Chemical Composition and Evolution of the Mantle

Abstract

A scheme of mantle evolution is proposed that involves extensive (~25%) partial melting of primitive mantle during accretion, followed by cumulate formation in the separated melt and transfer of late-stage fluids, similar to KREEP, from the deeper to the shallower cumulates. Midocean ridge basalts (MORB) form by remelting of incompatible element depleted garnet-rich cumulates; continental and ocean island basalts, including alkali basalts, form by partial melting of a shallow enriched peridotite layer. Forward calculations show that the initial magma and its cumulates have relatively unfractionated Rb/Sr and Sm/Nd and therefore will appear primitive in terms of isotopic ratios. Effective fractionation occurs relatively late in earth history when mantle cooling has reduced the amount of residual fluid in the cumulate layers. The transfer of an intercumulus fluid or partial melt is responsible for depletion of the MO RB reservoir and progressive enrichment of the continental/ocean-island basalt reservoir. The MORB reservoir appears to be an eclogite that earlier had lost a kimberlitic late-state melt. The eclogite, in turn, may have been the result of fractionation of a separated primary melt early in earth history. The large-ion lithophile (LIL) patterns of enriched magmas may be inherited from metasomatic fluids that had been in equilibrium with garnet, rather than indicating a garnet-rich composition for the immediate parent and the residue after partial melting. The composition of the mantle eclogite layer may be picritic. Although the omphacite-pyrope system has not been studied at sufficiently high pressure, results on related systems suggest that the eclogite-garnetite transformation may be responsible for the 400-km discontinuity. The density jump at this discontinuity is about 3%; it seems to be a second-order transition.