Memristors based on carbon dots for learning activities in artificial biosynapse applications

With the rapid development of information technology for big data, memristors have become more and more popular nanoscale devices for storing a large amount of information and neural learning. However, random formation of conductive filaments (CFs) in a memristor leads to a broad distribution of device parameters, which leads to a high error rate in the iteration process of neural network learning. In this work, carbon dots (CDs) are proposed to improve the uniformity of several different oxide memristor parameters, and obviously obtain more stable high and low resistances, lower power consumption, and fast response speed. What's more, three different spike-timing-dependent plasticity (STDP) learning rules, paired-pulse facilitation (PPF), supervised learning and interest-based learning activities are simulated by carbon dots based memristor devices (CDMDs). And the preview and review learning method simulation by the PQ4R strategy are also achieved faithfully for the first time. This work provides a new way to improve the performance of memristors and develop new neuromorphic functions, which could significantly facilitate the development of artificial nervous chip systems.