Is Glass a Liquid? What Happens When Something Stops Flowing Without Freezing

A liquid cooled slowly usually crystallises at a definite temperature, with the molecules arranging into a regular lattice, releasing heat and changing properties abruptly. Cooled fast enough, some liquids cannot organise quickly enough and remain disordered while continuing to thicken, with viscosity rising by enormous factors over a modest temperature range until the material no longer flows on any practical timescale. That point is the glass transition. Nothing has crystallised, the arrangement of molecules is essentially that of a liquid frozen in place, and the transition occurs over a range rather than at a sharp point, with the measured temperature depending on how fast the cooling was, which is unlike any ordinary phase change.

The contrasts are specific and each is measurable:

• •Freezing happens at a sharp temperature and this happens over a range
• •Freezing produces an ordered arrangement and this preserves a disordered one
• •Freezing releases a substantial quantity of heat and this does not
• •The transition temperature depends on the cooling rate, which a melting point does not
• •The resulting material continues to change very slowly over long periods, which a crystal does not
• •A glass is not a distinct phase in the strict thermodynamic sense, which is why the subject remains unsettled

The frequently repeated story that medieval window glass is thicker at the bottom because it has flowed over centuries is wrong, and the reason it is wrong is instructive. Estimates of the viscosity of window glass at ordinary temperatures give flow rates so slow that no measurable change would occur over the age of the universe, let alone several centuries. The observed thickness variation comes from the manufacturing method, since glass made by spinning a molten blob into a disc and cutting panes from it is uneven, and glaziers commonly installed the thicker edge downward for stability. Some old panes are thicker at the top, which the flow explanation cannot accommodate. The underlying claim that glass is a very slow liquid is a reasonable way of describing the disordered structure and a bad description of its behaviour.

The transition is frequently described as the deepest unsolved problem in the physics of ordinary matter, and the reason is worth stating. Viscosity rises by more than ten orders of magnitude over a temperature range in which no structural change is detectable by any method, so whatever slows the material down is not visible as a change in arrangement. Competing explanations propose growing regions of cooperatively rearranging molecules, an underlying transition that is never reached because the material falls out of equilibrium first, or purely dynamic explanations requiring no hidden transition at all. Experiments distinguishing them are extremely difficult, since the timescales become impractically long precisely where the answer lies. The problem has been actively worked on for decades and remains open, which is unusual for a phenomenon this common.

The transition governs the behaviour of an enormous range of materials. Every polymer has one, and whether a plastic is rigid or flexible at room temperature depends entirely on whether that temperature sits above or below its transition, which is why the same material can be a brittle sheet or a soft film depending on additives that shift it. Rubber becomes brittle when cooled below its transition, which was the mechanism behind the failure of a seal that destroyed a space shuttle in 1986. Food science uses it to explain the texture of boiled sweets, of frozen desserts and of dried products, and to predict shelf life. Cryopreservation of cells and tissues depends on cooling fast enough to form a glass rather than ice crystals, since crystals destroy the structures being preserved.

Cooling fast enough prevents crystallisation, and the liquid thickens until it stops flowing while keeping a disordered arrangement, over a range rather than at a point and at a temperature depending on the cooling rate. Medieval windows are not thicker at the bottom from flowing, since the calculated rate rules it out and some are thicker at the top.