【作者】 李煜；

【导师】 李瑞伟；

【作者基本信息】 清华大学 ， 电子科学与技术， 2004， 博士

【摘要】 论文首先考虑沟道效应,建立了离子注入机的设备模型,用Visual Basic编程实现了硅片表面离子注入工艺结果的分布图。主要成果:(1)用模拟试验的方法建立了各种注入条件下离子注入的模拟数据库。(2)编程实现了模拟数据库的排序以及模糊查询功能。(3)使用Visual Basic编程实现了离子注入工艺在硅片表面的均匀性模拟,即建立了离子注入的设备模型。论文第二部分建立了奇点分裂算法作为刻蚀剖面的模拟算法,采用Visual C++编程实现了干法刻蚀工艺的剖面形貌模拟。主要成果:(1)建立了用于刻蚀剖面模拟的奇点分裂数学模型,改进了刻蚀工艺剖面的模拟算法,对于刻蚀剖面模拟的准确性方面有较大意义。(2)通过理论分析以及典型刻蚀剖面的SEM试验照片,较好的解释了建立奇点分裂算法的实验依据及其优越性,提高了该算法的可信程度。(3)基于奇点分裂算法,使用Microsoft公司开发的Visual C++编程实现刻蚀剖面的刻蚀过程模拟,最终将刻蚀剖面结果以图形的方式直观的显示出来。论文的第三部分建立了刻蚀机的设备模型,并且使用matlab中的神经网络工具箱模拟了硅片表面的刻蚀工艺结果。主要成果:(1)选用cascade forward neural network作为刻蚀设备模拟的基本神经网络模型,试验得出优化的网络层次结构、传输函数。(2)将试验片的刻蚀设备输入参数以及工艺结果对应于上述神经网络模型的输入与输出,对优化后的神经网络进行训练,得到该刻蚀设备的神经网络工艺设备模拟模型。(3)将训练好的神经网络作为刻蚀设备的模型,模拟出硅片表面刻蚀速度工艺结果。以上一系列工作将前人所做的工作包括集成电路工艺控制以及神经网络算法结合起来向前发展了一步,得到了尚未给予足够重视的集成电路工艺的设备模型,丰富了TCAD领域的模拟模型。更多还原

【Abstract】 A equipment model of ion implant which thinks over the channel effects is established in This paper and it displays ion implant process result on silicon chip with Visual Basic programming. The main achievement is: (1) establish the simulation database in all implant condition by simulation experiment. (2) realize the function of database sorting and seeking. (3) simulate the uniformity of ion implant process on a chip by Visual Basic.In second part of this paper, Singular point Splitting Algorithm which can simulates the etch profile is set up by Visual C++. The main achievement is: (1) establish the Singular point Splitting Algorithm model which can simulate the etch profile, it improves the simulation algorithm of etch profile and has a quite meaning in veracity of etch profile simulation. (2) analyze the SEM photo of etch profile and explain the experimental support and advantage of Singular point Splitting Algorithm. (3) base on the Singular point Splitting Algorithm, we simulate the etch profile with Visual C++ and display the profile graphic.In the last part of this paper, equipment model of etch is established by neural network toolbox in Matlab. The main achievement is: (1) select the cascade forward neural network as the basic neural network model in etch equipment simulation and optimize the architecture and the transfer functions of neural network. (2) input the etch equipment parameter and the process results to train the neural network, then gain the neural network model of etch equipment. (3) simulate the etch rate on a chip by the trained neural network model.All the research combines the IC technology control with the neural network, achieves the equipment model in IC technology, enriches the simulation model in the field of TCAD.更多还原