^3,4; (2) with increasing MgO content in the liquid⁵; (3) with increasing Ni content in the liquid⁶, and (4) as crystallising olivines become more forsteritic⁷. These effects have been reviewed in detail elsewhere⁸ but the net effect is that for a liquid, rich in MgO and Ni, crystallising a forsteritic olivine at high temperature, K_D may be significantly lower than the figure currently accepted. Komatiitic lavas fall perfectly into this category of liquid as they have high MgO and Ni contents and it has been shown experimentally⁹ that they crystallise very magnesian olivines at temperatures well above the liquidus for normal basalts. Most of the studies which have indicated low K_D values have involved experimental work on synthetic samples^4–7, some of which^5–7 were iron-free and hence, although they represent an extremely important step towards understanding partition coefficients, quantitatively the K_D values may be suspect. We have analysed the Ni contents of komatiites with a wide range of compositions to determine the partition coefficient for these seldom-erupted lavas to test more rigorously the constraints imposed by Ni values on the fractionation model for basic magmas.