Carbon-saturated monosulfide melting in the shallow mantle: solubility and effect on solidus

Abstract

We present high-pressure experiments from 0.8 to 7.95 GPa to determine the effect of carbon on the solidus of mantle monosulfide. The graphite-saturated solidus of monosulfide (Fe0.69Ni0.23Cu0.01S1.00) is described by a Simon and Glatzel (Z Anorg Allg Chem 178:309–316, 1929) equation T (°C) = 969.0[P (GPa)/5.92 + 1]0.39 (1 ≤ P ≤ 8) and is ~80 ± 25 °C below the melting temperature found for carbon-free conditions. A series of comparison experiments using different capsule configurations and preparations document that the observed solidus-lowering is owing to graphite saturation and not an artifact of different capsules or hydrogen contamination. Concentrations of carbon in quenched graphite-saturated monosulfide melt measured by electron microprobe are 0.1–0.3 wt% in monosulfide melt and below the detection limit (<0.2 wt%) in crystalline monosulfide solid solution. Although there is only a small amount of carbon dissolved in monosulfide melts, the substantial effect on monosulfide solidus temperature means that the carbon-saturated monosulfide (Fe0.69Ni0.23Cu0.01S1.00) solidus intersects continental mantle geotherms inferred from diamond inclusion geobarometry at 6–7 GPa (~200 km), whereas carbon-free monosulfide (Fe0.69Ni0.23Cu0.01S1.00) solidus does not. The composition investigated (Fe0.69Ni0.23Cu0.01S1.00) has a comparatively low metal/sulfur (M/S) ratio and low Ni/(Fe + Ni), but sulfides with higher (M/S) and with greater Ni/(Fe + Ni) should melt at lower temperatures and these should have a broader melt stability field in the diamond formation environment and in the continental lithosphere. Low carbon solubility in monosulfide melt excludes the possibility that diamonds are crystallized from sulfide melt. Although monosulfide melt can store no more than 2 ppm C in a bulk mantle with 225 ppm S, melts with higher M/S could be a primary host of carbon in the deeper part of the upper mantle. For example, the storage capacity of C in sulfide melts in the deep upper mantle (~400 km) for a depleted mantle domain (MORB source, 120 ± 30 ppm S) is estimated to be \(57 \pm_{30}^{63}\) ppm, and so all the C could be in a sulfide melt. In an enriched (OIB source, 225 ± 25 ppm S) mantle domain, the C stored in sulfide melt in the deep upper mantle is estimated to be \(86 \pm_{44}^{92}\) ppm, which would amount to about half the available carbon.