Answers are grounded in this paper and cite their sources.
Most in-service failures of metal structural components occur as a result of metal fatigue from cyclic loading. As such, the ability to predict fatigue cracking and/or produce metals that resist these failure processes is of utmost importance. Unfortunately, the experimental data necessary for developing an understanding of the origins of fatigue crack processes is sorely lacking, especially at the nano- to atomic- scale. Recently, in situ high-cycle fatigue in the transmission electron microscope (TEM) discovered the phenomenon of autogenous fatigue crack healing in nanocrystalline Pt [1], attributed to grain-boundary migration. In the present work, we demonstrate that self-healing is not only possible in coarser grained metals – where dislocation plasticity should supplant grain-boundary-mediated mechanisms at lower stresses levels – but occurs nearly as frequently as crack progression itself. Coarse-grained pure gold films ∼100 nm thick with average grain sizes of 150-200 nm were grown via physical vapor deposition (PVD) on sacrificial polished NaCl (001) substrates. A ribbon of material was extracted and transferred to MEMS push-to-pull devices using a focused ion beam scanning electron microscope (FIB-SEM), where it was shaped into a single-edge notched tensile bar, after the manner detailed in ref [2]. In situ fatigue was performed using a Bruker Hysitron PI-95 picoindenter [3] cycling at 200 Hz in a Thermo Fisher Scientific Titan G2 Environmental TEM. By employing our recently developed experimental methodology for precisely controlling fatigue crack propagation over micrometer distances and millions of cycles, termed K*-control [2], we could obtain extremely low crack progression rates on the order of 10-12 m/cycle while maintaining a stable stress intensity at the crack tip. This level of control enabled the observation of a wide variety of fatigue damage processes; many never directly observed in TEM before, especially at the nanoscale. Two processes of particular note, discussed in greater depth herein, are the initial nucleation of fatigue cracks from the notch and fatigue crack self-healing. The case of the former may be the first time the entire series of events involved in the incubation of a fatigue crack has been observed in detail in the TEM. This turns out to be a complex process beginning with the formation of a stress-concentrating grain boundary cusp, followed by the formation of impinging extrusions and intrusions along slip planes. The competitive interplay between nucleation and plasticity-induced morphological changes can lead to the elimination of a nucleation site entirely before a stable crack could be formed. In practice, this resulted in the stochastic nucleation, evolution, and disappearance of multiple candidate nucleation sites before one eventually “wins out”. Once established, the fatigue crack meandered tortuously through the microstructure, probing different pathways in an organic-like manner. In contrast to the expectations of existing models, crack propagation did not proceed solely along single slip planes. Similarly, the extensive plasticity around the crack tip, even under these mild loading conditions, led to a generalized reshaping of the crack tip rather than a strongly crystallographic one. Most surprisingly, however, was the extent and frequency of crack healing. Far from an irreversible process, fatigue crack propagation proceeded in a back-and-forth manner (see Figures 1 and 2), with healing occurring on a scale comparable to the cracking itself. The grain-boundary migration mechanism proposed for the initial nanocrystalline study [1] would not be expected in this coarser-grained material, and indeed no evidence was found to support it. Observations instead indicate a plasticity-based mechanism wherein healing occurred in two ways: (1) extrusion-facilitated material transport to the crack tip region and (2) alteration of crack flank shape, promoting crack flank contact and subsequent cold-welding. Recent atomistic simulations [4] support the concept of material transport to the crack tip, and additional simulations are underway for the latter. The findings in this work provide illuminating insights into the manner in which fatigue damage occurs in metals at the nanoscale and strategies for its mitigation. While gold is not itself a structural metal, its fatigue behavior is highly relevant in microelectronics even beyond its value as a pedagogical study. Subsequent work on less-noble and microstructurally complex materials, such as commercial alloys, as well as under oxidizing environments (utilizing environmental TEM) are underway and will no doubt yield up even more surprises [5]. Still frames from an in situ fatigue video, wherein crack healing on the order of 100 nm was observed. After healing, the crack proceeded along a new path distinct from that prior to healing. Plot of fatigue crack growth and healing as a function of cycles for an example portion of an in situ fatigue test, showing how the two processes occurred on similar length and frequency scales. Each bar is 200 cycles.
No takes yet. Share an insight, caveat, or question.
House et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: