Deep carbon cycle constrained by carbonate solubility

Abstract

Earth's deep carbon cycle affects atmospheric CO₂, climate, and habitability. Owing to the extreme solubility of CaCO₃, aqueous fluids released from the subducting slab could extract all carbon from the slab. However, recycling efficiency is estimated at only around 40%. Data from carbonate inclusions, petrology, and Mg isotope systematics indicate Ca²⁺ in carbonates is replaced by Mg²⁺ and other cations during subduction. Here we determined the solubility of dolomite [CaMg(CO₃)₂] and rhodochrosite (MnCO₃), and put an upper limit on that of magnesite (MgCO₃) under subduction zone conditions. Solubility decreases at least two orders of magnitude as carbonates become Mg-rich. This decreased solubility, coupled with heterogeneity of carbon and water subduction, may explain discrepancies in carbon recycling estimates. Over a range of slab settings, we find aqueous dissolution responsible for mobilizing 10 to 92% of slab carbon. Globally, aqueous fluids mobilise [Formula: see text]% ([Formula: see text] Mt/yr) of subducted carbon from subducting slabs.

Fig. 1. Fluid–carbonate mineral interactions in the deep carbon cycle.

White headed black arrows indicate carbonate flux and blue arrows water flux. Blue shaded areas indicate water-rich regions. The melting of carbonated igneous oceanic crust is not shown as it starts at depths of 300 km. The image is to scale, apart from the thickness of oceanic sediments that has been exaggerated.

Fig. 2. Experimental determinations of carbonate mineral solubilities.

Solubility is reported as a function of pressure and temperature in water or aqueous solutions of different ionic strength: a present data, b literature data. Only studies at pressures of 1 GPa or more are included. Full symbols indicate upper limits for solubility. The solid line represents a solubility line for dolomite at 400 °C in 1 m NaCl aqueous solution. Since no magnesite crystals dissolved, the dashed line represents an upper solubility limit for magnesite at 400 °C. All solubility lines are guides to the eye. Error bars shown for our data represent estimated standard deviation. Note the log scale.

Fig. 3. Heterogeneity of subduction zone compositions and processes.

Differences in water (a) and carbon^, (b) flux among subduction zones and corresponding C recycling efficiency (c). The water flux estimates assume moderate mantle hydration.