First Demonstration of Yttria-Stabilized Hafnia-Based Long-Retention Solid-State Electrolyte-Gated Transistor for Human-Like Neuromorphic Computing

초록

Electrolyte-gated transistors have strong potential for high-performance artificial synapses in neuromorphic bio-interfaces owing to their outstanding synaptic characteristics, low power consumption, and human-like mechanisms. However, the short retention time is a hurdle to overcome owing to the natural diffusion of protons. Here, a novel modulation technique of ionic conductivity is proposed with yttria-stabilized hafnia for the first time to enhance the retention characteristic of a solid-state electrolyte-gated transistor-based artificial synapse. With the optimization of the ionic conductivity in yttria-stabilized hafnia, a high retention time of over 300 s and remarkable synaptic characteristics are accomplished by regulating channel conductance with precise modulation of the strength of the proton-electron coupling intensity along the input signals. Furthermore, pattern recognition simulation is conducted based on the measured synaptic characteristics, exhibiting 94.41% of operation accuracy, which implies a promising solution for neuromorphic in-memory computing systems with a high operation accuracy and low power consumption. The artificial synapse device with the YSH-based electrolyte-gated transistor is proposed. With the migration of protons, outstanding synaptic characteristics are accomplished in a human-like operation mechanism. Through optimization of ionic conductivity, long-term potentiation over 200 s is showed for potentiated, depressed, and intermediate states. Furthermore, pattern recognition simulation with measured synaptic characteristics showed 94.41% of operation accuracy for high-performance neuromorphic systems.image

키워드