A neotype self-rectifying Cu3SnS4-MoO3 synaptic memristor for neuromorphic applications

Self-rectifying memristor with integrated excellent bio-synaptic behaviors are great potential to realize high-density memristor neuromorphic networks with self-inhibition of stealth current effect, but it's still inferior and challenging yet achievable. Here, optoelectronic memristor in accordance with p-n heterostructured Cu3SnS4-MoO3 (CTS-MoO3) is developed. The Ag/CTS-MoO3/Mo memristor exhibits stable nonvolatile resistive switching with excellent spatial uniformity and high self-rectifying characteristics (rectification ratio >4000), which is beneficial to implement crossbar memristive synapse architectures. The memristor demonstrates not only concentrated Set/Reset voltage distribution (variation < 0.04 V/0.01 V), high On/Off ratio (>103) and long retention time (>104 s), but also continuously modulable conductance by applying electric pulses (triangular and square-wave) or various light (470–808 nm) stimulus. This behavior makes such memristor the ability to emulate vital bio-synaptic functionalities including excitatory and inhibitory, short-/long-term plasticity, spike-timing-dependent plasticity, as well as the learning-forgetting-learning process and Ebbinghaus forgetting rule. Moreover, the recognition rate for MNIST handwritten digits in such memristor based artificial neural network model is verified to be 89.3 % for neuromorphic simulations. The results dramatically facilitate the development of self-rectifying optoelectronic artificial synapse for future neuromorphic applications.