Engineering Synaptic Plasticity in WO3-Based Ion-Gated Transistors for Neuromorphic Systems

Neuromorphic computing, based on analog neural networks, aims to reduce computational power significantly by minimizing the need for data transfers between memory and logic components. Within neuromorphic computing, the artificial synapse serves a dual purpose as both a storage and computational unit, paving the way for resilient learning and power-efficient in-memory computing. Diverging from conventional computing paradigms, neuromorphic systems replicate the structural and functional attributes of the human brain. This emulation allows for parallel processing, improving efficiency in handling complex tasks.<br/>Metal Oxides (MOs)-based IGTs and the tailoring of their properties for synaptic behavior, achieved by leveraging their unique electronic structure, composition, crystallinity, morphology, intrinsic, and/or extrinsic defects, as well as doping, have gained significant attention.<br/>In this study, we used tungsten trioxide (WO₃) as channel material. WO₃ is earth-abundant with non-toxicity, high stability under ambient conditions, and biocompatibility. WO₃ can exist in different crystal structures, including orthorhombic, monoclinic, and hexagonal phases. The current modulation in MO IGTs arises from both electrostatic and electrochemical processes. The crystal structure of the metal oxide is anticipated to have an impact on ion permeability, leading to variations in doping, channel conductivity, and redistribution time of the anions in the IGT channel.<br/>In our WO3 synaptic transistor, the channel conductivity, analogous to the excitatory postsynaptic current (EPSC), was modulated by the amplitude, width, interval, and the number of applied input gate-source voltage (Vgs) pulses. The fluctuations in conductance correspond to the learning and forgetting processes of the artificial synaptic device, with an increase signifying learning and a decrease in signifying forgetting.<br/>Another fundamental aspect that we investigated in WO3-based IGTs is their plasticity, denoting the strengthen of synaptic weight or connection. Synaptic plasticity is typically divided into two categories, namely short-term (STP) and long-term (LTP), determined by the duration of their persistence.