ITR: Feedback Control of Thin Film Microstructure Using Multiscale Distributed Models

研究：本研究计划的目标是开发一个综合的理论、计算和实验程序，用于使用多尺度分布式模型对薄膜化学气相沉积中的薄膜微结构进行基于非线性模型的反馈控制。多尺度分布式模型的分布式控制理论将被用来产生新的解析反馈控制器和估值器设计，以加强多尺度闭环系统中期望的稳定性、性能和鲁棒性指标，并实现具有期望特性(例如，降低粗糙度)的薄膜微结构。这项工作的动机是：a)对薄膜微结构控制的需求日益增加，b)薄膜微结构对操作条件的任意变化具有高度的敏感性，以及c)缺乏系统和全面的反馈控制框架来在线形成薄膜微结构。为了实现预期的目标，拟议的研究将集中于以下项目：1.开发计算高效和准确的多尺度分布模型降阶算法。多尺度降阶估值器和非线性反馈控制器的设计，能够以理想的方式塑造材料的微观结构。在线诊断技术与多尺度分布式模型的结合以及基本控制理论问题的研究。控制算法在模拟薄膜生长问题中的应用。在加州大学洛杉矶分校的实验室规模等离子体增强化学气相沉积过程中开发和实验应用实时集成测量/控制系统，以控制薄膜微结构和空间变化。影响：该研究将为多尺度分布式系统的模型降阶、优化和控制问题的本质提供基本的理解，开发易于在实践中实施的具体控制算法，说明控制方法的应用，并为控制器的实现导出调整指南。开发一套软件和与公司合作将是将研究成果转移到工业部门的手段。将这项研究整合到教育中将使教授过程控制和半导体制造高级课程的教育工作者受益。新的控制算法有望使我们在薄膜化学气相沉积过程中形成在线材料微结构的能力取得重大进展。信息技术(IT)既是拟议研究的主要推动者，也是受益者。一方面，信息技术的发展，如高置信度计算、传感、数据处理和软件系统的进步，为实现项目目标提供了不可或缺的机制。反过来，这项研究有可能通过改进的半导体制造技术进一步推动IT发展，半导体制造技术代表着现代IT基础设施的组成部分。

Research:The objective of this research program is to develop a comprehensive theoretical, computational and experimental program for nonlinear model-based feedback control of film microstructure in chemical vapor deposition of thin films using multi-scale distributed models. Distributed control theory for multi-scale distributed models will be developed and employed to produce novel analytical feedback controller and estimator designs that enforce the desired stability, performance and robustness specifications in the multi-scale closed-loop system and achieve thin film microstructure with desired characteristics (e.g., reduced roughness). The motivation for this work is provided by: a) the increasing need for control of thin film microstructure, b) the high-sensitivity of thin film microstructure with respect to arbitrary variations in operating conditions, and c) the lack of systematic and comprehensive framework for feedback control that can shape thin film microstructureon-line.To realize the desired objective, the proposed research will focus on the following projects:1. The development of computationally efficient and accurate algorithms for order reduction of multi-scale distributed models.2. The design of multi-scale reduced-order estimators and nonlinear feedback controllers that can shape material microstructure in a desirable way.3. The integration of on-line diagnostic techniques with multi-scale distributed models and the study of fundamental control theoretic issues.4. The application of the control algorithms to simulated thin film growth problems.5. The development and experimental application of a real-time integrated measurement/control system to a lab-scale plasma-enhanced CVD process at UCLA to control thin film microstructure and spatial variations.Impact:The research will provide a fundamental understanding of the nature of the model reduction, optimization and control problems for multi-scale distributed systems, develop concrete control algorithms that can be readily implemented in practice, illustrate the application of the control methods and derive tuning guidelines for the implementation of the controllers. The development of a software package and the collaboration with companies will be the means for transferring the results of the research to the industrial sector. The integration of the research into education would benefit educators teaching advanced-level classes in process control and semiconductor manufacturing. The new control algorithms are expected to lead to significant advances in our ability to shape, on-line, material microstructure in the chemical vapor deposition of thin films. Information Technology (IT) serves both as a major enabler and beneficiary of the proposed research. On the one hand, IT developments, such as advances in high-confidence computing, sensing, data processing, and software systems, provide an indispensable mechanism for realizing the project goals. In turn, the research has the potential to further advance IT developments through improved semiconductor manufacturing technologies that represent an integral part of modern-day IT infrastructure.