講演情報
[17p-K307-13]Transition from Short-Term to Long-Term Memory in an Ag-Ag2S Atomic Switch Network
〇(P)SAMAPIKA MALLIK1,2, Thien Tan Dang1, Nakaoka Yusuke1, Usami Yuki1,2, Tanaka Hirofumi1,2 (1.LSSE, Kyutech, 2.Neumorph Center, Kyutech)
キーワード：
Nanoparticles、Atomic switch network、Synaptic plasticity
In recent years, mimicking the functionality of biological synapses has become a pivotal goal in developing neuromorphic devices and artificial intelligence systems. Among various material systems, atomic switch networks (ASNs) based on Ag-Ag2S have emerged as promising candidates for replicating synaptic behavior. Here, we investigated the short-term and long-term memory functions of an ASN using Ag-Ag2S core-shell nanoparticles. Our findings reveal that the network exhibited a transition from volatile to non-volatile switching, corresponding to a transition from short-term to long-term memory (LTM) in biological synapses.The device structure features 16 electrodes with nanoparticles drop-casted into the central gap. Figure 1a shows the typical current-voltage (I-V) characteristics of the network at room temperature under voltage sweep (0 V to 3 V, 3 V to -1 V and -1 V to 0 V). During the forward sweep from 0 to 3 V with a voltage step of 20 mV, the device exhibited a SET process at 1.6 V, transitioning from the OFF to the ON state, indicative of metal filament formation within the network. Upon reversing the sweep from 3 V to -1 V, a RESET process occurred at 0.6 V, resulting in filament dissolution and volatile switching behavior. This rapid return to the initial state upon voltage removal mirrors the characteristics of short-term memory (STM) in the human brain. Figure 1b illustrates the I-V characteristics over repeated sweeps from the 2nd to the 23rd cycle, revealing a transition from volatile to nonvolatile switching. With repeated voltage sweeps, the network gradually shifted from STM-like behavior to LTM-like behavior.