Pervasive subduction zone devolatilization recycles CO2 into the forearc

Abstract

The fate of subducted CO₂ remains the subject of widespread disagreement, with different models predicting either wholesale (up to 99%) decarbonation of the subducting slab or extremely limited carbon loss and, consequently, massive deep subduction of CO₂. The fluid history of subducted rocks lies at the heart of this debate: rocks that experience significant infiltration by a water-bearing fluid may release orders of magnitude more CO₂ than rocks that are metamorphosed in a closed chemical system. Numerical models make a wide range of predictions regarding water mobility, and further progress has been limited by a lack of direct observations. Here we present a comprehensive field-based study of decarbonation efficiency in a subducting slab (Cyclades, Greece), and show that ~40% to ~65% of the CO₂ in subducting crust is released via metamorphic decarbonation reactions at forearc depths. This result precludes extensive deep subduction of most CO₂ and suggests that the mantle has become more depleted in carbon over geologic time.

Fig. 1. Equilibrium fluid compositions.

Blue squares show maximum possible fluid CO₂ activity of each sample. The equivalent $X_{{\mathrm{CO}}_2}$ is shown on the right axis. Note the nonlinear relationship between activity of CO₂ and $X_{{\mathrm{CO}}_2}$. Red circles are CO₂ activities achieved in a closed system. The vast majority of samples record water-rich fluids, inconsistent with evolution in a closed system. Two samples are unconstrained, with any CO₂ activity < ~0.95 being plausible (blue arrows). The shaded area shows the 2σ standard deviation range of equilibrium activity values (0.06 ± 0.07), excluding the two unconstrained outliers.

Fig. 2. Carbon and oxygen isotopes.

The 211 isotope measurements of metamorphic carbonate minerals record a wide array of δ¹³C and δ¹⁸O values that differ from unaltered seafloor carbonate. Uncertainties of <0.1‰ (1σ) are smaller than plotted circles.

Fig. 3. Degree of decarbonation.

Blue symbols show the observed CO₂ loss of each sample whereas green symbols show the maximum potential CO₂ loss as a function of initial wt% CO₂. a Percent of CO₂ released. Nearly pure carbonate rocks degas the smallest proportion of CO_2. b The kg CO₂ released per kg rock. This shows a peaked distribution, with intermediate-CO₂-content rocks degassing the greatest mass of CO₂ in both observed and potential datasets.

Fig. 4. Carbon flux.

a The percentage of CO₂ degassed along two linear P–T paths. Both paths begin at 310 °C and 1.0 GPa; path A continues through peak field conditions of 525 °C and 1.5 GPa, while path B passes through 550 °C and 2.0 GPa. Paths A and B end at 650 °C and 1.8 or 2.4 GPa, respectively. A sharp pulse of decarbonation is seen around peak-T conditions. b The constructed moles of CO₂ subducting globally and the amount of CO₂ released at both peak conditions (observed) and up to 650 °C (observed + modeled). The arc volcanic CO₂ flux is shown for reference.

Fig. 5. Global mass balance.

This schematic representation shows the ~40% CO₂ loss recorded in the field and the total predicted 65% CO₂ loss by 650 °C in the forearc; most decarbonation is concentrated in a sharp peak at ~500–550 °C. The remaining ~35% of subducted CO₂ in the slab continues to greater depths where it could be released by processes such as dissolution or melting, or retained in the slab past the subarc. Global estimates are based on observations from the Cycladic Blueschist Unit.