Iron melts at 4,160 K under core-like pressure, hinting at a cooler Earth core

More than 300 melting measurements on pure iron trace a shallow melting curve that reaches 4,160 ± 80 K at 200 gigapascals, which the Carnegie-led team extends to infer a core-mantle boundary of about 3,430 K and only 9.8 terawatts of heat conducted through the outer core.

Peer reviewed · Laboratory experiment · Model-based extrapolation to the core

Iron melts at a lower temperature under core-like pressure than earlier experiments suggested, according to a study published online in Science on 8 November 2026. Zachary Geballe, Michael Walter and 12 colleagues, mostly at the Carnegie Institution for Science, heated tiny iron samples with microsecond pulses of electric current inside diamond anvil cells. Iron’s melting point anchors estimates of how hot Earth’s core is, and the core’s churning liquid iron generates the planet’s magnetic field. The catch: the abstract reports results up to 200 GPa, and the inner-core boundary sits near 330 GPa, so core temperatures are extrapolated.

Evidence at a glance

Is this a big deal?

For the measurement, yes: a new heating method and hundreds of consistent runs on one of the most argued-over numbers in geophysics. For what it says about the core, it is a solid step that still needs checking. The conclusions about a cool core and an easier-running early dynamo hang on extrapolations that other groups will want to test.

What scientists found

Along that curve, at 200 GPa, nearly 2 million times the air pressure at sea level, pure iron melted at 4,160 ± 80 K, or about 3,890 °C. That is still roughly 1,600 K cooler than the visible surface of the Sun, which NASA lists at 5,772 K. (For more on that comparison, see how hot the Sun is.)

The team calls the melting curve “surprisingly shallow.” By 200 GPa it rises only about 6 K per gigapascal. In plain terms, squeezing iron by another 10 GPa raises its melting point by only about 60 K.

The group also measured how strongly liquid iron resists electric current: 76 microhm-centimetres (µΩ cm). That number matters for heat, as the next sections explain.

How they found it

A diamond anvil cell squeezes a speck of material between the tips of two gem diamonds. The 2013 study discussed below heated its samples with lasers. The Carnegie group instead ran short bursts of electric current through the iron itself, a technique called Joule heating. It is the same effect that makes a toaster wire glow, compressed into pulses lasting millionths of a second.

In a 2025 conference abstract, Walter, Geballe and co-author Francesca Miozzi described the kit: an amplifier that drives pulses of up to 60 amperes, a five-colour light-measurement system that reads the sample’s temperature from its glow, and a four-point electrical probe for resistivity. The new paper’s abstract says the method both identifies melting and measures resistivity in solid and liquid iron. It does not describe the melting signature itself, so we don’t either.

Why it matters to you

Earth’s magnetic field is made by molten iron churning in the outer core, a process called the geodynamo. That field deflects charged particles from the Sun and makes a compass work. How hot the core is, and how fast it leaks heat into the rock above, decides how long and how strongly that engine can run. This study won’t change anything about daily life, but it changes the numbers scientists feed into models of the field’s past.

What the result implies for the core

The core has two layers: a liquid outer core and a solid inner core, which grows as the liquid freezes. The temperature at that boundary is expected to be close to the melting point of the core’s iron at about 330 GPa, so a lower melting curve points to a cooler core overall. (Our guide to the layers of Earth shows where each boundary lies.)

The abstract does not give a single temperature for the inner-core boundary. Instead, the team extrapolated in pressure and in alloy composition, since the core holds lighter elements as well as iron. From that, they infer a “relatively cold” core-mantle boundary of 3,430 K, about 3,160 °C.

The resistivity feeds into the heat budget. In a metal, electrons carry both current and heat, so a metal that resists current also conducts heat less well. Combining the 76 µΩ cm value with their temperatures, the authors calculate an adiabatic heat flow of 9.8 terawatts through the outer core. That is the heat conduction alone would carry out along the core’s natural temperature gradient. If the core loses heat faster than that, the extra heat can drive convection, so a lower figure makes heat-driven stirring easier.

Two conclusions follow, in the authors’ words, which use “suggests” and “could.” Thermal convection alone could have powered the geodynamo before the inner core formed. And the lowermost mantle today is fully solid.

What the evidence does not show

It does not measure the core. Nobody can sample it, so every core temperature comes from lab data plus models. Here the gap is large: the abstract reports measurements to 200 GPa, and the inner-core boundary is near 330 GPa.

It also measures pure iron. The core’s light elements lower the melting point by an amount that depends on which elements are present and how much of each, and the abstract does not list the composition the team assumed.

As a rough check, SciQuest extended the 6 K/GPa slope in a straight line from 200 to 330 GPa: 4,160 K + (6 K/GPa × 130 GPa) = about 4,940 K. If the curve keeps flattening, the true value for pure iron would be lower still. This is our arithmetic, not a number from the paper.

How this fits what we already knew

Iron’s melting point at core pressures has long been disputed; the 2013 paper itself noted little consensus. In 2013, a team led by Simone Anzellini of France’s Atomic Energy Commission (CEA) used laser-heated diamond cells and fast X-ray diffraction up to 200 GPa. Extrapolating, they put melting at the inner-core boundary at 6,230 ± 500 K and wrote that this favoured high heat flow out of the core and possible partial melting of the mantle above it.

Shock experiments approach the problem differently, slamming iron to high pressure for an instant. In 2004, J. H. Nguyen of Lawrence Livermore National Laboratory and N. C. Holmes reported that shocked iron starts to melt at 225 ± 3 GPa and 5,100 ± 500 K. The pressures differ, but the new curve runs lower than both: 4,160 K at 200 GPa sits under the 2004 onset at 225 GPa, and even our straight-line check gives only about 4,940 K at 330 GPa, against 6,230 K in 2013. Unlike the 2013 paper, it points toward low heat flow and a solid lowermost mantle. (Shock methods have also been used on other deep-Earth materials, such as magnesium oxide.)

Independent perspective

SciQuest has not yet found on-record comment from researchers outside the team, and we have not interviewed anyone for this story. We will add independent views if they become available.

How other headlines framed it

Only one outlet had covered the paper when we checked. Science‘s news section led with the idea that the core may be much cooler than scientists assumed. SciQuest could not open that article, so we have not used any detail from it. Our framing puts the measured number first, because the cooler core is an inference that rests on extrapolation.

What happens next

The obvious test is to push pulsed heating closer to 330 GPa and to run iron alloys with light elements, which would shrink the two extrapolations behind the core numbers. Independent groups can also reanalyse the raw spectroradiometry and electrical data, which the team has posted openly on Zenodo. Thermal-history models of the geodynamo will need rerunning with the lower temperatures and heat flow; we will cover those as they appear.

In one sentence

Hundreds of microsecond-heating runs trace a shallow melting curve for iron that reaches 4,160 K at 200 GPa, lower than earlier estimates, and if the team’s extrapolation holds, Earth’s core is cooler and its early magnetic field easier to explain than many models assumed.

Read next

• From crust to core: the layers of Earth, for where the core-mantle and inner-core boundaries sit.
• Shocked magnesium oxide melts at 13,200 K, another experiment on how deep-Earth materials melt.

Sources and disclosure

1. Geballe Z. M., Yang J., Dobrosavljevic V. V., Miozzi F., Sime N., Vitale S. M., Driscoll P. E., van Keken P. E., Tappero R., Liu W., Chariton S., Prakapenka V. B., Meng Y., Walter M. J. “Melting and resistivity of iron at Earth core conditions using pulsed Joule heating.” Science, published online 8 November 2026. doi:10.1126/science.aeh5267 (abstract and metadata via Crossref).
2. Geballe Z. M. et al. “Data for: Melting and resistivity of iron at Earth core conditions using pulsed Joule heating.” Zenodo, 26 October 2026. zenodo.org/records/23668895
3. Walter M., Geballe Z., Miozzi F. “Pulsed Joule heating of iron to >1.5 Mbar and >4000 K.” EMPG 2025 conference abstract. empg2025.sciencesconf.org/628948
4. Anzellini S., Dewaele A., Mezouar M., Loubeyre P., Morard G. “Melting of iron at Earth’s inner core boundary based on fast X-ray diffraction.” Science 340, 464–466 (2013). doi:10.1126/science.1233514
5. Nguyen J. H., Holmes N. C. “Melting of iron at the physical conditions of the Earth’s core.” Nature 427, 339–342 (2004). doi:10.1038/nature02248
6. NASA Goddard Space Flight Center. Sun fact sheet (effective temperature 5,772 K).

Disclosure: This article was drafted with AI assistance and checked against the sources above. SciQuest read the paper’s abstract, not the full text, which the publisher’s bot check blocked. Conversions (200 GPa ≈ 1.97 million atmospheres; 4,160 K ≈ 3,887 °C; 3,430 K ≈ 3,157 °C) and the 4,940 K straight-line check are SciQuest arithmetic, and the graphic is a SciQuest original drawn from the cited values. Spotted an error? See our corrections policy.