Jiyun Lee1,Seojun Heo1,Seongsik Jeong1,Hyeseon Jo1,Haejin Kim1
Gyeongsang National University1
Jiyun Lee1,Seojun Heo1,Seongsik Jeong1,Hyeseon Jo1,Haejin Kim1
Gyeongsang National University1
The human brain is a complex neural network with more than 100 billion neurons and 100 trillion synapses, which act as bridges to connect the neurons for signal communications. Moreover, it is a biocomputing system that covers advanced computation and cognitive processes while using only 20W of power. The development of neuromorphic devices to mimic the functions of such an efficient information processing and storage capability is emerging as a key technology in digital environments, which necessitate complicated information processing and high-density information storage functions. In particular, there has been a surge in related research on neuromorphic systems that mimic the electronic synapses with the operation of transmitting and computing the information signals. The electrical properties (e.g., electrical post-synaptic current (EPSC), paired-pulse facilitation (PPF)) of the developed synapse are characterized when the synapses connecting neurons exchange spike signals to process the information. The early stage of the research on developing synaptic devices was mainly based on a CMOS-based integrated circuit that requires numerous transistors to realize synapse characteristics, resulting in a limitation in the system complexity and reduced integration. Memristors were first proposed to overcome the limits however, high-power consumption per unit device was considered to be one of the disadvantages compared to the bio-synapse. Recently, the three-terminal synaptic transistors mimicking the human brain function were considered to be one of the outstanding candidates that fulfill various requirements to mimic the bio-synaptic systems. Also, synaptic devices with mechanical stretchability were developed to meet the increasing demand and interest in wearable device applications.<br/>In this work, elastomeric synaptic transistors were developed to perform various synaptic behaviors, including EPSC, PPF, long-term memory (LTM), and STM under various mechanical stimuli. Specifically, the elastomeric electrodes were manufactured by using the gold nanosheets (AuNSs) embedded within the elastomer polydimethylsiloxane (PDMS). Also, an organic semiconductor Poly(3-hexylthiophene-2,5-diyl) (P3HT) nanocomposites and ion-gel were utilized as elastomeric semiconductor and dielectric, respectively. Transfer characteristics of the devices were derived by applying a pulse-type voltage to the gate electrodes. Other synaptic characteristics, including EPSC, and PPF values were carried out under various levels of the mechanical strain, which resulted in stable performances regardless of the various level of the mechanical strain. Also, LTM was achieved by repeatedly learning for a short time while continually administering a pulse-type spike at 0.15s intervals. In addition, it was confirmed that the synapse characteristics were well maintained under the mechanical stimuli up to 100%. Overall, the developed elastomeric synaptic devices with advanced signal transmission and cognition functions while consuming less energy are expected to contribute to the advancement of soft robot engineering and the development of next-generation intelligent wearable devices.