An electronic synapse memristor device with conductance linearity using quantized conduction for neuroinspired computing

An electrochemical metallization memristor based on Zr0.5Hf0.5O2 film and an active Cu electrode with quantum conductance and neuromorphic behavior has been reported in this work. After electroforming in the Cu/Zr0.5Hf0.5O2/Pt device, linear conductance characteristics in low resistance states were found and the stepwise changes of conductance with the order of G(0) ((=2e(2))/h) multilevel states were obtained by varying pulse amplitude, width and adjacent-pulse time interval, which is beneficial for backpropagation learning algorithms belonging to deep neural networks, essentially using memristors as vector-matrix multiplication accelerators in image processing. The gradual resistance tuning served as the basis of memory and learning. Under the coactivity of pre- and post-synaptic spikes, bidirectional long-term Hebbian plasticity modulation was realized. The temporal difference, spike rate and size of the top and bottom electrode pulse voltage can strongly affect the sign and degree of Hebbian plasticity. Moreover, the quantum conductance phenomenon was ascribed to interstitial Cu in the dielectric layer forming single- and multi-atom chains. The results can provide multilevel storage and next-generation parallel neuromorphic computing architecture, promoting the development of functional plastic electronic synapses.