The nested toroidal shells of constant magnetic flux that organise a confined plasma—the scaffolding on which pressure, current, and transport are built.
A magnetic field line in a well-confined toroidal plasma does not wander freely through three-dimensional space. Instead, it stays on a two-dimensional toroidal surface, tracing out a helical path that, given enough transits, covers the entire surface ergodically. That surface is defined by a constant value of the poloidal magnetic flux function ψ and is called a flux surface. In an ideal axisymmetric equilibrium, these surfaces nest concentrically: the magnetic axis at the centre, then a continuous family of tori of increasing cross-sectional area, out to the separatrix at the plasma boundary.1
Charged particles in a magnetised plasma travel freely along field lines but cross them only slowly, through collisions or turbulent fluctuations. Because a field line remains on its flux surface, particles are effectively confined to thin shells. This means that pressure, temperature, and density are approximately constant on each flux surface—they are flux functions. The entire equilibrium can therefore be described by one-dimensional profiles (pressure vs. ψ, current density vs. ψ) rather than full three-dimensional fields, a simplification that underpins the Grad–Shafranov equation for axisymmetric equilibria.2
On most flux surfaces the rotational transform (or safety factor q) is irrational: a single field line covers the surface densely without ever closing on itself. On rational surfaces, where q equals a ratio of small integers (e.g., 1, 3/2, 2), field lines close after a finite number of transits. These surfaces are special because they can support magnetic islands—localised regions where the nested topology breaks down—driven by tearing modes or resonant magnetic perturbations. Large islands degrade confinement by short-circuiting the radial insulation between adjacent surfaces.3
In stellarators, where the magnetic geometry is fully three-dimensional, flux surfaces are not guaranteed. Careful coil optimisation is required to minimise regions of stochastic (destroyed) flux surfaces, particularly near the edge. Codes such as VMEC compute three-dimensional equilibria by assuming nested flux surfaces, while field-line tracing codes verify that the surfaces actually exist in the computed field.4
Whether in a tokamak, a stellarator, or a field-reversed configuration, the existence and quality of flux surfaces remain the first question any confinement concept must answer.