Realization of Tunable Plasticity in Porous Synaptic Memristors for Neuromorphic Computing

Abstract Brain-inspired neuromorphic computing is a promising way to implement artificial intelligence to overcome the issues of independent information processing and storage. An artificial synaptic device with tunable plasticity can perform learning and memorization by adjusting the weight of the synapse. In this work, a synaptic memristor composed of porous silicon oxide (PSiO x ) incorporated with MoS 2 quantum dots (QDs) is fabricated and excitatory paired-pulse facilitation (PPF), post-tetanic potentiation (PTP), learning-forgetting behavior, and spike-timing-dependent plasticity (STDP) are demonstrated. The short/long-term plasticity (SLTP) in biological synapses reveal the possibility of tunable synaptic plasticity with self-regulating functions under a series of excitation frequency between 200 μs and 10 ms. An artificial neural network (ANN) is designed theoretically according to the SLTP characteristic curves of the synapses and the recognition rate is observed to increase from 54.2% to 91.8% by simply adjusting the input frequency. The image recognition accuracy is improved by 6% in the presence of 20% noise at an input frequency of 1 ms. The excellent results and novel strategy reveal an important step for image recognition in next-generation neuromorphic computing systems.