Presets
Playback
Superconductor
Magnet and gravity
Cooling and actions
Camera
Layers
Quality
Model check
The science
What is shown
A small neodymium magnet floats above a bulk high-temperature superconductor (YBCO, Tc about 92 K) that has been cooled in liquid nitrogen (77.35 K). Above Tc the disc is an ordinary metal and the magnet simply lies on it. Below Tc the resistance drops to zero and the disc expels magnetic flux (Meissner effect), so the magnet is pushed up until the repulsion balances its weight.
Equations used
- Superconducting fraction through the transition:
S = smoothstep((Tc - T) / dT); superfluid fraction (two-fluid law)f = S (1 - (T/Tc)^4). - Resistance:
R = R_n(T) (1 - S)with a metal-likeR_n ~ 0.12 + 0.88 T/295; exactly zero below the transition. - London penetration depth:
lambda(T) = lambda0 / sqrt(1 - (T/Tc)^4), lambda0 = 150 nm (YBCO, illustrative). - Image method: the normal field component vanishes at a perfect diamagnet, so the screening currents act like a mirror magnet at depth -z with reversed magnetisation. A magnet is modelled as a solenoid current sheet
K = Br / mu0, cut into 10 loops. - Loop-to-loop mutual inductance
M = mu0 sqrt(a1 a2) [ (2/k - k) K(k) - (2/k) E(k) ],k^2 = 4 a1 a2 / ((a1 + a2)^2 + d^2); force between two loopsF = I1 I2 dM/dd. Total liftF(z) = s f (Br/mu0)^2 (L/N)^2 sum_ij ( -dM/dd )withd_ij = 2z + s_i + s_j. - Far-field limit (a check in this page): a point dipole m = Br V / mu0 gives
F = 3 mu0 m^2 / (32 pi z^4), so a free magnet hovers atz = (3 mu0 m^2 / (32 pi M g))^(1/4)and its vertical oscillation frequency isomega^2 = 4 g / z. - Motion:
M z'' = F(z) - M g - c z'integrated with 0.2 ms steps; lateral flux pinning is a phenomenological springF_x = - P (Mg / 4 mm) x. - Thermal model: Newton cooling
dT/dt = (T_bath - T)/tauwith tau = 8 s in the bath and 500 s in air; the bath evaporates in about 100 s.
Honest simplifications
The disc is treated as an infinite flat perfect diamagnet scaled by a shielding efficiency s. A real bulk YBCO disc is a type-II superconductor: the magnet's field at the surface is far above the lower critical field Bc1 (about 10-30 mT), so flux enters and is pinned in vortices. That pinning is what makes real levitation stable laterally and lets a magnet hang under the disc, which a pure Meissner model cannot do. Real hover gaps are therefore usually a few millimetres, smaller than the ideal-plane values shown for s = 1. The finite disc size, the field-cooled versus zero-field-cooled history, flux creep and heating by a moving magnet are not modelled. The field lines are those of a softened point dipole plus its image, drawn for orientation only.
Things to try
- Pour nitrogen on a warm disc (preset "Warm disc") and watch the magnet lift as T crosses Tc.
- Add brass weights until the magnet no longer lifts: the "Min shielding to lift" readout shows the limit.
- Pick niobium (Tc 9.25 K): nitrogen is not cold enough, nothing happens.
- Set pinning to 0 and nudge the magnet: it slides off the disc.
- Moon gravity: the magnet hovers higher (z scales as g^-1/4).