Mantle carbon fluxes and reservoirs: A status update

Abstract:

Carbon masses in and fluxes from the mantle are topics of intense community interest, but remain uncertain. Recent (2010-2015) estimates of ridge carbon fluxes and total mantle C concentrations each vary by a factors of 30 (0.3-8.4 X 10¹⁴ g CO₂/year) and 38 (20-765 ppm C), respectively. Early estimates of C fluxes at ridges depended chiefly on C/³He ratios, but the recognition of kinetic C/He fractionation during degassing renders these estimates unsound. Methods based on CO₂/Nb and CO₂/Ba ratios benefit from a greatly expanded set of observations of minimally degassed MORB glasses and melt inclusions, as well as new constraints on the Nb and Ba concentrations of mean oceanic crust. Quite liberal bounds on MORB CO₂/Nb and CO₂/Ba of 550±250 and 125±50, combined with the ALLMORB (Gale et al. 2013) or global average MORB of Arevalo and McDonough (2010) concentrations of Nb and Ba yield fluxes of 2.1±1.3 X 10¹⁴ g CO₂/year, thereby reducing the ridge flux uncertainty to a factor of 4. The C flux from oceanic islands, however, remains almost entirely unconstrained. For bulk mantle C, estimates based on He and Ar suffer from problems similar to those related to fluxes – C/He/Ar are strongly fractionated by degassing. As the upper mantle reservoir is well-constrained by sampling from MORB, uncertainty derives chiefly from the deep (OIB source) reservoir . High average mantle C (e.g. 300 or 765 ppm C; Dasgupta and Hirschmann, 2010; Marty, 2012) requires a deep C-rich reservoir that is a large fraction of the mantle (>30%), which means that it must be in part primitive (i.e. it is volatile-rich but cannot be chiefly recycled) and is too large to be “hidden”, so should be sampled by OIB. Assuming source Ba and Nb concentrations similar to primitive mantle, bulk mantle with 300 or 765 ppm C require OIB sources with CO₂/Ba >700-1400 and CO₂/Nb>2800-5200, and would produce magmas (with appropriate Ba and Nb concentrations for OIB) with 20-40 wt.% CO₂. These are far beyond any observations for OIB CO₂/Ba and CO₂/Nb, even with the most extensive corrections for deep degassing, and are petrologically inconsistent with the major element compositions of such lavas. The highest plausible C content of the mantle remains poorly constrained, but is undoubtedly far less than the high values among recent estimates, likely by a factor of at 3 or more.