Nucleation & Growth — Birth of a New Phase

A tug of war between surface and volume

Cool a liquid below its freezing point and it does not instantly turn solid everywhere at once. Somewhere in the bulk, a handful of atoms have to spontaneously arrange themselves into a small ordered cluster — a nucleus — before crystal growth can even begin. That first cluster faces two competing energy terms. Turning liquid into solid releases energy proportional to the volume of the new cluster, because the solid phase is more stable below the freezing point. But creating the boundary between the new solid and the surrounding liquid costs energy proportional to the cluster's surface area, because that interface is a region of disrupted bonding.

For a very small cluster, surface area dominates — a sphere's surface-to-volume ratio blows up as the radius shrinks — so tiny clusters cost more energy than they save and tend to dissolve back into the liquid. Only once a cluster is large enough for its favourable volume term to outweigh its unfavourable surface term does growing it further actually lower the total energy. That crossover point is the critical radius, and clusters below it are called embryos; clusters that randomly fluctuate past it become viable nuclei that grow without limit.

live demo · random clusters forming, dissolving, and occasionally taking off● LIVE

The classical nucleation theory equations

Classical nucleation theory writes down that competition explicitly for a spherical cluster of radius r, with ΔG_v the (negative) volume free-energy change per unit volume driving solidification and γ the surface energy per unit area of the new interface:

Delta_G(r)  =  (4/3) pi r^3 * Delta_G_v   +   4 pi r^2 * gamma
                \_________________/         \______________/
                  volume term (< 0)          surface term (> 0)

d(Delta_G)/dr = 0   solved for the maximum   -->

  r_critical = -2*gamma / Delta_G_v
  Delta_G_critical = 16 pi gamma^3 / (3 Delta_G_v^2)

Plotting ΔG(r) gives a hump: it rises through the surface-dominated small-r region, peaks at r_critical, then falls without limit as the volume term takes over for large r. ΔG_critical, the height of that hump, is the energy barrier that a random thermal fluctuation must climb before a nucleus can form. It behaves exactly like an activation energy in a chemical reaction, and it sets the nucleation rate through a Boltzmann-type exponential: rate ∝ exp(−ΔG_critical / k_BT). That exponential is brutal — small changes in undercooling, which shrinks r_critical and ΔG_critical together, can swing the nucleation rate by many orders of magnitude.

Homogeneous vs heterogeneous nucleation

Homogeneous nucleation — a cluster forming from nothing in a perfectly uniform bulk liquid — pays the full surface energy cost derived above, and needs deep undercooling to happen at any appreciable rate; pure water can be supercooled tens of degrees below 0°C in the laboratory before it freezes on its own. Heterogeneous nucleation forms instead on an existing surface: a dust particle, a container wall, a scratch, or a deliberately added seed crystal. Part of the new interface is replaced by the pre-existing solid-liquid contact, which lowers the effective surface energy penalty and can cut the nucleation barrier drastically — sometimes to a small fraction of the homogeneous value. This is why real-world freezing, crystallisation and precipitation almost always happen at far less extreme undercooling than the pure homogeneous theory predicts, and why seeding a supersaturated solution with a single crystal can trigger crystallisation almost instantly.

After nucleation: growth takes over

Once a nucleus is past the critical radius, the physics changes character. Growth is now governed not by a fluctuation problem but by how fast atoms can be delivered to and incorporated at the growing interface — diffusion of material through the surrounding liquid, and the kinetics of atoms attaching at the crystal surface. This handoff, from a rare stochastic nucleation event to a comparatively steady, deterministic growth process, is the reason materials solidified quickly tend to end up with many small grains (lots of nucleation events competing for limited undercooling) while materials cooled slowly end up with a few large ones (fewer nuclei, each with plenty of time and material to grow). Controlling that balance — how much undercooling, how fast the heat is removed, whether seed crystals or nucleating agents are added — is the entire basis of practical metallurgy, semiconductor crystal growth and even ice-cream texture.

Frequently asked questions

Why does water sometimes stay liquid below 0°C?

Because forming a tiny ice crystal costs surface energy before the volume energy can pay it back. Without a surface, dust particle, or container scratch to nucleate on (heterogeneous nucleation), pure water can supercool many degrees below freezing while it waits for a large enough random fluctuation to cross the critical radius on its own — homogeneous nucleation, which is statistically rare.

What is the difference between homogeneous and heterogeneous nucleation?

Homogeneous nucleation forms a new-phase cluster from nothing in the bulk of a uniform parent phase, and pays the full surface energy cost — it is rare and needs deep undercooling. Heterogeneous nucleation forms on an existing surface such as a dust particle, container wall, or seed crystal, which replaces some of the new surface with an existing interface and lowers the energy barrier dramatically, which is why it dominates in almost all real situations.

Why does more undercooling make crystals form faster?

Undercooling increases the volume free-energy gain per unit of new solid, which shrinks the critical radius and lowers the nucleation energy barrier, so viable nuclei appear more often per unit time and per unit volume. Push the undercooling far enough and the nucleation rate rises so steeply that a liquid solidifies almost everywhere at once instead of growing from a few isolated seeds.

Try it live

Everything above runs in your browser — open Nucleation & Growth and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.

▶ Open Nucleation & Growth simulation

Nucleation & Growth: How a New Phase Is Born