An artificial synaptic device based on 1,2-diphenylacetylene with femtojoule energy consumption for neuromorphic computing
Abstract
Organic small molecule memristors show great potential in the application of low-energy neuromorphic computing such as artificial synapses. In this study, based on the small molecule 1,2-diphenylacetylene (DPA), various biological synaptic functions have been imitated with subfemtojoule energy consumption (∼1.2 fJ), multilevel conductance states and highly linear conductance updates. Based on spike-rate-dependent plasticity (SRDP) and Bienenstock–Cooper–Munro (BCM) learning rules, the image edge detection has been simulated, which is helpful for real-time image processing. An accuracy rate of 94.7% is obtained when performing the classification task on the fashion-MNIST dataset, demonstrating high accuracy and low-energy consumption in brain-like pattern recognition for neuromorphic computing.