Page 1
The Pennsylvania State University
The Graduate School
College of Engineering
FROM CHIP-IN-A-LAB TO LAB-ON-A-CHIP: THE DEVELOPMENT OF A
PROTOTYPE FOR ACOUSTOFLUIDIC NANOPARTICLE SEPARATION
A Thesis in
Engineering Science and Mechanics
by
Joseph M. Rufo
© 2015 Joseph M. Rufo
Submitted in Partial Fulfillment
of the Requirements
for the degree of
Master of Science
May 2015
Page 2
ii
The thesis of Joseph M. Rufo was reviewed and approved* by the following:
Tony J. Huang
Professor of Engineering Science and Mechanics
Thesis Advisor
Jian Xu
Associate Professor of Engineering Science and Mechanics
Judith A. Todd
Professor of Engineering Science and Mechanics
P.B. Breneman Department Head
*Signatures are on file in The Graduate School.
Page 3
iii
ABSTRACT
Dynamic Light Scattering (DLS) is a commonly used analytical technique for measuring
the size distribution of particles in solution. DLS is an attractive technique because it is non-
invasive (a low power laser is used so as not to damage the samples), requires very little sample
preparation, and can extract information from small volume samples with relatively low particle
concentrations. For these reasons, DLS has become a widely used analytical technique in many
industries. For example, in the development of new biopharmaceuticals, one of the major
limitations is the large number of tests that must be performed on limited amounts of sample.
DLS allows researchers to extract size distribution information from as little as 2 μL of sample,
saving the rest of the sample for other required tests.
Although the size range of particles that can be analyzed via DLS is large (ranging from
.03 nm to 10 μm), careful attention must be paid to the concentration of particles larger than 500
nm. If the concentration of larger particles is too high, it can prevent accurate measurement of
the smaller sized particles. However, many applications produce samples containing particles
above and below 500 nm (i.e. proteins and protein aggregates). The goal of this thesis was to
develop an acoustic-based separation technique that could establish a tunable cutoff diameter and
remove all particles larger than the cutoff diameter. The concept of acoustic-based separation
was initially demonstrated in a laboratory setting with polystyrene beads. First, mixed samples
were analyzed by DLS. The samples were then passed through our acoustic filter, and the
smaller fraction was again analyzed by DLS and compared to both the initial measurements and
samples of known concentrations. Results showed that our acoustic filter drastically improved
the quality of the DLS measurements. Finally, we replaced the expensive lab equipment with
custom electronics and constructed a prototype for acoustic nanoparticle separation.
Page 4
iv
TABLE OF CONTENTS
List of Figures .............................................................................................................. v
List of Equations .......................................................................................................... vi
CHAPTER 1: INTRODUCTION ............................................................................... 1
1.1: Dynamic Light Scattering Theory ................................................................ 1
1.2: Problem Statement ........................................................................................ 9
1.3: Objective ....................................................................................................... 11
CHAPTER 2: LITERATURE REVIEW .................................................................... 13
2.1: Conventional Separation Techniques ........................................................... 13
2.2: Acoustic Particle Separation ......................................................................... 18
CHAPTER 3: METHODOLOGY .............................................................................. 24
3.1: Design Needs ................................................................................................ 24
3.2: Goals and Timeline ....................................................................................... 30
3.3: Experimental Setup ....................................................................................... 30
CHAPTER 4: RESULTS ............................................................................................ 33
4.1: Separation in the Lab .................................................................................... 33
4.2: Prototype Testing .......................................................................................... 42
CHAPTER 5: CONCLUSIONS ................................................................................. 47
5.1: Summary ....................................................................................................... 47
5.2: Future Work Suggestions ............................................................................. 47
Appendix A: Team ...................................................................................................... 49
Appendix B: IDT Fabrication ..................................................................................... 50
Appendix C: PDMS Fabrication ................................................................................. 51
Appendix D: PMMA Fabrication ............................................................................... 52
Appendix E: Supplementary Results .......................................................................... 53
References .................................................................................................................... 56
Page 5
v
LIST OF FIGURES
Figure 1.1: Polar plot arising from Mie scattering ...................................................... 3
Figure 1.2: Schematic of a typical DLS experimental setup ...................................... 5
Figure 1.3: Fluctuations in optical signals for large vs. small particles ...................... 6
Figure 1.4: Intensity plots and autocorrelation functions ........................................... 8
Figure 2.1: Schematic of SEC separation ................................................................... 13
Figure 2.2: Calibration curves for SEC....................................................................... 14
Figure 2.3: FFF underlying mechanism ...................................................................... 16
Figure 2.4: Normal vs. steric modes of FFF ............................................................... 17
Figure 2.5: Schematic of a typical SPLITT separation device ................................... 18
Figure 2.6: Acoustic tweezers for continuous particle separation .............................. 21
Figure 2.7: Tilted angle acoustic tweezers .................................................................. 22
Figure 3.1: Simulated and experimental microparticle trajectories ............................ 26
Figure 3.2: Simulated nanoparticle trajectories at different vibrational amplitudes ... 27
Figure 3.3: Simulated nanoparticle trajectories at different tilted angles ................... 28
Figure 3.4: Comparison of traditional and unidirectional IDT designs ...................... 29
Figure 3.5: Device immobilized on thermoelectric cooler ......................................... 31
Figure 4.1: Acoustic separation of 5 µm beads from 240 nm beads .......................... 33
Figure 4.2: Intensity plots obtained via DLS .............................................................. 34
Figure 4.3: Acoustic separation of 1.3 µm beads from 240 nm beads ....................... 35
Figure 4.4: Active region of separation device ........................................................... 36
Figure 4.5: Acoustic separation of 900 nm beads from 240 nm beads ....................... 37
Figure 4.6: Acoustic separation of 700 nm beads from 240 nm beads ....................... 38
Figure 4.7: Determining separation efficiency ........................................................... 39
Figure 4.8: Acoustic separation with PMMA chip ..................................................... 40
Figure 4.9: Intensity plots obtained via DLS for PMMA chip ................................... 41
Figure 4.10: Photos of prototype ................................................................................ 42
Figure 4.11: Internal layout of prototype .................................................................... 43
Figure 4.12: Intensity plots obtained via DLS for prototype ...................................... 44
Figure 4.13: USB camera images of particle separation............................................. 45
Figure 4.14: Comparison of experimental setups ....................................................... 46
Page 6
vi
LIST OF EQUATIONS
Equation 1.1: Rayleigh scattering ............................................................................... 2
Equation 1.2: Fick’s second law of diffusion ............................................................. 4
Equation 1.3: Stokes-Einstein relation ........................................................................ 4
Equation 1.4: Second-order autocorrelation function ................................................. 7
Equation 1.5: Siegert equation .................................................................................... 7
Equation 1.6: First-order autocorrelation function ..................................................... 7
Equation 1.7: Decay rate ............................................................................................. 7
Equation 1.8: Scattering vector ................................................................................... 7
Equation 1.9: Polydispersity index ............................................................................. 9
Equation 1.10: Number averaged molecular weight................................................... 9
Equation 1.11: Weight averaged molecular weight .................................................... 9
Equation 2.1: Calibration curve for size exclusion chromatography .......................... 14
Equation 2.2: Acoustic radiation force ....................................................................... 20
Equation 2.3: Acoustic contrast factor ........................................................................ 20
Equation 2.4: Stokes drag force .................................................................................. 20
Page 7
1
CHAPTER 1: INTRODUCTION
1.1: Dynamic Light Scattering Theory
Dynamic light scattering (DLS) is an analysis technique that allows the size distribution
of particles in solution to be determined by analyzing the manner in which the particles scatter
light. DLS is based on the combination of three fundamental theories from physics: Rayleigh
scattering, Mie scattering, and Brownian motion. Rayleigh scattering and Mie scattering both
describe how particles of different sizes scatter light, while Brownian motion describes how
particles of different sizes move in solution. In a typical DLS configuration, particles in solution
are illuminated with a monochromatic, coherent light source. The scattered light is then collected
and analyzed, and particle size information can be obtained. DLS is a label-free, reagent-free,
measurement technique that can be performed on very small sample volumes (as little as 2 μL of
sample) over a large range of particle sizes (ranging from .03 nm to 10 μm). In addition, the
measurement process is typically automated, and there are no difficult sample preparation
requirements (samples are simply loaded into a cuvette). Finally, DLS is capable of analyzing a
variety of solvents, enabling it to be employed in applications ranging from food
characterization1 to process validation in the manufacturing of biopharmaceuticals.
2 With the
aforementioned advantages, DLS has become a widely used analytical technique in both research
and industrial settings. This chapter will describe basic DLS theory and how particle size
information is extracted from scattered light signals.
In the late 19th
century, British physicist Lord Rayleigh published four seminal papers
that described the scattering of light by small particles.3-6
Rayleigh noted that when the diameter
(d) of the particle is much smaller than the wavelength (λ) of incoming light (d <
λ), the
incident photon can interact with the charges within the particle. Specifically, the incoming
Page 8
2
photon interacts with the electron cloud to create an oscillating dipole; as the dipole shifts,
energy is radiated equally in all directions.3 This isotropic radiation of energy is commonly
referred to as Rayleigh scattering. The intensity, I, of the light with an incoming intensity, Io,
scattered by any one particle with a refractive index, n, when measured at an angle, θ, and
distance, R, from the particle is given as:
(
)
(
)
(
)
(1.1)
The strong dependence of the scattering intensity on the wavelength of incoming light ( )
indicates that shorter wavelengths will be scattered much more strongly than longer wavelengths.
This exponential dependence explains why the sky appears blue, even though incoming solar
radiation spans the entire visible range.6 As the light passes through the atmosphere, atmospheric
molecules scatter blue light much stronger than any other visible wavelength. In DLS
instruments, the intensity of the scattered signal is analyzed to reveal information about the
diameter of the particles in solution. For a DLS instrument using a red HeNe laser with a
wavelength of 633 nm, Rayleigh scattering can provide size information for particles smaller
than around 60 nm.7 For larger sized particles, the manner in which light is scattered is slightly
different. Rather than isotropic scattering, the scattering intensity becomes distorted in the
forward direction. This form of scattering is called Mie scattering, named after German physicist
Gustav Mie. Mie developed his theory in an attempt to describe why the color of gold colloids
changes as the size of the gold nanoparticles is increases.8 Mie showed that the light extinction,
or the energy lost to the combined effects of absorption and scattering, is highly dependent on
the wavelength of incoming light and the size of the particle. Smaller sized gold particles show
extinction maxima in the blue and green wavelengths, causing them to appear red when
illuminated with white light; as the particle size is increased, the extinction maxima shift towards
Page 9
3
red wavelengths, and the particles begin to appear bluer in color. Mie theory was developed as a
by solving Maxwell’s equations. The solution involves an infinite series of partial wave
equations, and ultimately, it can be used to describe the complex scattering and absorption of
light by a spherical particle with respect to distance and angle.9 While the full derivation of Mie
scattering is beyond the scope of this thesis, Figure 1.1 shows a typical polar plot that is used to
determine the size of a particle based on how it scatters light. On the horizontal axis, a
logarithmic plot of the intensity is shown. The polar axes indicate the angle at which the light is
measured when the particle is being illuminated from 180°.
Figure 1.1: Polar plot of scattered light intensity for different sized particles.7
The polar plot shows that light is scattered much more strongly in the forward direction,
regardless of particle size. When comparing the light scattered by a 1000 nm particle (red) to that
of a 500 nm particle (green), the 1000 nm particle scatters light nearly 100 times stronger in the
Page 10
4
forward direction; however, for certain angles (i.e. 30°, 60°, 90°), the scattered light intensity is
practically identical for both particles. This can be problematic in DLS instruments because it
can make determining particle sizes very problematic when measuring from certain angles. As a
result, most DLS instruments utilize dual angle measurements to increase sensitivity and more
easily detect the presence of particles of different sizes.7 Unlike in Rayleigh scattering, careful
attention must be paid to the presence of maxima and minima in the scattering intensity plots
when using Mie theory to determine particle sizes.
The final factor that must be taken into account when determining the size distribution of
particles in solution is that the particles are constantly moving. This constant, random motion,
termed Brownian motion, is due to the constant collisions between the particles and molecules in
the solution.10
Brownian motion causes the concentration of particles in any given area to
fluctuate over time. Equation 1.2, Fick’s second law of diffusion, describes how concentration
changes with time (in one-dimension).
(1.2)
Where Φ(x,t) represents the concentration of a given substance as a function of location, x, and
time, t, and D is the diffusion coefficient. D is typically measured in units of m2/s, while
concentration is generally measured in units of mol/m3. For spherical particles in solution, the
diffusion coefficient can be determined by the Stokes-Einstein relation, shown in Equation 1.3.
(1.3)
Where kB is Boltzmann’s constant, T is the absolute temperature, η is the viscosity of the fluid,
and r is the radius of the particle in the fluid. Because the diffusion constant is inversely
proportional to the radius of the particle, smaller particles will undergo more rapid Brownian
Page 11
5
motion than larger particles in fluids with identical temperatures and viscosities. At higher
temperatures, particles undergo more rapid Brownian motion. Thus DLS measurement devices
must have some mechanism of reporting the temperature of the fluid. In addition, it is essential
that the temperature and viscosity of the fluid are kept constant throughout a sample
measurement.
While Rayleigh scattering, Mie scattering, and Brownian motion all relate to the size of
the particles in solution, it may not be immediately evident how DLS measurement tools
determine particle size distributions based on scattered light signals. In order to carry out these
measurements, sufficient optics and signal processing equipment are required. Figure 1.2 is a
schematic of a typical setup for DLS measurements.11
A laser is used to provide a highly
monochromatic, highly coherent light source. It is focused through a lens onto the sample.
Typically, a plastic or glass cuvette is used to hold the sample. At a fixed angle, θ, a lens and
photon detector collect the scattered light signals. This information is converted from analog to
digital and analyzed by autocorrelation software, which is able to determine particle size
distributions.
Figure 1.2: Schematic of a typical DLS experimental setup.11
Page 12
6
As previously mentioned, the intensity of the incoming optical signals is related to the
size of the particles (through Rayleigh and Mie scattering). The fluctuations in the optical signals
over time are due to the movement of particles in the solution, which is also related to the size of
the particles (through the Stokes-Einstein relation). Figure 1.3 compares the optical signals
collected from large and small particle samples.12
Because smaller particles are moving more
rapidly in the solution than larger particles, the optical signals collected for smaller particles
fluctuate more than those of the larger particles.
Figure 1.3: Fluctuations in optical signals for large vs. small particles.12
An autocorrelation function is used to quantify how slowly or rapidly an optical signal is
fluctuating. An autocorrelation function compares the intensity, I, of an optical signal at an initial
time, t, with the intensity at a later time, t + τ, and determines the degree to which they are
related. The normalized, second-order autocorrelation function, g2(q;τ), is shown below:
Page 13
7
( ) ⟨ ( ) ( )⟩
⟨ ( )⟩ (1.4)
Where q is the wave vector and the brackets, < >, represent the expected value operator.
Equation 1.4 is a referred to as a second-order equation because it involves intensity
measurements, which are the squares of electric fields. Over sufficiently short time periods, τ,
I(t+τ) and I(t) remain closely correlated, as the particles have not had sufficient time to move,
and the correlation coefficient is 1. However, as τ is increased, I(t+τ) and I(t) will no longer be
correlated due to the particle’s movement, and the autocorrelation coefficient diminishes to 0. By
determining how quickly the autocorrelation function decays, one can determine the size of the
particles in solution.13
The autocorrelation function will decay much faster for smaller particles
because they are moving more rapidly in solution. Because Equation 1.4 requires infinite time
limits of evaluation; an approximation, known as the Siegert equation, must be implemented.13
The Siegert equation is shown below:
( ) [ ( )] (1.5)
Where β is a correction factor that is based on the experimental setup and g1(q;τ) is the first-order
autocorrelation function. If treated as a monodiperse sample, the first-order autocorrelation
function follows the simple exponential decay relationship as follows:
( ) ( ) (1.6)
With:
(1.7)
And:
(1.8)
Page 14
8
Where Γ is the decay rate, q is the scattering vector, n is the refractive index of the fluid, and λ is
the wavelength of the laser. The decay rate of the autocorrelation function can therefore be used
to obtain the diffusion coefficient, which can be used in conjunction with Equation 1.3 to solve
for the particle radius. The diffusion coefficient is obtained by fitting the autocorrelation function
with a suitable algorithm, mainly cumulants analysis or distribution analysis.13
Figure 1.4 shows
intensity plots and corresponding autocorrelation plots for a sample with large particles and a
sample with small particles.
Figure 1.4: Intensity plots and autocorrelation functions for large and small particle samples.7
As expected, the correlation function for the large particles takes much longer to decay
than the correlation coefficient for the smaller particles. Subsequent cumulant analysis would
reveal that the diffusion coefficient in the case of the large particles is smaller than the diffusion
coefficient for the small particles. This diffusion coefficient is then converted to mean particle
size via the Stokes-Einstein relation. Many companies have utilized DLS theory to develop
Page 15
9
commercial devices and software that can be used to automatically acquire particle size
distribution information from solutions.
1.2 Problem Statement
There are several cases in which DLS may not be suitable for measuring samples. The
first is when the sample is highly polydisperse (i.e. there is a wide size distribution in the
sample). One of the assumptions made in DLS theory is that the first order autocorrelation
function obeys a simple exponential decay; however, this assumption only hold true for
monodiperse samples. Because most samples are polydisperse, the autocorrelation function
usually becomes a sum of the exponential decays for each species in the population.13
However,
once the polysdispersity crosses a certain threshold, the fitting algorithms will fail to extract
useful information about the size distribution of particles in the sample. The polydisperisty index
(PDI), a measurement that quantifies the heterogeneity of the sample, is given by the following
equation:
(1.9)
Where Mn represents the number average molecular weight and Mw represents the weight
average molecular weight. These two quantities can be obtained from:
∑
∑ (1.10)
And:
∑
∑ (1.11)
Page 16
10
Where Ni is the number of molecules and Mi is the molecular mass. For samples containing latex
standard spheres, PDI values are typically < 0.05.7 Distribution algorithms work well until the
PDI reaches 0.7; above this value, samples may not be suited for DLS measurements.7
The second factor that may deter the accuracy DLS measurements is particle aggregation.
Particle aggregation will increase the PDI, but even low levels of aggregation can negatively
impact a DLS measurement. Rayleigh scattering and Mie scattering are based on assumptions
that the particles are sphere-like. When particles aggregate, they are no longer sphere-like and
will scatter light in a different manner. These particle aggregates will be treated as larger
particles by the DLS software, resulting in inaccurate measurements. As a result, careful
attention must be paid to the dispersant (liquid medium) to make sure that it will not promote
particle aggregation.
Careful attention must also be paid to the concentration of particles. If the concentration
of particles is too high, particles will begin to interact with one another. Thus the movement of
the particles is no longer dictated by Brownian motion and the Stokes-Einstein equation. As a
result, the DLS software will incorrectly calculate the sizes of particles in the solution. Another
issue that arises at high particle concentrations is multiple scattering. DLS theory assumes that
each photon collected by detectors has only been scattered by one particle. When the
concentration is too high, photons begin to be scattered multiple times before reaching the
detectors. This will also lead to inaccurate results, as the initial conditions for Rayleigh scattering
and Mie scattering are no longer valid.
The final issue that can affect the accuracy of DLS measurements is particle
sedimentation. All particles will sediment to some degree; however, for DLS measurements, it is
essential that the rate of sedimentation is much slower than the rate of diffusion. If particles
Page 17
11
sediment to the bottom of the cuvette, they will no longer scatter light and will appear invisible
to the DLS equipment. If a sample containing sedimenting particles is measured multiple times
sequentially, the count rate obtained by the DLS software will decrease with each measurement.
Particle sedimentation becomes an issue with particles larger than about 500 nm. Above 1 µm,
the particles begin to sediment very rapidly, and careful attention must be made in preparing the
samples for DLS measurements. For example, if the sample has been sitting for some time, it is
essential that sample be vortexed immediately prior to insertion into the DLS machine.
1.3: Objective
To address the issues associated with the inability of DLS instruments to accurately
measure samples containing large particles, we proposed to develop a continuous-flow,
microfluidic, acoustic filtration system. The proposed system could establish an adjustable cutoff
diameter, and all particles larger than the cutoff diameter would be removed from the sample.
The samples would be analyzed pre and post filtration and compared to samples of known
concentrations to assess the ability of the acoustic filter to improve the accuracy of DLS
measurements. Not only did we want to conduct proof of concept demonstrations, but we also
sought to develop a commercially viable alpha prototype. Often times in the microfluidics
research community, researchers claim to develop “lab-on-a-chip” devices. That is, all of the
functions that are normally carried out with expensive laboratory equipment and highly trained
personnel are integrated onto an automated, microfluidic chip. While there have been many
successful examples of this process (i.e. microfluidic chips run by cell phones for the diagnosis
of syphilis14
and HIV15
), one of the major bottlenecks in microfluidics research is in the
transition from lab demonstrations to real-world commercial products. This is due to the fact that
many microfluidics laboratories utilize expensive, bulky equipment that cannot easily be
Page 18
12
replaced, rendering the technology commercially unviable. This approach is commonly referred
to as the “chip-in-a-lab” approach.16
Even though the technological feats are being carried out on
an inexpensive, portable, microfluidic chip, all of the peripheral equipment prevents the chip’s
use in practical, point-of-care (POC) applications. In this thesis, a strong emphasis was placed on
commercial development, cost reduction, and reliability engineering to ensure that the
technology developed could one day find real-world use and potentially be developed for other
applications.
Page 19
13
CHAPTER 2: LITERATURE REVIEW
2.1: Conventional Separation Techniques
One of the most commonly used techniques for separating nanoparticles is size exclusion
chromatography (SEC).17-19
SEC is a high resolution separation technique that uses porous beads
to separate particles. A separation column is first filled with beads that have well-defined pore
sizes. The particle containing solvent is then injected into the separation column. As the solvent
elutes through the column, smaller particles are able to enter the beads, while larger particles
flow past the beads. The smaller particles take longer to elute through the column, as they must
travel through a larger effective elution volume. Figure 2.1 shows a schematic of how separation
via SEC is achieved.
Figure 2.1: Schematic of SEC separation.18
Page 20
14
In SEC, both the solvent and beads must be tailored for the sample of interest. It is
important that the solvent does not dissolve the sample or damage the packing beads. In addition,
calibration curves for each column must be obtained prior to conducting measurements.
Molecular weight markers with known molecular weights (typically polystyrene) are injected
through the column to see how molecular weight (M) varies with elution volume (Ve).19
Equation 2.1 shows the equation used fit the calibration curve for SEC.
( ) (2.1)
Where b and c are constants related to the experimental setup. Once the constants are obtained
using known molecular weight samples, the molecular weights of unknown particles can be
found. Figure 2.2 shows the calibration curves obtained from packing materials with different
sized pores. The material with the larger pores (green) has a much higher slope than the material
with the smaller pores (blue).
Figure 2.2: Calibration curves for SEC.19
Particles larger than the exclusion limit will elute too quickly through the packing
material and cannot be quantified by the SEC instrument. Particles smaller than the penetration
Page 21
15
limit take too long to elute through the packing material and also cannot be quantified by the
SEC instrument. Figure 2.2 shows the importance of selecting the proper pore size for the
sample. Typically, beads with small pore sizes are used to separate smaller particles and beads
with large pore sizes are used to separate larger particles. When a sample that contains a broad
range of particle sizes is used, a combination of beads can be employed. By collecting different
volumes of eluted samples in different collection containers, the samples can be sorted by their
size. Common applications of SEC separation include the purification of biopharmaceuticals,20
separation of carbon nanotubes,21
and the removal of large contaminants from cell culture
fluids.22
Although a powerful separation technique for certain applications, the upper size limit for
SEC separation is around 60 nm.17
This eliminates SEC separation as a technology that can be
used to improve the resolution of DLS, as particles larger than 500 nm need to be separated. In
addition to the size limitation, optimizing the conditions for SEC (i.e. column dimensions, flow
rate, packing density, pore size, bead diameter, etc.) is a very tedious process that requires highly
trained personnel. The development of a separation technique that can easily be adjusted to
separate many different types of samples would fill an immediate need for many separation
applications.
Another commonly used separation technique is field flow fractionation (FFF).23
Unlike
SEC, which is limited to an upper size limit of around 60 nm, the upper size limit for FFF is
around 700 nm (although separation of particles as large as 10 µm has been demonstrated).24
The
underlying mechanism to FFF is similar to SEC in that particles of different sizes will migrate
through a narrow channel at different rates based on their sizes. However, FFF only requires a
Page 22
16
single aqueous phase, whereas SEC requires two phases: the solvent and the porous beads.
Figure 2.3 shows a schematic of how FFF is achieved.
Figure 2.3: FFF underlying mechanism.25
In the simplest embodiment of FFF, a solution containing particles is injected through a
narrow channel with a separation field acting perpendicular to the channel. The separation field
can be generated by various mechanisms including: cross flow, gravitational, centrifugal, thermal
gradient, electrical, and magnetic.24
The perpendicular field pushes the particles to different
heights in the channel based on the size of the particles. Within the channel, there exists a
parabolic flow profile. Thus particles located closer to the center of the channel will travel much
faster than particles located near the channel wall. These particles will elute through the channel
at a rate characteristic to their size, enabling size based separation.
FFF is complicated because the operating mode changes for particles found in different
size classes.26
For example, for particles under 500 nm, the external force will push the particles
towards the channel bottom. Upon reaching the channel bottom, particles will undergo Brownian
motion and diffuse towards the middle of the channel. As discussed earlier, smaller particles will
diffuse faster, allowing them to reach a higher final height and elute through the channel faster.
Page 23
17
Thus for particles under 500 nm, smaller particles elute through the channel faster than larger
particles. This operational mode is referred to as “normal” mode.26
However, for particles
between 500 nm and 10 µm, larger particles will elute through the channel faster than smaller
particles. Under these conditions, when the external field acts on the particles, they will also be
pushed towards the channel bottom. But at this size range, the Brownian motion of the particles
begins to become negligible. Rather than diffuse, the particles form a thin layer near the channel
bottom. The larger particles in the layer protrude into the channel slightly more than the smaller
particles, allowing them to reach a higher final height and elute through the channel faster. This
mode of operation is referred to as “steric” or “reversed” mode.26
Figure 2.4 shows the two
operating modes of FFF.
Figure 2.4: Normal vs. steric modes of FFF.26
In a slightly modified configuration, FFF can be used to continuously separate particles
of different sizes into different outlets (rather than collecting different volumes at different
times). This configuration greatly simplifies the separation process and is referred to as split flow
lateral transport thin (SPLITT) separation.26
In this configuration, two inlet fluids are required:
the sample and an exchange buffer. A splitter is placed downstream to divide the channel into
two outlets. The separation field pushes the particles to different heights within the channel, and
Page 24
18
the particles can be directly separated based on their final vertical position in the channel. Figure
2.5 shows a typical configuration of a SPLITT separation device.
Figure 2.5: Schematic of a typical SPLITT separation device.26
Many types of external fields have been applied to achieve separation, including
gravitational,27
magnetic,28
and dielectrophoretic forces.29
However, for gravitational separation,
only particles larger than about 20 µm can be separated at timescales comparable to other FFF
techniques.26
For magnetic and dielectrophoretic forces, only certain fluids with particular
magnetic or electric properties can be used. Furthermore, the separation is influenced by factors
other than size, complicating their use in purely size-based separation applications. In this regard,
the development of a separation technique that can be employed across a wide range of particle
sizes and is independent of the magnetic and electrical properties of the particles and solution
would be highly valuable.
2.2: Acoustic Particle Separation
In 2009, in an effort to create a more broadly applicable separation technique, the Penn
State Acoustofluidics Laboratory developed a standing surface acoustic wave (SSAW)
separation technology.30
This SSAW-based technology was based on the group’s previous work
on a particle manipulation technology called “acoustic tweezers”.31
Acoustic tweezers was the
Page 25
19
first technology to use SSAWs to manipulate and pattern cells and microparticles, but it has since
been applied to continuous flow applications.32-34
The term “surface waves” refers to the manner
in which the acoustic waves are generated. Surface acoustic waves are generated via the
piezoelectric effect, a phenomenon observed in certain crystals in which an applied electric field
is converted into a mechanical strain; likewise, in these crystals, an applied mechanical force is
converted into an electric charge. Surface waves have advantages over bulk waves in that they
can reach much higher frequencies and require less power to generate.35
In the fabrication of the
acoustic tweezers, interdigital transducers (IDTs) are evaporated onto a lithium niobate
(LiNbO3) piezoelectric substrate. A signal generator is used to apply a periodic voltage to the
IDTs. As a result, mechanical waves are generated on the surface of the LiNbO3 substrate.
Because the IDTs are placed on either side of the microfluidic channel, waves propagate in
opposite directions underneath the microfluidic channel. When waves propagate in opposite
directions, they constructively interfere to create a standing wave field. Figure 2.6 shows the
typical configuration of an acoustic tweezers device.
Figure 2.5: Schematic of a typical acoustic tweezers device.35
In this standing wave field, there is a periodic distribution of pressure nodes (minimum
pressure) and pressure antinodes (maximum pressure).36
This standing wave field is coupled into
Page 26
20
the liquid in the microchannel, resulting in periodic pressure fluctuations in the liquid. Pressure
fluctuations in the liquid result in acoustic radiation forces that act on particles to position them
at the pressure nodes or pressure antinodes, depending on the properties of the particle. The
acoustic radiation force (Fr) can be expressed as:36
(
) ( ) ( ) (2.2)
( )
(2.3)
In Equations 2.2 and 2.3, p0, Vp, λ, k, x, ρp, ρf, βp, and βf are acoustic pressure, volume of the
particle, wavelength, wave vector, distance from a pressure node, density of the particle, density
of the fluid, compressibility of the particle, and compressibility of the fluid, respectively.
Equation 2.3 describes the acoustic contrast factor, φ, which determines whether the particle
moves to pressure nodes or pressure antinodes in the SSAW field: the particle will move towards
pressure nodes if φ is positive and pressure antinodes if φ is negative. As the particle moves in
the fluid, it is opposed by Stokes drag force (Fd), which is given by the following equation:
( ) (2.4)
Where η, Rp, up, and uf are the viscosity of the fluid, radius of the particle, velocity of the
particle, and velocity of the fluid, respectively. Other forces acting on the particle include gravity
and buoyant forces, but they are almost balanced as they are generally similar in magnitude but
opposite in direction. Figure 2.6 shows how the acoustic tweezers can be employed to
continuously separate particles of different sizes.30
Fig 2.6 (a) shows a schematic of the device.
Particles are introduced from outlets on either side of the microchannel. Fig 2.6 (b) shows the
magnitude and direction of the different forces acting on particles of different sizes in the
channel. As the particles enter the acoustic region, the acoustic force drives the larger particles to
Page 27
21
the center of the channel. However, because the magnitude of the acoustic force is proportional
to volume of the particle, the smaller particles do not experience enough force to push them to
the center outlet, and they remain in the side outlets.
Figure 2.6: (a) Schematic of acoustic tweezers used for continuous particle separation. (b)
Comparison of forces acting on the particles at different locations within the acoustic field.30
The device in Figure 2.6 separated 13,000 PS beads of two different sizes (particle 1:
0.87 μm polystyrene (PS) beads, particle 2: 4.17 μm PS beads) using only 30 mW of power. The
low power consumption of the SSAW-based sorting renders it a highly biocompatible sorting
technique. In addition, a separation efficiency of 80% was achieved.
In the device shown in Figure 2.6, the maximum separation distance (the distance
between the particles being sorted) is limited to a quarter of the acoustic wavelength. This leads
to relatively low separation efficiencies (all of the target particles are not separated from the
undesired population) and low sensitivity (it is difficult to separate particles that are similar in
Page 28
22
size, density, and compressibility). To overcome these limitations, in 2014, a new configuration
was developed in which the channel was placed at angle to the IDTs. This approach, termed the
tilted-angle standing surface acoustic wave (taSSAW) approach, allows particles to be pushed
across multiple pressure nodes, thereby achieving separation distances of up to 10 times the
acoustic wavelength.34
The taSSAW-based separation device was used to successfully separate
MCF-7 cancer cells (~20 µm in diameter) from normal leukocytes (white blood cells, ~12 µm in
diameter). Figure 2.7 (a) shows a photo of the taSSAW-based separation device, while Figures
2.7 (b) and (c) show the device being used to separate 10 µm PS beads from 2 µm PS beads.34
Figure 2.7: (a) Photographic image of the ta-SSAW separation device. (b) The active region of
the sorting device and (c) the outlet, showing 10 µm beads being collected from the top outlet,
while the 2 µm beads remain in the bottom outlet.34
The taSSAW device achieved a separation efficiency of 99% for the separation of 10 µm
beads from 2 µm beads. In addition, the device separated 9.9 µm beads from 7.3 µm beads with
Page 29
23
97% efficiency. The device was also able to separate 15 µm PS beads from HL-60 cells (a
human promyelocytic leukemia cell line with a diameter ~15 µm), demonstrating the potential
for compressibility-based separation. When separating MCF-7 cancer cells from white blood
cells, 71% of the MCF-7 cancer cells were recovered with a purity of 84%. This approach shows
tremendous potential for cancer diagnostics, where it is important to separate a small number of
leukemia cells from normal white blood cells.37
The taSSAW design was proven to be highly
tunable. By changing the operating parameters (i.e. flow rate and input power) as well as the
design parameters (i.e. IDT wavelength, IDT angle, channel height, and channel width), the
device could be tailored to different separation application. The taSSAW device also possesses
other advantages, such as its low cost, compact nature, ease of use, and low requirements for
external equipment. Overall, with slight improvements to the throughput (~2 µL/min for cell
separation), the taSSAW device could become a powerful tool for point-of-care diagnostic
applications.
Page 30
24
CHAPTER 3: METHODOLOGY
3.1: Design Needs
There are many difficulties involved in transitioning from separating particles at the
micrometer scale to separating particles at the nanometer scale. From Equation 2.2, it can be seen
that the acoustic radiation force is directly proportional to the volume of the particle (Fr Vp). In
previous experiments, we had separated particles that were approximately 10 µm in diameter.
The objective of this project was to remove all particles larger than 500 nm in diameter. For an
acoustic separation device operating at a constant input power, frequency, and flow rate, the
acoustic radiation force experienced by a 500 nm particle is approximately 8,000 times less than
the acoustic radiation force experienced by a 10 µm particle. The magnitude of the acoustic
radiation force determines the lateral displacement of the particles. Therefore, in order to remove
all particles larger than 500 nm, we needed to generate sufficient acoustic radiation force to
move a 500 nm particle ~100 µm (the approximate distance required to move the particle to a
separate outlet) in the lateral direction. Many strategies were investigated to achieve this goal,
including: decreasing the flow rate (increases the amount of time particles spend in the acoustic
field and therefore increases the particles’ lateral migration), increasing the acoustic pressure, p0
(Fr p02), and decreasing the wavelength, λ (Fr λ
-1).
The most obvious way to increase the lateral migration of the particles is to decrease the
flow rate. The particles will spend a longer time in the acoustic field and attain a greater final
separation distance. However, decreasing the flow rate also decreases the throughput of the
device (the maximum volume of sample that can be processed per unit time). A minimum of ~
20 µL of sample is usually required for DLS measurements. As a result, we determined that the
minimum acceptable flow rate for our device should be ~ 1 µL/min in order to attain reasonable
Page 31
25
sample processing times. In previous experiments using taSSAW devices for the manipulation of
micrometer sized particles, flow rates of ~2-10 µL/min were used. We were therefore limited in
the reductions we could make to the flow rates.
In order to increase the acoustic pressure, one can increase the power applied to the IDTs.
The acoustic pressure (p0) is proportional to the square root of the applied power, P (p0 P1/2
).
However, as the applied power is increased, heat is generated via resistive heating. This results in
heating of the sample fluid. When the fluid reaches ~60 °C, dissolved gasses begin to be released
(gas solubility decreases at increased temperatures).38
This leads to the formation of bubbles in
the microchannel that become trapped against the sidewall. These bubbles will continue to
expand, disrupting the laminar flow in the microchannel. Once the laminar flow is disrupted,
particles are no longer focused and will randomly exit both outlets of the device. Another issue
that arises at high temperatures is irreversible damage to the substrate. The LiNbO3 substrate is
prone to cracking from thermal stresses, so precautions must be made to control the temperature
of the substrate.39
Due to the heat related issues, it was determined that a thermoelectric cooling
system should be implemented. This would allow the device to operate at high input powers and
increase the acoustic pressure without running into the problems that are associated with
increased temperatures.
The final step taken to increase the lateral migration of nanometer sized particles was to
decrease the wavelength of the IDTs. Previous experiments used a wavelength of ~300 µm;
however, with access to the PSU nanofabrication facility, we were able to decrease the
wavelength to 100 µm. After obtaining the minimum flow rate and minimum wavelength, a
series of simulations were conducted to determine the optimal angle for nanoparticle separation.
A custom MATLAB code was developed to determine the effect that varying the tilted angle of
Page 32
26
the IDTs, length of the IDTs, acoustic wavelength, input power, flow rate, and properties of the
sample fluid has on particle trajectories. In previous experiments, our simulations were able to
accurately predict particle trajectories at various operating conditions. Figure 3.1 shows the
simulation results (shaded regions) overlaid with experimental results (black dots) from previous
micrometer sized separation experiments.34
In Figure 3.1 (a), we see the trajectories of PS beads
of different sizes under a constant applied power, while Figure 3.1 (b) shows the trajectories of a
constant sized bead (15 µm) under different applied powers.
Figure 3.1: Comparison of simulated particle trajectories (shaded regions) and experimentally
observed particle trajectories (black dots) for (a) particles of different sizes and (b) different
input powers.34
The simulation results are generally in excellent agreement with the experimental results.
Similar simulations were used to obtain particle trajectories for nanometer sized particles. Figure
Page 33
27
3.2 shows the trajectories of 500 nm (blue) and 1000 nm (red) PS beads. The solid black lines
represent the channel walls, while the dashed red lines represent the acoustic field. In all three
cases, the tilted angle of the IDTs is 15°. The vibration amplitude is increased from 10 nm in Fig.
3.2 (a) to 15 nm in Fig. 3.2 (b) to 20 nm in Fig. 3.3 (c).
Figure 3.2: Simulation results for the trajectories of 500 nm (blue) and 1000 nm (red) PS beads
a tilted IDT angle of 15° vibration amplitudes of at (a) 10 nm, (b) 15 nm, and (c) 20 nm.
It can be seen from the simulations that the 1000 nm beads were predicted to achieve a
lateral migration distance of ~400 µm and be pushed to the top channel wall at a vibration
amplitude of only 20 nm. Under the same conditions, the 500 nm particles attained a separation
distance of ~50 µm. These simulations indicated that the taSSAW platform could in fact be used
for nanoparticle separation. Next, we conducted simulations to see the effect of the tilted angle.
Figure 3.3 shows the trajectories of 500 nm (blue) and 1000 nm (red) PS beads under a constant
vibration amplitude of 20 nm, but different tilted angles. Fig. 3.3 (a) shows the particle
trajectories under the optimal tilted angle of 15°, while Fig. 3.3 (b) shows the particle trajectories
under a tilted angle of 30°. A series of simulations were conducted at increments of 5°, ranging
from 5° to 85°, to determine the optimal tilted angle.
Page 34
28
Figure 3.3: Simulation results for the trajectories of 500 nm (blue) and 1000 nm (red) PS beads
a vibration amplitude of 20 nm and a tilted IDT angle of (a) 15° and (b) 30°.
While the simulations showed the potential for nanoparticle separation, we wanted to
increase the lateral displacement of the 500 nm particles. In order to achieve this, we looked into
ways of increasing the efficiency of the IDTs. Under our traditional design, surface waves
propagate in both directions from the IDTs. However, the microchannel is located in the center
of the two pairs of IDTs. Therefore, half of the acoustic energy never reaches the microchannel
when using a traditional IDT design. One way to increase the efficiency is to add reflectors at the
back end of each IDT to reflect acoustic energy back towards the microchannel. Another
approach is to design unidirectional IDTs, which only generate acoustic waves in the forward
direction.40-42
Unidirectional IDTs incorporate shorted floating electrodes with less acoustic
impedance than free space into their design, leading to higher vibrational amplitudes in the
forward direction.40
In addition, open floating electrodes with more acoustic impedance than free
space are incorporated to reflect waves towards the microchannel.40
Figure 3.4 (a) shows the
traditional IDT design,43
which generates waves that propagate in both directions, while Fig. 3.4
Page 35
29
(b) shows the unidirectional IDT design.40
In both cases, coherent AC signals are applied to the
active fingers; however, no signals are applied to the shorted strips and open strips in the
unidirectional design.
Figure 3.4: Comparison of (a) traditional and (b) unidirectional IDT designs.40,43
The unidirectional, reflector (not shown), and traditional designs of IDTs were all
fabricated with the same number of electrode pairs (n=50) and identical wavelengths (λ=120 µm)
and their performance was analyzed via a 2 port network analyzer. In all cases, the IDTs were
fabricated by depositing a metal double layer (Cr/Au, 50 Å/500Å) with an e-beam evaporator
(Semicore Corp). More information on the IDT fabrication process can be found in Appendix B.
The resonance frequency for all three designs was similar (λtraditional=32.96 MHz, λreflector=32.96
MHz, λunidirectional=32.51 MHz); however, the unidirectional IDT showed the highest return loss
(17.78 dB), indicating it was the most efficient design. Therefore, in our nanoparticle separation
experiments we elected to use unidirectional IDTs.
Page 36
30
3.2: Goals and Timeline
At the beginning of the project, a timeline was created, along with measures of success.
The project was divided into two phases. The first phase (6 months) was dedicated to
demonstrating the feasibility of nanoparticle separation, and the second phase (1.5 months) was
dedicated to prototype development. The goals for the first month of phase one were to
demonstrate the removal of 5 µm particles from smaller particles with 90% efficiency, and test
60-700 nm PS beads to check the loss of samples. In months 2-3, we would work to demonstrate
the removal of 5 µm particles from smaller particles with 95% efficiency, and demonstrate the
removal of 2 µm particles from smaller particles with 90% efficiency. In months 4-6, we sought
to demonstrate the removal of 1000 nm particles from smaller particles with 90% efficiency, and
demonstrate the removal of 500 nm particles from smaller particles with 90% efficiency. The
goals of phase 2 (months 7–8.5) were to use PMMA chips to replace PDMS chips, standardize
and streamline the fluidic design, develop a customized signal generator, amplifier, and other
electronics to drive SAW chips, and develop an Alpha prototype (an integrated system including
electronics in a box format). In all of our experiments, a DLS instrument (Zetasizer Nano,
Malvern Instruments) was used to quantify separation efficiency and sample loss. Videos of
experiments were obtained using a CCD camera (CoolSNAP HQ2, Photometrics) and analyzed
with ImageJ 1.46 software.
3.3: Experimental Setup
For initial experiments aimed at demonstrating the feasibility of nanoparticle separation,
polydimethylsiloxane (PDMS) was used as the material for the microchannel. PDMS is a good
material for rapid prototyping because devices can be made from standard soft lithography
techniques.44
This is an inexpensive process that enables the design of the device to be altered
Page 37
31
fairly easily. A detailed procedure for PDMS device fabrication can be found in Appendix C. To
carry out particle separation experiments, the device was first immobilized on the stage of an
inverted microscope (Nikon TE2000U). In cases where the heat could potentially damage the
substrate, the device was immobilized on the surface of a thermoelectric cooler (TE Technology
Inc. CP-031). The inlet tubing was connected to a computer controlled syringe pump
(neMESYS, Cetoni GmbH) to allow for precise fluid control. The waste outlet tubing was
connected to another computer controlled syringe pump that was operated in withdrawal mode.
This ensures a stable, laminar flow inside of the microchannel and fixes the flow rate from the
outlet that is being collected. Figure 3.5 shows the device immobilized on the stage of the
thermoelectric cooler.
Figure 3.5: Device immobilized on thermoelectric cooler.
After the fluidic components have been connected, an RF signal generator (Agilent Tech,
E4422B) that is amplified with a power amplifier (Amplifier Research, 100A250A) is connected
Page 38
32
to the IDTs. For the experiments using a cooling plate, an LED (MicroscopeNet, A92144L)
needed to be placed mounted underneath of the device to create a custom, reflection mode
microscope. Figure 3.6 (a) shows the signal generator and amplifier, Fig. 3.6 (b) shows the
cooling plate on the stage of the microscope and the syringe pumps, and Fig 3.6 (c) shows the
LED and connected leads of the IDTs.
Figure 3.6: Experimental setup including (a) signal generator and amplifier, (b) cooling plate
and pumps, and (c) LED.
The complex experimental setup is an example of the “chip-in-a-lab” approach referred
to in Chapter 1. In order for our acoustic separation technology to be commercially viable, the
development of a more compact experimental setup is required. Hence, the second phase of this
project was dedicated to prototype development.
Page 39
33
CHAPTER 4: RESULTS
4.1: Separation in the Lab
Our first goal was to demonstrate the removal of 5 µm particles from smaller particles
with 90% efficiency. A solution containing 200 µL of 5 µm PS bead solution (1% solids by
volume), 40 µL of 240 nm PS bead solution (1% solids by volume), and 1 mL deionized (DI)
water was injected into the device at a flow rate of 5 µL/min. DI water was used as sheath fluid
and was injected into the device at a flow rate of 15 µL/min. Figure 4.1 shows images taken at
the outlet of the device during the particle separation process. When the acoustic power was
turned off (Fig. 4.1 (a)), all of the beads exited the bottom outlet. When the acoustic power was
turned on (Fig. 4.1 (b)), the 5 µm particles were pushed to the top outlet, while the 240 nm
particles remained in the bottom outlet (not visible). In both cases (acoustic on and acoustic off),
the bottom outlet was collected for 5 minutes and subsequently analyzed via DLS.
Figure 4.1: Acoustic separation of 5 µm beads from 240 nm beads when (a) the acoustic power
is off and (b) the acoustic power in turned on.
Page 40
34
Although the particles are difficult to see, the DLS results indicate that nearly all of the 5
µm beads were removed from the sample. When the acoustic power was off, the Zetasizer
struggled to accurately measure the size distribution of the particles. This is due to the difficulties
involved in measuring particles larger than 500 nm (discussed in Chapter 1). There were three
separate peaks observed on the intensity plot, and each time the sample was analyzed, the
number and location of the peaks changed. The software reported an error message stating that
the “sample contains large particle/aggregates/dust” and the “sample is very polydisperse and
may not be suitable to DLS measurements”. For the sample collected when the acoustic power
was on, the Zetasizer produced a single peak near 240 nm that matched very well to the results
obtained from a control sample with the same concentration of 240 nm PS beads. Figure 4.2
compares the intensity plots from the acoustic off, acoustic on, and 240 nm control samples.
Figure 4.2: Intensity plots for the (a) unfiltered, (b) acoustic filtered, and (c) control samples.
Page 41
35
After successfully demonstrating the removal of 5 μm beads from 240 nm beads, we
moved to separating 1.3 μm beads from 240 nm beads. A solution containing 140 µL of 1.3 µm
PS bead solution (1% solids by volume), 40 µL of 240 nm PS bead solution (1% solids by
volume), and 1 mL deionized (DI) water was injected into the device at a flow rate of 3 µL/min.
DI water was used as sheath fluid and was injected into the device at a flow rate of 9 µL/min.
Figure 4.3 shows the separation results. When the acoustic power was turned off (Fig. 4.3 (a)),
all of the beads exited the bottom outlet. When the acoustic power was turned on (Fig. 4.3 (b)),
the 5 µm particles were pushed to the top outlet, while the 240 nm particles remained in the
bottom outlet (not visible). In both cases (acoustic on and acoustic off), the bottom outlet was
collected for 5 minutes and subsequently analyzed via DLS (Fig 4.3 (c)).
Figure 4.3: Acoustic separation of 1.3 µm beads from 240 nm beads when (a) the acoustic
power is off and (b) the acoustic power in turned on. (c) DLS results from bottom outlet.
Page 42
36
From the DLS data, it can be seen that nearly all of the 1.3 μm particles were removed
from the original (red) sample, while the acoustic filtered sample (green) matches reasonably
well with the 240 nm control (blue). Figure 4.4 shows the active region of the device when the
acoustic power is applied.
Figure 4.4: Active region of device.
The particles enter the active region focused in the center of the channel. As the 1.3 μm
particles enter the acoustic field, they are moved to pressure nodes (which cross the channel at a
15° angle). As a result, the majority of the 1.3 μm particles are pushed to the top of the channel.
The spacing, d, between the pressure nodes corresponds to half of the acoustic wavelength (λ =
100 μm, d = 50 μm). For the 240 nm particles, the acoustic radiation force is insufficient to push
the particles from their original streamline, and they remain focused in the center of the channel.
A similar experimental process was followed to demonstrate the separation of 900 nm beads
from 240 nm beads, 700 nm beads from 240 nm beads, and 500 nm beads from 240 nm beads.
The exact details for each case can be found in Appendix E. In each case, the bottom outlet was
collected and analyzed by DLS. Figure 4.5 shows the separation results for the separation of 900
nm beads from 240 nm beads. When the acoustic power was turned off (Fig. 4.5 (a)), all of the
beads exited the bottom outlet. When the acoustic power was turned on (Fig. 4.5 (b)), the 900 nm
particles were pushed to the top outlet, while the 240 nm particles remained in the bottom outlet
(not visible).
Page 43
37
Figure 4.5: Acoustic separation of 900 nm beads from 240 nm beads when (a) the acoustic
power is off and (b) the acoustic power in turned on. (c) DLS results from bottom outlet.
Again, from the DLS data, it can be seen that nearly all of the 900 nm particles were
removed from the original (red) sample, while the acoustic filtered sample (green) matches
reasonably well with the 240 nm control (blue). Figure 4.6 shows the separation results for the
separation of 700 nm beads from 240 nm beads. When the acoustic power was turned off (Fig.
4.6 (a)), all of the beads exited the bottom outlet. When the acoustic power was turned on (Fig.
4.6 (b)), the 700 nm particles were pushed to the top outlet, while the 240 nm particles remained
in the bottom outlet (not visible).
Page 44
38
Figure 4.6: Acoustic separation of 700 nm beads from 240 nm beads when (a) the acoustic
power is off and (b) the acoustic power in turned on. (c) DLS results from bottom outlet.
From the DLS data, it can be seen that nearly all of the 700 nm particles were removed
from the original (red) sample, while the acoustic filtered sample (green) matches reasonably
well with the 240 nm control (blue). The separation of 500 nm beads from 240 nm beads was
also performed (Appendix E); however, due to the difficulties in measuring bimodal samples, a
high concentration of 500 nm beads was required, which made it difficult to quantify the
separation efficiency. To measure the device’s separation efficiency, a sample containing only
600 nm beads was injected into the device. The bottom outlet was collected for 5 minutes while
the acoustic power was turned off and for 5 minutes when the acoustic power was turned on.
Page 45
39
DLS was then used to obtain count rates (rather than intensity plots) for each sample. When the
acoustic power was off, an average count rate (measured in kilo counts per second (kcps)) of
62,464 kcps was obtained; when the acoustic power was on, this count rate decreased to 8,942
kcps, indicating that approximately 86% of the 600 nm particles were removed. Figure 4.7 shows
a histogram of the count rates for the acoustic off (red) and acoustic on (green) samples.
Figure 4.7: Comparison of count rates for a filtered vs. unfiltered sample containing 600 nm PS
beads.
After demonstrating the ability of our device to successfully separate nanoparticles, we
began to develop a prototype. The construction of the prototype involved replacing the
expensive, bulky signal generator and amplifier used in the lab with a more compact,
inexpensive, custom signal generator and amplifier. In addition, the PDMS chips were replaced
with PMMA chips, a much harder, more robust material. Details of the PMMA chip fabrication
Page 46
40
can be found in Appendix D. To evaluate the functionality of the PMMA chip, we first
performed particle separation experiments using the lab equipment. A sample containing a
mixture of 5 μm, 1.3 μm and 240 nm PS beads was injected into the PMMA chip at a flow rate
of 2 μL/min. DI water was used as the sheath fluid and injected into the chip at 10 μL/min.
Figure 4.8 shows the separation results. When the SAW was turned off (Fig. 4.8 (a)), all of the
beads exited the bottom outlet. When the SAW was turned on (Fig. 4.8 (b)), the 5 μm and 1.3 μm
particles were pushed to the top outlet, while the 240 nm particles remained in the bottom outlet
(not visible).
Figure 4.8: PMMA chip for acoustic separation of 5 μm and 1.3 μm beads from 240 nm beads
when (a) the SAW is off and (b) the SAW is turned on.
In both cases, the bottom outlet was collected for 10 minutes and analyzed by DLS.
Figure 4.9 shows the intensity plots obtained by DLS. For the SAW OFF sample (Fig. 4.9 (a)),
the presence of large particles prevented accurate analysis by DLS. The Zetasizer produced many
error messages pertaining to the quality of the sample, and produced different results each time
the sample was analyzed. For the SAW ON sample (Fig. 4.9 (b)), the Zetasizer was able to
Page 47
41
accurately measure the particles, as evidenced by the favorable comparison to a 240 nm control
sample (Fig. 4.9 (c)).
Figure 4.9: Intensity plots for the (a) unfiltered, (b) acoustic filtered, and (c) control samples.
Although we were able to successfully separate particles using the PMMA chip, further
testing revealed that the PMMA chip was susceptible to leakage problems after multiple uses.
We are still in the process of optimizing the PMMA bonding process; however, due this issue,
PDMS chips were used to test the functionality of the prototype.
Page 48
42
4.2: Prototype testing
The main purpose of the prototype was to simplify the experimental setup. A custom
signal generator and amplifier were designed and constructed to replace their bench top
counterparts. An off the shelf power supply was purchased to power the signal generator and
amplifier. A commercially available cooling system was also purchased. All of the components
were placed in a custom built, metal chassis, and finally, a camera was mounted on top of the
cooling plate to allow users to observe the particle separation process and troubleshoot if
necessary. Figure 4.10 shows completed prototype.
Figure 4.10: Images of the (a) completed prototype, (b) cooling stage, and (c) outlet collection.
Page 49
43
Fig. 4.10 (a) shows the device when the lid to the platform is closed. The USB camera is
connected to a laptop or computer with image processing software capabilities. In Fig. 4.10 (b)
the lid to the platform is opened. The actual separation chip is placed on top of the cooling plate,
and connected to external pumps. Fig 4.10 (c) shows the waste and sample collection outlets.
Disposable microcentrifuges tubes, which can easily be inserted and removed from the fixed
support structure, are used to collect the sample and waste. Figure 4.11 shows a picture of the
prototype with the lid removed to expose the layout of the signal generator, amplifier, cooling
plate, and power supplies.
Figure 4.11: Internal layout of prototype.
We first tested the prototype’s ability to repeat the separation of 1.3 µm beads from 240
nm beads. The same parameters (sample concentration, flow rate, collection time) that were used
Page 50
44
in the laboratory experiments were used when testing the prototype. Figure 4.12 shows the DLS
results.
Figure 4.12: Intensity plots for the (a) unfiltered, (b) acoustic filtered, and (c) control samples.
In the unfiltered sample (Fig. 4.12 (a)), there are two distinct peaks; however, in both the
acoustic filtered sample (Fig. 4.12 (b)) and control sample (Fig. 4.12 (c)), there is only a single
peak. This indicated that the prototype was able to remove the majority of the 1.3 µm particles.
Although the quality of the images obtained using the USB camera is not very high (Figure
Page 51
45
4.13), they show the particles being pushed to the top outlet. When no power is applied (Fig.
4.13 (a)), the particles in the active region (the area between the gold L-shaped markings) travel
along the bottom wall. After the power is applied (Fig. 4.13 (b)), particles in the active region are
pushed towards the top wall.
Figure 4.13: Images taken from USB camera depicting the particle separation process on the
prototype, (a) acoustic power off and (b) acoustic power on.
We also separated 900 nm particles from 240 nm particles using the prototype. The DLS
results can be found in Appendix E. As in previous experiments, the peak observed in the
unfiltered sample was removed in the acoustic filtered sample. Overall, the prototype was proven
to successfully separate particles of different sizes, and it greatly simplified the experimental
setup. Further validation studies are in progress. Figure 4.14 compares the initial experimental
setup to the setup enabled by the prototype. The prototype setup is much more compact, allowing
for true “lab-on-a-chip” applications.
Page 52
46
Figure 4.14: (a) Laboratory vs. (b) prototype experimental setups for acoustic nanoparticle
separation.
Page 53
47
CHAPTER 5: CONCLUSION
5.1: Summary
This project demonstrated the feasibility of acoustic nanoparticle separation. Experiments
were first performed in a laboratory setting to demonstrate the ability of our microfluidic chips.
We successfully removed 5 µm, 1.3 µm, 900 nm, 700 nm, and 500 nm PS beads from 240 nm PS
beads. In each case, only one peak was observed on the intensity plots of the acoustically filtered
samples, indicating that the majority of the larger particles were removed. In addition, the
filtered samples compared favorably with control samples that contained only 240 nm particles.
These results showed that an acoustic filter could be used in conjunction with a DLS instrument
to improve the accuracy of DLS measurements. However, due to the requirements for expensive
equipment and complex experimental setups, the commercial viability of our technology was
limited. In addition, the material used to fabricate our microfluidic chips, PDMS, is a soft,
deformable polymer that is not suitable for many commercial applications. In order to overcome
these limitations, we developed a prototype that replaced the expensive laboratory equipment
with compact, custom electronics. We fabricated microfluidic chips from PMMA, a much more
durable, thermoplastic, and successfully demonstrated their ability to separate 5 µm, 1.3 µm, and
900 nm PS beads from 240 nm PS beads. Further testing of the prototype is still underway;
however, the prototype holds promise for future nanoparticle separation applications.
5.2: Future Recommendations
In the next phase of development, we should seek to incorporate micropumps into the
prototype. Currently, syringe pumps are the only external component required for the prototype
to operate. If pumps can be integrated into the next prototype, it will be able to operate as a
Page 54
48
standalone unit. This would enable portable applications and further simplify the experimental
setup. To further reduce the size of the device, the power supply for the cooling plate and the
power supply for the signal generator and amplifier should be integrated into a single power
supply. Aside from improvements to the prototype itself, I think it is important to demonstrate
the potential of acoustic nanoparticle separation in more diverse applications. For example, our
technology could potentially be used to separate components of the blood. This could be
particularly useful in the isolation of exosomes, 30-100 nm vesicles secreted by cells.45
Current
approaches to isolating exosomes require ultracentrifugation, which is time consuming, suffers
from low recovery rates (5-25% of initial exosome population)46
, and alters the morphology of
the exosomes.47
Our technique would eliminate the need for expensive centrifuges and we would
expect it to achieve higher recovery rates because the entire isolation process would take place
on a single chip. This differs from ultracentrifugation where multiple washing and resuspension
steps result in a loss of exosomes. Finally, we would expect our acoustic separation technology
to better preserve the integrity of the exosomes because the power intensity and frequency used
in our acoustic chip (0.1─3 W/cm2, 6─40 MHz)
30-34 are in a similar range as those used in
ultrasonic imaging (~0.5 W/cm2, 2─20 MHz)
48-51, which has proven to be extremely an
biocompatible monitoring technique.
Page 55
49
Appendix A
Team
Throughout this research opportunity, I had the pleasure of interacting with most
members of the Penn State Acoustofluidics lab. Shown below is a picture of the group, along
with Dr. Tony Huang (center of the front row).
I would especially like to thank Yuchao Chen and Feng Guo. I worked with Yuchao for
most of the SAW separation work. Yuchao was instrumental in arranging project meetings for
brainstorming and had many helpful suggestions that were needed for the success of the project.
Feng provided excellent instruction on how to fabricate microfluidic chips and provided
excellent guidance for understanding the underlying physics behind SAW separation.
Page 56
50
Appendix B
IDT Fabrication
The following IDT fabrication process, adapted from Shi et al., was followed:
“To achieve SAW with a high coupling coefficient, a Y + 128 X-propagation lithium
niobate (LiNbO3) wafer (500 mm thick) was used as the substrate for IDT deposition.
The LiNbO3 wafer was patterned with photoresist (SPR3012, MicroChem, Newton,
MA), a double metal layer (Cr/Au, 50 A˚ /800 A˚) was deposited (e-beam evaporator,
Semicore Corp) on the wafer, and a lift-off process was used to remove the photoresist
and the metal attached, thus obtaining the IDTs for SAW generation.”30
Figure A1: IDT Fabrication process.30
Page 57
51
Appendix C
PDMS Fabrication
The polydimethylsiloxane (PDMS) microfluidic channel was fabricated using a mold-
replica procedure. The master mold was obtained by performing deep reactive ion etching
(DRIE, Adixen, Hingham, MA) on a pre-patterned silicon wafer with spun-on photoresist. Once
the master mold was fabricated, the microfluidic chamber was made by simply filling the mold
with PDMS gel, allowing it to cure, and then removing it from the mold. In order to make the
removal of PDMS easier, the master mold was first coated with 1H, 1H, 2H, 2H-
perfluorooctyltrichlorosilane (Sigma Aldrich) to reduce the surface energy. In order to form the
PDMS gel, Sylgard184TM Silicone Elastomer base was mixed with curing agent (Dow Corning,
Midland, MI) according to a 10:1 weight ratio. The mixture was cured at 70°C for 17 min to
remove all bubbles. After the PDMS was degassed, it was cut and peeled from the mold. Inlets
and outlets were made using a 0.75 mm punch (Harris uni-core). Finally, the PDMS was bonded
onto a lithium niobate substrate to form a sealed microchannel, and plastic tubing was connected
to the inlets and outlets.
Page 58
52
Appendix D
PMMA Fabrication
The PMMA was fabricated using a laser cutter (Epilog Laser). The channel designs were
first drawn in SolidWorks, and then uploaded to the laser cutter. A blank sheet of PMMA was
loaded into the laser cutter, and multiple devices were simultaneously etched into the sheet.
Figure A2: PMMA sheet with multiple etched devices.
Each device was then cut from the sheet using a hand held acrylic cutter (Plaskolite).
Two different bonding methods were used to bond the PMMA channels to the LiNbO3 substrate:
UV epoxy (Thorlabs, NOA 61) and a double sided tape (3M, 524CW).
Page 59
53
Appendix E
Supplementary Results
For the separation of 900 nm PS beads from 240 nm PS beads, a solution containing 200
µL of 900 nm PS bead solution (1% solids by volume), 40 µL of 240 nm PS bead solution (1%
solids by volume), and 1 mL deionized (DI) water was injected into the device at a flow rate of 5
µL/min. DI water was used as sheath fluid and was injected into the device at a flow rate of 10
µL/min. The samples were collected for 10 minutes each (acoustic on, acoustic off) and diluted
with DI water to 1 mL.
For the separation of 700 nm PS beads from 240 nm PS beads, a solution containing 120
µL of 700 nm PS bead solution (1% solids by volume), 40 µL of 240 nm PS bead solution (1%
solids by volume), and 1 mL deionized (DI) water was injected into the device at a flow rate of 5
µL/min. DI water was used as sheath fluid and was injected into the device at a flow rate of 10
µL/min. The samples were collected for 10 minutes each (acoustic on, acoustic off) and diluted
with DI water to 1 mL.
For the separation of 500 nm PS beads from 240 nm PS beads, a solution containing 240
µL of 900 nm PS bead solution (1% solids by volume), 40 µL of 240 nm PS bead solution (1%
solids by volume), and 1 mL deionized (DI) water was injected into the device at a flow rate of 5
µL/min. DI water was used as sheath fluid and was injected into the device at a flow rate of 10
µL/min. The samples were collected for 10 minutes each (acoustic on, acoustic off) and diluted
with DI water to 1 mL.
The DLS data from the separation of 500 nm PS beads from 240 nm PS beads is shown
below in Figure A3.
Page 60
54
Figure A3: Acoustic separation of 500 nm beads from 240 nm beads when (a) the acoustic
power is off and (b) the acoustic power in turned on. (c) DLS results from bottom outlet.
Due to the difficulties in measuring bimodal samples and the high concentration of 500
nm beads used, the second peak appears close to 1000 nm rather than 500 nm in the unfiltered
sample. However, only one peak is observed in the acoustic filtered sample.
The DLS data from the separation of 900 nm PS beads from 240 nm PS beads using the
prototype is shown below in Figure A4.
Page 61
55
Figure A4: Intensity plots for the (a) unfiltered, (b) acoustic filtered, and (c) control samples.
When the acoustic power was off, the Zetasizer struggled to accurately measure the size
distribution of the particles. The location of the peaks changed each time the sample was
analyzed. For the sample collected when the acoustic power was on, the Zetasizer produced a
single peak near 240 nm that matched very well to the results obtained from a control sample
with the same concentration of 240 nm PS beads.
Page 62
56
REFERENCES
1. Xu, R. Particle characterization: light scattering methods. 13, (Springer Science &
Business Media, 2001).
2. He, F. et al. High-throughput dynamic light scattering method for measuring viscosity of
concentrated protein solutions. Anal. Biochem. 399, 141–143 (2010).
3. Strutt, J. W. XV. On the light from the sky, its polarization and colour. London,
Edinburgh, Dublin Philos. Mag. J. Sci. 41, 107–120 (1871).
4. Strutt, J. W. LVIII. On the scattering of light by small particles. London, Edinburgh,
Dublin Philos. Mag. J. Sci. 41, 447–454 (1871).
5. Rayleigh, Lord. X. On the electromagnetic theory of light. London, Edinburgh, Dublin
Philos. Mag. J. Sci. 12, 81–101 (1881).
6. Rayleigh, Lord. XXXIV. On the transmission of light through an atmosphere containing
small particles in suspension, and on the origin of the blue of the sky. London,
Edinburgh, Dublin Philos. Mag. J. Sci. 47, 375–384 (1899).
7. Shaw, Ryan, Dr., and Malvern Instruments. "Dynamic Light Scattering Training."
Atascientific.com. ATA Scientific. Web. 2 Feb. 2015.
8. Mie, G. Beiträge zur Optik trüber Medien, speziell kolloidaler Metallösungen. Ann. Phys.
330, 377–445 (1908).
9. Stratton, J. A. Electromagnetic theory. 33, (John Wiley & Sons, 2007).
10. Einstein, A. Investigations on the Theory of the Brownian Movement. (Courier
Corporation, 1956).
11. "Dynamic Light Scattering: Measuring the Particle Size Distribution." Dynamic Light
Scattering Technology. LS Instruments, 2015. Web. Feb. 2015.
Page 63
57
12. "Dynamic Light Scattering." Wikipedia. Wikimedia Foundation, Web. 04 Feb. 2015.
<http://en.wikipedia.org/wiki/Dynamic_light_scattering>.
13. Russo, Paul. "A Practical Minicourse in Dynamic Light Scattering." Lsu.edu. Louisiana
State University, Feb. 2012. Web. Jan. 2015.
<http://macro.lsu.edu/HowTo/DLS_Minicourse/DLS_Minicourse.pdf>.
14. Chin, C. D. et al. Microfluidics-based diagnostics of infectious diseases in the developing
world. Nat. Med. 17, 1015–1019 (2011).
15. Cheng, X. et al. A microfluidic device for practical label-free CD4+ T cell counting of
HIV-infected subjects. Lab Chip 7, 170–178 (2007).
16. Stone, H. A., Stroock, A. D. & Ajdari, A. Engineering flows in small devices:
microfluidics toward a lab-on-a-chip. Annu. Rev. Fluid Mech. 36, 381–411 (2004).
17. Mori, S. & Barth, H. G. Size exclusion chromatography. (Springer Science & Business
Media, 1999).
18. Barron, Andrew R., and Yen-Hao Lin. "Size Exclusion Chromatography and Its
Application in Polymer Science." OpenStax CNX., 13 Apr. 2010. Web. Jan. 2015.
<http://cnx.org/contents/dbc08b96-777f-4912-9caa-
04a31b974794@2/Size_Exclusion_Chromatography_>.
19. "Size Exclusion Chromatography." Shimadzu Corporation. Web. Jan. 2015.
<http://www.shimadzu.com/an/hplc/support/lib/lctalk/55/55intro.html>.
20. Kükrer, B. et al. Mass spectrometric analysis of intact human monoclonal antibody
aggregates fractionated by size-exclusion chromatography. Pharm. Res. 27, 2197–2204
(2010).
Page 64
58
21. Duesberg, G. S., Burghard, M., Muster, J. & Philipp, G. Separation of carbon nanotubes
by size exclusion chromatography. Chem. Commun. 435–436 (1998).
22. Kalbfuss, B., Wolff, M., Morenweiser, R. & Reichl, U. Purification of cell culture‐
derived human influenza A virus by size‐exclusion and anion‐exchange chromatography.
Biotechnol. Bioeng. 96, 932–944 (2007).
23. Giddings, J. C. Field-flow fractionation. Sep. Sci. Technol. 19, 831–847 (1984).
24. Caldwell, K. D. Field-flow fractionation. Anal. Chem. 60, 959A–971A (1988).
25. "Field Flow Fractionation." Wikipedia. Wikimedia Foundation, Web. Feb. 2015.
<http://en.wikipedia.org/wiki/Field_flow_fractionation>.
26. Lenshof, A. & Laurell, T. Continuous separation of cells and particles in microfluidic
systems. Chem. Soc. Rev. 39, 1203–1217 (2010).
27. Huh, D. et al. Gravity-driven microfluidic particle sorting device with hydrodynamic
separation amplification. Anal. Chem. 79, 1369–1376 (2007).
28. Inglis, D. W., Riehn, R., Sturm, J. C. & Austin, R. H. Microfluidic high gradient
magnetic cell separation. J. Appl. Phys. 99, 08K101 (2006).
29. Li, Y., Dalton, C., Crabtree, H. J., Nilsson, G. & Kaler, K. V. I. S. Continuous
dielectrophoretic cell separation microfluidic device. Lab Chip 7, 239–248 (2007).
30. Shi, J., Huang, H., Stratton, Z., Huang, Y. & Huang, T. J. Continuous particle separation
in a microfluidic channel via standing surface acoustic waves (SSAW). Lab Chip 9,
3354–3359 (2009).
31. Shi, J. et al. Acoustic tweezers: patterning cells and microparticles using standing surface
acoustic waves (SSAW). Lab Chip 9, 2890–2895 (2009).
Page 65
59
32. Shi, J. et al. Three-dimensional continuous particle focusing in a microfluidic channel via
standing surface acoustic waves (SSAW). Lab Chip 11, 2319–2324 (2011).
33. Li, S. et al. Standing surface acoustic wave (SSAW)-based cell washing. Lab Chip 15,
331–338 (2015).
34. Ding, X. et al. Cell separation using tilted-angle standing surface acoustic waves. Proc.
Natl. Acad. Sci. 111, 12992–12997 (2014).
35. Ding, X. et al. Surface acoustic wave microfluidics. Lab Chip 13, 3626–3649 (2013).
36. Gorkov, L. P. Forces acting on a small particle in an acoustic field within an ideal fluid.
Dokl. Akad. Nauk Sssr 140, 88 (1961).
37. Binet, J.-L. et al. Perspectives on the use of new diagnostic tools in the treatment of
chronic lymphocytic leukemia. Blood 107, 859–861 (2006).
38. Karlsson, J. M. et al. Active liquid degassing in microfluidic systems. Lab Chip 13,
4366–4373 (2013).
39. Vere, A. W. Mechanical twinning and crack nucleation in lithium niobate. J. Mater. Sci.
3, 617–621 (1968).
40. Takeuchi, M. & Yamanouchi, K. Coupled model analysis of SAW floating electrode type
unidirectional transducers. Ultrason. Ferroelectr. Freq. Control. IEEE Trans. 40, 648–
658 (1993).
41. Pradeep, D. J., Krishnan, N. R. & Nemade, H. B. Simulation of unidirectional interdigital
transducers in SAW Devices using COMSOL Multiphysics. in COMSOL Multiphysics
user Conference (Hannover, Germany) (2008).
42. Luo, J. K., Fu, Y. Q. & Milne, W. I. Acoustic Wave Based Microfluidics and Lab-on-a-
Chip. (2013).
Page 66
60
43. "Monolithic Crystal Filters." Epson. Web. Jan. 2015.
<http://www5.epsondevice.com/en/quartz/library/aboutfilter/filter_e.html>.
44. Whitesides, G. M., Ostuni, E., Takayama, S., Jiang, X. & Ingber, D. E. Soft lithography
in biology and biochemistry. Annu. Rev. Biomed. Eng. 3, 335–373 (2001).
45. Sokolova, V. et al. Characterisation of exosomes derived from human cells by
nanoparticle tracking analysis and scanning electron microscopy. Colloids Surfaces B
Biointerfaces 87, 146–150 (2011).
46. Lamparski, H. G. et al. Production and characterization of clinical grade exosomes
derived from dendritic cells. J. Immunol. Methods 270, 211–226 (2002).
47. van der Pol, E., Böing, A. N., Harrison, P., Sturk, A. & Nieuwland, R. Classification,
functions, and clinical relevance of extracellular vesicles. Pharmacol. Rev. 64, 676–705
(2012).
48. "Ultrasound." Wikipedia. Wikimedia Foundation, Web. Feb. 2015.
<http://en.wikipedia.org/wiki/Ultrasound>.
49. Martinez, Daniel A. "Therapeutic Ultrasound: A Review of the
Literature."ChiroACCESS. N.p., 01 Oct. 2001. Web. Feb. 2015.
<http://www.chiroaccess.com/Articles/Therapeutic-Ultrasound-A-Review-of-the-
Literature.aspx?id=0000210>.
50. Nelson, T. R., Fowlkes, J. B., Abramowicz, J. S. & Church, C. C. Ultrasound biosafety
considerations for the practicing sonographer and sonologist. J. Ultrasound Med. 28, 139
(2009).
51. Ter Haar, G. Ultrasound bioeffects and safety. Proc. Inst. Mech. Eng. Part H J. Eng.
Med. 224, 363–373 (2010).
Page 67
Joseph Rufo [email protected]
Campus Address Permanent Address
300 Waupelani Drive, Apt. 1045 701 Haviland Drive
State College, PA 16801 Bryn Mawr, PA 19010
(610) 420-0630 (610) 420-0630
OVERVIEW
Enthusiastic, disciplined student with a passion for continued learning. My strong
communication skills and technical writing ability have made me a valuable asset in a
multidisciplinary engineering lab. I am seeking a position where I can leverage these skills to add
value to a dynamic organization.
EDUCATION
The Pennsylvania State University, University Park, PA (May 2015)
M.S. Engineering Science and Mechanics
GPA 3.85/4.0
The Pennsylvania State University, University Park, PA (May 2013)
B.S. Engineering Science and Mechanics, Nanotechnology minor
GPA 3.43/4.0
PROFESSIONAL SKILLS
Experience with standard soft lithography fabrication techniques, signal processing, optical
system design, and computational fluid dynamics. Working knowledge of C++ and MATLAB to
create mathematical models. Skilled in using SolidWorks and AutoCAD to create 3D models.
EXPERIENCE
Penn State Acoustofluidics Lab, University Park, PA (August 2012 – current)
Graduate Student Researcher
Led an industry sponsored research project to design and develop a prototype for acoustic
nanoparticle separation.
Created a project timeline, established target goals and measures of success, assigned
various tasks to members of the groups, presented progress in weekly internal meetings,
and wrote bi-weekly status updates for our industry sponsor.
Collaborated with team to develop a low-cost, portable, flow cytometer.
Received second place for the P.B. Breneman Best Design in Research Award.
Asymmetric Therapeutics, State College, PA (January 2013 – May 2013)
Design Engineer / Senior Capstone Project
Worked with engineering team and medical professionals at the University of Pittsburgh
Medical Center, to design and build an affordable, non-paramagnetic chair for
magnetoencephalography (MEG) testing.
Performed patent searches, conducted feasibility studies, and carried out extensive
research on material properties.
Comcast, Downingtown, PA (June 2012-August 2012)
P&E Engineering Metrics and Analysis Intern
Page 68
Tested equipment from a variety of vendors to assess product performance across a wide
range of physical and environmental conditions.
Created reports that aided in qualifying products for use in the Comcast network. PUBLICATIONS
Po-Hsun Huang, Nitesh Nama, Zhangming Mao, Peng Li, Joseph Rufo, Yuchao Chen,
Yuliang Xie, Cheng-Hsin Wei, Lin Wang, and Tony Jun Huang, A reliable, programmable
acoustofluidic pump powered by oscillating sharp-edge structures, Lab on a Chip, Vol.
14, pp. 4319-4323, 2014.
Chenglong Zhao, Yuliang Xie, Zhangming Mao, Yanhui Zhao, Joseph Rufo, Shikuan Yang,
Feng Guo, John D. Mai, and Tony Jun Huang, Theory and experiment on particle
trapping and manipulation via optothermally generated bubbles, Lab on a Chip, Vol. 14,
pp.384-391, 2014.
Ahmad Ahsan Nawaz, Xiangjun Zhang, Xiaole Mao, Joseph Rufo, Sz-Chin Steven Lin, Feng
Guo, Yanhui Zhao, Michael Lapsley, Peng Li, J. Philip McCoy, Stewart J. Levine, and Tony
Jun Huang, Sub-micrometer-precision, three-dimensional (3D) hydrodynamic focusing
via “microfluidic drifting”, Lab on a Chip, Vol. 14, pp. 415-423, 2014.
Po-Hsun Huang, Yuliang Xie, Daniel Ahmed, Joseph Rufo, Nitesh Nama, Yuchao Chen,
Chung Yu Chan, and Tony Jun Huang, An acoustofluidic micromixer based on oscillating
sidewall sharp-edges, Lab on a Chip, Vol. 13, pp. 3847-3852, 2013.
Yanhui Zhao, Danqi Chen, Hongjun Yue, Jarrod B. French, Joseph Rufo, Stephen J.
Benkovic and Tony Jun Huang, Lab-on-a-chip technologies for single-molecule
studies, Lab on a Chip, Vol. 13, pp. 2183-2198, 2013.
Yuliang Xie, Chenglong Zhao, Yanhui Zhao, Sixing Li, Joseph Rufo, Shikuan Yang, Feng
Guo, and Tony Jun Huang, Optoacoustic tweezers: a programmable, localized cell
concentrator based on opto-thermally generated, acoustically activated, surface
bubbles, Lab on a Chip, Vol. 13, pp. 1772-1779, 2013.
ACHIEVEMENTS
P.B. Breneman Best Design in Research Award, Second Place, 2013 Dean’s List (3 semesters), 2009-2012 President’s Freshman Award, The Pennsylvania State University, 2009 Captain of Second City Troop Youth Rugby Football Club, 2009
ACTIVITIES
Second City Troop Rugby Football Club, 2008 – current
Teaching Assistant for Engineering Mechanics, January 2014 – January 2015