Statistics-enhanced multistage process models for integrated design &manufacturing of poly (vinyl alcohol) treated buckypaper

Carbon nanotube (CNT) is considered a promising engineering material because of its exceptional mechanical, electrical, and thermal properties. Buckypaper (BP), a thin sheet of assembled CNTs, is an effective way to handle CNTs in macro scale. Pristine BP is a fragile material which is held together by weak van der Waals attractions among CNTs. This dissertation introduces a modified filtration based manufacturing process which uses poly (vinyl alcohol) (PVA) to treat BP. This treatment greatly improves the handleability of BP, reduces the spoilage during transferring, and shortens the production time. The multistage manufacturing process of PVA-treated BP is discussed in this dissertation, and process models are developed to predict the nanostructure of final products from the process parameters. Based on the nanostructure, a finite element based physical model for prediction of Young's modulus is also developed. This accuracy of this physical model is further improved by statistical methods.;The aim of this study is to investigate and improve the scalability of the manufacturing process of PVA-treated BP. To achieve this goal, various statistical tools are employed. The unique issues in nanomanufacturing also motivate the development of new statistical tools and modification of existing tools. Those issues include the uncertainties in nanostructure characterization due to the scale, limited number experimental data due to high cost of raw materials, large variation in final product due to the random nature in structure, and the high complexity in physical models due to the small scale of structural building blocks. This dissertation addresses those issues by combining engineering field knowledge and statistical methods. The resulting statistics-enhanced physical model provides an approach to design the manufacturing process of PVA-treated BP for a targeting property and tailor the robustness of the final product by manipulating the process parameters. In addition, since the methodology of this study deals with the common issues in general nanomanufacturing processes, this work also serves as a case study of a potential framework of process modeling procedure for similar nanomanufacturing processes.;Several related topics are also investigated in this dissertation work. Those topics include a possible way to monitor the CNT dispersion process by observing the change in vibration structures using time series models, and an alternative method to handle the discrepancy between computer simulation and experimental data. Those topics, although not indispensable to the final goal, provide new angles to view the problem and a better understanding of the nanomanufacturing process.;Some possible extensions of future studies are discussed at the end of this dissertation, including an improvement of manufacturing process, a possible application of PVA-treated BP, and a further application of the prediction model. Those topics represent a broader impact of this work. Subsequent studies of this dissertation, both the manufacturing aspect and the application aspect, are meaningful and worthwhile. Only with continuous advances in every field of BP research can a full realization of the potential of CNTs be achieved.