Most of us are familiar with the four classical states of matter – solid, liquid, gas and plasma – but there’s a whole world of exotic states out there. Now, physicists at Radboud and Uppsala Universities have identified a new one named “self-induced spin glass,” which could be used to build new artificial intelligence platforms.
Magnetism usually arises when the electrons in the atoms of a material all spin in the same direction. But in a spin glass, the atomic magnets have no order, all spinning in random directions. The “glass” part of the name comes from the similarities to how atoms are arranged amorphously in a piece of regular old glass.
So far spin glasses have only been found in certain alloys, but now, researchers have discovered that the state occurs naturally in the pure element neodymium. To differentiate it from the alloy version, they’ve called the new state self-induced spin glass.
Neodymium is already well known for having some strange magnetic properties, so the team examined it using scanning tunneling microscopy (STM). In doing so, they found that the atomic spins in neodymium whirl like a helix, and even weirder, they all do so at different speeds, which means the shape of that helix is constantly changing.
“(STM) allows us to see the structure of individual atoms, and we can resolve the north and south poles of the atoms,” says Daniel Wegner, an author of the study. “With this advancement in high-precision imaging, we were able to discover the behavior in neodymium, because we could resolve the incredibly small changes in the magnetic structure. That’s not an easy thing to do.”
The team says that this observation implies that neodymium might not be the only self-induced spin glass – the behavior could be hiding in other elements too. As for what it might be used for, the researchers suggest that the neuron-like behavior could help in future artificial intelligence systems.
“The complex evolution of neodymium may be a platform to mimic basic behavior used in artificial intelligence,” says Alexander Khajetoorians, an author of the study. “All the complex patterns which can be stored in this material can be linked to image recognition. You could never build a brain-inspired computer with simple magnets, but materials with this complex behavior could be suitable candidates.”
The research was published in the journal Science.
Source: Radboud University
Source of Article