Subduction sediment-lherzolite interaction at 2.9 GPa: effects of metasomatism and partial melting

Abstract

We present the results of analogue experiments carried out in a piston–cylinder apparatus at 750–900°C and 2.9 GPa aimed to simulate metasomatic transformation of the fertile mantle caused by fluids and melts released from the subducting sediment. A synthetic H2O- and CO2-bearing mixture that corresponds to the average subducting sediment (GLOSS, Plank, Langmuir, 1998) and mineral fractions of natural lherzolite (analogue of a mantle wedge) were used as starting materials. Experiments demonstrate that the mineral growth in capsules is controlled by ascending fluid and hydrous melt (from 850°C) flows. Migration of the liquids and dissolved components develops three horizontal zones in the sedimentary layer with different mineral parageneses that slightly changed from run to run. In the general case, however, the contents of omphacite and garnet increase towards the upper boundary of the layer. Magnesite and omphacite (± garnet ± melt ± kyanite ± phengite) are widespread in the central zone of the sedimentary layer, whereas SiO2 polymorph (± kyanite ± phengite ± biotite ± omphacite ± melt) occurs in the lower zone. Clinopyroxene disappears at the base of lherzolite layer and the initial olivine is partially replaced by orthopyroxene (± magnesite) in all experiments. In addition, talc is formed in this zone at 750°C, whereas melt appears at 850°C. In the remaining volume of the lherzolite layer, metasomatic transformations affect only grain boundaries where orthopyroxene (± melt ± carbonate) is developed. The described transformations are mainly related to a pervasive flow of liquids. Mineral growth in the narrow wall sides of the capsules is probably caused by a focused flow: omphacite grows up in the sedimentary layer, and talc or omphacite with the melt grow up in the lherzolite layer. Experiments show that metasomatism of peridotite related to a subducting sediment, unlike the metasomatism related to metabasites, does not lead to the formation of garnet-bearing paragenesis. In addition, uprising liquid flows (fluid, melt) do not remove significant amounts of carbon from the metasedimentary layer to the peridotite layer. It is assumed that either more powerful fluxes of aqueous fluid or migration of carbonate-bearing rocks in subduction melanges are necessary for more efficient transfer of crustal carbon from metasediments to a mantle in subduction zones.