Geochemistry and models of mantle circulation

Geochemical data help to constrain the sizes of identifiable reservoirs within the framework of models of layered or whole-mantle circulation, and they identify the sources of the circulating heterogeneities as mainly crustal and/or lithospheric, but they do not decisively distinguish between different types of circulation. The mass balance between crust, depleted mantle and undepleted mantle based on 143Nd/144Nd, Nb/U and Ce/Pb, and the concentrations of very highly incompatible elements Ba, Rb, Th, U, and K, shows that ca. 25- 70% (by mass) of depleted mantle balances the trace element and isotopic abundances of the continental crust. This mass balance reflects the actual proportions of mantle reservoirs only if there are no additional unidentified reservoirs. Evidence on the nature and ages of different source reservoirs comes from the geochemical fingerprints of basalts extruded at mid-ocean ridges and oceanic islands. Consideration of Nd and He isotopes alone indicates that ocean island basalts (oibs) may be derived from a relatively undepleted portion of the mantle. This has in the past provided a geochemical rationale for a two-layer model consisting of an upper depleted and a lower undepleted (‘primitive’) mantle layer. However, Pb-isotopic ratios, and Nb/U and Ce/Pb concentration ratios demonstrate that most or all oib source reservoirs are definitely not primitive. Models consistent with this evidence postulate recycling of oceanic crust and lithosphere or subcontinental lithosphere. Recycling is a natural consequence of mantle convection. This cannot be said for some other models such as those requiring large-scale vertical metasomatism beneath oib source regions. Unlike other trace elements, Nb, Ta, and Pb discriminate sharply between continental and oceanic crust-forming processes. Because of this, the primitive mantle value of Nb/U = 30 (Ce/Pb = 9) has been fractionated into a continental crustal Nb/U = 12 (Ce/Pb = 4) and a residual-mantle (morb (mid-ocean ridge basalt) plus oib source) Nb/U = 47 (Ce/Pb = 25). These residual mantle values are uniform within about 20% and are not fractionated during formation of oceanic crust. By using these concentrations ratios as tracers, it can be shown that the possible contribution of recycled continental crust to oib sources is limited to a few percent. Therefore, recycling must be dominated by oceanic crust and lithosphere, or by subcontinental lithosphere. Oceanic crust normally bears a thin layer of pelagic sediment at the time of subduction, and this is consistent with oib sources that are dominated by subducted oceanic crust with variable but always small additions of continental material. Primordial 3He, 36Ar, and excess 129Xe, in oceanic basalts demonstrate that the mantle has been neither completely outgassed nor homogenized, but they do not constrain the degree of mixing or the size of reservoirs. Also, helium does not correlate well with other isotopic data and may have migrated into the basalt source from other regions. The high 3He/4He ratios found in some oibs suggest that, even though the basalts are not derived from primordial mantle, their sources may be located close to a reservoir rich in primordial gases. This leads to models in which the oib sources are in a boundary layer within the mantle. The primordial helium migrates into this layer from below. The interpretation of the rare-gas data is still quite controversial. It is often argued that the upper mantle is a well-homogenized reservoir, but the data indicate heterogeneities on scales ranging from 10° to 106 m. The 206Pb/204Pb ratios in the oceanic m antle range from 17 to 21, which is similar to the range in most continental rocks. The degree of mixing cannot be directly inferred from these data unless the size and composition of the heterogeneities and the time of their introduction into the system are known. The relative uniformity of Nb/U and Ce/Pb ratios in the otherwise heterogeneous morb and oib sources indicates that this reservoir was indeed homogenized after the separation of the continental crust, and that the observed isotopic and chem ical heterogeneities were introduced subsequently. Overall, the results are consistent with, but do not prove, a layered mantle where the upper layer contains both morb and oib sources, and the lower, primitive mantle is not sampled by present-day volcanism. Alternative models such as those involving a chemically graded mantle have not been sufficiently explored.