The Physics of Supercooling

Supercooling happens when water drops below its freezing point yet stays liquid—sometimes even resisting freezing after a smack. The key insight is a size-dependent tug-of-war at the ice–water boundary: the ice interior favors freezing, while the ice surface favors staying liquid. Because surface area scales with the square of a characteristic radius while volume scales with the cube, tiny ice embryos have proportionally more surface than interior. For those small crystals, the surface effect dominates, so any nascent ice melts back into liquid. Only when an ice crystal exceeds a “tipping point” size does the interior’s freezing drive win, allowing the crystal to grow and trigger runaway freezing.

For pure water, the tipping point depends on temperature. Around 4°C below the freezing point, the critical ice embryo size is roughly 20 nanometers—about 70 water molecules across. At colder conditions, the tipping point shrinks: near −10°C, a crystal larger than about 10 nanometers (around 30 molecules across) can become stable and keep growing. That explains why supercooled water doesn’t necessarily freeze immediately after crossing the freezing point: it can remain liquid until a sufficiently large ice nucleus forms. The waiting time can be shortened by cooling further (which lowers the critical size) or by providing nucleation sites such as minerals or dust particles that already contain “seed” structures large enough to jump-start ice formation.

Even mechanical agitation—like smacking a bottle—can sometimes supply the energy needed for an ice nucleus to surpass the tipping point and grow. But there’s a more extreme route to preventing freezing: pressure. In a rigid container, freezing water expands, building pressure that makes crystal growth harder. This pressure effect becomes crucial when only a tiny amount of water is present, because a single small ice crystal can raise pressure enough to suppress further growth. The transcript extends this idea to extremely small rigid containers: for container sizes below a few tens of micrometers, pressure’s “melt” influence overwhelms the interior’s “freeze” influence for all possible crystal sizes. In that regime, supercooled water can remain liquid regardless of how hard it’s struck.

The scale of this pressure-based prohibition is striking. For water about 1°C below zero, the rigid container must be smaller than roughly 200 nm in length; for water about 4°C below zero, the container needs to be even smaller, around 50 nm. The logic is consistent with the earlier tipping-point argument: at lower temperatures, the critical ice embryo size would normally be smaller, so the container must be small enough that pressure rises quickly enough to prevent any embryo from growing into a runaway freezing front. Finally, the restriction depends on rigidity: if the container can open and let pressure escape, then smacking can again trigger freezing.