Dry rocks may have helped create Earth’s oceans without bringing a single drop of water themselves.
As it grew, the young Earth would have captured countless rock fragments, ranging from a few millimeters to a few centimeters. These pebbles passed through a very hot, hydrogen-rich atmosphere. As they descended, they could melt before reaching the surface.

Formation of Earth / @IPGP.
The researchers recreated this situation in the laboratory using water-free volcanic rocks. They heated them to 1200 °C and 1300 °C, then exposed them to a stream of hydrogen. The goal was to see whether this encounter could create water where none had existed initially.
The result comes from a fairly simple reaction. Hydrogen captures the oxygen contained in the iron oxide of the molten rock. Metallic iron then forms at the same time as water vapor. This transformation occurred within a few hours in the experiments.
In other words, some of the water may not necessarily have been delivered already formed by wet materials. It could also have been produced locally during the planet’s growth. The early atmosphere would therefore not merely have been a backdrop: it would have participated directly in the chemistry of the young Earth.
The authors estimate that this mechanism can produce quantities of water equivalent to oceans. Above all, the experiment establishes that such a chemical pathway works under the simulated conditions.
The metal that forms also carries away phosphorus. If it then sinks toward the core, it removes this element from the rocks that will form the mantle. This behavior may help explain why Earth’s mantle contains relatively little phosphorus compared with Mars’s mantle.
The same process could occur on other rocky planets of roughly Earth’s size if they grow beneath a hydrogen-rich atmosphere. They could therefore produce large quantities of water very early on. However, the presence of water alone does not mean that a planet harbors life.
The experiment tracks neither the preservation of this vapor during the planet’s evolution nor its eventual condensation into oceans. These stages still need to be tested in models describing the formation of an entire planet.