“It seems natural that materials with nearly identical atomic arrangements should behave in similar ways.” In the world of glasses, however, this intuition often fails. What causes this mismatch between structure and dynamics has long remained unclear.
Now, the researchers from China and Denmark report a solution. Writing in National Science Review , they show that in certain glassy systems, structure–relaxation coupling is governed not simply by geometric similarity, but by how strongly — and how uniformly — atoms are bonded. The researchers argue that heterogeneity in chemical bonding and electronic interactions plays a decisive role.
A long-standing question in glass physics
Glasses, or amorphous materials, are central to condensed matter physics and materials science. They are widely used in structural materials, electronic devices, and functional components. Unlike crystals, which possess long-range atomic order, glasses are structurally disordered at the atomic scale. Their properties are highly sensitive to processing history and external conditions, and they exhibit complex relaxation dynamics spanning wide time and length scales.
For decades, researchers have explored the structure–dynamics relationship in glasses mainly through geometric structural descriptors . These include free-volume models, locally favored structures such as five-fold symmetry and icosahedral short-range order, atomic configuration matching, SOAP descriptors, and inherent-structure displacement analyses. While these approaches capture statistical features of atomic packing, they largely focus on where atoms sit rather than how atoms interact .
Increasingly, scientists have realized that geometric similarity alone often cannot explain why different glasses display dramatically different dynamic behaviors. This challenge is particularly evident in multicomponent metallic glasses, where structurally similar systems can show strikingly different relaxation characteristics.
Two similar glasses, two very different behaviors
To tackle this puzzle, the researchers carried out a comparative study on two classic Pd-based metallic glasses: Pd 40 Cu 40 P 20 and Pd 40 Ni 40 P 20 . The two materials are highly similar in atomic size, geometric packing, and conventional structural metrics.
Yet experiments reveal markedly different behaviors. Pd 40 Ni 40 P 20 more readily forms thermally stable glasses, while Pd 40 Cu 40 P 20 exhibits an unusually strong β relaxation, a secondary dynamic relaxation process. This phenomenon, nearly identical geometry but sharply different dynamics, has been difficult to reconcile using traditional geometric theories.
Looking beyond geometry — into bonding
To overcome this limitation, the team trained high-accuracy deep-learning interatomic potentials on large datasets generated by density functional theory. This approach enabled molecular dynamics simulations of large systems over long time scales with near-quantum accuracy, and allowed them to reproduce the experimental dynamic process in simulations.
Using bond-order analysis, the researchers quantified the strength of electronic interactions between atoms. The simulations showed that replacing Ni with Cu hardly changes the overall geometric structure, but it significantly alters the strength and spatial distribution of chemical bonds.
Cu–P bonds were found to contain a higher fraction of weaker bond components, leading to a more heterogeneous bonding network. In contrast, Ni–P bonds are generally stronger and more uniformly distributed. This bonding heterogeneity, rooted in electronic structure, directly regulates atomic mobility and cooperative rearrangement patterns.
As a result, Pd 40 Cu 40 P 20 more readily develops string-like cooperative motions, enhancing β relaxation. Meanwhile, the more robust bonding network in Pd 40 Ni 40 P 20 suppresses such localized rearrangements, producing markedly different macroscopic relaxation behavior.
A new degree of freedom for glass dynamics
The study demonstrates that bonding heterogeneity is a fundamental degree of freedom governing glassy dynamics. By moving beyond geometry-centered descriptions and incorporating electronic structure and chemical interactions, the work establishes a new physical framework for understanding structure–relaxation coupling in glasses.
These findings deepen insight into relaxation mechanisms in metallic glasses and provide a theoretical basis for tuning glass-forming ability and designing advanced amorphous materials.
In short, when it comes to glass behavior, it is not only about where atoms sit — but how they hold together.
National Science Review
Computational simulation/modeling