Nucleation and Growth Kinetics in Phase Transformations

Core Idea

Phase transformations occur through nucleation of stable new-phase embryos and subsequent growth into the parent phase. Classical nucleation theory predicts the critical nucleus size and activation barrier based on interfacial energy and thermodynamic driving force. Growth rate depends on atomic diffusion and interface kinetics.

Explainer

Binary phase diagrams tell you *whether* a transformation should happen — they show where two phases coexist at equilibrium. But phase diagrams say nothing about *when* the transformation starts, how fast it proceeds, or what microstructure results. A steel cooled below the eutectoid temperature *should* form pearlite according to the phase diagram, yet with fast enough cooling it forms martensite instead. Nucleation and growth kinetics bridge the gap between thermodynamic possibility and physical reality.

Why doesn't a liquid instantly crystallize the moment you cool it below its melting point? Because creating a new solid phase requires assembling a tiny embryo of the new phase inside the parent. That embryo has a surface — an interface with the surrounding liquid — and that interface has an energy cost proportional to its area. Simultaneously, forming the new phase releases bulk free energy proportional to the embryo's volume. For small embryos, the surface energy term (scaling as r²) dominates the volume energy term (scaling as r³), so the total free energy initially increases as the embryo grows. Only beyond the critical nucleus radius r* does the volume energy term win and further growth becomes thermodynamically downhill. Below r*, embryos spontaneously dissolve; above r*, they grow irreversibly into the new phase.

More undercooling below the equilibrium transformation temperature increases the driving force (the free energy difference between parent and product phases), which lowers r* and reduces the activation energy barrier ΔG*. A larger driving force means more embryos per unit time randomly fluctuate past the critical size — the nucleation rate rises sharply. But there is a competing constraint: atomic diffusion, which is required to rearrange atoms into the new phase structure, slows exponentially at lower temperatures. The interplay between a rising nucleation rate (favored by more undercooling) and a falling growth rate (limited by slower diffusion) creates the characteristic C-curve of time-temperature-transformation (TTT) diagrams: transformation is slowest near the equilibrium temperature (little driving force) and at very low temperatures (slow diffusion), with a nose of fastest transformation at intermediate undercooling.

Heterogeneous nucleation exploits pre-existing defects to bypass the surface energy barrier. At a grain boundary, the interface between two grains is already a high-energy region — forming the new phase there partially replaces that existing interface energy, effectively reducing the activation barrier. Dislocations, free surfaces, and inclusions play the same role. This is why phase transformations in real materials almost always initiate at grain boundaries and free surfaces rather than within grain interiors, and why finer-grained materials (more boundary area per unit volume) transform more readily and at higher temperatures.

Once nuclei exceed the critical size, growth is controlled by either diffusion or interface kinetics depending on the transformation. Diffusional transformations like pearlite formation require long-range atomic rearrangement: carbon must diffuse ahead of the growing pearlite front to partition between ferrite and cementite lamellae. Growth rate depends on how fast this diffusion can proceed, and the lamellar spacing gets finer at larger undercooling (faster growth, less time for diffusion). Displacive transformations like martensite formation involve coordinated shear of the lattice without diffusion — they proceed near the speed of sound and are essentially athermal, controlled by the temperature reached rather than time. The practical consequence is that you can suppress diffusional transformations by cooling fast enough (quenching), but you cannot suppress martensite once you've cooled into its formation range.