Short- and Medium-Range Order

The order you keep, and the order you lose

The previous guide left you with a puzzle. An amorphous solid is as rigid as a crystal, yet when you shine X-rays through it you get no sharp spots — only a few broad, blurry halos. It is tempting to read that blur as 'no order, just randomness'. That reading is wrong, and correcting it is the whole point of this guide. A glass is not a snapshot of a gas frozen mid-tumble; it is a highly correlated arrangement that has lost just one thing. The single most important sentence in this rung is this: a glass has no long-range order, but it keeps real short-range order and even medium-range order.

The key to seeing this is to stop treating order as a switch that is either on or off, and start treating it as something that fades with distance. Picture a crowd milling in a plaza. Each person keeps a definite arm's-length gap from their immediate neighbours — that is order at short range. Little knots and rings of friends form and hold together for a while — order at medium range. But there is no stadium-wide seating grid telling everyone exactly where to stand — no order at long range. A crystal IS the seating grid, repeating to infinity. A glass is the plaza crowd: locally polite, globally free.

Short-range order: the bond that survives

Short-range order (SRO) is the first shell of neighbours — how many there are, how far away they sit, and at what angles — and in a glass it is almost identical to the crystal's. Take vitreous (glassy) silica, SiO2. Every silicon still sits at the centre of a tidy SiO4 tetrahedron: four oxygens, each held by a strong directional covalent bond at a Si-O distance of about 1.6 angstrom, with the O-Si-O angles clustered near the ideal tetrahedral 109.5 degrees. Melt quartz into glass and you do not break these tetrahedra — the building block comes through intact. That is short-range order: the local unit is preserved.

This is exactly why a glass is stiff and why its structure gives one sharp signal amid all the blur. Every silicon has the same four oxygen neighbours at the same distance, so the very first shell of near neighbours is well defined — its coordination number is a clean 4. The same is true far beyond silica: in a metallic glass each atom hugs roughly 12 to 13 near neighbours, echoing the dense close packing of the parent crystal but jumbled; even in liquid water each molecule keeps its four hydrogen-bonded partners. Short-range order is nearly universal — every liquid and every glass has it, because atoms simply cannot occupy the same space and bonds have preferred lengths.

Be honest about how sharp this really is. The spread in the Si-O bond length in glassy silica is tiny — a fraction of a hundredth of a nanometre — so the first shell is as crisp as the crystal's first shell. Nothing is fuzzy here. What the glass loses is not the local unit but the instructions for how one unit connects to the next as you march outward. Hold onto that distinction: short-range order is about the tetrahedron; everything that follows is about how the tetrahedra are wired together.

Medium-range order: the subtle middle

Between the crisp first shell and the featureless far distance lies the hardest and most interesting regime: medium-range order (MRO), the correlations that survive out to roughly half a nanometre to a couple of nanometres — a few shells deep. In silica this is the level of how tetrahedra link up. Neighbouring tetrahedra share a corner oxygen, a bridging oxygen that belongs to two silicons at once, and those linked tetrahedra curl round into rings — most often rings of five, six, or seven tetrahedra. There is a real, non-random preference for certain ring sizes and linkage geometries; that preference IS medium-range order.

The reason long-range order dies while short-range order lives is a single hinge: the Si-O-Si angle where two tetrahedra meet at a bridging oxygen. That bridging angle is soft — in glass it ranges from about 120 to 180 degrees, peaked near 144 degrees — so each new corner-link can twist a little. Twist a rigid tetrahedron slightly at every joint and, after only a handful of links, all memory of a starting direction is gone. The tetrahedron (short range) stays perfect; the wandering hinge (medium range and beyond) erases the long-range repeat. Medium-range order even leaves a faint fingerprint in diffraction: a low-angle 'first sharp diffraction peak' whose position points to a characteristic length of a few angstrom — the scale of the inter-tetrahedral network.

Seeing disorder with statistics

How do you describe order that has no lattice to hang numbers on? You cannot index a glass with sharp reflections — with no repeating lattice there are no planes for Bragg's law to select, so instead of spots you get those broad diffuse halos in the structure factor S(Q). The trick is to stop asking 'where is each atom' and start asking a statistical question: on average, how likely am I to find another atom a distance r away from any given atom? That average is captured by the radial distribution function and its close relative the pair distribution function, written g(r).

You can read the three ranges of order straight off the shape of g(r), and the next guide is devoted to doing exactly that. For now, the picture below is the one to carry: a sharp first peak whose POSITION is the nearest-neighbour bond length and whose AREA is the coordination number (short-range order); a couple more bumps that blur out over a nanometre or two (medium-range order); and then a flat line at the average density, meaning any two far-apart atoms are uncorrelated (no long-range order). This way of working — keeping the diffuse scatter that a crystallographer usually throws away — is called total scattering, and g(r) is its natural output.

PAIR CORRELATION  g(r):  how likely is a neighbour at distance r?

 CRYSTAL  (long-range order)
   g |   |    |    |     |      |       |       sharp spikes,
     |   |    |    |     |      |       |       forever
     +---^----^----^-----^------^-------^----->  r
         1st  2nd  ...  peaks stay sharp to infinity

 GLASS / LIQUID  (short + medium range only)
   g |    /\
     |   /  \    __                          1  <- levels off
     |  /    \__/  \__     __________----------  (no LRO)
     +-/-------------------------------------->  r
        ^1st peak   ^2nd,3rd     blur out by ~1-2 nm
        SRO         MRO          then structureless

 first-peak POSITION = nearest-neighbour bond length  (~1.6 A, Si-O)
 first-peak AREA     = coordination number            (4 for Si in SiO2)

Two pictures, and why glasses form at all

If a glass keeps such definite short- and medium-range order, what does the whole structure look like? Two idealized model pictures answer this, one for each bonding style, and the last two guides in this rung build them out in full. For covalent oxide glasses like silica, the answer is the continuous random network — Zachariasen's beautiful idea that the glass uses the very same corner-sharing tetrahedra as the crystal, wired into one endless network, but with the bridging hinge randomized so the pattern never repeats. Add a modifier like soda and some bridging oxygens are cut into dangling ends, which is the difference between a network former and a network modifier — the subject of guide 4.

For metals, whose bonds are undirected, the picture is completely different: random close packing of hard spheres. Pour ball bearings into a jar and shake, and they jam at a packing fraction of about 0.64 — denser than a loose pile but short of the crystal's 0.74 — with each sphere touching roughly 12 to 13 others in a jumble rich in five-fold, icosahedral little clusters. That is the skeleton of a metallic glass, and guide 5 is devoted to it. Two very different models, one shared theme: identical local order to the crystal, no long-range repeat.

One question remains: why does a liquid freeze into this jammed, non-repeating state instead of doing the tidy thing and crystallizing? The answer is a race against the clock. Cool a liquid below its melting point and it WANTS to crystallize, but crystallizing means every atom finding its exact lattice site, which takes time and atomic motion. Cool fast enough and the liquid thickens — its viscosity soars — until the atoms freeze in place before they can find those sites. The temperature where this kinetic freezing happens is the glass transition, Tg, and the extra elbow room the atoms could not squeeze out is the trapped free volume that makes a glass slightly less dense than its crystal.

1. Start with the liquid above its melting point, atoms wandering freely and only short-range order present.
2. Cool it fast — for a metal, up to a million degrees per second by splatting a droplet onto a chilled wheel; this rapid quenching leaves no time to crystallize.
3. As temperature falls the viscosity climbs steeply and atomic rearrangement slows to a crawl.
4. At the glass transition Tg the atoms lock in place; the liquid's short- and medium-range order is frozen, its free volume trapped.
5. The result is a rigid solid with a liquid's structure — a glass — that never grew the long-range order of a crystal.

There is one more reason some liquids resist crystallizing, and it ties straight back to a fact from the symmetry rung: a periodic crystal cannot have five-fold symmetry. Yet the cheapest, densest local cluster of atoms is often the icosahedron — thirteen atoms with exactly that forbidden five-fold local order. A liquid full of comfy icosahedra faces frustration: its favourite local packing simply cannot tile space to build a crystal, so it dawdles, buying the time needed to freeze as a glass. Metals whose liquids lack this help crystallize almost instantly, which is why making a metallic glass demands such violent rapid quenching. The tools sketched here — g(r), the coordination number, and the diffuse S(Q) — are the instruments you will now pick up: the next guide turns the pair distribution picture into a working measurement.