The strength that never shows up
By now you can read a stress-strain curve and you know the deep secret of the last rung: real metals are far weaker than a perfect crystal, because slip lets dislocations glide instead of tearing whole planes of bonds apart at once. That story explains why metals yield so easily. But there is a second, separate gap between the strength a material should have and the strength it actually delivers — and this one is about breaking, not bending. Follow it and it leads straight into the whole subject of failure.
Add up the bond strengths and a perfect solid should hold roughly E/10 before it pulls apart — for glass, whose Young's modulus is about 70 GPa, that predicts a cohesive strength near 7000 MPa. Yet an ordinary glass rod snaps at maybe 50 to 100 MPa, about a hundred times weaker. Where did 99 percent of the strength go? It hid inside tiny flaws. Every real surface carries microscopic scratches, pores, and cracks, and a crack does something vicious: it acts as a lever for stress. This is stress concentration — the applied load gets funnelled and magnified at the sharp tip of the flaw, so the material there feels many times the average stress even though the bar as a whole is loaded gently.
Two ways to break
When a part finally gives way, it does so in one of two utterly different characters. In ductile fracture the material fights the whole way: it yields, flows, necks down, and only then tears apart, soaking up a lot of energy and giving you loud warning through visible bending. In brittle fracture a crack simply runs — it races across the section at close to the speed of sound with almost no plastic flow, no neck, and no warning at all. The everyday feel is the difference between slowly stretching a warm toffee until it thins and parts, versus a dropped teacup that is whole one instant and shards the next.
The difference is written at the microscopic scale. Ductile fracture is slip taken to the limit: under heavy plastic strain, tiny voids nucleate at inclusions and second-phase particles, grow as the metal stretches, and finally link up — microvoid coalescence — leaving a fibrous, dimpled surface. Brittle fracture is cleavage: bonds snap straight across specific crystal planes, splitting grain after grain along flat, mirror-like facets with barely any plastic work. This is why the two modes have such different toughness: ductility is the reservoir of plastic work that a ductile crack must pay for every millimetre it advances, and a brittle crack pays almost nothing.
DUCTILE (a tough metal) BRITTLE (glass, grey cast iron) load load | | v ___ v ___________ ========= =============== <- flat, ~90 deg | | <- necks down | | to the load | ( <- cup & cone | | | ) slant + fibrous =============== ========= dimpled centre crack ran straight across, ^ ^^^ shiny cleavage facets + | lots of plastic work chevron marks -> origin absorbs energy, WARNS you near-zero plastic work, NO warning
One caution against a common misconception: 'ductile' and 'brittle' are not permanent labels stamped on a material. The same steel can fracture in a ductile way when it is warm and slowly loaded, yet snap in a brittle way when it is cold, notched, or struck fast. Whether a part behaves like toffee or like a teacup depends on temperature, loading rate, and the sharpness of any notch — which is exactly the trap the rest of this guide unpacks.
Reading the wreckage: fractography
A broken surface is not garbage — it is a written confession. Fractography is the craft of reading it, and it is the heart of real-world failure analysis. The moment a crack moves it scribes its history into the surface it creates, so an investigator can often reconstruct where the failure began, which way the crack ran, how fast, and under what kind of load — sometimes from a part fished out of a river months later.
- Look at the overall shape first. A necked-down, slanted, grey and fibrous surface says ductile; a flat surface roughly at 90 degrees to the load, bright and granular, says brittle. This one glance already tells you which world you are in.
- Find the chevrons. Brittle surfaces often carry V-shaped chevron marks or a radiating fan pattern; both point back like arrows to the single spot where the crack was born — usually a flaw, weld defect, or sharp corner.
- Zoom in with a microscope. Under an electron microscope, ductile fracture shows a field of tiny cups (the dimples of microvoid coalescence), while brittle cleavage shows flat facets with river-like line patterns fanning downstream in the crack's direction.
- Watch for beach marks and striations. Concentric ripples like a clamshell, and at high magnification a ladder of fine parallel striations, are the fingerprint of fatigue — a slow crack that grew a little on each load cycle. That signature is so important it gets its own guide (guide 4).
The payoff is enormous: because the surface records the origin, fractography usually reveals whether a failure was a genuine overload, a hidden manufacturing defect, or slow fatigue — and that answer decides whether you redesign the part, tighten the factory inspection, or simply retire parts before they age out. It is detective work with real stakes, and every dramatic crash investigation leans on it.
When ductile turns brittle: the transition and the hammer test
Here is the treacherous fact that has sunk ships and cracked pipelines. Many metals — especially ordinary body-centred-cubic steels — undergo a ductile-to-brittle transition: above a certain temperature they fracture in a tough, energy-eating ductile way, but cool them below it and the very same steel turns glass-brittle, snapping with little warning. Face-centred-cubic metals like aluminium, copper, and austenitic stainless steel mostly do not show this transition and stay tough even in deep cold — which is precisely why cryogenic tanks are built from them, not from plain steel.
How do you measure this cheaply? Not with a slow tensile test but with a fast, brutal one: impact testing. In the Charpy test a swinging pendulum hammer strikes a small notched bar and shatters it; the energy the bar absorbed is read from how high the pendulum swings up on the far side — a high swing means little energy absorbed, so a brittle break. Run the test at a ladder of temperatures and plot absorbed energy against temperature, and a BCC steel shows a clear S-shaped drop: high-energy ductile at the top, low-energy brittle at the bottom, with the transition temperature in the fall between. The deliberately sharp notch matters — it is a built-in stress raiser that mimics the worst flaw a real part will meet.
The map of the whole rung
You now have the frame. Fracture is not one thing but a family, and the four guides ahead each take one member deeper. Guides 2 and 3 finish the crack story: the stress concentration a flaw creates gets turned into a single powerful quantity, the stress intensity factor K, which measures how fiercely the load is amplified at a crack tip. Compare K against the material's own fracture toughness K_IC — a genuine material property, as real as a modulus — and the Griffith criterion hands you the critical crack length: the exact flaw size a part can tolerate before it runs. That is how you design against fast fracture instead of hoping.
Then the rung pivots from the single catastrophic overload to the two slow killers. Guide 4 is fatigue: parts that fail after millions of cycles at a stress far below their static strength, because a crack initiates and creeps forward a striation at a time — the S-N curve plots how many cycles a stress survives, and for steels there is often an endurance limit below which it never fails. This matters more than any other item on the list, because most in-service failures — the majority of broken axles, wings, and shafts — are fatigue, not one big overload. Guide 5 is creep: at high temperature, metals slowly stretch under a constant load that would be perfectly safe when cold, and after enough time simply rupture — the reason a turbine blade has a service life measured in hours.
Hold on to the through-line as you climb: every one of these failures begins somewhere small — a flaw, a notch, a grain boundary, a single hot cycle too many — and grows until the part can no longer carry its load. Understanding fracture is really about respecting that small beginning: designing so the worst flaw stays below critical, keeping steel above its transition, and dividing every strength by an honest factor of safety. That respect is what keeps bridges standing and planes in the air.