The Geodynamo: How Earth's Core Makes a Magnet

The paradox: a magnet too hot to be magnetic

Start with a puzzle that stumped physicists for centuries. Hold a compass anywhere on Earth and the needle swings to point roughly north. In 1600 the English physician William Gilbert published De Magnete and drew the natural conclusion: the whole planet, he argued, is one giant spherical lodestone, magnus magnes ipse est globus terrestris — "the Earth itself is a great magnet." For three hundred years that picture seemed good enough.

The trouble is temperature. Every permanent magnet — a fridge magnet, a compass needle, a lump of lodestone — owes its field to countless electron spins locked into alignment inside solid iron. Heat the iron past its Curie point of about 770 °C and that alignment shakes apart; the material becomes an ordinary, non-magnetic metal. Earth's core sits at somewhere between 4,000 and 5,700 K — thousands of degrees above the Curie point. A permanent iron magnet down there could not exist for an instant.

So the field cannot be a frozen-in relic. It must be actively made, continuously, by something happening inside the planet right now. That something is the geodynamo: not a stored magnet but a running machine, converting the mechanical energy of a stirring liquid metal into electrical and magnetic energy — and it will keep running only as long as the stirring continues.

What is actually down there

Seismic waves passing through the planet let us map its interior the way an ultrasound maps a body. Below the rocky mantle, starting at a depth of about 2,890 km, lies the core-mantle boundary — a sharper contrast than the ground-to-air boundary at your feet. Beneath it the core divides into two very different parts:

• The outer core — a shell of molten iron-nickel alloy roughly 2,200 km thick, extending from ~2,890 km down to ~5,150 km. It is genuinely liquid: seismic shear waves, which cannot travel through a fluid, are blocked here, which is precisely how we know it is molten.
• The inner core — a solid ball of nearly pure iron about 1,220 km in radius, roughly 70% of the Moon's radius (~1,737 km). It is even hotter than the outer core but stays solid because the crushing pressure — over 3 million times atmospheric — raises iron's melting point above the local temperature.

Two facts about this arrangement make the dynamo possible. First, liquid iron is an excellent electrical conductor, so moving it can carry electric currents. Second, the outer core is convecting: as the planet slowly loses heat to space, the base of the outer core cools, iron crystallizes onto the growing inner core, and buoyant, lighter material is released and rises. This freezing-driven convection — plus the buoyancy of light elements like sulfur and oxygen left behind — is the engine. Radioactive heating and residual heat from Earth's formation top up the energy budget. Without a fluid, conducting, convecting layer, a planet gets no dynamo — which is why the Moon and Mars, whose cores have largely frozen or cooled, have no global field today.

How churning metal becomes a magnet

Here is the heart of it. Take a blob of conducting fluid that already sits in a weak magnetic field — even a whisper of a field left over from the young Solar System will do. Now move that fluid across the field lines. By the same physics that powers every generator in every power station, a moving conductor in a magnetic field drives an electric current. And every electric current, in turn, wraps itself in a magnetic field of its own.

A dynamo is self-exciting when the geometry is arranged so that the new field reinforces the original field instead of cancelling it. The flowing metal makes currents; the currents make field; the field, dragged by more flowing metal, makes still more current — a feedback loop that regenerates the field faster than electrical resistance can bleed it away. The seed field is amplified until losses and generation balance, and the machine settles into a steady, self-sustaining state.

Two ingredients shape the specific field we get. The first is convection: hot buoyant iron rises, cool dense iron sinks, in great columns. The second is rotation: Earth spins once a day, and the Coriolis effect twists those rising columns into corkscrews aligned roughly with the rotation axis. This is why the field is not a random tangle but a broadly dipolar one — two poles, near the geographic poles — that a compass can actually follow. The alignment of the magnetic axis with the spin axis is no coincidence; rotation organizes the flow. Numerical geodynamo simulations, first run successfully in the mid-1990s, reproduce this dipole-dominated field and even spontaneously produce reversals, which is strong evidence the basic picture is right.

The numbers, and how weak the field really is

For all its planetary importance, Earth's field is feeble up here where we live. At the surface it ranges from about 25,000 nT near the equator to 65,000 nT near the poles — that is roughly 25 to 65 microtesla (µT), or about half a gauss on average. A cheap refrigerator magnet, held against your skin, is on the order of a hundred times stronger locally. The planetary field only looks powerful because it fills all of space around Earth out to tens of thousands of kilometers, carving a protective bubble — the magnetosphere — that stands off the solar wind.

The field is also restless. The magnetic poles wander: the north magnetic pole has been sprinting across the Arctic toward Siberia at up to 50–60 km per year in recent decades, fast enough that aviation and navigation charts must be updated. And the overall dipole is weakening. Since systematic measurements began in 1840, the global field strength has fallen at roughly 5% per century; it is now about 9% weaker than in 1840. Much of that loss is concentrated in the South Atlantic Anomaly, a broad region over the South Atlantic and South America where the field is especially weak and low-orbiting satellites suffer more radiation glitches.

It is tempting to read the weakening as the onset of a reversal, but that leap is not justified. Studies from missions like the ESA Swarm satellite trio (launched 2013), building on earlier magnetic surveys (Ørsted and CHAMP), show the field has been weak-and-recovering before without flipping. The honest statement is that we are watching normal, if vigorous, variation — and we genuinely cannot yet predict whether a reversal is centuries or hundreds of thousands of years away.

Reversals: when north becomes south

The single strangest thing the geodynamo does is turn itself inside out. Roughly every few hundred thousand years — on average about 4 to 5 times per million years at present — the field weakens, becomes tangled and multipolar, and then re-establishes itself with north and south swapped. There have been on the order of several hundred documented reversals in the last ~160–170 million years — including roughly 183 reversals in the last 83 million years.

We know this because the ocean floor keeps the record. As new crust erupts along mid-ocean ridges and cools through the Curie point, iron-bearing minerals freeze in the direction of the ambient field like a magnetic tape. Spreading seafloor lays down symmetric stripes of alternating polarity on either side of each ridge — a discovery in the 1960s that both confirmed reversals and clinched the case for plate tectonics. The most recent full reversal, the Brunhes–Matuyama reversal (named for Bernard Brunhes and Motonori Matuyama), occurred about 780,000 years ago; the transition itself may have taken on the order of 1,000 to 20,000 years, brief by geological standards.

Crucially, reversals are not periodic — you cannot set a clock by them. Sometimes the field goes hundreds of thousands of years between flips; sometimes it holds a single polarity for tens of millions of years. These long quiet spells are called superchrons: during the Cretaceous Normal Superchron, from roughly 121 to 84 million years ago, the field did not reverse at all. Reversals also come in a milder flavor, the geomagnetic excursion, in which the field lurches and weakens dramatically but snaps back to its original polarity without completing the switch — the Laschamp excursion, about 41,000 years ago, is the best-studied example.

Why it matters, and what we still don't know

The geodynamo is not a curiosity — it is part of why Earth is habitable. The magnetosphere it builds deflects most of the solar wind and shields the atmosphere and surface from the worst of the charged-particle radiation. When the Sun hurls a coronal mass ejection at us, the field takes the punch: the Carrington Event of 1859, the most intense recorded geomagnetic storm, drove auroras seen near the tropics and set telegraph lines sparking. A storm of that size today could damage power grids and satellites worth trillions. The same field also produces the aurora borealis and australis, funnels particles into the Van Allen belts, and — via magnetized rocks — hands geologists a natural clock and a map of ancient continents.

A persistent misconception is worth killing: a reversal is not a doomsday. During a flip the field weakens and gets messier, but it does not vanish, and there is no reliable evidence in the fossil or geological record that past reversals caused mass extinctions. Compasses would misbehave and satellites would need better shielding, but life — which has weathered hundreds of reversals — carries on.

What remains genuinely open is deep and specific. We do not know exactly when the inner core began to freeze — estimates range from over a billion to as little as a few hundred million years ago, and the answer changes our whole story of how the dynamo has been powered over Earth's history. We cannot yet predict reversals or the next lurch of the poles. And even our best supercomputer models run at conditions far from the real core's turbulence, so they capture the physics in spirit rather than in exact detail. The geodynamo is one of the few planetary-scale machines humans have never seen and can only infer — running silently, 2,900 kilometers beneath our feet, keeping our compasses honest and our sky at bay.

A permanent magnet versus the geodynamo — why Earth cannot be a simple bar magnet

Property	Bar magnet (permanent)	Geodynamo (Earth's core)
Source of magnetism	Aligned electron spins locked in solid iron	Electric currents in flowing liquid metal
Survives above 770 °C?	No — loses magnetism at the Curie point	Yes — core is 4,000+ K yet still generates a field (via electric currents, not ferromagnetism)
Needs energy input?	No, but slowly demagnetizes	Yes — convection powered by cooling & freezing core
Can flip polarity?	Only if physically re-magnetized	Yes — reverses on its own, last time ~780,000 yr ago
Field geometry	Fixed and steady	Wanders; poles drift tens of km per year