<%BANNER%>

Microfabricated Silicon Microchannels for Cell Rheology Study


PAGE 1

MICROFABRICATED SILICON MICRO CHANNELS FOR CELL RHEOLOGY STUDY By KATHRYN ADELE MAIELLARO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Kathryn Adele Maiellaro

PAGE 3

ACKNOWLEDGMENTS I acknowledge my advisor Dr. Roger Tran-Son-Tay for 2 years of guidance. Especially at this years end, I have called upon his help and he has responded with encouragement and support. Though I began graduate school not knowing if research was quite right for me, with his guidance I have ended this experience with re-evaluated goals for a career in academic research. Along the same lines, I thank Dr. Mark Sheplak for reassuring me that I really am a smart person who has something to contribute to the engineering community. I also am greatly indebted to my lab colleagues who also deserve credit for helping complete this research. Melissa and Amrit are my stability and my sanity. I do not know what I would have done without them. Rebecca shared with me her knowledge and experience, for which I am a better researcher. Darren kept me in my place. Cecile, Eddie, Ethan, Brent, and Ben made the lab a great place to work. iii

PAGE 4

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURES..........................................................................................................vii ABSTRACT.......................................................................................................................ix CHAPTER 1 INTRODUCTION........................................................................................................1 1.1 Objective.................................................................................................................1 1.2 Specific Aims..........................................................................................................1 1.3 Significance............................................................................................................1 2 BACKGROUND..........................................................................................................4 2.1 Microfabrication.....................................................................................................4 2.1.1 Current microfabrication processes..............................................................5 2.1.2 Rapid prototyping.......................................................................................11 2.1.2.1 Replica molding...............................................................................12 2.1.2.2 Polymer hot embossing....................................................................12 2.2 BioMEMS.............................................................................................................13 2.2.1 Microchannels............................................................................................14 2.2.2 Current BioMEMS devices using microchannels......................................16 2.2.2.1 Cell sorting and counting.................................................................16 2.2.2.2 Macromolecular assays....................................................................17 2.3 Microfluidics.........................................................................................................18 2.4 Colon Cancer and Focal Adhesion Kinase...........................................................20 3 MATERIALS AND METHODS................................................................................22 3.1 Channel Design and Specifications......................................................................22 3.2 Computer Assisted Drawings...............................................................................25 3.3 Materials...............................................................................................................26 3.4 Process Flow for Microchannel Fabrication.........................................................28 3.5 Chip Cleaning.......................................................................................................33 iv

PAGE 5

3.6 Packaging..............................................................................................................34 3.7 Experimental Set-Up............................................................................................35 4 MICROCHANNEL CHARACTERIZATION...........................................................37 5 RESULTS AND DISCUSSION.................................................................................43 5.1 Microchannel System...........................................................................................43 5.1.1 Fabrication of microchannels.....................................................................43 5.1.2 Flow system................................................................................................49 5.1.3 Imaging.......................................................................................................54 5.2 Improvements to Microchannel System...............................................................54 6 CONCLUSION...........................................................................................................58 6.1 Effectiveness of Microchannel System................................................................58 6.2 Future Work..........................................................................................................59 APPENDIX A PROCESS FLOW......................................................................................................60 B BASE DIMENSIONS................................................................................................61 C MATLAB CODE.......................................................................................................62 LIST OF REFERENCES...................................................................................................63 BIOGRAPHICAL SKETCH.............................................................................................68 v

PAGE 6

LIST OF TABLES Table page 3-1 Summary of channel types and widths....................................................................23 4-1 Constant C1..............................................................................................................38 4-2 Values for dp/dx, Q, P, and R...............................................................................40 4-3 Values for Dh, Re, and Po....................................................................................41 4-4 Le (in m) for 3 m and 70 m wide channels........................................................41 vi

PAGE 7

LIST OF FIGURES Figure page 2-1 Computer generated pattern and resulting mask........................................................6 2-2 Photolithography schematic.......................................................................................7 2-3 Resulting cross-sections.............................................................................................7 2-4 Anisotropic etch profile..............................................................................................9 2-5 Isotropic etch profile..................................................................................................9 2-6 DRIE schematic........................................................................................................11 3-1 Sketch of the channel parts with dimensions...........................................................24 3-2 Coventor images.......................................................................................................26 3-3 Coventor drawings...................................................................................................26 3-4 Reference cross-section at the middle of the channels.............................................28 3-5 Initial wafer processing and photolithography of Channels mask...........................29 3-6 DRIE of channels.....................................................................................................30 3-7 Reference cross-section mid-wafer .........................................................................30 3-8 Wafer processing......................................................................................................30 3-9 Resist bond of support wafer....................................................................................31 3-10 Wafer processing......................................................................................................32 3-11 Thirty-four chips cleaved from each wafer..............................................................32 3-12 Packaging and fluid access of chip..........................................................................34 3-13 Chip package............................................................................................................34 3-14 Data acquisition set-up.............................................................................................36 vii

PAGE 8

4-1 Labeled cross-section of rectangular microchannel.................................................37 5-1 Selected pictures of microchannels..........................................................................44 5-2 Surface profile of the 50 m wide straight microchannels......................................45 5-3 Surface profile of the 12 m wide endothelial cell contoured microchannels.........45 5-4 Surface profile of 100 m wide channel..................................................................46 5-5 Surface profile of the 10 m wide bifurcated channel.............................................46 5-6 Perspective view of the 3 m wide channels...........................................................48 5-7 Perspective view of the 50 m wide channels.........................................................48 5-8 Fluid activation through a 50 m wide channel.......................................................49 5-9 Fluid flow through the 50 m wide bifurcated channel...........................................50 5-10 50 m endothelial cell contoured channel................................................................51 5-11 5 m wide endothelial cell contoured channels with blood flow.............................52 5-12 5 micron wide channel showing fluid traveling over the channel devices...............52 5-13 10 m straight channel with single cell flow...........................................................53 5-14 Input port of 50 m wide flow channel showing a fluid void on far right...............53 5-15 Schematic of improved base....................................................................................56 5-16 Schematic of improved channel parts......................................................................56 viii

PAGE 9

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering MICROFABRICATED SILION MICROCHANNELS FOR CELL RHEOLOGY STUDY By Kathryn Maiellaro August 2003 Chair: Roger Tran-Son-Tay Major Department: Biomedical Engineering The importance of the microcirculation is highlighted by the fact that most of the hydrodynamic resistance of the circulatory system lies in the microvessels; and most of the exchange of nutrients and waste products occurs in the capillaries. The main interests in microcirculation include pressure-flow relationship, flow of blood cells in capillaries, stress distribution in blood cells and vessels, and mass transfer across the capillary walls. Knowledge of the rheological properties of blood cells is also critical in understanding their role in health and diseases because the ability of a cell to flow into microvessels and to migrate through tissues is governed by those properties. There is also an increasing interest in understanding the mechanisms of flow regulation and molecular exchange in the microcirculation of organs and tissues under physiological and pathological conditions. Our objective was to develop sturdy microchannels for studying blood microcirculation and for characterizing cell rheological properties, function, and ix

PAGE 10

behavior. To demonstrate the capabilities of the proposed microfabrication technique, several features of the microcirculation system were produced and incorporated into a microfluidic device into which whole blood was introduced. To study the effect of vessel size on blood flow, microchannels with width ranging from 100 m to 3 m were designed. The largest of these channels mimic arterioles or venules, which have a diameter of about 30 m, while the smallest of these channels mimic capillaries, which have diameters of less than 8 m. Bifurcated microchannels were designed to provide a model for blood flow through a branching point from larger to smaller blood vessels, while the stenosed channels provide a model for flow through an occluded vessel, such as in cases of arteriosclerosis, or calcium or fatty deposits on the blood vessel interior wall. A process flow was developed to etch thirty-four 12 mm x 12 mm microchannel chips into a 100-mm-diameter single-crystal silicon wafer. Each chip is an individual microchannel system, including an input port, microchannel device, and an output port. Etching the channels into rigid silicon offered excellent geometric control to fabricate the network of microchannels for studying microcirculation. Results indicate that the microchannel system could extend deformational information offered by micropipette experiment in two ways. First the system can provide observation of cellular response to the endothelial lining of a vessel. Second, the system allows observation of blood and cell flow through an extensive network of microvessels similar to those found in the microcirculation. x

PAGE 11

1 CHAPTER 1 INTRODUCTION 1.1 Objective The objective of this research is to de velop rigid microchannels for the study of blood microcirculation and the characterization of cell rheological properties, function, and behavior. 1.2 Specific Aims To show the feasibility of using the deve loped procedure to fabricate a network of microchannels for the study of microcircula tion. Flow through a bifurcation and stenosis will be observed. To design a microdevice to characterize the rheological prope rties of individual cells. The flow of a blood cell through a capillary tube will be produced. Capillaries possessing different geometries will be created. To produce a microchannel that mimics th e lining of blood vesse ls. This design would also form a first order model for studying flow through porous media. To observe the effects of surface roughness on the behavior of whole blood and colon cancer cells. 1.3 Significance The significance of the microdevice is that it isolates specific aspects of the microcirculation system so that models of small vessel behavior and function can be obtained. The microcirculation is where oxyge n and nutrients diffuse into tissues and

PAGE 12

2 carbon dioxide and wastes are removed. It is a dynamic fluidic system whose vessels would span over 25, 000 miles if arranged end to end [1]. The specific parts of microcirculation that are isolated are small blood vessel size, blood vessel geometry, and blood vessel interior roughness. The vessel size is studied by creating a microchannel system that includes channels with diameters ranging from 100 m to 3 m. The largest of these channels mimic arteriole or venul e flow that possess a diameter of about 30 m, while the smallest of these channels is used to observe the single-file flow of red blood cells through the capillaries, which have diameter of about 8 m. Bifurcated and stenosed channels will be used to study vessel geometry. Observation of flow through the bifurcated vessels provides a model for blood flow through a branching point from larger to sm aller blood vessels. Observation of flow through the stenosed channels provides a model for arterios clerosis that which includes both calcium (focal calcification) and fatty pl aque (atherosclerosis) deposits on the vessel interior wall. Vessel interior roughness is studied by applying a surface contour to the walls of the channels. This surface pattern is comparable to that of the endothelial lining of the vessels. In addition, this pattern w ill supply a first order model of flow through porous media, which is a future study that will investigate diffusion processes in microcirculation. Saline solution, whole blood, and cells fr om a well defined 293 fetal kidney epithelial cell line will be introduced to the microdevice. Introducing whole blood to the system will provide images of blood flow thr ough the isolated microcirculation parts. Of specific interest will be the deformation of red blood cells and white blood cells through the smallest channels. Cell deformability is important because it affects cell motility,

PAGE 13

3 which then impacts the cells ability to maneuver into the capillary bed. Lastly, introducing the 293 cells into the system w ill provide observation of the effect of a protein called focal adhesion kinase (FAK) on cell deformation. To obtain specific experimental results, the cancer cells will be treated with doxycycline (Clontech, Palo Alto, CA) to observe on the deformation of the 293 cells as they maneuver through capillaries.

PAGE 14

CHAPTER 2 BACKGROUND 2.1 Microfabrication In the 1960s some microelectronics scientists had an idea to put sensors and circuitry on the same chip to reduce device size [2]. Also at this time the mechanical potential of single-crystal silicon for use in batch fabrication of miniaturized structures was established [3]. By the end of the decade silicon was being etched into thin films to utilize its piezoresistivity for pressure measurements. Eventually, mechanical processes were incorporated with the circuitry, creating tiny micron size systems now referred to as microelectromechanical systems, or MEMS. An interesting aspect of MEMS is that knowledge of pure electronics or mechanics is not sufficient to understand their capacity. The interdisciplinary nature of MEMS requires knowledge of a nonexhaustive list of materials, fluids, heat transfer, optics, feedback, statics, and dynamics. With so many systems acting in a tiny device, MEMS bring together researchers of different backgrounds to collaborate on a single project. No longer localized to electronics, MEMS technology now applies to acoustics, optics, electrostatics, mechanics, and, the focus of this research, to biotechnology. The process of making MEMS is called microfabrication. Microfabrication is a general term, which can be defined as the fabrication of miniature devices. It includes silicon wafer etching, film deposition, micromilling, microdrilling, and any other fabrication method that has the word micro in its title. By offering excellent geometric control and features in the submicron range (a feature that will decrease as new 4

PAGE 15

5 fabrication processes develop), microfabrication allows the creation of tools for molecular biology, biochemistry, medicine, and cell biology. Microfabrication not only opens the door for a wider range of cell biology studies, but it also enables cheaper and more efficient studies by offering the following advantages [4]: smaller device size less material use smaller sample size less waste geometrical control a viable environment for biomolecules single cell analysis As knowledge of microfabrication spreads, microfabricated devices will continue to appear in many different research fields. Biomedical engineering, biochemistry, chemistry, embryology, and biology are already benefiting from this growing technology. The result is more biologically significant studies utilized to yield better medical devices and health care in turn. 2.1.1 Current microfabrication processes Microfabrication processes are described in detail in the literature [2, 3, 5], but a brief overview is provided below. The major processes discussed here are photolithography and pattern etching. Photolithography is the process of transferring a pattern onto the substrate that is to be etched. The design or structure that will be etched into the substrate is first generated by computer and then drawn into a hardcopy called

PAGE 16

6 a mask, as shown in Figure 2-1. A mask is a transparent plate, commonly a 5 in x 5 in plate (depending on wafer size) made of soda lime or quartz, onto which the computer generated design is deposited in a thin chrome layer. There are two types of mask generation. Designs that consist of only rectangular features, called Manhattan features, are created into a mask by a process called pattern generation. Designs that include circular or non-Manhattan features must be made into masks by a more expensive electron beam process. Masks are either dark field or clear field. In a dark field mask, the entire field of the mask is covered with chrome while the pattern features remain transparent. Conversely in a clear field mask, chrome is only deposited in the patterned areas while the field of the mask remains chrome-free. A B Figure 2-1. A) Computer generated pattern that is to be transferred to silicon wafer. B) Dark-field chrome mask of the pattern, where the black is the chrome layer. Photolithography (Figure 2-2) includes four steps of photoresist spinning, aligning, exposure, and developing. Photoresist is a photosensitive polymer that acts as a photographic film, allowing the desired pattern to be transferred from the mask to the substrate. It is applied to the substrate in a thin layer, usually by a spin-coating process. There are two types of photoresist--positive and negativeeach with a different chemical

PAGE 17

7 reaction to ultraviolet light. Positive photoresist becomes more soluble when exposed to ultraviolet light, while negative resist cross-links and becomes less soluble in developing solution. If positive resist is selected, the patterned areas will be dissolved during the developing steps, exposing the substrate below, and allowing the pattern to be etched directly into the substrate (Figure 2-3A). However, if negative resist is chosen, the non-patterned areas will be dissolved by the developing agent, and the substrate will be etched down everywhere except the patterned areas. This method will leave the pattern standing above the substrate surface like a relief carving (Figure 2-3B). Figure 2-2. Photolithography schematic. The silicon wafer is coated with a thin layer of positive photoresist. The photomask is aligned over the wafer and the wafer/mask assembly is exposed to ultraviolet light. (A) (B) Figure 2-3. Resulting cross-sections created by A) positive photoresist and B) negative photoresist.

PAGE 18

8 After the photoresist coat, the mask is aligned over the substrate. Ultraviolet light is then directed over the mask/substrate assembly. This step is called exposure, wherein the ultraviolet light passes only through the clear areas of the mask to the substrate below. The final stage in photolithography is developing the photoresist. During this process the substrate is placed in developing solution to remove the soluble areas of photoresist. At this point the choice of photoresist becomes important. After the pattern has been transferred to the photoresist by photolithography, it is etched into the substrate by wet or dry etching. The reader is referred to the literature [2, 6, 7] for discussion of the specific etch systems and the resulting chemical reactions involved. Before the etchant system for a microfabrication process is chosen factors such as the etch rate of the system, material lattice orientation, and the etchant selectivity must be considered. In addition, cost is another significant factor since etching techniques range in price from tens to thousands of dollars. When using single crystal substrates, orientation-dependent etching, called anisotropic etching, often occurs. Alternatively, if the etch is orientation-independent the etch profile is called isotropic. It is therefore important to know the lattice orientation of the substrate involved. A common substrate material is (100) single-crystal silicon. (100) single-crystal silicon is composed of crystals that have the {100} plane at the surface. Etch profiles for anisotropic and isotropic etchants are shown in Figure 2-4 and Figure 2-5, respectively. Figure 2-4 shows the well characterized anisotropic etch created by potassium hydroxide (KOH). The profile makes a 54.7 angle with the horizontal. The non-directionality of the isotropic etch is shown in Figure 2-5. Isotropic etching

PAGE 19

9 often results in undercutting of mask features. Depending on the tolerance of the patterned features, undercutting may or may not be acceptable. Figure 2-4. Anisotropic etch profile. A) Rectangular mask. B) Cross section of silicon wafer. C) Diagonal etch profile. D) Top view of etched feature. Schematic adapted from Petersen [3]. Figure 2-5. Isotropic etch profile. A) Rectangular mask. B) Cross section of silicon wafer. C) Round cross section after isotropic etch. Schematic adapted from Petersen [3]. The etchant selectivity is a ratio that indicates the preference with which the etchant attacks a material. Etchant selectivity, along with etch rates at specified temperatures, are located in a database of etchants [6] so that the appropriate etchant system is selected for microfabrication processes. To define selectivity, consider the wet etchant KOH. Its selectivity to (100)/(111) silicon is 400:1, meaning that for every 400 units of (100) silicon etched, only 1 unit of (111) silicon will be etched [3]. However, though databases for microfabrication processes tabulate general etch rate and selectivity, temperature,

PAGE 20

10 concentration of etchant solution, and additives must also be considered when determining the etch conditions, as these influence the final fabrication results. Wet etching consists of submersing the wafer in a solution or combination of solutions that attack the exposed material. In silicon, etching occurs when silicon-silicon bonds are broken and replaced with a silicon-OH bond [2]. The reaction results in the consumption of water molecules and the formation of Si(OH)4 and hydrogen gas. A common wet etchant used to obtain an isotropic etch profile in silicon is hydrofluoric acid with nitric acid (HF/HNO3). KOH, ethylene diamine pyrcatechol (EDP), and tetramethyl ammonium hydroxide (TmAH) are common etchant systems to anisotropically etch silicon. Dry etchants are gaseous species that remove wafer material. They constantly introduce new species into the etch environment to reduce the chemical side-reactions and allow etching of tight patterns [6]. In addition, dry etching is significantly more expensive, requiring large machinery and trained personnel to operate. Dry etch systems are classified as vapor or plasma-assisted. One vapor etchant used widely in microfabrication is XeF2. It is often used to remove bulk silicon, when rough features are tolerable [5]. Plasma-assisted processes are more useful for obtaining smoother surfaces. Plasma processes introduce ions into the etch environment creating a reaction with the surface to chemically remove material. Surface atoms are then converted to volatile species that are removed by vacuum [2]. Deep vertical trenches or channels are achieved by a plasma process called deep reactive ion etching (DRIE). DRIE is a three step process as shown in Figure 2-6 that involves etching a shallow trench into silicon using SF6 plasma, passivating that newly formed trench with polymer created with the addition

PAGE 21

11 of C4F8 plasma, and then etching a subsequent and deeper trench with SF6 plasma [5]. Passivation with polymer prevents lateral etching of the sidewalls, while the hole becomes deeper. DRIE is often used to etch holes in a wafer or is used when sharp features are required. Figure 2-6. DRIE schematic showing SF6 etching, passivation, followed by SF6 etching. Schematic adapted from [2]. 2.1.2 Rapid prototyping While microfabrication processing offers superior geometric and dimensional control, the high cost and slow turn around time often preclude its use. In light of this difficulty, work to develop microstructure fabrication methods using common lab supplies and less expensive materials began. These fabrication methods are lumped under a term called rapid prototyping. Any discussion of rapid prototyping must first begin with a description of poly(dimethylsiloxane) (PDMS), likely the most widely used material in rapid prototyping. PDMS is an excellent material for rapid prototyping of microdevices for a variety of reasons. It is initially a liquid material that arrives in two components, including a silicone elastomer and a cross-linking agent. Our lab works with Sylgard 184 silicone elastomer kit (Dow Corning Corporation, Midland, MI). The liquid components are combined in a 10:1 elastomer to cross-linker ratio. The liquid is then poured over a surface called the master and left to cure into any desired shape.

PAGE 22

12 PDMS conforms to surface features on the micron scale and when cured releases from the master without damaging its surface. Cured PDMS is optically transparent which makes it suitable for microscopy. In addition, PDMS will cure at room temperature in about 72 hours or can be heated in a 60 oven to decrease the curing time to a few hours [8,9]. 2.1.2.1 Replica molding A common replica molding method uses a high-resolution printer to generate rapid-prototyped photomasks to be used in lieu of expensive soda lime/chrome masks [8,10,11]. With this process, a CAD design is photographically reduced onto 35mm film or microfiche film. This film acts as a mask that is aligned over a photoresist covered substrate and exposed to ultraviolet light. After the resist is developed, this substrate becomes the master for PDMS casting. The photoresist used in this type of rapid prototyping is called SU-8 (MicroChem Inc., Newton, MA). Compared to other photoresists that are spun on a substrate in thicknesses up to 2-3 m, SU-8 can be deposited in thicknesses up to 2 mm. Using this method the minimum feature obtained thus far is 1 m [12]. 2.1.2.2 Polymer hot embossing Another rapid prototyping method is polymer hot embossing. The impetus for polymer molding is that microfabrication does not lend itself to create 3-dimensional structures, at least without complicated fabrication protocols and high expense. Polymer hot embossing takes advantage of the viscoelastic properties of a polymer, often poly(methylmethacrylate) (PMMA), near its glass transition point, Tg, the point at which a polymer changes phase from hard to pliable. In this method a master pattern and polymer substrate are brought into contact, heated, and pressed together. The master

PAGE 23

13 pattern is then transferred, or embossed, into the deformable plastic substrate. After embossing the PMMA sheet is cooled below its Tg and the pattern is permanently formed in the plastic substrate. Hot embossing has shown dimensional fidelity of pillars that are 1 m deep and approximately 15 m in tip diameter [13]. PMMA embossing is used in the fabrication of many protein separation/detection systems, specifically the LabCardTM (ACLARA BioSciences, Mountain View, CA) [14]. One advantage of polymer embossing over PDMS molding is that PMMA is an easily handled and cleaned hard plastic, while PDMS is a flexible, sticky elastomer that collects ambient dust particles during experimentation. PDMS must be handled carefully to minimize fingerprint transfer and dust collection, both of which can interfere with experimental microscopy. Therefore, these permanent patterns in hard plastic are beneficial primarily due to handling ease. In addition, using plastics as alternative to soft elastmomers also has the advantage of minimizing the electroosmotic motility compared to microchannels etched in glass substrates [15]. 2.2 BioMEMS MEMS devices that are designed for biological applications are called BioMEMS. BioMEMS involves any use of microfabrication or rapid prototyping techniques with biotechnological applications. Other word combinations used in conjunction with BioMEMS are bioanalytical devices, microdevices, lab-on-a-chip, microchips, or micro-total analysis systems (TAS), coined by Manz et al. [16]. However, no matter what phrase is used, all of these terms encompass two components-small size and biological application.

PAGE 24

14 Harrison et al. [17] demonstrated the potential of the TAS concept in 1993. An extensive literature survey in Clinical Chemistry has most sufficiently illustrated the massive amount of investigation and research dollars pouring into the BioMEMS area [18]. In this review BioMEMS analytical microchips, yet another combination of words, were characterized into four key parts: Fabrication, integration of parts, and device modeling Microfluidic mechanics, fluid flow, and mixing Applications of microdevices in biotechnology Patents According to the review, as of 2002 there were 255 references on fabrication, integration and modeling, 194 references on flow at low Reynolds numbers, 291 references on BioMEMS applications, and 276 patents in this field. To limit a discussion that could fill many textbooks, only applications of microdevices in biotechnology that rely on microchannels in the areas of cell sorting and macromolecular assays will be reviewed here. However, first the microchannel will be introduced and defined. 2.2.1 Microchannels The microchannel is an integral part of a microdevice. A microchannel may be defined as a via to move cell suspensions or fluids from one part of a microchip to another part. Any port, groove, line, and etched or embossed feature become a microchannel when fluids are introduced. Microchannels often function to mimic capillaries in the micron scale, which allow observation of cells squeezing through the narrow passages of the vascular system. They are utilized for applications from the study of blood flow through vessels to protein analysis.

PAGE 25

15 The first microchannel system, consisting of an isotropically etched channel bonded with a glass cover, was developed by Terry et al. [19]. Kurt Petersen then detailed the utility of silicon as a mechanical material in 1982 [3]. His work reached beyond the microelectronic community, showing researchers in biological fields that microtechnology could benefit their work as well. In the late 1980s and 1990s, research groups were learning the basics of microfabrication and developing simple systems to determine whether the technology suited their needs. In 1989 Kikuchi et al. developed a system of straight anisotropically etched microchannels, which were used to view white blood cells [20-22]. These channels allowed single cell analysis and measurement of transit time, total flow rate of cell suspensions, and cell component aggregation. The group also emphasized the difficulty of introducing a bubble-free fluid solution into the channels, as the fluid set-up often is as difficult a task as fabricating the microdevice itself. In 1995 through 2000, Tracy et al. developed and improved a microchannel system for single cell analysis celled a haemorheometer [23-25]. Their chamber offered a correlation between cell length and cell velocity, and primarily analyzed the mechanical properties of red blood cells. The haemorheometer allowed detection of up to 1500 cells on a cell-by-cell basis. In addition, utilizing their haemorheometer, additional experiments with normal whole blood, chemically altered blood, and thalassaemic blood were performed to investigate the flow of each in terms of cell-membrane mechanical properties and extrinsic property of volume. In 1995 Brody et al. developed the first large microchannel array as part of a doctoral thesis [26]. This system was an attempt to place blood cells in an in vitro system that modeled the physiological environment. The cells flowed through a series of

PAGE 26

16 channels that mimicked a capillary bed. Cells were repeatedly deformed as they entered a microchannel smaller than the cell diameter and exited into a wide space, only to be deformed again in subsequent channels. Finally, in 1997 Ayliffe et al. [27] tested materials other than silicon in a microchannel system. They fabricated channels with glass tops and bottoms and polyimide sidewalls. What was unique about their study was that instead of driving fluid flow using a pressure system, they observed cell motility by chemotaxis. 2.2.2 Current BioMEMS devices using microchannels The ability to microfabricate bioanalytical devices allowed researchers to pin point areas in health care and medicine for this new technology to improve. Two areas of application are cell sorting/separation and macromolecular assays. These areas are addressed in the following sections. 2.2.2.1 Cell sorting and counting Microchannel systems that provide a method of separating and sorting cell solutions into cell types in order to target specific cells are widely researched in BioMEMS. Similarly, methods to separate whole blood into serum and cells to obtain concentrated reservoirs of erythrocytes, leucocytes, and platelets are being developed. Microfabrication allows creation of different geometries of microchannels with which to facilitate cell sorting and separation. To move cells with control and to count populations of cells, many groups have used microfabrication techniques to improve the existing Coulter Counter, developed in the 1950s [28]. The theory behind the counter is to correspond cell size to the voltage change measured as cells move individually through a small aperture. A number of

PAGE 27

17 groups have modified the Coulter Counter by changing the shape of the aperture using a silicon etching protocol [29] or by improving the voltage signal for the sensor [30]. Taking advantage of the impedance change in a circuit produced by cells passing through a small aperture, Gawad et al. [31] developed a novel on-chip flow-cytometer to create a model of an erythrocyte during flow. Their objective was to study the influence of cell size on the channel resistance. The purpose was to obtain a model for normal erythrocyte flow with which to compare other particles. The authors state that this micro-cell sorter may possibly be used to identify abnormal cell types such as cancer cells. A novel cell-sorting device developed in 2002 uses microfabricated gold-coated posts as electrodes to trap individual cells flowing through a microchannel [32]. The traps capitalize on dielectrophoresis to immobilize the cells in potential energy wells. The traps have yet to be fabricated on a large array scale, but the proof-of-concept design proves very effective at separating populations of cells. 2.2.2.2 Macromolecular assays In the late 1990s, after researchers had learned both of the benefits of microfabrication techniques and the basics of microchannel fabrication, they were able to modify the technology to create assays specific to protein analysis. In 2001, Shrewbury et al. [33] used silicon-etched rectangular microchannels to study the effect of flow on macromolecular molecules. Using microchannels they subjected the proteins to elongation flow along the centerline, as well as to shear flow along the channel walls. Microfabrication provided the ability to confine protein solutions in microchannels, subject these proteins to flow, and control their adsorption patterns. Controlling protein adsorption is critical to develop bioanalytical microdevices involved in applications from tissue engineering to the cell sorters discussed previously.

PAGE 28

18 Moreover, the ability to create well-defined patterns of protein adsorption allows selective attachment of cells, which is of utmost importance in the biomaterials field. Initial work in protein patterning was performed by Kleinfeld et al. in 1988 to manipulate the motion of neurons [34]. The group used a method now called photochemical patterning. The technique consisted of photolithographically patterning a substrate, then washing the exposed area with a protein solution which preferentially binds to the substrate and not to the remaining photoresist. The photoresist is then removed revealing a substrate with a defined protein coating. Photolithography techniques were also used to create hydrophilic and hydrophobic areas inside microchannels [35] by patterning the interior of the protein-coated channels. The ultraviolet light actually cleaves exposed protein molecules, changing the proteins from hydrophobic to hydrophilic. When flow was then activated through the channels, the fluid wetted only the patterned hydrophilic areas. Folch et al. also patterned proteins on a substrate using rapid-prototyping techniques with PDMS microchannels [36]. In their method the PDMS film was placed channel side down over a substrate and a protein solution was activated through the channels. Proteins from the solution absorbed to the substrate only. Removal of the PDMS channels left a protein-patterned substrate ready for cell culture, biocompatibility study, or other biological experiment. 2.3 Microfluidics While a system of microchannels can be considered a microdevice, a microdevice with fluid activation is considered a microfluidic device. These microfluidic systems have many advantages over macroscale fluidic systems. Due to their high surface area to volume ratio, they offer high rates of heat and mass transfer, making these systems

PAGE 29

19 suitable as mini-heat exchangers [37] and since their dimensions match those of biological samples microdevices may provide better manipulation and control of sample movement [14,15,17,31,32-34,38]. Microfluidic devices also require a small sample volume, which is favorable since reagents like blood, media, and antibodies are quite expensive. Finally, their portability and rapid analysis performance make microfluidic devices excellent for point-of-care service in clinical settings. However, as device dimensions decrease, the classical laws of fluid mechanics change as well. The critical dimensional parameter of a microfluidic device is the aspect ratio, which is the ratio between the width and height of the channels [39]. As the aspect ratio of microchannels decreases, the laminar friction constant increases [40]. This friction increase can induce phenomena that do not occur in macroscale fluid flow. Some of these phenomena are temperature variations through the system and transport effects in directions other than the fluid flow direction [40]. Many groups have investigated microscale fluid flow to explain microfluidic phenomena [40-44] and modify the classical Navier-Stokes equations accordingly. However, for Newtonian fluids, where the applied shear changes linearly with the fluid velocity, the classical Navier-Stokes equations can be utilized in microfluidic systems for simplicity [26]. In addition to the friction coefficient increase, another parameter that changes with aspect ratio is the Reynolds number, which decreases with decreasing aspect ratio. Re indicates the tendency of a fluid to become turbulent or laminar. It is a ratio of inertial versus viscous forces that act on a fluid. For definition, viscous forces scale in proportion to the surface forces, A in units of L2 (where L stands for length), and inertial forces scale in proportion to the body forces, V in units of L3. In the Reynolds number equation, Re =

PAGE 30

20 vD/, is the fluid density, v is the fluid velocity, D diameter of the microchannel, and the fluid viscosity. This dimensionless value can range as high as 105 in very turbulent fluid flow, such as flow across an airfoil, or as low as 10-5 in laminar fluid flow, as occurs in flow through small blood vessels [45]. In all microfluidic systems with microchannels less than 1 mm in width, Re is less than 1. At this low Reynolds number, as L goes to 0 in Equation 2-1, viscous forces become much larger than inertial forces and the fluid has no tendency to develop turbulence. 32~~~~LVforcesinertialLAforcesviscous (2-1) It is this range, Re<1, that is normally considered in microfluidic research [46]. 2.4 Colon Cancer and Focal Adhesion Kinase According to Cancer Facts and Figures 2003 ( www.cancer.org ) over 1 million Americans will be diagnosed with cancer in 2003. About 150,000 of these cancers will develop in the colon and rectum, as colorectal cancer is the third most common cancer in American men and women. While this type of cancer is treatable if polyps are diagnosed at an early stage, there is only a 9% survival rate that a person with secondary metastases will live an additional 5 years. Therefore, investigation into a potential marker for the beginnings of cancer is important to achieving new cancer therapies. The biology of a cancer cell is different from a normal cell. A healthy cell would normally be autoirradicated after detaching from the extracellular matrix (ECM) during a process called apoptosis [47]. However, a cancerous cell actually escapes this process and continues to mature. Investigating why a cancer cell escapes apoptosis is fundamental in cancer research. A starting point to investigating the functional differences between healthy and cancerous cells is to explore cell anchoring mechanisms,

PAGE 31

21 or focal adhesions. Focal adhesions affect physical attachment to the ECM and vessel linings, as well as regulate cellular functions such as growth, differentiation, migration, and survival [48]. A way to study focal adhesion is to utilize cellular and molecular biology techniques in combination with engineering techniques to correlate the effect of protein expression on the adhesion and motility of the cell. Ultimately, the microchannel system developed here will be used to study such a correlation between protein expression and cell motility. The protein focal adhesion kinase (FAK) and its effect on cell motility will be investigated. FAK is a protein tyrosine kinase that serves in the regulation of the flow of signals from the ECM to the action cytoskeleton. Since it is a regulator of the cytoskeleton, it is assumed that FAK may play a role in cell motility as well [47]. It has been shown that the amount of FAK is elevated, or overexpressed, in many cancers [49]. In the years following, FAK has been shown to be overexpressed in breast, colon, thyroid, prostate, ovarian, and liver cancers, and elevated in human papilloma virus Type 18 in human genital epithelial cells [47]. However, recent results from our lab show that though FAK may be overexpressed in metastatic and invasive tumors, migrating cancer cells, before they attach to form a secondary tumor, actually have decreased levels of FAK present. In other words, while FAK may be over expressed in the invasive tumor, the motile detached cancer cells have reduced expression of FAK. This difference in FAK expression between attached and detached colon cancer cells has prompted studies that investigate the role FAK plays in regulating cell function.

PAGE 32

CHAPTER 3 MATERIALS AND METHODS 3.1 Channel Design and Specifications To develop a microchannel device that benefits our current colon cancer study, as well as future microfluidic projects and computational modeling, a variety of channel designs were developed. These are smooth channels, channels with an endothelial cell contour on the walls, bifurcated channels, and channels with a constriction along their length. The smooth channels have straight and vertical sidewalls. These channels most closely mimic micropipette experiments, with the exception that the cell will travel a much longer length. The application of an endothelial cell contour to the walls of the next set of channels offers a better physiological picture of how cells traverse through a capillary, which is significant since it has been shown that surface patterning and roughness affect cell growth and motion [50,51]. The bifurcated channels allow observation of cell motion when traveling through a branched vessel and the constricted channels provide observation of how cells maneuver through a stenosis or an occluded vessel. After determining the shape of the channels, the dimensions were considered. The minimum width of the channels was limited by the masking making process of microfabrication. In creating the mask, the minimum thickness with which a line of photoresist could be deposited on the soda lime substrate was 3 m. Therefore, 3 m became the minimum channel width for this project. The other channel widths were chosen to accommodate a large range of widths. Small channels that would squeeze both 22

PAGE 33

23 red blood cells (RBCs), which are approximately 8 m in diameter, and white blood cells (WBCs), which range from 10 m in diameter we re designed. Also de signed were larger channels that would mimic ar terioles and venules, which can be as small as 20 m in diameter. The final channel types and widths are tabulated in Table 3-1. Table 3-1. Summary of channel types and widths Each channel is etch ed to 2 depths, 10 m and 25 m. Channel Type Straight Patterned Bifurcated Stenosed 3 3 10 5 10 5 10 5 5 25 12 25 12 25 10 10 50 25 50 25 50 12 12 100 50 100 50 100 15 15 20 20 25 25 30 30 35 35 40 40 50 50 60 60 Channel Width (m) 70 70 The channel length was chosen as 150 m to allow enough length for cell travel and maintain a relatively low channel resistan ce. The final dimension selected was the channel depth. Two depths were chosen to have a shallow depth that would promote the squeezing of RBCs and WBCs, and a larger de pth that would allow bulk flow through the larger channels. The solution was to etch ha lf the wafers to a depth of 10 m and the other half to a depth of 25 m. The next step in designing the channels was specifying the method of fabrication and characteristics of each channel part. Th e microdevice requires fluid access to and from the chip, reservoirs to contain the fluid as it enters and leaves the channels, and the microchannels, also called the channel device. A sketch of the chip is shown in Figure 3-

PAGE 34

24 1. The input and output fluid ports would be etched completely through the wafer thickness creating holes in the chip. These ports are square because a square mask feature is cheaper to create than a circular feature. The ports are 1mm across the diagonal. This diameter was chosen because it matches that of the tubing that will be used to pump fluid to and from the device. The ports are centered arbitrarily inside the reservoirs. The fluid reservoirs are simply a storage area for fluid before it is pushed into the channels and after it leaves the channels. They are the same depth as the channels. In addition, the reservoirs narrow sharply toward the chip center in order to direct the fluid flow into the channels. Figure 3-1. Sketch of the channel parts with dimensions.

PAGE 35

25 The reservoirs were chosen arbitrarily to create a large reservoir into which the fluid can fill. Circular reservoirs were chosen to prevent the fluid from making voids in corners, which would occur in square reservoirs. To complete the design, five channels will be etched on each chip so that the device is useful for multi-cell analysis. Furthermore, it allows the device to still be useful if cells were to lodge in a channel. If one channel becomes blocked, then there will be four more channels through which cells can travel and four more channels that can be observed. In addition, a scale bar was drawn above and below each the channel device. The bar consists of 3 m wide rectangles that are spaced 3 m apart. This allows measurement of cell length and travel distance. 3.2 Computer Assisted Drawings The 2 dimensional layout, or mask, of the channels was generated using a software package called CoventorWareTM (Coventor, Inc., Cary, North Carolina). The software converts the layout into GDSII format, which is a commonly accepted format for mask making. The Coventor images for both masks are shown in Figure 3-2 and the Coventor images of the channel configurations are shown in Figure 3-3.

PAGE 36

26 A B Figure 3-2. Coventor images of the masks used during photolithography. A) Channels mask. B) Ports mask. The images show the 34 devices per wafer and the circle around each mask indicates the silicon wafer. 150 m 150 m 100 m 200 m Figure 3-3. Coventor drawings of the channel configurations. Consider the flow as moving from left to right through the pink areas. A) Smooth channels, B) endothelial cell contoured channels, C) bifurcated channel, D) stenosed channel with recovery reservoir. 3.3 Materials In order to achieve the best dimensional detail the fabrication method chosen was silicon wafer etching. Silicon wafer processing contracted through MEMS Exchange

PAGE 37

27 (www.mems-exchange.org, Reston, VA). MEMS Exchange is a microfabrication network that acts as intermediary between the fabrication provider and the purchaser. The silicon wafers were provided by MEMS Exchange and the masks were fabricated at Photronics, Inc. (Brookfield, CT). The masks and wafers were then shipped to the Solid State Electronics Lab (SSEL) at the University of Michigan in Ann Arbor for processing. The SSEL is a part of the Department of Electrical Engineering and Computer Science. The wafers were 100 mm diameter, double-side polished, (100) n-type silicon wafers. Using double-sided wafers prevented any pattern distortion due to a rough wafer backside, though the potential distortion with this two mask process was minimal regardless. In addition, support wafers were also used to strengthen the wafer during the long DRIE of the fluid ports. These wafers were removed for reuse as test wafers for other projects. The process required two masks named Channels and Ports. Each mask was dark-field and made of soda lime glass with chromium coating. The Channels mask included the 2-D, top-down contour of the reservoirs and channels, while the Ports mask consisted of the 2-D squares comprising the ports. Due to the number of non-Manhattan (non-rectangular) features, the Channels mask was created using an electron beam process, while the Ports mask was fabricated using the standard pattern generation process. The resists used during photolithography were Shipley 1813 (Shipley Company, L.L.C., Marlborough, MA) and AZ P9260 (MicroChemicals, Ulm, Germany). Both are positive resists that become soluble when exposed to ultra violet light and are removed by the developing solution.

PAGE 38

28 The chip packaging was made out of aluminum. Aluminum was chosen because the metal is cheap, mills easily, and is adequately sturdy for this application. 3.4 Process Flow for Microchannel Fabrication Refer to Figure 3-4 through Figure 3below for step-by-step cross-sectional pictures of the wafer. The compact process flow is provided in Appendix A. The cross-section used is indicated by a corresponding figure. For pattern transfer and etching of the Channels mask the cross-section is shown in Figure 3-4 below. Figure 3-4. Reference cross-section at the middle of the channels. The etch process began with a pre-furnace cleaning which included removal of lint particles followed by a dip in hydrofluoric acid (Figure 3-5A). Next the wafers underwent wet oxidation to grow a 0.5 m layer of silicon dioxide (SiO2) (Figure 3-5B), which was followed by spectroscopic ellipsometry film thickness measurement. The SiO2 acts as a nested mask for the DRIE in the proceeding steps. To apply the pattern of the reservoirs and channels to the silicon wafer, contact photolithography of the Channels mask was performed. First the wafers were dehydrated in an oven and coated with a primer called hexamethyldisilazane (HMDS), which improved the adhesion between the SiO2 and photoresist (step not shown in figures). 1.3 m of Shipley 1813 photoresist was spun over the wafers (Figure 3-5C), which were then put in the oven for a softbake. Using an EV420 Series Mask Aligner (Electronic Visions Co., Pheonix, Arizona) the Channels mask was aligned over the front-side of the coated

PAGE 39

29 silicon wafers. The wafers were then exposed to ultra violet light for final pattern transfer (Figure 3-5D). After exposure the wafers were baked at 110C to strengthen the unexposed resist and then placed in developing solution to remove the exposed photoresist. In a etch process called a buffered oxide etch, the wafer was dipped in a buffered hydrofluoric acid (HF) solution, which etched the SiO2 only (Figure 3-5F). The wafers were then placed in an oxygen plasma environment to ash away the exposed photoresist (Figure 3-5F). (A) (D) (B) (E) (C) (F) Figure 3-5. Initial wafer processing and photolithography of Channels mask. A) Clean silicon wafer, B) grow the SiO2, C) spin photoresist, D) expose Channels mask, E) develop, F) buffered HF etch through SiO2 and photoresist removal. After the channels and reservoir pattern was transferred to the wafers by photolithography, the channels and reservoirs were etched into the silicon wafer by DRIE (Figure 3-6). In order to fabricate two sets of chips with two different channel depths, one wafer was placed in the STS Multiplex ICP (Inductively Coupled Plasma) Deep Reactive Ion Etcher (Surface Technology Systems, Redwood City, California) until an etch depth of 10 3 m was reached, while the other remained in the etcher until an etch depth of 25 3 m was reached. (Only one etch depth is illustrated in the figures). To complete the front-side etching of the channels and reservoirs, the remaining layer of photoresist was removed by ashing, while the silicon dioxide layer remained.

PAGE 40

30 Figure 3-6. DRIE of channels. At this point the wafers were flipped over for backside etching of the fluid access ports. For backside wafer processing description, the cross-sectional reference is changed and shown in Figure 3-7. Figure 3-7. Reference cross-section mid-wafer through the fluid access ports. Photolithography of the Ports mask (Figure 3-8A) was the same as the photolithography of the Channels mask, except that AZ P9260 photoresist was used instead of Shipley 1813. AZ P9260 was applied in a 12-m thick layer, which protects the silicon wafer during the lengthy through-wafer DRIE of the ports. After developing, the SiO2 was etched by buffered HF solution leaving the silicon underneath exposed (Figure 3-8B). (A) (B) Figure 3-8. A) Photolithography and B) buffered HF etch of the fluid access ports.

PAGE 41

31 After the Ports pattern was transferred to the backside of the wafers, support wafers were bonded to the front-side of the wafer (the side where the channel features were etched) by applying a thick coating of photoresist and pressing the wafers together (Figure 3-9A). Resist bonding consists of a dehydration bake of the wafers, HDMS prime, Shipley 1827, photoresist coat, wafer attachment, and a hardbake at 110C. Attaching the support wafers provides stability to the silicon wafer during etching and acts as a safeguard against possible wafer breakage. The fluid access ports then underwent DRIE down about 475 m to create holes through the entire thickness of the wafer (Figure 3-9B). device wafer device wafer (A) (B) support wafer support wafer Figure 3-9. A) Resist bond of support wafer to front-side of silicon wafer and B) DRIE of fluid access ports. For final wafer processing, the support wafer and photoresist coating were both removed by exposure to oxygen plasma (Figure 3-10A). The remaining SiO2 was removed by buffered HF solution leaving the finished and processed wafer cross-section (Figure 3-10B). In addition, the finished wafer, using the original wafer cross-section, is shown in Figure 3-10C.

PAGE 42

32 (A) (B) (C) Figure 3-10. Removal of A) support wafer and B) remaining silicon dioxide. C) The finished wafer from the mid-channel cross-section. The last step was to cleave the wafers using a scribe to release the 34 chips per wafer (Figure 3-11). Scribing was performed by nicking the edge of the wafer to create a defect. Care was taken to nick the wafer at a 90 angle to the (100) lattice orientation. After scribing, the wafer was snapped at the defect point to propagate a crack across the wafer and cleave it in two. Due to the lattice orientation, if cleaved properly, the newly formed edges were perpendicular and straight across the wafer. Figure 3-11. Thirty-four chips cleaved from each wafer. The number refers to the channel width in m, E refers to endothelial cell contour, bif refers to bifurcated channel, and sten refers to stenosed channels. The arrows indicate cleavage points.

PAGE 43

33 3.5 Chip Cleaning After cleavage the chips were cleaned in a series of solvents. Each chip was placed into a 15 mm outer diameter glass test tube (Fisher Scientific, Pittsburgh, PA). The tubes were placed in a test tube holder and inserted into an ultrasonic bath (Branson Cleaning Equipment Co., Shelton, CT). The tubes were then filled with subsequent solutions of tetrachloroethylene (J.T. Baker, Phillipsburg, NJ), acetone (Fisher Scientific, Pittsburgh, PA), methonol (Fisher Scientific, Pittsburgh, PA), and deionized water. After 2 mL of each solution was introduced, the ultrasonic bath was turned on for 5 minutes. The solutions work as degreasers to remove oils from the surface of the chips. The ultrasonic bath emits a rapid pulsatile effect to effectively scrub the chips. The final soak in deionized water removed any residual contaminants. After cleaning the chips were placed in labeled chip carriers (Fluoroware, Inc., Chaska, MN) for storage. To study different sealing methods, two sealing methods-pressure and plasma surface modification-were used. For the pressure sealing method, the chip and cover slip were pressed together by the packing tension arms discussed in 3.6. This sealing method is not permanent. In the second sealing method a permanent seal is created. As reviewed in the literature [52], the plasma surface modification causes an energy transfer from high-energy gaseous species to surface molecules. Evacuating a vessel, purging it with argon gas, and energizing the gas with radiofrequency energy creates the high-energy species. After exposure, the silicon chip and PDMS-coated cover slip were then brought into contact and pressed together. The new silicon-PDMS interface forms permanent Si-O-Si bonds which creates a tight, irreversible seal.

PAGE 44

34 3.6 Packaging A rigid assembly shown in Figure 3-12 and Figure 3-13 below was fabricated in order to use the chip for fluid study. The assembly not only stabilized the silicon chip, but also provided the fluid access to the microchannels. The parts of this package were the aluminum base, 2-3/64 x 9/64 Buna-N o-rings (McMaster-Carr Supply Co., Atlanta, GA), the chip itself, a PDMS coated 22 mm x 22 mm x 0.2 mm unbreakable plastic cover slip (Fisher Scientific, Pittsburgh, PA), and two tension arms. Figure 3-12. Packaging and fluid access of chip showing the aluminum base, o-rings and o-ring pockets, chip, cover clip, and tension arms. Figure 3-13. Chip package including base, chip, cover slip, and tension arms. The base is an aluminum rectangle that is 5 cm x 3 cm x 9 mm thick with a 22 mm x 22 mm x 1 mm square hole, called the well, milled into its center. This well is where

PAGE 45

35 the o-rings, chip, and cover slip sit. To create the input and output ports in the base, 1 mm diameter holes were drilled into opposite sides of the base. These holes stop directly under the location where the fluid ports of the chip will lie. Holes 2 mm in diameter were then drilled from the well down to meet the 1 mm diameter holes. A 3 mm diameter hole was then drilled over the 2 mm diameter ports to create pockets where the o-rings will sit. These holes complete the fluid pathway in the base. A dimensional drawing of the base is included in Appendix B. After the o-rings, chip, and cover slip are placed into the well, the two tension arms swings directly over the port holes to engage the o-rings and create a sealed fluid circuit. After the seal is created, 0.62 in inner diameter Silastic laboratory tubing (GlycoTech, Rockville, MD) connected to the fluid ports in the aluminum base by luer locks (ArkPlas Products, Flippin, AR). Such a packaging assembly allows chips to be changed easily with a ready fluid connection. The assembly is then ready to be placed on the microscope stage. 3.7 Experimental Set-Up The data acquisition set-up is shown in Figure 3-14. The fluid is introduced to the chip by a Pump 33 Dual Syringe Pump (Harvard Apparatus Inc, Holliston, Massachusetts), which provides a closed fluid circuit. Using a user defined flow rate in l/ min, one syringe pushes solution through the tubing, while a second syringe pulls the solution.

PAGE 46

36 Figure 3-14. Data acquisition set-up including fluid syringe pump, microchannel chip, microscope, and computer for image acquisition. For preliminary testing the solution used was a 1:12 ratio of whole blood and Hanks Buffered Saline Solution (HBSS) (Fisher Scientific, Pittsburgh, PA). Prior to inducing the blood or saline solution, the tubing, base, and channels were flushed with a 4% poly(vinylalcohol) solution (Sigma Chemical, St. Louis, MO). This surfactant solution primed and wetted the tubing and channels to provide a hydrophilic coating over which the blood solution will flow. Priming the channels will prevent the blood cells from sticking to the inside of the tubing and separating out of solution. After testing, the flow system was sterilized by flushing with a series of diluted bleach solution, deionized water, and, lastly, ethanol. Images are acquired with an AxioPlan2 IE Manual microscope (Carl Zeiss Microimaging, Inc., Thornwood, New York) set-up for incident light transmission. The image acquisition software was AxioVision 3.1. (Carl Zeiss Microimaging, Inc., Thornwood, New York).

PAGE 47

CHAPTER 4 MICROCHANNEL CHARACTERIZATION In order to obtain a first order approximation between pressure drop across the channels and flow rate, the case of the flow through a rectangular channel is used and the following assumptions are made. Newtonian fluid presence of blood cells is neglected Fully developed flow Incompressible fluid No-slip boundary condition 2-D flow Under these conditions a closed form solution to the governing flow equations can be derived [39]. The labeled cross-section is shown in Figure 4-1. aya bzb L = 150 m Figure 4-1. Labeled cross-section of rectangular microchannel. The equation for flow rate, Q, through a rectangular channel is ...5,3,15532/tanh192134iiabibadxdpbaQ (4-1) 37

PAGE 48

38 which can be written as dxdpCQ1 (4-2) which states that flow rate is proportional to pressure drop. Constant C1 is a function of the channel geometry and fluid viscosity. It was solved in MatLab (The MathWorks, Inc., Natick, MA) and are tabulated in Table 4-1. The MatLab program for calculating C1 is provided in Appendix C. For the calculation, viscosity, was 1.2 cp, the viscosity of blood plasma. Table 4-1. Constant C1. The channel width and height is specified. Channel Height (m) 10 25 Width (m) C1 10 -20 (m4/Pas) C1 10-20 (m4/Pas) 3 1.52 4.33 5 5.96 19.0 10 29.3 130 12 41.5 210 15 61.2 367 20 95.3 716 25 13.0 114 30 165 1620 35 200 2130 40 235 2650 50 306 3720 60 378 4810 70 451 5900 Next the pressure gradient, dxdp can be calculated using Uavg. By definition AQUavg (4-3) Plugging (4-2) into (4-3), the expression for dxdp is obtained.

PAGE 49

39 1CwhUdxdpavg (4-4) where w = channel with and h = height. From the rectangular channel geometry and assuming Uavg = 25 m/sec, dxdp can be calculated. This average velocity is chosen because it allows a cell to reside in the 150 m long channel for 6 seconds, which is an adequate residence time to observe and record the cell motion. Assuming that dxdp is constant across the channel length, pressure drop, P, which is the driving force for fluid flow can be calculated. As shown in Equation 4-5, P is the product of the pressure gradient and the total channel length, L (L = 150 m). LdxdpP (4-5) Furthermore, channel resistance, R, is the driving force divided by the flow rate as shown in Equation 4-6. QPR (4-6) The values of pressure gradient, flow rate, pressure drop, and channel resistance are provided in Table 4-2 for the two channel heights.

PAGE 50

40 Table 4-2. Values for dxdp Q, P and R. Channel Height (m) 10 25 Width (m) dxdp (Pa/m) Q (m3/s) P (Pa) R (Pas/m3) dxdp (Pa/m) Q (m3/s) P (Pa) R (Pas/m3) 3 -4.95x104 7.50x10-16 -7.42 -9.89x1015 -4.33x104 1.88x10-15 -6.49 -3.46x1015 5 -2.10x104 1.25x10-15 -3.15 -2.52x1015 -1.64 x104 3.13 x10-15 -2.47 -7.89x1014 10 -8.53x103 2.50x10-15 -1.28 -5.12x1014 -4.81 x103 6.25 x10-15 -0.722 -1.15 x1014 12 -7.22x103 3.00x10-15 -1.08 -3.61x1014 -3.58 x103 7.50 x10-15 -0.537 -7.16 x1013 15 -6.13x103 3.75x10-15 -.919 -2.45x1014 -2.56 x103 9.38 x10-15 -0.383 -4.09 x1013 20 -5.24x103 5.00x10-15 -.787 -1.57x1014 -1.75 x103 1.25 x10-14 -0.262 -2.10 x1013 25 -4.81x103 6.25x10-15 -.721 -1.15x1014 -1.37 x103 1.56 x10-14 -0.205 -1.31 x1013 30 -4.55x103 7.50x10-15 -.682 -9.10x1013 -1.16 x103 1.88 x10-14 -0.173 -9.24 x1012 35 -4.38x103 8.75x10-15 -.657 -7.51x1013 -1.03 x103 2.19 x10-14 -0.154 -7.04 x1012 40 -4.26x103 1.00x10-14 -.638 -6.38x1013 -9.42 x102 2.50 x10-14 -0.141 -5.65 x1012 50 -4.09x103 1.25x10-14 -.613 -4.90x1013 -8.39 x102 3.13 x10-14 -0.126 -4.03 x1012 60 -3.97x103 1.50x10-14 -.596 -3.97x1013 -7.80 x102 3.75 x10-14 -0.117 -3.12 x1012 70 -3.88x103 1.75x10-14 -.582 -3.33x1013 -7.42 x102 4.38 x10-14 -0.111 -2.54 x1012 In addition to calculating the pressure drop and resistance across the channels, we can also solve for the Reynolds Number, Re, based on hydraulic diameter, Dh, of the channels. ReDh was discussed in 2.3. It is the ratio between the inertial and viscous forces affecting the fluid flow. Dh is a factor used to approximate a non-circular channel to a circular channel. PADh4 (4-7) and havgDDUhRe (4-8)

PAGE 51

41 In Equation 4-7, A is the cross-sectional area of the channel and P is the wetted perimeter. In Equation 4-8, is the fluid density and is the fluid viscosity. The values for Dh and ReDh are listed in Table 4-3. Table 4-3. Values for wall shear stress, hydraulic diameter, Reynolds number, and Poiseulle number. Height (m) 10 25 Width (m) Dh (m) 10-4 ReDh Dh (m) 10-4 ReDh 3 4.62 1.15 5.36 1.15 5 6.67 1.43 8.33 1.79 10 10.0 2.15 14.3 3.07 12 10.9 2.34 16.2 3.48 15 12.0 2.58 18.8 4.02 20 13.3 2.86 22.2 4.77 25 14.3 3.07 25.0 5.36 30 15.0 3.22 27.3 5.85 35 15.6 3.34 29.2 6.26 40 16.0 3.43 30.8 6.60 50 16.7 3.58 33.3 7.15 60 17.1 3.68 35.3 7.57 70 17.5 3.76 36.8 7.91 Now that RDh has been calculated for each channel we can calculate the entrance length, Le, for the channels to determine the fully developed region of fluid flow [53]. 30eeRdL (4-9) where d is the channel depth. The entrance results using only the 3 m and 70 m wide channels are shown in Table 4-4. RDh values are obtained from Table 4-3. Table 4-4. Le (in m) for 3 m and 70 m wide channels. Channel depth (m) 10 25 3 3.83x10-5 9.58x10-5 Channel width (m) 70 1.25x10-4 6.59x10-4

PAGE 52

42 As expected for systems with very low Reynolds number, the entrance lengths for the channels are so small that they can be neglected. The flow becomes fully developed immediately as fluid enters the channels.

PAGE 53

CHAPTER 5 RESULTS AND DISCUSSION 5.1 Microchannel System The microchannel system developed satisfies the objective to develop rigid microchannels for the study of microcirculation and the characterization of cell rheological properties, function, and behavior. It has been shown that microfabrication techniques provide the geometric control and minimum dimensions required to create a microfluidic network. The microchannel system isolated the extensive microcirculation into three parts to study the effect of vessel size, vessel geometry, and vessel surface roughness on blood flow. To isolate vessel size, straight channels of various widths were designed. To isolate vessel geometry, bifurcated and stenosed channels were designed. And to isolate vessel surface roughness, another set of channels was designed with a bumpy contour modeled after endothelial cells applied to the channel walls. This chapter discusses the function of the current microchannel system. The strengths and weaknesses of the system are discussed and suggested improvements to a future microchannel system are provided. 5.1.1 Fabrication of microchannels Fabrication of the microchannels was successfully performed at the Solid State Electronics Laboratory at the University of Michigan. Pictures of the four types of channels are shown in Figure 5-1. These 2-dimensional pictures show that the channel configurations were etched sharply. 43

PAGE 54

44 (A) 20x (B) 40x Scale b ar stenosis (C) 10x (D) 10x Figure 5-1. Selected pictures of microchannels. A) 25 m wide straight channels. B) 12 m wide endothelial cell contoured channels. C) 25 m wide bifurcated channel. D) 25 m wide channel that constricts into a 12 m wide channel. The scale bar, however, did not etch with good integrity. The rectangles were designed to be 3 m wide, but were etched laterally to a width of about 5 m. They offer a rough estimate of cell length and travel distance, but not as accurate a scale as their design intended. While 2-dimensional channel images were obtained with incident light microscopy, 3-dimensional images were obtained with a Wyko NT1000 Profilometer (Veeco Metrology Group, Tucson, AZ). Surface topography images of the channels types are shown in Figure 5-2 through Figure 5-7. The scale to the right of each image correlates the depth variation with color change.

PAGE 55

45 Figure 5-2. Surface profile of the 50 m wide straight microchannels. The image shows a clean etch down to the channel floor. Figure 5-3. Surface profile of the 12 m wide endothelial cell contoured microchannels. The orange color at the top of the figure shows that the surface was etched unevenly. The top of these channels slope down approximately 4 m from left to right across the surface.

PAGE 56

46 Figure 5-4. Surface profile of 100 m wide channel that narrows into a 50 m wide stenosis. Figure 5-5. Surface profile of the 10 m wide bifurcated channel.

PAGE 57

47 The dimensions obtained from the surface profiles show good agreement with the design dimensions with the exception that the 10 m deep channels were over-etched to a depth of 16 m. In addition, the reservoir area was etched slightly deeper than the channels, due to the greater surface area exposed to the etchant system. Figure 5-5 shows the limitation of the profilometer, in that depth measurement of tight spaces can be difficult to obtain for reasons such as scanning too large a surface area or inadequate vibrational isolation. However, the feature that is best illustrated by the profilometer images is the profile of the channel walls. As shown in Figure 5-6 and Figure 5-7, the 10 m wide walls that separate each channel are not vertical, but are thicker at the base than at the terrace. The perspective views of the 3 m and 70 m wide channels show that the angled walls form channels that are not rectangular, but are actually slightly v-shaped. In addition, the wall terrace is not straight and smooth across the top, but is rough and, in addition, is etched approximately 1 m to 2 m below the surface of the chip. This roughness may affect the sealing of the channels prior to and during flow activation by preventing good contact between the PDMS-coated cover slip and the surface of the channels. Channel sealing is discussed in 5.1.2.

PAGE 58

48 top surface of chip terrace channel Figure 5-6. Perspective view of the 3 m wide channels. top surface of chip terrace channel Figure 5-7. Perspective view of the 50 m wide channels. It is noted, however, that the apparent surface roughness of the microchannels could be a artifact of vibrations during the optical surface scan of the profilometer. A scanning electron microscopy micrograph would provide a better picture of the true microchannel chip surface.

PAGE 59

49 5.1.2 Flow system Fluid flow was successfully activated through the microchannel system. The channels used to test the flow system were 3 large width and 3 small width channels. These were the 70 m, 50 m, & 50 m patterned channels and the 10 m, 5 m, & 5 m patterned channels. In addition, flow was activated through the stenosed and bifurcated channels as well as shown in Figure 5-8 and Figure 5-9. Figure 5-9 is included to show a non-wetted area of the channels. The channels are primed with poly(vinyl alcohol) solution, which coats the walls with a hydrophilic layer over which the solution preferentially flows. The system was tested at various flow rates to determine when leakage would occur. Leakage occurred at either the o-ring seal underneath the chip or between the chip and the cover slip as the fluid filled the input reservoir. 20x Figure 5-8. Fluid activation through a 50 m wide channel that constricts to a 25 m wide stenosis.

PAGE 60

50 20x Figure 5-9. Fluid flow through the 50 m wide bifurcated channel showing a non-wetted area. As expected, flow was activated easily through the larger channels. Figure 5-10A and Figure 5-10B below shows blood flowing through the 50 m wide channels at two magnifications. For these channels, the tension arms on the packaging provided enough pressure to engage the o-rings and create a no-leak seal up to the maximum flow rate of 125 l/min. flow direction (A) 10x

PAGE 61

51 flow direction (B) 20x Figure 5-10. 50 m endothelial cell contoured channel at A) 10x and B) 20x. While pressure sealing using the tension arms was effective for the larger width channels, effective pressure sealing was not effective with the smaller channels. When flow was activated through the pressure sealed 5 m patterned channels using a flow rate of 15 l/min, the system leaked slowly over time at both leakage sites. An image of these 5 m patterned wide channels is shown in Figure 5-11. The image shows the blood cells forming blood clots inside the channels, as indicated by the arrows. These clots most likely caused the fluid to backup and leak at both the input port o-ring and in the input reservoir between the chip and cover slip. However, though the sealing of this small channel was ineffective, it shows that indeed cell deformation can be observe with the microchannel system, which partially satisfies the objective of this system. Sealing the 5 m wide channels under argon plasma did not provide a more effective seal than the pressure seal. Figure 5-12 shows an image of the plasma treated 5 m wide channels. The arrows indicate areas where fluid is actually traveling over the channels, not through the channels. So in effect, this system did not leak at the o-rings or

PAGE 62

52 in the reservoir, but rather it leaked at the seal above the channels themselves. This leakage indicates that the 10 m wide walls separating each channel did not provide enough surface area for the plasma treatment to oxidize the surface groups and create a tight bond with the PDMS-coated cover slip. blood clots 20x Figure 5-11. 5 m wide endothelial cell contoured channels with blood flow. Fluid flowing over the channels instead of through the channels 20x Figure 5-12. 5 micron wide channel showing fluid traveling over the channel devices. In addition, the PDMS-coated cover slip does not provide a good seal over the 10 m wide channels either. The picture in Figure 5-13 shows single cell flow through the 10 m wide channels. This channel was sealed with pressure by the tension arms. This image was obtained after the fluid system leaked and the cover slip was removed.

PAGE 63

53 flow direction 20x Figure 5-13. 10 m straight channel with single cell flow. Finally, images of the flow channel showed that even though rounded fluid reservoirs were chosen, voids still occurred (Figure 5-14). Though voids did not hinder the function of the flow channels, modification of the channel design, as discussed in 5.2, would prevent their occurrence. fluid void 5x Figure 5-14. Input port of 50 m wide flow channel showing a fluid void on far right.

PAGE 64

54 5.1.3 Imaging The maximum magnification obtained was 20x. The 40x objective was not used because its working distance was not large enough to focus on the channels through the PDMS-coated cover slip. 5.2 Improvements to Microchannel System There are a few opportunities for improvement of this microchannel system in the areas of sealing and imaging. Sealing problems encountered in the current design could possibly be prevented in three ways. First, the silicon dioxide film thickness should be increased from 0.1 m to 0.5 m. This increase in thickness should better protect the microchannel features from over etching (Figure 5-6 and Figure 5-7). Second, the wall separating each channel should be at least 20 m wide, instead of 10 m wide. This increased separation distance will provide greater surface area with which PDMS-coated cover slip can bond under pressure or after plasma treatment. Third, a polysilicon RIE should be considered as an alternative to DRIE. Polysilicon RIE is a slower process that would etch sharper features, compared to DRIE, which is a faster etching system that may have caused the jagged channel tops and angled side-walls. An alternative bonding method is anodic bonding. Anodic bonding consists of placing a 500 m Pyrex 7740 glass wafer (Corning, Corning, New York) over the silicon wafer and applying a voltage of 300-700 V at 500C across the assembly [2]. The positive voltage causes sodium ions in the Pyrex glass to migrate away from the glass/silicon interface leaving a net negative charge on the glass surface. The opposing positively charged silicon surface is strongly attracted to the positive glass and the two surfaces chemically fuse together irreversibly. In addition to providing a better sealing

PAGE 65

55 method, anodic boding would also provide clearer pictures due to the optical quality of the glass. Another design modification to improve sealing is to replace the tension bars with a plate and four bolts as shown in Figure 5-15. The plate would provide equal pressure on all sides of the microchannel chip and tightening with a torque wrench would offer repeatable pressure acquisition. To preclude the formation of fluid voids, the size of the reservoirs should be minimized so that the reservoir wall makes a small rim around the port opening. The improved channel schematic is shown in Figure 5-16. In addition, the size of the chip could also be enlarged to match the size of the cover slip. This will help in aligning the chip to the part holes. Enlarging the chip, however, also means that fewer chips will be diced out of one wafer. In addition, the chips should be aligned in a square of straight columns and rows so that the dicing saw can be used to sacrifice the chips from the wafer. The dicing saw would cut perfectly square chips, which would make aligning the chips to the port holes in the fluid system even easier.

PAGE 66

56 Figure 5-15. Schematic of improved base with plate instead of tension bars. Figure 5-16. Schematic of improved channel parts with no reservoirs. Finally, the success is partly determined by the quality of the images obtained. Currently, the only objective utilized was the 20x objective. However, imaging the

PAGE 67

57 system using a LD ACHROPLAN 40x objective with a 0.6 mm working distance (Carl Zeiss Microimaging, Inc., Thornwood, New York), would focus through the PDMS-coated cover slip and provide more detailed pictures of fluid flow and cell deformation. This 40x, or even a 60x long working distance objective, is crucial to improving image collection and to creating a successful microchannel system.

PAGE 68

CHAPTER 6 CONCLUSION 6.1 Effectiveness of Microchannel System Rigid microchannels for the study of blood microcirculation and the characterization of cell rheological properties, function, and behavior were successfully designed and fabricated. Using microfabrication techniques, sufficient dimensional and geometric control was achieved. The silicon etching process developed allowed fabrication of a network of microchannels for the study of microcirculation. Using this process, microchannels that provide an in vitro method to isolate the varying vessel size, geometries, and surface roughness comprising the microvascular network were fabricated. Bifurcated, stenosed, smooth, and endothelial cell patterned channels were created, each mimicking a facet of the microcirculation. The endothelial cell patterned channel also provides a means to subject viscoelastic fluids to compression and relaxation. The microchannel fluid system was shown to be successful as well. Imaging showed that blood cell deformation was observed using the microchannel system. These images extend deformational information offered by micropipette experiment, in that they provide observation of cellular response to the endothelial lining of a vessel and, furthermore, provide information on cell migration through branching or occluded vessels. 58

PAGE 69

59 6.2 Future Work Immediate work will begin to improve the sealing and imaging capabilities of the microchannel system. A 40x long distance working objective will be obtained to acquire images of higher power and provide better observation of deformation and motility. In addition, Pyrex glass cover slips should be anodically bonded to the top of the microchannel chips to create permanent and irreversible seals. Tight sealing may allow larger flow rates that will prevent cells from clogging the channels. Moreover, obtaining higher flow rates through the channels would provide more physiologically relevant results. In addition, the microchannel system developed will be used to study cell deformation and motility as affected by protein expression in the growth and spread of colon cancer. Cells from a well-defined 293 fetal kidney epithelial cell line, the control cell line for the current colon cancer study, will be introduced into the system to study the role that focal adhesion kinase (FAK) plays in cell deformation and motility. The system can also be used to begin a study on the flow through porous media. The endothelial cell lined channel can be used to observe the compression and relaxation of viscoelastic fluids. This compression and relaxation provides a first order model for the type of flow created in foam or a porous system. Compression/relaxation models are used in the development of absorbent materials, such as diapers and paper towels.

PAGE 70

APPENDIX A PROCESS FLOW Step Description Starting material 100mm diameter double-side polish (100) n-type silicon 1. Clean Standard RCA clean with HF dip 2. Oxide Wet TCA oxidation 0.1 m SiO2 3. Photolithography Dehydration bake HDMS prime Photoresist coat 1.3 m (Shipley 1813) Photoresist software Contact front-front align Mask 1 I-line exposure Photoresist develop Photoresist hardbake (110 C) 4. Buffered oxide etch Etch Mask 1 features into oxide (0.1 m) 5. DRIE Etch Mask 1 features into silicon (10 m and 25 m) 6. Ash Oxygen plasma to remove photoresist 7. Photolithography Dehydration bake HDMS prime Photoresist coat 12 m (AZ P9260) Photoresist softbake Contact front-front align Mask 2 I-line exposure Photoresist develop Photoresist hardbake (110 C) 8. Buffered Oxide Etch Etch Ports features into oxide to create alignment marks for backside through-wafer etch of Ports 9. Resist Bond Dehydration bake HDMS prime Photoresist coat (Spoke patterb Shipley 1827) Resist bond Photoresist hardbake (110 C) 10. DRIE Through-device-wafer etch of ports 11. De-mount support wafer Oxygen plasma to remove photoresist 12. Ash Oxygen plasma to remove remaining PR 13. Bufferedn Oxide Etch Remove remaining oxide 60

PAGE 71

APPENDIX B BASE DIMENSIONS 61

PAGE 72

APPENDIX C MATLAB CODE %SOLVE FOR c1/viscosity for 25 micron deep channel% C1=[0 0 0 0 0 0 0 0 0 0 0 0 0]' a=[1.5 2.5 5 6 7.5 10 12.5 15 17.5 20 25 30 35]; b=12.5; for k=1:1:13 sum=0.0 for i=1:2:5 res=(tanh(i*pi*b/2/a(k))) /powe r(i,5); sum=sum+res; end C1(k)=4*b*a(k)^3/3* (1-192*a(k)/pi^5/b*sum) end %SOLVE FOR c2/viscosity for 25 micron deep channel% C2=[0 0 0 0 0 0 0 0 0 0 0 0 0]' a=[1.5 2.5 5 6 12.5 15 17.5 20 25 30 35] y=2*a b=12.5 z=2*b for k=1:1:13 sum=0.0 for i=1:2:5 res=(-1)^((i-1)/2)*(1-(cosh(i*pi*z/2/a(k)))/(cosh(i*pi*b/2/a(k))))*cos(i*pi*y(k)/2/a(k))/i^3 sum=sum + res end C2(k)=16*a(k)^2/pi^3*sum end 62

PAGE 73

LIST OF REFERENCES 1. Martini, F.H., Fundamentals of Anatomy and Physiology. 2001, Upper Saddle River, New Jersey: Prentice-Hall, Inc. 2. Senturia, S.D., Microsystem Design. 2001, Norwell, Massachusetts: Kluwer Academic Publisher. 3. Petersen, K.E., Silicon as a Mechanical Material. Proceedings of the IEEE, 1982. 70(5): p. 420-457. 4. Voldman, J., Gray, M.L., Schimdt, M.A., Microfabrication in Biology and Medicine. Annual Review of Biomedical Engineering, 1999. 1: p. 401-425. 5. Kovacs, G.T.A., Maluf, N.I., and Petersen, KE., Bulk Micromachining of Silicon. Proceedings of the IEEE, 1998. 86(8): p.1536-1551. 6. Williams, K.R., and Muller, R.S., Etch Rates for Micromachining Processing. Journal of Microelectricalmechanical Systems, 1996. 5: p. 256-269. 7. El-Kareh, B., Fundamentals of Semiconductor Processing Technologies. 1995, Norwell, Massachusetts: Kluwer Academic Publisher. 8. McDonald, J.C., Duffy, D.C., Anderson, J.R., Chiu, D.T., Wu, H., Schueller, O.J.A., and Whitesides, G.M., Fabrication of Microfluidic Systems in Poly(dimethylsiloxane). Electrophoresis, 2000. 21: p. 27-40. 9. Branham, M.L., Tran-Son-Tay, R., Schoonover, C., Davis, P.S., Allen, S.D., Shyy, W., Rapid Prototyping of Micropatterned Substrates using Conventional Laser Printer. Journal of Material Research, 2002. 17(7): p. 1559-1562. 10. Deng T., Wu, H., Brittain, S., and Whitesides, G., Prototyping of Masks, Masters, and Stamps/Molds for Soft Lithography Using an Office Printer and Photographic Reduction. Analytical Chemistry, 2000. 72: p. 3176-3180. 11. Kenis, P.J.A., Ismagilov, R.F., Takayama, S., Whitesides, G.M., Li, S., White, H.S., Fabrication InsideMicrochannels Using Fluid Flow. Account of Chemical Research, 2000. 33: pg. 841-847. 63

PAGE 74

64 12. Love, J.C., Wolfe, D.B., Jacobs, H.O., Whitesides, G.M., Microscope Projection Photolithography for Rapid-Prototyping of Masters with Micro-scale Features for use in Soft Lithography. Langmuir, 2001. 17: p. 6005-6012. 13. Schmitz, T., Dagata, J., Dutterer, B., and Sawyer, W., Nanolithography and Microfluidics: A Manufacturing Perspective, 2003. Submitted to the Journal of Manufacturing Processes. 14. Tan, W., Fan, Z.H., Qiu, C.X., Ricco, A.J., Gibbons, I., Miniaturized Electric Capillary Isoelectric focusing in Plastic Microfluidic Device. Electrophoresis, 2002. 23: p. 3638-3645. 15. Herr, A.E., Molho, J.I., Drouvalakis, K.A., Mikkelsen, J.C., Utz, P.J., Santiago, J.G., Kenny, T.W., On-chip Coupling of Isoelectric Focusing and Free Solution Electrophoresis for Multidimensional Separations. Analytical Chemistry, 2003. 75: p. 1180-1187. 16. Manz, A., Graber, N., Widmer, H.M., Miniaturized Total Chemical Analysis Systems: A Novel Concept for Chemical Sensing. Sensors and Actuators, 1990. B1(1-6): p. 244-248. 17. Harrison, D.J., Fluri K., Seiler, K., Fan, Z., effenhauser, C.S., Manz, A., Micromachining a Miniaturized Capillary Electrophoresis-based Chemical Analysis System on a Chip, Science, 1993. 261: p. 895-897. 18. Kricka, L.J., and Fortina, P., Microchips: An All Language Literature Survey including Books and Patents. Clinical Chemistry, 2002. 48(9): p.1620-1622. 19. Terry, S.C., Jerman, J.H., Angell, J.B., A Gas Chromatographic Air Analyzer Fabricated on a Silicon Wafer. IEEE Transactions on Electron Devices, 1979. ED-26: p. 1880-1886. 20. Kikuchi, Y., Ohki, H., Kaneko, T., and Sato, K., Microchannels Made on Silicon Wafer for Measurement of Flow Properties of Blood Cells. Biorheology, 1989. 26: p. 1055. 21. Kikuchi, Y., Sato, K., Kaneko, T., and Ohki, H., Optically Accessible Microchannels Formed in Single-Crystal Silicon Substrate for Studies of Blood Rheology. Microvascular Research, 1992. 44: p. 226-240. 22. Kikuchi, Y., Sato, K., and Mizuguchi, Y., Modified Cell Flow Microchannels in a Single-Crystal Silicon Substrate and Flow Behavior of Blood Cells. Microvascular Research, 1994. 47: p. 126-139.

PAGE 75

65 23. Tracey, M.C., Greenaway, R.S., Das, A., Kaye, P.H., and Barnes, A.J., A Silicon Micromachined Device for Use in Blood Cell Deformability Studies. IEEE Transactions on Biomedical Engineering, 1995. 42(8): p. 24. Sutton, N., Tracey, M.C., Johnston, I., Greenaway, R.S., and Rampling, M.W., A Novel Instrument for Studying the Flow Behavior of Erythrocytes through Microchannels Simulating Human Blood Capillaries. Microvascular Research, 1997. 53: p. 272-281. 25. Tracey, M.C., Sutton, N., Johnston, I., and Doeztel, W., A Microfluidics-Based Instrument for Cytomechanical Studies of Blood. 1st Annual International IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine and Biology, October 12-14, 2000. 26. Brody, J.P., Han, Y., Austin, R.H., Bitensky, M., Deformation and Flow of Red Blood Cells in a Synthetic Lattice: Evidence for an Active Cytoskeleton. Biophysics Journal, 1995. 68(6): p. 2224-2232. 27. Ayliffe, H.E., Rabbitt, R.D., Tresca, P.A., and Frazier, A.B., Micromachined Cellular Characterization System for Studying the Biomechanics of Individual Cells. Transducers 1997. 1997 International Conference on Solid-State Sensors and Actuators, Chicago, June 16-19, 1997. 28. Coulter, W.H., High Speed Automatic Blood Cell Analyzer. Proceedings of the National Electronics Conference, 1956. 12: p. 1034-1040. 29. Roberts, K., Parameswaran, M., Moore, M., and Muller, R., ASilicon Microfabricated Aperture for Counting Cells using the Aperture Impedance Technique. Proceedings of the IEEE Canadian Conference on Electrical and Computer Engineering. Shaw Conference center, Edmonton, Alberta, Canada May 9-12, 1999. 30. Satake, D., Ebi, H., Oku, N., Matsuda, K., Takao, H., Ashiki, M., and Ishida, M., A Sensor for Blood Cell Counter Using MEMS Technology. Sensors and Actuators B, 2002. 83: p. 77-81. 31. Gawad, S., Heuschkel, M., Leung-Ki, Y., Iuzzolino, R., Schild, L., Lerch, Ph., and Renaud, Ph., Fabrication of a Microfluidic Cell analyzer in a Microchannel Using Impedance Spectroscopy. 1st Annual International IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine and Biology, October 12-14, 2000. 32. Voldman, J., Gray, M.L., Toner, M., and Schmidt, M.A., A Microfabrication-Based Dynamic Cytometer. Analytical Chemistry, 2002. 74: p. 3984-3990.

PAGE 76

66 33. Shrewsbury, P.J., Muller, S. J., Liepmann, D., Effect of Flow on Complex Biological Macromolecules in Microfluidic Devices. Biomedical Microdevices, 2001. 3(3): p. 225-238. 34. Kleinfeld, D., Kahler, K.H., Hockberger, P.E., Controlled Outgrowth of Dissociated Neurons on Patterned Substrates. Neuroscience, 1988. 8(11): p. 4098-4120. 35. Zhao, B., Moore, J.S., Beebe, D.J., Surface-Directed Liquid Flow Inside Microchannels. Science, 2001. 291: p. 1023-1026. 36. Folch, A., Toner, M., Cellular Micropatterns on Biocompatible Materials. Biotechnology Progress, 1998. 14: p. 388-392. 37. Gad-el-Hak, M., The MEMS Handbook. 2002, London, New York, Washingtion D.C.: CRC Press. 38. Pellois, J.P., Zhou, X., Srivannavit, O., Zhou, T., Gulari, E., Gao, X., Individually Addressable Parallel Peptide Synthesis on Microchips. Nature Biotechnology, 2002. 20: p. 922-926. 39. White, Frank. Viscous Fluid Flow, 1991, New York: McGraw-Hill. 40. Papautsky, I., Gale, B.K., Mohanty, S., Ameel T.A., Frazier, A.B., Effects of Rectangular Microchannel Aspect Ratio on Laminar Friction Constant. SPIE Microfluidic Devices And Systems Conference, 1999. p. 147-159. 41. Eringen, A.C., Simple Microfluids. Int. J. Eng. Sci., 1964. 2: p. 205-217. 42. Eringen, A.C., Suhubi, E.S., NonlinearTtheory of Simple Micro-Elastic Solids. International Journal. Engineering Science, 1964. 2: p. 189-203. 43. Kang, C.K., Eringen, A.C., The Effect of Microstructure on the Rheological Properties of Blood. Bull. Math. Biol., 1976. 38: p. 135-158. 44. Wu, P., Little, W.A., Measurement of Friction Factors for the Flow of Gases in Very Fine Channels used for Microminiature Joule-Thomson Refrigerators. Cryogenics, 1983. 5: p. 273-277. 45. White, F.M., Fluid Mechanics, 1999, Boston, Massachusetts: WCB McGraw-Hill. 46. Bousse, L., Cohen, C., Nikiforov, T., Chow, A., Kopf-Sill, A.R., Dubrow, R., Parce, J.W., Electrokinetically Controlled Microfluidic Analysis Systems. Annual Review of Biophysical and Biomolecular Structures, 2000. 29: p. 155-181.

PAGE 77

67 47. Kornberg, L.J., Focal Adhesion Kinase in Cancer. Emerging Therapeutic Targets, 2000. 4(2): pg. 1-16. 48. Haier, J., Nicholson, G.L., Tumor Cell Adhesion under Hydrodynamic Conditions of Fluid Flow. APMIS, 2001. 49: pg. 155-166. 49. Weiner, T.M., Lui, E.T., Craven, R.J., Cance, W.G., Expression of Focal Adhesion Kinase Gene and Invasive Cancer. Lancet, 1993. 342: pg. 1024-1025. 50. Ferguson, D., Cellular Attachment to Implanted Foreign Bodies in Relation to Tumorigenesis. Cancer Research, 1977. 37: pg. 4367-4371. 51. Tran-Son-Tay, R., Branham, M., Glover, S., Perrault, C., Benya, R.V., Allen, S., Shyy, W., A Rapid Prototyping Technique for Studying Contribution of Extracellular Microtexture to Cell Shape, Attachment and Spreading, 11th Int. Congress of Biorheology and 4th Int. Conf. on Clinical Hemorheology, Antalya, Turkey, 2002). 52. Loh, I-H., Plasma Purface Modification in Biomedical Applications. AST Technical Journal, 2003. Available from URL: http://www.astp.com/technical_info/general_lit.cfm Site last visited May 2003. 53. Tritton, D.J., Physical Fluid Dynamics, 1988. Oxford: Clarendon Press.

PAGE 78

BIOGRAPHICAL SKETCH Katie Maiellaro was born in Pensacola, Florida, on June 29, 1978. Her parents Edward and Rachel Maiellaro offered her a loving environment to grow up happy and healthy. Katie has two older brothers Matt and Tom, whom she loves and admires incredibly. Her foundation of faith and strong values was created by her family, as well as by St. Pauls Catholic School, where she attended from kindergarten through 8th grade. She then attended the International Baccalaureate Program at Pensacola High School, graduating Valedictorian of the Class of 1996. In addition, Katie was a 4-year member of the PHS tennis team, winning her district title in 1993. Katie received her Bachelor of Science degree in engineering science at the University of Florida. With the thought of eventually attending medical school, Katie first decided to further her interest in biomechanics by remaining at UF and pursuing her Master of Engineering degree in biomedical engineering. However, during her tenure as a graduate student, she discovered her love of the academic environment and decided continue her education as an engineer and complete her doctorate in biomedical engineering. After receiving her M.E. in biomedical engineering at the University of Florida she will enroll in the Joint Biomedical Engineering Ph.D. Program at Georgia Tech and Emory in August 2003. 68


Permanent Link: http://ufdc.ufl.edu/UFE0001145/00001

Material Information

Title: Microfabricated Silicon Microchannels for Cell Rheology Study
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0001145:00001

Permanent Link: http://ufdc.ufl.edu/UFE0001145/00001

Material Information

Title: Microfabricated Silicon Microchannels for Cell Rheology Study
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0001145:00001


This item has the following downloads:


Full Text












MICROFABRICATED SILICON MICROCHANNELS FOR CELL RHEOLOGY
STUDY















By

KATHRYN ADELE MAIELLARO


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Kathryn Adele Maiellaro















ACKNOWLEDGMENTS

I acknowledge my advisor Dr. Roger Tran-Son-Tay for 2 years of guidance.

Especially at this year's end, I have called upon his help and he has responded with

encouragement and support. Though I began graduate school not knowing if research

was quite right for me, with his guidance I have ended this experience with re-evaluated

goals for a career in academic research. Along the same lines, I thank Dr. Mark Sheplak

for reassuring me that I really am a smart person who has something to contribute to the

engineering community.

I also am greatly indebted to my lab colleagues who also deserve credit for

helping complete this research. Melissa and Amrit are my stability and my sanity. I do

not know what I would have done without them. Rebecca shared with me her knowledge

and experience, for which I am a better researcher. Darren kept me in my place. Cecile,

Eddie, Ethan, Brent, and Ben made the lab a great place to work.
















TABLE OF CONTENTS
Page

A C K N O W L E D G M E N T S ................................................................................................. iii

L IST O F T A B L E S ........ ............................................................. .... ...... .... ....... vi

L IST O F F IG U R E S .... ...... ................................................ .. .. ..... .............. vii

ABSTRACT .............. .................. .......... .............. ix

CHAPTER

1 IN T R O D U C T IO N ................. .................................. .... ....... .. ............. .

1. 1 Obj active ................................ .............................. ......... 1
1.2 Specific Aim s ................. ............................. ...............
1.3 S ig n ifican ce ....................................................... 1

2 B A CK G R O U N D .........................................................................4

2.1 M icrofabrication .............................................................. 4
2.1.1 Current microfabrication processes .......... ............. .....................5
2.1.2 R apid prototyping .... .... ......... .... .... ...... ........ ..................... 11
2.1.2.1 Replica molding ............ ..................... ........... ............... 12
2.1.2.2 Polym er hot embossing ............. .................. ............... ....... 12
2 .2 B ioM E M S ................................................................... ............................ 13
2.2.1 M icrochannels .................................. ...................... 14
2.2.2 Current BioMEMS devices using microchannels ...................................16
2.2.2.1 Cell sorting and counting ................................ .......... .. .......... 16
2.2.2.2 M acromolecular assays .................................. .................. 17
2.3 M icrofluidics............................................................................. 18
2.4 Colon Cancer and Focal Adhesion Kinase .......... .......................................20

3 MATERIALS AND METHODS.............................. ............... ...............22

3.1 Channel Design and Specifications ..................................................22
3.2 Computer Assisted Drawings.............. ................................ ...............25
3 .3 M materials ............................................ ................. ................. ............ 2 6
3.4 Process Flow for Microchannel Fabrication...............................28
3 .5 C h ip C lean in g ......... ........................................... ............................................. 3 3










3.6 Packaging .............................................. 34
3.7 E xperim ental Set-U p .................................................. .............................. 35

4 MICROCHANNEL CHARACTERIZATION........... ..... .................37

5 RESULTS AND DISCUSSION ............ ..... ................................ 43

5.1 M icrochannel System .................................................. .............. ............... 43
5.1.1 Fabrication of microchannels ....................... ......... ...... ............... 43
5.1.2 Flow system .............. .... ......... ...... ........49
5.1.3 Imaging ..... ................... .......... ........ 54
5.2 Im provem ents to M icrochannel System ...............................................................54

6 CON CLU SION .................. ........................................ ............... 58

6.1 Effectiveness of M icrochannel System ............................................................ 58
6 .2 F u tu re W ork ................................................... ................ 59

APPENDIX

A PROCESS FLOW ........................................................... .............. 60

B BA SE D IM EN SION S ............................................................... ......61

C M ATLAB CODE .............. ................. ....................... ............ 62

L IST O F R E FE R E N C E S ............................................................................. .............. 63

B IO G R A PH IC A L SK E TCH ..................................................................... ..................68
























v
















LIST OF TABLES

Table pge

3-1 Summ ary of channel types and widths ............... ............................. ............... 23

4 -1 C o n sta n t C 1 .................................. ................................ ................ 3 8

4-2 Values for dp dx, Q, AP, and R. .............................. .... ................................ 40

4-3 Values for T, Dh, Re, and Po .................................... ....... ............... 41

4-4 Le(in kim) for 3 km and 70 kim wide channels........................................... ........... 41
















LIST OF FIGURES

Figure page

2-1 Computer generated pattern and resulting mask ....................... ................

2-2 Photolithography schem atic ......................................................... .............. 7

2-3 R resulting cross-sections .................................................. ............................... 7

2-4 A nisotropic etch profile.............................................................. ....................... 9

2-5 Isotropic etch profile ..................................... ............................. .9

2-6 DRIE schematic......... ....................................... ....... 11

3-1 Sketch of the channel parts with dimensions. ............ ..................................... 24

3-2 C oventor im ages ........ ........ ....... ......... ... ....................... 26

3-3 C oventor draw ings ......................... .............................. .. ........ .... ..... ...... 26

3-4 Reference cross-section at the middle of the channels.....................................28

3-5 Initial wafer processing and photolithography of Channels mask.........................29

3-6 D RIE of channels. ........................................... ........ .......... ... ..... 30

3-7 Reference cross-section m id-w afer .............................................. ............... 30

3-8 W after processing................. ....... ..................................... ..... ......... 30

3-9 Resist bond of support wafer........... .... .... ......... ............... 31

3-10 W after processing ............................................... ............. ................. 32

3-11 Thirty-four chips cleaved from each wafer ......................... ..................32

3-12 Packaging and fluid access of chip ............................................... ............... 34

3-13 C hip package ...................... .... ......... .. .. ..... .... .................................34

3-14 D ata acquisition set-up .................................................. ............................... 36









4-1 Labeled cross-section of rectangular microchannel ..................... .............. 37

5-1 Selected pictures of microchannels .... ............................ ........................ ............... 44

5-2 Surface profile of the 50 tm wide straight microchannels ...................................45

5-3 Surface profile of the 12 tm wide endothelial cell contoured microchannels.........45

5-4 Surface profile of 100 tm wide channel. ..................................... ............... 46

5-5 Surface profile of the 10 tm wide bifurcated channel. ........... ... ........... 46

5-6 Perspective view of the 3 tm wide channels. .................................. .................48

5-7 Perspective view of the 50 tm wide channels. ......................................................48

5-8 Fluid activation through a 50 tm wide channel................................................. 49

5-9 Fluid flow through the 50 am wide bifurcated channel .......................................50

5-10 50 tm endothelial cell contoured channel........... ......... ....... ............... 51

5-11 5 tm wide endothelial cell contoured channels with blood flow ...........................52

5-12 5 micron wide channel showing fluid traveling over the channel devices ..............52

5-13 10 tm straight channel with single cell flow. .................................. .................53

5-14 Input port of 50 tm wide flow channel showing a fluid void on far right ..............53

5-15 Schem atic of im proved base .............................................................................. 56

5-16 Schematic of improved channel parts. ........................................ ............... 56














Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Engineering

MICROFABRICATED SILION MICROCHANNELS FOR CELL RHEOLOGY
STUDY

By

Kathryn Maiellaro

August 2003

Chair: Roger Tran-Son-Tay
Major Department: Biomedical Engineering

The importance of the microcirculation is highlighted by the fact that most of the

hydrodynamic resistance of the circulatory system lies in the microvessels; and most of

the exchange of nutrients and waste products occurs in the capillaries. The main interests

in microcirculation include pressure-flow relationship, flow of blood cells in capillaries,

stress distribution in blood cells and vessels, and mass transfer across the capillary walls.

Knowledge of the theological properties of blood cells is also critical in understanding

their role in health and diseases because the ability of a cell to flow into microvessels and

to migrate through tissues is governed by those properties. There is also an increasing

interest in understanding the mechanisms of flow regulation and molecular exchange in

the microcirculation of organs and tissues under physiological and pathological

conditions.

Our objective was to develop sturdy microchannels for studying blood

microcirculation and for characterizing cell theological properties, function, and









behavior. To demonstrate the capabilities of the proposed microfabrication technique,

several features of the microcirculation system were produced and incorporated into a

microfluidic device into which whole blood was introduced.

To study the effect of vessel size on blood flow, microchannels with width

ranging from 100 [tm to 3 [tm were designed. The largest of these channels mimic

arterioles or venules, which have a diameter of about 30 [tm, while the smallest of these

channels mimic capillaries, which have diameters of less than 8 tm. Bifurcated

microchannels were designed to provide a model for blood flow through a branching

point from larger to smaller blood vessels, while the stenosed channels provide a model

for flow through an occluded vessel, such as in cases of arteriosclerosis, or calcium or

fatty deposits on the blood vessel interior wall.

A process flow was developed to etch thirty-four 12 mm x 12 mm microchannel

chips into a 100-mm-diameter single-crystal silicon wafer. Each chip is an individual

microchannel system, including an input port, microchannel device, and an output port.

Etching the channels into rigid silicon offered excellent geometric control to fabricate the

network of microchannels for studying microcirculation.

Results indicate that the microchannel system could extend deformational

information offered by micropipette experiment in two ways. First the system can

provide observation of cellular response to the endothelial lining of a vessel. Second, the

system allows observation of blood and cell flow through an extensive network of

microvessels similar to those found in the microcirculation.














CHAPTER 1
INTRODUCTION

1.1 Objective

The objective of this research is to develop rigid microchannels for the study of

blood microcirculation and the characterization of cell theological properties, function,

and behavior.

1.2 Specific Aims

* To show the feasibility of using the developed procedure to fabricate a network of

microchannels for the study of microcirculation. Flow through a bifurcation and

stenosis will be observed.

* To design a microdevice to characterize the theological properties of individual

cells. The flow of a blood cell through a capillary tube will be produced.

Capillaries possessing different geometries will be created.

* To produce a microchannel that mimics the lining of blood vessels. This design

would also form a first order model for studying flow through porous media.

* To observe the effects of surface roughness on the behavior of whole blood and

colon cancer cells.

1.3 Significance

The significance of the microdevice is that it isolates specific aspects of the

microcirculation system so that models of small vessel behavior and function can be

obtained. The microcirculation is where oxygen and nutrients diffuse into tissues and









carbon dioxide and wastes are removed. It is a dynamic fluidic system whose vessels

would span over 25, 000 miles if arranged end to end [1].

The specific parts of microcirculation that are isolated are small blood vessel size,

blood vessel geometry, and blood vessel interior roughness. The vessel size is studied by

creating a microchannel system that includes channels with diameters ranging from 100

[tm to 3 tm. The largest of these channels mimic arteriole or venule flow that possess a

diameter of about 30 am, while the smallest of these channels is used to observe the

single-file flow of red blood cells through the capillaries, which have diameter of about 8

am. Bifurcated and stenosed channels will be used to study vessel geometry.

Observation of flow through the bifurcated vessels provides a model for blood flow

through a branching point from larger to smaller blood vessels. Observation of flow

through the stenosed channels provides a model for arteriosclerosis that which includes

both calcium (focal calcification) and fatty plaque (atherosclerosis) deposits on the vessel

interior wall. Vessel interior roughness is studied by applying a surface contour to the

walls of the channels. This surface pattern is comparable to that of the endothelial lining

of the vessels. In addition, this pattern will supply a first order model of flow through

porous media, which is a future study that will investigate diffusion processes in

microcirculation.

Saline solution, whole blood, and cells from a well defined 293 fetal kidney

epithelial cell line will be introduced to the microdevice. Introducing whole blood to the

system will provide images of blood flow through the isolated microcirculation parts. Of

specific interest will be the deformation of red blood cells and white blood cells through

the smallest channels. Cell deformability is important because it affects cell motility,









which then impacts the cell's ability to maneuver into the capillary bed. Lastly,

introducing the 293 cells into the system will provide observation of the effect of a

protein called focal adhesion kinase (FAK) on cell deformation. To obtain specific

experimental results, the cancer cells will be treated with doxycycline (Clontech, Palo

Alto, CA) to observe on the deformation of the 293 cells as they maneuver through

capillaries.














CHAPTER 2
BACKGROUND

2.1 Microfabrication

In the 1960s some microelectronics scientists had an idea to put sensors and

circuitry on the same chip to reduce device size [2]. Also at this time the mechanical

potential of single-crystal silicon for use in batch fabrication of miniaturized structures

was established [3]. By the end of the decade silicon was being etched into thin films to

utilize its piezoresistivity for pressure measurements. Eventually, mechanical processes

were incorporated with the circuitry, creating tiny micron size systems now referred to as

microelectromechanical systems, or MEMS.

An interesting aspect of MEMS is that knowledge of pure electronics or mechanics

is not sufficient to understand their capacity. The interdisciplinary nature of MEMS

requires knowledge of a nonexhaustive list of materials, fluids, heat transfer, optics,

feedback, statics, and dynamics. With so many systems acting in a tiny device, MEMS

bring together researchers of different backgrounds to collaborate on a single project. No

longer localized to electronics, MEMS technology now applies to acoustics, optics,

electrostatics, mechanics, and, the focus of this research, to biotechnology.

The process of making MEMS is called microfabrication. Microfabrication is a

general term, which can be defined as the fabrication of miniature devices. It includes

silicon wafer etching, film deposition, micromilling, microdrilling, and any other

fabrication method that has the word micro in its title. By offering excellent geometric

control and features in the submicron range (a feature that will decrease as new









fabrication processes develop), microfabrication allows the creation of tools for

molecular biology, biochemistry, medicine, and cell biology.

Microfabrication not only opens the door for a wider range of cell biology studies,

but it also enables cheaper and more efficient studies by offering the following

advantages [4]:

* smaller device size

* less material use

* smaller sample size

* less waste

* geometrical control

* a viable environment for biomolecules

* single cell analysis

As knowledge of microfabrication spreads, microfabricated devices will continue to

appear in many different research fields. Biomedical engineering, biochemistry,

chemistry, embryology, and biology are already benefiting from this growing technology.

The result is more biologically significant studies utilized to yield better medical devices

and health care in turn.

2.1.1 Current microfabrication processes

Microfabrication processes are described in detail in the literature [2, 3, 5], but a

brief overview is provided below. The major processes discussed here are

photolithography and pattern etching. Photolithography is the process of transferring a

pattern onto the substrate that is to be etched. The design or structure that will be etched

into the substrate is first generated by computer and then drawn into a "hardcopy" called









a mask, as shown in Figure 2-1. A mask is a transparent plate, commonly a 5 in x 5 in

plate (depending on wafer size) made of soda lime or quartz, onto which the computer

generated design is deposited in a thin chrome layer. There are two types of mask

generation. Designs that consist of only rectangular features, called Manhattan features,

are created into a mask by a process called pattern generation. Designs that include

circular or non-Manhattan features must be made into masks by a more expensive

electron beam process.

Masks are either dark field or clear field. In a dark field mask, the entire field of

the mask is covered with chrome while the pattern features remain transparent.

Conversely in a clear field mask, chrome is only deposited in the patterned areas while

the field of the mask remains chrome-free.



UF
A







Figure 2-1. A) Computer generated pattern that is to be transferred to silicon wafer. B)
Dark-field chrome mask of the pattern, where the black is the chrome layer.

Photolithography (Figure 2-2) includes four steps of photoresist spinning, aligning,

exposure, and developing. Photoresist is a photosensitive polymer that acts as a

photographic film, allowing the desired pattern to be transferred from the mask to the

substrate. It is applied to the substrate in a thin layer, usually by a spin-coating process.

There are two types of photoresist--positive and negative-each with a different chemical










reaction to ultraviolet light. Positive photoresist becomes more soluble when exposed to

ultraviolet light, while negative resist cross-links and becomes less soluble in developing

solution. If positive resist is selected, the patterned areas will be dissolved during the

developing steps, exposing the substrate below, and allowing the pattern to be etched

directly into the substrate (Figure 2-3A). However, if negative resist is chosen, the non-

patterned areas will be dissolved by the developing agent, and the substrate will be etched

down everywhere except the patterned areas. This method will leave the pattern standing

above the substrate surface like a relief carving (Figure 2-3B).




SUV L.I, Source
5L The photoresisl s exposed here
SBut there

Thin layer photosensilive polymer
500 micron tck silicon wafer



Only the letters Jr haI e
been exposed to uv' I-n, {
The exposed poiyeer U
cialirig has been *ierniicdllly F
changed and now oir cli;i, ,.
waf er MA be el-hgI '.



Figure 2-2. Photolithography schematic. The silicon wafer is coated with a thin layer of
positive photoresist. The photomask is aligned over the wafer and the
wafer/mask assembly is exposed to ultraviolet light.




(A)l (B)

Figure 2-3. Resulting cross-sections created by A) positive photoresist and B) negative
photoresist.









After the photoresist coat, the mask is aligned over the substrate. Ultraviolet light

is then directed over the mask/substrate assembly. This step is called exposure, wherein

the ultraviolet light passes only through the clear areas of the mask to the substrate below.

The final stage in photolithography is developing the photoresist. During this

process the substrate is placed in developing solution to remove the soluble areas of

photoresist. At this point the choice of photoresist becomes important. After the pattern

has been transferred to the photoresist by photolithography, it is etched into the substrate

by wet or dry etching. The reader is referred to the literature [2, 6, 7] for discussion of

the specific etch systems and the resulting chemical reactions involved. Before the

etchant system for a microfabrication process is chosen factors such as the etch rate of the

system, material lattice orientation, and the etchant selectivity must be considered. In

addition, cost is another significant factor since etching techniques range in price from

tens to thousands of dollars.

When using single crystal substrates, orientation-dependent etching, called

anisotropic etching, often occurs. Alternatively, if the etch is orientation-independent the

etch profile is called isotropic. It is therefore important to know the lattice orientation of

the substrate involved. A common substrate material is (100) single-crystal silicon.

(100) single-crystal silicon is composed of crystals that have the {100} plane at the

surface. Etch profiles for anisotropic and isotropic etchants are shown in Figure 2-4 and

Figure 2-5, respectively. Figure 2-4 shows the well characterized anisotropic etch created

by potassium hydroxide (KOH). The profile makes a 54.7 angle with the horizontal.

The non-directionality of the isotropic etch is shown in Figure 2-5. Isotropic etching










often results in undercutting of mask features. Depending on the tolerance of the

patterned features, undercutting may or may not be acceptable.

1'11' 1111'' 11111' 1i i I111'


(100) silicon N






(d)


Figure 2-4. Anisotropic etch profile. A) Rectangular mask. B) Cross section of silicon
wafer. C) Diagonal etch profile. D) Top view of etched feature. Schematic
adapted from Petersen [3].



(a) I I (b)





(c)




Figure 2-5. Isotropic etch profile. A) Rectangular mask. B) Cross section of silicon
wafer. C) Round cross section after isotropic etch. Schematic adapted from
Petersen [3].

The etchant selectivity is a ratio that indicates the preference with which the etchant

attacks a material. Etchant selectivity, along with etch rates at specified temperatures, are

located in a database of etchants [6] so that the appropriate etchant system is selected for

microfabrication processes. To define selectivity, consider the wet etchant KOH. Its

selectivity to (100)/(111) silicon is 400:1, meaning that for every 400 units of (100)

silicon etched, only 1 unit of (111) silicon will be etched [3]. However, though databases

for microfabrication processes tabulate general etch rate and selectivity, temperature,


N1~0~10I


iii iii i~i~iiiiii ....... ....... ...iiii

. . . . .
.......... ............ ......... .... . . . . .
. . . . . . .
. . . . .









concentration of etchant solution, and additives must also be considered when

determining the etch conditions, as these influence the final fabrication results.

Wet etching consists of submersing the wafer in a solution or combination of

solutions that attack the exposed material. In silicon, etching occurs when silicon-silicon

bonds are broken and replaced with a silicon-OH bond [2]. The reaction results in the

consumption of water molecules and the formation of Si(OH)4 and hydrogen gas. A

common wet etchant used to obtain an isotropic etch profile in silicon is hydrofluoric acid

with nitric acid (HF/HNO3). KOH, ethylene diamine pyrcatechol (EDP), and tetramethyl

ammonium hydroxide (TmAH) are common etchant systems to anisotropically etch

silicon.

Dry etchants are gaseous species that remove wafer material. They constantly

introduce new species into the etch environment to reduce the chemical side-reactions

and allow etching of tight patterns [6]. In addition, dry etching is significantly more

expensive, requiring large machinery and trained personnel to operate. Dry etch systems

are classified as vapor or plasma-assisted. One vapor etchant used widely in

microfabrication is XeF2. It is often used to remove bulk silicon, when rough features are

tolerable [5]. Plasma-assisted processes are more useful for obtaining smoother surfaces.

Plasma processes introduce ions into the etch environment creating a reaction with the

surface to chemically remove material. Surface atoms are then converted to volatile

species that are removed by vacuum [2]. Deep vertical trenches or channels are achieved

by a plasma process called deep reactive ion etching (DRIE). DRIE is a three step

process as shown in Figure 2-6 that involves etching a shallow trench into silicon using

SF6 plasma, passivating that newly formed trench with polymer created with the addition









of C4Fs plasma, and then etching a subsequent and deeper trench with SF6 plasma [5].

Passivation with polymer prevents lateral etching of the sidewalls, while the hole

becomes deeper. DRIE is often used to etch holes in a wafer or is used when sharp

features are required.

SF@ etch Polymerized SF6 etch






Figure 2-6. DRIE schematic showing SF6 etching, passivation, followed by SF6 etching.
Schematic adapted from [2].

2.1.2 Rapid prototyping

While microfabrication processing offers superior geometric and dimensional

control, the high cost and slow turn around time often preclude its use. In light of this

difficulty, work to develop microstructure fabrication methods using common lab

supplies and less expensive materials began. These fabrication methods are lumped

under a term called rapid prototyping.

Any discussion of rapid prototyping must first begin with a description of

poly(dimethylsiloxane) (PDMS), likely the most widely used material in rapid

prototyping. PDMS is an excellent material for rapid prototyping of microdevices for a

variety of reasons. It is initially a liquid material that arrives in two components,

including a silicone elastomer and a cross-linking agent. Our lab works with Sylgard

184 silicone elastomer kit (Dow Corning Corporation, Midland, MI). The liquid

components are combined in a 10:1 elastomer to cross-linker ratio. The liquid is then

poured over a surface called the master and left to cure into any desired shape.









PDMS conforms to surface features on the micron scale and when cured releases

from the master without damaging its surface. Cured PDMS is optically transparent

which makes it suitable for microscopy. In addition, PDMS will cure at room

temperature in about 72 hours or can be heated in a 600 oven to decrease the curing time

to a few hours [8,9].

2.1.2.1 Replica molding

A common replica molding method uses a high-resolution printer to generate rapid-

prototyped photomasks to be used in lieu of expensive soda lime/chrome masks

[8,10,11]. With this process, a CAD design is photographically reduced onto 35mm film

or microfiche film. This film acts as a mask that is aligned over a photoresist covered

substrate and exposed to ultraviolet light. After the resist is developed, this substrate

becomes the master for PDMS casting. The photoresist used in this type of rapid

prototyping is called SU-8 (MicroChem Inc., Newton, MA). Compared to other

photoresists that are spun on a substrate in thicknesses up to 2-3 [tm, SU-8 can be

deposited in thicknesses up to 2 mm. Using this method the minimum feature obtained

thus far is 1 [tm [12].

2.1.2.2 Polymer hot embossing

Another rapid prototyping method is polymer hot embossing. The impetus for

polymer molding is that microfabrication does not lend itself to create 3-dimensional

structures, at least without complicated fabrication protocols and high expense. Polymer

hot embossing takes advantage of the viscoelastic properties of a polymer, often

poly(methylmethacrylate) (PMMA), near its glass transition point, Tg, the point at which

a polymer changes phase from hard to pliable. In this method a master pattern and

polymer substrate are brought into contact, heated, and pressed together. The master









pattern is then transferred, or embossed, into the deformable plastic substrate. After

embossing the PMMA sheet is cooled below its Tg and the pattern is permanently formed

in the plastic substrate. Hot embossing has shown dimensional fidelity of pillars that are

1 tm deep and approximately 15 tm in tip diameter [13]. PMMA embossing is used in

the fabrication of many protein separation/detection systems, specifically the LabCardTM

(ACLARA BioSciences, Mountain View, CA) [14].

One advantage of polymer embossing over PDMS molding is that PMMA is an

easily handled and cleaned hard plastic, while PDMS is a flexible, sticky elastomer that

collects ambient dust particles during experimentation. PDMS must be handled

carefully to minimize fingerprint transfer and dust collection, both of which can interfere

with experimental microscopy. Therefore, these permanent patterns in hard plastic are

beneficial primarily due to handling ease. In addition, using plastics as alternative to soft

elastmomers also has the advantage of minimizing the electroosmotic motility compared

to microchannels etched in glass substrates [15].

2.2 BioMEMS

MEMS devices that are designed for biological applications are called BioMEMS.

BioMEMS involves any use of microfabrication or rapid prototyping techniques with

biotechnological applications. Other word combinations used in conjunction with

BioMEMS are bioanalytical devices, microdevices, lab-on-a-chip, microchips, or micro-

total analysis systems ([tTAS), coined by Manz et al. [16]. However, no matter what

phrase is used, all of these terms encompass two components-- small size and biological

application.









Harrison et al. [17] demonstrated the potential of the (TAS concept in 1993. An

extensive literature survey in Clinical Chemistry has most sufficiently illustrated the

massive amount of investigation and research dollars pouring into the BioMEMS area

[18]. In this review "BioMEMS analytical microchips," yet another combination of

words, were characterized into four key parts:

* Fabrication, integration of parts, and device modeling

* Microfluidic mechanics, fluid flow, and mixing

* Applications of microdevices in biotechnology

* Patents

According to the review, as of 2002 there were 255 references on fabrication, integration

and modeling, 194 references on flow at low Reynolds numbers, 291 references on

BioMEMS applications, and 276 patents in this field. To limit a discussion that could fill

many textbooks, only applications of microdevices in biotechnology that rely on

microchannels in the areas of cell sorting and macromolecular assays will be reviewed

here. However, first the microchannel will be introduced and defined.

2.2.1 Microchannels

The microchannel is an integral part of a microdevice. A microchannel may be

defined as a via to move cell suspensions or fluids from one part of a microchip to

another part. Any port, groove, line, and etched or embossed feature become a

microchannel when fluids are introduced. Microchannels often function to mimic

capillaries in the micron scale, which allow observation of cells squeezing through the

narrow passages of the vascular system. They are utilized for applications from the study

of blood flow through vessels to protein analysis.









The first microchannel system, consisting of an isotropically etched channel

bonded with a glass cover, was developed by Terry et al. [19]. Kurt Petersen then

detailed the utility of silicon as a mechanical material in 1982 [3]. His work reached

beyond the microelectronic community, showing researchers in biological fields that

microtechnology could benefit their work as well. In the late 1980s and 1990s, research

groups were learning the basics of microfabrication and developing simple systems to

determine whether the technology suited their needs. In 1989 Kikuchi et al. developed a

system of straight anisotropically etched microchannels, which were used to view white

blood cells [20-22]. These channels allowed single cell analysis and measurement of

transit time, total flow rate of cell suspensions, and cell component aggregation. The

group also emphasized the difficulty of introducing a bubble-free fluid solution into the

channels, as the fluid set-up often is as difficult a task as fabricating the microdevice

itself. In 1995 through 2000, Tracy et al. developed and improved a microchannel

system for single cell analysis celled a haemorheometer [23-25]. Their chamber offered a

correlation between cell length and cell velocity, and primarily analyzed the mechanical

properties of red blood cells. The haemorheometer allowed detection of up to 1500 cells

on a cell-by-cell basis. In addition, utilizing their haemorheometer, additional

experiments with normal whole blood, chemically altered blood, and thalassaemic blood

were performed to investigate the flow of each in terms of cell-membrane mechanical

properties and extrinsic property of volume.

In 1995 Brody et al. developed the first large microchannel array as part of a

doctoral thesis [26]. This system was an attempt to place blood cells in an in vitro system

that modeled the physiological environment. The cells flowed through a series of









channels that mimicked a capillary bed. Cells were repeatedly deformed as they entered

a microchannel smaller than the cell diameter and exited into a wide space, only to be

deformed again in subsequent channels.

Finally, in 1997 Ayliffe et al. [27] tested materials other than silicon in a

microchannel system. They fabricated channels with glass tops and bottoms and

polyimide sidewalls. What was unique about their study was that instead of driving fluid

flow using a pressure system, they observed cell motility by chemotaxis.

2.2.2 Current BioMEMS devices using microchannels

The ability to microfabricate bioanalytical devices allowed researchers to pin

point areas in health care and medicine for this new technology to improve. Two areas

of application are cell sorting/separation and macromolecular assays. These areas are

addressed in the following sections.

2.2.2.1 Cell sorting and counting

Microchannel systems that provide a method of separating and sorting cell

solutions into cell types in order to target specific cells are widely researched in

BioMEMS. Similarly, methods to separate whole blood into serum and cells to obtain

concentrated reservoirs of erythrocytes, leucocytes, and platelets are being developed.

Microfabrication allows creation of different geometries of microchannels with which to

facilitate cell sorting and separation.

To move cells with control and to count populations of cells, many groups have

used microfabrication techniques to improve the existing Coulter Counter, developed in

the 1950s [28]. The theory behind the counter is to correspond cell size to the voltage

change measured as cells move individually through a small aperture. A number of









groups have modified the Coulter Counter by changing the shape of the aperture using a

silicon etching protocol [29] or by improving the voltage signal for the sensor [30].

Taking advantage of the impedance change in a circuit produced by cells passing

through a small aperture, Gawad et al. [31] developed a novel on-chip flow-cytometer to

create a model of an erythrocyte during flow. Their objective was to study the influence

of cell size on the channel resistance. The purpose was to obtain a model for normal

erythrocyte flow with which to compare other particles. The authors state that this micro-

cell sorter may possibly be used to identify abnormal cell types such as cancer cells.

A novel cell-sorting device developed in 2002 uses microfabricated gold-coated

posts as electrodes to trap individual cells flowing through a microchannel [32]. The

traps capitalize on dielectrophoresis to immobilize the cells in potential energy wells.

The traps have yet to be fabricated on a large array scale, but the proof-of-concept design

proves very effective at separating populations of cells.

2.2.2.2 Macromolecular assays

In the late 1990s, after researchers had learned both of the benefits of

microfabrication techniques and the basics of microchannel fabrication, they were able to

modify the technology to create assays specific to protein analysis. In 2001, Shrewbury

et al. [33] used silicon-etched rectangular microchannels to study the effect of flow on

macromolecular molecules. Using microchannels they subjected the proteins to

elongation flow along the centerline, as well as to shear flow along the channel walls.

Microfabrication provided the ability to confine protein solutions in microchannels,

subject these proteins to flow, and control their adsorption patterns.

Controlling protein adsorption is critical to develop bioanalytical microdevices

involved in applications from tissue engineering to the cell sorters discussed previously.









Moreover, the ability to create well-defined patterns of protein adsorption allows

selective attachment of cells, which is of utmost importance in the biomaterials field.

Initial work in protein patterning was performed by Kleinfeld et al. in 1988 to manipulate

the motion of neurons [34]. The group used a method now called photochemical

patterning. The technique consisted of photolithographically patterning a substrate, then

washing the exposed area with a protein solution which preferentially binds to the

substrate and not to the remaining photoresist. The photoresist is then removed revealing

a substrate with a defined protein coating.

Photolithography techniques were also used to create hydrophilic and

hydrophobic areas inside microchannels [35] by patterning the interior of the protein-

coated channels. The ultraviolet light actually cleaves exposed protein molecules,

changing the proteins from hydrophobic to hydrophilic. When flow was then activated

through the channels, the fluid wetted only the patterned hydrophilic areas.

Folch et al. also patterned proteins on a substrate using rapid-prototyping

techniques with PDMS microchannels [36]. In their method the PDMS film was placed

channel side down over a substrate and a protein solution was activated through the

channels. Proteins from the solution absorbed to the substrate only. Removal of the

PDMS channels left a protein-patterned substrate ready for cell culture, biocompatibility

study, or other biological experiment.

2.3 Microfluidics

While a system of microchannels can be considered a microdevice, a microdevice

with fluid activation is considered a microfluidic device. These microfluidic systems have

many advantages over macroscale fluidic systems. Due to their high surface area to

volume ratio, they offer high rates of heat and mass transfer, making these systems









suitable as mini-heat exchangers [37] and since their dimensions match those of

biological samples microdevices may provide better manipulation and control of sample

movement [14,15,17,31,32-34,38]. Microfluidic devices also require a small sample

volume, which is favorable since reagents like blood, media, and antibodies are quite

expensive. Finally, their portability and rapid analysis performance make microfluidic

devices excellent for point-of-care service in clinical settings.

However, as device dimensions decrease, the classical laws of fluid mechanics

change as well. The critical dimensional parameter of a microfluidic device is the aspect

ratio, which is the ratio between the width and height of the channels [39]. As the aspect

ratio of microchannels decreases, the laminar friction constant increases [40]. This

friction increase can induce phenomena that do not occur in macroscale fluid flow. Some

of these phenomena are temperature variations through the system and transport effects in

directions other than the fluid flow direction [40]. Many groups have investigated

microscale fluid flow to explain microfluidic phenomena [40-44] and modify the

classical Navier-Stokes equations accordingly. However, for Newtonian fluids, where

the applied shear changes linearly with the fluid velocity, the classical Navier-Stokes

equations can be utilized in microfluidic systems for simplicity [26].

In addition to the friction coefficient increase, another parameter that changes with

aspect ratio is the Reynolds number, which decreases with decreasing aspect ratio. Re

indicates the tendency of a fluid to become turbulent or laminar. It is a ratio of inertial

versus viscous forces that act on a fluid. For definition, viscous forces scale in proportion

to the surface forces, A in units of L2 (where L stands for length), and inertial forces scale

in proportion to the body forces, V in units of L3. In the Reynolds number equation, Re =









pvD/j, p is the fluid density, v is the fluid velocity, D diameter of the microchannel, and P

the fluid viscosity. This dimensionless value can range as high as 105 in very turbulent

fluid flow, such as flow across an airfoil, or as low as 10-5 in laminar fluid flow, as occurs

in flow through small blood vessels [45]. In all microfluidic systems with microchannels

less than 1 mm in width, Re is less than 1. At this low Reynolds number, as L goes to 0

in Equation 2-1, viscous forces become much larger than inertial forces and the fluid has

no tendency to develop turbulence.

viscous forces ~ A L2
(2-1)
inertial forces ~ V ~ L3

It is this range, Re
2.4 Colon Cancer and Focal Adhesion Kinase

According to Cancer Facts and Figures 2003 (www.cancer.org) over 1 million

Americans will be diagnosed with cancer in 2003. About 150,000 of these cancers will

develop in the colon and rectum, as colorectal cancer is the third most common cancer in

American men and women. While this type of cancer is treatable if polyps are diagnosed

at an early stage, there is only a 9% survival rate that a person with secondary metastases

will live an additional 5 years. Therefore, investigation into a potential marker for the

beginnings of cancer is important to achieving new cancer therapies.

The biology of a cancer cell is different from a normal cell. A healthy cell would

normally be autoirradicated after detaching from the extracellular matrix (ECM) during a

process called apoptosis [47]. However, a cancerous cell actually escapes this process

and continues to mature. Investigating why a cancer cell escapes apoptosis is

fundamental in cancer research. A starting point to investigating the functional

differences between healthy and cancerous cells is to explore cell anchoring mechanisms,









or focal adhesions. Focal adhesions affect physical attachment to the ECM and vessel

linings, as well as regulate cellular functions such as growth, differentiation, migration,

and survival [48]. A way to study focal adhesion is to utilize cellular and molecular

biology techniques in combination with engineering techniques to correlate the effect of

protein expression on the adhesion and motility of the cell.

Ultimately, the microchannel system developed here will be used to study such a

correlation between protein expression and cell motility. The protein focal adhesion

kinase (FAK) and its effect on cell motility will be investigated. FAK is a protein

tyrosine kinase that serves in the regulation of the flow of signals from the ECM to the

action cytoskeleton. Since it is a regulator of the cytoskeleton, it is assumed that FAK

may play a role in cell motility as well [47]. It has been shown that the amount of FAK is

elevated, or overexpressed, in many cancers [49]. In the years following, FAK has been

shown to be overexpressed in breast, colon, thyroid, prostate, ovarian, and liver cancers,

and elevated in human papilloma virus Type 18 in human genital epithelial cells [47].

However, recent results from our lab show that though FAK may be overexpressed in

metastatic and invasive tumors, migrating cancer cells, before they attach to form a

secondary tumor, actually have decreased levels of FAK present. In other words, while

FAK may be over expressed in the invasive tumor, the motile detached cancer cells have

reduced expression of FAK. This difference in FAK expression between attached and

detached colon cancer cells has prompted studies that investigate the role FAK plays in

regulating cell function.














CHAPTER 3
MATERIALS AND METHODS

3.1 Channel Design and Specifications

To develop a microchannel device that benefits our current colon cancer study, as

well as future microfluidic projects and computational modeling, a variety of channel

designs were developed. These are smooth channels, channels with an endothelial cell

contour on the walls, bifurcated channels, and channels with a constriction along their

length. The smooth channels have straight and vertical sidewalls. These channels most

closely mimic micropipette experiments, with the exception that the cell will travel a

much longer length. The application of an endothelial cell contour to the walls of the

next set of channels offers a better physiological picture of how cells traverse through a

capillary, which is significant since it has been shown that surface patterning and

roughness affect cell growth and motion [50,51]. The bifurcated channels allow

observation of cell motion when traveling through a branched vessel and the constricted

channels provide observation of how cells maneuver through a stenosis or an occluded

vessel.

After determining the shape of the channels, the dimensions were considered. The

minimum width of the channels was limited by the masking making process of

microfabrication. In creating the mask, the minimum thickness with which a line of

photoresist could be deposited on the soda lime substrate was 3 am. Therefore, 3 am

became the minimum channel width for this project. The other channel widths were

chosen to accommodate a large range of widths. Small channels that would squeeze both









red blood cells (RBCs), which are approximately 8 [m in diameter, and white blood cells

(WBCs), which range from 10 [tm in diameter were designed. Also designed were larger

channels that would mimic arterioles and venules, which can be as small as 20 [tm in

diameter. The final channel types and widths are tabulated in Table 3-1.

Table 3-1. Summary of channel types and widths. Each channel is etched to 2 depths, 10
[tm and 25 tm.
Channel Type Straight Patterned Bifurcated Stenosed
3 3 10-5 10-5-10
5 5 25-12 25-12-25
10 10 50-25 50-25-50
12 12 100-50 100-50-100
15 15
20 20
Channel Width
(tm) 25 25
30 30
35 35
40 40
50 50
60 60
70 70


The channel length was chosen as 150 [tm to allow enough length for cell travel

and maintain a relatively low channel resistance. The final dimension selected was the

channel depth. Two depths were chosen to have a shallow depth that would promote the

squeezing of RBCs and WBCs, and a larger depth that would allow bulk flow through the

larger channels. The solution was to etch half the wafers to a depth of 10 [tm and the

other half to a depth of 25 tm.

The next step in designing the channels was specifying the method of fabrication

and characteristics of each channel part. The microdevice requires fluid access to and

from the chip, reservoirs to contain the fluid as it enters and leaves the channels, and the

microchannels, also called the channel device. A sketch of the chip is shown in Figure 3-










1. The input and output fluid ports would be etched completely through the wafer

thickness creating holes in the chip. These ports are square because a square mask

feature is cheaper to create than a circular feature. The ports are Imm across the

diagonal. This diameter was chosen because it matches that of the tubing that will be

used to pump fluid to and from the device. The ports are centered arbitrarily inside the

reservoirs.

The fluid reservoirs are simply a storage area for fluid before it is pushed into the

channels and after it leaves the channels. They are the same depth as the channels. In

addition, the reservoirs narrow sharply toward the chip center in order to direct the fluid

flow into the channels.

12 mm





Fluid
Input reservoirs Output fluid
fluid port port
-Channel
device
2500 2350
12imm


Figure 3-1. Sketch of the channel parts with dimensions.









The reservoirs were chosen arbitrarily to create a large reservoir into which the fluid can

fill. Circular reservoirs were chosen to prevent the fluid from making voids in corners,

which would occur in square reservoirs.

To complete the design, five channels will be etched on each chip so that the

device is useful for multi-cell analysis. Furthermore, it allows the device to still be useful

if cells were to lodge in a channel. If one channel becomes blocked, then there will be

four more channels through which cells can travel and four more channels that can be

observed. In addition, a scale bar was drawn above and below each the channel device.

The bar consists of 3 pm wide rectangles that are spaced 3 pm apart. This allows

measurement of cell length and travel distance.

3.2 Computer Assisted Drawings

The 2 dimensional layout, or mask, of the channels was generated using a

software package called CoventorWareTM (Coventor, Inc., Cary, North Carolina). The

software converts the layout into GDSII format, which is a commonly accepted format

for mask making. The Coventor images for both masks are shown in Figure 3-2 and the

Coventor images of the channel configurations are shown in Figure 3-3.























A B---


Figure 3-2. Coventor images of the masks used during photolithography. A) Channels
mask. B) Ports mask. The images show the 34 devices per wafer and the
circle around each mask indicates the silicon wafer.


Adk+ A+kk
*+MW 14W


150 pm


150 pm


100 tpm


200 pm


Figure 3-3. Coventor drawings of the channel configurations. Consider the flow as
moving from left to right through the pink areas. A) Smooth channels, B)
endothelial cell contoured channels, C) bifurcated channel, D) stenosed
channel with recovery reservoir.

3.3 Materials

In order to achieve the best dimensional detail the fabrication method chosen was

silicon wafer etching. Silicon wafer processing contracted through MEMS Exchange


4b- a-* W-0 --









(www.mems-exchange.org, Reston, VA). MEMS Exchange is a microfabrication

network that acts as intermediary between the fabrication provider and the purchaser.

The silicon wafers were provided by MEMS Exchange and the masks were fabricated at

Photronics, Inc. (Brookfield, CT). The masks and wafers were then shipped to the Solid

State Electronics Lab (SSEL) at the University of Michigan in Ann Arbor for processing.

The SSEL is a part of the Department of Electrical Engineering and Computer Science.

The wafers were 100 mm diameter, double-side polished, (100) n-type silicon

wafers. Using double-sided wafers prevented any pattern distortion due to a rough wafer

backside, though the potential distortion with this two mask process was minimal

regardless. In addition, support wafers were also used to strengthen the wafer during the

long DRIE of the fluid ports. These wafers were removed for reuse as test wafers for

other projects.

The process required two masks named Channels and Ports. Each mask was dark-

field and made of soda lime glass with chromium coating. The Channels mask included

the 2-D, top-down contour of the reservoirs and channels, while the Ports mask consisted

of the 2-D squares comprising the ports. Due to the number of non-Manhattan (non-

rectangular) features, the Channels mask was created using an electron beam process,

while the Ports mask was fabricated using the standard pattern generation process.

The resists used during photolithography were Shipley 1813 (Shipley Company,

L.L.C., Marlborough, MA) and AZ P9260 (MicroChemicals, Ulm, Germany). Both are

positive resists that become soluble when exposed to ultra violet light and are removed by

the developing solution.









The chip packaging was made out of aluminum. Aluminum was chosen because

the metal is cheap, mills easily, and is adequately sturdy for this application.

3.4 Process Flow for Microchannel Fabrication

Refer to Figure 3-4 through Figure 3- below for step-by-step cross-sectional

pictures of the wafer. The compact process flow is provided in Appendix A. The cross-

section used is indicated by a corresponding figure. For pattern transfer and etching of

the Channels mask the cross-section is shown in Figure 3-4 below.

Mask 1
Wells & Channels






Figure 3-4. Reference cross-section at the middle of the channels.

The etch process began with a pre-furnace cleaning which included removal of lint

particles followed by a dip in hydrofluoric acid (Figure 3-5A). Next the wafers

underwent wet oxidation to grow a 0.5 [tm layer of silicon dioxide (Si02) (Figure 3-5B),

which was followed by spectroscopic ellipsometry film thickness measurement. The

Si02 acts as a nested mask for the DRIE in the proceeding steps.

To apply the pattern of the reservoirs and channels to the silicon wafer, contact

photolithography of the Channels mask was performed. First the wafers were dehydrated

in an oven and coated with a primer called hexamethyldisilazane (HMDS), which

improved the adhesion between the Si02 and photoresist (step not shown in figures). 1.3

[tm of Shipley 1813 photoresist was spun over the wafers (Figure 3-5C), which were then

put in the oven for a softbake. Using an EV420 Series Mask Aligner (Electronic Visions

Co., Pheonix, Arizona) the Channels mask was aligned over the front-side of the coated









silicon wafers. The wafers were then exposed to ultra violet light for final pattern

transfer (Figure 3-5D). After exposure the wafers were baked at 110C to strengthen the

unexposed resist and then placed in developing solution to remove the exposed

photoresist. In a etch process called a buffered oxide etch, the wafer was dipped in a

buffered hydrofluoric acid (HF) solution, which etched the SiO2 only (Figure 3-5F). The

wafers were then placed in an oxygen plasma environment to ash away the exposed

photoresist (Figure 3-5F).


(A) (D)


(B ) ( ) "
(E)


E 0 1


Figure 3-5. Initial wafer processing and photolithography of Channels mask. A) Clean
silicon wafer, B) grow the Si02, C) spin photoresist, D) expose Channels
mask, E) develop, F) buffered HF etch through Si02 and photoresist removal.

After the channels and reservoir pattern was transferred to the wafers by

photolithography, the channels and reservoirs were etched into the silicon wafer by DRIE

(Figure 3-6). In order to fabricate two sets of chips with two different channel depths,

one wafer was placed in the STS Multiplex ICP (Inductively Coupled Plasma) Deep

Reactive Ion Etcher (Surface Technology Systems, Redwood City, California) until an

etch depth of 10 3 [tm was reached, while the other remained in the etcher until an etch

depth of 25 3 [tm was reached. (Only one etch depth is illustrated in the figures). To

complete the front-side etching of the channels and reservoirs, the remaining layer of

photoresist was removed by ashing, while the silicon dioxide layer remained.











25 micron


Figure 3-6. DRIE of channels.

At this point the wafers were flipped over for backside etching of the fluid access

ports. For backside wafer processing description, the cross-sectional reference is

changed and shown in Figure 3-7.

Mask 3
Ports







Figure 3-7. Reference cross-section mid-wafer through the fluid access ports.

Photolithography of the Ports mask (Figure 3-8A) was the same as the

photolithography of the Channels mask, except that AZ P9260 photoresist was used

instead of Shipley 1813. AZ P9260 was applied in a 12-[m thick layer, which protects

the silicon wafer during the lengthy through-wafer DRIE of the ports. After developing,

the SiO2 was etched by buffered HF solution leaving the silicon underneath exposed

(Figure 3-8B).


I__ (A)

25 micron



r 7 (B)

Figure 3-8. A) Photolithography and B) buffered HF etch of the fluid access ports.









After the Ports pattern was transferred to the backside of the wafers, support wafers

were bonded to the front-side of the wafer (the side where the channel features were

etched) by applying a thick coating of photoresist and pressing the wafers together

(Figure 3-9A). Resist bonding consists of a dehydration bake of the wafers, HDMS

prime, Shipley 1827, photoresist coat, wafer attachment, and a hardbake at 110C.

Attaching the support wafers provides stability to the silicon wafer during etching and

acts as a safeguard against possible wafer breakage. The fluid access ports then

underwent DRIE down about 475 pm to create holes through the entire thickness of the

wafer (Figure 3-9B).


Device wafer device wafer
(A) M (B)


support wafel support wafer

Figure 3-9. A) Resist bond of support wafer to front-side of silicon wafer and B) DRIE of
fluid access ports.

For final wafer processing, the support wafer and photoresist coating were both

removed by exposure to oxygen plasma (Figure 3-10A). The remaining SiO2 was

removed by buffered HF solution leaving the finished and processed wafer cross-section

(Figure 3-10B). In addition, the finished wafer, using the original wafer cross-section, is

shown in Figure 3-10C.










(A) (B)







Figure 3-10. Removal of A) support wafer and B) remaining silicon dioxide. C) The
finished wafer from the mid-channel cross-section.

The last step was to cleave the wafers using a scribe to release the 34 chips per

wafer (Figure 3-11). Scribing was performed by nicking the edge of the wafer to create a

defect. Care was taken to nick the wafer at a 90 angle to the (100) lattice orientation.

After scribing, the wafer was snapped at the defect point to propagate a crack across the

wafer and cleave it in two. Due to the lattice orientation, if cleaved properly, the newly

formed edges were perpendicular and straight across the wafer.



j -3 -IE-| .< j.h \


F- E [i': 72E Id -" E L /


S___- it'- .T /




Figure 3-11. Thirty-four chips cleaved from each wafer. The number refers to the channel
width in [m, "E" refers to endothelial cell contour, "bif' refers to bifurcated
channel, and "sten" refers to stenosed channels. The arrows indicate cleavage
points.









3.5 Chip Cleaning

After cleavage the chips were cleaned in a series of solvents. Each chip was

placed into a 15 mm outer diameter glass test tube (Fisher Scientific, Pittsburgh, PA).

The tubes were placed in a test tube holder and inserted into an ultrasonic bath (Branson

Cleaning Equipment Co., Shelton, CT). The tubes were then filled with subsequent

solutions oftetrachloroethylene (J.T. Baker, Phillipsburg, NJ), acetone (Fisher Scientific,

Pittsburgh, PA), methonol (Fisher Scientific, Pittsburgh, PA), and deionized water. After

2 mL of each solution was introduced, the ultrasonic bath was turned on for 5 minutes.

The solutions work as degreasers to remove oils from the surface of the chips. The

ultrasonic bath emits a rapid pulsatile effect to effectively "scrub" the chips. The final

soak in deionized water removed any residual contaminants. After cleaning the chips

were placed in labeled chip carriers (Fluoroware, Inc., Chaska, MN) for storage.

To study different sealing methods, two sealing methods-- pressure and plasma

surface modification-- were used. For the pressure sealing method, the chip and cover

slip were pressed together by the packing tension arms discussed in 3.6. This sealing

method is not permanent. In the second sealing method a permanent seal is created. As

reviewed in the literature [52], the plasma surface modification causes an energy transfer

from high-energy gaseous species to surface molecules. Evacuating a vessel, purging it

with argon gas, and energizing the gas with radiofrequency energy creates the high-

energy species. After exposure, the silicon chip and PDMS-coated cover slip were then

brought into contact and pressed together. The new silicon-PDMS interface forms

permanent Si-O-Si bonds which creates a tight, irreversible seal.








3.6 Packaging

A rigid assembly shown in Figure 3-12 and Figure 3-13 below was fabricated in

order to use the chip for fluid study. The assembly not only stabilized the silicon chip,

but also provided the fluid access to the microchannels. The parts of this package were

the aluminum base, 2-3/64" x 9/64" Buna-N o-rings (McMaster-Carr Supply Co.,

Atlanta, GA), the chip itself, a PDMS coated 22 mm x 22 mm x 0.2 mm unbreakable

plastic cover slip (Fisher Scientific, Pittsburgh, PA), and two tension arms.


Figure 3-12. Packaging and fluid access of chip showing the aluminum base, o-rings and
o-ring pockets, chip, cover clip, and tension arms.


Figure 3-13. Chip package including base, chip, cover slip, and tension arms.

The base is an aluminum rectangle that is 5 cm x 3 cm x 9 mm thick with a 22 mm

x 22 mm x 1 mm square hole, called the well, milled into its center. This well is where


on :,









the o-rings, chip, and cover slip sit. To create the input and output ports in the base, 1

mm diameter holes were drilled into opposite sides of the base. These holes stop directly

under the location where the fluid ports of the chip will lie. Holes 2 mm in diameter were

then drilled from the well down to meet the 1 mm diameter holes. A 3 mm diameter hole

was then drilled over the 2 mm diameter ports to create pockets where the o-rings will sit.

These holes complete the fluid pathway in the base. A dimensional drawing of the base

is included in Appendix B.

After the o-rings, chip, and cover slip are placed into the well, the two tension

arms swings directly over the port holes to engage the o-rings and create a sealed fluid

circuit. After the seal is created, 0.62 in inner diameter Silastic laboratory tubing

(GlycoTech, Rockville, MD) connected to the fluid ports in the aluminum base by luer

locks (ArkPlas Products, Flippin, AR). Such a packaging assembly allows chips to be

changed easily with a ready fluid connection. The assembly is then ready to be placed on

the microscope stage.

3.7 Experimental Set-Up

The data acquisition set-up is shown in Figure 3-14. The fluid is introduced to the

chip by a Pump 33 Dual Syringe Pump (Harvard Apparatus Inc, Holliston,

Massachusetts), which provides a closed fluid circuit. Using a user defined flow rate in

[l/ min, one syringe pushes solution through the tubing, while a second syringe pulls the

solution.





















Figure 3-14. Data acquisition set-up including fluid syringe pump, microchannel chip,
microscope, and computer for image acquisition.

For preliminary testing the solution used was a 1:12 ratio of whole blood and

Hanks Buffered Saline Solution (HBSS) (Fisher Scientific, Pittsburgh, PA). Prior to

inducing the blood or saline solution, the tubing, base, and channels were flushed with a

4% poly(vinylalcohol) solution (Sigma Chemical, St. Louis, MO). This surfactant

solution primed and wetted the tubing and channels to provide a hydrophilic coating over

which the blood solution will flow. Priming the channels will prevent the blood cells

from sticking to the inside of the tubing and separating out of solution. After testing, the

flow system was sterilized by flushing with a series of diluted bleach solution, deionized

water, and, lastly, ethanol.

Images are acquired with an AxioPlan2 IE Manual microscope (Carl Zeiss

Microimaging, Inc., Thornwood, New York) set-up for incident light transmission. The

image acquisition software was AxioVision 3.1. (Carl Zeiss Microimaging, Inc.,

Thornwood, New York).














CHAPTER 4
MICROCHANNEL CHARACTERIZATION

In order to obtain a first order approximation between pressure drop across the

channels and flow rate, the case of the flow through a rectangular channel is used and the

following assumptions are made.

* Newtonian fluid presence of blood cells is neglected

* Fully developed flow

* Incompressible fluid

* No-slip boundary condition

* 2-D flow

Under these conditions a closed form solution to the governing flow equations can be

derived [39]. The labeled cross-section is shown in Figure 4-1.

Y





2a 0 --



-a< y
Figure 4-1. Labeled cross-section of rectangular microchannel.

The equation for flow rate, Q, through a rectangular channel is

S4ba3 -dp 192a tanh(ir b/2a)
Q = 3 dx 5 (4-1)
3,u dx z b 1=1,3,5 1









which can be written as


Q = (4-2)


which states that flow rate is proportional to pressure drop. Constant C1 is a function of

the channel geometry and fluid viscosity. It was solved in MatLab (The MathWorks,

Inc., Natick, MA) and are tabulated in Table 4-1. The MatLab program for calculating

C1 is provided in Appendix C. For the calculation, viscosity, [a, was 1.2 cp, the viscosity

of blood plasma.

Table 4-1. Constant C1. The channel width and height is specified.
Channel Height (tm)
10 25
CWidth C
Width ([im) 10 -20 (m4/Pa-s) 10-20 (m4/Pa-s)

3 1.52 4.33
5 5.96 19.0
10 29.3 130
12 41.5 210
15 61.2 367
20 95.3 716
25 13.0 114
30 165 1620
35 200 2130
40 235 2650
50 306 3720
60 378 4810
70 451 5900


dp
Next the pressure gradient, can be calculated using Uavg. By definition
dx


Uvg A (4-3)


Plugging (4-2) into (4-3), the expression for is obtained.
dx









dp whU (4ag 4)
-- (4-4)
dx C1

where w = channel with and h = height.

dp
From the rectangular channel geometry and assuming Uav,,g = 25 am/sec, can be
dx

calculated. This average velocity is chosen because it allows a cell to reside in the 150

tm long channel for 6 seconds, which is an adequate residence time to observe and

dp
record the cell motion. Assuming that is constant across the channel length,
dx

pressure drop, AP, which is the driving force for fluid flow can be calculated. As shown

in Equation 4-5, AP is the product of the pressure gradient and the total channel length, L

(L = 150 tm).


AP= dpL (4-5)
dx

Furthermore, channel resistance, R, is the driving force divided by the flow rate as shown

in Equation 4-6.

AP
R = (4-6)
Q

The values of pressure gradient, flow rate, pressure drop, and channel resistance are

provided in Table 4-2 for the two channel heights.











dp
Table 4-2. Values for Q, AP, and R.
dx
Channel Height
(pm)
10 25
Width dp dp/

(uWi) / Q AP R d Q AP R
(Pa/m) (m3/s) (Pa) (Pa s/m3) (Pa/m) (m3/s) (Pa) (Pa s/m3)


-4.95x104

-2.10x104

-8.53x103

-7.22x103

-6.13x103

-5.24x103

-4.81x103

-4.55x103

-4.38x103

-4.26x103

-4.09x103

-3.97x103

-3.88x103


7.51 11(kl

1 5\111"1'

2.50x10'15

3.00x10'15

3.75\111"1

5.00x10'15

6 25\111 1'

7.50x10-15

8.75\111 1

1.00xl0 14

1.25x10-14

1.50x10 14

1.75x10-14


-7.42

-3.15

-1.28

-1.08

-.919

-.787

-.721

-.682

-.657

-.638

-.613

-.596

-.582


-9.89x1015

-2.52x1015

-5.12x1014

-3.61x1014

-2.45\ 1111

-1.57x1014

-1 15\1111

-9.10x1013

-7.51x1013

-6.38x1013

-4.90x1013

-3.97x1013

-3.33x1013


-4.33x104

-1.64 x104

-4.81 xl03

-3.58 x103

-2.56 x103

-1.75 x103

-1.37 x103

-1.16 x103

-1.03 x103

-9.42 x102

-8.39 x102

-7.80 x102

-7.42 x102


1.88x10
15
3.13 x10
15
6.25 x10l
15
7.50 x10l
15
9.38 x10l
15
1.25 x10l
14
1.56 x10l
14
1.88 x10
14
2.19 x10
14
2.50 x10
14
3.13 x10
14
3.75 x10
14
4.38 x10
14


-6.49

-2.47

-0.722

-0.537

-0.383

-0.262

-0.205

-0.173

-0.154

-0.141

-0.126

-0.117

-0.111


-3.46x1015

-7.89x1014

-1.15 x1014

-7.16 x103

-4.09 x103

-2.10 x103

-1.31 x103

-9.24 x1012

-7.04 x1012

-5.65 xl012

-4.03 x1012

-3.12 x1012

-2.54 x1012


In addition to calculating the pressure drop and resistance across the channels, we


can also solve for the Reynolds Number, Re, based on hydraulic diameter, Dh, of the


channels. ReDh was discussed in 2.3. It is the ratio between the inertial and viscous


forces affecting the fluid flow. Dh is a factor used to approximate a non-circular channel

to a circular channel.


4A
D,h = (4-7) and ReD,
P


PUagDh (4-8)
/P









In Equation 4-7, A is the cross-sectional area of the channel and P is the wetted

perimeter. In Equation 4-8, p is the fluid density and u is the fluid viscosity. The values

for Dh and ReDh are listed in Table 4-3.

Table 4-3. Values for wall shear stress, hydraulic diameter, Reynolds number, and
Poiseulle number.
Height
(Iam)
10 25
Width h 10-4 ReDh Dh 10-4 ReDh
(Im) (Im) (Im)
3 4.62 1.15 5.36 1.15
5 6.67 1.43 8.33 1.79
10 10.0 2.15 14.3 3.07
12 10.9 2.34 16.2 3.48
15 12.0 2.58 18.8 4.02
20 13.3 2.86 22.2 4.77
25 14.3 3.07 25.0 5.36
30 15.0 3.22 27.3 5.85
35 15.6 3.34 29.2 6.26
40 16.0 3.43 30.8 6.60
50 16.7 3.58 33.3 7.15
60 17.1 3.68 35.3 7.57
70 17.5 3.76 36.8 7.91

Now that RDh has been calculated for each channel we can calculate the entrance

length, Le, for the channels to determine the fully developed region of fluid flow [53].

L R
Le (4-9)
d 30

where d is the channel depth. The entrance results using only the 3 tm and 70 tm wide

channels are shown in Table 4-4. RDh values are obtained from Table 4-3.

Table 4-4. Le (in tm) for 3 tm and 70 tm wide channels.
Channel depth (tm)
10 25
Channel 3 3.83x10-5 9.58x10-5
width 4 4
width 70 1.25x10-4 6.59x10-4
Gan)>






42


As expected for systems with very low Reynolds number, the entrance lengths for the

channels are so small that they can be neglected. The flow becomes fully developed

immediately as fluid enters the channels.














CHAPTER 5
RESULTS AND DISCUSSION

5.1 Microchannel System

The microchannel system developed satisfies the objective to develop rigid

microchannels for the study of microcirculation and the characterization of cell

theological properties, function, and behavior. It has been shown that microfabrication

techniques provide the geometric control and minimum dimensions required to create a

microfluidic network. The microchannel system isolated the extensive microcirculation

into three parts to study the effect of vessel size, vessel geometry, and vessel surface

roughness on blood flow. To isolate vessel size, straight channels of various widths were

designed. To isolate vessel geometry, bifurcated and stenosed channels were designed.

And to isolate vessel surface roughness, another set of channels was designed with a

bumpy contour modeled after endothelial cells applied to the channel walls. This chapter

discusses the function of the current microchannel system. The strengths and weaknesses

of the system are discussed and suggested improvements to a future microchannel system

are provided.

5.1.1 Fabrication of microchannels

Fabrication of the microchannels was successfully performed at the Solid State

Electronics Laboratory at the University of Michigan. Pictures of the four types of

channels are shown in Figure 5-1. These 2-dimensional pictures show that the channel

configurations were etched sharply.












*IaIII.II....s.I.. gI


( 10\l i l 10\








Figure 5-1. Selected pictures of microchannels. A) 25 tm wide straight channels. B) 12
[m wide endothelial cell contoured channels. C) 25 [m wide bifurcated
channel. D) 25 am wide channel that constricts into a 12 am wide channel.

The scale bar, however, did not etch with good integrity. The rectangles were


designed to be 3 am wide, but were etched laterally to a width of about 5 am. They offer


a rough estimate of cell length and travel distance, but not as accurate a scale as their


design intended.


While 2-dimensional channel images were obtained with incident light microscopy,


3-dimensional images were obtained with a Wyko NT 1000 Profilometer (Veeco


Metrology Group, Tucson, AZ). Surface topography images of the channels types are


shown in Figure 5-2 through Figure 5-7. The scale to the right of each image correlates


the depth variation with color change.


~.....rriLrC;i;;.

~.~~~.;;; ~~~.;;;;;;;-;;r;rlli.
~~~;;;;;;;~I~C~

~r~lY1.
----------~- -urr---

.1-*11**

































Figure 5-2. Surface profile of the 50 [tm wide straight microchannels.
a clean etch down to the channel floor.


- 14.7


- 12.0

-10.0

- 8.0

- 6.0

-4.0

2.0

-0.0


-2.8


The image shows


um
S12.9



5.0


O.10


-5.0


-1o.0


-15.0

-19.5


Figure 5-3. Surface profile of the 12 [tm wide endothelial cell contoured microchannels.
The orange color at the top of the figure shows that the surface was etched
unevenly. The top of these channels slope down approximately 4 tm from
left to right across the surface.




































Figure 5-4. Surface profile of 100 [m wide channel that narrows into
stenosis.


S5.2

3.0


0.0


-3.0


-6.0


-9.0



S-12.7



a 50 [tm wide


1.2


-1.0


-3.0


-5.0


-7.0


-9.0


-11.0



-14.1


Figure 5-5. Surface profile of the 10 [m wide bifurcated channel.









The dimensions obtained from the surface profiles show good agreement with the

design dimensions with the exception that the 10 ptm deep channels were over-etched to a

depth of 16 pm. In addition, the reservoir area was etched slightly deeper than the

channels, due to the greater surface area exposed to the etchant system. Figure 5-5 shows

the limitation of the profilometer, in that depth measurement of tight spaces can be

difficult to obtain for reasons such as scanning too large a surface area or inadequate

vibrational isolation.

However, the feature that is best illustrated by the profilometer images is the profile

of the channel walls. As shown in Figure 5-6 and Figure 5-7, the 10 [pm wide walls that

separate each channel are not vertical, but are thicker at the base than at the terrace. The

perspective views of the 3 [pm and 70 [pm wide channels show that the angled walls form

channels that are not rectangular, but are actually slightly v-shaped. In addition, the wall

terrace is not straight and smooth across the top, but is rough and, in addition, is etched

approximately 1 jpm to 2 jpm below the surface of the chip. This roughness may affect

the sealing of the channels prior to and during flow activation by preventing good contact

between the PDMS-coated cover slip and the surface of the channels. Channel sealing is

discussed in 5.1.2.









um
top surface of chip 8.1

terrace -
e5.0
channel


ski& 0.0

5.0


10.0

-13.4


Figure 5-6. Perspective view of the 3 pm wide channels.

um
14.7
top surface of chip

terrace 12.0

channel 10.0
A8.0
a-a- a m a a
6.0

-4.0

2.0

0.0


-2.8

Figure 5-7. Perspective view of the 50 pm wide channels.

It is noted, however, that the apparent surface roughness of the microchannels

could be a artifact of vibrations during the optical surface scan of the profilometer. A

scanning electron microscopy micrograph would provide a better picture of the true

microchannel chip surface.









5.1.2 Flow system

Fluid flow was successfully activated through the microchannel system. The

channels used to test the flow system were 3 large width and 3 small width channels.

These were the 70 [am, 50 [am, & 50 [am patterned channels and the 10 [am, 5 [am, & 5 Lam

patterned channels. In addition, flow was activated through the stenosed and bifurcated

channels as well as shown in Figure 5-8 and Figure 5-9. Figure 5-9 is included to show a

non-wetted area of the channels. The channels are primed with poly(vinyl alcohol)

solution, which coats the walls with a hydrophilic layer over which the solution

preferentially flows. The system was tested at various flow rates to determine when

leakage would occur. Leakage occurred at either the o-ring seal underneath the chip or

between the chip and the cover slip as the fluid filled the input reservoir.






.-









Figure 5-8. Fluid activation through a 50 [am wide channel that constricts to a 25 jam wide
stenosis.






50





















Figure 5-9. Fluid flow through the 50 pm wide bifurcated channel showing a non-wetted
area.

As expected, flow was activated easily through the larger channels. Figure 5-10A

and Figure 5-10B below shows blood flowing through the 50 [m wide channels at two

magnifications. For these channels, the tension arms on the packaging provided enough

pressure to engage the o-rings and create a no-leak seal up to the maximum flow rate of

125 kl/min.


V *'~
















.d i












Figure 5-10. 50 tm endothelial cell contoured channel at A) 10x and B) 20x.

While pressure sealing using the tension arms was effective for the larger width

channels, effective pressure sealing was not effective with the smaller channels. When

flow was activated through the pressure sealed 5 [tm patterned channels using a flow rate

of 15 al/min, the system leaked slowly over time at both leakage sites. An image of these

5 [tm patterned wide channels is shown in Figure 5-11. The image shows the blood cells

forming blood clots inside the channels, as indicated by the arrows. These clots most

likely caused the fluid to backup and leak at both the input port o-ring and in the input

reservoir between the chip and cover slip. However, though the sealing of this small

channel was ineffective, it shows that indeed cell deformation can be observe with the

microchannel system, which partially satisfies the objective of this system.

Sealing the 5 [tm wide channels under argon plasma did not provide a more

effective seal than the pressure seal. Figure 5-12 shows an image of the plasma treated 5

[tm wide channels. The arrows indicate areas where fluid is actually traveling over the

channels, not through the channels. So in effect, this system did not leak at the o-rings or






52


in the reservoir, but rather it leaked at the seal above the channels themselves. This

leakage indicates that the 10 [tm wide walls separating each channel did not provide

enough surface area for the plasma treatment to oxidize the surface groups and create a

tight bond with the PDMS-coated cover slip.


bloo I o,'


ANN .





Figure 5-11. 5 tm wide endothelial cell contoured channels with blood flow.




Fluid flowing over the
channels instead of -
through the channels




20x

Figure 5-12. 5 micron wide channel showing fluid traveling over the channel devices.

In addition, the PDMS-coated cover slip does not provide a good seal over the 10

[tm wide channels either. The picture in Figure 5-13 shows single cell flow through the

10 tm wide channels. This channel was sealed with pressure by the tension arms. This

image was obtained after the fluid system leaked and the cover slip was removed.











n- "" l .. .


St.5
Sf


me4~


dlli W '' tIoil


Figure 5-13. 10 [m straight channel with single cell flow.

Finally, images of the flow channel showed that even though rounded fluid

reservoirs were chosen, voids still occurred (Figure 5-14). Though voids did not hinder

the function of the flow channels, modification of the channel design, as discussed in

5.2, would prevent their occurrence.



V.. a .a*.a.,


1*




0k


fluid
void


Figure 5-14. Input port of 50 am wide flow channel showing a fluid void on far right.









5.1.3 Imaging

The maximum magnification obtained was 20x. The 40x objective was not used

because its working distance was not large enough to focus on the channels through the

PDMS-coated cover slip.

5.2 Improvements to Microchannel System

There are a few opportunities for improvement of this microchannel system in the

areas of sealing and imaging. Sealing problems encountered in the current design could

possibly be prevented in three ways. First, the silicon dioxide film thickness should be

increased from 0.1 [tm to 0.5 [tm. This increase in thickness should better protect the

microchannel features from over etching (Figure 5-6 and Figure 5-7). Second, the wall

separating each channel should be at least 20 [tm wide, instead of 10 [tm wide. This

increased separation distance will provide greater surface area with which PDMS-coated

cover slip can bond under pressure or after plasma treatment. Third, a polysilicon RIE

should be considered as an alternative to DRIE. Polysilicon RIE is a slower process that

would etch sharper features, compared to DRIE, which is a faster etching system that

may have caused the jagged channel tops and angled side-walls.

An alternative bonding method is anodic bonding. Anodic bonding consists of

placing a 500 [tm Pyrex 7740 glass wafer (Coming, Corning, New York) over the silicon

wafer and applying a voltage of 300-700 V at 500C across the assembly [2]. The

positive voltage causes sodium ions in the Pyrex glass to migrate away from the

glass/silicon interface leaving a net negative charge on the glass surface. The opposing

positively charged silicon surface is strongly attracted to the positive glass and the two

surfaces chemically fuse together irreversibly. In addition to providing a better sealing









method, anodic boding would also provide clearer pictures due to the optical quality of

the glass.

Another design modification to improve sealing is to replace the tension bars with

a plate and four bolts as shown in Figure 5-15. The plate would provide equal pressure

on all sides of the microchannel chip and tightening with a torque wrench would offer

repeatable pressure acquisition.

To preclude the formation of fluid voids, the size of the reservoirs should be

minimized so that the reservoir wall makes a small rim around the port opening. The

improved channel schematic is shown in Figure 5-16. In addition, the size of the chip

could also be enlarged to match the size of the cover slip. This will help in aligning the

chip to the part holes. Enlarging the chip, however, also means that fewer chips will be

diced out of one wafer. In addition, the chips should be aligned in a square of straight

columns and rows so that the dicing saw can be used to sacrifice the chips from the

wafer. The dicing saw would cut perfectly square chips, which would make aligning the

chips to the port holes in the fluid system even easier.





































Figure 5-15. Schematic of improved base with plate instead of tension bars.

L-- 18 mm-


Input
fluid port


Fluid
reservoirs


Output fluid
i port


18 mrn


Figure 5-16. Schematic of improved channel parts with no reservoirs.

Finally, the success is partly determined by the quality of the images obtained.

Currently, the only objective utilized was the 20x objective. However, imaging the






57


system using a LD ACHROPLAN 40x objective with a 0.6 mm working distance (Carl

Zeiss Microimaging, Inc., Thornwood, New York), would focus through the PDMS-

coated cover slip and provide more detailed pictures of fluid flow and cell deformation.

This 40x, or even a 60x long working distance objective, is crucial to improving image

collection and to creating a successful microchannel system.














CHAPTER 6
CONCLUSION

6.1 Effectiveness of Microchannel System

Rigid microchannels for the study of blood microcirculation and the

characterization of cell theological properties, function, and behavior were successfully

designed and fabricated. Using microfabrication techniques, sufficient dimensional and

geometric control was achieved. The silicon etching process developed allowed

fabrication of a network of microchannels for the study of microcirculation. Using this

process, microchannels that provide an in vitro method to isolate the varying vessel size,

geometries, and surface roughness comprising the microvascular network were

fabricated. Bifurcated, stenosed, smooth, and endothelial cell patterned channels were

created, each mimicking a facet of the microcirculation. The endothelial cell patterned

channel also provides a means to subject viscoelastic fluids to compression and

relaxation.

The microchannel fluid system was shown to be successful as well. Imaging

showed that blood cell deformation was observed using the microchannel system. These

images extend deformational information offered by micropipette experiment, in that

they provide observation of cellular response to the endothelial lining of a vessel and,

furthermore, provide information on cell migration through branching or occluded

vessels.









6.2 Future Work

Immediate work will begin to improve the sealing and imaging capabilities of the

microchannel system. A 40x long distance working objective will be obtained to acquire

images of higher power and provide better observation of deformation and motility. In

addition, Pyrex glass cover slips should be anodically bonded to the top of the

microchannel chips to create permanent and irreversible seals. Tight sealing may allow

larger flow rates that will prevent cells from clogging the channels. Moreover, obtaining

higher flow rates through the channels would provide more physiologically relevant

results.

In addition, the microchannel system developed will be used to study cell

deformation and motility as affected by protein expression in the growth and spread of

colon cancer. Cells from a well-defined 293 fetal kidney epithelial cell line, the control

cell line for the current colon cancer study, will be introduced into the system to study the

role that focal adhesion kinase (FAK) plays in cell deformation and motility.

The system can also be used to begin a study on the flow through porous media.

The endothelial cell lined channel can be used to observe the compression and relaxation

of viscoelastic fluids. This compression and relaxation provides a first order model for

the type of flow created in foam or a porous system. Compression/relaxation models are

used in the development of absorbent materials, such as diapers and paper towels.















APPENDIX A
PROCESS FLOW


Description


Step


Starting material

1. Clean
2. Oxide
3. Photolithography


4. Buffered oxide etch
5. DRIE

6. Ash
7. Photolithography








8. Buffered Oxide Etch


9. Resist Bond





10. DRIE
11. De-mount support wafer
12. Ash
13. Bufferedn Oxide Etch


100mm diameter double-side polish (100)
n-type silicon
Standard RCA clean with HF dip
Wet TCA oxidation 0.1 [tm SiO2
Dehydration bake
HDMS prime
Photoresist coat 1.3 [im (Shipley 1813)
Photoresist software
Contact front-front align Mask 1
I-line exposure
Photoresist develop
Photoresist hardbake (110 C)
Etch Mask 1 features into oxide (0.1 .im)
Etch Mask 1 features into silicon (10 tim
and 25 [tm)
Oxygen plasma to remove photoresist
Dehydration bake
HDMS prime
Photoresist coat 12 [tm (AZ P9260)
Photoresist softbake
Contact front-front align Mask 2
I-line exposure
Photoresist develop
Photoresist hardbake (110 C)
Etch Ports features into oxide to create
alignment marks for backside through-
wafer etch of Ports
Dehydration bake
HDMS prime
Photoresist coat (Spoke patterb Shipley
1827)
Resist bond
Photoresist hardbake (110 C)
Through-device-wafer etch of ports
Oxygen plasma to remove photoresist
Oxygen plasma to remove remaining PR
Remove remaining oxide

















APPENDIX B
BASE DIMENSIONS





U! t'EGO Ct



.C








alI I




66
-r-- ----








a,


L _

















61


It


-p


















APPENDIX C
MATLAB CODE


%SOLVE FOR cl/viscosity for 25
micron deep channel%


Cl=[0 0 0 0 0 0 0 0 0
0] '
a=[1.5 2.5 5 6 7.5 10
17.5 20 25 30 35];
b=12.5;


000

12.5 15


for k=1:1:13
sum=0.0

for i=1:2:5

res=(tanh(i*pi*b/2/a(k)))
/powe r(i,5);
sum=sum+res;
end

Cl(k)=4*b*a(k)"3/3*
(1-192*a(k)/pi^5/b*sum)


%SOLVE FOR c2/viscosity for 25
micron deep channel%

C2=[0 0 0 0 0 0 0 0 0 0 0 0]'
a=[1.5 2.5 5 6 12.5 15 17.5 20 25
30 35]
y=2*a
b=12.5
z=2*b

for k=1:1:13
sum=0.0

for i=1:2:5
res=(-l)^((i-1)/2)*(l-
(cosh(i*pi*z/2/a(k)))/(cosh(i*pi*b
/2/a(k))))*cos(i*pi*y(k)/2/a(k))/i
^3
sum=sum + res
end


C2(k)=16*a(k)^2/pi^3*sum
















LIST OF REFERENCES


1. Martini, F.H., Fundamentals ofAnatomy and Physiology. 2001, Upper Saddle
River, New Jersey: Prentice-Hall, Inc.

2. Senturia, S.D., Microsystem Design. 2001, Norwell, Massachusetts: Kluwer
Academic Publisher.

3. Petersen, K.E., Silicon as a Mechanical Material. Proceedings of the IEEE, 1982.
70(5): p. 420-457.

4. Voldman, J., Gray, M.L., Schimdt, M.A., Microfabrication in Biology and
Medicine. Annual Review of Biomedical Engineering, 1999. 1: p. 401-425.

5. Kovacs, G.T.A., Maluf, N.I., and Petersen, KE., Bulk Micromachining of Silicon.
Proceedings of the IEEE, 1998. 86(8): p.1536-1551.

6. Williams, K.R., and Muller, R.S., Etch Rates for Micromachining Processing.
Journal of Microelectricalmechanical Systems, 1996. 5: p. 256-269.

7. El-Kareh, B., Fundamentals of Semiconductor Processing Technologies. 1995,
Norwell, Massachusetts: Kluwer Academic Publisher.

8. McDonald, J.C., Duffy, D.C., Anderson, J.R., Chiu, D.T., Wu, H., Schueller,
O.J.A., and Whitesides, G.M., Fabrication of Microfluidic Systems in
Poly(liiet1y/l .ihin,\ue)i. Electrophoresis, 2000. 21: p. 27-40.

9. Branham, M.L., Tran-Son-Tay, R., Schoonover, C., Davis, P.S., Allen, S.D., Shyy,
W., Rapid Prototyping ofMicropatterned Substrates using Conventional Laser
Printer. Journal of Material Research, 2002. 17(7): p. 1559-1562.

10. Deng T., Wu, H., Brittain, S., and Whitesides, G., Prototyping of Masks, Masters,
and .I,i11mq,.\ lhlJ' for Soft Lithography Using an Office Printer and Photographic
Reduction. Analytical Chemistry, 2000. 72: p. 3176-3180.

11. Kenis, P.J.A., Ismagilov, R.F., Takayama, S., Whitesides, G.M., Li, S., White,
H. S., Fabrication InsideMicrochannels Using Fluid Flow. Account of Chemical
Research, 2000. 33: pg. 841-847.









12. Love, J.C., Wolfe, D.B., Jacobs, H.O., Whitesides, G.M., Microscope Projection
Photolithographyfor Rapid-Prototyping of Masters n i/h Micro-scale Featuresfor
use in Soft Lithography. Langmuir, 2001. 17: p. 6005-6012.

13. Schmitz, T., Dagata, J., Dutterer, B., and Sawyer, W., Nanolithography and
Microfluidics: A Manufacturing Perspective, 2003. Submitted to the Journal of
Manufacturing Processes.

14. Tan, W., Fan, Z.H., Qiu, C.X., Ricco, A.J., Gibbons, I., Miniaturized Electric
Capillary Isoelectric focusing in Plastic Microfluidic Device. Electrophoresis,
2002. 23: p. 3638-3645.

15. Herr, A.E., Molho, J.I., Drouvalakis, K.A., Mikkelsen, J.C., Utz, P.J., Santiago,
J.G., Kenny, T.W., On-chip Coupling of soelectric Focusing and Free Solution
Electrophoresis for Multidimensional Separations. Analytical Chemistry, 2003. 75:
p. 1180-1187.

16. Manz, A., Graber, N., Widmer, H.M., Miniaturized Total ChemicalAnalysis
Systems: A Novel Concept for Chemical Sensing. Sensors and Actuators, 1990.
Bl(1-6): p. 244-248.

17. Harrison, D.J., Fluri K., Seiler, K., Fan, Z., effenhauser, C.S., Manz, A.,
Micromachining a Miniaturized Capillary Electrophoresis-based Chemical
Analysis System on a Chip, Science, 1993. 261: p. 895-897.

18. Kricka, L.J., and Fortina, P., Microchips: An All Language Literature Survey
including Books and Patents. Clinical Chemistry, 2002. 48(9): p.1620-1622.

19. Terry, S.C., Jerman, J.H., Angell, J.B., A Gas Chromatographic Air Analyzer
Fabricated on a Silicon Wafer. IEEE Transactions on Electron Devices, 1979. ED-
26: p. 1880-1886.

20. Kikuchi, Y., Ohki, H., Kaneko, T., and Sato, K., Microchannels Made on Silicon
Wafer for Measurement ofFlow Properties of Blood Cells. Biorheology, 1989. 26:
p. 1055.

21. Kikuchi, Y., Sato, K., Kaneko, T., and Ohki, H., Optically Accessible
Microchannels Formed in Single-Crystal Silicon Substrate for Studies of Blood
Rheology. Microvascular Research, 1992. 44: p. 226-240.

22. Kikuchi, Y., Sato, K., and Mizuguchi, Y., Modified Cell Flow Microchannels in a
Single-Crystal Silicon Substrate and Flow Behavior of Blood Cells. Microvascular
Research, 1994. 47: p. 126-139.









23. Tracey, M.C., Greenaway, R.S., Das, A., Kaye, P.H., and Barnes, A.J., A Silicon
Micromachined Device for Use in Blood Cell Deformability Studies. IEEE
Transactions on Biomedical Engineering, 1995. 42(8): p.

24. Sutton, N., Tracey, M.C., Johnston, I., Greenaway, R.S., and Rampling, M.W., A
Novel Instrument for Studying the Flow Behavior ofEiy)n, I 1) e/i. through
Microchannels Simulating Human Blood Capillaries. Microvascular Research,
1997. 53: p. 272-281.

25. Tracey, M.C., Sutton, N., Johnston, I., and Doeztel, W., A Microfluidics-Based
Instrument for Cytomechanical Studies ofBlood. 1st Annual International IEEE-
EMBS Special Topic Conference on Microtechnologies in Medicine and Biology,
October 12-14, 2000.

26. Brody, J.P., Han, Y., Austin, R.H., Bitensky, M., Deformation andFlow of Red
Blood Cells in a Synthetic Lattice: Evidence for an Active Cytoskeleton. Biophysics
Journal, 1995. 68(6): p. 2224-2232.

27. Ayliffe, H.E., Rabbitt, R.D., Tresca, P.A., and Frazier, A.B., Micromachined
Cellular Characterization System for Studying the Biomechanics of Individual
Cells. Transducers 1997. 1997 International Conference on Solid-State Sensors
and Actuators, Chicago, June 16-19, 1997.

28. Coulter, W.H., High Speed Automatic Blood CellAnalyzer. Proceedings of the
National Electronics Conference, 1956. 12: p. 1034-1040.

29. Roberts, K., Parameswaran, M., Moore, M., and Muller, R., ASilicon
Microfabricated Aperture for Counting Cells using the Aperture Impedance
Technique. Proceedings of the IEEE Canadian Conference on Electrical and
Computer Engineering. Shaw Conference center, Edmonton, Alberta, Canada May
9-12, 1999.

30. Satake, D., Ebi, H., Oku, N., Matsuda, K., Takao, H., Ashiki, M., and Ishida, M., A
Sensor for Blood Cell Counter Using MEMS Technology. Sensors and Actuators
B, 2002. 83: p. 77-81.

31. Gawad, S., Heuschkel, M., Leung-Ki, Y., luzzolino, R., Schild, L., Lerch, Ph., and
Renaud, Ph., Fabrication of a Microfluidic Cell analyzer in a Microchannel Using
Impedance Spectroscopy. lst Annual International IEEE-EMBS Special Topic
Conference on Microtechnologies in Medicine and Biology, October 12-14, 2000.

32. Voldman, J., Gray, M.L., Toner, M., and Schmidt, M.A., A Microfabrication-Based
Dynamic Cytometer. Analytical Chemistry, 2002. 74: p. 3984-3990.









33. Shrewsbury, P.J., Muller, S. J., Liepmann, D., Effect of Flow on Complex
Biological Macromolecules in Microfluidic Devices. Biomedical Microdevices,
2001. 3(3): p. 225-238.

34. Kleinfeld, D., Kahler, K.H., Hockberger, P.E., Controlled Onigi ,I1 /th of
Dissociated Neurons on Patterned Substrates. Neuroscience, 1988. 8(11): p. 4098-
4120.

35. Zhao, B., Moore, J.S., Beebe, D.J., Surface-Directed Liquid Flow Inside
Microchannels. Science, 2001. 291: p. 1023-1026.

36. Folch, A., Toner, M., Cellular Micropatterns on Biocompatible Materials.
Biotechnology Progress, 1998. 14: p. 388-392.

37. Gad-el-Hak, M., The MEMS Handbook. 2002, London, New York, Washingtion
D.C.: CRC Press.

38. Pellois, J.P., Zhou, X., Srivannavit, O., Zhou, T., Gulari, E., Gao, X., Individually
Addressable Parallel Peptide Synthesis on Microchips. Nature Biotechnology,
2002. 20: p. 922-926.

39. White, Frank. Viscous Fluid Flow, 1991, New York: McGraw-Hill.

40. Papautsky, I., Gale, B.K., Mohanty, S., Ameel T.A., Frazier, A.B., Effects of
Rectangular Microchannel Aspect Ratio on Laminar Friction Constant. SPIE
Microfluidic Devices And Systems Conference, 1999. p. 147-159.

41. Eringen, A.C., Simple Microfluids. Int. J. Eng. Sci., 1964. 2: p. 205-217.

42. Eringen, A.C., Suhubi, E.S., NonlinearTtheory of Simple Micro-Elastic Solids.
International Journal. Engineering Science, 1964. 2: p. 189-203.

43. Kang, C.K., Eringen, A.C., The Effect of Microstructure on the Rheological
Properties ofBlood. Bull. Math. Biol., 1976. 38: p. 135-158.

44. Wu, P., Little, W.A., Measurement of Friction Factors for the Flow of Gases in
Very Fine Channels used for Microminiature Joule-Thomson Refrigerators.
Cryogenics, 1983. 5: p. 273-277.

45. White, F.M., Fluid Mechanics, 1999, Boston, Massachusetts: WCB McGraw-Hill.

46. Bousse, L., Cohen, C., Nikiforov, T., Chow, A., Kopf-Sill, A.R., Dubrow, R.,
Parce, J.W., Electrokinetically Controlled Microfluidic Analysis Systems. Annual
Review of Biophysical and Biomolecular Structures, 2000. 29: p. 155-181.









47. Kornberg, L.J., FocalAdhesion Kinase in Cancer. Emerging Therapeutic Targets,
2000. 4(2): pg. 1-16.

48. Haier, J., Nicholson, G.L., Tumor CellAdhesion under Hydrodynamic Conditions
ofFluidFlow. APMIS, 2001. 49: pg. 155-166.

49. Weiner, T.M., Lui, E.T., Craven, R.J., Cance, W.G., Expression ofFocalAdhesion
Kinase Gene andlnvasive Cancer. Lancet, 1993. 342: pg. 1024-1025.

50. Ferguson, D., Cellular Attachment to Implanted Foreign Bodies in Relation to
Tumorigenesis. Cancer Research, 1977. 37: pg. 4367-4371.

51. Tran-Son-Tay, R., Branham, M., Glover, S., Perrault, C., Benya, R.V., Allen, S.,
Shyy, W., A Rapid Prototyping Technique for Studying Contribution of
Extracellular Microtexture to Cell Shape, Attachment and Spreading, 11th Int.
Congress ofBiorheology and 4th Int. Conf. on Clinical Hemorheology, Antalya,
Turkey, 2002).

52. Loh, I-H., Plasma Purface Modification in Biomedical Applications. AST
Technical Journal, 2003. Available from URL:
http://www.astp.com/technicalinfo/generallit.cfm. Site last visited May 2003.

53. Tritton, D.J., Physical FluidDynamics, 1988. Oxford: Clarendon Press.















BIOGRAPHICAL SKETCH

Katie Maiellaro was born in Pensacola, Florida, on June 29, 1978. Her parents

Edward and Rachel Maiellaro offered her a loving environment to grow up happy and

healthy. Katie has two older brothers Matt and Tom, whom she loves and admires

incredibly. Her foundation of faith and strong values was created by her family, as well

as by St. Paul's Catholic School, where she attended from kindergarten through 8th

grade. She then attended the International Baccalaureate Program at Pensacola High

School, graduating Valedictorian of the Class of 1996. In addition, Katie was a 4-year

member of the PHS tennis team, winning her district title in 1993.

Katie received her Bachelor of Science degree in engineering science at the

University of Florida. With the thought of eventually attending medical school, Katie

first decided to further her interest in biomechanics by remaining at UF and pursuing her

Master of Engineering degree in biomedical engineering. However, during her tenure as

a graduate student, she discovered her love of the academic environment and decided

continue her education as an engineer and complete her doctorate in biomedical

engineering. After receiving her M.E. in biomedical engineering at the University of

Florida she will enroll in the Joint Biomedical Engineering Ph.D. Program at Georgia

Tech and Emory in August 2003.