← Latest papers
🔬 materials science

Brute-force molecular dynamics simulation of aging in amorphous Ge2_2Sb2_2Te5_5

Using machine learning interatomic potentials in long-timescale molecular dynamics simulations, this study reveals that aging in amorphous Ge2_2Sb2_2Te5_5 involves enhanced Peierls distortion and reduced atomic mobility that widen the electronic band gap, while homopolar bonds persist on the simulated timescale.

Original author: Elisa Petroni, Dario Baratella, Omar Abou El Kheir, Andrea Redaelli, Marco Bernasconi

Published 2026-11-13
📖 4 min read☕ Coffee break read

Original author: Elisa Petroni, Dario Baratella, Omar Abou El Kheir, Andrea Redaelli, Marco Bernasconi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

Modern computers rely on a type of memory that can instantly switch between two states: one that conducts electricity easily and another that resists it. This technology, known as phase-change memory, uses a special alloy of germanium, antimony, and tellurium. When heated, this material can melt and then freeze into a disordered, glassy state that blocks current, or it can be heated just enough to rearrange into an orderly crystal that allows current to flow. This ability to toggle between states allows the device to store information. However, there is a persistent problem: the glassy state is not perfectly stable. Over time, it slowly changes its internal structure, a process called aging. As it ages, the material becomes more resistant to electricity, which can cause the memory device to fail or lose its stored data. Understanding exactly how this material changes inside the glass is crucial for building more reliable computers, but these changes happen on such a tiny scale that they are nearly impossible to watch in a real laboratory.

To solve this puzzle, a team of researchers turned to a powerful form of digital experimentation. They used a computer model that mimics the behavior of atoms with extreme precision, powered by a machine-learning system trained on the laws of quantum physics. They created a digital box containing thousands of atoms of the alloy and watched them move and interact over a period of about one hundred nanoseconds. While this sounds like a blink of an eye, in the world of atomic motion, it is a long time—enough to see the material settle down and change its shape. The simulation was run at a temperature just below the point where the material would normally turn into a liquid, allowing the researchers to observe the slow, subtle shifts that occur as the glassy structure tries to find a more stable form.

The researchers discovered that as the material aged, the atoms did not simply sit still. Instead, a small group of atoms became much more active than the rest, moving significantly while their neighbors remained relatively calm. This uneven movement caused the bonds holding the atoms together to break and reform repeatedly. Over time, these bonds became much more stubborn, lasting longer before breaking, which made the entire network of atoms more rigid and less mobile. This slowing down of atomic movement is a key sign that the material is maturing into a more stable state. However, the researchers found that the most obvious structural defects—atoms bonded to the wrong neighbors or arranged in unusual shapes—did not disappear during this period. These "wrong" bonds remained largely unchanged, suggesting that the complete removal of such defects requires a much longer timeline than the one they observed.

What did change, and what the researchers found to be the most significant effect, was a subtle stretching and compressing of the bonds between specific pairs of atoms. In the disordered glass, some atoms are connected in a way that creates a slight imbalance, with one bond becoming longer and the other shorter. As the material aged, this imbalance grew stronger. The researchers linked this specific structural change to a widening of the energy gap that electrons must cross to move through the material. A wider gap means it is harder for electricity to flow, which perfectly explains why the material becomes more resistant over time. This finding suggests that the aging process is driven first by these bond adjustments that tighten the electronic structure, rather than by the immediate elimination of all structural mistakes.

The study also revealed that the material's internal landscape is filled with trapped electronic states that act like potholes for electricity. As the material aged, the Urbach tail of localized states near the edge of the energy gap increased, even as the gap itself widened. This indicates a complex sequence of events where the material first rearranges its bonds to widen the gap, but leaves behind a trail of localized electronic traps that may only be cleared out over a much longer period. The researchers concluded that the aging of this memory material is a multi-stage process. It begins with a rapid relaxation of internal stress, followed by a slower strengthening of the atomic network and a widening of the energy gap. The final step, which would involve the total removal of all structural imperfections, likely takes place over a timescale far beyond what current simulations can easily capture. By mapping out these early stages, the study provides a clearer picture of why these memory devices drift over time and points toward the specific atomic movements that engineers must manage to create more stable, long-lasting storage.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →