Somewhere between solid and liquid, between order and chaos, antimony is having a moment. And by "moment," I mean a phase transition that researchers just figured out might explain why certain materials are so absurdly good at remembering things.
The Plot Thickens at -1.5 Gigapascals
Here's the setup: phase-change memory (PCM) is the technology that stores data by rapidly switching materials between glassy (amorphous) and crystalline states. Think of it like a material that can toggle between being window glass and a diamond - except it does this in nanoseconds, which is frankly rude to other materials that take their sweet time with structural changes.
Antimony sits at the heart of many PCM alloys, but nobody had actually studied what the element does when you supercool it - that awkward temperature zone where a liquid stubbornly refuses to crystallize. It's the scientific equivalent of that friend who won't commit to plans.
A team of researchers from Sapienza University of Rome and University of Milano-Bicocca decided to fix that knowledge gap. They ran massive molecular dynamics simulations - we're talking 4,096 atoms bouncing around in a computational box - using neural network potentials trained on quantum mechanical data. Basically, they taught a machine learning model to mimic the laws of physics, then let it simulate antimony for longer than any human would have the patience to calculate by hand.
Water, But Make It Metal
What they found was weird. Good weird.
Antimony exhibits water-like anomalies. Yes, the same quirky behavior that makes ice float on your drink apparently shows up in a metalloid that's mostly known for being in flame retardants and old-timey mascara. The liquid has a density maximum - a temperature where it's at its heaviest before becoming lighter again as it cools. Water does this at 4°C, which is why lakes freeze from the top down and fish survive winter. Antimony does it too, just at much higher temperatures and pressures where things get spicy.
The researchers describe this using a "two-state model" - essentially, the liquid can't decide what it wants to be. At high temperatures, antimony atoms mill around chaotically. Cool it down, and patches of ordered, crystal-like structures start appearing within the liquid, like ice cubes forming in slow motion. These A17-like local structures (named after the crystal phase they resemble) compete with the disordered regions, creating a liquid with an identity crisis.
Fragile? More Like "Dramatically Temperature-Sensitive"
The other big finding involves something called fragility, which in materials science doesn't mean the liquid shatters easily - it describes how sharply viscosity changes with temperature near the glass transition.
"Strong" liquids like silica (window glass material) get thicker gradually as they cool. "Fragile" liquids suddenly go from runny to immobile over a narrow temperature range, like honey that decides to become concrete between the refrigerator and the kitchen counter.
Antimony? Extremely fragile. The fragility index ranges from 74 to 330 depending on how you measure it, putting it well into "dramatically temperature-sensitive" territory. This is actually useful - it means antimony transitions from liquid to glass very suddenly, which helps explain why PCM materials can switch states so fast. There's no wishy-washy intermediate zone. The material commits.
Why This Matters for Your Future Laptop
Phase-change memory is already in some commercial products, but the technology has bigger ambitions. PCM is a leading candidate for neuromorphic computing - computers that work more like brains than calculators. Instead of shuttling data between separate memory and processor chips (the digital equivalent of writing everything down on paper before doing math), neuromorphic systems compute directly where data lives.
The catch? You need materials that can reliably switch between states millions of times without degrading, maintain their state without constant power, and switch fast enough to be practical. Understanding why antimony and its alloy cousins are so good at this - combining amorphous stability with ultrafast crystallization - brings researchers closer to designing even better materials.
The two-state liquid behavior might be the key. Those structural fluctuations between ordered and disordered regions could be priming the material for both states simultaneously, like keeping a foot in two doors. When crystallization finally happens, the material already has local blueprints of where atoms should go.
The Simulation Revolution
What makes this study possible is the rise of machine-learned potentials - neural networks trained to reproduce quantum mechanical calculations at a fraction of the computational cost. Traditional first-principles simulations max out at maybe a few hundred atoms for short timescales. These researchers simulated thousands of atoms for long enough to watch viscosity evolve, observe crystallization at weird pressures, and map out phase diagrams that would have been computationally impossible a decade ago.
They even discovered that antimony spontaneously crystallizes into a different structure (A17, the "black phosphorus" type) at negative pressures - a phase nobody had considered relevant until the simulations showed atoms doing it without being asked. When your simulation surprises you with new physics, that's when you know the methodology is working.
The connection between liquid-state anomalies and phase-change functionality isn't just academic curiosity. If the two-state model holds, engineers could potentially tune material compositions to enhance or suppress specific behaviors, optimizing for particular applications. Faster switching? Longer retention? Lower power consumption? The liquid's internal conflict might be the design knob nobody knew they had.
References
-
Giuliani, F., et al. (2026). Liquid anomalies and fragility of supercooled antimony. Proceedings of the National Academy of Sciences, 123(13). DOI: 10.1073/pnas.2531605123
-
Ding, K., et al. (2024). Phase-Change Memory for In-Memory Computing. Chemical Reviews. DOI: 10.1021/acs.chemrev.4c00670
-
Nilsson, A. & Pettersson, L. G. M. (2015). The structural origin of anomalous properties of liquid water. Nature Communications, 6, 8998. DOI: 10.1038/ncomms9998
-
Martin-Barrios, R., et al. (2024). An overview about neural networks potentials in molecular dynamics simulation. International Journal of Quantum Chemistry, 124(7), e27389. DOI: 10.1002/qua.27389
-
Ritarossi, M., et al. (2025). Phase‐Change Heterostructures Based on Antimony. Physica Status Solidi (RRL). DOI: 10.1002/pssr.202500012
Disclaimer: This blog post is a simplified summary of published research for educational purposes. The accompanying illustration is artistic and does not depict actual model architectures, data, or experimental results. Always refer to the original paper for technical details.