A South Korean research team has developed a new type of storage device, potentially replacing current memory or facilitating neuromorphic computing in future artificial intelligence hardware.
Neuromorphic computing, a computing paradigm that simulates brain-like mechanisms, often uses Spiking Neural Networks (SNN) on neuromorphic chips. The goal is to replicate artificial intelligence (AI) by mimicking the human brain’s neurons and synapses. Benefiting from low power consumption, neuromorphic computing is considered a potential successor to traditional AI.
Recently, researchers from South Korea created a new storage device for potential use in neuromorphic computing for future AI hardware. This device has the benefits of low processing costs and ultra-low power consumption.
On April 4th, the Korea Advanced Institute of Science and Technology (KAIST) announced that a research team led by Professor Shinhyun Choi from the School of Electrical Engineering had developed a next-generation phase-change memory device. This device, which has ultra-low power consumption, could replace DRAM and NAND flash.
Phase Change Memory (PCM) stores and processes information by altering the crystalline state of a material using heat, thus changing its resistance state. PCM is considered a promising candidate to overcome the von Neumann bottleneck due to its low latency, non-volatile memory characteristics, and high integration density. However, existing PCMs face issues of high production costs and high operating power. To combat these, Professor Choi’s team developed an ultra-low power PCM device that forms small nanoscale phase-changeable filaments without needing expensive manufacturing processes.
DRAM, a commonly used memory type, is known for its speed but is volatile, as data disappears when power is cut off. NAND flash, on the other hand, retains data even when power is disconnected, although it has slower read/write speeds.
PCM unifies the advantages of DRAM and NAND flash by offering high speed and non-volatility. Therefore, it is considered a potential successor to existing memory and is actively researched as a brain-mimicking or neuromorphic computing technology.
However, traditional PCM devices require high power, creating difficulties in producing practical large-capacity storage products or neuromorphic computing systems. Previous research tried to maximize the thermal efficiency of storage device operation by minimizing the physical size of devices using advanced lithography techniques. However, the improvements in power consumption were minimal, and manufacturing costs and difficulties increased with each advancement.
To solve this issue, Professor Shinhyun Choi’s team developed a method to electrically form phase-change material in an extremely small area. This resulted in an ultra-low power PCM device that consumes 15 times less power than traditional PCM devices made using expensive lithography tools.
Professor Shinhyun Choi is optimistic about the future development of this research, stating, “Our developed phase-change memory device is significant as it provides a novel method to solve long-standing issues in memory production. It greatly improves manufacturing costs and energy efficiency. We expect our research findings to lay the foundation for future electronic engineering, enabling various applications including high-density three-dimensional vertical memory and neuromorphic computing systems.”
Doctoral students See-On Park and Seokman Hong from the School of Electrical Engineering at KAIST led the research. It was published in the April issue of the academic journal Nature (Paper title: Phase-Change Memory via a Phase-Changeable Self-Confined Nano-Filament).
This research was supported by South Korea’s Next-Generation Intelligent Semiconductor Technology Development Project, the PIM AI Semiconductor Core Technology Development (Device) Project, the Korea National Research Foundation’s Excellent Emerging Research Project, and the National Nanofab Center’s Semiconductor Process Nano Medical Device Development Project.
Disclaimer: This article is created by the original author. The content of the article represents their personal opinions. Our reposting is for sharing and discussion purposes only and does not imply our endorsement or agreement. If you have any objections, please contact us through the provided channels.
