Subducted Carbon in the Earth’s lower mantle: The fate of magnesite

Subduction of carbon-bearing phases throughout Earth’s history may be an important mechanism of sourcing carbon to the Earth’s lower mantle. As carbon has very low solubility in mantle silicates, it is primarily present in accessory phases such as carbonates, diamond, or metal carbides. Previous studies indicate that more than half of the carbonate contained in the oceanic crust may survive metamorphism and dehydration in the sub-arc and reach the lower mantle, even though the oxygen fugacity in the deep mantle may not favour their stability [1]. Indeed, the presence of carbonate in ultra-deep diamond inclusions provides evidence for carbonate subduction at least down to the transition zone [2].

The carbonate phases present in the lower mantle depend on their bulk composition, the oxygen fugacity, and on their stability at a given pressure and temperature. Results from high-pressure experiments show that magnesite (MgCO₃) can be stable up to deep lower mantle conditions (∼80 GPa and 2500 K) [3]. Accordingly, magnesite may be considered the most probable carbonate phase present in the deep Earth. Experimental studies on magnesite decarbonation in presence of SiO₂ at lower mantle conditions suggest that magnesite is stable along a cold subducted slab geotherm [4, 5]. However, our understanding of magnesite’s stability in contact with bridgmanite [(Mg,Fe)SiO₃],  the most abundant mineral in the lower mantle, remains incomplete.

Hence, to investigate sub-solidus reactions, melting, decarbonation, and diamond formation in the system MgCO₃-(Mg,Fe)SiO₃, we conducted a combination of high-pressure experiments using multi-anvil press and laser-heated diamond anvil cells (LH-DAC) at conditions ranging from 25 to 70 GPa and 1300 to 2100 K.

Multi-anvil experiments at 25 GPa and temperatures below the mantle geotherm (1700 K) show the formation of carbonate-silicate melt associated with stishovite crystallization, indicating incongruent melting of bridgmanite to stishovite, in accordance with the recent finding of Litasov and Shatskiy [4]. LH-DAC data from in situ X-ray diffraction show crystallization of bridgmanite and stishovite. Diamond crystallization is detected using Raman spectroscopy. A melt phase could not be detected in situ at high temperatures.

Our results suggest a two-step process that starts with melting at temperatures below the mantle geotherm, followed by crystallization of diamond from the melt produced.  Therefore, we propose that subducted carbonate-bearing silicate rocks will not remain stable in the lower mantle and will instead melt at upper-most lower mantle conditions, fostering diamond formation. Our study also provides additional evidence that diamond production is related to carbonated melt. Consequently, the melting of recycled crust and chemical transfer to the surrounding mantle will hinder the transport of carbon deeper into the lower mantle.

[1] Stagno et al. (2015) Contrib. Mineral. Petrol. 169(2), 16.
[2] Brenker et al. (2007) EPSL 260(1-2), 1-9.
[3] Binck, et al. (2020) Physical Review Materials, 4(5),1-9.
[4] Litasov & Shatskiy (2019) Geochemistry International, 57(9), 1024-1033.
[5] Drewitt, et al. (2019). EPSL, 511, 213-222.