Neuromorphic computing (brain-like computing) has become an important direction for the development of high-performance computing due to its advantages of low energy consumption, adaptive learning, and high parallel computing. Neuromorphic computing requires the construction of artificial neural networks. As the key elements of artificial neural networks, synaptic devices have been attracting more and more attention in recent years. In the same time, optogenetics that uses light to control and monitor neural activities in neuroscience has been making significant progress, inspiring the incorporation of light into synaptic devices to make optoelectronic synaptic devices. In terms of facile optoelectronic integration and large-scale application, it is advantageous to build optoelectronically integrated neuromorphic computing on the platform of silicon technologies in the future. Therefore, the development of silicon-based optoelectronic synaptic devices is of great importance.

Professor Xiaodong Pi’s group at Zhejiang University now fabricates a new type of silicon-based optoelectronic synaptic devices. These devices are made of two key materials: silicon nanocrystals and organometal halide perovskite. The former is technologically important nanostructured silicon. The latter is an emerging optoelectronic semiconductor with excellent optical absorption.

A series of synaptic functionalities have been simulated by using the synaptic devices based on the heterostructure of silicon nanocrystals and perovskite. Compared with previous optoelectronic synaptic devices only based on silicon nanocrystals, current devices show significantly enhanced optical responsivity and significantly reduced electrical energy consumption (both are in the order of magnitude of three). In addition, the synaptic plasticity of current devices may be employed to simulate important neural activities such as the well-known biased and correlated random-walk (BCRW) learning. This work is published in Research. 

https://spj.sciencemag.org/journals/research/2020/7538450/

Prof. Pi point out that the continued development of integrated circuits that modern computers rely on requires silicon-based optoelectronic integration. If the future neuromorphic computing (brain-like computing) can make full use of silicon-based optoelectronic integration, it’s performance and deployment should be effectively improved. The development of silicon-based optoelectronic synaptic devices and their integration into artificial neural networks is important and forward-looking work. Clearly, silicon-based heterostructures are critical structures for silicon-based optoelectronic synaptic devices, enabling silicon materials to also play a vital role in the future computing.

Tag:  Technological applications of all research