Hardware Implementation of Network Connectivity Relationships Using 2D hBN-Based Artificial Neuron and Synaptic Devices

Brain-inspired neuromorphic computing has been developed as a potential candidate for solving the von Neumann bottleneck of traditional computing systems. 2D materials-based memristors have been exponentially investigated as promising building blocks of neuromorphic computing because of their excellent electrical performance, simple structure, and small device scale. However, while many researchers have focused on looking into individual artificial neuromorphic devices based on memristors, only few studies on the integration of artificial neuron and synaptic devices have been reported. In this work, both volatile and nonvolatile memristors are fabricated by using a 2D hexagonal boron nitride film for artificial neuron and synaptic devices, respectively. The leaky-integrate-and-fire neuron performance and synaptic functions (e.g., synaptic weight plasticity and spike-timing-dependent plasticity) are well emulated with the fabricated volatile and nonvolatile devices. The MNIST image classification is conducted based on the experimental data. For the first time, an artificial neuron-synapse-neuron neural network is physically constructed using the artificial neuron and synaptic devices to mimic the biological neural networks. The synaptic connection strength modulation is experimentally demonstrated between the neurons depending on the conductance state of the synapse, paving the way for the development of large-scale neural network hardware. Volatile and nonvolatile memristors based on 2D hexagonal boron nitride are fabricated and the leaky-integrate-and-fire neuron and synaptic functions are demonstrated. The network connectivity relationships using 2D hBN-based artificial neuron and synaptic devices in hardware are investigated. The connection strength between the artificial neurons is well modulated via the different synaptic weights of the artificial synaptic device.image