Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
A Miniaturized Microfluidic Cytometer Platform for Point-of-Care Blood Testing Applications
by
James Jiahua Dou
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Department of Electrical and Computer Engineering
© Copyright by James Dou 2017
ii
A Miniaturized Microfluidic Cytometer Platform for Point-of-Care
Blood Testing Applications
James Jiahua Dou
Doctor of Philosophy
Department of Electrical and Computer Engineering
University of Toronto
2017
Abstract
In recent years, microfluidic lab-on-a-chip devices have made tremendous progress. Many new
features, new processes and fabrication methodologies have been proposed and realized in the
last two decades. In the field of clinical diagnostics, however, there still lack devices and
products that truly utilize microfluidic lab-on-a-chip technologies. My thesis is focused on
developing a novel blood testing platform that can be commercialized as a point-of-care
diagnostic tool using microfluidics and optical detection technologies. Access to laboratory
quality blood testing has been difficult and sometimes prohibitive for many populations around
the world, which lead to increased mortality, morbidity and healthcare cost. In this work, I
research and developed a particle imaging and detection system using microfluidics to control
sample motion. The system used fluorescence based detection methodology. A novel image
analysis algorithm is proposed and implemented to detect and track particles captured by the
optical detector. My thesis investigates the fluidic design, on-chip fluidic control, optical
detection system and particle/cell counting algorithm. The overall system is designed to be
capable of commercialized as a point-of-care portable blood analysis system deployed in remote
health settings. In addition, CD4 T cell counting, which is the current gold standard in
HIV/AIDS disease monitoring, is used as an example assay developed on the platform. A
iii
microfluidic chip, that can complete the cell counting analysis within 15 minutes directly from a
single finger prick blood sample is designed, fabricated and characterized. This work also studies
the sample preparation steps required for the cell counting assay and the impact of various
system parameters. The system described in this thesis highlights a portable blood testing
platform that can be further developed and commercialized as a clinical tool for cell, protein and
nucleic acid based laboratory tests conducted at point-of-care. The outcome of this work
demonstrates the feasibility and utility of such system or product in medical diagnostics in
remote health settings.
iv
Acknowledgments
I would like to express my gratitude to my supervisor, Prof. Stewart Aitchison, whose expertise,
understanding, and patience, added considerably to my graduate experience. I appreciate his vast
knowledge and skill in many areas (e.g., vision, aging, ethics, interaction with participants), and
his assistance in writing reports (i.e., grant proposals, scholarship applications and this thesis).
His passion, energy, relentless curiosity, and clear vision have defined research to me. This thesis
would not have been possible without the mentorship and tremendous support of Prof. Aitchison.
I would like to thank the other members of my committee, Dr. Axel Gunther, and Dr. Yu Sun for
the assistance they provided at all levels of the research project.
I have been blessed to work with friendly colleagues. Dr. Lu Chen and Rakesh Nayyar were a
pleasure to work with in the lab. I am grateful to Dr. Aju Jugessur for great discussion and lab
work on silicon photonics and nanofabrication. I would also like to thank Dr. Francis Mandy,
without whose motivation and encouragement I would not have considered developing novel
point-of-care technologies for global health. Appreciation also goes out to Jason Grenier, Luis
Fernandez, Stephen Ho, Dr. Jianzhao Li of the Department of Electrical and Computer
Engineering for all the wonderful support and fruitful discussions.
I would also like to thank my family for the support they provided me through my entire life and
in particular, I owe my parents Jinglie and Xinxuan, among a multitude of other things, for
giving me the curiosity to explore the exciting research of combining medicine with technology.
I must acknowledge my wife and best friend, Jenny, without whose love, encouragement and
patience, I would not have finished this thesis.
Finally, I recognize that this research would not have been possible without the financial
assistance of NSERC, the University of Toronto Graduate Studies, the Department of Electrical
and Computer Engineering at the University of Toronto (Teaching Assistantships, Graduate
Research Scholarships) and the Ministry of Research and Innovation, Government of Ontario,
and express my gratitude to those agencies.
v
Table of Contents
Acknowledgments .......................................................................................................................... iv
Table of Contents ............................................................................................................................ v
List of Tables ............................................................................................................................... viii
List of Figures ................................................................................................................................ ix
Chapter 1 Introduction ............................................................................................................ 1
1.1 Introduction ......................................................................................................................... 1
1.2 Flow Cytometry .................................................................................................................. 2
1.2.1 Working Principle ................................................................................................... 3
1.2.2 Data Analysis .......................................................................................................... 7
1.2.3 Labels ...................................................................................................................... 8
1.2.4 State of the Art ........................................................................................................ 9
1.2.5 Modern Cytometry Applications ............................................................................ 9
1.3 Aims of this Work ............................................................................................................. 10
1.4 Organization of Thesis ...................................................................................................... 12
Chapter 2 Background .......................................................................................................... 13
2.1 Microfluidic Lab-on-a-Chip .............................................................................................. 13
2.1.1 History of Microfluidics ....................................................................................... 13
2.1.2 Recent Advances ................................................................................................... 14
2.1.3 Microfabrication Technologies ............................................................................. 16
2.2 Point-of-Care Testing ........................................................................................................ 21
2.2.1 Microfluidics and POCT ....................................................................................... 23
2.2.2 Lab on a chip and Global Health .......................................................................... 25
2.2.3 Motivation of this Thesis ...................................................................................... 29
vi
Chapter 3 Passive Microfluidic Systems .............................................................................. 33
3.1 Introduction ....................................................................................................................... 33
3.2 Flow Control Methodology in Microfabricated Flow Cytometers ................................... 34
3.2.1 Acoustic ................................................................................................................ 34
3.2.2 Electrical ............................................................................................................... 35
3.2.3 Micro Structure ..................................................................................................... 36
3.2.4 Micro Flow Cytometers without Focusing ........................................................... 36
3.3 Microfluidics Design Concept .......................................................................................... 37
3.4 Capillary Microfluidic Systems ........................................................................................ 37
3.4.1 Governing equations of Fluid Mechanics ............................................................. 38
3.4.2 Numerical Modeling ............................................................................................. 40
3.4.3 Fluidic Resistance Calculation .............................................................................. 46
3.4.4 Microchannel Design ............................................................................................ 48
3.4.5 Capillary Microfluidic Device Fabrication ........................................................... 49
3.4.6 Capillary Microfluidic Device Characterization ................................................... 50
3.5 Conclusion ........................................................................................................................ 54
Chapter 4 Active Microfluidic System ................................................................................ 55
4.1 Introduction ....................................................................................................................... 55
4.2 Bellows Actuation System ................................................................................................ 56
4.3 Bellows Transport System Design .................................................................................... 57
4.3.1 Bellows Actuation Volume Calculation ............................................................... 58
4.4 Bellows Actuation System and Microfluidics Design ...................................................... 61
4.4.1 Material Characterization ...................................................................................... 62
4.4.2 Microfluidic Channel Design ................................................................................ 66
4.4.3 Bellows Slide Fabrication ..................................................................................... 66
4.4.4 Characterization .................................................................................................... 68
vii
4.5 On-Chip Sample Preparation ............................................................................................ 78
4.5.1 Reagents ................................................................................................................ 79
4.5.2 Reagent Handling and Incorporation .................................................................... 79
4.5.3 Reagent Drying ..................................................................................................... 81
4.5.4 Re-suspension ....................................................................................................... 82
4.5.5 Microfluidic Mixer ................................................................................................ 83
4.6 Conclusion ........................................................................................................................ 93
Chapter 5 Optics and Detection .......................................................................................... 95
5.1 Introduction ....................................................................................................................... 95
5.2 Particle Detection using Optics ......................................................................................... 95
5.2.1 Optical Detection Methodology Literature Review .............................................. 95
5.2.2 Dynamic Particle Detection and Counting ........................................................... 97
5.2.3 Cell Detection and Enumeration using Capillary Microfluidic Devices ............ 105
5.3 Beadarray Multiplexed Detection ................................................................................... 117
5.3.1 Underlying Strategy ............................................................................................ 118
5.3.2 Proof of Concept Demonstration ........................................................................ 121
5.4 Conclusion ...................................................................................................................... 128
Chapter 6 Conclusion and Future Work ......................................................................... 129
6.1 Thesis Work Summary ................................................................................................... 129
6.2 Impact ............................................................................................................................. 131
6.3 Future Outlook ................................................................................................................ 132
References ................................................................................................................................... 133
viii
List of Tables
Table 1 - Comparison between polymer and glass as substrate of microfluidic systems for bio
and chemical applications. ............................................................................................................ 20
Table 2 - Prevalent diseases in developing countries. DALY is a measure of the disease burden
on healthcare system. Type of test is the conventional testing methodology.102 .......................... 28
Table 3 - Materials used in autofluorescence characterization. .................................................... 65
Table 5 - Performance of passive micromixers in recent development141. ................................... 83
Table 6 - Performance summary of active micromixers published in literature141. ...................... 85
Table 7 - Mixing time characterization result. Relative mean fluorescence intensity was
calculated as the ratio between liquid control mean fluorescence intensity and measured mean
fluorescence intensity for each scenario. This characterization was completed using design 2 of
the resuspension prototype slide shown in Figure 4-21. ............................................................... 92
Table 8 - Mixing time characterization result. This characterization was completed using design
4 of the resuspension prototype slide shown in Figure 4-21. With the optimized reagent chamber
dimension, the mixing time can be further reduced. ..................................................................... 93
Table 9 - Mixing time characterization result. This characterization was completed using design
4 of the resuspension prototype slide shown in Figure 4-21 and surface treatment of the reagent
plug. .............................................................................................................................................. 93
Table 10 - Camera comparison summary. .................................................................................. 100
Table 11 – Comparison of results using wide field dynamic counting system and flow cytometer.
..................................................................................................................................................... 113
Table 12 - Concentration levels of IL-6 cytokine tested as proof of concept demonstration on
prototype. .................................................................................................................................... 125
ix
List of Figures
Figure 1-1 - Schematic of a flow cell in a conventional flow cytometer.1 ..................................... 4
Figure 1-2 - Optical setup in a typical modern flow cytometer.1 .................................................... 6
Figure 1-3 – A sample single parameter histogram of particles labeled with FITC dye. FITC
(fluorescein isothiocyanate) is a synthetic organic compound that is widely used as a fluorescent
dye in many applications. Its absorption peak is at 494 nm and emission maximum is at 512 nm.1
......................................................................................................................................................... 7
Figure 1-4 – Sample of a two-parameter (dual color fluorescence) histogram. Particles in this
examples are labeled with PE and FITC dyes. PE (Phycoerythrin) is another organic
fluorochrome whose absorption peak is at 488 nm and 532 nm, and emission maximum is at 585
nm. 1 ................................................................................................................................................ 8
Figure 3-1 - Schematic of liquid plug in a rectangular microfluidic channel. 117 ......................... 39
Figure 3-2 – Schematic illustration of two reservoirs and a straight microchannel connecting the
two reservoirs. ............................................................................................................................... 40
Figure 3-3 – Schematic of microfluidic setup in the numerical model. ........................................ 42
Figure 3-4 – Snapshots of flow in the microchannel at different times. The contact angle for this
simulation is set at 70 degrees. The length of the channel shown in this figure is 1 mm. ............ 43
Figure 3-5 – Axial velocity of liquid in the microchannel with respect to time. .......................... 44
Figure 3-6 – Axial velocity of liquid in the microchannel with respect to time. The cell size is
1.5µm (10 cells per height of channel). The contact angle is 30 degrees. All other conditions are
the same as those in Figure 3-5. .................................................................................................... 45
Figure 3-7 – An example of a capillary microfluidic device. (a) Top view of a capillary
microfluidic device with a circular sample inlet, serpentine microchannel and a tapering structure
to enhance capillary force. (b) Side view of the rectangular microfluidic channel. ..................... 47
x
Figure 3-8 - Graphical illustration of photolithography process used by Epigem to produce
capillary microfluidic devices. ...................................................................................................... 50
Figure 3-9 – Serpentine microfluidic structure for fluidic flow characterization. The channels
were designed with the same cross-sectional dimensions as the detection channels in the
cell/particle counting device. Channel length of each pass is 20 mm. .......................................... 51
Figure 3-10 – Experimental results on filling characterization of the microfluidic device
corresponding to the serpentine microfluidic structure shown in Fig. 4. (a) Observed filling time
of each linear section of the serpentine microfluidic device. (b) Fluid flow speed and channel
filling distance characterization results based on the filling time and the distance for each pass in
the serpentine microfluidic device. An inversely proportional relationship between flow speed
and the channel length can be seen. .............................................................................................. 52
Figure 3-11 - Capillary microfluidic chip design layout. This design relies on capillary forces to
manipulate sample flow. The device has a volumetric design to allow a sample volume of 2 µL to
be processed. ................................................................................................................................. 53
Figure 4-1 – Bellows transport concept. The soft elastomer is depressed under external force F.
The deflection of the elastomer induces a pressure change inside the chamber and actuates the
fluidic motion inside the microchannel that is connected to the chamber. ................................... 56
Figure 4-2 - Bellows slide concept. The entire device consists of a soft elastomer semi-sphere,
bonded to a plastic fluidic chip. The fluidic channels are connected to the bellows. When bellows
is depressed, the reduction in volume inside the bellows increases the pressure inside the
microchannel which subsequently pushes the liquid sample forward. ......................................... 57
Figure 4-3 – A graphical illustration of the coordinate system and variables used in bellows
volume change calculation. ........................................................................................................... 58
Figure 4-4 – Volume change induced by bellows deflection. Bellows deflection is actuated from
the top of the semi-sphere in this diagram and the amount of deflection is denoted by d. ........... 59
Figure 4-5 - Volume change of the bellows as a function of bellows deflection. ........................ 60
Figure 4-6 - Linear stepper motor from Haydon Kerk. (www.haydonkerkexpress.com) ............ 61
xi
Figure 4-7 - Microfluidic chip layout of the bellows slide. The bellows slide was fabricated using
injection molding and it was used to test the fluidic actuation and on-chip flow control. ........... 62
Figure 4-8 - Autofluoscence levels of different PMMA and COC material under different
excitation and emission optical setup. .......................................................................................... 64
Figure 4-9 - A photograph of the injection molded microfluidic bellows chip using the PMMA
material selected in previous section. The bellows chip is 1 inch wide and 3 inches long. ......... 67
Figure 4-10 - Schematics of the stepper motor setup used to test and characterize on-chip fluidic
actuation ........................................................................................................................................ 69
Figure 4-11 - Fluidic control eletronics setup block diagram. This setup was used to
experimentally characterize and test the fluidic actuation using bellows concept. ...................... 70
Figure 4-12 - LabView program interface. This tool was used to control the motion of the stepper
motor for on-chip fluidic actuation. .............................................................................................. 71
Figure 4-13 - A picture showing the mechanical setup of the bellows actuation. ........................ 73
Figure 4-14 - Schematics of microfluidic connections used in on-chip fluidic actuation testing
and characterization. ..................................................................................................................... 74
Figure 4-15 - Pictures of the bellows actuation experimental setup. (a) Side view of the motion
control board, stepper motor, bellows slide and re-suspension slide. (b) Top view of the
experimental setup. The engineered blood sample was introduced into the re-suspension slide
and actuated back and forth inside the device. ............................................................................. 75
Figure 4-16 - Fluidic linear flow speed measured as a function of time. Graph (a) is obtained
with a bead sample only while (b-d) were obtained with a sample of beads mixed with
Immunotrols. Three different pumping conditions were tested in this experiment: the stepper
motor pumping distance (2 µm, 10 µm, and 30 µm at a speed of 50). .......................................... 76
Figure 4-17 – Fluidic linear speed plotted as a function of time in the detection microchannel
using bellows actuation. The measurement was made on the resuspension slide described earlier
in this work. .................................................................................................................................. 77
xii
Figure 4-19 – 3D drawing of reagent plug used to handle and incorporate reagents into the
microfluidic cartridge. ................................................................................................................... 81
Figure 4-20 - Reagent plug coated with dried fluorescently labelled CD4 antibodies. The pink
color indicates the high concentration of antibodies. .................................................................... 82
Figure 4-21 - Microfluidic mixer design. The “wiggly” channels are mixing structures which
utilizes Dean’s flow to enhance mixing efficiency. ...................................................................... 86
Figure 4-22 - Dean flow. In curved channels, when inertia is important, faster moving fluid near
the channel center tends to continue outward, and to conserve mass, more stagnant fluid near the
walls re-circulates inward. This creates two counter-rotating vortices perpendicular to the
primary flow direction164 .............................................................................................................. 87
Figure 4-23 - Resuspension slide design layout. In this design, four different mixer were
proposed to optimize the mixing and fluidic motion in the microchannel. The circular holes are
reagent plug chambers where plugs are inserted. ......................................................................... 90
Figure 4-24 - Fully assembled microfluidic re-suspension prototype slides with reagent plugs
inserted. The re-suspension prototype slides were used to charaterize on-chip mixing and re-
suspension of dried reagents. ........................................................................................................ 91
Figure 5-1 – Optical imaging system of the cell/particle detection and analysis platform. ........ 103
Figure 5-2 – Illustration of the multiplexed detection system described in this paper. The optical
detector continually takes images as particles/cells move into the detection window.
Particles/cells labelled with different fluorophores can only be detected in the corresponding sub-
regions within the detection window depending on the filter setup. (a) a graphical illustration of
the underline principle of the technique; (b) transmission spectrum of the left sub-region of the
detection window; (c) transmission spectrum of the right sub-region of the detection window. 105
Figure 5-3 – CD3 antibody concentration titration curve for signal to background ratio. ......... 108
Figure 5-4 – CD4 antibody concentration titration curve for signal to background ratio. ......... 108
xiii
Figure 5-5 – Positive events intensity histogram for PE channel at 50ng CD4 antibody per 100µL
blood. .......................................................................................................................................... 109
Figure 5-7 – Linearity test result. Cell concentrations range from 150 to 720 per μL were tested
for the comparison with flow cytometer. (Each data point is an average value for 3 measurements
with standard deviation bar.) ....................................................................................................... 114
Figure 5-8 – Two color fluorescence image captured through a custom setup with two half-moon
shaped optical filters placed together side by side. The CD4 cells labelled with PE
(phycoerythrin) were shown in the left panel while the CD3 cells labelled with PE-Cy5.5 were
shown in the right panel. ............................................................................................................. 116
Figure 5-9 – Combined images is showing four detected MESF beads. .................................... 117
Figure 5-10 – Multiplexed beadarray detection process illustration.192 ..................................... 119
Figure 5-11 – Bead complexes and reagents explanation. .......................................................... 120
Figure 5-12 – Two color multiplexed beadarray detection concept illustration. The optical
imaging area is divided into two sections: left is the identification channel and the right is the
quantification channel. ................................................................................................................ 121
Figure 5-13 - Dilution steps performed to obtain different IL-6 concentration levels to be
measured. .................................................................................................................................... 124
Figure 5-14 – Histogram of multiplexed beads detected by the optical detection system described
and developed in this chapter. The beads fluorescence intensities are evenly distributed on a log
scale. ............................................................................................................................................ 126
Figure 5-15 - Correlation between fluorescence intensity of reporter or detection antibody and
target analyte concentration. In this experiment, bead type number 4 in the BD kit was used. The
target analyte was IL-6 cytokine as described in Section 5.3.2 on Page 79. .............................. 127
1
Chapter 1 Introduction
1.1 Introduction
The enumeration of cells, bacteria and viruses in human body fluids is of primary importance in
medicine and immunology. The human immune system is a network of white blood cells,
proteins, tissues and organs. It is a critical component of human body that prevents infection and
the growth of bacteria, viruses and parasites. It also prevents the unwanted growth of cancerous
cells. When compromised, the immune system does not perform its functions and infections
occur. On the other hand, the immune system is overactive and attacks the normal cells of the
body, leading to autoimmune disorder. Maintaining the proper functions of human immune
system is critical in keeping people healthy and preventing infections.
Examples of the clinical importance of the immune system include: counting of CD4 T-cells in
HIV positive subjects and of granulocytes/platelets in patients receiving chemotherapy1.
Currently, flow cytometry is the tool of choice for rapid blood cell analysis2. In this technique,
cells are suspended in a fluid stream and passed through the detection region where there are
illuminated by a laser. Light scattered in different directions can be used to distinguish
differences in size and internal complexity of the cells, whereas light emitted from fluorescently
labeled antibodies can identify a wide array of cell surface and cytoplasmic antigens. This
technique is widely used in cell counting, sorting, biomarker detection and protein engineering.
Moreover, multi-parametric analysis of the physical and/or chemical characteristics of up to
thousands of particles per second can be completed.
The objective of this work is to research and develop a miniaturized, compact imaging cytometer
system that can bring facility bound diagnostic work to the field. More importantly, it can help
doctors and physicians to develop personalized therapy for patients based on customized
biomarker testing for each individual. The model application is a CD4 cell counting assay for
HIV disease monitoring.
2
Figure 1-1 - Concept drawing of the portable blood testing workflow from a droplet of sample from a finger
prick.
Conceptually, the portable analysis instrument will consist of a disposable cartridge and a
handheld/portable electronic device as illustrated in Figure 1-1. The plastic cartridge contains
slow-dried reagents coated on the surface of a disposable microfluidic device during fabrication
and packaging. Disposability reduces the cost of cleaning and eliminates cross contamination
between tests. This point-of-care analysis system offers rapid, effective, accurate and low-cost
HIV monitoring that is suitable for remote areas and resource poor settings. In addition, the
proposed handheld cytometer will have built-in software and hardware quality control
mechanisms to ensure the proper operation of the system, which is vital for such portable
instruments deployed in the field that lacks well established infrastructure.
This chapter lays the background information and motivation of this thesis. It explains the
working principles of a flow cytometer and overviews the common applications of flow
cytometry in clinical setting. The second half of this chapter reviews current efforts to develop
portable flow cytometry solutions that can be deployed at the point of care.
1.2 Flow Cytometry
Flow cytometry is the method of choice for rapid analysis of multiple characteristics of single
cells. The information obtained is both qualitative and quantitative. Whereas in the past flow
cytometers were found only in larger academic centers, advances in technology now make it
possible for community hospitals to use this methodology. Contemporary flow cytometers are
3
much smaller, less expensive, more user-friendly, and well suited for high-volume operation.
Flow cytometry is used for immunophenotyping of a variety of specimens, including whole
blood, bone marrow, serous cavity fluids, cerebrospinal fluid, urine, and solid tissues.
Characteristics that can be measured include cell size, cytoplasmic complexity, DNA or RNA
content, and a wide range of membrane-bound and intracellular proteins1.
1.2.1 Working Principle
The working principles of a flow cytometer are shown schematically in Figure 1-2. The flow
cytometer measures the optical and fluorescence characteristics of single cells (or any other
particle, including nuclei, microorganisms, chromosome preparations, and latex beads). Physical
properties, such as size (represented by forward angle light scatter) and internal complexity
(represented by right-angle or side scatter) can be used to differentiate certain cell populations. A
beam of light of a single wavelength is directed onto a hydro dynamically focused stream of
liquid. A number of detectors are aimed at the detection point to image the subject. By using the
principles of light scattering, and the emission of light from fluorochrome molecules conjugated
to the subject, it is possible to generate specific multi-parameter data from microscopic particles
in the size range of 0.5 µm to 40 µm in diameter. Hydro-dynamic focusing of particles is
accomplished through the introduction of sheath flow surrounding the sample stream; ensuring
only one particle is interrogated at a time.
4
Figure 1-2 - Schematic of a flow cell in a conventional flow cytometer.1
Figure 1-2 also illustrates the principle of hydrodynamic focusing in a flow cytometer. The
sample is injected into the center of a sheath flow. The combined flow is reduced in diameter,
forcing the cells or particles into the center of the fluid stream. This allows the laser beam to
interrogate cells/particles one at a time. As the sheath fluid moves, it creates a drag force on the
narrowing central chamber, which alters the velocity of the central fluid stream whose flow front
becomes parabolic with greatest velocity at its center while effectively having zero velocity at
the wall. The net effect produces a single file of particles, known as hydrodynamic focusing.
Under optimal conditions, the fluid in the center will not mix with the surrounding sheath fluid.
The flow characteristic of the central column can be modeled using the Reynolds number (Re):
𝑅𝑒 =𝜌𝑉𝐷
𝜇 (1 – 1)
Where D is the tube diameter, v is the mean fluid speed, 𝜌 is the density of the liquid and 𝜇 being
the viscosity of the fluid. Laminar flow usually occurs when either the fluid is moving slowly or
the fluid is viscous, under which conditions Reynolds number (Re) is low. As Reynolds number
increases, the flow will transition from laminar to turbulent flow at a specific range of Reynolds
5
numbers. The laminar-turbulent transition range depends on the disturbance levels in the fluidic
system. Typically, laminar flow occurs when the Reynolds number is below a threshold value of
approximately 2000, while the transition range is usually between 1,800 and 2,100.3
After forming single file, particles or cells, are interrogated by a beam of light. Scattering and
fluorescence emission provides information about particle/cell properties. Light that is scattered
in the forward direction, up to 20◦ offset from the laser beam axis, is known as the forward
scatter channel signal (FSC). The FSC signal depends on the particle/cell size and can be used to
distinguish between debris and living cells. On the other hand, light measured at 90◦ angle to the
excitation is termed side scatter signal (SSC). The SSC collects information about the granular
content within a particle/cell. Combing FSC and SSC, a flow cytometer is able to differentiate
different cell types in a heterogeneous population.
Another dimension to the flow cytometer is the ability of carry out fluorescence measurements,
which can provide quantitative and qualitative data about fluorochrome-labeled cell surface
receptors, or intracellular molecules such as DNA and cytokines. Typically, in flow cytometry
different fluorescence channels are used to detect emission of different wavelength.
As cells or particles pass through the detection volume, fluorochromes are excited to a higher
energy state. When electrons of the fluorochrome relax to the ground state, photons of light are
emitted with specific spectral properties unique to different fluorochromes. One unique aspect of
flow cytometry is that it can measure fluorescence per cell or particle. This contrasts with
spectrophotometry, in which the percent absorption and transmission of specific wavelengths of
light is measured for a bulk sample.
Typically, the scattered optical signal and the emitted fluorescence light from cells and particles
are converted to electrical pulses by optical detectors such as silicon photodiodes or
photomultiplier (PMT) devices. Because of its higher sensitivity and lower noise, PMTs are
widely used in conventional flow cytometry.
Figure 1-3 is a graphical illustration of the optical components in a flow cytometer. Collimated
laser beam is focused at the interrogation region. Emitted light is directed to different optical
detectors with filters and mirrors.
6
Figure 1-3 - Optical setup in a typical modern flow cytometer.1
The electrical signals are then processed by a series of amplifiers, with both linear and log
amplification. Logarithmic amplification is commonly used in fluorescence imaging of
biological samples, since this technique expands the scale of weak signals and compresses the
scale for strong signals, resulting in a distribution that is easy to display on a histogram.
The measurement from each detector shown in Figure 1-3 is referred to as a “parameter”. The
data acquired in each parameter represent the number of occurrence of such parameter, which
usually means the number of particles or cells displaying the physical feature or surface marker
of interest. Finally, the amplified signals are processed using a standard analog-to-digital
converter (ADC) which in turn allows further data processing and analysis, and the generation of
histograms and scatter plots.
7
1.2.2 Data Analysis
An important principle of flow cytometry data analysis is to selectively visualize the
particles/cells of interest while eliminating results from unwanted subjects such as debris and
dead cells. This process is called gating.
Gating can be done based on the physical characteristics of particles/cells, surface markers or
intracellular contents. In addition to FSC /SSC gating, newer strategies utilizes fluorescence
parameters with scatter signals to distinguish monocytes, lymphocytes and granulocytes.1
In addition to gating, single parameter histograms are also often used to display a single
measurement with light intensity on the x-axis and the number of events on the y-axis. The
sample histogram shown in Figure 1-4 is useful for evaluating the total number of particles/cells
in the sample that possess the physical properties selected for or which express the surface
marker of interest.
Figure 1-4 – A sample single parameter histogram of particles labeled with FITC dye. FITC (fluorescein
isothiocyanate) is a synthetic organic compound that is widely used as a fluorescent dye in many applications.
Its absorption peak is at 494 nm and emission maximum is at 512 nm.1
8
A more complex and informative process is the two-parameter histogram shown in Figure 1-5,
one on the x-axis and the other on the y-axis. The cell counts of the measurements are shown as a
density plot, or contour map.
Figure 1-5 – Sample of a two-parameter (dual color fluorescence) histogram. Particles in this example are
labeled with PE and FITC dyes. PE (Phycoerythrin) is another organic fluorochrome whose absorption peak
is at 488 nm and 532 nm, and emission maximum is at 585 nm. 1
In this example, R2 encompasses the Phycoerythrin (PE)-labelled B cells whereas R5 contains
the fluorescein isothiocyanate (FITC)-labelled T cells. Both PE and FITC are organic fluorescent
dyes commonly used in microscopy and flow cytometry in biological applications. PE has an
absorption maximum at 488 nm or 532 nm, and an emission peak at 585 nm while FITC has an
absorption maximum at 494 nm and an emission peak at 512 nm.1,4 The top right quadrant
contains a few “activated T cells” that possess some HLA-DR (Human Leukocyte Antigen –
antigen D related) expression. HLA-DR is a type of cell surface receptor that mediates
acceptance or rejection of tissue or organ transplants.5,6 As the T cells with HLA-DR expression
are labeled with both antibody markers (PE and FITC), they are grouped in their own region
(R3). R4 contains cells negative for both FITC and PE, as shown in Figure 1-5.
1.2.3 Labels
To achieve specific detection and analysis of biological samples, target particles or analytes are
typically conjugated with labels. Most common labels used in flow cytometry are fluorescent
labels and isotope labels.1
9
1.2.3.1 Fluorescent Labels
A wide range of fluorophores can be used as labels in flow cytometry. Fluorophores are typically
conjugated to an antibody that recognizes a target feature on, or in the cell, they may also be
attached to a chemical entity with affinity for the cell membrane or another cellular structure.
Each fluorophore has a characteristic peak excitation and emission wavelength, and the emission
spectra often overlap. Consequently, the combination of fluorescence labels which can be used
depends on the wavelength of the excitation sources used and on the detectors available.
1.2.3.2 Isotope Labelling
Currently, due to the limited available bandwidth in fluorescence detection, the maximum
number of labels that can be used in a flow cytometer is 17. In another approach intended to
overcome the fluorescent labelling limit, lanthanide isotopes are attached to antibodies. This
method could theoretically allow the use of 40 to 60 distinguishable labels and has been
introduced into a plasma, ionizing them and allowing time-of-flight mass spectrometry to
identify the associated isotopes. Although this method permits the use of a large number of
labels, it currently has lower throughput capacity than traditional flow cytometers. In addition,
this labelling technique destroys the cells, precluding their recovery by sorting.
1.2.4 State of the Art
The state of the art flow cytometers can perform measurements on samples labelled with up to 17
fluorescence markers simultaneously. Adding the forward scattering and side scattering, that
implies a total of 19 parameters that can be detected simultaneously, making flow cytometry a
powerful tool in clinical and research tool in biology and medical applications. The measurable
parameters of today’s flow cytometers include total DNA content, nuclear antigens, chromosome
analysis and sorting, cell detection and enumeration, enzymatic activity and characterizing
multidrug resistance in cancer cells.
1.2.5 Modern Cytometry Applications
Modern flow cytometry has application in a number of fields, including molecular biology,
pathology, immunology, plant biology and marine biology. In medicine the flow cytometer is
the tool of choice for measuring or monitoring a wide range of conditions such as
transplantation, hematology, tumor immunology and chemotherapy, prenatal diagnosis, genetics
10
and sperm sorting for sex pre-selection. For example, in marine biology, the auto-fluorescence
properties of photosynthetic plankton can be exploited by flow cytometry in order to characterize
abundance and community structure. Similarly, in protein engineering, flow cytometry is used in
conjunction with yeast display and bacterial display to identify cell surface-displayed protein
variants with desired properties.
1.3 Aims of this Work
The flow cytometer has become an important tool in today’s clinical and research settings. The
introduction of novel labels such as quantum dots and nanoparticles enabled development of
novel probes and techniques to detect and characterize microscopic particles. On the other hand,
the development of new optics and electronics has opened doors for miniaturization and
advancement of flow cytometry instrumentation. The aim of this thesis is to develop a portable
particle detection and analysis system, which can be integrated with a handheld device, enabling
a more mobile and effective cell cytometry analysis.
Prior to this work, attempts at realizing portable/handheld cytometer were made. A
electrochemical based particle/cell detection system was developed by Li et al.7 where detection
was accomplished by measuring the impedance of the sample. Morgan et al.8 also developed a
microfabricated flow cytometer system where integrated photonic circuits and fibers were used
to excite and collect optical signals of target particles and molecules. However, these attempts
lacked a design approach that can be practically packaged in a commercial instrument for use in
clinical or low resource settings. In this thesis, microfluidic, lab-on-a-chip devices were
developed which allow various clinical and biological assays to be developed in a cartridge
format. Its size and physical properties brings significant advantages to developing new tools for
biological, chemical and medical analysis. It presents an ideal platform for the potable
cytometer/cell analysis.
11
Figure 1-6 – Rendering of the proposed handheld cytometer. The instrument is intended to be a multi-test
platform system that can be used for a range of applications.
The ultimate aim of this work is to develop a portable, cartridge based cytometer system, as
illustrated in the 3D rendering shown in Figure 1-6, that can be integrated with a commercial
grade instrument to produce a portable system which can be taken into a rural or remote setting.
The key challenges in this work include integration of optics, lyophilization, fluorescence
imaging and microfluidic lab-on-a-chip devices.
The target requirements for this portable cytometer are summarized as follows:
Sample collection: finger prick
Sample volume required: 10 µL
Requirement for metering: Yes
On-chip reagent storage: Yes
Format of on-chip reagent: Dried
Mixing and actuation: Active and/or passive
Sample analysis throughput: > 0.28 µL/min
Overall test completion time (for CD4 cell counting assay): < 20 min
Instrument hardware must be portable, less than 2 kilograms in weight
Optical detection methodology: fluorescence
Light source: LED or Laser with emission wavelength at 532 nm
Detection optics: must have a magnification of 10x and a field of view of 1mm by 1mm
The proposed system will advance HIV monitoring, treatment, patient care, and point-of-care
analytical tools in general. The process of prognosis will be extremely efficient by this platform’s
ability to deliver a patient’s current condition in a matter of minutes. By integrating the smart
12
electronic device functions including position tracking, communication, biometric identification
with advanced optical sensing technologies, this system can provide a complete and total
solution from patient identification, disease monitoring to patient care, tracking and treatment.
1.4 Organization of Thesis
The tremendous promise of microfluidic lab-on-a-chip devices, particularly in the application of
point-of-care testing, is the main motivation for this work.
Chapter 1 introduces the flow cytometry and its application in modern biology and medicine.
Chapter 2 focuses on the background of the work, including topics in point-of-care testing,
microfluidics, and current state-of-the-art in microfabricated and miniaturized flow cytometry
instruments and devices. Starting from Chapter 3, the thesis describes the main contribution of
this work: microfluidics of the disposable device, detection methodology, and image acquisition
and analysis. Chapter 3 and 4 presents on-chip fluidics developed for the miniaturized flow
cytometry application. In Chapter 5, a unique and novel imaging approach is proposed and
characterized while Chapter 6 is a summary of the thesis with a brief look ahead on potential
future work.
13
Chapter 2 Background
2.1 Microfluidic Lab-on-a-Chip
Microfluidics is the study of systems that analyzes small amount of fluids (10-9 to 10-18 liters),
using channels with dimensions of tens to hundreds of micrometers. Some of the first research
and scientific projects involving microfluidics began with the development of high pressure
liquid chromatography9 and capillary electrophoresis10,11.
2.1.1 History of Microfluidics
Towards the end of 1970s, the field of microfluidics emerged as a valuable tool for analytical
chemistry as well as pharmaceutical research. A silicon micro-machined gas chromatograph was
first developed in 1975 by Terry12,13. This remarkable gas chromatograph was capable of
separating simple mixture of compounds in a span of a few seconds. The device included a valve
to control sample injection, a column that is 1.5 meters long for separation, with all functional
components integrated in a single silicon wafer. A thermal conductivity detector was fabricated
on a separate silicon wafer and mechanically clamped to the wafer containing the column.
However, this research was not very well received by the academia community. Related research
were primarily focused on the fabrication of component such as micropumps14-18,
microvalves19,20, and chemical sensors21,22. The need to precisely control liquid flow led to the
development of micropumps and valves in the integrated systems.23-25 Furthermore, the concept
of miniaturized total chemical analysis system (µTAS), also known as Lab-on- Chip (LOC), was
first proposed by Manz et al. in 199026, in which silicon chip analyzers incorporating sample
pretreatment, separation, and detection played a fundamental role. At the time, µTAS was
envisioned as a new concept for chemical sensing, given the sensor technologies available were
not providing the necessary requirements in terms of selectivity and lifetime. In the beginning,
the main purpose and motivation for miniaturization was to enhance the analytical performance
of the device. However, it was realized that smaller overall sensor architecture also meant
smaller consumption of sample and reagent. The concept of µTAS enabled the integration of
separation techniques that could provide the monitoring of many components within a single
14
device. It was envisioned such devices were capable of performing multiple functions in the
same device20,21, including sample handling, transport, analysis, detection and measurement,
hence the name total analysis system.
Early theoretical considerations of the miniaturization concept demonstrated that pumping using
electro-osmosis was an effective and possible method to transport samples inside the on-chip,
interconnected channel systems, especially those where separation was required26. On the other
hand, conventional pumps utilizing high pressure systems showed problems in such systems
during fluid and sample transport when interconnecting channels were on the order less than 20
µm wide or deep. Research efforts continued to develop novel pumping methodologies16,27 and
sample introduction17,20 in such microsystems.
Electrophoresis in planar chips was successfully integrated and demonstrated for the first time in
1992 using silicon and glass substrates28,29. These results showed the feasibility of using
electroosmotic pumping for flow control in interconnected microchannels without the use of
valves and the use of glass and silicon as a potential material of choice for µTAS. The concept of
µTAS that integrates sample introduction, separation and detection within the same device was
first demonstrated. Subsequent efforts were made towards increasing the separation efficiency.29-
32, where amino acids and dyes were separated in less than 30 seconds with plate height of 0.3
µm. In addition, automated repetitive sample introduction and separation on a time scale of
seconds were also achieved. The most common detection method was laser induced
fluorescence, which was used to detect mixtures of fluorescein derivatives and fluorescein
isothiocyanate-labeled amino acids separated on-chip.33 In addition to the separation of
biological samples, applications related to the reaction and handling of biomolecules and cells
began to develop. These include the use of microfabricated chambers to carry out Polymerase
Chain Reaction (PCR) to amplify DNA strands,34 the measurement of cellular metabolism in
micromachined channels,35 and the use of microfabricated devices for flow cytometry
application.36
2.1.2 Recent Advances
Recent advances in microfluidics science and technology are enabling more functions and
features that can be integrated in a lab-on-a-chip device to complete more complex analysis.
15
Progress has been made on a multitude of fronts, from materials to fabrication techniques, from
fluidic manipulation to detection, and from integration to application. Iliescu et al.37 reviewed
silicon and glass microfluidic systems. A polymer/glass-hybrid approach was proposed and
presented 38,39 which led to a renaissance in silicon based detectors for microfluidic systems.
Droplet based PCR40, cellular culture41 and nanowire for label-free cardiac enzyme detection42
were also developed in recent research activities.
A great deal of research has also focused on the development of on-chip fluidic manipulation and
functionalities, including sample separation, mixing, fluidic management and detection.
Although microfluidic electrophoretic systems received more initial attention than
chromatographic ones, important progress has been made in both areas, as covered in greater
detail in recent review articles on microfluidic chromatography43,44 and electrophoretic
approaches.45,46
Optical detection methods have several advantages over other electrical based methods.
Typically, optical detection techniques have good detection limits, they can be isolated from the
fluid inside the fluidic channels, and can be used to monitor a wide variety of compounds.47
Mogensen et al. provided a very in-depth investigation on this topic in a recent review.48 There
are several approaches to optical detection that are currently being implemented in microfluidic
systems; these can be classified as label-based, such as fluorescence and chemiluminescence, or
label-free.49 More specifically, Laser-induced fluorescence (LIF) is the most frequently used
optical method in microfluidic systems because of its low detection limits and availability of off-
the-shelf components and devices to design such systems.50 In many instances, the detection
optics are not integrated in the fluidic chip itself. For LIF, a light source such as a laser is used
for excitation and a CCD or photodiode detector is used for detection.51,52 Integration of an LED
excitation source and a photodiode detector in a micro-system has been shown, producing
relatively high limits of detection of 100 nM for rhodamine 6G and 10 μM for fluorescein.53
Integration of pumps into microfluidic devices can significantly decrease external equipment
needs and reduces the dead volumes from interfacing with pumps. However, it could increase
complexity and difficulty in manufacturing of the microfluidic devices. A detailed review on
micropumps and microvalves has also recently been published by Au et al.54 Because flow inside
microfluidic channels is strictly laminar due to the microscopic channel size, the main driving
16
force for mixing in a microfluidic system is by diffusion. Several micromixer designs have been
developed in the pursuit of thorough and rapid mixing of multiple samples.55 Detailed reviews of
mixing systems for microfluidics have been written by Jeong et al.56 and Lee et al. 50
These are only few examples highlighting the recent advances in the field of microfluidic lab-on-
a-chip research. Nge et al. has presented a more thorough and complete in-depth summary of
microfluidics research in a recent review article.57
2.1.3 Microfabrication Technologies
The development of microfabrication technology for microfluidic lab-on-a-chip systems
stemmed from the readily available and recent advances in microelectronics manufacturing
processes. It is clear from the history of the development of microfluidics, the material (silicon
and glass) and fabrication process (photolithography) were directly taken from the
microelectronics industry.20,21 Gradually plastics became the dominant material selection for
microfluidics devices based on its excellent optical quality, biocompatibility and ease of
manufacturing at both prototyping stage and for large scale production. The following sections
outline some of the most common fabrications technologies currently used.
2.1.3.1 Laser Processing
Laser ablation was used by Grzybowski et al. for rapid fabrication of elastomeric masters
(poly(dimethylsiloxane) or PDMS) for micro-contact printing (µCP) and a technique named
controlled sagging micro-contact printing (CSµCP) was developed. This approach is capable of
patterning structures with dimensions of 1 µm in width without the need of a cleanroom
environment, or specialized photolithographic tools.58 With the advent of ultrafast photonics, Ho
et al.59,60 demonstrated the feasibility of creating 3D structures inside a fuse silica bulk. In this
approach, the fused silica was directly exposed to ultrafast laser pulses followed by chemical
etching with diluted hydrofluoric acid. Both woodpile structures and 3D microfluidic channel
networks were fabricated with a resolution of 5 µm.
17
2.1.3.2 Lithography
2.1.3.2.1 Photolithography
Photolithography is the process of transferring geometric shapes on a mask to the surface of a
silicon wafer.61-63 It uses light to transfer a geometric pattern from a photomask to a light-
sensitive chemical material known as “photoresist”. A series of chemical treatments then either
etches the exposure pattern into, or enables the deposition of new material in the desired pattern.
The steps involved in the photolithographic process are wafer cleaning; barrier layer formation;
photoresist application; soft baking; mask alignment; exposure and development; and hard-
baking.
Many microfluidic lab-on-a-chip devices have been fabricated using photolithography
techniques, particularly using SU-8 photoresist. For example, the preparation and
characterization of optical and mechanical properties of SU-8 negative photoresist for the
fabrication of high aspect-ratio structures were reported by Lorenz and co-workers.64 Using
alternating spin-coating of SU-8 photoresist and exposure steps followed by a single
development step to remove the unpolymerized resist, Guérin et al. have fabricated monolithic
SU-8 channels.65 Microchannels of 80-nm width on carbon-based resist were fabricated by
Johnson et al. on Si, SiO2, and gold substrates by exposing them to a metastable argon atom
beam in the presence of dilute vapors of trimethylpentaphenyltrisiloxane.66
2.1.3.2.2 Soft Lithography
Pioneered by Whitesides and Quake, soft lithography is a unique technique to fabricate structures
using elastomeric stamps, molds and confirmable photomasks.67 It is generally used to build
features measured typically on the order of nanometer to micrometer scale.68 The key advantage
of soft lithography include lower cost than traditional photolithography, more suitable for
biotechnology applications and processes,69 and ability to rapidly generate prototype device.70
Since its first publication in 199867, soft lithography has become the most widely used rapid
prototyping technique in the research and development of microfluidic systems. The work
summarized in Section 2.1.2 was predominantly completed using soft lithography fabrication
process.
18
2.1.3.3 Injection Molding
As research on microfluidic lab-on-a-chip devices grows and matures, many research activities
are focused on the manufacturing of integrated microfluidic devices on a mass-production scale
with relative low costs.71 This is especially critical for applications where disposable devices are
used such as medical diagnostics and analysis. For mass production, polymer material presents a
very attractive alternative material choice compared to glass and silicon given its cost and ease of
manufacturing.
Micro-injection molding is the process of transferring a thermoplastic material in the form of
granules from a hopper into a heated barrel so that it becomes molten and soft. The material is
then forced under pressure inside a mold cavity where it is subjected to holding pressure for a
specific time to compensate for material shrinkage. The material solidifies as the mold
temperature is decreased below the glass-transition temperature of the polymer. After sufficient
time, the material freezes into the mold shape and gets ejected, and the cycle is repeated. A
typical cycle lasts between few seconds to few minutes. The process has a set of advantages that
makes it commercially applicable with potential for further developments in the future.
Advantages include the wide range of thermoplastics available and the potential for full-
automation with short cycle times,72,73 cost-effectiveness for mass-production process, especially
for disposable products44,45,74-76, very accurate shape replication and good dimension control45,46,
low maintenance costs of capital equipment, when compared to lithographic methods77, and
applicability of the large amount of industrial information and know-how available from
conventional injection molding. Within certain limitations, this may be scaled down to micro-
injection molding.
A comparison between micro-injection molding and other techniques for microfluidic device
manufacturing, such as hot-embossing and PDMS casting, is also available in the literature76.
Other micromachining techniques have been presented using deep reactive ion etching (DRIE)
and surface fusion bonding (SFB)78 and an electron cyclotron resonance (ECR) source79,80 to
produce high-aspect-ratio narrow-gap silicon devices.
On the other hand, polymers have some limitations related to their properties or processing
techniques relative to glass. These include limited operation-temperature range, higher auto-
fluorescence and limited well-established surface modification techniques. Table 1 presents a
19
comparison between polymers and glass for manufacturing microfluidic devices, where
information is compiled from the literature.77,81-88 When it comes to processing, mass-production
processing techniques impose limitations on the moldable geometry of the microfluidic device.
These geometrical limitations restrict the flexibility of integrating external functional elements
within a mass-manufacturing technique.
20
Table 1 - Comparison between polymer and glass as substrate of microfluidic systems for bio and chemical
applications.
Polymers Glass
Manufacturing costs Lower costs than glass, especially
in mass volume
More expensive to
manufacture as the fabrication
process are more complex
Fabrication complexity Simpler fabrication process Time consuming and
expensive, and usually wet
chemistry is used.
Clean room facilities Cleanroom environment is
necessary
Clean-room facilities are
required.
Properties A wide range of polymer material
to choose from to tailor the
desired requirements
(mechanical, optical and
chemical)
Less variability in available
properties compared to
polymer.
Operation
temperature
Narrower range due to the low
glass transition temperature
Wider range of operation
temperature than polymer.
Optical properties and
fluorescence detection
Higher auto fluorescence in the
UV end of the spectrum and
lower transparency than glass
Superior optical property than
glass.
Bonding Different bonding options are
available, including adhesives,
thermal fusion and mechanical
welding
More time consuming than
polymer. Possible bonding
processes include thermal,
adhesive and anodic bonding.
21
Surface treatment Direct treatment available
including oxygen plasma
treatment
Established chemical
modification procedures for
glass are available.
Compatibility with
organic solvents or
strong acids
Generally not compatible with
most organic solvents and in
some cases, strong bases or acids
Excellent resistance to
solvents and acids.
Joule heating Significant Joule heating due to
low thermal conductivity
More resistant to Joule
heating.
Electro osmotic flow
(EOF)
Smaller EOF because of lack of
ionisable functional groups
Higher EOF compared to
polymers.
Geometrical flexibility More flexibility for geometrical
designs with a wide selection of
different cross-sections (curved,
vertical or V-groove); high aspect
ratio and arbitrary wall angle
Limited to 2D designs due to
the isotropic nature of the
etching process. Less flexibility
in cross-sections than
polymer.
Permeability to gasses Higher gas permeability relative
to glass
Does not meet the gas
permeability requirements for
some biological applications
such as living mammalian
cells.
2.2 Point-of-Care Testing
Point-of-care testing (POCT), also known as bedside testing, is defined as medical testing
conducted at, or near the site of patient care.89 Typically, these are simple blood or urine tests
which can be performed by nurses, healthcare workers or patients. Examples of such POCT
22
systems include rapid diagnostic tests for HIV, malaria, pregnancy test, blood glucose monitor,
blood chemistry and electrolytes analysis, and portable ultrasound just to name a few. The main
driving force in the development of POCT is to improve the accessibility of medical tests to the
patients. This increases the likelihood that the patient, physician and healthcare professionals to
receive the results faster, which leads to immediate clinical and patient care decisions.
Generally, POCT consists of a portable, or handheld instrument and a disposable test kit or
cartridge.90 The ultimate goal is to collect patient sample and obtain the test results in a very
short period of time at, or near the location of the patient so that clinical decisions can be made
faster. Cheaper, faster and smarter POCT devices have increased the use in a wide range of
disease diagnosis and testing, such as diabetes, acute coronary syndrome, malaria and
HIV/AIDS.91
A major class of POC diagnostic tests is the lateral flow test, which uses a membrane or paper
strip to indicate the presence of protein markers such as pathogen antigens or host antibodies. On
a membrane, addition of sample induces capillary action without user intervention. As the
sample flows across the membrane, it reacts with embedded labeling reagents in the membrane,
and flows over an area that contains capture molecules. The labeled, captured analytes are
usually read and interpreted by human eye to form a visible band. In the U.S., lateral flow tests
are used for diagnosis in a small number of indications, most notably pregnancy as well as
infections with streptococcus or flu. On the other hand, in developing countries or remote areas,
the lateral flow test is widely used to diagnose infectious diseases such as HIV and malaria.
Although the test is simple to perform, the single-flow action does not resemble the multi-step
procedures of laboratory-based assays that are crucial for producing highly reproducible,
quantitative, and sensitive results. As the lateral flow test comprises a multibillion dollar market,
with many of the technologies now consolidated at the company Alere (formerly Inverness
Medical Innovations), there has been significant industry interest in trying to improve their
performance over the last few decades, but so far without significant progress.
The other major class of successful POC tests is the blood glucose test, which can be considered
as a classic example for a high impact POC diagnostics product that has improved millions of
diabetic patients’ lives, and now acts as a pillar of the entire diagnostics industry. The glucose
test is also performed on membranes, but is typically classified differently from lateral flow
23
immunoassays as the rest of the analytical method is altogether different. The glucose test uses
signal amplification by a redox enzyme, which generally generates an electrochemical signal to
be detected by a reader.92 The glucose test, however, is by nature is somewhat unique. On the
technology side, the concentration of the analyte is in the mM range, which far exceeds the
concentration of most diagnostics markers. On the market side, the frequency of testing, typically
multiple times a day, outnumbers that of most other tests, which creates a billion-dollar market
world-wide. Both factors contribute to the rapid development and tremendous success of the
POC glucose testing.
POC testing has seen significant and rapid advances in the last decade.92 A number of
technologies and products were successfully developed and commercialized. The research and
development of POC tests often require interdisciplinary expertise, ranging from biology,
chemistry, to fluidics, electromechanical or optical detection. The availability of POC tests for
infectious and non-communicable diseases enabled faster and more affordable medical
diagnostics to be conducted at or near patient, significantly reducing the burden on healthcare
systems world-wide.
2.2.1 Microfluidics and POCT
With the emergence of new POC diagnostic technologies in the market, there has recently been a
resurgence in interest to develop novel and clever methods to re-invent both lateral flow
immunoassays and the glucose test, to significantly improve their detection limit, quality control
and readout systems, as well as expanding their range of targets. Examples include marrying
modern LOC concepts, such as sophisticated flow control, to diagnostic tests performed on paper
and membranes.93-95 In the current concept of LOC-based devices, the iSTAT handheld system
(now part of Abbott) was among the first commercially successful products. The iSTAT system
marries miniature fluidics and electrochemical detection to conduct clinical chemistry
measurements, such as electrolyte levels and limited immunoassays using a disposable test kit.
Another interesting hybrid of LOC technologies with lateral flow is the A1cNow® test for
diabetic patients formerly from Metrika (now Bayer Healthcare), which uses multiple strips
integrated with detection optics in a single package.
24
The clinical need for new POC diagnostic tests remains high, especially for tests that can detect
low concentrations of the target and with an ability to quantify the result. Currently, tests such as
lateral flow immunoassays can detect analytes present in a sample solution at high native
concentrations ranging from µM to mM. For targets present at low native concentrations that are
beyond the detection limit of the current detection methodologies, the assay systems require
amplification either of the signal or the target, which typically increases the complexity of the
testing device, and is not available in current POC tests.96 In addition, current POC can mostly
produce qualitative results. The challenges in designing and manufacturing quantitative tests
remain prohibitive. Finally, multiplexing is another desirable feature that current POC tests strive
to incorporate.
POC testing is beginning to benefit from the potential and recent advances in microfluidics
research. The degree of integration of a microfluidic technology can vary from having a
disposable microfluidic chip used with peripheral equipment (pumps, reader, etc.) to having all
functions needed for processing and analyzing a sample and reporting the results on a chip. The
main selection criteria are portability, time to result and cost per test. The time to result is
between seconds and minutes as devices are often used at the patient side and timely results are
key requirements. Multiplexing is usually done for a few analytes. The size and weight of the
device are minimal and affect the portability and energy consumption of the reader peripheral.
The cost per test must be low in order for the test to be performed routinely and fit into pricing
and reimbursement policies that are relevant for the geography where the tests are performed. As
can be seen from above, there is a large number of requirements that POC diagnostics must meet.
Generally, technologies for research and central laboratories meet these requirements by using a
variety of peripheral equipment, several sampling methods, flexible protocols, and a number of
signal detection formats. In contrast, a POC microfluidic device is optimized during
manufacturing for a particular application.
Central laboratory testing is done mostly on clinical analyzers. The main selection criteria are
throughput and cost per test. Samples are sent from the patient to the central laboratory and
placed in a queue with an option for high priority. The time to result can be from several minutes
to hours and is usually not critical. Clinical analyzers have a large variety of analysis capabilities
and can detect hundreds of analytes. Machines can be meters in size and weigh more than a ton.
25
Operators of clinical analyzers are usually trained technicians. In clinical laboratories that
perform a large number of tests, the instrument is often provided and the instrument cost may be
small compared to the running costs. The cost per test should be low enough to be done
routinely, but can be higher for less common analytes.
Microfluidics is, at its heart, a technology, with a primary goal of improving the performance of
end products. Lab-on-a-chip (LOC) technologies utilizes the manipulation of fluids and particles
in a microfluidic system, which can be fabricated in a glass or polymer substrate. LOC enables
the significant reduction in sample and reagent volumes compared with conventional bench
based, macroscopic analysis techniques. Secondly, chemical separation procedures are much
faster and more efficient at these small dimensions.11 Arrays of similar structures on one chip
allow for a large group of measurements to be made under the same conditions and at the same
time. Most importantly, since sample handling, reactions, separation, transport and detection all
take place on the same substrate, supplementary connecting interfaces between these different
functions are eliminated. For building POC diagnostic devices, there have been dazzling progress
in the development of individual LOC components, but unfortunately only very few microfluidic
technologies have made the leap to fully functioning integrated devices that provide real clinical
value.96
2.2.2 Lab on a chip and Global Health
Lab-on-a-chip (LOC) technologies have a tremendous potential to improve the efficiency of
healthcare system in Low and Middle Income countries (LMIC). Ever since the modern
inception of LOC and microfluidic technologies around 1990, use in remote settings has been
perceived as potentially one of the most powerful applications of the technology. Indeed,
portable LOC devices are now beginning to be used in remote settings, as a result of
developments in integrating fluid actuation, sample preparation, sample separation, signal
amplification, and signal detection into a single device.
There is an urgent need in developing countries for new and innovative health-related
technologies, and specifically, new technologies for health diagnostics, to improve patient care.
For example, in one survey of international scientists familiar with the public health programs of
26
developing countries, Singer et al. discovered that the overall top priority in new technology
development for global health was “modified molecular technologies for affordable, simple
diagnosis of infectious diseases”.97 Similarly, in a study by the Bill and Melinda Gates
Foundation and the NIH to identify “Grand Challenges for Global Health”, two of the 14
priorities involved diagnosis and measurement of patients’ health statuses.98 Microfluidic LOC
technology holds substantial potential for fulfilling these challenges by automating complex
diagnostic procedures that are normally performed in a centralized laboratory on a microfluidic
chip. This outcome could empower healthcare workers and patients with simple-to-use POC tests
to generate important health-related information in even the most remote settings. To this effect,
funding by philanthropic foundations (such as those from Doris Duke, Soros, and Gates) are
leading the development of microfluidics technologies for diagnostics in LMICs. The broad aim
of these scientific initiatives is to combine new diagnostic and prevention methods with
treatment to improve global health.99,100
In both developed and developing countries, early and accurate diagnosis for every disease is
critical for the well-being of patients: it permits prompt and proper treatment of patients, and
minimizes the waste of public resources on ineffective treatments.97 In developing countries, the
value of diagnosis for certain diseases is sometimes mitigated by the lack of available treatment.
Early diagnosis can allow patients to receive required therapy and medication on time, reducing
morbidity and mortality, and investments in diagnostics and prevention can be more cost-
effective than treatment.101 Moreover, point-of-care devices can enable epidemiological
surveillance of diseases,102 which is an especially challenging problem in developing countries.
For scientists and engineers who aim to design new diagnostic technologies, a crucial question
for achieving real world impact is which health conditions in developing countries are most in
need of diagnostic devices. In a study led by Murray and Lopez, the World Health Organization
conducted an unprecedented and comprehensive initiative to compile statistics for comparing the
relative burden of diseases, conditions, injuries, and risk factors on a global scale.103,104 Table 2
summarizes the most common diseases by disability adjusted life years (DALYs) in developing
countries, which is a metric that accounts for years of life lost due to premature death and
disability. Infectious diseases constitute a large burden of disease in developing countries
(32.1%; by comparison, they represent only 3.7% of total DALYs in developed countries). The
combination of HIV/AIDS, malaria, and tuberculosis (TB), constitutes an important 12% of
27
DALYs in developing countries. The social impact of these diseases stretches beyond the DALY
statistics, however, since HIV/AIDS (along with common co-infections of TB) targets healthy
adults, thereby leaving behind villages of orphans which destroy the underlying functioning of
entire communities.105
In addition to infectious diseases, the burden of non-communicable diseases in developing
countries is often underappreciated.106 The list of important non-communicable diseases include
cardiovascular disease (such as ischemic heart disease and stroke), cancer, neuropsychiatric
conditions (such as unipolar depressive disorder), and respiratory diseases (such as chronic
obstructive pulmonary disorder and asthma). As the standard of living in developing countries
improves and average life span increases, the burden of disease will gradually shift to the non-
communicable diseases. Already, obesity and diabetes are increasingly prevalent in developing
countries.107 As these trends develop, accessibility of the corresponding diagnostic technologies
in developing countries can be difficult in developing countries due to the lack of infrastructure
and inadequacy in the healthcare systems.
28
Table 2 - Prevalent diseases in developing countries. DALY is a measure of the disease burden on healthcare
system. Type of test is the conventional testing methodology.105
Disease % DALY
Type of Test
Communicable diseases 32.1
Respiratory infection 6.8 Immunoassay
HIV/AIDS 6.1 Immunoassay
Diarrheal disease 4.5 ELISA
Malaria 3.4 Microscopy; immunochromatography
Tuberculosis 2.5 Microscopy; PCR
Non-communicable diseases 43.5
Neuropsychiatric conditions 11.7 Hormone levels
Cardiovascular disease 9.5 ELISA
Sense order disease 4.6 clinical diagnosis
Cancer 4.2 Immunoassay
Respiratory diseases (asthma) 3.5 Spirometry
Digestive diseases 3.0 Complete blood count and blood chemistry
Maternal perinatal and nutritional conditions
11.8 No tests available
Perinatal conditions 7.0 Clinical diagnosis
Nutritional deficiencies 2.5 Immunoassay; cell count
Maternal condition 2.4 Hematology
Intentional injuries 3.3 Cell culture; immunoassay
Unintentional injuries 9.2 Analytical toxicology
29
To diagnose this wide array of diseases and conditions, assays with a variety of methodologies
will be needed. The types of assays that are currently used to diagnose them are listed in Table 1;
some assays are in great need of new diagnostic methods.
The use of point-of-care testing systems in the resource limited settings pose a series of design
challenges, including affordability, time to analysis result, ease of use and rugged system design.
In addition, the system must be operable without access to stable electricity and clean water.
These constraints hold direct pertinence to the design of the diagnostic technology. It is
necessary to take them into consideration at the earliest design stages to ensure the performance
of the product is satisfied.
2.2.3 Motivation of this Thesis
A key problem with the current healthcare system lies in the disease diagnostics. Most of the
medical devices and instrumentations are designed for use in the central laboratories and
hospitals where established infrastructure must be available to operate the equipment effectively.
As a result, for people living in resource limited settings or remote areas around the world,
access to medical diagnosis is very limited and sometimes non-existent. The aim of this thesis is
to develop novel platforms that enable easy access to state-of-the-art in vitro diagnostics by
utilizing recent advances in microfluidic lab-on-a-chip technologies. The target assay chosen in
this work is CD4 T cell count test, which is a measure of a human’s immune system strength.
2.2.3.1 CD4 T Cell Count and HIV/AIDS
The CD4 T-cell count (CD4 count) serves as the major laboratory indicator of immune function
in patients who have HIV infection. It is one of the key factors in determining both the urgency
of antiretroviral therapy (ART) initiation and the need for prophylaxis for opportunistic
infections.108 It is also the strongest predictor of subsequent disease progression and survival
according to findings from clinical trials and cohort studies109-111.
CD4 cells or T-cells are a type of white blood cells that play a major role in protecting your body
from infection. They send signals to activate your body’s immune response when they detect
“intruders,” like viruses or bacteria. Once a person is infected with HIV, the virus begins to
attack and destroy the CD4 cells of the person’s immune system. HIV uses the machinery of the
30
CD4 cells to multiply (make copies of itself) and spread throughout the body. This process is
called the HIV life cycle.
A CD4 count is a lab test that measures the number of CD4 cells in a blood sample. It is an
important indicator of how well the immune system is working. The CD4 count of a healthy
adult/adolescent ranges from 500 cells/mm3 to 1,200 cells/mm3. A very low CD4 count (less than
200 cells/mm3) is one of the ways to determine whether a person living with HIV has progressed
to stage 3 infection (AIDS). ART involves taking a combination of HIV medicines every day. It
prevents HIV from multiplying and destroying patient’s infection-fighting CD4 cells. ART
cannot cure HIV, but it can help patient live a longer, healthier life and reduce the risk of HIV
transmission by suppressing the virus.
When the amount of HIV in a patient blood is lowered by ART, it allows the CD4 cells to
reproduce and increase in number. The higher the CD4 count, the more capable patient is to fight
HIV and other infections. ART is recommended for everyone with HIV, but the urgency to start
ART is greater in people with low or rapidly falling CD4 counts. A falling CD4 count indicates
that HIV is advancing and damaging patient’s immune system.
Traditionally, a CD4 T cell count is completed using flow cytometry, a well-established
methodology that requires sophisticated infrastructure and technical expertise to operate. In
developing countries, the lack of healthcare infrastructure and resources make it impossible to
monitor the CD4 T cells in all HIV patients. There is an urgent need to develop simple to use,
affordable POC CD4 testing solutions with vigorous quality control protocols.
2.2.3.2 Current Gap and Proposed Concept
Flow cytometry is an important blood analysis technology that remains largely inaccessible for
clinical use in low and middle-income countries (LMICs) due to size, cost, and the lack of
infrastructure and skilled personnel. The result is a highly inefficient health system where a
patient’s blood must be sent to a central laboratory (when available). The patient must return
weeks later to get test results. This delay blocks clinician decision making; test results often do
not make it back to the clinician; and patients are often “lost to follow up” leading to increased
morbidity and death.
31
This thesis aims to develop a novel laboratory-quality diagnostic platform, which can be
described as a mini flow cytometer. The platform is intended to be mobile, simple to use, and
inexpensive. By leveraging advances in microfluidic and biomarker technologies, we plan to
research and develop a microfluidic based point-of-care system that enables the detection of cell-
surface and blood serum protein biomarkers. The ultimate objective is from a single drop of
blood, health workers in remote locations will rapidly and accurately perform tests to diagnose or
monitor a range of infectious and non-communicable diseases. The portable, or handheld
platform will allow for testing at community level facilities, mobile clinics and on hospital
wards. Simplicity of use will allow front-line nurses, laboratory technicians and advanced
community health workers to reliably perform accurate tests. Disposable microfluidic cartridges
do not require cold chain and have a minimum 12-month shelf life. Quantitative results will be
available within tens of minutes (depending on test). A rechargeable battery will allow for day-
long continuous use. The low cost of platform and tests will allow for the cost-effective uptake
of the platform in low-test throughput sites (5-20 tests per day). Cloud connectivity enables the
review of results from a central location for quality control, clinical decision-making support,
and facilitates electronic medical record (EMR) data aggregation. Diagnostic testing available at
the point-of-care, especially in remote or rural settings, will improve patient care and morbidity
and mortality outcomes, improve health worker motivation, increase health system efficiencies,
and significantly reduce health system costs.
Starting from Chapter 3, this thesis will present the design and development of a novel
microfluidic based imaging and detection system. This portable system forms the core of a
handheld cytometer that can be further integrated into a point-of-care tool for clinical testing and
diagnosis. As a first demonstration of its clinical application, a CD4 T cell counting test is part of
the objective of this thesis and described in the subsequent chapters. Microfluidic chips were
designed, fabricated and tested against the current state-of-the-art flow cytometers to validate the
performance of the detection methodology. Engineering challenges in designing on-chip
functions, such as reagent incorporation, blood cell labeling and mixing, on-chip fluidic actuation
and fluorescence detection were described in detail in Chapter 3, 4 and 5.
The same detection platform can be further expanded to image and analyze blood serum proteins
using beadarray technology. Chapter 5 also included an introduction the concept of the beadarray
32
system and provided a proof-of-concept demonstration and analysis towards building a portable
or handheld ELISA-like beadarray instrument.
33
Chapter 3 Passive Microfluidic Systems
In flow cytometry, precise control of fluid flow is critical to ensure the proper functioning of the
instrument. For microflow, or microfluidic-based cytometers, this is of utmost importance to the
overall system. Typically, conventional flow cytometers employ a sheath flow surrounding the
sample stream to form the particles/cells into a single file stream. This stream must be stable
without significant pulsation and well aligned with the detection optics for optimal sensing of
each particle/cell. The shape, width, and height of the detected optical signal of the particle/cell
passing through the interrogation point corresponds to its position, speed in the flow stream.
Hence, a slight variation in flow will result in variation of the detected signal.
3.1 Introduction
Capillary driven flow is ideal for disposable, on-site analytical systems, such as point-of-care
devices because of no external energy such as electricity or mechanical forces are required. The
complexity of the system is also reduced compared to other active fluidic systems. However, the
flow characteristics is heavily dependent on the channel geometry and material properties. In a
passive actuation system, system design and power requirements are significantly reduced as no
active components are needed. The interfacial energy, or surface tension, is the dominant driving
force that moves the liquid sample in the microfluidic channel.
This chapter discusses concept and strategies of control fluidic flow on-chip. The first sections
introduce various methodologies used in microfluidic based flow cytometry applications and
reviews the microfluidic system design concept for the proposed cell/particle imaging platform.
This chapter then describes the concept, theory and experimental results on capillary microfluidic
systems, one of two fluidic transport strategies, developed during this thesis.
34
3.2 Flow Control Methodology in Microfabricated Flow Cytometers
The fluidic system in a flow cytometer is used to control the motion of the particles suspended in
a liquid medium, transporting target particles to the interrogation point for optical
characterization. For optimal performance, the particles must be positioned at the center of the
optical beam and only one particle is being illuminated by the laser at any time during operation.
In conventional flow cytometry, hydrodynamic focusing is used to organize the particles and
cells into single file formation. A sheath flow of distilled water is pumped through the instrument
into the flow cell to create the hydrodynamic focusing effect. In portable systems aimed for
applications at point-of-care where microfluidic technologies are typically used, generation of
precise flow control to produce sheath flow is a challenge. In micro fabricated flow cytometer
devices and systems, other approaches were developed.
There are two distinct approaches in creating sheathless particle focusing in a microfluidic based
flow cell in cytometry: field based and flow assisted method. Field based approaches include the
application of external physical forces such as electric, acoustic and optical forces, whereas flow
assisted methods entails micro-channel or micro structures that form physical barriers to move
particles out of their streamlines and into desired focusing pattern.
3.2.1 Acoustic
Similar to electric fields, acoustic waves can generate pressure gradients in a fluid transporting
suspended particles either to the pressure nodes (minimum pressure amplitude) or the antinodes
(maximum pressure amplitude). Particles can be trapped or focused with resonating transducers
generating the acoustic wave field (confocal or planar fields). When a standing wave is generated
in a medium, the acoustic pressure at position x can be described by the following relationship:
∆𝑝(𝑥) = 𝑝𝑜 sin(𝑘𝑥) cos(𝑤𝑡) (3 – 2)
where 𝑝𝑜 is the acoustic pressure amplitude, 𝑘 is the wave number of ultrasonic radiation (𝑘 =
2𝜋/𝜆, λ is the wavelength), x is the distance from the nodal position in the medium, 𝑤 is the
angular frequency, and t is time. An acoustic radiation force on a particle can be expressed as the
following:
35
𝐹𝑎𝑐 = −4
3𝜋𝑅3𝑘𝐸𝑎𝑐𝐴 sin(2𝑘𝑥) (3 – 3)
where 𝑅 is the particle radius, 𝐸𝑎𝑐 is the averaged acoustic energy density, and A is the constant
given by material density, compressibility and the sound velocity in the medium and particle.
When A is positive, the particles move to the nodal position of the acoustic standing wave.112
3.2.2 Electrical
Dielectrophoresis can be used to focus micrometer and even nanometer sized particles in a flow
stream. A non-uniform electric field drives motion of dielectric particles in a certain direction.
The dielectrophoretic force acting on a particle in a non-uniform electric field is represented by
the following expression:
𝐹𝐷𝐸𝑃 = 𝜋𝑎3𝜖𝑚𝑅𝑒[𝑓𝐶𝑀]∇|𝐸|2 (3 – 1)
where 𝑎 is particle radius, 𝜖𝑚 is the permittivity of the suspending medium, 𝑅𝑒[𝑓𝐶𝑀] is the real
part of the Clausius-Mossotti factor, 𝑓𝐶𝑀, and E is the applied non-uniform electric field.113
The direction of the particle movement is determined by the sign of the real part of the Clausius-
Mossotti factor which depends on the permittivity and conductivity of the particle and the
suspending medium, and the frequency of the applied electric field. Particles move either toward
the region of high-electric field strength (positive DEP) or to the minimum field gradient
(negative DEP). 113
An example of using DEP to focus micro-particles is demonstrated by Yu et al.114 The elliptical
channel was fabricated by bonding of two soda-lime glass wafers after chemical etching and
electrode deposition. The electric field gradient was generated in the radial direction from the
electrode pattern and was minimal at the center of the channel. Therefore, particles were directed
towards the center from all directions by way of negative DEP. The etched channel of 50 µm in
depth, 250 µm in width and 100 µm center-to-center distance between two adjacent electrodes
focused micro-beads and human leukemia HL-60 cells to regions 10-15 µm in diameter at 15 V
peak to peak at a frequency of 10 kHz.
36
3.2.3 Micro Structure
Microfluidic channels can be fabricated with accurate dimensions and shapes. Recent advances
in micro/nanofluidics enable detailed understanding of fluid/particle transport which led to the
development of new techniques for self-ordering of biological particles without external forces.
Many researchers investigated micro-structures and their impact on fluid flow in a microfluidic
device to achieve the desired fluidic profile.
Choi et al. described a unique particle ordering principle called hydrophoresis that refers to the
movement of suspended particles under the influence of a microstructure induced pressure
field.115,116 The hydrophoretic ordering principle is governed by anisotropic obstacles, a kind of
the physical barrier. Upon application of a fluid flow into the channel, the anisotropic fluidic
resistance of the V-shaped obstacles generates rotational fluid streams117. Flow streams force
particles to migrate laterally and into the center of the microchannel. The streamlines starting at
the center move upward or downward in the z-direction along with the particles, and their
motions are determined by steric hindrance mechanism. The steric hindrance occurs when the
obstacles prevent rotational flows of large particles. A particle with a diameter that is similar to
the obstacle gap will steer its position toward the center of the z-axis due to the particle-wall
interaction. Therefore, the particle can be focused to the channel center and remain in its focused
position. The authors reported the sheathless focusing of 10 µm and 15 µm polystyrene particles
within the standard deviation of 22 µm and 18 µm in 1 mm wide channels. In addition, the
authors confirmed that the focusing effect was not affected by the flow rate in a range from 2 to 9
µL/min. It was discovered that the hydrophoretic focusing is dependent on particle size. The
smaller particles were found to have large focusing variation.
3.2.4 Micro Flow Cytometers without Focusing
In microfluidics based cytometry system, another approach is non-focusing flow in the detection
region. The target particles/cells will be moving inside a microfluidic device. Instead of being
interrogated one at a time, the particles/cells are probed by laser in a group. This approach
eliminates the complex fluidic controls that a conventional flow cytometer entails, and makes the
optical alignment less critical for the target fluorescent excitation and detection. However, the
optical signal to background ratio is compromised and the forward and side scattering signals are
lost in the measurement, yielding effectively an imaging flow cytometer system. Given the goal
37
of this work being development of a portable and affordable cell analyzer, designing a functional
cell analyzer with simplest system architecture and specification is desired.
3.3 Microfluidics Design Concept
Both passive and active fluidic actuation approaches were investigated in this thesis. In passive
microfluidics, sample transport was driven by the capillary forces generated between the liquid
and micro-channel structures. This approach further simplifies the system design and power
requirements on the overall system. Since there is no particle focusing or sheath flow required,
the main objective of microfluidics system is to transport fluid sample through the detection
region by capillary forces only. The key design considerations of the microfluidic structures in
this approach include uniform fluid flow, metering of sample processed and channel filling
pattern and rate.
In addition, active fluid transport was also studied. A pressure based fluidic actuation system was
developed to control the fluidic motion on-chip. In this approach, a pneumatic interface between
the actuation hardware and microfluidic device was designed. By varying the volume of the
microfluidic device, a pressure difference was created inside the microfluidic channels which
was the dominating force in driving the motion of liquid sample.
Note in a fluidic system, both of these mechanisms, capillary and pressure based, are present and
affecting the flow characteristics. Depending on the application and the design requirements, one
phenomena can be suppressed by tweaking the design parameters to achieve the desired
outcomes. The following sections describe the capillary and pressure based microfluidic
transport systems developed in this thesis. Strategies on how to manipulate, actuate fluid flow, as
well as parameters that can effectively change fluidic flow properties are discussed in this
chapter.
3.4 Capillary Microfluidic Systems
38
The physical laws describing, for example, the forces produced by inertia and interfacial tensions
are valid at any scale, but the relative magnitude of these forces reverses as the volume of the
liquid decreases and the surface/volume ratio increases. Solid-liquid-air interfacial phenomena
can be exploited and harnessed for controlling, guiding, and affecting the transport of liquids.
3.4.1 Governing equations of Fluid Mechanics
The flow of fluid through a control volume can be described by the complete Navier-Stokes
equations. These equations can be derived from the principles of conservation of mass,
momentum and energy. The conservation of mass equation states that at all times (for
incompressible, steady flow) the mass entering the control volume is equal to the mass leaving
the control volume,
𝜕𝑚
𝜕𝑡+ ∇ ∙ (𝜌𝑉) = 0 (3 – 4)
where m denotes the mass and ρ, v denote density and volume respectively. When liquid
penetrates into the capillary tube by capillary action at the gas-air interface, the fluidic motion is
driven completely by the capillary action. This is known as the capillary driven flow. In the last
two decades, extensive studies have addressed capillary force as the primary driving force
instead of electrical or mechanical means.118-121
39
Figure 3-1 - Schematic of liquid plug in a rectangular microfluidic channel. 120
To accurately calculate the capillary force in a microfluidic channel, it is necessary to obtain the
shape of the liquid/air interface. Although there is some work on calculating the interface shape
in rectangular microchannels,122-124 it is difficult to estimate the exact interface shape and
capillary force for microchannel with general cross sectional shape. To simplify the analysis, the
pressure difference at the interface in a rectangular microchannel can be described as the
following by assuming a constant interface curvature and constant contact angle on the channel
inner surfaces, as shown in Figure 3-1.120
∆𝑃 = 𝜎 (1
𝑅𝑤+
1
𝑅ℎ) (3 – 5)
Where 𝜎 is the surface tension, 𝑅𝑤, and 𝑅ℎ are radii of curvature in y- (width) and z- (height)
directions, respectively. Applying the following relations between curvature, channel size, and
contact angle, where
𝑅𝑤 =𝑤
2 cos 𝜃, 𝑅ℎ =
ℎ
2 cos 𝜃 (3 – 6)
The total capillary force, Fc, can be expressed as120
40
𝐹𝑐 = ∆𝑃𝑤ℎ = 2𝜎 cos 𝜃 (1
𝑤+
1
ℎ) 𝑤ℎ = 2𝜎 cos 𝜃 (𝑤 + ℎ) (3 – 7)
This approximation provides an analytical calculation on capillary forces experienced by the
liquid sample in a rectangular microchannel. It can be used to balance the capillary and fluidic
resistance when designing a capillary microfluidic system.
3.4.2 Numerical Modeling
To assist with the microfluidic chip design, a numerical model was constructed to predict the
fluid flow in the capillary system. A computational model for free surface flow that can
accommodate the presence of obstacles in the flow was developed by Simulent and used in this
work.125 This model was developed to provide an analytical model and to aid the design of the
capillary microfluidic devices for the cytometry application.
3.4.2.1 Objective
The objective of this phase of the project is to model the flow of liquid in a simple microchannel
geometry, as shown in Figure 3-2, to determine the feasibility of using the numerical modeling.
The first reservoir will be filled with liquid and due to the capillary forces, the liquid will move
through the microchannel to reach the second reservoir. The simulation can be used to determine
the effect of parameters such as the geometry of the channel (size, shape) and the properties of
the liquid (contact angle, surface tension) on the time required for the liquid to reach from one
side to the other side.
Figure 3-2 – Schematic illustration of two reservoirs and a straight microchannel connecting the two
reservoirs.
41
The reservoirs are 1.58 mm diameter by 1 mm height cylinders and are connected together
through a micro-channel of 15 microns high, 200 microns wide, and 15 mm long.
A single-phase algorithm was used to calculate the time required for liquid to flow from one
reservoir to another. One feature of the micro-channel geometry is that the maximum aspect ratio
is 1,000 to 1. This means for a good resolution, the number of nodes may be excessive. A rough
estimate shows at least that more that 15 million cells are needed to capture the simulation
domain. For this reason, some simplifications must be considered to lower the simulation time at
the same time to capture the capillary effect of the blood movement.
Simulations were run using a software tool developed by Dr. Hamideh Parizi at Simulent. Three
simulations with different mesh cell numbers were performed to find the effect of the cell sizes
on the accuracy of the simulations. The purpose of this simulation was to confirm and validate
the experimental results and observations. Hence only three mesh sizes were chosen in this
study. The results would also demonstrate how efficient the Simulent code was for such
problems and, if needed, what type, or types, of modifications it would require to make it faster
and more efficient.
Since surface tension is the dominant parameter in microfluidics, modelling surface tension and
contact angle play a very important role in all of the applications. A computational model for free
surface flows that can accommodate the presence of obstacles in the flow is developed by
Simulent and adapted for this work. The specific attributes of this numerical model are listed as
follows:
The numerical model uses a software tool for free surface flows and interfaces with no-slip
boundary conditions. Because of the dominance and effectiveness of surface tension force
applied in the code, it is very accurate and efficient for microfluidic analysis and it would be a
powerful tool for designing and prototyping new biochips and components. This mathematical
model has been extensively validated against experiments and has been already tested to model a
moving droplet in a microchannel, as a result of applying electro-capillary forces 125.
42
3.4.2.2 Simulation Results
In order to capture the capillary flow in the microchannel, we need to have enough number of
cells in the smallest dimension of the channel, which is 15 microns. On the other hand, the height
of liquid column in the reservoir is at least 500 microns and the length of the channel is 15 mm.
To reduce to total number of cells in the whole calculation domain, it was decided that only part
of the reservoir is considered in the calculation domain and it is assumed that the height of liquid
remains constant. (Please refer to Figure 3-2 for details). Also, since the width of the channel (w)
is much larger that its height (h) the two boundaries at the “y” direction were considered to be
symmetric, with minimum number of cells. In addition since the flow velocity is very low and in
order to reduce the number of cells further, the simulations were performed only for one
millimeter of the microchannel,( L = 1 mm).
Figure 3-3 – Schematic of microfluidic setup in the numerical model.
43
Figure 3-4 – Snapshots of flow in the microchannel at different times. The contact angle for this simulation is
set at 70 degrees. The length of the channel shown in this figure is 1 mm.
These simulations were used to confirm and validate the experimental observations and results.
As a result, only three different simulations were performed. To see the effect of the mesh size
on the simulation results, two different cell sizes, i.e., 3.5 and 2.5 microns were used. The total
number of mesh cells in each case was 264,552 and 681,952, respectively.
To characterize the effect of the contact angle, the simulations were performed at two contact
angles of 44.3 degrees and 70 degrees, respectively. These two contact angles were chosen to
model the material property of acrylic substrate such as PMMA. In Figure 3-4, snapshots of
simulation results at different time for a contact angle of 70 degrees are shown. In the last frame,
44
the maximum distance that the liquid has been able to travel is shown, which is less than one
millimeter.
Due to the limitation of computer memory, the length of channel must be kept short. In Figure
3-5, the simulation results for the above three case are shown. This Figure shows the variation of
the flow velocity with time. As expected the velocity rises to its maximum at the very beginning
of the liquid movement and then drops as the liquid flows more into the channel. Different
curves represent various contact angles and simulation conditions (small or large mesh size). The
solid lines were plotted from discrete outputs from numerical simulation while the dashed lines
were logarithmic fit of the simulated discrete data points.
Figure 3-5 – Calculated axial velocity of liquid in the microchannel with respect to time.
45
Clearly, the effect of mesh cell size in this range is not very significant. However, the effect of
the contact angle is more remarkable. At contact angle of 70 degrees, the flow has stopped in the
channel after about 0.04 second. This has been also shown from analytical investigation.
Figure 3-6 – Axial velocity of liquid in the microchannel with respect to time. The cell size is 1.5µm (10 cells
per height of channel). The contact angle is 30 degrees. All other conditions are the same as those in Figure
3-5.
One important observation is the oscillation of the flow velocity at the very beginning of the
capillary flow. To investigate the effect of the cell size and to eliminate the numerical error,
another simulation was performed with cell size of 1.5 microns (10 cells within the height of
channel). The preliminary results for a contact angle of 30 degrees are shown in Figure 3-6.
There are several important issues reported in the literatures that will affect the flow of liquid
in the microchannel:
46
The aspect ratio of the channel, height to width ratio, has great effect on the behaviour
of the liquid. In general, the maximum center line velocity decreases with decreasing
channel aspect ratio 126.
Contact angle has a great effect on the velocity of the liquid in the microchannel. The
greater the contact angle (less wettability), the lower the velocity. In present case in
which there is only 500-1,000 microns of liquid column in the reservoir, the capillary
forces are more remarkable than the pressure force. This makes the effect of contact
angle more important.
For the above reason, the effect of gravity may also be neglected. Due to very low liquid
velocity, the use of the equilibrium contact angle instead of the dynamic contact angle
must be investigated in more details.
3.4.3 Fluidic Resistance Calculation
In a capillary system, the fluidic resistance is dependent on two factors: cross sectional area of
the microchannel and the fluid viscosity. Extensive research efforts have been put into providing
analytical models to predict fluidic resistance of various types of microfluidic channels.
The flow rate Q of a liquid plug in a capillary fluidic system is determined by the wettability of
the microchannels, the viscosity of the liquid, the total flow resistance and the capillary pressure
in the capillary system127:
𝑄 =1
𝜂
Δ𝑃
𝑅𝑓=
Δ𝑃
𝑅 (3 – 8)
Where 𝜂 is the viscosity of the liquid, Δ𝑃 is the difference in pressure inside and in front of the
liquid, and 𝑅𝑓 is the friction factor and 𝑅 is the hydraulic fluidic resistance of the system.
47
Figure 3-7 – An example of a capillary microfluidic device. (a) Top view of a capillary microfluidic device
with a circular sample inlet, serpentine microchannel and a tapering structure to enhance capillary force. (b)
Side view of the rectangular microfluidic channel.
In a capillary microfluidic system shown in Figure 3-7, the capillary pressure Pc of a liquid-air
meniscus in a channel with rectangular cross-sectional profile can be expressed as the
following127:
𝑃𝑐 = −𝛾 (cos 𝛼𝑏+cos 𝛼𝑡
𝑎+
cos 𝛼𝑙+cos 𝛼𝑟
𝑏) (3 – 9)
Where 𝛾 is the surface tension of the liquid, 𝛼𝑡,𝑏,𝑙,𝑟 are the contact angles of the liquid on the top,
bottom, left and right side wall, respectively, a and b are the depth and width of the
microchannel.
The flow resistance of the rectangular microchannel shown in Figure 3-7 can be expressed with a
Fourier Series and can be approximated by a linear term:3
48
𝑅𝐹 = [1
12(1 +
5
6
𝑎
𝑏)
𝑎𝑏𝑅𝐻2
𝐿]
−1
(3 – 10)
When the condition of a < b is met. In the above equation, L is the length of the microchannel,
and 𝑅𝐻 denotes the hydraulic radius of the microchannel, which can be further expressed as the
following:
𝑅𝐻 =2𝐴
𝑃=
𝑎𝑏
𝑎+𝑏 (3 – 11)
Where P being the perimeter and A is the area of the cross section of the microfluidic channel.
Similar to fluid dynamics in macroscopic application, the flow in a microchannel can be
approximated by capillary pressure divided by the fluidic resistance, which continually increases
as the channel is being filled under the influence of capillary force. The filling of microchannels
is determined by the surface tension of the liquid, and the chemistry and geometry of the micro-
structures in a microfluidic system.
3.4.4 Microchannel Design
A typical lymphocyte has a size of 6 to 8 µm in diameter. To improve sensitivity, a mono layer
of cells or particles distribution is desired during detection. On the other hand, according to the
equations described in Section 3.4.1, the fluidic resistance of the microfluidic system is strongly
dependent on the channel cross section. A shallow channel, on the order of 10 µm, with a width
of 600 µm and 1 mm in length, has fluidic resistance of 2.04x1015 m-3. If the channel is 100 µm
deep while all other parameters remain the same, the fluidic resistance becomes 2.39x1013 m-3. A
10 fold increase in channel depth resulted 100 fold decrease in fluidic resistance. Small channel
depth of less than 10 µm also increases the likelihood of microchannel clogging during
flow2,4,128,129 as blood samples often contain lipids and other molecules in addition to the
cells.6,130,131
For the dynamic imaging approach proposed and developed in this work, the microfluidic
channels must match with the active area of the imaging sensor in order to capture all the
cells/particles in a sample. To ensure the entire cross-section of the microchannel at detection
completely lies within the image sensor field of view, the microchannels must have a width of
500 – 900 µm with a depth of ~ 20 – 50 µm at detection region.
49
To characterize and experimentally investigate flow properties of the capillary microfluidic
systems, testing structure shown in Figure 3-3 was designed and fabricated. This prototype
enables further understanding of the impact of fluidic channel geometry on the capillary flow.
3.4.5 Capillary Microfluidic Device Fabrication
The microfluidic chips were fabricated using standard photolithographic technique and the
wafers were packaged using a laminar press process, shown graphically in Figure 3-8. The base
layer, made of plastic acrylic, has fluidic structures defined in SU-8 negative photoresist
(Microchem Corp.) The microfluidic channels were patterned using a standard photolithography
technique – first an underlayer is deposited (spin coated then dried) and fully cured, then a
second layer is deposited in the same way and patterned by exposing through a photo mask.
Exposed samples were baked and developed to form the desired features. The lid or capping
layer, also made of plastic acrylic, has a partially cured SU-8 photoresist layer deposited with
mechanically drilled through holes to form the inlets and outlets. The base and the capping layers
are then assembled in registration in bonding jogs and heated under mechanical load in a
laminating press and held for a period of time to form the bond. The fabrication and packaging
were completed by Epigem, UK (http://www.epigem.co.uk).
50
Figure 3-8 - Graphical illustration of photolithography process used by Epigem to produce capillary
microfluidic devices.
3.4.6 Capillary Microfluidic Device Characterization
The microfluidic device was characterized in two approaches: fluidic flow characterization and
cell/particle counting characterization.
3.4.6.1 Fluid Transport
The microfluidic devices were designed based on capillary fluidic properties described
previously. Once fabricated using the process described in Section 3.4.5, the chips were first
characterized to determine their fluidic flow properties. The fluidic speed characterization chip
had a serpentine design shown in Figure 3-9.
To test the fluidic flow in the capillary system, colored food dye samples mixed with distilled
water were used to enhance the visualization. A blue dye was used in this case to characterize the
fluidic motion inside the microfluidic channels.
51
Figure 3-9 – Serpentine microfluidic structure for fluidic flow characterization. The channels were designed
with the same cross-sectional dimensions as the detection channels in the cell/particle counting device.
Channel length of each pass is 20 mm.
In a typical capillary microfluidic device, the flow resistance is dependent on the cross-sectional
area, the viscosity of the fluid moving inside the channel and the length of the channel. The
larger the cross-sectional area, the smaller the fluidic resistance seen by the liquid. The fluidic
channel has a cross sectional area of 800 x 20 µm and each pass is 20 mm long. The width of the
channel confines with the width of the optical detector, whereas the depth of the channel is
limited by the target cell dimensions. Ideally during detection a uniform, a single layer
distribution of cells inside the microfluidic device is desired. This eliminates potential errors that
may result in cell detection and tracking due to cells stacking and also minimizes the background
noise of blood plasma and other cell populations. On the other hand, if the channel depth is too
narrow, the cross-sectional area is reduced and fluidic resistance increases accordingly, thus
reducing the fluidic flow rate of the entire system.
52
Figure 3-10 – Experimental results on filling characterization of the microfluidic device corresponding to the
serpentine microfluidic structure shown in Fig. 4. (a) Observed filling time of each linear section of the
serpentine microfluidic device. (b) Fluid flow speed and channel filling distance characterization results based
on the filling time and the distance for each pass in the serpentine microfluidic device. An inversely
proportional relationship between flow speed and the channel length can be seen.
Figure 3-10 demonstrates the flow properties of the colored dye when the sample is introduced in
the PMMA device. As expected in the fluidic simulation, the flow speed decreases as the liquid
propagates inside the microchannel when only exposed to capillary forces.
From the microfluidic channel dimensions and filling time of each linear section of the
serpentine channel, the fluidic velocity can be calculated along the entire microfluidic device. It
is known that the flow rate of a liquid in a capillary fluidic system is determined by the viscosity
of the liquid, the total flow resistance and the capillary pressure of the capillary pump127,132.
Since the serpentine channel has the same capillary pressure throughout the whole channel due to
constant surface wettability and cross sectional area, the flow rate of a liquid in the channel will
53
be inversely proportional to the distance between entrance and the liquid filling front namely the
total flow resistance. The experimental results on filing characterization of the device are shown
in Figure 3-10. Because the particles/cells are imaged dynamically inside the microchannel and
fluidic motion is driven purely by capillary force, this investigation enables the validation of
capillary fluid system at experimental conditions. It also guides the design of the cell
enumeration device and the selection of detection window along the channel for the final design
so that a desired particle moving speed can be achieved during image capture; hence achieving
the required signal to background ratio of the captured images.
The result of this fluid flow characterization enabled the refinement of the capillary microfluidic
devices for cell counting. From this exercise, the length of the leading microfluidic channel can
be determined to yield the optimal fluid flow speed at the interrogation region. The PMMA
microfluidic chip was then used to characterize the fluidic flow.
3.4.6.2 Volume Metering
For many applications such as cell or particle enumeration, sample volume that has been
analyzed must be measured in order to produce an accurate volumetric concentration result. As a
result, the microfluidic cartridge must be designed to be able to calculate the exact volume of
sample that was imaged or processed.
Figure 3-11 - Capillary microfluidic chip design layout. This design relies on capillary forces to manipulate
sample flow. The device has a volumetric design to allow a sample volume of 2 µL to be processed.
In the passive microfluidic device designed in this thesis, the sample chamber is designed to have
a pre-determined overall volume. Figure 3-11 illustrates this design principle. The liquid sample
flow past the detection point will be imaged and analyzed. Hence, the volume downstream of the
detection point must be precisely set to a known value. Using this information, the cell or particle
concentration can be calculated based on the total number of positive events captured or
54
detected. In the case of CD4 T cell analysis, to produce a statistically accurate result, a minimum
of 1000 events must be acquired133,134. Since the CD4 T cell concentration varies from 500 to
1500 per microliter, a sample volume of 2 µL must be analyzed in order to yield an accurate cell
counting result. The microfluidic device shown in Figure 3-11 has a chamber size of exactly 2
µL so that when the chamber is completely filled, the system has processed 2 µL of sample. A
capillary stop valve135 was designed at the end of exiting microchannel (shown in Figure 3-11).
Once the chamber if filled and the liquid front reaches the capillary valve, the flow will stop, due
to capillary force and surface tension, which signals the end of the entire analysis.
3.5 Conclusion
In this chapter, we investigated capillary-driven fluidic transport mechanism. Capillary driven
flow is a passive technique where no external forces are required to manipulate the sample fluids
on-chip. This approach is strongly dependent on material property, more specifically surface
contact angle with the liquid. The smaller the contact angle, the higher wettability of the material
which translate to larger capillary forces. In order to optimize or enhance the fluidic flow on-
chip, surface treatment such as oxygen plasma is often required.
A detailed analysis on capillary-driven fluid flow was conducted in this work. The flow
characteristics were investigated and strategies to control the flow speed were presented in this
chapter. The on-chip fluidic management is a key component in the disposable cartridge that
completes the cell enumeration analysis. The knowledge gained in this part will be utilized in
developing the point-of-care cell analysis device as described in the subsequent chapters.
55
Chapter 4 Active Microfluidic System
In this chapter, the active microfluidic system is studied. A volumetric pressure based pneumatic
actuation concept is designed and developed in this part of the thesis. This approach is more
suitable for complex on-chip operations such as mixing, incubation and other similar chemical
reactions which require precise control of fluid flow.
4.1 Introduction
Capillary microfluidic devices offer the advantage of simple design since the fluidic actuation is
passive and no active components are required. The elimination of external energy sources
allows for reduction in the overall system complexity. However, the disadvantage of capillary
pumping in a microfluidic system is the strong dependency on surface tension between the liquid
and the channel walls. The surface tension is a function of the contact angle between the liquid
and material. Hence the hydrophobicity, also known as wettability, of the material is critical in
determining the fluidic transport inside the microchannel. For most applications, a hydrophilic
surface is desired to enhance the fluidic flow. To achieve that, it often requires surface treatment
of the material. This surface treatment could be a plasma oxygen treatment to increase the water
contact angle and improve the wettability of the surface. However, for mass production of
microfluidic devices and cartridges, this process is expensive and it is difficult to achieve
consistent fluidic transport. In addition, more complex operations such as mixing and incubation
require more consistent fluidic transport properties. Capillary force alone may not be sufficient
for these types of functions.
To overcome this drawback with capillary microfluidic systems, an alternate fluidic transport
approach was also investigated in this work. The alternative uses a volumetric pressure based
actuation approach. In this thesis, a soft elastomer – bellows – based pneumatic interface was
designed to generate the pressure change for fluidic transport.
56
The first section of this chapter describes the design principle of the active microfluidic system.
Microfluidic devices were fabricated and tested. Additional operations and functions for sample
preparation was also developed and characterized in this chapter, followed by a conclusion.
4.2 Bellows Actuation System
In collaboration with thinXXS Microtechnology AG (www.thinxxs.com), an alternative fluidic
actuation mechanism was developed to mechanically control volume change in the fluidic
system to actuate sample flow. Figure 4-1 illustrates the basic concept of the pumping
mechanism. In this approach, an elastic film, or bellows that contains certain amount of pumping
volume integrated with the cartridge. During actuation, the elastic film of the bellows deflects
under external mechanical force, leading to a volume change which subsequently changes the
pressure in the microfluidic channel for fluidic transport.
Figure 4-1 – Bellows transport concept. The soft elastomer is depressed under external force F. The deflection
of the elastomer induces a pressure change inside the chamber and actuates the fluidic motion inside the
microchannel that is connected to the chamber.
57
4.3 Bellows Transport System Design
The bellows transport system has two modes of operation: forward pumping which pushes the
liquid move the sample forward and reverse suction which retracts the sample backward.
To design the appropriate bellows transport system, a key parameter must be specified to ensure
proper control of fluidic motion in the microfluidic channel, namely: the total sample volume
that the bellows can transport.
Figure 4-2 - Bellows slide concept. The entire device consists of a soft elastomer semi-sphere, bonded to a
plastic fluidic chip. The fluidic channels are connected to the bellows. When bellows is depressed, the
reduction in volume inside the bellows increases the pressure inside the microchannel which subsequently
pushes the liquid sample forward.
In order to estimate the internal bellows volume needed, the following calculation steps are
applied:
1. Calculate the back-end gas spring pressure required to transport the sample trough the mixing
and detection zone depending on the size of the gas spring.
2. Calculate the front-end volume change of the air in the bellows vs. the internal bellows
volume for the pressure range selected.
58
3. Calculate the motion of a stepper motor actuation deflecting the pneumatic bellows depending
on the bellows volume and the diameter of the actuator in order to get an indication for the
pulsation of the sample flow through the detection zone.
4.3.1 Bellows Actuation Volume Calculation
The first design parameter is the bellows volume. The deflection of bellows membrane
compresses the air inside the bellows chamber, hence increasing the pressure inside. The
increased pressure will subsequently push the liquid plug to move forward in the direction shown
in Figure 4-1. As a result, the relationship between bellows membrane deflection and pressure
change inside the bellows chamber must be established. The following calculation establishes the
change in bellows chamber volume as a function of bellows deflection.
Figure 4-3 – A graphical illustration of the coordinate system and variables used in bellows volume change
calculation.
The volume of the semi sphere shown in Figure 4-3 can be calculated from the following
equation:
59
𝑉 = ∫ 𝜋𝑥2𝑟
0𝑑𝑧 (4 – 1)
Figure 4-4 – Volume change induced by bellows deflection. Bellows deflection is actuated from the top of the
semi-sphere in this diagram and the amount of deflection is denoted by d.
When the bellows is deflected, the integral is changed to the following equation to represent the
volume inside the bellows after deflection. (Also illustrated in Figure 4-4).
∫ 𝜋𝜌2𝑟 cos 𝜃
0𝑑𝑧 (4 – 2)
Applying the Pythagorean Theorem,
𝜌2 + 𝑧2 = 𝑟2
Or 𝜌2 = 𝑟2 − 𝑧2 (4 – 3)
Substitute Equation (4 – 3) in Equation (4 – 2), the volume of the deflected bellows can be
calculated using the following:
𝑉 =1
2∫ 𝜋(𝑟2 − 𝑧2)𝑑𝑧
𝑟 cos 𝜃
−𝑟 cos 𝜃 (4 – 4)
60
Integrating gives:
𝑉 =1
2𝜋 [𝑟2𝑧 −
𝑧
3
3
]−𝑟 cos 𝜃
𝑟 cos 𝜃
Resulting in the following expression for the volume:
𝑉 = 𝜋𝑟3 (cos 𝜃 −1
3cos3 𝜃) (4 – 5)
Where:
𝜃 = cos−1 (𝑟−𝑑
𝑟) (4 – 6)
Where V is the volume of air inside the bellows, r is the radius of the bellows and d is the amount
of deflection applied to the bellows. Results of this analysis are illustrated in Figure 4-5 below
where the volume change incurred as a result of bellows deflection is plotted.
Figure 4-5 - Volume change of the bellows as a function of bellows deflection.
0
200
400
600
800
1000
1200
0 1 2 3 4 5 6
Vo
lum
e ch
ange
(µ
L)
Bellows deflection d (mm)
Volume Change as a function of Bellows Deflection
r = 5 mm
r = 10 mm
r = 15 mm
61
From this analysis, it is clear under the same bellows deflection displacement, the larger bellows
will result in a larger volume change inside the microfluidic system, hence more samples
transported through the microchannel. On the other hand, since it produces smaller volume
change, a smaller bellows would have more precise control of fluidic motion inside the
microchannel per unit deflection of the bellows.
4.4 Bellows Actuation System and Microfluidics Design
The bellows slide, fabricated using injection molding process by thinXXS Microtechnology AG,
comprised two parts: a soft elastomer membrane and a hard-plastic substrate. The schematic of
this mechanical structure is shown in Figure 4-2.
The bellows deflection is driven by linear stepper motor shown in Figure 4-6. Linear stepper
motors have a compact foot print with low power consumption. They also offer very fine and
well controlled displacement, as small as 2 µm per step, which makes it ideal for microfluidic
application. An actuation pin can be installed at the tip of the shaft shown in the Figure 4-13 on
Page 73 to deflect the bellows membrane to drive the fluidic motion inside the microchannel.
Figure 4-6 - Linear stepper motor from Haydon Kerk. (www.haydonkerkexpress.com)
According to the analysis completed, to generate the same pressure change, a smaller bellows
size requires a smaller bellows deflection. In addition, a smaller actuation pin diameter requires a
bigger displacement to deflect the spherical bellows. The size and height of the deflection defines
the motor motion and therefore the number of motor steps.
62
Figure 4-7 - Microfluidic chip layout of the bellows slide. The bellows slide was fabricated using injection
molding and it was used to test the fluidic actuation and on-chip flow control.
For the selected range of bellows and actuation pins, the number of 2 µm motor steps per µl
ranges from 15 to 30 mm (this value would double in case of 1 µm motor resolution). Assuming
the flow rate through the microfluidic system is set to a target of 0.2 µl/min, this results into 5
steps/min (10 steps in case of 1 µm displacement per step), which is assessed to be critical in
terms of pulsation. In general, the design has to reflect a compromise between the tendency to
keep all elements as small as possible (due to cost) and allowing enough motor steps to transport
the sample through the detection zone at the required flow rate. Figure 4-7 is a picture of the
microfluidic chip layout of the bellows slide that was used to test and evaluate the on-chip fluidic
management and actuation.
4.4.1 Material Characterization
The active microfluidic devices were fabricated using injection molding process, as it is the
preferred method in mass production for plastic medical consumables. Since the fluidic
component of the cartridge is comprised of two parts: microfluidic channels fabricated in a
plastic substrate and a thin covering film. Both material must be characterized. Two most
commonly used thermo plastic material that can be used include polymethyl methacrylate
(PMMA) and cyclic olefin copolymer (COC).
63
PMMA is a transparent thermoplastic often used in sheet form. It is a clear, colorless polymer
with high mechanical strength, high Young’s modulus and low elongation at break. It has
excellent optical quality and is shatter-resistant. It is one of the hardest thermoplastics and is also
highly scratch resistant. It exhibits low moisture and water absorbing capacity. COC is an
amorphous polymer material that is widely used in medical device industry. It has excellent
optical transmission, high purity, and resistant to moisture. It is also resistant to chemicals as it is
becoming more and more important in microfluidics manufacturing.136,137
Since optical detection of target cells is used by measuring their fluorescence intensities, the
optical properties of the material used to fabricate the chip, specifically auto fluorescence,
becomes a critical evaluation parameter. A range of different PMMA and COC polymer material
were characterized by measuring their auto fluorescence in the 400 nm to 750 nm range to
determine the best material choice.
64
Figure 4-8 - Autofluoscence levels of different PMMA and COC material under different excitation and
emission optical setup.
The material label represents different common injection molding plastic material that was
supplied by the injection molding foundry thinXXS Microtechnology (http://www.thinxxs.com/)
in Germany. Table 3 summarizes the name and physical characteristics of each material
investigated in this work. For each substrate material, a 3 inch by 1 inch slide was provided.
Auto fluorescence measurement was made on a standard fluorescence microscope (Olympus
BX51). Excitation filters were chosen to match potential fluorescence dye that may be used in
0
500
1000
1500
2000
2500
3000
3500
dark field COC BlackPMMA
COP PC PP OpaquePC
PMMA Black COC Black COP OpaquePC
Au
tofl
uo
resc
ence
inte
nsi
ty
Material type
Plastic material autofluorescence measured under different excitation wavelength
ex:387/11, em:440/40 ex:485/20, em:525/30
ex:560/25, ex:607/36 ex:650/13, em:684/24
65
the detection system. Each data point was captured under 20 ms exposure time using a
Photometrics CoolSNAP HQ2 monochrome CCD camera (http://www.photometrics.com/) under
non-cooled operation mode. Dark field was a reference data point measured when no material
was present while all other excitation and emission setup were the same.
Table 3 - Materials used in autofluorescence characterization.
Label name Material full name Physical characteristics
COC Cyclic olefin copolymer Transparent polymer
PMMA Polymethyl methacrylate Transparent polymer
COP Cyclic olefin polymer Transparent polymer
PC Polycarbonate Transparent polymer
PP Polypropylene Transparent polymer
Opaque PC Polycarbonate White polycarbonate
Black PMMA Polymethyl methacrylate Black PMMA, non-transparent
Black COC Cyclic olefin copolymer Black COC, non-transparent
Black COP Cyclic olefin polymer Black COP, non-transparent
Opaque PP Polypropylene White PP, non-transparent
As expected, the shorter the excitation wavelength, the stronger the auto fluorescence. When
excited with longer wavelengths, the auto fluorescence of all these plastic material reduced
significantly. Of the material tested, black PMMA (see Figure 4-8) exhibited the least auto
fluorescence levels across all excitation wavelengths. Another transparent variant of PMMA,
also has minimal auto fluorescence level even when excited by 387 nm violet light source. The
experimental results suggest the 3 black materials have the lowest level of auto fluorescence (the
66
level of auto fluorescence are COP < PMMA < PC as shown in Figure 4-8). Based on the
current results and considering material property compatibility, black PMMA would be the ideal
material choice for the microfluidic device.
4.4.2 Microfluidic Channel Design
The microchannels used for fluid transport have a width of 800 microns and a depth of 400
microns. The width of the microchannel at detection zone was designed to be 200 microns to
match the size of the optical sensor.
Based on the analysis completed, a microfluidic device using bellows actuation was designed
that can meet the sample transport requirement for CD4 T cell enumeration. Since typical healthy
individual’s CD4 T cell concentration ranges from 700 to 1000 per microliter, according to
Poisson statistics133,138, the counting outcome is accurate only if 1000 events are encountered and
detected. That means a minimum of 2 µL sample must be analyzed, or processed. Typically, in
conventional flow cytometry, 100 µL of sample is prepared and analyzed. The larger volume can
make the sample preparation process less prone to statistical variations. On the other hand,
microfluidic systems are not well-suited for applications that require processing of large sample
volumes. To reduce the statistical uncertainty on the cell enumeration result, a larger volume (10
µL) is used for sample preparation while only a sub-set of the prepared sample (2 µL) is used for
image analysis to produce the final cell counting result.
4.4.3 Bellows Slide Fabrication
Microfluidic prototypes were fabricated using injection molding by thinXXS. Figure 4-9 is a
picture of the fabricated bellows slide using PMMA material as characterized in Section 4.4.1.
67
Figure 4-9 - A photograph of the injection molded microfluidic bellows chip using the PMMA material
selected in previous section. The bellows chip is 1 inch wide and 3 inches long.
Injection molding is the most commonly used manufacturing process for producing parts by
injecting liquid material into a mold and allowing it to set. A wide variety of products are
manufactured using injection molding, which vary greatly in their size, complexity, and
application. Material for the part is fed into a heated barrel, mixed and forced into a molding
cavity. Once cooled, the material will harden and conforms to the shape of the cavity. 139-141
Injection molding is used to produce thin-walled plastic parts for a wide variety of applications,
including electronics, biomedical, automotive and industrial. Injection molding process is widely
used to manufacture parts and components with a variety of sizes from microscopic devices to
entire body panels of cars.
Parts to be injection molded must be very carefully designed to facilitate the molding process;
the material used for the part, the desired shape and features of the part, the material of the mold,
and the properties of the molding machine must all be taken into account. The versatility of
injection molding is facilitated by this breadth of design considerations and possibilities. It is the
most common modern method of manufacturing parts; it is ideal for producing high volumes of
the same object.142
68
Adding to the capabilities of micro-molding is the ability to mold two different materials on the
same part at the same time. The two different thermoplastic resins are injected in synchronization
into the molding cavity so that it requires only one mold cycle.
4.4.3.1 Micro injection molding for microfluidics
To produce very small components, such as microfluidic devices where the critical features are
on the order of tens of micrometers, the injection molding process requires maximum possible
accuracy and precision. From the material and machine to the mold, everything must be
streamlined to this objective.
Micro-molding is defined as a very unique Injection Molding process requiring specialized
molding machine capable of delivering high injection speed, high injection pressure, precise shot
control, uniform melt temperature and ultra-fine resolution using servo-electric drives and
sophisticated controls.
4.4.4 Characterization
A bellows slide was designed according to the design principles described in previous sections.
The microfluidic device was fabricated using injection molding in PMMA to minimize auto
fluorescence. A thin PMMA film was used to cover the injection molded microfluidic channel
network. The bellows had a size of 15 mm diameter, which is sufficient to transport 50 µL of
sample on-chip.
4.4.4.1 Experimental Setup
A linear stepper motor was used to deflect the bellows membrane. A metal pin was placed at the
end of stepper motor shaft to push the bellows membrane. The experiment was executed using
the setup according to schematics shown in Figure 4-10.
69
Figure 4-10 - Schematics of the stepper motor setup used to test and characterize on-chip fluidic actuation
4.4.4.1.1 Electronics
A universal motion controller (TMC1110) from Trinamic (www.trinamic.com) was used to
control the motion of the stepper motor. Figure 4-11 on Page 70 is a block diagram of the
electronics setup.
70
Figure 4-11 - Fluidic control eletronics setup block diagram. This setup was used to experimentally
characterize and test the fluidic actuation using bellows concept.
A LabVIEW program was developed to interface the motion controller via USB. All hardware
was integrated in a mechanical structure to obtain a compact format as shown in Figure 4-13. A
screenshot of the LabVIEW program is shown in Figure 4-12. The motion controller can drive
the stepper motor in 256 microsteps per full step. This enabled precise control of bellows
deflection. The stepper motor can drive the shaft to move at a wide range of speeds from 0.16
µm/second to 0.341 mm/second. The wide range of speeds allows the fluidic actuation for
different application requirements. The motor can also move in both forward and backward
direction to drive the liquid sample back and forth inside the microfluidic channel for mixing and
incubation purposes, as described in Section 4.2 and Section 4.3.
71
Figure 4-12 - LabView program interface. This tool was used to control the motion of the stepper motor for
on-chip fluidic actuation.
The following list is a detailed description of the functionality of the LabVIEW program on how
to control the stepper motor.
1. Communication port interfacing the motion controller circuit board
2. Electronic pulse initialization where the program will establish the initial configuration of
the electronic and stepper motor system.
3. Stopping the stepper motor.
4. This function drives the stepper motor forward until it touches the bellows. Then the
motor steps. The stepper motor moves at a pre-defined speed.
72
5. This function drives the stepper motor backward until it reaches the original or home
position. The motion speed is pre-defined.
6. Drives the stepper motor forward
7. Drives the stepper motor backward.
8. Specify a given amount of stepper motor travel distance. The distance is defined in
micrometers.
9. Speed of the stepper motor movement. Default value is 200, which corresponds to a
linear speed of 2mm/min.
10. This function specifies direction of motion: forward or backward.
11. Motion starts once this button is pressed. The button returns to the original unpressed
state once the pre-defined displacement is reached.
4.4.4.1.2 Mechanics
Figure 4-15 is a picture that illustrates the mechanical setup of the bellows actuation and
controller. The linear actuator had a travel distance of 9 mm. According to calculation carried out
in Section 4.3.1 on Page 58, for bellows with a diameter of 15 mm, 3 mm deflection of bellows
membrane would induce approximately 50 µL volume change, while at full 9 mm motor travel it
can transport up to 200 µL of sample inside the microfluidic channel.
.
73
Figure 4-13 - A picture showing the mechanical setup of the bellows actuation.
4.4.4.1.3 Testing Slide
To test the fluidic actuation and detection, a separate testing slide was designed and fabricated.
The testing slide included a sample inlet, an outlet, and a detection zone for optical interrogation.
The design of the device is shown in Figure 4-7 on page 62.
74
Figure 4-14 - Schematics of microfluidic connections used in on-chip fluidic actuation testing and
characterization.
During fluidic transport testing, samples were introduced at the inlet. A flexible, plastic tube was
used to connect the bellows slide with the testing slide via fluidic connector. Figure 4-14 is a
schematic of the experimental setup of the microfluidic connections while Figure 4-15 are
pictures of the experiment setup.
75
Figure 4-15 - Pictures of the bellows actuation experimental setup. (a) Side view of the motion control board,
stepper motor, bellows slide and re-suspension slide. (b) Top view of the experimental setup. The engineered
blood sample was introduced into the re-suspension slide and actuated back and forth inside the device.
4.4.4.1.4 Fluidic Transport Observations
Using experimental setup described, the fluidic transport properties of the bellows actuated
microfluidic system were characterized. The liquid test sample used was colored food dye mixed
with water. The following experimental observations were made:
The bellows slide was initially actuated at a linear speed of 16.7 µm/second. After the liquid
sample breaks into the microchannel, the speed was reduced to a linear speed of 0.167
µm/second. The narrow taper in the testing slide means a large increase in the fluidic resistance
in the system, as resistance is proportional to the cross-sectional dimension of the microchannel
as described in Section 3.4.3. Once the steady state is reached, the flow rate of liquid sample was
very stable at 0.615 (±0.016) μL/min. Figure 4-16 is a plot of the speed characterization on
fluidic transport measurement.
76
The narrow detection region means large fluidic resistance. To overcome this resistance, there is
a pressure built-up inside the channel before liquid breaks into the detection channel. This means
there will be wasted or dead volume of sample during analysis. However, the high resistance is
desirable as it results more precise fluidic speed control. The large fluidic resistance makes the
system less responsive to the stepper motor actuation, as illustrated by the 200 seconds settling
time shown in Figure 4-16, resulting more precise fluidic actuation during analysis.
In the second part of the chip characterization, Immuno Trol, a standard flow cytometry control
sample, was used to test the fluidic actuation. The Immuno Trol sample has the same physical
property, such as particle density, viscosity, cell distribution as a real blood sample. The results
from this analysis will reveal how blood sample would behave inside the microchannels during
analysis.
Figure 4-16 on Page 76 is a graph on speed measurements completed with beads in immunotrol
blood when pumping the blood through the detection channel and then stop once the sample has
passed the detection zone, the fluid linear speed stabilized around the range of 0.2-0.8 mm/s after
the first 200 seconds.
Figure 4-16 - Fluidic linear flow speed measured as a function of time. Graph (a) is obtained with a bead
sample only while (b-d) were obtained with a sample of beads mixed with Immunotrols. Three different
77
pumping conditions were tested in this experiment: the stepper motor pumping distance (2 µm, 10 µm, and 30
µm at a speed of 50).
Figure 4-16 showed a graph on speed measurements done with fluorescently labeled beads in an
Immunotrol blood sample. In this experiment, the fluorescent beads were mixed with the blood
sample and their individual speeds were measured using image analysis techniques. The x, y
positions of the particles were recorded in each frame. Since the frame rate is fixed, by
calculating the spatial distance between two subsequent frames, the linear speed of particle in the
sample plug can be deduced. After the sample breaks into the detection channel, the sample was
continuously pumped at a very slow speed. (Stepper motor was moving at a speed of 0.167
µm/s.) By actively control the blood, the settling time for fluidics to stabilize can be shortened to
less than 100 seconds.
Figure 4-17 – Fluidic linear speed plotted as a function of time in the detection microchannel using bellows
actuation. The measurement was made on the resuspension slide described earlier in this work.
0.00
2.00
4.00
6.00
8.00
10.00
12.00
0.0 100.0 200.0 300.0 400.0 500.0 600.0 700.0 800.0 900.0 1000.0
Flu
idic
lin
ear
spee
d (
mm
/s)
Time lapsed from sample entrance (seconds)
Fluidic flow speed measured in the microfluidic device using bellows actuation methodology
0415 bead blood flow rate20um@2000 followed by 4um@2 permin
0417 bead blood flow rate
78
The same fluidic actuation experiment was also performed on wider microchannels with width
of 700 µm). Wider channels had less pressure build-up before the liquid entered the detection
channel area as shown in Figure 4-18. And the flow was easier to control and more responsive to
the operation of the bellows deflection. By repeatedly pushing the bellows by 2 µm at a pumping
speed of 0.167 µm/second, it is feasible to control the flow speed to be within 0.3-0.5 mm/second
range. The particle speed within the flow slowed down to the 0.3-0.5 mm/second range within
100 seconds. In addition, there was large fluctuation in particle speed due to the fact that
individual particles travel at different speeds even under the same actuation conditions.
Figure 4-18 – Fluidic speed plotted as a function of time for the wider 700 µm channel using bellows
actuation.
4.5 On-Chip Sample Preparation
To develop practical point-of-care devices, sample preparation is a key function that must be
designed and integrated in the microfluidic chip. For CD4 T cell enumeration, a single step
sample preparation of staining the blood sample is required. Other blood testing that involves
79
probing of intracellular protein, or DNA molecules often require more complex and cumbersome
sample preparation procedure such as washing, cell lyse and amplification. This section
describes the design and development of the on-chip sample preparation sub-system that includes
reagent and reagent incorporation, reagent re-constitution and mixing functions for the cell
enumeration assay.
4.5.1 Reagents
The reagents required to complete a CD4 T cell absolute count are anti-human CD3+ antibody
and anti-human CD4+ antibody. Each antibody is labeled with a specific fluorescent dye to
enable optical detection. The dye PE-Cy5 is used to conjugate with CD3 antibody while PE is
used to conjugate with CD4 antibody. This arrangement enables the use of single wavelength to
excite both fluorescent molecules so that both the CD3 and CD4 cell populations can be counted
simultaneously. The fluorescently labeled antibodies were purchased from BioLegend
(www.biolegend.com), catalogue number 344605/344606 (PE anti-human CD4), and catalogue
number 300410 PE/Cy5 anti-human CD3.
4.5.2 Reagent Handling and Incorporation
A number of different approaches for the incorporation of the dry reagents into the microfluidic
cartridge have been investigated:
a) The application of the reagent directly in the microfluidic channel prior to bonding the chip
via liquid spotting; and drying of reagent after bonding
b) Drying of the reagent on a plastic plug, followed by the application of the plug on bonded chip
by a heat staking process (allows separate batch manufacturing of functionalization process and
cartridge manufacturing process)
c) Drying of reagent on a plastic disk, and followed by insertion of the disks in to the chip prior
to bonding and packaging of the chip
d) Drying of reagent as a pellet, and deposit the pellet inside the fluidic chip prior to bonding and
packaging
80
A qualitative analysis was conducted to evaluate the different strategies of incorporating reagents
into the disposable cartridge. The following table summarizes the advantages and disadvantages
of each approach.
Table 4 - Qualitative analysis of five, reagent drying approaches investigated in this thesis.
Reagent
Drying
Direct in Channel Plug Carrier Disk Carrier Pellet Comments
Coated
material
Must be microfluidic
chip material;
potential surface
treatment may be
required and its
impact on bonding is
unknown.
Plug material
could be
different from
microfluidic
chip material
Disk material
could be
different
from
microfluidic
chip material
No media
required
Must be
compatible with
future assays
and
requirementsw
Drying
process
Tested and proven Tested and
proven with
sugar sample
Tested and
proven
Tested and
proven
Either freeze
drying or slow
drying
Integrati
on
Prior to microfluidic
chip bonding and
assembly
After bonding
of the
microfluidic
chip
Prior to
microfluidic
chip bonding
and assembly
Prior to
microfluidic
chip
bonding
and
assembly
Preferably
integration
completed
before bonding
Re-
suspensi
on
Need to verify Tested for
selected
reagents; need
verification on
Need
verification
Need
verification
81
target
reagents
Handling No handling
required
Pick and place
handling
Difficult to
handle
Delicate
handling,
difficult to
automate
Since during bonding and assembly, the microfluidic chips are exposed to high temperatures ( >
65 degree Celsius), the reagent integration is best to be completed afterwards to prevent loss of
protein activity. Given the analysis conducted, it is clear that the reagent plug is the best
approach to introduce dried reagents to the microfluidic chip in terms of both handling and
preserving function.
4.5.3 Reagent Drying
Reagent drying was completed at Reametrix using their proprietary slow drying process, which
has received US Food and Drug Administration (FDA) approval. Both CD3 and CD4 antibody
reagents were dried on a plastic transfer plug
Figure 4-19 – 3D drawing of reagent plug used to handle and incorporate reagents into the microfluidic
cartridge.
A stock solution primarily comprised of sugar additives was used to protect the protein
molecular structure and chemical activity during the drying process. Five microliters of the
82
antibody reagents were first dispensed on a plastic reagent plug, shown in Figure 4-19. The
reagent plug was then slow dried under room condition in a vacuum. Stock sugar solution was
also added to the reagent plug to preserve the chemical activity of the antibody. Figure 4-20 is a
picture of reagent plug after drying. The pink color exhibited by the dried pellet indicates a high
concentration of antibody.
Figure 4-20 - Reagent plug coated with dried fluorescently labelled CD4 antibodies. The pink color indicates
the high concentration of antibodies.
4.5.4 Re-suspension
The sample preparation component allows the on-chip dried reagents to be mixed with the blood
sample. This part of the microfluidic system must accomplish two main objectives: re-suspend
the dried powder reagents and conjugate of the target biomolecule with appropriate fluorescent
83
tags. Several design concepts were proposed and the based on the functionality, one concept was
chosen and a prototype microfluidic device was fabricated and characterized.
4.5.5 Microfluidic Mixer
The mixing channel must first suspend the dried powder reagents in the blood sample. The re-
suspended reagents will then be mixed with the sample. After incubation, the stained sample will
be ready for detection. Typically, the same sample preparation process takes 30 minutes to
complete in a traditional lab setting under the standard flow cytometry protocol. Using a
microfluidic device, the reaction time is significantly reduced. The entire sample preparation: re-
suspension, mixing and incubation can be completed in less than 10 minutes.
The aim of a microfluidic mixer is to thoroughly mix multiple samples, or material in a
microchannel. In these devices, mixing is essentially achieved by enhancing the diffusion effect
between different samples. Typical microfluidic mixer can be divided into two categories:
passive mixer and active mixer. In passive mixer, capillary force driven diffusion is the primary
driving force for mixing. Geometries and structures are often used to introduce turbulence in
otherwise laminar microfluidic flow. Common passive microfluidic mixers include split and
combine, T or Y shaped microchannels, and barriers or obstacles structure embedded in the
channels. Table 5 is a summary of passive microfluidic mixer that were studied in recent years in
literature143.
Table 5 - Performance of passive micromixers in recent development143.
Categories Mixing Technique Mixing Time (ms) Mixing Length (µm)
Lamination Wedged shaped
inlets144
1 1
90 rotation145 - -
Zigzag channels Elliptic-shape
barriers146
- 10,000
Folding structure147 489 -
84
3D serpentine
structure
Creeping structure148 - -
Stacked shim
structure149
- -
Multiple splitting,
stretching and
recombining flows150
- -
Unbalanced driving
force151
- -
Embedded barriers SMX152 - -
Multidirectional
vortices153
- 4,255
Twisted channels Split-and-
recombine154
730 96,000
Surface chemistry Obstacle shape155 - 1,000
T-/Y- mixer156 - 1,000
On the other hand, an external energy source is applied to perturb the sample flow and enhance
mixing in the active mixers. In active mixing, an external force is used to enhance mixing, such
as mechanical, acoustic or thermal. Table 6 outlines the recent development in passive
microfluidic mixer143.
85
Table 6 - Performance summary of active micromixers published in literature143.
Categories Mixing Technique Mixing Time
(ms)
Mixing Length
(µm)
Acoustic/Ultrasonic Acoustically driven sidewall-trapped
microbubbles157
120 650
Acoustic streaming induced by
surface acoustic wave158
600 10,000
Dielectrophoretic Chaotic advection based on Linked
Twisted Map159
- 1,000
Electrokinetic time-
pulsed
Chaotic electric fields 100 Width*5.0
Periodic electro-osmosis flow160 - 200
Electrohydrodynamic
force
Staggered herringbone structure161 - 825
Staggered herringbone structure162 - 2300
Thermal actuation Thermal163 - 6,000
Magneto-
hydrodynamic flow
High operating frequency164 1,100 500
Electrokinetic
instability
Low Reynolds number139 - 1,200
Low Reynolds number165 - 1,200
For a point-of-care device, the key design considerations are low power consumption and ease of
integration with the detection and sample introduction protocols. Robustness and reliability are
two other major design inputs as well. As a result, an active microfluidic mixer, shown in Figure
4-21 was designed and developed in this thesis to meet the desired requirements.
86
Figure 4-21 - Microfluidic mixer design. The “wiggly” channels are mixing structures which utilizes Dean’s
flow to enhance mixing efficiency.
This mixer design relies on Dean’s flow to enhance the turbulence in the microchannels. This
improves the mixing efficiency to reduce the reaction time and device geometry. The sample
plug will be actuated and moves back and forth around the center point similar to a pendulum,
where dried reagent is encapsulated via a plastic reagent plug. An external pneumatic pumping
system, as described in Section 4.2 actuates the sample plug movement in the microfluidic
channel. As the blood samples moves back and forth inside the microchannel, it re-suspends and
mixes with the dried reagent. The sample, along with reagent, moves through the serpentine
structure. This design utilizes Dean’s flow to introduce turbulence into the flow, hence
improving the mixing efficiency.160
4.5.5.1 Physics of Dean Flow
In microfluidic channels, since the dimensions of the channels are on the order of hundreds of
micrometers, the flow is typically laminar. This is beneficial for certain applications but is a
significant limitation for applications such as mixing. To have improved mixing results, turbulent
flow is required to allow thorough mixing of different liquid samples. Typically, diffusion is the
phenomena that was exploited in on-chip mixing in microfluidic systems. To further enhance the
87
rate of mixing and ensure a thorough re-constitution of the dried reagent on-chip, other physical
phenomena were often utilized as well such as Dean’s flow166.
Figure 4-22 - Dean flow. In curved channels, when inertia is important, faster moving fluid near the channel
center tends to continue outward, and to conserve mass, more stagnant fluid near the walls re-circulates
inward. This creates two counter-rotating vortices perpendicular to the primary flow direction166
Secondary flow arises when a fluid flows through a curved channel because of a mismatch of the
velocity in the downstream direction between fluid in the center and near-wall regions of a
channel. Therefore, fluid elements near the channel centerline have larger inertia than fluid near
the channel walls, and would tend to flow outward around a curve, creating a pressure gradient in
the radial direction of the channel. Because the channel is enclosed, relatively stagnant fluid near
the walls re-circulates inward due to this centrifugal pressure gradient, creating two symmetric
vortices as shown in Figure 4-22. A dimensionless number that describes the magnitude of this
flow was first established by W. R. Dean167, and a more generally accepted form of this Dean
number is described by Berger et al168 as 𝜅 = (𝐻
2𝑅)
1
2𝑅𝑒, where H is the width of the curved
channel, R is the radius of the channel curvature and Re is the Reynolds number. Berger et al.
noted that the ratio of the channel dimension to the radius of curvature, defined by the parameter
𝛿 =𝐻
2𝑅, also has important effects on the shape of the secondary flow168. Following Squires and
Quake119 the secondary flow velocity scales as 𝑈𝐷~𝜅2 𝜇
𝜌𝐻. Besides giving a measure of the Dean
88
flow velocity, increases in Dean number are associated with changes in shape of the secondary
flow, with the centers of the symmetric vortices moving towards the outer wall and development
of boundary layers with increasing κ.168
4.5.5.2 Applications of secondary flow in microfluidic systems
Secondary flows have been employed in microfluidic systems primarily for applications in fluid
mixing. Mixing in microfluidic systems has been extensively explored because of the difficulty
in quickly mixing fluid streams without the aid of turbulence.169,170
Most techniques are based on the concept of increasing the interfacial area for diffusive mixing
to occur. Often the concept of chaotic advection is used, whereby fluid interfaces are stretched
and folded to increase the interfacial area to an extreme level171. Because of the exponential
growth in stretching of fluid interfaces in these systems, the positions of individual fluid
elements cannot be confidently assigned, recapitulating an aspect of turbulent flow that leads to
good mixing.
Secondary flows in curved microfluidic channels have been used to increase the interfacial area
for diffusive mixing. One of the first examples is the use of three-dimensional ‘‘twisted’’
channels analogous to macro-scale systems that take advantage of chaotic advection172.
Effective mixing in microfluidic channels requires that fluids be manipulated to increase the
interfacial surface area between initially distinct fluid regions so that diffusion can complete the
mixing process in a reasonable time. Unfortunately, the rapid mixing produced by turbulent
flows is usually not available at the micro-scale because the Reynolds number (Re)173 is typically
below the critical value for transition to turbulence. Thus, some other mechanism must be
employed to enhance mixing.
It has been shown that a “twisted pipe” has the potential to enhance mixing even at low Reynolds
numbers174. This mixing enhancement is possible because of the phenomenon known as chaotic
advection175,176, in which simple regular velocity fields produce chaotic particle trajectories.
Dynamical systems theory shows that chaotic particle motion can occur when a velocity field is
either two-dimensional and time-dependent or three-dimensional (with or without time
dependence)170. The occurrence of chaotic advection typically indicates rapid distortion and
89
elongation of material interfaces. This process significantly increases the area across which
diffusion occurs, which leads to rapid mixing.
Advantages of using secondary flow in curved channels for mixing include: (i) the relatively
simple design and operation and ease of fabrication, (ii) enhancement of mixing with increasing
flow rate, and (iii) applicability to a range of different fluids of varying viscosities, densities, and
conductivities. However, it should be noted that these techniques are often not appropriate for
many microfluidic lab-on-a-chip applications dealing with small volumes of fluid, since mixing
enhancement becomes negligible for lower flow rates where κ < 1.
In the mixer design shown in Figure 4-21, secondary flow in serpentine channels was utilized to
enhance mixing. Active fluidic control was also implemented to reduce the mixing time and
increase mixing rate. This approach enables rapid and effective, thorough mixing of blood
sample with dried reagents on-chip, which is a critical requirement for the point-of-care CD4 T
cell counting test.
4.5.5.3 Mixing Testing
A re-suspension microfluidic chip was designed to test and characterize the on-chip dried reagent
re-constitution and mixing. Figure 4-23 is a picture of the re-suspension slide designed and
fabricated using injection molding process. The microfluidic device was made from PMMA. The
channels are 800 µm wide and 400 µm deep. The circular hole in the middle of the reagent
chamber is where reagent plugs were inserted after drying.
90
Figure 4-23 - Resuspension slide design layout. In this design, four different mixers were proposed to optimize
the mixing and fluidic motion in the microchannel. The circular holes are reagent plug chambers where plugs
are inserted.
The plastic COC plugs, coated with dried antibody reagents, are inserted and assembled with the
re-suspension microfluidic chips via a snap fit. Figure 4-24 illustrates an assembled chip. The
testing and characterization of re-suspension, mixing and incubation was completed using the
bellows actuation setup shown in Figure 4-15.
91
Figure 4-24 - Fully assembled microfluidic re-suspension prototype slides with reagent plugs inserted. The re-
suspension prototype slides were used to characterize on-chip mixing and re-suspension of dried reagents.
The same experimental setup that was used in bellows actuation characterization was used in
mixing and re-suspension testing. Figure 4-14 on Page 74 illustrates the block diagram of the
fluidic setup and connections used in mixing testing. The re-suspension slide was connected to
the bellows actuation slide via a soft, tygon tube. Plastic microfluidic connectors were used to
connect the tube and the plastic microfluidic device with an air tight seal. The bellows membrane
deflection induced a volume change in the microfluidic system, which results in a pressure
change inside the microchannels to drive the liquid sample forward. When the bellows returned
to the original state, the volume inside the microfluidic system expanded, thus liquid sample
retracts and moves backward inside the microchannel. Figure 4-15 on Page 75 is a photograph of
the experimental setup used in this part of the work.
4.5.5.4 Testing Procedure
Fluorescently labeled antibodies were purchased from BioLegend as described in Section 4.5.1.
The liquid reagent was used as the control sample on the flow cytometer for re-suspension
testing. Ten micro liters of Immunotrol blood control sample, purchased from Becton Dickinson,
was injected into the microfluidic slide via the sample port. After connecting and sealing the
Reagent plug
92
bellows slide with the re-suspension slide, the stabilized blood sample was then pushed forward
to the reagent chamber. The blood sample was then actuated back and forth around the reagent
chamber to re-constitute and mix with the dried reagent. After mixing is complete, the stabilized
blood sample was then pushed forward to the sample outlet, which was transferred to a flow
cytometer for cell counting using a pipette.
4.5.5.5 Experimental Observations and Results
4.5.5.5.1 Mixing Time
The stabilized blood sample was actuated back and forth inside the microfluidic mixer for 10, 15,
20, and 30 passes using the experimental setup shown in Figure 4-15 on Page 75. Table 7 below
summaries the mean fluorescence intensity measured by flow cytometer as a function of varying
mixing time. From this investigation, it was clear that 15 passes of back and forth actuation, (~4-
5 minutes) was required to re-suspend the dried reagents thoroughly.
Table 7 - Mixing time characterization result. Relative mean fluorescence intensity was
calculated as the ratio between liquid control mean fluorescence intensity and measured
mean fluorescence intensity for each scenario. This characterization was completed using
design 2 of the resuspension prototype slide shown in Figure 4-23.
Number of Passes Mixing Time (sec) Relative Mean Fluorescence Intensity
5 passes 50 0.2
10 passes 100 0.5
15 passes 150 0.7
20 passes 200 0.98
25 passes 250 1
30 passes 300 1
4.5.5.5.2 Reagent Chamber Design
In Figure 4-23 on Page 90, there were three different micro-mixer designs in the re-suspension
slide. As shown in the figure, design 2 had a reagent chamber that was wider than the reagent
plug while in design 4, the reagent chamber has the same width as the reagent plug. It was
observed design 4 worked much better than design 2, since liquid tend to flow by reagent plug on
the side instead of underneath the plug with reagent if the chamber was wider than the reagent
plug. This significantly reduced the interaction time between dried reagent and liquid sample
during re-suspension, hence leads to longer re-suspension time.
93
Table 8 - Mixing time characterization result. This characterization was completed using
design 4 of the resuspension prototype slide shown in Figure 4-23. With the optimized
reagent chamber dimension, the mixing time can be further reduced.
Number of Passes Mixing Time (sec) Relative Mean Fluorescence Intensity
5 passes 50 0.3
10 passes 100 0.7
15 passes 150 0.98
20 passes 200 1
25 passes 250 1
30 passes 300 1
4.5.5.5.3 Reagent Plug
Some of the reagent plugs underwent oxygen plasma treatment prior to reagent drying. The
oxygen plasma treatment improved the adhesion of the dried reagents: no reagent came off the
plug surface but re-suspension took longer (up to 30 passes to dissolve the dried reagents). The
oxygen plasma treatment made the plastic plug hydrophilic, especially along the circular walls of
the tip. During mixing, the liquid tends to flow through the reagent chamber around the plug,
hence reducing contact with the dried reagent and increasing the required re-suspension time.
Table 9 - Mixing time characterization result. This characterization was completed using
design 4 of the resuspension prototype slide shown in Figure 4-23 and surface treatment of
the reagent plug.
Number of Passes Mixing Time (sec) Relative Mean Fluorescence Intensity
5 passes 50 0.25
10 passes 100 0.65
15 passes 150 0.78
20 passes 200 94
25 passes 250 1
30 passes 300 1
4.6 Conclusion
In this chapter, we investigated the two fluidic transport mechanisms: capillary driven flow and
volumetric pressure driven flow. Capillary driven flow is a passive technique where no external
forces or energies are required to manipulate the sample fluids on-chip, whereas volumetric
pressure driven flow requires an external energy source and components to generate the pressure
94
needed to achieve the desired flow control. Capillary driven fluid flow is strongly dependent on
material property, more specifically surface contact angle with the liquid. Hence to optimize or
enhance the fluidic flow on-chip, surface treatment such as oxygen plasma is often required. The
active fluidic manipulation method is less dependent on the material of the microfluidic device
and it produces more consistent and precise fluidic transport outcome, although more energy
consumption is necessary and the overall system is more complex.
A detailed analysis on each fluidic actuation approach: capillary driven and volumetric pressure
driven fluid flow was conducted in this work. The flow characteristics were investigated and
strategies to control the flow speed were presented in this chapter. The on-chip fluidic
management is a key component in the disposable cartridge that completes the cell enumeration
analysis, whereas volumetric pressure driven flow requires external energy sources and
components to generate the pressure needed to achieve the desired flow control.
95
Chapter 5 Optics and Detection
After the fluidic system of the proposed portable cell/particle detection system is finalized, the
optical detection sub-system can be designed for cell/particle detection and counting. Since a
dynamic counting approach is used in this work, maintaining a uniform flow speed is critical in
ensuring accurate cell counting results. The detailed analysis and results of particle detection and
enumeration are shown in this chapter.
5.1 Introduction
This chapter focuses on the development of an optical based imaging module to detect and
enumerate microparticle and cells populations. A brief literature review explains the current
common approaches and research results, followed by a detailed description of the dynamic cell
counting method and experimental results of the proposed detection concept. The second part of
this chapter discusses the utility of using this imaging technique to develop a beadarray detection
platform for the analysis of proteins and DNA/RNA molecules.
5.2 Particle Detection using Optics
As described in the introduction chapter, there are a number of detection approaches. Optical and
fluorescent detection is the most commonly used approach and widely adopted in laboratory-
based technologies such as flow cytometry and microscopy. It offers better sensitivity, accuracy
and precision, but suffers on weight, size and instrumentation cost.
5.2.1 Optical Detection Methodology Literature Review
Optical detection of particles and cells can be classified in two main categories: far field and near
field imaging. Far field imaging is a direct imaging approach where lenses are used to capture
96
directly the particles/cells whereas in near field imaging, the diffraction pattern of the subject is
captured and original image is re-constructed from the diffraction pattern.
5.2.1.1 Static Imaging
The most common optics based approach is far field static imaging of cells or microparticles.
The working principle is the same as a modern microscope. Typically if fluorescence is utilized
to specifically target cells of interest, expensive and bulky optical components, such as filters,
dichroic mirrors and condensers, are required177.
With the advances in microfluidic lab-on-a-chip technologies, researchers began to investigate
mobile platforms that can detect and analyze cells at point of care. Ozcan et al.178 engineered an
imaging module that can be attached to a mobile phone to analyze and image cells. The system
comprises affordable off-the-shelf optical components that are readily obtainable. The imaging
system has a field of view of ~81 mm2 and a depth of field of 2 mm, permitting analyzing
volumes > 0.1 mL for high throughput applications.
Static imaging offers the best optical quality and flexibility in detecting target particles/cells. The
optical detector also has the freedom to vary the exposure time to optimize the acquired signal
intensity level. The main disadvantage related to this type of imaging technique is the ability to
multiplex multiple target particle/cell population detection. The overall system will get
significantly more complex mechanically and becomes impractical to be miniaturized to a
portable, point-of-care analytical or diagnostic tool.
5.2.1.2 Microfabricated Flow Cytometry
Microfabricated flow cytometry is another common approach179-181. Recent development in
micro/nano fabrication in the semiconductor industry has accelerated research efforts and
progress in this area. In these approaches, the underlying system design follows that of a
conventional flow cytometer where particle focusing is utilized to form a single file of
cell/particle before being interrogated by a laser beam. Different detectors have been used to
capture individual emission wavelength of the target fluorescence. Morgan et al.8 developed a
device where all the optical devices were monolithically integrated in a Si wafer while the optical
routing was accomplished via integrated waveguides.
97
A microfabricated flow cytometry, in essence, employs the same technology and process as
conventional flow cytometry. Without the necessary fluidic system, it is not feasible to achieve
the required flow rate and analysis throughput for practical portable and point-of-care
applications.
5.2.1.3 Near Field Imaging
Far field imaging offers excellent image quality. However, it requires use of filters and lenses,
which makes the instrument larger and more expensive. Another approach is that of near field or
contact imaging, where the subject is placed in close proximity of the detector (tens of microns).
A diffraction pattern is generated and captured by the optical sensor. Upon numerical
reconstruction, the original subject image can be re-constructed182-186.
Coskun et al.187 developed a lens-free imaging system that can detect fluorescently labeled cells.
In this work, the optical sensor was placed directly on top of the microfluidic channel with target
cells, hence the name lens free imaging. A similar method was described in a patent
application188, where homogeneous particles in a solution can be detected in a lens-less setup
using the holography technique.
The main problem associated with near field imaging is the requirement of placing the target
cells/particles close to optical sensor. This is also detrimental to any microfluidic based portable
analytical systems that can be useful and effective in clinical and commercial applications.
5.2.2 Dynamic Particle Detection and Counting
A dynamic counting/detection approach was developed in this work to improve the mechanical
stability and reduce the active components in the system. When combined with capillary
microfluidic devices described earlier, this forms a simple and flexible multiple fluorescence
detection system. This could be miniaturized on a handheld platform as a portable cytometer or
cell analysis device. The following sections describe the dynamic imaging, its principle and
implementation.
98
5.2.2.1 Design Concept
The static imaging requires stitching of multiple images to cover the entire imaging area due to
the limitation on the size of field of view. To image multiple biomarkers with difference
fluorescence antibodies, filter wheels must be used to choose the appropriate fluorescence filter
just as in a fluorescence microscope. On the other hand, in conventional flow cytometry,
individual particles/cells are interrogated one by one. Although different fluorescent dyes can be
distinguished through complex optical systems involving dichroic mirror and filters, the entire
setup is bulky and prone to mechanical instability. The fluidic system is also very complex to
produce single column particles/cells so that only one particle/cell is being examined at a time.
To improve the detection, a dynamic imaging approach is designed and developed in this work.
This dynamic imaging approach uses wide microfluidic microfluidic channel structures to
regulate and control the fluid flow during analysis. The particles are moving in multiple streams
in parallel passing through the optical detector. The use of microfluidic device eliminates
complex fluidic control system present in a typical flow cytometer. The multi-stream imaging
also improves the throughput of this approach so that the final particle/cell analysis platform is
practical and feasible to become a commercial cell analyzing device. The dynamic imaging,
while particles are moving inside the microchannel, further eliminates the need of filter wheels in
multi wavelength fluorescence detection. The reduction of the required mechanical components
makes this an ideal platform for a portable or handheld cell analysis device.
5.2.2.2 Optical Detector
There is a wide range of optical detectors that are used in biomedical and scientific imaging
applications, including one dimensional avalanche photodiode, photomultiplier, and two
dimensional CCD (charge coupled device) and CMOS (complementary metal-oxide
semiconductor) sensors. For the imaging methodology described in the previous section
(5.2.2.1), a two dimensional image sensor is required.
99
5.2.2.2.1 CCD vs CMOS
Much has been written about the relative advantages of CMOS versus CCD image sensors. Both
technologies and markets have been evolving, affecting not only what is technically feasible, but
also what is commercially available. Some applications are best served by CMOS sensors while
for others a CCD sensor might be the superior choice. In most visible imaging application,
CMOS outperforms CCD imagers, while in Near Infrared (NIR) application, CCD may be a
preferred.189,190 In scientific applications where sensitivity is critical to capture weak
fluorescence signal, CCD is also the preferred device. It also has hardware binning features to
enable combination of adjacent pixels to act as one giant pixel, hence significantly reducing
readout noise and improving signal to background ratio.
5.2.2.2.2 Optical Sensor Selection and Characterization
Several CCD and CMOS cameras that are designed for scientific imaging applications were
identified and tested. For the target application, sensitivity and dynamic range are the two most
important parameters. Recent advances in electronics and semiconductor manufacturing process
have enabled the realization of wide dynamic range in all the optical image sensors. The real
limiting factor in this characterization becomes the sensitivity. The other consideration is form
factor, since the eventual goal is to develop a portable, handheld cell analyzer platform for global
health. Affordable devices with small foot print, low power consumption are also important
considerations in choosing the image detector.
Four different Lumenera cameras have been tested. These low cost, scientific grade Lumenera
modules are compared with a more expensive Pixelfly USB camera which was used as the bench
mark. An inverted fluorescence microscope was used for the experiments. The microscope
attached halogen lamp was used for the transmitting light. The testing parameter was chosen to
be close to the condition for CD4 cell detection.
The testing conditions are described as follows: 10X objective lens (NA = 0.25) was focused on
a cover glass slide, the lamp was adjusted so that the average intensity reading on Pixelfly USB
is about 2000 when using 20 ms exposure time and 1x1 binning. The lamp light setting was kept
the same throughout the tests.
100
Within the four Lumenera cameras, only LW110 is a CMOS camera. Other three cameras are all
CCD cameras with hardware binning capabilities. LW160 camera uses the same SONY
ICX285AL sensor (type 2/3) as Pixelfly USB camera. LW130 camera uses a smaller SONY
ICX205AL sensor (type 1/2). Images were always acquired using 16 bits bit depth. The gamma
is set to 1 to keep the original data untouched. Whenever available, the smallest gain that can still
cover the whole bit depth of the camera at strong light condition is used. The actual gain used are
(LW160, LW110 gain=1), (LW130, Lt365 gain=0.7). The background was acquired using the
same 20 ms condition except the camera was fully covered. The following table listed the testing
results.
Table 10 - Camera comparison summary.
Camera model binning exposure I average stdev var(I) SNR*
Pixelfly USB 1x1 20ms 2076.81 130.2 8476.02 27.1
2x2 20ms 8422.04 268.24 35976.35 32.9
Lumenera LW160RM-sci 1x1 20ms 934.36 80.45 3236.10 24.3
2x2 20ms 3739.44 155.1 12028.01 30.7
4x4 20ms 15039.43 316.56 50105.12 36.5
Lumenera LW130RM-sci 1x1 20ms 268.18 54.3 1474.25 16.9
2x2 20ms 1075.1 97.49 4752.15 23.9
4x4 20ms 4341.97 183.39 16815.95 30.5
lumenera LW110 1x1 20ms 344.77 52.33 1369.21 19.4
Lumenera lt365 1x1 20ms 684.85 73.27 2684.25 22.4
2x2 20ms 2738.76 140.85 9919.36 28.8
3x3 20ms 6173.01 208.62 21761.15 32.4
4x4 20ms 10994.44 275.6 37977.68 35.0
8x8 20ms 44235.2 554.81 153907.07 41.0
101
* The signal to noise ratio (SNR) here is based on the conditions used above, which will be
different for the actual CD4 cell signal detection.
5.2.2.2.3 Camera Characterization Results
The commonly used setting on the Pxelfly USB camera for CD4 cell detection are 2x2 binning
with 20 ms exposure time (blue color). Other camera settings that can achieve comparable SNR
have been identified (green color). LW110 has a rather low SNR number. Combined with its
small pixel size at 3.6 μm, it is unlikely it can meet the requirements for CD4 cell detection.
Lt365 is a high-end camera. Its great performance has a high price tag. Since LW160 is
comparable to Pixelfly USB that can detect CD4 cells. It is unlikely that we will choose Lt365.
However, its superior performance will be useful for an application requiring better sensitivity
such as beads assay.
Both LW160 and LW130 have settings that comparable with the Pixelfly USB. The LW130
camera is especially interesting. At 4x4 binning, its SNR is at 30.5. Although the ICX205 sensor
has a smaller pixel size of 4.65 μm comparing to the 6.45 μm pixel size for ICX285 sensor, its
4x4 binning has an effective pixel size of 18.6 μm while the 2x2 binning of ICX285 is only 12.9
μm. It is reasonable to say that the LW130 can detect CD4 cells at the same condition as Pixelfly
USB even considering that the ICX205 sensor has a smaller quantum yield comparing to ICX285
sensor. In fact, a test using CD4 PE antibody stained Immunotrol sample has been conducted.
The results demonstrated comparable detection sensitivity to the Pixelfly USB camera.
5.2.2.3 Imaging Lens
At high numerical apertures of the microscope, depth of field is determined primarily by wave
optics, while at lower numerical apertures, the geometrical optics dominates the phenomenon.
Using a variety of different criteria for determining when the image becomes unacceptably sharp,
several authors have proposed different formulas to describe the depth of field in a microscope.
The total depth of field is given by the sum of the wave and geometrical optical depths of fields
as:191
𝑑𝑡𝑜𝑡 =𝜆𝑜𝑛
𝑁𝐴2 +𝑛
𝑀⋅𝑁𝐴𝑒 (5 – 1)
102
Where 𝑑𝑡𝑜𝑡 represents the depth of field, 𝝀 is the wavelength of illuminating light, 𝑛 is the
refractive index of the medium (usually air (1.000) or immersion oil (1.515)) between the
coverslip and the objective front lens element, and 𝑁𝐴 equals the objective numerical aperture.
The variable 𝑒 is the smallest distance that can be resolved by a detector that is placed in the
image plane of the microscope objective, whose lateral magnification is 𝑀. Using this equation,
the depth of field 𝑑𝑡𝑜𝑡 and wavelength 𝜆 must be expressed in similar units. For example, if
𝑑𝑡𝑜𝑡 is to be calculated in micrometers, 𝜆 must also be formulated in micrometers (700
nanometer red light is entered into the equation as 0.7 micrometers). Notice that the diffraction-
limited depth of field (the first term in the equation) shrinks inversely with the square of the
numerical aperture, while the lateral limit of resolution is reduced in a manner that is inversely
proportional to the first power of the numerical aperture. Thus, the axial resolution and thickness
of optical sections that can be attained are affected by the system numerical aperture much more
so than is the lateral resolution of the microscope.192
5.2.2.4 Optics Setup
A handheld green laser pointer, emission wavelength 532 nm with power of 18 mW was used to
excite the fluorescent dyes. Emission filters used in the fluorescence detection system were
purchased from Semrock (585/40 product number FF01-585-40 and 708/75 product number
FF01-708-75). An Olympus microscope objective lens (10x, NA 0.30) for emission light
collection was purchased from Spectra Physics. The CCD camera (Pixelfly USB) was purchased
from PCO-TECH Inc.(Kelheim, Germany). The fluorescence signals were acquired in a 2x2
binning configuration to enhance the signal to noise ratio. The optics was assembled in a lens
tube (SM1L, SM1V and SM1ZM, Thorlabs) configuration.
5.2.2.4.1 Measurement Setup
This imaging system is based on a modified fluorescence microscopy imaging approach, where
the excitation filter and dichroic mirrors are not required. Figure 5-1 is a schematic diagram of
the optical detection system. Upon illumination, fluorescently labelled particles/cells of interest
will be excited and emit light. The imaging system, a microscope objective lens, magnifies the
103
target particles/cells, collects emitted light and projects to the optical detector. The light then
passes through emission filter and is captured by the optical detector. The samples under testing
are loaded onto a custom designed disposable plastic microfluidic chip. A thin blood smear is
formed once the sample fills inside the microfluidic channel, whereby particles are imaged and
characterized based on their optical properties. Furthermore, an image analysis program is used
to analyze and process the acquired optical images for particle and cell detection and
enumeration. The only optical elements in this system are the imaging lens and emission filters.
Unlike a conventional fluorescence microscope setup, the dichroic mirrors and excitation filters
are eliminated, which reduces the complexity of the platform and makes the optical system easier
to miniaturize.
Figure 5-1 – Optical imaging system of the cell/particle detection and analysis platform.
For multi-color detection, a custom designed emission filter was used in the optical detection
system. The filter set, (optical property shown in Figure 5-2) incorporates two half-moon shaped
fluorescence filter into a single round cell. This specific custom filter allows for two-color
fluorescence detection side by side simultaneously.
The imaging system uses wide field dynamic imaging approach that does not require any moving
components, such as filter wheels and rotation stages, needed in standard fluorescent detection
for multi-color (or multiplexed) analysis, a key feature and advantage of this analytical platform.
104
During analysis, a time-series images of particles moving inside the microfluidic device was
captured. An optical detection area, defined by the size of the CCD detector and the
magnification of objective lens, is used to examine and measure the fluorescence signal. The
CCD detector starts data acquisition when the sample enters the analysis chamber. As the fluidic
sample moves into and fills the analysis chamber, the detector continues capturing optical images
over time until the chamber is full and fluidic flow is stopped. By analyzing the acquired images,
the number of fluorescent particles of interest which pass under the detection window can be
counted, generating the final cell count. The captured images are also analyzed to render the
intensity of particles and to generate counting statistics. The microfluidic chamber is designed
and fabricated with a precise volume; hence the final results are well calibrated. An optical filter
with a single filter or an array of filters is placed in front of the optical detector. When an array of
filters is used, the system has the ability to do multi-color (or multiplexed) fluorescent detection
without any additional (moving) components, a key feature of this analytical platform. Figure 5-2
is a graphical illustration that further describes the underlying principle of this technique.
105
Figure 5-2 – Illustration of the multiplexed detection system described in this paper. The optical detector
continually takes images as particles/cells move into the detection window. Particles/cells labelled with
different fluorophores can only be detected in the corresponding sub-regions within the detection window
depending on the filter setup. (a) a graphical illustration of the underline principle of the technique; (b)
transmission spectrum of the left sub-region of the detection window; (c) transmission spectrum of the right
sub-region of the detection window.
5.2.3 Cell Detection and Enumeration using Capillary Microfluidic Devices
In order to demonstrate the proof of concept of this cell analysis platform, CD4 T cell counting
was chosen as the initial test, in direct comparison with clinical flow cytometry. CD4 T cells are
a specific type of white blood cell and its concentration can be a measure of human immune
system strength. This assay is one of the most predictive tests for HIV/AIDS monitoring in
infected individuals and is used to determine the timing for the initiation of anti-retroviral drug
therapy. The test of our cell analysis system is conducted using Immuno-Trol control cell
samples for single population CD4 T cell detection.
106
5.2.3.1 Sample Preparation
Samples were stained with 40 μL PE (excitation 532 nm, emission 585 nm) labeled CD4 antibody
and 6 μL of PE-Cy5.5 (excitation 532 nm, emission 670 nm) labelled CD3 antibody. A single
droplet of stained Immuno-Trol, approximately 5 μL, was transported to the inlet reservoir of the
microfluidic device shown in Figure 3-11. The fluid inside the microfluidic chip was then
completely driven by capillary force. The microfluidic channel can be divided up into three
zones: flow restrictor, detection window and capillary pump. The flow restrictor is a linear
microchannel, with a cross sectional area of 15 μm x 100 μm. This zone is to reduce the fluid
flow to a reasonable rate so that particle/cell motion can be captured by the optical detector. The
detection window has an area of 500 μm x 800 μm, which matches the size of the 1M pixel CCD
sensor through magnification of the objective. The acquired fluorescence images will be
processed and analyzed to produce the final cell/particle counting statistics. The final section of
the microchannel serves a dual purpose: a volumetric reservoir and a capillary pump. As the fluid
sample fills the microfluidic device, the fluidic resistance increases and the flow speed reduces.
To make the whole instrument functional and practical, the sample flow needs to maintain a
uniform rate so that the acquired images can be analyzed to produce an accurate cell
concentration measurement within a reasonable time frame. The widening of the fluidic channel
reduces the fluidic resistance seen by the sample fluid inside the microchannel and effectively
acts as a capillary pump. This capillary actuation counteracts the decrease in the fluidic flow in
the previous sections along the microfluidic device, maintaining a uniform flow profile.
5.2.3.2 Antibody Concentration
In flow cytometry, detailed sample preparation procedures were involved in each analysis,
including CD4 T cell counting. Blood samples are typically lysed, then diluted with Phosphate
buffered saline (PBS) buffer solution. During analysis, distilled water was added to the fluidics
to create hydrodynamic focusing which further dilutes the blood sample. These steps
significantly reduce the unbound antibody concentration remaining in the blood sample. Hence,
in flow cytometry, there is no significant impact on the recorded mean fluorescence intensity.
In the proposed microfluidic based point-of-care system, the only sample preparation involved is
staining with fluorescent dyes. As a result, the background fluorescent signal of the blood sample
became the dominant source of noise.
107
The background fluorescence primarily came from two sources: auto-fluorescence from the
blood and unbound antibodies in the solution. To avoid any complex on-chip sample washing
processes, the auto-fluorescence of blood is difficult to be further reduced. The unbound
antibody in the solution, however, can be optimized. A range of antibody concentration was used
to study its impact on background and signal to background ratio. 100 microliter of stabilized
blood control sample was separated into two vials: one to be run on flow cytometer while the
other on the microfluidic platform. The flow cytometry sample preparation followed the standard
protocols, including lysing, adding buffer solution and staining while the microfluidic sample
only included staining of fluorescently labeled CD3 and or CD4 antibodies.
CD3 and CD4 antibody concentration optimization was conducted. The optimization was based
on the concentration at which the CD3 and CD4 cell intensity plateau was reached. The deciding
factor here is the average intensity of CD3 and CD4 cells in their respective fluorescent channel,
since the background intensity is low in flow cytometry. However, the same is not necessarily
true for the dynamic wide-field imaging detection approach developed in this thesis. The correct
identification of CD3 and CD4 cell event in the wide-field imaging detection is based on whether
the signal can be well separated from the rest of the background. As a result, the optimization
will need to be based on the signal to background ratio. Experiments using a series of different
concentration of CD3 and CD4 antibody were conducted. The results of their respective average
signal to background ratio was used to identify ideal antibody concentration range.
The concentration range used on the flow cytometer was used as a starting point. The
concentration we chose for CD3 antibody testing were 10, 25, 50, 100, 250, 500 and 2500 ng per
100 µL blood. The concentration range used for CD4 antibody testing were 2.5, 5, 10, 25, 50 and
250 ng per 100 µL blood. Two optical channels, PE and PE-Cy5, were used to observe CD4 and
CD3 respectively. The results are shown in Figure 5-3 and Figure 5-4 where the average signal to
background ratio is plotted against the logarithm of each antibody concentration. Based on the
results, the best average signal to background was achieved at 100 ng and 50 ng per 100 µL
blood for CD3 and CD4 antibody respectively.
108
Figure 5-3 – CD3 antibody concentration titration curve for signal to background ratio.
Figure 5-4 – CD4 antibody concentration titration curve for signal to background ratio.
1.00
1.50
2.00
2.50
3.00
3.50
1.00 1.50 2.00 2.50 3.00 3.50 4.00
Sig
/BG
(A
U)
log of antibody quantity (ng)
CD3 Antibody titration curve
1
1.5
2
2.5
3
3.5
0.50 1.00 1.50 2.00 2.50
Sig
/BG
(A
U)
log of antibody quantity (ng)
CD4 Antibody titration curve
109
Figure 5-3 and Figure 5-4 are titration curves for both CD3 and CD4 antibody concentrations.
From the range of reagents tested in this part of the thesis project, there is an optical level of
antibody concentration for each cell population as shown. Each data point in these two figures
represents average values of three separate measurements. The error bars indicate the variation in
the experimental measurements.
When the antibody concentration is low, there are not enough antibody molecules in the solution
to bind with antigens on cell surface, resulting a weaker fluorescence signal. On the other hand,
if the antibody concentration is too high, all the cell surface binding sites are occupied and there
remains a large number of antibody molecules in the solution, resulting a high background signal
and a low signal to background ratio.
Figure 5-5 – Positive events intensity histogram for PE channel at 50ng CD4 antibody per 100µL blood.
In addition, it was also observed that the population of positive event separation in the CD4
antibody concentration tests that are at or close to optimal concentration. Figure 5-5 shows the
positive events intensity histogram in PE channel when 50ng PE CD4 antibody per 100 µL was
used for the test. There is a clear separation between two intensity populations. It is most likely
that the two populations are CD4 cells and monocytes since the CD4 antigen is also weakly
110
expressed on monocytes. This capability to differentiate cell populations is not crucial for the
CD4 assay. However, it will be critical for other cell counting assays or even beadarray assays.
5.2.3.3 Image Analysis
An image analysis program was developed in collaboration with an undergraduate project
student James Durst, to produce the counting results. The program takes the fluorescence images
acquired by the CCD camera, and based on the intensity level of pixels it detects and tracks the
cells of interest. The entrance region of the detection window is scanned in every captured frame
to search for fluorescent events. Fluorescent objects are detected if groups of pixels display
intensity levels exceeding a pre-determined threshold. Since the fluidic flow inside the
microchannels are strictly laminar, the program would look for small regions where a fluorescent
particle is expected in subsequent frames. Since only a small portion of the entire images are
processed, it significantly improves the processing speed and reduces computing power
requirements. This is very important for real time analysis especially at a high frame rate. To
track fluorescently labelled subjects’ location within each frame, a virtual bounding box is placed
on each particle/cell, recording its maximum and minimum x and y coordinates. The center
position of each box is then computed and marked as the current position of the cell. This
process will be repeated for each frame captured and analyzed to produce a final enumeration
result. In addition to the entrance scanning of target population, particles/cells of interest are also
detected and counted upon exiting the detection window. This secondary analysis is compared
with the entrance scanning to ensure the accuracy of counting. Finally, the detection of
fluorescent subjects by continuously tracking them through multiple image frames increases the
reliability of the counting results, which makes the technique readily applicable for health-related
measurements such as CD4 T cell counts in human blood.
5.2.3.4 Detection Optimization
Since dynamic counting is used in determining the cell/particle statistics, the exposure time of
the optical detector is one of the most important parameter in optimizing the signal to
background ratio of the acquired images and such counting accuracy. The ultimate goal is to
maximize the signal to background ratio, but the ability to this maximization is affected by the
111
exposure time and flow speed. If the CCD integration time is too short, the fluorescence signal
captured by the detector would be low and compromising signal to background ratio. On the
other hand, if the integration time is too long, the particles/cells might be travelling too fast for
the detectors such that it can create motion blur in images or even been missed by detector
completely. Motion blur will compromise the signal to background ratio, while cells flowing
through the detection window will be and missed registering on the CCD and will also lead to
errors in counting and analysis results. The optimal integration time should allow the
fluorescently labelled particles/cells to produce bright signal comparing with background, and at
the same time the electronic circuitry that drives the CCD is fast enough to capture all the
cells/particles of interest with proper sampling rate.
112
Figure 5-6 – Snapshots illustrating the effect of optical detector exposure time. (a-d) shows a steady decrease
in detector exposure time from 50ms to 10ms. The particles/cells shape appears more circular at shorter
exposure, while the signal to background ratio is lower.
5.2.3.5 Comparison with Flow Cytometry
Using our microfluidic approach, enumeration measurements were first conducted using 10-um
Flow-CountTM polystyrene microspheres in phosphate buffered saline (PBS) solutions. The
initial experiments were performed on an Olympus BX50 upright fluorescence microscope. Band
pass filter sets were used for fluorescence excitation and emission measurements. An average
count of 970 ±70 particles/μL was recorded with our microfluidic chips. Using the same testing
material the flow cytometer recorded a count of 1,007 ± 100 particles/μL.
(a) 50 ms exposure, S/B: 4200/2000 (b) 25 ms exposure, S/B: 1700/912
(c) 15 ms exposure, S/B: 858/540 (d) 10 ms exposure, S/B: 695/500
Direction of fluid flow
Ch
an
ne
l wid
th
100 μm
113
This dynamic approach was then tested using Immuno-Trol Control Cells, commonly used
stabilized blood samples for ensuring clinical labs can count CD4 cells at two concentrations.
Immuno-Trol High CD4 counts are in the range of a normal blood sample, and Immuno-Trol
Low is at the level of an immune-compromised individual. A clinically validated commercial
CD4 antibody with Phycoerythrin (PE) dye was used in testing these stabilized blood samples.
The testing produced an average count of 620 ± 15 cells/μL and 184 ± 17 cells/μL for high and
low concentrations whereas the flow cytometer measured 670 ± 70 cells/μL and 158 ± 28
cells/μL respectively. For each sample, the counting test was repeated 10 times.
Table 11 – Comparison of results using wide field dynamic counting system and flow
cytometer.
FlowCount Fluorospheres
(#of particles/μL)
Immuno-Trol High
(# of cells/μL)
Immuno-Trol Low
(# of cells/μL)
Flow Cytometer 1007 100 670 70 158 28
Prototype 970 70 620 15 184 17
5.2.3.5.1 Linearity
To demonstrate the performance of this platform in real world applications, a range of cell
concentrations of interest to CD4 enumeration assay was tested on this platform, from 180 per μL
to 720 per μL as plotted in Figure 5-7. Two types of Immuno-Trol samples were directly used in
this analysis: low (180/ μL) and high (700/ μL) concentration controls while the intermediate cell
concentrations were obtained through dilution from the high concentration Immuno-Trol sample.
In total, there were four concentrations tested in this part of the characterization: 180/μL, 380
/μL, 500 /μL and 700 /μL. The same samples were tested through our portable system and clinical
flow cytometer in parallel and cell statistics were plotted in Figure 5-7. The direct comparison
showed great agreement between the two methods, proving the accuracy of the portable cell
analysis platform under a wide range of cell concentration that is of clinical importance.
114
Figure 5-7 – Linearity test result. Cell concentrations range from 150 to 720 per μL were tested for the
comparison with flow cytometer. (Each data point is an average value for 3 measurements with standard
deviation bar.)
5.2.3.5.2 Mixed Population
In clinical relevant assays, the ability of detecting multiple biological targets is of primary
importance. The multiplexing provides additional information for each sample being studied and
allows identification and characterization of multiple cell populations or cell subpopulations. In
the particular case of CD4 analysis, both CD3 and CD4 cells need to be enumerated to eliminate
possible inclusion of monocytes which also weakly expresses the CD4 biomarker on its cell
surface. To develop a portable cell analyzer that can produce both statistics, a two-color
detection system must be implemented. As a demonstration of the multiplexing capability of this
platform, fresh whole blood sample was used to count cells with both CD3+ and CD4+
biomarkers for absolute CD4 T cell counts. Similar to the schematics shown in Figure 5-1 and
Figure 5-2 where the optical detection window was divided into two sub-regions, two half-moon
shaped fluorescence filters were combined and placed in front of the optical detector. Particles
y = 0.9611xR² = 0.9825
0
100
200
300
400
500
600
700
800
0 200 400 600 800 1000
pro
toty
pe
CD
4 c
ou
nts
flowcytometer CD4 cell counts
Series1
Linear (Series1)
115
labelled with specific fluorophores will be present in the corresponding panel/region of the
optical detector as shown in Figure 5-2. Fresh whole blood sample was prepared as described in
the material section. In the mixed population experiment, cells with the CD3+ antigen were
bound with antibody conjugated PE-Cy5.5 (excitation 532nm/emission 705nm) molecules,
corresponding to the right panel of the detection window shown in Figure 5-2, while CD4+ cells
were bound with antibody conjugated PE (excitation 532nm/emission 585nm) molecules, the left
panel of the detection window in Figure 5-2. In a single run, statistics of cells expressing CD3
and CD4 surface biomarkers could be obtained. Cells that express both CD3 and CD4
biomarkers on the cell surfaces are leukocytes that we are interested in counting in absolute CD4
counting analysis since monocytes do not express CD3+ antigens on their surface.
Similar to previous experiments, the stained samples were run through both the clinical flow
cytometer and our portable cell analyzer system in parallel. This direct comparison showed
excellent agreement between these two approaches and demonstrated the functionality and
accuracy of this dynamic cell counting technique. The imaging system reported in this work can
be further miniaturized and integrated into a handheld electronic device, such as a smart phone,
to be deployed in the field as a point-of-care testing instrument.
116
Figure 5-8 – Two color fluorescence image captured through a custom setup with two half-moon shaped
optical filters placed together side by side. The CD4 cells labelled with PE (phycoerythrin) were shown in the
left panel while the CD3 cells labelled with PE-Cy5.5 were shown in the right panel.
5.2.3.5.3 Limit of Detection
The sensitivity (or limit of detection) of the system is very important for the detection especially
when the signal is weak. Experiments have been conducted using beads that have known
fluorescence intensity based on their Molecules of Equivalent Soluble Fluorochrome (MESF)
value. The beads with PE dyes used here have MESF of 14975. The magnified images of
detected MESF beads during our acquisition were shown in Figure 5-9.
585 nm +/- 20 nm emission channel 705 nm +/- 70 nm emission channel
117
Figure 5-9 – Combined images is showing four detected MESF beads.
Using the signal and noise values acquired on these images, the limit of detection of the system
was determined to be 6400 MESF unit based on that the minimum detectable signal need to be 3
times of noise. The sensitivity of a typical flow cytometer is in the 900 MESF unit range for
fluorescein. Currently, the sensitivity of our system is not as good as flow cytometer, but
additional optimization could help to improve the limit of detection further. Both a more
powerful light source and a better sensor such as backside illuminated sensor can be used to
increase the sensitivity of the system. In addition, our detection is based on image recognition.
The signal can be reliably detected even before the signal reaches the level of 3 times of the
noise. This can further increase our sensitivity. By proper optimization, it is reasonable to expect
this detection system can reach sensitivity in 1,000 MESF unit range.
5.3 Beadarray Multiplexed Detection
A further development of the detection system described in this work is multiplexed beadarray
platform. Similar to the multiplexed beadarray technology from Luminex Corporation, it can
simultaneously detect multiple protein, or enzyme molecules. This approach uses the sensitivity
of amplified fluorescence detection to measure soluble analytes in a particle-based immunoassay,
118
which will significantly widen the detection capacity of the detection platform. Target analytes
can range from blood serum proteins, enzymes to DNA/RNA molecules.
5.3.1 Underlying Strategy
The method is based on two-dimensional fluorescence detection system to identify the bead array
populations in solution. Each bead population can be functionalized with an antibody, or antigen
for a specific test. This detection methodology is similar to the Luminex xMAP® technology
where multiplexed discrete assays are performed on the surface of color coded beads or
microspheres. These beads are read in a desktop analyzer in which lasers or LEDs, optical
detectors and high speed digital signal processors combine to generate fluorescence intensity
based analytical data. The fluorescence intensities of each color of beads are differentiated and
can be correlated to analyte concentration.
In this detection approach, the beads are mixed with blood or serum sample, with a reporter
fluorochrome label in a single or multiple mixing step, followed by necessary washing step. We
may be able to mix the beads, blood or serum, with a reporter fluorochrome tag in a single
mixing step, followed by one washing step. One or more bead populations can be included to
measure unspecific binding. An additional bead can be deployed with a specific fluorescent
intensity, the same as our reporter fluorochrome, to establish a threshold range, positive for all
the analytes measured, as a control signal to the analysis.
The multiplexed beadarray detection concept that is proposed is further illustrated in Figure 5-10.
119
Figure 5-10 – Multiplexed beadarray detection process illustration.193
The fluorescent beads are conjugated with antibodies on their surfaces. Several types of beads,
each with its unique fluorescence intensity levels, are mixed together. The bead solution is then
mixed with a sample solution where target analyte would react and bind to the antibodies on the
surface of fluorescent beads. The binding is specific and would only occur if the target analyte is
present in the solution. Otherwise no binding will take place and a negative signal will be
detected. A second reporting reagent with detector antibody is then added to the mixture. Similar
to the previous reaction, if the target analyte is present in the sample solution, specific binding
would occur and the detector antibodies would bind to the target analyte and form a bead
complex as shown in Figure 5-11.
120
Figure 5-11 – Bead complexes and reagents explanation.
Figure 5-11 illustrates an example of the bead complex after staining is complete. In the case
shown, the fluorescent bead emits in the red channel, at 670 nm. This fluorescent bead is
functionalized with the antibodies, which will capture target antigen floating in the sample
solution. If the antigen is present in the solution, it will conjugate to the surface of the bead via
antibody-antigen binding. A second reagent containing detection or reporter antibody, will then
be added to the mixed solution. In this illustration, the reporter antibody is also pre-labeled with
PE dyes that emits in 585 nm. The reporter antibody will bind to the bead complex, forming the
final bead complex as shown in Figure 5-11. From the graphic illustration, it is clear that the
more antigen molecules in the solution, the more reporter antibody will bound to the surface of
the bead complex, yielding a stronger fluorescence signal in 585 nm channel. The presence of the
target antigen will be decided by the fluorescence signal in 670 nm channel and 585 nm channel,
while the concentration of the target antigen can be determined by the fluorescence intensity
level in 585 nm channel.
Figure 5-12 graphically shows how the bead complex are detected and the target antigen
concentration can be measured and calculated. In a two channel detection scheme a, the first
channel can be used to detect the presence of fluorescent beads as described in Figure 5-11. The
presence of a specific type of bead can be determined by its unique fluorescence intensity. The
fluorescence intensity captured in the second channel is used to quantify the analyte
concentration in the solution. As shown in the illustration, the higher the analyte concentration,
121
the more bound bead complex will be formed and hence yielding a stronger fluorescence signal.
Using this approach, it is possible to use the first channel to identify the presence of a specific
analyte in a solution and using the second channel to quantify its concentration in the solution.
Figure 5-12 – Two color multiplexed beadarray detection concept illustration. The optical imaging area is
divided into two sections: left is the identification channel and the right is the quantification channel.
5.3.2 Proof of Concept Demonstration
To further demonstrate the multiplexed beadarray concept proposed in previous sections, a
proof-of-principle experiment was designed and implemented. The experiment was designed
based on a commercially available assay kit, BC Cytometric Bead Array (CBA) Mouse
Inflammation Kit (Category number 552364), from Becton Dickinson. This kit can be used to
quantitatively measure Interleukin-6 (IL-6), Interleukin-10 (IL-10), Monocyte Chemoattractant
Protein-1 (MCP-1), Interferon-γ (IFN-γ), Tumor Necrosis Factor (TNF) and Interleikin-12p70
(IL-12p70) protein levels in a single sample. This kit has been optimized for analysis of specific
proteins in tissue culture, EDTA plasma and serum samples.
122
5.3.2.1 Experimental Setup
As shown in Figure 5-12, the optical detection area is divided into two sub regions. The one on
the left is coated with a band pass filter with center wavelength of 670 nm. This sub region is
used to identify the presence of the target analyte. The sub region on the right-hand side of
Figure 5-12 has a band pass filter with center wavelength of 585 nm. This region is used to
measure target analyte concentration. The detection optics were the same as that used for
CD3/CD4 cell counting, as illustrated in Figure 5-1.
5.3.2.2 Interleukin 6 Protein
Interleukin-6 (IL-6) protein, a blood serum protein, is used in this demonstration. It is an
inflammation marker that includes information on human health. It is an interleukin194-196 that
acts as both a pro-inflammatory cytokine5,197,198 and an anti-inflammatory myokine199. In
humans, it is encoded by the IL6 gene.200
IL-6 is secreted by T cells and macrophages201-204 to stimulate immune response, for example,
during infection and after trauma, especially burns or other tissue damage leading to
inflammation. IL-6 also plays a role in fighting infection, as IL-6 has been shown in mice to be
required for resistance against bacterium Streptococcus pneumoniae.205
A sample with IL-6 proteins is mixed with the fluorescent beads first. The beads are also
conjugated with IL-6 antibodies on their surfaces. IL-6 proteins in suspended in the sample
solution will then bind to the functionalized fluorescent beads. In a second mixing step, detection
antibodies are added to the solution. The detection antibodies are also fluorescently labeled as
shown in Figure 5-10. Upon mixing and incubation, the detection antibodies (anti-IL-6) will bind
to the IL-6 – bead complex as shown in Figure 5-10, forming a complex in Figure 5-11. This
complex has two fluorescence signatures: one is emitting at 585 nm from fluorescent beads while
the other is emitting at 670 nm from the functionalized detection antibodies. The higher the
concentration of IL-6 in the original sample solution, the more binding of IL-6 with antibodies
will take place and hence the fluorescence intensity in the 670 nm channel will be stronger. If
there are no IL-6 present in the solution, then no antibody-antigen binding will occur during
incubation and reaction and hence no fluorescence signal will be observed in the 670 nm
channel. The 585 nm channel is used to identify and confirm the presence of the IL-6 protein
123
molecules in the solution as shown in Figure 5-12. When a bead complex travels from the first
detection region to the second detection region, if an IL-6 protein is captured by the bead, the
fluorescent signal will be present in both sub panels. Otherwise, only the first sub region, or the
identification sub panel, will show fluorescence signal while the quantification channel will
remain dark since there are no binding between detection antibodies with the target analyte.
5.3.2.3 Sample Preparation
The IL-6 cytokine standards, purchased from BD (http://bd.com/), catalogue number 552364, has
a known concentration of 5,000 pg/mL. The solution was diluted a number of times to obtain
different concentration levels, and incubated for over 3 hours to ensure all antibodies are
conjugated with the fluorescent beads. The dilution procedure that was used to obtain various
concentrations were outlined as follows and also illustrated graphically in Figure 5-13:
Opened one vial of lyophilized Human Inflammatory Cytokine Standards. Transferred the
standard to a 15-mL polypropylene tube. Labeled the tube “Top Standard.”
Reconstituted the standards with 2 mL of Assay Diluent.
Allowed the reconstituted standard to equilibrate for at least 15 minutes at room
temperature.
Gently mixed the reconstituted protein by pipette only. Did not vortex or mix vigorously.
Labeled eight 12 × 75-mm tubes and arranged them in the following order: 1:2, 1:4, 1:8,
1:16, 1:32, 1:64, 1:128, and 1:256.
Pipetted 300 μL of Assay Diluent in each of the 12 ×75-mm tubes.
Performed a serial dilution:
Transferred 300 μL from the Top Standard to the 1:2 dilution tube and mixed thoroughly
by pipette only. Did not vortex.
Continued making serial dilutions by transferring 300 μL from the 1:2 tube to the 1:4 tube
and so on to the 1:256 tube.
Prepared one 12 × 75-mm tube containing only Assay Diluent to serve as the 0 pg/mL
negative control.
124
Figure 5-13 - Dilution steps performed to obtain different IL-6 concentration levels to be measured.
5.3.2.4 Assay
The prepared samples with different analyte concentrations were then divided into two groups:
one group was processed on a clinical flow cytometer while the other was analyzed using the
optical detection setup shown in Figure 5-1. Each concentration was processed individually after
which data were collected and processed together. The experiment was performed based on the
procedure described below:
Vortexed the mixed Capture Beads and add 50 µL to all assay tubes.
Added 50 µL of the Human Inflammatory Cytokine Standard dilutions to the control
tubes as listed in the following table.
Added 50 µL of each unknown sample to the appropriately labeled sample tubes.
Incubated the assay tubes for 1.5 hours at room temperature, protected from light.
Added 1 mL of Wash Buffer to each assay tube and centrifuge at 200g for 5 minutes.
Carefully and consistently aspirated and discarded the supernatant, leaving approximately
100 µL of liquid in each assay tube.
125
Added 50 µL of the Human Inflammatory Cytokine PE Detection Reagent to all assay
tubes. Gently agitated the tubes to re-suspend the pellet.
Incubated the assay tubes for 1.5 hours at room temperature, protected from light.
Added 1 mL of Wash Buffer to each assay tube and centrifuge at 200g for 5 minutes.
Acquire standards from lowest (0 pg/mL) to highest (Top Standard) concentration,
followed by the test samples
CBA samples must be acquired on the same day they are prepared. Prolonged storage of
samples, once the assay is complete, can lead to increased background and reduced sensitivity.
Table 11 summarizes the concentration levels that were tested after a series of dilution
described in Section 5.3.2.3.
Table 12 - Concentration levels of IL-6 cytokine tested as
proof of concept demonstration on prototype.Concentration
Tube label (pg/mL) Standard dilution
1 0 (negative control) no standard dilution(Assay Diluent only) 2 20 1:256 3 40 1:128 4 80 1:64 5 156 1:32 6 312.5 1:16 7 625 1:8 8 1,250 1:4 9 2,500 1:2
10 5,000 Top Standard
Samples were run on BD FACSCanto II and a benchtop prototype using the dynamic
imaging approach. Concentration curves were generated from both instruments and compared.
5.3.2.5 Results and Discussions
The BD Cytometric Bead Array Kit included a 6-peak bead populations and a negative control
sample. The 6 peak beads were injected onto the microfluidic devices described in Section 4.5.4
on Page 82. These beads are fluorescent polystyrene microspheres with different intensities.
Their excitation wavelength is 532 nm and emission is centered at 650 nm. The beads were
processed one by one using the optical detection system shown in Figure 5-1, with bellows
126
actuation described in Section 4.3 on Page 57. The fluorescence intensity of each bead was
detected and recorded. Figure 5-14 shows the histogram of the captured bead population.
Figure 5-14 – Histogram of multiplexed beads detected by the optical detection system described and
developed in this chapter. The beads fluorescence intensities are evenly distributed on a log scale.
Each column in Figure 5-14 represents a bead population that is differentiated by its fluorescence
intensities. From this experimental demonstration, it is clear that the detection system designed
and developed in this thesis can differentiate up to 6 different types of bead population based on
their intensity levels. In a multiplexed bead array implementation with a two wavelength
excitation configuration, this system can be used to detect up to 36 markers simultaneously.
A second set of experiments related to multiplexed beadarray platform was performed to
correlate the detected fluorescence intensity with analyte concentration. The same beadarray
0
2
4
6
8
10
12
4.00 4.20 4.40 4.60 4.80 5.00 5.20 5.40 5.60 5.80 6.00
FREQ
UEN
CY
LOG OF RELATIVE INTENSITY
Beads population identification
Beads 1
Beads 2
Beads 3
Beads 4
Beads 5
Beads 6
127
standard kit was used in these experiments. Bead number 4 in Figure 5-14 was chosen for this
experiment.
Figure 5-15 - Correlation between fluorescence intensity of reporter or detection antibody and target analyte
concentration. In this experiment, bead type number 4 in the BD kit was used. The target analyte was IL-6
cytokine as described in Section 5.3.2 on Page 121.
From the experiments performed in this work, it is clear that a mulplexed beadarray detection
platform can be implemented to detect and measure protein concentrations in a liquid sample.
With a two laser excitation configuration, this platform can detect up to 36 biomarkers or
analytes simultaneously. The detection capacity can be further expanded by introducing
additional laser sources and optical components. A second excitation wavelength at 405 nm can
be added to the optical system to detect and measure additional markers and target analytes. The
current detection platform has a detection limit of <50 pg/mL. Further optimization can be done
to the detection system to improve the detection limit. The laser source used was a green DPSS
laser pointer emitting at 532 nm with 10 mW output power. A more powerful laser source can
significantly improve the fluorescence signal of the beads. Additional optical components can be
y = 0.0015xR² = 0.9957
0
1
2
3
4
5
6
7
8
10 100 1000 10000
Rel
ativ
e Fl
uo
resc
ent
Inte
nsi
ty
IL-6 concentration (pg/mL)
IL-6 calibration curve
128
introduced to improve the signal to background ratio as well. The optical beam of the laser
pointer has a Gaussian profile. For multiplex beadarray application, a uniform illumination is
preferred for data processing. Beam shaping components such as microlens arrays or diffractive
optical element should be used to homogenize the beam intensity to achieve a top-hat profile that
is desired for beadarray detection and fluorescence intensity differentiation.
5.4 Conclusion
An optical imaging methodology for particle detection and analysis that functions much like a
handheld cytometer was proposed and demonstrated. By utilizing a unique wide field dynamic
imaging technique, we have eliminated all moving components in a conventional multi-color
fluorescence detection system, such as filter wheels and rotating stages. These mechanical
components are prone to wear and failure. Extreme mobility in a point of care device is a key
technological advantage. An arrayed set of filters placed in front of an optical detector that
enables multi-color fluorescence detection without excess mechanical components. This multi-
color capability makes our platform versatile and simple to combine/implement new tests.
The imaging and detection platform can not only capture cells, but also smaller biomolecules
such as proteins and DNA/RNA molecules. By using beadarray technology, an ELISA like assay
can be developed on this platform. A proof-of-concept demonstration of this expansion was
presented in the last part of this thesis. IL-6, a blood serum protein biomarker, was measured and
quantified in this experiment. This work would enable the development of a handheld or portable
ELISA system that can conduct much wide range of blood testing at point-of-care, such as
cardiovascular diseases, sepsis and diarrhea diseases just to name a few.
129
Chapter 6 Conclusion and Future Work
A portable imaging and detection system was designed and developed in this work. This system
could be used as the foundation of a handheld point-of-care testing platform for cell, protein and
DNA/RNA analysis. As a demonstration, a CD4 T cell counting test was developed using this
imaging and detection platform. Microfluidic devices were designed, fabricated and
characterized. A direct comparison was also made against the state-of-the-art flow cytometry to
demonstrate the performance of the system.
6.1 Thesis Work Summary
The following list summarizes the key findings and outcomes made in this work:
A capillary microfluidic cell detection and counting device was designed and fabricated. The
microfluidic device relies on surface tension and capillary pumps to actuate sample flow inside
the microchannel. The channel cross section was designed in a way such that uniform flow rate is
maintained throughout the detection and analysis.
Passive and active fluidic actuation mechanisms were investigated in this work and an active
volumetric based actuation approach was designed and developed. A soft elastomer membrane
was used to act as the pneumatic interface between the microfluidic cartridge and the instrument.
The underlying principle was the same as bellows where external force was applied to compress
the air inside the microfluidic device. The reduction in volume resulted in increase in pressure
inside which in turn drove the fluidic sample forward. Similarly, an increase in volume would
lower the pressure inside the microchannel system and hence retract the fluidic sample. With this
approach, it was feasible to actuate fluidic samples inside the microfluidic device backward and
forward as required. Using this approach, an active mixer was designed and fabricated to label
the blood cells with fluorescent dyes on-chip by utilizing Dean’s flow. The active actuation
combined with curved microchannel structures greatly enhanced mixing efficiency. For CD4 T
cell counting test, a conventional labeling process which takes more than 30 minutes in a
benchtop setting was successfully reduced to less than 5 minutes.
130
A common and critical challenge in POC devices and technologies is about functionalization and
reagent incorporation. Usually this step requires cumbersome and complex protocols which
makes the overall microfluidic device design more difficult to manufacture, hence increasing the
cost of production and reduce the reliability. In this work, a novel reagent handling and
integration process was proposed and demonstrated. A plastic reagent plug had a concave cone
shape end where a droplet of reagent solution with a volume of 5 uL was dispensed on. The
reagent was slow-dried to a pellet form, and the plug was then inserted into the microfluidic
cartridge by snap fit or pressure. This entire process did not require any wet chemistry or surface
treatment on microchannels. The reagent plugs can be transferred and inserted into the device
after the entire microfluidic device is fabricated and packaged, which reduced the risks of
damage of antibody affinity during the packing and bonding process. Finally, the entire process
could be automated to reduce the cost improve ease of integration at high volume.
Antibody concentration is another important parameter in the detection of the target
biomolecules. In the CD4 counting test designed in this work, there was no washing steps
involved in order to reduce the complexity of the fluidic system. Hence the signal-to-background
ratio was higher than for a sample with a washing step. To improve the signal-to-background, a
very effective method was to reduce or optimize the background signal. Since the background
was dominated by the blood sample, where unbound antibody was floating freely, the antibody
concentration had a large impact on the assay performance. Larger antibody concentration would
ensure signal from target cells were bright, but at the same time, it would increase the
background fluorescence as excess antibody would result in a higher background signal. An
optimized level of antibody concentration was determined based on the measured signal-to-
background ratio for a number of concentration levels.
An optical imaging methodology for particle detection and analysis that functions much like a
handheld cytometer was proposed and demonstrated. By utilizing a unique wide field dynamic
imaging technique, we have eliminated all moving components in a conventional multi-color
fluorescence detection system, such as filter wheels and rotating stages. These mechanical
components are prone to wear and failure. Extreme mobility in a point of care device is a key
technological advantage. An arrayed set of filters placed in front of an optical detector that
enables multi-color fluorescence detection without excess mechanical components. This multi-
color capability makes our platform versatile and simple to combine/implement new tests.
131
The imaging and detection platform can not only capture cells, but also smaller biomolecules
such as proteins and DNA/RNA molecules. By using beadarray technology, an ELISA like assay
can be developed on this platform. A proof-of-concept demonstration of this expansion was
presented in the last part of this thesis. IL-6, a blood serum protein biomarker, was measured and
quantified in this experiment. This work would enable the development of a handheld or portable
ELISA system that can conduct much wide range of blood testing at point-of-care, such as
cardiovascular diseases, sepsis and diarrhea diseases just to name a few.
6.2 Impact
Access to decentralized diagnostics can save millions of lives globally, yet the lack of
supervision and training opportunities for overburdened nurses and community health workers in
remote settings can negatively affect quality of testing and care. This thesis has built the
foundation of portable or handheld point-of-care testing and diagnostic system that will bring lab
quality clinical diagnostic testing to patients at point of care.
Insufficient access to simple, accurate and affordable diagnostic testing in remote health settings
makes it difficult to provide timely, evidence-based clinical care. The result is millions of
preventable deaths from infectious and non-communicable diseases worldwide, reduced
economic growth and limited human development. Remote health is defined as settings that do
not have access to 24-hour laboratory facilities including low and middle-income countries, rural
Ontario hospitals, aboriginal health settings, intensive care units, ambulances, point of entry
diagnostics and military applications.
The system designed and characterized in this thesis can be further developed into a mobile,
rugged, inexpensive and easy-to-use diagnostic platform that produces laboratory quality results.
From single drops of blood, healthcare workers in remote locations can rapidly and accurately
perform tests to diagnose or monitor a range of infectious and non-communicable diseases. The
platform does not require infrastructure such as refrigeration, stable electricity or highly trained
technicians. Simplicity of use allows community health workers with limited training to test
effectively. Affordability makes the platform cost effective in remote health facilities anywhere
in the world. Its small size and weight facilitate patient testing in remote locations, use in mobile
clinics and deployment in hospitals where a platform can be shared between hospital wards and
outpatient clinics.
132
Currently, in many countries and communities, people do not have access to lifesaving
diagnostic testing because often times there is none: labs that do exist do not operate 24/7 while
sophisticated diagnostic platforms cannot be operated because of lack of trained personnel and
stable electricity and refrigeration, or people live too far from health settings that do offer testing.
The result is millions of preventable deaths from infectious and non-communicable diseases
globally, reduced economic growth, and limited human development. By integrating smart
electronic device functions, including position tracking, bar code scanning and two-way
communication, the envisioned system has the potential to provide a complete and total solution
from patient identification and disease monitoring to patient care, tracking and treatment. This
tool could change the challenges in global health by providing rugged, mobile and simple-to-use
diagnostics in remote health settings that reduce morbidity, decrease healthcare costs, and save
lives.
6.3 Future Outlook
The handheld cytometer developed in this thesis is a very powerful tool. Novel and unique point-
of-care assays can be developed using this platform as demonstrated in the experiments
performed. Additional work is required to further expand the detection capability of the
instrument, as well as development of new assays and tests to target more diseases and
applications.
Further optimization to the optical system can improve the detection limit, sensitivity and
dynamic range, which are all important parameters for many clinical tests. Additional modalities
may also be incorporated into the system such as Raman detection and holographic imaging.
These functionalities could allow more parameters be characterized, reduction of optical
system’s physical size and detection limit.
To develop new and innovative assays, more on-chip fluidic functionalities must also be
researched and developed to satisfy more complex sample preparation processes. On-chip
washing, filtration, dilution, temperature cycling,fluidic valves and more involved fluidic
systems must be incorporated into the current microfluidic architecture to complete the required
steps. In addition, by adding additional sample processing features or capabilities, such as
isothermal amplification, this detection methodology can be useful in developing a portable
nucleic acid test that can be engineered as a molecular diagnostic tool at point-of-care.
133
References
1 Brown, M. & Wittwer, C. Flow cytometry: principles and clinical applications in
hematology. Clinical chemistry 46, 1221-1229 (2000).
2 Marti, G. E., Stetler-Stevenson, M., Bleesing, J. J. & Fleisher, T. A. in Seminars in
hematology. 93-99 (WB Saunders).
3 Aksel, J. H. S. N. Fluid Mechanics. 2 edn, 164-167 (Springer-Belin, 2004).
4 Van Dilla, M. A. Flow cytometry: instrumentation and data analysis. (Academic Pr,
1985).
5 Schall, T. J. Biology of the RANTES/SIS cytokine family. Cytokine 3, 165-183 (1991).
6 Lodish, H. F. et al. Molecular cell biology. Vol. 4 (WH Freeman New York, 2000).
7 Wang, Y.-N. et al. On-chip counting the number and the percentage of CD4+ T
lymphocytes. Lab on a Chip 8, 309-315, doi:10.1039/B713932B (2008).
8 Barat, D., Benazzi, G., Mowlem, M. C., Ruano, J. M. & Morgan, H. Design, simulation
and characterisation of integrated optics for a microfabricated flow cytometer. Optics
Communications 283, 1987-1992 (2010).
9 Martin, A. J. P. & Synge, R. L. M. A new form of chromatogram employing two liquid
phases: A theory of chromatography. 2. Application to the micro-determination of the
higher monoamino-acids in proteins. Biochemical Journal 35, 1358-1368 (1941).
10 Kunkel, H. G. & Tiselius, A. ELECTROPHORESIS OF PROTEINS ON FILTER
PAPER. The Journal of General Physiology 35, 89-118 (1951).
11 Kuschel, M., Buhlmann, C. & Preckel, T. High-throughput protein and DNA analysis
based on microfluidic on-chip electrophoresis. Journal of the Association for Laboratory
Automation 10, 319-326 (2005).
12 Terry, S. C. A gas chromatography system fabricated on a silicon wafer using integrated
circuit technology. PhD thesis, Stanford, (1975).
13 Terry, S. C., Jerman, J. H. & Angell, J. B. A gas chromatographic air analyzer fabricated
on a silicon wafer. Electron Devices, IEEE Transactions on 26, 1880-1886,
doi:10.1109/T-ED.1979.19791 (1979).
14 van Lintel, H. T. G., van De Pol, F. C. M. & Bouwstra, S. A piezoelectric micropump
based on micromachining of silicon. Sensors and Actuators 15, 153-167,
doi:http://dx.doi.org/10.1016/0250-6874(88)87005-7 (1988).
15 Pol, F. C. M. v. d., Wonnink, D. G. J., Elwenspoek, M. & Fluitman, J. H. J. A thermo-
pneumatic actuation principle for a microminiature pump and other micromechanical
devices. Sensors and Actuators 17, 139-143 (1989).
134
16 Smits, J. G., Dalke, S. I. & Cooney, T. K. The constituent equations of piezoelectric
bimorphs. Sensors and Actuators A: Physical 28, 41-61 (1991).
17 Shoji, S., Nakagawa, S. & Esashi, M. Micropump and sample-injector for integrated
chemical analyzing systems. Sensors and Actuators A: Physical 21, 189-192 (1990).
18 Van de Pol, F., Van Lintel, H., Elwenspoek, M. & Fluitman, J. A thermopneumatic
micropump based on micro-engineering techniques. Sensors and Actuators A: Physical
21, 198-202 (1990).
19 Shoji, S., Esashi, M. & Matsuo, T. Prototype miniature blood gas analyser fabricated on a
silicon wafer. Sensors and Actuators 14, 101-107 (1988).
20 Esashi, M. Integrated micro flow control systems. Sensors and Actuators A: Physical 21,
161-167 (1990).
21 van der Schoot, B. H. & Bergveld, P. ISFET based enzyme sensors. Biosensors 3, 161-
186 (1988).
22 Verheggen, T. P., Beckers, J. & Everaerts, F. Simple sampling device for capillary
isotachophoresis and capillary zone electrophoresis. Journal of Chromatography A 452,
615-622 (1988).
23 Thomas, L. J. J. & Bessman, S. P. PROTOTYPE FOR AN IMPLANTABLE
MICROPUMP POWERED BY PIEZOELECTRIC DISK BENDERS. ASAIO Journal
21, 516-522 (1975).
24 Spencer, W. J., Corbett, W. T., Dominguez, L. R. & Shafer, B. D. An electronically
controlled piezoelectric insulin pump and valves. Sonics and Ultrasonics, IEEE
Transactions on 25, 153-156, doi:10.1109/T-SU.1978.31006 (1978).
25 Woias, P. Micropumps—past, progress and future prospects. Sensors and Actuators B:
Chemical 105, 28-38 (2005).
26 Manz, A., Graber, N. & Widmer, H. á. Miniaturized total chemical analysis systems: a
novel concept for chemical sensing. Sensors and actuators B: Chemical 1, 244-248
(1990).
27 Van de Pol, F. C. M., Van Lintel, H. T. G., Elwenspoek, M. & Fluitman, J. H. J. A
thermopneumatic micropump based on micro-engineering techniques. Sensors and
Actuators A: Physical 21, 198-202, doi:http://dx.doi.org/10.1016/0924-4247(90)85038-6
(1990).
28 Manz, A. et al. Planar chips technology for miniaturization and integration of separation
techniques into monitoring systems:: Capillary electrophoresis on a chip. Journal of
Chromatography A 593, 253-258 (1992).
135
29 Harrison, D. J., Manz, A., Fan, Z., Luedi, H. & Widmer, H. M. Capillary electrophoresis
and sample injection systems integrated on a planar glass chip. Analytical chemistry 64,
1926-1932 (1992).
30 Effenhauser, C. S., Manz, A. & Widmer, H. M. Glass chips for high-speed capillary
electrophoresis separations with submicrometer plate heights. Analytical Chemistry 65,
2637-2642 (1993).
31 Harrison, D. J. et al. Micromachining a miniaturized capillary electrophoresis-based
chemical analysis system on a chip. Science 261, 895-897 (1993).
32 Harrison, D. J., Glavina, P. & Manz, A. Towards miniaturized electrophoresis and
chemical analysis systems on silicon: an alternative to chemical sensors. Sensors and
Actuators B: Chemical 10, 107-116 (1993).
33 Seiler, K., Harrison, D. J. & Manz, A. Planar glass chips for capillary electrophoresis:
repetitive sample injection, quantitation, and separation efficiency. Analytical Chemistry
65, 1481-1488 (1993).
34 Northrup, M. A., Ching, M.T., White, R.M., Watson, R.T. in Transducers '93. 924-926.
35 Bousse, L. M., R.J.; Kirk, G.; Dawes, T.; Lam, P.; Bemiss, W.R.; P. J. in Transducers.
916-919.
36 Sobek, D. Y., A.M.; Gray, M.L.; Senturia, S.D. in Micro Electro Mechanical Systems
Proceedings. 219-224.
37 Iliescu, C., Taylor, H., Avram, M., Miao, J. & Franssila, S. Biomicrofluidics 6, 016505
(2012).
38 Washburn, A. L., Luchansky, M. S., Bowman, A. L. & Bailey, R. C. Anal. Chem. 82, 69
(2009).
39 Anderson, R. R. et al. Lab Chip 11, 2088 (2011).
40 Wang, F. & Burns, M. A. Biomed. Microdevices 11, 1071 (2009).
41 Sung, J. H. & Shuler, M. L. Lab Chip 9, 1385 (2009).
42 Chua, J. H., Chee, R. E., Agarwal, A., Wong, S. M. & Zhang, G. J. Anal. Chem. 81, 6266
(2009).
43 Lavrik, N. V., Taylor, L. T. & Sepaniak, M. J. Anal. Chim. Acta 694, 6 (2011).
44 Kutter, J. P. J. Chromatogr., A 1221, 72 (2012).
45 Breadmore, M. C. J. Chromatogr., A 1221, 42 (2012).
46 Kenyon, S. M., Meighan, M. M. & Hayes, M. A. Electrophoresis 32, 482 (2011).
136
47 Angelescu, D. E. Highly Integrated Microfluidics Design. (2011).
48 Mogensen, K. B. & Kutter, J. P. Optical detection in microfluidic systems.
Electrophoresis 30, S92-100, doi:10.1002/elps.200900101 (2009).
49 Baker, C. A., Duong, C. T., Grimley, A. & Roper, M. G. Bioanalysis 1, 967 (2009).
50 Myers, F. B. & Lee, L. P. Lab Chip 8, 2015 (2008).
51 Mitra, I. et al. Anal. Chem. 84, 3621 (2012).
52 Mainz, E. R. et al. Anal. Methods 4, 414 (2012).
53 Pais, A., Banerjee, A., Klotzkin, D. & Papautsky, I. Lab Chip 8, 794 (2008).
54 Au, A. K., Lai, H., Utela, B. R. & Folch, A. Micromachines 2, 179 (2011).
55 Johnson, T. J., Ross, D. & Locascio, L. E. Anal. Chem. 74, 45 (2002).
56 Jeong, G. S., Chung, S., Kim, C. B. & Lee, S. H. Analyst 135, 460 (2010).
57 Nge, P. N., Rogers, C. I. & Woolley, A. T. Advances in microfluidic materials, functions,
integration, and applications. Chemical reviews 113, 2550-2583 (2013).
58 Grzybowski, B. A., Haag, R., Bowden, N. & Whitesides, G. M. Generation of
Micrometer-Sized Patterns for Microanalytical Applications Using a Laser Direct-Write
Method and Microcontact Printing. Analytical Chemistry 70, 4645-4652,
doi:10.1021/ac9807621 (1998).
59 Ho, S., Haque, M., Herman, P. R. & Aitchison, J. S. Femtosecond laser-assisted etching
of three-dimensional inverted-woodpile structures in fused silica. Optics Letters 37,
1682-1684, doi:10.1364/OL.37.001682 (2012).
60 Ho, S., Herman, P. & Aitchison, J. S. Single- and multi-scan femtosecond laser writing
for selective chemical etching of cross section patternable glass micro-channels. Appl.
Phys. A 106, 5-13, doi:10.1007/s00339-011-6675-7 (2012).
61 Berkowski, K. L., Plunkett, K. N., Yu, Q. & Moore, J. S. Introduction to
Photolithography: Preparation of Microscale Polymer Silhouettes. Journal of Chemical
Education 82, 1365, doi:10.1021/ed082p1365 (2005).
62 Smith, B. & Suzuki, K. Microlithography: Science and Technology, 2nd ed. Vol. 126
(CRC Press, 2007, 2007).
63 Sheats, J. R. & Smith, B. W. Microlithography: Science and Technology. (Marcel
Dekker, 1998).
64 Lorenz, H. et al. SU-8: a low-cost negative resist for MEMS. Journal of Micromechanics
and Microengineering 7, 121 (1997).
137
65 Guerin L.J.; Bossel M.; Demierre M., C. S. R. P. in Transducers '97. 1419-1421.
66 Johnson, K. et al. Using neutral metastable argon atoms and contamination lithography to
form nanostructures in silicon, silicon dioxide, and gold. Applied physics letters 69, 2773-
2775 (1996).
67 Xia, Y. & Whitesides, G. M. Soft Lithography. Angewandte Chemie International
Edition 37, 550-575, doi:10.1002/(SICI)1521-3773(19980316)37:5<550::AID-
ANIE550>3.0.CO;2-G (1998).
68 Xia, Y. & Whitesides, G. M. SOFT LITHOGRAPHY. Annual Review of Materials
Science 28, 153-184, doi:doi:10.1146/annurev.matsci.28.1.153 (1998).
69 Quake, S. R. & Scherer, A. From Micro- to Nanofabrication with Soft Materials. Science
290, 1536-1540, doi:10.1126/science.290.5496.1536 (2000).
70 Rogers, J. A. & Nuzzo, R. G. Recent progress in soft lithography. Materials Today 8, 50-
56, doi:http://dx.doi.org/10.1016/S1369-7021(05)00702-9 (2005).
71 Attia, U., Marson, S. & Alcock, J. Micro-injection moulding of polymer microfluidic
devices. Microfluidics and Nanofluidics 7, 1-28, doi:10.1007/s10404-009-0421-x (2009).
72 L., N. M. E. W. W. in Proceedings of SPIE - The International Society for Optical
Engineering.
73 Donggang, Y. & Byung, K. Simulation of the filling process in micro channels for
polymeric materials. Journal of Micromechanics and Microengineering 12, 604 (2002).
74 Wimberger-Friedl, R. Injection Molding of Sub-jam Grating Optical Elements.
Specialized Molding Techniques: Application, Design, Materials and Processing, 149
(2008).
75 Pirskanen, J. et al. Replication of sub-micrometre features using microsystems
technology. Plastics, rubber and composites 34, 222-226 (2005).
76 Mazzeo, A., Dirckx, M. & Hardt, D. in ANTEC-CONFERENCE PROCEEDINGS-.
2931.
77 De Mello, A. Focus: Plastic fantastic? Lab on a Chip 2, 31N-36N (2002).
78 Klaassen, E. H. et al. Silicon fusion bonding and deep reactive ion etching: a new
technology for microstructures. Sensors and Actuators A: Physical 52, 132-139 (1996).
79 Juan, W. & Pang, S. High‐aspect‐ratio Si etching for microsensor fabrication. Journal
of Vacuum Science & Technology A 13, 834-838 (1995).
80 Snow, E., Juan, W., Pang, S. & Campbell, P. Si nanostructures fabricated by anodic
oxidation with an atomic force microscope and etching with an electron cyclotron
resonance source. Applied physics letters 66, 1729-1731 (1995).
138
81 Becker, H. & Gärtner, C. Polymer microfabrication technologies for microfluidic
systems. Analytical and bioanalytical chemistry 390, 89-111 (2008).
82 Becker, H. & Locascio, L. E. Polymer microfluidic devices. Talanta 56, 267-287 (2002).
83 Inc., M. < http://www.micralyne.com> (2012).
84 Gerlach, A. et al. Microfabrication of single-use plastic microfluidic devices for high-
throughput screening and DNA analysis. Microsystem Technologies 7, 265-268 (2002).
85 Erickson, D. & Li, D. Integrated microfluidic devices. Analytica Chimica Acta 507, 11-
26 (2004).
86 Soper, S. A. et al. Surface modification of polymer-based microfluidic devices. Analytica
Chimica Acta 470, 87-99 (2002).
87 Whitesides, G. M. The origins and the future of microfluidics. Nature 442, 368-373
(2006).
88 Soper, S. A. et al. Peer Reviewed: Polymeric Microelectromechanical Systems.
Analytical chemistry 72, 642 A-651 A (2000).
89 Kost, G. J. Guidelines for point-of-care testing. Improving patient outcomes. American
journal of clinical pathology 104, S111-127 (1995).
90 College of American Pathologists POCT toolkit,
<http://www.cap.org/apps/cap.portal?_nfpb=true&cntvwrPtlt_actionOverride=%2Fportle
ts%2FcontentViewer%2Fshow&_windowLabel=cntvwrPtlt&cntvwrPtlt%7BactionForm.
contentReference%7D=committees%2Fpointofcare%2Fpoct_toolkit.html&_state=maxim
ized&_pageLabel=cntvwr> (
91 Publications, T. Point of Care Diagnostic Testing World Markets. 493 (2015).
92 Yager, P., Domingo, G. J. & Gerdes, J. Point-of-care diagnostics for global health. Annu.
Rev. Biomed. Eng. 10, 107-144 (2008).
93 Fu, E., Ramsey, S. A., Kauffman, P., Lutz, B. & Yager, P. Transport in two-dimensional
paper networks. Microfluidics and nanofluidics 10, 29-35 (2011).
94 Martinez, A. W., Phillips, S. T., Butte, M. J. & Whitesides, G. M. Patterned paper as a
platform for inexpensive, low‐volume, portable bioassays. Angewandte Chemie
International Edition 46, 1318-1320 (2007).
95 Martinez, A. W. et al. Programmable diagnostic devices made from paper and tape. Lab
on a Chip 10, 2499-2504 (2010).
96 Chin, C. D., Linder, V. & Sia, S. K. Lab Chip 12, 2118 (2012).
139
97 Daar, A. S. et al. Top ten biotechnologies for improving health in developing countries.
Nature genetics 32, 229-232 (2002).
98 Varmus, H., Klausner, R., Zerhouni, E. & Acharya, T. Grand challenges in global health.
Science 302, 398 (2003).
99 Christianson, A., Streetly, A. & Darr, A. Lessons from thalassaemia screening in Iran:
Screening programmes must consider societal values. BMJ: British Medical Journal 329,
1115 (2004).
100 Sachs, J. & Malaney, P. The economic and social burden of malaria. Nature 415, 680-685
(2002).
101 Laxminarayan, R. et al. Advancement of global health: key messages from the Disease
Control Priorities Project. The Lancet 367, 1193-1208 (2006).
102 Robertson, B. H. & Nicholson, J. K. New microbiology tools for public health and their
implications 1. Annu. Rev. Public Health. 26, 281-302 (2005).
103 Murray, C. J. & Lopez, A. D. Alternative projections of mortality and disability by cause
1990–2020: Global Burden of Disease Study. The Lancet 349, 1498-1504 (1997).
104 Hunter, J. World Bank, World Development Report 1993: Investing in Health. ANNALS-
ASSOCIATION OF AMERICAN GEOGRAPHERS 85, 774-775 (1995).
105 Chin, C. D., Linder, V. & Sia, S. K. Lab-on-a-chip devices for global health: Past studies
and future opportunities. Lab on a Chip 7, 41-57 (2007).
106 Beaglehole, R. & Yach, D. Globalisation and the prevention and control of non-
communicable disease: the neglected chronic diseases of adults. The Lancet 362, 903-908
(2003).
107 Browning, J. D. et al. Prevalence of hepatic steatosis in an urban population in the United
States: impact of ethnicity. Hepatology 40, 1387-1395 (2004).
108 Mellors, J. W. et al. Plasma viral load and CD4+ lymphocytes as prognostic markers of
HIV-1 infection. Annals of internal medicine 126, 946-954 (1997).
109 Badri, M., Lawn, S. D. & Wood, R. Utility of CD4 cell counts for early prediction of
virological failure during antiretroviral therapy in a resource-limited setting. BMC
Infectious Diseases 8, 89, doi:10.1186/1471-2334-8-89 (2008).
110 Peter, T. et al. Challenges in implementing CD4 testing in resource-limited settings.
Cytometry. Part B, Clinical cytometry 74 Suppl 1, S123-130, doi:10.1002/cyto.b.20416
(2008).
111 Thairu, L., Katzenstein, D. & Israelski, D. Operational challenges in delivering CD4
diagnostics in sub-Saharan Africa. AIDS care 23, 814-821,
doi:10.1080/09540121.2010.541416 (2011).
140
112 MASUDO, T. & Okada, T. Ultrasonic radiation-novel principle for microparticle
separation. Analytical Sciences/Supplements 17, i1341-i1344 (2002).
113 Morgan, H., Holmes, D. & Green, N. G. in Nanobiotechnology, IEE Proceedings-. 76-81
(IET).
114 Yu, C. et al. A three-dimensional dielectrophoretic particle focusing channel for
microcytometry applications. Microelectromechanical Systems, Journal of 14, 480-487
(2005).
115 Choi, S. & Park, J.-K. Continuous hydrophoretic separation and sizing of microparticles
using slanted obstacles in a microchannel. Lab on a Chip 7, 890-897,
doi:10.1039/B701227F (2007).
116 Choi, S., Song, S., Choi, C. & Park, J.-K. Continuous blood cell separation by
hydrophoretic filtration. Lab on a Chip 7, 1532-1538, doi:10.1039/B705203K (2007).
117 Choi, S., Song, S., Choi, C. & Park, J.-K. Sheathless Focusing of Microbeads and Blood
Cells Based on Hydrophoresis. Small 4, 634-641, doi:10.1002/smll.200700308 (2008).
118 Haeberle, S. & Zengerle, R. Microfluidic platforms for lab-on-a-chip applications. Lab on
a Chip 7, 1094-1110, doi:10.1039/B706364B (2007).
119 Squires, T. M. & Quake, S. R. Microfluidics: Fluid physics at the nanoliter scale. Reviews
of Modern Physics 77, 977-1026 (2005).
120 Ichikawa, N., Hosokawa, K. & Maeda, R. Interface motion of capillary-driven flow in
rectangular microchannel. Journal of Colloid and Interface Science 280, 155-164,
doi:http://dx.doi.org/10.1016/j.jcis.2004.07.017 (2004).
121 de Gennes, P.-G. B.-W., Francoise; Quere, David. Capillarity and Wetting Phenomena.
(Springer-Verlag New York, 2004).
122 Ajaev, V. S. & Homsy, G. M. Steady Vapor Bubbles in Rectangular Microchannels.
Journal of Colloid and Interface Science 240, 259-271,
doi:http://dx.doi.org/10.1006/jcis.2001.7562 (2001).
123 Ajaev, V. S. in Interfacial Fluid Mechanics 125-140 (Springer US, 2012).
124 Ajaev, V., Gatapova, E. & Kabov, O. A. Two-phase viscous flows in channels with
chemically patterned walls. Bulletin of the American Physical Society 58 (2013).
125 Mohseni, K., Arzpeyma, A. & Dolatabadi, A. Behaviour of a Moving Droplet under
Electrowetting Actuation: Numerical Simulation. The Canadian Journal of Chemical
Engineering 84, 17-21, doi:10.1002/cjce.5450840104 (2006).
126 Mukhopadhyay, S., Roy, S. S., Mathur, A., Tweedie, M. & McLaughlin, J. A.
Experimental study on capillary flow through polymer microchannel bends for
141
microfluidic applications. Journal of Micromechanics and Microengineering 20, 055018
(2010).
127 Zimmermann, M., Schmid, H., Hunziker, P. & Delamarche, E. Capillary pumps for
autonomous capillary systems. Lab on a Chip 7, 119-125 (2007).
128 Watson, J. V. Introduction to flow cytometry. (Cambridge University Press, 2004).
129 Raveche, E. et al. Introduction to Flow Cytometry. Flow Cytometry in Drug Discovery
and Development, 3-21 (2011).
130 Alberts, B. et al. Molecular Biology of the Cell (3rd edn). Trends in Biochemical
Sciences 20, 210-210 (1995).
131 Cooper, G. M. & Hausman, R. E. The cell. (Sinauer Associates Sunderland, 2000).
132 Juncker, D. et al. Autonomous microfluidic capillary system. Analytical chemistry 74,
6139-6144 (2002).
133 Dalgaard, P. Introductory Statistics with R. 2nd Edition edn, (Springer, 2008).
134 Mandel, J. The Statistical Analysis of Experimental Data (Dover Publications, 1984).
135 Zimmermann, M., Hunziker, P. & Delamarche, E. Valves for autonomous capillary
systems. Microfluidics and Nanofluidics 5, 395-402 (2008).
136 Klein, R. Material Properties of Plastics. 1st Edition edn, (Wiley-VCH Verlag GmbH &
Co. KGaA, 2011).
137 Malloy, R. A. Plastic Part Design for Injection Molding. 2nd edn, (Hanser Gardner
Publisher, 2010).
138 Pishro-Nik, H. Introduction to Probability, Statistics and Random Processes. (Kappa
Research, LLC, 2014).
139 Lam, Y. C., Gan, H. Y., Nguyen, N. T. & Lie, H. Micromixer based on viscoelastic flow
instability at low Reynolds number. Biomicrofluidics 3, 014106, doi:10.1063/1.3108462
(2009).
140 Bryce, D. M. Plastic Injection Molding: Manufacturing Process Fundamentals. (Society
of Manufacturing Engineers, 1996).
141 Rosato, D. V., Rosato, D. V. & Rosato, M. G. Injection Molding Handbook. (Kluwer
Academic Publishers, 2000).
142 Wikipedia. Injection Molding,
<http://en.wikipedia.org/wiki/Injection_moulding#cite_note-2> (2015).
142
143 Lee, C.-Y., Chang, C.-L., Wang, Y.-N. & Fu, L.-M. Microfluidic Mixing: A Review.
International Journal of Molecular Sciences 12, 3263-3287, doi:10.3390/ijms12053263
(2011).
144 Buchegger, W., Wagner, C., Lendl, B., Kraft, M. & Vellekoop, M. J. A highly uniform
lamination micromixer with wedge shaped inlet channels for time resolved infrared
spectroscopy. Microfluidics and Nanofluidics 10, 889-897 (2011).
145 Tofteberg, T., Skolimowski, M., Andreassen, E. & Geschke, O. A novel passive
micromixer: lamination in a planar channel system. Microfluidics and Nanofluidics 8,
209-215 (2010).
146 CY Lee, C. L., MF Hung, T Chin, LM Fu. Experimental and numerical investigation into
mixing efficiency of micromixers with different geometric barriers. Material Science
Forum, 391-396 (2006).
147 Chen, Z. et al. Performance analysis of a folding flow micromixer. Microfluidics and
nanofluidics 6, 763-774 (2009).
148 Kang, T. G., Singh, M. K., Anderson, P. D. & Meijer, H. E. A chaotic serpentine mixer
efficient in the creeping flow regime: from design concept to optimization. Microfluidics
and nanofluidics 7, 783-794 (2009).
149 Moon, D. & Migler, K. B. Forced assembly and mixing of melts via planar polymer
micro-mixing. Polymer 51, 3147-3155, doi:10.1016/j.polymer.2010.04.070 (2010).
150 Neerincx, P. E., Denteneer, R. P. J., Peelen, S. & Meijer, H. E. H. Compact Mixing Using
Multiple Splitting, Stretching, and Recombining Flows. Macromolecular Materials and
Engineering 296, 349-361, doi:10.1002/mame.201000338 (2011).
151 C.H. Lin, C. H. T., L.M. Fu. A rapid three-dimensional vortex micromixer utilizing self-
rotation effects under low Reynolds number conditions. Journal of Micromechanics and
Microengineering 15, 935-943 (2005).
152 Singh, M. K., Anderson, P. D. & Meijer, H. E. Understanding and Optimizing the SMX
Static Mixer. Macromolecular rapid communications 30, 362-376,
doi:10.1002/marc.200800710 (2009).
153 Tsai, R.-T. & Wu, C.-Y. An efficient micromixer based on multidirectional vortices due
to baffles and channel curvature. Biomicrofluidics 5, 014103, doi:10.1063/1.3552992
(2011).
154 Hardt, S., Pennemann, H. & Schönfeld, F. Theoretical and experimental characterization
of a low-Reynolds number split-and-recombine mixer. Microfluidics and Nanofluidics 2,
237-248 (2006).
155 Jain, M., Yeung, A. & Nandakumar, K. Induced charge electro osmotic mixer: Obstacle
shape optimization. Biomicrofluidics 3, 022413, doi:10.1063/1.3167279 (2009).
143
156 Jain, M. & Nandakumar, K. Novel index for micromixing characterization and
comparative analysis. Biomicrofluidics 4, 031101, doi:10.1063/1.3457121 (2010).
157 Daniel Ahmed, X. M., Bala Krishna Juluri, Tony Jun Huang. A fast microfluidic mixer
based on acoustically driven sidewall-trapped microbubbles. Microfluidics and
Nanofluidics, 727-731, doi:10.1007/s10404-009-0444-3 (2009).
158 T-D. Luong, V.-N. P., N-T. Nguyen. High-throughput micromixers based on acoustic
streaming induced by surface acoustic wave. Microfluidics and Nanofluidics 10, 619-625,
doi:10.1007/s10404-010-0694-0 (2010).
159 Michele Campisi, D. A., Francesco Damiani, Paolo Dario. A soft-lithographed chaotic
electrokinetic micromixer for efficient chemical reactions in lab-on-a-chips. Journal of
Micro-Nano Mechatronics 5, 69-76, doi:10.1007/s12213-010-0024-3 (2010).
160 Lim, C. Y., Lam, Y. C. & Yang, C. Mixing enhancement in microfluidic channel with a
constriction under periodic electro-osmotic flow. Biomicrofluidics 4, 014101,
doi:10.1063/1.3279790 (2010).
161 Du, Y., Zhang, Z., Yim, C., Lin, M. & Cao, X. A simplified design of the staggered
herringbone micromixer for practical applications. Biomicrofluidics 4, 024105,
doi:10.1063/1.3427240 (2010).
162 Zhang, Z., Yim, C., Lin, M. & Cao, X. Quantitative characterization of micromixing
simulation. Biomicrofluidics 2, 034104, doi:10.1063/1.2966454 (2008).
163 Xu, B., Wong, T. N., Nguyen, N.-T., Che, Z. & Chai, J. C. K. Thermal mixing of two
miscible fluids in a T-shaped microchannel. Biomicrofluidics 4, 044102,
doi:10.1063/1.3496359 (2010).
164 Y. Wang, J. Z., B. T. F. Chung, P. Dutta. A rapid magnetic particle driven micromixer.
Microfluidics and Nanofluidics, 375-389, doi:10.1007/s10404-007-0188-x (2008).
165 Huang, M.-Z., Yang, R.-J., Tai, C.-H., Tsai, C.-H. & Fu, L.-M. Application of
electrokinetic instability flow for enhanced micromixing in cross-shaped microchannel.
Biomed Microdevices 8, 309-315, doi:10.1007/s10544-006-0034-z (2006).
166 Di Carlo, D. Inertial microfluidics. Lab on a Chip 9, 3038-3046 (2009).
167 Dean, W. R. Note on the motion of fluid in a curved pipe. Philosophical magazine series
7 4, 208-223 (1927).
168 S.A. Berger, L. T., L.S. Yao. Flow in curved pipes. Annual Review of Fluid Mechanics
15, 461-512 (1983).
169 Bringer, M. R., Gerdts, C. J., Song, H., Tice, J. D. & Ismagilov, R. F. Microfluidic
systems for chemical kinetics that rely on chaotic mixing in droplets. Philosophical
transactions. Series A, Mathematical, physical, and engineering sciences 362, 1087-
1104, doi:10.1098/rsta.2003.1364 (2004).
144
170 Dinh, T. X. & Ogami, Y. Mixing Enhancement by Microrotor in Step Channel. Journal
of Fluids Engineering 133, 021101-021101, doi:10.1115/1.4003420 (2011).
171 Ottino, J. M. Mixing, Chaotic Advection, And Turbulence. Annual Review of Fluid
Mechanics 22, 207-253 (1990).
172 Liu, R. H. et al. Passive mixing in a three-dimensional serpentine microchannel.
Microelectromechanical Systems, Journal of 9, 190-197 (2000).
173 R.C. Anderson, G. J. B., A. Puski, X. Su. in Proceedings Solid-State Sensors and
Actuator Workshop. 7-10.
174 Jones, S. W., Thomas, O. M. & Aref, H. Chaotic advection by laminar flow in a twisted
pipe. Journal of Fluid Mechanics 209, 335-357, doi:doi:10.1017/S0022112089003137
(1989).
175 Aref, H. Stirring by chaotic advection. Journal of Fluid Mechanics 143, 1-21,
doi:doi:10.1017/S0022112084001233 (1984).
176 Aref, H. Chaotic Advection of Fluid Particles. Philosophical Transactions of the Royal
Society of London A: Mathematical, Physical and Engineering Sciences 333, 273-288
(1990).
177 Lin, B. et al. (Google Patents, 2010).
178 Zhu, H., Mavandadi, S., Coskun, A. F., Yaglidere, O. & Ozcan, A. Optofluidic
Fluorescent Imaging Cytometry on a Cell Phone. Analytical Chemistry 83, 6641-6647,
doi:10.1021/ac201587a (2011).
179 Yang, S.-Y. et al. A cell counting/sorting system incorporated with a microfabricated
flow cytometer chip. Measurement Science and Technology 17, 2001 (2006).
180 Fu, A. Y., Spence, C., Scherer, A., Arnold, F. H. & Quake, S. R. A microfabricated
fluorescence-activated cell sorter. Nature biotechnology 17, 1109-1111 (1999).
181 Godin, J. et al. Microfluidics and photonics for Bio‐System‐on‐a‐Chip: A review
of advancements in technology towards a microfluidic flow cytometry chip. Journal of
biophotonics 1, 355-376 (2008).
182 Benton, S. A. & Bove Jr, V. M. Holographic imaging. (John Wiley & Sons, 2008).
183 Collier, R. Optical holography. (Elsevier, 2013).
184 Hariharan, P. Optical Holography: Principles, techniques and applications. Vol. 20
(Cambridge University Press, 1996).
185 Marquet, P. et al. Digital holographic microscopy: a noninvasive contrast imaging
technique allowing quantitative visualization of living cells with subwavelength axial
accuracy. Optics letters 30, 468-470 (2005).
145
186 Sung, Y. et al. Optical diffraction tomography for high resolution live cell imaging.
Optics express 17, 266-277 (2009).
187 Coskun, A. F., Su, T.-W. & Ozcan, A. Wide field-of-view lens-free fluorescent imaging
on a chip. Lab on a Chip 10, 824-827, doi:10.1039/B926561A (2010).
188 Vykoukal, J., Vykoukal, D. M., Stone, G. P. & Alt, E. U. (Google Patents, 2010).
189 Dalsa. CCD vs CMOS, <https://www.teledynedalsa.com/imaging/knowledge-
center/appnotes/ccd-vs-cmos/> (2015).
190 Access, I. CCD or CIS: The Technology Decision. 6 (2012).
191 Chartier, G. Introduction to optics. (Springer Science & Business Media, 2005).
192 Pedrotti, F. L. & Pedrotti, L. S. Introduction to Optics 2nd Edition. Introduction to Optics
2nd Edition by Frank L. Pedrotti, SJ, Leno S. Pedrotti New Jersey: Prentice Hall, 1993 1
(1993).
193 Luminex xMAP technology, <https://www.luminexcorp.com/clinical/our-
technology/xmap-technology/> (2015).
194 Wikipedia. Interleukin, <http://en.wikipedia.org/wiki/Interleukin> (2015).
195 Schindler, R. & Dinarello, C. A. in Growth Factors, Differentiation Factors, and
Cytokines (ed Andreas Habenicht) Ch. 7, 85-102 (Springer Berlin Heidelberg, 1990).
196 di Giovine, F. S. & Duff, G. W. Interleukin 1: the first interleukin. Immunology Today 11,
13-20, doi:http://dx.doi.org/10.1016/0167-5699(90)90005-T (1990).
197 Lackie, J. A dictionary of biomedicine. (Oxford University Press, 2010).
198 McCoy, C. E. 1. Abstract 2. Introduction 3. Cytokine regulation of miRNA 3.1.
Interleukin-1 3.2. Tumor necrosis factor 3.3. Interleukin-6. Frontiers in Bioscience 16,
2161-2171 (2011).
199 Pedersen, B. K., Åkerström, T. C., Nielsen, A. R. & Fischer, C. P. Role of myokines in
exercise and metabolism. Journal of applied physiology 103, 1093-1098 (2007).
200 Ferguson-Smith, A. C. et al. Regional localization of the interferon-β2B-cell stimulatory
factor 2/hepatocyte stimulating factor gene to human chromosome 7p15-p21. Genomics
2, 203-208 (1988).
201 North, R. An introduction to macrophage activation. Lymphokines 3, 1-10 (1981).
202 Zwilling, B. S. & Eisenstein, T. K. Macrophage-pathogen interactions. (M. Dekker,
1994).
203 Natsuka, S. et al. Macrophage differentiation-specific expression of NF-IL6, a
transcription factor for interleukin-6. Blood 79, 460-466 (1992).
146
204 Cromwell, O. et al. Expression and generation of interleukin-8, IL-6 and granulocyte-
macrophage colony-stimulating factor by bronchial epithelial cells and enhancement by
IL-1 beta and tumour necrosis factor-alpha. Immunology 77, 330 (1992).
205 van der Poll, T. et al. Interleukin-6 gene-deficient mice show impaired defense against
pneumococcal pneumonia. Journal of Infectious Diseases 176, 439-444 (1997).