In 2014, a cloudy and commercially worthless diamond from Brazil convinced a lot of people that Earth was hiding an ocean. Twelve years later, a steel press in Japan has redrawn the map of where that water sits.

The diamond that started it

In March 2014, a team led by Graham Pearson of the University of Alberta published a short paper in Nature about a battered stone from Juína, Brazil. Trapped inside it was a grain of ringwoodite, a high-pressure form of olivine that until then had only been seen in meteorites and laboratory furnaces. The grain held about 1 percent water by weight, bound into the crystal as hydroxide rather than sitting there as liquid.

Olivine takes that form in the mantle transition zone, roughly 410 to 660 kilometres down. Scale one damp grain up to an entire layer and the arithmetic turns spectacular. The headlines wrote themselves: an ocean's worth of water, perhaps several, hidden under our feet.

The idea stuck. But the transition zone is only the lid. Below 660 kilometres the lower mantle begins, and in twelve years no diamond has surfaced with a clean water reading for the bulk of it.

Meet davemaoite, the suspect

The lower mantle runs on three minerals. Two of them, bridgmanite and ferropericlase, are known to hold almost no water at all. That leaves the third: davemaoite, calcium silicate perovskite, CaSiO3.

Davemaoite has an unusually accommodating crystal structure. It will take in awkward, oversized atoms that the other two reject, which is why it ends up hoarding much of the lower mantle's uranium and thorium, the elements that keep the planet warm from the inside. Water was assumed to be no exception. A 2022 study in American Mineralogist modelled 0.99 to 1.96 weight percent water sitting in the crystal structure and concluded that CaSiO3 was an important phase to consider for water storage in the lower mantle. If that were right, davemaoite alone could be the deep Earth's reservoir.

The mineral has only had a name since 2021, when researchers confirmed it inside a diamond from the Orapa mine in Botswana, identifying it at the Advanced Photon Source, the Argonne National Laboratory synchrotron in Illinois, on its 34-IDE beamline. They named it for Ho-kwang "Dave" Mao, a geophysicist at the Carnegie Institution's Geophysical Laboratory since 1968, who pushed diamond anvil cells past a megabar and founded China's HPSTAR high-pressure centre in 2013.

An answer in two hundredths of a percent

Takayuki Ishii of Okayama University's Institute for Planetary Materials and his collaborators made davemaoite twice at BL04B1, the high-pressure beamline at SPring-8 in Hyogo Prefecture: once dry, once wet, then compared the volumes of the two crystals. They grew it at up to 30 gigapascals and 2,100 kelvin, about 300,000 atmospheres and 1,830 degrees Celsius, and read the lattice with synchrotron X-rays without taking it out of those conditions. The work brought together Okayama, Hiroshima, Ehime, Teikyo University of Science and Kyoto University, along with JASRI, which runs SPring-8, and HPSTAR in Beijing.

Water forced into a crystal leaves a fingerprint in its volume. If davemaoite really carried 1 percent water, the hydrous crystal should have come out 0.2 to 0.5 percent smaller than the dry one. The measured difference was 0.02 percent. Essentially nothing.

Run that backwards and you get a ceiling on how much water the mineral can hold at all: 0.08 percent by weight, at most. A tenth of the old estimate, or less. The paper appeared in Communications Earth & Environment on 24 July, and the universities announced it on 31 August.

So where did the water go?

It did not go anywhere. It was never spread evenly through the rock to begin with.

If all three main lower-mantle minerals are close to dry, then deep water must be riding in something that is not mantle rock: the crust that gets dragged down with it. Ocean floor sinking at subduction zones carries hydrous silica phases that survive to great depth, and the paper concludes that this subducted crustal material, not the surrounding mantle, dominates deep water storage.

Picture wet streaks threaded through a dry body, following the tracks of old seafloor on its way down. Where the water sits governs how rock melts, how it flows, and how the volume of the oceans has been balanced over billions of years.

Why the verdict came out of Hyogo

There is a reason this particular answer came from Japan, and it goes back to a physicist named Naoto Kawai.

Kawai moved to Osaka University in 1962 and built a press whose eight anvils, cut from a tungsten carbide sphere, closed on an octahedral sample holder from every side at once. The inner assembly is still called the Kawai cell. He and his colleagues then added a second stage of compression and published it in 1973 in the Proceedings of the Japan Academy, as "A New Device for Pressure Vessels". The design passed to the University of Tokyo, Nagoya and Okayama in the 1970s, and was then exported to the United States, Canada, Australia and Germany. The two Kawai-type presses at BL04B1 pass 120 gigapascals with sintered diamond anvils and reach 65 with tungsten carbide, at temperatures up to 3,000 kelvin, all of it watched live in white X-rays. The 2024 review that sets out that history was written by Tetsuo Irifune of Ehime University, who is also a co-author of the new paper.

The people circulate as much as the technique does. Ishii took his doctorate at Gakushuin University in 2015, was at the Bayerisches Geoinstitut in Bayreuth, Germany from 2016 to 2020, and then, from 2021 to 2023, at the Beijing centre Mao founded. He arrived at Okayama in 2023, and in April 2026 he received the Young Scientists Prize in Japan's national science and technology commendations. In the announcement Ishii described the result as a domestic research group fusing what he called synchrotron and multi-anvil press technology that Japan has led the world in, and wrote that he wants to hand the method to the next generation of researchers.

As of the August 2026 announcement, SPring-8 is due to shut down from August 2027 for about a year while its storage ring is rebuilt as SPring-8-II. Measurements like this one have a window.

An American synchrotron found davemaoite. A Japanese one weighed its water. Does your country run a machine on that scale, and do you ever hear what comes out of it?

References