Hydrogen Isotope Studies of Submarine Basalts

California Institute Of Technology, Pasadena CA

Investigators

Abstract

Eiler EAR-0208127 Hydrogen is arguably the most important minor element in the solid earth: It dominates the melting behavior and rheology of mantle rocks; it is the principle solvent for metasomatic transport in the crust and mantle; its concentration in silicate minerals controls rates of diffusion of other elements; and its abundance in silicate melts strongly influences crystallization-differentiation. However, the geochemistry of hydrogen is less well described than that for virtually any other element (and certainly any of similar importance): Reasonable estimates of the total abundance of H in the earth span a range of nearly an order of magnitude; concentrations, residence times, and sources and sinks of hydrogen in major mantle reservoirs are poorly known or only guessed at; and partitioning of hydrogen between minerals and melts has yet to be described with the same experimental sophistication given to studies of other minor elements. We believe that our understanding of the global hydrogen cycle can be advanced by studying the hydrogen isotope composition (i.e., D/H ratio) of well characterized suites of submarine basalts whose origin and evolution is influenced by mantle hydrogen. The potential for such studies has been recognized for many years, but has not been realized due to sparse data, the large proportion of isotopic measurements on samples that are not characterized for other important geochemical parameters, and lack of inter-laboratory standardization. We propose to characterize the variation in D/H ratio among submarine basalts associated with the Mariana arc and its back arc basin, which are a key set of samples for our current understanding of the role of water in magma genesis at convergent margins. We will approach this study using a new analytical method for determining D/H ratios and H contents of small quantities (10's to hundreds of micrograms) of hydrous solids and glasses at a rate approximately 10x that of conventional analyses but with similar accuracy and precision. This method will let us produce relatively large, well standardized and well replicated data sets in a reasonable time frame.