The study, led by Hongsheng Yuan of the Center for High Pressure Science and Technology Advanced Research in Shanghai, together with Lianjie Man and Leonid Dubrovinsky of the Bayerisches Geoinstitut at the University of Bayreuth and an international team of collaborators, is published in Nature Geoscience. Using laser-heated diamond anvil cells—devices that squeeze samples between two gem-quality diamond tips while laser beams raise temperatures to thousands of degrees—the researchers created conditions rivaling those near the core–mantle boundary, where pressures exceed one hundred gigapascals and temperatures climb above several thousand kelvin.

What emerged from these experiments were two hexagonal iron oxyhydroxides with the nominal compositions Fe5O12Hx and Fe7O12Hx, where x denotes a variable hydrogen content. These are not exotic compounds dreamed up in a computer simulation; they are crystalline phases whose structures were determined directly from in situ single-crystal X-ray diffraction measurements at synchrotron facilities, including the European Synchrotron Radiation Facility, PETRA III at DESY in Hamburg, and the Shanghai Synchrotron Radiation Facility. The crystallographic data underpinning the structure determinations have been deposited in the Cambridge Structural Database, and the broader experimental datasets are openly available through figshare, giving other researchers the means to scrutinize and extend the findings.

The significance of these phases lies in two properties highlighted by the team. First, they can form under water-undersaturated conditions, meaning they do not require an abundant free water reservoir to crystallize. This is crucial because the deep mantle is generally thought to be far drier than the shallow regions from which volcanic gases escape. Second, the new oxyhydroxides are remarkably dense. Density matters enormously in a planet whose interior is sorted by gravity: materials less dense than their surroundings tend to rise, while denser ones sink. The high densities of Fe5O12Hx and Fe7O12Hx mean that once formed, they would gravitationally settle and remain stably parked near the base of the mantle rather than escaping upward.

That combination of stability and density provides what the authors describe as a plausible mineralogical basis for the seismically observed ultralow-velocity zones. These zones, which can stretch for hundreds of kilometers horizontally but are only a few to tens of kilometers thick, reduce seismic shear and compressional wave speeds so severely that ordinary mantle minerals alone struggle to explain them. Hydrogen-bearing iron-rich phases, with their altered elastic properties and potentially partial molten character, have been increasingly invoked as candidates. Earlier work had identified a hydrogen-bearing iron peroxide and pyrite-type FeOOH as possible deep water hosts; the new hexagonal oxyhydroxides extend that family of compounds and demonstrate that iron, oxygen, and hydrogen can combine into dense crystalline architectures at the pressures and temperatures of the lowermost mantle.

The context for why water might be down there in the first place reaches back to the planet’s earliest history. Isotopic anomalies in helium and hydrogen recorded in deep-sourced volcanic rocks have long hinted that some primordial water—water inherited from the materials that accreted to form Earth—survives in isolated reservoirs deep in the mantle rather than having been lost entirely to space or mixed into the surface water cycle. A leading scenario involves a basal magma ocean: a dense, iron-rich melt that ponded at the base of the mantle during planetary differentiation and crystallized slowly over geological time. The question that has frustrated researchers is which minerals, if any, could have sequestered hydrogen during that crystallization. Most candidate hydrous silicates decompose at the relevant temperatures, and common mantle minerals such as bridgmanite hold very little water at depth. The new oxyhydroxides fill that gap, offering mineral hosts capable of locking hydrogen into the deep interior both from primordial sources and from water recycled downward by subducting slabs.

Subduction supplies the second pathway. Oceanic plates carry hydrated minerals down into the mantle, and laboratory studies in recent years have shown that some hydrous phases can survive transit to the lowermost mantle, potentially delivering water to the core–mantle boundary billions of years after Earth formed. If that arriving water reacts with iron-bearing mantle material—or with metallic iron from the core itself—the dense oxyhydroxides identified in this study could crystallize and accumulate. The result would be chemically distinct, hydrogen-rich lumps at the base of the mantle, exactly the kind of heterogeneity that seismic imaging keeps revealing.

The implications ripple outward from mineralogy into planetary evolution. The authors note that the stability of these phases provides a mechanism for transporting volatiles into Earth’s core, where hydrogen and oxygen could act as light elements dissolved in liquid iron alloy, or conversely for their episodic release back into the mantle. Mantle plumes—hot, buoyant upwellings that rise from the core–mantle boundary and feed surface hotspots such as Hawaii and Iceland—appear to be rooted in some of the same deep regions. If dense oxyhydroxide accumulations are destabilized or partially melted at the edges of plumes, the water they carry could be injected into the upwelling flow, eventually influencing the melting behavior, chemistry, and gas emissions of hotspot volcanoes at the surface. In this way, minerals formed nearly three thousand kilometers down could modulate the volatile budget of the planet across geological time.

The experimental achievement behind these conclusions should not be understated. Diamond anvil cell experiments at lowermost mantle conditions are notoriously difficult: samples are measured in volumes smaller than the width of a human hair, laser heating must be controlled with precision to avoid catastrophically melting the diamond anvils, and extracting single-crystal diffraction data from polycrystalline samples squeezed to megabar pressures demands sophisticated analysis software and high-brightness synchrotron beams. The team combined these techniques with electron microscopy and chemical analysis of recovered samples to characterize the new phases from multiple angles, including constraints on their hydrogen contents and densities at high pressure. Extended data on unit-cell parameters and refined atomic positions support the structural assignments, and the authors report that chemical analysis of recovered material found no evidence of contamination from the rhenium gasket or other experimental artifacts that might otherwise mimic the observed compositions.

Much remains to be tested. Whether the new oxyhydroxides can account quantitatively for the seismic properties of ultralow-velocity zones, how much water the deep mantle could realistically store in them, and how they interact with surrounding mantle minerals at the atomic scale are questions for future experiments and modeling. But the study reframes a long-standing problem in Earth science: the search for the planet’s missing deep water no longer needs to rely on sparse hydrous silicates that barely survive at depth. Iron, the most abundant element in Earth’s core and a major component of the deep mantle, combined with oxygen and hydrogen, may instead form its own family of dense reservoir minerals. If those reservoirs exist at the core–mantle boundary today, they could be simultaneously a vault preserving water from the planet’s birth, a sink for water recycled from its surface, and a chemical trigger for the plumes that reshape its surface—tying together, in a single crystal structure, processes that span the whole depth and history of our planet.

Subject of Research: Dense iron oxyhydroxide phases as potential water reservoirs in the deep lower mantle

Article Title: Dense iron oxyhydroxides as possible reservoirs of water under deep mantle conditions

Article References: Yuan, H., Man, L., Kupenko, I., Yin, Y., Aprilis, G., Pantousas, A., Pakhomova, A., Zhou, W., Pierru, R., Yang, K., Glazyrin, K., Rodrigues, J. E., Garbarino, G., Ball, J. A. D., Bykova, E., Hu, Q., Sossi, P. A., Frost, D. J., & Dubrovinsky, L. (2026). Dense iron oxyhydroxides as possible reservoirs of water under deep mantle conditions. Nature Geoscience. https://doi.org/10.1038/s41561-026-02067-1

Image Credits: AI Generated

DOI: 10.1038/s41561-026-02067-1

Keywords: iron oxyhydroxides, deep mantle water, core–mantle boundary, ultralow-velocity zones, diamond anvil cell, high-pressure experiments, basal magma ocean, mantle plumes, subduction, hydrogen storage, volatile cycles, Nature Geoscience

The study rests on a simple but technically demanding question: what happens to molten rock under the conditions that prevail at Mars’ core–mantle boundary, roughly 2,000 kilometers beneath the surface, where pressures reach tens of gigapascals and temperatures climb above 2,000 kelvin? Earlier in the planet’s history, Mars was almost certainly covered by a global magma ocean. As that ocean cooled, minerals crystallized and settled, and the question of what the last dregs of melt looked like—and where they went—has hung over Martian science for decades. If the final melts were dense enough, they could have drained downward and pooled at the base of the mantle, forming a long-lived silicate layer sandwiched above the liquid iron core.

To test that possibility, the researchers performed high-pressure experiments on synthetic compositions modeled after the silicate melts expected to be produced during the crystallization of a Martian magma ocean. Using multi-anvil press apparatus capable of reaching the gigapascal-scale pressures of the deep Martian interior, they equilibrated samples at controlled temperatures, quenched them rapidly to preserve their textures and chemistry, and then analyzed them with electron microscopy and other microanalytical techniques. From these recovered samples, the team determined the density of the melts as a function of pressure, temperature and composition, along with the melting behavior of the deep mantle assemblage.

The central result concerns buoyancy. For a magma layer to persist at the core–mantle boundary, it must be denser than the overlying solid mantle so that it does not rise and disperse, yet it must remain molten rather than freezing solid. The experiments show that late-stage melts enriched in iron and incompatible elements—components that preferentially remain in the liquid as crystals grow—become sufficiently dense under Martian deep-interior conditions to be gravitationally stable at the base of the mantle. In other words, the last liquids of a crystallizing magma ocean would naturally sink, collect and stagnate at the very bottom of the mantle, exactly where a basal magma layer has been hypothesized.

Just as important, the measurements constrain how much heat such a layer would have to exchange with its surroundings to remain molten. A magma layer at the core–mantle boundary sits atop a hot liquid iron core and beneath crystalline mantle rock, so its survival depends on a delicate thermal balance. The experimental data on melting temperatures and melt densities allowed the team to build thermal evolution models of the Martian interior, tracking the layer’s fate over the planet’s 4.5-billion-year history. Those models indicate that, for plausible core temperatures and mantle heat flows, the layer does not necessarily freeze; instead, it can be maintained as a thin, stable reservoir of molten silicate for geologically long timescales, sustained by heat flowing out of the core.

The implications extend directly to seismology. NASA’s InSight lander, which recorded Martian seismic activity until late 2022, produced the first direct glimpses of the planet’s interior, and analyses of its data have hinted at structure near the base of the mantle—some studies even proposed an entirely molten or partially molten layer above the core to explain the observed seismic velocities and attenuation. A basal silicate magma layer offers a physical mechanism for such signals: molten rock attenuates seismic waves strongly and slows shear waves dramatically compared with solid rock. The new experiments provide the compositional and thermodynamic anchor that seismic modelers need, linking a specific melt composition and thickness to specific seismic signatures rather than treating the layer as a free parameter.

The results also speak to the long-term thermal and chemical evolution of Mars. A persistent magma layer at the core–mantle boundary acts as an insulating blanket between the mantle and the core, regulating how efficiently heat escapes the core. That, in turn, influences whether the core can generate a dynamo-driven magnetic field and how the mantle convects over time. Mars lost its global magnetic field billions of years ago, and understanding the thermal insulation provided by a basal melt layer helps explain the timing and efficiency of core cooling. The layer would also sequester heat-producing and incompatible elements—uranium, thorium, potassium and others—concentrating them at the base of the mantle and altering the planet’s internal heat budget in ways that standard models, which assume a chemically uniform mantle, do not capture.

There are caveats, and the authors are careful about them. The experiments constrain the behavior of candidate melt compositions under simplified conditions; the real Martian core–mantle boundary may host heterogeneous materials, partial melting of mantle rock in contact with the core, or mixtures of silicate melt with light elements transferred from the core. The thickness of any surviving layer depends sensitively on the initial sulfur and iron content of the core, the efficiency of mantle convection and the exact crystallization sequence of the ancient magma ocean. Still, the experimental demonstration that deep Martian melts are gravitationally stable at the base of the mantle removes one of the biggest objections to the basal-magma-layer hypothesis: that such melt would have been buoyant and would have risen away long ago.

The work also carries lessons for other planets. Magma ocean crystallization is a universal stage in terrestrial planet formation, and analogous basal melt layers have been proposed for early Earth, where dense iron-rich melts may have accumulated above the core–mantle boundary and influenced the chemistry of plumes that rise from that region today. If a small planet like Mars can retain a molten basal layer for billions of years, the same physics may operate on Venus, Mercury and rocky exoplanets, where the presence or absence of such layers could alter volcanic activity, magnetic-field generation and long-term habitability. Each new laboratory dataset on melt density and melting curves thus becomes a tool for reading the interiors of worlds no spacecraft will ever drill into.

Future observations could settle the question. Additional seismic data, whether from a future Mars geophysical network or from advances in analyzing the existing InSight archive, could be compared directly with the experimentally derived elastic and attenuation properties of the candidate melt compositions. Improved constraints on Mars’ core size and density, its tidal response and its moment of inertia will further narrow the permissible layer thickness. For now, the experiments turn an intriguing speculation into a physically grounded scenario: beneath the cold, dusty surface of Mars, at the deepest reach of its rocky mantle, the last traces of a primordial magma ocean may still be glowing—molten, dense and stubbornly enduring at the edge of the planet’s iron heart.

Subject of Research: Experimental constraints on the existence of a long-lived basal magma layer at the Martian core–mantle boundary

Article Title: Experimental constraints on a long-lived magma layer at the Martian core–mantle boundary

Article References: Pierru, R., Gréaux, S., Dominijanni, S., Man, L., Kono, Y., Kakizawa, S., Higo, Y., Badro, J., Frost, D. J., & Antonangeli, D. (2026). Experimental constraints on a long-lived magma layer at the Martian core–mantle boundary. Nature Geoscience. https://doi.org/10.1038/s41561-026-02104-z

Image Credits: AI Generated

DOI: 10.1038/s41561-026-02104-z

Keywords: Mars, core–mantle boundary, magma ocean, high-pressure experiments, silicate melt density, InSight seismology, planetary interiors, thermal evolution, Martian mantle, basal magma layer, geophysics, Experimental

Water is not merely a surface phenomenon. Although oceans, ice caps, and groundwater form the visible part of the hydrological cycle, a hidden reservoir of water is stored within the rocks of the mantle, bonded into the crystal structures of nominally anhydrous minerals and discrete hydrous phases. Geophysicists have long recognized that the Earth’s interior exchanges water with its surface over geological time, a process that regulates volcanism, seismicity, and the evolution of the oceans. Estimates of how much water resides in the deep mantle, however, span an enormous range, precisely because researchers cannot agree on which minerals can retain hydrogen under lower-mantle conditions of extreme pressure and temperature.

The commentary frames the problem against the backdrop of the Earth’s early history. Work on ancient zircons, melt inclusions, and primitive basalts suggests that the planet acquired much of its volatile inventory very early, possibly from primordial sources rather than exclusively from late-arriving comets or asteroids. If a substantial fraction of the Earth’s water was incorporated into the mantle during accretion, then the deep water cycle must be capable of retaining and redistributing that water over billions of years. Conversely, if the mantle is largely dry, then subduction must efficiently return most of the water it carries back to the surface. Both scenarios carry profound implications for mantle convection, the oxidation state of the deep Earth, and the long-term stability of the oceans.

The traditional carriers of water in the mantle are well catalogued down to the transition zone. Serpentine, a hydrous magnesium silicate formed when seawater alters oceanic crust and lithosphere, decomposes at shallow depths. Its breakdown products, dense hydrous magnesium silicates such as phase D, were once thought to shepherd water downward. Laboratory experiments showed that these phases can persist to the top of the lower mantle, roughly 660 kilometers deep, and even deeper in colder subducting slabs. Yet as temperature and pressure rise through the lower mantle, most of these hydrous silicates destabilize, releasing water into melts or fluids. Whether that water can be fixed into a stable crystalline host deeper down has been a puzzle for decades.

A key development came with the discovery that water itself, in the form of hydroxyl, can enter the crystal lattice of the lower mantle’s dominant minerals. Experiments by Li and colleagues reported in Nature Geoscience in 2020 demonstrated that bridgmanite, the most abundant mineral in the Earth, can incorporate substantial hydroxyl under lower-mantle conditions, while studies of the high-pressure ice phase Ice VII, synthesized and observed in inclusions in diamonds, showed that free water can exist as a crystalline solid at transition-zone depths. Zhang and coworkers had earlier identified a hydrous layer in silicon-deficient bridgmanite, while Nishi and colleagues reported a pyrope-type hydrous phase stable near the core-mantle boundary. Each discovery suggested multiple potential reservoirs, but none settled the question of what actually transports water across the 660-kilometer boundary in the first place.

The new experiments reviewed in the commentary focus on iron oxyhydroxides, phases closely related to the familiar surface minerals goethite and hematite but squeezed and heated to lower-mantle conditions. Under extreme compression, FeOOH phases adopt crystal structures stabilized by hydrogen bonds and can retain hydrogen at pressures corresponding to depths greater than 1,500 kilometers. Crucially, when combined with aluminum or other elements, these phases appear to remain stable at the elevated temperatures of the deep mantle, something the pure iron endmember cannot achieve. In slabs cold enough to suppress thermal breakdown, iron oxyhydroxides could therefore act as water ferries, carrying hydrogen from the seafloor to the base of the mantle.

Experiments led by researchers including Ohtani and Ishii, and summarized in recent work in Progress in Earth and Planetary Science, show that FeOOH synthesized in the laser-heated diamond anvil cell retains its hydroxyl content even when subjected to lower-mantle temperatures, particularly when the starting materials reflect the compositions of subducted banded iron formations—iron-rich sediments deposited on ancient seafloors. Ishii and colleagues also identified post-stishovite and other dense silica phases that can host aluminum and hydrogen, extending the suite of potential deep-water carriers. Together, these results sketch a pathway in which chemically heterogeneous, cold subducted material delivers water into the lower mantle where earlier models predicted it should escape.

Independent lines of evidence lend the experimental picture geological credibility. Ultra-deep diamonds, formed at depths exceeding 600 kilometers, frequently trap inclusions of Ice VII and hydrous ringwoodite, direct snapshots of free and bound water at transition-zone depths. Seismological observations of discontinuities, attenuation anomalies, and low-velocity features near the core-mantle boundary have been interpreted as signs of partial melting triggered by hydrous phases releasing water at the base of the mantle. Studies by Karato and colleagues have clarified how even small amounts of hydrogen alter the viscosity, electrical conductivity, and seismic anisotropy of mantle minerals, providing geophysical tools to test where water accumulates. The consistency between laboratory phase diagrams and geophysical signatures strengthens the case that the deep water cycle is active and dynamic rather than a geological curiosity.

Still, the commentary emphasizes that the puzzle is far from solved. The stability fields of iron oxyhydroxides depend sensitively on temperature, oxygen fugacity, and composition, and the conditions inside subducting slabs vary enormously. Estimates of the hydrogen-storage capacity of bridgmanite and ferropericlase continue to be revised, and the partitioning of water between minerals, melts, and fluids at lower-mantle pressures remains poorly constrained. Moreover, the total mass of subducted iron-rich sediments available to host FeOOH may limit how much water the oxyhydroxide pathway can deliver. Resolving these uncertainties will require new experiments at even higher pressures and temperatures, advanced synchrotron and spectroscopic techniques capable of detecting hydrogen in situ, and closer integration with seismic imaging of slabs penetrating the lower mantle.

What is at stake is nothing less than the budget of the Earth’s water. If iron oxyhydroxides and related phases can pump water into the deep mantle, the planet’s interior may hold an ocean’s worth of water or more, slowly exchanging it with the surface through volcanism and degassing. That reservoir would influence the oxidation state of the mantle, the generation of deep melts, and possibly the long-term evolution of the atmosphere. It would also mean that the familiar water cycle of rain, rivers, and evaporation is only the shallow skin of a vastly larger circulation that reaches within a few hundred kilometers of the Earth’s core. The new experimental results, as highlighted by Wilson, bring scientists closer to tracing the full arc of that hidden cycle, and remind us that one of the most familiar substances on Earth still keeps its deepest secrets hidden in plain sight, locked within minerals no one has ever seen with the naked eye.

Subject of Research: Transport and storage of water in the Earth's lower mantle by hydrous minerals such as iron oxyhydroxides

Article Title: The Earth’s puzzling deep water cycle

Article References: Wilson, A. J. (2026). The Earth’s puzzling deep water cycle. Nature Geoscience. https://doi.org/10.1038/s41561-026-02088-w

Image Credits: AI Generated

DOI: 10.1038/s41561-026-02088-w

Keywords: deep water cycle, iron oxyhydroxide, lower mantle, subduction, hydrous minerals, bridgmanite, FeOOH, high-pressure experiments, mantle water reservoir, geodynamics, mineralogy, dense hydrous phases