FLORIDA THE STATE UNIVERSITY
COLLEGE OF ENGINEERING
CHARACTERIZATION OF MICROFLUIDIC CHANNELS FOR BIODIAGNOSTICS
By ANDREW MICHEAL ZWOLINSKI
A Thesis submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Master of Science
Degree Awarded Spring Semester, 2004
i
The members of the Committee approve the Thesis of Andrew Michael Zwolinski defended on April, 15 2004.
____________________________________ Yousef Haik Professor Directing Thesis
____________________________________ Ching-Jen Chen Committee Member
____________________________________ Chiang Shih Committee Member
Approved:
Chiang Shih, Chairman, Department of Mechanical Engineering
Ching-Jen Chen, Dean, College of Engineering The office of Graduate Studies has verified and approved the above named committee members. ii
ACKNOWLEDGEMENTS
I would like to first thank the supporters of this research Sandia National Laboratories (SNL) with their Lab Directed Research and Development (LDRD) grant number 64709 as well as the Florida State University (FSU) Research Foundation. With out their support this research would not be possible. In addition to financial support I would like to thank all of those who provided guidance during the span of my engineering career at FSU. This includes but is not limited to Dr. Yousef Haik, Dean Ching-Jen Chen, Dr. Jhunu Chatterjee and all my colleagues at the Center for Nanomagnetics and Biotechnology (CNB). I would like to acknowledge Dr. Paul Galambos of the Intelligent Micromachine Group at SNL Albuquerque, NM, without his guidance and expertise this research would not have been as fruitful. I would also like to single out Dr. Weston (CNB participating scientist) at the Department of Chemistry, FSU, and his laboratory, especially Kalyan K. Kuricheti, for assisting me with the use of their fluorescence correlation spectroscopy (FCS) setup which was an integral part of this work. I would like to thank Dr. Chang Shih for his assistance on my work as a member of my committee. I wish to thank my fianc, Katy Spears, whose unrelenting support through the ups and downs of this work provided me with the strength and motivation to conclude my work to the best of my ability, thanks sweetie. I also thank my parents, Danuta and Aleksander, for their dedication, love and guidance without which I could not have made it to this point. Finally I would like to thank all of those at SNL and the Department of Mechanical Engineering that are behind the scenes making sure that I received my parts and financial support from FSU while at SNL. I cannot thank each of you enough.
iii
TABLE OF CONTENTS
LIST OF FIGURES .......................................................................................................... vii ABSTRACT...................................................................................................................... xii CHAPTER 1: INTRODUCTION ........................................................................... 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 Acute Myocardial Infarction........................................................................... 2 Diagnosis......................................................................................................... 3 Current Diagnosis Techniques........................................................................ 4 Newly Developed Diagnosis Technique......................................................... 6 Microfluidic Integration.................................................................................. 8 Study Objective............................................................................................... 9 Accomplished work ........................................................................................ 9 Scope............................................................................................................. 10
CHAPTER 2: ACUTE MYOCARDIAL INFARCTION DETECTION ..........................12 2.1 Super Paramagnetic Microspheres................................................................ 12 2.1.1 Background ............................................................................................... 13 2.1.2 Synthesis ................................................................................................... 13 2.1.3 Characterization ........................................................................................ 14 2.1.3.1 Scanning Electron Microscopy ......................................................... 15 2.1.3.2 Atomic Force Microscopy ................................................................ 16 2.1.3.3 Superconducting Quantum Interference Device............................... 17 2.1.3.4 Conclusion: HSA Microsphere Characterization.............................. 17 2.1.4 Surface Modification ................................................................................ 19 2.1.5 Conclusion: Super Paramagnetic Microspheres ....................................... 20 2.2 Acute Myocardial Infarction Detection ........................................................ 20 2.2.1 Detection Setup......................................................................................... 20 2.2.1.1 Antibody Biotinylation ..................................................................... 21 2.2.1.2 Alkaline Phosphatase Labeling......................................................... 21 2.2.1.3 Anti-Myo 908/SPM Coupling........................................................... 22 2.2.1.4 Anti-Myo 4E2/AP Coupling ............................................................. 22 2.2.1.5 Anti-FABP 10E1/SPM Coupling...................................................... 22 2.2.1.6 Anti-FABP 9F3/AP Coupling........................................................... 23 2.2.1.7 Human Serum Albumin (HSA) Saturation ....................................... 23 2.2.1.8 Conclusion: Detection Setup............................................................. 23
iv
2.2.2 AMI Detection Experiments ..................................................................... 24 2.2.2.1 Overview of AMI detection Process................................................. 24 2.2.2.2 Myoglobin Detection ........................................................................ 25 2.2.2.3 FABP Detection ................................................................................ 26 2.2.2.4 Conclusion: AMI Detection Experiments......................................... 26 2.3 AMI Detection Applied to Microfluidics ..................................................... 26 CHAPTER 3: MICROFLUIDICS .....................................................................................29 3.1 Microchannel Fabrication .................................................................................. 29 3.1.1 SUMMiT V TM .......................................................................................... 30 3.1.2 SwIFT TM .................................................................................................. 33 3.1.3 Poly(dimethylsiloxane)-glass Microchannels.......................................... 35 3.2 Microfluidics ...................................................................................................... 36 3.3 Microchannel Characterization Methods ........................................................... 41 3.3.1 Pressure Flow Characterization ................................................................ 42 3.3.2 Micro-Particle Image Velocimetry Characterization................................ 42 3.3.3 Fluorescence Correlation Spectroscopy.................................................... 43 3.4 Conclusion.......................................................................................................... 45 CHAPTER 4: Flow Characteristics in Micro Channels ....................................................47 4.1 Preparation of Magnetic Particles................................................................. 47 4.2 SwIFT TM Microchannel Characterization.................................................... 48 4.2.1 Pressure Microflow................................................................................... 48 4.2.1.1 Experimental Results Pressure Flow ................................................ 56 4.2.1.2 Study 1: Sandia National Laboratory Experiments .......................... 58 4.2.1.3 Study 2: Florida State University Experiments ................................ 62 4.2.1.4 Comparison between SNL and FSU ................................................. 62 4.2.1.5 Study 3: Repeatability (A) ................................................................ 64 4.2.1.6 Study 4: Repeatability (B) ................................................................ 66 4.2.1.7 Study 5: Fluid Dependency............................................................... 67 4.2.1.8 Conclusion: Pressure Microflow....................................................... 69 4.2.2 Microsphere Compatibility ....................................................................... 70 4.2.2.1 Set Up for Microsphere Introduction................................................ 70 4.2.2.2 Study: Microsphere Introduction ...................................................... 71 4.2.2.3 Conclusion: Microsphere Compatibility........................................... 73 4.2.3 Fluorescence Correlation Spectroscopy.................................................... 74 4.2.3.1 SwIFTTM Microchannels................................................................... 74 4.2.3.2 FCS SwIFTTM Microchannel Set Up ................................................ 75 4.2.3.3 Study 1: 200nm FluoroSpheres...................................................... 75 4.2.3.4 Study 2: 40nm FluoroSpheres........................................................ 76 4.2.3.5 Study 3: Time Dependence and Blockage ........................................ 78 4.2.3.6 Conclusion: FCS ............................................................................... 81 4.3 PDMS-Glass Microchannel Characterization............................................... 82 4.3.1 Pressure Microflow................................................................................... 82
v
4.3.1.1 Study 1: DI Water ............................................................................. 83 4.3.1.2 Study 2: 1M NaOH ........................................................................... 83 4.3.1.3 Comparison DI Water and 1M NaOH .............................................. 85 4.3.1.4 Conclusion: Pressure Microflow....................................................... 87 4.3.2. Fluorescence Correlation Spectroscopy.................................................... 88 4.3.2.1 FCS PDMS-Glass Microchannel Setup ............................................ 88 4.3.2.2 Study 1: 40nm FluoroSpheres........................................................ 88 4.3.2.3 Study 2: Velocity Profile .................................................................. 89 4.3.2.4 Conclusion: FCS ............................................................................... 94 4.4 Conclusion: Flow Characteristics in Micro Channels .................................. 94 CHAPTER 5: SUMMARY AND CONCLUSIONS.........................................................96 REFERENCES ................................................................................................................100 BIOGRAPHICAL SKETCH ...........................................................................................104
vi
LIST OF FIGURES
Figure 1.1
Schematic of a IgG antibody. The antibody has physical disulfide bonds that help hold it together. The IgG antibody has variable regions on both the heavy and light chains. A typical ELISA showing four steps involved. (1) Attach an antibody to a solid anchor, (2) introduce the antigen, (3) introduce the enzyme linked antibody, and wash excess, (4) introduce a known quantity of reagent, react and quantify using a spectrophotometer. Modified HSA microsphere attached to a erythrocyte, red blood cell. SEM micrograph of HSA microspheres in dispersion [11]. Histogram showing particle size distribution of HSA microspheres found in dispersion [11]. (a) AFM of a HSA microsphere in phase mode, (b) AFM of HSA microsphere in the height enhanced mode showing crosslimking due to a heat stabilization process. The temperature susceptibility plot for heat stabilized HSA microspheres. Applied field versus magnetization plot for heat stabilized HSA microspheres showing a super paramagnetic nature. Schematic of AMI detection process. Schematic of the AMI detection process that needs to be applied to the micro scale. SUMMiT V TM layer terminology and order. SwIFT TM layer terminology and order vii
Page 5
Figure 1.2
Page 6
Figure 1.3 Figure 2.1 Figure 2.2 Figure 2.3
Page 7 Page 15 Page 15 Page 16
Figure 2.4 Figure 2.5 Figure 2.6 Figure 2.7 Figure 3.1 Figure 3.2
Page 18 Page 18 Page 25 Page 27 (b) Page 31 Page 34
Figure 3.3
Bosch backside etch example. The picture is in focus on the nitride layer, the center of the holes. The portion of the picture that is not in focus is the underside of the MEMS based microchannel, single crystal silicon. (1) Photolithography is preformed to produce the master, (2) the PDMS positive is created using the master, (3) remove the polymerized PDMS from the master, (4) seal the PDMS against a glass cover slide to create the microchannels. Typical representation of APD data. As the velocity increase the number of peaks within a period of time increases as well, the opposite is true when the flow decreases. Confocal optics setup where the APD is the avalanche photodiode. 44 Typical MEMS Microchannel used in this research. AutoCAD SwIFT TM based drawing the colors signify different layers. Microchannel with silicon oxide blocking flow. Microchannel with silicon nitride etched through. Photograph showing foreign obstructive material thought to be quick dry glue. Pressure head set up on probe station. A close up of the set up. The pressure versus Mass Flow Rate relationship found at SNL, the circled points were redone values to check for consistency. This graph shows the relationship between pressure and Reynolds number, notice the laminar flow shown by the low Reynolds number.
Page 34
Figure 3.4
Page 36
Figure 3.5
Page 44
Figure 3.6 Figure 4.4 Figure 4.5 Figure 4.3 Figure 4.4 Figure4.6 Figure 4.7 Figure 4.8 Figure 4.9
Page 45 Page 49 Page 49 Page 50 Page 50 Page 54 Page 57 Page 57 Page 58
Figure 4.10
Page 60
viii
Figure 4.11
Relationship between mass flow rate and resistance. The resistance for the data points corresponding to SNL 1 (phobic) were out of scale and shown as figure 4.12. Relationship between mass flow rate and resistance for the data set corresponding to SNL 1 (phobic). 61 Pressure versus flow rate for COE Tallahassee, FL. The behavior is relatively linear even when at a higher pressure then those achieved at SNL in Albuquerque, NM. Pressure versus Reynolds number for COE Tallahassee, FL, with numbers ranging from 0.6 2. Comparison between the SNL and FSU pressure versus flow rate. Pressure versus mass flow rate same channel using DI water, two separate experiments to determine if results are reproducible. Pressure versus mass flow rate same channel using PBS water, two separate experiments to determine if results are reproducible. The figure above shows pressure versus mass flow rate relationship between two different liquids within the same channel. FSU 7(A) has a pH of 2 and FSU7(B) is DI water. (a) A single free flowing HSA microsphere, (b) agglomeration and adsorption of HSA microspheres onto the walls of the channel. Above is a picture of the FCS setup. Noticeable in the foreground are the mirrors and to the right the laser. To the far left is the stage setup with the piezoelectric actuation device mounted on top. (a) This scan was taken at the top nitride layer of microchannel, (b) this scan was taken at a point inside of the microchannel. These figures show particles stuck within the channel.
Page 61
Figure 4.12 Figure 4.13
Page 61 Page 63
Figure 4.14 Figure 4.15 Figure 4.16
Page 63 Page 64 Page 65
Figure 4.16
Page 67
Figure 4.18
Page 68
Figure 4.19
Page 73
Figure 4.20
Page 74
Figure 4.21
Page 77
ix
Figure 4.22
(a) This is a scan of the inlet region to the microchannel. There is massive agglomeration and adsorption occurring and blocking the flow into the channel. (b) This figure is a picture of the actual channel with some key dimensions. Scan of neck region of channel prior to introducing bead seeded solution. The arrows are pointing to debris on top of the channel. The channel walls can be seen faintly in the upper left portion of the figure. Scan taken five minutes after flow had exited the microchannel. Notice adsorption occurring at the side walls of the channel and agglomeration occurring in the central area of the neck region. Scan taken 1 hour after flow had started. The signal was getting stronger and started overwhelm the photodiode detectors. Final scan taken 2 hours after initial flow, adsorption to side walls is very pronounced and agglomeration and adsorption prevalent in the neck region of the entrance to the channel. Schematic of PDMS-glass channel, shown with dimensions (not to scale). Head pressure versus mass flow rate data for the PDMS-glass channel with DI water as the fluid. DI Water 1 was the first experiment preformed and DI Water 2 is the second experiment preformed. Pressure versus flow data for 1M NaOH solution. The first run is labeled as NaOH1 and the second run is labeled. Pressure versus mass flow rate comparison between DI water and 1M NaOH, with 1M NaOH having a lower relative flow rate. Head pressure versus Reynolds number for the two different fluids introduced into the PDMS-glass channel.
Page 77
Figure 4.23
Page 78
Figure 4.24
Page 79
Figure 4.25
Page 80
Figure 4.26
Page 80
Figure 4.27 Figure 4.28
Page 84 Page 84
Figure 4.29
Page 85
Figure 4.30
Page 86
Figure 4.31
Page 87
x
Figure 4.32
Preliminary scan of PDMS-glass microchannel. The scan was taken at half the height of the channel, 20m from the glass interface. Data collection program interface. The lower middle line scan images show the data collection window for velocity measurement. An example of the spectral reading for flowing beads within the channel, the number of peaks increase with increase in velocity. The normalized autocorrelation function versus tau (time). This data is un adjusted, therefore raw. The line is a best fit curve and the points are the actual data points collected. This graph represents a refined set of data points in the normalized autocorrelation function whose average is used to calculate the velocity of a specific point in the channel. (a) Graph showing anomalous behavior of data points, (b) corresponding velocity profile of channel with data points missing where anomalies were detected. Velocity profile at a driving pressure of 4850Pa for the PDMS-glass microchannel. Pressure versus flow rate comparison between SwIFTTM and PDMS-glass microchannels. Pressure versus Reynolds number comparison between SwIFTTM and PDMS-glass microchannels.
Page 89
Figure 4.33
Page 90 Page 90
Figure 4.34
Page 91
Figure 4.35
Page 92
Figure 4.36
Page 93
Figure 4.37
Page 93 Page 98 Page 98
Figure 4.38 Figure 5.1 Figure 5.2
xi
ABSTRACT
Characterization of fluid with suspended nanoparticles in microchannels has been studied as a part of a microfluidic based acute myocardial infarction (AMI) detection device. The AMI detection process uses heat stabilized human serum albumin (HSA) magnetic microspheres and specific antibodies to create a magnetic immunoassay used in the detection of AMI. Microanalysis systems have several advantages over conventional analysis systems due to their sensitivity, reliability and the amount of anlaytes needed for the test. The microchannels used in this work were fabricated at Sandia National Laboratories (SNL) using a SwIFTTM microfabrication surface micromaching process. Micro channels made of Poly(dimethylsiloxane)-glass (PDMS-glass) designed and fabricated at the Department of Chemistry at the Florida State University were also used in this work. The SwIFTTM microchannels had dimensions of 6m in height, 20m in width and 200m in length where as the PDMS-glass microchannels had dimensions of 40m in height, 200m wide and 13mm in length. Characterization of the microchannels was accomplished using a variety of techniques. The first method used to characterize the microchannels was to used a head pressure-flow set up to determine the pressure and flow characteristics of the SwIFTTM microchannels with the different fluids that the biodiagnostic process calls for, with average mass flow rate being 1.9x102
g/s and Reynolds number of 1.45 at a pressure of 23kPa for a typical channel, these values
approach the upper limit of the work accomplished. Since the HSA microspheres, 1m in diameter and less, play a critical role in the detection protocol their compatibility to the SwIFTTM microchannels was investigated. Results showed the HSA microspheres agglomerated and adsorbed to the walls of the channels. Fluorescence correlation spectroscopy (FCS) was attempted on the SwIFTTM microchannels with 200nm and 40nm beads and the same conclusion of agglomeration and adsorption was reached which made these channels not suitable for
xii
adaptation in the microanaylsis system considered for AMI detection. PDMS-glass microchannels head pressure-flow rates were also investigated showing an average mass flow rate of 1.76x10-1g/s and a Reynolds number of 1.03 at a pressure of 4.5kPa. FCS was preformed on these channels successfully without any signs of agglomeration, though some adsorption of the beads to the walls of the channel was evident. FCS measured max velocity was equal to approximately 6.6 cm/s. Thus it is concluded that microchannels of similar sizes of the PDMS-glass will be needed in the microanalysis system that is being developed to detect for AMI markers.
xi
CHAPTER 1 INTRODUCTION
Current processes used to determine protein levels quantitatively require specialized expensive laboratory equipment. This study contributes to the creation of a portable self contained affordable microsystem. The system incorporates the technology that was developed at the Center for Nanomagnetics and Biotechnology for detection of the levels of proteins in liquid media. The detection process is achieved without the use of specialized laboratory equipment, with the exception of a spectrophotometer, eliminating the need for a costly centrifuge and other large bench top devices that detract from system portability. The objective of this study is to test the applicability of this technology in Micro Electro Mechanical Systems (MEMS) to allow for the miniaturization of the entire detection device. The detection technology incorporate the use of nanomagnetic particles that are prepared to label desired proteins in the heterogeneous sample then with the application of external magnetic field these coupled nanoparticles with the specific protein can be isolated from the sample, thus replacing the function of the centrifuge in the large systems. The benefits of miniaturization include reduced cost, increased sensitivity, parallel processing and portability to name a few. The developed processes for detecting AMI markers depends on moving bodies of fluids with suspended media into channels and conduits, mixing bodies of fluids and then separating the magnetically labeled proteins from the test sample. In order to assess the adaptation of the developed process in a microanalysis system it is important to characterize the fluid motion, particularly the magnetic fluid, in the microchannels. The study concentrates on understanding the behavior of fluid with suspended nanomagnetic particles at the micro level
1
including pressure flow characteristics, fluid flow characterization and mixing characteristics. Investigating and understanding these phenomena will allow for basic knowledge needed to create a viable miniaturized protein detection device. In this case the final device will be a system that will be used for the quantification of biomarkers in a Point of Care (POC) device. The biomarkers chosen for this particular device will be the markers associated with Acute Myocardial Infarction (AMI), more commonly known as a heart attack.
1.1
Acute Myocardial Infarction
Acute Myocardial Infarction (AMI) is a disorder in which damage to an area of heart muscle
occurs because of an inadequate supply of oxygen [1]. The decrease of oxygen to the area is a function of the blood flow. In an AMI situation the blood flow is either dramatically reduced or all together stopped. The blockage is found in a coronary artery, which is responsible for supplying the myocardium with blood, and is due to the narrowing of the artery because of plaque build up or arteriosclerosis. The end result of the blockage and subsequent AMI is the damaging or death of a portion of the myocardium. The damaged tissue in turn permanently affects the heart's ability to function normally [1]. Factors that can lead to an AMI include, but are not limited to, smoking, hypertension, diabetes mellitus, a high fat diet, high Low Density Lipoprotein levels, obesity, male gender, being over the age of 65 and heredity. The primary sign of an AMI is chest pain with numbness in the left upper extremity, but in many cases these symptoms are not present in the individual experiencing the AMI, this is called a silent AMI. Approximately 1 of every 500 people will experience an AMI per year making this incident a major cause of sudden death in adults [1]. Currently there is a thrust to decrease the time of AMI determination in patients entering the emergency department with the symptom of chest pains. The faster the diagnosis could be made the faster the patient can either be admitted to the emergency department or be safely sent home. The main treatment for an AMI is the administering of thrombolitic therapy, the introduction of drugs to induce the dissolving of the clot, is most effective in the early stages of the AMI, thus minimizing the damage to the myocardium. [2,3]. According to the American College of Cardiology/American Heart Association guidelines for the management of patients with AMI, a proper goal for an emergency department AMI protocol would be that which yields
2
a targeted clinical examination and a 12-lead electrocardiogram within 10 minutes and a door-toneedle (as in treatment for AMI) time that is less than 30 minutes [3]. There are approximately 5 million individuals in America that are admitted into an emergency department with symptoms of acute chest pains. Of this 5 million 3 million are admitted due to the fact that the precursory screening is not accurate enough in roughly 50% of the admitted patients. A fact that is more important is that 2-8% of patients with acute chest pains that are discharged by an emergency department end up having an AMI resulting in adverse conditions for the patient and malpractice action against the department [3,4].
1.2
Diagnosis
The world health organization states that in order to diagnose an AMI two of three criteria
must be present, these criteria are: clinical history, electrocardiographic tracing and the rise and fall of cardiac markers [5]. This work is concerned with the detection and quantification of these cardiac markers. As the heart is experiencing the infarction proteins such as myoglobin, troponin, creatine kinase, fatty acid binding protein and glycogen phosphorylase are released into the blood stream at varying time post infarction. The time of release is bases on the size of the protein. The smaller the protein the faster it will be introduced into the blood stream and subsequently detected. The detection of these proteins is instrumental in both the diagnosis, detecting the presence of an increase in the concentration of the proteins in question, and sizing the AMI by taking serial measurements to estimate the extent of the damage [3]. Among the common protein utilized for the detection of the AMI is the MB isoenzyme of creatine kinase (CK-MB). This protein is relatively large in size, 86 kDa, and is concentration in the blood stream peaks from 3g/L to 25g/L 16 to 20 hours post AMI. The concentration of this biomarker is insignificant in the early stages of the AMI rendering the use of this protein useless for detecting the AMI in its early phases. In the case of skeletal muscle trauma CK-MB is also released into the blood stream leading to the use of CK-MB as a confirmation biomarker in the case of an AMI [6]. A proper protein marker to use in an assay is one that is present in measurable concentrations in the early post AMI stage as well as having high clinical sensitivity and specificity [7]. The National Academy of Clinical Biochemistry (NACB) recommends utilizing
3
two protein markers: a marker that appears within 6 hrs after onset and a later marker that has high cardiac specificity and elevates between 6-9 hrs after onset [8]. In this instance the use of proteins that are smaller in size would lead to an earlier detection time. Studies have shown that Fatty Acid Binding Protein (FABP, 15 kDa), and Myoglobin (Myo, 17kDa) are two relatively small proteins that become elevated in the blood stream shortly after an infarction. The protein concentrations become elevated after two hours and peak four to six hours post infarction. The normal levels of Myo and FABP are approximately 32g/L and 3g/L respectively, these levels peak to a concentration of over 200g/L for Myo and over 100g/L for FABP, in many cases within five hours of the incident [8]. A problem with using Myo and FABP for the diagnosis of an AMI is the lack of cardiac specificity as is the case with CK-MB the levels of Myo and FABP are elevated in the blood stream after skeletal muscle injury [2]. Studies have shown that the ratio of Myo to FABP in the blood stream post skeletal muscle injury differ from the ratio observed post AMI. In post skeletal muscle injury a Myo/FABP ratio of 20:70 is observed and post AMI a Myo/FABP ratio of 4:5 is observed allowing for the ability to discern between skeletal muscle injury and injury to the myocardium [7].
1.3
Current Diagnosis Techniques
A common technique used to determine an unknown concentration of a cardiac protein in
a given sample is through the use of immunoassays. Of these immunoassays a common form is called an Enzyme-Linked Immunosorbent Assay otherwise known as an ELISA. An ELISA is basically described as a protein sandwich, consisting of two antibodies, an antigen, an enzyme and a base of attachment. An antibody (Ab) is a protein that has a high affinity for a specific antigen (Ag), which is another protein. An antigen has many different antibody binding sites on it; these binding sites are called epitopes. There are many classes of antibodies ranging from a basic IgG (Immunoglobulin G), composed of a single Y shape, to an IgM that is composed of five Y shapes attached to each other [9]). The basic anatomy of an antibody stays the same through the variety of types. Antibodies are composed of two arms and a base. The antibody that is utilized in this research is the IgG antibody. The IgG antibody is composed of four polypeptides, two identical heavy
4
chains, approximately 55,000 daltons, and two identical light chains, approximately 25,000 daltons, please refer to figure 1.1. There are both constant and variable regions on the heavy and light chains. The variable region of one heavy chain and one light chain form to combine one antigen binding site. This is how a particular antibody has the specificity that it has for a specific antigen. The antibodies are produced in B-cell clones and it is due to the diversity of these Bcell clones that we see diversity in antibodies, one B-cell produces one type of antibody [9]. It is on this premises that an ELISA functions.
Variable Region
Variable Region Light Chain
Heavy Chain
Figure 1.1 Schematic of a IgG antibody. The antibody has physical disulfide bonds that help hold it together. The IgG antibody has variable regions on both the heavy and light chains.
An ELISA consists of two antibodies (Ab), an enzyme, an antigen (Ag) and base of attachment. One of the antibodies is attached to a solid surface and provides the attaching base for the assay. This antibody is used to separate an antigen, or target protein, from the rest of the background media. The anchored antibody binds with a specific epitope on the antigen and been isolates it from the background media. The next step introduces the second antibody that is conjugated with an enzyme. This antibody-enzyme complex then binds with another specific
5
epitope on the antigen while the antigen is already attached to the first anchoring antibody. This creates an immobilized complex that is ready to react with a solution in which there is a substrate that the enzyme can hydrolyze. The extent of the reaction provides a measurement of the quantity of the antigen, or target protein, within the given sample, as schematic of the process is shown as figure 1.2.
Legend
Ab Epitope 1 Ag Enzyme Ab Epitope 2 Anchor Anchor Anchor Anchor Reagent
Step
1
2
3
4
Figure 1.2 A typical ELISA showing four steps involved. (1) attach an antibody to a solid anchor, (2) introduce the antigen, (3) introduce the enzyme linked antibody, and wash excess, (4) introduce a known quantity of reagent, react and quantify using a spectrophotometer.
Earlier emergency department assays were preformed with the use of full laboratory equipment such as centrifuges, pipetters, and repeated washings along with substrate addition. This process makes a bedside diagnosis nearly impossible. Newer techniques incorporate self contained units that are disposable and automatic and provide qualitative rather than quantitative results.
1.4
Newly Developed Diagnosis Technique
Currently at the Center for Nanomagnetics and Biotechnology (CNB) a diagnostic tool
has been developed by the Bio-diagnostics Group to determine key biomarkers that are the tell
6
tale sign of an AMI [10]. The diagnosis method is a modified ELISA utilizing superparamagnetic microspheres as the solid substrate onto which the base antibody can attach to. The remainder of the process closely resembles a traditional ELISA with a few minor adjustments. CNB has experience with magnetic immunoassays in the past by isolating erythrocytes from non-coagulated samples. The magnetic microspheres are first coated with a protein envelope that is capable of binding with a modified antibody. This antibody/magnetic microsphere complex allows for the selective capture of a target cell, the erythrocyte. When the magnetic microsphere solution is combined with blood serum the microspheres selectively bind to the erythrocytes, an example of the binding can be seen in figure 1.3. As this heterogeneous solution passed over a magnetic field the erythrocyte magnetic microsphere complex is separated from the background media with out the use of current techniques such as centrifugation with the added bonus of higher efficiency. This allows for similar procedures once requiring an entire laboratory to be completed with this new magnetic microsphere procedure.
Figure 1.3 Modified HSA microsphere attached to a erythrocyte, red blood cell.
In the process to capture and quantify AMI markers from a solution the magnetic microspheres are enveloped with a protein in the same way as was done with the erythrocyte 7
capture. Once the microspheres are coated they are complexed with a modified antibody. The super-paramagnetic microspheres can then be easily dispersed in a solution providing more direct interaction between the antibody and the antigen. When the microsphere complex needs to be isolated from the background media an external magnetic field is applied. [10] The procedure for the modified ELISA starts with the addition of a microsphere-antibody complex to an antigen solution. By inducing mixing the antibody coupled microspheres are allowed to interact and capture the antigen in an easier fashion, as opposed to be anchored to the bottom of a micro well dish. Once a suitable time has passed and the antigen is completely coupled to the antibody-microsphere complex the microsphere complex can be subsequently isolated from the background media via a magnetic field. This allows for the background media to be discarded without the loss of the microsphere complex. The next step would be to remove the magnetic field and add a solution containing the antibody-enzyme complex. Once the new solution has been added, mixed with the microspheres and enough time has elapsed to insure the coupling of the antibody-enzyme to the microsphere-antibody-antigen complex an external magnetic field is applied to isolate the newly formed sandwich from the background media. The isolated microsphere sandwich is then washed repeatedly to remove any excess enzyme. Upon completion a solution containing a substrate is added to the microsphere sandwich and allowed to react for a given amount of time. After which a reading of the concentration can be made via a spectrophotometer.
1.5
Microfluidic Integration
The next step is to take this bench top process and to miniaturize it to a "lab on a chip" to
reduce cost and for parallel processing. Collaboration with Sandia National Laboratories (SNL) allows for work to be done using Micro Electro Mechanical Systems (MEMS) technology to achieve this goal using SUMMiT TM V and SwIFTTM technologies. At the CNB one of the focuses is to incorporate the technology developed with in the center to create such a "lab on a chip" device. The benefits of creating a MEMS based device are [11]: The ability to work with smaller samples, leading to less expensive biological analyses. Better performance with less power input.
8
-
The analysis is done on a single silicon chip. Batch production makes for disposable use insuring sterility and reducing cleaning costs.
By incorporating the magnetic microsphere modified ELISA technique and MEMS based novel micro-pump with micro-channels it is hoped that such a device can come to fruition. MEMS technology is closely related to Integrated Circuit (IC) technology. The IC concept first evolved from the production of a monolithic circuit by RCA in 1955 and the first IC was produced in 1958 by Texas Instruments. The MEMS technology used in this work was produced by using surface micromachining techniques developed at SNL namely SwIFT. The SwIFT process was chosen because of the fact that the top layer in the die is made of Silicon Nitride which is transparent. This property allows for the quantification of the analyte via a spectrophotometer device. In order to successfully create a MEMS based device that incorporates the AMI detection process the behavior of the components at such a reduced scale must first be investigated.
1.6
Study Objective
The main focus of this work is to determine the flow characteristics of the fluid
components involved in the AMI detection process developed at the CNB at the micro level. This includes the characterization of fluid flow and mixing in characteristic micro machined silicon nitride micro channels using head pressure flow analysis as well as Fluorescent Correlation Spectroscopy. By concentrating on characterizing the flow of various fluids required by the AMI detection process in micro-channels the design of a microfluidics based AMI detection system can be put forth in future studies. By using a microfluidics based AMI detection system as a POC device it is thought that patient's lives may be saved and malpractice costs for hospitals would decrease.
1.7
Accomplished work
The following work was achieved during the course of this study:
9
1. Synthesis of heat stabilized human serum albumin (HSA): HSA magnetic particles were synthesized at the Center for Nanomagnetic and Biotechnology [Ref]. Adopting to protocol new particles were synthesized for testing in micro channels. 2. Preparation of antibody labeled magnetic particles: The steps to produce magnetic immunoassay were adopted according to the work of Manuel [ref]. These particles were then used to detect for AMI markers. 3. Training at Sandia National Laboratory: Training on microsystems was provided through an LDRD funding from Sandia National Lab. Over a three month span at SNL training on micro electro mechanical systems (MEMS) design and overall handling and manipulating of such devices was achieved. Handling of the devices included, but no limited to, probe station setup, voltage application via micropositioners, voltage manipulation via Labview, dry and wet testing of pumping mechanisms as well as failure analysis of mechanical and electrical components in MEMS devices. The most relevant work, as related to thesis work, came from flow measurement experiment done on micropumps and eventually microchannels. It was at SNL that an idea of how difficult a matter it is to introduce fluids to such dimensions is. 4. Characterization of fluid flow in microchannels: Pressure-flow rate characteristics for different fluids with and without suspended particles were studied at SNL and FSU. FCS for the microchannels was also conducted using the facilities in the department of chemistry at FSU.
1.8
Scope
The research presented in this thesis is concentrates on investigating the behavior of
homogeneous and heterogeneous solutions at the micro level. The first step was to develop a test protocol on the bench top detection mechanism so that it can be applied to a microsystem. Once a understanding of the processes involved in detection of AMI was achieved the vital fluid components were determined and applied to microchannels to determine their behavior at a microscale. This investigation of the behavior of liquids in microchannels became the main focus of this thesis. In AMI detection the behavior of the main fluidic components, water, phosphate buffered saline (PBS) solution and microspheres was examined in microfluidic
10
channels. The early stages of the investigation utilized MEMS based microchannels with the focus of the research ending up with the study of fluid behavior using PDMS-glass channels. The subsequent chapters discuss the components used in the studies undertaken by this thesis. The production of human serum albumin (HSA) magnetic microspheres, the process involved in the detection of acute myocardial infarction (AMI) and its potential application to microfluidic systems is discussed in chapter 2. Chapter three focuses on the general behavior of fluids within microchannels, the fabrication of microchannels and, microflow characterization techniques used in measuring flow velocities in microchannels. The work accomplished, results and discution of the results is represented in chapter 4 of this thesis. This chapter includes pressure driven flow studies in both MEMS based micro channels as well as PDMS-glass microchannels. Attempts to characterize flow using fluorescence correlation spectroscopy and microsphere interaction is discussed in this chapter as well. The content in chapter 5 provides a conclusion of the thesis.
11
CHAPTER 2 ACUTE MYOCARDIAL INFARCTION DETECTION
This chapter discusses the magnetic immunoassay used for detection of cardiac markers. The process for detection using manual bench top testing procedure was first developed by the CNB [10]. The objective here was to develop a stream lined process that allows for integration in the microsystem. The process utilizes magnetic immunoassay. The magnetic immunoassay used for the detection of AMI markers utilizes a standard solid-phase enzyme linked immunosorbent assay (ELISA). The ELISA takes advantage of the multiple epitopes that are found on the target protein, in this case myoglobin and FABP. A sandwich is formed by attaching two of the same target protein antibodies, each having their own specific epitope, to the same target protein. The epitopes were chosen because of their wide spacing from each other. One antibody is attached to an enzyme, alkaline phosphatase (AP), and the other is attached to a solid surface, a superparamagnetic microsphere (SPM). The use of the SPMs allows for easy separations and concentrations in large volumes. This allows for fast assays and possibly greater sensitivity when compared to current immunoassays. In this assay the antibody that is attached to the SPM is used to separate the target protein from the surrounding media and the enzyme attached to the second antibody hydrolyzes a substrate, para-nitrophenolphosphate (pNPP). The concentration of the reactant determines the concentration of target protein found in the sample.
2.1
Super Paramagnetic Microspheres
This section introduces magnetic Human Serum Albumin (HSA) microspheres by
providing a brief history, current trend of use at the CNB, as well as their synthesis and characterization. The goal of the characterization of HSA microspheres is to show size
12
distribution, surface features and magnetic properties. All of these characteristics are vital for proper HSA microspheres integration in a detection mechanism. 2.1.1 Background
Human Serum Albumin (HSA) microspheres have been used as a diagnostic tool since the mid 1960's [12]. They were first used to measure blood flow in implanted cancerous tumors with the help of radioactive tracers [12]. The trend of using HSA microspheres as a diagnostic tool reverberates to this day. They have been modified with and without the addition of iron oxide as passive and active targeted drug carriers [13-15]. The CNB has demonstrated the ability to modify the surface of HSA magnetic microspheres to allow the use of the microspheres as a tool to selectively separate a target from a heterogeneous liquid media by incorporating an external magnetic field [16]. At the CNB this work was furthered into the diagnostics with the modification of the HSA microspheres to quantify the concentration of biomarkers associated with an AMI [10]. 2.1.2 Synthesis
The synthesis and characterization of the HSA microspheres was carried out at the CNB and is well understood [17,18]. In this study the protocol that was reported in [17,18] is utilized in order to produce microspheres to carry out the experimental work to develop a consistent process that will be implemented later in the micro device. The microspheres used in this study are composed of maghemite, iron oxide, and Human Serum Albumin (HSA) then they are modified with an avidin protein envelope. The synthesis of the HSA microspheres is done by using a heat stabilized technique. The components needed to synthesize the microspheres are:[17] Human Serum Albumin (HSA) obtained from Sigma Chemical Company Cotton seed oil obtained from Sigma Chemical Company Iron Oxide (maghemite, ) synthesized in house Sorbitan Sesquioleate obtained from Sigma Chemical Company. Diethyl Ether obtained from Sigma Chemical Company. 3.0m, 1.2m, 0.8m, 0.65m, 0.45m and 0.3m nylon filter membranes obtained from Pall Specialty Chemicals
13
The procedure for the synthesis of HSA microspheres is as follows:[ref] 1. 2. 250mg of HSA was dissolved in a dispersion 75mg maghemite in 1ml of distilled water then added to 30ml of cotton seed oil containing 200l of Sorbitan Sesquioleate. To create the primary emulsion the mixture was subsequently shaken vigorously and sonicated using a Cole Parmer ultrasonic homogenizer for three 30 second intervals at an amplitude of 60% in a 4C ice-water bath. 3. The primary emulsion was then added drop-wise to 100ml of cotton seed oil heated to 130C while stirred at 1500rpm using a 3inch diameter Teflon paddle. The addition of the primary emulsion must be completed within 10 minutes. While keeping the mixture at 130C the solution was stirred at 1500rpm for 15 additional minutes. 4. The microsphere solution was allowed to cool and extraction proceeded. Extraction was accomplished by washing the solution, addition of diethyl ether and centrifuging then removing the supernatant for a total of three times. 5. After the final washing the dispersion of microspheres in diethyl ether was filtered successively using nylon filter membranes. All membranes containing microspheres were kept with the exception of the 0.3m membrane which was discarded. 6. The microspheres were subsequently dried and stored in a vacuum desiccator. Characterization
2.1.3
The characterization of the HSA microspheres was carried out in order to measure the average size of the particles and to ensure the magnetic particles encapsulation. Size and surface characteristics were investigated with the use of Scanning Electron Microscope (SEM). The surface characteristics, i.e. protein structure and nucleic acids, were investigated with the use of Atomic Force Microscope (AFM). A Superconducting Quantum Interference Device (SQUID) was utilized to determine the magnetic properties of the particles. Since these HSA microspheres will be dispersed in a fluid these properties must be understood because the magnetic force needed for isolation is proportional to the magnetic material volume, the magnetic properties, magnetic field strength and the magnetic field gradient.[17].
14
2.1.3.1 Scanning Electron Microscopy The micrograph showing the morphology of the microspheres shown in figure 2.1 was obtained from a dispersion. Figure 2.1 shows the round morphology as well as the size distribution associated with the microspheres. The size distribution ranged from 0.2m to 2m, the size histogram is shown as figure 2.2 [17]. To obtain a more uniform distribution of microspheres the dispersion can be vacuum filtered serially through a filter membranes that corresponds with the dimensions desired.
Figure 2.1 SEM micrograph of HSA microspheres in dispersion at 5000x [17]
14
12
10
Number of particle
8
6
4
2
0 0.2 0.6 0.7 0.8 1 1.4 1.8 2
Average particle diameter ( m)
Figure 2.2 Histogram showing particle size distribution of HSA microspheres found in dispersion [17].
15
2.1.3.2 Atomic Force Microscopy Atomic force microscopy allows for the investigation of the surface of the HSA microspheres in a much finer detail. A D-3000 (Digital Instruments, Santa Barbara) atomic force microscope was used to study the HSA microspheres immobilized on glass slides by poly(Llysine). Figure 2.3(a) shows the surface structure of the HSA microsphere in phase mode and figure 2.3(b) shows the surface structure of the same sphere in height mode on an enhanced scale. The cracks and crevices are due to heat cross-linking accomplished by heat stabilization [17]. The surface of the HSA microsphere plays a role in antibody conjugation that will be discussed in section 2.1.4.
(a)
(b)
Figure 2.3 (a) AFM of a HSA microsphere in phase maode, (b) AFM of HSA microsphere in the height enhanced mode showing crosslimking due to a heat stabilization process [17].
16
2.1.3.3 Superconducting Quantum Interference Device When analyzing the heat stabilized HSA microsphere in the SQUID there was an initial increase in susceptibility with temperature in the absence of a magnetic field. This was due to the fact that the particles had a random orientation of their magnetizations. When a magnetic field is applied along with an increase in thermal energy is experienced by the sample, the particles will have energy comparable to their energy barrier thus aligning their magnetizations in the direction of the applied field. It is this phenomenon that increases the magnetization as an increase in thermal energy is applied to the sample, this phenomenon is shown in figure 2.4. Due to the distribution of particle size, submicron to mare than 1m, there will be a range of blocking temperatures dependant on the particle size. This is why there was no observance of a blocking temperature. As the sample's thermal energy was being reduced and the magnetic field was kept constant the particles magnetic moment was fixed in the direction of the magnetic field. This phenomenon explains why there is almost no observable change in the susceptibility. The magnetization curve, found as figure 2.5 demonstrates, a symmetrical hysteresis loop about the origin of the graph. This phenomenon is characteristic of a material that is super paramagnetic in nature [17]. The super paramagnetic behavior states that the HSA microspheres will become magnetized in the presence of a magnetic field. Once the field is taken away from the particles there will be minimal residual magnetization left within the particle. In the case of the a magnetic based immunosorbent assay this is ideal because once an external magnetic field is removed and the microspheres still had residual magnetization they would agglomerate together making the assay useless. 2.1.3.4 Conclusion: HSA Microsphere Characterization The characterization section of the HSA microspheres showed that the size distribution for heat stabilized HSA microspheres ranges from 0.2 to 2m as determined through observing SEM micrographs. Through the use of an AFM protein cross linking due to a heat stabilization process could be witnessed and using a SQUID a blocking temperature for the microspheres could not be established but super paramagnetic behavior was observed.
17
6.00E-05
5.00E-05
4.00E-05
Susceptibility
3.00E-05
2.00E-05
1.00E-05
0.00E+00 0.00E+00 5.00E+01 1.00E+02 1.50E+02 2.00E+02 2.50E+02 3.00E+02 3.50E+02 T(K)
Figure 2.4 The temperature susceptibility plot for heat stabilized HSA microspheres [17].
2.50E+01 2.00E+01 1.50E+01 At 300K At 5K
Magnetization (emu/gm)
1.00E+01 5.00E+00 0.00E+00 -5.00E+00 -1.00E+01 -1.50E+01 -2.00E+01 -2.50E+01 -8000
-6000
-4000
-2000
0
2000
4000
6000
8000
Applied Field (Oe)
Figure 2.5 Applied field versus magnetization plot for heat stabilized HSA microspheres showing a super paramagnetic nature [17].
18
2.1.4
Surface Modification
In order to produce the magnetic immunoassay the surface of the magnetic particles will need to be modified in order to facilitate the coupling with the antibodies. The HSA magnetic microspheres left alone after production will not bind to the antibodies. In order to conjugate an antibody with a microsphere the microspheres must be encapsulated in an avidin layer. The antibodies are then biotinylated and the covalent interaction between the biotin and avidin provides the coupling. The binding capacity between avidin and biotin is a strong covalent interaction with a large binding capacity (Ka=1015/M). The avidination procedure was optimized by the CNB for these particular microspheres. The components needed are: - Avidin - SATA obtained from Pierce Biochemical - Dimethylsulfoxide (DMSO) - sulfo-SMCC obtained from Pierce Biochemical - Human Serum Albumin microspheres The procedure for the avidination of the HSA magnetic microspheres is as follows: 1. In order to conjugate HSA and avidin they must be reacted in a 4:1, HSA to avidin, molar ratio. The avidin was dissolved in Phosphate Buffered Saline (PBS) at a pH of 7.4 to a concentration of 10mg/ml 2. 3. 4. Prepare a SATA stock solution with a concentration of 13mg/ml in DMSO. 25l of the SATA stock solution is added to each ml of the avidin solution and reacted at room temperature in a vortex at low speed. (a) The HSA magnetic microspheres were dispersed in PBS, pH 7.4, at a concentration of 10mg/ml. (b) Add 3.3 mg of sulfo-SMCC to each ml of HSA magnetic microspheres and reacted for 30 minutes at room temperature in a vortex at low speed. 5. 6. Add the sulfo-SMCC/HSA magnetic microsphere solution to the thiolated avidin solution and vortex, at a low speed, at room temperature for a period of two hours. The mixture is then centrifuged and the first supernatant is saved for a BCA protein protocol (Pierce Biochemical), the remaining HSA magnetic microspheres enveloped in avidin are suspended in PBS and washed three more times. 19
In order to determine the amount of avidin incorporated onto the HSA magnetic microsphere a BCA protein test must be conducted. This will insure the proper amount of biotinylated antibodies are added to conjugate with the microspheres. The BCA test includes the protocol used and can be purchased from Pierce Biotechnology 2.1.5 Conclusion: Super Paramagnetic Microspheres
HSA microspheres were first introduced as a clinical diagnostic tool when Blanchard et al used them to investigate blood flow patterns in cancerous tumors. Currently HSA microspheres are utilized by the CNB in the detection of an AMI. In this AMI detection process HSA microspheres play a critical role. Determining the size, surface and magnetic characteristics are crutial to understand how they will perform once incorporated into the AMI detection system. In order for the HSA microspheres to be a useful tool in AMI detection their surface must be modified in order for antibodies to bind to them, this modification encapsulates the microsphere with avidin to allow biotinylated antibodies to bind with them.
2.2
Acute Myocardial Infarction Detection
The other necessary component in the detection scheme is the antibody. The antibodies
(Ab) are conjugated with either an enzyme or avidinated HSA magnetic microspheres. The antibodies utilized in this process are two complementary clones of monoclonal mouse antihuman cardiac myoglobin, mouse isotype IgG1: 908 and 4E2, as well as two complementary clones of monoclonal mouse anti-human cardiac FABP, mouse isotype IgG1: 10E1 and 9F3. These particular isotypes were chosen because their epitope proximity is roughly opposite from each other on the protein antigen (Ag). The clones were derived from hybridization of SP2/O myeloma cells with spleen cells of bald/c mice immunized with either human cardiac myoglobin or human cardiac FABP. The antibodies were produced and purchased from Research Diagnostics Incorporated. 2.2.1 Detection Setup In order to detect an AMI using the process developed at the CNB there are some steps need to be taken in order to make the microspheres and antibodies useful in the detection
20
process. These steps are discussed in the subsequent sections and involve biotinylation of antibodies to take advantage the biotin-avidin interaction and bind the antibodies to the microspheres. The surface modification, or avidination, of the microspheres was discussed earlier in section 2.1.4. Also discussed in this section is the addition of an enzyme to specific antibodies to complete the ELISA. 2.2.1.1 Antibody Biotinylation The use of a biotin-avidin or biotin-streptavidin coupling mechanism for coupling the labels to the antibodies provides ease and economy as well as a binding capacity, Ka, of 1015/M. Biotinylation was achieved using the EZ-link Sulfo-NHS-LC Biotinylation Kit purchased from Pierce Biotechnology. The long chain arm biotin was chosen to minimize any possible effects of steric hinderance, defined as the effect that the initial binding causes physical occlusion of binding sites within binding substrate, when conjugating biotin to the SPM as well as increasing sensitivity. The antibodies used in this assay are two complementary clones of monoclonal mouse anti-human cardiac myoglobin, mouse isotype IgG1: 908 and 4E2 type, and cardiac FABP, mouse isotype IgG1 10E1 and 9F3 type. The antibodies were chosen because their epitopes are placed widely apart on the myoglobin protein. The antibodies were purchased from Research Diagnostics, Inc. After biotinylation of the antibodies is complete it is necessary to determine the average number of biotin molecules found on the antibody to insure proper conjugation with the avidinated HSA magnetic microspheres. The determination of biotin incorporation is achieved through the use of the HABA method. Once the average number of biotin molecules is determined the antibodies are now ready to be coupled to their respective label: a super paramagnetic microsphere or an AP enzyme. 2.2.1.2 Alkaline Phosphatase Labeling This assay utilized Calf Intestinal Alkaline Phosphatase (AP), molecular mass of 140 kd, to hydrolyze pNPP, molecular mass of 371 kd, to produce a yellow reactant, absorbance 405 nm. The reaction is easily stopped by the addition of 2M NaOH. This provides the possibility to stop all reaction at exactly the same time and to store samples for future comparison and verification. The AP was purchased from Prozyme Inc. and had been conjugated with streptavidin.
21
Streptavidin is closely related to avidin, they are both tetrameric proteins with four binding sites that are similar to each other. Combine 800l of AP solution (800g) with 375l of biotinylated Ab (40g) and vortex while incubating at 30C. For more details please refer to [10]. 2.2.1.3 Anti-Myo 908/SPM Coupling In order to couple the antibody to the avidin coated SPM a few simple steps were followed. 1ml of anti-myo 908 biotinylated antibody solution, at a concentration of 0.1mg/ml, was combined with 7.6 ml of avidin coated SPM solution that had a concentration of 0.01g/ml. These figures were found by calculating the minimum number of biotin groups associated with each antibody using the HABA method (Pierce Biotechnology). In this case a minimum number of 15 biotin groups were associated per antibody giving 10nmoles of Avidin needed to couple with 0.1 mg of 908 antibodies. Each 0.076 g of Avidin coupled SPM solution has 10 nmoles of Avidin, or 7.6 ml of solution at 0.01 g/ml. Once the antibody and SPM solutions were mixed, the combination was then incubated at 300C for thirty minutes while mixing[10]. 2.2.1.4 Anti-Myo 4E2/AP Coupling Streptavidin coupled AP was purchased from Prozyme, Inc. in a 1mg/ml concentration. 800 l of Streptavidin coupled AP (800 g) solution with 378 l of anti-myo 4E2 biotinylated antibody (40g) solution. These figures were found by calculating the minimum number of biotin groups associated with each antibody using the HABA method. In this case a minimum number of 15 biotin groups were associated per antibody giving 4 nmoles of streptavidin needed to couple with 40g of anti-myo 4E2 biotinylated antibody. Assuming that streptavidin reacts in a 1:1 ratio with AP then 4 nmoles of streptavidin is equal to 0.8 mg of streptavidin which in turn is equal to 800 l of streptavidin conjugated AP solution, 1mg/ml concentration. Once the antibody and AP solutions were mixed they were subsequently incubated at 300C while mixing [10]. 2.2.1.5 Anti-FABP 10E1/SPM Coupling The coupling of the avidin coated SPM to the antibody followed some simple steps. 1ml of anti-FABP 10E1 type at a concentration of 0.1 mg/ml was combined with 88 l of avidin coupled SPM solution. These figures were found by calculating the minimum number of biotin 22
groups associated with each antibody using the HABA method (Pierce Chemical Co.). In this case a minimum number of 20 biotin groups were associated per antibody giving 13.2 nmoles of avidin needed to couple with 0.1 mg of anti-FABP 10E1 type. The volume of avidin coated microspheres was determined to be 88 l due to the fact that 13.2 nmoles of avidin were found in the SPM solution. Once the solutions were mixed together they were incubated at 300C while mixing.[10] 2.2.1.6 Anti-FABP 9F3/AP Coupling Streptavidin coupled AP was purchased from Prozyme, Inc. in a 1mg/ml concentration. 1000l of streptavidin coupled AP (1000g) solution with 375 l of anti-myo 4E2 biotinylated antibody (37.5g) solution. These figures were found by calculating the minimum number of biotin groups associated with each antibody using the HABA method. A minimum of 20 biotin groups were calculated per antibody giving 5.5 nmoles of streptavidin needed to couple with 37.5g of 9F3 antibody. Assuming that streptavidin couples in a 1:1 ratio with AP then 5.5 nmoles equals 1mg of streptavidin (1000l of AP solution). Once the antibody and AP solutions were mixed they were subsequently incubated at 300C while mixing.[10] 2.2.1.7 Human Serum Albumin (HSA) Saturation The HSA was added to minimize a secondary signal that was seen in preliminary experiments. What the HSA does is inhibit the avidin biotin interaction between the enzyme labeled antibody and the avidinated surface of the microsphere. The HSA was purchased from Sigma-Aldrich and was added to the antibody label conjugates. The HSA saturation was achieved by incubating the test sample amounts with 1ml of HSA, 50g/ml dissolved in TBS, at 300C while mixing. The HSA was added to block the enzyme linked antibody from interaction with the microsphere causing a secondary signal [10]. 2.2.1.8 Conclusion: Detection Setup Many different components must come together in order for the AMI detection system to work properly. Each of these components must undergo some sort of modification and once they are modified they can be combined together to create the working principle for AMI detection. The previous section, 2.2.1, dealt with the protocols utilized to setup the AMI detection 23
components. Once these components have been synthesized they are ready for the use in the detection process. 2.2.2 AMI Detection Experiments
The AMI detection process uses the modified components described in section 2.2.1. The separate components combine to for the detection system with the use of general items found in a typical biological laboratory, i.e. pipetors, spectrophotometer, cuvettes and a magnet. When used in a designed protocol the system can determine the concentration of myoglobin and FABP found in an unknown solution. 2.2.2.1 Overview of AMI detection Process The immunoassay consisted of the AP labeled antibody, SPM coupled antibodies and an antigen solution consisting of either myoglobin or FABP. The tests were performed using 3.5ml disposable acrylic cuvettes (Perfector Scientific, Inc. Altascadero, CA) used for spectrophotometry. The spectrophotometer used was a Turner model sp-830. The spectrophotometer measured the quantity of pNPP hydrolyzed by AP at a wavelength of 405nm. Five minutes was allotted to insure that the formation of the microsphere-antigen-enzyme complex took place. Two 1.2 Tesla magnets were used to isolate the reactants from solution. The reactants were then repeatedly washed with tris buffer saline (TBS), total volume of the wash equaled 9ml, to eliminate background media, example of process shown as figure 2.6. The standard assay starts by introducing 2ml of a known concentration of Ag, either myoglobin or FABP solution. At this point 1ml of the Ab-SPM coupled complex solution, HSA saturated, was added to the Ag solution. The solution was then manually agitated. Post agitation 500 l of Ab-AP conjugate, HSA saturated, was added and manually agitated. The solution was allowed to react for 5 minutes. Through the application of the 1.2 Tesla magnets the SPM complex segregated to the side of the cuvette where the magnets were located. While keeping the magnets engaged the solution was decanted and subsequently 3 ml of TBS was added. The magnets were removed from the cuvette and the complex was mixed in the TBS. The magnets were once again applied and the
24
Process Overview
Blood sample with Myoglobin
Add MS-908 Immunoassay
Apply field
Red Blood Cell Platelette Myoglobin Microsphere908 immunoassay AP-402 immuno-assay
Wash out
Add AP-402 Immunoassay
Apply field and wash out
Add PNPP
Add 2M NaOH after two minutes
Spectrometry
405 nm
Figure 2.6 Schematic of AMI detection process
solution was decanted. The washing process was comprised of this procedure and was applied several times until a total wash volume of 9ml was reached. The addition of 50l of pNPP solution was added to the washed solution, mixed and allowed to react for 2 minutes. After the two minute reaction period, 100l of 1M NaHO was added to the solution to inactivate the AP. A spectrophotometer reading at 405nm was subsequently taken. 2.2.2.2 Myoglobin Detection To determine the efficacy of the myoglobin detection aspect of the assay, 5 different concentrations of myoglobin were used, 0.0g/ml, 0.05g/ml, 0.10g/ml, 0.15g/ml, and 0.20g/ml. These concentrations were then used in the aforementioned experimental setup using the 908-SPM conjugate, HSA saturated, as well as the 4E2-AP conjugate, HSA saturated [10].
25
2.2.2.3 FABP Detection To determine the efficacy of the FABP detection aspect of the assay 3 different concentrations of FABP were used, 0.0g/ml, 0.005g/ml and 0.020g/ml. These concentrations were then used in the aforementioned experimental setup using the 10E1-SPM conjugate, HSA saturated, as well as the 9F3-AP conjugate, HSA saturated [10]. 2.2.2.4 Conclusion: AMI Detection Experiments In order to determine the concentration of Myo or FABP in an unknown solution the AMI detection protocol must be followed. The protocol was designed using equipment readily available in a biological science laboratory. The protocol introduces microsphere bound antibodies to a blood solution. This heterogeneous solution is the introduced to a magnetic field and washed. Upon washing the magnetic field is removed and enzyme labeled antibodies are added to the solution, a magnetic field is applied and the solution is washed a few times. A pNPP solution is added and the solution reacts for a given amount of time and then neutralized with 1M NaOH solution and subsequently the transmission characteristic at 405nm is determined. Once the concentration curves are determined an unknown solutions concentration of Myo and FABP can be determined. It is on this principle that a micro AMI detection system will be based on.
2.3
AMI Detection Applied to Microfluidics
Now that the process for AMI detection has been reviewed the process must be packaged to produce a viable detection device. This is one of the goals at the Center for Nanomagnetics and Biotechnology. The process applied to the macro scale will be the same process that is applied to the micro scale, a schematic of this process is shown as figure 2.9. One of the differences between the macro detection protocol and the micro detection protocol will be the fact the the volume of the working fluids will be much lower and therefore a separate protocol with adjusted volumes must be devised. In the mean time the first basic step involved in realization of this device is to determine how fluid and microspheres will behave at the micro
26
level. For this to occur microchannels are needed and their flow characteristics must be investigated.
Flow Diagram
Blood sample with Myoglobin
PBS Water
AP-4E2
PBS Water
PNPP
2M NaOH
MYO- MS Magnetic Washout Station Magnetic Washout Station
MYO- MS-AP
Two Minutes
Spectrometry
Microsphere 908 Immunoassay
RBCs, WBCs, Plasma, other Proteins
Excess AP-402
Figure 2.9 Schematic of the AMI detection process that needs to be applied to the micro scale.
There are many differences associated with working with fluids in a micro environment when compared to a macro environment. For example mixing becomes a crucial issue, because of the small dimensions the Reynolds number is low, but how low is the Reynolds number in the channels that will be attempted to be used in the micro AMI detection device. Since the detection protocol's vital component are the HSA microspheres, how will the microspheres behave in these micro channels? Since the device will be developed at the micro level will there be behavior not witnessed before when the flow in the channel is pressure driven? The next chapters of this thesis attempt to provide a better understanding of field of microfluidics and how it differs from macro flow regimes. A discussion on the type of channels used in this work includes the fabrication techniques how to work with introducing fluids to micro scale channels. Finally a representation of the studies conducted in this thesis will be presented showing how the different fluids and suspensions interact with micro channels. These studies included qualitative and quantitative studies carried out at both Sandia National
27
Laboratories in Albuquerque, NM as well as studies conducted at the FAMU-FSU College of Engineering and the Department of Chemistry at the Florida State University located in Tallahassee, FL.
28
CHAPTER 3 MICROFLUIDICS
This chapter introduces the microfluidic devices developed for the purpose of testing the behavior of fluid motion in the miniaturized devices. The main factors governing the need for miniaturization of clinical diagnostics are low sample consumption, fast analysis time, device integration, and the minimizing of cross contamination through the utilization of single use devices [19]. In order to achieve miniaturization there are many technical issues that must be tackled such as fluid behavior, in particular when it has suspended nanomagnetic particles, at the micron level, the delivery of the fluids and mixing of fluids at the micro scale. These issues require more investigation, but the fabrication of these micro-devices is no longer an issue. There are many different approaches to the fabrication of microfluidic devices ranging from the simple, no need for extremely specialized equipment, to the complex, the use of silicon based technologies requiring clean rooms and highly specialized processes as well as equipment. The following sections found in this chapter discuss the fabrication techniques used in creating two types of microchannels, MEMS based and Poly(dimethylsiloxane)-glass (PDMS-glass). This chapter also includes microfluidics theory and schemes used in the characterization of microchannels.
3.1
Microchannel Fabrication
This section aims to introduce microchannel fabrication by providing a history of MEMS
devices and then an introduction to PDMS-glass microchannel fabrication. Micro Electro Mechanical Systems (MEMS) technology is closely related to Integrated Circuit (IC) technology. The IC concept first evolved from the production of a monolithic circuit
29
by RCA in 1955 and the first IC was produced in 1958 by Texas Instruments. Wet etching techniques for the use on silicon wafers in the IC industry were first used in the 1960 and 1967, isotropic and anisotropic etching respectively [11]. These types of etches are instrumental in the production of surface micro-machined silicon based MEMS devices. There are many different ways to create microstructures that can be incorporated into MEMS devices. Some of the most common forms of micromachining are bulk micromachining, surface micromachining, wafer bonding, LIGA/SLIGA, micro EDM, 3-D lithography, laser micromachining and focused ion beam milling [11]. The technology used in the production of the channel used in this work is surface micromachining technology developed at Sandia National Laboratories (SNL). The general form of the fabrication technique is as follows; a single crystal silicon n-type wafer with a crystal orientation of <110> is used as the substrate, a thin film is produced on its surface, positive resist photolithography is used to pattern the underlying thin film, etching of the thin film ensues, these steps can be repeated a number of times, once complete the silicon crystal with surfaces modifications can then be diced and packaged accordingly [20]. To fabricate MEMS devices by using surface micro-machining techniques there must be structural, sacrificial, electrical isolation and substrate components. In general the structural component must have desirable mechanical and electrical properties, such as low stress, conductivity and insulation. The sacrificial layer must be stable through the deposition process and be readily etched in an etchant that will not etch the mechanical component. Both these materials must be compatible with an Integrated Circuit (IC) fabrication facility. The following subsections will discuss the ways that SNL can produces MEMS based microchannels as well as a way to produce microchannels that does not involve surface micromachining of silicon. 3.1.1 SUMMiT V TM
At SNL the process used is titled with the acronym SUMMiT V TM, which stands for Sandia Ultra-planar, Multi-level MEMS Technology 5. In fabrication one wafer produces approximately 63 die, a die is composed of 8 2.82mm by 6.34mm modules. A photolithographic process is used to pattern thin film layers. The masked used for the photolithography defines a single die, this mask is then stepped across the wafer and provides a resolution of 0.3m. [20].
30
The SUMMiT V TM process stylizes single crystal silicon as the substrate, polysilicon as the structural component, electrical isolation is achieved through the use of silicon nitride and the sacrificial layer is silicon oxide [20]. The use of this process enables the designer to create complex systems because of the fact that there are five levels for the designer to work with, please refer to figure 3.1. In the case of microfluidic channels each polysilicon layer, four in total, has the possibility of becoming either a channel top or bottom means that there are 4!=24 different configurations possible [21].
2.25 m MMpoly4
2.0 m SacOx4
2.25 m MMpoly3
2.0 m SacOx3 1.5 m MMpoly2 0.3 m SacOx2 2.25 m SacOx1 0.3 m MMpoly0 0.3 m Silicon Nitride 0.63 m Thermal SiO2 1.0 m MMpoly1
6 inch Substrate, 675 m thick Wafer, <100>, n-type
Figure 3.1 SUMMiT V TM layer terminology and order
The process starts with single crystal silicon 6 inch, 675m thick, wafer substrate commonly used in IC technology making it readily available for this process. The silicon substrate plays a role in the overall process by providing a foundation onto which all subsequent layers will be deposited onto. Since it is used as the foundation layer the silicon must be precisely flat and devoid of surface imperfections that could be exaggerated through the process. 31
To ensure that the deposited layers added to the substrate are flat the sacrificial oxide layers three and four are chemically and mechanically polished post deposition [20] and etched with hydrofluoric acid, HF [22]. The first layer to be deposited onto the silicon substrate is silicon dioxide. The layer is deposited via Low Pressure Chemical Vapor Deposition (LPCVD) and is 0.63m thick. This layer is not used on its own but in concert with the layer directly on top of it to provide electrical isolation of the upper layers. The layer found directly on top of the silicon dioxide layer is a layer of silicon nitride. This layer is deposited on top of the silicon dioxide layer using LPCVD and is 0.8m thick. Upon the deposition of the second layer, the silicon nitride layer, the first structural layer is deposited. This polysilicon layer is deposited upon the electrical isolation layers and acts as the ground plane onto which the designed structures can be attached to. In figure 3.1 this layer is named MMpoly0, is precisely 0.3m thick and is deposited using LPCVD. This ground plane must be placed under structures to prevent electrostatic attraction between structures. A sacrificial silicon dioxide layer, SacOx1, is encountered next, it has a thickness of 2.0m and is deposited using LPCVD. This layer is primarily used to space the next structural layer from the ground plate. The SacOx1 layer can be cut into as well, this cut will provide a way for the structural layer on top of it to attach to the MMpoly0 ground plane. The parts of the SacOx1 layer that are to be removed are done so by using a positive photoresist. The process uses a positive photoresist, this means that a layer of the resist is placed on the oxide, the part of the layer that is exposed to UV radiation is removed allowing for an etchant to interact with the oxide layer. The area where the photoresist is present remains protected until it is stripped off of the oxide layer. Located on top of the SacOx1 layer is 1m thick layer of structural polysilicon called MMpoly1, deposited by LPCVD. Found directly on top of the MMpoly1 layer is a 0.3m layer of silicon dioxide. This layer, SacOx2, is in most instances removed immediately after it is deposited, allowing for the next layer, a structural polysilicon layer MMpoly2, to be directly deposited onto the MMpoly1 layer. The combined height of the two layers is 2.5m making the MMpoly2 layer 1.5m thick. The next layers are SacOx3, MMpoly3, SacOx4 and MMpoly4 are located above the MMpoly2 layer. The silicon dioxide layers SacOx3 and SacOx4 are both 2.0m thick and are
32
deposited using Plasma Enhanced Chemical Vapor Deposition (PECVD). The structural polysilicon layers, MMpoly3 and MMpoly4, are both 2.25m thick and are deposited by using LPCVD. They have the corresponding roles discussed earlier.[20] 3.1.2 SwIFT TM
Though the use of the SUMMiT V TM process allows for elaborate and complex designs, the use of polysilicon makes the designs opaque. Besides not being transparent a recent study has shown that silicon to be the least biocompatible in a test group that consisted of gold, silicon, silicon nitride, silicon oxide and a SU-8 a photoresist [23]. SNL has developed a modified SUMMiT V TM process and named it SwIFT TM, which stands for SUMMiT V TM with Integrated Fluid Technology. SwIFT TM utilizes the insulating and optical properties of silicon nitride to create transparent microfluidic channels that are completely isolated from polysilicon or silicon [21]. The properties of SwIFT TM and the SUMMiT V TM process remain the same, what changes is the addition of a SacOx3 deep cut. There is also a layer of silicon nitride that is deposited via LPCVD on top of the SacOx3 layer [21]. When the SacOx3 deep cut is present and silicon nitride is deposited on the SacOx3 layer the silicon nitride reaches the silicon nitride layer found below SacOx1 as shown in figure 3.2. This feature allows the designer to create a square tube of a fixed height but variable width and geometry. In order to allow a fluid to enter in these channels they must have a Bosch backside etch figure 3.3. What this provides is an entrance or exit for the flow. Without this feature the channel itself would be left with sacrificial oxide left within it because the etchant would have no way of entering the channel. Post fabrication half of the channels were coated with a hydrophobic layer, self assembling monolayer. This was done to investigate the differences between the two different types of channels, hydrophobic and hydrophilic, in hopes that one type may aid in the detection process in some unforeseen way. The self assembling monolayers were originally used in stiction issues associated with fabrication. Up to this point it thought that this is the first time this method was applied to these channels.
33
2.25 m MMpoly4
2.0 m SacOx4
2.25 m MMpoly3 0.8 m Silicon Nitride 2.0 m SacOx3 SacOx3 Deep Cut 1.5 m MMpoly2 0.3 m SacOx2 1.0 m MMpoly1
2.0 m SacOx1 0.3 m Silicon Nitride
0.8 m Silicon Nitride 0.63 m Thermal SiO2
Substrate, 6 inch wafer, 675 m thick, <100>, n-type
Figure 3.2 SwIFT TM layer terminology and order
Figure 3.3 Bosch backside etch example. The picture is in focus on the nitride layer, the center of the holes. The portion of the picture that is not in focus is the underside of the MEMS based microchannel, single crystal silicon.
34
3.1.3
Poly(dimethylsiloxane)-glass Microchannels The fabrication of microchannels through the use of polymers is simpler, easier and less
costly then fabrication techniques that utilize surface micromachining of silicon. One of the draw backs of using polymers is the fact that dimensions achieved by using surface micromachining can not be achieved using polymers. Though this may be the case poly(dimethylsiloxane)-glass (PDMS-glass) microchannels are widely used in biodetection, microreactors and microfluidics to name a few [24]. The production method for creating PDMS-glass microchannels is referred to as soft lithography. This process starts with the creation of a master. The master could be produced in several ways, the method used in these experiments utilizes a typical office laser printer to reproduce a designed channel onto plain white paper. This image is then reduced by photographing the design onto 35mm black a white film. The film is then used as a 1:1 contact lithography mask on a silicon wafer, three inches in diameter 500m thickness, that is coated with photoresist. The wafers are etched prior to use, then a negative photoresist, coating that is exposed to light polymerizes (negative resist), is diluted and spun onto the wafer, at two different rpms. It is the thickness of this layer that will determine the depth of the microchannel. The spin coated wafer is then heated on a contact hot plate. Following this step the 35mm film contact lithography mask is placed directly onto the spin coated wafer and exposed to ultra violet (UV) radiation. Post exposure the spin coated, UV irradiated wafer is heated on a contact hot plate, cooled and the unexposed resist is removed via propanol. Once the excess photoresist is removed the master is created. The actual negative of what the microchannels will become. `To create the microchannels PDMS is poured over the master while in a Petri dish. The PDMS resembles an epoxy where two parts are mixed together, a base and curing agent, in a 10:1 ratio, respectfully [24]. After a couple of hours in a vacuum at 65C the PDMS is completely polymerized and is ready to be cut out of the Petri dish. The next step requires the newly made PDMS microchannels and a 45mm x 50mm glass coverslip to be placed together in air plasma for 45s. This process modifies the surfaces of the PDMS and the glass so when they come into contact a seal is created. A schematic of this procedure can be found as figure 3.4. [28]
35
LIGHT
35mm Film Photoresist
1
Master
2 PDMS
3 PDMS 4 PDMS
Microchannel Glass
Figure 3.4 (1) photolithography is preformed to produce the master, (2) the PDMS positive is created using the master, (3) remove the polymerized PDMS from the master, (4) seal the PDMS against a glass cover slide to create the microchannels.
3.2
Microfluidics
The content of this section is the theory behind microfluidics and its similarity to macro
flow regimes. Equations used in the subsequent studies are presented and discussed as well as explained in the following section. When dealing with fluids at the macro scale, many behaviors of a fluid are well know and follow well know modeled equations, this case being true for laminar flow regimes. The characteristics of fluid behavior in submicron system are still the subject of many studies. Surface tension is a great factor in the design and characterization of micro-channels, it is because of surface tension that special steps must be taken in order to etch and characterize correctly. The continuum assumption must also be checked to make sure that macro equations remain valid for the micro flow regimes. When dealing with the macro flows it can be generally assumed that one can treat the fluid as a continuum. Meaning that quantities such as velocity, pressure, density are assumed to 36
be defined everywhere within the flow. Does this assumption still hold for micro flow regimes? An easy way to answer this question is that if the numbers of molecules in a certain length scale are closely packed then the general answer is yes. This may be the case for many liquids but may not be the case in terms of a gas flow. In order to determine an appropriate point value corresponding to a continuum flow there are varying length scales for liquids and gasses. To determine these length scales stationary statistics of less then 1% variation must be achieved. The stationary statistic is achieved when 104 molecules in a sampling is used [11]. The values found in the denominators of the fractions are the number densities of N2 for the gas and the number density of H2O for the liquid. The number density is simply the number of molecules divided by the volume that those molecules are found in.
3
L gaspt
10 3 . 10
25
4
m
3
9 70. 10 m
Equation 3.1
3
L liquidpt
10
28
4 3
2 . 10 m
8 . 10 m
9
Equation 3.2
Equations 3.1 ad 3.2 show that the corresponding length scale associated with a valid continuum assumption is equal to 70nm for gas flows and 8nm for liquid flows. The length scale is associated with a fluids kinematic properties, such as velocity and acceleration, as well as thermodynamic properties, such as pressure and density. But in order for a fluid to be considered a continuum all of its properties must be valid.[25] The continuum assumption must hold true in the case of viscosity and diffusivity, transport quantities, as well. In order to determine weather the transport quantities behave as a continuum the molecules of the fluid must interact with each other more than they interact with their flow boundaries. For this calculation an arbitrary cube length 10 times greater then the molecules interaction length will be used as the measurement point. For a gas the interaction length is determined to be the molecules mean free path, which for most molecules is on the order of 100nm. On the other hand liquid molecules are constantly interacting with each other
37
therefore the mean free path will not work as a good estimate; in a liquid case we use the number density of the fluid.
3
L gastr
3 10 . 100 nm 10
6
m
Equation 3.3
3
L liquidtr
10 2 . 10
3
4 . 10 m
3
9
m Equation 3.4
28
In order for the continuum assumption to hold both point and transport quantities must hold, keeping this in mind the greater of the two equations corresponding with gaseous and liquid flow is chosen. For a gas the mean length is equal to 10-6 m, or 1m, and for a liquid the length scale is equal to 810-9 m, or 8nm so to approximate to value the length scale for liquids to be treated as a continuum will be 10nm. These values provide a guideline for what to use and do not provide a black and white answer.[25] In the case of the channels used in these studies the smallest length scale is 6.8m well above the guideline of 10nm therefore the continuum assumption will be appropriate to assume in these studies. Another major factor influencing liquid flows on the micro scale is surface tension. In macro scales surface tension is usually negligible and is overcome by the force exerted by the liquid fluid quite easily. In surface tension there are liquid-liquid interactions, liquid- gas interactions, as well as liquid-solid interactions. In manufacturing silicon based microstructures etching of the sacrificial oxide layers is a necessary and it is in this instance that surface tension plays a major factor. When the oxide layer is etched out the etcha nt, a liquid, remains in its place. The surface tension between the liquid and solid makes it difficult to remove; in many instances the surface tension present is so great that the manufactured microstructure fail due to the enormous pressures generated. To counteract this phenomenon surface micromachining techniques have added dimples to structures and to long tubes etch release holes have been added. In pressure driven microflows of an incompressible fluid such as water and no other body forces are present with the exception of gravity the general continuum expression is given as the following incompressible continuity equation:
38
u j x j
=0
Equation 3.5
And the momentum equation is:
u u i + u j i t x j
ij = + bi x j
Equation 3.6
where ? is the density, ui is the ith component of the velocity vector u(x,t), t ij id the stress tensor and bi is the body force per unit mass. If flow in a channel where the velocity field is unidirectional and there is no acceleration of the fluid the fluid is said to be fully developed. As long as the temperature is constant along with the fluid properties and the flow is steady, the equation for the streamwise velocity profiles become a Poisson equation, where we can assume no slip on the wall [29]: Due to the large heat capacity of liquids internal heating due to viscous dissipation is less significant in liquid flows and therefore can be ignored [29].
2u 2 u 1 d + = ( p - g x x ) y 2 z 2 dx
Equation 3.7
Flow resistance in a microchannel is another way to characterize performance. Fluid resistance is defined as the ratio of the pressure drop to the volumetric flow rate, shown as equation along the length of the channel [11].
R
P Q
Equation 3.8
39
Where the pressure drop, P, is equal to equation 3.9 where it can be calculated as a function of channel dimensions as well of the mean stream velocity. The change in pressure can also be determined experimentally by applying equation 3.13. :
P Re .f .
L 2Dh
2
.u
Equation 3.9
Where is equal to the dynamic viscosity of the fluid, L is the length of the channel, u is the mean velocity of the fluid within the channel, Re is the Reynolds number of the flow (equation 3.10), f is the Fanning friction factor (equation 3.11) and Dh is the hydraulic diameter (equation 3.12).
Re D h . u D h . u
Equation 3.10
Where is the density of the fluid, is the dynamic viscosity of the fluid and is the kinematic viscosity of the fluid.
f 64 Re
Equation 3.11
Dh 4
A P wet
Equation 3.12
Equation 3.13 shows the relationship between the height a fluid above a certain datum and the pressure that it creates in relationship to that datum. When dealing with gauge pressure the pressure differential, P, would simply equal equation 3.13.
P gh P
Equation 3.13
Where A is the cross sectional area of the channel and Pwet is described as the perimeter of the channel that is in direct contact with the flow [25]. 40
These equations hold true until the shear rates experienced by the fluid in the channel reach an upper limit and the fluid begins to behave in a non-Newtonian fashion, as far as the concern in this work the flow regime is considered Newtonian. This occurs when the strain rate approximately exceeds twice the molecular frequency scale, refer to equation 3.14.
d u = 2T -1 dt y
Equation 3.14
Where T is the molecular time scale expressed by equation 3.15.
m 2 2 T =
1
Equation 3.15
Where m is the molecular mass and and are the characteristic length and energy scale for the molecules. For liquids such as water the time scale T is extremely small and the threshold for non-Newtonian behavior is extremely high [30]. This is not the case for high molecular weight polymers though and the linear stress-strain relationship breaks down [30].
3.3
Microchannel Characterization Methods
To determine the characteristics of any type of internal flow many methods can be
implemented from determining pressure flow characteristics to seeding the flow solution and using external devices to determine the flow patterns and incorporating sensors within the channel to determine laminar flow characteristics. Through characterizing novel channels a greater understanding of microflow regimes can be reached. This section will discuss some methods that are implemented in this study and are becoming standard in the literature for microchannel characterization.
41
3.3.1 Pressure Flow Characterization
The most basic way to characterize a channel is by determining the pressure flow characteristics of the channel. This experimental procedure is important as the first step to characterize a novel channel or fluidic system. Many microchannels are made by using a poly(dimethylsiloxane) PDMS-photoresist method and seen in 1988 for use as microdetection tool [26]. The PDMS channels are more robust and can be handled in a different manor then silicon based devices. PDMS channels cannot achieve the small dimensions that are capable with surface micromachining. On the other hand PDMS channels cost a few dollars to produce and silicon based surface micromachined channels cost many times that number, on the order of $100.00 when small batch productions are used. As the batch number is increased the cost of silicon based microchannels drop to cents per device [11]. Since silicon based microchannels are not as prevalent, steps must be taken to first see have much pressure is too much and what the corresponding flow rates are for a given pressure head.
3.3.2 Micro-Particle Image Velocimetry Characterization
The principle behind particle image velocimetry (PIV) is that fluid velocities can be evaluated by recording the position of images produced by small tracers suspended in the flow at successive time instants. The images are obtained by illuminated the seeded flow by a light sheet generated by a laser. Two images are subsequently taken in a short period of time. When these images are compared a velocity vector can be determined [31] PIV methods for characterizing the flow within channels use either a seeded fluid with laser illumination or sensors based within the channel with a seeded flow to determine the characteristics. The most common method for characterizing the flow within a microchannel is to use a method name micro-particle image velocimetry (micro-PIV). Micro-PIV is considered to be a Full-Field method, as opposed to a point wise method i.e. laser doppler velocimetry, determining P and providing a detailed view of the flow. These measurements play a useful role in optimizing mixing, pumping and filtering. Micro-PIV works on the same concepts as PIV with the setup such that the investigation of microflows is made possible. PIV uses a seeded fluid that is photographed at two different times. These images are then sectioned into smaller interrogation regions and the motion is determined using a statistical technique called cross-correlation. Recently this technique has 42
been coupled with resistance temperature detectors coupled within the channel to determine the effect that temperature has on microflows [27].
3.3.3 Fluorescence Correlation Spectroscopy
There has been another development in flow characterization using the full field method where fluid flow characterization in microchannels is accomplished through the use of fluorescence correlation spectroscopy (FCS) [ref]. FCS uses fluorescent encapsulated nanospheres to seed the flow. Then through the use of a confocal microscope set up, which includes but is not limited to an excitation laser, a PC for control as well as data capture and photon detection devices. This technique is available at FSU at the Department of Chemistry at a lab under the direction Dr. of Kenneth Weston. Dr. Weston is a participating faculty at the CNB. In the system the excitation laser activates the nanosphere that travels through a known confocal detection volume, the displacement and special distribution of the fluorophone is monitored in time. Fluorescence intensity is recorded continuously and the transit time of single particles through the control volume of the laser can be extracted from the decay time of an autocorrelation curve [28]. The benefits of such a system are apparent in the size of the fluorescent particles used, which are on the order of 10 to about 100 nanometers, compared to PIV systems that utilize particles of several hundred nanometers. The PIV particles will start to interfere with the flow by distortion or by blockage. There are some drawbacks to using the FCS system for flow characterization. The nanospheres have a tendency to adsorb to the walls of the channels making near wall velocity determination very difficult, not to mention the expense of purchasing the fluorescent nanospheres and the need for the channel in question to be transparent to UV light. The typical FCS set up includes optics, scan control as well as data collection and analysis as shown in figure 3.6. The optics direct the output beam of a frequency doubled Nd:YAG (=532nm) into the back side of an oil immersion objective that is rigidly mounted to an optical table. The objective not only emits the laser radiation but also collects the fluorescence emitted by the region that is illuminated. The emitted fluorescence is then directed through a diachronic beam splitter, where the incoming laser radiation gets reflected into the objective, through a 150m pin hole to a holographic notch filter and a long pass edge filter.
43
The fluorescence signal is then split into two beams using a second diachronic beam splitter. The resulting separated fluorescent signals are then focused onto two separate silicon avalanche photodiode (APD) photon counting modules. A schematic of the optic set up can be seen as figure 3.4. Scan control is achieved through the use of DC servo motorized actuators mounted to a XYZ translation stage onto which the specimen is placed. The control of these servo motors is achieved by using a PC. For high precision scans a servo controlled piezoelectric XY translator system is used with a limited scan range of 100m. Data collection and analysis starts with the silicon avalanche photodiode photon counting modules that produce a voltage when a photon is detected. This voltage is then routed to a counter-timer board and recorded with 50ns resolution. Data acquisition as well as real time data analysis is achieved through the use of Labview 6.1 a typical representation of the instantaneous data is shown in figure 3.5.
Figure 3.5 Typical representation of APD data. As the velocity increase the number of peaks within a period of time increases as well, the opposite is true when the flow decreases.
In FCS the confocal detection is used to limit the detection region to extremely small dimensions. The dimensions are so small that the detection volume only allows for one particle excitation at a time, given the correct concentration of fluorescent particle. Therefore the intensity fluctuates in time according to when the particle is in the detection volume. This fluctuation is correlated to the diffusion or flow rate of the fluid within the microchannel. If flow rates are high enough diffusion can be considered a minor factor and flow the dominant. For more information please refer to [28].
44
PC- Control and Data processing
Band Pass Focusing Lens
XYZ Scan Control
Stage Objective
Laser
Dichronic
APD
Photon Counting Interface
Mirror Pin Hole Focusing Lens APD
Figure 3.6 Confocal optics setup where the APD is the avalanche photodiode.
3.4
Conclusion
This chapter introduced the microfluidics and the system of characterization that is usedin
this study. Through Sandia Funding and in collaboration with Scientists at SNL microfluidic devices using SNL fabrication techniques were produced. Sandia National Laboratories has two such systems at their disposal, SUMMiT V TM and SwIFTTM that allows designers to create multilevel MEMS devices. The SwIFT TM process introduces the capability to create transparent silicon nitride channels, an important feature for bio-detection procedures that require quantification via spectroscopy or fluorescence. The creation of micron-sized channels creates a new area of study, namely microfluidics. Another way to produce microchannels is by utilizing a polymer called PDMS. These channels are usually larger than their silicon based counterparts,
45
but the uses of PDMS channels cover a wide variety of applications. In many instances a continuum assumption may be used, but it is necessary to determine if the assumption will hold for each particular flow regime. If the continuum assumption holds then the use of macro flow equations is permitted strain rate reaches an upper limit. If the strain rate is extremely high the fluid will begin to behave in a non-Newtonian manner. Ways to characterize microchannels are quite similar to those for in the study of macrochannels. The first step is to determine the pressure/flow characteristics of a novel channel so a determination of the max allowable pressure could be made since microchannels are on the whole fragile especially MEMS based channels. Then applications to determine flow characteristics can be made using micro-PIV or FCS.
46
CHAPTER 4 Flow Characteristics in Micro Channels
This chapter discusses the experiments conducted to characterize the flow in the microchannels for the purpose of developing a microanalysis system. A macro scale process that utilizes magnetic particles to capture cardiac proteins from heterogeneous samples was discussed in previous chapters. The goal here is to explore the possibility of utilizing the process developed for macro scale devices in micro scale systems. This study investigates the application of nanomagnetic particles in micro channels. The chapter starts with a summary of the preparation protocols for the component needed to detect for AMI using a micro device. The production of magnetic particles with a suitable label to capture the cardiac markers are first discussed then the characterization of pressure driven flow in micro channels is also discussed.
4.1
Preparation of Magnetic Particles
The study of using magnetic particles in macro scale system was first conducted at the
BMEL by Manuel [10, 16]. In his protocol the size of the magnetic particle was not an issue. While for this study the size of the particles must be less than the size of micro channels. Additionally, the effect of surface charges on the particles and device surface was not explored in Manuel's work. A particle charge study was conducted and concluded that the HSA microspheres had a zeta potential of -24.05 mV. Prior to running the test the HSA super paramagnetic microspheres must be produced, the surface must be modified then conjugated with the proper antibody. Then another proper antibody must be labeled with alkaline phosphatase. The working solutions of phosphate buffered saline and 1M NaOH must be prepared as well. Now that the components are 47
assembles how with they individually behave within the confines of a MEMS based microchannel. A look must be taken at how PBS flows within the channel as well as the HSA super paramagnetic microspheres react and flow within the channel. Finally how will the 1M NaOH solution flow within the channel, a schematic of the procedure was shown as figure 2.6 on page 25. All of these behaviors must be witnesses since one can not take their behavior at the microscale for granted. A review of the procedures required for running the AMI detection protocol provided a clear path for what needed to be investigated.
4.2
SwIFT TM Microchannel Characterization
MEMS based SwIFT TM microchannels were obtained from Sandia National Laboratories
(SNL), in Albuquerque, NM. Through funding from Sanida, I was able to spend the summer at the Sandia National Lab. The first work done for characterization of microchannels was done at SNL. Due to the limited amount of time allocated to work on this project a channel could not be designed and fabricated. The usual fabrication time period for the SwIFT TM process is approximately six months, then there are some post fabrications modifications made in order to provide and entrance and exit for the fluid. Due to these factors a preexisting design had to be used for the characterization, the channel was designed by Dr. Paul Galambos of the Intelligent Micromachine Group at SNL (Supervisor while at Sanida). A major issue facing engineers is the ability to mix fluids at such small dimensions. The small dimensions bring about low volumetric flow rates which in turn bring about low Reynolds numbers. The mixing issue has been addressed by many researchers [32-36] and this section investigates the flow phenomena occurring in these channels for a better overall understanding of the mixing possibilities of these particular channels.
4.2.1 Pressure Microflow
The time spent at SNL was of great value and experience for completing and correctly performing the flow experiments at such a small scale. It is at SNL where the setup procedure for using the MEMS based microchannels was first learned and procedures for mounting and handling. An actual representation of the channel used in this work is found as figure 4.1 and the AutoCAD, SwIFT TM computer drawing of the same channel found as figure 4.2. The overall
48
length of the channel is 500m, the straight portion is equal to 200m, the width is 20m and the height is approximately 6.8m, there were two different types of channels investigated; a hydrophilic and a hydrophobic channel Before the module can be handled and mounted each chip must first be inspected through an optical microscope to insure that the etching was done correctly. Since the top layer of the channels is transparent left over oxide is readily seen through a microscope. The tell tale sign of left over oxide is a contrast difference fo und within the channel, as seen in figure 4.3 common failure in MEMS based SwIFT TM microchannels is an etch that goes through the top nitride layer. This renders the channel useless because the fluid will not be bound to the channel, as shown in figure 4.4 Once it has been determined that the channel is clear then the device can proceed to the mounting stage.
500m
200m
20m
Figure 4.1 Typical MEMS Microchannel used in this research.
Figure 4.2 AutoCAD SwIFT TM based drawing the colors signify different layers.
49
Oxide
Nitride Hole
Figure 4.3 Microchannel with silicon oxide blocking flow.
Figure 4.4 Microchannel with silicon nitride etched through.
As mentioned previously in chapter 3 the dimensions of the MEMS device is approximately 2.82mm by 6.34mm. This in itself is relatively small and very cumbersome to work with. Therefore the first step was to find a way to make dealing with the module easier. The method used at SNL was to create a cantilever style setup of the module. To accomplish the cantilever the module was rigidly attached to a basic single glass microscope with the use of a quick dry glue. The modules come in a "Gel Pak" that protects them from incidental bumps and jars and removing the modules from the "Gel Pak" is a bit tricky. If grabbed with tweezers and not removed properly damage could render it useless. Due to their small mass MEMS devices can experience high accelerations without failing, F = m a, but once an object comes into contact with the surface, damage is highly likely. Now that the module is mounted there needs to be a way to introduce fluid into the microchannels. The initial way to set the modules up for fluid introduction was the way SNL originally used. The original set up called for setting up the mounted module on a couple of slides in the up side down position, the Bosch etch holes were visible. While keeping in mind which channels were useful, straight cut, 30 gauge, hypodermic needles (capillary tube) with an
50
outside diameter of 304m were inserted into the proper etch hole. The insertion was accomplished through the use of a stereo dissecting microscope with adjustable magnification. The capillary tube was grabbed with the use of forceps, while looking through the oculars of the microscope the capillary tube was guided into position, free hand, and placed in the proper Bosch etch hole. This is a very delicate process and much care needed to be taken to not damage the device. While the hypodermic needles were in place a quick drying glue was used to rigidly attach the capillary tubes to the module. A few minutes was allocated for the glue to completely dry. Once the glue had dried a silicone sealant was placed over the bottom of the modules to insure that a good seal was present. Once the silicon was dry the module was ready for fluid introduction and flow experiments. This procedure took a couple hours at best and was very fragile. Any bump or sang of the capillary tubes would cause them to dislodge from the module rendering the device useless. Once the capillary tubes were rigidly in place and sealed the module complex was carefully brought to a probe station. Once at the probe station the module complex was placed on the edge of the stage to allow the capillary tubes to protrude from underneath the module. The complex was taped to the stage in such a manor as to not allow any movement. Once in place an attachment was then placed directly under the module to help in isolating the capillary tubes from any type of motion. The setup was now ready for connection to macro tubing. The macro tubing is the conduit for which the liquid to flow into the channel. The tubing was inserted through the isolation device and attached to the entrance and exit of the microchannel. An important thing to note is the fact that surface tension plays an important role in the behavior of microfluidic devices. It is due to this fact that care was taken to make sure that no air bubbles were trapped in the fluid line. If an air bubble were to enter the channel the inlet pressure would have to be increased in order to force the gas through the channel. This was achieved by allowing the fluid, the fluid introduction technique will be discussed in the next paragraph, to flow to the end of the macro tube then it was allowed to retreat into the tube. This would eliminate any air bubbles and allow for attachment to the capillary tubes. The macro tubing was slipped over the capillary tubes with extreme care. Any sudden movements made while slipping the macro tubing over the capillary tubes would cause the capillaries to dislodge and render the device useless. To insure that capillary action would not
51
pull the glue into the capillary and block the channel, the macro tubing was brought to the closest proximity to the bottom of the module. Once the macro tubing was slipped over the capillary tubes they were held in place by putting a small amount of quick dry glue. This provided the initial seal and rigidly attached the macro tubing to the capillary. Once the quick dry glue had dried the silicon sealant was placed over the quick dry glue bond to insure that the set up would not leak. Once the silicon sealant had dried the device was ready for the fluid flow experiment. The approximate length of time that it took to set up the macro tubing connection to the module complex was two hours, quite a long period of time and when added to the set up time the entire length of time that elapsed for a flow experiment was approximately four hours. Now to start the flow experiment, head pressure versus flow rate, a mechanism had to be devised so that a variable pressure flow could be provided at the inlet to the channel. There were two options available, and one that was not. The first was to use a syringe pump and pressure transducers, the second was to use a liquid reservoir whose height could be adjusted in reference to the microchannel. On considering the use of a syringe pump pressure transducer system there were some concerns that the fluid flow would be of a pulsate nature, raising the concern if that effect would influence the flow so that option was disregarded. Using a reservoir in which an external pressure could be applied via a nitrogen cylinder. This external pressure would exert a pressure on a sealed reservoir into which the inlet tube to the channel could be introduced to causing flow out of the reservoir and into the microchannel. This option proved to be too complex for the capabilities of the particular lab where the experiment was being conducted. This left the final option, the creation of a variable height liquid reservoir. By varying the height of the liquid reservoir in reference to the microchannel a variable inlet pressure could be achieved. This entire process was done while the microchannel was being observed from a probe station monitor. The channel would be monitored to insure that the channel would not break in the case of high pressure. The fluid reservoir used in the experiment was a 60ml syringe with the plunger removed. The macro tubing was interfaced with the fluid reservoir via Luer Loks, providing a liquid tight seal. A pulley system was devised to adjust the height of the reservoir and consisted of two Cclamps one of which was connected to a ceiling rafter and the other was connected to the table onto which the probe station was located. The clamp on the table provided a tie down and the ceiling clamp was used as a pulley to raise and lower the fluid reservoir via a wire. A wire was
52
used to minimize the amount of dust created, which would have been the case if a fiber rope was used. The inlet pressure could be determined by using the pressure equation described in chapter 3 equation 3.13.
P gh P
Equation 3.13
A measuring tape was attached to the C-clamp that was fixed to the ceiling rafter and ran the entire length to the ground. Measurements were taken to determine the relationship of the microchannel to the rafter so that an accurate value could be determined. To introduce the fluid into the microchannel the liquid reservoir would be raised slowly from a position that resulted in a height of zero from the micro channel. This would slowly increase the pressure head of the fluid. The macro tubes used in the experiment were transparent, allowing the meniscus, the liquid gas interface, to be monitored as it moved through the tube. It is this property of the macro tubing that allowed the flow rate to be monitored. Once the meniscus progressed to the capillary tubing the height of the liquid reservoir with respect to the microchannel was decreased to reduce the inlet pressure, this insured that the top nitride layer would no break with the sudden pressure change. Once the liquid was seen entering the microchannel the pressure was increased and the set up was slowly brought to max pressure, which was governed by the height of the ceiling rafter, and allowed to flow until the meniscus was approximately 6inches from the outlet capillary. Once it was determined that there were no trapped air bubbles in the flow the experiment progressed. On occasion an air bubble would be trapped in the fluid line and enter the channel, the air bubble increases the surface tension and causes the flow to cease. When this situation would arise a male Luer Lok would be fixed to the end of the macro tubing that was fixed to the outlet capillary. To overcome the surface tension negative pressure would be applied to the outlet macro tubing. The negative pressure would be applied through the use of a 5ml syringe. Applying a 1ml equivalent of negative pressure would allow the air bubble to pass through the microchannel. Once the flow had been established it was possible to proceed with measuring the flow rate with respect to the pressure head. This was accomplished by finding the meniscus and determining how far it has traveled within a certain length of time. To achieve this a piece of tape was used to mark the location of the meniscus at time 0 at the maximum achievable pressure 53
head. A stop watch was used to determine the 5 minute interval that was used for each reading. After the 5 minutes had expired a caliper was used to determine the distance traveled by the meniscus. This process would be repeated for a minimum of 4 readings, spanning 20 minutes, at the same pressure. Once the set of readings was completed, for a particular pressure, the pressure would decrease and subsequent readings were taken. The entire experiment would span a period of hours and in this span it appeared that the channels would degrade. There was evidence of residue entering the channel and inhibiting the flow, as shown in figure 4.6. It was thought that the residue was due to the method used in attaching the capillary tubes to the module. To see if this method was the culprit another method for attachment and sealing of the capillary tubes was sought out.
Figure 4.6 Photograph showing foreign obstructive material thought to be quick dry glue.
Knowing that the original process took hours to set up and once set up the protruding capillaries were held in place by a weak bond that appeared to introduce foreign elements into the microchannel an alternative to this process was sought out. Ideally this alternative would have to provide rigidity to the capillary tubes as well as seal them in one step. By finding a bonding and sealing mechanism in one would decrease to time spent on setting up microchannels
54
for fluid flows thus allowing for more experimentation in the same amount of time. A possible solution that had the capability of sealing and rigidly bonding in a short amount of time was an epoxy. A 5-minute acrylic epoxy was chosen as a test. The capillaries were inserted in the same fashion as previously discussed. Instead of using a quick dry glue to rigidly attach the capillary tubes a small batch of resin and hardener was mixed up and used. When applying the epoxy care must be taken to not dislodge the capillaries from the Bosch etch holes. After a brief polymerization period the module was ready for fluid testing. The module complex was taken to a probe station and macro tubing was applied to the capillaries. Instead of using the quick dry glue another small batch of epoxy was used to adhere the macro tubing to the capillary tubes. The epoxy provided a rigid bond as well as a good seal. Using this method was quicker, approximately half an hour for the entire process as opposed to a couple hours, and provided consistent results with no introduction of foreign material into the channel. The overall process for the pressure-flow experiment remained the same with the exception of the use of an epoxy instead of a quick dry glue coupled with a silicon sealant. Overview of Micro Fluidic Setup: Part A: Module complex 1. 2. Inspect the modules for signs of trapped oxide and breach of top nitride layer. Disregard any channels that appear to have these qualities. Remove the module from the Gel-Pak with tweezers and mount the module onto a glass slide with the use of any instant glue, "Elmer's Instant Krazy Glue Advanced Formula" was used in these experiments, in a cantilever style. 3. Place capillary tubes, Microgroup 30 gauge by 1 inch stainless steel hyperdermic needle straight cut, into the appropriate Bosch etch holes for the particular channel under investigation. 4. With the 5-minute acrylic epoxy, the epoxy used in these experiments was Devcon 5 Minute Epoxy In DevTubeTM, number 14250, cover the backside of the module to bond and seal the capillary tubes, wait until the epoxy has completely polymerized before attempting to relocate. Part B: Pressure Head 1. Attach a C-clamp to a ceiling rafter and another C-clamp, for anchoring purposes, near by so that a guide wire could be attached to it.
55
2.
A wire was run through the C-clamp located at the ceiling rafter. One end was attached to the C-clamp anchor and the other was attached to the fluid reservoir, in this case a Becton Dickinson 60cc Luer Lok syringe with the plunger removed.
3.
At one end of the macro tubing, Upchurch Scientific PFA Tubing I.D. 0.020 inches O.D. 1/16 inch, a male Luer Lok, Upchurch Scientific 10-32 Peek Luer Lok with Finger Tight I Peek, was attached so that it could couple with the 60cc syringe. The opposite end was kept free so that it was able to be bonded to the inlet capillary tube of the microchannel. Another length of tubing was set aside to connect to the outlet capillary tube.
4. 5.
The macro tubing was then slid over the capillary tubing and epoxy was used to bond and seal them into place. The height of the liquid reservoir was raised slowly to a given height. The flow in the microchannel was observed to make sure that there were no air bubbles within it. If air bubbles were present a 5ml syringe was attached to the outlet macro tubing with a Luer Lok and a vacuum was applied.
6.
Once the flow was stable and no air bubbles were present within the flow readings were taken every five minutes for 20 minutes at every predetermined pressure.
A picture of the set up can be seen as figure 4.7 and a close up as figure 4.8.
4.2.1.1 Experimental Results Pressure Flow
The experimental portion of this work spanned over a period of eight months in two different geographical locations in three different laboratories. The initial work took place at Sandia National Laboratories (SNL) in Albuquerque, NM and the later portion of the work took place in Tallahassee, FL at the College of Engineering (COE) and the Department of Chemistry at the Florida State University. This section will focus on the work done at both SNL and the COE with a brief discussion of the work done at the Department of Chemistry.
56
Reservoir Measuring Tape
Figure 4.7 Pressure head set up on probe station.
Microchannel
Macro Tubes
Figure 4.8 A close up of the set up.
57
4.2.1.2 Study 1: Sandia National Laboratory Experiments
The first graph of results is data collected at SNL. The graph shows the channels relationship between inlet pressure and the mass flow rate of the fluid, this graph is shown as figure 4.9. The fluid used in these experiments was DI water.
Pressure vs Mass Flow Rate
20000.00
18000.00
1 4 2 3
SNL 1 (Phobic) SNL 2 (phobic) SNL 3(phylic) SNL 3(A) (phylic)
16000.00
Pressure (N/m )
14000.00
12000.00
10000.00
8000.00 0.00E+0 2.00E-09 4.00E-09 6.00E-09 8.00E-09 1.00E-08 1.20E-08 1.40E-08 1.60E-08 1.80E-08 2.00E-08 0 Mass Flow Rate (kg/s)
Figure 4.9 The pressure versus Mass Flow Rate relationship found at SNL, the circled points were redone values to check for consistency.
The mass flow rate was determined by knowing the inner diameter of the tube used, marking the starting point and ending point of the flow over a certain amount of time, therefore the volume over a period of time was known and with a little more math the mass flow rate could
58
be determined, each point represents the average of a minimum of four values taken at 5 minute intervals. The relationship does not appear to be linear, though when looked at a little closer the lower pressures appear to be linear and when the relatively higher pressures are reached the behavior of the fluid becomes non-linear. The order in which the data was taken is numerically resembled in the titles of the data. After SNL 1 and SNL 2 were done and the graphs viewed it was determined that a hydrophilic channel should be investigated to see if the same behavior could be found. While the experiment was in progress the data points collected were converted and plotted out so that a real time plot could be seen. There was an anomaly present at the higher pressure to see if this was an experimental error or if the results were repeatable after the lowest pressure was reached data was collected at the same high pressure that the experiment started with. The circled points in the graph signify points that were revisited and they are in numerical order, at each revisited point a total of three readings were taken over a period of 15 minutes.. Point 1 shows great deviation from its corresponding point at the same pressure. Points 2 and 3 were found at a corresponding pressure that is higher, point 2 (17kPa), and lower, point 3 (16.5kPa), then point 4 (16.75kPa), . Point 4 was the original point from the first run of the experiment, the initial run that went from a pressure of 19.6kPa to 10.3kPa. Points 2 and 3 are relatively consistent with point 4 when compared with the difference found between point 1 and its corresponding original data point, the relationship appears to be more linear. These readings are consistent with unpublished results reached by another reasearcher at SNL. There was no appreciable difference between the hydrophobic and hydrophilic channels. There also is a lack of consistency between the same type of channel as shown between SNL1 and SNL2. These differences will be discussed once again in the light of the experiments done at FSU an it will be at this point that a hypothesis will be put for to explain this phenomenon. The next graph, figure 4.10, will show the relationship between the pressure and Reynolds number for the channels investigated at SNL. The velocity of the flow within the channel was determined by taking the volumetric flow rate (m3/s) divided by the cross-sectional area of the channel. Then with the use of equation 3.10 the Reynolds number can be determined. The main feature to pay attention to in the figure labeled 4.10 is the fact that the Reynolds numbers are low ranging from 0.2 to 1.4. In most situations turbulent flow begins at a value around 2000; therefore the flow in these channels is laminar. Mixing problems exist when using
59
channels of these dimensions and ways of dealing with this issue have been looked at, though at a somewhat larger scale then the channels investigated in this work [25-29].
Pressure vs Rynolds Number
20000.00
18000.00
16000.00
Pressure (Pa)
SNL 1 (phobic) 14000.00 SNL 2 (phobic) SNL 3 (phylic)
12000.00
10000.00
8000.00 0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00 1.40E+00 1.60E+00 Reynolds Number
Figure 4.10 This graph shows the relationship between pressure and Reynolds number, notice the laminar flow shown by the low Reynolds number.
Figure 4.11 shows the relationship between the mass flow rate and the resistance. Resistance was calculated using equation 3.8 which states that resistance is equal to the difference in pressure over the volumetric flow rate.
R
P Q
Equation 3.8
60
Mass Flow Rate vs Resistance
2.00E-08 1.80E-08 1.60E-08 1.40E-08
Mass Flow Rate (Kg/s)
1.20E-08 1.00E-08 8.00E-09 6.00E-09 4.00E-09 2.00E-09 0.00E+00 7.00E+11 SNL 2 (phobic) SNL 3 (phylic)
1.20E+12
1.70E+12
2.20E+12
2.70E+12
3.20E+12
3.70E+12
Resistance (Pa / [m3/s])
Figure 4.11 Relationship between mass flow rate and resistance. The resistance for the data points corresponding to SNL 1 (phobic) were out of scale and shown as figure 4.12.
Mass Flow Rate vs Resistance
1.40E-08
1.30E-08
1.20E-08
Mass Flow Rate (kg/s)
1.10E-08
1.00E-08
SNL 1 (phobic)
9.00E-09
8.00E-09
7.00E-09
6.00E-09 1.00E+15
1.10E+15
1.20E+15
1.30E+15
1.40E+15
1.50E+15
1.60E+15
1.70E+15
1.80E+15
Resistance (Pa/[m3/s])
Figure 4.12 Relationship between mass flow rate and resistance for the data set corresponding to SNL 1 (phobic).
61
There is a general trend that shows as pressure increases the resistance decreases, with some anomalies. The first experiment SNL 1 (phobic) is a bit of an anomaly as well for this set of data, figure 4.12 shows this data. The resistance of this channel is much higher then the two following experiments. A reason for this phenomenon is not understood.
4.2.1.3 Study 2: Florida State University Experiments
In order to see the broader picture it was necessary to gather more data on pressure flow relationships on yet more channels. Since there was no clear difference between the hydrophobic and hydrophilic channels this was investigated further. The rest of the experiments were carried out at the COE in Tallahassee, FL. This would compare if the data would be the same between to different geographical and therefore climatologically different areas. The experiments were carried out in the same fashion as those at SNL with the difference being that a greater head pressure was achieved at the COE. It was thought that the greater the pressure the more inconsistent the flow measurements would be, as was seen in the data recorded at SNL. Figure 4.13 shows the data collected at FSU in regards to pressure versus mass flow rate. The comparison between the pressure and the Reynolds number for the data taken at FSU can be seen in figure 4.14.
4.2.1.4 Comparison between SNL and FSU
By studying the FSU data it clearly does not display the same phenomenon as seen in the SNL data. A comparison between the SNL and FSU data can be seen in figure 4.15. The data shows no correlation between SNL and FSU. One difference does become apparent though. This is the difference at the relatively higher pressures, at higher pressures greater flow rates are achieved and therefore higher Reynolds numbers, even though they still may be laminar. In terms of mixing issues this phenomenon is important. . The random fluctuations are not seen in the FSU data when compared to the SNL data. These random fluctuations may add to some unknown portion to mixing at these small scales, but could not be reproduced at FSU. It is also thought that the flow in these given set of circumstances is showing a new previously unwitnessed behavior. Unpublished accounts of this behavior have been witnessed at SNL as well. There is also no solid relationship between the same type of channel, the hydrophobic
62
Pressure vs Mass Flow Rate
25000.00
23000.00
21000.00
19000.00
Pressure (Pa)
17000.00
FSU 1 (phylic) FSU 3 (phylic) FSU 4 (phobic)
15000.00
13000.00
11000.00
9000.00 5.00E-09
1.00E-08
1.50E-08
2.00E-08
2.50E-08
3.00E-08
3.50E-08
4.00E-08
4.50E-08
Mass Flow Rate (kg/s)
Figure 4.13 Pressure versus flow rate for COE Tallahassee, FL. The behavior is relatively linear even when at a higher pressure then those achieved at SNL in Albuquerque, NM.
Pressure vs Reynolds Number
26500.00
24500.00
22500.00
20500.00
Pressure (Pa)
18500.00
16500.00
FSU1 (phylic) FSU 3 (phylic) FSU 4 (phobic)
14500.00
12500.00
10500.00
8500.00 5.00E-01
1.00E+00
1.50E+00
2.00E+00 Reynolds Number
2.50E+00
3.00E+00
3.50E+00
Figure 4.14 Pressure versus Reynolds number for COE Tallahassee, FL, with numbers ranging from 0.6 to 2.
63
Head Pressure Vs Flow Rate, SNL and FSU
25500.00
23500.00
21500.00
19500.00 SNL 1 (phobic) SNL 2 (phobic) SNL 3 (philic) FSU 1 (philic) FSU 3 (philic) FSU 4 (phobic)
Pressure (N / m2)
17500.00
15500.00
13500.00
11500.00
9500.00
7500.00 2.60E-10
5.26E-09
1.03E-08
1.53E-08
2.03E-08
2.53E-08
Mass Flow Rate (kg/s)
Figure 4.15 Comparison between the SNL and FSU pressure versus flow rate.
channels do not flow consistently at a higher or lower rate the same can be said for the hydrophilic channels. This topic is further discussed in the conclusion of this section.
4.2.1.5 Study 3: Repeatability (A)
The next set of experiments will try to answer the question of repeatability within the same channel. The experiments here should not be directly compared with the previous experiments because of the fact that Luer Loks were used as quick connects to save time setting up the experiments. Therefore there will be loses associated with the couplings that are not taken into account with the previous experiments since they were not used. In the repeatability experiment an initial reading was taken using distilled water. The initial reading was only taken at 4 different pressures to get an idea of what the curve would look like; the data is shown in figure 4.16. The data from this initial experiment is shown as FSU 6(A) (phobic). The next day readings were taken starting at the same pressure and down to a low 64
pressure using DI water in place of HPLC water. The data from this experiment is shown as the line labeled FSU 6(B) (phobic).
Head Pressure Vs Flow Rate, DI
25500.00
23500.00
21500.00
Pressure (N / m2)
19500.00
FSU 6(A) (phobic) FSU 6(B) (phobic)
17500.00
15500.00
13500.00 9.00E-09
1.10E-08
1.30E-08
1.50E-08
1.70E-08
1.90E-08
Mass Flow Rate (kg/s)
Figure 4.16 Pressure versus mass flow rate same channel using DI water, two separate experiments to determine if results are reproducible.
The data points in the figure above are for the same channel used twice, once for FSU6(A) (phobic) and the next day for FSU6(B) (phobic). The goal of this study was to see if there would be reproducibility of results when the same exact channel would be used twice. The reasoning behind this was because the data shown in the previous experiments did not show consistency within the hydrophobic or hydrophilic subsets. The consistency is also missing from repeating the exact same experiment on the same channel showing that variability is significant
65
in these experiments. The exact reason for this behavior is unknown but it is a trend seen through the data. When a liquid flow is introduced to the channel its initial flow rate is lower then the flow rate of a following liquid. A reason for this occurrence may be seen as a capacitive in nature. Due to the ability of the nitride layer to flex in the presence of a pressure, the deformation may in some way be permanently imparted on the microchannel and affecting its ability to hold its previous non introduced flow dimensions. So when the liquid flow is introduced again the layers flex further allowing a for a greater volume that in turn makes for a higher flow rate.
4.2.1.6 Study 4: Repeatability (B)
Since the results were not consistent when repeating the study on the same channel it was thought that the flow characteristics may be fluid dependant. In the next set of experiments repeatability was investigated with fluids other than DI water. The first study done with a fluid other than DI water was an experiment that used a phosphate buffered saline solution (PBS), figure 4.17. This experiment proved to be interesting because of the anomaly found at the relatively higher pressure. The first run of the experiment FSU 7(A) (phobic) was run with three different pressures on one day. The next day, FSU 7(B) (phobic), the experiment was repeated by first revisiting the previous pressures, to check for repeatability, and then proceeding to lower pressures. This data is interesting because of the fact that both sets of data follow the same overall trend but with an offset. As previously witnessed the flow rate of the second run has increased when compared to the flow rate of the first run. The same explanation could hold true as well. As far as the anomaly experienced at 23kPa the reason is not understood, but was repeated.
66
Head Pressure Vs Flow Rate
25500.00
23500.00
21500.00
Pressure (N / m2)
19500.00
FSU 7(A) (phobic) FSU 7(B) (phobic)
17500.00
15500.00
13500.00 6.00E-09 7.00E-09 8.00E-09 9.00E-09 1.00E-08 1.10E-08 1.20E-08 1.30E-08 1.40E-08 1.50E-08
Mass Flow Rate (kg/s)
Figure 4.17 Pressure versus mass flow rate same channel using PBS water, two separate experiments to determine if results are reproducible.
4.2.1.7 Study 5: Fluid Dependency
The next experiment used two different types of fluids that were run in the same channel. It appeared that consistency was a hit or miss and by varying the fluid within the same channel may either exaggerate this difference or prove that the fluid plays no role in the flow behavior. In this experiment an acidic fluid was chosen because as the number of free hydrogen ions increase in concentration in a fluid within a silicon nitride channel the zeta potential of the layer changes [37]. This paper states that as the pH of the fluid interacting with silicon nitride decreases the zeta potential of the silicon nitride increases. At a pH of roughly 2.5 the value is approximately +22.5 zeta potential/mV and at a pH of 6 the value is approximately -27.5 zeta
67
potential/mV. The two fluids used in this experiment was HPLC water mixed with HCl to a pH of roughly 2 and DI water. The results are shown as figure 4.18.
Pressure vs Mass Flow Rate
26000.00
24000.00
22000.00
Pressure (Pa)
20000.00 FSU 8(A) (phobic) FSU 8(B) (phobic) 18000.00
16000.00
14000.00
12000.00 6.00E-09
1.10E-08
1.60E-08
2.10E-08
2.60E-08
3.10E-08
3.60E-08
4.10E-08
Mass Flow Rate (kg/s)
Figure 4.18 The figure above shows pressure versus mass flow rate relationship between two different liquids within the same channel. FSU 8(A) has a pH of 2 and FSU8(B) is DI water.
The results from this experiment do not show an appreciable difference between the results of the other repeatability studies and basically conclude that the phenomenon experienced is not fluid dependent. Once again a difference with regards to flow rate between the first run and second run is witnessed. The flow rate of the second run is higher than the flow rate for the first run.
68
4.2.1.8 Conclusion: Pressure Microflow
When reviewing all the data collected in the experiments performed both at SNL and FSU there is big difference and a similarity. The similarity is that the flow experiments appear to have not too much consistency between them this includes both the hydrophobic and hydrophilic channels. Experiments have shown that in hydrophobic microchannels have a slip velocity at the wall that is approximately equal to 10% of the free stream velocity due to the presence of nanoair bubbles present on the hydrophobic surface[29]. This in turn equates to a greater flow rate when compared to hydrophilic microchannels, which was not necessarily the case in the work accomplished here. A big difference between the data collected at SNL and FSU is the difference in the flow at the relatively high pressures. This maybe explained in a couple ways. The first is the difference in humidity between Tallahassee and Albuquerque. In a review the author states that due to the surface roughness of silicon based channels that adsorption of water onto the surface is a common phenomenon. In macro studies this phenomenon does not play a big factor but at the micro scale the effects could be more pronounced. The adsorption of water molecules onto a micro surface would affect an internal gas flow but its ability to effect an internal liquid flow by means of an external nitride layer have not been determined [29]. This factor may play some unknown role in the physical properties of the top nitride layer. A possibility is that the adsorbed water molecules my change the physical characteristics of the layer and make it more rigid, thus making it resist flexure and random fluctuations. The second factor that could influence the behavior of fluids in the microchannels is the physical properties of the channel itself. It has been observed in these experiments that the nitride layers flex and deflect. A difference between a pressurized and non-pressurized channel can be seen in figure 4.15. Since layers of the channels can flex and not break the possibility of a change in external pressure upon the channel may affect the underlying liquid flow. Let us say that Albuquerque is approximately at 6000ft above sea level and that Tallahassee is roughly near sea level. The pressure difference between Tallahassee and Albuquerque would approximately be equal to 2.92psi. This difference in pressure would not affect the flow per say because it is assumed to be gauge pressure. But this would affect the amount of pressure on the upper nitride layer of the channel. This would possibly allow the top layer greater freedom of movement and
69
fluctuate more at a higher altitude then at a lower altitude, this behavior could possible have an effect on the fluid flowing through the channel. The previous studies on SwIFTTM microchannels show Reynolds numbers ranging from 0.6 to 2. This is most definitely in the realm of laminar flow and ways to introducing mixing need to be addressed one way to induce mixing is to use the microspheres themselves. If an alternating magnetic field could be applied across the channel and the microspheres could be pulled across perpendicular to the flow then that would increase the chances of the antigen to come into contact with the antibodies attached to the microspheres. To investigate the nonlinear flow behavior seen at SNL a bit further at FSU a device would need to be fabricated. This device would need to be able to reduce the external pressure exerted onto the surface of the microchannel. This way variable external pressures could be investigated with their effects on the flow rates of pressure driven liquid flow. This would answer the question of whether or not the flow phenomenon witnessed at SNL is due to a difference in pressure.
4.2.2 Microsphere Compatibility
The other aspect of the research was to determine how the microspheres created at the Center for Nanomagnetics and Biotechnology would interact with the MEMS based SwIFTTM microchannels. The microspheres used in this work were super-paramagnetic HSA magnetic microspheres that were 1m in diameter or smaller. The diameter of the microspheres was determined by a staff scientist at the CNB using SEM micrographs. The goal of this work was to determine if these microspheres were suitable for these channels, to determine how to introduce the microspheres into the channel and if it was possible to magnetically capture a HSA magnetic microsphere. The majority of this work was done at SNL.
4.2.2.1 Set Up for Microsphere Introduction
The overall set up for the module complex is the same protocol that was summarized in section 4.2.1, Overview of Micro Fluidic Setup Part A: Module complex. The part that differs is that the pressure generating mechanism was a syringe pump coupled with a syringe. There was no need to keep track of the outlet flow since this was a qualitative study.
70
Once the module complex was completed and set up on the probe station inlet and outlet macro tubing was bonded to the corresponding capillary tubes. Male Luer Loks were placed on the free end of both the inlet and outlet macro tubing so that they could connect with a syringe. A stock solution of HSA microspheres, approximately 1 m in diameter, was obtained at a concentration of 10mg/ml from the Center for Nanomagnetics and Biotechnology. A working solution was made by dispersing the microspheres into solution and removing an aliquot of 1ml and combining it with 29ml of PBS. The working solution was then sonicated for a period of 5 minutes to make sure that there were no agglomerated particles in solution. The solution was then transferred to a 5ml syringe and the syringe was subsequently placed in a syringe pump. An important characteristic of the microspheres that is somewhat negligible in a macro system is the charge of the avidinated microsphere. Tests were run on a 0.57mg/ml solution of avidinated microspheres in PBS and repeated a total of 3 times with 5 runs within each test, total of 15 data points. These points were averaged with a result of -24.05 mV, zeta potential. Even though these particles were large in reference to the channel dimensions they were readily visible using the probe station and made the qualitative study readily available. The microspheres would impede the flow, but the main purpose of this study was to determine compatibility.
4.2.2.2 Study: Microsphere Introduction
In the first two runs the syringe pump was in a horizontal orientation. The flow rates were noted and were on the order of 0.1l/min to 10l/min at the end of the experiment. These first two experiments yielded very few microspheres entering the channel. In the first run 1 HSA microsphere passed through the channel in 30 minutes. In the second experiment the number was closer to 10 HSA microspheres passing through the channel. Upon inspection of the inlet tube and syringe it appeared that the HSA microspheres were settling out of solution before they could enter into the microchannel. One solution tried for this problem was to introduce a tiny air bubble into the inlet macro tubing with hopes that the surface tension would push the particles into the channel, but there was no difference between the air bubble introduction and the previous experiment. In the next experiment the syringe pump was adjusted from the conventional horizontal position to a vertical position, on the outer portion of the probe station. This allowed the 71
microspheres to settle and was successful in introducing a relatively large number of microspheres into the channel. At first the microspheres flowed freely through the channel, refer to figure 4.19(a), but after a period of time the particles began to agglomerate in the inlet to the channel, refer to figure 4.19(b). It appeared that the individual and agglomerated microspheres where adsorbing to the walls of the channels. The agglomeration was occurring before the microspheres entered the channel and in the channel as well. Reasons for this occurrence were not understood because the particles were sonicated before introduction. Ways that agglomeration could have occurred are; bacterial contamination, and the introduction of microspheres in a relatively high pressure. The microspheres were never exposed to pressure driven flow and this could have forced the microspheres together upon entering the channel. In an attempt to dislodge the particles a vacuum was applied to the outlet of the channel. This did not aid in the dispersal of the particles. A closer look at the system was taken. The overall charge of the HSA microsphere was negative, -24.05 mV zeta potential, and the HSA microspheres were in a solution of PBS the pH was close to 7.4. With the knowledge already known that the walls of silicon nitride will have an overall negative charge above a pH of 4 there was no chance that the attraction would be due to surface charge attraction. In some instances the microspheres appeared to be sheared apart and turned into a type of slurry that moved through the channel close to resembling a lava lamp. The experiment was run 5 more times with varying levels of success. In some instances straight PBS was pumped into the channel to try and dislodge the particles but nothing happened. In the brief times when particles were flowing in the channel an attempt was made to attract the particles with a magnetic tip. The magnetic tip was coupled to a 0.25T NdFeB rare earth magnet which was then mounted to a micropositioner. The magnetic tip was tested for viability by inserting the tip into a solution of HSA magnetic microspheres. This test showed that the tip was capable of attracting the magnetic microspheres. The next series of test done were to see if the tip could be used to attract the HSA magnetic microspheres while they were in the microchannel. Much of this work was done using channels other than the characterized one. This was due to quantity issues; there were not many straight channels available. The results were inconclusive. On a few occasions the tip pierced through the top nitride layer allowing for the fluid to escape from with in the newly created opening. With the channels that stayed in tact
72
the issue of agglomeration was still present so these set of experiment were not successful and a few hours of video documentation were taken.
(a)
(b)
Figure 4.19 (a) A single free flowing HSA microsphere, (b) agglomeration and adsorption of HSA microspheres onto the walls of the channel.
4.2.2.3 Conclusion: Microsphere Compatibility
Using the HSA magnetic microspheres attained from the Center for Nanomagnetics and Biotechnology the results showed that there would be many problems associated with working with these particular microspheres in these channels. Some of the issues brought up by these experiments were; introduction of the microspheres to the channel, agglomeration of microspheres, adsorption of the microspheres to the walls of the silicon nitride channel, and difficulty magnetic manipulating the microspheres within the microchannel. Applying a magnetic immunoassay to these channels would prove to be a very difficult process in which all of the aforementioned obstacles would have to be conquered.
73
4.2.3 Fluorescence Correlation Spectroscopy
Once the pressure flow characteristics of the microchannel were determined as well as the fact that the HSA magnetic microspheres used in the microsphere introduction studies would not work in there current iteration the next step was to characterize the flow within the microchannel. Characterizing the flow and getting a velocity profile of the flow within the channel would create a greater understanding of the flow physics involved. The flow characterization chosen for these sets of experiments was a method named Fluorescence Correlation Spectroscopy (FCS) mentioned earlier in section 3.3.3 of this thesis. The set up used in this experiment is shown in figure 4.20.
Figure 4.20 Above is a picture of the FCS setup. Noticeable in the foreground are the mirrors and to the right the laser. To the far left is the stage setup with the piezoelectric actuation device mounted on top.
4.2.3.1 SwIFTTM Microchannels
Silicon nitride channels have never been investigated using this particular set up. Due to fact that this was a novel approach for fluid characterization there were some issues with the set up. The system was originally designed to be used with transparent PDMS-glass channels. Not 74
only were these channels transparent but they were mush larger making it easy for the laser light passing through the objective to be seen in passing through the material. The optical properties of the SwIFTTM microchannels are such that the top layer of silicon nitride is transparent but the underlying silicon layer did not allow for the laser to pass through since the silicon base is not transparent. The set up has an ocular type device that allows on to see through the objective and it is this feature that allowed for the scanning of the module for the microchannel.
4.2.3.2 FCS SwIFTTM Microchannel Set Up
The initial set up of the SwIFTTM microchannels was identical to the set up found in section 4.2.1 Pressure Microflow Setup, Overview of Micro Fluidic Setup: Part A: Module complex. This initial set up allowed for the capillary tubes to be rigidly connected to the module. Once the module was complete the complex was placed upside down away from the FCS setup. A pressure head set up similar to the previously mentioned set up in section 4.2.1 Pressure Microflow Setup, Overview of Micro Fluidic Setup: Part B: Pressure Head. The working solution was placed in the fluid reservoir and allowed to travel to within 6 inches of the end of the capillary. Then the inlet and outlet macro tubing was slid into place over the capillary tubes. The capillary tubes were then bonded using the same 5-minute epoxy that was previously used. The set up was the placed on the stage in emersion oil and fixed to the piezoelectric actuator via Scotch tape. The solution of FluoroSpheres comes concentrated so a dilution must be made of the solution to get something that is manageable by the system. The stock solution was diluted 50,000x 75,000x depending on study. For details on running the data collection and autocorrelation please refer to [21].
4.2.3.3 Study 1: 200nm FluoroSpheres
After finding the channel, which was no easy feat and took hours, an initial test was preformed with 200nm beads (Molecular probes, FluoroSpheres, orange (540/560)) at a dilution of 50,000x from a 2% solid stock solution then sonicated for 15 minutes, to see how the channel would behave with these spheres. The initial readings showed some promise, but after a period of minutes the flow stopped. Upon further investigation the channel was completely clogged with these nanospheres. The experiment was repeated two more times with the same
75
result, clogged channels. It was concluded that the next step should include reducing the size of the FluoroSpheres from 200nm to 40nm.
4.2.3.4 Study 2: 40nm FluoroSpheres
After knowing that the 200nm beads were clogging the channel 40nm beads were ordered from the same company as the 200nm beads. Since these beads were incredibly small their diffusion rate would be much higher then the 200nm beads. A quick calculation using the Stokes Einstein diffusion equation showed that as long as the flow velocity was above 0.115mm/s diffusion could be considered negligible. Upon having this knowledge a preliminary experiment was conducted to determine if the 40nm beads would clog as the 200nm beads did. The set up was the same as was described in section 4.4.1.1. After a few minutes of steady flow the flow began to subside slowly. Upon inspection of the complex, the 40nm beads fell to the same fate as the 200nm beads. The stock solution came as 5% solid and was diluted to 50,000x for this experiment. The working solution was then sonicated for 15 minutes. They agglomerated and adsorbed to the walls of the channel, deduced from visual inspection and inspection via the FCS setup. Reasons for the agglomeration can range from bacterial contamination of the seeded flow to the protein encapsulation of the fluorescent dye adhering to the walls of the channel. A feature of the FCS set up allows one to make a picture of the scanning area that would be under velocity investigation. Once the flow had stopped this feature was used to take a closer look at what was occurring in these channels. Since the set up utilized a confocal arrangement it is possible to scan the channel at different depths. Since the channel's depth was known, 6.8m, the actuators were set to step in 1m increments so that the top, middle and bottom of the same section could be investigated. The initial scan, figure 4.21 (a) , was done to slightly above the top nitride layer. Though it appeared that some beads adsorbed to the layer the results were not conclusive. Figure 4.21 (b) investigates the inside of the channel and shows that the beads in this area are agglomerated and appear to be stagnant. Another scan was taken at the entrance to the channel to see what had occurred there, figure 4.22(a). A figure of the channel and its dimensions is shown as figure 4.22(b). The neck leading into the channel is completely blocked with beads disrupting the flow to the channel. Another interesting feature is that the walls appeared to fluoresce as well. This
76
(a) Top
(b) Inside
Figure 4.21 (a) This scan was taken at the top nitride layer of microchannel, (b) this scan was taken at a point inside of the microchannel. These figures show particles stuck within the channel.
500m
200m
20m
(a)
(b)
Figure 4.22 (a) T...