Why Do Some Substances Expand When Freezing?

Most liquids shrink as they freeze, but a handful of substances do the opposite, expanding as they solidify. Water is the most familiar example, growing by roughly nine percent in volume when it turns to ice. The reason traces back to how molecules arrange themselves in the solid state: in certain substances, the crystal structure that forms during freezing is more open and spacious than the disordered jumble of the liquid. That structural openness, driven by the geometry of molecular bonds, is the core explanation for why your water pipes can burst in winter and why ice floats on a lake.

How Water’s Crystal Structure Forces Expansion

In liquid water, molecules are in constant tumbling motion, loosely attracted to their neighbors through hydrogen bonds that form and break billions of times per second. Each water molecule can form up to four hydrogen bonds with surrounding molecules, but in the liquid state, those bonds are fleeting and point in all directions. The result is a relatively compact arrangement where molecules can pack fairly tightly.

When water freezes at ordinary pressure, those same hydrogen bonds lock into a rigid hexagonal lattice known as Ice Ih (the “h” stands for hexagonal, and this is the ice you see in everyday life). In this structure, every molecule sits at a fixed position and maintains four hydrogen bonds simultaneously, each angled in a way that creates wide, six-sided rings. The crystal structure is fairly loose compared to the liquid: the bonds between neighboring molecules (O:H bonds) become longer and weaker, while the bonds within each molecule (H–O bonds) become shorter and stiffer.¹ That six-sided ring geometry is also why snowflakes always have six-fold symmetry.

The key point is that these hexagonal rings contain a lot of empty space at their centers. In the liquid, molecules can crowd into those gaps because the hydrogen bonds are flexible and transient. Once the lattice locks in, the gaps become permanent. The solid ends up occupying more volume than the same mass of liquid, making ice less dense than water. That density difference is about nine percent, which translates to real mechanical force when water freezes in a confined space.

The Density Peak Near 4°C

Water’s expansion on freezing is actually the dramatic endpoint of a trend that starts well before the freezing point. If you cool liquid water from, say, room temperature, it behaves like most liquids at first: it contracts and gets denser. But this only continues down to about 4°C (roughly 39°F). Below that temperature, the water begins expanding again even though it is still liquid.

This happens because as the temperature drops toward freezing, more and more clusters of molecules begin adopting that hexagonal, ice-like geometry. These transient clusters are bulkier than the surrounding disordered liquid. By around 4°C, the expansion caused by these growing ice-like clusters overtakes the normal thermal contraction, and the liquid starts becoming less dense. This is why the bottom of a deep lake in winter sits at about 4°C: that water is the densest, so it sinks, while slightly colder (and less dense) water floats above it. The phenomenon was documented experimentally as far back as the early nineteenth century, when researchers observed that water contracts with heating at ice-cold temperatures until about 40–41°F, barely changes for the next degree or two, and then begins expanding with further warming.²

That density maximum has enormous ecological importance. Because the densest water sinks to the bottom of a lake well above freezing, ice forms at the surface rather than from the bottom up. Aquatic life survives winter beneath an insulating lid of ice rather than being frozen solid from the lakebed upward.

Other Substances That Expand When They Freeze

Water gets all the attention, but it is not alone. Several elements and compounds also expand upon solidification, and the reasons vary.

Bismuth is the most commonly cited metallic example. Like water, bismuth’s solid crystal structure is more open than its liquid phase. Bismuth crystallizes in a layered, rhombohedral arrangement with significant gaps between layers, making the solid less dense than the melt. Early crystallographic work on bismuth focused on understanding the unusual directional dependence of its thermal expansion, since it does not crystallize in a regular cubic pattern like most metals.³ This expansion behavior has practical value: bismuth-based alloys are used in precision casting and micro-fabrication because the expansion during solidification helps the metal fill every corner of a mold before it hardens.⁴

Silicon, germanium, and gallium also expand on freezing. All three share a common structural trait: in their solid crystalline form, the atoms bond in a tetrahedral, diamond-like geometry that is geometrically open, much as water’s hexagonal lattice is. In the liquid state, the atoms pack more efficiently because they are no longer constrained to those rigid bond angles. The result is the same as with water: the ordered solid takes up more room than the disordered liquid. This is why silicon wafers used in electronics are fabricated with great care during the cooling step: uncontrolled expansion can introduce defects.

Plutonium offers a more exotic example. This element has six distinct solid phases at atmospheric pressure, and the volume changes between them are striking. Its largest-volume solid phase (the δ phase) is roughly 25 percent more voluminous than its lowest-temperature α phase. That dramatic volume collapse is driven by competing electronic instabilities in plutonium’s outer electron shell, which push the metal’s volume in opposing directions depending on temperature.⁵ Plutonium’s behavior is less about simple crystal geometry and more about the unusual way its electrons interact, making it one of the most structurally complex elements known.

Why Most Substances Shrink Instead

To understand why expansion on freezing is unusual, it helps to know what happens in the vast majority of materials. Most liquids are disordered, with molecules or atoms bouncing around and occupying more space than they need. When such a liquid freezes, the molecules lock into a tightly packed crystal lattice. Think of the difference between a pile of tennis balls thrown randomly into a box versus the same balls stacked in neat rows: the stacked version takes up less room. For most metals, salts, and organic compounds, the solid crystal is denser than the liquid, and the substance contracts as it freezes.

Expansion on freezing happens only when the geometry of the bonds in the solid forces molecules apart more than the randomness of the liquid pushes them together. That requires a crystal structure with built-in empty space, which in turn requires directional bonding. Water’s hydrogen bonds and silicon’s covalent bonds both have strong directional preferences that create open frameworks. Metals bonded without a strong directional preference, like iron or copper, pack more tightly in the solid and contract normally.

Frost Wedging and the Splitting of Rock

The roughly nine percent volume expansion of water upon freezing generates enormous pressure when the water is confined. One of the most visible consequences is frost wedging, the process by which water seeping into rock cracks and then freezing can split stone apart. The pressure from the water-to-ice transition inside a fissure acts like a hydraulic wedge, prying the crack open a little wider with each freeze-thaw cycle.⁶

Over geological time, frost wedging is one of the dominant forces shaping landscapes in cold regions. Mountain peaks are shattered into fields of angular boulders, cliff faces shed slabs, and road surfaces develop potholes. Experimental studies on fractured granite have directly measured crack opening under controlled freeze-thaw conditions, confirming that even low-porosity rock can be progressively broken apart when water fills narrow fractures.⁷ The same mechanism is responsible for damage to concrete infrastructure, gravestones, and building facades anywhere winter temperatures cycle above and below freezing.

Cell Damage and the Challenge of Cryobiology

The expansion of water upon freezing is a serious problem for living cells. When ice crystals form inside a cell, they do not just take up extra space; they physically puncture and tear membranes. Intracellular ice formation is one of the primary causes of cell death during freezing. In studies of human dental stem cells, membrane damage after complete intracellular ice formation reached about 86 percent in isolated cell suspensions.⁸ Cells that were in contact with neighbors in a monolayer fared somewhat better, with damage around 26 percent, likely because cell-to-cell connections provide some structural reinforcement.

This is the central challenge of cryopreservation, whether you are freezing embryos for fertility treatment, blood products, or organ tissue for transplant. The goal is to get cells cold enough to stop all biological activity without allowing damaging ice crystals to form inside them. One approach is vitrification, in which the liquid is cooled so rapidly, often in the presence of cryoprotectant chemicals, that it solidifies into a glass-like state rather than forming crystals. The result is a solid with the disordered molecular arrangement of a liquid, achieved when viscosity climbs high enough that molecules cannot rearrange into a crystal lattice.⁹ Vitrification sidesteps the expansion problem entirely because no ice forms at all.

Ice on Other Worlds

The expansion of water ice has consequences well beyond Earth. Jupiter’s moon Europa and Saturn’s moon Enceladus both have thick ice shells overlying liquid water oceans. As those ice shells gradually thicken over time, the volume expansion of freezing water builds pressure in the ocean below. On Europa, the freezing of just one to ten kilometers of additional ice generates enough pressure to produce tangential stresses that exceed the tensile strength of the ice shell itself.¹⁰ That mechanism is one plausible explanation for how liquid water could erupt through the surface of these moons, forming the plumes and fractured terrain that spacecraft have observed.

The same physics also means that bodies of ice in the outer solar system are not static. Any icy moon with an internal ocean that is slowly freezing is under a kind of slow-motion hydraulic stress. The ice shell cracks, water fills the cracks, it freezes and expands further, and the process continues. In some respects, the same frost-wedging mechanism that crumbles mountain cliffs on Earth operates on a planetary scale in the outer solar system.

Ice Under Extreme Pressure

The familiar hexagonal ice that forms in your freezer is just one of at least twenty known solid phases of water. Under high pressure, water molecules can be forced into denser and denser crystal arrangements, some of which are actually denser than liquid water. These high-pressure ice phases are not just laboratory curiosities; they are thought to exist in the deep interiors of ice-giant planets like Uranus and Neptune.

At extremely high pressures, above roughly 300 billion pascals, the hydrogen atoms in ice are no longer bound to a specific oxygen atom; they sit midway between two oxygens in a structure called ice X. At even higher pressures, a structural transition to yet another phase (called Pbcm) occurs, involving a sharp structural change that, strangely, has no apparent energy barrier at the transition point. At higher pressures still, the barrier between the two phases gradually increases.¹¹ These exotic ices bear little resemblance to the airy hexagonal structure that causes everyday expansion. They are crushed into tighter arrangements by sheer force, and many of them are denser than liquid water.

This means that the familiar expansion-upon-freezing behavior is actually specific to ordinary conditions. Under enough pressure, water freezes into a form that is denser than the liquid, just like most other substances. The anomalous expansion is a feature of Ice Ih and its open hexagonal rings, not a universal property of water at all temperatures and pressures.

Water Confined in Tiny Spaces

The behavior of water changes in surprising ways when it is confined to nanometer-scale spaces. Inside carbon nanotubes, for instance, water molecules cannot form the same hexagonal rings they would in an open container. Instead, they arrange themselves into entirely different ice structures, including square and pentagonal tubes of ice, depending on the diameter of the nanotube. Simulations suggest there are at least nine distinct ice phases that can form inside nanotubes, each one maximizing the number of hydrogen bonds possible under the cylindrical constraint. Some of these confined ices have melting points well above 0°C.¹²

This matters beyond pure curiosity. Biological systems are full of nanoscale channels and pores where water behaves differently than it does in bulk. Understanding how water freezes in confined geometries is relevant to everything from cryopreservation of tissues (where ice can form inside cellular compartments only a few nanometers wide) to the design of anti-icing coatings and water-purification membranes. The expansion-on-freezing rule of thumb applies to bulk water, but at the nanoscale, the rules can change dramatically.

Supercooled Water and the Two-Liquid Hypothesis

Water can be cooled well below 0°C without freezing if it is pure enough and undisturbed. This supercooled liquid water becomes increasingly strange as the temperature drops: its compressibility, heat capacity, and other properties start to diverge from what simple models predict. For decades, researchers have debated whether this anomalous behavior points toward a hidden critical point deep in the supercooled region, a temperature and pressure at which two distinct forms of liquid water (one denser, one less dense) would become indistinguishable.

Recent experimental work has provided the most direct evidence yet for this idea. Researchers observed a rapid increase in heat capacity and enhanced density fluctuations consistent with a critical point at around 210 K (about −63°C), along with a crossover from a discontinuous to a continuous structural transition and the kind of slow structural fluctuations expected near a critical point.¹³ Earlier theoretical modeling had predicted this critical point would fall near 223 K and around 50 megapascals of pressure, and experimental measurements of supercooled water’s volume under pressure have been broadly compatible with that prediction.¹⁴

If confirmed, the existence of two distinct liquid forms of water would provide a deeper explanation for why water expands when it freezes. The low-density liquid form, which predominates at low temperatures, already has a structure resembling ice’s open framework. Freezing, in a sense, just locks in the arrangement that the liquid was already trending toward. The expansion upon freezing would then be the final step in a gradual structural shift that begins in the liquid state, the same shift responsible for the density maximum near 4°C and all of water’s other anomalous thermal properties. The research is ongoing, but it hints that the expansion of ice may be just one visible symptom of a much more fundamental strangeness in the way water molecules interact.