ɫɫÀ²

News

Physicists use machine learning to find out how layered gases and metals melt

The research team, which included a Nobel Prize winner, found out how layered materials confined in two dimensions transition between states.
Two superimposed grid layers arranged on a green background, with one grid appearing transparent and complex.
Materials pressured between two graphene layers melt in complex ways. Illustration: Ville Heirola/Aalto University.

In physics, a phase transition is a transformation of a substance from one form to another. They happen everywhere from beneath the earth’s crust to the cores of distant stars, but the classic example is water transitioning from liquid to gas by boiling. Things get much more complex when physicists zoom in to the miniscule quantum realm or work with exotic matter. Understanding phase transitions rewards both increased knowledge of fundamental physics and future technological applications. 

Now researchers have found out how thin layers of noble gases like helium and metals like aluminium melt in confined spaces by topological excitations. In the study, the layers were confined between two graphene sheets at high pressures. The team consisted of Aalto University’s Department of Applied Physics Prof. Tapio Ala-Nissilä, famed Prof. Roberto Car from Princeton University and physicists from Nanjing University in China. The team also included the 2016 Nobel Prize in Physics winner, Prof. Michael Kosterlitz from Brown University, who landed the prize with his work in the similar vein of topological phase transitions.

The paper was published recently in PNAS:

Researchers focused on the melting of relatively simple layered materials that, when heated up and put under the pressure of two graphene layers, melt in ways that do not adhere to prior ideas of how they should transition via melting. 

‘We found that how the materials melt depends heavily on how many layers of them there are between the graphene sheets. A two-step process known as the Kosterlitz-Thouless-Halperin-Nelson-Young (KTHNY) theory predicts how a single layer melts. We found that to ring true but also discovered that additional layers surprisingly arrange themselves in a more complex manner than expected and melt in unexpected ways. We were able to identify the reason for the latter due to changes in the topological excitations that drive melting in layered systems,' says Ala-Nissilä, who is also a Professor at Loughborough University in the UK.

The team found a way to cut down on the computationally heavy process of identifying workable materials. They used machine learning methods that probed down to the molecular level while also retaining quantum-mechanical accuracy of the various interactions. 

‘As the number of material layers increases, the modelling calculations become too demanding  and complex. Our method allowed us to run the detailed calculations up to an unprecedented 12 material layers.’

Ala-Nissilä says that the team are going to keep pushing the envelope for exploring phase transitions.

“Theoretically we know that an infinite number of material layers results in a completely different melting that does not feature any of the hallmarks of the KTHNY theory. Though computationally unfeasible, it would be interesting to study whether that can happen with a finite number. Another avenue to pursue is to make the melting to occur abruptly even for a single layer and see what that would do to the entire process,’ Ala-Nissilä envisions.

While Ala-Nissilä submits that the new results  could be useful for future devices, the main thrust of the effort is fundamental.

‘Back when the theories for topological matter were first formed decades ago, nobody knew whether they were useful at all. But now we see that topological matter might be what our quantum computation is based on in the future. That shows just how vital it is to carry out fundamental basic research even when there’s no immediate practical application,’ Ala-Nissilä says.

Read more:

Fun problem solving rewarded with the Nobel Prize in physics (2016)

  • Updated:
  • Published:
Share
URL copied!

Read more news

An illustrative figure comparing disease-induced immunity (left) and randomly distributed immunity (right) in the same network. Illustration: Jari Saramäki's research group, Aalto UIniversity.
Research & Art Published:

Herd immunity may not work how we think

A new study from researchers at Aalto University suggests that our picture of herd immunity may be incomplete — and that understanding how people are connected could be just as important as knowing how many are immune.
AI applications
Research & Art Published:

Aalto computer scientists in ICML 2025

Department of Computer Science papers accepted to International Conference on Machine Learning (ICML)
Close-up of a glowing dual processor on a dark motherboard with futuristic light effects and detailed circuitry.
Press releases, Research & Art Published:

New quantum record: Transmon qubit coherence reaches millisecond threshold

The result foreshadows a leap in computational capabilities, with researchers now inviting experts around the globe to reproduce the groundbreaking measurement.
Aerial view of a coastal city with numerous buildings, a marina, and boats docked. Trees and water surround the city.
Press releases, Research & Art Published:

Study: 70% of emissions from new buildings come from construction – and this is often overlooked

While energy efficiency and the use of renewable energy have reduced the life cycle emissions of new buildings, emissions from construction have not decreased. Preserving green areas and prioritizing timber construction would make construction more sustainable, researchers emphasize.