Flow of complex biofluids in microfluidic devices

Flow of Complex Biofluids in Microfluidic Devices Zhu Liang NATIONAL UNIVERSITY OF SINGAPORE 2006 Flow of Complex Biofluids in Microfluidic Devices Zhu Liang (B.Eng, Southeast University) A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE GRADUATE PROGRAMME IN BIOENGINEERING NATIONAL UNIVERSITY OF SINGAPORE 2006 Acknowledgement Firstly I’d like to express my sincere appreciations to my supervisor A/P Lim Siak Piang in Department of Mechanical Engineering, National University of Singapore, for his great help on the guidance of research as well as my life in Singapore. Prof. Lim is the most kind and nice person I have ever met and has given me as much love as my father. His great knowledge and novel ideas in scientific research has impressed me deeply. I feel extremely lucky to have Prof. Lim as my supervisor. I am also very grateful to Dr. Guillaume Chaidron, my co-supervisor during the two lab rotations. He not only taught me about research but also gave me so many advices of conducting myself. The past two years I spent with him is an unforgettable happy memory throughout my life. Thirdly, let me thank Ms. Ji Hongmiao from IME and Mr. Liu Yang from NUS for their generous help in experimental studies. I’d like to thank Dr. David Whyte from IHPC and Dr. Xie Wenfeng from NUS for their help in simulation work as well. Fourthly, I acknowledge all my classmates in GPBE and all the lab officers in Dynamic and Vibration Lab for their help and care to me. They are so lovely and friendly to me that I never feel alone even far away from my hometown. Lastly, I owe my thanks to my families – my father, mother and husband who have supported me for so many years. I can always feel their love with me and I will never give up because of their deep love. I Table of Content Acknowledgement.................................................................................................I Table of Content.................................................................................................. II Summary ............................................................................................................. V List of Figures ................................................................................................. VIII Nomenclature .....................................................................................................XI Chapter 1 Introduction ...................................................................................... 1 1.1 Microfluidic Devices in Biological Use – An Overview ......................... 1 1.2 Thesis Objective....................................................................................... 2 1.3 Thesis Organization ................................................................................. 3 Chapter 2 Literature Review ................................................................................ 5 2.1 Cell Sorting and Separation ..................................................................... 5 2.1.1 Counter Current Distribution (CCD) Method .................................. 5 2.1.2 Continuous Flow Electrophoresis (CFE) Method............................ 7 2.1.3 Fluorescence Activated Cell Sorter (FACS) .................................... 8 2.1.4 Cell Sorting Applications – Separation of Blood Cells ................... 9 2.1.5 Summary and Current Work .......................................................... 11 2.2 Mixing Techniques in Microfluidic Devices ......................................... 12 2.2.1 Passive Mixer................................................................................. 14 2.2.2 Active Mixer .................................................................................. 15 2.2.3 Summary and Current Work .......................................................... 20 2.3 Effect of Ultrasound on Flow of Complex Biofluid in Microdevices ... 21 2.3.1 Radial oscillations of the insonified microbubble.......................... 22 2.3.2 Translation of the insonified microbubble ..................................... 25 II 2.3.3 Acoustic Radiation Pressure of Suspended Particulates in Ultrasonic Standing Wave........................................................................... 29 2.4 Bubble Problem in Microfluidics........................................................... 30 2.4.1 How a Gas Bubble is Formed in Microfluidics ............................. 30 2.4.2 Bubble Problems in Microfluidic Devices..................................... 31 2.4.3 In Vivo Bubble problems ............................................................... 32 Chapter 3 Micro-flows under Ultrasonic Standing Wave .............................. 35 3.1 Background ............................................................................................ 35 3.2 Principles of the Design ......................................................................... 36 3.3 Materials & Methods.............................................................................. 38 3.3.1 The micro-parts and chip design .................................................... 39 3.3.2 Particles and working fluids........................................................... 41 3.4 Results and Discussions ......................................................................... 42 3.4.1 Flow of polystyrene particle suspension in the ultrasonic standing wave field .................................................................................................... 43 3.4.2 The frequency range for the polystyrene particles to converge ..... 45 3.4.3 Flow of Melamine particles suspended in water under ultrasonic standing wave .............................................................................................. 46 3.4.4 Effect of ultrasonic standing wave on the flow of milk-powder solution ........................................................................................................ 47 3.4.5 Flow of the complex fluid of milk-powder solution suspended with polystyrene particles in ultrasonic field....................................................... 48 3.4.6 Numerical Simulation Using CFD Software.................................. 49 3.4.7 Flow of two miscible fluids in magic-cross channel...................... 51 3.4.7.1 Flow of water and water................................................................. 51 III 3.4.7.2 Flow of saturated NaCl solution and water.................................... 53 3.4.7.3 Flow of ethanol and water.............................................................. 54 3.4.7.4 Flow of milk and water .................................................................. 58 3.5 Effect of Ultrasonic Standing Wave for Mixing Purpose ...................... 59 Chapter 4 Bubble Problems in Microfluidic Devices..................................... 62 4.1 Principle ................................................................................................. 62 4.2 Method and Materials ............................................................................ 63 4.3 Results .................................................................................................... 65 4.4 Discussion .............................................................................................. 66 4.5 Brief Summary....................................................................................... 69 Chapter 5 Conclusions and Recommendations .............................................. 70 5.1 Conclusions ............................................................................................ 70 5.2 Recommendations .................................................................................. 71 Reference List .................................................................................................... 72 IV Summary Along with the development of microelectronic chip design and manufacturing technologies, MEMS (Micro-Electro-Mechanical System) becomes more widely used in biotechnology field, which forms a specific subset namely BioMEMS. Moreover, integration of all the micro scale components on a single chip in order to fulfil the whole work flow of a biological process leads to another exciting classification of microdevices named Lab-on-a-chip. The potential of these devices is enormous in the field of diagnostics, medicine, therapeutics, biological research and military area. Specifically for biomedical purpose, they are potential powerful tools for precision surgery, rapid and cheap diagnosis, autonomous therapeutic management and early identification of diseases like cancer. When designing BioMEMS or Lab-on-a-chip, microfluidics is an important component to consider. Just as its name implies, the term “microfluidics” refers to the analysis and manipulation of fluids in structures with micrometer scale. At this dimension, many dominant forces are different from those in macro devices. Moreover, when the device is used for biological or medical applications, the flow of complex biofluids will be involved, which makes the prediction and analysis of the flow field even more complicated. Therefore, experimental observations on the microflows are especially important to characterize the factors mentioned above in order to make reliable and efficient microdevices. In particular, a microfluidic device conducts a certain process of the target fluids in a designed container such as micro-chamber or micro-channel and also V provides the coupling of the fluids into and out of this container. Hence separation and mixing are the two essential topics in microfluidics. The major work in this thesis is to experimentally observe and analyze the flow of complex fluids in microfluidic chips under various conditions for the purpose to optimize the design of BioMEMS or Lab-on-a-chip devices. Ultrasound standing wave is used as the external force to act on the microflows and the phenomena are recorded with CCD camera and microscope. It is hypothesized that “nodal shift” may occur when two fluids are driven in parallel laminar flow into a microchannel. Series of experiments are designed and implemented to verify this hypothesis. It is found that the ultrasound standing wave field distorts the shape and length of the interface between two miscible fluids, which obscures the observation of nodal shift in current experimental setup. However, the distortion phenomenon points to possible utilization of ultrasonic standing wave in micro mixing process. It may be concluded that the acoustic wave accelerates the mixing by increasing the diffusivity between the fluids rather than causing turbulence in the flow domain. This mixing method is important for biomedical devices because it avoids the harsh environment which may cause damage to cells. Due to the high ratio of surface to volume in the above-mentioned microfluidic devices, bubble problems occur almost everywhere and intensively affect the working efficiency. This thesis also covers the research on the bubble reduction mechanisms and their relation to fluid properties such as surface tension and gas concentration. Through experimental observation, analysis and quantification, it is concluded that choosing working fluids with less surface VI tension or degassing the fluid before flowing can greatly alleviate bubble problems. VII List of Figures Figure 1 Silicone-based microchannel array built up as a cytometry to separate WBCs from whole blood dilutions. Figure 2 Classification scheme for micromixers. Figure 3 Schematic drawings of selected passive and active micromixing principles. Figure 4 Schematic of the three-dimensional serpentine, square-wave and straight channel. Figure 5 Normalized average intensity max in each channel 18 mm beyond the T-junction for various Reynolds numbers. Figure 6 Inverted fluorescent views (except (e)) of the mixing process. Figure 7 Mixing effects measured near the outlet Figure 8 Schematic of acoustic microstreaming induced by an air bubble resting on a solid wall. Figure 9 Snapshots showing multi-bubble induced (9 top bubbles) acoustic mixing in a 12×15×0.125 mm chamber at time Figure 10 Schematics of two micromixers working with EKI principle. Figure 11 The variation of normalized bubble position with normalized time. Figure 12 Standing wave and particles concentrated at either the node or antinode. Figure 13 Simulation of a single bubble through a microchannel. Figure 14 The bubble absorption process. Figure 15 Comparison of absorption process with the experimental and simulation results. VIII Figure 16 A principal cross-section drawing of the node and anti-node of ultrasonic standing wave (dashed) formed in the microchannel with the piezoceramic element. Figure 17 Hypothesis of nodal shift when two fluids (A and B) are flowed side by side into the Magic-cross microchannel. Figure 18 Experimental setup of whole system and the Block (at the bottom) which could fill in the piezoceramic. Figure 19 Magic-cross microchannel Figure 20 Setup of the microparts to put into the block slots Figure 21 The trace of particle flowed in distilled water in ultrasonic field using simple device. Figure 22 The trace of several particles flowed in distilled water in ultrasonic field using better coupled device. Figure 23 Frequency range for concentration. Figure 24 The channel after flowed with Melamine particles suspended in water. Figure 25 Deposition of milk lipid particles on the wall of magic-cross microchannel corresponding to the effect of ultrasonic standing wave. Figure 26 Separation of polystyrene particles (the black spheres in the central plane of the channel) and the milk lipid particles (deposit on the channel wall) Figure 27 Simulation results in the magic-cross chip (non-slip boundary conditions). Figure 28 Flow blue and red water from the two entries under ultrasonic standing wave. Figure 29 Trace of the particle flowed in blue & red water. IX Figure 30 Interface between water (dyed in red, enter from top left entry) and satirized salt solution (enter from bottom left entry) Figure 31 Interface between water (up, red) and 99.5% pure ethanol (down, blue) Figure 32 Interface between 50% ethanol solution (up flow, blue) and water (bottom flow, red). Figure 33 The interface when ultrasound is switched on. Figure 34 Closer view of the interface: Figure 35 Polystyrene Particles suspended in water (down, red) flowed with 50% ethanol (up, blue) under ultrasound standing wave. Figure 36 The interface under ultrasound standing wave (freq = 3.88 MHz) when flow milk (upper fluid, appears red color in the photo) and water (lower fluid, dyed with blue). Figure 37 Focusing gradually to the bottom of the microchannel. Figure 38 Moving interfaces between miscible liquids Figure 39 Design of the microfilter for blood cell sorting. Figure 40 Experiment set-up. Figure 41 Scale the photo and measure the displacement for velocity calculation Figure 42 When flowing alcohol (99.5%), the rapid change of an incoming big bubble within 4 sec. in the zigzag filter region. Figure 43 The velocity of the air-liquid interface with different liquid flowing through the filter branch. Figure 44 From the top view of the chip, the surface area between different phases were increased, leading to higher impact of surface tension X Nomenclature BioMEMS Bio-Micro-Electro-Mechanical System CAM Cell Adhesion Molecule CCD Counter Current Distribution CFD Computational Fluid Dynamics CFE Continuous Flow Electrophoresis E.Coli Esherichia coli bacteria EKI Electrokinetic Instability FACS Fluorescence Activated Cell Sorter HCl Hydrochloric acid MEMS Micro-Electro-Mechanical System µFACS microfabricated fluorescence-activated cell sorter PBS Phosphate Buffer Saline PEG poly(ethylene glycol) PZT Piezo Lead Zirconate Titanate RBC Red Blood Cell WBC White Blood Cell XI Chapter 1 1.1 Microfluidic Devices Introduction in Biological Use – An Overview As the manufacturing technologies of silicon microelectronic chip get more and more mature, MEMS (Micro-Electro-Mechanical System) becomes more and more widely used in the field of biotechnology. This trend results in a specific subset named BioMEMS. Generally BioMEMS includes micron-scale sensors, actuators, motors, conjunctive channels and chambers, etc. When these micro-components are integrated into a single chip, “Lab-on-a-chip” is set up, which can complete the whole process of a planned analysis automatically. The applications of these devices are so diverse that they absorbed a lot of researchers from different fields such as diagnostics, medicine, therapeutics, biological research, military and so on. Specifically speaking, it is of great prospect for the chip to be used for precision surgery, rapid and cheap diagnosis of both common and genetic disease, autonomous therapeutic management and early identification of cancer. The advantages of these microdevices lie in low cost, disposability, low weight, and low power consumption. [3~12] When it comes to the design of BioMEMS or Lab-on-a-chip, microfluidics is an inevitable subject to be considered. It is involved in manipulating and analysing fluids in structures on micrometer scale.[13] At this dimension, many forces become dominant other than those experienced in macro devices. Additionally, when the device is used in biological or medical applications, biofluids will be involved, which makes the flow field even more complicated. 1 Therefore, all the factors mentioned above have to be taken into account when a microfluidic device is designed. Specifically, a microfluidic device must contain the fluid to be studied and provide couplings to let the fluid flow into and out of this container. Hence separating and mixing various fluids are essential functions of microfluidic devices. However, because of the high ratio of surface to volume in these devices, bubbles can easily form. Thus careful optimization shall be done to eliminate air bubbles. All the above topics will be further explored in this thesis. 1.2 Thesis Objective A series of experiments are carried out to investigate the flow of complex fluids in microfluidic devices. Ultrasound is used as the external force to excite the flow field so as to seek for its potential use in micro-separating and micromixing. Besides, in order to reduce bubbles in microfluidic devices so as to optimize the micro-flow, the mechanism of air bubble formation in micro-filter is also explored through the experiments. In the study of micro-flows in the ultrasound field, the frequency of ultrasound actuator is tuned to form ultrasonic standing wave between the 2side walls of a rectangular microchannel. The standing wave is used as the external force to act on the microflows and the phenomena are recorded with CCD camera and microscope. It is hypothesized that “nodal shift” may occur when two fluids are driven into a microchannel in the form of parallel laminar flow. A series of experiments are designed and implemented to verify this hypothesis. It is found that the ultrasound standing wave field distorts the shape and length of the interface between the two miscible fluids, which obscures the 2 observation of nodal shift in current experimental setup. However, the distortion phenomenon points to possible utilization of ultrasonic standing wave in micro mixing process. It may be concluded that the acoustic wave accelerates the mixing by increasing the diffusivity between the fluids rather than causing turbulence in the flow domain. This mixing method is important in biomedical devices because it avoids the harsh environment of high shear stresses which may cause damage to cells and macromolecules. Additionally, this thesis also covers the research on the bubble-reduction mechanisms and their relation to fluid properties such as surface tension and gas concentration. Through experimental observation, analysis and quantification, it is concluded that choosing working fluids with less surface tension or degassing the fluid before flowing can greatly alleviate the bubble problem. 1.3 Thesis Organization This thesis is mainly made up of 5 chapters. Chapter 1 gives a brief introduction to this thesis. It comprises the overview of microfluidics as well as the objective and organization of the whole thesis. Chapter 2 is the literature review on general cell sorting & separating and micro-mixing technique. Specific methodologies by using ultrasound to act on microflows are reviewed. Besides, the air bubble problem in the above microfluidic devices as well as its in vivo consequences is also introduced in this chapter. 3 In chapter 3, a series of experimental work are executed based on the effect of ultrasonic standing wave on microflows. This chapter explores the application of ultrasound on micro-separating and micro-mixing process. In chapter 4, mechanism of bubble formation in microflows is presented. At the end, solutions are proposed according to the analysis aforesaid. Chapter 5 draws conclusions from the whole project and recommendations for future work are made accordingly. 4 Chapter 2 Literature Review This chapter reviews the early works on these subjects in order to have a global view on what will happen when certain fluid, especially complex biofluid is flowed into a microfluidic device and how to optimize the device design or flowing conditions so as to achieve higher working efficiency. 2.1 Cell Sorting and Separation Cell separation is to take advantages of multiple physical parameters to separate cells out of a bulk sample and recover their morphology and function afterwards. When more precise cell sub-types are desired, cell sorting is used. It may analyse and select certain types of cells from a whole cell population and enrich them. Modern commercial cell sorters are able to sort cells at speeds exceeding 20,000 per second with purities of over 99%. [2, 8, 9, 14~26] Various cell sorting and separation methods are reviewed below. The pros and cons of these methods are compared at the end of Section 1.2. Based on this background, the proposal of a new separator design using ultrasound standing wave is raised, which is the main idea of this Thesis project. 2.1.1 Counter Current Distribution (CCD) Method The Counter Current Distribution (CCD) separation is carried out in a two-phase aqueous system. Generally speaking, two polymers such as poly(ethylene glycol) (PEG) and dextran, or one polymer and one salt are added into aqueous solution so as to form two phases. The parameters “partition ratio” (defined as mass ratio of the partitioned material in the upper phase to that in 5 the lower phase, P=mu/ml) and “partition coefficient” (defined as the concentration ratio of partitioned material in the upper phase to that in the lower phase, Kpart=Cu/Cl) are commonly used for quantifying the efficiency of separation using CCD method. [21] According to the forces utilized to “pull” cells from the interface to one phase so as to separate them, there are 3 types of partitioning: [27] 1) Charge-sensitive partitioning. When salts are added to create uneven electrostatic potential (∆Ψ) between the two phases, the cells, which are negatively charged under this condition, will move into the positive-charged phase to some extent; whereas others will remain at the interface. For example, when phosphate is introduced, the top phase becomes relatively positive compared to the bottom phase. 2) Non-charge-sensitive partitioning. Some salts don’t bring in ∆Ψ. But when decreasing the concentration of polymer, there will be a critical point for cell to partition. There are many parameters governing this separation and the mechanism is still unclear. NaCl is a commonly used salt in this case. 3) Affinity partitioning. When there is no ∆Ψ plus the critical point is not reached, the ligand between the cells and the polymer or modified polymer plays the dominant role in separation. Early examples utilized hydrophobic affinity of cells to link to PEG, whereas more recently, immunoglobulins are used for highly selective bioseparation. 6 The yield and selectivity of CCD method depends on many parameters such as the choice of polymer, purity of phase forming reagent, centrifuge speed, etc. The cell viability and recovery after separation is also closely related to these factors. P. Eggleton et al [28] reported in their comparative studies on the stimulatory effect of the polymers on neutrophils in whole blood. They found that when pre-treating the dextran with an antibiotic, the stimulation is severely reduced whereas the treatment for PEG has no such effect. They also discovered that long lasting and high speed centrifuge result in many neutrophils lodged between red blood cell aggregates, which then lead to the loss of up to 80% neutrophils. The leftover cells also tend to clump and aggregate. 2.1.2 Continuous Flow Electrophoresis (CFE) Method Based on the different surface charge of cells, the Continuous Flow Electrophoresis (CFE) method allows cell populations to be sub-fractionated. This is originally a method to separate ions contained in a sample continuously injected into a laminar flow of electrolyte. Since the cells are also electriferous, their behaviour in CFE system is similar to the ions thus can be separated accordingly. P. Eggleton et al[28] compared CFE with CCD method to separate white cells. For CFE method, they used NH4Cl lysis to remove red cells followed by differential centrifugation in Hank’s balanced salt solution (HBSS). One advantage of CFE over CCD is that CFE avoid long and fast centrifuge thus keeps better cell yield (CFE 65~89%; CCD 30~40%) and morphological recovery. Additionally, CFE avoid the possibility of cell absorption by polymers. 7 Fiedler S et al[25] invented a different method to improve the cell manipulation and miniaturize the system. A high-frequency AC-field was applied to a dielectrophoresis apparatus, so that when the cells/particles passed through the micro-electrodes housed in microchannels, they could be manipulated by modification of the AC-drives. However, as with FACS (see 1.2.4 below), cells/particles were passed through a nozzle to generate droplets and optical detection was combined to obtain a high-throughput cytometry. 2.1.3 Fluorescence Activated Cell Sorter (FACS) In macro-flow cell sorting, the best performing system available today is the “fluorescence activated cell sorter” (FACS). Invented in the late 60s, FACS instruments represent a powerful way of measuring multi-parametric characteristics of individual particles as they rapidly flow past a laser beam in single file. Light scattered by the passage of a cell through the focused laser beam is measured by precisely positioned photodiodes, revealing information on cell size and structure, and signals from fluorescent markers of the cells are amplified by photomultiplier tubes. Moreover, flow cytometers have the additional capability of extracting defined cell populations, based on the above parameters of size, structure, and fluorescence. After passing through the laser beam, the cell flow is fragmented into droplets with a piezoelectric actuator, which is based on ink jet-printing technology. The flow characteristics (flow rate, cell concentration) ensure that each droplet statistically contains a maximum of one cell. Based on the chosen values for the above parameters, drops are charged and deviated from the mainstream to sort out the drops containing the defined cell population. [29] 8 The miniature system using this method is called the microfabricated fluorescence-activated cell sorter (µFACS), which was reported in 1999 (See Ref. 9). It is a disposable chip based on silicone elastomer. Connected as Tshape, three 100µm wide microchannels each linked to a chamber at one end were mounted together at a 3µm sorting junction. This chip has been used to separate fluorescent-tagged latex beads as well as E.coli bacteria cells and it is capable of two modes of flow algorithm. [22] Fiedler S et al[25] invented a different method to improve the cell manipulation and miniaturize the system. A high-frequency AC-field was applied to a dielectrophoresis apparatus, so that when the cells/particles passed through the micro-electrodes housed in microchannels, they could be manipulated by modification of the AC-drives. However, as with FACS, cells/particles were passed through a nozzle to generate droplets and optical detection was combined to obtain a high-throughput cytometry. 2.1.4 Cell Sorting Applications – Separation of Blood Cells The sorting of White Blood Cells (WBCs) is an application where cell sorting is of great importance. Human blood is a complex biofluid containing three kinds of cell components – (a) Erythrocytes or Red Blood Cells (RBCs); (b) Leukocytes or White Blood Cells (WBCs); (c) Thrombocytes or platelets Erythrocytes make up 95% of all cells within whole blood. RBCs do not contain cell nuclei and thus are not suitable for DNA analysis. WBCs (essential in DNA duplication and analysis) are the only normal blood cells that contain 9 nuclei. Furthermore, WBCs can be subdivided into 5 different types: neutrophils, basophils, eosinophils (generally called granulocytes), lymphocytes and monocytes. Various diseases may cause the changing of ratio and count of the WBC subtypes hence cytometry has long been used as a standard clinical diagnostic means. Carlson RH [2] has built a silicone- based micro channel array to trap the white blood cells while the diluted whole blood (40µl/3~5 drops blood into 400 µl buffer) flowed through (Figure 1). Clear pictures of the cells trapped in the array (stained by either normal or fluorescent dye) were taken to show separation of the WBCs. A mathematical model was then given to explain the activated Figure 1. Silicone-based microchannel array built up as a cytometry to separate WBCs from whole blood dilutions. (B) 2 WBC trapped by the microchannels. (C) a single WBC stained by normal dye collected from the array. [2] adhesion of WBCs in the array. Another technique is to mimic the capillary in vivo with a lattice of channels. When a drop of diluted human blood containing red and white blood cells is forced to move via hydrodynamic forces through this lattice, the white blood cells can self-fractionate into the different types subtypes. The pattern of WBCs that forms is due to a combination of stretch-activated adhesion of cells with the walls, stochastic sticking probabilities, and hetero-avoidance between granulocytes and lymphocytes.[16] Apart from mechanical methods, specific antibodies can be bound to WBCs can also be used to sort cells. A group at the University of California-- 10 Berkeley coated the microchannel surface with E-selectin lgG to separate the leukocytes from erythrocytes [20] . Selectin is a Cell Adhesion Molecule (CAM) protein specific for leukocyte-vascular interactions. The selectin protein binds to the receptors on the surface of the leukocytes and restricts the rolling speed of the WBC on the inner surface of the blood vessel when leukocytes are passed from blood vessels to external tissue (extravasation). Coating a silicone-based microchannel with this protein is a good attempt for cytometry via a biomimetic (mimic of biological characteristics) method. 2.1.5 Summary and Current Work Most of the separation methods mentioned above are based on macro devices and cannot finish separation in line. The disadvantages of these methods are stated below: 1. The methods based on the size and shape of the droplets passing through the device might be affected by the cell size or charge. Failure to ensure the uniformity of these droplets will lead to detection bias. 2. The methods based on centrifugation cannot be integrated thus limit the range and place of application. 3. When these macro devices are operated in successive batches, the backflow of previous sample makes the sterilization of system timeconsuming and lowers the possibility of high-throughput. [29] Hence there are good reasons to pay attention to the design of micronscale devices. Building up cell sorters in BioMEMS (Bio-Micro-ElectronicMechanical System) is an exciting application. [30] An important part of this 11 thesis is making effort on new method to separate cells in silicon-based microfluidic chip and it will be elaborated in Chapter 2. 2.2 Mixing Techniques in Microfluidic Devices Biological processes such as cell activation, enzyme reactions, and protein folding often involve reactions that require mixing of reactants for initiation. While doing biochemistry analysis, drug delivery, and sequencing or synthesis of nucleic acids, it is also essential to complete the mixing within few seconds. Generally speaking, mixing is realized by effects of turbulence and interdiffusion. However, when it comes to microfluidic devices targeted for the above processes, Reynolds number in these systems are usually much lower than 2000 and laminar flow is dominant in the flow fields.[31] Therefore, molecular diffusion often dominates mixing mechanism in these devices. The difficulty lies in inefficient mixing by pure diffusion in laminar bulk flow. When channel cross section is of tens of microns (Re commonly less than 1), mixing via pure diffusion can be completed within a few seconds; while the dimension is increased to several hundreds microns, the time for mixing is increased to tens of seconds. The time is even longer if macromolecules or cells are involved in the diffusion. [32] There are many research paper reporting their designs of micromixer. These designs can be classified into 2 Figure 2 Classification scheme for micromixers. species (Figure 2 & 3): active and 12 passive mixer.[31, 33, 34, 1, 35, 32] According to diffusion principle, the diffusive flux (e.g. of a solute) equals D·A· c, which is diffusion coefficient (D) times the interfacial surface area (A) times the gradient of species concentration ( c). Thus diffusive mixing can be optimized by maximization these above 3 factors. [1] Many designs such as micronozzle arrays concentrate to increase the contact surface (i.e. factor A) of two fluids. Some other designs such as tendril-, whorl- or striation-like shapes attempted to increase both interfacial area and concentration gradient by deforming flows into lamellas. These designs only add complexity without introducing external energy to the system thus are categorized as “passive mixer”.[32] However, there are several problems with passive mixers. Firstly, the mixing time is affected greatly by flow rates and ratio of flowing liquids. Secondly, the pressure drops through the flow path greatly affect the homogeneity of mixing product. Thirdly, these devices are difficult to prime and sensitive to gas bubbles, which is the same as most micropumps.[36] Many focused turbulence researchers on or generating secondary flows in microfluidics by means of introducing timedependent forces into the system. This sort of mixers Figure 3 Schematic drawings of selected passive and active micromixing principles. [1] is defined as “active mixer”. [31] 13 2.2.1 Passive Mixer Passive mixing is based on adding complexity to the system, such as “laminating” the fluid or generating “chaotic advection” in the fluid field. This category of mixer does not introduce external energy except those used for delivering fluid into the system. Hence high complexity of wafer or channel is essential for rapid mixing. In order to be more comparable to the current research project using ultrasonic energy input, more attention is paid to the active Figure 4 Schematic of the threedimensional serpentine (top), square-wave (middle) and straight channel (bottom). “Viewing windows” in the channel are labeled 1–10 for serpentine channel design. mixer and only an example of passive mixer is given below. A review focused mainly on passive micromixers is available by Nguyen N-T. and Wu Z. recently.[35] Liu et al[32] reported a three dimensional design of microchannel to generate secondary flow as well as unsteady flow at the bends of channel so as to raise chaotic advection for mixing purpose (Figure 4). This paper compared mixing process in the 3D Figure 5. Normalized average intensity serpentine microchannel with the I / I max in each channel 18 mm beyond the Tjunction for various Reynolds numbers. square-wave and straight channel 14 in the 6 ≤ Re ≤ 70 range. From the experimental graph it is clearly indicated that mixing in straight channel is realized by pure diffusion and an inverse relationship of mixing and Re is shown because residence time is decreased as Re increases. For square-wave channel, mixing increases dramatically after Re = 70 indicates that unsteady flow is raised after this critical point so the efficiency is enhanced compared to mixing with diffusion mechanism in low Re range. In contrast, 3D serpentine channel is able to raise chaotic advection even when Re ≤ 70. That’s why mixing product in this channel is 16 times more than the straight channel and 1.6 times more than the square-wave channel when Re=70 (Figure 5). The drawback of this serpentine mixer design is the “necking” part when fabricate the turnings. However, these narrow openings are not necessary in design and can be avoided by using different fabrication process (such as deep reactive ion etching (DRIE) instead of KOH anisotropic etching). Another limitation for this design to be used in biomedical field is its high drop in pressure. Since the geometry of the channel gets complicated, the flow resistance is raised, which leads to the increase in driving pressure. This high pressure might be harmful to biological samples. 2.2.2 Active Mixer Active mixer works based on time-pulsing flow due to a periodical change of pumping energy or electrical fields, acoustic fluid shaking, ultrasound, electro-wetting-based droplet shaking, microstirrers and others.[1] It should be tolerant of gas bubbles and the mixing effects should be adjustable by changing the level of input energy.[36] 15 However, there are also disadvantages for active mixers. Firstly the structure of active micromixer is relatively complicated and requires complex fabrication processes because of the components generating external fields. Secondly the external power sources will increase the cost and difficulty to integrate active mixers into systems. energy microfluidic Thirdly, might external raise the temperature of the system or be Figure 6. Inverted fluorescent views (except (e)) of the mixing process. (a) Water & fluorescent dye (uranine) flow at standby state; (b) ultrasound ON; (c) 7 s after ultrasonic irradiation; (d) ultrasound OFF; (e) photograph in bright field. harmful to biological samples thus careful design and test should be made especially when active micromixer is used in BioMEMS.[35] Yang Z. et al[31] reported an active mixer using the force generated by PZT (Figure 6). Ethanol and water were introduced into the mixing chamber which has a dimension of 6×6×0.06 (mm) at the flow rate of 6.9 ml/h and 6ml/h, respectively. The PZT is working under the voltage of Vpeak-peak = 150 V at 48 kHz. Figure 7 Mixing effects measured near the outlet at positions on the line AA’ shown in Fingure 6e before and after ultrasonic radiation. Position 0 is defined as the top of the mixing chamber on line AA’. 16 The video showed red-tinted ethanol and water flowed into the chamber in laminar flow when the PZT is at standing by. After turned on the vibration,, the flow became turbulent immediately and the ethanol-water mixing almost occupied the whole chamber within 2 seconds. When the vibration was turned off, laminar flow resumed. In his later more detailed report, Yang showed quantified results of this mixer using fluorescence. Figure 7 showed the intensity measurement of mixing results along the AA’ line, which is near the outlet in Figure 6 (flow direction from right to left). The qualitative results of proper frequency range for mixing showed that ~15 kHz to 90 kHz mixing ability is roughly stable and effective. The mixed area decreased at 90 kHz and recovered at 130 kHz. Before 8 kHz there was no significant the active mixing observed. There was no linear relation between the excitation frequency and mixing ability. Besides, the input power showed positive relation with the mixed area (the mixing spread the whole mixing chamber is achieved at 90 V) whereas the mixing time was not changed as the voltage increased. The drawback of this work is that the high working voltage and frequency needed to drive PZT. Thus the author suggested the usage of focused ultrasound as well as the combination of this design with the passive mixer so as to decrease the driven voltage and frequency. It also has problem of heat generation but because of the good thermal conductivity of silicon, the mixer maintain the temperature from 28.3 ºC to 31.4 ºC. This temperature might be harmful for some enzyme catalyzed reaction so additional cooling process would be needed. [36] 17 Compared to using PZT as the ultrasound source, Liu R. H. et al used the surface of air bubbles in liquid medium as the vibrating membrane to actuate acoustic field. [37] The bubble actuation is mainly determined by the bubble resonance characteristics and will cause a bulk fluid flow around the air bubble surface which is called “cavitation microstreaming” or “acoustic microstreaming” (shown in Figure Figure 8 Schematic of acoustic microstreaming induced by an air bubble resting on a solid wall. 8). These circulatory flows lead to global convection flows with “tornado”-type pattern which quicken mixing. Since the insonation frequency has to match the bubble resonance frequency which is strongly dependent on the bubble radius, the bubble has to be fixed at a solid boundary. Using the air bubble to generate sonic irradiation may cause considerable gross fluid motion for a solution. Through design of uniform distribution of the air pockets, mixing can be induced in the complete microchamber. Simulation of optimal mixing results was obtained for staggered bubble distribution that minimizes the number of internal flow stagnation regions. Experimental results also showed that the increase of bubble numbers both increased the mixing area and Figure 9 Snapshots showing multi-bubble induced (9 top bubbles) acoustic mixing in a 12×15×0.125 mm chamber at time (a) 0 s; (b) 28 s; (c) 1 min 7 s; (d) 1 min 46 s. decreased the mixing time significantly (Figure 9). In this paper, cell viability after exposed to acoustic field and microstreaming was tested through double 18 staining of green fluorescence & peopidium iodide red fluorescence. Both blood cells and bacteria cells (e.g. Esherichia coli K12) were shown intact after the mixing process. Besides acoustic field, electrokinetic instability can also be used for micromixing. M. H. Oddy et al[38] developed a process to rapidly stir microflow streams by initiating flow instability (Figure 10). In fact, stirring process is a mechanical resulting in a redistribution of material such that the net inter-material area increases. The instability phenomenon occurring in the electro-osmotic channel flows driven by oscillating electric fields, namely “electrokinetic Figure 10 Schematics of two micromixers working with EKI principle. (a) Micromixer I. (b) Micromixer II instability” (EKI) may rapidly stretch and fold the material lines in the fluids thus stir fluid streams. EKI has been visualized in both submicrometer tracer particles and high-concentration fields of injected fluorescent dyes. The driven frequency is relatively low (below ~ 100 Hz) but with the micron-scale channel dimension (above ~ 50 µm), the electric field strengths is above 100 V/mm. Accordingly, the authors designed and fabricate two micromixers using EKI (as shown in Figure 10 (a) & (b)). Through timestopped image, the fluids in Micromixer I became unstable within 2 s after the 19 AC field was turned on leading to rapid deformation of the initial seeded/unseeded fluid interface which in turns quickly stretch and fold material lines in the flow. At time ~ 13.3s, the image clearly showed a random redistribution of the flow tracer transverse to the applied ac field. For Micromixer II, the mixing efficiency is even higher. Within ~ 2.5 s after the application of the ac field, the mixing chamber fluid was already qualitatively well stirred and the output stream of the mixer showed approximately homogeneous fluorescence intensity. The trajectories of 490-nm particles were also found in three-dimensional motion in micromixer II which is not expected in a stable flow field because the electric field is strongly two-dimensional in this uniform depth, fluidic network. The diffusion and stirring of fluoresce observed experimentally under CCD camera were quantified using Ensembleaveraged probability density functions and power spectra of the instantaneous spatial intensity profiles, completing the information given by direct imaging.[38] There are many other kinds of active mixers using principles of periodic flow switching, electrowetting-induced droplet shaking, microimpellers and magneto-hydrodynamics. For more detailed review of these mixers, one may refer to V. Hessel et al (2005)[1]. 2.2.3 Summary and Current Work Whether it is biological processes, medical diagnoses, drug development or chemical analytics, the pivotal step is the reactions of the sample and the reagent. Therefore, rapid and thorough mixing techniques in microfluidic devices are essential for the purpose to miniaturize and integrate the analysis systems in life science and chemistry. Many new design and fabrication 20 techniques have come out in recent years which may be categorized into “passive mixers” and “active mixers”. Section 1.3 has reviewed some of these works. The pros and cons to choose either one of the two kinds of micromixers are also stated. Special attention has been given to active micromixer using ultrasonic fields because the phenomena found in current experiments showed potential usage as a novel acoustic mixer. Notably, Yang Z.’s paper[31] stated that high voltage is difficult to achieve in microdevices thus decrease of mixing chamber dimension is an effective way to decrease the voltage for ultrasound driving. In current work, this method will lead to change of resonance frequency so that the frequency to drive the acoustic actuator is even higher, which will result in another difficulty in the microdevice. Detailed discussion will be given in Chapter 2, Section 2.4. 2.3 Effect of Ultrasound on Flow of Complex Biofluid in Microdevices When the fluid containing compressible particulates such as microbubbles is exposed to ultrasonic field, the acoustic radiation force at the interface of bubble/fluid will make the bubble possessing both translational motions and radial oscillations. Both of these two motions can be described by governing equations. [14, 23, 36, 39~ 49] This section elaborates the theoretical work from three reports so as to give a detailed description of how the ultrasonic standing wave affect the complex biofluids containing cells, labouring particles or micron-scale bubbles, 21 reagent encapsulations in microdevices. These suspended particulates are normally assumed to be spherical for simplicity. The main difference between them is whether the particulate is deformable or not. Different assumptions are made due to different prospects and purpose of calculation so one should carefully choose the algorithm according to the real work in which the analysis are applied. 2.3.1 Radial oscillations of the insonified microbubble Watanabe and Kukita[39] used Rayleigh–Plesset equation as the model to solve the radial motion of spherical gas bubble in an incompressible liquid: 2 d 2 R 3  dR  4µ dR 1  2σ  =  Pg − Pl + Pv − R 2 +   +  2  dt  ρ l R dt ρ l  dt R  (1.1) where R is the bubble radius, Pv is the vapour pressure which is assumed to be constant, µ is the liquid viscosity, ρl is the liquid density and and σ is the surface tension of liquid. Besides the governing equation, there are 3 other equations to represent the variables in the equation: R  Pg = Pg , 0  0  R 3γ Pl = Pl , 0 + ∆P sin (ωt )sin (kx ) Pg ,0 = 2σ + Pl , 0 − Pv R0 (1.2) (1.3) (1.4) where the suffix 0 denotes the quiescent (equilibrium) conditions, γ is the polytropic exponent of the gas, ∆P, ω and k are the amplitude, angular frequency and wave number of the driving pressure, respectively. Notably, the 22 wavelength of the pressure variation is assumed to be large compared with the bubble radius. This model didn’t consider the compressibility of the liquid or the temperature field in the gas phase. However the error of this simplified model is small compared with the detailed model reported by other researchers when the amplitude of driving pressure is relatively small. Hence it is sufficient in many cases. From the calculated results, the radial response is closely related to the ratio of bubble radius R0 to the resonance radius Rres which is obtained by the linear theory for the driving frequency and obeys the following equation: 2 res R 1  2σ 4µ 2    = 2  3γPg ,0 − − 2  Rres ρl Rres ω ρl   (1.5) When the bubble responded at the main, 2nd and 3rd harmonics (i.e. the region around R0 / Rres = 1, 0.5 and 0.33, respectively), there are peaks of resonances and the response curve coincides with the experimental results closely. This is a strong evidence for the capability of this model. The response curve has large difference from the curve taking in account the coupling of translational and radial motions after R0 / Rres > 1 and will be discussed later in Section 1.4.2. Dayton et al[44] used the modified Rayleigh–Plesset equation to describe the radius–time oscillations of insonified microbubbles, which is shown below: 23 γ  3     R0 − R03  b   V   3   2σ 2 χ  R d   m  + p (R , t ) ρl  RR&& + R& 2  =  P0 + +  2   R0 R0  3 c dt  3 b   R (t ) − R0    (1.6) V     m   2 4µR& 2σ 2 χ  R0  R& − − − − (P0 + Pdriv (t ))   − 12µ shε R R R R R(R − ε ) γ  3     R0 − R03  b    V   2σ 2 χ R 2  2σ 2 χ   0  m   p(R, t ) =  P0 + + − −      R0 R0  3 R R R 3 b       R (t ) − R0  V    m   (1.7) This model describes the bubble radius as a function of time in response to the specified driving pressure, which is also as a function of time. It takes in further consideration for the shell properties and acoustic radiation damping, which is different from the Rayleigh–Plesset equation. In Equation (1.6), the first and second terms on the right describe the gas pressure in the bubble and acoustic radiation damping; the third term accounts for the viscosity of the surrounding liquid; the fourth term approximates the surface tension of the bubble; and the fifth term approximates shell elasticity; viscous losses due to the shell are described with the sixth term; and finally the seventh term describes the driving pressure. The above equations were solved numerically using MATLAB with the initial conditions R = R0 and R& = 0 at the time t = 0 . 24 2.3.2 Translation of the insonified microbubble According to the model established by Watanabe and Kukita[39], the translational motions of the microbubble in the acoustic standing wave field can be described using the following equation: mb dub dP 1 d 1 = −V l − ρl (Vur ) − ρ l ur ur ACd dt dx 2 dt 2 (1.8) where mb is the mass of the gas and vapour inside the bubble, ub is the translational velocity, V is the volume of the bubble, ur is the relative velocity between the bubble and the liquid, A is the projected area of the bubble and Cd is the drag coefficient. There are three more equations describing the variables in the governing equation: dx = ub dt (1.9) u r = ub − ul ul = k∆P ωρ l cos(ωt ) cos(kt ) (1.10) (1.11) In order to solve the above equations, there are two prerequisite assumptions stated as (1) the wavelength of the pressure variation is large compared with the bubble radius; (2) the pressure field of the ambient liquid is unaffected by the bubble motion. 25 As shown in Figure 11, there are two special points which is decided by the translational force of the standing wave – node and antinode. When the bubble radius is sufficiently R0 Rres Anti-node x/y R0 Rres = 0.87 = 1.03 smaller than Rres (such as the doted line with R0 / Rres = 0.87), the Node R0 Rres bubble moves to the antinode; whereas when it is = 0.93 larger 1/T Figure 11 The variation of normalized bubble position with normalized time. than or equal to Rres (like the solid curve when R0 / Rres = 1.03), it will concentrate at the nodal plane. The most complicated situation is when R0 / Rres is slightly smaller than 1, the bubble will move to the nodal plane and oscillate chaotically near the node. The author also showed the Poincare map of the bubble position and the power spectra of the translational motion to show how the motion of the microbubble becomes chaotic as R0 / Rres reduces from 1.00 to 0.97. Notably, the chaotic oscillations may amplify the radial response of the bubble in both translational and the radial directions. Through comparing the calculation results before and after the linearization of Equation (1.1), the author concludes that this chaotic radial and translational motion is because of the nonlinearity in Equation (1.1) Through the above analysis one may see that the translational motion of a spherical bubble is affected by the radial oscillation in a way that the latter will affect the radius of the bubble, just as shown in the previous Section 1.4.1. Combining equations (1.1) to (1.4) with the equations (1.8) to (1.11), one could 26 get the coupled motion of the microbubble. The case when bubble radius is slightly smaller than Rres has just been discussed. Remarkably, when the bubble radius is sufficiently small (i.e. when R0 / Rres [...]... when it comes to microfluidic devices targeted for the above processes, Reynolds number in these systems are usually much lower than 2000 and laminar flow is dominant in the flow fields.[31] Therefore, molecular diffusion often dominates mixing mechanism in these devices The difficulty lies in inefficient mixing by pure diffusion in laminar bulk flow When channel cross section is of tens of microns (Re... out to investigate the flow of complex fluids in microfluidic devices Ultrasound is used as the external force to excite the flow field so as to seek for its potential use in micro-separating and micromixing Besides, in order to reduce bubbles in microfluidic devices so as to optimize the micro -flow, the mechanism of air bubble formation in micro-filter is also explored through the experiments In the... after the mixing process Besides acoustic field, electrokinetic instability can also be used for micromixing M H Oddy et al[38] developed a process to rapidly stir microflow streams by initiating flow instability (Figure 10) In fact, stirring process is a mechanical resulting in a redistribution of material such that the net inter-material area increases The instability phenomenon occurring in the electro-osmotic... general cell sorting & separating and micro-mixing technique Specific methodologies by using ultrasound to act on microflows are reviewed Besides, the air bubble problem in the above microfluidic devices as well as its in vivo consequences is also introduced in this chapter 3 In chapter 3, a series of experimental work are executed based on the effect of ultrasonic standing wave on microflows This chapter... Through design of uniform distribution of the air pockets, mixing can be induced in the complete microchamber Simulation of optimal mixing results was obtained for staggered bubble distribution that minimizes the number of internal flow stagnation regions Experimental results also showed that the increase of bubble numbers both increased the mixing area and Figure 9 Snapshots showing multi-bubble induced... Microfluidic Devices Biological processes such as cell activation, enzyme reactions, and protein folding often involve reactions that require mixing of reactants for initiation While doing biochemistry analysis, drug delivery, and sequencing or synthesis of nucleic acids, it is also essential to complete the mixing within few seconds Generally speaking, mixing is realized by effects of turbulence and interdiffusion... loss of up to 80% neutrophils The leftover cells also tend to clump and aggregate 2.1.2 Continuous Flow Electrophoresis (CFE) Method Based on the different surface charge of cells, the Continuous Flow Electrophoresis (CFE) method allows cell populations to be sub-fractionated This is originally a method to separate ions contained in a sample continuously injected into a laminar flow of electrolyte Since... drops through the flow path greatly affect the homogeneity of mixing product Thirdly, these devices are difficult to prime and sensitive to gas bubbles, which is the same as most micropumps.[36] Many focused turbulence researchers on or generating secondary flows in microfluidics by means of introducing timedependent forces into the system This sort of mixers Figure 3 Schematic drawings of selected passive... passive and active micromixing principles [1] is defined as “active mixer” [31] 13 2.2.1 Passive Mixer Passive mixing is based on adding complexity to the system, such as “laminating” the fluid or generating “chaotic advection” in the fluid field This category of mixer does not introduce external energy except those used for delivering fluid into the system Hence high complexity of wafer or channel is... line AA’ shown in Fingure 6e before and after ultrasonic radiation Position 0 is defined as the top of the mixing chamber on line AA’ 16 The video showed red-tinted ethanol and water flowed into the chamber in laminar flow when the PZT is at standing by After turned on the vibration,, the flow became turbulent immediately and the ethanol-water mixing almost occupied the whole chamber within 2 seconds .. .Flow of Complex Biofluids in Microfluidic Devices Zhu Liang (B.Eng, Southeast University) A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE GRADUATE PROGRAMME IN BIOENGINEERING NATIONAL... generating secondary flows in microfluidics by means of introducing timedependent forces into the system This sort of mixers Figure Schematic drawings of selected passive and active micromixing principles... of the flow geometry in microfluidics is of the order of micron, the effect of interfacial tension becomes comparable to the effect of the flow kinematics.[51] Weber (We) number of the flow becomes