Where did Earth get its water?

Anat Shahar makes planets for a living. In her lab at Carnegie Science Earth and Planets Laboratory (EPL), she subjects slivers of silicate to the pressures and temperatures of a newborn planet, then sees what comes out. Lately, what keeps coming out is water.

Down the hall, Conel Alexander studies something a bit more ancient: meteorites, some older than Earth itself. He measures the ratios of their atoms and reads them like diaries of the early Solar System. For about 20 years, those diaries have told him the same thing: Earth was born dry, and water had to be shipped in from farther out.

Both scientists work at EPL and they disagree about one of the biggest questions in planetary science. Ask EPL Director, Mike Walter, which of his colleagues is right, and he will tell you it’s probably both. 

"It's almost certainly some combination of the two," he says.

The Case for Special Delivery

The traditional explanation for Earth's water goes like this: The planet formed from the collision and accretion of rocky debris that would have been too close to the Sun and too hot to hold on to volatile materials like water. Water was delivered later, riding in on asteroids and comets that bombarded our young planet. Scientists call this chaotic phase of our Solar System’s development the late veneer.

"I still think the delivery by a relatively 'small' number of volatile-rich objects from the outer Solar System—beyond Jupiter's orbit, like some meteorites, not comets—ticks the most boxes,” says Alexander. 

The strongest evidence comes from isotopes. The ratio of deuterium (a heavier form of hydrogen) to ordinary hydrogen in Earth's water does not match what one would expect if the planet made its own water from solar nebula gas—the cloud of hydrogen that surrounded the young Sun. 

Instead, it looks more like the signature of carbonaceous chondrites, ancient water-rich meteorites that formed in the cold outer Solar System before being flung inward by the gravitational jostling of Jupiter and Saturn. Recent work by Alexander and colleague Megan Newcombe, a former EPL postdoc and current associate professor at the University of Maryland, showed that melted meteorites, the closest analogs to those early planetesimals that built Earth, are remarkably water-poor. 

In essence, they argue that if the building materials were dry, the water must have come from elsewhere.

Or Maybe Earth Just Made Its Own

Shahar takes a different approach. Rather than ask what fell from the sky, she asks what was already here.

"We don't actually think about it like water coming to Earth," she says. "We think about it as: If I were Earth, could I make water just in the process of planet formation?"

In 2023, Shahar and her colleagues published a paper in Nature modeling what happens when a young, molten planet is enveloped by a hydrogen-rich atmosphere. They ran the chemistry and out came an enormous quantity of water. No exotic ingredients required, no lucky collisions. 

Then they tested the idea in the lab. In a study led by Francesca Miozzi, then an EPL postdoctoral researcher and now a Marie Curie Global Fellow at ETH Zurich, the team surrounded a silicate sample with hydrogen gas at extreme pressures and temperatures. Water formed.

The only requirements in Shahar's model are a rocky body big enough to melt and a hydrogen atmosphere still present when it does. 

The hitch? The solar nebula dissipates within a few million years of the Solar System's formation, which is a tight deadline. But Shahar argues the timing is doable. 

"If you look at the oldest differentiated things we know, they're iron meteorites," she says. "Iron meteorites have been dated to be one or two million years post-Solar System formation. So that's well within the time that the solar nebula is still around."

The Sticking Point

The largest obstacle for Shahar's model is, again, those isotopes. Earth's deuterium-to-hydrogen ratio does not match solar nebula gas, and the mismatch may be the delivery camp's strongest card. 

Alexander argues the tests go beyond hydrogen: "It is the collateral effects on other elements that are the real tests for both models." 

Carbon, nitrogen, and noble gases all have to fit the story as well.

Shahar does not dispute the isotope evidence. She wants to understand it. The experiment she most wants to see would measure how hydrogen isotopes sort themselves between Earth's silicate mantle and its iron core. If the partitioning shifts the planet's overall signature to match observations, the case for outside delivery weakens. 

"If it then matches the current estimate of hydrogen isotopes on Earth," she says, "then there is no reason to suggest that water came from anywhere else."

Bigger Than the Oceans

The origins debate is bound up with another, more basic question, and the answer helps constrain both models: How much water does Earth actually have, and where is it?

Walter has spent years on that problem, and his answer is counterintuitive. For a planet famous for its oceans, Earth is remarkably dry. 

"Earth is a pretty dry planet," Walter says. "Even though it's covered with water on the surface. Jupiter's moon Europa, only about 1% the mass of Earth, has a similar amount of water.” 

Scientists count Earth’s water in “ocean masses.” All the water on the surface is one ocean mass. The mantle likely holds roughly one more ocean mass according to Walter’s and others calculations, though there are competing estimates for this number as well. 

Then there is the core. About a third of Earth's mass is iron-rich metal, and hydrogen loves to dissolve into it. In fact, recent work by Wenzhong Wang, a former EPL postdoc who worked with Walter and is now a professor at the University of Science and Technology of China, showed that hydrogen may be in a superionic state inside the inner core—meaning the atoms move freely through the crystal lattice, almost like electrons. That could explain anomalies in how seismic waves travel through that region. Wang's calculations here suggest there could be one ocean mass of hydrogen in the inner core alone. 

If you’re counting, that’s three ocean masses so far. 

Because the far more massive outer core should hold similar concentrations, there could be roughly 20 ocean masses in the outer core. If the numbers hold, Walter thinks that the Earth may contain 25 to 30 ocean masses of water in total—most of it locked inside iron at the center of the planet. 

So is the water coming from inside the house? We’re still not sure. 

Both delivery and internal water formation could have supplied that much water, so the total alone does not settle the origin question. But the distribution offers a test. If Shahar's model is right and hydrogen was absorbed from a nebular atmosphere early enough, much of it should have sunk into the segregating iron core, which is roughly where Wang's numbers place it. 

It also means that Earth has held almost exactly one ocean mass on its surface for 4 billion years, cycling water downward through subduction zones, where one tectonic plate slides beneath another, and back upward through volcanism. 

"The fact that we have water delivered, and we can get it out, and we can keep it at the surface over time," Walter says. "That's habitability. That’s where the word really carries weight.” 

Walter is sure to caution that research findings on the water stored in Earth's mantle and core vary widely throughout the geoscience community. His own calculations have been presented in scientific talks, but are not yet published. 

Honing in on an Answer

The two models are not necessarily mutually exclusive. Shahar is careful on this point. 

"We can just say it's not necessary," she says of asteroid delivery. "Earth can make enough water on its own. It doesn't need that late veneer. But we can't say it didn't happen." 

Her model removes the luck from the equation. It does not rule out the asteroids.

The disagreement has already proved productive. A mutual colleague, working with both Shahar and Alexander, recently found water in a very old meteorite from a region of the Solar System previously thought to be dry. The discovery complicates both stories, and it has brought the two researchers together on a joint paper.

Water Everywhere?

If Shahar is right that water production is baked into rocky planet formation, the implications reach well beyond Earth. 

"It would mean that water formation is a ubiquitous step of planet formation," she says. 

In fact, JWST has already detected water vapor in the atmospheres of distant worlds, which fits.

Alexander agrees the landscape has shifted. 

"The 'meteorite' and the massive atmosphere models can probably account for all of Earth's water," he says. "It is the collateral effects on other elements that are the real tests for both models." 

He credits exoplanet science with driving much of the shift, noting that the discovery of hydrogen-rich atmospheres on distant worlds has led to "a resurrection of the massive atmosphere hypothesis."

The question is still far from settled. But for now, three scientists are working from the same small campus, closing in on the answer from different directions, which, for the Earth and Planets Laboratory, feels about right.