Phase Transition
Everything changes when the temperature changes. Not gradually. Not always. Sometimes the change is sudden, sharp, dramatic. This is a phase transition, and it is the most interesting thing that happens in macroscopic systems.
A first-order phase transition is discontinuous. The order parameter — magnetization, density difference, crystal structure — jumps. There is latent heat: you must add energy without changing temperature, and that energy goes into breaking bonds, not raising kinetic energy. Water boiling at 100°C is a first-order transition. The density of liquid water is about 1000 kg/m³. The density of steam at the same temperature is about 0.6 kg/m³. The jump is nearly three orders of magnitude. You can see it. You can hear it. The kettle screams when it happens.
Second-order (continuous) phase transitions are more subtle and more beautiful. The order parameter goes to zero continuously as you approach the critical point. Magnetization decreases as you heat an iron magnet and vanishes at the Curie temperature. There is no latent heat. But there is something else: the correlation length diverges. Fluctuations at all scales become important. The system becomes scale-invariant. Near the critical point, a magnet cannot tell whether it is made of atoms or of tiny magnetic domains behaving like magnets themselves. This is why critical phenomena are universal — a topic we will return to.
The Landau theory of phase transitions provides a framework: write down the free energy as a function of the order parameter, constrained by symmetry, and minimize it. When the coefficient of the quadratic term changes sign, the minimum shifts, and the system transitions. This works remarkably well for many systems. It fails spectacularly for others — specifically, those where fluctuations are strong enough to destroy the mean-field description, which happens below the upper critical dimension. This is the boundary between simple physics and interesting physics.
Hysteresis accompanies first-order transitions. If you cool a liquid past its freezing point, it may not freeze immediately. It supercools. The system is trapped in a local minimum of the free energy landscape, separated from the global minimum by a barrier. The barrier scales with the interfacial free energy between phases. To cross it, you need a nucleation event — a fluctuation large enough to create a critical droplet of the new phase. This is why cloud seeding works, why ice forms on supercooled planes, and why your glass of water in the freezer doesn't instantly become a block of ice.
Supercooling, superheating, metastability — these are not exceptions. They are predictions of the thermodynamic framework, and they reveal something deep: the free energy landscape is a topography, and systems get stuck in valleys. The phase diagram is a map of where the deepest valleys are. As you change temperature and pressure, you move the terrain. When two valleys become degenerate, the system can switch. When the old valley disappears entirely, it must.
There are topological phase transitions — Kosterlitz-Thouless transitions in two-dimensional systems, driven by the unbinding of topological defects rather than symmetry breaking. These don't fit the Landau paradigm at all. The order parameter never vanishes discontinuously or continuously in the usual sense. Instead, vortex-antivortex pairs dissociate, and the system goes from a topologically ordered state to a disordered one. The transition is essential singular: the correlation length grows as exp(c/√(T-Tc)), not as a power law. This is a phase transition that lives on a different mathematical planet.
Phase transitions are where simplicity meets complexity. They are described by a few parameters. They produce phenomena at every length scale. They are exactly solvable in some cases (the 2D Ising model, Onsager's triumph) and opaque in most others. They are simple and hard at the same time. They are one of the things that make condensed matter physics beautiful.