Boron release and transfer induced by phengite breakdown in subducted impure metacarbonates

Abstract White mica is an important reservoir of boron (B) in subducted metasedimentary rocks. To quantify the effect of phengite breakdown on B release and transfer, we investigated the replacement processes of phengitic mica and associated element redistribution in exhumed ultra-high pressure (UHP) impure marbles from the Dabie terrane, China. Three types of microstructures, which are related to the liberation, transport, and precipitation of B, are recognized: (1) Type-I is the pseudomorphic replacement of phengite by the assemblage of biotite + plagioclase ± quartz ± epidote. The assemblage maintains a flaky shape of pre-existing phengite, which only rarely occurs as a relict phase in the product assemblages. Phase equilibria modelling indicates that the conditions of replacement reactions are 1.2–1.3 GPa at ~600 °C, with a relatively low XCO2 of 0.04–0.16. (2) Type-II is characterized by the local occurrence of fine-grained tourmaline (Tur-F) along the outer margin of Type-I texture domains, which represents the in situ precipitation of B released by the replacement reactions. (3) Type-III texture involves the irregularly shaped intergrowth of tourmaline (Tur-In) + quartz in the calcite matrix, isolated from the Type-I and Type-II texture domains. The Type-III minerals are in equilibrium with diopside and plagioclase and most likely represent the precipitation from a “transported”, B-bearing fluid. Mineral trace element analyses indicate that relict phengite in Type-I texture has significantly higher B contents (284–464 μg/g) than all of the product minerals (2–26 μg/g). A mass balance estimate indicates that the majority (>90%) of B in the reactants is lost during the process of phengite replacement. The calculations in terms of mineral-fluid partition indicate that the fluid involved in the replacement reactions has B contents of 1893–6858 μg/g. In situ B isotopic analyses indicate that Tur-F and Tur-In have a similar variation range of δ11B (+4 ~ +7‰). Both types of low-pressure tourmaline have lower δ11B values than the previously reported eclogite-facies tourmaline (reaching +15‰) from phengite-bearing marbles in the same study area. This result indicates two episodes of tourmaline growth and B transfer. Thus, a two-stage model of B loss from phengite during slab exhumation is proposed. Our results reveal that the breakdown of white mica in exhumed metasedimentary rocks can cause a strong release of B, which would largely influence the B budget at convergent plate margins.