Weiss, 1907
Pierre Weiss proposed that iron is strongly magnetized everywhere, but divided into regions whose directions differ. An unmagnetized piece of iron is then one whose regions cancel.
P. Weiss, J. Phys. Theor. Appl. 6, 661 (1907).
Quantum Spin Lab · Magnetic domains
In iron, the spins of neighbouring electrons line up. The aligned regions are called domains. In an ordinary piece of iron they point different ways and cancel. A field makes the right ones grow. Heat destroys the order.Apply a field, then heat it.
A model of the ideaEach cell is a small block of iron with one magnetization direction. Neighbours prefer to agree, a net magnetization costs energy, and random defects hold the walls in place. The colours fade as heat weakens the alignment inside each domain. Real domains are far smaller than drawn.
Net magnetization
Start here: make a magnet, then destroy the order
The plate is not magnetized: its domains point both ways and cancel.
From the two coils, along the plate. Switch it off again and part of the magnetization stays: defects hold the walls. That is how a permanent magnet keeps its strength.
Iron loses its magnetic order at 770 °C, its Curie point.
Two levels of order. Inside each domain, electron spins line up with their neighbours. That alignment weakens with heat and is gone at 770 °C. The domains themselves point different ways until a field makes the favourable ones grow.
Real experiments
Domains were first an idea that explained how iron can be magnetic inside and still show no net magnetism. Then they were heard, and then photographed.
Pierre Weiss proposed that iron is strongly magnetized everywhere, but divided into regions whose directions differ. An unmagnetized piece of iron is then one whose regions cancel.
P. Weiss, J. Phys. Theor. Appl. 6, 661 (1907).
Heinrich Barkhausen wound a coil around iron, slowly changed the field, and listened through an amplifier. The magnetization grew in small jumps, heard as a crackle: walls between domains breaking free of the defects that hold them.
H. Barkhausen, Phys. Z. 20, 401 (1919). A modern account: J. P. Sethna, K. A. Dahmen and C. R. Myers, Nature 410, 242 (2001).
Francis Bitter spread a fine magnetic powder on polished iron. It gathered along lines where the magnetization changes direction and made the pattern visible under a microscope.
F. Bitter, Phys. Rev. 38, 1903 (1931).
The reasoning, in the open
Inside a domain the alignment m (1 = complete) follows the mean-field equation for spin one-half, with Tc = 1043 K (770 °C) for iron:
m = tanh(m Tc / T), m = 0 above Tc
The plate is a lattice of 64 by 40 cells, each with a direction s = +1 (right) or −1 (left). Its energy is
E = m² [ −Σ sisj + (N/2n)(Σ si)² ] − m Σ (H + hi) si
The first term is the cost of a wall between neighbours that disagree. The second is the stray-field cost of a net magnetization, which is why an untouched plate prefers to cancel. H is the applied field and hi a fixed random field that stands for defects. Cells flip one at a time with the usual Metropolis rule. The net magnetization shown is m times the average of s.
A model of the idea, not of an iron sample. Each cell stands for a block of many atoms. Real domains are micrometres across, and their shapes depend on the crystal and on the stray field in detail. The alignment inside a domain uses the mean-field formula, which is close to measured iron but not the same. Field strength and time are in model units.
Above the Curie point the model shows no magnetization at all. Real iron is paramagnetic there: a field still produces a weak magnetization along it, stronger the closer the temperature is to the Curie point. That response is left out.
Use Tab to reach each control, arrow keys for sliders, and Enter or Space for buttons. Pause freezes the illustration only. Motion starts paused when your system asks for reduced motion. Results are written as text. The page runs entirely in your browser with no accounts, trackers or external code. See the privacy notice.