Cloudy ice cubes

Why does a home ice cube turn white in its middle when the water it froze from was perfectly clear?

Fill a tray from the tap. The water is clear — hold it to the window and there is nothing in it to see. Six hours later you have cubes with a white core, sharp-edged and opaque, sitting inside a rim of glass-clear ice.

The standard explanation is impurities: minerals in the tap water, left behind as the ice forms. It sounds right, and there is a clean way to test it. Freeze distilled water in the same tray. It goes cloudy too. Then freeze tap water — minerals and all — in a small insulated box with only its top face exposed to the cold. It comes out clear, all the way through.

So the cloudiness is not tracking what is in the water. It is tracking something about how the water froze.

Ask first what the white is. It cannot be a white substance, because nothing white went in. Anything that scatters light in every direction at every interface looks white — crushed clear glass, sea foam, snow. So the core is not stained; it is full of interfaces. What kind? Once you look at a cloudy cube in the light, you can see them: thousands of tiny bubbles.

Bubbles of what? Air. Water in contact with the atmosphere dissolves it. At around 0 degrees Celsius, saturated water holds roughly 15 milligrams of oxygen per litre and rather more nitrogen — those are recalled figures, so take them as an order of magnitude. Call the total something like 38 milligrams of air per litre. Air at ordinary pressure has a density near 1.3 grams per litre, so those 38 milligrams occupy about 30 millilitres of gas: about three per cent of the water's volume, hidden in it, invisible.

Now the crucial fact. Ice is an ordered lattice, and a nitrogen molecule has no place in it. Freezing water does not carry its dissolved air along; it refuses it. This is the same move that sea ice makes with salt, worked through in Sea-ice brine rejection — a growing crystal front takes the water and pushes the dissolved passenger into the liquid ahead of it. The difference here is what the passenger does next. Salt stays dissolved and drains away as heavy brine. Air cannot: the water was already close to saturated, so the moment it is concentrated even slightly it is supersaturated, and it comes out as gas.

That gives us a race, and the race decides everything. A bubble forming at the advancing front can drift away into the remaining liquid, or the front can catch it and freeze around it. Freeze slowly and the bubbles escape; freeze fast and they are engulfed where they stand.

But the deeper question is where the rejected air can go, and that is set by the shape of the front. A domestic tray sits in cold air with cold plastic underneath. Freezing begins on every exposed face at once and closes inward, so the last liquid is a shrinking pocket in the middle — and every face of that pocket has spent hours pushing air into it. The air is not so much trapped as herded. There is no exit, because the exits froze shut first.

Change the geometry and the whole thing changes. Insulate a container on all sides but the top, and the front travels one way only, downward, with open liquid above it the entire time. Now the rejected air is pushed ahead of the front into water that still has a free surface, and simply leaves. The bottom of the block, frozen last, may finally go cloudy — which is why people saw off the cloudy end — but everything above it is clear. The minerals, note, are still in there.

Two corrections to the folk account, and one to my own.

"It's the minerals" is not simply wrong, it is testing the wrong variable. Dissolved calcium and magnesium are rejected by the front too, and in very hard water they can leave a genuine white deposit or a slight haze in the last-frozen region. But they are a garnish. The main event is gas, and the decisive proof is that the same minerals go through directional freezing and produce clear ice.

"Boil the water first" is real but oversold. Boiling drives dissolved air out, so freshly boiled and covered water starts with less to reject and gives a clearer cube. Left open while it cools, it re-absorbs air steadily — cold water dissolves more gas, not less — and much of the benefit goes back. Compared with changing the direction of freezing, degassing is the weaker lever.

And a caveat on my own account. The bubble story explains the milky texture, but not every white feature in a home cube. Rapid freezing also sets up stresses as the ice expands against ice already frozen, and the cracks that result scatter light as well. If your cubes show white planes or radiating fractures rather than a soft cloud, that is a different defect with the same colour.

• Why clear ice melts more slowly in a drink than cloudy ice of the same mass.
• How the same rejection-at-a-front process is used deliberately to purify metals and semiconductors.

Every term the collection defines is gathered in the glossary.