Development of data logging system for chemical mechanical polishing and its application for process control

Chemical-mechanical polishing (CMP) has been widely used in semiconductor production for a certain length period, but the process control of CMP still has a number of issues to be resolved. The process control of CMP has not been well established. A good process model is essential to a successful CMP integration into mass production. This paper describes the development of a data logging network system for a CMP process. This system helps engineers build proper process control models. This system consists of the following entities: CMP process equipment, several analog/digital I/O devices built into the equipment, film thickness and scratch measurement instrument, particle measurement facility, and an infrared thermometer. All of these are connected with a local area network.     

基于神经网络的CMP过程智能R2R预测控制

针对化学机械研磨(CMP)过程非线性、时变和不易在线测量的特性,提出了基于径向基函数(RBF)神经网络和微粒群(PSO)算法的CMP过程run-to-run(R2R)预测控制器NNPR2R。首先通过样本数据用减聚类算法和最小二乘法构建CMP过程的RBF神经网络预测模型,解决了复杂CMP过程难以建立精确数学模型的难题和提高了预测模型的精度。然后通过PSO算法滚动优化求取控制律,解决了基于导数的优化技术易于陷入局部最优的问题并提高了控制精度。仿真结果表明,CMP过程NNPR2R控制器的性能优于常规的EWMA方法,有效抑制了过程漂移和减小了不同批次间产品的差异,显著降低了材料去除率(MRR)的均方根误差。     

CMP抛光机抛光盘温度的精确控制

CMP的加工过程,是对晶圆表面进行平坦化的过程,在对晶圆表面材料进行去除当中产生了热量,造成温度的上升,为了获得均匀的抛光去除率,达到全局的平坦化,必须对温度采取精确控制,分析研究了CMP加工过程中热量的产生,并介绍了一种对温度进行控制的方法。     

Cu CMP过程中背压对膜厚一致性影响

在极大规模集成电路Cu布线的化学机械平坦化过程中,抛光后表面薄膜厚度的一致性是检验平坦化能力的重要参数。从抛光过程中晶圆所受背压方面着手,研究了在不同背压参数情况下抛光后表面薄膜的一致性,得出了在工作压力为103 mdaN/cm2(1 kPa=10 mdaN/cm2)时,最佳的背压参数为108 mdaN/cm2。采用此工艺参数进行抛光后得到的晶圆表面薄膜非均匀性为5.04%,即一致性可以达到94.96%,而且此时表面粗糙度为0.209 nm,从而得到了良好的抛光效果,为极大规模集成电路Cu布线化学机械平坦化的进一步发展提供了新的途径。     

Modeling of Wafer Topography's Effect on Chemical–Mechanical Polishing Process

Topography on the wafer surface has a great effect on chemical-mechanical polishing (CMP). In this paper, the performance of a CMP system is demonstrated to be approximately an linear time invariant (LTI) system. The effects of the low-frequency components and high-frequency components in the wafer topography on CMP process are investigated. The magnitude spectra and phase spectra of the system are obtained. A model for the effect of more complex topography is established based on Fourier transform. The influences of down force and pad stiffness on the performance of CMP with the same topography are also studied.     

基于神经网络MIV算法的烟叶加料均匀性预测分析

为了预测分析不同工况下烟叶加料的均匀性,文中提出一种基于神经网络MIV算法的烟叶加料工艺参数优化方法。通过对烟叶加料的工艺参数进行神经网络训练,利用MIV计算出各参数与加料均匀性的相关性,确定影响加料均匀性的关键工艺指标。文中进行了相应的试验证明,结果显示相对误差被控制在5%以内。由此结果推出影响加料均匀性的关键工艺指标为:排潮开度、工艺流量、气体压力及料液流量。其中,排潮开度、工业流量及料液流量与加料均匀性负相关,气体压力与加料均匀性正相关;经过MIV进行变量剔除的神经网络的预测性较好。     

Optimization of tribological properties of silicon dioxide during the chemical mechanical planarization process

Chemical mechanical planarization (CMP) has been proved to achieve excellent global and local planarity, and, as feature sizes shrink, the use of CMP will be critical for planarizing multilevel structures. Understanding the tribological properties of a dielectric layer in the CMP process is critical for successful evaluation and implementation of the materials. In this paper, we present the tribological properties of silicon dioxide during the CMP process. A CMP tester was used to study the fundamental aspects of the CMP process. the accessories of the CMP tester were first optimized for the reproducibility of the results. The coefficient of friction (COF) was measured during the process and was found to decrease with both down pressure and platen rotation. An acoustic sensor attached to this tester is used to detect endpoint, delamination, and uniformity. The effects of machine parameters on the polishing performance and the correlation of physical phenomena with the process have been discussed.     

神经网络动态逆在歼击机安全着陆中的控制

给出了基于神经网络动态逆的自适应跟踪控制方法,用以解决飞机着陆过程中的复杂非线性和出现舵机故障的情况.应用神经网络直接对非线性系统故障模型求逆,使得所设计的逆系统能够包含故障信息,克服了传统的控制设计中将过程模型线性化,从而将不可忽视的非线性关系用线性关系代替或忽略的弊端.对由于建模误差、不确定性因素等引起的非线性系统逆误差,通过自组织模糊小脑模型关节控制器(SOFCMAC)神经网络在线进行修正.并在此基础上对3个通道分别设计了参考模型和线性控制器,以实现对伪线性系统进行跟踪控制.通过将这种方法用于某型歼击机在着陆过程中发生平尾卡死故障控制的过程仿真,验证了该方法的可行性.     

On the Wafer/Pad Friction of Chemical–Mechanical Planarization (CMP) Processes—Part I: Modeling and Analysis

Friction characteristics between the wafer and the polishing pad play an important role in the chemical–mechanical planarization (CMP) process. In this paper, a wafer/pad friction modeling and monitoring scheme for the linear CMP process is presented. Kinematic analysis of the linear CMP system is investigated and a distributed LuGre dynamic friction model is utilized to capture the friction forces generated by the wafer/pad interactions. The frictional torques of both the polisher spindle and the roller systems are used to monitor in situ the changes of the friction coefficient during a CMP process. Effects of pad conditioning and patterned wafer topography on the wafer/pad friction are also analyzed and discussed. The proposed friction modeling and monitoring scheme can be further used for real-time CMP monitoring and process fault diagnosis.     

嵌入式闪存中浮栅多晶硅CMP制程的研究与改善

随着CMP技术的日益发展和闪存特征尺寸的越来越小以及对多晶硅表面形态及前后层次间套准要求的提高,这一技术也被用于嵌入式闪存产品中浮动栅多晶硅的平坦化。浮动栅多晶硅厚度及表面形态对器件的电性参数及后续工艺影响较大,因此怎样得到一个稳定、厚度均匀及表面形态佳的浮动栅多晶硅显得至关重要。文章就以在90nm嵌入式闪存开发浮动栅...     

High-quality conformal silicon oxide films prepared by multi-step sputtering PECVD and chemical mechanical polishing

High-quality conformal oxide films were obtained by using multi-step sputtering (MSSP) plasma enhanced chemical vapor deposition (PECVD) process with argon ion sputtering and chemical mechanical polishing (CMP). The repeated deposition by plasma enhanced chemical vapor deposition (PECVD) and anisotropic etching of oxide films by multi-step sputtering PECVD improve the step coverage and gap filling capability significantly. The argon plasma treatment enhances the binding energy of Si-O in the SiO₂ network, and the temperature dependence of stress for MSSP oxide film showed no hysteresis after the heating cycle up to 440 °C. The stress-temperature slope of MSSP oxide film was found to be much less than that of conventional PECVD oxide film. The slope for 1.1 μm thick film is about 5.8×10⁵ dynes/cm²/°C which is smaller than that of thermally grown oxide film. It seems that MSSP oxide film reduces stress-temperature hysteresis and becomes more dense and void-free in the narrow gaps with inter-metal spacing of 0.5 μm. After filling of the narrow gap, we adopted the CMP process for global planarization and obtained good planarization performance. The uniformity of the film thickness was about 4% and the degree of the planarization was over 95% after CMP process.     

Intelligent control of via formation by photosensitive BCB forMCM-L/D applications

Via formation is a critical process sequence in multichip module (MCM) manufacturing, as it greatly impacts yield, density, and reliability. To achieve low-cost manufacturing, modeling, optimization, and control of via formation are crucial. In this paper, a model-based supervisory control algorithm is developed and applied to reduce undesirable behavior resulting from various process disturbances. A series of designed experiments are used to characterize the via formation workcell (which consists of the spin coat, soft bake, expose, develop, cure, and plasma descum unit process steps). The output characteristics considered are film thickness, uniformity, film retention, and via yield. Sequential neural network process models are used for system identification, and hybrid genetic algorithms are applied to synthesize process recipes. Computer simulation results show excellent control of output response shift and drift, resulting in a reduction of process variation. The performance limits of the supervisory control system are investigated based on these simulation results. The control algorithm is verified experimentally, and the results show 82.6, 64.4, and 17.3% improvements in maintaining target via yield, film thickness, and film uniformity, respectively, as compared to open loop operation     

等离子体刻蚀过程的APC技术研究进展

先进过程控制(APC)是一个多层次的控制系统,其包含了实时的设备及工艺控制和非实时的RtR控制.APC引入等离子体刻蚀过程控制可极大提高刻蚀机的使用效率.针对等离子体刻蚀过程,本文分别从实时控制和RtR控制两个不同层次展开了论述,概述了RtR控制器中所使用的线性回归模型和神经网络模型.最后,讨论了APC技术的发展趋势.     

Application of neural network inverse control system in turbo decoding

Adaptive inverse control system can improve the performance of turbo decoding, and modeling turbo decoder is one of the most important technologies. A neural network model for the inverse model of turbo decoding is proposed in this paper. Compared with linear filter with its revision, the general relationship between the input and output of the inverse model of turbo decoding system can be established exactly by Nonlinear Auto-Regressive eXogeneous input （NARX） filter. Combined with linear inverse system, it has simpler structure and costs less computation, thus can satisfy the demand of real-time turbo decoding. Simulation results show that neural network inverse control system can improve the performance of turbo decoding further than other linear control system.     

直线电机教学实验系统的研制

随着城市轨道交通的发展,牵引用直线感应电机在我国得到越来越广泛的应用。本文阐述了一种直线感应电机实验系统,包括组成结构、DSP控制系统、电压、电流、速度和力测量等部分。在该实验系统上可实现速度、电流的闭环控制和对电机的力特性进行研究。作为例子,本文给出了一个基于神经网格控制算法的研究实例。     

CMP抛光运动机理研究

通过对硅晶片化学机械抛光过程中抛光运动机理的理论分析,研究了硅晶片的抛光运动特性,探讨了主要工艺参数对硅晶片化学机械抛光后晶片表面粗糙度和表面平整度、抛光均匀性的影响规律。     

化学机械抛光（CMP）过程中抛光盘温度控制的分析研究

在化学机械抛光（CMP）过程中,温度是影响晶片最终抛光效果的主要因素之一,采用合理的温度控制方法,把温度控制在一定的范围内才能满足化学机械抛光的工艺要求。通过化学机械抛光机理分析阐述了抛光过程中热量产生的根源,介绍了一种温度控制系统的设计原理,并通过抛光实验验证并说明了温度控制的必要性。     

飞机自动着陆的一种非线性鲁棒控制器设计

将一种直接基于非线性模型的带神经网络补偿信号的逆系统方法用于具有强非线性和受不确定扰动干扰的飞机自动着陆控制,并对神经网络补偿的方式进行了改进。采用神经网络补偿动态逆反馈线性化后伪系统的逆误差,使得非线性系统在参数受到摄动或外部扰动的情况下仍能保持良好的控制效果。可以证明该方法在理论上是收敛的,对于任意的状态初值和给定的期望输出信号,能控制到指定的精度。神经网络的权值是局部收敛的,同时大量仿真表明,经过较少的神经网络离线训练,即能够获得很好的控制效果。最后通过飞机着陆下滑段的仿真验证表明,该方法具有强的鲁棒性和良好的跟踪精度。     

Widrow-Hoff神经网络学习规则的应用研究

基于线性神经网络原理,提出线性神经网络的模型,并利用Matlab实现Widrow-Hoff神经网络算法。分析Matlab人工神经网络工具箱中有关线性神经网络的工具函数,最后给出线性神经网络在系统辨识中的实际应用。通过对线性神经网络的训练,进一步验证Widrow-Hoff神经网络算法的有效性,以及用其进行系统辨识的高精度拟合性。     

Linear and nonlinear ARMA model parameter estimation using anartificial neural network

Addresses parametric system identification of linear and nonlinear dynamic systems by analysis of the input and output signals. Specifically, the authors investigate the relationship between estimation of the system using a feedforward neural network model and estimation of the system by use of linear and nonlinear autoregressive moving-average (ARMA) models. By utilizing a neural network model incorporating a polynomial activation function, the authors show the equivalence of the artificial neural network to the linear and nonlinear ARMA models. They compare the parameterization of the estimated system using the neural network and ARMA approaches by utilizing data generated by means of computer simulations. Specifically, the authors show that the parameters of a simulated ARMA system can be obtained from the neural network analysis of the simulated data or by conventional least squares ARMA analysis. The feasibility of applying neural networks with polynomial activation functions to the analysis of experimental data is explored by application to measurements of heart rate (HR) and instantaneous lung volume (ILV) fluctuations