The outer core of Earth spans a temperature range from roughly 3,500 to 4,500 K at its top, where it meets the mantle, to about 5,950 to 6,400 K at its base, where it borders the solid inner core. That works out to something like 5,800°F at the top and over 11,000°F at the bottom. These numbers come with real uncertainty, though, because no instrument has ever been lowered into the core. Instead, scientists squeeze tiny samples of iron and rock to extreme pressures in the lab and compare the results with what seismic waves reveal about the deep Earth, and the estimates from different research groups do not always agree.
Why the Outer Core Is a Range, Not a Single Number
The outer core is not a thin shell with one temperature. It is a roughly 2,260-kilometer-thick layer of liquid iron alloy sitting between about 2,900 and 5,150 kilometers below your feet. Pressure rises enormously with depth, from around 136 gigapascals at the core-mantle boundary (CMB) at its top to roughly 330 gigapascals at the inner-core boundary (ICB) at its bottom. Because the liquid metal is vigorously convecting, its temperature follows what geophysicists call an adiabatic gradient: the temperature rises with depth at a rate set by the compression of the fluid under its own weight. That gradient means the outer core is hottest where the pressure is greatest, at its base, and coolest at the top. Any single “temperature of the outer core” is misleading without specifying which depth you mean.
Temperature at the Top of the Outer Core
The core-mantle boundary sits at roughly 2,900 kilometers depth and about 136 GPa of pressure. Estimating the temperature here is tricky because it depends on the melting behavior of the overlying mantle rock as much as it depends on the iron below. If the lowermost mantle were partially molten everywhere, that would set a floor on the temperature; the fact that it is mostly solid sets an upper bound instead.
A 2014 study published in Science measured the melting point of primitive mantle rock (pyrolite) at pressures near the CMB and found a solidus temperature of about 3,570 ± 200 K. Because the deep mantle is not globally molten, the CMB temperature sits below that threshold.1PubMed Central. Low core-mantle boundary temperature inferred from the solidus of pyrolite A seismological and mineral-physics joint model published in PNAS arrived at a CMB temperature of roughly 3,800 ± 200 K, with lateral variations of a few hundred kelvins explaining much of the seismic heterogeneity observed in the deepest mantle.2Europe PMC. Temperature profile in the lowermost mantle from seismological and mineral physics joint modeling Meanwhile, a 2022 study using seismic shear-wave velocity and attenuation beneath Central America and the Northern Pacific suggested a range of roughly 3,470 to 3,880 K at 95 percent likelihood.3CrossRef (Frontiers in Earth Science). Estimating core-mantle boundary temperature from seismic shear velocity and attenuation
A separate approach looks at the problem from the iron side: if you measure the melting temperature of iron at CMB pressure, you get a value around 4,300 K.4CrossRef (Geophysical Research Letters). Shock Melting Curve of Iron: A Consensus on the Temperature at the Earth’s Inner Core Boundary But the real outer core is not pure iron. It contains lighter elements like sulfur, silicon, oxygen, and possibly hydrogen, which depress the melting point substantially. That depression is exactly why the outer core remains liquid at temperatures below the melting point of pure iron. The low CMB temperature from the pyrolite experiments implies that the impurity-driven melting-point depression is large, a conclusion the authors of that study explicitly noted.1PubMed Central. Low core-mantle boundary temperature inferred from the solidus of pyrolite
Pulling these lines of evidence together, most recent work places the CMB temperature somewhere in the range of about 3,500 to 4,000 K, with a central estimate around 3,700 to 3,800 K. That translates to roughly 5,800 to 6,700°F.
Temperature at the Base of the Outer Core
The bottom of the outer core is defined by the inner-core boundary, where the liquid freezes into solid iron at a pressure of roughly 330 GPa. Pinning down the temperature at this depth means measuring or calculating the melting point of iron (adjusted for impurities) at that extreme pressure.
This has been attempted with several techniques over the past three decades, and the numbers have shifted considerably. A classic 1994 diamond-anvil-cell study in Science reported that iron melts at 6,350 ± 350 K at central-core pressure, with the inner-outer core boundary temperature around 6,130 K.5Science. Temperatures in Earth’s Core Based on Melting and Phase Transformation Experiments on Iron Shock-compression experiments published in Nature found that iron melting is complete by about 260 GPa at 6,100 ± 500 K.6Nature. Melting of iron at the physical conditions of the Earth’s core A more recent shock-wave synthesis study brought these strands closer together, finding the melting temperature at the ICB to be about 5,950 ± 400 K.4CrossRef (Geophysical Research Letters). Shock Melting Curve of Iron: A Consensus on the Temperature at the Earth’s Inner Core Boundary
On the computational side, a 2024 study using machine-learning-driven molecular simulations found most of its data clustering around 6,370 ± 200 K, consistent with recent experimental results reporting about 6,230 ± 500 K at 330 GPa.7ScienceDirect. Melting temperature of iron under the Earth’s inner core condition from deep machine learning The spread across all these studies is roughly 5,500 to 6,600 K, but the central cluster is around 6,000 to 6,400 K. Again, impurities lower the actual freezing temperature of the outer core alloy relative to pure iron, so the real ICB temperature is likely a few hundred kelvins below the pure-iron melting curve.
How Scientists Pin Down Temperatures They Cannot Measure Directly
Nobody can stick a thermometer into Earth’s core. The deepest borehole ever drilled barely scratched 12 kilometers, and the outer core begins at 2,900 kilometers. Every temperature estimate relies on indirect methods, each with its own strengths and weaknesses.
Diamond-anvil cells are the workhorse of high-pressure mineral physics. Two gem-quality diamonds compress a pinhead-sized sample while lasers heat it to thousands of degrees. Researchers watch for the telltale signs of melting, a sudden change in the X-ray diffraction pattern or a jump in the way light scatters, and record the temperature. The technique can reach pressures equivalent to the CMB, but getting all the way to ICB pressure is extremely difficult, and temperature measurements at those extremes carry large error bars.
Shock-compression experiments take a different approach. A high-velocity projectile slams into an iron sample, generating a brief but intense pressure-temperature pulse. By measuring how fast sound travels through the shocked iron, scientists can tell whether it has melted.6Nature. Melting of iron at the physical conditions of the Earth’s core The advantage is that these experiments reach core pressures routinely. The disadvantage is that the conditions last for microseconds, and temperatures must be inferred from equations of state rather than measured with a sensor.
On the computational side, quantum-mechanical simulations model hundreds or thousands of iron atoms on a computer and determine the pressure-temperature point at which the simulated crystal collapses into a liquid. Density functional theory has been the main computational tool for decades.8PubMed Central. Melting of iron under Earth’s core conditions from diffusion Monte Carlo free energy calculations More recently, machine-learning potentials trained on quantum-mechanical data have enabled much larger simulations at a fraction of the computing cost, which is how the 2024 study mentioned earlier arrived at its estimate.7ScienceDirect. Melting temperature of iron under the Earth’s inner core condition from deep machine learning Newer frameworks also assess how sensitive the melting point is to the treatment of electrons at high temperature, a source of systematic uncertainty that earlier studies sometimes glossed over.9CrossRef. Melting point of iron at high pressure: An assessment of uncertainties and effect of electronic temperature
None of these methods alone is definitive. The convergence of diamond-anvil, shock-wave, and computational results over the past decade is what gives the current temperature estimates their weight. Where the three approaches overlap, confidence is highest; where they diverge, the debate continues.
Why the Outer Core Stays Liquid
It might seem paradoxical that the outer core, under tremendous pressure, is liquid while the inner core below it is solid. The explanation is straightforward once you know two things: pressure raises the melting point of iron, and the actual temperature profile of the core rises more slowly with depth than the melting-point curve does. At the inner-core boundary, the two curves cross. Above that crossing point the temperature exceeds the melting point, so the iron alloy stays liquid. Below it the pressure is high enough that the melting point wins, and iron freezes into a solid.
The light elements dissolved in the outer core are crucial to this picture. Pure iron at CMB pressure would have a melting point around 4,300 K, as noted above. The actual CMB temperature appears to be hundreds of kelvins lower than that. That gap is possible only if the impurities push the melting curve down far enough that the alloy remains above its liquidus at the observed temperature. In other words, the outer core is liquid not despite its extreme conditions but because of its chemistry.
Heat Flowing Out of the Core
The temperature of the outer core is not a static number. The core has been cooling since Earth formed, and the rate of that cooling is governed by how efficiently heat escapes upward through the core-mantle boundary. This heat flux matters for everything from mantle convection to the strength of Earth’s magnetic field.
Estimates of total CMB heat flux range widely, from about 5 to 17 terawatts (TW) depending on the method. Numerical models of mantle convection that track the supercontinent cycle give CMB heat fluxes of roughly 13 to 15 TW.10Oxford Academic. Changes in core–mantle boundary heat flux patterns throughout the supercontinent cycle A study using experimentally measured thermal conductivity of lower-mantle minerals arrived at about 11 ± 1.4 TW.11Europe PMC. Lattice thermal conductivity of lower mantle minerals and heat flux from Earth’s core Those numbers are large enough to matter for the planet’s overall heat budget but small enough that the core cools very slowly on human timescales.
Part of the core’s internal heat may come from radioactive decay. Potassium-40, though mostly associated with the rocky mantle and crust, can dissolve into iron-sulfur melts at high pressure and temperature. Experiments have shown that this process is strongly temperature-dependent, meaning potassium could serve as a meaningful heat source inside the core.12Nature. Experimental evidence that potassium is a substantial radioactive heat source in planetary cores If the core contains enough potassium-40, it would slow the rate of cooling and push the age of the inner core’s formation earlier in Earth’s history.
A Stratified Layer at the Top of the Outer Core
One of the more surprising recent findings is that the uppermost few hundred kilometers of the outer core may not be convecting at all. If the thermal conductivity of the liquid iron alloy is high enough, heat can escape by conduction alone, without needing convective overturning. That would produce a thermally stratified layer, a zone where temperature increases with depth too gently for buoyancy-driven flow to kick in.
First-principles calculations published in 2012 found both the thermal and electrical conductivity of iron at core conditions to be two to three times higher than older estimates. The resulting adiabatic heat flux at the CMB would be around 15 to 16 TW, potentially exceeding the actual CMB heat flux driven by mantle convection. If the heat conducted down the temperature gradient exceeds the total heat being removed by the mantle, convection in the uppermost core shuts off.13arXiv. Thermal and electrical conductivity of iron at Earth’s core conditions A subsequent experimental study on iron-silicon alloys found thermal conductivity near the topmost outer core on the order of 100 to 110 W/m/K, supporting a thermally stratified layer roughly 400 to 500 kilometers deep below the CMB.14PubMed Central. Thermal conductivity of Fe-Si alloys and thermal stratification in Earth’s core
A stratified layer has real consequences. It affects the way seismic waves travel through the outermost core, and it complicates models of the geodynamo, the process that generates Earth’s magnetic field. If the top of the core is stagnant, any convection needed to drive the dynamo must be chemical rather than thermal, powered by the release of light elements as the inner core freezes, or by lateral variations in how much heat the mantle pulls out at different locations.
The Core-Mantle Boundary Is Not the Same Temperature Everywhere
It is tempting to think of the CMB as a smooth surface with a single temperature, but the reality is messier. The mantle above the boundary is not uniform: dense slabs of former ocean floor sink down to the CMB in some regions, while hot mantle plumes rise from it in others. These structures create significant lateral variation in how quickly heat is extracted from the core.
Numerical models tracking the last billion years of Earth’s history show that the ratio of peak-to-peak heat-flux variation to the average heat flux at the CMB (a parameter geophysicists call q*) is at least 2 and can easily exceed 30, depending on material properties and the configuration of subduction zones at the surface.10Oxford Academic. Changes in core–mantle boundary heat flux patterns throughout the supercontinent cycle This means the heat leaving the core beneath a cold, recently subducted slab could be many times higher than beneath a hot mantle plume root. The joint seismological model noted earlier found that lateral temperature variations of roughly 200 to 300 K are enough to explain much of the seismic heterogeneity in the deep mantle.2Europe PMC. Temperature profile in the lowermost mantle from seismological and mineral physics joint modeling
These variations change over geologic time. As supercontinents assemble and break apart, the pattern of subduction shifts, and with it the map of heat flow at the CMB. The shape and number of hot and cold structures at the boundary fluctuate through the supercontinent cycle and may have looked very different in Earth’s past.10Oxford Academic. Changes in core–mantle boundary heat flux patterns throughout the supercontinent cycle That means the “temperature of the outer core” is not just a range because of measurement uncertainty; it is genuinely different from place to place and from era to era.
What the Outer Core’s Chemistry Tells Us About Its Temperature
Because the outer core is a liquid alloy rather than pure iron, its temperature profile depends on what else is dissolved in it. The leading candidates for light elements in the core are oxygen, silicon, sulfur, hydrogen, and carbon, but their proportions remain debated. Each element depresses the melting point by a different amount and changes the density and sound speed of the liquid differently.
Sound-velocity measurements offer one way to constrain the chemistry. Inelastic X-ray scattering experiments on a liquid iron-carbon alloy found that the sound speed of the outer core, about four percent faster than that of pure liquid iron, is consistent with about four to five atomic percent carbon in the mix.15Nature. Carbon-depleted outer core revealed by sound velocity measurements of liquid iron–carbon alloy Silicon is another strong candidate; the thermal-conductivity experiments on iron-silicon alloys described earlier tested compositions with significant silicon content precisely because geochemical models favor it.14PubMed Central. Thermal conductivity of Fe-Si alloys and thermal stratification in Earth’s core
The identity and amount of these impurities feed back into temperature estimates in two ways. First, they set how far below the pure-iron melting curve the actual freezing point of the outer core alloy falls. A large depression means the inner core could have formed relatively recently in Earth’s history, because the core had to cool further before solidification could begin. Second, as the inner core grows and expels light elements into the liquid above, it changes the composition of the outer core over time, which in turn shifts the melting curve. The temperature at which the outer core will eventually freeze completely is a moving target that depends on both how fast the core is cooling and how the chemistry is evolving.
Powering the Magnetic Field Before the Inner Core Existed
Earth’s magnetic field is at least 3.5 billion years old, but the inner core is thought to be much younger, perhaps only one to two billion years old by most estimates. Before the inner core began to crystallize, there was no release of latent heat and no chemical buoyancy from light elements being rejected at the ICB. The geodynamo had to run on thermal convection alone.
This puts tight constraints on what the outer core’s temperature was doing in the distant past. The energy available to drive the dynamo was limited to the secular cooling of the core and any radiogenic heat from elements like potassium-40.16Elsevier. The inner core and the geodynamo Without enough heat flux across the CMB, the dynamo would have stalled. Maintaining a magnetic field for billions of years before the inner core formed requires either that the early core was significantly hotter (providing a large thermal driving force) or that an additional energy source, such as radioactive decay or exsolution of light oxides, supplemented the budget.
Recent large-scale atomistic simulations using machine-learning potentials have explored one such supplementary mechanism: the precipitation of magnesium oxide from the liquid core. As the core cooled below the saturation limit for dissolved magnesium, solid magnesium-rich ferropericlase crystals would have formed and floated upward, releasing gravitational energy in a manner analogous to the chemical buoyancy from inner-core growth. These simulations suggest that the precipitate takes the form of crystalline, iron-poor ferropericlase rather than a liquid oxide phase, which affects how much energy the process delivers and over what timescale.17CrossRef (Geophysical Research Letters). Large‐Scale Atomistic Simulations of Magnesium Oxide Exsolution Driven by Machine Learning Potentials: Implications for the Early Geodynamo If this mechanism worked efficiently, it could have kept the dynamo running during the long interval before the inner core began to freeze, without requiring the early outer core to have been unrealistically hot.