Diamagnetic Levitation: How Graphite Floats Above Magnets

The tiny currents that push graphite away

Diamagnetism is universal. Put any material in a magnetic field and the orbital motion of its electrons responds by the atomic-scale version of Lenz's law: the electron orbits precess (Larmor precession), setting up microscopic current loops whose magnetic moment opposes the applied field. The material is therefore pushed away from regions of strong field. The classical Langevin result gives a negative susceptibility χ = −μ₀ N e²⟨r²⟩ / 6mₑ, proportional to the mean-square orbital radius ⟨r²⟩ and independent of temperature.

For most substances this is vanishingly weak — water sits at χ ≈ −9×10⁻⁶. Graphite is special. Its delocalised π electrons circulate in large, aromatic-ring-like orbits spread across the graphene planes, so when the field points perpendicular to the sheets it drives wide ring currents with a huge effective ⟨r²⟩. The response is strongly anisotropic: along the c-axis (field crossing the layers) χ⊥ ≈ −4.5×10⁻⁴ (sources range from about −4 to −6 ×10⁻⁴), while in-plane it is only ≈ −10⁻⁵, some 40–50× smaller. That c-axis value makes pyrolytic graphite one of the most strongly diamagnetic materials at room temperature — though still a world away from a superconductor's perfect χ = −1. A free flake even rotates itself layers-horizontal, so its strongest diamagnetic axis points straight up out of the magnets.

The force law and how strong a magnet you need

A linear magnetic material in a field carries energy density U/V = −(χ/2μ₀) B². For a diamagnet χ < 0, so this is +(|χ|/2μ₀) B² — positive, and minimised where B is smallest. The flake is therefore pushed down the gradient of B², toward weak field, with a force per unit volume

f = −(|χ|/2μ₀) ∇(B²).

Levitation means this upward push balances gravity, ρg per unit volume:

(|χ|/2μ₀) |∂(B²)/∂z| = ρg → |∂(B²)/∂z| = 2μ₀ρg / |χ|.

Plugging in graphite's ρ ≈ 2200 kg/m³ and |χ| ≈ 4.5×10⁻⁴ gives a required gradient of about 120 T²/m. A neodymium (NdFeB) magnet has a surface field of roughly 0.3–0.5 T, so B² ≈ 0.16 T², and above a small-magnet checkerboard that field collapses over only a millimetre or two. That yields ∂(B²)/∂z of order 100 T²/m right at the surface — just enough, which is exactly why the graphite floats a fraction of a millimetre up and no higher: climb any further and the gradient, which falls off steeply, can no longer hold the weight.

Step by step: from magnet array to floating flake

1. The array. Four or more NdFeB cube magnets are set in a checkerboard, poles alternating N-up / S-up. This maximises the field near the surface, steepens its gradient, and — crucially — creates local minima of B² in the seams between adjacent magnets.
2. The response. The graphite's electrons develop opposing moments (χ < 0), so the flake carries energy proportional to +B².
3. The lift. Being energetically pushed toward weak field, it rises away from the strong-field surface.
4. Vertical balance. It settles at the gap where the gradient force exactly cancels gravity.
5. Sideways trapping. The B² minima over the seams pin it laterally, and torque aligns its c-axis vertical — so it hovers flat and centred rather than sliding off.
6. Passive and static. No current, no cooling, no moving parts — a ball resting in an invisible magnetic valley.

Why it doesn't fall — beating Earnshaw's theorem

Here is the deep part. Samuel Earnshaw proved in 1842 that you cannot hold a paramagnet or a permanent magnet in stable static equilibrium using only magnetostatic forces in free space — there is no local energy minimum, so it always slides or flips away. That is why you can't balance one magnet in mid-air above another (the Levitron cheats with a spinning gyroscope), and why maglev needs active feedback or motion.

Diamagnets are the loophole. In a current-free region the quantity B² is subharmonic: ∇²(B²) ≥ 0. That inequality forbids a local maximum of B² but explicitly allows a local minimum. Since a diamagnet's energy is proportional to +B², it seeks that minimum and sits there stably in all three directions. Werner Braunbek worked this out and demonstrated it in 1939, levitating small pieces of bismuth and graphite in an electromagnet — the first genuinely stable magnetic levitation, and living proof that Earnshaw's theorem quietly exempts materials with χ < 0.

A worked comparison: graphite, bismuth, and a levitating frog

Whether something floats over a given magnet depends only on the ratio |χ|/ρ — a material figure of merit, since both the lifting force and the weight scale with volume. Compare:

• Pyrolytic graphite: |χ|/ρ ≈ 4.5×10⁻⁴ / 2200 ≈ 2.0×10⁻⁷ m³/kg
• Bismuth: ≈ 1.66×10⁻⁴ / 9780 ≈ 1.7×10⁻⁸ m³/kg — about 12× weaker (dense metal, modest χ)
• Water (a frog is mostly water): ≈ 9×10⁻⁶ / 1000 ≈ 9×10⁻⁹ m³/kg — about 23× weaker

Graphite's edge is why a desktop NdFeB checkerboard suffices, while bismuth barely manages and water needs monstrous fields. Run the numbers for a frog: 2μ₀ρg/|χ| ≈ 2.7×10³ T²/m. Geim and Berry's frog levitated in the Nijmegen 16 T Bitter magnet, where the field product B·(dB/dz) ≈ 1400 T²/m — that is ∂(B²)/∂z ≈ 2800 T²/m, right on the money. Graphite's equivalent requirement is only about 60 T²/m of B·(dB/dz), a job cheap permanent magnets do easily.

History, real uses, and a common misconception

Sebald Brugmans first noticed bismuth and antimony being repelled by a magnet in 1778; Michael Faraday coined the word 'diamagnetic' in 1845 while studying bismuth and 'heavy glass'; Braunbek achieved stable levitation in 1939; and in 1997 Michael Berry and Andre Geim famously floated a live frog (and strawberries, and water) in a 16 T magnet — earning Geim the 2000 Ig Nobel Prize a decade before his graphene Nobel. Pyrolytic-graphite-on-magnets has since become a classic desktop demo.

It is more than a toy. Levitated graphite is a contactless, frictionless force transducer — the basis for sensitive accelerometers, gravimeters, and micro-bearings. Because graphite's susceptibility depends on temperature, a laser spot that warms one edge lowers its diamagnetism locally, unbalancing the forces and driving optically powered graphite rotors and actuators. Strong-field diamagnetic levitation also simulates microgravity for cells and protein-crystal growth.

The misconception: that this is superconductivity or the Meissner effect. It is not. Graphite is a weak diamagnet (χ ≈ −10⁻⁴) at room temperature, not a perfect one (χ = −1); there is no cooling, no zero resistance, no flux pinning — only the faint push of orbital electrons. And a neat subtlety people miss: the effect is size-independent. Since force and weight both scale with volume, a large flake floats exactly as readily as a tiny one — provided it stays close to the surface, where the field gradient is steep enough to hold it.

Three ways to levitate something with magnetism, and where diamagnetic graphite fits

Property	Diamagnetic (pyrolytic graphite)	Superconductor (Meissner + pinning)	Eddy-current (electrodynamic)
Physical origin	Orbital-electron diamagnetism, χ < 0	Perfect diamagnetism, flux expulsion + pinning	Faraday/Lenz currents induced by motion
Susceptibility χ	≈ −4.5×10⁻⁴ (weak)	−1 (ideal, perfect diamagnet)	Not applicable (motional)
Cooling needed	None — room temperature	Yes (below Tc; e.g. 77 K for YBCO)	None
Passive static stability	Yes (stable at a B² minimum)	Yes (flux pinning locks it)	No — needs relative motion or AC
Field / drive required	Strong permanent magnets, 0.3–0.5 T, steep gradient	Moderate static field	Strong AC field or fast motion
Everyday example	Graphite on an NdFeB checkerboard	YBCO puck on a magnet track	Maglev trains, the jumping-ring demo