Observation of Magnéli Phase Filament Formation in MoO x Artificial Synapse

Artificial synapses are the basic building block of artificial neural networks capable of neuromorphic computing, which might overtake conventional digital computing in areas like artificial intelligence, deep learning, and in-memory computation. Neuromorphic computing with artificial synapses opens a new window for fast in-memory computing and image processing. In this paper, an MoOx-based artificial synapse that mimics almost all characteristics of bio-synapses is demonstrated. The fabricated device shows excellent synaptic properties such as potentiation, depression, forgetting, paired-pulse facilitation, and spike-timing-dependent plasticity. The mechanism behind all the observed characteristics has been well explained. A lateral-type device is also designed and tested in order to confirm the conductance change during the application of electrical pulses. Energy-dispersive X-ray analysis and Raman analysis reveal that the switching mechanism is primarily due to the formation and rupture of conducting filaments composed of Magnéli phases of MoOx. Since the fabricated ITO/MoOx/Ag artificial synapse exhibits excellent synaptic characteristics and the mechanism is thoroughly explored, this study will contribute to the fabrication of neural networks and future neuromorphic computing.