Why Metal Bends and Glass Shatters

Calculate the strength of a perfect metal crystal from the bonds between its atoms and you get a figure roughly a hundred times what real metals actually withstand. For a long time this was a genuine embarrassment: the theory was straightforward and the answer was wrong by two orders of magnitude.

The resolution, proposed in the 1930s and confirmed by electron microscopy two decades later, is that real crystals are not perfect. They contain line defects called dislocations: imagine an extra half-plane of atoms inserted into an otherwise regular stack, with a distortion running along its edge. That distortion can move under modest stress, and as it moves it carries deformation with it, letting one plane of atoms slip past another a row at a time rather than all at once.

The analogy that sticks is moving a heavy rug. Dragging the whole thing at once requires overcoming friction across the entire area. Pushing a ripple along it moves the same rug the same distance while only ever lifting a small strip. Dislocations are ripples in a crystal, and they are why metals bend instead of snapping, why they can be rolled, drawn and forged, and why civilisation is built out of them.

Metals are therefore weak because they are imperfect, and useful for exactly the same reason. This has a practical corollary that governs the entire discipline of metallurgy: almost every technique for strengthening a metal works by obstructing dislocation movement. Alloying inserts atoms of a different size that distort the lattice and impede slip. Work hardening multiplies dislocations until they tangle and jam each other. Grain refinement adds boundaries they cannot easily cross. Precipitation hardening places hard particles in their path. Four different-looking processes, one mechanism.

The cost is always the same, and it is worth knowing before anyone specifies a material. Strength is bought with ductility. A metal that cannot slip readily cannot deform before it breaks, so the strongest alloys tend to fail suddenly rather than bending first, and a structure that bends visibly before failing is a structure that warns you.

Glass sits at the far end of this. It has no crystal planes, being an amorphous network, so there is no slip mechanism at all. Applied stress cannot be relieved by deformation and concentrates instead at the tip of whatever flaw happens to be present until a bond breaks, and then the next, and the crack runs. There is no intermediate state between intact and fractured.

Griffith showed in 1921 what this implies: the strength of a brittle material is set by its worst existing flaw rather than by the bulk material. Freshly drawn glass fibres are extraordinarily strong until something touches the surface. Toughened glass exploits the same principle from the other side, by rapid cooling that leaves the surface in compression, so a crack must first overcome the compressed layer. It is also why such a pane shatters entirely rather than cracking locally: penetrate the compressed skin anywhere and the stored tension in the interior has an escape route through the whole sheet.