Boron Clues Show Carbon Sinks Deep in Hot Subduction Zones

Deep beneath volcanic arcs, the slow collision of oceanic plates drives one of Earth’s great chemical conveyor belts: the recycling of carbon between the planet’s surface and its interior. A new study published in Nature Communications by Hai-Quan Liu of the Guangzhou Institute of Geochemistry, Yi-Gang Xu and colleagues argues that in unusually hot subduction zones, a surprisingly large share of the carbon locked in the descending slab escapes the shallow breakdown that many models predict, and instead travels far deeper into the mantle. The evidence comes from an unexpected source: tiny droplets of magma trapped inside crystals of olivine in volcanic rocks from Panama and southwestern Japan, whose boron isotope fingerprints point to melted carbonate-bearing rocks rather than fluids released from them.

The debate at the heart of the study concerns the efficiency of deep carbon recycling in hot subduction systems. Hot slabs, which are young, warm and buoyant, were long expected to lose their carbonate cargo early. Contrasting models make opposite predictions: one holds that extensive decarbonation occurs at shallow depths, releasing carbon dioxide into the forearc and arc regions before the slab can carry carbon any deeper; the other holds that substantial carbonate survives within the subducted slab and is transported into the deep mantle. Which of these scenarios dominates matters enormously, because subduction zones are the principal route by which carbon returns from the surface reservoir to the mantle, balancing the carbon vented by volcanoes and mid-ocean ridges over geological time.

To test these competing pictures, the team examined olivine-hosted melt inclusions in two distinctive suites of volcanic rocks. The first are adakites from western Panama, erupted above a modern hot subduction system where young oceanic crust descends beneath Central America. The second are high-magnesium andesites from southwest Japan, which the researchers treat as a window into Archean-like hot subduction, the kind of tectonic regime thought to have operated on the early Earth more than two and a half billion years ago, when the mantle was hotter and slabs warmer. Melt inclusions are precious to geochemists because they seal a sample of the parental magma inside a host crystal as it grows, protecting it from later degassing and alteration that would otherwise erase the volatile record.

The key measurement is the ratio of boron’s two stable isotopes, expressed as the δ11B value. Boron has an unusually light isotope, boron-10, and a heavier one, boron-11, and the two are fractionated strongly at low temperatures near the seafloor, where seawater alteration loads the oceanic crust with boron enriched in the heavy isotope. When a slab is cold, this heavy boron is flushed out early by fluids, so arc magmas in cool subduction zones typically carry moderate δ11B signatures. In the melt inclusions from Panama and Japan, however, the team found strikingly elevated δ11B values reaching up to +23.2 per mil, among the heaviest recorded in arc magmas, indicating that the boron source had been heated to the point of melting rather than simply dewatering.

Crucially, the heavy boron isotope signal does not stand alone. The same melt inclusions show high ratios of cerium to lead, lanthanum to ytterbium, europium to titanium, and thorium to zircon. Each of these trace-element ratios is a diagnostic tracer of contributions from subducted sediment or carbonate-bearing lithologies that have been melted, rather than merely fluxed by aqueous fluids. Together, the elevated δ11B values and the enriched trace-element patterns indicate that the arc magmas incorporated melts derived from carbonate-bearing parts of the subducted slab. In other words, the slab was not shedding its carbon as fluid at shallow depth; it was carrying it down intact until temperatures grew high enough to melt the carbonate-bearing material directly.

The team then integrated these melt-inclusion observations with regional estimates of carbon dioxide flux from western Panama. The comparison yields a striking quantitative conclusion: more than roughly 60 percent of the carbonate in the subducted slab escapes shallow decarbonation. That figure challenges the intuition that hot slabs should decarbonate efficiently, and the authors propose a mechanism to explain the paradox. Extensive dehydration in the forearc and beneath the arc, they argue, causes such substantial fluid loss early in the subduction journey that the slab is left essentially dry by the time it reaches the deeper forearc and the arc front. Because fluid-mediated decarbonation depends on water-rich fluids infiltrating and reacting with carbonate, a dehydrated slab is a protected slab: with little fluid left to drive the reactions, the carbonate survives.

This mechanism reframes how scientists think about the fate of carbon in warm subduction zones. Rather than heat alone determining whether carbon is released, the balance between dehydration and decarbonation becomes decisive. A hot slab loses its water quickly and thoroughly, and in doing so it loses the very agent needed to strip away its carbonate. The result is that substantial carbon can bypass shallow decarbonation and continue its descent, in both modern hot subduction systems like Panama and Archean-like systems like the one recorded by the southwest Japan andesites. The finding extends the conclusion back into deep time, suggesting that early Earth’s hotter tectonics did not necessarily vent more carbon to the atmosphere; it may instead have buried more of it.

The implications ripple outward across the Earth sciences. Global carbon cycle models, which attempt to balance volcanic outgassing against subduction inputs over millions of years, depend on assumptions about how much carbon each slab delivers to the deep mantle. If more than half of slab carbonate survives shallow processing in hot subduction zones, the deep carbon flux may be considerably larger than many current estimates allow. That, in turn, bears on reconstructions of atmospheric carbon dioxide and climate across geological eras, including the Archean, when the interplay of a hotter mantle, faster plate recycling and a different biosphere shaped the planet’s early climate system. It also informs the study of carbonated mantle peridotites and other proposed hidden reservoirs of subducted carbon, which recent work has highlighted as potentially significant sinks.

Methodologically, the study demonstrates the power of melt inclusions as archives of slab inputs. Because the inclusions trap the magma before degassing, they preserve volatile and trace-element information that whole-rock samples, altered by eruption and weathering, often lose. Pairing boron isotopes with a suite of trace-element ratios gives the approach redundancy: the isotope signal identifies a melted, carbonate-bearing source, while the element ratios independently corroborate it. Applying the same toolkit to both a modern arc and an Archean analogue gives the conclusion a temporal reach that single-site studies cannot achieve, allowing the authors to argue that the behavior is a general property of hot subduction rather than a local peculiarity of Panama.

Questions remain, as they always do in deep-Earth science. The carbon budget of any single subduction zone is difficult to constrain, and regional flux estimates carry uncertainties that propagate into the 60 percent figure. Extending the melt-inclusion approach to other hot subduction systems around the world will be needed to test whether the Panama and Japan results are representative. But the core message of the study is clear and consequential: the hottest, most aggressive subduction zones may be among the most efficient at smuggling carbon into the deep mantle, not the least. In the planet’s long-term carbon ledger, the descending slab appears to be a far more faithful courier than many had assumed, carrying its carbonate cargo past the shallow gauntlet and into the depths where it may reside for hundreds of millions of years before returning, if ever, to the surface.

Subject of Research: Deep carbon recycling and carbonate retention in hot subduction zones traced by boron isotopes in melt inclusions

Article Title: Heavy boron isotopes reveal deep carbon recycling in hot subduction zones

Article References: Liu, H.-Q., Tian, F., He, M.-H., Cui, Z.-X., Zhang, L., He, P.-L., Hong, L.-B., Huang, X.-L., Bindeman, I., Hoernle, K., & Xu, Y.-G. (2026). Heavy boron isotopes reveal deep carbon recycling in hot subduction zones. Nature Communications. https://doi.org/10.1038/s41467-026-78096-6

Image Credits: AI Generated

DOI: 10.1038/s41467-026-78096-6

Keywords: subduction zones, boron isotopes, carbon cycle, melt inclusions, adakites, decarbonation, deep carbon recycling, Archean, geochemistry, volcanic arcs, carbonate, Panama