238U–230Th disequilibria and mantle melting processes: a discussion

This paper discusses the implications of 238U–230Th disequilibria in MORB and OIB on mantle melting. It is shown that melting rates can only be calculated if the melting process is not batch melting. At present, however, the uncertainties in U and Th partition coefficients are such that, in most cases, batch melting cannot be discarded. Moreover, a simple argument based on the Th–Sr isotope correlation, suggests that batch melting might best approximate the actual melting process in the mantle, at least in the case of OIB. The existence of mantle heterogeneities (e.g., the “marble-cake” model of Allegre and Turcotte [Allegre, C.J., Turcotte, D., 1986. Implications of a two-component marble-cake mantle. Nature 323, 123–127]), when taken into account, suggests that complex melting models might not be necessary to interpret 238U–230Th data. In particular, preferential melting of garnet pyroxenites could explain the origin of OIB, whereas MORB could derive from larger degrees of melting of garnet lherzolites. However, the large 231Pa and 226Ra excesses found in MORB are better explained by dynamic melting models, if they are actually produced through crystal-melt partitioning during mantle melting. In any case, it is worth emphasizing that derivation of melting rates is strongly model dependent, and should be taken with caution.