How does capillarity interact with freezing and phase change in confined geometries?

Why does water in a tiny pore freeze differently from water in a bucket?

Two things change at once when water sits in a narrow space. First, the walls curve the water's surface, and that curvature lowers how readily the water can rearrange into ice; a model of freezing in saturated sandstone describes pore geometry as inducing curvature in confined water and reducing its activity [5]. Second, the walls themselves interact with the water molecules, and in very heterogeneous hosts such as metal-organic frameworks these host-guest interactions produce atypical phase behavior, including capillary condensation [2].

The consequence is that confinement does not simply delay freezing by a fixed amount. A statistical model of fluids in mesoporous materials found that fluids in larger pores undergo a discontinuous, first-order transition, while those in smaller pores undergo a continuous, higher-order one [2]. The same model found the energy barrier between low-density and high-density adsorbed states is lower in confined fluids than in bulk, which pushes condensation to a lower pressure than the bulk saturation pressure [2]. In plain terms: the tighter and more interactive the surroundings, the more the phase change is smeared out and shifted rather than happening all at once at a sharp temperature.

Does confinement make freezing expansion better or worse?

For anyone storing cold energy or protecting a structure, the worry is expansion. When water freezes it grows, and in a sealed or rigid space that growth can crack containers and cause leakage. Soaking water-based phase-change materials into a cellulose sponge through capillary action, then sealing each piece in a plastic film bag, sharply reduced the measured expansion: the composite materials showed maximum expansion coefficients of about 1.99% for pure water and 2.4% for a salt solution, compared with about 9% for pure water alone [1].

That is a large practical difference, and it suggests the sponge's pore network gives the ice somewhere to go and spreads the strain rather than concentrating it. The same study found the composite's effective thermal conductivity stayed above 0.2 W/(m·K) once liquid saturation exceeded about 52.87%, with sponge porosity and thermal conductivity being the dominant factors at a given saturation [1]. So the capillary shaping does not just contain expansion; it also sets how well the material moves heat, which matters for any storage application.

Capillary forces can also block and reroute flow

Capillarity is not only a passive container. In porous media it acts as a gate on where fluids can move. Pore-scale simulations of water displacing oil found that as water advances, some water-oil interfaces stall at pores or throats because of capillary force, and the flow paths through those spots disappear [4]. Which surfaces are wet matters: in oil-wet and intermediate-wet media capillary forces always act as resistance and interfaces stop at throats, while in water-wet media the interface can reverse and become blocked at widening areas, typically moving from throats into pores [4].

This is relevant to freezing because a phase change in a confined geometry is also a story about where liquid sits and how it is connected. If capillary forces pin interfaces and cut off pathways, then the local supply of liquid, and the space available for expansion, are unevenly distributed. The flow-path study did not examine freezing, so it cannot say directly how these stalled interfaces affect ice formation; it shows the mechanism by which confinement makes fluid distribution patchy [4].

Where this is being used, and where the picture is still thin

The most concrete application is modular cold storage: capillary adsorption lets water-based phase-change materials be shaped into pieces that can be handled and installed, with the sponge limiting expansion and the encapsulation preventing leakage [1]. A separate line of work uses confinement in electronics rather than thermal storage: a phase-change memory device built around a self-confined nano-filament reached a reset current of roughly 10 microamps, one to two orders of magnitude below highly scaled conventional phase-change memories, while keeping a large on/off ratio, fast speed, and multilevel memory behavior [3]. That device work addresses phase change in a confined geometry but does not involve capillarity, so it is a parallel case rather than the same mechanism [3].

What remains genuinely open is the quantitative link between pore size, surface chemistry, and freezing behavior in real materials. The sandstone model treats capillarity and adsorption together but is a theoretical model of one rock type [5]; the mesoporous model is statistical and aimed at framework materials [2]; the sponge result is an experimental measurement of expansion and conductivity in a specific composite [1]. These are consistent with a single picture, in which curvature and wall interactions shift the phase transition and redistribute where the material goes, but they do not yet give a general rule for predicting the expansion of an arbitrary confined water system.

Sources used in this answer

Related research questions