Reports: G8

47664-G8 Carbon Sequestration in the Earth's Mantle

Wendy R. Panero, Ohio State University

Recent experiments have shown that the Earth's upper mantle can dissolve trace amounts of carbon into the silicate minerals of the mantle, with concentrations increasing with increasing pressure. These results have led to growing interest in the planet's bulk carbon budget and implications for the mantle's control of our climate. Although it is well established that the vast majority of the Earth's oil and natural gas resources form from surface-derived organic carbon, it has long been recognized that the carbon in the Earth's interior has direct control on surface carbon budgets.

We report results on the potential for carbon sequestration in deep mantle materials, such as those found in the Earth's lower mantle. The participation of the deep mantle in the global carbon cycle and its ability to sequester carbon over billion-year time scales depends upon the mineralogical host for carbon. Density-functional theory calculations for MgCO3-magnesite and structures with tetrahedrally coordinated carbon reveal the stability of magnesite up to ~80 GPa, with a bulk modulus of 110 (±2) GPa. Magnesite undergoes a structural transition to a pyroxene-like structure at ~80-100 GPa, with a density increase of 4.5-7.1%. Combined with thermodynamic models for the MgSiO3-MgCO3 system, the inter-solubility of MgCO3 with MgSiO3 orthoenstatite and perovskite constrains the carbon content in the silicates to an upper bound of 4 and 20 ppm (wt), respectively. The carbon content in lower mantle silicates is estimated to be no more than 1% of the mantle's total carbon budget for degassed regions, such that in even the mantle's most depleted regions, most carbon must be stored in carbonates or diamond.

The results of this work highlights the importance of the oxidation state of carbon in the deep mantle, paired with ongoing debate as to the oxidation state of iron in the deep mantle. Recent experimental results indicating that iron in perovskite may undergo an "autoredox" reaction of 3Fe2+=2Fe3++Fe0 has led to the suggestion that the lower mantle is buffered by reduced iron. Alternatively, because this autoredox reaction is independent of oxidation potential of the system, this reaction may lead to unexpected behaviors of those minor and trace elements in the mantle with multiple oxidation states. As such, we have initiated a series of experiments to probe the relative oxygen buffers: Siderite-Diamond-Iron (SDI) compared to Iron-WŸstite (IW). As the primary mineral host, is likely controlled by the oxidation state of carbon, the first-order constraint of carbon relative to the dominant iron is crucial. These can be approximated through relative equations of state of the buffer minerals, but uncertainty in the equations of state of diamond and of siderite and wŸstite across the iron spin transition introduces significant uncertainty. The volume collapse of siderite through the iron spin transition at mid-mantle depths opens the possibility for stabilization of carbonate at depths greater than ~1000 km.

Experiments at pressures of 10-70 GPa and temperatures of 1500-2500 K of combined buffers show the formation of diamond in all experiments. This result would indicate that the volume collapse of siderite is insufficient to stabilize carbonate under these conditions. However, combined x-ray diffraction of the bulk experiment, combined with micro Raman spectroscopy and TEM of cross-sections of the samples show significant spatial variations, with those portions of the sample at the highest temperatures showing carbonate in equilibrium with iron oxides and diamond, and no remaining free iron. This indicates that at temperatures above ~2200 K (representative of mantle temperatures at that depth) that SDI is above IW. As such, the stability of carbonate in the mantle is then a function of how high above the IW buffer the mantle at those depths may be.