Fully memristive neural networks for pattern classification with unsupervised learning

A memristor is a two-terminal electronic device whose conductance can be precisely modulated by charge or flux through it. Here we experimentally demonstrate a nanoscale silicon-based memristor device and show that a hybrid system composed of complementary metal-oxide semiconductor neurons and memristor synapses can support important synaptic functions such as spike timing dependent plasticity. Using memristors as synapses in neuromorphic circuits can potentially offer both high connectivity and high density required for efficient computing.

Memory is believed to occur in the human brain as a result of two types of synaptic plasticity: short-term plasticity (STP) and long-term potentiation (LTP; refs 1-4). In neuromorphic engineering, emulation of known neural behaviour has proven to be difficult to implement in software because of the highly complex interconnected nature of thought processes. Here we report the discovery of a Ag(2)S inorganic synapse, which emulates the synaptic functions of both STP and LTP characteristics through the use of input pulse repetition time. The structure known as an atomic switch, operating at critical voltages, stores information as STP with a spontaneous decay of conductance level in response to intermittent input stimuli, whereas frequent stimulation results in a transition to LTP. The Ag(2)S inorganic synapse has interesting characteristics with analogies to an individual biological synapse, and achieves dynamic memorization in a single device without the need of external preprogramming. A psychological model related to the process of memorizing and forgetting is also demonstrated using the inorganic synapses. Our Ag(2)S element indicates a breakthrough in mimicking synaptic behaviour essential for the further creation of artificial neural systems that emulate characteristics of human memory.

Nanoscale resistive switching devices, sometimes termed memristors, have recently generated significant interest for memory, logic and neuromorphic applications. Resistive switching effects in dielectric-based devices are normally assumed to be caused by conducting filament formation across the electrodes, but the nature of the filaments and their growth dynamics remain controversial. Here we report direct transmission electron microscopy imaging, and structural and compositional analysis of the nanoscale conducting filaments. Through systematic ex-situ and in-situ transmission electron microscopy studies on devices under different programming conditions, we found that the filament growth can be dominated by cation transport in the dielectric film. Unexpectedly, two different growth modes were observed for the first time in materials with different microstructures. Regardless of the growth direction, the narrowest region of the filament was found to be near the dielectric/inert-electrode interface in these devices, suggesting that this region deserves particular attention for continued device optimization.