What is glass?

The question

Glass is one of the oldest materials people make and one of the most familiar. Yet in 1995 the Nobel laureate Philip Anderson wrote that the deepest and most interesting unsolved problem in solid-state theory is probably the nature of glass and the glass transition.

The puzzle is this. When most liquids cool, their molecules lock into an ordered crystal at a sharp freezing point. But if a liquid is cooled fast enough, or if its molecules have trouble ordering, it skips crystallisation. It stays liquid below its freezing point, becoming thicker and thicker, until it effectively stops flowing. The result is glass: as rigid as a solid, but with its atoms in the disordered arrangement of a liquid. Is glass a separate state of matter, or just a liquid that has become too slow to move?

What we know

• Glass is not only window glass. Many substances can form glasses: sugar (in hard candy), plastics, metals, even water, if it is cooled extremely fast or deposited as vapour onto a very cold surface. Being a glass is a matter of how a material solidifies, not what it is made of.
• The slowdown is extraordinary. Near the glass transition, the viscosity of a liquid rises by more than ten orders of magnitude over a fairly small range of temperature, while its structure barely changes. Nothing visible in the arrangement of the molecules explains why they suddenly can hardly move.
• The temperature at which a glass forms depends on the cooling rate: cool faster, and the liquid falls out of equilibrium at a higher temperature. That is why the glass transition looks more like a gradual loss of mobility than a sharp phase transition.
• Old church windows that are thicker at the bottom did not flow. Calculations from the viscosity of medieval glass show that noticeable sagging would take far longer than all of human history; the windows were simply made unevenly.
• In 1948 Walter Kauzmann noticed a paradox. If the trend in a supercooled liquid's entropy continued below the glass transition, the liquid would reach lower entropy than the crystal at a finite temperature, which seems impossible. Something must intervene, and what it is remains debated.
• New methods have made glasses easier to study. Glasses grown slowly by depositing vapour layer by layer, first reported in 2007, are far more stable than ordinary ones, equivalent to glass aged for thousands of years. Since 2017, computer algorithms have been able to equilibrate simulated liquids at temperatures that used to be out of reach.

Main hypotheses

A hidden phase transition

According to the random first-order transition theory, developed from 1989, a true thermodynamic transition lies beneath the glass transition, at a temperature that can never be reached in practice. As the liquid cools, regions of molecules become cooperatively locked into one of an enormous number of possible disordered arrangements, and these regions grow.

Purely dynamic slowdown

The opposite view, called dynamic facilitation: there is no hidden thermodynamic transition. Molecules can only move if their neighbours have moved, and mobility spreads through the liquid in rare, localised bursts. The slowdown is a matter of kinetics, not of an underlying state of matter.

Growing structure

Perhaps there is a subtle form of order in the liquid, invisible in standard measurements, whose growth controls the slowdown. Researchers have looked for such order with new statistical tools, with partial success.

What would settle it

• Evidence of a growing length scale of cooperatively moving molecules, measured directly in real liquids as they approach the glass transition.
• Ultrastable glasses or new simulations reaching low enough temperatures to show whether the Kauzmann paradox is resolved by a phase transition or avoided by dynamics.
• A theory that predicts, from the properties of the molecules, how a given liquid will slow down — and is confirmed for liquids it has not been fitted to.