Perovskite-Enhanced Silicon-Nanocrystal Optoelectronic Synaptic Devices for the Simulation of Biased and Correlated Random-Walk Learning.

Yiyue Zhu,Yiqiang Zhang,Yifei He,Xiaodong Pi,Lei Yin,Wen Huang,Deren Yang

doi:10.34133/2020/7538450

Yiyue Zhu, Yiqiang Zhang + Show 5 more

Open Access

https://doi.org/10.34133/2020/7538450

Copy DOI

Abstract

Silicon- (Si-) based optoelectronic synaptic devices mimicking biological synaptic functionalities may be critical to the development of large-scale integrated optoelectronic artificial neural networks. As a type of important Si materials, Si nanocrystals (NCs) have been successfully employed to fabricate optoelectronic synaptic devices. In this work, organometal halide perovskite with excellent optical asborption is employed to improve the performance of optically stimulated Si-NC-based optoelectronic synaptic devices. The improvement is evidenced by the increased optical sensitivity and decreased electrical energy consumption of the devices. It is found that the current simulation of biological synaptic plasticity is essentially enabled by photogating, which is based on the heterojuction between Si NCs and organometal halide perovskite. By using the synaptic plasticity, we have simulated the well-known biased and correlated random-walk (BCRW) learning.

Highlights

Computers have gained worldwide popularity over the past few decades
For instance excitatory postsynaptic current (EPSC), pairedpulse facilitation (PPF), spike-number-dependent plasticity (SNDP), and spike-rate-dependent plasticity (SRDP) [48], can be all mimicked by using synaptic transistors which are based on the hybrid structure of organometal halide perovskite and silicon nanocrystals (Si NCs)
From the cross-section scanning electron microscopy (SEM) image (Figure 1(e)), it is found that the perovskite film and Si NC film are ~500 nm and ~850 nm thick, respectively

Summary

Introduction

Computers have gained worldwide popularity over the past few decades. The von Neumann architecture on which computers are based, has been increasingly limiting the further development of computers [1,2,3,4]. It is noteworthy that silicon nanocrystals (Si NCs) have been successfully employed in the fabrication of synaptic devices as a type of important Si materials [36,37,38,39] They exemplified the great promise for the development of Si-based optoelectronically integrated ANNs, which would facilitate widely deployable neuromorphic computing. The electronic coupling between organometal halide perovskite and Si NCs likely helps to reduce the electrical energy consumption of Si-NC-based optoelectronic synaptic devices. For instance excitatory postsynaptic current (EPSC), pairedpulse facilitation (PPF), spike-number-dependent plasticity (SNDP), and spike-rate-dependent plasticity (SRDP) [48], can be all mimicked by using synaptic transistors which are based on the hybrid structure of organometal halide perovskite and Si NCs. it is demonstrated that biased and correlated random-walk (BCRW) learning can be simulated by our synaptic devices

Results and Discussion

Conclusions

Experimental Section

Conflicts of Interest