the continuous random network
/ Zachariasen: zah-kar-ee-AH-sen /
How can a material have no repeating pattern and yet be a rigid, strong solid rather than a loose powder? In 1932 W. H. Zachariasen answered this for glasses like silica with a beautifully simple picture: the continuous random network. Imagine a vast three-dimensional jungle gym built from identical rigid units, each firmly bolted to its neighbours — but where the connecting joints are allowed to swivel to slightly different angles at every junction. The result is one single, fully connected, space-filling framework that is perfectly bonded everywhere yet never repeats. It is 'continuous' because the bonding runs unbroken through the whole solid, and 'random' because the linking angles vary.
For silica (SiO2) the rigid unit is the SiO4 tetrahedron: one silicon bonded to four oxygens, exactly as in crystalline quartz. In both the crystal and the glass every silicon has four oxygen neighbours and every oxygen bridges two silicons, so the short-range order is essentially identical. The difference lies entirely in the joints: the silicon-oxygen-silicon angle where two tetrahedra meet is locked to one value in the crystal but is free to spread over a range (roughly 120 to 180 degrees, peaking near 144 degrees) in the glass, and the tetrahedra can also rotate about the bonds. This modest freedom, repeated at joint after joint, destroys long-range periodicity while keeping every bond satisfied — no dangling bonds, no wrong coordination. Zachariasen even set out rules for which oxides could do this, essentially requiring small, low-coordination cations that share only corners, not edges or faces.
The continuous random network is the foundational structural model for covalent and oxide glasses — silica, boron oxide, germania, and by extension amorphous silicon and germanium. Its great virtue is honesty: it explains why glasses are rigid (a fully bonded network resists deformation) and why they have sharp short-range order but no diffraction spots (identical local units, random linkage) without pretending there is any hidden crystallinity. Its main limitation is that the strict Zachariasen picture is an idealisation: real network glasses also contain network modifiers that snip the network, small voids, and a non-random distribution of ring sizes, so the network is neither perfectly continuous nor perfectly random. It is a superb first model, not the last word.
Crystalline quartz and silica glass are built from the very same SiO4 tetrahedron, corner-sharing so that each oxygen bridges two silicons. In quartz the Si-O-Si bridging angle is fixed and the tetrahedra repeat in perfect helical order. In the glass that bridging angle wanders from junction to junction, so the identical tetrahedra assemble into one continuous but non-repeating network — same bricks, same mortar, but no blueprint.
Identical rigid units, joined at randomly varying angles: one connected network, no repeat.
The continuous random network keeps the crystal's local coordination but discards its long-range periodicity — it is the structural embodiment of 'short-range order without long-range order.' Do not read 'random' as chaotic: bond lengths and coordination are tightly fixed; only the linkage angles are free.