Characterization and Real Time Defect Mitigation in Chemical/Mechanical Polishing of Microelectronic Wafers Using Decision Theory and MultiSensor Fusion

这项研究的目的是应用多传感器融合和决策理论的原理，推导出量化关系，将微电子晶片化学机械平坦化过程中收集的各种传感器的信号特征，包括力、温度、振动和其他特征，与特定的缺陷演变联系起来，用于实时表面缺陷缓解和工艺控制。该方法将首先在晶片上诱导某些基本的缺陷图案，如压痕、使用纳米压痕和纳米划痕技术的划痕。然后，这些晶片将在装有多个无线微电子机械系统传感器的化学机械平整机上进行抛光。预计将针对化学机械平坦化过程中产生的不同缺陷开发缺陷敏感传感器特征档案。运用决策理论对缺陷进行预测和控制。每个时间点的控制动作(例如，下压力和压板速度的具体调整)将通过最大化效用函数来估计，从而使控制动作对噪声具有鲁棒性。化学机械平坦化平台将在可能的情况下配备多个基于有线和无线微电子机械系统的传感器，以监测温度、振动、化学(PH)和用于缺陷检测和缓解的力。如果成功，这项研究将促进行业采用基于微电子机械系统传感器的方法来解决缺陷和此类晶片产量的关键障碍。传感器网络化的化工机械平面化平台可以作为学生和行业人员的教学和培训的试验台。首席调查员在吸引来自不同背景的学生方面有着良好的记录，并计划与当地少数族裔机构合作，招收符合条件的学生参加这项研究项目。学生将通过大学和行业之间的相互访问，接触到先进制造工艺中的基础多学科研究和行业实践。研究的重点将在首席调查员的网站上公布，研究结果将在各种期刊、行业杂志上发表，并在国内和国际会议上发表演讲。

The objective of this research is to apply the principles of multi-sensor fusion and decision theory for deriving quantitative relationships connecting signal features from various sensors, including, force, temperature, vibration, and others, gathered during chemical mechanical planarization of microelectronics wafers, with specific defect evolutions, towards real-time surface defect mitigation and process control. The approach will be to first induce certain basic defect patterns on wafers, such as indentation, scratching using nanoindentation and nanoscratching techniques. These wafers then will be polished on a chemical mechanical planarization machine instrumented with multiple wireless microelectromechanical systems sensors. It is anticipated that an archive of defect-sensitive sensor features for different defects generated in chemical mechanical planarization will be developed. Decision theory will be used to predict and control the defects. The control action (for example, specific adjustment of down force and platen speed) at each time-epoch will be estimated by maximizing a utility function, so that the control actions are robust to noise. The chemical mechanical planarization platform will be equipped with multiple wired- and wireless-microelectromechanical systems-based sensors, where possible, to monitor temperature, vibrations, chemistry (pH), and forces for defect detection and mitigation.If successful, this research will facilitate industry adoption of micro-electromechanical systems sensor-based approaches to address defects and such critical impediments to wafer yield. The sensor-networked chemical mechanical planarization platform can be used as a test bed for instruction and training of students and industry personnel. The Principal Investigators have a track record for attracting students from diverse backgrounds, and plan to work with local minority institutions to recruit qualified students into this research project. The students will be exposed to fundamental multi-disciplinary research and industry practices in advanced manufacturing processes through mutual visits between university and industry. The highlights of the research will be made available on the website of the Principal Investigators and the results will be published in various journals, trade magazines, and presentations at national and international conferences.