2022
DOI: 10.1039/d2na00146b
|View full text |Cite
|
Sign up to set email alerts
|

A flexible dual-gate hetero-synaptic transistor for spatiotemporal information processing

Abstract: Artificial synapses based on electrolyte gated transistors with conductance modulation characteristics have demonstrated their great potential in emulating the memory functions in the human brain for neuromorphic computing. While previous...

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
16
0
1

Year Published

2023
2023
2024
2024

Publication Types

Select...
8

Relationship

2
6

Authors

Journals

citations
Cited by 18 publications
(19 citation statements)
references
References 38 publications
(56 reference statements)
0
16
0
1
Order By: Relevance
“…NNs are usually constructed based on artificial synaptic devices, and then further applied in neuromorphic computing. Pattern recognition is a typical neuromorphic computing [70,71], which requires that the NNs constructed by synaptic devices can identify the typical features of digital images, and the recognition accuracy is high [72,73]. Both NW-based synaptic memristors and NW synaptic transistors have been used for pattern recognition.…”
Section: Neuromorphic Computing Based On Nanowire Synaptic Devicesmentioning
confidence: 99%
“…NNs are usually constructed based on artificial synaptic devices, and then further applied in neuromorphic computing. Pattern recognition is a typical neuromorphic computing [70,71], which requires that the NNs constructed by synaptic devices can identify the typical features of digital images, and the recognition accuracy is high [72,73]. Both NW-based synaptic memristors and NW synaptic transistors have been used for pattern recognition.…”
Section: Neuromorphic Computing Based On Nanowire Synaptic Devicesmentioning
confidence: 99%
“…(g) EPSCs excited by the presynaptic spikes applied on the gates. (h) Time-dependent read current of the electrolyte gated transistor after different stimuli conditions [ 91 ].…”
Section: Neuromorphic Functions Emulated With Egtsmentioning
confidence: 99%
“…The multi-terminals can also act as separate synaptic inputs that interact with each other to complete the synergistic learning and memory task. Liu et al [ 91 ] reported an artificial heterosynapse based on coplanar dual-gate with spatiotemporal information integration and storage capability ( Figure 3(e,f) ). They show that the electric pulses applied on the single gate or unsynchronized electric pulses applied on the dual gates only induce volatile conductance modulation effect for short-term memory emulation.…”
Section: Neuromorphic Functions Emulated With Egtsmentioning
confidence: 99%
“…PET 具有极高的透明度, 其长期使用温度在 80 ℃ 左右 [38] 。图 4(a)展示了基于 PET 衬底构建的突触晶 体管的实物照片 [39] 。2013 年,Wan 等通过简单的自 组装方法在 PET 衬底上构建了基于铟锌氧 (IZO)的 柔性突触晶体管 [40] of classification accuracy of two-gate heterogeneous synaptic transistor and single-gate homogeneous synaptic transistor networks [41] 2.4 基于云母衬底的突触晶体管 云母是一种众所周知的天然无机晶体,具有成 本低、柔韧性好、透光性高、介电强度高、热稳定 性好等优点 [42][43] ,已被广泛应用于电子工业领域。 例如,云母被用作射频设备的电容器、高压电气设 备的绝缘体以及薄层低热阻绝缘体 [44] 。有趣的是, 云母在原子水平上是平的,被视为原子力显微镜 (AFM)校准的标准样品 [43] 。更重要的是,云母耐高 温((600±20)℃),很多铁电薄膜需要经过高温 退火(>400℃),才能结晶成铁电相,所以云母作为 铁电晶体管器件的理想衬底, 引起了广泛关注 [45][46][47][48] 。 2020 年, Li 等 [45] 基于云母衬底制备了一个全无机柔 性人工突触晶体管(云母/SRO/PZT/IGZO 异质结 构)。从图 5(a)可以发现,该器件具有出色的突触 特性。不仅如此,该器件还显示出显著的机械灵活 性和高温可靠性。当弯曲半径减小到 4.0 mm,循环 测试达到 400 次;当测试温度达到 100 ℃时,器件 的突触行为几乎没有变化。此外,该器件的 MNIST 手写数字识别精度可达 94.4%。这项工作为实现灵 活的、低功耗的和高效的模拟大脑的神经形态计算 提供了一种很有前途的方法。2021 年,Duan 等 [49] 在云母衬底上构建了一个柔性透明的 VO2 Mott 晶 体管,用于模拟生物突触的功能。图 5(b)展示了该 突触晶体管的光学照片。随着多种基本突触功能的 实现,该突触晶体管成功模拟了一种重要的感觉神 经元。 图 5(c)为 EPSC 计数图, 用于评估疼痛强度。 更重要的是,该突触晶体管对弯曲变形具有较高的 耐受性,多导态在增强和下降特性上的周期变化保 持在 4.0%以内。而且图 5(d)的仿真结果显示,由该 突触晶体管构建的卷积神经网络可以达到较高的手 写数字识别精度(>95%),进一步表明该柔性装置适 用于神经形态计算。最近,Jeon 等 [50] 基于云母衬底 沉积的 Hf0.5Zr0.5O 薄膜成功制备了柔性铁电突触晶 体管。该晶体管经过 10,000 次弯曲循环后,电学性 能的变化可以忽略不计,具有优异的机械灵活性。 而且,该柔性突触具有线性和对称的增强和抑制特 性,模式识别准确率高达 91.44%。 Fig. 5 Research on synaptic transistor on mica substrate (a) Synaptic plasticity of synaptic transistors constructed on mica substrate [45] ; (b) Optical photographs of flexible VO2 transistors constructed on mica substrate; (c) EPSC count chart to assess pain intensity; (d) Simulation identification accuracy of ideal devices an d VO2 transistor devices [49] 2.5 基于纸基衬底的突触晶体管 纸作为人类历史上最伟大的发明之一,一直在 人类历史中发挥着重要的作用。随着科技的进步, 纸又被开发出了新的功能,比如作为柔性光电器件 的衬底 [51][52] ]。纸衬底具有柔软可弯曲、可循环利用、 价格低廉和可生物降解等优异特性。但也有一个致 命的缺陷--不耐高温,所以使用其制备突触晶体 管时,不仅在制备器件过程中,而且在测试使用过 程中,均以室温下为宜。早在 2014 年,就有研究者 在纸衬底上构建突触晶体管了 [52] 。Jiang 等 [52] 以纸 为衬底,制备了以质子酸掺杂的壳聚糖为栅介质, 氧化铟锌(IZO)为沟道材料的突触晶体管。图 6(a) 展示了该突触晶体管的原理示意图,图 6(b)为纸衬 底上弯曲突触晶体管的光学照片。由于具有脉冲电 压响应的质子在壳聚糖电介质中的运动,纸突触晶 体管成功模拟了生物突触功能,包括兴奋性突触后 电流、双脉冲易化、动态滤波和时空相关信号处理 (图 6(c,d))。他们的研究结果有力地证明了这种 纸基突触晶体管不仅对构建廉价和生物友好的人工 神经元网络很重要,而且对实现智能生物材料也很 有意义。另外,Wan 等 [53] 也基于纸衬底制备了以 SiO2 纳米颗粒薄膜为栅介质,IZO 为沟道材料的双 电层薄膜晶体管,研究结果表明,这种柔性 IZO 晶 体管可作为人工突触装置,并证明了突触的刺激响 应和短程突触可塑性功能。这种基于 IZO 的纸基双 电层薄膜晶体管可能在低功率纸电子、柔性人工突 触和神经形态系统中有潜在的应用。 图 6 基于纸衬底的突触晶体管的研究工作 [52] Fig.…”
Section: 基于 Pet 衬底的突触晶体管unclassified
“…Fig.4 Research on synaptic transistor on PET substrate (a) Physical photographs of the construction of synaptic transistors on a PET substrate; (b) The change of current with time after 20 (4.0 V, 20 ms) pulses were applied to the gate of the synaptic transistor when VDS = 0.5 V [39] ; (c) Schematic diagram of spatio-temporal signal integration and memory performed by heterogeneous synapses (with the introduction of noise signals during the transmission of the original signal); (d) Comparisonof classification accuracy of two-gate heterogeneous synaptic transistor and single-gate homogeneous synaptic transistor networks[41] …”
mentioning
confidence: 99%