Microfluidic cell sorting with a high frequency ultrasound beam

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content

MICROFLUIDIC CELL SORTING
WITH A HIGH FREQUENCY ULTRASOUND BEAM

by

Changyang Lee

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillments of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOMEDICAL ENGINEERING)

December 2014

Copyright 2014 Changyang Lee
ii

DEDICATION

With genuine thanks to my beloved parents, Seongsik Lee and Yeongja Kwon;
my precious sister and brother, Myounghee Lee and Changgak Lee
for their unconditional and endless love and unflagging support

iii

ACKNOWLEDGEMENTS
Over the past six years I have received tremendous support and encouragement from a
great number of individuals directly or indirectly. In retrospect, it would have been
impossible for me to finish without their help. Thank you.
During these graduate years at University of Southern California, it has always been my
pleasure to work with good people and colleagues in NIH Resource Center of Medical
Ultrasonic Transducer Technology. I have been able to have countless opportunities
making for a full and enriching experience. In the first place, I would like to show my
deepest appreciation and admiration to my supervisor Dr. K. Kirk Shung whose warm
encouragement, excellent support, and thoughtful guidance from the initial to end of this
dissertation. His guidance has made this a thoughtful and rewarding journey for me.
Especially, I have been inspiring and enriching my growth as a Ph.D. from his
generousness, patience and his trust.
I gratefully acknowledge my committee members Dr. K. Kirk Shung, Dr. Megan McCain,
and Sc.D. Herbert Meiselman for their insightful comments to allow me to finish this
dissertation. In addition, I would like to express my thanks and appreciation to all my lab
members for their friendship as well as discussion on my research. Dr. Jungwoo Lee
deserves my special thanks for his advice and support during graduate years. I also thank
Dr. Hyung Ham Kim, Dr. Jin Ho Chang, Dr. Jae Yooon Hwang, and Dr. Jong-Seob
Jeong for their encouragement and warm advice during my graduate study and life. Dr.
Jinhyoung Park and Mr. Bong Jin Kang fertilized my life with their helps and truthful
iv

words. Lunch time and soccer games with Dr. Sangpil Yoon, Dr. Changhan Yoon, Mr.
Hayong Jung, Mr. Chi Woo Yoon, Mr. Mingon Kim, Mr. Hae Gyun Lim, and Mr. Kyo
Suk Goo were a fountain of joy, concern, and new ideas for all of us. I would like to
thank Dr. Ying Li, Dr. Kwok Ho Lam, Mr. Nestor E. Cabrera-Munoz, and Chi Tat (Harry)
Chiu for their support.
My thanks also go to Dr. Abraham Lee at University of California at Irvine and Dr.
Yingxiao Wang at University of California at San Diego for their precious advice and
support on this dissertation and other researches.
I give my thanks to Dr. Eunjin (Grace) Lee, Dr. Yongjin Cho, and Dr. Soyun Kim.
During my early years in California, they have given my both academic advice and
wisdom to live in Los Angeles.
I would like to give my heartily thanks to my adored parents and precious sister and
brother for their endless love and unconditional devotion. I could not have completed any
piece of accomplishment without my family’s support and love. I specially wish to
express my deepest thanks to my mother, Yeongja Kwon. Since August 2008, my mother
has been always with me as a mentor, strong supporter, and friend. I always feel sorry not
showing any reward for her sacrifice except for this humble dissertation. My mother has
been sick for a long time and has always taken care of my family even with her critical
conditions. I wish that her health may improve even a little and hope that she would also
take care of herself and spend time for her own as well. I love you. I also thank my father,
v

Seongsik Lee, and my precious sister and brother, Myounghee Lee and Changgak Lee for
their unconditional and endless love and unflagging support. I love my family.
Lastly, I would like to dedicate this dissertation to Pastor’s family, Pastor, Seung-ho Synn,
his wife, Nackkum Synn, and Heidi for supporting me to settle down in Los Angeles and
University Southern California. Without his support I could not have continued to study
in the states and would have returned back home. I also give my thanks to grandmother,
Subok (Sue) Kim, as my great supporter in Los Angeles who gave me her strength and
experience for my life in the United States.
Life with The Well’s family has always been a great joy and refreshment when I was in
stress. I give my thanks to the Choi’s sisters, Ms. Jung Hyun Choi, Ms. Hae Lin Choi, Ms.
Green Choi, and to Mr. Jingul Kim, Mr. Haejun Lee, Mr. Kyo Suk Goo, Mr. Woonkee
Kim, Mr. Juman Lee, Ms. GaHyun Cho, Ms. Yura Lee.
Everyone has always given me their unquestioning faith and has encouraged me to keep
moving forward to my dream. I would like to dedicate this dissertation to you all of you
with my heart.

vi

TABLE OF CONTENTS
DEDICATION .................................................................................................................... ii
ACKNOWLEDGEMENTS ............................................................................................... iii
LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... ix
ABSTRACT ..................................................................................................................... xvi
CHAPTER 1 INTRODUCTION ..................................................................................... 1
1.1 High Frequency Ultrasound ................................................................................. 1
1.2 Micro Total Analysis Systems ............................................................................. 6
1.3 Cell Sorting .......................................................................................................... 8
1.4 Objective of Research ........................................................................................ 13
1.5 Overview ............................................................................................................ 15
CHAPTER 2 HIGH FREQUENCY ULTRASOUND CELL SORTING DEVICE ..... 17
2.1 Introduction ........................................................................................................ 17
2.2 Material and Methods......................................................................................... 20
2.2.1 Device Concept and Mechanism ................................................................ 20
2.2.2 High Frequency Ultrasound Transducer ..................................................... 23
2.2.3 Synthesis of Lipid Micro Droplets .............................................................. 27
2.2.4 Microfluidic Channel Design and Fabrication ............................................ 29
2.2.5 Experimental Configuration........................................................................ 31
2.2.6 Experimental Procedure .............................................................................. 32
2.3 Results and Discussion ....................................................................................... 35
2.3.1 Sensing Sensitivity ...................................................................................... 35
2.3.2 Sensing Efficiency ...................................................................................... 36
2.4 Conclusion .......................................................................................................... 40
CHAPTER 3 SIZE-BASED MICROFLUIDIC DROPLET SORTING WITH HIGH
FREQUENCY ULTRASOUND BEAM ................................................ 41
3.1 Introduction ........................................................................................................ 41
3.2 Material and Methods......................................................................................... 44
3.2.1 Synthesis of Lipid Micro Droplets .............................................................. 44
3.2.2 Microfluidic Channel Design and Micro Flow Cytometry ......................... 45
3.2.3 Multi-parameter Sorting using Integrated Backscatter Coefficient ............ 50
3.2.4 Experimental Configuration and Procedure................................................ 53
vii

3.3 Results and Discussion ....................................................................................... 57
3.4 Conclusion .......................................................................................................... 65
CHAPTER 4 MATERIAL PROPERTIES-BASED MICROFLUIDIC MICRO-
PARTICLE SENSING ............................................................................ 66
4.1 Introduction ........................................................................................................ 66
4.2 Material and Methods......................................................................................... 69
4.2.1 High Frequency Ultrasound Transducer ..................................................... 69
4.2.2 Polymer Micro Spheres and Lipid Droplets ............................................... 71
4.2.3 Ultrasonic Spectral Parameters ................................................................... 72
4.2.4 Experimental Configuration and Procedure................................................ 74
4.3 Results and Discussion ....................................................................................... 76
4.4 Conclusion .......................................................................................................... 79
CHAPTER 5 A LIVE SINGLE CELL SENSING AND DISCRIMINATION ............ 80
5.1 Introduction ........................................................................................................ 80
5.2 Material and Methods......................................................................................... 80
5.2.1 Live Cells Preparation................................................................................. 80
5.2.2 High Frequency Ultrasound Cell Sorting System ....................................... 81
5.2.3 Experimental Configuration and Procedure................................................ 87
5.3 Results and Discussion ....................................................................................... 91
5.4 Conclusion ........................................................................................................ 100
CHAPTER 6 SUMMARY AND FUTURE WORKS................................................. 101
6.1 Summary .......................................................................................................... 101
6.2 Future Works .................................................................................................... 103
BIBLOGRAPHY ............................................................................................................ 106

viii

LIST OF TABLES
Table 3.1: Summarized results for peak-to-peak echo amplitude, integrated backscatter
coefficient, sensing and sorting efficiencies for 50 µm/100 µm droplets ......................... 61

ix

LIST OF FIGURES
Figure 1.1: The scientifically useful frequency spectrum of acoustic excited or detected
elastic waves. (White, 1997) ............................................................................................... 2

Figure 1.2: A classification scheme for acoustic waves by frequencies. (Maldovan, 2013)
............................................................................................................................................. 2

Figure 1.3: Ultrasound medical applications at low frequencies range (Mitragotri, 2005) 4

Figure 1.4: Macro-scale clinical laboratory and micro-scale systems. (a) conventional
clinical laboratory (Courtesy of Stony Brook, Health Technology and Management), (b) a
molecular diagnostic system by Siemens, the fraunhofer Institutem and Infineon
Technologies (Courtesy of Medical Technology Business Europe) and (c) New microchip
technology performs 1,000 chemical reactions at once (Courtesy of UCLA Newsroom) . 7

Figure 1.5: Conventional cell separation and sorting methods. (a) bulk separation
methods (magnetic separation (left) and centrifuge methods (right), (b) single-cell-based
sorting methods ................................................................................................................... 9

Figure 1.6: Fluorescence-Activated Cell Sorter (FACS). (a) Optical structure and sorting
feature of FACS (Recktenwald et al., 1998), (b) early stage FACS model manufactured
by the Becton Dickinson Electronics Laboratory (Herzenberg et al., 1976), and (c)
current FACS commercial product by Beckman Coulter (Courtesy of Beckman Coulter,
Inc.) ................................................................................................................................... 11

Figure 2.1: A concept diagram of high frequency ultrasound sorting device. The device
uses echo signals for sensing and its analysis by calculating ultrasonic spectral parameters
provides guide information for recognizing bioparticles. After making decision from
analyzing echo, acoustic radiation force is used for sorting and separation of
subpopulation of sample by pushing. A single high frequency ultrasound transducer
x

combines sensing with sorting for the advantages of being more simple and compact
device. ............................................................................................................................... 21

Figure 2.2: Device schematics and mechanism. High frequency ultrasound transducer is
placed on outside of microchannel. Position of focus of transducer is important for better
echo signals from the bio-particles. (a ~d, f) show sensing mechanism of device and (e)
illustrates its sorting mechanism. (a) A red sample object is approaching to sensing zone
in sensing mode of device. (b)(c) Echo singles of the red object are received and analyzed,
and if the one is not targeted, it passed through to zone. (d) When a targeted green object
is placed in sensing area, the system senses targeted one by calculated ultrasonic spectral
parameters, such as IB value, slope, intercept and midband fit. (e) The system changes
operation mode from sensing to sorting, and generate acoustic radiation force for pushing
the green one. (f) Then, the device comes back to sensing mode for continuously working
as a sorting device. These processes are automatically controlled by customized
LabVIEW program. .......................................................................................................... 22

Figure 2.3: Schematic diagram of a 30MHz LiNbO
3
single element ultrasonic transducer
and fabricated transducer. The transducer consists of a LiNbO
3
piezoelement, a backing
material, and two matching layer for sound transmission. ............................................... 23

Figure 2.4: Pulse-echo waveform from quartz at the focus of 30 MHz and its frequency
response............................................................................................................................. 25

Figure 2.5: Lateral pressure beam profiles of 30 MHz transducer. The measured 3 dB
beam width is 36 µm and peak pressure is 3.24 MPa at the focus of the transducer. ...... 26

Figure 2.6: (left) Oleic acid droplet formation microfluidic device. The two phases, oleic
acid and 5% pluronic solution, meet at the flow focusing junction where oleic acid is
continuously sheared into droplets by the two aqueous phases. (right) Monodispersed
population of oleic acid droplets dyed oil Red (Dupont, Wilmington, DE, USA) in
visualization (Lee et al., 2010a) ........................................................................................ 28

xi

Figure 2.7: Microfluidic channel fabricated in poly(dimethyl) siloxane (PDMS) using
conventional soft lithography techniques ......................................................................... 29

Figure 2.8: Experimental configurations for high frequency ultrasound cell sorting device
........................................................................................................................................... 32

Figure 2.9: The sensing zone of acoustic particle sorting. (a) the particles pass through
portion of the sample solution with width of about 80 µm and flow rate of 1 µl/min. (b)
focal point of the transducer are located at the point which are moved with distance 25
µm from left wall of PDMS channel and distance of 20 µm above center line of the
channel in order to be positioned at the center of sensing zone........................................ 34

Figure 2.10: Scattered echo signals from lipid droplets. (a) Reflected signal from channel
well (b) 30 µm lipid droplet echo signal (c) 75 µm lipid droplet echo signal .................. 37

Figure 2.11: A mode signals difference between 30 µm and 75 µm. The peak-to-peak
voltage levels made from reflection at surface of lipid particles had different signal
amplitude, which enabled to distinguish size of lipid droplets. The A mode signals from
small particles of 30 µm, was 7.9 ± 1.9 mV
pp
, while the reflected signals levels of large
particles of 75 µm was 16.4 ± 1.6 mV
pp
, both based on A mode signals from a single
excitation condition. .......................................................................................................... 38

Figure 2.12: Sensing efficiency of each particle - small particles (30 µm) / large particles
(75 µm). The detecting efficiency was calculated based on this data, which showed a
detecting sensing rate of 52.50 ± 7.68% for the 30 μm droplets and 79.88 ± 7.82% for the
75 μm droplets. ................................................................................................................. 39

Figure 3.1: Histogram of size distribution of droplet sample of (a) 50 µm and (b) 100 µm
in diameter. ....................................................................................................................... 44

Figure 3.2: Layout of the PDMS microfluidic device fabricated by conventional soft
lithography and the microfluidic channel wall thickness. (a) The upper and lower inlets
xii

provide sheath (or guiding) flows that hydrodynamically focus the droplet stream to the
center of the channel, while the middle inlet delivers the droplet flow. (b) The channel
wall thickness measured by echo signal for positioning ultrasound transducer was 148
µm ..................................................................................................................................... 47

Figure 3.3: Plot of focused stream width as a function of the ratio of sheath flow rate (fsh)
to droplet flow rate (fd). (a) fsh is adjusted from 2 µl/min to 14 µl/min with a step
increment of 2 µl/min, whereas fd remains constant at 2 µl/min (b) The stream width for
uniform droplet flow is found to be 60 µm when fd = 2 µl/min and fsh = 6 µl/min, as a
triangular marker indicates in (a). ..................................................................................... 49

Figure 3.4: (a) Experimental set-up for IB measurement (Lee et al., 2011b) (b)
Comparison of IB coefficients between experimental results and theoretical T-matrix
values (Lee et al., 2011b) IB coefficients were experimentally measured as a function of
droplet size. The mean IB coefficient of a single droplet was -107 ± 0.9 dB for cluster A,
and -93 ± 0.9 dB for cluster B. The measured results were in agreement with the
theoretical values based on the T-matrix method. The results indicate that the IB
coefficient may be used as a useful index to distinguish one particle from the others based
on several physical properties of the particle, including size and acoustic impedance. This
work was performed as an initial attempt to realize ultrasonic cell sorting in which both
cell sensing and sorting are carried out by sound beams. ................................................. 52

Figure 3.5: Experimental arrangement for acoustic droplet sensing and sorting device. (a)
Top view (b) side view. .................................................................................................... 55

Figure 3.6: Experimental set up ........................................................................................ 56

Figure 3.7: Illustration of (a) sensing and (b) sorting signals driving the transducer. ...... 56

Figure 3.8: Acquired echo signals and their IB coefficients for droplets passing through
the sensing beam are plotted over time. ............................................................................ 58

xiii

Figure 3.9: Relation between echo amplitudes and integrated backscatter (IB) coefficients
corresponding to differently sized droplets ....................................................................... 59

Figure 3.10: The droplet size is more linearly related to the IB coefficient than to the echo
amplitude........................................................................................................................... 60

Figure 3.11: The 100 µm droplets sorting processes from a mixture of different size
droplets. (a) The targeted 100 µm droplet is approaching to defined sensing zone. (b)
When the droplet passes through the sensing zone in red box, the 100 µm droplet is
pushed by acoustic radiation force. In the contrast, non-targeted droplet, 50 µm droplet
(yellow triangle mark) is not sorted, following the sensing zone. Note that green arrow
indicates sample flow, and blue arrow shows waste channel, and red arrow means
collection channel. ............................................................................................................ 62

Figure 4.1: Pulse-echo signal and its frequency spectrum analysis and fabricated high
frequency single element ultrasound transducer ............................................................... 70

Figure 4.2: Histogram of size distribution of micro lipid droplet sample of 45 µm in mean
diameter (CV < 17%). ....................................................................................................... 72

Figure 4.3: The calculation processes of ultrasonic spectral parameters. (a) RF spectrum
from sample (solid line) and glass plate spectrum for calibration procedure from a flat
reflector (dash line) (b) Normalized calibrated spectrum (solid line) calculated by RF
spectrum over glass plate spectrum in spectral band width and linear regression line (dash
line) (c) Spectral slope (slope of regression line) and intercept (INT) at zero frequency by
extrapolation, and midband fit (MBF) at center frequency, f
0
. Note the Figure 4.3 is
modified from Figure 3 and 4 (Lizzi et al., 2003) and Figure 2(b) (Lizzi et al., 2006). ... 74

Figure 4.4: Calibration of transducers initial position through pulse-echo signals. The
channel wall thickness by echo signal was 148 µm. ......................................................... 75

xiv

Figure 4.5: Integrated Backscatter coefficient of lipid and polystyrene micro-particle. The
IB coefficient was -79.37 ± 3.4 dB for 45 μm lipid droplets, while it was -76.40 ± 3.07
dB for 45 μm polystyrene micro-beads, which was not able to clearly discriminate
different material properties in similar diameter size in Figure 4.5. Note the graph was
drawn by positive value for better appearance. ................................................................ 77

Figure 4.6: Spectral slope of each particle.The spectral slope value in dB/MHz was
3.04E-07 ± 1.19E-07 for 45 μm lipid droplets while it was -2.08E-07 ± 1.13E-07 for 45
μm polystyrene micro-beads. Those droplets can be readily identified by spectral slope.78

Figure 5.1: Block diagram of designed module ................................................................ 82

Figure 5.2: Schematic of 2-stage power amplifier ............................................................ 83

Figure 5.3: Schematic of 2-stage pre-amplifier ................................................................ 84

Figure 5.4: Schematic of diode-based expander for protecting power amplifier from low
voltage echo signals .......................................................................................................... 85

Figure 5.5: Schematic of diode-based limiter for protecting pre-amplifier from high
voltage signals of power amplifier .................................................................................... 85

Figure 5.6: Designed power amplifier module performance. Power amplifier gain is
measured at 86 MHz and different input voltages. ........................................................... 86

Figure 5.7: Designed high frequency ultrasound cell sorting system ............................... 86

xv

Figure 5.8: A pulse echo signal from agar surface for location of ultrasound transducer.
Echo signal from agar surface is indicated in red box. ..................................................... 89

Figure 5.9: The experimental configuration ..................................................................... 89

Figure 5.10: Leukemia cells (K562) and red blood cells. The 10 µm was placed on agar
surface as reference. .......................................................................................................... 90

Figure 5.11: Shows measured echo signals from single live cell, (a) K562 cell and (b)
RBC in beam center of focal plane of transducer. Note that beam center was indicated as
a pink circle of screen center. ........................................................................................... 92

Figure 5.12: Amplitude of echo signals of K562 cell and RBC. ...................................... 94

Figure 5.13: The integrated backscatter of K562 cell and RBC. Note the graph was drawn
by positive value for better appearance. ........................................................................... 95

Figure 5.14: The spectral slope in dB/MHz of K562 cell and RBC. ................................ 96

Figure 5.15: The spectral intercept in dB of K562 cell and RBC. Note the graph was
drawn by positive value for better appearance. ................................................................ 97

Figure 5.16: Correlation between parameters ................................................................... 99

xvi

ABSTRACT
Ultrasound has been used as diagnostic imaging tools in medicine for a long time due to
its real-time capability and mobility as well as nonionizing radiation and safety. High
frequency ultrasound (above 20 MHz) has opened up new biomedical applications thanks
to its fine spatial resolution by sacrificing the depth of penetration due to increasing
attenuation. Shung’s group has shown the potential of high frequency at biomedical
engineering fields such as backscattering measurement for studying material properties
and micro devices for micro-engineered platforms. These new approaches can be
potentially powerful tools in the biology and medicine such as cell identifying and sorting
for high quality sample in a miniaturized Micro Total Analysis System (µTAS).
Different types of high frequency ultrasound cell sorting micro-device, which combines
high frequency ultrasound transducer with microfluidic channel, was proposed and
developed as simple and cost-effective bioparticle sorting device. Despite the fact that the
basic concept and sorting mechanism is straightforward and intuitive, it has several
advantages such as high sensitivity without pre-treatment of samples, much simpler
processes of sorting, and benefit by scaling down compared to conventional cell sorter.
In this research, two different size droplets in same material and different material micro-
particles in similar size were clearly discriminated from spectral parameters such as
spectral slope and intercept with echo amplitude and IB coefficient. In size-based case,
50 µm and 100 µm diameter lipid droplets were separated by these suggested methods in
real time. Acoustic radiation force of high frequency ultrasound is shown to produce
xvii

forces sufficient for sorting micro bioparticles by pushing through PDMS channel wall.
In material properties-based case, it was able to sense and identify flowing polystyrene
microbeads and oleic acid droplets in 45 µm diameter inside microchannel by spectral
slope value in dB scale.
Lastly, it was carried out a live cell sensing and identifying experiments with human
leukemia and red blood cells through multi-parameters. The potential and performances
of devices as a live cell sorter are verified by experimental results. For further clinical
applications, the necessity of more research related to bioeffect of high frequency
ultrasound beam was briefly described although there exist currently no guidlines or
criteria for high frequency ultrasound.
In this research, the ultimate goal is to develop a simple and yet cost-effective micro
sorting device with high frequency ultrasound microbeam for individual particulates. To
improve presented acoustic cell sorting method for real cell sorting applications is needed.
Especially for higher frequency ultrasound, new microfluidic channel using an alternative
material with lower acoustic attenuation for microfluidic channel is required due to high
acoustic energy loss of PDMS for improving sensing sensitivity (>99%). Better
hydrodynamic focus in new designed device could improve performance of ultrasound
cell sorting devices. In this acoustic sorter, multi-parameter data analysis including
statistical methods for sorting and separation should be applied with this ultrasound
sensing and identifying technique in real time with high throughput. These multi-
dimensional analysis and selectively visualization of data are useful for extracting
xviii

physical characteristics of cells without pre-treatments of samples. It maybe enables the
real time acoustic cell sorting device to identify and separate bioparticles as simpler
method. Current high frequency device is still relatively bulky and has low sorting
performance. The system for ultrasound sorter as alternative tools is needed to be
simplified with integrated micro-chip circuit design replacing outside customized
LabVIEW program, which enable an ultrasound sorter to reduce the operating time,
leading high throughput of current system.

1

CHAPTER 1 INTRODUCTION
1.1 High Frequency Ultrasound
Sound is an elastic wave based on mechanical vibrations in a medium and can be
classified according to frequency range. Acoustic elastic wave spectrum of scientifically
useful frequency range goes from 0.01 Hz to 10 THz, and is shown in Figure1.1 (White,
1997). Specifically, the infrasound range in nature is below the limits of human hearing
(0.01 ~ 10 Hz). In general, audible sound which most humans can hear has a frequency in
the range of 20 Hz to approximately 20 kHz, which represents the upper frequency limit
of human hearing. Acoustic waves which have a frequency above audible sound are
defined as ultrasound. It is generated by ultrasonic transducers that can transform an
electrical energy into acoustic energy. Its frequency range is generally from 20 kHz to 10
GHz. The narrow range of 20 kHz to 10 MHz has been used for medical applications
including diagnostic medical imaging. Acoustic elastic waves are categorized according
to frequency range as above mentioned in Figure 1.2.
2

Figure 1.1: The scientifically useful frequency spectrum of acoustic excited or detected
elastic waves. (White, 1997)

Figure 1.2: A classification scheme for acoustic waves by frequencies. (Maldovan, 2013)
3

The history of ultrasound is long, dating from the early 1900’s. Ultrasound has been used
as diagnostic imaging tools in medicine for a long time due to its real-time capability and
mobility as well as nonionizing radiation and safety (Shung, 2006). Ultrasound frequency
for conventional medical imaging is from 2 MHz to 10 MHz, which has the spatial
resolution of the order of a few millimeters and penetration depth from 40 mm to 240 mm
depending on the ultrasound frequency. In addition, medical ultrasound has been adopted
in various biomedical applications such as therapeutic applications and combinations
with drug delivery. Figure 1.3 shows other medical applications at low frequencies (< 10
MHz) ultrasound range (Mitragotri, 2005).
High frequency ultrasound (above 20 MHz) has opened up new clinical applications in
diagnostic medical imaging field: visualizing blood vessel wall, anterior segments of the
eye and skin, and imaging the heart of a small animal (Lockwood et al. 1996; Shung et al.,
2009; Shung, 2011), because high frequency ultrasound has the spatial resolution
improved to several tens of micrometers by sacrificing the depth of penetration due to
increasing attenuation in a material medium. It has been shown that the high frequency
range above 10 MHz is useful in drug delivery and tissue engineering (Bommannan et al.,
1992a; Bommannan et al., 1992b).

4

Figure 1.3: Ultrasound medical applications at low frequencies range. (Mitragotri, 2005)

5

Recently, ultrasound has been used in a remarkably diverse set of high frequency
ultrasonic devices for biomedical engineering applications thanks to its fine spatial
resolution: acoustic tweezers for trapping and manipulating micro particle and a single
cell (Lee et al., 2009a; Lee et al., 2010a; Lee et al., 2011a; Lee et al., 2014), cell stimuli
for studying cellular response from external stimuli and understanding their mechanisms
(Lee et al., 2012a; Hwang et al., 2012; Hwang et al., 2013; Hwang et al., 2014), radiation
force imaging of zebrafish (Park et al., 2012) and backscattering measurement (Lee et al.,
2011b) for studying material properties and size, and micro devices for micro-engineered
platforms (Jeong et al., 2011; Lee et al., 2012b). These achievements of Shung’s group
have shown the potential of high frequency at diagnostic medical imaging and other
biomedical engineering fields. High frequency ultrasound applications have been being
researched, which are termed ultrasound microbeam project in his group. These new
approaches at high frequencies (20 MHz ~ 1 GHz) can be potentially powerful tools in
the biology and medicine such as cellular machonotransduction, intercellular kinetics
studies and cell fusion control.

6

1.2 Micro Total Analysis Systems
The novel concept of a Micro Total Analysis System (µTAS) was first presented
conceptually in 1990 by Manz, Graber and Widmer for chemical sensing (Manz et al.
1990). µTAS, based on lab on a chip (LOC) device using micro-fabrication techniques,
has become an important tool and technology platform for the development of point-of-
care laboratory tests in the field of biology and medicine. The system has a wide range of
its applications in genomics, proteomics, clinical diagnostics, drug discovery and
biosensors (Lee et al., 2004).
Particularly in analytical chemistry, there are many reasons why scaling down in size is
needed and beneficial for analyses in research. Such miniaturized micro-devices offer
numerous advantages such as reducing sample or reagent volume, low power
consumption, faster sample processing, high sensitivity and spatial resolution, low device
cost, and increased portability, compare with conventional macro-scale methods at the
laboratory. Figure 1.4 shows conventional macro-scale laboratory and developed micro
analysis systems.
In the past decade, advancements in biology and medicine have led to a significant
increase of the number of micro fabrication techniques, which provide the advantages of
being more compact and cost-effective devices. The systems have integrated sequential
operations in a single microfluidic device with analytical processes such as sample
acquisition, pre-treatment, separation or sorting, post-treatment, detection, and analysis
(Lee et al., 2004; Yi et al, 2006). These processes include analytical technology such as
7

biological and chemical reaction (Lee et al., 2004; Yi et al, 2006), for onsite monitoring
and diagnosis of various pathogens with bath processing. Most of analytical processes,
including micro-PCR (polymerase chain reaction) chips, micro-DNA chips, micro-DNA
biosensor, micro-CE (capillary electrophoresis) chips, and micro-protein chip, require
simple and yet effective methods of obtaining high quality samples in many biological
and medical assays (Erickson et al., 2004).
The sample preparation for analysis is one of the crucial functions that need to be
performed in the micro analytical systems, and for analytical improvements, automated
and rapid sample preparations for pure and high quality were required as key components
in developing µTAS associated with the scaling down of the size.

Figure 1.4: Macro-scale clinical laboratory and micro-scale systems. (a) conventional
clinical laboratory (Courtesy of Stony Brook, Health Technology and Management), (b) a
molecular diagnostic system by Siemens, the fraunhofer Institutem and Infineon
Technologies (Courtesy of Medical Technology Business Europe) and (c) New microchip
technology performs 1,000 chemical reactions at once (Courtesy of UCLA Newsroom).
8

1.3 Cell Sorting
One of the important techniques for laboratory analysis is the purification or isolation of
sample subpopulations based on separation and sorting of small particles. During the long
history in the development of technologies for the separation of small particles,
purification or isolation of sample has been a major goal for basic research in cell biology,
molecular genetics to diagnostics, and therapeutics. In the early stage, available
parameters for tiny particle sorting and separation were their physical characteristics or
biochemical characteristics such as density, and selectable enzymes (Recktenwald et al.,
1998).
The techniques previously reported can be broadly grouped as being bulk separation and
single-cell sorting methods depending on its feature (Orfao et al, 1996) as shown in
Figure 1.5. Bulk separation, including filtration, sedimentation, and cell charge-based
method, has a common feature for the isolation of group of target cells in a relatively
simple manner. In contrast to it, single-cell-based sorting is more sophisticated for
purification or isolation of samples than bulk separation in that the method allows a rapid,
objective, and sensitive multi-parametric separation. This type of sorting for cells is
known as a form of flow cytometry. Single particulate sorting devices, based on this
method, have frequently been used in investigating bioassay compartmentalization, an
encapsulated microenvironment for inducing various cell expressions (Doerr, 2005; He et
al., 2005).
9

Figure 1.5: Conventional cell separation and sorting methods. (a) bulk separation
methods (magnetic separation (left) and centrifuge methods (right), (b) single-cell-based
sorting methods.

Conventional methods for single particle sorting are categorized by the mechanisms used
for separation and sorting. Fluorescence-activated cell sorting (FACS) utilizes
fluorescence resulted from dyes attached or absorbed by cells upon illumination by a
laser for cell sensing and static electric charges carried by the cells for sorting (Bonner et
al., 1972; Herzenberg et al., 1976). Alternatively, magnetically activated cell separation
(MACS), microfludic channel separation using non-inertia force such as dielectrophoretic
(DEP) force, magnetic force, optical gradient force, and acoustic primary radiation force
all have also been used (Tsutsui et al., 2009). However, these methods are by no means
perfect. They have limitations in accuracy and speed. Improvements and new approaches
are constantly sought. Also, several different approaches, combining microfluidics and
10

flow cytometer including hydrodynamic focusing, have been employed for the sorting of
particles to overcome the drawbacks. Among the findings it appears that the optical
gradient force and acoustic radiation force in continuous flow, which allows both bulk
and single cell separation, are the most attractive. Especially, in particle separation via
acoustic primary radiation force, Icí ar González and his co-workers performed particle
sorting by ultrasound in a polymeric chip (Gonzalez et al., 2010), and Thomas Laurell et
al published results on acoustic separation and manipulation of cells and particles using
standing wave (Laurell et al., 2007). Moreover, researchers have shown that these
techniques can be applied to biomedical fields. For example, Henrik Jonsson reported
that particle separation using ultrasound can be applied in medical surgery (Jonsson et al.,
2004), and M. Wilklund and H. M. Hertz applied ultrasonic enhancement of particles to
bioaffinity assays (Wiklund et al., 2006). Also, Otto Manneberg’s group used ultrasound
resonances in order to generate ultrasound force fields in microfluidic channels
(Manneberg et al., 2009). These researchers all showed that separation and sorting of
particles are possible using ultrasound.
Among these single-cell sorting methods, fluorescence-activated cell sorting (FACS)
using light scattering of laser and flow cytometer has been the gold standard, due to
mature engineering development, single cell level sensitivity, and high throughput
(5000~40,000 cells∙s
-1
) (Bhagat et al., 2010; Leary, 2005). Although FACS has been the
gold standard for high quality samples, the method still has several limitations such as a
minimum number of cells (> 100,000) required for operation (Wang et al., 2005a),
additional sample treatment of fluorescent labeling, complicated and costly expensive
11

device, and contamination of biohazardous samples due to open operation environments,
and need of skilled operator. Also, FACS frequently produces aerosols that cause serious
biological side effects on cells (Oberyszyn et al., 2001). Structure and sorting features of
FACS are shown through early stage model and current commercial product in Figure 1.6.

Figure 1.6: Fluorescence-Activated Cell Sorter (FACS). (a) Optical structure and sorting
feature of FACS (Recktenwald et al., 1998), (b) early stage FACS model manufactured
by the Becton Dickinson Electronics Laboratory (Herzenberg et al., 1976), and (c)
current FACS commercial product by Beckman Coulter (Courtesy of Beckman Coulter,
Inc.).

(a)
(b) (c)
12

To overcome these drawbacks, fluorescence activated droplet sorting (FADS) (Baret et
al., 2009) have been demonstrated by encapsulating each cell within droplet emulsion
integrated with a microfluidic flow system in a conventional FACS. Although allowing
sterile and rapid sample processing without compromising cell viability, FADS mainly
relies on fluorescent detection and dielectrophoresis driven by complex electronic circuits.
Other microfluidic sorting techniques proposed so far include hydrodynamic flow
switching (Fu et al., 2002) and acoustic standing waves (Petersson et al., 2004) that are
more appropriate tools for bulk sample screening. Hence a simple and yet cost-effective
sorting method for individual particulates is needed for precise bioassay analysis.

13

1.4 Objective of Research
Previously reported cell sorters that utilized ultrasound resonances were made with
standing wave at lower frequencies from 1 KHz to 10 MHz. Acoustic force fields formed
by standing wave are not capable of performing single cell sorting because they use
pressure nodes which are affected by channel size and ultrasound frequency. In addition,
two transducers or one transducer and a strong reflector are needed for such an approach,
making its actual implementation quite difficult if not impossible in practical situations.
In this research, a novel method using radiation forces produced by highly focused high
frequency ultrasound beams is proposed. Cell or particle sensing can be carried out with
conventional approaches ultrasonically via ultrasonic scattering measurements.
In fact ultrasonic sensing would eliminate one major limitation that has plagued
conventional devices and methods for single particle or cell sorting i.e., the particles or
cells have to be pre-treated. A good example is fluorescence-activated cell sorting (FACS)
in which the particles or cells have to be pretreated with fluorescent dye. The proposed
device may allow sensing and sorting of single bioparticles to be achieved without pre-
treatment. It can make the processes of sorting much simpler. The device has added
advantages in size reduction and ease of operation. FACS device is bulky due to laser
sources. The operation of the instrument requires a person with specialized training
because of its complexity and the requirement for sample pre-treatment. Further, this
technique can be employed for sorting and separation in light opaque media, thanks to the
nature of ultrasound.
14

In comparison to optical radiation force based devices, the proposed technology
possesses the following merits: greater radiation forces and penetration of light opaque
media. Ultrasonic sensing can be achieved via a measurement of particle or cell scattering
properties which are related to the size as well as acoustic properties of the particles or
cells, namely compressibility and density. A number of ultrasonic scattering properties
including backscattering, angular scattering, and scattering dependence on frequency may
be measured for particle or cell characterization. From these limitation of conventional
devices and needs of better device for single cell level sorting, real time acoustic cell
sorter in single cell level has thus been researched to satisfy this need by combining high
frequency ultrasound beam with micro flow cytometry in microfluidic technology.
The primary objective is to experimentally show that micro-particles can be sensed and
identified by high frequency ultrasound microbeam in a continuous stream in the
microfluidic channel. The secondary objective is to demonstrate that the particles can be
sorted by acoustic radiation force in microfluidic device. Lastly, the proposed acoustic
cell sorter as an important and valuable analytical instrument in a biomedical laboratory
and as an important component of µTAS is discussed

15

1.5 Overview
The dissertation consists of six chapters including the introduction of general background
of high frequency ultrasound, micro total analysis systems, and cell sorting in Chapter 1.
Chapter 2 describes ultrasound cell sorting device, which combines high frequency
ultrasound transducer with microfluidic channel. This chapter introduces device concept
and sorting mechanism and evaluates performance of sorting device for supporting
potential as cell or particle sorting device in biomedical fields.
In Chapter 3, a microfluidic channel newly designed and fabricated is described,
providing improved sensing sensitivity and carrying out more accurate micro flow
cytometry. For more precise identifying different micro-particles size, multi-parameter
method was employed, including amplitude of echo signals and Integrated Backscatter
(IB) coefficient. The device have discriminated size difference of micro-particles by
multi-parameter and separated in real time by applying acoustic radiation force.
Chapter 4 shows extended function of high frequency ultrasound sorting device by
applying ultrasonic spectral analysis, which high frequency ultrasound can identify
difference of material properties of single micro particle in similar size.
In Chapter 5, the capability of current ultrasound technology is discussed, as single cell-
based sorting device for sorting a live cell. Highly focused single element high frequency
ultrasound was able to distinguish human red blood cells from human leukemia cells on
agar phantom surface. The potential and performances of devices as a live cell sorter are
16

verified by experimental results. Bioeffect of high frequency ultrasound beam is briefly
mentioned.
Chapter 6 summaries the works in this dissertation, and discusses future works.

17

CHAPTER 2 HIGH FREQUENCY ULTRASOUND CELL
SORTING DEVICE
2.1 Introduction
For the past twenty years, a number of engineers and scientists have been continuously
researching to develop miniaturized devices due to needs of analytical batch processes in
the field of biology and medicine using microelectromechanical systems and micro-
fabrication techniques. Conventional macro-scale analytical systems can be made into
miniaturized micro devices by combining electrical, optical, and microfluidic
components through rapid advances of micro-technologies. Fabricated micro-scale
devices and systems have several advantages over macro-scale devices and systems such
as lower cost, lower reagent and sample consumption, disposability, portability and lower
consumption.
Most of analytical processes in micro-scale analytical devices and systems require simple
and effective methods of obtaining pure and high quality samples, which provide more
efficient monitoring and diagnosis of various illnesses. Automated and rapid sample
preparation parts in µTAS are all important components in the development of an
integrated system for prompt detection and monitoring of all pathogens and environment
in onsite.
A variety of technologies for these core functions in µTAS have been being employed,
such as micro flow cytometry and microfluidic technologies with optical, magnetic and
18

electrical force. The pure and high quality samples resulting from these core parts are
prepared by sorting or isolation techniques of subpopulations of raw samples in the
systems. Ultrasonic devices, which use standing waves and acoustic force fields of low
frequency range of 1 KHz to 10 MHz, also have been used as an essential part in the
micro systems and independent tool. Acoustic force fields made by standing wave cannot
perform accurate sorting of single cell level due to limitation of the methods, which use
pressure node of ultrasound based on channel size and ultrasound frequency in sorting
devices. As mentioned in chapter 1, both accuracy and high speed are key elements in
cell sorting and separation devices. Cell sorting rate of 10,000 ~ 25,000 cells per second
is performed by single cell level sorting using flow cytometric soring (Recktenwald et al.,
1998). However, this is considered relatively low speed compared to bulk sorting and
separation. Such slow sorting capability is due to the sensing function of flow cytometric
methods.
In this chapter, the proposed method that involves the design of highly focused single
high frequency ultrasound transducers from 20 MHz to 100 MHz and the utilization of
ultrasound radiation force generated by highly focused ultrasound beams to manipulated
cells or particles is described. This method has different approach compared with other
ultrasonic devices in that one highly focused ultrasound transducer is used as the sensing
or sorting source with a microfluidic channel in order to sort micro-particles. Not only
that, this method may be used in the standing wave made with continuous signal, like
previous ultrasound devices. This provides us with a more flexible device with various
applications in biological, chemical, medical research fields, which allow multifunctional
19

devices to perform cell sorting and specific target separation from applied radiation force
based on micro-particle size.

20

2.2 Material and Methods
2.2.1 Device Concept and Mechanism
The basic device concept is an acoustically driven cell sorting device integrated with a
poly(dimethyl) siloxane (PDMS) microfluidic channel. Hydrodynamically focused
bioparticles flowing in the channel are non-invasively probed by a high frequency
ultrasonic beam which measures quantitatively their backscattering properties. A high
frequency ultrasound transducer is placed outside the channel, aiming at the droplet flow.
The overall system consists of two independent and sequential processes, acoustic
sensing and sorting. In sensing, a series of single-pulse emitted from the transducer forms
a sensing beam by which individual biopartcles are remotely interrogated inside the
channel. Ultrasonic spectral parameters, such as integrated backscatter (IB) coefficients
in dB (O’Donnell et al., 1979), spectral slope (dB/MHz) and spectral intercept (dB;
extrapolation to zero frequency) (Lizzi et al., 1976; Lizzi et al., 1987) and midband fit
(dB; the value of the regression line at the center frequency of the spectral band) (Lizzi et
al., 1997; Feleppa et al., 1996; Feleppa et al., 1997), are then calculated in the frequency
domain, by analyzing received RF echoes from the cells passing though the beam. From
the measured spectral parameters and echo amplitudes, a custom-developed LabVIEW
program immediately triggers the same transducer to produce a sorting beam that
separates a mixture of difference bioparticles.
The most important component of this sorting device is a highly focused high frequency
ultrasound transducer which is needed for both sensing and sorting of bioparticles. Other
21

components of the experimental arrangement for feasibility demonstration include a
microfluidic channel, a microscope, a linear stage, a data acquisition board, an amplifier
and a function generator. A high frequency ultrasound transducer and a microfluidic
channel were custom made for this purpose. The concept and schematic diagram of an
acoustic sorting device is illustrated in Figure 2.1 and Figure 2.2 to depict device concept
and operating mechanism.

Figure 2.1: A concept diagram of high frequency ultrasound sorting device. The device
uses echo signals for sensing and its analysis by calculating ultrasonic spectral parameters
provides guide information for recognizing bioparticles. After making decision from
analyzing echo, acoustic radiation force is used for sorting and separation of
subpopulation of sample by pushing. A single high frequency ultrasound transducer
combines sensing with sorting for the advantages of being more simple and compact
device.
22

Figure 2.2: Device schematics and mechanism. High frequency ultrasound transducer is
placed on outside of microchannel. Position of focus of transducer is important for better
echo signals from the bio-particles. (a ~d, f) show sensing mechanism of device and (e)
illustrates its sorting mechanism. (a) A red sample object is approaching to sensing zone
in sensing mode of device. (b)(c) Echo singles of the red object are received and analyzed,
and if the one is not targeted, it passed through to zone. (d) When a targeted green object
is placed in sensing area, the system senses targeted one by calculated ultrasonic spectral
parameters, such as IB value, slope, intercept and midband fit. (e) The system changes
operation mode from sensing to sorting, and generate acoustic radiation force for pushing
the green one. (f) Then, the device comes back to sensing mode for continuously working
as a sorting device. These processes are automatically controlled by customized
LabVIEW program.
23

2.2.2 High Frequency Ultrasound Transducer
A 30 MHz lithium niobate (LiNbO
3
) single element transducer was designed by KLM
modeling software and fabricated. Its aperture size and proper thickness of acoustic
stacks, such as LiNbO
3
single crystal, silver epoxy 1
st
matching layer, and parylene 2
nd

matching layer, were optimized by a KLM modeling software (PiezoCAD, Sonic
Concepts, Woodinvill, WA, USA). Along with the piezo-element, the transducer
consisted of a backing and two matching layers for efficient sound transmission as
depicted in Figure 2.3.

Figure 2.3: Schematic diagram of a 30 MHz LiNbO
3
single element ultrasonic transducer
and fabricated transducer. The transducer consists of a LiNbO
3
piezoelement, a backing
material, and two matching layer for sound transmission.

A 36º rotated Y-cut lithium niobate plate (Boston Piezo-Optics, Bellingham, MA, USA)
was lapped to its designed thickness of 77 µm and electroplated with 1500 A°
24

chrome/gold (Cr/Au) layer on both sides, by an NSC-3000 automatic sputter coater
(Nano-Master, Austin, TX, USA). The silver epoxy inner matching layer was made from
a mixture of Insulcast 501 epoxy (American Safety Technologies, Roseland, NJ, USA)
and 2 to 3 µm silver particles (Aldrich Chemical Co., St. Louis, MO, USA) with the
weight ratio of 1.25 to 3. The layer was cast after an adhesion promoter (Chemlok AP-
131, Lord Corp., USA) to one side of the LiNbO
3
plate, and cured over the plate. After
curing, the first matching layer was lapped down to the designed thickness of 12 µm. The
lapped LiNbO
3
and1
st
matching stacks were mechanically diced into square pieces, with
the size that would encompass a circular aperture. A backing layer of conductive silver
epoxy (E-Solder 3022; Von Roll Isola Inc., Schenectady, NY, USA) was then cast and
centrifuged onto the back side of the electro plated LiNbO
3
. After curing, the acoustic
stack was turned down to the designed diameter, 4 mm, and was concentrically placed in
the brass housing. The stack and the housing are electrically insulated by filling the gap
with an epoxy (Epo-Tek 301, Epoxy Technologies, Billerica, MA, USA).
The stack was press focused (Lockwood et al., 1993; Lockwood et al., 1994; Cannata et
al. 2003) at 3 mm to obtain an F# of 0.75. Hence, the convergence angle of acoustic
beams was formed to be 84º at the focus. The transducer surface was then sputtered with
Cr/Au electrodes to set ground contact between the stack and the brass housing. A 14 µm
parylene layer, as a second matching layer, was deposited and coated over the aperture
using a PDS 2010 Labcoater (SCS, Indianapolis, IN, USA). The finished transducer
element was connected to a subminiature version A (SMA) connector.
25

The performance of the transducer is evaluated by measuring the beam characteristics
and the pressure level at the focus with a needle type hydrophone (HPM04/01, Precision
Acoustics, UK). The 6 dB bandwidth of the beam is 54.9% as shown in Figure 2.4. The
lateral resolution is 36 µm and the peak pressure is 3.24 MPa at 30 MHz for both the
sensing and the pushing beams as shown in Figure 2.5.

Figure 2.4: Pulse-echo waveform from quartz at the focus of 30 MHz and its frequency
response
26

Figure 2.5: Lateral pressure beam profiles of 30 MHz transducer. The measured 3 dB
beam width is 36 µm and peak pressure is 3.24 MPa at the focus of the transducer.

27

2.2.3 Synthesis of Lipid Micro Droplets
In this study, two different size lipid spheres with average diameter of 30 and 75 µm were
used for the ultrasound sorter device. Oleic acid (Fisher Scientific, Pittsburgh, PA, USA)
lipid droplets were synthesized using droplet-based microfluidic devices, a robust method
of forming monodispersed droplets and particles in the nanometer to micrometer size
range (Tan et al., 2006; Teh et al., 2008). Microfluidic devices were fabricated in
poly(dimethyl) siloxane (PDMS) using conventional soft lithography techniques (Xia et
al., 1998; Whitesides et al., 2001). A hydrophilic surface treatment was applied to render
the channel surfaces hydrophilic (Kozlov et al., 2003). As PDMS is inherently
hydrophobic and the continuous phase used for the droplet generation is water, the
hydrophilic surface treatment ensures complete wetting of the walls by the aqueous
solution. The solution phase consists of a 5 wt. % mixture of Pluronic F-68 (Sigma
Aldrich, USA) and ultra-pure water (Millipore, Billerica, MA, USA). Pluronic F-68
stabilizes oleic acid droplets during storage and transport. As shown in Figure 2.6 (left),
the two liquid phases meet at the shear junction and oleic acid is sheared into droplets by
the aqueous phase. The flow-focusing nozzle geometry creates a local shear maximum
that ensures repeatable breakup of the fluid stream (Tan et al., 2006). The droplets were
formed at a rate of approximately 50 droplets per second and had a monodispersed size
distribution as shown in Figure 2.6 (right). The droplet size can be controlled by changing
the relative flow rates of the solutions. The greater the oleic acid flow rate, the larger the
droplet. Oleic acid and aqueous flow rates of 0.5 to 1 µL/min were applied to form oleic
28

acid droplets of 75 to 300 µm in diameter. The resultant droplets tended to float because
the density of oleic acid, 900 kg/m
3
, is known to be less than that of the water.

Figure 2.6: (left) Oleic acid droplet formation microfluidic device. The two phases, oleic
acid and 5% pluronic solution, meet at the flow focusing junction where oleic acid is
continuously sheared into droplets by the two aqueous phases. (right) Monodispersed
population of oleic acid droplets dyed oil Red (Dupont, Wilmington, DE, USA) in
visualization. (Lee et al., 2010a)

29

2.2.4 Microfluidic Channel Design and Fabrication
The sorting platform is fabricated in a poly(dimethyl) siloxane substrate. As shown in
Figure 2.7, the device has two narrow inlet channels leading into a main channel which
then splits into two outlet channels. One of two inlet channels serves as flow of sample
solution, the other inlet channel serves as flow of buffer solution, providing
hydrodynamic positioning of the lipid spheres to the detection area of the main channel.
The height of all the channels is 100 μm. The width of the sheath flow channels are 250
µm, the bead inlet channel is 250 µm, main channel is 500 µm, the sorting channel is 250
µm and outlet channels is 250 μm. As there is not much literature on the acoustic
transmissibility of the transducer through PDMS, the sorting capability of the transducer
was tested for varying PDMS wall thicknesses. The thickness of the wall was 250 μm. It
was determined that even with a wall thickness of 250 μm, the high frequency ultrasound
pressure was sufficient to push the lipid droplet into the sorting channel.

Figure 2.7: Microfluidic channel fabricated in poly(dimethyl) siloxane (PDMS) using
conventional soft lithography techniques
30

Sorting channels were fabricated in poly(dimethyl) siloxane (PDMS) using conventional
soft lithography techniques (Xia et al., 1998; Whitesides et al., 2001). First, 3 inch silicon
wafers were spin-coated with a 100 μm layer of SU8-50 (MicroChem) photoresist, baked
to improve the adhesion of the SU-8 to the silicon wafer and then patterned by exposure
to UV light through a high resolution photomask containing the channel design. After
post-exposure baking, the wafer is then submerged in SU8 developer to expose the
channel pattern. The remaining crosslinked SU-8 resist forms a positive mold for the
silicone polymer. PDMS (Sylgard 184, Dow Corning) was mixed at a 10:1 prepolymer
base to curing agent ratio and poured over the patterned wafer. The polymer mix was
cured at 65 º C for at least 4 hours. After curing the device were peeled off the mold, cut
into individual devices and connection holes were bored into the device using flat end
dispensing needles (Integrated Dispensing Solutions Inc.). The devices were then cleaned
before bonding via oxygen plasma treatment to a cleaned 5 mm thick slab of PDMS. The
oxygen plasma activates the surfaces of the PDMS and allows for irreversible bonding
between the two surfaces.
A hydrophilic surface treatment is applied to the channels to minimize bubble formation
and to match surface wettability since an aqueous continuous phase is used. Polyvinyl
alcohol (PVA) hydrophilic treatment is applied to the channels as it has been shown to
maintain the PDMS surface hydrophilic for multiple weeks (Kozlov et al., 2003). Briefly,
the channels are incubated in a 1 wt. % PVA solution for 5 minutes at room temperature.
Then excess solution is removed by vacuum, and the device is incubated in a 120 º C oven
31

for 5 minutes to promote adhesion of the PVA monomers to the PDMS surface. This
process can be repeated multiple times to ensure even coating to the surface.

2.2.5 Experimental Configuration
To study acoustic particle sorting with highly focused ultrasound transducers, the
experimental equipment was set under distilled water in a chamber. Micro-fluidics
channel was fixed in the water chamber. Each flow rate of sample and buffer in
microfluidics channel was controlled by syringe pump (NE-1000 Multi-PhaserTM; New
Era Pump System lnc., NY, USA). The ultrasound transducer was assembled at a three-
axis motorized linear stage attached two axis goniometer (LMG26 T50MM; OptoSigma,
Santa Ana, CA, USA; Newport Corporation, Irvine, CA, USA) in order to manipulate
and locate its position. The transducer was positioned at the side of microfluidics channel
in order to detect small particles with A mode signals and to push them with radiation
force. The positioner was operated with customized LabVIEW program with RS232C
connection. The schematic diagram of sorting device is illustrated in Figure 2.8. The
highly focused ultrasound transducer was driven by function generator (AFG3251;
Tektronix, Anaheim, CA, USA) and 200MHz computer controlled pulser / receiver
(Model 5900PR; Panametrics-NDT, USA), and then amplified by a 50dB power
amplifier (325LA; ENI, Rochester, USA). A mode echo signals were monitored by
oscilloscope (Waverunner 104MXi; LeCory, USA). The video was recorded by a CCD
camera (InfinityX; Lumenera, USA) assembled at a microscope (SMZ1500; Nikon, Japan)
32

in order to check out motion of particles related to detect and pushing signals as well as
area of detection.

Figure 2.8: Experimental configurations for high frequency ultrasound cell sorting device.

2.2.6 Experimental Procedure
Micro-particle sorting device was designed with A-mode scan method via single
elements transducers, which uses amplitude of reflection signals from particles in order to
recognize lipid particles of different size. Generally, large particles reflect bigger
amplitude than small ones. That is basic and crucial concept of this sorting system. Peak
to peak voltage levels of reflected signals from the particles in microfluidic channel were
monitored and analyzed by specialized LabVIEW program. For collecting better signals,
the focal point was located at the center of detection area and then particles were passed
through that area by adjusting flow rate of sample and buffer solution. Each flow rate of
33

sample and buffer solution was 1 µl/min and 3 µl/min. The area and position of focal
point of transducer are exemplified by Figure 2.9.
For the evaluation of this sorting device, movie clips were monitored and recorded with
peak-to-peak voltage value of A mode signals by using CCD camera (InfinityX;
Lumenera, USA) attached to microscope (SMZ1500; Nikon, Japan) and LabVIEW
program. After the experiment, the total number of lipid particles which passed through
the sensing zone was counted manually through the clips, while the detected particles
were numbered based on recorded signals and movies.

34

Figure 2.9: The sensing zone of acoustic particle sorting. (a) the particles pass through
portion of the sample solution with width of about 80 µm and flow rate of 1µl/min. (b)
focal point of the transducer are located at the point which are moved with distance 25
µm from left wall of PDMS channel and distance of 20 µm above center line of the
channel in order to be positioned at the center of sensing zone.
Reference signals
Microfluidic Channel Section
DI water
Transducer
~80µm
Sensing Zone
Lipid Droplet
(b)
Buffer Flow rate : 3.0ul/min
80µm
Sample Flow rate : 1.0ul/min
(a)
Side View
Top View
Center line
35

2.3 Results and Discussion
To test the sorting capability of the acoustic transducer, lipid droplets are flowed down a
channel with a bifurcating outlet. The channels are designed such that without transducer
actuation, the lipid droplets will continue down and exit through the main channel to
initiate droplet sensing. The transducer is activated by 30 MHz and one cycle signal of
amplitude 32 V
pp
with 1 KHz pulse repetition frequency.

2.3.1 Sensing Sensitivity
For the purpose of this study, the detection signals reflected from lipid droplets of two
different sizes, 30 µm and 75 µm, were measured in the microfluidic channel with a very
low flow rate using acoustic particle device. The peak-to-peak voltage levels made from
reflection at surface of lipid particles had different signal amplitude, which enabled to
distinguish size of lipid droplets. The A mode signals from small particles of 30 µm, was
7.9 ± 1.9 mV
pp
, while the reflected signals levels of large particles of 75 µm was 16.4 ±
1.6 mV
pp
, both based on A mode signals from a single excitation condition. When
droplets pass through defined sensing zone, echo signals from lipid droplets are captured
by specialized LabVIEW program. Recorded signals of both droplets are shown in Figure
2.10

36

2.3.2 Sensing Efficiency
In order to evaluate detecting efficiency of the acoustic particle sorting device and its
method, a mixture containing oleic acid droplets of 30 μm and 75 μm was passed through
the sensing zone and recorded into movie clips, to manually count the total and sensing
number of particles based on movie clips and peak-to-peak voltage levels of A mode
signals. The operation conditions for applied voltage and flow rate of sample and buffer
solution were same as described above. The total number of particles counted in the
experiments was 1512 small droplets (30 μm) and 243 large droplets (75 μm),
respectively. Among this total number, the small and large particles, which were sensed
by the device, were 709 and 187 spheres, respectively. The detecting efficiency was
calculated based on this data, which showed a detecting sensing rate of 52.50 ± 7.68% for
the 30 μm droplets and 79.88 ± 7.82% for the 75 μm droplets.
Although the rate of this detection method needs to be improved above 90%, the results
show that the suggested method with highly focused ultrasound transducer in this
research is capable of sorting target particles. There may be various methods in
improving the sensing rate. First, the rate may be improved by applying other analyzing
methods of A mode signals such as spectrum and integrated backscattering analysis.
Second, a method of more precise flow cytometry may be employed since the position of
lipid particles play a key role in amplitude of reflected signals from particles. However,
the low sensing efficiency in this device is mainly due to data loss resulting from low
data transfer rate from oscilloscopes to customized LabVIEW program through Ethernet
37

connection. This could be solved by improving data transfer rate between LabVIEW
program and data Acquisition board.

Figure 2.10: Scattered echo signals from lipid droplets. (a) Reflected signal from channel
well (b) 30 µm lipid droplet echo signal (c) 75 µm lipid droplet echo signal

Small Particles Signals (30µm) Reference signals
Microfluidic Channel Section
Large Particles Signals (75µm)
(c) (b) (a)
38

Figure 2.11: A mode signals difference between 30 µm and 75 µm. The peak-to-peak
voltage levels made from reflection at surface of lipid particles had different signal
amplitude, which enabled to distinguish size of lipid droplets. The A mode signals from
small particles of 30 µm, was 7.9 ± 1.9 mV
pp
, while the reflected signals levels of large
particles of 75 µm was 16.4 ± 1.6 mV
pp
, both based on A mode signals from a single
excitation condition.

7.92
16.37
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
Signal amplitude(mV
pp
)
Echo signal amplitude
Small(30µm) particles
Large(75µm) particles
39

Figure 2.12: Sensing efficiency of each particle - small particles (30 µm) / large particles
(75 µm). The detecting efficiency was calculated based on this data, which showed a
detecting sensing rate of 52.50 ± 7.68% for the 30 μm droplets and 79.88 ± 7.82% for the
75 μm droplets.

52.50
79.88
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
Sensing Efficiency (%)
Sensing Efficiency
Small(30µm)
Large(75µm)
40

2.4 Conclusion
The testing results demonstrated that the sorting method with a highly focused ultrasound
transducer in this research could sort targeted particles based on the detection signals
reflected from lipid droplets. In order to evaluate performance of the acoustic particle
sorting device, the sensing rate is measured by manually counted total and sensing
number of particles. Particle detecting rate of the device needs to be improved by higher
data transfer rate between LabVIEW program and data acquisition board. However, the
results indicate that the suggested method with the ultrasound transducer has potential as
particle sorting device.
A major advantage of flow cytometric sorting can apply multi-parameter to identify the
small particles to be separated. In the sorting device, accuracy and speed are very
important factors in order to be applied in practical application. However, a trade-off
between accuracy and speed is one of the difficult challenges. The suggested acoustic
particle sorting device using acoustic radiation force could generate bigger radiation force
than optical radiation force with a less possibility of cell damage, which allows faster
sorting in flow cytometric method with faster flow rate and sorting of large particles. Not
only that, but also the device using highly focused ultrasound beam could rapidly sense
the particles in specific sensing zone. In addition, the detection method has much simpler
device than optical sensing and pushing. As mentioned advantages above, the acoustic
particle sorting has great potential in various engineering and medical applications.

41

CHAPTER 3 SIZE-BASED MICROFLUIDIC DROPLET
SORTING WITH HIGH FREQUENCY ULTRASOUND BEAM
3.1 Introduction
Single particulate sorting devices have frequently been used in investigating bioassay
compartmentalization, an encapsulated microenvironment for inducing various cell
expressions (Doerr et al., 2005; He et al., 2005). Individual compartments, in contrast to
bulk sample separation techniques, e.g. filtration and sedimentation (Orfao et al., 1996),
are screened by manipulating small volumes (typically < 1 µl), and the sorting efficiency
depends on their bio/physical properties. Fluorescence activated cell sorting (FACS) has
thus been developed to satisfy this need by combining light scattering with flow
cytometry (Bonner et al., 1972; Leary, 2005). Despite its high throughput capability
(5,000 ~ 40,000 cells∙s
-1
), FACS requiring additional sample treatment of fluorescent
labeling often produces aerosols that cause serious biological side effects on cells
(Oberyszyn et al., 2001). To overcome these drawbacks, fluorescence activated droplet
sorting (FADS) (Baret et al., 2009) has been demonstrated by encapsulating each cell
within a droplet emulsion, integrated with microfluidic flow systems in conventional
FACS. Although allowing sterile and rapid sample processing without compromising cell
viability, FADS mainly relies on fluorescence detection and dielectrophoresis driven by
complex electronic circuits. Other microfluidic sorting techniques proposed so far include
hydrodynamic flow switching (Fu et al., 2002) and acoustic standing waves (Petersson et
al., 2004), which are more appropriate tools for bulk sample screening. A surface
42

acoustic wave (SAW) actuated cell sorting (SAWACS) system (Franke et al., 2010) has
recently been developed to separate human keratinocytes, fibroblasts, and melanoma cells.
The system complexity is still high, because a SAW generation unit is integrated with a
poly(dimethyl)siloxane (PDMS) device. Hence a simple and yet inexpensive sorting
method for individual particulates is needed for precise bioassay analysis.
Size-based sorting approaches have often been used for low throughput separation
applications without antibody tags e.g. sorting of stem cells that express few protein
markers. Mesenchymal stem cells have been sorted from epithelial progenitor cells by
being injected into a ribbon-like capillary device in continuous flow (Roda et al., 2009).
Myocytes and non-myocytes have also been isolated by size from rat cardiac cell
populations in microfluidic channels (Murthy et al., 2006). Parenchymal cells
(hepatocytes) and non-parenchymal cells in liver have been separated by their difference
in size via microfluidic filtration (Yamada et al., 2007). In particular, large hepatocytes
from liver are used for toxicological assessment and cell transplantation, whereas small
non-parenchymal cells are essential in liver reconstruction (Kane et al., 2006; Katayama
et al., 2001). The separation of each liver cell by size is critical for carrying out
pharmacological and metabolic studies.
This chapter presents an acoustically driven size-based droplet sorting device, integrated
with a PDMS microfluidic channel. In this work, hydrodynamically focused lipid
droplets flowing in the DI water-filled channel are non-invasively probed with a high
frequency ultrasonic beam, through a quantitative measurement of their backscattering
43

properties. A 30 MHz single element lithium niobate (LiNbO3) transducer is placed
outside the channel, aiming at the droplet flow. The overall system consists of two
independent and sequential processes, acoustic sensing and sorting. A series of short
pulses emitted from the transducer forms a sensing beam by which individual droplets are
remotely interrogated inside the channel. From analyzing echo amplitudes and integrated
backscatter (IB) coefficients corresponding to those droplets, the transducer is switched
from the sensing to the sorting mode via a custom-built LabVIEW routine. A sorting
beam of 30 MHz sinusoidal bursts then drives the transducer to push 100 µm droplets
from the center stream. The performance of the proposed method is evaluated by
measuring the sensing and the sorting efficiencies for a mixture of these droplets, and its
potential applications are discussed.

44

3.2 Material and Methods
3.2.1 Synthesis of Lipid Micro Droplets
Oleic acid (Fisher Scientific, USA) lipid droplets are synthesized through droplet-based
PDMS microfluidic devices using a soft lithographic technique (Whitesides et al., 2001),
a robust method of forming monodispersed droplets in the nanometer to micrometer size
range (Tan et al., 2006). Droplets of 50 µm and 100 µm in diameter are made at a rate of
50 droplets per second, as oleic acid is sheared into a spherical shape by the aqueous
phase at the flow focusing nozzle. The droplets of the specified sizes have a very tight
size distribution of 96–99% for each batch prepared. Figure 3.1 shows size distribution of
randomly extracted droplet sample of 50 µm (86 droplets) and 100 µm (60 droplets) in
diameter.

Figure 3.1: Histogram of size distribution of droplet sample of (a) 50 µm and (b) 100 µm
in diameter.
44 45 46 47 48 49 50 51 52 53 54 55 56 57
0
5
10
15
20
25
30
Frequency
Droplet Diameter (µ m)
Diameter Distribution - 50µm Droplets
90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112
0
1
2
3
4
5
6
7
8
9
Frequency
Droplet Diameter (µ m)
Diameter Distribution - 100µm Droplets
(b) (a)
45

3.2.2 Microfluidic Channel Design and Micro Flow Cytometry
The same lithography processes are employed to fabricate PDMS based sorting devices.
Three-inch silicon wafers are spin-coated with a 100 µm thick photoresist layer (SU8-50,
MicroChem, Newton, MA). The wafers are baked to promote the adhesion of the SU-8,
patterned by UV exposure through a photomask corresponding to the device structure and
followed by the post-exposure baking. The final device pattern is then exposed by
submerging the wafers in the SU8 developer. After the residual SU-8 resist is cross-
linked, a positive mold is formed for the silicone polymer. A prepolymer base and curing
agent are mixed at a ratio of 10 : 1, poured over the patterned wafer and cured at 65 °C
for nearly 4 h. Individual devices are cut, and punched to make connecting holes into
each device with flat end dispensing needles (Integrated Dispensing Solutions Inc.,
Agoura Hills, CA). Oxygen plasma treatment is used to bond the devices to a 5 mm thick
PDMS substrate, because this procedure activates the PDMS surfaces and yields
irreversible bonding between the two surfaces. A hydrophilic surface treatment using
polyvinyl alcohol (PVA) is applied to the devices to minimize bubble formation and to
match surface wettability. The devices are incubated in a 1 wt. % PVA solution for 5 min
at room temperature, briefly followed by additional incubation at 120 °C. This procedure,
repeated until an even coating over the device surface is achieved, is necessary to
facilitate PVA adhesion to the device surface after the excess solution is removed by
vacuum.
46

The sorting device (Figure 3.2 (a)) has three inlets merged into a 500 µm wide main
channel. Droplets enter through the middle inlet, while the upper and the lower inlets
provide sheath flows that confine the droplet flow to the center of the channel. The
channel is then bifurcated to two outlets where the sorted droplets are collected. The
channel height is 100 µm, equal to those of the inlets and outlets. The width is 200 µm
for each sheath inlet, and 100 µm for the droplet inlet. The outlet widths are designed
differently, 150 µm and 300 µm, respectively. The front wall thickness where the
ultrasonic beam enters the channel is set to 250 µm, to ensure that sufficient sound
energy can be transmitted to those droplets flowing in the channel. The channel wall
thickness measured by echo signal for positioning ultrasound transducer was 148 µm, in
Figure 3.2 (b).
47

Figure 3.2: Layout of the PDMS microfluidic device fabricated by conventional soft
lithography and the microfluidic channel wall thickness. (a) The upper and lower inlets
provide sheath (or guiding) flows that hydrodynamically focus the droplet stream to the
center of the channel, while the middle inlet delivers the droplet flow. (b) The channel
wall thickness measured by echo signal for positioning ultrasound transducer was 148
µm.
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
3.50E-06 3.70E-06 3.90E-06 4.10E-06 4.30E-06 4.50E-06 4.70E-06 4.90E-06
148 µm 518 µm
200ns 700ns
Transducer(30MHz)
Micro Fluidic channel wall
Wall width Channel width
Pulse-Echo Signal
Time (s)
Amplitude (V)
Sample Inlet Channel Sorting Channel
Outlet Channel
Sheath inlet channel
Main channel 0.5 mm
10 mm
0.3 mm
0.15 mm
0.1 mm
0.2 mm
Top View
(b)
(a)
48

Hydrodynamic focusing is employed to maximize the ultrasonic exposure to flowing
droplets during the sorting operation. It is very important to make sure that the focused
stream is centered along the main channel. In order to find a proper stream width where
steady flows of 50 µm and 100 µm droplets are formed, the sheath flow rate (fsh) is
varied from 2 µl/min to 14 µl/min with a step increment of 2 µl/min, while the droplet
flow rate (fd) is fixed at 2 µl/min. Each flow rate is adjusted by two syringe pumps (NE-
1000 Multi-Phaser, New Era Pump System Inc., NY, USA). Figure 3.3 illustrates the
change in the focused stream width as a function of flow rate ratio (r = fsh/fd). The
focused widths at r = 1 and 7 are 120 µm and 20 µm, respectively. fd and fsh are 2 µl/min
and 6 µl/min respectively, determined by the size and position of the droplets. With fsh
set at 6 µl/min, a 60 µm focused stream width is achieved. In particular, a uniform
droplet flow centered along the middle of the channel was experimentally observed under
these conditions, even for 100 µm droplets larger than the stream width. As the width
gets larger than 60 µm, the flow pattern of 50 µm droplets becomes more irregular within
the stream width.
49

Figure 3.3: Plot of focused stream width as a function of the ratio of sheath flow rate (fsh)
to droplet flow rate (fd). (a) fsh is adjusted from 2 µl/min to 14 µl/min with a step
increment of 2 µl/min, whereas fd remains constant at 2 µl/min (b) The stream width for
uniform droplet flow is found to be 60 µm when fd = 2 µl/min and fsh = 6 µl/min, as a
triangular marker indicates in (a).
Sample Inlet Channel Sorting Channel
Outlet Channel
Sheath inlet channel
Main channel 0.5 mm
10 mm
0.3 mm
0.15 mm
0.1 mm
Top View
60 μm
Ratio of Sheath Flow Rate to Sample Flow Rate
0 1 2 3 4 5 6 7 8
Focused Width ( m)
0
20
40
60
80
100
120
140
(b)
(a)
50

3.2.3 Multi-parameter Sorting using Integrated Backscatter Coefficient
Backscattered echo signal from the droplet by transmitted the 30-MHz single pulse,
sensing beam was received by the same transducer, and then the Integrated Backscatter
(IB) coefficient of a single micro-droplet was calculated.
The IB coefficient is defined as the ratio of backscattered energy from a scatterer volume
to that from a planar quartz reflector, expressed in the frequency domain as (Shung,
2006):

(

∫
| |

| |

)
where V(f) and R(f) are the Fourier transforms of an echo signal v(t) from a scatterer and
the reference signal r(t) from the reflector, respectively. fc and Δf are the resonant
frequency of r(t) and its half-width bandwidth. 2Δ f is the band width. The instrument
dependence of power spectrum analysis is eliminated by normalizing to the reference
spectrum of known reflectivity. Note that is measured prior to the experiment.

51

Theoretical IB coefficients H(f) can be approximately calculated by T-matrix analysis
method, and H(f) was calculated as (Waterman, 1969; Varadan et al., 1979; Kuo et al.,
1994):
[

|∑

|]

where kw is the ultrasonic wave numbers at the surrounding media, and m is the orders of
the spherical functions.
Lee and his coworkers showed IB coefficients experimentally obtained as a function of
droplet radius (Lee et al., 2011b). They fabricated a P[VDF-TrFE] broad band transducer
with a bandwidth of 112% for backscattering measurement from a single microdroplet.
The transducer has f-number = 1 and center frequency of 24 MHz. Oleic acid droplets of
64 µm and 90 µm diameter was fabricated in microfluidic channels. Backscattered echo
signal from the droplet by ultrasound bema driven the 15 MHz single pulse was received
by the same transducer, and then IB coefficient of a single microdroplet was calculated.
Figure 3.4 shows their experimental set-up and results.

52

Figure 3.4: (a) Experimental set-up for IB measurement (Lee et al., 2011b) (b)
Comparison of IB coefficients between experimental results and theoretical T-matrix
values (Lee et al., 2011b) IB coefficients were experimentally measured as a function of
droplet size. The mean IB coefficient of a single droplet was -107 ± 0.9 dB for cluster A,
and -93 ± 0.9 dB for cluster B. The measured results were in agreement with the
theoretical values based on the T-matrix method. The results indicate that the IB
coefficient may be used as a useful index to distinguish one particle from the others based
on several physical properties of the particle, including size and acoustic impedance. This
work was performed as an initial attempt to realize ultrasonic cell sorting in which both
cell sensing and sorting are carried out by sound beams.

53

3.2.4 Experimental Configuration and Procedure
The schematic diagram of an acoustic sorting device illustrates the location of the
transducer’s focus and the focused stream with respect to the main channel (Figure 3.5).
The sensing and the sorting beams are sequentially transmitted from the transducer,
interrogating and pushing the moving droplets from the outside of the channel. Both the
microfluidic structure and the transducer are immersed in a DI water chamber. The
transducer, whose focal length is 3 mm, is mounted on a three-axis manual linear stage
coupled to a goniometer (OptoSigma, Santa Ana, CA, USA) for precise alignment with
the main channel. The beam axis of the transducer is perpendicular to the channel wall in
order to effectively measure the IB coefficient of these streaming droplets and to
simultaneously sort them with acoustic radiation forces. The sorting beam is initiated by
the LabVIEW to direct the 100 µm droplets to the upper sheath flow. Focused droplet
motion is monitored via a CCD camera (InfinityX, Lumenera, USA) assembled with a
stereomicroscope (SMZ1500, Nikon, Japan). For signal transmission and reception, the
transducer is equipped with an arbitrary waveform generator (AFG3252, Tektronix,
Anaheim, CA, USA), a pulser/receiver (Model 5900PR, Panametrics-NDT, USA), and a
50 dB power amplifier (525LA, ENI, Rochester, USA).
A single cycle of a 30 MHz sinusoidal wave for sensing is input to the transducer from
the pulser/receiver. The peak-to-peak amplitude is 63 V
pp
and the pulse repetition
frequency (PRF) is 2 kHz (Figure 3.7a). Echo signals scattered from the continuous
droplet flow are received by the same transducer and digitized by a 12-bit analog-to-
54

digital converter (ADC) board (CS12400, GaGe, USA), at a sampling rate of 400 MS∙s
-1
.
A custom-programmed LabVIEW routine calculates the IB values corresponding to those
echoes, identifying the droplet size in real time. Based on the echo amplitude and the IB
coefficient obtained, the LabVIEW program switches the transducer from the sensing to
the sorting mode by sending a trigger signal to the waveform generator. In the sorting
mode, the transducer is driven by 2,000-cycled 30 MHz sinusoidal bursts whose peak-to-
peak amplitude and PRF are 63 V
pp
and 200 Hz (Figure 3.7b). A specific droplet is then
diverted away from the center channel by acoustic forces emitted from the transducer. In
order to find the driving voltage required for the sorting, the voltage is slowly increased
from 16 V
pp
, until 100 µm droplets begin to be pushed into the sheath flow.

55

Figure 3.5: Experimental arrangement for acoustic droplet sensing and sorting device. (a)
Top view (b) side view.

56

Figure 3.6: Experimental set up.

Figure 3.7: Illustration of (a) sensing and (b) sorting signals driving the transducer.
57

3.3 Results and Discussion
Pulse echo measurements are made at the focus of the transducer, 100 µm deep into the
channel. The center of the focused stream width (or the droplet center) is located slightly
beyond the focus by 150 µm, (1) to satisfy the mirror theory (Yuan et al., 1986) for
validating the backscattering measurement, and (2) to ensure that the sensing beam is
wide enough to encompass a droplet, which can be as large as 100 µm.
The temporal variations of typical RF echo magnitudes and their corresponding IB
coefficients are displayed in Figure 3.8. In particular, the peak amplitude is 13.5 ± 5.0
mV
pp
for 50 µm droplets and 38.0 ± 10.9 mV
pp
for 100 µm droplets. The ambient noise
level is 7.3 ± 0.5 mV
pp
. The measured IB coefficients are -103.6 ± 3.1 dB for 50 µm and
-95.4 ± 3.2 dB for 100 µm, whereas the coefficient is -114.0 ± 1.0 dB for the background
noise (Figure 3.9). It is shown that the droplet size is linearly related to the IB coefficient
on the logarithmic scale (Figure 3.10). The sensing capability of this sorting device is
evaluated by the sensing efficiency, the number of size-identified droplets divided by the
total number of droplets. The number of those droplets is manually counted from the
recorded movies.

58

Figure 3.8: Acquired echo signals and their IB coefficients for droplets passing through
the sensing beam are plotted over time.

Time (sec)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Echo Signal Amplitude (V
pp
)
0.00
0.01
0.02
0.03
0.04
0.05
0.06
Integrated Backscatter Coefficient (dB)
-120
-115
-110
-105
-100
-95
-90
-85
Echo signal Amplitude (V
pp
)
Integrated Backscatter Coefficient (dB)
59

Figure 3.9: Relation between echo amplitudes and integrated backscatter (IB) coefficients
corresponding to differently sized droplets.
Echo Signal Amplitude (V
pp
)
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Integrated Backscattering Coefficient (dB)
-120
-115
-110
-105
-100
-95
-90
Noise
50 m
100 m
60

Figure 3.10: The droplet size is more linearly related to the IB coefficient than to the echo
amplitude.

Noise 50 µm 100 µm
Echo Signal Amplitude (V
pp
)
0.00
0.02
0.04
0.06
0.08
Integrated Backscattering Coefficient (dB)
-120
-115
-110
-105
-100
-95
-90
Echo Signal Amplitude (V
pp
)
Integrated Backscattering Coefficient (dB)
61

For 50 µm and 100 µm droplets, the acquired sensing efficiencies are 98.6% and 99.0%,
respectively. The sorting efficiency is determined as in the sensing mode, with 234
droplets being tested (139 for 50 µm droplets and 95 for 100 µm droplets). The measured
sorting efficiencies are 99.3% and 85.3%, respectively.

Table 3.1: Summarized results for peak-to-peak echo amplitude, integrated backscatter
coefficient, sensing and sorting efficiencies for 50 µm/100 µm droplets.

Incomplete sorting efficiency (<100%) for 50 µm droplets may occur when a group of
droplets flows closely together. Since the PRF of the sensing beam (= 2 kHz) is higher
than that of the sorting beam (= 0.2 kHz), the sorting beam may miss some of the droplets,
while all the droplets in the group are scanned by the sensing beam. The experimental
data also confirm that one droplet in every 139 miss the sorting beam. The sensing and
sorting results are summarized in Table 3.1. The degraded sorting efficiency for 100 µm
droplets may be because part of the channel is intermittently blocked by those 100 µm
droplets agglomerated along the channel wall, resulting in a non-uniform stream width.
This type of two-dimensional hydrodynamic focusing has also been known to be
Parameter 50 µm droplets 100 µm droplets Ambient noise
Peak-to-peak echo amplitude (mV
pp
) 13.5 ± 5.0 38.0 ± 10.9 7.3 ± 0.5
Integrated backscatter coefficient (dB) -103.6 ± 3.1 -95.4 ± 3.2 -114.0 ± 1.0
Sensing efficiency (%) 98.6 99.0 Not applicable
Sorting efficiency (%) 99.3 85.3 Not applicable
62

intrinsically problematic in flow cytometry, causing a non-uniform velocity distribution
of small particles, e.g. cells or molecules, in the vertical direction of the channel (Wolff et
al., 2003). As such vertically spread targets pass through a sensing zone in the channel,
they may leave undetected droplets behind (Wang et al., 2005b). The sorting beam, due
to its increased number of burst cycles, generates strong enough acoustic radiation forces
to push the target droplets towards the channel wall. In our recent work, the applied force
produced by the same transducer was reported to be in the range of a few tens of nano-
Newtons (Lee et al., 2009b; Lee et al., 2010b) much stronger than by other sorting
techniques. Note that optical or dielectrophoretic forces have mostly been in the pico-
Newton range (Brouhard et al., 2003; Wei et al., 2009). The 100 µm droplets are then
directed to the upper sheath flow, while no forces are applied to the 50 µm droplets in
Figure 3.11.

Figure 3.11: The 100 µm droplets sorting processes from a mixture of different size
droplets. (a) The targeted 100 µm droplet is approaching to defined sensing zone. (b)
When the droplet passes through the sensing zone in red box, the 100 µm droplet is
pushed by acoustic radiation force. In the contrast, non-targeted droplet, 50 µm droplet
(yellow triangle mark) is not sorted, following the sensing zone. Note that green arrow
indicates sample flow, and blue arrow shows waste channel, and red arrow means
collection channel.
63

After passing the channel bifurcation, the droplets are collected in two separate outlets.
The sorting efficiency can be improved by making the channel’s cross-sectional area
larger than the droplet size. Higher frequencies and PRFs of the sensing beam, along with
faster flow rates, are needed to expedite the sorting speed and to offer fine spatial
resolutions, essential for separating small biological cells. On the other hand, plant cells
and protoplasts whose diameters are as large as 95 µm, have been sorted with flow
cytometers (Harkins et al., 1987). Since the proposed technique is developed for targets
of similar size, it can also be useful to separate those large cells. The calibrated pressure
applied by the transducer at the focus is 4.7 MPa under the sorting conditions, but the
actual pressure inside the channel may be lower than that value because of reflection at
the fluid–channel interface. Given the much larger droplet size than that of typical cells,
e.g. erythrocytes or leukocytes, a lower pressure level arising from the higher frequency
transducers may be sufficient in the case of cell sorting. The whole experimental
apparatus, including the transducer and the microfluidic device, is immersed in a water
bath to promote ultrasonic transmission into the main channel, making the proposed
sorting system relatively bulky. The current system can be simplified by building an
additional guiding branch to the main channel. The transducer can be integrated with the
main channel through the guiding structure, which is filled with a small amount of the
liquid medium. The sound beam can directly interrogate the target without requiring a
large space of liquid immersion in between. This type of design can also reduce reflection
at the fluid–channel interface of the proposed system. For increasing the operating speed
of the sorting process, an advanced circuit design is necessary to replace a series of bulky
64

electronic instruments. Higher frequency transducers and microfluidic channels with
better hydrodynamics will also be constructed in the near future to apply the proposed
approach to biological cell sorting. Typically, the two-dimensional hydrodynamic
focusing utilized in this work is implemented only in the horizontal plane of the channel,
and needs to be improved by additionally compressing the droplet stream in the vertical
direction (Mao et al., 2007). When the target is detected, the LabVIEW program
commands trigger signals for pushing the droplet within 5 µs. The process, including
signal acquisition, data analysis, and final sorting, is completed within 18 ms (about 60
Hz in terms of sorting throughput) for one droplet, considerably slower than conventional
cell sorting techniques, e.g. FACS (40 kHz) or FADS (300 Hz). Because separate
electronic equipment is utilized for data processing and communication between the
LabVIEW panel and the transducer during the sorting, the delayed response time
(approximately) between them results in such a low throughput. Integrated circuit design,
replacing those bulky units, is thus required to further reduce the operating time,
consequently increasing the throughput of the proposed system. Such enhancements will
enable the proposed method to be used for droplet-based applications, e.g. bioassay
compartmentalization, where individual droplets need to be analyzed and separated for
studying the variety of cell expressions induced within each droplet compartment.

65

3.4 Conclusion
In this chapter, an acoustic droplet sensing and sorting device for microparticles and cells
is reported. Non-contact detection and sorting of lipid microspheres by a high frequency
ultrasound beam in a microfluidic device is demonstrated in real time via hydrodynamics
and acoustic radiation forces. The sensing beam is generated by short pulses emitted from
a high frequency ultrasonic transducer, probing individual droplets in the stream.
Frequency-dependent backscattering properties are exploited as a sorting criterion, where
the IB coefficients for 50 µm and 100 µm droplets are measured. The sorting beam of
sinusoidal bursts produced by the same transducer is transmitted to divert 100 µm
droplets away from the main channel. The sorting beam is at a higher intensity level,
producing strong enough acoustic radiation forces to push them towards the outer wall
near the bifurcating point. The instantaneous transition between the sensing and the
sorting modes is accomplished by a LabVIEW controlled ADC board, depending on the
acquired IB coefficients. The efficiency in each mode is evaluated for mixed-sized
droplets and is found to be comparable to those of other conventional methods. Hence,
the results suggest that this ultrasonic technique may have the potential to be further
developed as a fast cell or particle sorter by increasing the flow rate and the ultrasound
frequency.

66

CHAPTER 4 MATERIAL PROPERTIES-BASED
MICROFLUIDIC MICRO-PARTICLE SENSING
4.1 Introduction
Conventional medical ultrasound, which has frequency range of 2 MHz to 10 MHz and
the spatial resolution of the order of a few millimeters, is compatible in tissue level
characterization. In the contrast, high frequency (35 MHz to 100 MHz) has the spatial
resolution improved to several tens of micrometers with larger attenuation, which
decreases penetration depth of ultrasound. The resolution can also be improved to a few
micrometers through highly focused high frequency ultrasound transducers. The
maximum penetration depth is approximately 10 mm for 20 MHz and 1.5 ~ 3 mm for 75
MHz ~ 100 MHz, even though the exact depth is determined by the acoustic attenuation
properties of the tissues (Passmann et al., 1996; Foster et al., 2002; Jasaitiene et al., 2011).
The improved fine spatial resolution and significant decreased penetration depth confine
high frequency ultrasound to specialized applications such as ophthalmic, dermatologic,
intravascular, small animal, and molecular imaging. The fine spatial resolution satisfies
their requirement of image quality regardless of relative low penetration depth of high
frequency ultrasound compared to that of conventional medical ultrasound frequency.
The higher temporal resolution could make them characterize micro size tissues, even a
single cell.
Micro-particle separation and sorting technologies combined with high frequency
ultrasound could have been employed in such biomedical applications as cancer
67

diagnosis and study of cell gene expression by tissue characterization techniques. Among
conventional sorting devices and systems, Fluorescence Activated Cell Sorter (FACS)
using light scattering with flow cytometry has been popular for these applications due to
its high throughput and accurate sorting performance. However, FACS is bulky and
complicated and needs well trained users for pre-treatment of a sample. Micro systems
and MEMS technology has been investigated to overcome these drawbacks.
High frequency ultrasound microbeam integrated with micro-fluidics channel has been
reported as one of alternatives. Its feasibility as a single cell sorter has been demonstrated
at the previous chapter’s researches. The capability to accurately measure scattering from
single cell is crucial if ultrasound scattering is to be used as the sensing mechanism as
well in a cell sorter. Lee and Shung showed size-based single droplet discrimination for
cell sorting applications at 30 MHz by analyzing scattered echo signals from single cell
(Lee et al., 2012b). They had calculated integrated backscatter coefficient and amplitude
of echo signals in real time for sensing micro-particle characterization.
Spectrum analysis of ultrasound echoes gives us useful information for ultrasonic
characterization of biological tissues, based on different acoustic properties of tissues.
With the long ultrasound history, ultrasonic spectrum analysis techniques have been
being developed, such as integrated backscatter (IB) coefficients in dB (O’Donnell et al.,
1979), spectral slope (dB/MHz) and spectral intercept (dB; extrapolation to zero
frequency) (Lizzi et al., 1976; Lizzi et al., 1987) and midband fit (dB; the value of the
regression line at the center frequency of the spectral band) (Lizzi et al., 1997; Feleppa et
68

al., 1996; Feleppa et al., 1997) in the frequency domain, by analyzing received RF echoes
from biological tissues.
In this chapter, an ultrasound cell sorter device was evaluated as material properties-
based cell sorter without any pre-treatment of samples. Two type micro objects,
polystyrene micro-spheres and oleic acid lipid droplets, were applied for this experiment,
which have quite different material properties and similar size. Highly focused 86 MHz
lithium niobate single element ultrasound transducer was designed and built for
identifying difference of two micro-particles by analyzing scattered echo signals with the
fine resolution.

69

4.2 Material and Methods
4.2.1 High Frequency Ultrasound Transducer
An 86 MHz lithium niobate press-focused single beam acoustic transducer was designed
by a Krimholtz, Leedom and Matthaei (KLM) model and fabricated using conventional
transducer technology. Depending on the material properties of lithium niobate single
crystal, the optimized thickness of the active element is 34 µm with an aperture size of
2.0 mm. As the first acoustic matching layer, a λ/4–thick (3 µm) silver epoxy made from
a mixture of silver particles (2-3 µm diameter) and electrical insulating epoxy was cast
onto the negative side of the chrome / gold sputtered element. A very lossy (an
attenuation of 112 dB/mm at 30 MHz) conductive silver epoxy (E-Solder 3022) was
served as a backing layer, which was cast onto the positive side of the sample. The
acoustic stack was then press-focused at a focal length of 1.5 mm so that the f–number of
the device was as low as 0.75. A 3.2 µm-thick parylene layer was vapor-deposited on the
front face of the transducer to serve as the second acoustic matching layer and a
protective layer as well. The transducer was finally assembled in an SMA connector for
further experiments. The performance of the transducer was measured by pulse-echo
method from quartz plate reflector with a pulser/receiver (Model 5900PR, Panametrics-
NDT, USA) in 0 dB amplifier gain because commercial hydrophone measurement over
60 MHz high frequency is not available and reliable. Resonance center frequency of built
single element transducer was 86 MHz and the 6 dB bandwidth of the beam is ~ 37% as
70

shown in Figure 4.1. The axial and lateral resolution, beam width at focus, is theoretically
calculated as the following equations.

(1)

(2)
where λ is the wavelength defined as the ratio of the sound speed to the center frequency
of ultrasound transducers (c / f
0
), f
#
is the f-number defined as the ratio of focal length to
aperture size of ultrasound transducers (Z
f
/ D), and BW
-6dB
(6dB bandwidth) represents
the width of the frequency response curve.
Expected axial and lateral resolution of high frequency ultrasound transducer based on
theoretical calculation was ~ 23 µm and ~13 µm respectively.

Figure 4.1: Pulse-echo signal and its frequency spectrum analysis and fabricated high
frequency single element ultrasound transducer
71

4.2.2 Polymer Micro Spheres and Lipid Droplets
Monodisperse polystyrene microspheres and oleic acid lipid droplet was prepared for
sensing and identifying two different material micro-particles and similar size.
Polybead® Microspheres of 45 µm mean diameter (Polysciences, Inc., Warrington, PA,
USA) contained a slight anionic charge form sulfate ester were purchased, which has
coefficient of variance (CV): ≤10% due to precise monodisperse particle size
distributions. The microspheres were made by crosslinking with divinylbenzene (DVB),
and stored in an economical 2.5% solids (w/v) aqueous suspension with minimal
surfactant. Due to high concentration of 4.99 × 10
5
particles/ml, the polystyrene
microsphere was diluted for final experimental use.
Oleic acid lipid droplets were synthesized in poly (dimethyl) siloxane (PDMS)
microfluidic channels described in previous chapters for the micro-particle
characterization experiments by scattering echo singles. Micro lipid droplets of mean
diameter of 45.4 µm were used. Size distribution of oleic acid lipid droplet sample of 100
droplets was shown in Figure 4.2 with coefficient of variance < 17%. The droplets and
polystyrene micro-spheres are similar in size with large difference of material properties
such as density, sound speed, acoustic impedance and elasticity coefficient.

72

Figure 4.2: Histogram of size distribution of micro lipid droplet sample of 45 µm in mean
diameter (CV < 17%).

4.2.3 Ultrasonic Spectral Parameters
Ultrasonic spectrum analysis procedures in the frequency domain can extract more detail
information from diagnosis medical ultrasound images. Calibration procedures in the
analysis can reduce system artifacts and noise, and allow quantitative measurements of
back scatted signals (Lizzi et al., 1983). These parameters can be applied to single micro-
particle sorting due to their advantages in characterization or identification of tissue
microstructure. Three type ultrasonic spectrum parameters were employed for material
properties-based micro-particle discrimination in ultrasonic micro devices.
23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71
0
5
10
15
20
25
30
Frequency
Droplet Diameter (µ m)
Diameter Distribution - 45um Droplets
73

(1) Integrated backscatter (IB) coefficients in dB use a frequency domain averaging
technique and can reduce the effects of frequency dependent fluctuations due to
inhomogeneous scattering distributions in tissues (O’Donnell et al., 1979). IB coefficient
is defined as the ratio of the frequency average of the backscatter energy from a scatter
volume over the bandwidth of the signal to the one from a flat quartz or glass reflector
(Shung, 2006).

(

∫
| |

| |

)
(2) Spectral slope in dB/MHz and spectral intercept in dB (Lizzi et al., 1976; Lizzi et al.,
1987) utilize the regression line of the spectral band. Spectral slope is slope of the
regression line, which is computed by calibrated power spectrum. The calibrated power
spectrum is plotted by RF spectrum from sample echo over calibration spectrum of echo
signal of a flat reflector. Spectral intercept can be calculated by the extrapolation of the
regression line to zero frequency.
(3) Midband fit in dB is defined the value of the regression line at the center frequency of
the analyzed spectral band (Lizzi et al., 1997; Feleppa et al., 1996; Feleppa et al., 1997).
The midband fit value is directly related to IB coefficient (Lizzi et al., 2006).
Figure 4.3 show the calculation processes of spectral slop and intercept, and midband fit
(Lizzi et al., 2003; Lizzi et al., 2006).

74

Figure 4.3: The calculation processes of ultrasonic spectral parameters. (a) RF spectrum
from sample (solid line) and glass plate spectrum for calibration procedure from a flat
reflector (dash line) (b) Normalized calibrated spectrum (solid line) calculated by RF
spectrum over glass plate spectrum in spectral band width and linear regression line (dash
line) (c) Spectral slope (slope of regression line) and intercept (INT) at zero frequency by
extrapolation, and midband fit (MBF) at center frequency, f
0
. Note the Figure 4.3 is
modified from Figure 3 and 4 (Lizzi et al., 2003) and Figure 2(b) (Lizzi et al., 2006).

4.2.4 Experimental Configuration and Procedure
The same experimental setup was used in Chapter 3. A PDMS device, immersed in a
distilled water chamber, consisted of two sheath channels on the side and one main
channel providing sample of polymer micro-spheres and lipid droplets at the center. The
transducer was positioned outside the channel, perpendicularly aiming at the channel wall.
To increase sensing sensitivity, a highly focused high frequency ultrasound transducer
was positioned at the location for being able to cover entire sensing zone. The reference
position was located and precisely adjusted by a three-axis linear and tilt stages, based on
pulse-echo signals in Figure 4.4. After finding inside channel wall from the location of
transducer, the transducer moved 200 μm toward inside channel. Flow rates of the sample
and sheath solutions were 0.5 and 1.5 in μl/min, respectively. Echo signals scattered from
(a)
(b)
(c)
75

these streaming micro objects of different material and similar size were acquired by
LabVIEW program through a data acquisition board with fast data transfer rate 200 MB/s
in real time. 12 bits high resolution data acquisition board (CS122G1; GaGe, USA) of
PCI type with sample rate 2 GS/s and 500 MHz -3dB Input bandwidth was purchased
because of an 86 MHz high frequency ultrasound. After collecting data, the integrated
backscatter (IB) coefficient and slope was analyzed by a custom-built MATLAB program.

Figure 4.4: Calibration of transducers initial position through pulse-echo signals. The
channel wall thickness by echo signal was 148 µm.

148 µm 518 µm
200ns 700ns
Wall width Channel width
Pulse-Echo Signal
Time (s)
Amplitude (V)
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.000002 0.0000022 0.0000024 0.0000026 0.0000028 0.000003 0.0000032
Transducer(86MHz)
Micro Fluidic channel wall
76

4.3 Results and Discussion
For sensing and identifying microparticles, single pulse microbeam of 32 V
pp
and 86
MHz sinusoidal signals, which was generated and amplified by function generator and
amplifier was driven with high Pulse Repetition Frequency (PRF) of 2 KHz to prevent
from missing particles, and the echo signals were averaged to reduce noise effects before
analyzing amplitude and calculating IB coefficient in real time. The device was enabled
to average echo signals reducing noise by low flow rate.
The IB coefficient was -79.37 ± 3.4 dB for 45 μm lipid droplets, while it was -76.40 ±
3.07 dB for 45 μm polystyrene micro-beads, which was not able to clearly discriminate
different material properties in similar diameter size in Figure 4.5. However, the spectral
slope value in dB/MHz was 3.04E-07 ± 1.19E-07 for 45 μm lipid droplets while it was
-2.08E-07 ± 1.13E-07 for 45 μm polystyrene micro-beads. Those droplets can be readily
identified by spectral slope, which is shown in Figure 4.6.

77

Figure 4.5: Integrated Backscatter coefficient of lipid and polystyrene micro-particle. The
IB coefficient was -79.37 ± 3.4 dB for 45 μm lipid droplets, while it was -76.40 ± 3.07
dB for 45 μm polystyrene micro-beads, which was not able to clearly discriminate
different material properties in similar diameter size in Figure 4.5. Note the graph was
drawn by positive value for better appearance.
79.37
76.40
60.00
65.00
70.00
75.00
80.00
85.00
Integrated Backscatter (IB) Coefficient (dB)
Integrated Backscatter (IB) Coefficient
Lipid
Polystyrene
78

Figure 4.6: Spectral slope of each particle.The spectral slope value in dB/MHz was
3.04E-07 ± 1.19E-07 for 45 μm lipid droplets while it was -2.08E-07 ± 1.13E-07 for 45
μm polystyrene micro-beads. Those droplets can be readily identified by spectral slope.

3.04E-07
-2.08E-07
-4.0E-07
-3.0E-07
-2.0E-07
-1.0E-07
0.0E+00
1.0E-07
2.0E-07
3.0E-07
4.0E-07
5.0E-07
Spectral Slope (dB/MHz)
Spectral Slope
Lipid
Polystyrene
79

4.4 Conclusion
This research reports experimental results from scattering measurements carried out on
different materials and similar size at a frequency of 86 MHz. These results show that it
is possible to characterize micro-particles, in this case two different materials and similar
size from scattering.

80

CHAPTER 5 A LIVE SINGLE CELL SENSING AND
DISCRIMINATION
5.1 Introduction
In this Chapter 5, the capability of current ultrasound technology is discussed, as single
cell-based sorting device for sorting a live cell. Highly focused single element high
frequency ultrasound was able to distinguish human red blood cells from human
leukemia cells on agar phantom surface. The potential and performances of devices as a
live cell sorter are verified by experimental results. Bioeffect of high frequency
ultrasound beam is briefly discussed.

5.2 Material and Methods
5.2.1 Live Cells Preparation
Human leukemia cells (K-562 cell line, ATCC CCL-243) were cultured in Dulbecco’s
modified eagle medium (DMEM, GIBCO, Invitrogen) supplemented with 10% fetal
bovine serum (FBS) and 1% Penicillin–Streptomycin–Neomycin (PSN, GIBCO,
Invitrogen) in an incubator supplied with 5% CO
2
and set at 37 °C. Human red blood
cells (RBC) were prepared and stored in a mixed solution of alsever’s solution and
phosphate-buffered saline (PBS) for cell sensing experiments with high frequency
ultrasound.
81

5.2.2 High Frequency Ultrasound Cell Sorting System
The system for previous experiments had components connected by BNC coaxial cables,
providing undesirable spurious noises. A high frequency ultrasound cell sorting system
was custom designed and built for reducing noise and better mobility.
The ultrasound cell sorting system requires several modules such as power amplifier and
pre-amplifier and protection circuits. The power amplifier (PA) is used to drive the
ultrasonic transducers and the pre-amplifier is to amplify the echo signal larger to allow
detection. As the operating frequency of the transducer increases, the PA design would be
more challenging due to higher demanding specification such as sensitive noise level and
wide bandwidth of the ultrasound system. Another issue is that the commercial
equipment is quite bulky so the long coaxial cables need to be used to be integrated with
other components and to minimize the cable loading generated effect on the high
frequency transducer performances. Therefore, a low-noise and wideband portable all-in-
one system is developed. Figure 5.1 shows block diagram of designed module for system.
82

Figure 5.1: Block diagram of designed module

A wide-band Class A power amplifier was designed using LDMOS (PD57018). The
topology of the PA is a 2-stage common-source amplifier with inductor-capacitor
matching circuits. In order to estimate the performance of the power amplifier, the PA
was tested with a function generator and displayed the waveforms on the oscilloscope
and spectrum analyzer. The measurement results show that PA has a gain of 39.3-42.6 dB
in the range of 40 to 120 MHz and its current consumption of 1 A with 36 V supply
voltage. The measured 1-dB compression point was as a 45.35 dBm. The strongest
advantage is to generate noise amplitude up to 20 mV compared to commercial power
amplifier’s noise amplitude (>100 mV).
83

Figure 5.2: Schematic of 2-stage power amplifier

Since initial received echo-signal from transducer is quite small, we need to amplify the
signal for ultrasonic based analysis. The pre-amplifier mainly requires two specifications
- sufficient broadband gain and low noise figure. In order to satisfy the requirement,
PSA4-5043+ E-PHEMT based low noise amplifier is selected. The pre-amplifier is made
up of 2-stage cascaded common source topology with wide-band bias tee. The total
current consumption is 60 mV with 3 V supply voltage. The gain is 31-32 dB and the
noise figure is less than 3 dB.
84

Figure 5.3: Schematic of 2-stage pre-amplifier

Diode-based expander and limiter are used for high voltage and low voltage protection.
The function of an expander is to prevent the noise signal from the transmitter so that it
does not pass into the transducer and receiver. Additionally, the received small signal
from the transducer is blocked by the expander because the general echo signal’s
amplitude is quite small compared diode threshold voltage. Limiter can protect the
receiver from the high voltage signal generated from transmitter while low voltage echo-
signal can pass through the limiter. These two conventional topologies are used widely
for ultrasound’s pulse-echo measurement because of simple structure and low cost. In
these design, PMBD7000, double high-speed switching diodes, is used. These parallel
85

diodes have characteristics such as high switching speed, high reverse voltage, and short
reverse recovery time.

Figure 5.4: Schematic of diode-based expander for protecting power amplifier from low
voltage echo signals

Figure 5.5: Schematic of diode-based limiter for protecting pre-amplifier from high
voltage signals of power amplifier
86

Figure 5.6: Designed power amplifier module performance. Power amplifier gain is
measured at 86 MHz and different input voltages.

Figure 5.7: Designed high frequency ultrasound cell sorting system
0
5
10
15
20
25
30
0 1 2 3 4
Gain (dB)
Input Voltage (Vpp)
Power Amplifier Gain (at 86MHz)
1 cycle
N-cycles(2000 cycles) or CW
87

5.2.3 Experimental Configuration and Procedure
A highly focused 86 MHz single element transducer of f-number 0.75 was fabricated for
single cell scattering or sensing experiments.
In order to assess the performance of sensing and identifying live single cells, a tissue
mimicking phantom, agar substrate of 500 µm thickness, was fabricated in 30 mm culture
dish because acoustic impedance of agar substrate (1.50 ~ 1.55 Mrayls) is similar to that
of water (1.48 Mrayls) (Partanen et al., 2009).
The phantom materials were manufactured with gelatin and graphite powder. Agar gel
powder (Agar A360-500, Fisher Scientific) was hydrated with deionized water and n-
propanol. The solution was slowly stirred in order to avoid clumps during the hydrating,
and vacuumed (66 to 73 cmHg) to degas the solution for few minutes. After this process,
it was stirred to avoid forming a thin film on surface and heated on a hotplate around
70 °C in water bath in order to clarify the solution by dispersing colloid, and to release
gasses. After over 70 °C, the mixture was heated directly on hotplate until temperature
reached 90 °C with vigorously stirring to prevent burning the agar solution at the bottom
of beaker. Then, the mixture was vacuumed for short sessions to draw trapped air bubbles
to surface. When the gel solution was liquidified by these processes, graphite powder
(Silicon dioxide, Sigma-Aldrich, USA) of particle size between 0.5 µm and 10 µm was
added to increase scattering and absorption in typical tissue mimicking phantom.
However, the graphite powder was not added to the transparent window of optical
microscopy here. The phantom material was then rotated to be homogenously mixed in
88

the solution, and heated to prevent gelation. In the last process, the mixture was cooled to
allow cross-linking with slowly stirring, in the water bath until around 45 °C, and then
the solution was poured into the mold to produce phantom materials. The designed
phantom as a measurement platform was acoustically and optically transparent.
The cultured K562 cells and RBCs were suspended in a mixed solution of alsever’s
solution and phosphate-buffered saline, and then placed on the top of an optically
transparent agar substrate in 30 mm culture dish, where cell motions were monitored by
an inverted microscope (IX-71, Olympus, Japan). The transducer then interrogated the
cells from above, driven in a sinusoidal burst mode whose peak-to-peak voltage
amplitude was as high as 40 V. Single pulse was applied and the pulse repetition
frequency was set to 200 Hz. Prior to the cell sensing experiment, a pulse echo test was
performed to ensure that the cells were located on the focal plane (Figure 5.8). A single
cell was then randomly targeted within the microscope’s field of view and laterally
scanned by micro-stage of microscope in parallel with the focus. The experimental
schematic is shown in Figure 5.9 and mixed samples view in microscope in Figure 5.10.

89

Figure 5.8: A pulse echo signal from agar surface for location of ultrasound transducer.
Echo signal from agar surface is indicated in red box.

Figure 5.9: The experimental configuration.
890ns (~ 680 µm)
Agar - 1540 m/s
520ns (~ 620 µm)
Polystyrene 2400m/s
Microscope
86MHz Transducer
Agar Phantom
Alsever’s Solution + DPBS
Petri Dish
Cells
90

Figure 5.10: Leukemia cells (K562) and red blood cells. The 10 µm was placed on agar
surface as reference.

10um micro-particle
91

5.3 Results and Discussion
Acoustic particle discrimination on agar substrate surface was carried out by analyzing
back scattered signals using highly focused high frequency ultrasound beam.
A focused transducer was driven with a sinusoidal single cycle signal at the resonance
frequency. When a single live cell on the substrate surface was moved to the focal
position of ultrasound beam, echo signals were collected and analyzed with a microscope
and oscilloscope. Figure 5.11 shows measured echo signals from single live cell, K562
cell and RBC in beam center of focal plane of transducer. Note that beam center was
indicated as a pink circle of screen center.

92

Figure 5.11: Shows measured echo signals from single live cell, (a) K562 cell and (b)
RBC in beam center of focal plane of transducer. Note that beam center was indicated as
a pink circle of screen center.

(a)
(b)
93

Amplitudes of measured scattered signals are shown in Figure 5.12. Amplitude of echo
signals of K562 cell and RBC were 48.25 ± 11.98 mV
pp
and 56.97 ± 7.53 mV
pp
,
respectively. The integrated backscatter of the two different types of cells on the agar
surface is -89.39 ± 2.44 dB for K562 cell and -89.00 ± 1.19 dB for RBC, respectively
(Figure 5.13). The spectral slope in dB/MHz (Figure 5.14) and intercept in dB (Figure
5.15) were 3.00E-07 ± 1.91E-07 dB/MHz and -56.07 ± 17.17 dB for K562 cell while
7.75E-07 ± 9.21E-08 dB/MHz and -98.18 ± 8.80 dB for RBC. Factors of both sensing
and identifying were dependent on cell type, except IB coefficient. These spectral
parameters showed statistically significant difference by student t-test, including echo
amplitude except IB value. These spectral parameter values can be applied to sorting
decisions, and correlation between parameters was indicated in Figure 5.16.
Those signal analysis suggest that this acoustic sensing method, based on spectrum
analysis processes resulting from echo signal induced by highly focused high frequency
ultrasound micorbeam, might be able to be applied for single-particle sorting in real time
and high throughput.

94

Figure 5.12: Amplitude of echo signals of K562 cell and RBC.

48.25
56.97
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
Singal Amplitude (mV
pp
)
Echo signal Amplitude
K562cell
RBCs
95

Figure 5.13: The integrated backscatter of K562 cell and RBC. Note the graph was drawn
by positive value for better appearance.
89.39
89.00
84.00
85.00
86.00
87.00
88.00
89.00
90.00
91.00
92.00
93.00
Integrated Back Scatter (IB) Coefficient (dB)
Integrated backscatter (IB) Coefficient
k562cell
RBCs
96

Figure 5.14: The spectral slope in dB/MHz of K562 cell and RBC.

3.00E-07
7.75E-07
0.0E+00
1.0E-07
2.0E-07
3.0E-07
4.0E-07
5.0E-07
6.0E-07
7.0E-07
8.0E-07
9.0E-07
1.0E-06
Spectral Slope (dB/MHz)
Spectral Slope
k562cell
RBCs
97

Figure 5.15: The spectral intercept in dB of K562 cell and RBC. Note the graph was
drawn by positive value for better appearance.

56.07
98.18
0.0E+00
2.0E+01
4.0E+01
6.0E+01
8.0E+01
1.0E+02
1.2E+02
Spectral Intercept (dB)
Spectral Intercept
k562cell
RBCs
98

The biological effect of highly focused high frequency ultrasound is critical issue of
concern after live cells are exposed. Before it can be applied to in vivo studies, study
needs to be carried out to examine the effect of ultrasound on cell damages due to heating
and high pressure generated by high frequency ultrasound. Mechanical index (MI) and
thermal index (TI) have been employed for many years to evaluate the exposure levels of
conventional medical ultrasound frequency (2 MHz to 20 MHz) because of their
quantitative nature to estimate the risk of adverse ultrasonic effects due to the nonthermal
and thermal mechanisms. However, there are still no general guidelines for the safe use
high frequency ultrasound. It appears that it is prudent to study carefully the bioeffects of
acoustic beam at high frequency and highly focused acoustic intensity before it is used
for in vivo studies.
99

Figure 5.16: Correlation between parameters.

100

5.4 Conclusion
In previous chapters, we have presented acoustic devices using acoustic radiation force
for single-cell based sorting, which integrated the advantages of microfluidic channel and
highly focused high frequency ultrasound microbeam. The advantages of these devices
are their simpler structure and higher sensitivity for sensing, identifying and sorting on
flowing individual particles in a microchannel with minimal time delay between sensing
and sorting steps. We evaluated sensing and identifying performance of this method
through the use of two different live cell, leukemia (K562) and red blood cell. These
results demonstrated that the feasibility of ultrasonic scattering as a sensing mechanism in
the development of ultrasonic cell sorters.

101

CHAPTER 6 SUMMARY AND FUTURE WORKS
6.1 Summary
Different types of high frequency ultrasound cell sorting micro-devices are reported for
sample and cost-effective cell sorting device. Despite the fact that the basic concept and
sorting mechanism is straightforward and intuitive, acoustic sorting device have several
advantages such as high sensitivity without pre-treatment of samples and much simpler
processes of sorting. Not only that, but also the device has advantages of miniaturized
micro device scaling down by combining high frequency ultrasound with microfluidic
channels including micro flow cytometery, compared to conventional cell sorter and
methods.
Acoustic radiation force of high frequency ultrasound is shown to produce forces
sufficient for sorting or separating micro bioparticles by pushing through PDMS channel
wall. Ultrasonic spectral analysis procedures enable this proposed device to sense or
screen subpopulations of sample with high sensitivity and efficiency.
In this research, two different size droplets in same material and different material micro-
particles in similar size were clearly discriminated from spectral parameters such as
spectral slope and intercept with echo amplitude and IB coefficient. In size-based case,
50 µm and 100 µm diameter lipid droplets were separated by these suggested methods in
real time. In material properties-based case, it was able to sense and identify flowing
102

polystyrene microbeads and oleic acid droplets in 45 µm diameter inside microchannel
by spectral slope value in dB scale.
Lastly, it were carried out a live cell sensing and identifying experiments with human
leukemia and red blood cells through multi-parameters, which were verified by sensing
and discriminating different size in same material and different material properties in
similar size. The potential and performances of devices as a live cell sorter are verified by
experimental results. For further clinical applications, the necessity of more research
related to bioeffect of high frequency ultrasound beam was briefly described although
there exist currently no guidlines or criteria for high frequency ultrasound.

103

6.2 Future Works
In this research, the ultimate goal is to develop a simple and yet cost-effective micro
sorting device with high frequency ultrasound microbeam for individual particulates is
needed for precise bioassay analysis. To improve presented acoustic cell sorting method
for real cell sorting applications is needed. First, new microfluidic channel will be
designed and fabricated for hydrodynamic focusing of smaller bioparticles for improving
sensing sensitivity (>99%) due to vertically spread targets passing through sensing zone
in microchannel. Especially for higher frequency ultrasound, an alternative material with
lower acoustic attenuation for microfluidic channel is required due to high acoustic
energy loss of PDMS. Mechanical properties of PDMS, including acoustic attenuation,
could be improved by changing the ratio of a polymer to cross-linker without effect of
surface chemistry (Mata et al., 2005). For example, transparent RTV-2 silicone (P4,
Silicones Inc., NC, USA) showed better acoustic properties, including less attenuation of
high frequency ultrasound in our group’s experiments. Microfluidic channels built by
alternative materials could improve performance of ultrasound cell sorting devices.
In addition, flow system will be upgraded in order to do better flow control with custom
built LabVIEW program. Also, Higher frequency ultrasound microbeam will be applied
in this method for better sensitivity and approaching at much smaller sample size thank to
better resolution of higher frequency ultrasound beam. For less mechanical and thermal
damage of samples in single cell-based sorting device, sensing and sorting beam will be
evaluated by cell conditions monitored after sorting.
104

In this acoustic sorter, multi-parameter sorting and separation should be applied with this
ultrasound sensing and identifying technique in real time with high throughput. Multi-
parameter data analysis of conventional flow cytometry, including gating for
identification of homogenous cell populations and interpretation for finding correlations
between some characteristics of the identified cell populations (Bashashati et al., 2009),
is employed for identifying cells with similar properties in populations. Also, collected
data from sample are presented by various display methods, including histogram, dot or
density plot, and contour diagrams (Recktenwald, 1993), with specifically designed scale
such as logarithmic and logicle (Herzanberg et al., 2006). Cluster and classification of
collected data using statistical analysis could be applied for improvement of sensitivity
and calibration of device. Multi-parameters such as angular scattering, multi-frequency
and pulse width of sensing beam could be additionally employed for discrimination of
cells. These multi-dimensional analysis and selectively visualization of data are useful for
extracting physical characteristics of cells without pre-treatments of samples. It maybe
enables the real time acoustic cell sorting device to identify and separate bioparticles as
simpler method.
Current high frequency device is still relatively bulky and has low sorting rate with
outside equipment such as water chamber, axis stage, function generator, and microscope
for monitoring in spite of integrated new ultrasound system in Chapter 5 including pre-
amplifier and power amplifier for improving system sensitivity by reducing system and
outside noise. The system for ultrasound sorter as alternative tools is needed to be
simplified with integrated micro-chip circuit design replacing outside customized
105

LabVIEW program, which enable an ultrasound sorter to reduce the operating time,
leading high throughput of current system.

106

BIBLOGRAPHY
Baret, Jean-Christophe, Oliver J. Miller, Valerie Taly, Michaël Ryckelynck, Abdeslam
El-Harrak, Lucas Frenz, Christian Rick, et al. 2009. Fluorescence-activated droplet
sorting (FADS): Efficient microfluidic cell sorting based on enzymatic activity. Lab on a
Chip 9 (13): 1850-1858.

Bashashati, Ali, and Ryan R. Brinkman. 2009. A survey of flow cytometry data analysis
methods. Advances in Bioinformatics 2009 : 584603-19.

Beckman Coulter. Inc. 2014. MoFlo™ XDP, Retrieved from
http://www.beckmancoulter.com/wsrportal/wsr/research-and-discovery/products-and-
services/flow-cytometry/cell-sorters/moflo-
xdp/index.htm#2/10//0/25/1/0/asc/2/ML99030///0/1//0/,

Bhagat, Ali Asgar S., Hansen Bow, Han Wei Hou, Swee Jin Tan, Jongyoon Han, and
Chwee Teck Lim. 2010. Microfluidics for cell separation. Medical & Biological
Engineering & Computing 48 (10): 999-1014.

Bommannan, D., Hirohisa Okuyama, Paul Stauffer, and Richard H. Guy. 1992a.
Sonophoresis. I. the use of high-frequency ultrasound to enhance transdermal drug
delivery. Pharmaceutical Research 9 (4): 559-564.

Bommannan, D., Gopinathan K. Menon, Hirohisa Okuyama, Peter M. Elias, and Richard
H. Guy. 1992b. Sonophoresis. II. examination of the mechanism(s) of ultrasound-
enhanced transdermal drug delivery. Pharmaceutical Research 9 (8): 1043-1047.

Bonner, W. A., H. R. Hulett, R. G. Sweet, and L. A. Herzenberg. 1972. Fluorescence
activated cell sorting. Review of Scientific Instruments 79 (3): 404-409.

107

Brouhard, Gary J., Henry T. Schek III, and Alan J. Hunt. 2003. Advanced optical
tweezers for the study of cellular and molecular biomechanics. Biomedical Engineering,
IEEE Transactions on 50 (1): 121-125.

Cannata, Jonathan M., Timothy A. Ritter, Wo-Hsing Chen, Ronald H. Silverman, and K.
Kirk Shung. 2003. Design of efficient, broadband single-element (20-80 MHz) ultrasonic
transducers for medical imaging applications. Ultrasonics, Ferroelectrics and Frequency
Control, IEEE Transactions on 50 (11): 1548-1557.

Doerr, Allison. 2005. The smallest bioreactor. Nature Methods 2 (5): 326.

Erickson, David, and Dongqing Li. 2004. Integrated microfluidic devices. Analytica
Chimica Acta 507, (1): 11-26.

Feleppa, Ernest J., Andrew Kalisz, Jone B. Sokil-Melgar, Frederic L. Lizzi, Tian Liu,
Angel L. Rosado, Mary C. Shao, et al. 1996. Typing of prostate tissue by ultrasonic
spectrum analysis. Ultrasonics, Ferroelectrics, and Frequency Control, IEEE
Transactions on 43 (4): 609-619.

Feleppa, Ernest J., Tian Liu, Andrew Kalisz, Mary C Shao, Neil Fleshner, Victor Reuter,
and William R. Fair. 1997. Ultrasonic spectral-parameter imaging of the prostate.
International Journal of Imaging Systems and Technology 8 (1): 11-25.

Foster, F. S., M. Y. Zhang, Y. Q. Zhou, G. Liu, J. Mehi, E. Cherin, K. A. Harasiewicz, et
al. 2002. A new ultrasound instrument for in vivo microimaging of mice. Ultrasound in
Medicine & Biology 28 (9): 1165-1172.

Franke, T., S. Braunmüller, L. Schmid, A. Wixforth, and D. A. Weitz. 2010. Surface
acoustic wave actuated cell sorting (SAWACS). Lab on a Chip 10 (6): 789-794.

108

Fu, Anne Y., Hou-Pu Chou, Charles Spence, Frances H. Arnold, and Stephen R. Quake.
2002. An integrated microfabricated cell sorter. Analytical Chemistry 74 (11): 2451-2457.

González, Icí ar, Luis José Fernández, Tomás Enrique Gómez, Javier Berganzo, Jose Luis
Soto, and Alfredo Carrato. 2010. A polymeric chip for micromanipulation and particle
sorting by ultrasounds based on a multilayer configuration. Sensors & Actuators:
B.Chemical 144 (1): 310-317.

Harkins, Kristi R., and David W. Galbraith. 1987. Factors governing the flow cytometric
analysis and sorting of large biological particles. Cytometry 8 (1): 60-70.

He, Mingyan, J. Scott Edgar, Gavin D. M. Jeffries, Robert M. Lorenz, J. Patrick Shelby,
and Daniel T. Chiu. 2005. Selective encapsulation of single cells and subcellular
organelles into picoliter- and femtoliter-volume droplets. Analytical Chemistry 77 (6):
1539-1544.

Herzenberg, Leonard A., Richard G. Sweet, and Leonore A. Herzenberg. 1976.
Fluorescence-activated cell sorting. Scientific American 234 (3): 108-117.

Herzenberg, Leonore A., James Tung, Wayne A. Moore, Leonard A Herzenberg, and
David R. Parks. 2006. Interpreting flow cytometry data: A guide for the perplexed.
Nature Immunology 7 (7): 681-685.

Hwang, Jae Youn, Jungwoo Lee, Changyang Lee, Anette Jakob, Robert Lemor, Lali K.
Medina-Kauwe, and K. Kirk Shung. 2012. Fluorescence response of human HER2+
cancer- and MCF-12F normal cells to 200 MHz ultrasound microbeam stimulation: A
preliminary study of membrane permeability variation. Ultrasonics 52 (7) (9): 803-808.

Hwang, Jae Youn, Nan Sook Lee, Changyang Lee, Kwok Ho Lam, Hyung Ham Kim,
Jonghye Woo, Ming-Yi Lin, et al. 2013. Investigating contactless high frequency
ultrasound microbeam stimulation for determination of invasion potential of breast cancer
cells. Biotechnology and Bioengineering 110 (10): 2697-2705.
109

Hwang, Jae Youn, Changyang Lee, Kwok Ho Lam, Hyung Ham Kim, Jungwoo Lee, and
K. Kirk Shung. 2014. Cell membrane deformation induced by a fibronectin-coated
polystyrene microbead in a 200-MHz acoustic trap. Ultrasonics, Ferroelectrics, and
Frequency Control, IEEE Transactions on 61 (3): 399-406.

Jasaitiene, D., S. Valiukeviciene, G. Linkeviciute, R. Raisutis, E. Jasiuniene, and R.
Kazys. 2011. Principles of high-frequency ultrasonography for investigation of skin
pathology. Journal of the European Academy of Dermatology and Venereology : JEADV
25 (4): 375-382.

Jeong, Jong Seob, Jung Woo Lee, Chang Yang Lee, Shia Yen Teh, Abraham Lee, and K.
Kirk Shung. 2011. Particle manipulation in a microfluidic channel using acoustic trap.
Biomedical Microdevices 13 (4): 779-788.

Jönsson, Henrik, Cecilia Holm, Andreas Nilsson, Filip Petersson, Per Johnsson, Thomas
Laurell, Metabolism and Endocrinology Diabetes, et al. 2004. Particle separation using
ultrasound can radically reduce embolic load to brain after cardiac surgery. The Annals of
Thoracic Surgery 78 (5): 1572-1577.

Kane, Bartholomew J., Michael J. Zinner, Martin L. Yarmush, and Mehmet Toner. 2006.
Liver-specific functional studies in a microfluidic array of primary mammalian
hepatocytes. Analytical Chemistry 78 (13): 4291-4298.

Katayama, Shigeru, Chise Tateno, Toshimasa Asahara, and Katsutoshi Yoshizato. 2001.
Size-dependent in vivo growth potential of adult rat hepatocytes. The American Journal
of Pathology 158 (1): 97-105.

Kozlov, Mikhail, Mamle Quarmyne, Wei Chen, and Thomas J. McCarthy. 2003.
Adsorption of poly(vinyl alcohol) onto hydrophobic substrates. A general approach for
hydrophilizing and chemically activating surfaces. Macromolecules 36 (16): 6054-6059.

110

Kuo, I. Y., and K. Kirk Shung. 1994. High frequency ultrasonic backscatter from
erythrocyte suspension. Biomedical Engineering, IEEE Transactions on 41 (1): 29-34.

Laurell, Thomas, Filip Petersson, and Andreas Nilsson. 2007. Chip integrated strategies
for acoustic separation and manipulation of cells and particles. Chemical Society Reviews
36 (3): 492-506.

Leary, James F. 2005. Ultra high-speed sorting. Cytometry.Part A : The Journal of the
International Society for Analytical Cytology 67 (2): 76-85.

Lee, Changyang, Tae-Jin Kim, Jungwoo Lee, Jae Youn Hwang, A. Jakob, R. Lemor,
YingXiao Wang, and K. Kirk Shung. 2012a. Ultrasonic stimulation of single bovine
aortic endothelial cells at 1GHz. Ultrasonics Symposium (IUS), 2012 IEEE
International, : 21-23

Lee, Changyang, Jungwoo Lee, Hyung Ham Kim, Shia-Yen Teh, Abraham Lee, In-
Young Chung, Jae Yeong Park, and K. Kirk Shung. 2012b. Microfluidic droplet sorting
with a high frequency ultrasound beam. Lab on a Chip 12 (15): 2736-2742.

Lee, Changyang, Jong Seob Jeong, Jae Youn Hwang, Jungwoo Lee, and K. Kirk Shung.
2014. Non-contact multi-particle annular patterning and manipulation with ultrasound
microbeam. Applied Physics Letters 104 (24): 244107,244107-3.

Lee, Jungwoo, Shia-Yen Teh, Abraham Lee, Hyung Ham Kim, Changyang Lee, and K.
Kirk Shung. 2009a. Single beam acoustic trapping. Applied Physics Letters 95 (7):
073701,073701-3.

Lee, Jungwoo, Changyang Lee, and K. Kirk Shung. 2009b. Calibration of acoustic
trapping forces by fluid drag forces. Ultrasonics Symposium (IUS), 2009 IEEE
International, :410-413.

111

Lee, Jungwoo, Shia-Yen Teh, Abraham Lee, Hyung Ham Kim, Changyang Lee, and K.
Kirk Shung. 2010a. Transverse acoustic trapping using a gaussian focused ultrasound.
Ultrasound in Medicine & Biology 36 (2) (2): 350-355.

Lee, Jungwoo, Changyang Lee, and K. Kirk Shung. 2010b. Calibration of sound forces in
acoustic traps. Ultrasonics, Ferroelectrics and Frequency Control, IEEE Transactions on
57 (10): 2305-2310.

Lee, Jungwoo, Changyang Lee, Hyung Ham Kim, Anette Jakob, Robert Lemor, Shia-
Yen Teh, Abraham Lee, and K. Kirk Shung. 2011a. Targeted cell immobilization by
ultrasound microbeam. Biotechnology and Bioengineering 108 (7): 1643-1650.

Lee, Jungwoo, Jin Ho Chang, Jong Seob Jeong, Changyang Lee, Shia-Yen Teh, Abraham
Lee, and K. Kirk Shung. 2011b. Backscattering measurement from a single microdroplet.
Ultrasonics, Ferroelectrics, and Frequency Control, IEEE Transactions on 58 (4): 874-
879.

Lee, S. J., and S. Y. Lee. 2004. Micro total analysis system (micro-TAS) in
biotechnology. Applied Microbiology and Biotechnology 64 (3): 289-299.

Lizzi, Frederic L., Michael A. Laviola, and D. Jackson Coleman. 1976. Tissue signature
characterization utilizing frequency domain analysis. Ultrasonics Symposium (IUS),
1976 IEEE International, : 714-719.

Lizzi, Frederic L., Michael Greenebaum, Ernest J. Feleppa, Marek Elbaum, and D.
Jackson Coleman. 1983. Theoretical framework for spectrum analysis in ultrasonic tissue
characterization. The Journal of the Acoustical Society of America 73 (4): 1366-1373.

Lizzi, Frederic L., Michael Ostromogilsky, Ernest J. Feleppa, Mary C. Rorke, and
Mykola M. Yaremko. 1987. Relationship of ultrasonic spectral parameters to features of
tissue microstructure. Ultrasonics, Ferroelectrics and Frequency Control, IEEE
Transactions on 50 (3): 319-329.
112

Lizzi, Frederic L., Michael Astor, Tian Liu, Cheri Deng, D. Jackson Coleman, and
Ronald H. Silverman. 1997. Ultrasonic spectrum analysis for tissue assays and therapy
evaluation. International Journal of Imaging Systems and Technology 8 (1): 3-10.

Lizzi, Frederic L., Ernest J. Feleppa, S. Kaisar Alam, and Cheri X. Deng. 2003.
Ultrasonic spectrum analysis for tissue evaluation. Pattern Recognition Letters 24 (4):
637-658.

Lizzi, Frederic L., S. Kaisar Alam, Samuel Mikaelian, Paul Lee, and Ernest J. Feleppa.
2006. On the statistics of ultrasonic spectral parameters. Ultrasound in Medicine &
Biology 32 (11): 1671-1685.

Lockwood, Geoffrey R., Daniel H. Turnbull, and F. Stuart Foster. 1993. High frequency
(>20 MHz) spherically shaped ceramic transducers. Ultrasonics Symposium (IUS), 1993
IEEE International, : 495-498.

Lockwood, Geoffrey R., Daniel H. Turnbull, and F. Stuart Foster. 1994. Fabrication of
high frequency spherically shaped ceramic transducers. Ultrasonics, Ferroelectrics and
Frequency Control, IEEE Transactions on 41 (2): 231-235.

Lockwood, G. R., D. H. Turnball, D. A. Christopher, and F. S. Foster. 1996. Beyond 30
MHz [applications of high-frequency ultrasound imaging]. IEEE Engineering in
Medicine and Biology Magazine 15 (6): 60-71.

Maldovan, Martin. 2013. Sound and heat revolutions in phononics. Nature 503 (7475):
209-217.

Manneberg, Otto, S. Melker Hagsäter, Jessica Svennebring, Hans M. Hertz, Jörg P.
Kutter, Henrik Bruus, and Martin Wiklund. 2009. Spatial confinement of ultrasonic force
fields in microfluidic channels. Ultrasonics 49 (1): 112-119.
113

Manz, A., N. Graber, and H. M. Widmer. 1990. Miniaturized total chemical analysis
systems: A novel concept for chemical sensing. Sensors & Actuators: B.Chemical 1 (1):
244-248.

Mao, Xiaole, John Robert Waldeisen, and Tony Jun Huang. 2007. "Microfluidic
drifting"- implementing three-dimensional hydrodynamic focusing with a single-layer
planar microfluidic device. Lab on a Chip 7 (10): 1260-1262.

Mata, Alvaro, Aaron J. Fleischman, and Shuvo Roy. 2005. Characterization of
polydimethylsiloxane (PDMS) properties for biomedical micro/nanosystems. Biomedical
Microdevices 7 (4): 281-293.

Medical Technology Business Europe. 2014. Retrieved from
http://mtbeurope.info/news/2005/506024.htm

Mitragotri, Samir. 2005. Healing sound: The use of ultrasound in drug delivery and other
therapeutic applications. Nature Reviews Drug Discovery 4 (3): 255-260.

Murthy, Shashi K., Palaniappan Sethu, Gordana Vunjak-Novakovic, Mehmet Toner, and
Milica Radisic. 2006. Size-based microfluidic enrichment of neonatal rat cardiac cell
populations. Biomedical Microdevices 8 (3): 231-237.

Oberyszyn, Andrew S. and Fredika M. Robertson. 2001. Novel rapid method for
visualization of extent and location of aerosol contamination during high-speed sorting of
potentially biohazardous samples. Cytometry 43 (3): 217-222.

O'Donnell, M., D. Bauwens, J. W. Mimbs, and J. G. Miller. 1979. Broadband integrated
backscatter: An approach to spatially localized tissue characterization in vivo. Ultrasonics
Symposium (IUS), 1979 IEEE International, : 175-178.
114

Orfao, Alberto, and Alejandro Ruiz-Argüelles. 1996. General concepts about cell sorting
techniques. Clinical Biochemistry 29 (1) (2): 5-9.

Park, Jinhyoung, Jungwoo Lee, Sien Ting Lau, Changyang Lee, Ying Huang, Ching-Ling
Lien, and K. Kirk Shung. 2012. Acoustic radiation force impulse (ARFI) imaging of
zebrafish embryo by high-frequency coded excitation sequence. Annals of Biomedical
Engineering 40 (4): 907-915.

Partanen, Ari, Charles Mougenot, and Teuvo Vaara. 2009. Feasibility of agar-silica
phantoms in quality assurance of MRgHIFU. AIP Conference Proceedings 1113, : 296-
300.

Passmann, Christian, and Helmut Ermert. 1996. A 100-MHz ultrasound imaging system
for dermatologic and ophthalmologic diagnostics. Ultrasonics, Ferroelectrics and
Frequency Control, IEEE Transactions on 43 (4): 545-552.

Petersson, Filip, Andreas Nilsson, Cecilia Holm, Henrik Jonsson, and Thomas Laurell.
2004. Separation of lipids from blood utilizing ultrasonic standing waves in microfluidic
channels. The Analyst 129 (10): 938-943.

Recktenwald, Diether J. 1993. Introduction to flow cytometry: Principles, fluorochromes,
instrument set-up, calibration. Journal of Hematotherapy 2 (3): 387-394.

Recktenwald, Diether and Andreas Radbruch. 1998. Cell Separation Methdos and
Applications. Marcel Dekker, Inc. NY: CRC Press.

Roda, Barbara, Pierluigi Reschiglian, Andrea Zattoni, Francesco Alviano, Giacomo
Lanzoni, Roberta Costa, Arianna Di Carlo, et al. 2009. A tag-less method of sorting stem
cells from clinical specimens and separating mesenchymal from epithelial progenitor
cells. Cytometry.Part B, Clinical Cytometry 76 (4): 285-290.
115

Shung, K. Kirk. 2006. Diagnostic Ultrasound: Imaging and Blood Flow Measurements.
Boca Raton, FL: CRC Press.

Shung, K. Kirk, Jonathan Cannata, Qifa Zhou, and Jungwoo Lee. 2009. High frequency
ultrasound: A new frontier for ultrasound. 2009 Annual International Conference of the
IEEE Engineering in Medicine and Biology Society 2009, : 1953-1955.

Shung, K. Kirk. 2011. Diagnostic ultrasound: Past, present, and future. Journal of
Medical and Biological Engineering 31 (6): 371-374.

Stony Brook, Health Technology and Management. 2014. Retrieved from
http://healthtechnology.stonybrookmedicine.edu/programs/clinical

Tan, Yung-Chieh, Vittorio Cristini, and Abraham P. Lee. 2006. Monodispersed
microfluidic droplet generation by shear focusing microfluidic device. Sensors &
Actuators: B.Chemical 114 (1): 350-356.

Teh, Shia-Yen, Robert Lin, Lung-Hsin Hung, and Abraham P. Lee. 2008. Droplet
microfluidics. Lab on a Chip 8 (2): 198-220.

Tsutsui, Hideaki, and Chih-Ming Ho. 2009. Cell separation by non-inertial force fields in
microfluidic systems. Mechanics Research Communications 36 (1): 92-103.

UCLA Newsroom. 2014. New microchip technology performs 1,000 chemical reactions
at once. Retrieved from http://newsroom.ucla.edu/releases/new-microchip-technology-
performs-97160?link_page_rss=97160

116

Varadan, Vasundara V., and Vijay K. Varadan. 1979. Low-frequency expansions for
acoustic wave scattering using Waterman’s T-matrix method. The Journal of the
Acoustical Society of America 66 (2): 586-589.

Wang, Mark M, Eugene Tu, Daniel E Raymond, Joon Mo Yang, Haichuan Zang, Norbert
Hagen, Bob Dees, Elinore M Mercer, Anita H Forster, Ilona Kariv, Philippe J Marchand,
and William F Butler. 2005a. Microfluidic sorting of manmmalian cells by optical force
switching. Nature biotechnology 23 (1): 83-87.

Wang, Tza-Huei, Yahui Peng, Chunyang Zhang, Pak Kin Wong, and Chih-Ming Ho.
2005b. Single-molecule tracing on a fluidic microchip for quantitative detection of low-
abundance nucleic acids. Journal of the American Chemical Society 127 (15): 5354-5359.

Waterman, P. C. 1969. New formulation of acoustic scattering. The Journal of the
Acoustical Society of America 45 (6): 1417-1429.

Wei, Ming-Tzo, Joseph Junio, and H. Daniel Ou-Yang. 2009. Direct measurements of the
frequency-dependent dielectrophoresis force. Biomicrofluidics 3 (1): 12003, 012003-8.

White, Richard M. 1997. Introductory lecture - acoustic interactions from faraday's
crispations to MEMS. Faraday Discussions 107 : 1-13.

Whitesides, George M., Emanuele Ostuni, Shuichi Takayama, Xingyu Jiang, and Donald
E. Ingber. 2001. Soft lithography in biology and biochemistry. Annual Review of
Biomedical Engineering 3 (1): 335-373.

Wiklund, M., H. M. Hertz. 2006. Ultrasonic enhancement of bead-based bioaffinity
assays. Lab on a Chip 6 (10): 1279-1292.

117

Wolff, A., I. R. Perch-Nielsen, U. D. Larsen, P. Friis, G. Goranovic, C. R. Poulsen, J. P.
Kutter, and P. Telleman. 2003. Integrating advanced functionality in a microfabricated
high-throughput fluorescent-activated cell sorter. Lab on a Chip 3 (1): 22-27.

Xia, Younan, and George M. Whitesides. 1998. Soft lithography. Annual Review of
Materials Science 28 (1): 153-184.

Yamada, Masumi, Kyoko Kano, Yukiko Tsuda, Jun Kobayashi, Masayuki Yamato,
Minoru Seki, and Teruo Okano. 2007. Microfluidic devices for size-dependent separation
of liver cells. Biomedical Microdevices 9 (5): 637-645.

Yi, Changqing, Cheuk-Wing Li, Shenglin Ji, and Mengsu Yang. 2006. Microfluidics
technology for manipulation and analysis of biological cells. Analytica Chimica Acta 560
(1): 1-23.

Yuan, Y. W., and K. Kirk Shung. 1986. The effect of focusing on ultrasonic backscatter
measurements. Ultrasonic Imaging 8 (2): 121-130.

Asset Metadata

Creator Lee, Changyang (author)

Core Title Microfluidic cell sorting with a high frequency ultrasound beam

Contributor Electronically uploaded by the author (provenance)

School Andrew and Erna Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 05/18/2015

Defense Date 10/24/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cell sorting,high frequency ultrasound,integrated backscattering,microfluidic device,OAI-PMH Harvest,ultrasonic spectrum analysis

Format application/pdf (imt)

Language English

Advisor Shung, Kirk Koping (committee chair), McCain, Megan L. (committee member), Meiselman, Herbert J. (committee member)

Creator Email changyal@usc.edu,changyanglee@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-518163

Unique identifier UC11297652

Identifier etd-LeeChangya-3094.pdf (filename),usctheses-c3-518163 (legacy record id)

Legacy Identifier etd-LeeChangya-3094.pdf

Dmrecord 518163

Document Type Dissertation

Format application/pdf (imt)

Rights Lee, Changyang

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Abstract (if available)

Abstract Ultrasound has been used as diagnostic imaging tools in medicine for a long time due to its real-time capability and mobility as well as nonionizing radiation and safety. High frequency ultrasound (above 20 MHz) has opened up new biomedical applications thanks to its fine spatial resolution by sacrificing the depth of penetration due to increasing attenuation. Shung’s group has shown the potential of high frequency at biomedical engineering fields such as backscattering measurement for studying material properties and micro devices for micro-engineered platforms. These new approaches can be potentially powerful tools in the biology and medicine such as cell identifying and sorting for high quality sample in a miniaturized Micro Total Analysis System (µTAS). ❧ Different types of high frequency ultrasound cell sorting micro-device, which combines high frequency ultrasound transducer with microfluidic channel, was proposed and developed as simple and cost-effective bioparticle sorting device. Despite the fact that the basic concept and sorting mechanism is straightforward and intuitive, it has several advantages such as high sensitivity without pre-treatment of samples, much simpler processes of sorting, and benefit by scaling down compared to conventional cell sorter. ❧ In this research, two different size droplets in same material and different material micro-particles in similar size were clearly discriminated from spectral parameters such as spectral slope and intercept with echo amplitude and IB coefficient. In size-based case, 50 µm and 100 µm diameter lipid droplets were separated by these suggested methods in real time. Acoustic radiation force of high frequency ultrasound is shown to produce forces sufficient for sorting micro bioparticles by pushing through PDMS channel wall. In material properties-based case, it was able to sense and identify flowing polystyrene microbeads and oleic acid droplets in 45 µm diameter inside microchannel by spectral slope value in dB scale. ❧ Lastly, it was carried out a live cell sensing and identifying experiments with human leukemia and red blood cells through multi-parameters. The potential and performances of devices as a live cell sorter are verified by experimental results. For further clinical applications, the necessity of more research related to bioeffect of high frequency ultrasound beam was briefly described although there exist currently no guidelines or criteria for high frequency ultrasound. ❧ In this research, the ultimate goal is to develop a simple and yet cost-effective micro sorting device with high frequency ultrasound microbeam for individual particulates. To improve presented acoustic cell sorting method for real cell sorting applications is needed. Especially for higher frequency ultrasound, new microfluidic channel using an alternative material with lower acoustic attenuation for microfluidic channel is required due to high acoustic energy loss of PDMS for improving sensing sensitivity (>99%). Better hydrodynamic focus in new designed device could improve performance of ultrasound cell sorting devices. In this acoustic sorter, multi-parameter data analysis including statistical methods for sorting and separation should be applied with this ultrasound sensing and identifying technique in real time with high throughput. These multi-dimensional analysis and selectively visualization of data are useful for extracting physical characteristics of cells without pre-treatments of samples. It maybe enables the real time acoustic cell sorting device to identify and separate bioparticles as simpler method. Current high frequency device is still relatively bulky and has low sorting performance. The system for ultrasound sorter as alternative tools is needed to be simplified with integrated micro-chip circuit design replacing outside customized LabVIEW program, which enable an ultrasound sorter to reduce the operating time, leading high throughput of current system.