Why Does Water Expand When It Freezes, and What Does That Do to Rock?

Some years ago I put a full glass jar of chicken stock in the freezer and shut the door, feeling rather organised. A neighbour had once told me to leave a gap at the top. I had nodded politely and forgotten all about it. By morning the jar had a clean crack down one side, and the dog was sitting beside the freezer with great interest. I spent the next hour with a tea towel and a rueful sense that I had just been taught something about geology.

Here is the short answer. Water expands when it freezes because its molecules lock into an open, spacious pattern as ice. That pattern holds them slightly further apart, on average, than they sit in liquid water. A litre of water becomes roughly 1.09 litres of ice. If that water is trapped in a crack in rock, the growing ice pushes on the walls. Repeat this night after night and the crack widens, bit by bit, until pieces of rock come away.

The rest of this article is about the "why" behind each part of that answer.

Why does ice take up more room than liquid water?

Start with a single water molecule: one oxygen atom with two hydrogen atoms attached, bent into a shape a bit like a boomerang. The oxygen end carries a slight negative charge and the hydrogen ends a slight positive charge. Opposite charges attract, so the hydrogen of one molecule is drawn towards the oxygen of its neighbour. This attraction is called a hydrogen bond. It is much weaker than the bonds holding each molecule together, but there are a great many of them, and they matter enormously.

Each water molecule can form up to four hydrogen bonds: two through its hydrogens and two through the oxygen end. When water freezes, every molecule settles into exactly that four-neighbour arrangement. The result is a rigid lattice of hexagonal rings, a bit like a very tidy honeycomb viewed in three dimensions. The geometry is fixed by the angles at which the bonds point, and those angles leave a lot of empty space in the middle of each ring.

In liquid water the molecules are still hydrogen-bonding, but the bonds are constantly breaking and re-forming, many times a second. The tidy rings are distorted and broken, and molecules can slip into some of those gaps. On average they sit closer together than in the lattice. So melting ice is a kind of collapse: the open framework gives way, and the same molecules pack into less room.

That is why ice is about 9 per cent more voluminous than the water it came from, and why it is less dense. Ice has a density of roughly 917 kilograms per cubic metre, against about 1000 for liquid water near freezing. Work it through: a thousand kilograms of water occupies one cubic metre, while a thousand kilograms of ice occupies about 1.09 cubic metres. If you have ever noticed that an ice cube floats with only a small part above the surface, that is the same number in disguise. About one-twelfth of it sits above the water, and the rest is submerged.

There is a small footnote that people often get wrong. Water does not simply keep expanding as it cools. As you cool liquid water from room temperature, it contracts, like most liquids, until it reaches about 4°C. That is where it is densest. Below that, the structure starts organising itself towards the open lattice, and the water expands slightly as it approaches 0°C. Then, at the freezing point itself, there is a sudden jump of about 9 per cent as the lattice locks in.

How unusual is this, really?

Quite unusual, among everyday substances. Most liquids contract when they freeze, because molecules that stop tumbling around can generally pack more tightly. A handful of other materials behave like water, including silicon, germanium, gallium and bismuth. You are unlikely to find any of them in your kitchen, which is partly why water's habit seems so ordinary and is easy to take for granted.

It has large consequences. Because ice is lighter than water, it floats, and a pond freezes from the top down rather than from the bottom up. The densest water, at about 4°C, sinks to the bottom, and the ice on top acts as a lid. In a cold winter the fish and the frogs underneath still have liquid water to live in. If ice sank, ponds in cold climates would tend to fill from the bottom with ice, and life there would be a very different business.

People noticed that ice floats long before they understood why. I like the story of Galileo and the floating ice. In the early 1600s he took part in an argument with followers of Aristotle, who held that ice floats because of its flat shape. Galileo's reply, as the story is usually told, was that ice is simply water that has become lighter by rarefying. He pointed out that you can push a thin slab of ice down and it will still bob back up, and that shape alone could not explain it. I am hazy on the exact details of the dispute, but the heart of it is well known, and I find it rather lovely. A question that sounds like a child's puzzle about ice cubes was once a serious dispute about how to reason about nature.

How does freezing water break rock?

Rock looks solid, but it is rarely whole. Most rock in the ground and on mountain faces is threaded with cracks, joints, thin planes between layers, and tiny pores between grains. Some of these were formed when the rock cooled, some when it was squeezed or stretched by movements of the earth, and some by the removal of the weight of rock that once lay above it.

Rain, snowmelt and groundwater seep into these gaps. Then the temperature drops below 0°C, and the water in the crack begins to freeze, usually from the outside inwards. As the ice forms, it needs more space than the water did. The walls of the crack have to give a little, or the water has to be pushed out somewhere. Rock is very strong when squeezed but much weaker when pulled apart, and a crack widening is a pulling-apart job. So even a tiny extra push at the tip of the crack can lengthen it, or widen it a hair.

The widening itself is minute. A hair's breadth, however, accumulates. When the ice thaws, the crack stays a little wider, and the meltwater, or the next rain, can run a little deeper in. Then it freezes again. Geologists call this process frost wedging, or freeze-thaw weathering, and the repeated cycle is what does the work. Over many winters, or in some places many hundreds of nights, a block that was once part of the cliff is wholly detached.

This is a good moment to connect to the story of how people worked out that rocks have a long past. James Hutton and the thinkers who followed him argued that the land is slowly worn down and rebuilt, and that processes small enough to watch, if you were patient, could account for great changes over long ages. Frost is a fine example of such a process. No single frosty night does anything you could photograph. Yet the sum of many nights can bring down a cliff face, and the gravel and sand it makes are the raw material of future rocks.

Is it really just the 9 per cent expansion?

Here is a point I had wrong for a long time. I used to picture frost wedging exactly as I have just described it, as if the 9 per cent expansion did all the pushing. That is a reasonable first picture, and it is not wholly wrong, but it is incomplete.

First, expansion only builds up real pressure if the water is trapped. If a crack is open and the water can squeeze out along it, the extra volume has somewhere to go, and very little force results. For ice to push hard, the crack has to be nearly full of water, and the water has to be sealed in by ice forming at the opening, or by the shape of the crack.

Second, there is a thought experiment worth doing. Imagine a perfectly rigid, perfectly sealed space full of water, cooled a little below 0°C. Thermodynamics tells us that the water cannot freeze as easily when it is squeezed, because the pressure itself lowers the freezing point. The pressure that builds up is, roughly speaking, in the region of 13 megapascals for every degree below zero, which is more than a hundred times atmospheric pressure. That is an upper limit for an ideal sealed box, and real cracks are not like that. But it shows that, in principle, freezing water can push with more force than many rocks can resist when pulled. The sums are not the problem; the leaks are.

Third, and this was a surprise to me, scientists found that frost damage does not depend only on expansion. Careful experiments in the early twentieth century showed that soil can heave upward on freezing even when the liquid in it is one that contracts on freezing. So something besides the 9 per cent was going on. The explanation, which is now well established for soils, is that ice crystals growing in a pore or crack can draw more liquid water towards them from the surrounding material, growing into thicker lenses and sheets of ice. These push the ground or rock apart as they grow. This is sometimes called ice segregation. In rock, scientists are still working out how much of the cracking on real cliffs comes from simple expansion and how much from this slow suction-and-growth process, and the answer probably differs from place to place and rock to rock.

I find this a good reminder that a tidy explanation can be a useful first step and still not the whole story. The 9 per cent figure is real and does matter. It is just not the only thing at work.

What does this mean for cliffs, roads and scree slopes?

Cliffs. On steep rock faces, frost works on the joints and layers already there. Over time, blocks loosen and fall. The fresh, pale, angular face where a block has just come away looks quite different from the weathered, darker surface around it. Rocks that are more porous or more heavily jointed tend to suffer more, because they offer water more places to sit. Rocks with few cracks, and little water in them, do better.

Mountain scree. Walk beneath an alpine cliff in Aotearoa, or in nearly any cold mountain range, and you will often find a long apron of broken, angular stones sloping away from the base. That is scree, sometimes called talus. Much of it is frost-shattered rock that has fallen and tumbled down. The stones are sharp-edged because they have been broken off rather than rolled along a river and rounded. Scree typically settles at an angle in the region of thirty-odd degrees, steep enough to feel precarious underfoot and not much steeper, because steeper piles slide down until they relax. If you have ever shuffled across a scree slope, you have been standing on the result of thousands of frosts.

Frost needs the temperature to cross freezing, and the most active places are those where it does so often, and where there is water about. In many parts of our country's high country, that means frequent cycles on clear nights. Lowland places that seldom see a frost are much less affected.

Roads. The same story plays out on a smaller scale in tar seal. Water gets in through small cracks in the surface, and it also lingers in the layers beneath. It freezes, expands, and lifts or loosens the material. Traffic then breaks up the weakened patch, and a pothole is born. Good drainage helps, because a road with no standing water in it has little for the frost to work on. This is one reason road crews spend so much effort on sealing cracks and keeping ditches clear. It is a very practical application of the physics above, and also a good example of why the best remedy often is to keep the water out in the first place.

Small things to look for, and a tip for your freezer

You don't need a mountain to see this at work. A few places to look:

• Terracotta pots and garden ornaments. If a pot is left out and soaked in rain, it may flake or split after a hard frost. Porous materials hold water, and water in the pores freezes.
• Needle ice on bare soil. On a frosty morning you may find slender columns of ice pushing up soil crumbs. That is ice growing by drawing in water, the same process mentioned earlier, and in a form you can see on a lawn or garden bed.
• Kerbs, paths and old concrete. Look for cracks that are wider at the top and for flaking surfaces. These often show where water has been working.
• Fresh rock on roadside cuttings. Pale, sharp-edged fragments at the base usually point to recent frost shattering.

And a practical tip, from someone who learned the hard way: if you freeze liquids at home, leave headroom. Because water gains about a tenth of its volume, fill containers to around nine-tenths or less, and let any liquid cool before sealing. Plastic containers with a bit of flex cope better than rigid glass jars, though this is only what has worked in my kitchen, and I have no tests to offer beyond a sad jar and a hopeful dog. Pipes are a bigger version of the same problem. Lagging exposed ones in cold places gives them a better chance. If you are unsure about your own plumbing, a plumber will know far more than I do.

A question I'm still turning over

The thing I keep coming back to is how much of the landscape's slow change hangs on one oddity of one molecule. If water behaved like most liquids, ponds would freeze from the bottom, scree slopes might look different, and the mountains might wear down in other ways. A small quirk in how hydrogen bonds arrange themselves ends up shaping the ground under our boots.

I am also still learning about the parts that scientists haven't settled. How much of the cracking in a given cliff is simple expansion, and how much is ice slowly growing by drawing in water? How does that balance change with rock type, temperature and how quickly the cold arrives? Those are live questions, and I find it rather comforting that something so familiar as a frosty morning still has corners nobody has fully mapped.

Next time you are out on a clear, cold evening, perhaps listen for a quiet moment. Somewhere, in a crack you can't see, a little water may be turning to ice and nudging the world a hair's breadth along. Thank you for reading, and if you have noticed frost doing something curious near you, I'd be glad to hear about it.