Argonne Researchers: Tiny Magnetic Vortexes Could Be Game Changer for HPC Memory Technology – High Performance Computing News Analysis

Magnetic fields created by skyrmion in two-dimensional sheets of material composed of iron, germanium and tellurium. (credit: Argonne Lab)

Scientists from the United States Department of Energy (DO) Argonne National Laboratory is investigating the possibility that tiny magnetic vortices called “skyrmions,” which are magnetic vortices as tiny as billionths of a meter, could transform memory in future high-performance computers.

Skyrmions, say the Argonne researchers, display characteristics that could overcome the shortcomings of microscopic bar magnets, whose magnetic fields can each store one bit of memory as a zero or a one — the language of computers.

Why Skyrmions? We estimate that skyrmion’s energy efficiency could be 100 to 1000 times better than current memory in high-performance computers used in research,” said Arthur McCray, a Northwestern University graduate student who works in the materials science division of Argonne (msd extension).

Energy efficiency is essential for the next generation of microelectronics. Today’s microelectronics already account for a substantial fraction of world energy consumption and could consume almost 25% of it within the decade. More energy efficient electronics need to be found.

We still have a long way to go before skyrmions find their way into any future low-power computer memory,” said Charudatta Phatak, materials scientist and group leader in the msd extension.​However, this kind of radically new way of thinking about microelectronics is the key to next-generation devices.”

Skyrmions, unlike bar magnets, have potential for computers because, in McCray’s words, “The bar magnets in computer memory are like shoelaces tied in a single knot; it takes almost no energy to undo them. This means that any magnet bars that malfunction due to one outage will affect the others.

In contrast, skyrmions are like shoelaces tied in a double knot,” McCray said. “No matter how far you pull a strand, the shoelaces stay tied.” Skyrmions are therefore extremely stable to any break.

Another important feature is that scientists can control their behavior by changing the temperature or by applying an electric current.

Scientists have a lot to learn about the behavior of skyrmions under different conditions. To study them, the team led by Argonne developed an artificial intelligence (TO THE) program that works with a high-power electron microscope at the Center for Nanoscale Materials (CM extension), a DO User facility of the Office of Science in Argonne. The microscope can view skyrmions in samples at very low temperatures.

The team’s magnetic material is a mixture of iron, germanium and tellurium. In structure, this material is like a stack of paper with many sheets. A stack of such sheets contains many skyrmions, and a single sheet can be detached from the top and parsed into structures such as CM extension.

The CM extension electron microscope coupled with a form of TO THE called machine learning allowed us to visualize skyrmion sheets and their behavior at different temperatures,” said Yue Li, a postdoctoral fellow in msd extension.

Our most intriguing finding was that skyrmions are arranged in a highly ordered pattern at minus 60 degrees Fahrenheit and beyond,” Phatak said.​But as we cool the sample, the arrangement of the skyrmions changes. Like bubbles in beer foam, some skyrmions have grown larger, some smaller, some merged, and some vanished.

Changing groupings of skyrmion from highly ordered to messy with temperature from -92 F (204 kelvin) to -272 F (104 kelvin). The bright dots indicate the order. (credit Argonne National Lab)

At minus 270, the layer reached a state of nearly complete disorder, but order returned when the temperature returned to minus 60. This order-disorder transition with temperature change could be exploited in future microelectronics for storage of memory.

This research was supported by the DO Office of Basic Energy Sciences. The team’s machine learning program ran on supercomputing resources at the Argonne Leadership Computing Facility, a DO Facility for users of the Office of Science.

This research appeared in Nano Letters. In addition to Phatak, Li, and McCray, Argonne authors include Amanda K. Petford-Long, Daniel P. Phelan, and Xuedan Ma. Other authors include Rabindra Basnet, Krishna Pandey, and Jin Hu of the University of Arkansas.

Leave a Comment

%d bloggers like this: