Magnetic materials and ferromagnetic hysteresis

Every material responds to an applied magnetic field, just not by the same amount or in the same direction. Diamagnetic materials — most of them, including water and wood — develop a weak magnetization opposing the field, so they're very weakly repelled; it's a tiny effect that's easy to miss without sensitive equipment. Paramagnetic materials have atoms with their own small magnetic moments that partially align with an applied field, giving a weak attraction that vanishes the moment the field is removed. Iron, nickel, cobalt and a handful of others are different in kind, not just degree: they're ferromagnetic, meaning neighbouring atomic moments lock into alignment with each other through a quantum-mechanical effect strong enough that whole regions — domains — stay aligned even without any external field at all.

That persistence is what makes a ferromagnet's response depend on its history, not just on the field applied right now. Plot magnetization against applied field as the field is cycled up, down and back up again, and the curve doesn't retrace itself — it traces a loop, called a hysteresis loop, because the magnetization at any given field strength depends on which direction you were sweeping from. The loop's width along the field axis is the coercivity, how hard you have to push in the opposite direction to erase the existing magnetization; a "hard" magnetic material (good for permanent magnets) has a wide loop, and a "soft" one (good for transformer cores, which get cycled every AC period) has a narrow one, since the enclosed area is energy lost to heat on every cycle.

Ferromagnetism itself isn't permanent. The alignment between neighbouring domains is a competition between the interaction trying to align them and thermal agitation trying to randomize them, and above a material-specific temperature — the Curie temperature — thermal agitation wins outright. Heat a piece of iron past roughly 770°C and its ferromagnetic order collapses; it becomes merely paramagnetic, responding weakly to a field like any other paramagnet, until it's cooled back down.