Earth’s core contains the planet’s largest subsurface reservoir of carbon. The element sank there during Earth’s accretion because of its siderophile (iron-loving) character, leaving very little in the mantle. But the abundance of carbon in today’s mantle—roughly 120 ppm—is much greater than Earth scientists had predicted (1–5 ppm), and it’s never been clear how it was added to the mantle.
A decade ago, Rice University’s Rajdeep Dasgupta and his collaborators speculated that water—in the form of point defects in the mantle’s silicate minerals—released at the core–mantle boundary (CMB) during subduction may chemically react with iron in the core to liberate carbon atoms. Verifying the reaction and the exchange of carbon between the core and mantle would directly link the carbon and water cycles in Earth’s core to key planetary processes, such as mantle melting, chemical differentiation, and advection.
Led by Byeongkwan Ko (then a doctoral student at Arizona State University, now a postdoc at Michigan State University) and Dan Shim (his adviser at ASU), a team of geophysicists has now verified the hypothesis. Conducting experiments at the Advanced Photon Source at Argonne National Laboratory, the group compressed iron–carbon alloys and water together in diamond anvil cells at the pressure and temperature expected at the CMB. The reaction produced iron oxide, iron hydroxide, and carbon.
According to those experiments, the stable form of carbon at 70–140 GPa and up to 4050 K turns out to be not oxidized carbon, as Dasgupta and colleagues had assumed, but diamond. The hottest spots (3) in the figure gave rise to a sharp peak at the end of a Raman spectral feature (1470 cm−1)—from diamond crystals formed from the reaction of Fe3C and H2O—whereas the unheated spots (1 and 2) do not show such a peak.
Prior experiments showed that hydrogen is more siderophile than carbon, and the two elements affect each other’s solubility in the reaction zone. The solubility of carbon therefore decreases locally in the presence of hydrogen. The team’s new x-ray diffraction results indicate that if water released by the subducting slab reacts with liquid iron at the CMB, hydrogen would alloy with the iron metal and most likely be incorporated into the liquid outer core. With its now lower solubility, carbon, by contrast, would be incorporated into the mantle, where it solidifies into diamond.
That transfer of carbon from the core to the mantle is thought to have taken place over a significant fraction of Earth’s history. It’s unclear how much subducted water reaches the CMB. But if some 3–30% of it does and reacts with the core, as much as 160 ppm of carbon can eventually be added to the mantle over 3 billion years. Indeed, the 120 ppm of today’s carbon content can mainly be explained by water-induced transfer from the core. (B. Ko et al., Geophys. Res. Lett. 49, e2022GL098271, 2022.)
