Ultraflexible and Energy‐Efficient Artificial Synapses Based on Molecular/Atomic Layer Deposited Organic–Inorganic Hybrid Thin Films. (9th December 2022)

• Record Type: Journal Article
• Authors: Liu, Chang; Zhu, Lin; Weng, Jing‐Kai; Li, Yu‐Chen; Chen, Zhuo; Lei, Jin; Zi, Tao‐Qing; Wu, Di; Fa, Wei; Chen, Shuang; Liang, Bin; Li, Ai‐Dong
• Is Part Of: Advanced Electronic Materials. Volume 9:Number 2(2023)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-09
• Subjects: flexible artificial synapses -- low energy consumption -- molecular/atomic layer deposition -- organic–inorganic hybrid materials -- PEDOT:PSS electrodes; Materials -- Electric properties -- Periodicals; Materials science -- Periodicals; Magnetic materials -- Periodicals; Electronic apparatus and appliances -- Periodicals; 537
• DOI: 10.1002/aelm.202200821
• Languages: English
• ISSNs: 2199-160X

Abstract

Abstract: Flexible artificial synapses, a conjunctive product of brain‐inspired neuromorphic computing and wearable electronics, arouse enormous interest in highly connected and energy‐efficient neural networks. The organic–inorganic hybrid materials hold great potential in flexible devices due to versatile properties. Here, an organic–inorganic hybrid synaptic device consisting of 2 nm Al2 O3 and 22 nm Al‐based hydroquinone (Al‐HQ) sandwiched between Pt and poly(3, 4‐ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) electrodes is prepared on highly flexible cellophane by molecular/atomic layer deposition (MLD/ALD). The vertically integrated Pt/Al2 O3 /Al‐HQ/PEDOT:PSS device exhibits reliable resistive switching with an ON/OFF ratio greater than 10 3 . Several important bio‐synaptic functions, such as long‐term potentiation, long‐term depression, paired‐pulse facilitation, and spike‐time‐dependent plasticity, are realized in this device with the extremely low energy consumption of ≈25.2 fJ per reset operation, which is ascribed to the unique electron trapping/detrapping and tunneling mechanism. Remarkably, the excellent flexibility and robustness of this hybrid synaptic device is confirmed under the minimum curvature radius of ≈0.7 mm after 10 4 bending cycles. A pattern recognition computation based on these hybrid synapse devices shows a 90.2% learning accuracy. This research paves a way for the MLD/ALD‐derived organic–inorganic‐hybrid‐based artificial synapse applications in flexible energy‐efficient neuron network systems toward system‐on‐plastics. Abstract : Molecular/atomic layer deposited ultraflexible organic–inorganic hybrid synaptic devices are developed and evaluated for energy‐efficient neural network applications. These devices exhibit exceptional flexibility and robustness with extremely low energy consumption. The simple bilayer‐architecture and manufacture technology compatible with semiconductor processing enable hybrid devices to have great prospects in future flexible neuromorphic computing and wearable devices.