Glassy materials are everywhere around us, with applications far exceeding windowpanes and drinking glasses. They range from bioactive glasses for bone repair and amorphous pharmaceuticals that boost drug solubility to ultra-pure silica optics used in gravitational-wave detectors. In principle, any substance can become glass if its hot liquid is cooled fast enough to avoid forming an ordered crystal.
A distinguishing feature of a glass is that its atoms freeze into an irregular, disordered arrangement. This stands in contrast to crystals, where atoms sit in a regular pattern. This disorder gives glass many of its unique and useful properties, but scientists still struggle to understand how atomic-scale disorder produces the properties we observe in everyday glasses.
Many material properties are governed by how atoms vibrate when the temperature is not very high. In ordered crystalline materials, atomic vibrations take the form of waves that extend over very large distances. Glasses, however, do not support only wave-like vibrations. In recent years, it has become clear that glasses also host disorder-induced atomic vibrations that are localized in space, and that these vibrations significantly affect their thermal properties. Moreover, recent research indicates that when a glass is deformed strongly enough, the same localized vibrations give rise to glass flows--a phenomenon termed yielding. That is, large deformations force glasses to flow like a fluid, but in a way that is distinctly different from the yielding of crystalline solids.
Strikingly, repeated deformation can even "train" a glass to store mechanical memories--information about past forcing--which can later be read out from its response. Bringing these pieces together--vibrations, yielding, and memory--within one coherent theory has remained difficult. A recent model, published in Physical Review Letters, shows how such unification can be achieved.
In the study, Makoto Suda (a doctoral student at Tohoku University), Edan Lerner (University of Amsterdam), and Eran Bouchbinder (Weizmann Institute of Science) extended a theoretical model that had previously been shown to reproduce spatially localized vibrations in glasses. Earlier works suggested that when a glass is deformed, these vibrations align to form flow-like patterns, potentially giving rise to yielding and memory. To test this scenario, the team set out to analyze the model under controlled external forcing.
The researchers applied repeated back-and-forth forcing, known as cyclic loading, while varying the strength of each cycle. At small forcing amplitudes, the system quickly stopped accumulating new deformation and, after a few cycles, settled into a repeatable loop. In this stable regime, the model also developed a memory of the forcing strength, which could be extracted at a later time. At larger forcing amplitudes, the behavior changed qualitatively: the system continued accumulating new deformation patterns and never returned to previous states--a hallmark of yielding in glasses. In this regime, the memory of the forcing strength became increasingly obscured.
These results support the idea that "how glass shakes" and "how it starts to flow" are not separate mysteries, but rather two outcomes of the same underlying physics rooted in glasses' disordered nature. This connection is captured here by a single predictive model.
Looking ahead, the team aims to pinpoint the precise mechanism that sets the yielding threshold and to explore whether other hallmark properties of glasses can be traced within the same theoretical framework. By testing the model's predictions against atomistic simulations and experiments on real glasses, this approach could deepen scientists' understanding of the glassy state of matter and help establish more reliable routes for durability assessment and, ultimately, materials design.
These findings were published in Physical Review Letters on 18 December 2025.
- Publication Details:
Title: Yielding and memory in a driven mean-field model of glasses
Authors: Makoto Suda, Edan Lerner, Eran Bouchbinder
Journal: Physical Review Letters
DOI: 10.1103/vpmn-sw7n
