What 'amorphous' really means
Every rung you have climbed so far leaned on one quiet assumption: that a solid is a crystal — an endless three-dimensional wallpaper pattern, the same motif stamped at every lattice point, describable by a single unit cell and one of the 14 Bravais lattices. That assumption bought us everything: Miller indices, the reciprocal lattice, Bragg's law. Now we throw it away. An amorphous (non-crystalline) solid has no lattice at all — no unit cell, no repeating pattern, no long-range order. Window glass, a silica optical fibre, most plastics, and a rapidly frozen metal are all amorphous. The obvious question is: if there is no lattice, is there any structure left to talk about?
The answer, emphatically, is yes — and getting this right is the whole point of the rung. 'Amorphous' comes from the Greek for 'without form', but it does NOT mean 'without order'. Look closely at any single atom in silica glass and you find it is not adrift: each silicon still sits at the centre of a tidy tetrahedron of four oxygens, with a silicon-oxygen bond length of about 1.6 angstrom, exactly as in crystalline quartz. Every atom has a definite number of nearest neighbours at definite distances and definite bond angles. This is genuine short-range order — sharp, real, and chemically identical to the crystal's. What the glass has lost is only the long-range repetition: follow the tetrahedra outward and, after two or three neighbours, small variations in the linking angles accumulate and all memory of a regular pattern washes away.
Glass: a liquid caught in the act
The most useful picture of a glass is a liquid caught in the act of freezing. Cool any liquid and its atoms slow and draw closer, so its volume shrinks steadily. At the melting point Tm the atoms would normally snap into their lowest-energy arrangement — a crystal — releasing latent heat and dropping abruptly to a smaller, denser volume. But crystallising takes TIME: atoms must find and file into their proper lattice sites. If you cool fast enough to outrun that, the liquid sails straight past Tm as a supercooled liquid, still disordered, still contracting smoothly, with no sudden jump.
Keep cooling and something has to give. As the supercooled liquid gets colder and denser, its atoms have less and less free volume — the little pockets of extra space they need in order to shuffle past one another — and they move ever more sluggishly. At a temperature called the glass transition Tg the rearrangements become so slow (viscosity soars past about 10^12 pascal-seconds) that, within any human timescale, the atoms simply stop moving. The structure is frozen: you now have a glass, a solid that holds the disordered, liquid-like arrangement it had at Tg. Crucially, Tg is a KINETIC, not a thermodynamic, landmark — it marks where atomic motion falls behind the clock, not a true phase transition. Cool faster and the atoms freeze sooner, so Tg shifts slightly higher; the glass you get literally depends on how fast you made it.
COOLING A LIQUID: TWO FATES (volume vs. temperature)
high T .......... liquid .......... low T
|
reach Tm (melting point)
/ \
crystallise keep cooling
(given enough time) (fast: outrun Tm)
| |
SUDDEN volume DROP supercooled liquid
ordered CRYSTAL |
packing ~ 0.74 reach Tg (glass transition)
|
gentle KINK; motion frozen
GLASS: liquid-like disorder
packing ~ 0.64The diffraction signature: halos, not spots
How do we actually SEE that a solid is disordered? Point a diffraction camera at it. Recall the crystal's trick: its planes act like evenly spaced cliff walls, and the scattered waves — echoes — only add up in phase at the special angles set by Bragg's law, so a crystal throws sharp, discrete spots, a direct photograph of its reciprocal lattice. A glass has no planes and no reciprocal lattice, so those pinpoint echoes never build up. Instead you get a few broad, diffuse rings — halos — smeared across a range of angles. That halo pattern is the diffuse structure factor S(Q), and its very broadness is the fingerprint of disorder: no sharp spots means no long-range periodicity.
But a halo is not empty of information — its position and width still encode the short-range order. The trick is to measure ALL the scattered intensity (Bragg peaks and diffuse background alike — 'total scattering') over a wide range of the scattering vector Q, and Fourier-transform it. Out drops the radial distribution function, usually written g(r): a curve of how the atom density around a typical atom rises and falls with distance r. For a glass, g(r) is a series of blurry peaks that fade out within a nanometre — order that is sharp up close and gone far away, exactly what 'short-range order' should look like. Its first peak marks the nearest-neighbour shell, and the AREA under that peak counts the coordination number — for silica glass, close to 4, the same four oxygens around each silicon as in the crystal. The next guide builds this function properly; here just hold the idea that disorder is measured not by sharp spots but by a smooth, decaying g(r).
- Shine X-rays, neutrons, or electrons on the glass and record the scattered intensity at every angle — not just where peaks might be, but the whole diffuse background too.
- Convert angle to the scattering vector Q, then correct and normalise the data to obtain the structure factor S(Q) — those broad halos.
- Fourier-transform S(Q) over Q into real space; the result is the radial distribution function g(r).
- Read structure straight off g(r): the first peak's position is the nearest-neighbour bond length, and the area under it is the coordination number.
Two ways to be disordered: networks and packings
If a glass has real local structure, we ought to be able to model it — and there are two great pictures, one for each broad family of glass. Covalent glasses, like silica and everyday window glass, are captured by Zachariasen's continuous random network (1932). The idea is elegant: keep the rigid, directional building block — the SiO4 tetrahedron — intact, because the strong covalent bonds fix the O-Si-O angle near 109.5 degrees. But let neighbouring tetrahedra join corner-to-corner through shared oxygens (bridging oxygens), with the Si-O-Si linking angle FREE to vary by ten or twenty degrees from one junction to the next. Those small, random angle choices let the network fill space forever without ever repeating — a covalent scaffold that is fully connected yet perfectly aperiodic. Add a network modifier like soda (Na2O) and it snaps some of those bridges, creating loose non-bridging oxygens that soften the glass and lower its melting point — which is exactly why we melt sand with soda to make bottles. Guide 4 is devoted to this network.
Metals are the opposite case. A metallic bond has no direction to satisfy — the atoms behave much like hard spheres that simply want to touch as many neighbours as possible. So a metallic glass is modelled not as a network but as random close packing: hard spheres poured together and jammed as densely as they can go WITHOUT settling into an ordered stack. Compare the numbers with the grocer's orange pyramid from the crystal rung — an ordered close packing (FCC) reaches a packing fraction of 0.74 with each atom touching 12 neighbours. Random close packing jams at only about 0.64, a little looser, with a slightly lower and more scattered coordination. That small gap — the few percent of empty space that ordered stacking would have squeezed out — is the structural signature of the metallic glass, and it is the source of the free volume we met earlier. Guide 5 develops random close packing and how metals are frozen into it.
Why some liquids freeze into glass
So which liquids actually become glasses? In principle any of them — glass-forming is a race between cooling and crystallising, and if you cool fast enough you always win. What differs enormously is HOW fast you must go. Silica is a champion glass-former: near its melting point it is already a stiff, tar-like network whose atoms can barely move, so crystals nucleate hopelessly slowly and it turns to glass even when cooled by hand over minutes. Pure metals are the nightmare case — their mobile, non-directional atoms crystallise almost instantly, so the first metallic glasses demanded rapid quenching on the order of a million degrees per second, achieved by squirting the melt onto a spinning chilled copper wheel (melt-spinning) to freeze a thin ribbon before a single crystal could form.
Why do some liquids resist crystallising far better than others? A deep part of the answer is geometric frustration. In many metallic liquids the arrangement each atom LOCALLY prefers — the densest little cluster, an icosahedron of twelve neighbours — carries a five-fold flavour of symmetry. But recall the shock of quasicrystals and the crystallographic restriction: a periodic crystal cannot have five-fold symmetry, so those cosy icosahedral clusters simply CANNOT tile space into a crystal. The locally favoured packing is frustrated — it cannot grow into a periodic solid — so crystallisation stalls and the supercooled liquid is stabilised, giving atoms more time to freeze into a glass. Frustration is one ingredient among several (a big spread of atomic sizes and deep eutectics help too), and modern bulk metallic glasses stack the deck by mixing four or five elements so that no single crystal can easily form, letting them glass at a gentle one-to-a-hundred degrees per second.