Thermal Nanostructuring for Rectifying Resistive Switching Behaviors of Cobalt Oxide Neuromorphic Devices

Abstract

Inspired by biosynapses, memristor devices have gained considerable attention as a vital step toward high-performance artificial neuromorphic computation. Despite recent dramatic advances in this field, there are still technical challenges such as low-power switching, robustness, and well-stabled devices for the practical applications of artificial neural networks, all of which are critical to achieve ideal neuromorphic functioning. Herein, we demonstrate a facile approach to produce a thermally nanostructured cobalt-oxide-based memristor with high reproducibility. Due to the asymmetric interfaces, our memristor device exhibits a remarkable rectifying resistive switching behavior with distinctive synaptic functions, including bulk (104pulses) and nanoscale synaptic weight (long-to short-term plasticity) behaviors, paired-pulsed potentiation/depression, and abilities of learning and forgetting like those of the human brain. Furthermore, the Pavlovian associative learning behaviors are successfully imitated by our nanostructured memristor device. Overall, our presented results demonstrate a potential pathway for advancing artificial neuromorphic computing.