Earth’s inner core is often described as a solid iron ball. That is a useful first approximation, but it leaves out the lighter elements mixed into the metal and a basic question that is surprisingly difficult to answer: how did the first solid crystal form?

A 2025 study in Nature Communications argues that carbon may be central to the answer. Molecular-dynamics simulations of iron-carbon liquids under inner-core pressures found that increasing the carbon content made the first stable solid nuclei easier to form. An extrapolation to roughly 4 percent carbon by mass reduced the required supercooling to 266 kelvins.

This is one study, not settled consensus. The 4 percent result was extrapolated beyond the carbon concentrations directly validated in the model, and the authors do not claim that Earth’s core is a simple iron-carbon mixture.

Freezing has to get past an atomic barrier

I had always assumed that a liquid freezes as soon as it cools below its melting temperature. In practice, reaching that temperature only makes the solid phase energetically favorable. The first crystal still has to appear.

A tiny cluster of ordered atoms creates an interface between solid and liquid, and that interface costs energy. Unless the cluster grows beyond a critical size, it is more likely to dissolve back into the surrounding liquid. Cooling a liquid below its nominal freezing point increases the odds that a large enough nucleus will form and keep growing.

This extra cooling is called supercooling. At Earth’s center, the amount required matters because the planet’s thermal history and the present size of the inner core limit how far below the melting point the liquid could plausibly have fallen before freezing began.

A 2025 review of inner-core formation put the maximum available supercooling at about 420 kelvins, and suggested that a stricter reading of the geophysical constraints makes less than 100 kelvins more probable. A proposed core composition that requires far more cooling has a serious timing problem.

Carbon changed the nucleation calculation

Lead author Alfred Wilson of the University of Leeds worked with colleagues at Leeds, Oxford and University College London. Their team modeled iron-carbon alloys at pressures of 330 to 360 gigapascals, roughly the range inside the inner core.

The researchers first trained an interatomic model against more computationally demanding calculations of small systems. They then used much larger simulations, some containing 128,000 atoms, to examine melting and the formation of crystal-like clusters. Those clusters are the possible seeds from which a solid inner core could grow.

As the carbon fraction increased, the critical nuclei became easier to form. At 10 mol percent carbon, equivalent to about 2.4 percent by mass, the model required about 481 kelvins of supercooling, with an uncertainty range that overlaps the most generous geophysical limit.

The model could not be validated above 10 mol percent because higher-carbon simulations showed possible phase separation. The team therefore extrapolated the trend to 15.2 mol percent, close to 4 percent by mass and near the upper end proposed by earlier work on Earth’s formation. That produced the 266-kelvin figure behind the headline.

So “3.8 percent carbon” is not a direct sample measurement from the core. It is a model-dependent estimate of a composition that could bring spontaneous nucleation closer to geophysically acceptable conditions.

The core is almost certainly more complicated

We cannot collect material from 5,000 kilometers below our feet. Instead, researchers infer core composition from meteorites, models of Earth’s formation, high-pressure experiments and the speed of seismic waves passing through the planet.

Those seismic measurements show that the core is less dense than pure iron. Candidate lighter ingredients include carbon, oxygen, silicon, sulfur and hydrogen, while nickel is expected to account for a substantial part of the metal.

The new paper adds nucleation as another filter for judging that chemical recipe. Carbon helps with the freezing problem in the simulations, while previously studied iron-sulfur and iron-silicon systems required less plausible supercooling.

An iron-carbon alloy still does not reproduce every seismic observation. The authors write that at least one additional light element is needed to match both the core’s total mass and the density change at the inner-core boundary. Their next-step suggestion is to study more realistic mixtures containing carbon and nickel.

The magnetic field is generated in the liquid outer core

The phrase “the solid heart powers our magnetic field” captures part of the relationship, but the geometry matters. Earth’s field is generated by moving electrically conducting fluid in the liquid outer core, not by the inner core behaving like a permanent bar magnet.

As the solid inner core grows, it releases latent heat and rejects lighter elements into the surrounding liquid. These sources of thermal and chemical buoyancy help drive convection. That moving conductive fluid sustains the geodynamo, and inner-core growth is thought to provide its dominant power source today.

There is an important timeline hidden here. Geological evidence indicates that Earth had a magnetic field long before the inner core formed. Other energy sources, possibly including the precipitation of compounds near the boundary between the core and mantle, must have maintained the earlier dynamo.

The inner core did not create Earth’s first magnetic field. Its continuing growth helps power the field now.

The link to life is elemental, not biological

The same element does sit at both ends of this story. Carbon forms the molecular framework of every known living organism at the surface, and carbon dissolved in metallic iron may have helped the center of Earth crystallize.

That does not mean the core’s carbon came from life or resembles organic material. An element is defined by its atoms, not by the compounds those atoms happen to form. Carbon in a strand of DNA and carbon alloyed with iron under 360 gigapascals share nuclei containing six protons, but almost nothing about their physical setting.

What I like about the study is the new test it gives theories of Earth’s interior. A proposed composition must not only match meteorites and seismic waves. It must also explain how a planet-sized ocean of liquid metal managed to produce its first lasting crystal.