Experimental Constraints on CO2 Solubility in Terrestrial Magma Oceans

Abstract:

In order to constrain the distribution of volatiles, i.e., carbon, in the terrestrial planets today, we must understand its fate during the magma ocean (MO) stage. In particular, we must know the solubility of C in the silicate MO because this has key bearing on the relative distribution of C in all the major reservoirs, i.e., the metallic core, the early atmosphere, and the mantle [1]. In an MO, C can dissolve as oxidized (CO₃^2-) and reduced (CO, CH₄) species, which varies as a function of fO₂; yet, for a MO with fO₂ between IW-2 to +2, the dominant form of C is CO₃^2- [2]. It is therefore critical to determine the CO₂ solubility in ultramafic/peridotitic melt at reducing conditions. Previous work attempted to constrain the C content of ultramafic melt chiefly by extrapolating measured CO₂ solubility in mafic/basaltic compositions using NBO/T [3]. However, using only NBO/T may not reliably take into account the effects of different network modifiers, like Ca²⁺ and Mg²⁺, on CO₂ solubility, as their relative roles remain ambiguous [4,5].

To constrain the CO₂ solubility in ultramafic compositions at graphite saturation, we determined CO₂ contents of 4 silicate compositions spanning between natural basalts and peridotite (NBO/T from 0.8 to 1.5, MgO from 8 to 21 wt.%, and Ca# from 0.4 to 0.2), using Pt/Gr capsules and a piston cylinder device. Experiments were conducted at 1.0 GPa and 1600 °C. FTIR was used to measure dissolved CO₃^2- and H₂O (≤0.5 wt.%) contents. For graphite-saturated glasses at ~IW+1.6, CO₂ contents increase from ~0.4 to 0.9 wt.% with increasing NBO/T and decreasing Ca#.

These experimental results, combined with previous data, allow us to construct a more complete CO₂ solubility model of reduced silicate melts as a function of P, T, fO₂, and composition. By determining CO₂ solubility on a continuum of compositions between basalt and peridotite, we will more accurately constrain the tradeoffs between core-MO partitioning and MO-atmosphere interactions and begin to place constraints on the initial carbon content of the mantle soon after MO crystallization.

[1] Dasgupta (2013) RiMG. [2] Li et al. (2015) EPSL. [3] Brooker et al. (2001) CG. [4] Chi et al. (2014) GCA. [5] Stanley et al. (2012) AM.