Elsevier

Nano Energy

Volume 52, October 2018, Pages 422-430
Nano Energy

Full paper
Broadband optoelectronic synaptic devices based on silicon nanocrystals for neuromorphic computing

https://doi.org/10.1016/j.nanoen.2018.08.018Get rights and content

Highlights

  • Silicon nanocrystals (Si NCs) are used to fabricate optoelectronic synaptic devices.

  • Main functionalities of biological synapses are emulated by using broadband light.

  • The operation is basically governed by the electronic and optical behavior of Si NCs.

Abstract

Optically stimulated synaptic devices are critical to the development of neuromorphic computing with broad bandwidth and efficient interconnect. Although a few interesting materials have been employed to fabricate optically stimulated synaptic devices, the use of silicon (Si) that is the material of choice for very large-scale integration circuits in the conventional von Neumann computing has not been explored for optically stimulated synaptic devices. Here we take advantage of one of the most important nanostructures of Si — Si nanocrystals (NCs) to make synaptic devices, which can be effectively stimulated by light in the unprecedented broad spectral region from the ultraviolet to near-infrared, approaching the wavelength of ∼ 2 µm. These optically stimulated Si-NC-based synaptic devices demonstrate a series of important synaptic functionalities, well mimicking biological synapses. The plasticity of Si-NC-based synaptic devices originates from the dynamic trapping and release of photogenerated carriers at defects such as dangling bonds at the NC surface. The current facile use of Si NCs in broadband optoelectronic synaptic devices with low energy consumption has important implication for the large-scale deployment of Si in the emerging neuromorphic computing.

Graphical abstract

Silicon nanocrystals (Si NCs) are used to fabricate optoelectronic synaptic devices whose energy consumption may be rather low. Essential synaptic functionalities have been realized in these devices by using broadband light to stimulate them.

fx1
  1. Download : Download high-res image (243KB)
  2. Download : Download full-size image

Introduction

In the past half century computers with the von Neumann architecture have been evolving rapidly due to their advantages in solving structured mathematical problems [1]. However, their potential is now nearly exhausted, struggling to meet the growing demand for highly energy-efficient and intelligent computing [2]. This is mainly because frequent data transfer between the physically separated processor and memory in the von Neumann architecture consumes enormous energy. The von Neumann architecture can only work with pre-specified programs, disabling unstructured and real-time problems to be solved. As the most powerful information processor in nature, a brain is capable of high-speed computation with ultralow power dissipation because numerous interconnected neural circuits in the brain enable distributed parallel processing [3], [4]. A brain is also good at intelligent activities such as self-learning, future event prediction and language comprehension [4]. Clearly, brain-like computing is more energy-efficient and intelligent than the von Neumann computing. Therefore, neuromorphic computing that mimics the functionalities of a brain is emerging as one of the most important choices for the next-generation computing [5], [6], [7].

The realization of neuromorphic computing critically depends on the development of synaptic devices since the transmission of information between neurons is enabled by synapses in a neural system [5], [8], [9], [10]. Although all kinds of synaptic devices based on electrically stimulated resistive changes have been demonstrated [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], their all-electronic nature renders limited bandwidth and considerable interconnect issues such as delay and power loss. It has been recently proposed that optical stimulation may be employed to significantly broaden the bandwidth and mitigate the interconnect issues of a synaptic device [26], [27], [28], [29], [30], [31]. For example, synaptic devices fabricated by using hybrid structures based on carbon nanotubes were shown to effectively respond to optical stimulation in the ultraviolet (UV) to visible region (405 – 532 nm) [26], [27]. Synaptic devices based on perovskite materials were proved to effectively respond to optical stimulation in the visible region (500 – 635 nm) [28], [29]. Li et al. [30] and Lee et al. [31] demonstrated synaptic devices based on indium-gallium-zinc oxide (IGZO) films, which worked with the UV and visible light stimulation (365 – 630 nm). Given the fact that silicon (Si) is the material of choice for very large-scale integration (VLSI) circuits in the conventional von Neumann computing [1], it is interesting to ask if Si can also play a critical role in the emerging neuromorphic computing by enabling the recently proposed optically stimulated synaptic devices.

As one of the most important nanostructures of Si, Si nanocrystals (NCs) exhibit remarkable optical properties [32], [33], [34], [35], [36], [37], [38], [39]. Especially, it has been recently found that Si NCs can effectively absorb light in a broad wavelength region from the UV to near-infrared (NIR) after they are heavily doped [40], [41], [42]. This lays a solid foundation for the exploration of using Si NCs to fabricate synaptic devices that can be optically stimulated in the broad UV-to-NIR region. In addition, it is well known that the wavelength of light signals in optical communication is usually in the NIR region [43]. Si-NC-based synaptic devices that can be stimulated by NIR light may significantly facilitate the coupling of neuromorphic computing with optical communication. Hence, it is of great importance to develop optically stimulated synaptic devices based on Si NCs, fulfilling the full potential of the material of Si in neuromorphic computing.

In this work we have used Si NCs heavily doped with boron (B) to fabricate optically stimulated synaptic devices. When pulsed optical signals in the broad UV to NIR region are used as input spikes, these Si-NC-based synaptic devices exhibit important synaptic functionalities such as excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), short-term plasticity (STP) to long-term plasticity (LTP) transition and spike-timing-dependent plasticity (STDP). It is found that the device operation is basically governed by the electronic and optical behavior of Si NCs.

Section snippets

Results and discussion

Fig. 1a schematically shows a biological neural system with typical synaptic structures. Information may be transmitted from a presynaptic axon terminal to a postsynaptic dendrite terminal through electrical or electrochemical signals [7], [8]. The connection strength between the presynaptic axon terminal and postsynaptic dendrite terminal is called synaptic weight. The change of the synaptic weight is known as synaptic plasticity, which underpins the implementation of important cognitive

Conclusions

In summary, we have fabricated Si-NC-based optoelectronic synaptic devices, which effectively work upon the spiking of light with the wavelength broadly ranging from the UV to NIR (∼ 2 µm). A series of important synaptic functionalities have been realized in these optically stimulated Si-NC-based synaptic devices, the energy consumption of which may be rather low. Especially, the successful demonstration of STDP has the implication for the development of an artificial neural network containing

Synthesis of Si NCs

Si NCs were synthesized by using a nonthermal plasma system. A mixture of 12 sccm (standard cubic centimeter per minute) SiH4 (20% in SiH4/Ar), 157 sccm B2H6 (0.5% in B2H6/Ar) and 105 sccm Ar was introduced into the nonthermal plasma system. The pressure of the plasma was maintained at ~ 6 mbar during the synthesis of Si NCs. The plasma was generated with a 13.56 MHz power source and a matching network. The power coupled into the plasma was ∼150 W. The colloid of Si NCs with a concentration of

Acknowledgements

Prof. Qing Wan at Nanjing university is thanked for insightful discussion. The High Magnetic Field Laboratory of the Chinese Academy of Sciences (CAS) is acknowledged for the EPR measurement. Mr. Hongjiang Li, Mr. Guoliang Wang and Prof. Weijie Song at the Ningbo Institute of Materials Technology and Engineering of CAS are thanked for the assistance in the device fabrication. This study is mainly supported by the National Key Research and Development Program of China under Grant 2017YFA0205700

Hua Tan received his B.S. degree in the Department of Materials Science and Engineering from Huazhong University of Science and Technology in 2015. He is currently a M.S. student of Prof. Xiaodong Pi in State Key Laboratory of Silicon Materials at Zhejiang University. His research interest is focused on silicon nanocrystals and their applications for optoelectronic devices such as synaptic devices.

References (74)

  • C. Rocks et al.

    Nano Energy

    (2018)
  • Y. Ding et al.

    Nano Energy

    (2014)
  • S. Zhao et al.

    Nano Energy

    (2016)
  • W. Xu et al.

    Nano Energy

    (2018)
  • M.M. Waldrop

    Nature

    (2016)
  • N.K. Upadhyay et al.

    Sci. China Inf. Sci.

    (2016)
  • G. Indiveri et al.

    Proc. IEEE

    (2015)
  • D.A. Drachman

    Neurology

    (2005)
  • L.F. Abbott et al.

    Nature

    (2004)
  • Z.G. Cheng et al.

    Sci. Adv.

    (2017)
  • D. Kuzum et al.

    Nanotechnology

    (2013)
  • R.S. Zucker et al.

    Annu. Rev. Physiol.

    (2002)
  • D.O. Hebb et al.
    (1949)
  • L.F. Abbott et al.

    Nat. Neurosci.

    (2000)
  • S. Choi et al.

    Nat. Mater.

    (2018)
  • Z. Wang et al.

    Nat. Mater.

    (2017)
  • H. Tian et al.

    ACS Nano

    (2017)
  • T. Ohno et al.

    Nat. Mater.

    (2011)
  • T. Tuma et al.

    Nat. Nanotechnol.

    (2016)
  • A. Chanthbouala et al.

    Nat. Mater.

    (2012)
  • L.Q. Zhu et al.

    Nat. Commun.

    (2014)
  • X. Yan et al.

    Adv. Funct. Mater.

    (2018)
  • J. Shi et al.

    Nat. Commun.

    (2013)
  • L. Wang et al.

    Adv. Electron. Mater.

    (2017)
  • Z. Wang et al.

    Adv. Electron. Mater.

    (2017)
  • L. Wang et al.

    Adv. Electron. Mater.

    (2017)
  • F. Alibart et al.

    Adv. Funct. Mater.

    (2012)
  • C. Diorio et al.

    T. Electron Dev.

    (1996)
  • M.R. Azghadi et al.

    Proc. IEEE

    (2014)
  • S. Qin et al.

    2D Mater.

    (2017)
  • K. Pilarczyk et al.

    Adv. Electron. Mater.

    (2016)
  • Z. Xiao et al.

    Adv. Electron. Mater.

    (2016)
  • X. Zhu et al.

    ACS Nano

    (2018)
  • H.K. Li et al.

    J. Appl. Phys.

    (2016)
  • M. Lee et al.

    Adv. Mater.

    (2017)
  • C.M. Hessel et al.

    Chem. Mater.

    (2011)
  • U.R. Kortshagen et al.

    Chem. Rev.

    (2016)
  • Cited by (147)

    • ZnO photoconductive synaptic devices for neuromorphic computing

      2023, Materials Science in Semiconductor Processing
    View all citing articles on Scopus

    Hua Tan received his B.S. degree in the Department of Materials Science and Engineering from Huazhong University of Science and Technology in 2015. He is currently a M.S. student of Prof. Xiaodong Pi in State Key Laboratory of Silicon Materials at Zhejiang University. His research interest is focused on silicon nanocrystals and their applications for optoelectronic devices such as synaptic devices.

    Zhenyi Ni received his Ph.D. degree in the School of Materials Science and Engineering at Zhejiang University in 2016. He is now a postdoctor working with Prof. Xiaodong Pi in the State Key Laboratory of Silicon Materials and the School of Materials Science and Engineering at Zhejiang University. His current research is mainly focused on silicon-nanocrystal-based optoelectronic devices.

    Wenbing Peng got his B.S. degree in Materials Physics at Nanchang University in 2015. He is now a Ph.D. student of Prof. Xiaodong Pi in State Key Laboratory of Silicon Materials, Zhejiang University. He mainly works on the fabrication and physical analysis of low dimensional semiconductor materials by the use of scanning tunneling microscope.

    Sichao Du received his B.S. degree in Materials Physics from Northwestern Polytechnical University in 2005, Xi’an, China, and M.Phil. and Ph.D. degrees in Optoelectronics Devices and Materials from The Australian National University and The University of Sydney in 2009 and 2014, Australia. He is currently an assistant professor at the College of Information Science and Electronic Engineering, Zhejiang University, China. His research interest includes the design and fabrication of broadband infrared optoelectronics devices based on 2D materials and low-dimensional III-V semiconductors.

    Xiangkai Liu received his B.S. degrees in the Department of Materials Science and Engineering at Zhejiang University. He continued to be a M.S. degree student in the State Key Laboratory of Silicon Materials at Zhejiang University under the direction of Prof. Xiaodong Pi. His research interest is focused on the fabrication of silicon nanocrystals and their applications for light-emitting devices.

    Shuangyi Zhao obtained his B.S. degree in the Department of Material Science and Engineering at Shandong University in 2013. He is now a PhD student under the supervision of Prof. Xiaodong Pi in the State Key Laboratory of Silicon Materials and the School of Materials Science and Engineering at Zhejiang University. He currently explores the use of silicon nanocrystals for optoelectronics.

    Wei Li received his B.S. degree from University of Electronic Science and Technology of China in 2015. Now he is a M.S. degree student under the supervision of Prof. Yang Xu in the College of Information Science and Electronic Engineering at Zhejiang University. He is currently interested in 2D materials for applications in photodetectors and imaging sensors.

    Zhi Ye received the B.S. and M.S. degree from Xiangtan University, Hunan, China, in 2004 and 2007, respectively, and Ph.D degree in the Department of electronic and computer engineering from Hong Kong University of Science and Technology, Kowloon, Hong Kong, in 2012. His research interest concerns ZnO-based thin-film transistor structure, methods of fabrication, and ferroelectric thin-film memory.

    Mingsheng Xu earned his Ph.D. degree in Department of Electronic Engineering from The Chinese University of Hong Kong in 2003. Then he worked as a researcher at The University of Tokyo, a lecturer at Chiba university, and a researcher at National Institute for Materials Science (NIMS), Japan, respectively. Now he is a full professor in the College of Information Science and Electronic Engineering at Zhejiang University. His currently works on two-dimensional materials and devices, clean energy, and sensors.

    Yang Xu is a professor at the College of Information Science and Electronic Engineering, Zhejiang University, China. He received his BS degree in ECE from Tsinghua University in 2000, Beijing, and MS and PhD degrees in ECE from the University of Illinois at Urbana and Champaign in 2005 and 2009. He is a Fellow of Churchill College at the University of Cambridge, UK, and a visiting professor at UCLA. He is an IEEE Senior Member of Electron Devices Society. His current research concerns silicon-based invisible photodetectors and image sensors.

    Xiaodong Pi received his PhD degree in the Department of Physics at the University of Bath in 2004. He is now a Professor in the State Key Laboratory of Silicon Materials and the School of Materials Science and Engineering at Zhejiang University. His current research mainly concerns low-dimensional silicon and germanium materials and their applications for optoelectronics.

    Deren Yang, an academician of Chinese Academy of Sciences, is director of the State Key Laboratory of Silicon Materials, and director of the Institute for Semiconductor Materials at Zhejiang University. He received his PhD degree in 1991 in the State Key Laboratory of Silicon materials at Zhejiang University. In 1990's he worked in Japan, Germany and Sweden for several years as a visiting researcher. He has been engaged in the research on silicon materials for microelectronic devices, solar cells and nano-devices.

    1

    These authors contributed equally to this work.

    View full text