Solution-based Transition Metal Dichalcogenides for Flexible Neuromorphic Electronics

点击翻译按钮获取中文摘要

The human brain is an amazing computing machine that can process and store vast amount of information using no more energy than a 20-watt lightbulb. Modern computing systems, despite out-performing the human brain in examples like Alpha go, consume far too much energy in shuffling data between separated storage and computing units. Inspired by the human brain, where over 10^12 neurons and 10^15 synapses can process and store data concurrently with extremely low energy consumption, electronic devices that emulate biological elements, such as synapses and neurons, demonstrate great promise for neuromorphic computing. Such a neuromorphic paradigm is even more meaningful in the development of many new-concept flexible electronic systems such as wearable electronics, bionic sensors and brain-inspired chips where the energy budget is significantly constrained. Like graphene, transition metal dichalcogenide (TMDC) are a family of 2D materials whose three-atom thick unit cell is formed by a layer of transition metal atoms (Mo, W, etc.) sandwiched between two layers of chalcogen atoms (S, Se, Te). TMDCs exhibit extraordinary electronic and optical properties, making them appealing for applications spanning from nanoelectronics and nanophotonics to nanosensing. In addition, these atomic sheets can withstand mechanical strains of 10%, which makes these materials particularly suitable for flexible electronic devices, a market expected to be worth more than £10B in the next five years. Inspired by the human brain, this project will develop TMDC-based memristor devices that combine state-of-the-art performance together with scalable, industrially acceptable processing on flexible substrates. These memristors, fabricated from state-of-art fabrication technologies in nanoscales, can server as artificial synapses and neurons to faithfully mimic the biological neuronal system and perform a variety of computing tasks such as pattern and voice recognition, data analysis, process optimisation at extremely low power. The capability of integrating them onto flexible substrates further opens their application in the fast-growing world where real-time data of human and items are increasingly demanded. One key challenge here is to demonstrate the feasibility in large-scale deposition of low-dimensional TMDCs at a low temperature that is compatible with flexible substrates. Our recent work has demonstrated a breakthrough in this area through the development of a novel group of chalcogenide chemical precursors that decompose at low temperature. We will use the same chemical group to develop precursors that are capable of depositing TMDCs via low-temperature solution-processed approaches. We will also apply the high quality, large-scale and ultra-low dimensional TMDCs films into developing two-terminal memristors that can faithfully emulate the synaptic behaviour of human synapse and neurons. We will further improve memristor performance and enrich its functionality through defect engineering and composition modulation. The resulting device would have a significant impact on flexible neuromorphic electronics, and open up new and interesting applications for their deployment in skin-attachable and implantable neuromorphic electronics for wearable computing, health monitoring, and sensorimotor neural signal transmission.