Skip Navigation | ANU Home | Search ANU | Directories
The Australian National University
Research School of Earth Sciences
Printer Friendly Version of this Document
RSES SITE SEARCH
Untitled Document

Interaction of hydrous granitic melts with carbonates at 4.5 GPa: Implications for metamorphic diamond formation and devolatilisation in subduction zones

J. Hermann and D.H. Green

Interlayered pelites and carbonates (marls) are common in subducted continental crust and oceanic sediments. The interaction of hydrous granitic melts produced in metapelites with carbonates has been investigated by piston cylinder sandwich experiments at 4.5 GPa. The carbonate layer consists of natural dolomite and a synthetic simplified pelite composition (KCMASH) with additional 1.2 wt% H2O was used for the pelite layer. At 1000°C, 4.5 GPa the paragenesis in the carbonate-free system consists of phengite + garnet + clinopyroxene + kyanite + coesite + hydrous granitic melt. Dolomite embedded in the pelite reacts with the hydrous silicate melt and produces the paragenesis garnet + clinopyroxene + dolomite + liquid. Qualitative mass-balance constraints the liquid composition to 60 wt% CO2, about 10 wt% of H2O, K2O and CaO and very small amounts of SiO2, MgO and Al2O3. It is likely that significant amounts of the carbon are dissolved as (CO3)2-. This liquid therefore reflects rather a carbonatite-like liquid than an aqueous CO2-H2O fluid.

In the pelite adjacent to the dolomite layer, an increasing abundance of garnet, kyanite and melt and a decreasing amount of clinopyroxene, phengite and coesite has been observed. This suggests that the presence of carbonate enhances melt production in the pelite. No clear separation between the hydrous granitic melt and the carbonatite-like liquid has been observed indicating that they are most probably completely mixable at the experimental conditions. The presence of a carbonatite-like liquid in the carbonate layer and a hydrous granitic melt in the pelite domain could be a clue to understanding formation of metamorphic diamond in subducted crust. The interaction of the hydrous granitic melt with dolomite produces CO2 and/or (CO3)2-. It is likely that a hydrous granitic melt is more reduced than carbonates. Because of the curvature of the carbon saturation surface in a f(O2) vs liquid composition diagram, the liberated CO2/(CO3)2- drives the hydrous granitic melt into the "liquid+diamond" field, leading to the precipitation of diamond. Another way of diamond formation could include the reaction of the carbonatite-like liquid with the hydrous granitic melt. Because the carbonate—like liquid is likely to retain a much higher solubility of carbon than a hydrous granitic melt, diamond could precipitate during mixing of the two different liquids.Further carbonate solubility experiments were carried out in a natural pelite composition with addition of different amounts of carbonate.

At 900°C, 4.5 GPa, 10, 20, and 30% of a carbonate mix consisting of 20% calcite and 80% dolomite were run together with a pelite containing 6.8% H2O. All runs contained phengite + garnet + onphacite + coesite + liquid ± kyanite. The 10% of carbonate was completely dissolved in the liquid. In the runs with 20% and 30% of carbonate added, dolomite but no aragonite was found. This indicates that the whole CaCO3 component of the carbonate was dissolved in the liquid. The liquid is characterised by a quenched hydrous granitic glass containing numerous bubbles, probably exsolved CO2, and small quench carbonates. This further supports the hypothesis, that a carbonatite-like liquid and a hydrous silicate melt are mixable at 4.5 GPa and 900-1000°C. The melting reaction in marls can be summarised asphengite + omphacite + coesite + aragonite + H2O -> garnet + liquidMass balance indicates that 1.5-2 times more CO2 than H2O can be dissolved in hydrous granitic melts. We therefore propose that hydrous granitic melts are capable of transferring not only H2O but also significant amounts of CO2 from subducted sediments to the mantle wedge. The physical properties, the ability to transport trace elements and the interaction with the mantle wedge of such melts are yet to be determined.