Isotopic evolution of the mantle: a model

We investigate a model of mantle evolution which acknowledges the importances, in early Earth history, of large degrees of partial melting and crystal fractionation rather than, as in recent models, small degrees of partial melting operating on primary mantle. The major source regions are in the upper mantle and evolve from a primitive mantle melt by crystal fractionation and the formation of chemically distinct cumulate layers. The Rb/Sr and Sm/Nd ratios in melts are relatively unfractionated in processes involving large (> 15%) degrees of partial melting. The ratios are even less fractionated in orthocumulates forming from these melts. Small degrees of partial melting, or removal of residual fluids after nearly complete crystalization of cumulate layers, are the processes that lead to significant fractionation. The importance of these processes increases as the upper mantle cools. The observation of nearly primitive isotopic ratios does not necessarily imply the presence of ancient primitive undifferential reservoirs. The transfer of intercumulus fluids from a crystalizing cumulate layer gives time-dependent Rb/Sr and Sm/Nd ratios in both the donor and receiving reservoirs since the late-stage fluids become more enriched as crystalization proceeds. Enrichment in such a system is progressive with time. Time-integrated Nd/Sm and Rb/Sr ratios, therefore, need not depart much from primitive ratios even though current values for LILE concentration, Rb/Sr, Nd/Sm, etc., may be large. The time constant for LILE enrichment appears to be 2 Gyr, about the same as required for the U/Pb system and the same as the characteristics time associated with mantle cooling. The mantle reservoirs may be much more ancient than generally assumed