Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust

The average chemical compositions of the continental crust and the oceanic crust (represented by MORB), normalized to primitive mantle values and plotted as functions of the apparent bulk partition coefficient of each element, form surprisingly simple, complementary concentration patterns. In the continental crust, the maximum concentrations are on the order of 50 to 100 times the primitive-mantle values, and these are attained by the most highly incompatible elements Cs, Rb, Ba, and Th. In the average oceanic crust, the maximum concentrations are only about 10 times the primitive mantle values, and they are attained by the moderately incompatible elements Na, Ti, Zr, Hf, Y and the intermediate to heavy REE. This relationship is explained by a simple, two-stage model of extracting first continental and then oceanic crust from the initially primitive mantle. This model reproduces the characteristic concentration maximum in MORB. It yields quantitative constraints about the effective aggregate melt fractions extracted during both stages. These amount to about 1.5% for the continental crust and about 8-10% for the oceanic crust. The comparatively low degrees of melting inferred for average MORB are consistent with the correlation of Na20 concentration with depth of extrusion [1], and with the normalized concentrations of Ca, Sc, and AI (= 3) in MORB, which are much lower than those of Zr, Hf, and the HREE ( = 10). Ca, A1 and Sc are compatible with clinopyroxene and are preferentially retained in the residual mantle by this mineral. This is possible only if the aggregate melt fraction is low enough for the clinopyroxene not to be consumed. A sequence of increasing compatibility of lithophile elements may be defined in two independent ways: (1) the order of decreasing normalized concentrations in the continental crust; or (2) by concentration correlations in oceanic basalts. The results are surprisingly similar except for Nb, Ta, and Pb, which yield inconsistent bulk partition coefficients as well as anomalous concentrations and standard deviations. The anomalies can be explained if Nb and Ta have relatively large partition coefficients during continental crust production and smaller coefficients during oceanic crust production. In contrast, Pb has a very small coefficient during continental crust production and a larger coefficient during oceanic crust production. This is the reason why these elements are useful in geochemical discrimination diagrams for distinguishing MORB and OIB on the one hand from island arc and most intracontinental volcanics on the other. The results are consistent with the crust-mantle differentiation model proposed previously [2]. Nb and Ta are preferentially retained and enriched in the residual mantle during formation of continental crust. After separation of the bulk of the continental crust, the residual portion of the mantle was rehomogenized, and the present-day internal heterogeneities between MORB and OIB sources were generated subsequently by processes involving only oceanic crust and mantle. During this second stage, Nb and Ta are highly incompatible, and their abundances are anomalously high in both OIB and MORB. The anomalous behavior of Pb causes the so-called "lead paradox", namely the elevated U / P b and Th/Pb ratios (inferred from Pb isotopes) in the present-day, depleted mantle, even though U and Th are more incompatible than Pb in oceanic basalts. This is explained if Pb is in fact more incompatible than U and Th during formation of the continental crust, and less incompatible than U and Th during formation of oceanic crust.