Quantifying the nature of Earth’s deep carbon cycle is one of DCO’s most complex yet critical decadal goals. Research addressing this goal involves myriad diverse measurements, predictions, and calculations, as well as an appreciation for geological changes on Earth through deep time.
In a new paper appearing in the Proceedings of the National Academy of Sciences this month, DCO’s Peter Kelemen (Lamont-Doherty Earth Observatory, USA) and Craig Manning (University of California Los Angeles, USA) present new analyses of Earth’s deep carbon cycle . The paper is part of a special series of Inaugural Articles by members of the National Academy of Sciences, and celebrates Kelemen’s election to the Academy in 2014. Their paper refines previous estimates of Earth's carbon flux, and suggests that, contrary to some previous estimates, the total concentration of carbon in the mantle lithosphere, atmosphere, oceans, and crust may be gradually increasing over time.
To address deep Earth flux of carbon, Kelemen and Manning looked at how carbon behaves in a variety of tectonic settings. They assessed how much carbon is taken up by Earth’s crust during hydrothermal processes at or near mid-ocean ridges, and how much carbon is released into the atmosphere through volcanic degassing and diffuse venting. They also considered how carbon in fluids and melts moves during subduction and what alteration processes affect carbon flux near tectonic boundaries.
Kelemen and Manning write “We were inspired to undertake this review by new data on carbon mobility in subduction zones, new data and theory on the solubility of carbon in aqueous fluids at elevated pressure and temperature, and new insights into the formation of buoyant diapirs of carbon-rich metasediments in subduction zones. Also, as erstwhile metamorphic petrologists, we hoped to quantify the amount of carbon stored in the lower crust and lithospheric mantle, which has been omitted from many previous studies. Thus, from the outset, we were focused on refining the ‘lower bound’ estimate for the flux of subducted carbon that is recycled into the convecting mantle. And indeed, we found that the ‘lower bound’ is close to zero. This expands the range of allowable models for the evolution of the Earth’s carbon cycle over geologic time.”
Their work includes compilation of existing data on samples taken at sites around the world. Kelemen and Manning added to these datasets new analyses of partially serpentinized peridotites from the Samail ophiolite in the Sultanate of Oman, and new understanding of carbon mobility in aqueous fluids and buoyant diapirs at elevated pressure and temperature under subduction zone condition. Their results suggest that mantle carbon released at mid-ocean ridges and ocean island volcanoes is not necessarily returned to the mantle through subduction. Thus, over the course of Earth’s history, the total carbon content of the mantle lithosphere, crust, oceans, and atmospheres has increased.
One of the most scientifically important mantle reservoirs of carbon is diamond. Diamonds form in the mantle from both primordial carbon (carbon that has never travelled to Earth’s surface) and recycled carbon (carbon that has cycled from the mantle to the surface and back again). Diamonds at the surface offer valuable insight into the composition of Earth’s interior. Isotopic measurements of the carbon in diamond provide information about the mineral’s age, formation conditions, and whether the carbon is primordial or recycled. Kelemen and Manning corroborated their findings with a calculation based on the known characteristics of Earth’s diamonds, and confirmed that the amount of recycled carbon in diamonds is orders of magnitude (300 thousand times) less than the amount of carbon found in near-surface reservoirs.
This work represents an important step in refining our understanding of Earth’s deep carbon cycle, and will inform numerous research efforts ongoing as part of DCO.
Images: header: The Samail ophiolite, Oman. Credit: Katie Pratt. Above center: Major fluxes of carbon estimated in this paper, with values from Dasgupta and Hirschmann  for comparison. Adapted from Figure 5 in Kelemen and Manning by Josh Wood.