CAREER: Data Rectification, Process Monitoring, Fault Diagnosis, and their Integration by Multiscale Empirical Modeling

化工过程的有效运行依赖于从测量数据中提取有关过程的信息，整合各种操作任务，以及接受过先进工艺操作技术培训的工程师。高效的操作还需要工程师接受过高级过程操作、统计和随机建模方法、沟通和团队合作技能以及将化学过程视为一个综合系统的能力等方面的培训。该职业奖项目致力于通过针对实现这些目标的综合研究和教育活动来提高化学过程操作的效率。将开发一种新的通用多尺度方法，用于从测量数据中提取经验模型。该方法将任何线性或非线性经验建模方法转化为多尺度建模方法。多尺度建模将用于开发比现有方法性能更好的多尺度数据校正、过程监控和故障诊断技术。多尺度方法将允许这些操作任务的有效集成，从而更好地利用测量数据，更有效地操作，以及大规模操作的高效计算。这些方法将应用于与不同行业和联邦实验室合作的几个实际例子。本研究代表了开发多尺度经验建模的一般方法的努力，小波的大多数现有用途是用于多尺度分析或随机建模。研究结果将通过俄亥俄州立大学化学工程系现有的和新的课程、工业短期课程传播，并可能列入工艺操作教科书。将开发新的课堂和实验室实验以及网络资源，以改进工艺操作教育。如何提高学生的沟通和团队合作技能也将被研究。将开发模块来整合整个课程的过程设计和模拟原理，并训练化学工程师以系统的观点来看待过程，而不是单元操作的观点。当前和未来微电子和光电子器件制造的需求对金属沉积方案提出了严格的要求。这包括高薄膜纯度、低温和快速、可控的沉积速率。PI假设这些目标可以通过化学流体沉积（CFD）来实现，这是一种新的金属沉积方法，涉及在低温（40-80℃）的超临界二氧化碳中对可溶性有机金属化合物进行化学或热还原，以在无机或有机基底上产生连续薄膜。SCF溶剂位于液体和气体的中间，CFD利用SCF溶剂独特的、可调节的物理化学性质，规避了蒸汽和液相技术的局限性。在CFD中，与化学气相沉积（CVD）等气相技术相比，前驱体在更低的温度和更高的试剂浓度下在溶液中传输和还原。虽然CFD是一种基于溶液的过程，但SCF的“类气体”传输特性及其与气态还原剂（如氢）的混溶性，使得该过程不受与液相还原相关的传质差和沉积速率差的问题的影响。初步实验表明，在比CVD温度低170℃的温度下，SCF溶液可以在硅片和其他无机衬底上沉积高纯度、连续的铂和钯薄膜。该研究项目将重点关注二氧化碳溶液中氢解二甲基环氯丁定铂（II）沉积薄膜，以及铜（II）二（六氟乙酰丙酮）和铜（II）二（2,2,6,6-四甲基-3,5-庚二酸）还原沉积铜薄膜。选择前驱体是为了便于将CFD中的膜质量和还原动力学与现有技术的膜质量和还原动力学进行直接比较，以及金属沉积在微电子铜和催化铂器件中的潜在用途。工作的教育部分是：(1)培养将在工程和材料化学领域工作的研究生；(2)为本科生提供研究经验的机会；(3)开发材料加工的两门系列课程，以满足学生的兴趣，并通过化学工程系不断扩大的材料研究工作得到加强。(4)将研究问题和主动学习原则纳入课堂，并协助在整个课程中实施交互式教学工具。

Abstract - Bakshi - 9733627 Efficient operation of chemical processes rely on the extraction of information about the process from measured data, the integration of various operations tasks, and engineers trained in advanced techniques for process operation. Efficient operation also requires engineers trained in topics such as advanced process operation, statistical and stochastic modeling methods, communications and teamwork skills, and the capability to view a chemical process as an integrated system. This CAREER award project strives to improve the efficiency of chemical process operations by integrated research and educational activities directed towards achieving these goals. A new and general multiscale method will be developed for extracting empirical models from measured data. This method will transform any linear or nonlinear empirical modeling method to a multiscale modeling method. The multiscale modeling will be used to develop multiscale data rectification, process monitoring, and fault diagnosis techniques that perform better than existing methods. The multiscale approach will allow efficient integration of these operations tasks, leading to better utilization of measured data, more efficient operation, and efficient computation for large-scale operation. These methods will be applied to several practical examples in collaboration with various industries and federal laboratories. This research represents an effort to develop a general approach for multiscale empirical modeling, most existing uses of wavelets are for multiscale analysis or stochastic modeling. The results of the research will be disseminated via existing and new courses in the Chemical Engineering Department at Ohio State University, in industrial short courses, and may be included in a textbook on process operation. New in-class and laboratory experiments and web-based resources will be developed for improved education in process operation. Ways of improving students' communication and teamwork sk ills will also be studied. Modules will be developed to integrate the principles of process design and simulation across the curriculum, and to train chemical engineers to take a systems view of a process, as opposed to a unit operations view. Abstract - Watkins - 9734177 The demands of present and future microelectronic and optoelectronic device fabrication place stringent requirements on metal deposition schemes. These include high film purity, low temperatures and rapid, controllable deposition rates. The PI postulates that these objectives can be met via Chemical Fluid Deposition (CFD), a new approach to metal deposition that involves the chemical or thermal reduction of soluble organometallic compounds in supercritical carbon dioxide at low temperatures (40-80oC) to yield continuous films on inorganic or organic substrates. CFD exploits the unique, and adjustable, physicochemical properties of SCF solvents, which lie intermediate to those of liquids and gases, to circumvent the limitations of both vapor and liquid phase techniques. In CFD, precursor transport and reduction occurs in solution at significantly lower temperatures and higher reagent concentrations than those of vapor phase techniques such as chemical vapor deposition (CVD). While CFD is a solution-based process, the "gas-like" transport properties of the SCF and its miscibility with gaseous reducing agents such as hydrogen, render the process unencumbered by issues of poor mass transfer and poor deposition rates associated with liquid phase reductions. Preliminary experiments demonstrate that high-purity, continuous platinum and palladium films can be deposited from SCF solution onto silicon wafers and other inorganic substrates at temperatures up to 170oC below those employed in CVD. The research program will focus on the deposition of thin films from carbon dioxide solution by the hydrogenolysis of dimethylcyclooctadine platinum (II) and the deposition of copper films by reduction of copper(II)bis (hexafluoroacetylacetone) and copper(II)bis(2,2,6,6-tetramethyl-3,5-heptanedionate). The precursors were chosen to facilitate a direct comparison of film quality and reduction kinetics in CFD to those of existing techniques and the potential utility of the metal deposits in microelectronics copper and catalytic platinum devices. The educational portion of the work are to: (1) train graduate students who will work at the interface of engineering and materials chemistry, (2) provide undergraduates with opportunities for research experience, (3) develop a two-course series in materials processing that addresses the interests of students and is reinforced by the expanding materials research efforts in the Department of Chemical Engineering, and (4) incorporate research problems and active learning principles into the classroom and assist in the implementation of interactive teaching tools across the curriculum.