Freezing-point depression

Why does scattering salt on ice melt it in cold that would otherwise keep it frozen?

A gritting lorry scatters salt on a road at minus five, and the ice goes. It is tempting to assume the salt is doing something warm — reacting, releasing heat, attacking the ice. But dissolving salt in water cools the mixture slightly; the road gets colder, and the ice still melts. So whatever the salt did, it did not add heat. What else could it have changed?

Ask first what "frozen" really describes. Not stillness — at the boundary between ice and liquid water, molecules are constantly leaving the ice, and molecules from the liquid are constantly landing on it and being captured. Zero degrees is not where motion stops. It is where those two rates happen to be equal, so nothing changes overall even though everything is in traffic.

What sets each rate? Leaving depends on how vigorously the ice's molecules are jiggling, which is temperature. Joining depends on how often a water molecule arrives at the surface from the liquid side. Hold that distinction, because it is the whole answer.

Now dissolve salt in the liquid. The salt does not enter the ice crystal — the growing lattice rejects it, which you can taste in the fresh meltwater from sea ice. So the liquid at the boundary is partly ions, and a smaller share of arrivals are water molecules that could be captured. The joining rate falls; the leaving rate has not changed at all. What must happen? Net melting.

Where does it stop? Cool the mixture and the leaving rate falls too, since that is the rate temperature governs. At some lower temperature the two balance again, and that new balance point is the freezing point of the salty water. The salt did not melt the ice — it moved the temperature at which melting and freezing are in balance, leaving the ice above its own melting point.

Here is the turn that surprises people. If what matters is the share of arrivals that are not water, the identity of the intruder should barely matter — only how many separate particles it contributes. And so it is: sugar depresses the freezing point too, and a unit of table salt contributes two particles on dissolving, a sodium and a chloride, so mole for mole it does roughly twice the work. Properties that turn on the number of dissolved particles rather than what they are are called colligative, and they are one of the few places where the chemistry genuinely does not matter.

Three refinements. First, the salt needs a little liquid to dissolve into — a brine film, normally present on ice, or supplied by traffic, sun, or pre-wetted grit. On genuinely dry ice well below freezing, dry salt does very little.

Second, melting absorbs heat, and that heat comes out of the road, the ice, and the air, so salting cools its surroundings while it works. An old ice-cream churn packed with salted ice uses the same effect deliberately.

Third, and this is what limits the technique: the more salt you dissolve the lower the balance point goes, but not without end. A saturated sodium chloride brine freezes at about minus twenty-one degrees Celsius, and below that no amount helps. The practical limit bites long before that — around minus ten the salt dissolves sluggishly and the quantities needed become unreasonable, which is why cold-climate services switch to calcium or magnesium chloride, or to sand, which melts nothing and simply adds grip.

• Why boiling point rises for the same reason freezing point falls, and why osmosis belongs in the same family.
• How antifreeze proteins in polar fish work by a different route — binding to ice crystals rather than diluting the liquid.
• Why sea ice is nearly fresh, and what happens to the salt the growing ice rejects.

Every term the collection defines is gathered in the glossary.