The Hole That
Became an Exit

A metal component can survive a load once and still be doomed by its repetition. Each rotation of an axle, vibration of a turbine and bend of a wire adds another cycle. Plastic deformation becomes concentrated, a microscopic crack appears, and the crack advances until the remaining section can no longer carry the load. The final break may be sudden. Its history is not.

That history is why designers dislike holes. Remove material and the load must flow around an edge, raising local stress. A machining mark, notch or pore can become the place where a fatigue crack starts. The instinct to smooth, reinforce or eliminate such features has been built into mechanical design for generations.

The new work, published online Aug. 23 in Advanced Materials, does not repeal stress concentration. It adds a competing effect that becomes unusually powerful when every point in the metal is only a few hundred nanometers from a free surface. At that scale, mobile crystal defects can leave the material before they organize into the structures that localize cyclic strain. The hole remains a geometric stress raiser; its surface also becomes an escape boundary.

The project was led by Professor Takashi Sumigawa and doctoral student Kota Sugisaka of Kyoto University’s Graduate School of Energy Science. The author team includes Yamato Ishizaka, Hiroki Ikeda and Masataka Abe of Kyoto University; Shigeo Arai of Nagoya University; Professor Yoshitaka Umeno of the University of Tokyo’s Institute of Industrial Science; and Professor Hiroyuki Shima of the University of Yamanashi. The universities and the Japan Science and Technology Agency describe the nanoscale tension–compression fatigue test as a world first.

First, make a 300-nanometer crystal tired

The experiment had two linked parts. The first asked a basic question: what does cyclic loading do to a single crystal so small that its dislocations are never far from a surface?

The researchers began with 99.999% pure nickel single crystal. After polishing, annealing and electropolishing it, they used a focused ion beam to cut hourglass-shaped specimens. The narrow gauge was approximately 300 nanometers square and one micrometer long, oriented to favor a single slip system. Against a human hair roughly 50 to 100 micrometers wide, the gauge width was about 1/170 to 1/330 as large.

Gripping such a specimen without damaging it is itself difficult. The team developed an in-situ system that alternated tension and compression while a field-emission scanning electron microscope tracked motion. Because the nanoscale setup did not provide an ordinary direct stress reading, the researchers reconstructed geometry from images and used finite-element analysis to infer stress. The paper treats those nanospecimen stress values as estimates, an important uncertainty beneath the headline numbers.

At lower strain amplitudes, fine slip lines appeared by 100 cycles but did not develop visibly, and no crack was seen through 10,000 cycles. The team estimated an average resolved shear-stress amplitude of about 800 megapascals at the neck. Persistent slip bands and crack initiation in bulk or micrometer-scale nickel single crystals typically appear around 40 to 50 MPa over 1,000 to 100,000 cycles. The nanoscale specimen therefore tolerated an order of magnitude more cyclic shear stress under those particular conditions. At higher strain, it cracked.

Dislocations: the defects that let metal bend

A crystal is not a flawless stack of atoms. A dislocation is a line defect where the regular arrangement is offset. When dislocations move, atomic planes can slip incrementally; that motion lets a metal deform plastically instead of requiring an entire plane of bonds to break at once. The defect is not simply damage. It is also the carrier of metal’s useful ductility.

Cyclic loading changes the population. Dislocations travel, multiply, tangle and interact. In a single crystal, where there are no grain boundaries to impede their collective motion, they can self-organize into walls and ladder-like persistent slip bands. Slip concentrates there. Intrusions and extrusions form at the surface, and a fatigue crack can nucleate.

Nanoscale dimensions change the balance. A tiny volume contains few pre-existing dislocations. A moving dislocation has only a short distance to travel before it reaches a free surface and escapes. Continued cycling can leave the gauge nearly empty of dislocations—a state called dislocation starvation.

The team watched this happen in a one-million-volt ultra-high-voltage transmission electron microscope. At a lower loading condition, dislocation density fell in the specimen’s center, a nearly dislocation-free region emerged after 100 cycles and expanded through 10,000. Dislocations accumulated and tangled near the constrained ends instead. Under heavier loading, a crack appeared at an end. Those observations make the mechanism more persuasive than an after-the-fact claim that a very small specimen happened to begin with fewer flaws.

The architecture does not make a hole harmless. It makes surface so abundant—and so close—that dislocations struggle to preserve a memory of repeated deformation.

Scaling a size effect into an architecture

A single nanoscale bridge can be strong without being useful. The second part of the study therefore tried to distribute the beneficial size effect through a larger, micrometer-scale specimen.

The team patterned elliptical holes into nickel single crystal. Each ellipse had a major axis of one micrometer and a minor axis of 300 nanometers. The holes repeated every 400 nanometers along the loading direction, with about 350 nanometers between neighboring holes across the other directions. Narrow, hourglass-like ligaments remained. The purpose was geometric: keep all load-bearing nickel within several hundred nanometers of a free surface, so dislocations throughout the structure would have an exit.

The resulting metamaterial had a relative density of roughly 50%. That is a ratio of solid volume to the fully dense material occupying the same envelope. It does not mean a usable component delivered equal stiffness, load capacity and safety at half the weight. The specimens were microscopic structures cut serially by a focused ion beam.

Researchers loaded the architected specimens and solid references in fully reversed tension and compression at one hertz. Unlike the initial nanoscale tests, these larger specimens used a load cell. Nominal stress was calculated over the total envelope cross-section, including the vacant area—a stringent and practically relevant convention for comparing load capacity per overall size.

In a representative comparison, the solid reference saw a stress amplitude of 70.2 ± 0.2 MPa; the nano-architected specimen saw 124.3 ± 0.1 MPa, 1.8 times higher. The solid specimen developed surface undulations within 100 cycles, roughly 300-nanometer intrusions and extrusions by 1,000, and multiple cracks by 10,000. The perforated specimen did not show those features at the corresponding endpoint despite the higher nominal stress.

Across the tested specimens, the fatigue-crack-initiation threshold was approximately twice that of the homogeneous reference, whose value was close to the roughly 50 MPa familiar from bulk nickel single crystal. At about twice that baseline, no distinct crack was observed through 100,000 cycles in the reported runout data. Postmortem electron microscopy found dislocation walls and ladder structures in the solid reference but extremely low dislocation density in the architected gauge. Walls remained outside the gauge, where the geometry no longer enforced nanoscale proximity to surface.

From railway axles to designed internal boundaries

Systematic fatigue science emerged with the 19th-century railway. German railway engineer August Wöhler investigated why axles could fail under loads far below their static breaking strength. His repeated-load experiments connected stress amplitude with cycles to failure. The S–N curve—also called the Wöhler curve—remains a foundation of fatigue testing.

Engineering responses accumulated over the next century and a half: reduce stress concentrations; improve surface finish; introduce compressive residual stress; select alloys and heat treatments; use precipitates, solute atoms and fine grains to impede dislocations; inspect for cracks; and design with damage tolerance. When holes are unavoidable, their contours, finish and surrounding reinforcement are controlled because geometry directs stress.

Metamaterials add a different intellectual tradition. The term became prominent around 2000 for subwavelength structures that manipulated electromagnetic waves in ways unavailable from an ordinary bulk substance. Japan’s National Institute of Advanced Industrial Science and Technology summarizes the principle as designing function by designing structure. The concept has since spread to acoustics, vibration and mechanics.

This nickel metamaterial is not a new alloy. Its chemistry is deliberately plain. The designed function comes from arranging internal free surfaces at the distance scale where dislocation behavior changes. A hole that would be classified as a flaw at the macroscale becomes part of the mechanism at the nanoscale—not because stress concentration vanished, but because dislocation escape overcame it in the tested regime.

Why “twice the fatigue resistance” needs a denominator

The most important editorial qualification is the endpoint. The paper reports resistance to fatigue-crack initiation under specific fully reversed loading. That is not a blanket twofold improvement in fatigue life. A structure’s service life also depends on how a crack propagates after it starts, how rapidly ligaments fail, whether the architecture buckles, and how stiffness, impact, wear and corrosion interact.

The tests reached mostly up to 100,000 cycles at one cycle per second. Many turbines, vehicles and electronic interconnects encounter millions or billions of cycles, variable amplitudes, pauses, thermal gradients and environmental attack. A room-temperature, high-purity nickel microstructure cannot stand in for a hot single-crystal superalloy turbine blade merely because both lack grain boundaries.

Single-crystal nickel is also not ordinary engineering metal. Commercial alloys contain grain boundaries, precipitates, inclusions, residual stress and manufacturing texture. Grain boundaries can block dislocations—the very population the architecture is trying to expel—and can introduce other fatigue mechanisms. The authors note that dislocation starvation has been observed in several face-centered-cubic and body-centered-cubic metals and alloys, suggesting broader potential. They did not demonstrate it here.

Nor is 300 nanometers an established optimum. The paper explicitly warns against that interpretation. Hole size, ligament width, orientation and spacing must be mapped systematically, including the point where stress concentration wins, where surface damage dominates, or where manufacturing variability leaves a dangerously thin ligament.

A design principle with a manufacturing problem

The most immediate applications may be those already near the demonstrated scale: metallic interconnects, microsystems and small devices where free surfaces dominate and a limited volume must survive repeated strain. The team’s public release also points toward transport and energy equipment, but explicitly lists polycrystalline alloys, scaling, manufacturing, long-duration fatigue and reliability as work still required.

Focused ion beams are excellent for fabricating and editing research specimens. They are serial, slow and can alter the near-surface material. A practical route might eventually involve lithography, templating, electrodeposition or advanced additive manufacturing, but the paper does not demonstrate a production process. A flawless square millimeter is not a flawless square meter; every missed hole, rough edge and joined boundary can create the defect the architecture was meant to control.

The result nevertheless changes the question. Conventional strengthening often puts more obstacles in a dislocation’s path. This architecture gives the dislocation less path before it exits. Instead of trapping the carriers of plastic history until they form a persistent fatigue pattern, it uses surface to empty the gauge repeatedly.

Sumigawa and Sugisaka said in the official university statement that two long-standing assumptions—material strength does not depend on component size, and stress-concentrating shapes should be avoided—do not necessarily hold once matter is reduced to the nanoscale. The careful version is not that the assumptions became false. It is that a size-dependent mechanism became strong enough to reverse their outcome under a narrow, observed set of conditions.

Fatigue science has spent more than 160 years counting how repeated load becomes remembered damage. This study proposes an elegant countermeasure: shorten the distance to an exit so that some of that microscopic memory cannot remain. Whether industry can manufacture that exit everywhere a real component needs it is now the harder—and more consequential—experiment.