Oxygen Isotope Fractionation Among Melts, Minerals and Rocks

Abstract

The key factor in interpreting oxygen isotope measurements of natural materials is knowledge of the relative partitioning (or fractionation) of the oxygen isotopes between coexisting phases and its dependence on temperature. Statistical mechanical and empirical chemical-bond calculations stress the importance that cation-oxygen bond type and mineral structure plays in governing the oxygen isotope fractionation behaviour of crystalline materials. The early studies of Taylor and Epstein (1962) and Garlick (1966) on igneous and metamorphic rocks showed that isotopic fractionations in silicates could be substantially understood in terms of linear combinations of the various bond-types (Si-O; A1-O; M-O) present in each silicate structure. An extension of this reasoning is that melts should generally exhibit fractionations similar to their chemically identical crystalline phase, and therefore that the partitioning behaviour of normative major minerals would be sufficient to describe that of melts. On the other hand, naturally occurring partial melts will be compositionally different from their source rocks and thus melt/ rock isotopic fractionations could arise from the chemical differentiation during melting. Even so, it has generally been assumed that at the temperatures of mantle melting, melt/residue fractionation factors will be small (less than a few tenths of a per mil) and not petrologically significant. This simple view of melt/crystalline solid fraetionation behaviour needs to be critically examined. The increased level of detail available using new techniques such laser fluorination means that effects that were previously unnoticed or too difficult to characterize (such as small isotopic differences between magma types), are now becoming recognized, and they cannot be confidently interpreted without equally detailed understanding of partitioning behaviour. Figure 1 summarizes the results of experiments we have made 1 atm and 650-950~ on exchange between CO2 and minerals and melts/glasses of identical composition. Experimentally measured CO2-albitic glass/melt and CO2-crystalline albite are identical within experimental error and preliminary experimental work on the CO2-anorthitic glass also suggests a close correspondence between the fractionation properties of anorthitic glass and crystalline anorthite. Moreover, the CO2-glass/mineral fractionation factors for these phases are reasonably well approximated by