Partial Molar Volume of CO2 in Peridotitic Melt at High Pressure and Applications to Melt Mobility in the Mantle

The partial molar volume of CO2 (VCO2) is a quantity that may be used for elucidating the role of volatiles in silicate melt at high pressure, such as magma mobility during mantle differentiation when CO2 is present. Because CO2 tends to decrease silicate melt density, its presence should cause crystal-liquid density crossovers to occur at higher pressures than for non-carbonated silicate melts with the same major element composition. Experimental data presented here are crucial to understanding the driving force of carbonated magma eruption, such as kimberlite and nephelinite, especially in the pressure range of 0-3 GPa where extreme changes are observed in the compressibility and solubility of CO2 in peridotite partial melt.High pressure, sink/float experiments were conducted on a synthetic peridotite composition with ~5 wt.% CO2 added as CaCO3 (DG-5), and no CO2 added (DG-N). The mixtures were placed in Mo capsules with two synthetic forsterite spheres or two San Carlos spheres near the top and bottom of the capsule. Experiments were run in a Walker style multi-anvil using 8 mm TEL.Using the known compressibility of the spheres, the density (ρ) of each melt was determined at neutral buoyancy pressures and temperatures using the third order Birch-Murnaghan EOS. The forsterite crossover occurred at 4.6 GPa for DG-5 (ρ=3.14 g/cm3), and 4.0 GPa for DG-N (ρ=3.12 g/cm3). The San Carlos crossover occurred at 6.1 GPa for DG-5 (ρ=3.29 g/cm3), and ~5.0 GPa for DG-N (ρ=3.26 g/cm3).The VCO2 was determined using a modified version of the Bottinga and Weill (1970) equation: ρ=Σ XiMi / XiVi, the calculated ρ, the analyzed compositions, and the assumption that CO2 remained in the melt during sink/float. Preliminary values for VCO2 are 25.35 cm3/mol at 4.3 GPa and 22.92 cm3/mol at 5.6 GPa, both corrected to 1850°C. Based on the VCO2 at 1 bar (Liu & Lange, 2003) and at 19.5 GPa (Ghosh et al., 2007), the compressibility curve for CO2 may now be better constrained. The calculated curve shows a rapid decrease in VCO2 at low pressures indicating a high compressibility in the upper mantle.Currently, we are exploring electron microprobe and FTIR transmission and reflectance methods to better quantify the CO2 in our experimental run products which contain quench crystals and small pockets of glass. We believe the assumption that CO2 remains in the melt during sink/float experiment is valid because crossover positions differ between the DG-5 and DG-N samples. However, we are investigating this assumption because CO32-, CO2 and CO are not observed in the run products with FTIR transmission and reflectance, and the microprobe results for C are ambiguous. There are several possibilities that we are testing at this time.