In-situ neural network process controller for copper chemical mechanical polishing

Process control is one of the key methods to improve manufacturing quality. This research proposes a neural network based run-to-run process control scheme that is adaptive to the time-varying environment. Two multilayer feedforward neural networks are implemented to conduct the process control and system identification duties. The controller neural network equips the control system with more capability in handling complicated nonlinear processes. With the system information provided by this neural network, batch polishing time (T) an additional control variable, can be implemented along with the commonly used down force (p) and relative speed between the plashing pad and the plashed wafer (v). Computer simulations and experiments on copper chemical mechanical polishing processes illustrate that in drafting suppression and environmental changing adaptation that the proposed neural network based run-to-run controller (NNRTRC) performs better than the double exponentially weighted moving average (d-EWMA) approach. It is also suggested that the proposed approach can be further implemented as both an end-point detector and a pad-conditioning sensor.     

Developing a neural network-based run-to-run process controller for chemical-mechanical planarization

A new neural network-based run-to-run process control system (NNRtRC) is proposed in this article. The key characteristic of this NNRtRC is that the linear and stationary process estimator and controller in the exponentially weighted moving average (EWMA) run-to-run control scheme are replaced by two multilayer feed-forward neural networks. An efficient learning algorithm inspired by the sliding mode control law is suggested for the neural network-based run-to-run controller. Computer simulations illustrate that the proposed NNRtRC performs better than the EWMA approach in terms of draft suppression and adaptation to environmental change. Experimental results show that the NNRtRC can precisely trace the desired target of material removal rate (MRR) and keep the within wafer non-uniformity (WIWNU) in an acceptable range.     

A neural-Taguchi-based quasi time-optimization control strategy for chemical-mechanical polishing processes

Besides the major factors such as the down force, back pressure and the rotating speed of wafer carrier, effect of polishing time is also an important issue in CMP processes. In this study, a neural-Taguchi method based on a cost-effective quasi time-optimisation technique for chemical-mechanical polishing (CMP) processes is developed. The key concept of this new technique is that an optimal process parameter set is obtained through a neural-network-simulated CMP process model. Under such an optimal parameter set, the desired material removal rate within-wafer-nonuniformity can be reached with the optimal polishing time. It has been proved by experiment that the proposed method can offer a better polishing performance while reducing the polishing time by 1/3.     

Flatness control lapping/polishing machine

A lapping/polishing machine has been newly designed and constructed to produce precise flat surfaces efficiently. The machine can control the shape of a base plate to within a few microns by varying pressure distribution and rotating speed. Stable floating and vibration suppression to less than 1 μm for the base plate are attained by using hydrodynamic bearings. Accurate and large loading are also attained by using parallel leaf springs. The base plate surface is lapped within a few microns by controlling the pressure distribution. In silicon wafer polishing, work plates are deformed due to polishing heat. Therefore, two-layer discs with a larger coefficient of thermal expansion in the upper layer were used. As a result, it has been proved that highly accurate polishing may be attained efficiently by using the machine     

Investigation of pad wear in CMP with swing-arm conditioning and uniformity of material removal

Chemical mechanical polishing (CMP) process plays the role of planarizing and smoothing the uneven layers after the material deposition process in the semiconductor industry. In this process, pad conditioning using a diamond disk is inevitable to attain a high material removal rate (MRR) and to ensure the stability of the process. Pad conditioning is performed for providing uniform surface roughness and opening up the glazed surfaces of the polishing pad. However, the uneven pad wear resulting from pad conditioning leads to changes in the uniformity of MRR and productivity of the device. In this study, we investigate the pad wear profile after swing-arm conditioning of the pad, based on measurements performed using a pad measurement system (PMS). Conditioning experiments are conducted with seven cases of profiles of the conditioner's duration time (PCDT). In all the cases, “W”-shaped pad profiles are generated through swing-arm conditioning. It is observed that a concave-shaped PCDT results in the lowest value of maximum pad wear rate. The average depth of pad wear (h_avg) is mainly related to the MRR, and the maximum depth of pad wear (h_max) and the horizontal distance from the wafer center to the position (e) where the maximum pad wear occurs affect the within-wafer non-uniformity (WIWNU). A concave-shaped PCDT results in longer life of the polishing pad by minimizing the variation in pad wear. This paper can provide a technical assistance in selecting the conditioning recipe and improving the lifetime of the polishing pad in the CMP process.     

Developing a neural network-based run-to-run process controller for chemical-mechanical planarization

A new neural network-based run-to-run process control system (NNRtRC) is proposed in this article. The key characteristic of this NNRtRC is that the linear and stationary process estimator and controller in the exponentially weighted moving average (EWMA) run-to-run control scheme are replaced by two multilayer feed-forward neural networks. An efficient learning algorithm inspired by the sliding mode control law is suggested for the neural network-based run-to-run controller. Computer simulations illustrate that the proposed NNRtRC performs better than the EWMA approach in terms of draft suppression and adaptation to environmental change. Experimental results show that the NNRtRC can precisely trace the desired target of material removal rate (MRR) and keep the within wafer nonuniformity (WIWNU) in an acceptable range.     

Mathematical modeling based on contact mechanism due to elastic and plastic deformation of pad asperities during CMP

Technologies in semiconductor industry have been developed into a three-dimensional multilayer wiring for high integration of devices. Chemical mechanical planarization (CMP) process is one of the key technologies for achieving multilayer wiring, which enables global planarization. In addition, highly integrated devices can be realized by increasing the depth of focus in the photolithography process. However, in the inter-layer dielectric (ILD) CMP of the transistor, the uppermost oxide layer has the step due to the arrangement of the devices. The ideal material removal mechanism is to gradually remove materials from the top of the step height which allows for global planarization. However, in the CMP of the patterned wafers, simultaneous polishing of the upper and lower layers occurs when the step height reaches a certain height. This means that the polishing is strongly dependent on the structural characteristics of the pattern. Especially, the difference in the material removal rate depending on the pattern density acts as a constraint in terms of device layout. Therefore, it is essential to develop an accurate prediction model of material removal rate as a function of pattern density, size and arrangement. This study aims to define the mathematical planarization model according to contact mode between a polishing pad and patterned wafer. Considering that the real contact area between the actual polishing pad and the wafer is about 1 %, the mathematical model is derived based on the microscopic deformation of the pad asperities, not the macroscopic deformation of the bulk pad. Finally, we describe the verification between the theoretical material removal rate model and step height reduction and the actual CMP results. The root mean square error of the upper layer material rate, the lower material removal rate, and the step height reduction were 24.59 nm/min, 22.03 nm/min and 22.6 nm, respectively. Compared with the previous studies, the new model of this study improved the error by up to 50.9 %.

     

Viscoelastic behavior of polishing pad: Effects on edge roll-off during silicon wafer polishing

In semiconductor device fabrication, surface flatness of silicon wafers has a significant impact on the chip yield. Hence, there is a strong demand to prevent the deterioration in surface flatness near the wafer edge due to edge roll-off during polishing. In the present study, we investigate the viscoelastic behavior of polishing pads and its effects on the uniformity of material removal distribution near the wafer edge. On the basis of the findings, we propose polishing conditions required to improve surface flatness near the wafer edge. The double-sided polishing experiments performed using silicon wafers reveal that the proposed polishing conditions effectively reduce edge roll-off.     

A study of material removal amount of sapphire wafer in application of chemical mechanical polishing with different polishing pads

The study mainly explores the fabrication mechanism for fabricating sapphire wafer substrate, by using chemical mechanical polishing (CMP) method. A slurry containing the abrasive particles of SiO₂ is used to contact with the sapphire substrate polish and to produce chemical reaction for removal of sapphire wafer substrate when CMP method is used. The study observes the changes of the removal amount of sapphire wafer substrate when the pattern-free polishing pad and hole-pattern polishing pad are used under different down forces, polishing velocities, abrasive particle sizes and slurry concentrations. Employing regression analysis theory, the study makes improvement of the equation of material removal rate (MRR) to be the material removal height per 30 minutes (MRRh), and develops a compensation parameter C_rv of the error caused by the volume concentration of slurry. The results of experimental analysis show that under a certain down force, if the polishing velocity is greater, the material removal amount will be greater. Generally speaking, the material removal amount of hole-pattern polishing pad is greater than that of pattern-free polishing pad. As to the relationship between abrasive particle size and slurry concentration, when particle size is smaller, the volume concentration of slurry will be higher, and the number of abrasives for polishing wafer will be greater. As a result, a better material removal depth can be acquired. Through the above analytical results, considerable help is offered to the polishing of sapphire wafer.     

Hydrodynamic analysis of chemical mechanical polishing process

Chemical Mechanical Polishing (CMP) refers to a material removal process done by rubbing a work piece against a polishing pad under load in the presence of chemically active abrasive containing slurry. The CMP process is a combination of chemical dissolution and mechanical action. The mechanical action of CMP involves hydrodynamic lubrication. The liquid slurry is trapped between the work piece (wafer) and pad (tooling) forming a lubricating film. For the first step to understand the mechanism of the CMP process, hydrodynamic analysis is done with a semiconductor wafer. Slurry pressure distribution, resultant forces and moments acting on the wafer are calculated in typical conditions of the wafer polishing, and then nominal clearance of the slurry film, roll and pitch angles at the steady state are obtained.     

超声椭圆振动-化学机械复合抛光工具及实验研究

针对硅片的传统化学机械抛光,特别是随着硅片直径不断增大,硅片抛光表面质量和抛光效率成为一个亟待解决的问题。介绍一种超声椭圆振动-化学机械复合抛光新技术,并对抛光机理,抛光工具设计,性能及检测进行了详细论述和说明。在上述研究工作基础上,建立了超声椭圆振动辅助抛光实验系统,并进行了实验研究。实验结果显示,在抛光加工中引入超声椭圆运动后,工件抛光表面粗糙度值由传统抛光法的Ra0.077μm降低到Ra0.032μm,材料去除率最多可提高24%,且工件表面形貌有明显改善。     

Composite adaptive control of belt polishing force for aero-engine blade

The existing methods for blade polishing mainly focus on robot polishing and manual grinding.Due to the difficulty in high-precision control of the polishing force,the blade surface precision is very low in robot polishing,in particular,quality of the inlet and exhaust edges can not satisfy the processing requirements.Manual grinding has low efficiency,high labor intensity and unstable processing quality,moreover,the polished surface is vulnerable to burn,and the surface precision and integrity are difficult to ensure.In order to further improve the profile accuracy and surface quality,a pneumatic flexible polishing force-exerting mechanism is designed and a dual-mode switching composite adaptive control(DSCAC) strategy is proposed,which combines Bang-Bang control and model reference adaptive control based on fuzzy neural network(MRACFNN) together.By the mode decision-making mechanism,Bang-Bang control is used to track the control command signal quickly when the actual polishing force is far away from the target value,and MRACFNN is utilized in smaller error ranges to improve the system robustness and control precision.Based on the mathematical model of the force-exerting mechanism,simulation analysis is implemented on DSCAC.Simulation results show that the output polishing force can better track the given signal.Finally,the blade polishing experiments are carried out on the designed polishing equipment.Experimental results show that DSCAC can effectively mitigate the influence of gas compressibility,valve dead-time effect,valve nonlinear flow,cylinder friction,measurement noise and other interference on the control precision of polishing force,which has high control precision,strong robustness,strong anti-interference ability and other advantages compared with MRACFNN.The proposed research achieves high-precision control of the polishing force,effectively improves the blade machining precision and surface consistency,and significantly reduces the surface roughness.     

A study on the stress and nonuniformity of the wafer surface for the chemical-mechanical polishing process

In this paper, a two-dimensional axisymmetric quasic-static model for the chemical-mechanical polishing process (CMP) was established. Based on the principle of minimum total potential energy, a finite element model for CMP was thus established. In this model, the four-layer structures including the wafer carrier, the carrier film, the wafer and the pad are involved. The von Mises stress distributions on the wafer surface were analysed, and the effects of characteristics of the pad and the carrier film and the load of the carrier on the von Mises stress and nonuniformity on the wafer surface were investigated. The findings indicate that the profile of the von Mises stress distributions correlates with the removal rate profile. The elastic modulus and thickness of pad and carrier load would significantly affect the von Mises stress and nonuniformity, but those of the film did not affect very much.     

硅片边缘化学机械抛光的微小去除量测量

针对硅片化学机械抛光工艺的材料去除量非常微小并难以测量的问题,本文介绍一种采用表面粗糙度测量仪,对硅片边缘化学机械抛光的材料去除量进行一种简易、快速的测量方法,且该方法同时还可准确地测量硅片边缘抛光表面粗糙度值。检测结果表明,本方法较好地解决了硅片边缘化学机械抛光表面检测问题。     

Estimating the mechanical properties of polyurethane-impregnated felt pads

Chemical mechanical polishing (CMP) is an essential process in semiconductor fabrication. The results of CMP process are determined with the selection of consumables and process parameters. The polishing pad transports the slurry to the interface between the polishing pad and wafer and obtains material removal planarity. The mechanical properties of the polishing pad should be studied to analyze the material removal mechanism of CMP because polishing pad deformation is directly related to material removal rate and its uniformity. Various studies have investigated the stress distribution of the CMP process by using the elastic modulus and Poisson’s ratio of the polishing pad. However, these aspects of polishing pad have not been fully elucidated. In this study, we estimated the mechanical properties of commercial polyurethane-impregnated felt pads by comparing the experimentally measured compressive deformation amounts with finite element analysis results.     

Numerical analysis of hydrodynamic process of circular-translational-moving polishing (CTMP)

By keeping a pad moving relative to a wafer along a circular path without rotation, we developed a polishing technique called circular-translational-moving polishing (CTMP), which permits multidirectional polishing of the work piece and thus bears the advantage of isotropic polishing and a potential increase of material removal rate (MRR) on the wafer. To illuminate the mechanisms of CTMP and determine the optimum process variables in a CTMP process, a three-dimensional hydrodynamic lubrication model for CTMP with a smooth and rigid pad under a quasi-stable state is established in a polar coordinate system. The model equations are further calculated numerically by the finite difference method. The instantaneous distribution of fluid pressure is obtained, which shows that a negative pressure exists. The reason for negative pressure in CTMP and its effect on polishing is discussed. Moreover, the nominal clearance of the fluid film, roll, and pitch angles under different working conditions are obtained in terms of the applied load, moments, and polishing velocity. The obtained numerical analysis can be used as guidance for choosing operation parameters in a practical CTMP application.     

Numerical analysis of hydrodynamic process of circular-translational-moving polishing (CTMP)

By keeping a pad moving relative to a wafer along a circular path without rotation, we developed a polishing technique called circular-translational-moving polishing (CTMP), which permits multidirectional polishing of the work piece and thus bears the advantage of isotropic polishing and a potential increase of material removal rate (MRR) on the wafer. To illuminate the mechanisms of CTMP and determine the optimum process variables in a CTMP process, a three-dimensional hydrodynamic lubrication model for CTMP with a smooth and rigid pad under a quasi-stable state is established in a polar coordinate system. The model equations are further calculated numerically by the finite difference method. The instantaneous distribution of fluid pressure is obtained, which shows that a negative pressure exists. The reason for negative pressure in CTMP and its effect on polishing is discussed. Moreover, the nominal clearance of the fluid film, roll, and pitch angles under different working conditions are obtained in terms of the applied load, moments, and polishing velocity. The obtained numerical analysis can be used as guidance for choosing operation parameters in a practical CTMP application.     

Grey forecasting run-to-run control system in copper chemical mechanical polishing

In this paper, an online grey forecasting run-to-run control system was proposed with the integration of run-to-run control system, recursive least-squares (RLS) algorithm, and grey forecasting model (GFM). One of the objectives of this study is to explore the possibility and feasibility of applying GFM to run-to-run control system in copper chemical mechanical polishing. Under the condition of limited experiment data, GFM is excellent at estimating and forecasting error of the next batch online. To keep the process under control, the controllers are then employed to adjust the process parameters in order to compensate the error. In addition, the RLS algorithm is used to construct dynamically a system estimation matrix for the purpose of stating precisely the relationship between process quality and process parameters, and to consequently improve processing performances. From the computer simulation and the experiment results, the proposed new method developed in this study was able not only to confine the processing performances’ error within the range of 5% but also to supplement, when the process parameters are saturated, the control capability through adjusting other unsaturated process parameters, thus maintaining good processing performances.     

Abrasive removal depth for polishing a sapphire wafer by a cross-patterned polishing pad with different abrasive particle sizes

The paper establishes a new theoretical model for abrasive removal depth for polishing a sapphire wafer using chemical mechanical polishing with a polishing pad that has a cross pattern. The theoretical model uses binary image pixel division to calculate the pixel polishing times. An abrasive contact model for single-pixel multiple abrasive particles, to estimate the contact force between a single abrasive particle and the wafer, is then established. When the contact force is calculated, it is possible to calculate the abrasive depth of a single abrasive particle on the surface of the sapphire wafer. Using this theoretical model, carring a numerical simulation with a slurry of the same concentration, but with different abrasive particle diameters, determines the removal volume and average abrasive removal depth at each pixel position and the surface condition of the wafer. The simulation result is also compared with experimental data, in order to verify that the new model is feasible.     

光学玻璃磁性复合流体抛光应力与材料去除率的实验研究

根据光学玻璃元件超精密加工技术的需求,研究自旋转式和行星旋转式磁性复合流体(MCF)抛光的应力分布和材料去除率。首先,设计可实现自旋转和行星旋转抛光装置,搭建抛光实验平台;然后,进行自旋转式和行星旋转式MCF抛光实验,通过自行设计抛光应力分布测试实验分析了两种抛光方式的应力分布规律;最后,通过定点抛光实验,对抛光前后的工件表面轮廓进行检测,计算并分析两种抛光方式的材料去除率。实验结果表明,立式的两种抛光方式,正应力均明显大于剪切应力,工件外侧受到的剪切应力大于中心受到的剪切应力,行星式抛光的材料去除率明显大于自旋转式抛光的材料去除率。