New study shows that catastrophic fractures begin not with an explosive crack, but with tiny two-dimensional patches that creep almost imperceptibly before suddenly racing through a material. The findings offer a new picture of material failure, with implications ranging from detecting fractures to understanding how earthquakes begin.

We have all seen something suddenly break: a phone screen cracks, a plastic object snaps, or a piece of glass shatters. To our eyes, the failure seems to happen all at once.

But what if the most important part of the break happens long before the final snap?

A new study presents a new physical picture of how materials fail. The researchers found that cracks can begin as tiny, two-dimensional patches that grow extraordinarily slowly, at speeds of microns to millimeters per second. Only after these patches grow to span the thickness of the material do they transform into the rapidly moving cracks associated with sudden, explosive fracture.

The research was conducted by Yuval Paz and Prof. Jay Fineberg of the Racah Institute of Physics at The Hebrew University of Jerusalem, together with Meng Wang of the Beijing Institute of Technology and Mokhtar Adda-Bedia of CNRS, ENS de Lyon and Université de Lyon.

For decades, scientists have used Linear Elastic Fracture Mechanics, or LEFM, to understand why materials break. Classical fracture theory says that a crack smaller than a certain critical size, known as the Griffith length, remains stable.
Once it grows beyond that size, it becomes unstable and rapidly accelerates, eventually reaching speeds approaching the characteristic speed of sound in the material.

The theory is remarkably successful at describing cracks once they are racing through a material. But it leaves a fundamental question unanswered: How does the crack get started?

The team set out to investigate this missing beginning of the fracture story.

Their experiments revealed that, in three-dimensional materials, fracture can begin when the material reaches a critical stress. A tiny broken region nucleates, not as the familiar line-shaped crack, but as a two-dimensional patch.

That patch expands extremely slowly in a process known as creep. Using a specially designed experimental system and high-speed imaging, the researchers followed these fractured regions from their first appearance through slow creep and, finally, rapid rupture.

The experiments revealed a remarkably consistent sequence: a tiny two-dimensional crack patch appears, it creeps, and when it approaches the thickness of the material, everything changes.

The key is geometry.

As the patch grows to span the thickness of the plate, it undergoes a geometrical (or ‘topological’) transition, effectively changing from a two-dimensional patch into the one-dimensional, through-going crack described by classical fracture mechanics. Only then does explosive acceleration begin.

The slow stage is far from insignificant. In the experiments, creep could occupy at least 75% of the total fracture process. Once the transition occurred, however, the final rupture could sweep through the material on microsecond-to-millisecond timescales.

The new theory therefore does not replace classical fracture mechanics, it extends it. By accounting for the interplay between the two-dimensional geometry of an emerging crack and the thickness of a three-dimensional material, the framework connects three stages that previously lacked a single mechanical description:
crack initiation → extremely slow creep → explosive fracture.

Once the patch becomes a through-crack, the framework naturally returns to the rapid fracture behavior predicted by classical LEFM.

The experiments also challenge the idea that a crack must reach a particular length before failure can begin. Instead, they support a critical stress for crack initiation that depends on geometry. In the experiments, the classical Griffith length was about 1 millimeter, while cracks were observed nucleating at scales of roughly 0.1 millimeters, about ten times smaller.

Understanding this hidden, slow stage could eventually help researchers think differently about detecting material failure before the final break. Instead of looking only for cracks that have already become dangerously large, it may also be important to understand the tiny, slowly expanding patches that precede them.

The findings also point to a much larger phenomenon: earthquakes. The new framework builds on related research into frictional ruptures, the processes that occur when two surfaces begin sliding against one another. The researchers show that tensile fracture and frictional rupture can reflect the same underlying geometrical mechanism.

This suggests that the physics governing a slowly growing crack in a laboratory material may also offer insight into how ruptures initiate and evolve along frictional interfaces, including those involved in earthquakes.

A catastrophic fracture may appear instantaneous. But beneath that explosive event can lie a long, nearly invisible preparation: a tiny patch forms, creeps slowly outward and changes geometry until it spans the material's thickness.
Then the crack takes off.

By connecting these stages within one theory, the researchers provide a new mechanical picture of failure, from its almost imperceptible beginnings to the explosive fracture we see in everyday life.