Practice with mock tests, learn from structured notes, and get your questions answered by a global forensic community, all in one place.
How to tell ductile from brittle fracture using macroscopic and microscopic features of the fracture surface, with the Alexander Kielland platform collapse as the central case study.
Last updated:
Pick up two broken pieces of metal and look at the fracture faces. One surface is grey and rough, like torn bread, with a dimpled texture visible under a hand lens. The other is flat, slightly shiny, and covered in fine radiating lines. These two appearances are not cosmetic. They record completely different physical processes, one component absorbed energy by deforming plastically before it broke, and the other cleaved suddenly with almost no plastic work at all. Reading these fracture surface features correctly is the core skill of fractography, and it is often the first thing a failure analyst does with a broken component.
The distinction between ductile and brittle fracture matters to courts and regulators for a direct reason. Ductile fracture requires large plastic deformation, which means the component was loaded well beyond its yield stress before it broke. Brittle fracture can occur at stresses well below yield, which means a small pre-existing crack or a material deficiency could be responsible. The same overload looks very different under those two hypotheses, and the liability picture shifts accordingly.
This topic covers the macroscopic and microscopic signatures of both failure modes, the scanning electron microscope workflow that takes the analysis from visual examination to fractographic confirmation, the intermediate cases of intergranular fracture and mixed-mode failure, and the Alexander Kielland platform collapse of 1980, which remains the most thoroughly documented brittle fracture failure in offshore engineering history.
Metal does not simply pull apart; it tears, one microvoid at a time.
When a ductile metal is loaded to failure, the fracture surface reveals the history of a process that began at a microscopic scale. The first stage is void nucleation. Voids form at second-phase particles or inclusions, either by decohesion of the particle-matrix interface or by fracture of the particle itself. The density and spacing of inclusions in the microstructure controls when this starts.
Once voids have nucleated, they grow as the surrounding matrix stretches and thins under continued loading. When neighbouring voids grow close enough, the ligament between them becomes unstable and ruptures, linking two voids into one. This void coalescence continues across the cross-section until the part separates. The resulting fracture surface, examined under SEM, is covered with bowl-shaped depressions called dimples. Each dimple is the imprint of one half of a void; the matching half is on the opposing fracture face. The dimple floor often contains a small particle: the original nucleation site.
Dimple morphology gives additional information. Equiaxed dimples form under pure tensile loading, where voids open symmetrically. Elongated or parabolic dimples form under shear loading, where one face drags past the other. Tear dimples are elongated in the opposite direction on opposing faces. Recognising these patterns lets an analyst determine whether the load was tensile, shear, or a combination, which speaks directly to what force caused the failure.
Grains fracture along crystallographic planes, and the cracks leave a map of their path.
Brittle fracture by cleavage is the opposite of the void-growth picture. The crack moves through the grain along a specific low-index crystallographic plane, almost without any plastic work. In iron and low-carbon steel the preferred cleavage plane is {100}. The crack front moves so fast that by the time you see the fracture it has crossed the component in a fraction of a second.
At the grain scale the fracture looks flat and glassy. But grains are not all oriented the same way, and adjacent grains have their {100} planes tilted slightly relative to each other. When the crack front jumps from one grain to the next, it has to deflect slightly to stay on the cleavage plane in the new grain. This deflection creates a step on the fracture surface. Many of these steps join up to form river marks, lines that flow across the grain surface converging in the upstream direction, toward the crack origin. Following river marks back to their convergence point takes the examiner to the fracture origin, exactly as following rivers upstream leads to a watershed.
At the macroscopic scale, cleavage fractures in thick sections often show chevron marks, broad V-shaped or herringbone patterns visible to the naked eye. These also point toward the fracture origin. The combination of macroscopic chevron marks to find the approximate origin location and microscopic river marks to locate it precisely is the standard fractographic workflow for brittle fracture investigation.
When the grain boundary is the weakest path, the crack takes it.
In normal circumstances grain boundaries in metals are stronger than the grain interiors, so cracks go through grains (transgranular fracture) rather than around them. Intergranular fracture happens when this relationship reverses, when the boundary has been weakened to the point where it becomes the preferred crack path. This inversion is a flag that something has gone wrong with the material or its environment.
Under SEM, intergranular fracture has a faceted, rock-candy appearance. The grain surfaces are relatively smooth, reflecting the rounded shape of the grains, and adjacent facets meet at angles reflecting the grain boundary geometry. Energy-dispersive X-ray spectroscopy (EDS) can detect segregated embrittling elements on the grain faces, providing direct chemical evidence of the embrittlement mechanism.
From naked-eye examination to electron microscopy, step by step.
The scanning electron microscope resolves fracture surface features at scales impossible with optical microscopy. Its large depth of field means complex topography stays in focus across the image, and its back-scattered and secondary-electron imaging modes reveal both topographic and compositional contrast. But SEM fractography only pays off if the sample preparation is done correctly.
An undocumented weld attachment, a fatigue crack, and 123 deaths in the North Sea.
At 18:30 on 27 March 1980, the semi-submersible drilling rig Alexander Kielland capsized in the Ekofisk field, North Sea, in a force 6 gale. Of the 212 people on board, 123 died. The Norwegian government commission that investigated the accident produced one of the most detailed fracture analyses in offshore engineering history.
The rig consisted of five pontoons supporting five buoyancy columns, connected by bracings. Bracing D-6 connected column D to the central platform. An external hydrophone had been attached to D-6 by a fillet weld through a hole drilled in the bracing tube wall. This attachment was not shown on the original design drawings. The weld created a severe stress concentration at the edge of the hole.
Fatigue cracks initiated at the weld toe, driven by the cyclic wave-loading the bracing experienced during every sea state. The cracks grew over what the commission estimated as approximately two years of service. When they reached a critical size, D-6 fractured. The fracture surface showed flat brittle cleavage over the pre-existing fatigue crack region, confirming that final failure was by brittle fracture once the remaining ligament was too small to sustain the applied stress. The sudden loss of D-6 overloaded the adjacent bracings, which failed in rapid succession within seconds, and the platform capsized.
Most real failures sit between the clean textbook extremes.
The ductile-brittle distinction is cleaner in a textbook than in a failure laboratory. Real fractures frequently show mixed zones: a central flat cleavage region surrounded by ductile shear lips, or a fracture surface that transitions from dimple fracture in warm service to cleavage at a stress concentration where local temperatures are lower. Recognising mixed-mode fracture is important because the blend of features carries its own diagnostic information.
| Feature | Ductile fracture | Brittle fracture |
|---|---|---|
| Macroscopic appearance | Grey, rough, fibrous, with necking | Flat, shiny or granular, no necking |
| Fracture surface angle | 45-degree shear lips at free surfaces | Perpendicular to tensile stress axis |
| Microscopic mechanism | Void nucleation, growth, coalescence | Cleavage along crystallographic planes or grain boundaries |
| SEM features | Equiaxed or elongated dimples | River marks, cleavage facets, fan patterns |
| Energy absorbed | High (large area under stress-strain curve) | Low (little plastic deformation) |
| Temperature sensitivity | Less sensitive above DBTT | Increases sharply below DBTT in BCC metals |
One category deserves special mention: quasi-brittle fracture in high-strength alloys. An aerospace aluminium alloy or a high-strength steel fastener can fracture with a flat surface and little visible deformation, and it can look brittle macroscopically. But SEM reveals dimples, just very small ones, because the material did deform plastically, just over a much smaller scale than mild steel would. Calling this fracture brittle without the SEM evidence leads to an incorrect failure analysis. The implication is that quasi-brittle fracture is a high-energy process, similar to overload rather than embrittlement, which changes how liability is assessed.
Under SEM, a fracture surface is covered with equiaxed bowl-shaped depressions, each with a small particle at the bottom. Which fracture mechanism does this confirm?
Test yourself on Forensic Engineering with free, timed mocks.
Practice Forensic Engineering questionsSpotted an error in this page? Report a correction or read our editorial standards.