仿神经元突触行为的碳纳米管晶体管制备技术的研究
发布时间：2018-11-25 13:35
【摘要】：随着科学技术的飞速发展,人们需要处理的问题日趋复杂,所需要处理的数据量呈指数型增长,因此人们对计算机性能要求也飞速提高,计算机需要面对越来越多的数据结构不清晰、逻辑结构不明确的问题,而基于冯诺依曼体系的传统计算机在处理这种类型的问题时表现的十分低效和无力。因此一种可以模拟人脑计算行为的计算机概念在被提出后便立即吸引了全球各界的广泛关注。这种新型的类脑计算机可以通过反复的学习来快速掌握海量数据中的内在规律,从而高效的解决传统计算机不能很好解决的问题,故而被广泛认为是计算机未来的发展趋势。神经元是人脑的最基本单元,而连接神经元的突触结构是人脑实现信号处理和记忆功能的关键。故而突触晶体管作为类脑计算的基础单元,将成为构架类脑计算机的底层基础。本文围绕基于碳纳米管网络的薄膜晶体管进行深入研究,成功制备出具有类似神经元突触功能的碳纳米管晶体管。本工作有三个突出特点:一是采用无结型的薄膜晶体管,源漏电极和沟道之间没有PN结,减小了短沟道效应、降低了阈值电压,并且制作工艺简单;二是使用电双层栅介质,通过室温条件下PECVD法制作掺磷的二氧化硅纳米颗粒薄膜,这种结构薄膜内嵌有大量的离子具有电双层效应;三是制备的基于碳纳米管网络的薄膜晶体管,可以模拟生物神经元突触的基本功能,如兴奋性突触后电流和记忆特性等。基于以上这些特点,这种碳纳米管晶体管具有开态电流大、开关比高、器件噪声小、制备简单、记忆窗口大等优势。
[Abstract]:With the rapid development of science and technology, the problems that people need to deal with are becoming more and more complicated, and the amount of data needed to be processed increases exponentially. The computer has to face more and more problems of unclear data structure and unclear logical structure, while the traditional computer based on von Neumann system is very inefficient and powerless in dealing with this kind of problems. Therefore, the concept of computer which can simulate human brain computing behavior has attracted wide attention from all over the world immediately after it was proposed. This new type of brain-like computer can quickly grasp the internal rules of massive data through repeated learning, thus solving problems which can not be solved very well by traditional computer, so it is widely considered as the future development trend of computer. Neuron is the most basic unit of human brain, and the synaptic structure of connecting neuron is the key of signal processing and memory function in human brain. Therefore, synaptic transistors, as the basic unit of brain-like computing, will become the underlying foundation of the framework brain-like computer. In this paper, the thin film transistors based on CNT network are studied, and the CNTs with neuronal synaptic function are successfully fabricated. There are three outstanding features in this work: first, there is no PN junction between the source and leakage electrode and the channel, which reduces the short channel effect and the threshold voltage, and the fabrication process is simple; The second is to fabricate the phosphorus-doped silicon dioxide nanocrystalline film by PECVD method at room temperature in the electrically double-layer gate dielectric. A large number of ions embedded in the structure film have the electric double-layer effect. Thirdly, the thin film transistors based on CNT network can simulate the basic functions of synapses of biological neurons, such as excitatory postsynaptic currents and memory properties. Based on these characteristics, the CNT has the advantages of high on-state current, high switching ratio, low device noise, simple fabrication and large memory window.
【学位授予单位】：清华大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN321.5
本文编号：2356289
[Abstract]:With the rapid development of science and technology, the problems that people need to deal with are becoming more and more complicated, and the amount of data needed to be processed increases exponentially. The computer has to face more and more problems of unclear data structure and unclear logical structure, while the traditional computer based on von Neumann system is very inefficient and powerless in dealing with this kind of problems. Therefore, the concept of computer which can simulate human brain computing behavior has attracted wide attention from all over the world immediately after it was proposed. This new type of brain-like computer can quickly grasp the internal rules of massive data through repeated learning, thus solving problems which can not be solved very well by traditional computer, so it is widely considered as the future development trend of computer. Neuron is the most basic unit of human brain, and the synaptic structure of connecting neuron is the key of signal processing and memory function in human brain. Therefore, synaptic transistors, as the basic unit of brain-like computing, will become the underlying foundation of the framework brain-like computer. In this paper, the thin film transistors based on CNT network are studied, and the CNTs with neuronal synaptic function are successfully fabricated. There are three outstanding features in this work: first, there is no PN junction between the source and leakage electrode and the channel, which reduces the short channel effect and the threshold voltage, and the fabrication process is simple; The second is to fabricate the phosphorus-doped silicon dioxide nanocrystalline film by PECVD method at room temperature in the electrically double-layer gate dielectric. A large number of ions embedded in the structure film have the electric double-layer effect. Thirdly, the thin film transistors based on CNT network can simulate the basic functions of synapses of biological neurons, such as excitatory postsynaptic currents and memory properties. Based on these characteristics, the CNT has the advantages of high on-state current, high switching ratio, low device noise, simple fabrication and large memory window.
【学位授予单位】：清华大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN321.5
【相似文献】
相关会议论文 前2条
1 张涛;张显峰;周秀萍;于如同;;Geranylgeranyltransferase Ⅰ对神经元突触发育的作用研究[A];2011中华医学会神经外科学学术会议论文汇编[C];2011年
2 刘深泉;;神经元突触模型的建立和数值分析[A];第十一次中国生物物理学术大会暨第九届全国会员代表大会摘要集[C];2009年
相关博士学位论文 前1条
1 陈婵;Rho激酶在中枢炎性脱髓鞘病变轴突生长和突触形成中的作用机制研究[D];复旦大学;2012年
相关硕士学位论文 前2条
1 李兆强;基于忆阻器和FPGA的仿生电路设计[D];河北工业大学;2015年
2 尹长青;仿神经元突触行为的碳纳米管晶体管制备技术的研究[D];清华大学;2015年
，本文编号：2356289
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2356289.html
