近10年来,在神经形态工程的飞速发展进程中,基于硬件途径实现人造突触得到了充分的重视和研究,从模拟生物突触或神经元行为,到复杂的神经形 态计算 ,都取得了显著的进展。其中三端口晶体管由于具有稳定性好、测试参数相对可控、运行机制清晰、可由多种材料构成等优点在众多被构筑出来用以模拟突触行为的电子器件中脱颖而出。通过适当的材料选择和结构设计,三端/多端突触晶体管可以将外界的刺激(光、压力、温度等)转换成电信号,从而实现对外界环境直接响应的人工突触。此外,三端/多端突触晶体管还具有同时执行信号传输和自学习的优势,而且模拟需要多端操作的并行学习和树突整合可以很容易地在基于三端/多端突触晶体管的人工突触中实现,这为开发具有较少神经元件 的神经网络开辟了可能性。 本文讨论了包括铁电突触晶体管、双电层/电化学突触晶体管和光电突触晶体管在内的三端/多端晶体管近年来的最新进展。它们有各自的优缺点,其中铁电场效应晶体管具有编程速度快、无损读出、开关比大、低功耗等优点。然而部分改变铁电材料的极化状态通常需要较大的工作电压,铁电材料的稳定极化状态使其易于实现LTP,但是很难实现STP;双电层/电化学突触晶体管在实现逻辑功能、树突整合和人工树突神经元方面优于其他类型的器件。 此外,双电层/电化学晶体管的低压工作特性也为实现超低能耗的突触器件提供了可能。然而,器件的耐用性和电解质的不稳定性可能是双电层/电化学突触晶体管的主要限制;在光电突触晶体管方面,以光为输入信号的光电神经形态器件,不仅将视觉、信息处理和记忆结合在一起,而且具有带宽高、鲁棒性强、并行性好等优点,适用于模拟人眼视网膜神经元等功能。然而,利用光信号实现抑制性突触仍然是一个很大的挑战。这些器件各有优缺点,根据特定应用程序的要求,一种类型的器件可以优先于其他类型的器件。 神经形态工程旨在构建具有超低能耗的鲁棒类脑计算机,利用新兴的突触装置来实现人工神经网络已经有了很多成功的尝试,然而神经形态器件领域所面临的一些主要问题和挑战仍需要我们不断进取努力。例如:目前所报道的突触器件模拟的小部分突触行为,完整的人造突触功能还需进一步研究完善;对于能够模拟生物神经元信息处理功能的神经形态器件的研究还仅限于少数报道,迫切需要更 深入的研究;目前科学对人脑的功能和运行机制的了解还在初步阶段。虽然目前这一领域还处于起步阶段,但是在未来,期待利用大量人造突触晶体管模拟类似于感官(视觉、听觉、运动、嗅觉)这样复杂的人类神经系统。相信通过物理、化学、材料学、计算机和医学等科学领域的跨学科交流,类脑神经形态工程将获得不断的革新进步和应用,先进的人工智能系统将促进人类在服务行业、个人医疗、教育和交通等领域的生活。 文献引用: 朱力,万青. 神经形态晶体管研究进展[J]. 微纳电子与智能制造, 2019, 1(4): 39-50.Research progress of neuromorphic transistors[J]. Micro/nano Electronics and Intelligent Manufacturing, 2019, 1(4): 39-50.《微纳电子与智能制造》刊号:CN10-1594/TN主管单位:北京电子控股有限责任公司主办单位:北京市电子科技科技情报研究所北京方略信息科技有限公司投稿邮箱:tougao@mneim.org.cn(网站:www.mneim.org.cn) 参考文献: [1] BACKUS J. Can programming be liberated from the von Neumann style? a function style and its algebra of programs[J]. Communications of the ACM, 1978, 21(8): 613-641. [2] DRACHMAN D A. Do wo have brain to spare[J]. Neurology, 2005, 64 (12):2004-2005. [3] FURBER S. To build a brain[J]. Spectrum IEEE, 2012, 49(8): 44-49. [4]MARKRAM H. The blue brain project[J]. Nature Reviews Neuroscience, 2006, 7(2): 153-160.[5] MARKRAM H, GERSTNER W, SJÖSTRÖM P J. A history of spike-timing-dependent plasticity[J]. Frontiers in Synaptic Neuroscience, 2011, 3(4): 4. [6] CAPORALE N, DAN Y. Spike timing-dependent plasticity: a hebbian learning rule[J]. Annual Review of Neuroscience, 2008, 31(1): 25-46. [7] RACHMUTH G, SHOUVAL H Z, BEAR M F, et al. A biophysically- based neuromorphic model of spike rate- and timing-dependent plasticity[J]. Proceedings of the National Academy of Sciences, 2011, 108(49): E1266-74. [8] PREZIOSO M, MERRIKH-BAYAT F, HOSKINS B D, et al. Training and operaion of an integrated neuromorphic network based on metal- oxide memeristors[J]. Nature, 2015, 521(7550): 61-64. [9 ] XU W, LEE Y, MIN S Y, et al. Simple, inexpensive, and rapid approch to fabricate cross- shaped memristors using an inorganic- nanowire- alignment technique and a one- step reduction process[J]. Advanced Materials, 2016, 28(3): 527-532. [10] NISHITANI Y, KANEKO Y, UEDA M, et al. Three-terminal ferroelectric synapse device with concurrent learning function for artficial neural networks[J]. Journal of Applied Physics, 2012, 111(12): 124108-1-124108-6. [11] ZHU L Q, WAN C J, GUO L Q, et al. Artificial synapse network on inorganic proton conductor for neuromorphic systems[J]. Nature Communications, 2014, 5: 3158. [12] DAI S, WU X, LIU D, et al. Light- stimulated synaptic devices utilizing interfacial effect of organic field-effect transistors[J]. ACS Applied Materials & Interfaces, 2018, 10 (25): 21472-21480. [13] WANG K, DAI S, ZHAO Y, et al. Light-stimulated synaptic transistors fabricated by a facile solution process based on inorganic perovskite quantum dots and organic semiconductors[J]. Small, 2019, 15(11): 1900010-1-19000010-8. [14] MILLER S L, MCWHORTER P J. Physics of the ferroelectric nonvolatile memory field effect transistor[J]. Journal of Applied Physics, 1992, 72(12): 5999-6010. [15] KANEKO Y, NISHITANI Y, UEDA M. Ferroelectric artificial synapases for recognition of a multishaded image [J]. IEEE Transactions on Electron Devices, 2014, 61 (8): 2827-2833. [16] KIM M K, LEE J S. Ferroelectric analog synaptic transistors[J]. Nano Letters, 2019, 19(3): 2044-2050. [17] JANG S, JANG S, LEE E -H, et al. Ultrathin comformable organic artificial synapses for wearable intelligent device applicatios[J]. ACS Applied Materials & Interfaces, 2019, 11(1): 1071-1080. [18] MIN Y, AKBULULUT M, SANGORO J R, et al. Measurement of forces across room temperature ionic liquids between mica surfaces[J]. The Journal of Physical Chemistry C, 2009, 113(37): 16445-16449. [19] KIM S H, HONG K, XIE W, et al. Electrolyte-gated transistors for organic and printed electronics[J]. Advanced Materials, 2013, 25(13): 1822-1846. [20] KIM K, CHEN C L, TRUONG Q, et al. A carbon nanotube synapse with dynamic logic and learning[J]. Advanced Materials, 2013, 25(12): 1693-1698. [21] SHARBATI M T, DU Y, TORRES J, et al. Low- power, electrochemically tunable graphene synapses for neuromorphic computing[J]. Advanced Materials, 2018, 30 (36): 1802353. [22] JIANG J, GUO J, WAN X, et al. 2D MoS2 neuromorphic devices for brain- like computational systems[J]. Small, 2017, 13(29): 1900933-1-1700933-11. [23] HE Y L, NIE S, LIU R, et al. Spatiotemporal information processing emulated by multiterminal neuro- transistor networks[J]. Advanced Materials, 2019, 31(21): 1900903-1-19000903-8. [24] GKOUPIDENIS P, SCHAEFER N, GARLAN B, et al. Neuromorphic functions in PEDOT: PSS organic electrochemical transistors[J]. Advanced Materials, 2015, 27 (44): 7176-7180. [25] QIAN C, SUN J, KONG L A, et al. Artificial synapses based on in- plane gate organic electrochemical transistors[J]. ACS Applied Materials & Interfaces, 2016, 8 (39): 26169-26175. [26] KURAMOCHI E, NOZAKI K, SHINYA A, et al. Largescale integration of wavelength- addressable all- optical memories on a photonic crystal chip[J]. Nature Photonics, 2014, 8(6): 474-481. [27] GHOLIPOUR B, BASTOCK P, CRAIG C, et al. Amorphous metal- sulphide microfibers enable photonic synapses for brain-like computing[J]. Advanced Optical Materials, 2015, 3(5): 635-641. [28] LI H K, CHEN T P, LIU P, et al. A light-stimulated synaptic transistor with synaptic plasticity and memory functions based on InGaZnOx- Al2O3 thin film structure [J]. Journal of Applied Physics, 2016, 119(24): 244505- 1-244505-5. [29] DAI S L,WU X H, LIU D P, et al. Light-stimulated synaptic devices utilizing interfacial effect of organic fieldeffect transistors[J]. Applied Materials & Interfaces, 2018, 10(25): 21472-21480. [30] WANG Y, LV Z Y, CHEN J R, et al. Photonic synapses based on inorganic perovskite quantum dots for neuromorphic computing[J]. Advanced Materials, 2018, 30 (38): 1802883-1-1802883-9.