Study of Metal and Oxide based Synaptic Transistors by Ionic liquid Gating for Neuromorphic Computing

Abstract

Neuromorphic Computing (NC), which emulates neural activities of the human brain, is considered for low-power implementation of artificial intelligence. Towards realizing NC, fabrication, and investigations of synaptic devices replicating the functionalities of the biological counterparts are essential. Emulation of synaptic plasticity, which is the basis for learning and memory in human brain is viewed as a key step for hardware implementation of NC. In this work, we have demonstrated three-terminal synaptic transistors by ionic-liquid gating, based on various materials and mainly focused on emulating essential synaptic functionalities. In the first work, we have studied the emulation of synaptic properties in cobalt-based synaptic transistor by ionic-liquid gating. The transfer curve exhibited a giant hysteresis loop demonstrating a nonvolatile and reversible change of channel conductance, suitable for emulation of synaptic functionalities. This is the first demonstration of a synaptic device using a metallic channel. We have emulated several functions, including excitatory/inhibitory postsynaptic conductance, paired-pulse facilitation/depression in the device, demonstrating short-term plasticity. Furthermore, a transition from short-term to long-term plasticity is shown by tuning the gate pulse amplitude, duration, and number. We have mimicked an important cognitive behavior, learning-relearning -forgetting, showing resemblance to the human brain. Along with memory and learning, the device showed dynamic filtering behavior. The second work deals with the study of permalloy-based synaptic transistor by ionic-liquid gating. We have realized a conductance modulation of 11% at room temperature, one of the highest among metals. The multilevel, nonvolatile conductance states were realized by applying gate pulses of different amplitudes that are crucial for emulation of synaptic properties. In the permalloy-based synaptic transistor, we mimicked the multistore model of the human memory system...