What is dark oxygen on the deep seafloor?

“Dark” means darkness where no light reaches. The 2024 paper’s title was “Evidence of dark oxygen production at the abyssal seafloor.” In Japanese, the term is translated as “ankoku sanso,” literally “darkness oxygen.”

In the sea, below about 200 meters there is rarely any significant light, and photosynthesis is no longer possible. Even faint light can be detected only down to about 1,000 meters. Much of Earth’s oxygen is made by photosynthesis, but the seabed in this study is about 4,000 meters down and pitch dark. Because this oxygen would be made without light, it is called “dark.”

It has nothing to do with “dark matter” in space. The word “dark” is the same, but this is a story about the deep sea, not about space.

It turned up in experiments meant to measure how much oxygen life on the seabed uses. The site was the Clarion–Clipperton Zone, which stretches across the northeastern Pacific. Polymetallic nodules lie scattered over wide areas of the seafloor here. A nodule is not a lump of pure metal but a stone-like lump made of a mix of minerals that contain metals. Nodules range in size from tiny particles to about the size of a potato. On three voyages in 2021 and 2022, the team lowered an instrument platform to the deep seafloor 16 times. It carries three square boxes about 22 centimeters on a side, which are pushed into the seabed to trap mud, nodules and the seawater above them. Sensors in the boxes’ lids recorded how the oxygen changed for 47 hours.

Living things in the mud use oxygen, so the oxygen in the boxes was expected to fall. Indeed, another method that measures oxygen in fine detail within the mud showed that oxygen was being used up in the mud. Yet in 25 box experiments, the oxygen rose within 47 hours, and in the highest box it reached more than four times the starting level (the average of the 25). Even after subtracting what was used, oxygen was being made, the calculation shows.

Andrew Sweetman, who led the research, said in a university news release that at first he thought the sensors were faulty. But similar readings kept showing up over 10 years. In this study, the team also confirmed the rise with a separate method, in which a chemical is added to water drawn from the box to measure its oxygen. Rechecking records from experiments done the same way in 2015 and 2018 at other places in the same zone (about 4,000 to 5,200 meters deep) also showed oxygen rising (the sensors had been calibrated months earlier, so those records show only how the oxygen changed, not its exact level).

First, they checked whether living things were responsible. On the ship, they kept mud taken from the seabed in the dark at 1.6 °C, the same as on the seabed, and measured the oxygen. Oxygen still rose when a poison that kills microbes was added, and also in tests with nodules alone and no mud. Some microbes can make oxygen in the dark, but this poison kills them. So the paper concluded that most of the added oxygen was not made by living things and that the nodules were involved. The team also calculated how much oxygen could come from radiation splitting water or from chemical changes in the nodules’ metals, and found it came to less than 0.5% of the rise.

Next, they measured voltages (the size of the push that drives electricity, in volts) on the nodules’ surfaces. In the lab, they took 153 readings of the voltage between two spots on the surfaces of 12 nodules; the highest was 0.95 volts. Nodules are built of stacked layers containing metals. The paper proposed that voltages arising from differences between the metals in these layers might split seawater into oxygen and hydrogen (electrolysis). Franz Geiger, a chemist on the study, calls the nodules a natural “geobattery.”

But there is a weak point: the voltage is too low. The paper itself says that splitting water by electrolysis needs about 1.6 volts. The minimum in theory is 1.23 volts, but a real reaction needs some extra voltage to get going (about 0.37 volts, by the paper’s working figure). The measured voltage was 0.95 volts at most, which falls short.

On this gap, the paper writes that, depending on how the reaction proceeds, the voltage needed could be lower, and the university news release explains that when many nodules cluster together, the voltage could add up, as when batteries are connected. Both say only that this could happen. The paper itself also says it is not known where the energy to keep producing electricity comes from. A battery runs down as it is used. If nodules keep splitting water like batteries, the energy behind that electricity must be coming from somewhere.

After the paper came out, people raised doubts about the measurements. In December 2025, the marine science journal Frontiers in Marine Science published an opinion article (Downes et al.) that had passed review by other researchers (peer review). Of its 13 authors, six are staff of The Metals Company, a deep-sea mining company that holds rights to explore for nodules in the study area (one of them a former employee); four are researchers at other companies, and three are at universities and research institutes. This company had also helped fund the 2024 study.

What the opinion article suspects first is air left in the boxes. Before a box is sealed on the seabed, the water inside is supposed to be fully swapped with the surrounding seawater. But the article points out that the starting oxygen in the boxes was always higher than in the surrounding seawater. It takes this as a sign that the water was not fully swapped, and that air taken in at the sea surface, among other things, was left in the boxes (in another study, trapped bubbles raised the oxygen in a box).

It also gives points such as these, and argues that the rise was an experimental error, not a process on the seabed. Oxygen rose even in tests with boxes closed at the bottom that did not enclose the seabed, and, in 2018 experiments elsewhere that the paper used as support, in boxes said to have held no nodules. The sensors and the method that tests drawn water did not agree on the starting oxygen levels. The 0.95 volts is an outlier; in the paper’s main figure the highest is only 0.24 volts. Hydrogen, which electrolysis should make along with oxygen, was not measured. Five studies that used similar methods on nodule-rich seafloor found no rise in oxygen.

In May 2026, six current and former staff of the same company (Webber et al.) published a paper saying no link can be seen between nodules and how the oxygen rose. It appeared in F1000Research, a journal that posts papers first and then has other researchers review them, with the verdicts also made public. The 2024 paper had reported a link: the larger the nodules’ surface (on average, per nodule) in a box, the more oxygen formed. This paper says that after re-measuring the nodules’ area from photos and adding three more boxes, the link weakens until it cannot be told apart from chance. The peer reviewers’ verdicts are split.

The 2024 paper had already answered the bubble question. Each box has two valves that let air out while it sinks. And even if a bubble remained, at 4,000 meters its oxygen would dissolve into the water in less than a second, so it could not explain a slow, steady rise over two days, the paper calculates. It also says that because oxygen rose by the separate method of testing drawn water too, the sensors were not faulty. The opposing article replies that this calculation takes for granted that a bubble dissolves completely into the box water. When a bubble collapses under high pressure, it argues, water rich in oxygen can collect inside the valves and tubes attached to the box, then seep slowly into the box, so the oxygen could rise in much the same shape as the records.

On April 8, 2026, Nature Geoscience, the journal that published the paper, added a note to it saying that aspects of the article are subject to concerns being considered by the editors. It says a further response will follow once the issues are resolved.

If it is real, the deep sea gets oxygen not only from above but also from a source on the seabed. Deep-sea oxygen is usually brought by seawater that cools and sinks near the poles, and oxygen used up in the deep sea can take more than a thousand years to be replaced. The paper says dark oxygen might be used for breathing by life on the seabed.

It also bears on deep-sea mining. Nodules contain cobalt, nickel, copper, lithium, manganese and other metals, and mining them is being considered because these metals are vital to batteries. The paper wrote that it is necessary to study how dark oxygen might be affected when mining stirs up seabed mud that then settles over wide areas. Geiger said in a university news release that the way these materials are mined needs rethinking, so as not to deplete the oxygen source for deep-sea life. The opposing opinion article, on the other hand, writes that research money spent in the wrong places on the strength of claims without solid evidence is also a plain, visible cost.

Sweetman, who led the team, said in the university news release that questions such as where life that uses oxygen might have begun need revisiting. The paper itself, though, says that because the results come from limited places and times, scaling up the amount to longer periods or wider areas of seabed calls for caution.

What exists now is “a record of oxygen rising inside boxes on the deep seafloor.” Whether that is because oxygen is really being made on the seabed, or because of a measurement error, is not yet settled. Getting closer to an answer needs checks like the ones below, raised by the paper’s authors and by the opposing article.

First, whether other researchers, in other places and with other instruments, also see the oxygen rise. The opposing article writes that, as of 2025, the result had not yet been confirmed by other studies. Second, whether boxes without nodules, checked the same way, show no rise. The opposing article points to the lack of this comparison as a problem. Third, whether hydrogen, which electrolysis should make along with oxygen, also builds up in the boxes. The opposing article writes that finding hydrogen is essential to confirm that electrolysis took place. Fourth, where the energy that keeps the nodules making electricity comes from; the paper itself says this is not yet known.

Sweetman’s team also plans to check again in other places. In 2025, the Nippon Foundation and the Scottish Association for Marine Science (Sweetman’s institute) announced a three-year joint project, with the foundation providing £2 million. It plans to build autonomous instrument platforms for the work and to study whether dark oxygen forms, and how it affects microbes, in the deep Pacific, including parts of the hadal zone in ocean trenches down to 11,000 meters. As of November 11, 2026, Nature Geoscience, the journal that published the paper, had not yet announced the outcome of its review.