View Article Online / Journal Homepage / Table of Contents for this issue

PAPER www.rsc.org/loc | Lab on a Chip

Inertial microfluidics for continuous particle separation in spiral

microchannels
Sathyakumar S. Kuntaegowdanahalli,a Ali Asgar S. Bhagat,a Girish Kumarb and Ian Papautsky*a
Received 24th April 2009, Accepted 9th July 2009
First published as an Advance Article on the web 21st July 2009
DOI: 10.1039/b908271a
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34.

In this work we report on a simple inertial microfluidic device that achieves continuous multi-particle
separation using the principle of Dean-coupled inertial migration in spiral microchannels. The
dominant inertial forces coupled with the Dean rotational force due to the curvilinear microchannel
geometry cause particles to occupy a single equilibrium position near the inner microchannel wall. The
position at which particles equilibrate is dependent on the ratio of the inertial lift to Dean drag forces.
Using this concept, we demonstrate, for the first time, a spiral lab-on-a-chip (LOC) for size-dependant
focusing of particles at distinct equilibrium positions across the microchannel cross-section from
a multi-particle mixture. The individual particle streams can be collected with an appropriately
designed outlet system. To demonstrate this principle, a 5-loop Archimedean spiral microchannel with
a fixed width of 500 mm and a height of 130 mm was used to simultaneously and continuously separate
10 mm, 15 mm, and 20 mm polystyrene particles. The device exhibited 90% separation efficiency. The
versatility of the device was demonstrated by separating neuroblastoma and glioma cells with 80%
efficiency and high relative viability (>90%). The achieved throughput of 1 million cells/min is
substantially higher than the sorting rates reported by other microscale sorting methods and is
comparable to the rates obtained with commercial macroscale flow cytomerty techniques. The simple
planar structure and high throughput offered by this passive microfluidic approach make it attractive
for LOC devices in biomedical and environmental applications.

1. Introduction microscale leading to the development of numerous membrane-

less separation techniques.6
High throughput microparticle separations are becoming indis- Electrophoresis7,8 and dielectrophoresis9 have been used on the
pensable in many lab-on-a-chip (LOC) systems for biomedical1 microscale for achieving high resolution particle separation. As
and environmental2 applications. In most LOCs developed for the separation principle is size-based, separation of two or more
biomedical applications, microparticle separators are used in particle sizes can be achieved simultaneously. However, these
size-based separation and sorting of cells. For example, efficient techniques require an external power source and are unable to
separation of human T-lymphocytes (CD4+) from whole blood process large sample volumes due to batch mode operation. This
is a critical step in the diagnosis and treatment of HIV disease.3 has resulted in the development of several passive continuous
Alternatively, separation of neuroblastoma and glioma cells flow separation techniques such as pinched flow fractionation
may have potential applications in cell replacement therapy (PFF),10 sedimentation, hydrodynamic chromatography11,12
of neurodegenerative disorders (e.g., Parkinson’s disease, (HDC) and deterministic lateral displacement.13,14
Alzheimer’s disease, Multiple sclerosis) and cancer.4,5 Environ- Passive separation of two or more particle components
mental applications of microparticle separators include extrac- simultaneously has been successfully demonstrated on the
tion of harmful bacteria or metal nanoparticles in water quality microscale using PFF10 and deterministic lateral displacement
analysis.3 (DLD)13,14 based techniques. In PFF, fluid flows with and
Traditional size-based separation techniques on the macro- without particles are introduced into microchannels comprising
scale include the use of porous membrane filters. By employing of a pinched and broadened segment. In the pinched segment,
membranes with different pore sizes, multi-component particle particles are aligned along the channel sidewall by controlling the
filtration can be achieved. However, on the microscale, flow rate of the fluid flow without particles. As the particles travel
membrane filtration techniques suffer from various disadvan- from the pinched segment to the broadened microchannel
tages including fabrication of complex 3-D structures to define segment, smaller particles experience a force directed towards the
the pore sizes and issues arising from membrane clogging. These channel sidewalls while the bigger particles experience a force
factors have limited the popularity of this technique at the towards the channel center, thus achieving separation. In deter-
ministic lateral displacement, micro-pillars are suitably placed in
a
Department of Electrical and Computer Engineering, 814 Rhodes Hall, the microchannel such that particles larger than the critical
ML030, University of Cincinnati, Cincinnati, OH, 45221, USA. E-mail:
[email protected]; Fax: (513) 556-7326; Tel: (513) 556-2347
diameter dc ¼ 20%  2w (w is the separation between two pillars)
b
Department of Chemical Engineering, University of Cincinnati, follow a deterministic path leading to the formation of multiple
Cincinnati, OH, 45221, USA particle streams based on size. The method was successfully used

This journal is ª The Royal Society of Chemistry 2009 Lab Chip, 2009, 9, 2973–2980 | 2973
 View Article Online

to demonstrate efficient sorting of 0.8 mm, 0.9 mm and 1 mm with ap/Dh $ 0.07 (where, ap is particle diameter and Dh is the
diameter microspheres. Although these techniques operate in microchannel hydraulic diameter), the lift forces dominate and
continuous mode, the need for narrow channel geometries in are responsible for the particles equilibrating within the micro-
PFF and presence of obstructions in DLD may lead to channel channel.17–19 Herein, we demonstrate that the position at which
clogging and particle–particle interactions. Additionally, the particles of different sizes equilibrate is dependent on the ratio of
reported throughputs are insufficient for cell sorting and lift and Dean drag forces which varies with the third power of the
cytometry applications. particle diameter. By modulating the flow parameters, individual
Recently, inertial migration based particle filtration techniques particle streams can be generated for separation applications as
have been reported to achieve high throughput particle separa- shown in Fig. 1.
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34.

tion.15–19 Under the influence of the inertial forces, neutrally In a plane Poiseuille flow, the parabolic nature of the velocity
buoyant particles flowing in a microchannel migrate to stable profile results in a shear gradient induced inertial lift force that
equilibrium positions along the channel periphery. Di Carlo drives the suspended microparticles away from the microchannel
et al.18 showed that by employing a curvilinear channel the Dean center towards the channel walls. As the particles move closer to
vortices can be used to reduce the number of equilibrium posi- the channel wall, the rotational wake around the particles is
tions to one. This single particle stream can later be extracted by disturbed due to the presence of the wall inducing a lift force on
employing bifurcated outlets. Using this technique rapid micro- the particles directed away from the wall.18,23,24 These oppositely
particle filtration can be achieved.17–22 For example, we recently directed forces exert a net lift force (FL) on the particles equili-
demonstrated complete separation of a two particle mixture brating them into focused streams around the microchannel
(7.32 mm and 1.9 mm diameter) by ensuring that the smaller perimeter. Segre and Silberberg were the first to report this effect
particle motion was affected solely by Dean forces and the larger demonstrating that a uniformly distributed suspension of
particle migration by inertial lift forces.17 neutrally buoyant particles form a narrow band at 0.6R from the
In our previous work,17 we introduced a design to achieve channel center in a circular channel of radius R.25,26 Chun and
complete separation of two different particle sizes by exploiting
the effects of Dean drag and inertial lift forces. The larger
particles equilibrated under the combined influence of the inertial
lift forces and the Dean force at the inner microchannel wall. The
inertial lift forces however did not affect migration of the smaller
particles, which were transposed to the outer microchannel wall
due to Dean drag. Since particle separation was not dependent
on the ratio of the two forces, a major limitation of the design
was its inability to separate more than two particle mixtures
irrespective of the number of outlets.
In this paper, we demonstrate for the first time the use of Dean
coupled inertial migration to simultaneously separate multi-
particle mixtures. In a curvilinear microchannel, a combination
of Dean drag and inertial lift forces result in particle equilibra-
tion at the inner microchannel wall. The position at which the
particles equilibrate is dependent on the ratio of these two forces.
In this work we have developed a design which exploits the
particle size dependence of the ratio of the two forces to form
segregated, focused particle streams which can be extracted by
employing a suitable outlet system. Although the design reported
herein is similar to the previously reported spiral separators,17 the
working principle for achieving separation is different and is the
first demonstration of a multi-particle separation using inertial
microfluidics. Additionally, the system reported in this work is
simple, easy to fabricate, and is capable of continuously sepa-
rating particles with a high throughput over a wide dynamic
range. The proposed principle was also used to demonstrate size-
based cell separation between SH-SY5Y neuroblastoma cells and
C6 glioma cells.
Fig. 1 (a) Schematic of the spiral microparticle separator. The randomly
dispersed particles equilibrate at different equilibrium positions along the
2. Design principle inner wall (IW) of the spiral microchannel under the influence of FL and
FD. Separation between individual particle streams is enhanced by
In our recent work,17 we showed that particles flowing in a spiral opening the spiral channel into a wider straight channel before extracting
microchannel with rectangular cross-section experience the individual streams using a multiple outlet design. (b) Microchannel
a combination of inertial lift and Dean drag forces. The magni- cross-section illustrating effects of FL and FD on particles. The ratio of
tude and direction of these forces depends on the particle size and forces (FL/FD) is the determining factor in where a particle of given size
position across the microchannel cross-section. For particles (diameter) equilibrates.

2974 | Lab Chip, 2009, 9, 2973–2980 This journal is ª The Royal Society of Chemistry 2009
 View Article Online

Ladd showed the preferential equilibration of particles with streams which can be independently extracted by designing
ap/Dh  0.1 in square cross-sectional channels.27 In our previous appropriate outlets.
work, we showed the equilibrium positions in square and rect-
angular microchannels at finite Re (1 < Re < 50) flows.15,16
3. Experimental methods
The net lift force acting on the particle suspended in a plane
Poiseuille flow is given by24 The microchannels were fabricated in polydimethylsiloxane
(PDMS, Sylgard 184, Dow Corning) using standard soft
FL ¼ rG2CLap4 (1) lithography methods. Briefly, a negative silicon master was
fabricated by patterning SU-8 photoresist (2075, Microchem
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34.

where r is the density of fluid medium, G is the shear rate of the Corp.) of desired height using the conventional photolithog-
fluid given by G ¼ Umax/Dh, Umax is the maximum fluid velocity raphy process.32 Subsequently, PDMS polymer mixed in the
and CL is the lift co-efficient which is a function of the particle ratio of 10:1 with the curing agent was cast onto the fabricated
position across the channel cross-section and channel Reynolds silicon master to form a replica of the device. Following curing
number (Re). The magnitude of lift coefficient and hence the lift on a hotplate for 2 h at 80  C, the PDMS molds were peeled and
force rises from zero at the channel center to a maximum value input and output ports were cored using a 14 gauge syringe
before falling back to zero again at a distance of 0.2Dh away needle. The PDMS mold was then bonded to a 1 mm thick
from the channel wall, equilibrating the particles.24 Beyond this microscopic glass slide to complete the channel using corona
point, the lift coefficient changes sign indicating a dominance of wand treatment (BD-20AC, Electro-Technic Products Inc.).
wall-induced lift force. Fluorescently labeled polystyrene particles purchased from
The curvilinear geometry of a spiral microchannel introduces Bangs Laboratories were diluted in DI water before testing
a centrifugal acceleration component directed radially outward (0.05% volume fraction). A syringe pump (KDS101, KD
resulting in the formation of two symmetric-counter rotating Scientific Inc.) was used to drive syringes filled with the particle
vortices known as the Dean vortices in the top and bottom halves solution. To evaluate particle stream positions within the
of the microchannel.28,29 The magnitude of these Dean vortices microchannel, high speed images of the microchannel were
can be determined by a dimensionless number known as the captured using an inverted epi-fluorescence microscope (IX71,
Dean Number (De), which is given by Olympus Inc.) equipped with a 12-bit CCD camera (Retiga EXi,
rffiffiffiffiffiffi rffiffiffiffiffiffi QImaging). Using ImageJ software, Z-stacked composite
rUf Dh Dh Dh images were generated by overlaying a stack of 300 images. The
De ¼ ¼ Re (2)
m 2R 2R particle stream position was determined by analyzing the gray-
scale line scans across the channel width in the composite
For a straight microchannel De ¼ 0. Increasing the channel image.33
cross-section (Dh) or the flow rate increases De, resulting in To demonstrate separation of particle mixtures, a solution
stronger Dean forces. A particle suspended in a spiral micro- containing 10 mm (s ¼ 0.54 mm), 15 mm (s ¼ 0.98 mm), and 20 mm
channel experiences a transverse drag force due to these Dean (s ¼ 0.5 mm) diameter polystyrene particles labeled with DAPI,
vortices. The expression for the Dean drag force experienced by FITC, and TRITC fluorophores respectively was used. The
the particle can be arrived at by assuming Stokes drag.17,30,31 particle streams were viewed and captured separately using
appropriate filter cubes. The individual images were super-
FD ¼ 3pmUDeanap ¼ 5.4  104pmDe1.63ap (3) imposed to create a composite image to display the formation of
three separate focused particle streams.
where UDean is the average Dean velocity given by UDean ¼ To quantify separation efficiency, the outlet streams at each of
1.8  104De1.63. the eight outlets were collected and analyzed by flow cytometry.
Particle equilibration in rectangular microchannel cross- The BD LSR II (BD Biosciences) flow cytometer was used to
section is independent of Dh and rather depends on the shortest perform the analysis. Individual particle solutions were used as
channel dimension (microchannel height, H) due to varying controls to draw the gates on a FSC vs. SSC plot. The collected
shear rates across the channel cross-section.15,16 Hence, the samples were then run through the flow cytometer to determine
criterion for particle focusing is ap/H $ 0.07. We use this result to counts for each of the three particles in all the eight outlets.
design low aspect ratio spiral microchannels to enhance the To demonstrate cell separations, SH-SY5Y neuroblastoma
separation between individual particle streams over a wide range cells 15 mm (s ¼ 5 mm) in diameter were cultured in Opti-MEM
of particle sizes. The dependence of ratio of the lift force (FL) to medium containing 10% fetal bovine serum supplemented with
Dean drag force (FD) on particle size (FL/FD f ap3) and the shear L-glutamine. The C6 rat glioma cells 8 mm (s ¼ 3 mm) in
rate modulation in low aspect ratio microchannels have been diameter were cultured in Ham’s F-12 medium containing 15%
exploited to demonstrate separation of 10 mm, 15 mm, and 20 mm horse serum and 5% fetal bovine serum supplemented with
diameter particles. The dominant inertial lift forces align the 0.25 mg/mL amphotericin B, 100 U/mL penicillin, and 100 mg/mL
randomly distributed particles at the inlet near the inner micro- streptomycin. Both cultures were maintained at 37  C in
channel wall as the flow progresses downstream. On the other a humidified atmosphere containing 5% (v/v) CO2. Cells were
hand, the significant Dean drag force move these focused streams cultured in sterile 25 cm2 flasks (Corning). Cells were sub-culti-
farther away from the channel wall depending on the particle vated (1:4) three times a week and media was replaced every
size, with the largest particle being closest to the inner channel 48 hours. Sub-confluent monolayers were dissociated with 0.01%
wall. This results in the evolution of three distinct particle trypsin and 5.3 mM EDTA solution, re-suspended in fresh basal

This journal is ª The Royal Society of Chemistry 2009 Lab Chip, 2009, 9, 2973–2980 | 2975
 View Article Online

media with 5% serum for further subculture. Cells were never

allowed to reach 100% confluent. Prior to testing, the cells were
mixed together and diluted in 1 phosphate buffered
saline (PBS) solution (0.05% volume fraction). The SH-SY5Y
neuroblastoma cells were labeled with CellTracker Green
(Invitrogen Corp.) for better visualization and to confirm sepa-
ration. The PDMS based devices were thoroughly flushed with
PBS and antibiotics (1  PSN) before running the cell mixture
and all experiments were conducted in a sterile environment.
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34.

4. Results and discussion

The fabricated devices consisted of a five loop Archimedean
spiral microchannel with two inlets and eight equally spaced
outlets (Fig. 2). The spiral designs had an initial radius of
curvature of 1 cm, with spacing between the successive spiral
loops fixed at 500 mm. Width of microchannels was fixed at
500 mm, while microchannel height was varied from 90 mm to
Fig. 3 Fluorescent images (pseudo-colored purple) illustrating the
140 mm. At the outlet, the 500 mm wide channel opened into position of the focused 10 mm diameter particle stream in spiral micro-
a 1 mm wide segment to increase spacing between particle channels of different heights at varying De.
streams before splitting into eight 100 mm wide outlets (Fig. 2(b)).
In this work, we take advantage of the particle size dependence
both inlets and all images were captured in the 500 mm section
of the ratio of FL and FD in spiral microchannels to separate
just prior to the outlet. For a given channel height, the focused
10 mm, 15 mm and 20 mm diameter particles. Before testing the
particle stream moved away from the channel wall at increasing
particle mixture, particles were tested individually in channels of
De, indicating the dominance of the Dean force. Although an
varying heights and De. For each channel height, De was
increase in flow velocity results in a greater lift force (FL f Uf2) as
increased by increasing the flow rate until a single focused
compared to the Dean drag (FD f Uf1.63), the stream movement
particle stream formed at the outlet. Increasing the flow rate
away from the wall can be explained by the decrease in the lift co-
further, resulted in migration of the focused particle stream away
efficient with increasing fluid flow velocity.24 Hence, the net lift
from the channel inner wall towards the channel center. As the
force reduces with increasing flow velocity and the particle
flow rate increased further, the Dean forces begin to dominate
stream moves away from the channel wall. Increasing the
the inertial lift forces resulting in de-focusing of the particle
channel height also resulted in focusing of particle streams
stream. Similar behavior was also noted by Di Carlo et al.18 The
further away from the channel inner wall, which may be
equilibrium positions of the particle streams were recorded as
explained by the fact that for a given flow velocity Uf the Dean
a function of De and these results were used to select the
drag force increases with channel height, while the lift force
optimum channel height and flow rate required to demonstrate
decreases with increasing channel height. Thus, the particle
multi-particle separation.
stream position can be altered by either increasing the De or
Fig. 3 shows composite images indicating the position (x) of
increasing the channel height.
the 10 mm diameter particle streams as a function of De in
Fig. 4(a) plots position of particle streams relative to the
microchannels of varying height (H ¼ 90 mm–140 mm). The
channel width (W) inside microchannels of different height as
suspension of 10 mm diameter particles was introduced through
a function of De. For a 10 mm diameter particle in a 90 mm high
channel (ap/H  0.11), the distance of the particle stream from
the wall remained constant up to De ¼ 8.8 before rising linearly.
The lift force acting on the particle below De ¼ 8.8 was very
much greater than FD and hence the particle stream was unaf-
fected by an increase in the fluid velocity. However, beyond
a critical De (flow rate), FD increases to the same order as FL and
hence the particle stream position varied linearly with the flow
rate indicating a dominance of Dean drag. Increasing micro-
channel height reduced the critical De required for particle
stream migration, as seen in Fig. 4(a), due to a larger Dean force
with increase in channel height. Hence, in the case of 130 mm and
140 mm high channels (ap/H  0.07), the particle stream position
varied for all flow conditions tested.
Fig. 2 (a) Photograph of the 5-loop spiral microchannel with two inlets
and eight outlets fabricated in PDMS (the microchannel is filled with dye For the larger 15 mm diameter particles, the higher ap/H ratio
for visualization). The outlets are numbered 1–8, starting from the inner in the case of the 90 mm and 110 mm microchannels resulted in
channel wall. (b) SEM image illustrating the outlet section of the spiral a larger lift force and hence the particle stream positions
microchannel. remained predominantly constant for increasing De. However,

2976 | Lab Chip, 2009, 9, 2973–2980 This journal is ª The Royal Society of Chemistry 2009
 View Article Online

force in the case of the larger particles is primarily due to

the strong dependence of the lift force on the particle diameter
(FL f ap4 vs. FD f ap). Thus, in the 90 mm channel, the 20 mm
diameter particle stream position was constant for almost all
flows tested.
Results from these individual particle tests indicate that
a complete separation between the three particle sizes can be
achieved in microchannels of different heights and a range of
flow conditions. Thus, in the next set of experiments we tested
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34.

a homogenous mixture of 10 mm, 15 mm, and 20 mm diameter

polystyrene particles. Particle separation was experimentally
observed by using 10 mm, 15 mm, and 20 mm diameter particles
labeled with DAPI, FITC and TRITC fluorophores. Corre-
sponding filter cubes were used to capture images of each of the
particle streams at the channel outlet. The captured images were
then superimposed to create a composite image showing the
three particle streams (Fig. 5). To extract the three individual
particle streams, the mixture was tested using the 130 mm high
spiral microchannel at De ¼ 14.4 (flow rate 3 mL/min). The
500 mm wide channel opened into a 1 mm wide section prior to
the outlet to achieve better separation by taking advantage of the
laminar flow profiles in the outlet channels.
The composite images of the inlet and outlet sections are
shown in Fig. 5. The image clearly indicates the formation of
three distinct particle streams at the microchannel outlet. The
position of the 10 mm, 15 mm, and 20 mm particle streams in the
500 mm wide section of the channel were 180 mm, 120 mm, and
65 mm respectively from the inner microchannel wall. Thus, the
three particles were collected at the first, second, and third outlets
respectively. Although the channel height determines whether the
particles focus into a single stream, it is the microchannel width
that greatly influences the spacing between individual particle
streams. Thus, wider channels result in larger spacing between
particle streams, which in turn permits separation of particles
with closely spaced diameters. The influence of the microchannel
width on the resolving capability of this separation technique is
currently being investigated.
The particle streams from each of the eight outlets were
collected and analyzed by flow cytometry to quantify the sepa-
ration efficiency. Fig. 6 presents the cytometry data indicating
the particle concentration at the inlet and outlets. Nearly 98% of
the particles were filtered out at outlets 1, 2 and 3 (the outlets are
numbered starting from the inner wall) suggesting a high degree
of particle focusing. Separation efficiency of 90% was observed
between the three particles. Even higher separation efficiencies
may be achieved by using mixtures of mono-dispersed particles.
Fig. 4 Plots illustrating position of the (a) 10 mm, (b) 15 mm, and (c) Although only separation of three particles sizes has been
20 mm diameter particle streams from the inner channel wall for demonstrated in this work, a larger number of particle sizes can
increasing De in microchannels ranging from 90 mm to 140 mm in height. be simultaneously separated by using wider spiral microchannels
to increase spacing between the multiple focused particles
streams.
similar to the 10 mm diameter particles, an increase in channel Application of the developed technique to high-throughput
height resulted in a lower ap/H  0.1 ratio and hence the cell sorting was demonstrated by separation of SH-SY5Y
position of the particle stream was strongly influenced by the neuroblastoma and C6 glioma cells. Complete separation of
Dean force. Similarly, the 20 mm diameter particles yielded these neural stem cells is of critical importance to understanding
higher ap/H, resulting in significant FL acting on particles in each the specific and unique functions these cells play in the central
of the channel heights considered. Hence, a prominent nervous system (CNS), and potential applications in cell
flat (constant position) region is observed in all plots for the replacement therapy in many neurodegenerative disorders (such
20 mm diameter particles. The increased dominance of the lift as Parkinson’s, Alzheimer’s, or Multiple sclerosis) and cancer.4,5

This journal is ª The Royal Society of Chemistry 2009 Lab Chip, 2009, 9, 2973–2980 | 2977
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34. View Article Online

Fig. 5 Superimposed fluorescent images illustrating distribution and position of the 20 mm (pseudo-colored orange), 15 mm (pseudo-colored green), and
10 mm (pseudo-colored purple) diameter particles. The panels represent the inlet, a 500 mm wide channel section just prior to the outlet, and the
bifurcated outlet of the 130 mm high spiral microchannel at De ¼ 14.4. The randomly distributed particles at the inlet form ordered focused streams
which are then collected separately at outlets 1, 2 and 3.

Fig. 6 Flow cytometry data indicating concentration of the three particle mixture. The dot plots indicate concentration of the 10 mm, 15 mm, and 20 mm
diameter particles at (a) the inlet, and (b–d) the first three outlets of the spiral channel. The gates in the side scatter vs. forward scatter panels were drawn
using pure particle samples in order to discriminate between the three particle sizes. (e) Particle counting results illustrating particle distribution across
the first four outlets of the spiral microchannel. A 90% separation efficiency between the three particle sizes was achieved.

2978 | Lab Chip, 2009, 9, 2973–2980 This journal is ª The Royal Society of Chemistry 2009
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34. View Article Online

Fig. 7 Bright-field and epifluorescent images illustrating the distribution of the bigger 15 mm diameter SH-SY5Y cells (pseudo-colored green) and the
smaller 8 mm diameter C6 glioma cells at the inlet and the first two outlets of the spiral microchannel (scale bar 100 mm). Also shown in the figure are the
cell counting results clearly indicating 80% separation efficiency between the two cells types, with the SH-SY5Y cells collected at outlet 1 and the C6
glioma cells at outlet 2.

Based on the size of these cells, the mixture was passed through than the sorting rates achievable by other microscale sorting
a 120 mm high microchannel at De ¼ 11.8 to collect the bigger methods. In fact, the throughput of our system is quite compa-
15 mm diameter cells from outlet 1 and the smaller 8 mm rable to the sorting rates obtained with commercial flow
diameter cells from outlet 2. Fig. 7 shows the bright-field and cytometry techniques, which have a maximum throughput of
epifluorescent images illustrating the distribution of the bigger 2.4 million cells/min.36 Since only a 0.05% volume fraction was
SH-SY5Y cells (fluorescently labeled) and the smaller C6 glioma used in our experiments, the throughput may be increased even
cells (unlabeled) at the inlet and at the first two outlets of the further, nearly an order of magnitude, by simply increasing the
spiral microchannel. The inlet solution consisted of the two cell cell volume fraction and fluid flow rates (volume fraction values
mixture with equal cell concentrations (500,000  2 cells/mL). of 0.1 to 0.3% have been reported by others34,35).
Nearly 90% of the cells were collected at outlets 1 and 2, indi-
cating a high degree of cell focusing. The cell separation effi-
5. Conclusions
ciency was >80% for both the bigger SH-SY5Y cells at outlet
1 and the smaller C6 glioma cells at outlet 2, but was limited by the In this work, we introduced an inertial microfluidic system for
large variations in the cell sizes (s ¼ 5 mm). A potential concern passive simultaneous separation of a microparticle mixture. The
when separating cells is the possibility of damage to the cells due effect of Dean coupled inertial migration in spiral microchannels
to the high shear forces. Cell viability following separation was was exploited to generate focused streams of individual particles at
confirmed by bringing both SH-SY5Y neuroblastoma and C6 distinct positions within the microchannels based on their size.
glioma cells back into culture using the procedure described in the Tests with individual particles were conducted to identify the effects
methods section with >90% cell recovery after 24 hrs. of microchannel height and the Dean number on particle stream
As Pamme6 discusses in a recent review, continuous flow position. Based on the results obtained from individual particle
separation techniques are becoming popular on the microscale experiments, a 130 mm high spiral channel was employed to sepa-
due to their ability to achieve high sample throughputs. rate a mixture of 10 mm, 15 mm, and 20 mm diameter particles. The
The average throughput of many of these microfluidic systems is separation efficiency of the device was calculated to be around 90%.
2,000 cells/min,6 although higher throughputs have been recently The device was also used to show a high throughput (1 milion
reported. For example, Fu et al.34 demonstrated a throughput of cells/min) separation of neural cells with high viability, demon-
1,200–7,200 cells/min while separating Escherichia coli cells strating application of this technique in microscale cell sorting. The
expressing green fluorescent protein from a background of passive separation principle and the planar nature of the described
non-fluorescent E. coli cells in microfluidic channels using fluo- design will permit easy integration with existing LOC systems
rescently-activated sorting. A similar throughput (5,400 cells/ requiring high-throughput separations.
min) was demonstrated using the PFF approach by Takagi et al.35
who separated erythrocytes from diluted blood.
The developed spiral microparticle separator is ideal for
Acknowledgements
achieving high throughput separations due to the fact that This work was supported by the University of Cincinnati Insti-
inertial and Dean drag forces acting on particles increase with tute for Nanoscale Science and Technology and the National
increasing flow rates. For the flow rates tested (in the mL/min Institute of Occupational Safety and Health (NIOSH) Health
range), a sorting rate of 1 million cells/min was achieved in our Pilot Research Project Training Program of the University of
system. This throughput is substantially higher (more than 100) Cincinnati Education and Research Center (T42/OH008432-04).

This journal is ª The Royal Society of Chemistry 2009 Lab Chip, 2009, 9, 2973–2980 | 2979
 View Article Online

References 17 A. A. S. Bhagat, S. S. Kuntaegowdanahalli and I. Papautsky, Lab

Chip, 2008, 8, 1906–1914.
1 M. Toner and D. Irimia, Annu. Rev. Biomed. Eng., 2005, 7, 77–103. 18 D. Di Carlo, D. Irimia, R. G. Tompkins and M. Toner, PNAS, 2007,
2 G. Blankenstein and U. D. Larsen, Biosens. Bioelect., 1998, 13, 104, 18892–18897.
427–438. 19 D. Di Carlo, J. F. Edd, D. Irimia, R. G. Tompkins and M. Toner,
3 M. Kersaudy-Kerhoas, R. Dhariwal and M. P. Y. Desmulliez, IET Anal. Chem., 2008, 80, 2204–2211.
Nanobiotechnol., 2008, 2, 1–13. 20 I. Gregoratto, C. J. McNeil and M. W. Reeks, Proc. SPIE, 2007,
4 Z. Wu, K. Hjort, G. Wicher and A. Fex Svenningsen, Biomed. 6465, 646503.
Microdevices, 2008, 10, 631–638. 21 J. Seo, M. H. Lean and A. Kole, Appl. Phys. Lett., 2007, 91,
5 E. Hedlund, J. Pruszak, A. Ferree, A. Vinuela, S. Hong, O. Isacson 033901.
and K. S. Kim, Stem Cells, 2007, 25, 1126–35. 22 J. Seo, M. H. Lean and A. Kole, J. Chromatogr. A, 2007, 1162,
Published on 21 July 2009. Downloaded by University of Texas Libraries on 19/10/2014 08:00:34.

6 N. Pamme, Lab Chip, 2007, 7, 1644–1659. 126–131.

7 X. Xu, K. K. Caswell, E. Tucker, S. Kabisatpathy, K. L. Brodhacker 23 R. Eichhorn and S. Small, J. Fluid Mech., 1964, 20, 513–527.
and W. A. Scrivens, J. Chromatogr. A, 2007, 1167, 35–41. 24 E. S. Asmolov, J. Fluid Mech., 1999, 381, 63–87.
8 F.-K. Liu, Y.-Y. Lin and C.-H. Wu, Anal. Chim. Acta, 2005, 528, 25 G. Segre and A. Silberberg, J. Fluid Mech., 1962, 14, 136–157.
249–254. 26 G. Segre and A. Silberberg, Nature, 1961, 189, 209–210.
9 M. Durr, J. Kentsch, T. Muller, T. Schnelle and M. Stelzle, 27 B. Chun and A. J. C. Ladd, Phys. Fluids, 2006, 18, 031704.
Electrophoresis, 2003, 24, 722–731. 28 W. R. Dean, Phil. Mag. Ser. 7, 1927, 4, 208–223.
10 M. Yamada, M. Nakashima and M. Seki, Anal. Chem., 2004, 76, 29 W. R. Dean, Phil. Mag. Ser. 7, 1928, 5, 673–695.
5465–5471. 30 S. Ookawara, R. Higashi, D. Street and K. Ogawa, Chem. Eng. J.,
11 E. Chmela, R. Tijssen, M. T. Blom, H. J. G. E. Gardeniers and A. van 2004, 101, 171–178.
den Berg, Anal. Chem., 2002, 74, 3470–3475. 31 S. Ookawara, D. Street and K. Ogawa, Chem. Eng. Sci., 2006, 61,
12 M. T. Blom, E. Chmela, R. E. Oosterbroek, R. Tijssen and A. van den 3714–3724.
Berg, Anal. Chem., 2003, 75, 6761–6768. 32 A. A. S. Bhagat, E. T. K. Peterson and I. Papautsky, J. Micromech.
13 L. R. Huang, E. C. Cox, R. H. Austin and J. C. Sturm, Science, 2004, Microeng., 2007, 17, 1017–1024.
304, 987–990. 33 A. A. S. Bhagat and I. Papautsky, J. Micromech. Microeng., 2008, 18,
14 W. Inglis, J. A. Davis, R. H. Austin and J. C. Sturm, Lab Chip, 2006, 085005.
6, 655–658. 34 A. Y. Fu, C. Spence, A. Scherer, F. H. Arnold and S. R. Quake,
15 A. A. S. Bhagat, S. S. Kuntaegowdanahalli and I. Papautsky, Nature Biotech., 1999, 17, 1109–1111.
Microfluid. Nanofluid., DOI: 10.1007/s10404-008-0377-2. 35 J. Takagi, M. Yamada, M. Yasuda and M. Seki, Lab Chip, 2005, 5,
16 A. A. S. Bhagat, S. S. Kuntaegowdanahalli and I. Papautsky, Phys. 778–784.
Fluids, 2008, 20, 101702. 36 T. D. Chung and H. C. Kim, Electrophoresis, 2007, 28, 4511–4520.

2980 | Lab Chip, 2009, 9, 2973–2980 This journal is ª The Royal Society of Chemistry 2009