A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
At criticality, the system has no intrinsic length scale: correlations scale as power laws with system size, and the structure is fractal-like. Universality states that exponents depend only on spatial dimension and order-parameter symmetry, not microscopic details. This explains why vastly different systems (binary fluids, ferromagnets, superconductors) have identical exponents.
From critical exponents, you know that near a critical point thermodynamic quantities diverge as power laws: the correlation length ξ ~ |t|^{−ν}, the susceptibility χ ~ |t|^{−γ}, the magnetization m ~ |t|^β, where t = (T − T_c)/T_c. What you may not yet have a geometric picture of is *why* power laws appear — and why exponents from completely different physical systems are identical. Scaling invariance answers the first; universality answers the second.
Scaling invariance at T_c means the system has no characteristic length scale. Ordinarily a ferromagnet has two relevant length scales: the lattice spacing a (microscopic) and the correlation length ξ (mesoscopic, measuring how far spins tend to align). At T_c, ξ → ∞ — spin correlations extend across the entire system. With no finite ξ to compare distances to, the system looks statistically the same at every scale: zoom in by a factor of 2 and the spin configuration is statistically indistinguishable from the original. This self-similarity is the defining property of fractals. Correlation functions that decay exponentially ~e−r/ξ for finite ξ must, when ξ → ∞, decay instead as power laws: ⟨S(0)S(r)⟩ ~ r−(d−2+η), where η is a critical exponent. Power laws are the only functional form that is scale-free — they have no preferred length encoded in an exponent.
Universality says that the critical exponents depend only on (1) spatial dimension d and (2) the symmetry of the order parameter — not on the microscopic Hamiltonian, lattice structure, or interaction details. Water near its liquid-gas critical point and an iron magnet near its Curie temperature both belong to the 3D Ising universality class and share identical exponents (β ≈ 0.326, ν ≈ 0.630, γ ≈ 1.237), even though their microscopic physics is completely different. The deep explanation comes from the renormalization group: when you systematically coarse-grain a system — averaging over short-distance degrees of freedom — the Hamiltonian flows through a space of possible Hamiltonians. Near a critical point, this flow converges to a fixed point. The universal exponents are properties of the fixed point, not of the microscopic starting Hamiltonian. All systems whose coarse-graining flows to the same fixed point share the same exponents — they are in the same universality class.
The universality classes are organized by symmetry content. The Ising class (Z₂ symmetry, scalar order parameter) covers liquid-gas transitions, binary alloy order-disorder transitions, and uniaxial ferromagnets. The Heisenberg class (O(3) symmetry, 3-component vector order parameter) covers isotropic ferromagnets. The XY class (O(2) symmetry) covers superfluid helium-4 and describes the Kosterlitz-Thouless transition in two dimensions. Within each class, critical exponents are not independent — they are related by scaling relations such as the Rushbrooke relation α + 2β + γ = 2 and the Fisher relation γ = ν(2 − η). These relations reduce the number of independent exponents to two, reflecting the fact that the fixed point is characterized by only two relevant scaling fields (temperature and the ordering field). Scaling relations are the mathematical signature of the underlying scale invariance: they follow from demanding that the singular part of the free energy obeys a generalized homogeneity law near T_c.