Uranium–thorium disequilibria and partitioning on melting of garnet peridotite

Abstract

THE abundances of isotopes in the ²³⁸U decay series can be used as both tracers and chronometers of magmatic processes. In the subsolidus asthenosphere, the activity of each daughter isotope (defined as the product of its concentration and decay constant, and denoted by parentheses) is assumed to be equal to that of its parent. By contrast, (²³⁰Th/²³⁸U) is greater than unity in most recent mid-ocean-ridge and ocean-island basalts¹, implying that thorium is more incompatible (that is, it is partitioned into the melt phase more strongly) than uranium. Melting of spinel peridotite cannot produce the (²³⁰Th) excesses, because measured partition coefficients for pyroxenes and olivine demonstrate that uranium is more incompatible than thorium for this rock². Here I report garnet–melt partitioning data which show that for this mineral-melt pair thorium does behave more incompatibly than uranium, thus supporting the suggestion that mid-ocean-ridge basalts (MORB) are produced by melting initiated at depths where garnet is stable^3–6. Using these data, I show that the observed (²³⁰Th/²³⁸U) ratios of MORB and most ocean-island basalts can be explained by slow, near-fractional melting initiated in the garnet stability field.