Work hardening

Why does bending a paperclip make it stiffer exactly where you bent it, right up until the moment it breaks there?

Bend a paperclip back and forth at one spot. The second bend takes more force than the first, the fifth more than the second, and then, without ceremony, the wire parts at exactly the place that had been getting stronger. Two descriptions of that region seem true at once: it is the toughest part of the wire, and the part about to fail.

We normally treat "strengthening" and "damaging" as opposites, and here one process is doing both. So the assumption worth questioning is not what the bending adds — it adds nothing, the composition being identical before and after — but whether strength and durability were ever the same property.

Start from why a metal is soft at all. Its atoms sit on a lattice, and it yields not by breaking every bond across a plane at once but by moving a line defect — a dislocation — across that plane one atomic row at a time. Cheap, local, and the reason real metals give way at a fraction of their theoretical strength. Hardening a metal does not strengthen its bonds; it obstructs that line.

In an alloy the obstacles are foreign atoms deliberately added. But the paperclip gains no atoms. What obstacle appears?

Follow the naive prediction first, because it is instructive when it fails. Dislocations glide to a free surface and leave the crystal, so deforming the metal enough should run it out of dislocations and make it softer. It does the opposite, so they must be created faster than they escape. They are: a line pinned at two points bows out under stress like a soap film in a wire loop, wraps around, pinches off a closed loop and restores the original segment, ready to do it again. Deformation is a dislocation factory. Annealed metal carries something like ten billion metres of dislocation line per cubic metre and heavily cold-worked metal a million times more — both recalled as orders of magnitude, varying with the metal.

Now the pivot. Why should more of the thing that permits slip make slip harder? Because each dislocation drags an elastic strain field around it, and one gliding on a given plane must cut through every other dislocation threading that plane. The population obstructs itself. The classical relation says the extra stress required rises with the square root of the dislocation density — so quadrupling the density only doubles the added strength. That single square root explains the shape of the whole phenomenon: the first bend hardens far more than the tenth, and the wire's resistance climbs steeply and then flattens.

Why at the bend, and only there? Hardening's natural tendency is the reverse: a hardened patch resists further strain, so the next increment goes to its softer neighbours, which is why a bar in tension stretches uniformly for a long while instead of thinning immediately. In the paperclip, geometry overrules that — once a kink exists it is where the wire is most curved and where your fingers apply the largest moment, so deformation returns to the same millimetre no matter how hard it has become.

And why does it end? Because the hardening capacity is finite. As the density climbs, dislocations of opposite sign meet and annihilate, until annihilation balances creation and the density saturates. The metal stops hardening. Separately and in parallel, deformation has been pulling the lattice away from hard inclusions, opening microscopic voids that grow and link. Once the material can no longer harden fast enough to push strain elsewhere, strain concentrates in one band, the voids there coalesce, and the wire parts.

One more question decides how well we understand this. A tangle of stored defects holds energy — a small fraction of the work you did, a few percent by most published estimates, the rest leaving as heat. Stored energy means the cold-worked metal is thermodynamically unstable with respect to the annealed metal. So why does a bent paperclip stay hard in a drawer indefinitely?

Because thermodynamics only says which way a state would go if it could move. Rearranging that tangle needs atoms to hop — dislocations to climb and cancel, then strain-free grains to nucleate and eat the deformed ones — and at room temperature the metal has nothing like enough atomic mobility. Heat steel or copper to roughly four-tenths of its absolute melting temperature (a rule of thumb, not a constant) and recrystallisation wipes the whole deformation history out in minutes. Kinetics, not thermodynamics, is what makes work hardening permanent in practice. Read the other way, the same fact explains hot rolling: deform a metal above that temperature and recovery keeps pace, so it never hardens at all.

Three refinements hold the picture together.

Hardening and damage run in parallel; they are not one process. The dislocation tangle is fully reversible — anneal it and the metal is as it was, same composition, same shape. The voids at inclusions are not reversible by heat. Early bends are mostly the first; the last bend is decided by the second.

"Stronger" here means a higher stress is needed to yield further. It does not mean the part will survive more. Energy absorbed before fracture — toughness — falls the whole time the yield stress rises, and toughness is what a paperclip needs.

And there is a clean falsification test. Take the hardened bend, heat it above its recrystallisation temperature, cool it, and re-test. If hardening really is stored dislocation structure, the region must return to near its original softness and ductility with no change in composition or geometry; if the hardening survived annealing, the account is wrong and something deeper has happened. The metals that quietly confirm it are the low-melting ones: lead and tin sit above four-tenths of their melting point at room temperature, so they anneal themselves as you bend them and barely work-harden at all.

• How rolling schedules are designed around the saturation point, with anneals between passes.

Every term the collection defines is gathered in the glossary.