A Clue to Improving the Completeness Level of Neuromorphic Devices Has Been Discovered

A Korean research team has succeeded in securing a basic technology for further improving the completeness level of neuromorphic devices.

The Korea Research Institute of Standards and Science (KRISS, President: Lee, Ho Seong) announced that it has become Korea’s first institute observing the finestructure of magnon*, which is attracting attention as a key material for neuromorphic devices. As areas that are approximately 1,000 times finer than before were observed successfully, it is expected that the observation results will enable the design of more sophisticated neuromorphic devices.
^{*Magnon: A wave that is generated when quantum spins (a characteristic of electrons with magnetic properties) in magnetic materials influence each other. Just as when one domino block falls, the others fall in turn, when energy is applied to one spin, the energy is transferred to the other spins in a ripple-like manner.}

Neuromorphic devices are next-generation semiconductors designed to mimic the structure of the human brain. They process information by mimicking the way neurons generate signals and transmit them to other neurons through synapses. Unlike classical semiconductors where data processing devices and storage devices exchange information with each other, neuromorphic devices perform both data storage and processing simultaneously, allowing for rapid processing of massive amounts of information with little power. This is why neuromorphic devices are considered an innovative technology that will drastically reduce the power consumption of artificial intelligence (AI), which has been rapidly increasing recently.

Magnons are a promising material for implementing neuromorphic devices. This is because they are capable of simultaneously sending multiple signals at ultra-low power by utilizing their unique characteristics of transmitting energy to other spins in a ripple-like manner when energy is applied to one quantum spin. However, with the existing level of technology, only a few areas with large bandwidths in the entire structure of magnons can be investigated; therefore, the implementation of high-performance magnon-based neuromorphic devices devices has been limited.

The KRISS Quantum Magnetic Sensing Group has become Korea’s first research group that observed the entire structure of magnons in the frequency domain. Using VNA* equipment, the research group discovered that numerous fine frequency structures are present around the previously known frequency domain of magnons. The research group was able to confirm the entire structure of magnons by transmitting electric signals and analyzing the reflected and penetrated spectrum.
^{* Vector network analyzer (VNA): Equipment that measures the frequency response characteristics (S-parameters) of electronic circuits or materials. By measuring the amplitude and phase change of the signal output when a specific frequency is input, the characteristics of the material can be analyzed. The research group successfully observed the hidden finestructure of magnons by using the frequency offset function, which is a special function of the VNA equipment.}

Magnons are typcially measured in the gigahertz (GHz) range. The newly found fine structure of magnons in the megahertz (MHz) range can enhance their functionality. Just as the stronger the connections between neurons, the more active the brain becomes, it is expected that finely adjusting the frequency of magnons will allow for more sophisticated design of neuromorphic devices, further enhancing their performance.

In particular, the magnon observation technology used by the research group in this study is expected to be widely used in research and development of related devices, because it is an electrical method that is faster and simpler than the conventional optical method of converting photon signals in a specific area.

Kyongmo AN, a visiting researcher of KRISS’s Quantum Magnetic Sensing Group, said, “In addition to neuromorphic devices, magnons are also drawing attention as a material for implementing quantum spin qubits, quantum ultra-high-speed networks, and next-generation high-precision sensors.” He added, “We will accelerate the development of application devices based on the structure of the magnons found through our study.”

This research result was supported by the Sejong Science Fellowship Grant Program of the Ministry of Science and ICT and was published in August in the internationally renowned journal, Nature Communications (IF: 14.7).