Professor Khalil Najafi for his all-time support, encouragement and advice. If it was not for his deep understanding and patience, as well as respectful and humane attitude towards his students, going through hardships of the research would totally be impossible for me. My sincere acknowledgement goes to Dr. Rebecca Peterson, my supervisor and mentor, for all her unbelievably responsible, respectful and kind support of my research over the past years. I would also like to thank my other thesis committee members, Professor Luis Bernal, Professor Kensall Wise, Professor Edward Zellers and Professor Yogesh Gianchandani for their guidance and support. I would like to acknowledge my former and present group-mates and colleagues from Najafi Research Group and from Wireless Integrated MicroSensing and Systems (WIMS) Center, for their help, friendship and useful discussions. My special thanks goes to my colleague Karthik Kumar from Aerospace Engineering for his help and assistance in device testing and modeling. I would like to thank the staff and students of the Solid-State Electronics Laboratory (SSEL) and Luire Nanofabrication Facility (LNF) at the University of Michigan for all their help, in particular, sharing their expertise with me. I would also like to thank Professor Karl Grosh research group at Mechanical Engineering and the staff of Polytec facility at Dexter, MI, for their help with laser vibration analysis. My Ann Arbor experience remains a great chapter in my life thanks to all wonderful friends I made, or I enjoyed the company of, here. I would like to recognize and thank them all. In particular, I would like to thank Ms. Mahta Mousavi for all her support, love and encouragement over the past several months. This dissertation has been dedicated to my parents and my sister, who disserve the biggest appreciation of all. If it was not for their all-time love, sacrifice, support and encouragement, I would not be where I am now. iv TABLE OF CONTENTS DEDICATION ii ACKNOWLEDGMENTS iii LIST OF FIGURES vii

Micro pumps are required in gas chromatographs and other gas analysis systems. Thermodynamic and reduced order models reported in the literature consider only the deflection of pumping membranes which is insufficient to study the performance characteristics of more complicated multistage peristaltic pumps with multiple coupled chambers, active checkerboard valves and single or multiple electrodes for actuating the pumping membranes and valves. Specifically the volume displaced by the valves membranes can be significant compared to the volume displaced by the pumping membranes. In this paper a multiphysics model is extended to include the effect of valves as well the pump membranes deflections. The viscous losses in the flow through the valves are modeled using CFD while the structural behavior of the membrane is modeled using COMSOL in quasi-steady analyses and the results incorporated into the multiphysics unsteady reduced-order model. The model is validated by comparing computed flow-rate as a function of frequency with experimental results for a 4-stage micropump which are found to be within 7%. At a high flow rate condition the model results show that for a 4-stage pump the deflection of the valves significantly affects the performance of the pump.

This paper introduces a new class of compact, highly scalable and efficient membrane-based gas micropumps in which both stroke and flow-control actions are realized by checkerboard microvalves. Realized based on a novel scalable and modular microfabrication technology, the micropump reported here utilizes an unwanted secondary pumping effect, valve pumping, as the main pumping principle, eliminating the need for any pumping membrane. This reduces down the entire pumping structure to a collection of active microvalves, and hence, significantly simplifies the configuration and operation of mechanical gas micropumps and minimizes their dead volume (often the Achilles heel for gas micropumps). Valve-only pumping can produce a flow rate as high as 160 µL/min using only two valve membranes and pressure accumulations of 500 Pa, for a 4-stage device, while it also improves the total device size by 50%.

This paper presents the results of a theoretical analysis of dynamic valve timing on the performance of a multistage peristaltic vacuum micropump. Prior work has shown that for optimum steady state performance a fixed valve timing which depends on the operating pressure can be found. However, the use of a fixed valve timing could hinder performance for transient operation when the pump is evacuating a fixed volume. At the beginning of the transient the pump operates at low pressure difference and a large flow rate would be desirable. As the pump reaches high vacuum the pressure difference is large and the flow rate is necessarily small. Astle and coworkers1–3 have shown using a reduced order model that for steady state operation short valve open time results in lower inlet pressure and flow-rate and conversely. Here we extend the model of Astle and coworkers to include transient operation, multiple coupled stages and non-ideal leaky valves, and show that dynamic valve timing (DVT) reduces the transient duration by 30% compared to high vacuum pressure valve timing. The results also show a significant reduction in resonant frequency of the pump at low pressures, and quantify the effect of valve leakage.

No abstract