Optoelectronic Synaptic Devices Based On Silicon Nanocrystals

With the booming of Internet of Things(IoT),Big Data(BD)and Artificial Intelligence(AI),the demand for the low-energy-consumption and high-adaptability computing is gradually growing.The strategy of information processing in the neuromorphic computing is total different from that in the conventional von Neumann computing,which mimicks the structure of human brains to improve the thinking capacity and reactivity of computing.Therefore,neuromorphic computing may play a leading role in the post-von Neumann era.The first step to realizing neuromorphic computing is to fabricate synaptic devices that mimic the functionalities of basic units in the human brain-Synapses.Although significant progress has been made in electronic synaptic devices,they may face limited bandwidth and crosstalk issues when they are inter-coupled.Optoelectronic synaptic devices introduce optical mechanism to process information,well solving the above obsessions.Nowadays,a few interesting materials have been employed to fabricate optically stimulated synaptic devices,the use of silicon(Si)that is the material of choice in the conventional von Neumann computing has not been explored for optically stimulated synaptic devices.In this paper,we take advantage of one of the most important nanostructures of Si-Si nanocrystals(NCs)doped with boron(B)to make synaptic devices,which can be effectively stimulated by light in the unprecedented broad spectral region from the ultraviolet(UV)to near-infrared(NIR),approaching the wavelength of?2 ?m.We use pulsed lasers in a broadband(UV-NIR)to act as synaptic spikes,these Si-NC-based synaptic devices exhibit a series of major synaptic plasticity.It is also found that the synaptic plasticity arising in our Si-NC-based synaptic devices is governed by the dynamic trapping and releasing of photo-carriers induced by dangling bonds at the surface of Si NCs.The demonstration of Si-NC-based optoelectronic synaptic devices in this paper is of great importance to promote Si implemented on the neuromorphic computing.The main contents of this paper include:1.Preparing Si NCs by the non-thermal plasma method.We also obtain boron-doped Si NCs(B-doped Si NCs)by introducing B2H6 in the gas phase.The XRD,Raman and TEM are performed to analyze the structure and morphology of B-doped Si NCs.The results of STM,ERP,FET-and temperature-dependent conductance test show that there are two obvious states located in the electronic structure of B-doped Si NCs.The trap states are related to the dangling bonds at the surface of Si NCs,and the band-tail states are introduced by boron doping.The UV-vis-NIR absorption spectra also demonstrates the band-tail states induced decrease of the optical bandgap of B-doped Si NCs,which will improve the bandwidth of optoelectronic synaptic devices.2.Fabricating Si-NC-based synaptic devices on glass substrates by using the sandwich structure of ITO/Si-NC/Al.Employing pulsed lasers as synaptic spikes,we successfully simulate the main synaptic plasticity of EPSC,PPF,STP and LTP in our devices.Then we use inter-coupled synaptic devices to simulate the STDP by stimulating them with two spikes separately.The observed asymmetric STDP attributes to the Schottky barrier existed at the interface of Al/Si-NC.The difference in carrier transport between the pre-and post-synaptic neuron in the inter-coupled devices also contributes to the asymmetry STDP.Combining the electronic and optical properties of B-doped Si NCs into consideration,we conclude that the trapping and releasing of carriers induced by trap states in Si NCs is the critical reason that leads to the rising of synaptic plasticity in our synaptic devices.The non-volatile effect observed on the UV light-stimulated device is related to the optical absorption property of Si NCs and the UV-induced ionization of oxygen vacancies at the Al/Si-NC interface.