Diamond growth from subducted carbon implied by correlated δ18O-δ13C variations in diamonds and garnet inclusions

Much of our knowledge of the deep-Earth carbon cycle is derived from studies of diamond. The sources of carbon in the mantle and the mechanisms of transport and precipitation as diamond, however, are not entirely understood. Due to the chemical purity of diamond, scientific effort has focussed on syngenetic mineral inclusions and their relationship to their diamond hosts. For example, it is well known that, on a worldwide scale, diamonds with eclogitic inclusions have a distinct δ13C distribution when compared to more abundant peridotitic diamonds. Eclogitic diamonds have a distribution that extends from mantle-like δ13C values (ca. -5%), to very light carbon (<-20%). Strong 13C depletion has been explained by either invoking subducted organic carbon, or through high temperature isotopic fractionation of mantle carbon. Here we report high-precision SIMS δ18O measurements (2σ±< 0.3%) of eclogitic garnet inclusions in diamonds from the Damtshaa mine (Orapa cluster, Botswana). The δ13C values of the host diamond were determined to have a wide range (-4.4% to -18%; Deines et al. 2009; Lithos v.112 p776). From 15 inclusions, the δ18O variations range from +4.8 to +8.8 %. The relative 18O abundances are negatively correlated with the δ13C of the host diamonds, suggesting a link between high δ18O host rocks and low δ13C diamonds. Although fractionation of δ13C values is possible at high temperature, δ18O values are susceptible only to very small high temperature fractionations. For example, Cartigny et al. (2001, EPSL v.185 p85) suggested that CO2 degassing from a carbonate-bearing melt prior to diamond precipitation may be responsible for a δ13C distribution of eclogitic diamonds worldwide that is skewed to 13C depleted compositions. Our data place new constraints on that model. Depending on the C/O ratio of the melt, CO2 degassing will either have a negligible effect on the δ18O of the residual melt, or (at high C/O) induce a positive correlation between δ18O and δ13C, contrary to the negative correlation documented here. We suggest instead that low δ13C diamonds at Damtshaa are associated with sources whose protoliths were exposed to low-temperature alteration (increasing their δ18O). We relate the connection between low δ13C and elevated δ18O to an association between low-temperature alteration and the concentration of reduced organic carbon in the oceanic protolith. The spread in δ13C values, then, is a consequence of mixing between mantle carbon and a 13C depleted carbon reservoir concentrated in eclogite that has been exposed to extensive sea floor weathering (reflected in elevated δ18O ).