As you read this, intricate networks of neurons are firing at lightning speed, processing sensory information from the eyes almost instantaneously. These incredible capabilities of the human brain have served as inspiration for next-generation computing platforms.
In the future, mobile technologies enabled by such neuromorphic computing would have the ability to sense and make real-time decisions, right where data are collected.
However, attaining this characteristic, known as edge intelligence, is tough—it requires easy-to-manufacture, energy-efficient, and highly dynamic memory components, of which current industry offerings fall short. As a solution, researchers are pinning their hopes on tiny, bead-like magnetic structures called skyrmions.
“Skyrmions are nanometre-scale magnetic structures comprising a winding arrangement of electron spins,” explained Ho Pin, a Research Scientist with the Spin Technology for Electronic Devices (SpEED) team at A*STAR’s Institute of Materials Research and Engineering (IMRE). “Like tiny abacus beads that can be used to represent discrete values at defined positions, skyrmions can serve as room-temperature information carriers. They can be moved controllably by electrical currents in one or two dimensions.”
Whether skyrmions have the power to transform next-generation computing in real-world applications is still an open question. In search of answers, the SpEED team created a suite of innovative analytic approaches to take a closer look at skyrmions in action.
The researchers developed a specialised experimental setup using a magnetic force microscope to track skyrmion activity within electronic devices. They watched the dynamics of over 20,000 skyrmions under different conditions, to create a novel framework for how the size, geometry, and configuration of skyrmions affect their motion, and may influence their computing performance.
This was no small feat, considering that the experimental setup had to be incrementally tailored to a range of devices and circuits.
“We had to repeatedly refine our procedures for device fabrication, electrical pulses, and imaging to ensure the integrity of the devices and skyrmion images,” said Ho.
With the data collection mastered, they then developed custom image processing and computational tools to consolidate and analyse their massive dataset of individual skyrmion movements. The team found that skyrmions move exponentially faster with increasing applied current, largely impervious to their size, position, and imperfections within the device. In addition, they observed the transverse motion of skyrmions along the wire’s width under a lateral current—a behaviour known as the skyrmion Hall effect—was reduced at the edges of the wire and was less pronounced for smaller skyrmions.
“Our findings suggest that device geometries could be designed to achieve bespoke motion of individual skyrmions according to application needs,” Ho concluded. “Our results provide a jumping-off point for future skyrmionic devices as building blocks of next-generation computing architectures.”
The A*STAR-affiliated researchers contributing to this research are from the Institute of Materials Research and Engineering (IMRE).