Sediment subduction: Geochemical Implications for Mantle Metasomatism

Abstract

Sediment subduction is an important contributor to the geochemical refertilisation of the lithospheric mantle, with several implications for the origin of arc magmas. The trajectory of subduction carries continent-derived (terrigenous) sediment into a pressure-temperature (P-T) regime whereby volatile-rich phases are broken-down to release volatile-rich fluids and/or hydrous silicate melts. These fluids/melts interact with the overlying sub-arc mantle peridotite during metasomatic reactions that impart an overall 'crustal' geochemical signature to the mantle wedge. Despite the acknowledged importance of subducted terrigenous sediments in this process, fundamental questions regarding the mechanisms and consequences of the metasomatic reactions involving sediment-derived fluids/melts, remain unanswered.

The first part of this experimental study examines how differences in source composition, mineralogy and volatile content (i.e. H2O, CO2) manifest in the nature and composition of 'pristine' sediment-derived fluids/melts and assesses the extent of chemical fractionation during partial melting. The second part of this study examines how these same sediment-derived fluids/melts might react with the sub-arc mantle wedge. To that end, a series of piston-cylinder experiments were conducted at P-T conditions relevant to the shallow sub-arc mantle (800-1100 degC, 2.0 GPa) using natural, starting materials representative of either: (i) water-rich, carbonate-poor sediment (W-Sed), (ii) carbonate-rich, water-poor sediment (C-Sed), and (iii) water-poor, carbonate-poor 'average' continental crust (D-sed) with a harzburgite xenolith as the depleted mantle analogue. The sediment melting experiments not only serve to define the nature of fluids/melts generated from compositionally variable sedimentary sources, but also provide an initial composition for the fluids/melts that react with the overlying peridotite layer in the 'sandwich' experiments.

'Sediment melting' begins between 750 and 820 degC at 2.0 GPa, and produces peraluminous, granitic (sensu stricto), high-K to shoshonitic melts that contain up to 10.3 wt.% volatile content; they are enriched in LILE, Th and U and depleted in HFSE and HREE relative to the whole-rock sediment, but their trace-element abundance patterns remain broadly similar-to their source.

In the 'sandwich' experiments, the hydrous silicate melts that are produced in the sediment layer are immediately consumed in olivine dissolution reactions that form metasomatic orthopyroxene as the main product phase, but more importantly lead to the formation of a relatively thin polymineralic reaction zone composed of either phlogopite-bearing orthopyroxenite or websterite. This layer is impervious to infiltration by the hydrous silicate-melts and effectively 'shields' any remaining sediment melt from further reacting with peridotite. Furthermore, consumption of melt during melt-rock reactions give rise to a C-O-H-rich 'daughter' fluid, which - as well as any 'primary' C-O-H fluid or CO2 vapor phase produced in the sediment layer - pervasively infiltrates the peridotite layer, carries along dissolved solutes (e.g. Cl-, Na+, Si4+, K+, Ca2+, Al3+ and S2-) and results in the precipitation of 'secondary' metasomatic minerals in the peridotite (e.g. orthopyroxene, magnesite, dolomite, phlogopite, diopside and spinel) via modal metasomatism, and the subtle chemical modification of 'primary' minerals (e.g. olivine, pyroxenes, spinel) via cryptic metasomatism. Specific chemical signatures in minerals within the peridotite layer - for example, Zn in olivine and Ti-Ba-Pb in orthopyroxene - can be used to 'fingerprint' cryptic metasomatism by continental crustal-derived C-O-H fluids. More importantly, metasomatism by sediment-derived hydrous silicate melts enriches peridotite in B, LILE, HFSE, Li, Th and U, whereas C-O-H fluids enrich the mantle wedge in H2O, Cl, Be and chalcophile elements (Ag, As, Cu, Sb, Sn, Zn, Pb). Show more