UNIVERSITY OF CALGARY. Development of a Low Cost Micro-pump. Sina Khalilian A THESIS SUBMITTED TO THE FACULTY OF GRADUATE STUDIES

Transcription

1 UNIVERSITY OF CALGARY Development of a Low Cost Micro-pump by Sina Khalilian A THESIS SUBMITTED TO THE FACULTY OF GRADUATE STUDIES IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF MECHANICAL AND MANUFACTURING ENGINEERING CALGARY, ALBERTA Aug, 2012 Sina Khalilian 2012 i

2 UNIVERSITY OF CALGARY FACULTY OF GRADUATE STUDIES The undersigned certify that they have read, and recommend to the Faculty of Graduate Studies for acceptance, a thesis entitled Development of a Low Cost Micro-pump submitted by Sina Khalilian in partial fulfilment of the requirements for the degree of the degree of Master of Science. Supervisor, Dr. Theodor Freiheit Department of Mechanical & Manufacturing Engineering Dr. Robert W. Brennan Department of Mechanical & Manufacturing Engineering Dr. Alejandro Ramirez-Serrano Department of Mechanical & Manufacturing Engineering Dr. Colin Dalton Department of Electrical & Computer Engineering Date ii

3 Abstract There is a growing demand for micro-pumps as a subcomponent in micro-fluidic systems. In order to make micro-pumps more suitable for most of cost-sensitive mass applications such as micro cooling systems, fuel cells and disposable biomedical applications, a design targeted toward lowest manufacturing cost, high endurance, and reliability is devised for the micropump. In this work, a check-valve micro-pump with a piezoelectric motivator was selected as a low cost design. By following a modular design approach, the total number of parts to be manufactured is minimized. This design is composed of four parts, the upper and lowers casing halves, a single plate inlet and outlet valve, and a piezoelectric disc. To achieve high durability and reliability, stainless steel shim stock was selected as the valve material. Microinjection moulding and chemical etching processes were selected for the fabrication processes of the casings and the valve, respectively. Also, to facilitate the assembly process and for sealing the micro-pump casing, a heat press method of assembly was designed and tested. To predict performance of the micro-pump, a modular simulation model was developed. In this model, results from static flow testing of the micro-valves, piezoelectric displacement test, and finite element structural analysis were incorporated in the model. As a result, a practical model for predicting the micro-pump performance at operating frequencies below 50 Hz is achieved. This simulation model can be useful in parameter selection, optimization and also micro-valve selection for different application requirements of the micro-pump. As a result of this work, by applying design for manufacturing and mass fabrication methods to the overall micro-pump design, a low cost micro-pump designed for high performance and useful to a variety of applications is obtained. iii

4 Acknowledgements I would like to thank Dr. Theodor Freiheit and Dr. Simon Park for their constant support, guidance and encouragements. At all the stages of this project I benefited from their advices. Dr. Theodor Freiheit s positive outlook, patience and confidence in my research inspired me and gave me confidence. His careful editing contributed enormously to the production of this thesis. I want to express my thanks to Mr. Mehdi Mahmoodi, Mr. Chaneel Park and Mr. Liam Hagel and other members of the Micro Engineering, Dynamics and Automation Laboratory for their support to this project. I also would like to thank the Micro Systems Technology Research Initiative (MSTRI) of the Governments of Canada and Alberta, and the University of Calgary for their support of this project. Finally, I would like to thank the Advanced Micro-nanosystems Integration Facility (AMIF at the University of Calgary) for assistance in fabrication the valves, as well as the Alberta Innovates Technology Futures (AITF), nanoalberta program, to AMIF for equipment-related financial assistance. iv

5 Table of Contents Approval Page... ii Abstract... iii Acknowledgements... iv Table of Contents...v List of Tables...ix List of Figures... xi Chapter 1: INTRODUCTION Introduction Motivation Objectives and Approaches Organization... 9 Chapter 2: Review of Micro-pump Technology Displacement - Reciprocating micro-pumps Driving mechanisms Piezoelectric Motivators Electrostatic Motivators Electromagnetic Motivators Shape memory alloys (SMA) motivator Pneumatic and Thermo-pneumatic Motivators Micro-valve mechanisms for controlling backflow Passive valves Active valves Dynamic Valves Displacement- Rotary and Peristaltic Micro-pumps v

6 2.2.1 Micro Gear Pump Peristaltic Micro-pump Other displacement micro-pumps Dynamic- Non Mechanical Micro-pumps Electro-hydrodynamic (EHD) Electro-osmotic Magneto-hydrodynamic Ultrasonic Comparison and evaluation of the Micro-pump driving mechanisms Diaphragm micro-pump simulation and analysis Micro-pump Manufacturing Methods Photolithographic processes Wet (Chemical) Etching Dry etching methods Micro-Injection Moulding Micro-milling Chapter 3: Micro-Pump Design and Simulation Design specifications of the micro-pump Concept Design Micro-pump Simulation Analytical modelling of the micro-pump Detailed Design Piezoelectric Disc Selection Valve Design FEM Analysis vi

7 Multi-Physics Analysis Micro-valve vibrational modeling Design of the Experiment for Valve Prototyping Micro-pump Casing Pumping Chamber Volume Design for Manufacturing and Assembly Injection Moulding Plastic Material Properties MoldFlow TM Analysis Other materials Heat press sealing, assembly process Chapter 4: Manufacturing and assembly process Valves Wet (Chemical) etching Dry Laser Etching Machined Prototype Injection Moulding the Casing Moulds Machining Inserts alignment pins and datum surfaces for machining Injection moulding Dimensional analysis of micro injection moulded parts Heat Press Bonding Feasibility Tests Heat Press Fixture Strength and seal test vii

8 4.5 Cost Analysis Chapter 5: Testing, Results and Discussion Micro-valve static flow testing Simulation results Micro-pump testing Simulation verification Design verification Chapter 6: Conclusion Contributions to Engineering Science Innovations associate with the micro-pump design Simulations and analysis Fabrication and Assembly methods Recommendations for Future Work References: Appendices A: Engineering Drawings and Data Sheets Appendices B: Micro-valve Dynamic test results Appendices C: Simulation vs. Tests Results viii

9 List of Tables Table 1.1 Micro-pump target requirements... 9 Table 2.1 Comparison of micro-actuation mechanism of typical micro-pumps Table 3.1 Micro-pump design specifications Table 3.2 Membrane volume displacement for differential pressure of various PZT discs Table 3.3 Maximum displacement vs. frequency for the third PZT disc Table 3.4 Multi-physics analysis results of valve flow Table 3.5 Designed Experiment Parameter Levels Table 3.6 Casings features description Table 3.7 Chamber volume selection constraints Table 3.8 Selected injection moulding parameters for the TOPAS COC Table 3.9 Injection moulding analysis results for the TOPAS COC Table 3.10 Selected injection moulding parameters for PMMA (Acrylic) Table 3.11 Injection moulding analysis results for PMMA (Acrylic) Table 3.12 Selected injection moulding parameters for Polystyrene Table 3.13 Injection moulding analysis results for Polystyrene Table 3.14 Selected injection moulding parameters for Polyethylene (LDPE) Table 3.15 Injection moulding analysis results for Polyethylene (LDPE) Table 4.1 First injection moulding sample process conditions COC Table 4.2 Injection moulding process conditions COC Table 4.3 COC 8007 Dimensions and Measurements Table 4.4 Costs associated with injection moulding of the casings Table 4.5 Chemical etching costs for the initial design of the valve Table 4.6 Micro-pump production variable costs ix

10 Table 4.7 Micro-pump production fixed costs Table 4.8 Micro-pump overall cost for different production volume Table 5.1 Average flow rate at 10kPa pressure head Table 5.2 Valve pieces combinations used for micro-pump testing Table 5.3 p-values for valve parameters interaction effects Table 5.4 Comparison of the micro-pump performance Table 5.5 micro-pump design verification Table 6.1 Innovations with the micro-pump design, advantages and limitations Table 6.2 Recommended reliability tests x

11 List of Figures Figure 2.1 Schematic of a diaphragm micro-pump [Woias 2005] Figure 2.2 Schematic view of a PZT micro-pump with the inlet valve open Figure 2.3 piezoelectric actuator configurations: a) lateral-strain configuration, the piezoelectric is attached to a clamped plate from bottom and free on top b) axial-strain configuration, the piezoelectric is sandwiched between to clamped plates Figure 2.4 Schematic view of an electrostatically actuated diaphragm micro-pump [Zengerle et al. 1995] Figure 2.5 Schematic view of an electromagnetically actuated diaphragm micro-pump Figure 2.6 Schematic of a SMA diaphragm micro-pump [Xu et al. 2001] Figure 2.7 Schematic view of a pneumatically actuated diaphragm micro-pump [Laser and Santiago 2004] Figure 2.8 Schematic view of a thermo-pneumatic driven diaphragm micro-pump [Laser and Santiago 2004] Figure 2.9 Structure of various passive valves [Yang 2004, Shen et al. 2008] Figure 2.10 Schematic of active check-valves used in micro-pump Figure 2.11 Principle of Nozzel/diffuser valves [Stemme E and Stemme G 1993] a) Increasing chamber volume b) Decreasing chamber volume Figure 2.12 Principle of Tesla valves [Bendib and Olivier 2001] a) forward flow b) reverse flow Figure 2.13 Driving mechanism of vane and disc rotary micro-pumps a. Rotary vane micropump driver [Ahn 1995] b. Rotary disc micro-pump driver [Blanchard 2004] Figure 2.14 Schematic of operation principle of a rotary gear micro-pump [Dewa et al. 1997] Figure 2.15 Peristaltic micro-pumps [Nguyen et al. 2002] a) piezoelectric driven [Smith 1988] b) Electrostatic driven [Judy et al. 1991] Figure 2.16 Schematic of an electrowetting micro-pump [ Yun et al. 2001] Figure 2.17 Schematic of basic principle of an electroosmotic micro-pump [Chen and Santiago 2002] Figure 2.18 Schematic of a Magnetohydrodynamic driven micro-pump [Nisar 2008] Figure 2.19 Model of ultrasonic driving of micro fluid [Nuyen and White 1999] xi

12 Figure 2.20 Equivalent electrical network of micro-pumps with a) pneumatic actuation b) electrostatic actuation [Bourrouina and Grandchamp 1996] Figure 2.21 Wet etching lithography process a) substrate covered with photoresist b) exposing photoresist to U.V. light using a lithography mask c) removing exposed area by dipping substrate in alkaline solution Figure 2.22 Etch profile a) isotropic etching b) perfectly anisotropic etching Figure 3.1 Schematic of selected micro-pump design concept Figure 3.2 Simple cantilever beam valve micro-pump (Valve Design #1) Figure 3.3 Simple cantilever beam valve micro-pump, pre-stressed valve in the middle of the casings by a pre-stress feature (Valve design #2) Figure 3.4 Three plate cantilever valve. [Kan et al. 2005] Figure 3.5 Valve design of ThinXXS TM micro-pump [Weber 2005] Figure 3.6 Schematic of a simple 2D flap check valve Figure 3.7. Valve Design a) one piece module for inlet and outlet, with a simple cantilever valve b) bent valve module c) detailed view of holes forming bend line Figure 3.8 Different shapes designed and evaluated for valve moving parts Figure 3.9 Model of the initial concept selected for prototyping Figure 3.10 Model for the micro-pump with injection moulded casing design Figure 3.11 Modular system for analysis of the micro-pump Figure 3.12 Micro-pump chamber pressure as a function of time Figure 3.13 Micro-pump outlet flow as a function of time Figure 3.14 BUZZER Piezo, CEB-20D Figure 3.15 Test setup for the PZT disc deformation measurement Figure 3.16 Deformed ThinXXS TM valve; material: TOPAS COC Figure 3.17 FEM analysis of various valve designs Figure 3.18 a) a simple cantilever valve b) two cantilevers in series Figure D fluid-structure interaction analysis of the micro-valve Figure 3.20 Velocity integral at inlet and outlet vs. time Figure 3.21 Main effect plots of hole diameter and valve length on flow rate Figure 3.22 Main effect plot of hole diameter and valve length on backflow Figure 3.23 Model for calculating dynamic response of the micro-valve xii

13 Figure 3.24 Valve displacement by time obtained from the vibrational model Figure 3.25 Lithography mask designed for making the prototyping samples Figure 3.26 Final design of the micro-pump Casings a) upper casing upper side, b) upper casing lower side (pumping chamber), c) lower casing Figure 3.27 Machined prototype casing design a,b) upper casing, c) lower casing Figure 3.28 Key features of design for manufacturing and assembly considered in the design Figure 3.29 TOPAS COC-8007 mechanical properties and recommended processing conditions Figure 3.30 TOPAS COC-8007 a) effect of mould temperature on viscosity b) effect of injection or holding pressure on the specific volume Figure 3.31 Selected dimensions for the runners and gates Figure 3.32 Meshed model for injection moulding analysis Figure 3.33 Time required for filling stage of the injection moulding Figure 3.34 Volumetric shrinkage distribution at the end of the injection moulding Figure 3.35 Residual stress distribution of injection moulded parts Figure 3.36 Time to reach ejection temperature (cooling time) distribution Figure 3.37 Weld line formation locations during injection moulding Figure 3.38 Particle orientation at skin of the casings Figure 3.39 Schematic of the heat press assembly process of the micro-pump Figure 3.40 Fixture setup for the heat press assembly process Figure 4.1 two valves from the first batch of etched valves on steel shim stock, Pictures on the left are showing the top sides and on right are showing back sides Figure 4.2 a valve from the second batch of the etched valves on steel shim stock a) front side, 0.17mm etch line thickness b) backside, 0.07mm etch line thickness Figure 4.3 Chemical etching on shim stocks roll, a) front side of the etched valve b) back side of the etched valve Figure 4.4 Chemical etching on shim stocks roll, a) front side of the etched valve b) backside of the etched valve Figure 4.5 A thickness wet-etched valve Figure 4.6 a) a thickness valve fabricated with laser etching method b) front side of the valve c) back side of the valve xiii

14 Figure 4.7 a) a thickness valve fabricated with laser etching method b)topside of the valve c) Backside of the valve Figure 4.8 a thickness bent valve, fabricated by dry laser etching Figure 4.9 a 0.001" thickness valve with modified geometry, fabricated with laser etching. 96 Figure 4.10 Machined casings a) upside of the aluminum upper casing b) Brass lower casing Figure 4.11 Brass upper casing a) piezoelectric disc glued to the casing b) valve and tubes placed in the upper casing Figure 4.12 KERN Micro, a high precision CNC machining device Figure 4.13 Injection moulding Inserts, machined on machining wax Figure 4.14 Two injection moulding inserts, common datum surfaces Figure 4.15 Machining of the injection moulding inserts Figure 4.16 Two machined injection moulding inserts Figure 4.17 BOY 12A injection moulding machine Figure 4.20 COC 8007 injection moulded casings Figure 4.21 Defective COC8007 samples; a) improper filling of the thin edge around the lower casing b) Improper filling of the upper casing Figure 4.22 Polystyrene injection moulded sample Figure 4.23 PMMA injection moulded sample Figure 4.24 COC 5013 injection moulded sample Figure 4.25 Micro-pump Upper Part Casing Figure 4.26 Micro-pump Lower Part Casing Figure 4.27 Horizontal dimensional analysis of COC 8007 samples Figure 4.28 Vertical dimensional analysis of COC 8007 samples Figure 4.29 a) before the bonding b) after the bonding at C Figure 4.30 Heat press bonding a) heat press fixture b) fixture to clamp the micro-pump c) setup for heat press bonding assembly Figure 4.31 CARVER heat press with the fixtures assembled on it Figure 4.32 Schematic of the casings profile a) before the bonding b) predicted profile after bonding c) actual profile after bonding xiv

15 Figure 4.33 Heat press bonding results for COC 8007, formed at 130 o C heat press fixture temperature Figure 4.34 PMMA casings after the bonding, formed at 145 o C heat press fixture temperature Figure 4.35 micro-pump casings bonding test setup Figure 4.36 a) before gluing b) after gluing, the spot where the leakage first occurred is highlighted Figure 4.37 Comparison between the size of the initial design of the micro valves (the larger valve) and the modified version (smaller valves in this figure) Figure 5.1 Micro-valves static testing fixture Figure 5.2 Static pressure flow characteristic of valves with different valve seat hole diameters Figure 5.3 Static pressure flow characteristic of valves with different valve stiffness Figure 5.4 Static pressure backflow for valves with different valve seat hole diameters Figure 5.5 Static pressure backflow for valves with different valve stiffness Figure 5.6 Instantaneous flow for complete cycle at 30Hz and 10kPa pressure head, plotted for: a) different hole diameters b) different valves stiffness Figure 5.7 Effect on the average flow rate of seat hole diameter at 30 and 80 Hz Figure 5.8 Effect of valve stiffness on average flow rate Figure 5.9 Micro-pump test setup schematic Figure 5.10 Micro-pump prototype for dynamic testing, a) Lower part of the casing and a gasket, b) upper part of the casing c) assembled micro-pump Figure 5.11 Separated valve pieces used in the micro-pump testing Figure 5.12 Flow vs. pressure curve for tests number 4, 6, 8 and 9 in Table 5.2 at 50 Hz frequency Figure 5.13 Main effects plots for valve parameters at: a) frequencies between 10 to 50 Hz b) frequencies between 60 to Figure 5.14 Interaction effects plots for valve parameters at: a) frequencies between 10 to 50 Hz b) frequencies between 60 to Figure 5.15 Micro-pump test results for two different outlet hole diameters with inlet valve fixed at: a) D=0.6mm, K=79 N/m b) D=0.8mm, K=142 N/m Figure 5.16 Micro-pump test results for two different inlet valve stiffness s with outlet valve fixed at: a) D=0.6mm, K=79 N/m b) D=0.8mm, K=79 N/m xv

16 Figure 5.17 Static flow test results for valves with D=0.8mm,K=142N/m and D=0.6mm, K=79N/m Figure 5.18 Comparison between simulation results and micro-pump experimental test results at 10kPa pressure head Figure 5.19 a method for determination of dynamic effects correction factors for valve static flow properties a) correction Factors II= AxI, III=B+I b) Minimization objective parameter, φ c) The algorithm Figure 6.1 The suggested modified design for the injection moulding casings of micro-pump a) upper casing, b) lower casing Figure 6.2 Suggested design for the valve a) flatten pattern b) bent valve c) valve placed inside the upper casing xvi

17 1 Chapter 1: INTRODUCTION 1.1 Introduction Together with advances in the micro electro mechanical systems (MEMS), during the past few decades, fluidic systems in areas such as biotechnology, electronic cooling systems and medical, defence and aerospace are also undergoing effort for miniaturization. In the micro-fluidic systems, because of the changes in the basic physics, conventional macroscopic devices cannot simply be scaled down [Ouellette 2003]. When functional dimensions of a system gets small enough, despite that the fluid properties remain the same, effects such as viscosity, surface tension, electrostatic, electromagnetic forces and also suspended particles in the fluid becomes much more important. Therefore micro-fluidics systems require different methods of design and fabrication from macroscopic devices. The small fluid volumes in micro-fluidic systems need to be accurately pumped, controlled, or circulated in the system. Since early in the development of MEMS, micropumps as the actuation mechanism in micro-fluidic systems have been the subject of much of research, due to their advantages in mobility, low power consumption, and improved accuracy [Amirouche et al. 2009]. Generally speaking, any small pump based on MEMS technology is referred as a micro-pump. However, others define a micro-pump as pumps with one or more functional dimensions in the range of 100 micrometers or less. Due to the wide range of applications such as drug delivery systems, electronic cooling systems, inkjet printers and fuel cells, many different types of micro-pumps have been developed. In most of the applications, high reliability and durability are essential qualities of a micro-pump, as the micro-pump becomes a constraining sub-system for the total system life and reliability. In some applications, such as bio-medical or chemical injection, the chemical resistance of the pump materials to reaction with the working fluids can be important. Such pumps may also require a capability to work with two phase flow with micro-particles in the flow. In drug delivery applications, precise control of flow rate is an essential requirement, and for microelectronics cooling systems, having a high flow rate is a very demanding quality. 1

18 The majority of the micro-pumps reported are diaphragm micro-pumps, in which moving surfaces or boundaries do pressure work on the working fluid in a cyclic manner [Laser and Santiago 2004]. The operational principle of a diaphragm micro-pump is based on the interactions between various mechanics, such as piezoelectric materials, electrostatic forces, structural mechanics, and fluid mechanics. The growing demand for micro-pumps in micro-fluidic systems is driving a need for a low cost design for mass produced applications. Designing a micro-pump for low cost requires ease of component fabrication and assembly, as material costs are usually not a driver due to their small size. A design compatible with the most appropriate fabrication processes and the application of design for manufacturing (DFM) and assembly (DFA) methods provides a low cost approach. Currently a variety of fabrication methods are being used for micro-fluidic devices, including wet etching, plasma etching, conventional machining, photolithography, soft lithography, hot embossing, injection molding, and laser ablation [Fiorini and Chiu 2005]. The choice of fabrication method should be compatible with the design, material, and fabrication volume. Also, in micro-manufacturing, due to small component sizes, it is very important to design parts for ease of handling and assembly. Designing a micro-pump for low cost and competitive performance requires that various design concept alternatives be considered early in the process. Accurate simulations and optimization methods should be employed in the design process. This approach will result in a faster and smoother transition to manufacturing and lower life cycle cost for the product. 1.2 Motivation In an increasing number of engineering systems, from biology and medicine to space exploration and microelectronic cooling, very small fluid volumes exist [Amirouche et al. 2009]. Micro-pumps, the driver of micro-fluidics, will play an important role in micro-fluidic applications [Brand 2006]. According to a research report published in 2012 by Yole [2012], micro electro mechanical systems (MEMS) market is expected to have an average annual growth of 20% in units and 13% in revenues, which will make this market an estimated $21 billion market by It is expected that motion sensing devices such as accelerometers, gyros and magnetometers will be 25% of this market, while micro-fluidics will be 23%, of this 2

19 market in Although currently a number of different micro-pumps are commercially available, the growing demand for micro-pumps in micro-fluidic systems is driving a need for low cost micro-pump designs for mass produced applications. In order to have a successful design which is capable of satisfying various application requirements, further to meeting the desired application performance requirements, a micropump should be reliable, durable, and have a low cost. Some of the requirements that a micropump design should address are: a. Performance In different applications, a micro-pump is required to provide a certain flow rate at a specified pressure head. Regardless of the application, a micro-pump has to provide an adjustable flow rate within a reasonable range. Adjustable flow rate Depending to the application, there can be different requirements for precision and the range of the flow rate. For example, micro-pumps used for circulating coolant in microelectronic systems are normally required to have a high flow rate but the controllability and accuracy are not very crucial [Laser and Santiago 2004, Tuckerman and Peace 1981]. In a study performed by Zhang et al. [2002], it was shown that two phase convective cooling of a 100W microchip requires a flow rate more than 10 ml/min. However in biomedical and medical applications and devices, high volumetric flow rates are not likely to be required, but precise metering is of great importance [Laser and Santiago 2004, Dash and Cudworth 1998]. Minimum backflow In diaphragm check-valve micro-pumps, a major challenge which can significantly reduce the efficiency is backflow. Backflow is the flow passed through the check-valve in the opposite direction of the valve. The amount of backflow is mainly related to flow behaviour of the micro-pump check valves. In some designs, by utilizing active micro-valves the backflow is very well controlled but adds to the complexity of the system. While modelling and analysis of the backflow is addressed in few previous works [Kan J. et al. 2005, Yun Y. et al. 2000, Stemme E. and Stemme G. 1993], a practical model for predicting backflow in a micro-pump 3

20 at different frequencies and pressure heads and which could be used for micro-valve design and optimization is not yet available. To use the simple cantilever check valves at different working conditions, backflow minimization is required. Pressure head In diaphragm micro-pumps, the maximum attainable pressure for liquids is mainly dictated by the driver and valve backflow characteristics. The pressure generation requirements for implantable micro-pumps are moderate for micro-pumps, as the maximum blood pressure does not exceed 25 kpa [Laser and Santiago 2004]. However, pressures as high as 100kPa or greater might be required to force flow through micro-channels or the jet structures found in micro heat sinks. b. Low cost To expand the application of micro-pumps in fields such as portable micro-fluidic devices like mobile fuel cells or disposable micro-pumps in medical applications, i.e., those used with bodily fluids such as blood, the cost of the micro-pumps should significantly be reduced. Designing a micro-pump for low cost requires ease of component fabrication and assembly, as material costs are usually not a driver due to their small size. Also "the cost of a micro-component is not exclusively defined by the fabrication technology itself, but mainly by an intelligent and fabrication-oriented design" [Woias 2001]. Therefore, a design compatible with the most appropriate fabrication processes and the application of design for manufacturing (DFM) and assembly (DFA) methods provides a low cost approach. DFM requires that various design concept alternatives be considered early in the process, putting more effort into the early stages of the design considering both manufacturing and functional requirements. However, this approach results in faster and smoother transition to manufacturing and lower life cycle cost for the product. In the design for manufacturing process, after the conceptual design phase, the design should be reviewed several times, simplified, and the number of parts be reduced. A modular design approach can minimize the total number of parts to be manufactured and facilitate the assembly process. Also, in micro-manufacturing, due to small component sizes, it is very 4

21 important to design parts for ease of handling and assembly. It can clearly be seen in a large number of scientific publications on micro-pump design that design for manufacturing and low cost aspects have often been left out of consideration [Woias 2001]. c. Reliability and Durability High reliability and durability is an essential quality for a micro-pump used in many different devices, as the micro-pump becomes a constraining sub-system for total system life and reliability. According to the U.S. military quality control and reliability handbook [MIL- HDBK ], to be reliable, a engineered system should be able to perform its required function under defined operating conditions for an specified amount of time. To ensure the reliability of a micro-pump, situations which might interrupt it from performing its function properly should be considered and the micro-pump should be able to handle them. As an example, a micro-pump should be insensitive to gas bubbles, as bubbles are present at many micro-fluidic systems [Laser and Santiago 2004]. Similarly, it may need to be able to handle small solid particles in the flow. When a micro-pump is used in portable devices, it is required to have a uniform performance at different orientations, even under a shock [Xiao 2008]. Self priming is also an important characteristic of a micro-pump performance, i.e., having the capability of providing sufficient vacuum to suck the fluid in and then empty the pumping chamber of gas bubbles. Durability can be defined as the ability to work for a long time without significant deterioration in the performance. Durability of a micro-pump is mainly dictated by fatigue and degradation of its dynamic parts. Designing for longer fatigue life and using corrosion resistant material can improve the durability of a micro-pump. d. Other criteria: Further to the above mentioned demand criteria, depending on the application, a micropump design should have biocompatibility, i.e. to be able to work with bio-fluids or other chemicals without contaminating them, have compactness, and have low power consumption. In some applications such as bio-medical or chemical injection, the chemical resistance of the pump materials to reaction with the working fluids is very important. Such pumps may also 5

22 require the capability of working with two phase flow with micro-particles in the flow. Miniaturization and low power consumption are very important in mobile applications [Laser and Santiago 2004]. Also, having the inlet and outlet of the micro-pump in-plane can make the micro-pump thinner and further allow it to easily be integrated in micro-fluidic circuits. Despite the development of new micro-pumps with various driving mechanisms in the last 25 years, application and performance requirements are now driving a need for low cost, reliable micro-pumps with an improved performance. Furthermore, after a survey through the research literature, it was found that a practical methodology that can be used for design and optimization of a diaphragm micro-pump does not exist. Therefore a modular analysis and modeling method based on a comprehensive diaphragm micro-pump analytical model suggested by Zengerle and Richter [1994] has been adopted with the aim of designing and optimizing the micro-pump. 1.3 Objectives and Approaches The initial issue in development of a new product is to set design goals [Otto and Wood 2000] to guide the development. As mentioned in the design motivation section, the application requirements of a micro-pump include delivery of the fluid with an adjustable flow rate at a specified pressure head, minimize the backflow, have a low cost, and have sufficient reliability and durability. Depending on the application area, the importance of these requirements can be different, and based on these requirements the goals for design will be set. In this work our main objective was to design and develop a low cost micro-pump having a competitive performance. Because of the diversity of the applications for micro-pumps, it is not possible to develop a micro-pump that satisfies all the performance requirements for all applications. In continuing the previous work on development of a novel micro-pump system by Xiao [2008], the objective for this work is to meet the performance requirements established by Xiao for most of the common application requirements, e.g., fuel cells, chemical injection, cooling system, etc. These requirements are a desired maximum flow rate (Q max ) of 5 ml/min or higher and a maximum pressure head ( P max ) of about 50kPa. The performance achieved by Xiao micro-pump was, P max =13.4 kpa and Q max =5.2 ml/min for water. In comparison, these values 6

23 for the ThinXXS MDP2205 are P max =50 kpa and Q max =9.2 ml/min. Both of these micropumps are piezoelectric driven diaphragm micro-pumps with cantilever check-valves. Generally piezoelectric driven micro-pumps have a very low power consumption (< 1 W) [Amirouche et al. 2009]. In this work, similar to the previous design by Xiao, the piezoelectric was preselected for the micro-pump actuation mechanism to keep the power consumption about or below 1 W. It is also desired to keep the dimensions of the micro-pump comparable or smaller than the ThinXXS micro-pump, i.e., about 30 x 25 x 10 mm (package size < 7.5 cm 3 ), to fit in smaller electronics applications. A reduced cost micro-pump will have a reduced number of component parts that use fabrication processes suitable for mass production. Further, in micro-technology the assembly process forms a large percentage of the overall cost of a product [Hsu 2005]. So the micropump should be designed for ease of assembly and be suited for having an automatic assembly process with minimum human intervention. In this work, in addition to the conventional product design stages (e.g. concept design, detail design, prototyping, and testing), a design for manufacturing and assembly (DFM and DFA) methodology was followed, meaning that fabrication and assembly methods are considered from the earliest stage in the design. By modularizing and integrating parts, the total number of parts for manufacturing can be minimized and assembly process will be facilitated. The desired overall price for the micro-pump is no more than $50, less than half of the price of commercially available micro-pumps such as ThinXXS MDP2205 with 100 in 2006 and TAKASAGO SDMP320 with 13,630 JPY in 2012 for a single unit of the micro-pump. As a rough estimation, the fabrication and assembly cost of the micro-pump should be about 1/4 th or 1/5 th of the desire price, i.e. about 10 to 15 dollars. In a diaphragm micro-pump, micro-valves are the smallest moving parts in direct interaction with the fluid. Therefore, to achieve better reliability and durability, stainless steel shim stock was selected for the micro-valve material. Compared to plastic micro-valves, this will allow the achievement of a micro-valves fatigue life of more than 2x10 7 cycles more easily. This number of cycles will allow more than 100 hours of operation at an average of 50 Hz. Further, it is important that the micro-pump be flexible in working with fluids at different 7

24 temperatures. This requires the selection of components and bonding methods which are not significantly affected by temperature. In this work, however, the objective is to ensure the proper operation of the micro-pump at about room temperature with a working fluid of water, and selecting the components and bonding method that have the potential of working at other temperatures. The further study of the micro-pump reliability over a range of temperatures and working fluids is beyond the scope of this work. In addition, the self priming and handling air bubbles in the working fluid capability of the micro-pump is desirable. Finally, the micropump should be properly sealed up to the maximum operational pressure with a safety factor of at least 2. It is desired that the micro-pumps have no more than 0.5% failure rate in 100 hours of operation, corresponding to $0.25 per unit warranty cost based on the desired $50 cost of the micro-pump. To test this reliability criterion, a large number of micro-pumps should be tested to failure, which is beyond the scope of this work. Therefore this objective can be considered during the mass fabrication of the micro-pump. 8

25 Table 1.1 Micro-pump target requirements Objectives Descriptions Target Specifications Flow rate > 5ml/min for water Performance Pressure Head 50 kpa for water Controllable flow rate Yes Back flow Minimized Compactness The overall dimensions/package size 30x25x10mm/ <7.5 cm 3 Service life >100 hours Reliability No more than 0.5% failure rate in 100 hours Reliability and Operating temperature Room temperature Durability Self priming Unassisted start up Sealing No leakage Production cost per unit of the micro-pump < $ 15 Low Cost Fabrication processes suitable for mass production Yes Ease of assembly and suitable for automatic assembly Yes The objectives are summarized in Table 1.1. Details of the design requirements developed for achieving the objectives are discussed and listed in the Design specifications section. Also as this work was a continuation of a previous work by Xiao [2008], the following design constraints were considered in this project: Using a piezoelectric disc for the actuation mechanism Injection moulding of the micro-pump casings Stainless steel micro-valves Chemical etching of micro-valves 1.4 Organization The thesis is organized as follows. The introduction chapter is followed by a review of current micro-pump technology in Chapter 2. Chapter 3 describes the design developed and 9

26 the analytical modeling of the micro-pump. Also in this chapter, various analysis methods such as finite element analysis, fluid-structure interaction analysis in COMSOL, and vibrational analysis are discussed. At the end of this chapter, design for manufacturing and assembly of the micro-pump and related analysis are described. Chapter 4 discusses the fabrication and assembly of the micro-pump prototype using micromechanical machining, wet and laser etching, injection moulding, and heat press bonding. In Chapter 5, tests procedures and tests results are presented and discussed. The last chapter provides the summary of the work and recommendations for future work. 10

27 2 Chapter 2: Review of Micro-pump Technology In this chapter a review of current micro-pump technology is performed. In the first three sections, operational principles and important features of various reported micro-pump designs are briefly reviewed. Later in this chapter, a detailed review of major simulation and analysis methods of diaphragm micro-pumps is performed. In the last section current and potential manufacturing methods for micro-pumps are discussed. As a supplement to this chapter, the reader may wish to refer to other references reviewing micro-pumps [Amirouche et al. 2009, Nguyen et al. 2002; Laser and Santiago 2004; Woias 2005; Yun and Yoon 2006]. Like any other complex system, various classifications are used to study and design micro-pumps. Similar to traditional pumps, micro-pumps are categorized based on the manner that the flow and fluid pressurization are generated [Laser and Santiago 2004]. Pumps are generally classified in two major categories: I) displacement and II) dynamic pumps [Krutzch and Cooper 2001]. Displacement micro-pumps, apply pressure forces to the fluid through one or more moving boundaries. Some examples of displacement micro-pumps are various types of diaphragm, peristaltic, and rotary micro-pumps. In dynamic micro-pumps, momentum and pressure of the fluid are continuously increased [Laser and Santiago 2004]. An increase in the momentum can be turned into pressure increase by the pressure load applied on the micropump. Centrifugal, electro hydrodynamic, magneto-hydrodynamic, electro-osmotic and so on, are examples of dynamic micro-pumps. The majority of the existing micro-pumps are diaphragm displacement micro-pumps, in which the moving boundaries, normally called a diaphragm or membrane, apply pressure directly to the fluid [Laser and Santiago 2004]. Diaphragm micro-pumps usually consist of inlet and outlet valves, pumping chamber and driver, which could consist of piezoelectric, electrostatic, or pneumatic actuation, and so on [Du et al. 2009]. Other than centrifugal pumps, dynamic micro-pumps are mostly non-mechanical pumps, in which there is no boundary work imposed on the fluid. Centrifugal pumps which are miniaturized to a limited extent are not very effective at low Reynolds numbers [Laser and 11

28 Santiago 2004]. For non-mechanical dynamic micro-pumps, the geometry and fabrication process are comparatively simple, as there is no moving parts [Amirouche et al. 2009]. However because their performance and operation are influenced by the properties of operating fluid and the micro-pump surface material, their application, material and fabrication process are constrained [Amirouche et al. 2009]. In most cases, the common nonmechanical micro-pumps do not have high flow rate, pressure head, and fast response time. Typical flow rate of non-mechanical micro-pumps are about 10µl/min or less, whereas this value can be as high as several millilitre per minute for mechanical micro-pumps [Nguyen and Wereley 2002]. 2.1 Displacement - Reciprocating micro-pumps Reciprocating displacement micro-pumps, which normally consist of a pumping chamber, inlet and outlet valves, and an actuated diaphragm, work on the same principle as macro-sized diaphragm pumps. These micro-pumps utilize oscillatory movement of mechanical parts to increase the pressure and displace fluid. Figure 2.1 illustrates schematic of a diaphragm micro-pump. In this figure, the micro-pump chamber is enclosed by a flexible diaphragm on top. Using an actuation mechanism, the moving boundary alternatively increases and decreases the pump chamber volume. As a result, pressure changes inside the chamber occur. When the diaphragm is moved upward, the pressure decreases to a value less than inlet pressure and suction from inlet occurs. Similarly as the diaphragm moves downward, first the inlet valve closes and then the pressure increases to a value higher than the outlet pressure, after which the valve opens and fluid flows. During this cycle, inlet and outlet valves direct the flow into and out of the pumping chamber, respectively. Backflow is flow passing through the valve in the opposite direction. Ideally, the micro-pump check valves should prevent the backflow completely; however, there is always some amount of backflow. 12

29 Figure 2.1 Schematic of a diaphragm micro-pump [Woias 2005] Reciprocating displacement micro-pumps are the most widely designed and used micropumps [Laser and Santiago 2004]. While most diaphragm micro-pumps have a configuration similar to Figure 2.1, a number of micro-pumps were designed with multiple chambers, either in series or parallel, e.g. Shoji et al.1990, Grosjean and Tai 1999, Folta et al Various types of actuation mechanism were used in these micro-pumps. Piezoelectric, electrostatic, electromagnetic, pneumatic, and others are reported. Also, various valve designs, such as active or passive flap valves, were developed for controlling the backflow. Further, in a number of micro-pumps, structures are employed which use fluid inertia for rectifying the flow instead of check-valves [Stemme E. and G. 1993, Olsson et al. 1995, Foster et al. 1995]. Later in this section, various actuation mechanism and valve types for diaphragm micropumps will be further discussed Driving mechanisms In diaphragm micro-pumps, oscillatory movement of the diaphragm is influenced by three major factors: actuation force and driving mechanism, chamber pressure and the load on the micro-pump, and membrane geometry and stiffness. Considering the superposition principle, the effects of these factors can be individually analysed then superposed to obtain the effective volume displacement of the diaphragm. As mentioned before, various actuation mechanisms have been proposed. Here a brief review of the key features, advantages, drawbacks, and performance aspects of these drivers will be given. 13

30 Piezoelectric Motivators Piezoelectric actuation of the micro-pump is based on the physical phenomenon of piezo-crystalline materials. The most commonly used piezoelectric material is lead zirconate titanate (PZT) [Ikeda 1990]. Figure 2.2 shows a schematic of a typical piezoelectric driven micro-pump. Figure 2.2 Schematic view of a PZT micro-pump with the inlet valve open [Iverson and Garimella 2008] For most piezoelectric applications, a layer of piezoelectric material of various thicknesses is deposited on a passive plate normally made of brass, stainless steel, or polymers. This plate is commonly used as the micro-pump diaphragm itself. However, in some designs, due to application requirements and for a better sealing, a separate membrane is used and the PZT is attached on it [Lutz Weber 2006]. The maximum free strain of a piezoelectric is determined by the applied voltage and polarization limit of the piezoelectric material [Laser and Santiago 2004]. In turn, the piezoelectric maximum strain, the clamping manner of the plate, and the plate geometry and stiffness determine the maximum displacement of the piezoelectric disc. Also, the free displacement of the piezoelectric disk when there is no pressure difference across it places a limit on the micro-pump s maximum stroke volume. Figure 2.3 shows two major configurations of the piezoelectric disc used in micro-pumps [Laser and Santiago 2004]. The polarization of both configurations is normal to the piezoelectric plate. In Figure 2.3a, a piezoelectric disc is attached at its bottom to the plate and free to move at its top. In this configuration, after applying an electrical field, both lateral and axial strains occur. The axial strain does not cause deflection in the plate; however, the plate bends in accordance to the 14

31 lateral strain. When the strain is expanding, it bends upward, and when it is contracting, it bends downward. This configuration of the piezoelectric disc is most commonly used, however, the configuration shown in Figure 2.3b has also been reported [Esashi et al.1989]. In this second configuration, the piezoelectric disc is constrained between a relatively rigid plate on the top, and a membrane below. The axial strain caused by the applied electric field in the piezo directly deforms the membrane. a) b) Figure 2.3 piezoelectric actuator configurations: a) lateral-strain configuration, the piezoelectric is attached to a clamped plate from bottom and free on top b) axial-strain configuration, the piezoelectric is sandwiched between to clamped plates [Chen et al. 2007]. The piezoelectric actuation method, due to its flexibility in operating at various frequencies and voltages, and good efficiency, is the most widely used actuation mechanism in micro-pumps [Amirouche et al. 2009]. Piezoelectric micro-pumps with operating frequencies as high as 5000Hz, voltage between 20 to 1,200 V and efficiency of about 30% have been reported. Using this actuation mechanism, micro-pumps with a maximum pressure heads as high as 200kPa for water [Kamper et al. 1998] and flow rates as high as 16 ml/min have been reported [Olsson et al. 1995]. Also the piezoelectric micro-pumps normally have a low power consumption, i.e. between a few mw to less than 1W [Amirouche et al. 2009] Electrostatic Motivators Electrostatic forces are widely used in MEMS for actuation purpose. Typically, an electrostatic actuator is composed of two parallel plates acting as two electrodes. Figure 2.4 shows schematic view of an electrostatically driven micro-pump. By applying a supply voltage between the pump diaphragm and the counterelectrode, the membrane deforms back and forth as a result of the electrostatic force. The supplied voltage is normally between V at frequencies as high as several hundred hertz [Zengerle et al. 1995]. Zengerle et al. 15

32 [1992] has reported an electrostatic driven diaphragm micro-pump, with outer geometry of 7x7x2mm 3, operating at 200V and 300Hz and having a maximum flow rate of 160µl/min and a 29kPa maximum pressure head. Figure 2.4 Schematic view of an electrostatically actuated diaphragm micro-pump [Zengerle et al. 1995] According to [Griffiths 1999], the electrostatic force between two parallel plates, can be expressed by equation 2.1, in which C is the capacitance between the two plates, V is the applied voltage, and S is the distance between the plates. ( 2.1) The capacitance between the two circular plates and consecutively the electrostatic force (F el ) are calculated in equation 2.2, where is the permittivity of the medium between the plates and d m is the diameter of the plates [Griffiths 1999, Laser and Santiago 2004]. ( 2.2) This formula gives an estimate for the electrostatic force between the plates and can be used to find the deflection of the membrane. However, accurate calculation of the electrostatic 16

33 force and deflection can only be found using numerical methods or finite element multiphysics analysis (electrostatic-structural analysis). Electrostatically driven micro-pumps, compared to other actuation mechanisms, have higher efficiency, i.e., about 50%, and they can operate at frequencies as high as 10,000Hz.[Zengerle et al. 1992, Laser and Santiago 2004]. However, they normally have a small actuation stroke limited to about 5 µm for an applied voltage of 200V [Nisar et al. 2008, Amirouche et al. 2009] Electromagnetic Motivators Figure 2.5 shows an electromagnetically actuated diaphragm micro-pump. As shown in this figure, an electromagnetic actuation mechanism is composed of an electromagnetically induced coil wound around a soft iron part and a permanent magnet attached to the micropump membrane [Zhou and Amirouche 2011]. The electromagnetic field generated by the AC current passing through the coil causes a periodically attractive and repulsive force on the permanent magnet and the attached membrane. Figure 2.5 Schematic view of an electromagnetically actuated diaphragm micropump [Yin et al. 2007] Depending on the design and the permanent magnet and electromagnet adopted, electromagnetic actuation has the potential of generating large deflections and actuation forces [Fu et al. 2005]. As an example, Yin et al. [2007] described the development of an electromagnetic membrane actuator with semi embedded coil and a permanent magnet fixed 17

34 above it, for pumping applications [Yin et al. 2007]. Using a flexible PDMS membrane with ø7mm x 150µm, a maximum deflection of 55 µm with an applied current of 0.5A was achieved. The amount of the normal electromagnetic force acting on the membrane through the permanent magnet is given by [Shen et al. 2008, Feustel et al. 1998]: ( 2.3) Where M y is the magnetisation of the permanent magnet and V is its volume. B y is vertical component of the magnetic flux caused by the coil. Electromagnetically actuated micro-pumps normally operate a lower voltage compared to electrostatic and piezoelectric actuators, i.e. with a typical value of 3-14V at frequencies up to about 1000 Hz [Amirouche et al. 2009]. Using electromagnetic actuation, micro-pumps with flow rates of more than 1 ml/min and a maximum pressure head as high as 28kPa have been reported. Electromagnetic actuators have a relatively lower efficiency compared to electrostatic and piezoelectric actuations, i.e., about 1% and less Shape memory alloys (SMA) motivator An actuation mechanism based on shape memory alloy materials (SMA) is provided by a transformation between material phases, i.e., between austenitic phase at high temperature and martensitic phase at low temperature [Amirouche et al. 2009]. The martensitic phase at low temperature is much more ductile than the austenitic phase and can allow large deformation. When pre-stressed and mechanically constrained SMA materials undergo a phase change that results in a mechanical deformation that can be used as an actuation mechanism. Figure 2.6, shows schematic of a SMA micro-pump, developed by Xu et al.[2001]. In this pump, a NiTi thin film is deposited on the silicon membrane. By heating and cooling the NiTi alloy, the reciprocating action of the membrane takes place. This pump with outer geometry of 6x6x1.5 mm 3, could provide 340µl/min at 1kPa pressure head. The operating frequency of this pump is up to 100 Hz, which is significantly improved compared to a previous SMA micro-pump developed by Benard et al. [1998] with an operating 18

35 frequency of 0.9 Hz. Also, a fatigue life of more than 4x10 7 cycles is reported for this micropump. Figure 2.6 Schematic of a SMA diaphragm micro-pump [Xu et al. 2001] Due to repetitive heating and cooling of the SMA alloys, these micro-pumps normally have a low efficiency and operation frequency, i.e. of about 1% and <100 Hz [Amirouche et al. 2009] Pneumatic and Thermo-pneumatic Motivators Figure 2.7, illustrates a pneumatically driven diaphragm micro-pump. As can be seen in this figure, a pneumatic diaphragm micro-pump requires an external source of pressurized gas and the driver is not fully integrated in the micro-pump [Begley and Mackin 2003, Rapp et al. 1994]. Therefore, it can only be used in a system where the required infrastructure is available [Laser and Santiago 2004]. The operating principle of these micro-pumps is that when high pressure gas flows into the secondary chamber, it pressurizes the operating fluid and a discharge stroke occurs. Then the pneumatic pressure acting on the membrane is released and the diaphragm relaxes. The pneumatic pressure, diaphragm structural properties, and high speed valves determine the diaphragm deflection and frequency of the actuator. 19

36 Figure 2.7 Schematic view of a pneumatically actuated diaphragm micro-pump [Laser and Santiago 2004] Pneumatic micro-pumps are capable of generating relatively large stroke volumes and therefore materials with low stiffness are usually used for the diaphragm to allow a larger displacement. However pneumatic actuation has a low response time and therefore their operating frequency is normally less than 100Hz [Amirouche et al. 2009]. Another type of actuation mechanism based on a similar principle is thermo-pneumatic actuation, which was first developed by Van de Pol et al. in As shown in Figure 2.8, the secondary fluid in this actuation mechanism is heated with a thin film heater. Heating and expansion of the secondary fluid cause the diaphragm to deflect and then discharging the primary working fluid. Figure 2.8 Schematic view of a thermo-pneumatic driven diaphragm micro-pump [Laser and Santiago 2004] 20

37 The response of thermo-pneumatic actuation is limited to the heat transfer rate. As a result, thermo-pneumatic micro-pumps operate at relatively very low frequencies, i.e., typically less than 5Hz. In some designs, to decrease the cooling time of the secondary fluid, the secondary chamber is vented Micro-valve mechanisms for controlling backflow Displacement micro-pumps which work in a periodic manner need to incorporate microvalves to control the backflow. The performance, ease of fabrication, reliability, and robustness of a diaphragm micro-pump rely highly on the micro-valve mechanism selected. Figures of merit for a micro-valve include diodicity, which can be defined as the ratio between the forward and reverse pressure drops across the valve, maximum operating pressure, ease of fabrication, dynamic response, and reliability [Laser and Santiago 2004]. Current micro-pumps employ different micro-valve designs for fluid control. These valves can be categorized into the following three major types [Oh and Ahn 2006, Amirouche et al. 2009]: passive micro-valves, which are driven by fluid pressure and viscous forces [Laser and Santiago 2004] and are used in most of the reported micro-pumps; active microvalves, which unlike the passive type are controlled and driven by an external electrical input signal; and valveless or dynamic valves, which instead of having a movable or flexible structure, incorporate a dynamically designed fluid channel that favours flow in one direction [Stemme and Stemme 1993, Xiao 2008] Passive valves Passive check valve are used for directing the flow from the inlet to the outlet. Normally a passive check valve consists of two major parts: a flexible (or moving) element which reacts to the fluid pressure, and a valve seat, which together with the flexible element, seals the valve and prevent the backflow [Zengerle and Richter 1994]. When the pressure difference is in the direction of the valve, the flexible part moves and flow occurs. If motion is in the opposite direction, the flexible part is pushed against the valve seat and blocks the passage of flow. However, in reality micro-valves can be leaky due to improper sealing of the valve and its seat or dynamic effects such as thin film damping. Many types of passive micro-valves are reported, such as cantilever valve (Koch et al. 1997), membrane (Nguyen and Truong 2004), 21

38 or ball (Pan et al. 2005). Some of the commonly used passive valve geometries are show in Figure 2.9. a) b) c) d) Figure 2.9 Structure of various passive valves [Yang 2004, Shen et al. 2008] Cantilever and circular diaphragm micro-valves, shown in Figure 2.9 a, b and c, are the most commonly reported types of the micro-valves in diaphragm micro-pumps. Other check valves, such as the ball valve shown in Figure 2.9 d, have been used in many designs. Other than the driver, check valves are the only moving components inside the micro-pump. Therefore the total performance, durability, and biocompatibility of the micro-pump rely on the check valves. The main challenges associated with the passive check valve are providing proper sealing, robust dynamic response, their sensitivity to solid particles, and the wear and fatigue of moving valves [Stemme and Stemme 1993, Amirouche et al. 2009] Active valves In an active valve, the moving element is actuated by an external force and not by the fluid pressure difference across the valve. Due to better controllability, active valves offer improved performance at the expense of fabrication and manufacturing complexity [Laser and Santiago 2004]. Active valves with electrostatic (Figure 2.10c), bimetallic (Figure 2.10b), thermopneumatic, piezoelectric (Figure 2.10 a), electromagnetic, and other actuators have been reported. Figure 2.10 shows the principle active valves. The operation principle of all of 22

39 these valves in directing the flow is pretty similar. However, the main difference between the performance of these mechanism is response time [Xiao 2008]. Amongst the active valves, piezoelectric and electrostatic valves have the shortest respond time. a)piezoelectric [Chakraborty et al. 2000] b) Bimetallic [Jerman 1994] c) Electrostatic [Ohnstein et al. 1990] d) Electromagnetic [Kwang and Chong 2006] Figure 2.10 Schematic of active check-valves used in micro-pump As the operation of active valves is easily controllable, better dynamic response and sealing and therefore improved flow characteristics can be achieved. Despite the fact that better performance can be achieved by incorporating active valves, they add much more complexity to the manufacturing and control of a micro-pump. A more detailed review of the active micro-valves can be found in Kwang and Chong [2006] Dynamic Valves Fluid flow in diaphragm micro-pump can also rectified by means of direction-dependent flow behaviour in specially designed channels. Micro-pumps with flow rectifying channels instead of check valves are also referred to as valveless, fixed geometry, or dynamic valves. A number of valveless micro-pumps have been reported using flow-rectifying channels such as 23

40 nozzle/diffuser [Stemme and Stemme 1993, Olsson et al.1997] and the Tesla principle [Morris and Forster 2003, Feldt and Chew 2002] in the inlet and outlet of diaphragm micro-pumps. Figure 2.11 Principle of Nozzel/diffuser valves [Stemme E and Stemme G 1993] a) Increasing chamber volume b) Decreasing chamber volume Figure 2.11 illustrates the principle of nozzle/diffuser valves. This valve utilizes the principle that the diffuser and nozzle can be designed to have a lower pressure drop in the diffuser rather than nozzle at the same velocity. Therefore it will favour the flow in one direction. In the supply mode, the inlet acts as a diffuser with low pressure drop and outlet acts as a nozzle with higher pressure drop. But in the pumping mode, the opposite of this occurs. Another type of dynamic valve is the Tesla valve. The principle of this valve is to use the momentum of the flow itself to increase the pressure drop in one direction, Figure As it is shown in Figure 2.12 b, when the fluid flows in the reverse direction, a portion of it enters the curved channel which changes the direction of its momentum, then when it re-enters the main channel, it causes a large pressure loss and flow separation in the channel. As a result, the flow rate in the reverse direction is much less than forward direction. In general, Tesla valves can provide higher diodicity than diffuser type valves [Xiao 2008]. 24

41 a) b) Figure 2.12 Principle of Tesla valves [Bendib and Olivier 2001] a) forward flow b) reverse flow Because dynamic valves do not have moving mechanical parts, they are able to avoid major problems encountered in check valves, such as high pressure losses, the sensitivity to solid particles, and the wear and fatigue of the moving valves [Stemme and Stemme 1993]. Further, another merit of using dynamic valves is ease of manufacturing and integration with the micro-pump. However, valve efficiency in a reverse flow direction is relatively much lower than the check valves since these channels are always open and leaking in the backward direction [Kwang and Chong 2006]. 2.2 Displacement- Rotary and Peristaltic Micro-pumps Rotary and peristaltic micro-pumps are two other categories of displacement micropumps. A few microscale rotary displacement pumps, mostly micro gear pumps, have been reported [Laser and Santiago 2004]. The operation principle of micro-pumps based on rotary elements such as gears or vanes is similar to the macro version of these pumps. Figure 2.13 and Figure 2.14 show three different rotary driving mechanisms. In Figure 2.13.a, which shows a rotary vane driving mechanism, the fluid is trapped between the vanes at the inlet and then pushed outward due to the centrifugal forces at the outlet. In this micro-pump, vanes, 25

42 inlet, outlet, and housing should be designed in accordance with the rotation direction to guide the flow in proper direction. Another type of rotary driving mechanism, shown in Figure 2.13.b, is a rotary disc, which was developed by Blanchard, Ligrani and Gale [2004]. In this method, instead of trapping the fluid between vanes or gears, the momentum is transferred from one or more rotating discs by means of fluid viscosity. Therefore this type of mechanism is more suited for high viscosity fluids. To obtain a reasonable amount of flow, the disc should also rotate at high speed. The simplicity of this driving mechanism makes it easier to manufacture than rotary gears or vanes. A common drawback of rotary micro-pumps is the need for an external driver, which influences the size of these micro-pumps. a. b. Figure 2.13 Driving mechanism of vane and disc rotary micro-pumps a. Rotary vane micropump driver [Ahn 1995] b. Rotary disc micro-pump driver [Blanchard 2004] Peristaltic micro-pumps are micro-pumps with multiple chambers in series and no valves [Laser and Santiago 2004]. They operate in a manner similar to macro scale peristaltic pumps. In the following, the operational principle and major features of rotary gear and peristaltic micro-pumps are described Micro Gear Pump Figure 2.14 shows the operating principle of a rotary gear micro-pump. The two rotary gears rotate in opposite directions. As a result, the gear teeth carry the fluid from inlet to outlet, as it is indicated in Figure However there is always an opposite flow from the 26

43 high pressure outlet to inlet through the meshed gear teeth. When the gears are well coupled together, due to the reduce space between the teeth, the amount of this backward flow is much lower than the forward flow. However, performance of a gear pump is highly dependent on having tight tolerances between the gears themselves and the housing. Figure 2.14 Schematic of operation principle of a rotary gear micro-pump [Dewa et al. 1997] An analytical formula for calculating the constant flow rate of a rotary gear micro-pump by considering the geometry of the gear wheels was developed by Dopper et al. [1997]: 2.4) Where d h is the diameter of the gears, a k is the center distance between the two gears, α 0 is the engagement angle of the gears, m is the module of the gears, h is the height of the gears, and n is the number of teeth. Due to the tolerances of the housing, e.g. the distance between the shafts, an internal leakage is inevitable. This leak flow limits the maximum achievable working pressure. The amount of this leakage can be estimated using equation ( 2.5,in which w and h are the width and height of the clearance, l is the length of the gap, Δp is the pressure difference, and η is the viscosity of the fluid. It is also assumed that w >>h and the flow is laminar, which are reasonable assumptions. ( 2.5) 27

44 Dopper et al. [1997] has reported a rotary gear micro-pump having 0.6 mm in-line gears fabricated by LIGA and using an electromagnetic motor as a driver. At operational speed of 2250 rpm, a maximum back pressure of 100 kpa and a maximum flow rate of 180μl/min were observed for a glycerine water solution. The maximum back pressure at which a gear pump can operate generally scales with the viscosity of the fluid. Dropper et al. [1997] also reported on the testing of a slightly larger gear micro-pump (1.2 mm diameter gears), with both pure water and glycerine water solution which has a higher viscosity. [Laser and Santiago 2004]. A gear micro-pump can generate relatively high vacuum pressures, which makes it capable of self priming [Dewa et al. 1997]. Another advantage of a gear micro-pump is that the output flow is continuous and smooth. One important challenge of using gear micropumps is the external driver, which other than increasing the size can cause external leakage around the shaft. Dewa et al. [1997] has adopted an integrated magnetic motor, which effectively decreased external leakage Peristaltic Micro-pump As mentioned before, peristaltic micro-pumps do not need check valves for rectifying the flow. The pumping principle is based on peristaltic movement of the pump chamber, which squeezes the fluid in the desired direction [Nguyen et al. 2002]. Theoretically, peristaltic micro-pumps require at last three consecutive chambers with reciprocating membranes. Peristaltic micro-pumps can be optimized by increasing the compression ration and the number of chambers. Most of the aforementioned reciprocating actuation mechanisms have been applied to peristaltic micro-pumps. Piezoelectric [Smith 1988, Shinohara et al.2000], electrostatic [Judy et al. 1991, Cabuz et al. 1999], thermopneumatic [Folta et al. 1992, Grosjean and Tai 1999] and electromagnetic [Pan et al. 2005] actuation mechanisms for peristaltic micro-pumps have been reported. In Figure 2.15, two of these peristaltic micropumps are shown. Both micro-pumps in this figure guide the flow from inlet to outlet by the peristaltic motion of the reciprocating membranes. 28

45 a) b) Figure 2.15 Peristaltic micro-pumps [Nguyen et al. 2002] a) piezoelectric driven [Smith 1988] b) Electrostatic driven [Judy et al. 1991] An electrostatically driven peristaltic micro-pump, presented by Cabuz et al. [1999], used numerous pumping chamber and a flexible membrane, with a three-dimensional array structure of the membrane, to optimize the micro-pump performance. The micro-pump reached a compression ratio of 10 and it was able to deliver 8 ml/min with only 4 mw electrical power Other displacement micro-pumps In addition to the described displacement micro-pumps, there are many other displacement micro-pumps with miscellaneous driving mechanisms such as bimetallic [Yang et al. 1996], electrowetting [Yun et al. 2001], thermocapillary [Sammarco and Burns 1999] and electrochemical [Bohm et al. 2000] displacement micro-pumps. Other than the described reciprocating and rotary micro-pumps, a number of different micro-pumps have been reported that despite having a moving surface performing pressure work on the fluid, the work is not performed in a periodic manner [Laser and Santiago 2004]. Aperiodic displacement micropumps based on electrochemical actuation [Bohm et al. 2000], thermocapillary [Sammarco and Burns 1999], and syringe pumps controlled by pneumatic or electromagnetic have been reported. Aperiodic micro-pumps are only suitable for pumping a finite amount of liquid. 29

46 Yun et al. [2001] has reported development of an electrowetting micro-pump based on the surface tension of the driven liquid metal (Hg) in an electrolyte. Figure 2.16 shows the schematic of this micro-pump. When voltage is applied between the two electrodes in this micro-pump, a potential difference between the liquid metal and electrolyte variable occurs [Yun et al. 2001]. This induces a charge redistribution in the liquid metal. As a result, the liquid metal will be more wettable (i.e. have lower surface tension), where the charges are more densely located. Therefore the surface tension in left and right of the liquid metal will be different and this gradient in the surface tension induces the motion of the liquid metal. Figure 2.16 Schematic of an electrowetting micro-pump [ Yun et al. 2001] A related class of micro-pumps based on thermally inducing surface tension gradient has also been reported [Sammarco and Burns 1999]. 2.3 Dynamic- Non Mechanical Micro-pumps Non-mechanical dynamic micro-pumps are based on a direct transformation of nonmechanical energy to fluid momentum. Devices with electro-hydrodynamic, electro-osmotic, magneto-hydrodynamic, which are all based on a direct transformation of energy from an electromagnetic field to the fluid, and ultrasonic actuation methods have been developed. Generally, because these micro-pumps do not have any mechanical moving parts, they have a simple structure. However, in most cases their applications are restricted to only certain types of fluids, as their performance relies on fluid properties [Woias 2005]. As the design and 30

47 operation principles of this class of micro-pumps are quite different and depend on their physical and chemical principles, only a brief overview of them will be given here Electro-hydrodynamic (EHD) When an electrostatic field is applied to a dielectric fluid with low conductivity it can result in fluid flow and pressure. Electro-hydrodynamic (EHD) micro-pump operation is based on the electro-chemical formation and movement of charged ions [Richter and Sandmaier 1990]. Since 1989, various EHD micro-pumps have been reported [Woias 2005]. They use the EHD induction effect [Bart et al. 1989], i.e., the generation and movement of charges at boundary layer, and the EHD injection effect [Richter and Sandmaier 1990, Richter et al. 1991], which is based on electrochemical formation and movement of charges. As an example of the performance capability of these micro-pumps, the micro-pump of Richter and Sandmaier [1991]achieved a maximum outlet pressure of 24 mbar at 700V and 14 ml/min with ethanol. In this micro-pump by reversing the poles, bidirectional pumping can be achieved. Despite the simple concept of these micro-pumps, all EHD micropums rely highly on the electric properties of the fluid, i.e., its permitivity and conductivity [Woias 2005]. Therefore, they cannot be used with non-conductive and non-ionic fluids Electro-osmotic The operational principle of an electro-osmotic micro-pump is based on the fact that most surfaces acquire finite electrical charges when they come into contact with an aqueous solution [Hunter 1981]. Figure 2.17, illustrates a schematic of the basic principle of an electroosmotic (EO) micro-pump. When glass (or silica) contacts an aqueous electrolyte, deprotonation of the surface silanol groups charges the glass surface [Chen and Santiago 2002]. As shown in Figure 2.17, the charged surface attracts counter-ions close to the surface and forms a charged double layer. By applying an external electrical field, mobile ions in the charged double layer are driven and then, due to viscous effects, a bulk liquid motion results. In this actuation method, the pressure difference along the channel applies a resisting force to the flow. In order to obtain the net flow rate the flow caused by such a pressure difference can be linearly subtracted from the electro-osmotic flow. 31

48 Figure 2.17 Schematic of basic principle of an electroosmotic micro-pump [Chen and Santiago 2002] Performance of an electro-osmotic micro-pump depends on the strength of the electric field, the dimension of the channel, and the fluid and wall properties. Typically they require a very high voltage (more than 1000V) and operate at very small flow rates. To improve the performance and make their manufacture easier, most electro-osmotic micro-pumps use a porous structure rather than a simple channel [Yao et al. 2006]. As an example, using porous media, Paul et al. [1998] reported an electro-osmotic micro-pump with 4000kPa pressure head and 0.04µl/min flow operating at 1500V. Similar to electro-hydrodynamic micro-pumps, a electro-osmotic micro-pumps has a smooth flow. But unlike EHD micro-pumps, which are able to induct or inject ions into the fluid, EO micro-pumps can only work with ionic conductive fluids Magneto-hydrodynamic The basis of a magneto-hydrodynamic (MHD) micro-pump is that when current-carrying ions in a fluid are subjected to a magnetic field perpendicular to its motion, Lorenz forces are applied to the ions. The force imparted will be perpendicular to the direction of both the electric and magnetic fields. As the motion of the positive and negative ions in the electric field and are opposite, the applied force will be in the same direction for both charges. As a result, ions and consequently the fluid will gain momentum. Figure 2.18 illustrates a schematic of a magneto-hydrodynamic driven micro-pump. 32

49 Figure 2.18 Schematic of a Magneto-hydrodynamic driven micro-pump [Nisar 2008] A MHD micro-pump generates only a small amount of flow and pressure head, i.e. typically less than 0.1 ml/min and 0.2 kpa. Jang and Lee [2000] reported on a magnetohydrodynamic micro-pump with a 40 mm chamber length, 1 mm hydraulic diameter, and having a magnetic flux and maximum total current of 100 μa. The performance achieved was maximum flow of 63 μl/min, with a maximum pressure of 170 Pa. A major problem of MHD micro-pump is bubble generation due to ionization [Nisar 2008]. This problem can be solved by using AC current and an electro-magnet instead of DC current and a permanent magnet [Lemoff and Lee 2000] Ultrasonic Ultrasonic pumping is based on the phenomenon of acoustic streaming [Nuyen and White 1999]. When a mechanical wave propagates in a thin membrane at a liquid-solid interface, a high intensity acoustic field appears in the fluid near the membrane which exerts a drag force in the fluid in the direction of wave propagation [Moroney et al. 1991, Nuyen and White 1999]. 33

50 Figure 2.19 Model of ultrasonic driving of micro fluid [Nuyen and White 1999] 2.4 Comparison and evaluation of the Micro-pump driving mechanisms Table 2.1 compares the previously discussed micro-pump actuation mechanisms. The data in this table is based on previous reviews of micro-pumps and shows typical ranges and values of the actuation mechanism properties of the reported micro-pumps [Amirouche et al. 2009, Laser and Santiago 2004, Tsai and Sue 2007, Chen et al. 2008]. For more information, one may refer to these papers that reviewed performance and properties of micro-pumps with various actuation mechanisms. According to the literature survey on driving mechanisms of micro-pumps, nonmechanical micro-pump performance is strongly dependant on the electrical properties of the fluid. These micro-pumps normally have a very small flow rate (< 1 ml/min), and other than the electro-osmotic, they have very low pressure head, i.e. less than 1 kpa. However electroosmotic micro-pumps have achieved maximum pressures as high as 20,000kPa for a very small flow rate, about 0.2 μl/min, by using porous media [Paul et al. 1998]. Electro-osmotic actuation mechanisms typically require a very high voltage (>1000V), and magnetohydrodynamic actuation generates very intensive magnetic field in the fluid, which may influence the fluid properties [Xiao 2008]. Other than ultra-sonic actuation in which the flow profile is affected by acoustic waves propagating in the fluid-structure boundary, nonmechanical actuation mechanisms normally have a flat flow profile with time. 34

51 Displacement Dynamic Table 2.1 Comparison of micro-actuation mechanism of typical micro-pumps Applicable flow Flow profile Selfpriming Freq. Voltage Maximum Pressure Flow rate Electrohydrodynamic dielectric fluid, low conductivity flat no n/a V 0-1 kpa < 14 ml/min Electro-osmotic electrolytes flat no n/a V 0 2x10 4 kpa < 100 μl/min Magnetohydrodynamic electrolytes flat no n/a n/r kpa < 1 ml/min Ultrasonic Liquids oscillating n/r 2-3x 10 6 Hz n/r n/r < 2 μl/min Rotary Gear Liquid/air flat yes n/r kpa <10 RPM ml/min Pneumatic/ Liquid/air oscillating no-yes V kpa < 4 thermo-pneumatic /10 Hz ml/min Piezoelectric Liquid/air oscillating no-yes , < 16 Hz V kpa ml/min Electrostatic Liquid/air oscillating no-yes kpa < 1 Hz V ml/min Electromagnetic Liquid/air oscillating no-yes kpa < 3 Hz V ml/min Shape memory Liquid/air oscillating n/r kpa < 0.1 alloys Hz V ml/min n/a: not applicable n/r: not required Gear micro-pumps can generate relatively high vacuum pressures, which make them capable of self priming and they can tolerate bubbles [Dewa et al. 1997]. Another advantage of gear micro-pumps is that the output flow is continuous and smooth. Due to internal leakage between the walls and gears, the maximum achievable pressures of these micro-pumps are highly depended on the viscosity of the fluid. One important challenge of using gear micropumps is the external driver, which other than increasing the size, can also cause external leakage around the shaft. Unlike rotary and dynamic micro-pumps, reciprocating displacement micro-pumps have an oscillating flow and require a flow control mechanism [Xiao 2008]. However, reciprocating displacement micro-pumps have the advantage of working with a wider variety of fluids (e.g., with various viscosity and even two phase flow or gases). They, unlike rotary 35

52 micro-pumps, do not have the leakage problem to outside of the pump. They also do not require very high voltages as it is required by dynamic pumps. Amongst the reciprocating actuation mechanisms, a piezoelectric can provide a wider range of flow rate and pressure. Due to its capabilities in operating at various frequencies and providing a high actuation force and good deformation (normally higher than electrostatic and shape memory alloys), the piezoelectric activator is suitable for most of applications. Another important advantage of the piezoelectric actuator is that they are commercially available as standard parts in many different sizes and materials. 2.5 Diaphragm micro-pump simulation and analysis The operational principle of a diaphragm micro-pump is based on the interactions between various mechanics, such as piezoelectric materials, electrostatic forces, structural mechanics, and fluid mechanics. Basically, the problem is dealing primarily with fluidstructure interactions and further structural-actuation method mechanics interactions. Unlike macroscopic reciprocating pumps which have a forced volume stroke, structural displacement of the diaphragm is coupled with both an actuation force and fluid dynamic behaviours. Furthermore, micro fluid dynamic effects such as squeezed film damping forces add to its complexity [Starr 1990]. Despite these complications, in most of the publications on the micro-pumps, partial simulations have been performed. But because of the complex nature of micro-pump simulation, comprehensive and accurate models are not readily available. One of the earliest micro-pump simulations was performed by Van der Pol et al.[1990], in which he presented a computational model for the dynamics of an electro-thermopneumatic micro-pump based on bond graph technique. A bond graph is mathematical system modeling method using power and energy relationships for lumped elements. Later, Zengerle et al. [1993] developed a general theoretical model for diaphragm micro-pumps based on the continuity equation. By combining the mass flow in and out of the chamber with the time dependent change in pump chamber volume due to valve and membrane displacements, the basic differential which describes the pump chamber pressure P is obtained [Zengerle et al.1993]. 36

53 ( 2.6) In this equation, and are the valves flow properties, V 0,, and are the total pump chamber volume, the membrane displaced volume, and the volume displacement of inlet and outlet valves, respectively. After solving this equation by integrating or with respect to the pressure, the net flow rate can be obtained. This equation was solved with dedicated software and validated by measurement [Zengerle et al. 1994, 1995]. An important advantage of this model is that no simplification is applied to the equation and therefore there is no error from the equation itself. However, the methods used for predicting the flow properties of the valves ( ), the membrane displacement, and the membrane and valves stiffnesses ( ) can be possible sources of error. Also, a bubble trapped in the chamber or chamber flexibility, which is considered in the term, cannot be easily estimated. Despite the advantages of this model for modular analysis of micro-pumps, only a few publication from the same research group have reported the use of this method. Also, in these publications, large deviation from the experimental results was observed at some frequencies due to not considering the valves variable response to frequency and their natural frequency [Zengerle et al. 1995]. An alternative to analytical simulations are numerical methods such as finite element methods (FEM) or computational fluid dynamics (CFD) analysis. The use of multi-physics analysis software such as COMSOL and ANSYS has been extensively reported in performing coupled analysis of micro-pumps. Koch et al. [1995] reported dynamic flow simulation for a simple cantilever valve, and more recently, Abdul [2010] has performed flow simulation and shape optimization for a simplified 2D micro-pump by linking MATLAB and COMSOL. However, increased model complexity or finer meshing makes numerical multi-physics analysis highly time consuming, and in some cases when modeling the whole micro-pump, infeasible. Bourrouina and Grandchamp [1996] reported an electrical analogy method for the simulation of micro-pumps. This method is based on describing the system by a network of 37

54 mechanical impedances (possibly nonlinear) and subsequently carrying out an analysis using electrical simulation tools [Bourrouina and Grandchamp 1996]. Sub divisions of the system are considered as lumped elements and then, on the basis of physical and mathematical relevance, are described as electrical elements. In this model, flow rate and pressure (P) are analogous to electrical current and voltage respectively. For each lumped element, an electrical impedance is considered, which is defined as the pressure drop to flow rate ratio. ( 2.7) As shown in equation ( 2.7, viscous, inertial and elastic effects of each elements are considered in the model. Similar to the parameters in Zengerle et al. s [1993] method, the impedances in this model must be calculated separately by means of other analytical, experimental, or numerical methods. Figure 2.20 illustrates the equivalent networks of a pneumatic and an electrostatic driven micro-pumps, as developed by Bourrouina and Grandchamp [1996]. 38

55 Figure 2.20 Equivalent electrical network of micro-pumps with a) pneumatic actuation b) electrostatic actuation [Bourrouina and Grandchamp 1996] Results from the electrical equivalent of pneumatic micro-pump shown in Figure 2.20 a have been compared to Zengerle et al. s [1993] model and were validated. As can be seen in Figure 2.20 b, this method also gives the capability of combining mechanical and electrical subsystems together. As described in this section, most of the reported simulation methods of micro-pumps have incorporated a modular approach. In these simulations, use of electrical [Bourrouina and Grandchamp 1996] and vibrational [Accoto et al. 2000] analogies, bond graphes [Van der Pol et al.1990], and basic fluid mechanics equations, e.g. Navier-Stokes [Pan et al. 2003] and continuity [Zengerle et al. 1993] equations, as the core of their modular simulations have been reported. Despite the differences between modular simulation approaches, these simulation methods need to have the actuation mechanism and the membrane and valves properties 39

56 specified or obtained by some other analytical, numerical (such as FEM), or experimental method. As a result, several papers have discussed passive and active flap valves [e.g. Luharuka et al. 2007, Voigt et al.1998, Carmona et al. 2001], and many others have studied fixed geometry valves. Similarly, a large number of publications have reported numerical and analytical simulations for the dynamic behaviour of various actuation mechanisms. 2.6 Micro-pump Manufacturing Methods As reported by van Lintel et al. [1988], the high precision of silicon micromachining combined with glass bonding of layers makes it one of the most commonly used fabrication methods since the earliest micro-pump designs [Laser and Santiago 2004]. Woias et al.[1998] and Kluge et al. [2001] have shown in long term tests that wear and fatigue do not occur in flexible silicon micro-pump parts. The micromachining process of silicon is based on photolithography methods [Judy 2001], such as etching, lithography, and thin-film growth. The rather high fabrication costs and restriction of material to silicon has motivated the development of alternative fabrication methods [Woias 2005]. Microinjection moulding and hot embossing of polymers have been demonstrated as proper methods for mass fabrication. The micro-stereolithography method has been reported as an alternative fabrication method, useful for prototyping and with an accessible resolution of 5-10 µm [Piotter et al. 2002]. Precise conventional micromachining has also been used for the fabrication process of a number of micro-pumps. Stemme and Stemme [1993] reported fabrication of a piezoelectric driven micro-pump fabricated by precision machining of brass. Recently, the increasing use of injection moulding process has been facilitated by high precision micro milling for the mould machining. Selecting the best manufacturing method for a micro-pump depends on its material, thickness, design, and its suitability for mass fabrication. Despite the fact that the fabrication methods can significantly affect micro-pump costs, the cost is not exclusively defined by the fabrication technology itself, but mainly by an intelligent and fabrication-oriented design [Woias 2005]. To achieve a low-cost micro-pump, after selecting proper manufacturing 40

57 methods for each part, the design should be tailored to be within the selected processes capabilities Photolithographic processes Photolithography is a series of processes which are used to selectively remove material of a thin substrate. Conventional photolithography processes are normally based on initially forming the desired pattern on a thin layer of a photoresist layer on top of the substrate material, by using a lithographic process and then using a deposition (an additive process) or an etching (a subtractive process) process to add or remove the desired pattern on the substrate material [Judy 2001]. Using lithography with an etching process for micro-component fabrication has the advantage of precise control of dimensions and its suitability in processing very thin materials. Furthermore, by using batch fabrication in this method, a large number of micro components can be made from one lithography mask at a low cost Wet (Chemical) Etching The wet etching process as illustrated in Figure 2.21 can be briefly described as follows: a thin sheet is flat mounted and the surface is cleaned and dried. A thin layer of a photosensitive polymer (photoresist) is applied and exposed to UV light through a mask, the mask previously being designed and manufactured to have the pattern of the component. The UV light causes the exposed area of a positive photoresist to become acidic. Dipping the wafer into an alkaline solution removes the more acidic parts exposing a planar pattern and the remaining photoresist acts as a protective layer from the etchant solution during wet etching. The initial etching occurs normal to the plane, but as the etching process is isotropic, after some time the side walls become exposed to the etchant. Therefore, as shown in Figure 2.22a, an undercut will occur. To reduce the amount of undercut, anisotropic wet etching methods were developed, which are capable of having very different etching rates at different directions, depending upon the crystalline orientation of the material and the etchant solution [Seidel et al. 1990]. 41

58 Figure 2.21 Wet etching lithography process a) substrate covered with photoresist b) exposing photoresist to U.V. light using a lithography mask c) removing exposed area by dipping substrate in alkaline solution Figure 2.22 Etch profile a) isotropic etching b) perfectly anisotropic etching In many cases, depending on the substrate material and the etchant solution, prediction and control of the wet etching progress cannot be easily performed. For an impure material like stainless steel shim stock with anisotropic grain structure that causes different rates of material removal at different directions and areas, process optimization is required to achieve a good quality etch [Rao and Kunzru 2007] Dry etching methods Dry etching covers a family of methods for selectively removing material from a solid surface by using bombardment of ions from a plasma of reactive gases [Madou 2002]. Unlike most of the wet etching processes, which progress in an isotropic manner, dry etching processes typically progress directionally or anisotropically. 42

59 In dry etching, as ion can be directed by an electric field, a better control over the etch profile can be obtained over wet etching [Wilkinson and Rahman 2003]. Normally in this process ions can remove material in two major ways. Surface material can be removed due to direct momentum transfer from ions. Also unstable products from a reaction between the reactive gas composition and the material can be removed by the ions. Depending on ion generation, control methods and using reactive or non-reactive gases, various methods of dry etching exists. Further to the wet and dry etching methods mentioned, other micromachining methods useful in micro component fabrication have been reported, such as laser etching [Nath 1988] and micro-electrical discharge machining (EDM). Micro-EDM is a newly developed method for fabrication of micro-parts with a relatively lower cost than laser etching [Mahendran et al. 2010]. Currently a minimum machineable diameter of 5µm at best, has been reported using micro-edm, which indicates further improvements are required for this method. Also, residual stress caused by EDM may result in distortion of the micro-parts Micro-Injection Moulding Micro-Injection moulding is the process of melting and forcing small amounts of material into a mould cavity where it cools and takes the shape of the cavity. The capability of fabricating very small and geometrically complicated components with a high accuracy and in a very cost efficient way makes micro-injection moulding a very promising method for microcomponent fabrication. Micro injection moulding allows manufacturing of plastics having a minimum wall thickness of less than 20 μm, with a maximum aspect ratio as high as 20, the smallest features in the range of a tenth of a micrometer, and a surface roughness of about Rz < 0.05 μm [Piotter 2001]. Other than polymers, micro injection moulding can be applied to ceramics (Ceramic Injection Molding, CIM), metals (Metal Injection Molding, MIM) and fibre enhanced polymers [Piotter et al.2003]. In any manufacturing process, when materials undergo a phase change, shrinkage occurs as a result of the decrease in the part s specific volume. The polymer packed in the mould shrinks from the time it fills the mould in its molten state until it reaches room temperature. Because of variable cooling times, non-uniform thicknesses, and material orientations, 43

60 shrinkage will not normally occur uniformly in the part. Non-uniform shrinkage can cause residual stresses and strains (warpage) in micro components. In order to have a low shrinkage injection moulding process, an amorphous polymer such as cyclic olefin copolymers (COC) or PMMA is advantageous over semi-crystalline or crystalline polymer, as the additional contraction caused by crystallization of polymer chains while cooling does not occur in amorphous polymers [Whynott 2010, Van der Beek et al. 2008]. Also, by controlling injection moulding parameters such as moulding temperatures, injection time and the holding pressure, shrinkage and warpage could be reduced [Kramschuster et al. 2005]. In order to apply injection moulding to micro components, the moulding process should be considered during the part design process. Injection moulding parameters such as melt temperature, injection pressure, mould temperature, and injection time should be determined considering the properties of the melt material and the geometrical characteristics of the part. In order to have complete mould fill, low and uniform shrinkage, and minimum part warpage, the process parameters should be optimized either experimentally or using a modeling software such as "Mold Flow", although generally both methods are utilized. The moulds should have a proper draft angle to all surfaces normal to the mould parting line and the right shrinkage factor. Suitable runners and gates system should also be designed to ensure complete and uniform filling of the mould cavities Micro-milling Micro-milling is a form of high-speed CNC machining that is perceived by many as the most versatile method of fabricating miniaturized components in complex three-dimensional shapes from a variety of engineering materials [Malekian 2010]. In micro-components fabrication, micro milling can be an alternative to the lithography techniques for direct fabrication of parts or it can be used for mould machining in micro-injection moulding. Unlike most of the lithography techniques, micro-milling can be performed out of most materials. Achieving optimal results in micro-machining operations depends on variety of factors. A micro-cnc machine should be able to control positioning in each individual axis within submicrometer accuracy [Xiao 2008]. Furthermore, reaching very higher spindle speeds rather 44

61 than normal CNC machines is required, which allows obtaining a desired speed for using cutting tools as small as 100 µm in diameter. As an example, KERN has developed micromilling machines with a spindle speed as high 160,000 rpm, positioning precision of less than 0.5 μm and a surface quality with Ra <0.1 μm [KERN Evo 2012]. To achieve the desired accuracy and speed, the CNC should have a very high static and dynamic stability and rigidity. In addition, micro cutting tools are very fragile and can easily worn or subject failure due to excessive forces and vibrations [Malekian et al. 2009]. Therefore cutting forces monitoring and prediction are very important. Finally despite the challenges in micro-milling technology, it is a very versatile method for both direct fabrications of micro-components or in micro-mould machining applications. 45

62 3 Chapter 3: Micro-Pump Design and Simulation At the beginning of the design process, a number of design requirements and specifications were defined based on application requirements. Specifications for the target design were established by reviewing existing micro-pumps and identifying future customer needs. Learning from the last iteration of the micro-pump design by Xiao [2008] was applied to improve the current design, as well as the application of design for manufacturing methods. In order to achieve the objective of having a low cost, design for manufacturing and assembly (DFM and DFA) were considered in defining the design requirements and developing the micro-pump design. To ensure achieving the desired performance, the micro-pump performance is simulated and various tests and analysis are performed on the micro-pump components. The micropump simulation and detailed design parameter selection was performed by using a modular analysis process with a core analytical model that brings different analysis and tests results together. This process is described in the micro-pump simulation and analysis section. The most influential design factors found from the results of this analysis using DOE techniques, and the best parameter values were selected. A design of experiment was also performed on the micro-valves to improve the micro-pump analytical model and can be used in later design iterations to optimize the micro-pump design. Tolerance design was also performed on each part considering its manufacturing process and the overall performance requirements. 3.1 Design specifications of the micro-pump In the beginning of a product design process, the design objectives and application requirements are normally in the customer's language and are used for defining design requirements and specifications. Clearly defined design requirements can provide guidelines for the design and facilitate the design process. Table 3.1 presents the specification and design requirements derived from the application requirements. The driving mechanism is one of the major parameters that influences the performance, overall dimension, reliability and cost of the micro-pump. In this work, despite that a piezoelectric was required for the driving mechanism in the project definition, it allows a 46

63 relatively simple and compact design (compared to electromagnetic, electrostatic, pneumatic, etc.) for the micro-pump. By adjusting the operating frequency, the flow rate can be controlled. The size and type of piezoelectric diaphragm significantly influences the flow rate, pressure head and the overall size of the micro-pump. Standardized piezoelectric discs can be obtained with different standard diameters (e.g. 10, 15, 20 mm) and diaphragm material (e.g. brass, steel or even plastic). The diameter of the piezoelectric disc used in both of the ThinXXS and Xiao micro-pumps, for achieving a comparable performance, was 20 mm. The overall dimensions of the micro-pump are mainly affected by the diaphragm/ pumping chamber dimensions and layout of the micro-pump. Having the inlet and outlet in the same plane as the body of the micro-pump, will make the micro-pump more suited to be integrated in different fluidic systems. However it can increase the package size of the micro-pump. To keep the price of the micro-pump low, the cost for fabrication and assembly processes should be reduced by minimizing the number of parts and designing them for low cost mass fabrication methods, such as injection moulding, chemical etching or embossing. The capability of fabricating very small and geometrically complicated components in a cost efficient manner makes micro-injection moulding a very promising method for microcomponent fabrication. Injection moulding also gives the flexibility of using different materials for different application requirements. In addition, using an standardized piezoelectric disc has the advantages of having the mass fabrication cost for the piezoelectric. Generally the assembly processes form a large portion of the overall cost of a microsystem. Therefore design for assembly should be considered in the micro-pump design. Designing the micro-pump for one directional and automatic assembly, modularizing parts, and mistake proofing the assembly by considering alignment features can reducing the assembly process costs. To keep the micro-pump suitable for most of the disposable and low cost applications such as drug injection and fuel cells, at this stage of design a reliable service life of 100 hours is considered. In future work, by quality control and testing of the micro-pump, this service life can be improved. Designing the micro-pump for a service life of higher than 100 hours at 50Hz requires that the dynamic parts of the micro-pump have a fatigue life of higher than 47

64 2x10 7. To ensure the fatigue life of the micro-valves, as the smallest oscillating parts inside the micro-pump, it is desired to keep the maximum stresses under the endurance limit of the stainless steel shim stock with a factor of safety of higher than 1.5. Generally, the fatigue life of the standard piezoelectric actuators is much higher than the desired value of 2x10 7 [Piezo Systems Inc. 2012]. However accurate estimation of the fatigue life of the piezoelectric actuators under the actual loading and applied voltage requires a long test time for the micropump. Therefore the fatigue life of the piezoelectric, together with the reliability objective of having less than 0.5% failure rate in 100 hours, are beyond the scope of this design work and should be tested prior to mass fabrication. 48

65 Table 3.1 Micro-pump design specifications Application Requirement Design Specifications Performance Flow rate > 5 ml/min for water Adjustable flow Controlling the flow rate by adjusting operating frequency Pressure head 50 kpa for water Power consumption < 1 W Working fluid North American Tap water Overall dimensions/ Package size < (30 x 25 x 10 mm) / <7.5 cm 3 Compactness Micro-pump layout Inlet and outlet in the same plane as the body Fabrication process Casings: injection moulding Fabrication process Chemical etching /embossing of the micro-valves Low Cost Fabrication process Using standardized piezoelectric Fabrication and Assembly: minimum number of parts Modularizing and integrating the components, e.g. integrating the microvalves parts in one piece Assembly process One directional assembly process Mistake proofing the assembly process Using alignment features Service Life >100 hours Valve Fatigue Life > 2x10 7 cycles Reliability Reliability No more than 0.5% failure rate in 100 hours /Durability Operating Temperature Room Temperature Self priming < 20 sec. Having no leakage Considering proper sealing features and reliable bonding method. Work in different orientations, during movement 3.2 Concept Design The earliest stages of design are the most critical and sensitive, as they are highly influential to cost, manufacturability, and development time. Therefore many new ideas were discussed in group meetings at this stage of the design process. The final design concept for the micro-pump is shown in Figure

66 Figure 3.1 Schematic of selected micro-pump design concept This design is composed of four parts, the upper and lower casing halve, a single plate inlet and outlet valve, and a piezoelectric disc. Casing parts are moulded of plastic and the valve plate is etched from stainless steel shim. The valves are made in one piece by etching on a single plane and then bending at its mid section. The valve plate is assembled between the upper and lower casings which secure it from above and below. The outer section of the upper casing, which overlaps with the lower casing and the valve, seals the micro-pump casing from leakage to outside. Valve performance is an important part a micro-pump design, with the valve configuration and casing interface affecting performance. The valve should be capable of passing sufficient flow, minimizing back flow, have a proper dynamic response, and satisfy the other micro-pump design requirements. Many new micro-valve designs were suggested to achieve the dual objective of low price and good performance. Some of the different ideas for the valves and casings, considered in the configuration design process are shown in Figure 3.2 to Figure 3.5. Considering injection moulding as a desired manufacturing process for the casings, the designs in Figure 3.2 and Figure 3.3 are suggested. Both of these designs can be injected as two pieces and then assembled together. The valves are placed in the middle of the casings and provide a sealing by covering the valve seats that are formed by the casing. However in the first design, due to the clearance required for placing the valve, the inlet and outlet are not 50

67 well separated. In the design in Figure 3.3, this problem would probably be solved by applying a pre-stress caused by the deformed position for the valve. However, from previous experience placing a valve in the deformed pocket, this design would have handling difficulties for assembly. Furthermore, considering design for assembly, it is desirable to have one directional assembly, which cannot be achieved from these designs. Figure 3.2 Simple cantilever beam valve micro-pump (Valve Design #1) Figure 3.3 Simple cantilever beam valve micro-pump, pre-stressed valve in the middle of the casings by a pre-stress feature (Valve design #2) Using the design concept described in Kan et al. [2005], a modular valve design which allows one directional assembly as shown in Figure 3.4 was considered. Later on, this design was changed and improved to a more compatible concept for low cost manufacturing. 51

68 Figure 3.4 Three plate cantilever valve. [Kan et al. 2005] Also, benchmarking an existing micro-pump design, ThinXXS [Weber 2005], additional design ideas were developed. Figure 3.5 shows a model made from the ThinXXS micro-pump valve. This model was obtained by using profilometer and taking high precision photograph from the valve. Figure 3.5 Valve design of ThinXXS TM micro-pump [Weber 2005] Figure 3.6 Schematic of a simple 2D flap check valve As illustrated in Figure 3.6, the valve selected for development was a simple flap check valve. It is composed of a cantilever beam and a valve seat. A planar illustration of the valve configuration is shown in Figure 3.7. This new modular concept for this valve was developed 52

69 by focusing on design for manufacturing (DFM) and design for assembly principles, where the inlet and outlet valves are integrated into one piece, as illustrated in Figure 3.7a. By fabricating the both the cantilever valves and their valve seats on a single planar piece, and then bending the sheet, a three dimensional valve structure is formed. As can be seen in Figure 3.7c, a bend line composed of small holes is formed into the part, easing its fabrication and ensuring the relative orientation of the valve to its seat. Having these holes also provides a stress concentration at the bend line and ensures bending occurs on the proper line without causing deflection or undue stress at more sensitive parts of the valve. The bent valve in Figure 3.7b is sandwiched between the upper and lower casings, as shown in Figure 3.1, where the two halves of the housing provide the required force for closing the gap in the middle of the valve, and thereby sealing and secluding the inlet and outlet valves. Orientation features at the circumference of the valve plate and their correspondent features in the pump casings ensure the correct orientation of the valve when assembled inside the pump. a) b) c) Figure 3.7. Valve Design a) one piece module for inlet and outlet, with a simple cantilever valve b) bent valve module c) detailed view of holes forming bend line. Figure 3.8 shows, a number of different shapes considered for the cantilever design of the valve. Many different analyses were performed on these valves, which will be described in more detail in the detailed design section. In Figure 3.8, symmetric designs in b), c) and d) will cause a symmetric flow, but also require more than one hole. After discussion, it was decided to design a valve which can guide the flow toward the center of the pumping 53

70 chamber. Therefore the design in Figure 3.8e was developed, and after optimizing it for fatigue and stress distribution, the design in Figure 3.8a was developed. After discussing and evaluating these ideas, the design in Figure 3.8a was selected for the micro-pump design. A detail design of the micro-pump illustrated in Figure 3.9 was modelled and analysed from the concept illustrated in Figure 3.1. a) b) c) d) e) f) Figure 3.8 Different shapes designed and evaluated for valve moving parts 54

71 Figure 3.9 Model of the initial concept selected for prototyping The pins and holes in the casings and the edge of the valve are designed for facilitating and mistake proofing the pump assembly process and aligning the valve in the right place. The valve can be assembled between the casings only in the correct direction. A first prototype was made by machining the casing from a block of plastic. After making the first prototype, a number of modifications were made to the casing design to prepare it for injection moulding. The final design made for injection moulding is illustrated in Figure Figure 3.10 Model for the micro-pump with injection moulded casing design 55

72 3.3 Micro-pump Simulation A modular system for analysis of the micro-pump was developed to simulate the micropump behaviour and develop a comprehensive micro-pump model. This modular simulation was based on an analytical model for a diaphragm micro-pump adopted from Zengerle et al.[1993]. The modular model of the micro-pump is mainly composed of three correlated modules: a main analytical model of a diaphragm micro-pump, representing the fundamental fluid dynamics continuity equation, and two models for relating the valves and PZT disc performance to the corresponding terms in the continuity equation. There is no error in the model due to the continuity equation itself. However, sources of model error derive from uncertainty in the model for relating the flow and pressure difference across the valve and from the piezo-electric disk performance. Figure 3.11 illustrates the connections between the various analysis and test results used in the simulation. Figure 3.11 Modular system for analysis of the micro-pump Analytical modelling of the micro-pump The diaphragm micro-pump dynamics can be modeled by using of the continuity equation, which combines the mass flow into the pump chamber with the transient change in pump chamber content: 56

73 ( 3.1) where j is the time dependant mass flow intensity into and out of the pump chamber, and m is the mass content in the chamber. If the inertia of the check valves can be neglected, the mass flow is the integral of the chamber and can be written: ( 3.2) In this equation and are the flow passed through the valves, which is a function of the differential pressure across inlet and outlet valves, respectively, (, ). The time dependent volume change of the elastic components can be calculated for the time dependent mass change in the pump: ( 3.3) where V 0,V iv and V ov are the total pump chamber volume and the volume displacement of inlet and outlet valves, respectively. An important part of this equation is, the volume displacement of the pump diaphragm (piezoelectric disc), which is a function of the pump chamber pressure P(t) and the driver actuation parameter A(t). As described by Zengerle et al. [1993], the pressure distribution inside the pump chamber can be found by replacing equation ( 3.2 and ( 3.3 into equation ( 3.1: ( 3.4) By solving this equation ( 3.4, the pressure distribution inside the micro-pump chamber can be obtained. In this equation φ iv (P iv ) and φ ov (P ov ) are the respective flows passed through the inlet and outlet valves as a function of the pressure difference across them. An orifice model with ideal diodic performance was used initially as there was no model available for characterizing backflow early in the design process. 57

74 The pressure distribution was calculated considering a 50 kpa pumping pressure head and a frequency of 30 Hz for the micro-pump piezoelectric disc, as illustrated in Figure This pressure distribution is used for the loading in a transient multi-physics analysis and vibrational modeling of the valve. Figure 3.12 Micro-pump chamber pressure as a function of time Figure 3.13 shows the flow through the outlet valve. The outlet flow starts as soon as the chamber pressure exceeds the total outlet pressure (150kPa absolute, or a 50kPa pressure head) and stops when the valve completely closes when the pressure falls below this value. As can be seen, back flow is not observed in this figure because of the ideal diodic model used for the valves. Flow characteristics of the valves will be modeled from experimental results obtained from static flow testing of the valves and the results for pressure and flow will be updated, as will be discussed in the results and discussion section. 58

75 Figure 3.13 Micro-pump outlet flow as a function of time 3.4 Detailed Design Influential design parameters should be selected to ensure that the pump is capable of meeting the performance requirements after the basic pump configuration and external dimensions (size) are determined. As described in the micro-pump simulation section, the approach taken was to develop an analytical model at an appropriate level of detail to prototype this complex system. As shown in Figure 3.11, a modular analysis model is developed. This approach can significantly decrease the complexity of the analysis when selecting parameters, where the characteristics of different parts of the micro-pump are inputs to this model. To develop these inputs, a number of physical and analytical experiments were performed. For example, experiments were made on commercially available piezoelectric discs for selecting the proper piezo-driver, and multi-physics analysis using CFD software and FEM structural analysis were performed to predict the behaviour of the micro-valve. The final values for the parameters are selected using design of experiment methods. By changing the influential factors in each modular analysis, the performance characteristics of each part can be investigated. While parameter selection can provide a good starting point for a prototype design, due to the errors in modeling and analysis, it does not in practice provide the best design point. Therefore after a physical micro-pump prototype is fabricated and experimental 59

76 results are obtained, the micro-pump model will be improved and parameter optimization for the valve will be performed Piezoelectric Disc Selection A number of numerical and experimental tests were performed on the piezoelectric disc to find its behaviour at different conditions and to select a proper driver for the micro-pump. In the first test, the stiffness of the different PZT discs was evaluated using the finite element method (FEM). Table 3.2 Membrane volume displacement for differential pressure of various PZT discs Type of PZT disc Material Diameter (mm) Resonant Frequency Displaced Volume for applied pressure (dv/dp) 1 BUZZER Piezo AB2040B 2 BUZZER Piezo CEB-20D64 3 BUZZER Piezo AB2072S 4 BUZZER Piezo AB1541 Brass 20mm 4.0kHZ 1e-13 Brass 20mm 6.5kHZ 4e-14 Steel 20mm 7.2 khz 3e-14 Steel 15mm 4.1kHz 8e-14 The change in micro-pump chamber volume due to the driver displacement with respect to the pressure applied, dvm/dp, is an inverse measure of the PZT disc stiffness. When it is higher, the PZT disc displacement is more sensitive to internal pressure in the pumping chamber. Having a minimum dvm/dp is desired, as it is the least sensitive to driver motion as this pressure is caused by a force on the PZT in a direction opposite of the driver motion. Therefore, of the discs measured, the second and third piezoelectric discs in Table 3.2 are selected. By considering other criteria such as the maximum displacement of the Pizeo, the third PZT disc was selected. However for special applications like pumping corrosive or biological liquids, the second stainless steel PZT disc can also be used. 60

77 The unit price for the selected pizeo with wires attached to it as shown in Figure 3.14 (April 2012), if bought in lots of 100 is US$, and if bought lots of 1000, is US$. Further information about the selected piezoelectric disc can be found in Appendix A, in its datasheet. A test was performed to measure the maximum displacement of the PZT disc without applying any differential pressure using a laser vibrometer. The deformation at the center point of the disc was found as a function of time for different frequencies of oscillation. A control unit driving the PZT disc provided a relatively fixed voltage that did not change during the test. The deformation measured was then used to calculate the displaced volume for the disc assuming a sinusoidal shape of deformation for the disc. Figure 3.14 BUZZER Piezo, CEB-20D64 Figure 3.15 Test setup for the PZT disc deformation measurement 61

78 Table 3.3 Maximum displacement vs. frequency for the third PZT disc Frequency (Hz) Maximum Displacement (W max ) (µm) Swept Volume (V m ) (m 3 ) E E E E-09 Note that the swept volume of the disc is defined as the volume between the static position of the piezoelectric and the position with maximum displacement Valve Design Ideally valves should be designed and analyzed as a component interacting with the rest of the micro-pump. However, at the early stages of design, it is not possible to design a valve considering all structural-fluid interactions. Therefore valve design was initiated by conducting a competitive benchmarking of the ThinXXS TM micro-pump valve. Using FEM analysis, properties of this valve such as stiffness and spring ratio, minimum factor of safety, opened area at different pressures, displaced volume at different pressures and natural frequencies were estimated. For the valve designs shown in Figure 3.8, similar analyses were iteratively performed to adjust their shape and dimensions to values that give similar or better performance to the ThinXXS TM valve. The material selected for the valve was thin steel shim stock instead of 100 micron COC polymer used in the ThinXXS TM valve. Further, a design of experiment was performed on the valve to investigate the effect of different parameters and for optimization. In this design of experiment, 0.001, and thicknesses of A303 stainless steel shim stock were investigated FEM Analysis In the valve design, FEM tools were used to estimate valves stiffness, fatigue analysis, and its natural frequencies. Despite that the FEM analysis are performed for static load, the results are useful for comparison of different valves; even though a static analysis is not a 62

79 good assumption for a dynamic system. Using this analysis, the valve shapes were adjusted to have a uniform stress distribution and maximum fatigue life. The uniform load representing the pressure acting on the valve was applied to the area of the valve hole. As shown in Figure 3.16, initially the FEM analysis was performed on the model of an existing micropump valve (ThinXXS). The linear spring ratio (stiffness) of the valve was found. This valve was made of 100µm thickness, TOPAS COC polymers. While using a polymer can make the manufacturing process much easier, it can reduce the fatigue life of the valve significantly. Figure 3.16 Deformed ThinXXS TM valve; material: TOPAS COC As shown in Figure 3.17, similar analyses were performed on the other valve concepts shown in Figure 3.8. The geometrical dimensions of these valves were modified to provide stiffness close to the ThinXXS valve stiffness. For structural analysis of the valves, the maximum differential pressure across the outlet valve (P chamber - P outlet ), which as shown in Figure 3.12 is normally less than 20kPa was applied to one side of the valve. Also a factor of safety of 1.5 has been applied to this pressure load. By performing iterative finite element analysis, the geometries of the micro-valves were modified to reach the desired stiffness. To ensure durability of the designed valves, maximum stress occurring in the valve was tried to be kept below the endurance limit of the A303 stainless steel shim stock,i.e. 240MPa. The resulted maximum stress values, for different valve designs, on the 0.001" thickness stainless steel shim, were between MPa. Therefore, with a total factor of safety above (, it is predicted that the designed valves have a life of >10 7 cycles. 63

80 a) b) c) d) Figure 3.17 FEM analysis of various valve designs The final valve configuration was decided to facilitate an experimental analysis on the effect of valve the parameters to reduce the number of experimental tests required, which determined that a valve with only one circular hole and one stiffness value would be pursed. Amongst the valves shown in Figure 3.17, valves a) and b) requires more than one valve hole and similarly have more than one stiffness value. By comparing the valves shown in Figure 3.17c) and d) to the ThinXXS valve and other valves, it was found that the valve shown Figure 3.17c) shows better fatigue and stress distribution and torsional stiffness behaviour. Further, this valve can also provide a wider range of stiffnesses which is required for the design of experiment Multi-Physics Analysis A fluid-structure interaction (FSI) analysis of the valve was conducted using the MEMS module of the COMSOL TM multi-physics finite element analysis software. The best approach for finding valve performance is to conduct a FSI analysis with a 3-dimensional (3D) model. However, convergence problems with the COMSOL solvers were found while making a 3D model that limited its ability to solve large, complex 3D FSI models. Therefore, it was decided 64

81 to make a 2D FSI model of the valve. Figure 3.18b shows a simplified model of a two cantilever valve in series. However, this valve is difficult to model as a 2D valve. Thus a 2D simple cantilever valve as shown in Figure 3.18a is considered and a transient movement is applied to its base. This movement is the displacement of the cantiliver II beam shown in Figure 3.18b under transient loading. By giving this movement to the base of the 2D model of a single beam valve in Figure 3.19, the opened area of the valve and the volume displaced by the valve become similar to the 3D model in Figure 3.18b, and therefore the results of this analysis are more consistent with the design intent. a) b) Figure 3.18 a) a simple cantilever valve b) two cantilevers in series 65

82 Figure D fluid-structure interaction analysis of the micro-valve Note 1: Contours are showing constant pressure lines Note 2: Effect of the other parts of the valve not shown in 2D model are assumed to be as a movement to the base of this simple cantilever valve. For the analysis, the cross-sectional profile of the valve hole was scaled so that it would represent the same area as a circular hole. The boundary conditions applied to the fluid above and below the valve are the inlet and outlet pressures of the valve. The outlet back-pressure is assumed to be zero and the inlet pressure applied is obtained from an analytical model of the micro-pump, which is quadratic function of time. To find valve dimensions that provide sufficient passed flow, the diameter of the hole and valve length are changed and the inlet and outlet flow for the equivalent of one piezo-driver stroke are measured. The results of this analysis are shown for a driver-cycle in Figure The oscillation of the valve close to the valve seat could not be simulated in this simulation due to convergence problems in the computational fluid dynamics analysis. As a result, the backflow caused by this instability and the effect of squeezed film damping are not included. However, as listed in Table 3.4, the backflow can be estimated from this analysis without thin film damping and valve inertia dynamics. A vibrational model of the valve was made, however, to investigate the dynamic behaviour and backflow of the valve, as described in the next section. 66

83 Flow (ml/s) 5.0E E E E E E E E E-04 time (s) Valve Inlet Flow Valve Outlet flow Figure 3.20 Velocity integral at inlet and outlet vs. time Note: the minus sign indicates the direction of the flow, which is the same as the valve direction Table 3.4 Multi-physics analysis results of valve flow Hole Diameter Valve Length Passed Flow (ml/min) Back Flow (ml/min) * * Note: positive values of the back flow are in a direction opposite to the direction of the valve motion. * Analysis did not converge for a complete cycle, backflow not determined. Table 3.4 provides the results obtained from multi-physics analysis. The results were then analysed using DOE methods in MINITAB TM software for the effect of hole diameter 67

84 and valve length on flow variance, without considering interaction effects between these two parameters. As shown in Figure 3.21 and Figure 3.22, the analysis results predicts that the hole diameter is very important to flow, but not very influential to backflow, and backflow is mainly affected by the valve length and stiffness. While valve length is much less important to the mean flow rate than hole diameter, as seen in Figure 3.21,, increasing valve length slightly increases the flow rate, which could be a result of an improved flow pattern. Figure 3.21 Main effect plots of hole diameter and valve length on flow rate Figure 3.22 shows that hole diameter does not have a significant effect on the back flow. However, later experimental investigation indicated that hole diameter is an important parameter to backflow, as it highly affects the valve sealing, which was not modeled in this analysis. Backflow is also influenced valve length, valve stiffness, valve inertia, and the flow pattern. As can be seen in Figure 3.22, a shorter valve has less backflow due to its lower inertia and faster dynamic response. 68

85 Figure 3.22 Main effect plot of hole diameter and valve length on backflow Micro-valve vibrational modeling When a surface moves in close proximity of another solid surface, the fluid in the space between the moving surface and the solid surface can have a significant effect on the dynamics of the moving plate. This effect is called squeezed film damping. Squeezed film damping and valve inertia can affect the valve s dynamic behaviour and cause backflow for the micro-pump, which leads to a loss in system performance. A model of the dynamic response of the valve was made to explore backflow. As shown in Figure 3.23, the valve was modeled as a loaded mass, spring, and damper system, with the mass the equivalent to the effective mass of the valve, and the spring ratio the same as the valve stiffness. Two types of damping forces were considered: squeezed film and viscous damping [Huang et al. 2010]. The loading applied to the system was the same as the pressure distribution obtained from the micro-pump analytical model. A sudden change in valve stiffness was used to model contact between the flap valve and its seat, with the stiffness change becoming very large at contact. 69

86 Figure 3.23 Model for calculating dynamic response of the micro-valve The equation describing the model in Figure 3.23 is: ( 3.5) The equation for viscous and inertia forces caused by squeezed film damping for a rectangular plate moving in proximity of a fixed plate described in Huang et al. [2010] are used to estimate the fluid forces on the valve. Therefore effective damping and mass are estimated for equation ( 3.5, as notated by an index e on the mass and damping coefficients. As shown in Figure 3.24, the steady state dynamic response for displacement of the valve is found. In this figure, the valve starts oscillating at a high frequency when the pressure difference across the valve changes its direction as it comes into contact with the valve seat. This instable valve motion prevents proper sealing against its seat and can be a major cause for dynamic backflow. An attempt was made to improve the dynamic response of the valve by adjusting valve stiffness and damping by changing the valve geometry. However, the amount of change estimated for the damping to improve its dynamics was smaller than the model error. Therefore, no conclusion could be made for changing the damping without a more accurate estimate of the effect of squeezed-film damping. As can also be seen in Figure 3.24, the maximum displacement of the valve is about 350µm, which should be high enough to allow particles as big as 100 µm to easily pass through it. Note that the small variation in displacement seen between valve closures in Figure 3.24 is due to numerical errors in calculating the displacement. 70

87 Figure 3.24 Valve displacement by time obtained from the vibrational model Design of the Experiment for Valve Prototyping Determining the best value for a design parameter from an analytical model is difficult due to the large sensitivity to variation in the fabrication and assembly tolerances of very small dimensions in micro-components, as well as the effect of simulation errors and exclusion of physical effects like thin film damping forces that are difficult to estimate exactly. Therefore, a design of experiment (DOE) was performed around the best design parameter values determined from the analysis to determine the most robust values of the design parameters experimentally and further investigate the effect of different design parameter on valve performance. In this DOE, three design parameters, valve stiffness, valve seat hole diameter, and valve geometry were investigated, in a total of eight experimental tests. The number of tests was limited by the number of valve variations that could be fit in one photo-lithography mask. More valve variation could be made in more than one photo-lithography mask, however due to cost constraints only one mask could be made. For both the seat hole diameter and valve stiffness, three different levels were considered. All of the valve samples had the same geometry except one, which had a slightly different geometry combined with a mid-level hole diameter and stiffness parameters in order to observe the effect of geometry. Figure 3.25 shows the 4 x4 lithography mask, designed for etching the prototype valves. 71

88 The design parameter levels, shown in Table 3.5, are based on using 0.001" thickness stainless steel shim stock. However, the same geometry valves were also etched on a and 0.002" thickness shim stocks, providing significantly increased valve stiffness. Changing the plate thickness gives an opportunity for having a wider range of valve stiffness and to investigate the effect of shim thickness on the valve sealing. The main goal of this design of experiment is to investigate the flow, backflow, sealing, and the dynamic response of the valves as the parameter considered are varied. As a result, more optimized values for the design parameters can be determined. Table 3.5 Designed Experiment Parameter Levels Test # Valve Stiffness (N/m) Hole Diameter (mm) Geometry I I I I I I I II Figure 3.25 Lithography mask designed for making the prototyping samples Micro-pump Casing The micro-pump casings design must perform the following functions: 72

89 Form the pumping chamber and inlet and outlet passages of the pump Locate the micro-valve and piezoelectric disk Seal the micro-pump and fully seclude the inlet (low pressure) and outlet (high pressure) flows Compress the micro-valve plates As mentioned earlier, further to these requirements, the micro-pump casing should be designed for injection moulding. Many designs were considered for the casing to fully perform these tasks. Figure 3.26 illustrates the design for the micro-pump casing, with blue and red colours indicating the outlet and inlet, respectively. Alignment features inside the casing allow valve assembly only in its correct orientation. This casing is also designed to provide the required compression force for sealing the valve. As can be seen in Figure 3.26, for a planar inlet and outlet to the micro-pump, the flow must turn 90 degree in order to enter the pumping chamber. a) b) c) Figure 3.26 Final design of the micro-pump Casings a) upper casing upper side, b) upper casing lower side (pumping chamber), c) lower casing 73

90 The casing halves were designed for injection moulding, but a machined casing was desired for the first prototype and a simplified design based on an earlier design version shown in Figure 3.27 was fabricated. Table 3.6 presents a brief description of the casings features. Detailed drawings of the casing halves in Figure 3.26 with major dimensions can be found in Figure 4.25 and Figure a) b) c) Figure 3.27 Machined prototype casing design a,b) upper casing, c) lower casing 74

91 Table 3.6 Casings features description Casing feature description 1 Inlet channel 2 Outlet channel Casings datum surfaces, the valve is 3 placed between these two surfaces Alignment pin, for mistake proofing 4 the assembly 5 Sealing features (labyrinth seals) Features for the heat sealing 6 assembly of the micro-pump 50µm thickness feature for applying 7 a local pressure to the valve Outlet from the micro-pump 8 chamber 9 Inlet to the micro-pump chamber Pumping Chamber Volume From the micro-pump simulation it was concluded that the chamber volume is not influential to the flow rate for an incompressible flow. However, if the compressibility of the fluid is decreased significantly (e.g. for gasses or two phase flows) or for when the micropump the primed, its volume becomes important and it is desirable that it be minimum to provide a better compression and priming for the micro-pump. On the other hand, decreasing the volume reduces the clearance between the inlet valve and PZT disc and the maximum deflection of the inlet valve and PZT should checked to estimate the minimum required volume of the chamber. Further, the minimum volume of the chamber should be higher than the maximum flow in each cycle (maximum flow rate/(frequency 60)). These considerations are summarized in Table 3.7 and a V c =2e-7 m 3 was selected that satisfies both constraints and is also minimal. 75

92 Table 3.7 Chamber volume selection constraints Constraints Result Maximum flow rate V c > 2 max[(flow rate)/(f 60)] V c >2*(1e-8) Piezo and valve V c > 3 (max. inlet valve deflection)*(pzt disc clearance area) V c >3*(4e-8) Compressible flow Minimize V c V c be minimal 3.5 Design for Manufacturing and Assembly When designing a micro-pump for ease of manufacture and low cost, the parts should be designed based on a proper selection of mass fabrication techniques and assembly methods. Figure 3.28 summarizes the key features of design for manufacturing and assembly considered in the micro-pump design. In this design, plastic injection moulding and chemical etching are selected as the fabrication methods for the casings and valves, respectively. Both of these methods are having advantage of fabricating large number of parts at the same time. Also as mentioned before, the piezoelectric disc with wires connected to it, can be purchased as a standard part for less than $ 0.68 for higher than 1000 units. Figure 3.28 Key features of design for manufacturing and assembly considered in the design 76

93 By integrating and modularizing components together the total number of parts are minimized. In this design inlet/ outlet cantilever valves and their valve holes are integrated together, thus instead of 4 pieces, they are made in one piece. Having minimum number of parts is an important design for manufacturing criterion for reducing fabrication and assembly costs. Considering a one directional assembly process together with having alignment features, facilitates accurate alignment of the components and mistake proofs the assembly process, i.e. micro-pump could be assembled only with right positioning of the casings and micro-valve. Also the overlapping features around the micro-valve and inlet/outlet channels of the pump can prevent the internal leakage. Internal leakage inside the pump, i.e. leakage from high pressure points to low pressure points, can significantly reduce the performance of a micropump. A heat press sealing process was developed for assembling the two halves of the micropump casing. This method can be fully automated and eliminates the addition and expense of gluing the casing halves together Injection Moulding In order to injection mould the pump casings, the micro-pump design must have a 2 degree draft angle on all surfaces normal to the mould parting plane, and the dimensions must be scaled by a factor of 1.02 to compensate for shrinkage when the plastic cools Plastic Material Properties The main material selected for the injection moulding of the casings is TOPAS COC 8007 S-04. According to the TOPAS materials brochure [2006], The TOPAS COC family, in contrast to the partially crystalline polyolefins PE and PP, consists of amorphous, transparent copolymers based on cyclic olefins and linear olefins. COC 8007 is a suitable choice for micro-injection moulding as this material has high transparency, very low water absorption, high rigidity, and good strength. As an amorphous material, it also has lower shrinkage than crystalline materials and good flowability characteristics. Figure 3.29 and Figure 3.30, obtained from the Moldflow TM software material library, show characteristics and suggested process condition for COC

94 Recommended Processing Condition Figure 3.29 TOPAS COC-8007 mechanical properties and recommended processing conditions a) b) Figure 3.30 TOPAS COC-8007 a) effect of mould temperature on viscosity b) effect of injection or holding pressure on the specific volume MoldFlow TM Analysis The effect of runner and gate dimensions and injection moulding parameters were analysed and the best values are selected to ensure the complete filling of the mould cavities using Autodesk MoldFlow TM software. Figure 3.31 illustrates the dimensions selected for the runners and gates as a result of this analysis. 78

95 Figure 3.31 Selected dimensions for the runners and gates After determining the runner and gate dimensions, it was observed that injection time is the most important parameter, while the other injection parameters are kept within the recommended range. Therefore a minimum 0.2 second injection time for the injection moulding machine (BOY12A) was estimated. The meshed model of the parts, runner and gates used for this analysis is illustrated in Figure Table 3.8 and Table 3.9 show the processing conditions selected and injection moulding analysis results for COC Figure 3.32 meshed model for injection moulding analysis Table 3.8 Selected injection moulding parameters for the TOPAS COC-8007 Mould Surface Temp. Melt Temp. Injection Pressure Cooling Time Injection Time 70 0 C C 16 MPa 20 s 0.6 s 79

96 Table 3.9 Injection moulding analysis results for the TOPAS COC-8007 Total weight (part + runners) g Maximum Volumetric Shrinkage 7.83 % Maximum Line Shrinkage % Max. Clamping Force Max. Pressure at injection location Filling Time Maximum Residual Stress 0.86 tonne (8.44KN) 8.9 MPa s 45 MPa The MoldFlow TM analysis results are shown in Figure 3.33 to Figure Figure 3.33 indicates the cavities completely fill in 1.45 seconds with the set of injection moulding parameters of Table 3.8. Shrinkage is caused by the phase change of the material from liquid to solid. As can be seen in Figure 3.30b, the amount of shrinkage is reduced by increasing injection and holding pressures. The volumetric shrinkage distribution is illustrated in Figure Note that the design scale factor applied to the model for shrinkage compensation was a linear shrinkage factor and can be obtained by applying the following relationship: Shrinkage Scale factor= (1+Average Volumetric Shrinkage) (1/3). Non uniform shrinkage will cause residual stresses in the part, which can lead to part deformation. The residual stress distribution is illustrated in Figure

97 Figure 3.33 Time required for filling stage of the injection moulding Figure 3.34 Volumetric shrinkage distribution at the end of the injection moulding Figure 3.35 Residual stress distribution of injection moulded parts 81

98 Figure 3.36 shows the completion time of the cooling stage, which is the time to reach ejection temperature measured from the beginning of the process. Despite a maximum time indicated in this figure of 149 seconds, both casings are sufficiently cooled by 40 second, and thus the total process time is about 40 seconds. Figure 3.36 Time to reach ejection temperature (cooling time) distribution Figure 3.37 Weld line formation locations during injection moulding Figure 3.37 shows possible locations of weld lines during injection moulding. Weld lines are normally the weakest locations in moulded parts. However, due to the short filling time (1.4 s) and that the flow temperature is sufficiently high during filling, it is predicted that the flow fronts will be sufficiently mixed and no visible weld line will occur in these locations. Figure 3.38 also shows the particle orientation on the surface of the part. In the case 82

99 that a non-isotropic material is used or the moulding temperature is sufficiently high, particle orientation can be an influential factor in the strength of the casings. The isentropic material selected for the pump makes particle orientation less important, but may be important later if composite materials are used, and then particle orientation should be considered. Figure 3.38 Particle orientation at skin of the casings Other materials During the injection moulding analysis, a number of other materials were also considered. Table 3.10 and Table 3.15 summarize the results and process conditions for injection moulding of three other materials. Amongst these materials, acrylic (PMMA) and polystyrene (PS) are amorphous polymers like COC. However, low density polyethylene (LDPE) is a semi-crystalline polymer. 1. PMMA (Acrylic) Table 3.10 Selected injection moulding parameters for PMMA (Acrylic) Mould Surface Temp. Melt Temp. Injection Pressure Cooling Time Injection Time 80 0 C C 16 MPa 20 s 0.6 s 83

100 Table 3.11 Injection moulding analysis results for PMMA (Acrylic) Total weight (part + runners) g Maximum Volumetric Shrinkage 7.48 % Maximum Line Shrinkage Max. Clamping Force Max. Pressure at injection location Filling Time Maximum Residual Stress tonne 7.5 MPa 0.84 s 40 MPa 2. Polystyrene (PS) Table 3.12 Selected injection moulding parameters for Polystyrene Mould Surface Temp. Melt Temp. Injection Pressure Cooling Time Injection Time 95 0 C C 16 MPa 20 s 0.6 s Table 3.13 Injection moulding analysis results for Polystyrene Total weight (part + runners) 6.06 g Maximum Volumetric Shrinkage 9.5 % Maximum Line Shrinkage 1.03 Max. Clamping Force Max. Pressure at injection location Filling Time Maximum Residual Stress 1.5 tonne 15 MPa 0.86 s 60 MPa 3. Polyethylene (LDPE) 84

101 Table 3.14 Selected injection moulding parameters for Polyethylene (LDPE) Mould Surface Temp. Melt Temp. Injection Pressure Cooling Time Injection Time 70 0 C C 16 MPa 20 s 0.6 s Table 3.15 Injection moulding analysis results for Polyethylene (LDPE) Total weight (part + runners) 4.63 g Maximum Volumetric Shrinkage 15 % Maximum Line Shrinkage Max. Clamping Force Max. Pressure at injection location Filling Time Maximum Residual Stress tonne 3 MPa 0.86 s 3 MPa A larger volumetric shrinkage occurs in LDPE due to its semi-crystalline structure. However, as the shrinkage is more uniform and the material is very soft, the maximum residual stress is very small. From these results, it can be seen that the predicted shrinkage and flowability of PMMA is comparable to the COC-8007, therefore PMMA can be considered as an alternative material for the casing. However, as described later, it was observed that PMMA and PS samples are more brittle than the COC-8007 samples after the injection moulding tests were performed. Therefore, at this stage of the design of the micro-pump, only COC-8007 should be considered for the casing material Heat press sealing, assembly process A micro-pump designed for ease of manufacturing and low cost requires a proper method for assembling and sealing the casings. A heat press method for assembling and sealing the casings together will provide a good quality seal and can reduce the cost of the assembly process by simplifying the production equipment and using a process suitable for automatic assembly. Further, having a uniform material for the whole casing can be an 85

102 advantage in many applications. Figure 3.39 shows a schematic of the heat press assembly process design. As shown in the figure, a thin tapered feature is included in the lower casing. Using a heat press, this small feature will be deformed and fused to the upper casing, sealing the micro-pump casing together. Note that the arrows shown in the figure are the contact points where heat and pressure are applied to the casing. Figure 3.39 Schematic of the heat press assembly process of the micro-pump To prevent the upper casing from being pushed upward when applying the clamping pressure, a fixture is required to keep the casings together. A fixture composed of three parts was designed. Two of these parts are for positioning and applying a vertical load to ensure the part is in the correct vertical position before bonding when the two casing halves are at room temperature. The circular part, illustrated in Figure 3.40, transfers the pressure and heat to the casing joint. 86

103 Figure 3.40 Fixture setup for the heat press assembly process A CARVER TM heat press is used to apply the pressure and heat to the circular fixture for the prototypes. The heat press fixture is attached to the CARVER s upper platen using strong magnets. As shown in Figure 3.39, the geometry and dimensions of the fixture are designed such that the heat and pressure is only transferred to the joint lip area. When the lower edge of the part touches the lower platen of the CARVER, the heat sealing process is completed and a further increase in the hydraulic pressure of CARVER won't be transferred to the micro-pump casings. 87

104 4 Chapter 4: Manufacturing and assembly process In this work, amongst different manufacturing methods, injection moulding and wetetching are selected for the manufacturing processes of the casings and the valve. An existing commercially available piezoelectric motivator disk was also selected. In order to reach a final low cost and reliable micro-pump, these methods are adapted and optimized for manufacturing of the micro-pump components after analysing and testing the compatibility of these methods. Further, fixtures for the assembly method were made and the process was validated by testing a number of the injected micro-pump casings. 4.1 Valves Several methods are being used for the manufacturing micro-valves, including micromachining, dry etching, chemical wet etching, micro-moulding, micro-electro-discharge machining, and laser cutting. The method selected should be compatible with the design, material and fabrication volume. The material selected for the micro-valve is stainless steel shim stock, which has a very good resistivity to most corrosive or alkaline liquids, and further, it can increase the fatigue life of the valves. When considering a 2-dimensional valve design on a thin plate, a chemical etching process can be the best choice for manufacturing stainless steel micro-valves. In comparison to other methods, this chemical (wet) etching has advantages in its ability to process very thin materials and precise control of dimensions. Economies of scale can also be obtained in wet etching, as a large number of valves can be made from one lithography mask. In this work, a number of valves were etched using chemical etching, but for the procurement of prototypes, the non-ideal yield and a time consuming optimization of the wet etching process indicated that a dry etch process should be adopted instead for prototyping the devices, a process that could take advantage of previously published work that used different materials [6]. The advantage of laser etching is that it can provide a perfectly anisotropic etch profile (similar to Figure 2.22b) and thus does not require optimization for achieving good quality. This process provided fast and accurate patterning of the different valve designs, as 88

105 well as accurately cutting the relevant circular sections and the bend line region from the bulk material. However dry laser etching, due to serial processing of the features, is not suited for large scale fabrication. Both of the chemical etching and dry laser etching processes were performed in the Advanced Micro-nanosystems Integration facility (AMIF) at The University of Calgary Wet (Chemical) etching The stainless steel shim stock was exposed to a pattern (patterning) using a modified photolithography process. Photolithography typically deals with very thin layers of a sample material that is adhered to a thicker substrate material, and thus only one side of the material is exposed to chemical etchants. For cutting the stainless steel shim stock samples, one side of the sample is coated with a protective layer and then mounted onto a 4-inch glass wafer, with the unprotected side face up, before traditional photolithography processing could commence. This was to protect the backside of the stainless steel shim stock from etchant wicking between the glass wafer and the mounted stainless steel shim stock during the immersive etching process. The glass wafer and sample were then spin coated with a photoresist prior to patterning in a mask aligner (Karl Suss MA/BA6, Karl Suss, Germany). The photomask of the valve design was manufactured by Photosciences Inc., USA. After developing the photoresist to selectively expose the underlying stainless steel, ferric chloride was used to etch the required features in the stainless steel shim stock. For a more complete description, see Rao and Kunzru [2007]. Once etching is complete, the valve pieces float free in the ferric chloride solution, another reason a backside protective layer is necessary. The valve pieces were removed from the ferric chloride and cleaned with deionized water prior to removal of the remaining photoresist and protective layer. Due to the limited size of the 4-inch sample holder and the processing equipment, only 4 valve designs could be fabricated in each process run. The fabrication process of sample preparation, photolithography patterning, chemical etching and sample cleaning took approximately 6 hours. Multiple thickness of stainless steel shim stock were processed in this manner, with the thicker samples requiring longer etch times as appropriate. It was found that the thicker stainless steel shim stock was easier to process, as it had less tendency to flex during the many processing steps. 89

106 Figure 4.1 shows two valves from the first batch of the etched valves (each batch is composed of 4 valve designs). As can be seen in the pictures on right of Figure 4.1, the back side of both valves are affected by ferric chloride. This likely caused by defective protective coating of the backside of these samples. The small thickness (0.001 ) with its tendency to flex caused problems during the coating steps of the etching process. However, process improvement for future batches resolved this problem. Furthermore, in Figure 4.1, despite having two samples etched simultaneously in the same batch, one was thoroughly etched before the other had completed its etch. This non-uniformity in the etch time was probably caused by the substrate not being sufficiently flat. Also, the non-isotropic variation in material grain alignment of the stainless steel shim stock can cause non-uniformity in the etch time. In later batches, process changes provided better flatness and coating, which reduced the nonuniformity in the etch time at different location in the batch. a) b) Figure 4.1 two valves from the first batch of etched valves on steel shim stock, Pictures on the left are showing the top sides and on right are showing back sides. 90

107 In the second batch of valves, all four valves were etched properly and completely. However, as it can be seen in Figure 4.2, the etched line thickness is not uniform and can be improved further. For this batch, the average etched line thickness was 150 µm for the front side (top) and 70 µm for the back side; while the desired etch line thickness was 100 µm. An optimized etching process will provide a line thickness closer to the desired value. In order to obtain this lower line thickness value, the mask line thickness should be reduced to 50 µm. As this modification requires the manufacture of a new mask, it is suggested for future work. Note that in Figure 4.2 the different signs etched on the valves, such as +, are used for identification different valve designs. a) b) Figure 4.2 a valve from the second batch of the etched valves on steel shim stock a) front side, 0.17mm etch line thickness b) backside, 0.07mm etch line thickness The thickness shim stock was not commercially available as flat plates, so a roll of this material was ordered. This rolled material had a high tendency to flex in the fixture which resulted in an improper coating of the photoresist on the material. As shown in Figure 4.3, the etched line thickness was very non-uniform, and due to very large thickness of the etch lines at the base of the valves, four of the valves in batch 3 and 4 were detached from their base and thus not viable for testing. Flexing of the thinner stainless steel shim stocks (0.001, ) also resulted in pinholes being formed in the photoresist, causing unwanted areas of the stainless steel to be etched. These pinholes can be seen in Figure

108 a) b) Figure 4.3 Chemical etching on shim stocks roll, a) front side of the etched valve b) back side of the etched valve The thickest material (0.002") had no flexing problem due to its higher stiffness, and the chemical etching steps were performed much easier. However, due to the isotropic nature of the wet etching, the average thicknesses were 230 µm and 80 µm thicknesses for top and bottom side of the plate respectively. Due to the longer time required for etching this thickness, surface pinhole pits were observed in some places. Figure 4.4 shows an etched valve on 0.002" material. Figure 4.4 Chemical etching on shim stocks roll, a) front side of the etched valve b) backside of the etched valve It should also be mentioned that, the small design bending holes could not be etched, as shown in Figure 4.5, as they would cause the two valve plate to separate before bending. In the mask design should be modified to make those holes smaller to include these features. The pinholes noted earlier can also be seen in Figure

109 Figure 4.5 A thickness wet-etched valve In conclusion, implementing photolithography to fabricate these thin-film stainless steel prototypes was a challenging process, e.g. difficulties in fixing very thin and flexible substrate and also achieving good quality etch for stainless steel. Still, substantial optimization will be required to take advantage of bulk processing using wet etching which was out of the scope of this project. Etch tests showed the profiles and required time for etching the different stainless steel configurations. This can be fed back into designing a new mask with optimized geometry. In the next development phase, where numerous samples must be procured, it is recommended that the wet etching process be optimized Dry Laser Etching Dry laser etching permitted each prototype design to be micro-fabricated in a reasonable time (~35 minutes per valve design) in a single-step process, and thereby reduced processing overhead cost. The desired line thickness for the laser etching was set to 50μm and, due to the nature of the laser etching, a fully non-isotropic etch line was obtained. The equipment used to implement the selective dry-etching was an in-house designed, custom-built, and assembled laser micro-fabrication system located within the Advanced Micro-nanosystems Integration facility (AMIF) at the University of Calgary. The main advantages of this selective dryetching process are its minimal distortion from a single-step dry process, and that parts can easily be replicated in large numbers with minimal variation (microns or less). This is in contrast to conventional photolithography-based processes, which typically require selective masking, chemical wet-etching, and numerous photolithographic masks, and extensive process optimization for each design. 93

110 Figure 4.6 and Figure 4.7 show valves fabricated with laser etching on and thicknesses, respectively. As can be seen these figures, the valves are fully etched and the etched line is very uniform. The bending holes in the middle of the valves are also present. a) b) c) Figure 4.6 a) a thickness valve fabricated with laser etching method b) front side of the valve c) back side of the valve The thickness of the etched line on top and bottom of the valve shown in Figure 4.6 is measured at 60μm and 35μm, respectively. The line thickness in Figure 4.7 is 80μm and 45μm respectively for top and bottom sides. Despite the non-isotropic nature of the dry etching, the difference between top and bottom etch line thicknesses is caused by an etching rate higher than what would produce a uniform line thickness etching. This difference does not have a significant effect on the quality of the etched valve, but does reduce the time required for etching the valve. 94

111 a) b) c) Figure 4.7 a) a thickness valve fabricated with laser etching method b)topside of the valve c) Backside of the valve Figure 4.8 illustrates a thickness laser etched bent to its final configuration. The bend holes on the dry etched samples provided the design intent of very easy bending exactly on the bend line. Figure 4.8 a thickness bent valve, fabricated by dry laser etching As it will be described in the testing section, after performing an initial micro-pump testing, outer geometry of the micro-valves was modified. Figure 4.9 shows the new micro- 95

112 valve, manufactured using, dry laser etching. In addition to the performance advantages of this design, it has a much smaller size and therefore for mass fabrication, a larger number of these valves can be etched at the same time. Figure 4.9 a 0.001" thickness valve with modified geometry, fabricated with laser etching 4.2 Machined Prototype As described in the design section, the prototype was adapted for assembly using bolts and the machining process was simplified by simplifying the initial design illustrated in Figure 3.9 to the design shown in Figure Three sets of casings were machined using three different materials: aluminum, brass and acrylic. Figure 4.10 shows an aluminum upper casing, a), and a brass lower casing, b). These casings were machined in the Engineering Machine Shop at the University of Calgary. 96

113 a) b) Figure 4.10 Machined casings a) upside of the aluminum upper casing b) Brass lower casing The assembly process for this prototype was composed of the following steps: 1) Glue the piezoelectric into the upper casing 2) Placing a bent valve and fluid connection tubes into the upper casing 3) Place a gasket or use silicon glue around the bolt holes to seal the casing halves 4) Add the lower casing and close the micro-pump using screws Figure 4.11 illustrates the machined prototype assembly with the piezoelectric disc attached to the casing top, a), and the valve and tube assembly inside the casing, b). a) b) Figure 4.11 Brass upper casing a) piezoelectric disc glued to the casing b) valve and tubes placed in the upper casing 97

114 After testing the machined prototype, for the injection moulded casings, a number of modifications were applied to design of the machined prototype, such as changing the upper casing opening from single hole to double smaller holes, changing the assembly method and removing the sealing feature considered around the inlet and outlet channels, and modifying dimensions to eliminate the gasket between the casings. 4.3 Injection Moulding the Casing Injection moulding is the process of melting and forcing the molten material into a mould cavity where it cools and takes the shape of the cavity. Because of the high production rate of this process, injection moulding is a popular mass production method. When this process is applied to micro-components such as the micro-pump casings, where features can have dimensions of about 50 micron and much smaller tolerances, the process is called microinjection moulding. Three main phases are performed when implementing the micro-injection moulding processes: 1) part design for micro-injection moulding, 2) design and machining of the mould cavities, and 3) injection moulding process optimization. The first phase is described in detail in the design section under design for manufacturing section. After the mould cavities are machined into the mould inserts, the inserts are placed in the moulding clamp in the injection moulding machine Moulds Machining The KERN TM Micro high precision CNC machining station was used to machine the mould inserts, shown in Figure An accurate machining program (CNC G-Code), with minimal deviation from the designed parts, was developed using Esprit TM CAM software. For the machining process, two different tools were used: a 0.7mm 2 0 tapered end mill and a 0.5 mm ball end mill. The machining process developed was tested initially on machining wax as illustrated in Figure It should be noted that unlike shown in the wax, the actual inserts are machined into two different inserts. 98

115 Figure 4.12 KERN Micro, a high precision CNC machining device Figure 4.13 Injection moulding Inserts, machined on machining wax 99

116 Inserts alignment pins and datum surfaces for machining One major issue in achieving a good quality moulded parts is the alignment of the two halves of the mould inserts, illustrated in Figure Cavities for the upper casing need to be machined on both inserts. Thus, the two inserts should be aligned with each other with alignment pins and holes of very tight tolerances to ensure accurate alignment of the inserts. Figure 4.14 Two injection moulding While the alignment pins can align the two inserts, common datum surfaces inserts accurately, they cannot be used as a datum for machining the cavities. Therefore, in order to provide a common datum surface, the two raw inserts were assembled with their alignment pins and side surfaces were milled flat. As a result, the x and y surfaces shown in Figure 4.14 provide two datum surfaces for cavity machining. Figure 4.15 shows the machining of one of the inserts. The nozzles seen in this figure constantly sprayed liquid coolant reducing friction and wear on the very small size milling tools. For accurate positioning of the tool relative to the datum surfaces, an acoustic sensor was used. 100

117 Figure 4.15 Machining of the injection moulding inserts Figure 4.16 shows the two machined aluminum mould cavity inserts. The numbered pins and holes in this figure indicate the right relative orientation of the inserts. The two other holes in each insert are for attaching the inserts to the clamping unit of the injection moulding machine. Figure 4.16 Two machined injection moulding inserts Injection moulding The injection moulding process was performed using a BOY12A micro-injection moulding machine, shown in Figure This machine is mainly designed for very small samples and is capable of injecting small shot volumes. 101

118 Figure 4.17 BOY 12A injection moulding machine The maximum clamping force for the BOY12A is 120kN, which is a relatively small for injection moulding machines. If a clamping is not sufficiently high, complete internal sealing of the mould inserts for parts with large surface area at high injection and dwell pressures cannot be ensured. Ideally the center of pressure of the moulding cavities should be placed at the middle of the insert. Having the center of pressure slightly off center is not unusual. However, it can aggravate the insert sealing issues. In this mould, due to the different length of the runners which are obtained from injection moulding optimization analysis and other geometrical constraints, the center of pressure is slightly off center. The first injection moulding was performed using COC 8007and the injection conditions shown in Table 4.1. Due to problems with the mould heater elements, injection moulding was performed without heating the insets. However, as one of the inserts is attached to the injection nozzle whose temperature is normally above 50 0 C, which is only 20 0 C lower than the design value in Table 3.8. To compensate, the injection time was reduced to 0.4s. The injection pressure used in this test was also only 8 MPa, slightly lower than the maximum required pressure in Table 3.9 (8.9 MPa). 102

119 Table 4.1 First injection moulding sample process conditions COC 8007 Injection Pressure Nozzle Temperature (Melt Temperature) Injection Time Cooling Time Mold Temperature 38MPa C 0.4 s 20 s 50 0 C Figure 4.18 The first sample injected of COC 8007, using conditions in Table 4.1 The first injection moulding sample is shown in Figure In this sample, all cavities were completely filled. However, due to insufficient clamping force, molten material flashed out of the cavities. Because the cavity for the lower casing is in only one insert, the flash of material between the two inserts does not affect any critical dimension on this casing. The material flash on the upper casing, however, is in the middle of the part and influences its critical dimensions. Therefore, the injection pressure was reduced for future injections. The injection pressure could be reduced to 4 to 4.5 MPa, where it was observed that the material flashing around the upper casing was completely eliminated and significantly reduced for the lower casing. It was observed that further reduction beyond this injection pressure can led to incomplete cavities fill. To summarize, Table 4.2 shows the final setting yielding a relatively good injection moulding process using the BOY machine. Figure 4.20 also illustrates the micro-pump casings obtained from this injection moulding process. 103

120 Table 4.2 Injection moulding process conditions COC 8007 Injection Pressure Nozzle Temperature (Melt Temperature) Injection Time Cooling Time Mold Temperature 54MPa C 0.4 s 20 s 50 0 C Figure 4.19 COC 8007 injected sample, using Conditions in Table 4.2 Figure 4.20 COC 8007 injection moulded casings In Figure 4.21, two defective samples in the right of the pictures are compared with properly injected samples on the left of the pictures. The defect in Figure 4.21a, is caused by dirt in the insert and a low injection pressure. The defect in Figure 4.21b, is caused by low injection pressure and temperature. The weld line cause by low melt temperature can also be seen in Figure 4.21b. This weld line indicates the casing would have lower strength and may burst under pressure. 104

121 a) b) Figure 4.21 Defective COC8007 samples; a) improper filling of the thin edge around the lower casing b) Improper filling of the upper casing Three other materials beyond the COC8007 were used during the injection moulding tests. A brief description of the samples and comparison between them and COC8007 is presented here. Polystyrene (PS) Figure 4.22 Polystyrene injection moulded sample 105

122 These samples were very brittle and not very transparent. The melt and insert temperature required for PS was much higher than the COC8007. Since a heater was not used for the inserts, the physical properties of the samples might have been affected. PMMA Figure 4.23 PMMA injection moulded sample These samples are very similar to the COC8007 samples. The processing conditions only needed a higher melt temperature, but are otherwise similar to those of the COC8007 material. In separating the runners and gates from the casings, it was observed that the PMMA samples are relatively more brittle than COC8007, but they also showed similar low shrinkage and good flowability. COC5013 Figure 4.24 COC 5013 injection moulded sample TOPAS COC5013, also requires a higher melt and inserts temperature than COC8007. The sample quality was again affected by not using heater for insert. In comparison to grade COC8007 samples, they were more brittle and having less flowability. 106

123 Dimensional analysis of micro injection moulded parts The dimensional deviation of the sample casings parts from the design intent is caused by two sources: machining error and injection moulding error. Variation between the injected parts and the mould cavities can be either due to shrinkage, material flashing between two inserts, or incomplete filling of the mould cavities. Table 4.3, shows the values obtained from the measurement of the injected parts, machined cavity inserts, and design values. These measurements are performed for three different samples and the corresponding dimensions used in this table are shown in Figure 4.25 and Figure Definitions: Part Measurement: values obtained from measuring the injected sample parts Designed part: nominal dimension from design model (desired design intent values) Designed Insert: shrinkage factor (1.02) x the nominal designed part dimension Measured Insert: values obtained from measurement of machined inserts cavities U: indicates upper casing L: indicates lower casing S#: sample number 107

124 Dimension ID Horizontal measurements, COC 8007 Part Measurements S#1 S#2 S#3 Designed part Measured Insert Designed Cavity U U U U L L L L L Vertical measurements, COC 8007 U U L L Table 4.3 COC 8007 Dimensions and Measurements 108

125 Figure 4.25 Micro-pump Upper Part Casing Figure 4.26 Micro-pump Lower Part Casing 109

126 Deviation from the designed part % Using the following definitions, values for shrinkage, machining error and total deviation of the sample casings from the designed casings are obtained. Note that the values for shrinkage are also affected by overfill, flash, and incomplete fill. For the dimensions which are normal to parting line plane, the value for shrinkage is highly affected by overfill and therefore is called "shrinkage and overfill" U 27.5 U 27 U 4.6 U 1.4 L 30 L 28.4 L 0.76 L 1.6 L 20 Shrinkage Machining Error Total Deviation from the designed part Figure 4.27 Horizontal dimensional analysis of COC 8007 samples As it can be seen in Figure 4.27, the shrinkage value, which can be more clearly identified from the larger dimensions, is less than 1%. For smaller features, machining errors play a more important role. The horizontal measurements indicate all total deviations are within an acceptable range except for L1.6 and U1.4, in which their measured dimensions 110

127 Deviation from the designed part % from Table 4.3 are 1.8mm and 1.25mm respectively. These two features are actually designed to seat on each other and require good sealing and as both deviations are in a direction that increases the clearance between these features, the deviation may cause sealing problem. Therefore this machining error should be rectified when the mould cavity is machined for the next design iteration U 3.2 O.F. U 1.95 O.F. U 1.25 L 3.5 O.F. L 1.2 Shrinkage and Over filling Machining Error Total Deviation from the designed part Figure 4.28 Vertical dimensional analysis of COC 8007 samples Figure 4.28 shows the shrinkage and overfill dimensions that are affected by flashing the material between the two mould cavity inserts when they have large negative values. The negative value for shrinkage and overfill indicates that the dimension of the injected part will be larger than the insert and is caused mainly by separation of the two inserts due to high injection pressure. Because the lower casing cavity is in only one insert, critical dimensions such as L1.2 are not affected by material flash. However, this error increases the thickness of the thin membrane in the middle of the upper casing and can cause the upper casing thickness to be non-uniform. It is recommended that an injection moulding machine with higher clamping force be used to address these problems, as the120 KN maximum clamping force in the BOY is insufficient. The other materials used for the injection moulding (PMMA, PS, COC 5013) have similar results to those shown in Figure 4.27 and Figure 4.28 for COC

128 4.4 Heat Press Bonding As shown in Figure 3.39, the thin circular feature around the lower casing should melt and form a bond to the outside of the upper casing in the heat press. In order to prevent excessive thermal loading on the upper casing and to ensure a strong bond, a proper heating temperature and time should be selected and the heating fixture should be machined to high accuracy. In this work, a number of tests were performed to investigate the appropriate assembly process conditions and test the seal and strength of the final bond Feasibility Tests Feasibility tests were performed initially to check the bonding process and to investigate the assembly process condition. The goal of the tests was to find the proper temperature and contact time of the fixture with the thin plate, which can lead to a strong joint without causing any deflection on the thicker plate. A 2D heat transfer analysis using COMSOL multi-physics software found that a C and C fixture held for approximately 8 or 10 seconds, respectively, is required for the surface temperature of the thicker plate to the reach the glass transition temperature, T g (80 0 C). A number of thin plate samples of the selected material (TOPAS COC 8007) were made with various thicknesses (3 mm, 1 mm, 0.6 mm) using pressure moulding method. As an example, as shown in Figure 4.29, a 0.6 mm thick sample was place over the thickest plate (3mm) and pressed in the CARVER heat press by directly applying the smallest possible load. In these tests, while keeping the temperature of the bottom jaw of the fixture a constant room temperature, various upper jaw temperatures were tested, starting from the glass transient temperature of the material. After testing different temperatures, a temperature of 110 to C was found to be appropriate for the bonding. The 3mm plate did not deflect at the joint, but its thickness reduced about 0.1mm, which shows the surface of the thicker plate has increased to slightly above the glass transient temperature, which provides proper mixing of the plates material to form a strong joint. However, an overload of the hydraulic load applied directly on the two plates caused the thin part to flash. 112

129 Figure 4.29 a) before the bonding b) after the bonding at C Heat Press Fixture Figure 4.30 a) and b) illustrate the heat press bonding fixtures described in the "design for manufacturing and assembly" section. Initially, the fixture was designed to use magnets in the middle of the fixture to hold it against the heat press platen. However, the magnets were found not to provide enough strength for connecting the fixture to the upper jaw, and so the magnets were placed directly between the fixture and the CARVER jaw. The fixture was connected very well attached to the CARVER without the need for permanent changes to the heat press. It was also observed that the upper jaw temperature doesn't cause any problem for the magnets. 113

130 b) c) Figure 4.30 Heat press bonding a) heat press fixture b) fixture to clamp the micro-pump c) setup for heat press bonding assembly Figure 4.30 c) shows the setup for the heat press assembly process. The fixtures are accurately aligned to each other and the micro-pump is clamped between the two room temperature pieces of the fixtures. As shown in schematic view of the process in Figure 3.39, the geometry and dimensions of the heating fixture are designed such that the heat and pressure is transferred only to the joint lip area. When the lower edge of the fixture touches the lower jaw of the CARVER, the heat sealing process is complete and further increase in the hydraulic pressure does not cause an increase in the applied load to the micro-pump. Figure 4.31 shows the CARVER heat press with the fixtures assembled into it. 114

131 Figure 4.31 CARVER heat press with the fixtures assembled on it The process was performed for three, three second successive steps without causing cracking at the connection joint. The temperature of the hot jaw of the CARVER was 250 F to 285 F (121 to 140 o C) and the lower jaw temperature was at room temperature. Figure 4.32 shows a schematic of the casings, predicted bonding, and actual bonding formed between them. Figure 4.32 Schematic of the casings profile a) before the bonding b) predicted profile after bonding c) actual profile after bonding 115

132 Figure 4.33 shows the results of the heat press bonding for COC8007 casings. The same bonding tests were performed on PMMA casings and the resulting bonding is shown in Figure Because of the higher glass transient temperature of the PMMA than COC 8007, the CAVER jaw temperature used for its assembly was a slightly higher 145 o C. Figure 4.33 Heat press bonding results for COC 8007, formed at 130 o C heat press fixture temperature 116

133 Figure 4.34 PMMA casings after the bonding, formed at 145 o C heat press fixture temperature Strength and seal test A pressure test was performed the bonded casings in order to ensure that the bonding was sealing properly and was strong enough to withstand high pressures. Inlet and outlet pipes were glued to the casing using superglue, which performs well for these types of plastics, and then the bonded casings were pressurized using the test setup shown in Figure The pressure was controlled a pressure control valve shown in the figure. The test showed that no leakage occurred for both PMMA and COC casings until about gage pressure of 2.2 bar. This pressure is more than 4 times the maximum operating pressure (50kPa) inside the micropump. On both of the tests the leakage first occurred at the inlet and outlet of the casing, at the labyrinth seals, which were glued and not heat pressed, as shown in Figure It can be concluded the heat press bonding provides a strong joint and good seal, and further, this joint should be better than gluing. 117

134 Figure 4.35 micro-pump casings bonding test setup Figure 4.36 a) before gluing b) after gluing, the spot where the leakage first occurred is highlighted Summarizing from the results obtained from the heat press bonding tests and heat transfer analysis, the injection moulded micro-pump casings made out of COC 8007 should be bonded at 130 o C. The process should be performed in three successive steps of three seconds of contact to prevent cracking the casing. However if the jaw movement rate of the fixture can be controlled, the process can be performed in one step in 6 to 8 seconds. The quality and strength of the heat press assembly process for the micro-pump is validated. A fully automated assembly process is feasible and a very strong bond can be obtained without glue. Further, because of extremely low water absorption, very good biological and blood compatibility, and 118

135 very good resistance to acids and alkalis of COC 8007, the elimination of glue between the casings enhances the suitability of the micro-pump for many applications. 4.5 Cost Analysis To estimate the cost of the micro-pump, various expenses that occurred during the development, fabrication and assembly processes should be considered. These costs can be classified as variable and fixed costs. Variable costs are those costs that will vary in direct relation to production volume of the micro-pump. Material costs, labour, packaging, handling and assembly are examples of these costs. Fixed costs are those that are almost fixed, regardless of the production rate. Examples of fixed costs are plant, capital equipment, moulds, etc. Estimating these costs, the overall cost of a micro-pump can be calculated using equation 4.1. ( 4.1) In this analysis, to minimize the cost per unit at smaller production volumes (1,000 to 10,000), the micro-valves and casings should be purchased from a supplier rather than purchasing equipment to make them. Therefore while the fixed cost of the fabrication equipment is not required, the variable cost for fabricating these parts will be higher. The unit price for the selected piezoelectric with wires attached to it was shown in Figure 3.14 (April 2012), if bought in lots of 100 is US$, and if bought lots of 1000, is US$. To estimate the fabrication costs of a pair of casings, a preliminary injection moulding cost estimation tool of BASF [2012] is used. Table 4.4 shows the various costs per unit, associated with injection moulding of the casings. To estimate these costs the following consideration are made: 60 second injection moulding cycle 8 cavities per mould (4 pair of casings at each injection) A labour rate of $30/hr for injection moulding machine operators Material cost of $8.5/kg of COC Production volume of 10,000 units 119

136 At the prototyping stage, the moulds used for injection moulding were machined as part of this project. However, for machining a mould with 8 cavities (four pair of parts) for the production, an estimated cost of $1,400 is considered. This value is based on 6 hours of CNC programming, 5 hours of machining, and an hour of polishing, with raw mould material for $200. The technical labour cost of mould machining is considered 100 $/hr. Table 4.4 Costs associated with injection moulding of the casings Costs Mould Material Processing 30% Press cost Net variable design and cost per cost per pair overhead per pair of cost of one machining pair costs casings pair USD) $1400 $0.03 $0.12 $0.04 2x$0.50 $ 1.19 As described in the micro-valves manufacturing section ("4.1 Valves"), stainless steel and chemical etching were selected as the main material and the fabrication method. The laser etching process, due to its ability to follow a pattern and serial cutting, is not suitable for larger scale fabrication and therefore its fabrication cost will not be significantly reduced by increasing the fabrication volume. Table 4.5 shows the development costs of the chemical etched valves with the initial outer geometry design. These prices are obtained using invoices received from the AMIF lab in the University of Calgary where the prototype valve fabrication was performed. In this table, the initial setup, process testing, mask fabrication, material preparation, and the required technician time costs are included in the initial setup and testing costs. As a result, the cost per unit for the fabrication of these micro valves using this method was estimated at $55. Table 4.5 Chemical etching costs for the initial design of the valve Chemical Etching costs Initial setup and Quantity Cost Per R&D Total testing costs Unit Discount Cost $ $ 55 $ $ 2720 In the chemical etching process, depending on the capabilities of photolithography instruments and regardless of the pattern to be etched, a constant area of the work piece can be etched at one time. Therefore if the valve design is modified to the one shown in Figure 4.9, 120

137 which has a smaller area relative to the initial design, a larger number of valves can be processed simultaneously. Figure 4.37 shows that almost 6 units of the new valve design can be produced in the same area of a single unit of the previous design. Dr. Colin Dalton from AMIF lab estimates that in future cost per unit of valve can be reduced by approximately 25%. Therefore, the cost per unit of $6.87 (0.75*55/6) is predicted for the chemical etching of these valves. In a production environment, through process optimization, it is assumed that higher yields and larger mask area will permit this unit cost to be further reduced. Figure 4.37 Comparison between the size of the initial design of the micro valves (the larger valve) and the modified version (smaller valves in this figure) In this project, for the assembly process, a Carver Manual Press model 12-12H was purchased for $9,335 in April Fixtures mounted on the Carver machine were machined in University of Calgary engineering machine shop for $580. To attach the piezoelectric discs to the casing, gluing process is currently being used. However in future design work, a heat press bonding will be applied for assembling the piezoelectric. The estimated labour time required for this process is 30 second and based on $30/hr labour cost, it will have a labour per unit of $0.25. It is approximated that about 0.3 gram of Loctite 495 Super Glue are being used for attaching the piezoelectric and, based on the currently purchased price for glue, it will cost $ If this process is modified to be performed by heat press bonding, it is estimated that it 121

138 will only have about $ 0.05 cost for labour. The required floor space for storing received parts and performing the assembly process, including the required aisle space, is estimated to be 30m 2. For the required space, amortization cost of $30/ m 2 per month for one year is considered. At the end of fabrication, each micro-pump is inspected for self priming, any leakage, maximum flow, and pressure head. A fixed cost of $1000 for a multi-fixture functional testing setup is considered which will allow testing a number of micro-pumps at the same time before shipment. The estimated required labour time for inspecting each micro-pump is about a minute and therefore $0.5 labour cost is allocated for it. Also at this point, as a rough estimation, 3% of the micro-pump will be rejected in quality control. Further to the aforementioned fixed costs, $2000 is also considered for miscellaneous expenses, such as initial assembly line setup, buying required tables and test instrumentation. In Table 4.6 and Table 4.7, different costs of the micro-pump production are summarized as variable and fixed costs. Table 4.6 Micro-pump production variable costs Variable Costs Description Labour per Material per cost per unit unit (USD) unit (USD) (USD) Piezoelectric Buy $0.68 Casings Injection Buy, Costs for one pair of $1.19 Moulding Casings Micro-valves chemical etching Buy $6.87 Assembly and Heat $0.50 _ $0.50 press bonding Assembly-gluing Based on 0.3gr glue $0.25 $0.09 $0.34 piezoelectric*,loctite 495 Super Glue Quality control and Testing each micro-pump for $0.50 $0.50 testing self-priming, maximum flow and pressure head Packaging $0.10 $0.05 $0.15 Warrenty Set-aside $0.25 *Current assembly process; future design work modification to use heat press bonding 122

139 Table 4.7 Micro-pump production fixed costs Fixed costs Description Cost (USD) Plant and Utilities $ 30/m 2 per month, for a year $10,800 Mould machining Assuming $100/hr for required $1,400 technical labour Chemical etching R&D After considering the R&D $1,180 ( ) discount (-$1740) by AMIF Heat press sealing machine $9,335 Heat press fixtures machining $580 Quality control test setup costs $1000 Miscellaneous costs e.g. initial assembly plant setup $2, ) and Table 4.8 show the overall production cost per unit of the micro-pump for different production volumes. To consider the cost of rejected products (~3%) in quality control, a factor of 1.03 is multiplied by the net variable costs per unit. 4.2) Table 4.8 Micro-pump overall cost for different production volume N: Production Volume (units) Cost Per Unit (USD) 1,000 $ ,000 $ ,000 $ ,000 $ ,000 $ 10.8 Based on the equation 4.2), if the annual production volume exceeds about 6,000 units, the cost per unit will be less than $15 per micro-pump. 123

140 5 Chapter 5: Testing, Results and Discussion Different micro-valves designs and a number of assembled micro-pumps were tested to ensure feasibility and quality of the micro-pump beyond the manufacturing and assembly tests described earlier. Micro-valves manufactured for a design of experiment were tested by applying constant static pressure to obtain their static flow characteristics. The data from the static flow characteristics obtained from the micro-valves was used in the micro-pump simulation and optimization. The valves were then assembled and tested in the micro-pump injection moulding casings to demonstrate their functionality and investigate their dynamic behaviour. While a number of successful tests were performed on the micro-pump, many of the performance tests were unsuccessful due to inconsistent internal leakage between the outlet and inlet of the pump. The cause of this problem was investigated and modifications to the micro-pump casings and to the outer geometry of the valves were applied, as it was shown in Figure 4.9. Finally, a new casing prototype was CNC machined to perform dynamic testing of the micro-pumps, and modifications to the final injection moulding casing was left as future work. Based on the static testing and the simulations results, a revised test design for the new outer geometry micro-valves was performed. The new micro-valves were then assembled and tested in the final micro-pump prototype. As a result, the effect of inlet and outlet valve parameters on micro-pump performance was investigated for the revised micro-valves. Finally, simulation results using the results from the static testing of the new micro-valves were compared to the prototype micro-pump performance and the simulation method was verified. 124

141 5.1 Micro-valve static flow testing A two piece fixture, shown in Figure 5.1, was designed to perform static flow tests of the micro-valves. The micro-valve plate was enclosed and sealed between the brass fixture plates. In this test, pressure was applied to the valve using a column of water and the static flow rate was measured. By applying a column of water in a similar manner to the opposite side of the valve, the static backflow and the sealing characteristics of the valves was also obtained. Figure 5.1 Micro-valves static testing The flow and backflow characteristics of the fixture valves are highly affected by the valve seat hole diameter, valve stiffness, and valve shape. However, it was observed that the backflow behaviour of the valves was highly influenced by the quality of the valves. Small magnitude flexure of the valve during handling affected its residual stresses and thus its stiffness and could cause valve sealing problems that increased backflow. Figure 5.2 compares the static test results for valves with the same stiffness and different valve seat hole diameters. As can be seen, the flow increases as the seat hole area increases, while the flow trend maintains its pattern. In Figure 5.3, valves with different stiffness are compared, indicating that the stiffness at the levels tested does not significantly affect the flow. However, at relatively higher pressures, stiffer valves have higher flow rates, which could be caused by the pattern of flow passed through the valve. 125

142 Figure 5.2 Static pressure flow characteristic of valves with different valve seat hole diameters Figure 5.3 Static pressure flow characteristic of valves with different valve stiffness Similarly, applying pressure to the opposite direction of the valves provided a measure of the stationary backflow characteristics. Ideally, this flow is zero, however in passive valves this flow will be present except when there is sufficient load applied on the valve to create a proper seal. The backflow characteristics of valves with different valve seat hole diameters and stiffness are shown in Figure 5.4 and Figure 5.5, respectively. In Figure 5.4, it can be observed that a 0.5 mm valve seat hole diameter is very good at preventing the backflow, but the two other valves have significant backflow, particularly at lower pressures, and a certain amount of constant leakage occurs at higher pressures. For a 0.75 mm valve seat hole diameter, by increasing the pressure the amount of leakage increases. By further increasing the 126

143 pressure, the amount of backflow decreases until about -10kPa, as increasing the load on the valve also improves the sealing of the valve. Above -10kPa, the backflow becomes constant, as the valve reaches a point where there is a balance between improved sealing and increased flow due to the pressure increase. For the valve with the largest seat hole diameter, the backflow increases until it reaches its balance point at about -15kPa, thereafter backflow is relatively constant. Figure 5.4 Static pressure backflow for valves with different valve seat hole diameters Figure 5.5 shows no specific trend for the effect of stiffness on the backflow. However, for 0.5 and 1mm hole diameters, which are not shown in the figure, a very slight improvement (reduction) in the backflow was observed for higher stiffness. Therefore it can be concluded that the stiffness does not affect backflow significantly. The variation in the backflow, which can be observed in Figure 5.5, could be caused by variation in valve quality and the small flexure that could occur to the valve during handling. 127

144 Figure 5.5 Static pressure backflow for valves with different valve stiffness 5.2 Simulation results The flow passed through the valve for a full motivator cycle can be estimated by applying the static flow properties of the valve into the analytical model of the micro-pump. For this purpose, partial regressions between flow and pressure, of the test results presented in Figure 5.2 to Figure 5.5 are performed and the regression parameters were used for the flow properties of the inlet and outlet valves in the analytical model. Figure 5.6 shows the instantaneous flow through the valve for a complete cycle of the micro-pump, together with the differential pressure across the valve at 30 Hz and a 10 kpa pressure head for the micro-pump. The dashed line in the figures is the harmonic differential pressure applied to the valve. The effect of seat hole diameter and valve stiffness can be seen in Figure 5.6a and Figure 5.6b, respectively. In Figure 5.6a, it can be observed that while under positive pressure, a relatively large amount of flow is passed through the valve with a 1mm diameter valve seat hole, but there is a large amount of leakage through the remainder of the cycle. In contrast, the back flow for the valve with smallest hole diameter is almost eliminated, but the overall flow is also reduced. The 0.75 mm valve seat hole diameter has both higher flow than the valve with a 1mm seat hole diameter and a significantly reduced backflow. As can be seen in Figure 5.6b, valves with higher stiffness have sharper peaks and respond faster to the pressure changes and provide higher forward flow. However, the 128

145 backflow variation seen in Figure 5.6b may be due to variation in parameters taken from tested valves which have quality variation in their geometry, and this backflow variation may result from valve quality influenced by handling more than by variation in their designed stiffness. In order to obtain the flow rate of the valves for each cycle, the integral of the flow curves shown in Figure 5.6 is calculated. Multiplying the single cycle integral by the number of cycles per minute, an average flow rate is calculated. Table 5.1 shows the average flow rates at a 30Hz against a 10kPa pressure for the valves considered in the designed experiment. Table 3 also shows the average flow rate for the three valves with 142 N/m stiffness calculated at 80Hz and the same pressure head. 129

146 a) b) Figure 5.6 Instantaneous flow for complete cycle at 30Hz and 10kPa pressure head, plotted for: a) different hole diameters b) different valves stiffness 130

147 Table 5.1 Average flow rate at 10kPa pressure head D (mm) K (N/m) Average flow Rate (ml/min) 30 Hz 80 Hz In Figure 5.7, the effect of valve seat hole diameter on the average flow rate with a 142 N/m stiffness valves is plotted. Despite the different performances of the valves with 0.5mm and 0.75 mm valve seat hole diameters, under the positive pressure difference also shown in Figure 5.6a, their net flow is close. These results also shows that a 1mm valve seat hole diameter produces a relatively small flow at 30 Hz due to a large amount of backflow in each cycle. However, at higher operating frequencies, the flow rate of the 1mm seat hole diameter valve exceeds that of the 0.75mm seat hole diameter valve. This is because of the difference in the pattern of the differential pressure across the valve that changes with increasing frequency. The positive differential pressure causes forward flow to increase, while the negative differential pressure across the valve decreases, reducing the importance of backflow to the average flow rate. This point was also verified in dynamic testing of the micro-valves. Figure 5.7 Effect on the average flow rate of seat hole diameter at 30 and 80 Hz 131

148 Figure 5.8, illustrates the effect of valve stiffness on the average flow. As can be observed, the valve with mid-level stiffness has a higher flow rate and it is expected that the optimum valve stiffness can be found between the 57 and 142 N/m. Also, because back flow performance is not significantly affected by the stiffness, the trend of the curve in Figure 5.8 is not expected to significantly change with frequency. Although the results in Figure 5.6 are based on parameter values derived from experimental test results, this modeling approach and the trends shown in the curves are helpful for predicting the flow characteristic of valve. However, valve instability is not considered in the model and this deficiency could affect the overall conclusions. Figure 5.8 Effect of valve stiffness on average flow rate 5.3 Micro-pump testing The assembled micro-pumps were tested using the test setup illustrated in Figure 5.9. The micro-pump piezoelectric disc is driven and controlled using a commercially available micro-pump controller (Thinxxs micro-pump controller) which provides a 5V alternating current at variable frequencies. In this setup, a short circuit is used for priming the micropump at start-up and a bubble trapper is used before the micro-pump. 132

149 Figure 5.9 Micro-pump test setup schematic Figure 5.10 shows the final casing prototype and the assembled micro-pump prototype. These casings were only designed for the dynamic testing of the micro-pump and they are having the advantage of allowing the test of different combinations of separated valve pieces shown in Figure In the assembly process, micro-valve plates are placed in the inlet and outlet pockets, as shown in Figure 5.10b, and form a datum surface between the upper and lower casings. Despite a tight tolerance between the valves and pockets, variation in alignment of the two valve pieces was one major cause of error in the testing. 133

150 a) b) c) Figure 5.10 Micro-pump prototype for dynamic testing, a) Lower part of the casing and a gasket, b) upper part of the casing c) assembled micro-pump As shown in Figure 5.11, two valve seat hole diameters (0.6 and 0.8mm) and two valve stiffness's (K=79, 142 N/m) are used in these tests. A total of 16 (2 4 ) possible combinations of these four pieces can be used. In this test, a Taguchi design of experiment with a total of 8 tests was selected that enabled the investigation of all main effects and first order interactions. Further, to provide more degrees of interaction and provide a better estimate of the variance, two more tests were added to the Taguchi design of experiment. Therefore a total of the 10 experiments, as shown in Table 5.2, were run. 134

151 Figure 5.11 Separated valve pieces used in the micro-pump testing. Table 5.2 Valve pieces combinations used for micro-pump testing Test Number Inlet Outlet D i K i D o K o Comments L8 Main Effects and All First Order Interactions. Note: First order interactions confounded with higher order interactions. Additional Tests Flow rates for the valve combinations in Table 5.2 were measured at three different back pressures of 0, 11, and 20 kpa. The operating frequency of the piezoelectric driver was varied from 5 to 120 Hz. Detailed results of these tests can be found in Appendix B. General observations obtained from the testing are listed below: Flow rate of 11 ml/min at 11kPa pressure head. Flow rate of 7.4 ml/min at 20kPa pressure head. Maximum flow rate of 17 ml/min. 135

152 Flow (ml/min) An estimated maximum pressure head above 40 kpa. Maximum flow rates normally occur at frequencies between 40 and 60 Hz. Figure 5.12 shows the flow vs. pressure at a 50 Hz frequency for test numbers 4,6,8 and 9 in Table 5.2, which demonstrated relatively better performance T# 4 T# 6 T# 8 T# Pressure (kpa) Figure 5.12 Flow vs. pressure curve for tests number 4, 6, 8 and 9 in Table 5.2 at 50 Hz frequency Despite that the flow rates were not measured at back-pressures higher than 20kPa, the maximum pressure head can be estimated using linear regression. For the combination of valve parameters used in test number 4, the maximum pressure head is estimated to be about 42 kpa. Main and interaction effects of the inlet and outlet valves parameters were also investigated. To analyse the test results, the general linear model ANOVA function in the MINITAB TM software is used. In this analysis, the frequency is considered as a covariate parameter. In order to improve the accuracy of analysis and have a normal distribution of residuals, the analysis was performed separately over two different ranges of frequencies. The detailed result of this analysis is included in Appendix B. 136

153 Figure 5.13 and Figure 5.14 respectively show the main and interaction effects of the valves parameters at 10 to 50 Hz and 60 to 120 Hz. As it can be seen in Figure 5.13a, the outlet valve seat hole diameter (D o ) is the major significant parameter (p-value< 0.001) affecting the flow rate at the lower range of frequencies. Increasing the outlet diameter from 0.6 to 0.8 mm increases the micro-pump flow rate. The inlet valve stiffness (K i ) (pvalue=0.051) is the next most significant parameter in this range of frequencies, with a relatively less importance than the outlet diameter. Similarly, at the higher frequency range seen in Figure 5.13b, the outlet diameter and inlet stiffness are significant parameters. However, at this frequency range, inlet valve stiffness (p-value<0.001) has larger variance and is contrary to the lower frequency range, where increasing it will positively affect the flow rate. 137

154 a) 10 to 50 Hz frequency range b) 60 to 120 Hz frequency range Figure 5.13 Main effects plots for valve parameters at: a) frequencies between 10 to 50 Hz b) frequencies between 60 to120 Figure 5.14 is the interaction effects plots for both frequency ranges. While almost all parameters are interacting, these differences could be due to either variances or test errors. Therefore, to identify significant interactions, p-values obtained from ANOVA analysis will be used. Table 5.3 summarizes interactions that are significant with 95% confidence: 138

155 Table 5.3 p-values for valve parameters interaction effects Parameter Interaction Low Frequency (10-50 Hz) High Frequency ( Hz) Interaction p-value Interaction p-value Hole x Hole D i x D o D i x D o Stiffness x Hole K i x D o K i x D i Stiffness x Stiffness K i x K o K i x K o <0.001 The interactions between the outlet diameter and inlet valve parameters (D i x D o, K i x D o ) indicate that using the lower stiffness and smaller hole diameter for the inlet will reduce the sensitivity of the flow to the outlet diameter at lower frequencies. However at higher frequencies, as shown in Figure 5.14b, the inlet stiffness and outlet diameter do not interacting with each other anymore. Inlet and outlet stiffness interaction (K i x K o ) also indicates that at low frequencies having a lower stiffness reduces the sensitivity to the inlet stiffness. On the contrary, as can be seen Figure 5.14b, smaller outlet stiffness enhances the inlet stiffness effect at higher frequency. 139

156 a) 10 to 50 Hz frequency range b) 60 to 120 Hz frequency range Figure 5.14 Interaction effects plots for valve parameters at: a) frequencies between 10 to 50 Hz b) frequencies between 60 to120 In Figure 5.15 and Figure 5.16, a direct comparison between the valve parameter combinations gives insight into some of the main and interaction effects on the performance of the micro-pump. In both Figure 5.15a) and b), an increase in the outlet hole diameter from 0.6 to 0.8 mm increases flow rates, starting at a frequency about 30 Hz. However, due to an 140

157 Flow (ml/min) Flow (ml/min) interaction effect between the outlet hole diameter and the inlet valve parameters, the pattern of flow in mid-frequencies (30 to 70 Hz) are different. In Figure 5.15a both valves have the inlet diameter and stiffness at their lower levels, which, as it was mentioned before in the interaction effect plots, will reduce the sensitivity of the flow to outlet hole diameter and causes smoother increase in flow. a) Inlet: D=0.6mm, K=79 N/m Outlet: D 0.8,K 79 Outlet: D 0.6,K Freq. (Hz) b) Inlet: D=0.8mm, K=142 N/m Outlet: D 0.8,K 79 Outlet: D 0.6,K Freq. (Hz) Figure 5.15 Micro-pump test results for two different outlet hole diameters with inlet valve fixed at: a) D=0.6mm, K=79 N/m b) D=0.8mm, K=142 N/m Figure 5.16 compares the micro-pump flow rate for two different inlet valve stiffnesses. As can be seen, by increasing the inlet valve stiffness, at frequencies above 50 Hz, higher flow rates are achieved. The stiffer inlet valve appears to be less sensitive to the undesirable 141

158 Flow (ml/min) Flow (ml/min) dynamic effects occurring at higher frequencies in the micro-pump, which can provide a smooth increase in flow rate even in the frequencies higher than 80 Hz. Furthermore, due to an interaction between the smaller outlet hole diameter and the inlet stiffness, the maximum flow rate occurs at a relatively higher frequency for the micro-pump with stiffer inlet valve (Figure 5.16a). In another words, for applications which need smooth flow and therefore a higher operating frequency, using a stiffer inlet valve with a smaller outlet hole diameter is suggested. a) Outlet: D=0.6mm, K=79 N/m Inlet:D0.8, K142 Inlet:D0.8, K Freq. (Hz) b) Outlet: D=0.8mm, K=79 N/m Inlet:D0.6, K142 Inlet:D0.6, K Freq. (Hz) Figure 5.16 Micro-pump test results for two different inlet valve stiffness s with outlet valve fixed at: a) D=0.6mm, K=79 N/m b) D=0.8mm, K=79 N/m 142

159 Flow (ml/min) 5.4 Simulation verification The simulation model for the micro-pump was based on static testing of the microvalves and the Zengerle et al.[1993] analytical model of a diaphragm micro-pump. In order to verify this simulation method, static flow testing was performed on two sets of the new valves, one with a valve seat hole diameter D=0.6 mm and valve stiffness K=79N/m, and the other one having D=0.8mm and K=142N/m. Figure 5.17 shows the static test results for both forward-flow and back flow of these two valves. Similar to the previous simulations, as described in the section 5.2, a regression of the data characterized these two valves and the regression parameters were used for the flow properties of the inlet and outlet valves in the analytical model. The performance of the four combinations of the two valve parameters in both the inlet and outlet of the micro-pump (11, 12, 21 and 22) was simulated and compared to the actual micro-pump test results D 0.8, K 142 D 0.6, K Pressure (kpa) Figure 5.17 Static flow test results for valves with D=0.8mm,K=142N/m and D=0.6mm, K=79N/m Figure 5.18 shows a comparison between the simulation results and micro-pump experimental test results at 10kPa pressure head. The valves used in this figure are D=0.8mm and K=142N/m for both the inlet and outlet. A similar comparison between the simulation and test results are provided for the other combination of valve parameters in Appendix C. 143

160 Flow (ml/min) The simulation results predict an approximately linear increase of the flow rate with increasing frequency. However, the micro-pump test result shows non-linear behaviour, where flow rate growth above 50Hz stops increasing with frequency. As seen previously in Figure 5.15 and Figure 5.16, the maximum flow rate is normally achieved at a frequency between 40 and 60Hz. Dynamic effects, such as squeezed film damping, and then stops the flow increase with frequency. This effect is not accounted for in the simulation of the micropump. In Figure 5.18, at frequencies lower than 30 Hz, a good estimation of the flow is given by the simulation. The small difference between the empirical and simulation results is probably due to testing errors or errors in modeling the piezoelectric driver Frequency (Hz) Simulation Experiment Figure 5.18 Comparison between simulation results and micro-pump experimental test results at 10kPa pressure head The slight increase in the difference between the simulation and experimental results at frequencies around 30 Hz can be due to the variation in the piezoelectric displacement, which was considered to be constant in the modeling. It was also observed in Table 3.3 that the maximum displacement of the piezoelectric disk slightly increases at around the 30 Hz. Therefore, a better accordance to the empirical data can be achieved by improving the piezoelectric driver model in the simulation. The simulation method and its use of the static behaviour of the micro-valve in this work provides a good estimation of the micro-pump performance at frequencies under 50 Hz. For 144

161 higher frequencies the micro-valves performance is highly influenced by dynamic effects. Therefore, better performance estimation at higher frequencies should include a model of the dynamic behaviour of the valves, which requires further investigation. Further, the current micro-pump test results can be used for determination of correction factors for the static flow of the micro-valves at higher frequencies. Figure 5.19 shows an algorithm suggested for calculating the correction factors. Assuming that the squeezed film damping force prevents the micro-valve from normal closure at higher frequencies and that it is the main influence on the backflow properties of the valve, correction factors need only be applied to the backflow. Figure 5.19a shows that a constant backflow value B (III=I+B) and a scale factor A ( II=I x A) are applied to the backflow properties of the valve, in which A and B are both functions of frequency. In Figure 5.19b, a parameter φ is defined which is the difference between the simulation flow rate and experimental results. The objective is to minimize values of the φ at different frequencies. In Figure 5.19c, an algorithm is suggested which could be used to obtain factors A and B as a function of frequency. The simulation model can be significantly improved at higher frequencies by obtaining the correction factors described for different valve designs. Furthermore, these results could be very useful for the study of the squeezed film damping forces on the flow behaviour of the micro-valves. Currently, only few studies on squeezed film damping forces for very simple geometries have been reported in the literature. At the end, it should be mentioned that the method presented in Figure 5.19, is not yet validated and it is suggested as a future work. 145

162 a) b) c) Figure 5.19 a method for determination of dynamic effects correction factors for valve static flow properties a) correction Factors II= AxI, III=B+I b) Minimization objective parameter, φ c) The algorithm 5.5 Design verification In the micro-pump testing, it was observed that the maximum flow rate and pressure head normally occur between Hz frequencies and at a fixed pressure head, the flow rate can be adjusted by changing the operating frequency of the piezoelectric. By testing different 146

163 combinations of the valves components, it was observed that the valve with inlet/outlet hole diameters 0.8mm and stiffnesses 79 N/m can provide a maximum flow rate of 17 ml/min of water at 50 Hz frequency. For the same valve, based on the flow-pressure curves shown in Figure 5.12, by a linear approximation a maximum pressure of 31 kpa is predicted. For the valve diameter/stiffness combination of inlet of 0.6mm / 142N/m and outlet of 0.8mm / 79N/m, a maximum pressure of 42kPa and 14.5 ml/min at 50Hz frequency can be achieved. As a result, by using different valve designs, the micro-pump can have the required flexibility in meeting different performance requirements. A maximum flow rate of 17ml/min achieves a much higher flow rate than the minimum target value of 5 ml/min. For the maximum pressure head (42kPa), despite it being lower than our objective of 50 kpa, it is expected that 50kPa can be easily achieved by adjusting the valve diameter/stiffness combinations. Table 5.4 compares the maximum pressure head and flow rate of the micro-pump with the micro-pump designed by Xiao [2008], two micro-pumps developed by ThinXXS, two Japanese micro-pump fabricated by TAKASAGO, and a double piezoelectric actuator micro-pump (mp6) developed by Bartels Mikrotechnik. It shows that the new design is competitive in performance. Table 5.4 Comparison of the micro-pump performance P Max (kpa) Q Max (ml/min) New Design 42* 17 XIAO Design ThinXXS ThinXXS MDP TAKASAGO SDMP TAKASAGO SDMP Bartels Mikrotechnik mp In the current testing of the micro-pump, the micro-valves were separated to four pieces, to allow testing of the different combinations of hole diameters and stiffnesses. Due to the fact that the micro-pump performance is highly affected by valve quality, the production version 147

164 that uses a one piece valve is expected to have improved fabrication quality and better performance is expected. In this work, ThinXXS-EDP0604 micro-pump controller was used for driving the piezoelectric disc. According to the specification of the controller, it provides a stabilized supply DC voltage of 5V DC with maximum current consumption of the 200mA. Therefore the maximum power consumption of the micro-pump is less than 1 W ( P=VI=0.2*5 ). The overall package size of the micro-pump is 4 cm 3, which is composed of a circular base with diameter of 30.1mm and height of 5.6mm. Having the inlet/outlet in the same plane as the micro-pump base makes this design easier to integrate in fluidic circuits rather than micro-pumps having normal-to-plane inlet/outlet. Based on the preliminary cost estimation of the micro-pump, for production volume higher than 5,670 units, the cost of the micro-pump will be less than $15 USD and for production rates of 1,000, 10,000 and 100,000 the price will be, $35.83, $13.06 and $ At lower production volumes (<2,000), the cost is mainly dictated by fixed production costs (initial production setup costs). By increasing the production volume, the micro-pump price gets close to the net variable production cost (costs associated directly with fabrication and assembly of each micro-pump) per unit, which was estimate to be $ Currently, the micro-valve chemical etching forms 53% and 64% of the overall micro-pump cost at 10,000 and 100,000 production volumes, respectively. However, the $6.87 per unit micro-valve cost considered in the cost estimation analysis is expected to be reduced through process optimization, higher yields, and larger mask area. To further reduce the cost of the micropump for disposable applications which do not need high durability, similar valves with higher thickness can be made from plastic using hot embossing process. Due to the very low cost of hot embossing process it is expected that the cost of micro-pump fabrication can be reduced to under $ 5 per unit. During micro-pump testing it was observed that the micro-pump normally primed between 7 to 15 second after start-up. In this time, air trapped inside the micro-pump was evacuated. It was also observed that its flow rate is not influenced by orientation when pumping at a specified pressure. Using finite element analysis, stresses occurring on the valve 148

165 during the operation were calculated and by controlling the geometry of the valves, the highest stresses were kept under the endurance limit of the stainless steel (i.e., the stress limit under which fatigue would not occur) with a factor of safety of All micro-pump testing was conducted using room temperature water for the working fluid. However, according to the specifications of the selected actuation mechanism (included in Appendix A), the piezoelectric disk has an operating temperature range of -20 to 70 o C. Further, TOPAS COC s has a deflection temperature (i.e., the temperature that the polymer distorts under its normal loading) at 75 o C, and the micro-valve, due to it being made of stainless steel, can work at different temperature. Therefore, the selected component materials and the bonding method (which is the same material as the casings) has the potential of working at a wider range of fluid temperatures (-20 to 70 o C). However, whether the micropump as a system can work at other than room temperatures requires additional testing and analysis, which is an area suggested for future work. Table 5.5 compares the performance achieved to the specifications of the micro-pump derived from the desired design requirements. As mentioned in the design objectives, it is desired that the micro-pump have no than 0.5% failure rate in 100 hours of operation, corresponding to $0.25 per unit warranty cost based on the desired $50 cost of the micropump (4 to 5 times of the desired manufacturing cost of less than $15). To test this reliability criterion, a large number of micro-pumps should be tested to failure, in various situations which could prevent the micro-pumps from normal performance. This amount of testing is beyond the scope of this work. Also, for the durability test a number of micro-pumps (e.g. 10 samples), should be tested in their normal operating condition to failure. Therefore this objective can be considered during the mass fabrication of the micro-pump. In the future work section, the different conditions which could be tested for during the reliability tests are mentioned. 149

166 Reliability /Durability Low Cost Compactnes s Performance Table 5.5 micro-pump design verification Application Requirement Design Specifications Micro-pump Performance Maximum flow rate > 5 ml/min for water 17 ml/min Adjustable flow Controlling the flow rate by adjusting Yes operating frequency Pressure head 50 kpa for water 42 kpa * Power consumption < 1 W Yes Working fluid water Yes Overall dimensions/ Package size Micro-pump layout < (30 x 25 x 10 mm) / < 7.5 cm x30.1x5.6 mm, 4 cm 3 Inlet and outlet in the same plane as the body Yes Fabrication process Casings: injection moulding Yes Fabrication process Chemical etching /Embossing of the micro-valves Chemical etching: yes Embossing: future work Fabrication process Using standardized piezoelectric Yes minimum number of parts Modularizing and integrating the components Valve pieces are integrated in piece, Casing is composed of only two pieces Assembly process One directional assembly process Yes Mistake proofing the assembly process Using alignment features Yes Service Life >100 hours Not addressed Valve Fatigue Life >2x10 7 cycles Yes Reliability No more than 0.5% failure rate in 100 Not addressed hours Operating Temperature Room Temperature Yes Self priming < 20 sec sec. Having no leakage Considering proper sealing features Overlapping features for sealing and reliable bonding method. Heat press bonding (good for more than * estimate value 4 times of internal pressure of 50kPa) 150

167 6 Chapter 6: Conclusion In this work a novel micro-pump was designed, analysed and prototyped. This design was influenced by design for manufacturing and assembly (DFM and DFA) concepts. Having the aim of designing a micro-pump for low cost, the total number of parts was reduced and appropriate manufacturing methods were adopted. The micro-pump is composed of a total of four parts, the upper and lower casing halves, a single plate inlet and outlet valve, and a piezoelectric disc. Casing parts were successfully moulded of COC8007 and the valve plate was etched from stainless steel shim stock. The valve was designed as a single piece to minimize the number of parts and after etching is bent to form the inlet and outlet cantilever valves and valve seats. A heat press method for assembling and sealing the casings together was designed and tested. This method can reduce the cost of assembly process by simplifying the production equipment, using a process suitable for automatic assembly, and has the advantage in many applications of having a uniform material for the whole casing. From the test results, it was observed that the assembly method provides good sealing and a high strength bond. Combining different methods of analysis and testing, a modular model for micro-pump simulation based on static test results of valves was developed, including making use of experimental exploration using a design of experiment. This model can be used in micro-valve and piezoelectric selection and also for the further optimization of the micro-pump. This model was verified by testing the micro-pump, however, the integration model should be further improved to achieve a more accurate performance prediction at higher frequencies, i.e. above 50Hz, where the valves performance are more influenced with dynamic effects. Valves were tested statically to obtain the flow characteristics of different valve designs. While a number of successful tests were performed on the injection moulded micro-pump, many of the performance tests were unsuccessful due to inconsistent internal leakage between the outlet and inlet of the pump. Therefore, the design was modified to reduce the prevalence of internal leakage and a testing prototype was developed. As a result, a maximum flow rate 151

168 of 17 ml/min and an estimated maximum pressure head of above 40 kpa were achieved. Also, flow rates as high as 11 and 7.4 ml/min at 11 and 20 kpa, respectively, were achieved. In conclusion, a competitive low cost micro-pump designed for high performance and longer fatigue life, useful to a variety of applications is obtained by applying design for manufacturing and mass fabrication methods to the overall micro-pump design. 6.1 Contributions to Engineering Science In this work, a comprehensive product design for manufacturing process was followed. During this process various innovative ideas, fabrication methods proper for mass production, and a practical simulation method were developed Innovations associate with the micro-pump design Some of the innovations and novel ideas developed in this work, together with their advantages and limitations, are listed in Table 6.1. These innovations resulted in a micropump that can better satisfy the design requirements. Table 6.1 Innovations with the micro-pump design, advantages and limitations Design Innovation Advantages Limitations One piece valve design Stainless steel micro-valves Casings with minimum number of parts with in plane inlet/outlet, Designed for micro injection moulding Heat Press Bonding Reduces manufacturing cost and facilitates assembly Better reliability and fatigue life Reduces manufacturing cost. In plane inlet/outlet is more suited for most of applications Very strong bonding, No need for gluing and suitable for automotive assembly. Valve plate bending is required. Increases the susceptible area for internal leakage Higher fabrication cost Increases the susceptible area for internal leakage 152

169 Some of the results of this work were presented as a paper entitled Low cost micropump valve design in ASME International Design Engineering Technical Conferences (IDETC 2012), Chicago, IL, USA, August 12-15, Simulations and analysis In this work, a modular simulation model for the micro-pump was developed. Due to the complexity of the fluid-structure analysis of the micro-valves, analytical modeling and computational fluid dynamic cannot provide accurate predictions about the valves flow behaviour. In this model, results from static flow testing of the micro-valves, piezoelectric displacement test results, and finite element structural analysis results were incorporated in the model. As a result, a practical model of micro-pump performance is achieved. This model can provide a good prediction of the flow at frequencies under 50 Hz. However, due to the limitation of the static micro-valve flow properties model, dynamic effects such as squeezed film damping are not considered. As a result, the simulation predictions are not valid at higher frequencies. Despite the deviations observed between the simulation results and experiments, the modular nature of the simulation method allows it to be easily improved. The core of the modular simulation is based on the fluid dynamics continuity equation, which itself doesn't cause any error. Therefore, obtaining more accurate models for the valves flow properties, piezoelectric behaviour, and considering possible internal leakages, the simulation error can be significantly reduced Fabrication and Assembly methods The material selected for the micro-valve is stainless steel shim stock, which has a very good resistivity to most corrosive or alkaline liquids, and further, it can increase the fatigue life of the valves. A chemical wet etching process to which economies of scale can be applied was selected and tested. Despite performing a number of studies on stainless steel wet etching, their non isotropic material structure and high resistivity to etching required a modified wet etching process be developed in the University of Calgary AMIF lab. The use of micro-injection molding has been reported in the literature for a number of micro-pumps. In this work, injection moulding was investigated due to the minimization of 153

170 the number of the parts, in plane inlet/outlet, heat press sealing, and the use of features with very small dimensions, tolerances, and high aspect ratio features. This required mold flow simulation to ensure the complete mold fill and a relatively uniform shrinkage in the injection moulding. A novel heat press assembly method for micro-pumps was designed and tested. Advantages of this method include: suitable for automotive assembly, high strength and good sealing, and removing the need for glue. In the future, a similar method can also be developed for attaching the piezoelectric disc. 6.2 Recommendations for Future Work While the latest prototype of the micro-pump was successfully tested, this prototype is not yet adapted for the desired fabrication and assembly processes. Therefore, a modified design, shown in Figure 6.1 and Figure 6.2 is suggested for future works. In this design, the valve plate is kept as one piece and casings are modified to provide a better sealing around the valve plate and between the inlet and outlet. Also the outer geometries of the casings remain unchanged. Despite the application advantages of in-plane inlet and outlet for the micro-pump, it increases the possible leakage points and adds to the number of assembly datum surfaces in the injection moulded casings. In the proposed design, together with reducing possible leakage area, overlapping features and tight tolerances are considered in the design to have a better sealing around these channels. The proposed design revision will eliminate the need for having precision contact between the green surfaces shown in this figure. Similarly, overlapping features around the valve plate will enhance sealing. The main datum surface area is reduced and the only datum surface required is the yellow surface shown in Figure 6.1. However, a better seal can be achieved by providing a tight tolerance between the green surfaces. It should be mentioned that insert machining and injection moulding considerations have not yet been applied to the design shown in Figure

171 a) b) Figure 6.1 The suggested modified design for the injection moulding casings of micro-pump a) upper casing, b) lower casing Figure 6.2 a), b) and c) respectively illustrate the flat pattern, bent form of the new valve, and the valve placed in the upper casing. In the valve shown in Figure 6.2, the relative position of the inlet and outlet are the slightly (0.5mm) moved apart from the latest etched valves, to allow a better sealing between the valves. a) b) c) Figure 6.2 Suggested design for the valve a) flatten pattern b) bent valve c) valve placed inside the upper casing 155

172 Currently the simulation method by using static flow properties of the micro-valve, provides a good estimation of the micro-pump performance at frequencies under 50 Hz. At higher frequencies the micro-valves performance is highly influenced by dynamic effects. Therefore, to improve the simulation, further study of micro-valves dynamic flow behaviour at higher frequencies relative to the static flow properties are needed. Using the results of this study and a method such as the one presented in Figure 5.19, a better simulation model of micro-pump can be achieved based on modified static flow properties for the higher frequencies. As shown in the cost estimation section, one of the major manufacturing costs of the micro-pump is chemical etching of the micro-valves, which forms 53% and 64% of the overall micro-pump cost at 10,000 and 100,000 production volumes, respectively. To achieve a manufacturing cost of less than $5 for disposable applications, in future work it is recommended that the valves be manufactured from high strength polymers such TOPAS COCs using a hot embossing process. Due to the very low cost of hot embossing process, each valve can be fabricated at a very low cost. Also, the hot embossing process can be performed using the same heat press machine that seals the casings together. This process will require micro-machining a very accurate embossing mould with mills as small as 0.05 mm, which is within the capabilities of commercially available micro-machining stations like the KERN micro-cnc machining device in MEDAL lab. Using plastic will require thicker valves (about instead of current thickness of ), and will also require the valves to be made in two pieces, which will eliminate the bending process but will slightly add to the assembly costs, e.g., about $ 0.08 for 10 seconds of human intervention. Also, plastic valves won t provide durability similar to the stainless steel micro-valves. As a result, it is predicted that micro-valves could be obtained using this method at less than $0.20 per unit and about $2,000 of fixed cost of machining the embossing mould. Similar to the casings, the heat press bonding for the assembly of the piezoelectric can also be applied, which can both improve the strength of the bonding and reduce the assembly costs. The fixture required for the heat press bonding of piezoelectric can be integrated with the fixture for bonding the casings. Therefore both of the processes can be performed at the same time further reducing costs. 156

173 In the future, to test and improve the durability and reliability of the micro-pump, a large number of samples, should be tested to determine the service life and percentage of failure at different conditions. Table 6.2 shows a number of recommended reliability tests. In these tests, different conditions that might adversely affect the performance of the micro-pump are considered. By performing these tests, the micro-pump reliability in handling these conditions can be verified or the condition in which the micro-pump can perform reliably, e.g., maximum particle size or maximum amount of salt dissolve in water, can be specified. Table 6.2 Recommended reliability tests Reliability tests Typical method for testing Flow with particle Using water containing particles Sediment formation Using saltwater with different dilutions Bubble tolerance Using gas dissolved water such as carbonated water Shock, as in portable devices Testing the micro-pump on an oscillating table and at different orientations Impact resistance Dropping the micro-pump during and off the operation Fluid temperature Testing the micro-pump at different fluid temperatures To test the micro-pump for durability, a number of micro-pump (i.e. about 10 units), should be tested to failure under normal operating condition. 157

174 7 References: Abdul A. M., ME Thesis Integrating finite element and optimization software for shape optimization: a micro-pump design case study, 2010, Department of Mechanical and Manufacturing Engineering, University of Calgary. Accoto D., Carrozza M. C., and Dario P., Modelling of Micro-pumps Using Unimorph Piezoelectric Actuator and Ball Valves, 2000, Micromechanics and Microengineering, v 10, pp Amirouche F., Zhou Y., and Johnson T. Current micro-pump technologies and their biomedical applications, 2009, Microsystem Technologies, v 15, n 5, pp Bart S.F., Mehregany M., Tavrow L.S. and Lang J.H., Microfabricated electrohydrodynamic pumps, 1989, Transducers 89 Switzerland, p Bartels microtechnik, mp6 micro-pump, last visited Aug , Barth P. W., Silicon Micro-valves for Gas Flow Control, 1995, Transducer 95, Stockholm, Sweden, V. 2, pp Barth P. W., Beatty C. C., Field L. A., Baker J. W., and Gordon G. B., A robust normally closed silicon micro-valve, 1994, Technical Digest of the Solid-State Sensor and Actuator Workshop, Hilton Head, SC, USA, pp BASF chemical company, Visited July 2012, Begley M. R. and Mackin T. J., Dynamic characteristics of common micro actuator System, 2003, Mech. Phys. Solids, V. 51, pp Benard W.L., Kahn H., Heuer A.H. and Huff M.A., A titanium nickel shape-memory alloy actuated micropump, 1997, Inter. Conf. on Solid-State Sensors and Actuators, Proceedings, v. 1, pp Bendib S. and Olivier F., Analytical study of microchannel and passive micro-valve application to micro-pump simulator, 2001, SPIE 4593, pp Blanchard D. B., Ligrani P. M., and Gale B. K., Single-Disk and Double-Disk Viscous Micro-pumps, 2005, Sensors and Actuators V. 122, No. 1, pp

175 Bohm S., Burger G. J., Korthorst M. T. and Roseboom F., A micromachined silicon valve driven by a miniature bi-stable electro-magnetic actuator, 2000, Sensorsand actuators A80, pp Bohm S., Timmer B., Olthuis W. and Bergveld P. A closed-loop controlled electrochemically actuated micro-dosing system, 2000, Micromechanics Micro engineering 10, pp Bosch D., Heimhofer B., Muck G., Seidel H., Thumser U., and Welser W., A Silicon Micro-valve with Combined Electromagnetic/Electrostatic Actuation, 1993, Sensors and Actuators A, V , pp Bourrouina T. and Grandchamp J. P., Modeling micro-pumps with electrical equivalent networks", 1996, Micromechanics Micro engineering 6, pp Brand S, Microdosing systems: micro-pumps the beating heart of micro-fluidics., 2006,visited April 2012, Cabuz C., Cabuz E. I., Herb W.R.,Rolfer T. and Zook D., Measoscopic Sampler Based on 3D Array of Ellectrostatically Activated Diaphragms, 1999, Inter. Conf. on Solid-State Sensors and Actuators,pp Cao L., Mantell S. and Polla D., Design and simulation of an implantable medical drug delivery system using micro-electromechanical systems technology, 2001, Sensors and Actuators A 94, pp Carmona M., Marco S., Samitie J. r, Acero M. C., Plaza J. A., and Esteve J., An Analytical Model for Passive Micro-valves,, 2001, Sensors and Materials, V. 13, pp Carrozza M.C., Croce N., Magnani B. and Dario P., A piezoelectricallydriven stereolithography-fabricated micro-pump, 1995, Micromechanics and Microengineering 5,pp Chakraborty I., Tang W. C., Bame D. P., and Tang T. K., MEMS Micro-Valve for Space Applications, 2000, Sensors and Actuators A-Physical, v. 83, pp Chen C. and Santiago J. G., A Planar Electroosmotic Micro-pump, 2002, J. MEMS, V. 11, pp Chen S. C., Cheng C. H. and Lin Y. C., Analysis and experiment of a novel actuating design with a shear mode PZT actuator for microfluidic application, 2007, Sensors and Actuators A Physical, V. 135, pp

176 Dash A. K. and Cudworth G. C., Therapeutic applications of implantable drug delivery systems, 1998, J. Pharmacol. Toxicol. Methods, pp Dewa A.S., Deng K., Bonham C., Ritter D.C., Guckel H. and Masood-Ansari S., Development of LIGA-fabricated, self-priming, in-line gear pumps, 1997, 9th Int. Conf. Solid-State Sensors and Actuators, pp Dopper J., Clemens M., Ehrfeld W., Jung S., Ka mper K. P. and Lehr H., Micro gear pumps for dosing of viscous fluids, 1997, Micromechanics Microengineering, pp Du M.,Ye X., Wu K. and Zhou Z., A Peristaltic Micro-pump Driven by a Rotating Motor with Magnetically Attracted Steel Balls, 2009, Sensors 9, no. 4, pp Esashi M., Shoji S., and Nakano A., Normally Closed Micro-valve and Micro-pump Fabricated on a Silicon-Wafer, 1989, Sensors and Actuators, v. 20, pp Fahrenberg J., Bier W., Maas D., Menz W., Ruprecht R., and Schomburg W. K., A Microvalve System Fabricated by Thermoplastic Molding, 1995, J. of Micromechanics and Microengineering, v. 5, pp Feldt C. and Chew L., Geometry-based macro-tool evaluation of non-moving-part valvular microchannels 2002, J. Micromechanics Microengineering, pp Feustel, A., Krusemark, O. and Müller, J., Numerical simulation and optimization of planar electromagnetic actuators, 1998, Sensors and Actuators A Physics 70, pp Fiorini G. S. and Chiu D. T., Disposable micro-fluidic devices: fabrication, function, and application, 2005, BioTechniques 38, pp Folta J. A., Raley N. F. and Hee E. W. Design, fabrication, and testing of a miniature peristaltic membrane pump, 1992, Proc. Solid-State Sensor and Actuator Workshop, Hilton Head Island,pp Foster F. K., Bardell R. L., Afromowitz M. A., Sharma N. R. and Blanchard A., Design, fabrication and testing of fixed-valve micro-pumps, 1995,Proc. the ASME Fluids Engineering Division, IMECE 95, V. 234, pp Fu Y., Ghantasala M. K., Harvey E. and Qin L.," Design and fabrication of a hybrid actuator", 2005, Smart Mater. Structure 14, pp Gong Q. L., Zhou Z. Y., Yang Y. H., and Wang X. H., "Design, Optimization and Simulation on Micro-electromagnetic Pump", 2000, Sensors and Actuators A-Physical 83, pp

177 Griffiths D. J., Introduction to Electrodynamics, 1999, Upper Saddle River, N.J.: Prentice Hall. Grosjean C. and Tai Y C., A thermopneumatic peristaltic micro-pump, 1999, Proc. Transducers,Sendai, Japan, pp Hayden C.J. and Dalton C., Direct patterning of microelectrode arrays using femtosecond laser micromachining, 2010, Applied Surface Science 256, pp Huang S., Borca-Tasciuc D.-A. and Tichy J. A., A simple expression for fluid inertia force acting on micro-plates undergoing squeeze film damping, 2010, Proceedings of the Royal Society A, A 8,V. 467, No. 2126, pp Hunter R. J., Zeta Potential in Colloidal Science: Principles and Applications, 1981, London: Academic, pp Hsu T. R., Micro Assembly - A technology on the frontier of new industrial automation, 2005, 8th International Conference on Automation Technology, Taiwan, pp Ikeda T., Fundamentals of Piezoelectricity. New York, Oxford University Press., Iverson B. D. and Garimella S. V., Recent Advances in Microscale Pumping Technologies: A Review and Evaluation, 2008, Microfluid Nanofluid, V5, pp Jang J. and Lee S.S., Theoretical and experimental study of MHD (magnetohydrodynamic) micro-pump, 2000, Sensors and Actuators 80, pp Jerman H., Electrically-Activated, Normally-Closed Diaphragm Valves, 1994, J. Micromechanics and Microengineering, V. 4, pp Judy J. W, Microelectromechanical systems (MEMS): fabrication, design and Applications, 2001, Smart Materials and Structures, V. 10, No. 6, pp Judy J. W., Tamagawa T. and Polla D. L., Surface Machined Micromechanical Membrane Pump, 1991, IEEE 4th Int. Workshop on MEMS, pp Kamper K. P., Dopper J., Ehrfeld W. and Oberbeck S., A self-filling low-cost membrane micropump, 1998, Proc. 11th Annual Int. Workshop on Micro Electro Mechanical Systems, Heidelberg, Germany, pp Kan J., Yang Z., Peng T., Cheng G. and Wu B., Design and test of a high-performance piezoelectric micro-pump for drug delivery, 2005, Sensors and Actuators A 121, pp

178 Keppeler C. R., Micromilling for Mold Fabrication, 2003, LMA Reports, University of California at Berkeley, pp KERN Evo, micro feinwerktechnik gmbh & co. kg, visited June 2012, KERN, micro feinwerktechnik gmbh & co. kg, visited June Kluge S., Neumayer G., Schaber U., Wackerle M., Maichl M., Post P., Weinmann M. and Wanner R., Pneumatic silicon micro-valves with piezoelectric actuation, 2001, Proc. of Transducers 01/Eurosensors XV, Munich, Germany, pp Koch M, Evans A. G. R. and Brunnschweiler A. Coupled FEM simulation for the characterization of the fluid flow within a micromachined cantilever valve, 1995, Proc. Micro Mechanics Europe 96, pp Koch M., Harris N., Evans A.G.R., White N.M. and Brunnschweiler A., A novel micromachined pump based on thick-film piezoelectric actuation", 1998, Sensor Actuator A Phys A 70, pp Kramschuster A., Cavitt R., Ermer D., Chen Z., Turng L. S., Quantitative Study of Shrinkage and Warpage Behavior for Microcellular and Conventional Injection Molding, 2005, polymer engineering and sciences, pp Krutzch W. C. and Cooper P., Introduction: classification and selection of pumps, Pump Handbook ed I, J. Karassik et al,new York: McGraw-Hill, Kwang W. O. and Chong H. A., A review of micro-valves, 2006, J. Micromechanics Microengineering, V. 16, pp Laser D. J. and Santiago J. G., A review of micropumps, 2004, J. Micromechanics Microengineering, V. 14, pp Laser D. J., PhD Thesis Design, fabrication, and applications of silicon electroosmotic micropumps, 2005, department of mechanical engineering, University of Stanford. Lee S.W., Sim W.Y. and Yang S.S., Fabrication and in vitro test of a microsyringe, 2000, Sensors and Actuators A 83, pp Lemoff A.V. and Lee A.P., An AC electrohydrodynamic micro-pump, 2000, Sensors and Actuators B 63,pp

179 Luharuka R., LeBlanc S., Bintoro J. S., Berthelot Y. H. and Hesketh P. J., Simulated and experimental dynamic response characterization of an electromagnetic micro-valve, 2008, Sensors and Actuators A 143, pp Madou M. J., Fundamental of microfabrication, chapter 2: Pattern Transfer with Dry Etching Techniques, 2002, ISBN Mahendran S., Devarajan R., Nagarajan T. and Majdi A., A Review of Micro-EDM,2010, Proc. IMECS V. II, Hong Kong, pp Malekian M., PhD thesis, Investigation of micro milling processes, 2010 department of mechanical engineering, University of Calgary. Malekian M., Park S., Jun M. B.G., Modeling of dynamic micro-milling cutting forces, 2009, Int. J. of Machine Tools and Manufacture, V. 49, pp MIL-HDBK-217, Sampling procedure and tables for life and reliability testing, 1960 Moroney R.M., White R.M., Howe R.T., Ultrasonically induced microtransport, 1991,Proc. of the MEMS 91, Nara, Japan, 30, pp Morris C. J. and Forster F. K., Low-order modeling of resonance for fixed-valve micropumps based on first principles pump, 2003, J. Microelectromech. Syst. 12, pp Nath A. K., Laser etching and dry processing, Bull, 1988, Material Science., V. 11, Nos 2 & 3,1988, pp Nguyen N. T.,Huang X. and Chuan T.K., MEMS-Micro-pumps: A Review, 2002, journal of fluid engineering, V. 124, pp Nguyen N. T. and Wereley S.T., Micro-fluidics for internal flow control: micro-pumps. Fundam Appl Micro-fluidics, 2002, Artech House,pp Nguyen N.T. and Truong T.Q., A fully polymeric micro-pump with piezoelectric actuator, 2004, Sensor Actuator B Chemical 97,pp Nisar A., Afzulpurkar N., Mahaisavariya B. and Tuantranont A., MEMS-based micropumps in drug delivery and biomedical applications, 2008, Sensors and Actuators B:Chemical 130, pp Nisar A., Afzulpurkar N., Mahaisavariya B. and Tuantranont A., MEMS-based micropumps in drug delivery and biomedical applications, 2008, Sensors and Actuators B: Chemical, V. 130, pp

180 Nuyen N.T. and White R.M., Design and optimization of an ultrasonic flexural plate wave micro-pump using numerical simulation, 1999, Sensors and Actuators 77, pp Oh K.W. and Ahn C.H., A review of micro-valves, 2006, J. Micromechanics and Microengineering 16,pp Ohnstein T., Fukiura T., Ridley J. and Bonne U., Micromachined Silicon Micro-valve, 1990, MEMS 90, Napa Valley CA, pp Olsson A., Enoksson P., Stemme G. and Stemme E., Micromachined flat-walled valveless diffuser pumps, 1997, J. Microelectromech. Syst. 6, pp Olsson A., Enoksson P., Stemme G. and Stemme E., A valve-less planar pump in silicon, 1995, Proc. of the Transducers 95/Eurosensors IX, Stockholm, Sweden, pp Olsson A., Stemme G. and Stemme E. A valve-less planar fluid pump with 2 pump chambers, 1995, Sensors Actuators A 47, pp Otto K., Wood K., Product Design, 2000, Prentice Hall, New York. Ouellette J., A New Wave of Micro-fluidic Devices, 2003, visited Apr. 2012, The Industrial Physicist American Institute of Physics, 9/iss-4/p14.html Pan L. S., Ng T. Y., Wu X. H. and Lee H. P., Analysis of valveless micro-pumps with inertial effects, 2003, J. Micromech. Microeng. 13,pp Pan T., Kai E., Stay M., Barocas V. and Ziaie B., A magnetically driven PDMS peristaltic micro-pump. Engineering in Medicine and Biology Society",2004, IEMBS th annual Int. Conf. IEEE, pp Pan T., McDonald S.J., Kai E.M. and Ziaie B., A magnetically driven PDMS micro-pump with ball check-valves.,2005, Micromechanics and Microengineering 15,pp Paul P. H., Arnold D. and Rakestraw D. J., Electrokinetic generation of high pressures using porous microstructures, 1998, Proc. Micro Total Analysis Systems 1998 Banff, Canada, pp Piezo systems Inc., Piezoceramic materials and properties, visited Aug. 2012, 164

181 Piotter V. R., Hanemann T., Ruprecht R. and Hausselt J., Microinjection Moulding of Medical Device Components, 2002, Business Briefing: Medical Device Manufacturing & Technology, pp Piotter V., Micro Injection Moulding - current status and perspectives, 2001, Proceedings of Injection Moulding 2001 Conference, Copenhagen, pp Piotter V., Gietzelt T. and Merz L., Micro powder-injection moulding of metals and ceramics, 2003, Sadhana V. 28, pp Rao P. N. and Kunzru D., Fabrication of microchannels on stainless steel by wet chemical etching, 2007, J. Micromechanics and Microengineering. 17, pp Rapp R., W. Schomburg K., Maas D., Schulz J. and Stark W., LIGA micro-pump for gases and liquids, 1994, Sensors Actuators A, V. 40, pp Richter A. and Sandmaier H., An electrohydrodynamic micro-pump, 1990, Proc. of the MEMS 90, Napa Valley, USA, pp Richter A., Plettner A., Hoffmann K.A., Sandmaier H., Electrohydrodynamic pumping and flow measurement, 1991,Proc. of the MEMS 91, Nara, Japan, 30, 1991, pp Sammarco T. S. and Burns M. A., Thermocapillary pumping of discrete drops in microfabricated analysis devices, 1999, AIChE J. 45, pp Schneider P., Huck E., Reitz S., Parodat1 S., Schneider A. and Schwarz P., A modular approach for simulation-based optimization of MEMS, 2000, Design Modeling and Simulation in Microelectronics, Singapore, pp Seidel H., Csepregi L., Heuberger A. and BaumgSrtel H., Anisotropic Etching of Crystalline Silicon in Alkaline Solutions, 1990, J. Electrochem. Soc.,pp Shen M., Yamahata C. and Gijs M.A.M., Miniaturized PMMA ball-valve micro-pump with cylindrical electromagnetic actuator, 2008, Microelectronic Engineering 85 (5-6), pp Shen M., Yamahata C. and MAM G., "A high-performance compact electromagnetic actuator for a PMMA ball-valve micro-pump", 2008, J. Micromech. Microeng., pp Shinohara J., Suda M.,Furuta K. and Sakuhara T., A High Pressure Resistance Micro-pump Using Active And Normally Closed Valves, 2000, IEEE 13th Int. Workshop on MEMS, pp

182 Shoji S., Nakagawa S. and Esashi M., Micro-pump and sample-injector for integrated chemical analyzing systems, 1990, Sensors Actuators A 21,pp Smith J. G., "Piezoelectric Micro-pump With Three Valves Working Peristaltically", 1988, Sensors and Actuator, pp Starr J. B., Squeeze-film damping in solid-state accelerometers,1990, Solid-State Sensor and Actuator Workshop the 4th Technical Digest, IEEE, pp Stemme E. and Stemme G., A valveless diffuser/nozzle-based fluid pump, 1993, Sensors Actuators A 39, pp Stolzenberg S. J. and Linkenheimer W. H., Osmotic fluid reservoir for osmotically activated long-term continuous injector device, 1971, US Patent no. 3,604,417. TAKASAGO electronics Inc., SDMP 320 and SDMP 306 micro-pumps, visited Aug , Tuckerman D. B. and Pease R. F. W., High-performance heat sinking for VLSI, 1981, Electron Device Lettes, IEEE, V. 2, pp Van de Pol F. C. M., Van Lintel H. T. G., Elwenspoek M., and Fluitman J. H. J., A Thermopneumatic Micro-pump Based on Micro-Engineering Techniques, 1989, presented at 5th International Conference on Solid-State Sensors and Actuators/Eurosensors III, pp Van der Beek M.H.E., Peters G.W.M. and Meijer H.E.H., Characterization of polymers for improved shrinkage prediction in micromolding, 2008, Multi-material Micro Manufacture, pp Van der Pol F. C., Breeveld P. C. and Fluitman J. H. J., Bond graph modeling of an electrothermo-pneumatic micro-pump Technical Digest of MME, 1990, 2nd Workshop on Micromachining and Microsystems technical digest, pp Van Lintel H. T. G., Van de Pol F. C. M. and Bouwstra S., A piezoelectric micro-pump based on micromachining of silicon, 1988, Sensors Actuators 15, pp Voigt P., Schrag G., and Wachutka G., Electrofluidic full-system modelling of a flap valve micro-pump based on Kirchhoffian network theory, 1998, Sensors and Actuators A- Physical, V. 66, pp

183 Weber L., Zweibrucken (DE),ThinXXs GmbH, Micro-pump and method for production thereof, 2005, U.S. patent 7,771,176 B2. Whynott J., The Critical Questions When Micromolding for an MIS Device, visited 2011 May, Wilkinson C. D. W. and Rahman M., Dry etching and sputtering, 2003, Phil. Trans. R. Soc. Lond. A 362, pp Woias P., Micro-pumps - past, progress and future prospects, 2005, Sensors and Actuators B 105, pp Woias P., Micro-pumps summarizing the first two decades, 2001, Micro-fluidics and BioMEMS, SPIE, V. 4560, pp Woias P., Linnemann R., Richter M., Leistner A. and Hillerich B., Micro Total Analysis Systems, 1998, Kluwer Academic Publishers, Dordrecht, pp Xiao X., M.Sc. Thesis, Development of a novel micro-pump system, 2008, department of mechanical and manufacturing engineering, University of Calgary. Xu D., Wang L., Ding G., Zhou Y., Yu A. and Cai B., Characteristics and fabrication of NiTi/Si diaphragm micropump, 2001, Sensors and Actuators A Phys. 93, pp Yang K., PhD thesis, The fabrication and analysis of piezoelectric valveless micro-pump, 2004, National Yunlin University of Science and Technology. Yang Y., Zhou Z., Ye X. and Jiang X., Bimetallic Thermally Actuated Micropump, 1996, American Society of Mechanical Engineers: Dynamic Systems and Control Division, V.59, pp Yao S., Myers A. M., Posner J. D., Rose K. A. and Santiago J. G., Electroosmotic Pumps Fabricated From Porous Silicon Membranes, 2006, Journal of Microelectro mechanical Systems, V. 15, No. 3, pp Yin H.L., Huang Y.C., Fang W. and Hsieh J., A novel electromagnetic elastomer membrane actuator with a semi-embedded coil, 2007, Sensorse and Actuators A Phys 139, pp Yole, Status of the MEMS Industry, visited July 2012, pdf. 167

184 Yun K. S., Cho I. J., Bu J. U., Kim G. H., Jeon Y. S., and Kim C. J., A micropump driven by continuous electrowetting actuation for low voltage and low power operation, 2001, The 14th IEEE International Conference on MEMS, Interlaken Switzerland, pp Yun Y., Lee K. M., Lim G, Lee J H and Pak Y E, Modeling of diffuser/nozzle micro-pump under electrohydrodynamic backflow control, 2000, ASME FEDSM 00,Boston. Zhang L, Koo J and Jiang L, Measurements and modeling of two-phase flow in microchannels with nearly constant heat flux boundary conditions 2002, J. Microelectromech. Systems, pp Zengerle R., Geiger W., Ritcher M., Ulrich J., Kluge S. and Ritcher A., Application of micro diaphragm pumps in microfluid systems, 1994, Proc. Actuator 94, pp Zengerle R. and Richter M., Simulation of microfluid systems, 1994, J. Micromech. Microeng., V. 4, No. 4, pp Zengerle R., Richter A., and Sandmaier H., A micro membrane pump with electrostatic actuation, Micro Electro Mechanical Systems, 1992, MEMS 92, pp Zengerle R., Richter M., Brosinger F., Richter A., Sandmaier H., Performance simulation of Microminiaturized Membrane Pumps, Proc. Transducer 93; Yokohama,Japan, PP Zengerle R., Ulrich J., Kluge S., Richter M., and Richter A. A Bidirectional Silicon Micropump Sensors and Actuators A-Physical, V. 50, pp Zhou Y. and Amirouche F., An Electromagnetically-Actuated All-PDMS Valveless Micropump for Drug Delivery, 2011, Micromachines 2, pp

185 8 Appendices A: Engineering Drawings and Data Sheets Piezoelectric Disc: Figure A.1 Dimensions for the selected piezoelectric, dimensions are in millimetre 169