IBS Research Team developed "Resembling the human brain memory device"
- 자연과학대학
- Hit3349
- 2016-11-15
Resembling the human brain memory device development
IBS Nanostructures Physics Research Team developed 'tunneling memory' by two-dimensional nanomaterials
In March, the artificial intelligence(AI) 'Alpha-Go' of Google that beat Lee Se-dol caught attention for its low-power, energy-consuming and high-performance operating 1200 computer CPUs and comsuming 56 kilowatts (kW) power per hour, This is a special Software (SW) applied to it. Existing computer is fast in the sequentially calculating process such as mathematical calculations, but it is not suitable for parallel computation AI required. However, Alpha-Go was designed by applying a special SW to enable parallel computing. If it can be made in memory, it will also be able to implement the function of AI in a small smartphones. Korean researchers have developed a next-generation memory device for realizing these.
A synapse (left) to deliver and store information in the human brain and two electrodes memory device that researchers developed (bottom right) / data = IBS
Basic Science Institute (IBS) nanostructures Physics Research Center (Director Lee Young-hee, Sungkyunkwan University Department of Physics, Professor), Yu-Woojong Sungkyunkwan University Department of Electrical Engineering and co-researchers announced that they has succeeded in implementing a semiconductor device.
The human brain synapses (site to send and receive thousands, tens of thousands of signals attached to the ear of the nerve cells) send and receive signals to the two projections corresponding to the electrode of the device, and stores the memory to leave a residual image signal. The human brain can rapidly process a high degree of parallel operation with little energy based on the same synaptic system.
Researchers newly design the existing memory hardware structure to develop a memory that mimics the structure of the synapse.
Researchers first eliminate the 'storage electrode' (gate) in a flash memory structure having a conventional three electrodes. Instead, they developed tunneling memory that works like a synapse performing signal transmission and storage at the same time.
According to researchers, tunneling memory is a two-dimensional nanomaterial made with Graphene, hexagonal boron nitride (h-BN), and molybdenum disulfide (MoS2).
Graphene is a good carbon material with electric conductivity and thermal conductivity, and hexagonal boron nitride is a new material that can maintain full performance to materials and devices minimizing the interaction between the substrate and the electronic materials and devices. Molybdenum disulfide is new material that can be applied to electronic and optoelectronic devices.
According to researchers, tunneling memory structure can be directly applied to a silicon memory. It also has low power consumption and high material stability with only a two-dimensional nanomaterials with good elasticity so that can provide elastic supports. Tunneling memory is expected to be applied to future wearable device.
Professor Lee said, "We have plan to develop system using a tunneling memory as synapse like a signal transmission system of the brain cells.“ Hee added, "Further research will be completed in small digital devices, such as smartphones."
The study results were published in the international journal "Nature Communications".