Dislocations: Why Metals Bend and Ceramics Shatter

Dragging a heavy rug across a floor takes enormous force, because every square metre of contact resists at once. Kick a ripple into one end and walk it across, and the same displacement costs almost nothing: only the strip under the ripple is moving at any instant.

A dislocation is the ripple. It is a line along which the crystal's atomic planes are misregistered, and moving that line one step forward breaks and reforms one row of bonds rather than an entire plane's worth.

Definition: A dislocation is a one-dimensional crystal defect. In an edge dislocation an extra half-plane of atoms is inserted, and the line is its edge. In a screw dislocation the planes are sheared into a helix. Real dislocations are mixed, curving between the two characters along their length.

The force to move one is a few orders of magnitude below the force to shear a plane rigidly, which is exactly the gap the previous lesson could not account for.

A dislocation does not glide in an arbitrary direction. It moves on the most densely packed planes, in the most densely packed directions, because those are where the atomic misfit costs least to shuffle along. A plane-and-direction pair is a slip system.

Von Mises showed that a polycrystal needs five independent slip systems to deform to an arbitrary shape, since each grain must accommodate its neighbours. FCC metals have plenty. Magnesium has three easy ones, which is why magnesium sheet cracks when cold-formed and must be worked hot.

Key idea: Ductility is not softness. It is having enough independent ways for a crystal to shear that every grain can follow whatever its neighbours do.

This is the sharpest illustration of the whole subject. Bond strength sets the ceiling, dislocation mobility decides whether you get anywhere near it, and a material with strong bonds and no mobility is a material that fails without warning.

Glass has no crystal structure at all, so it has no slip planes either, and the same conclusion applies for a different reason.

If plastic deformation is dislocations moving, strength is whatever stops them. There are four standard obstacles and every commercial alloy uses some combination.

A modern high-strength steel or aluminium alloy is all four at once, tuned against each other. That is what an alloy designation encodes: not just a composition but a processing route, which is why 6061-T6 and 6061-O are the same metal with a factor of three between their yield strengths.

Grain boundaries obstruct dislocations because slip directions do not line up across them: a dislocation reaching a boundary cannot simply continue into the neighbouring grain. More boundaries per unit volume means more obstruction, and the relationship is empirical but reliable:

$\sigma_y = \sigma_0 + \frac{k}{\sqrt{d}}$

where $d$ is the average grain diameter. This is the Hall-Petch relation. Halving the grain size raises yield strength by a factor of about $\sqrt{2}$.

Key idea: Grain refinement is the one strengthening mechanism that improves toughness at the same time, because finer grains also make a crack turn more often as it propagates. Every other mechanism on the table trades one against the other. This is why controlled rolling and rapid solidification are worth their process cost.

The relation reverses at very small grain sizes, roughly below 10 to 20 nanometres, where boundaries become so numerous that grains slide past each other instead of deforming internally. That is the inverse Hall-Petch regime, and it sets a floor on how far the trick can be pushed.

Bend a paperclip back and forth and it gets stiffer, then harder to move, then snaps. Each of those stages is dislocations accumulating: the density can rise from around $10^{10}$ lines per square metre in an annealed metal to $10^{15}$ or more after heavy cold work.

Dislocations obstruct each other, so more of them means a higher stress is needed to move any of them. The material gets stronger and simultaneously runs out of capacity to deform further, which is why the paperclip's last bend is a fracture rather than a bend.

In practice: Annealing reverses it. Heat the metal enough for atoms to diffuse and the dislocation tangle reorganises into fresh, low-density grains, restoring ductility and giving back the strength. Every sheet-metal process that involves multiple forming steps has anneals scheduled between them for exactly this reason, and forgetting one is how a part cracks in the press rather than in service.

Dissolve an alloying element at high temperature, quench so it has no time to leave solution, then hold at a moderate temperature and let it come out as a fine dispersion of hard particles. Dislocations must either cut through those particles or bow around them, and both cost stress.

Alfred Wilm found the effect in 1906 while trying to harden aluminium by quenching, in the way steel is hardened. It did not work. The samples were left over a weekend and were measurably harder on Monday, and the alloy that resulted, duralumin, made aluminium a structural material and eventually made aircraft possible.

Gotcha: Precipitates are only strengthening at the right size and spacing. Hold the alloy hot for too long and the particles coarsen, spacing widens, dislocations bow through more easily, and the alloy softens. This is over-ageing, and it happens in service to any precipitation-hardened part running at elevated temperature, which is why an aluminium alloy has a maximum service temperature quoted alongside its strength.

Iron changes crystal structure with temperature: FCC austenite above about 910 degrees Celsius, BCC ferrite below. Carbon dissolves readily in the FCC form and barely at all in the BCC one, and that mismatch is the entire basis of heat treating steel.

Martensite forms because the transformation is too fast for carbon to diffuse out, so the lattice shears into a strained structure with the carbon stuck inside it. It is hard because that strain field obstructs every dislocation everywhere.

In practice: Quenched martensite is too brittle to use. Tempering, reheating to a few hundred degrees, lets a little carbon precipitate out and trades some hardness for a large gain in toughness. The tempering temperature is the dial, and essentially every load-bearing steel component in the world has been set on it.

Everything so far assumed dislocations move only when pushed. Above roughly 40 percent of a material's absolute melting temperature, atoms diffuse fast enough that a dislocation blocked by an obstacle can simply climb around it, given time.

The consequence is creep: slow, permanent deformation under a load well below the yield strength. A turbine blade at 1,000 degrees Celsius is not close to yielding, and it stretches anyway, over thousands of hours.

Gotcha: Creep breaks the mental model that a part is safe if the stress is below yield. Below the creep threshold that model holds; above it, life is measured in time under load rather than in load. Steam pipes, turbine blades, solder joints at room temperature (tin melts at 232 degrees Celsius, so room temperature is already 60 percent of its absolute melting point) and lead roofing all creep.

The defences are grain boundaries, again, but inverted: since creep runs partly along boundaries, creep-resistant designs use very large grains, or none. Single-crystal turbine blades are grown as one grain precisely so there are no boundaries to slide along.

Dislocation engineering is why a modern alloy can be ten times stronger than the pure metal it is based on. It is also, entirely, a theory of deformation.

None of it explains why a component that never approached its yield strength falls apart anyway. A pressure vessel operating at a third of yield can burst; an axle running for years at a safe stress can snap without warning; a bracket that passed every static test can fail after a million cycles.

Key idea: Dislocations govern how a material yields. Cracks govern how it breaks. They are different physics with different mathematics, and a design checked only against yield strength is checked against one of the two.

Most catastrophic structural failures in the historical record are crack failures, not yield failures, and the field that studies them was created after two specific disasters. That is the next lesson.