Carbon nanotube based ferroelectric gated transistors for artificial synaptic devices

Abstract

The advancement of neuromorphic computing necessitates the development of synaptic devices that can mimic the functionalities of biological synapses with high efficiency and low power consumption. This dissertation presents a comprehensive study on the development of carbon nanotube (CNT)-based ferroelectric gated synaptic transistor. It integrates high-mobility CNTs with high-quality ferroelectric gate oxide (HfZrO2) to achieve synaptic functions. The study aims to optimize deposition techniques for CNT films and HfZrO2 thin films, ensuring uniformity and high-quality interfaces that are critical for effective gate control. Detailed experimental analyses demonstrate the CNT-based ferroelectric gated synaptic transistors can operate at low voltages with an on/off current ratio of approximately six orders of magnitude and a subthreshold swing of around 80 mV/dec. The synaptic functionalities, characterized through excitatory post-synaptic current (EPSC) measurements, exhibit clear stepwise current increases, highlighting the device's capability to replicate synaptic plasticity. These findings indicate that the proposed CNT based ferroelectric gated synaptic transistors offer substantial potential for integration into neuromorphic systems, providing insights into future optimization strategies and broader applications in artificial intelligence.