Microsyst Technol (2009) 15:11571162

DOI 10.1007/s00542-008-0718-9

TECHNICAL PAPER

Microfluidic device for continuous magnetophoretic separation

of white blood cells
Ciprian Iliescu Guolin Xu Elena Barbarini
Marioara Avram Andrei Avram

Received: 30 June 2008 / Accepted: 3 October 2008 / Published online: 25 October 2008
Springer-Verlag 2008

Abstract This paper presents a microfluidic device for

magnetophoretic separation of red blood cells from blood
under continuous flow. The separation method consists of
continuous flow of a blood sample (diluted in PBS) through
a microfluidic channel which presents on the bottom
dots of ferromagnetic layer. By applying a magnetic
field perpendicular on the flowing direction, the ferromagnetic dots generate a gradient of magnetic field
which amplifies the magnetic force. As a result, the red
blood cells are captured on the bottom of the microfluidic
channel while the rest of the blood is collected at the outlet.
Experimental results show that an average of 95% of red
blood cells is trapped in the device.

1 Introduction
Manipulation and characterization of small quantities of
biological samples is one of the targets for lab on chip
devices. One approach is using electric fielddielectrophoresisfrequently used for cell trapping [Kua et al.
C. Iliescu (&)
G. Xu
Institute of Bioengineering and Nanotechnology,
31 Biopolis Way, The Nanos, #04-01,
Singapore 639798, Singapore
e-mail: ciliescu@ibn.a-star.edu.sg
E. Barbarini
Electronics Department, Politecnico di Torino,
24, Corso Duca degli Abruzzi, 10129 Torino, Italy
M. Avram
A. Avram
National Institute for Research and Development in
Microtechnologies, Erou Iancu Nicolae Str.,
32 B, Bucharest, Romania

2007, Iliescu et al. 2007b, Wang et al. 2007, Holmes et al.

2003) or cell separation (Hughes 2002, Yu et al. 2007,
Gascoyne and Vykoukal 2002) or even for electrofusion of
small lipid containersliposomes(Tresset and Iliescu
2007). The main disadvantage of the method is related to
the huge electric field gradient that can generate an
increase temperature (Tay et al. 2007). Meanwhile, the
dielectrophoretic separation methods are mainly based on
the difference between the dielectric properties of the cell
populations or the difference in cell size. Not all the time,
these conditions are satisfied. Another solution can be
using magnetic field. Immunomagnetic cell separation in
which magnetic particle are selectively attach to cell beads
(Ramadan et al. 2006b, Deng et al. 2002, Inglis et al. 2004,
Han and Frazier 2004), can be a very performing method
especially for separation of rare cell types. Other research
(Inglis et al. 2006, Han and Frazier 2006, Svoboda 2000)
used the native magnetic properties of biological cells
(deoxyhemoglobin red blood cellsRBCs) for separation.
Separation of red blood cells (RBCs) from whole blood is
often an essential step before the application of many clinical and molecular diagnostic tests on blood samples. The
procedure is also useful in blood transfusion applications as
it allows the selective transfusion of particular cell types.
Separation can be possible due to hemoglobin. Hemoglobin is a conjugated metal-protein comprising of four
polypeptide globins chains containing a ring structure
covalently bonded with a central ferrous iron atom (Fe2?),
which binds reversibly with oxygen. If is deoxygenated,
each of the four iron atoms contains four unpaired electrons,
giving the protein and the cell a substantial paramagnetic
moment. White blood cells (WBCs) do not contain hemoglobin and are diamagnetic particles. These properties are
the starting point for the trapping of RBCs particles using
high gradient magnetic separation.

123

1158

Here, we report a simple microfluidic device for

extraction of RBCs from blood under continuous flow. The
device consists of a microfluidic channel with ferromagnetic dots deposited on its bottom. The geometry of the
ferromagnetic layer and the application of an external
magnetic field perpendicular to the flow direction generate
a magnetic force on the RBCs present in the blood. The
RBCs are collected inside the channel. The glass structure
allows the complete visibility and analysis of the sample.

Microsyst Technol (2009) 15:11571162

the wire. Therefore, blood cells flowing near the ferromagnetic wire experience a magnetic force created by the
high gradient magnetic field near the ferromagnetic wire.
The magnetic force of a rectangular wire on blood cells,
located around the x-axis in Figure, can be calculated as:

 
2kl0 DvVa2 w a2
w 2
Fm 
k

cos
2u
H ar
h r2
h 0
r3
w
2kl0 DvVBC a2

sin
2u
1
H 2 a/ ; r [ a
h 0
r3
with

2 Device design
k
The structure of the device is presented in Fig. 1. A glass
die with inlet/outlet holes and a 60 lm-depth microfluidic
channel is bonded to another glass die, on which a ferromagnetic structuresquare dots (2 9 2 lm) of Ni
was patterned. A permanent magnet generates an external
magnetic flux of 0.2 T perpendiculars on the flow direction.
The blood is diluted with PBS and is flown through the
microfluidic channel. The ferromagnetic dots generate a
gradient of magnetic field which amplifies the magnetic
force that acts on RBCs. As a result the RBCs are trapped
by the ferromagnetic layer while the WBCs are flushed out
with the plasma and the other blood components.

3 Analytical considerations
A uniform external magnetic field applied normal to the
axis of a ferromagnetic wire is deformed near the ferromagnetic wire, and generates a high gradient magnetic
field, experienced by the magnetic particles moving around
Fig. 1 Schematic view of the
microfluidic device for RBCs
trapping

123

lw  lB
lw lB

where Dv is the relative magnetic susceptibility of a blood

cell relative to the buffer solution; lw and lB are the permeabilities of the ferromagnetic layer and the buffer
solution, respectively; V is the volume of the blood cell; a
is the lateral dimension of the ferromagnetic structure; r
and u are the cylindrical coordinates of the distance and
angle; H0 is the external magnetic field; and ar and au are
unit vectors for the distance and angle in the cylindrical
coordinate.
If the ferromagnetic structure is magnetically saturated,
the first term of the magnetic force in Eq. 1 is independent
of the external magnetic field, H0, and proportional to the
square of the saturation magnetization field and it. The
second and third terms of the magnetic force are linearly
proportional to the saturation magnetization field and the
external magnetic field.
For magnetic particles placed on the x-axis (u & 0 in
Figure), sin2u & 0, cos2u & 1, the ferromagnetic material attracts paramagnetic particles (RBCs) and repels

Microsyst Technol (2009) 15:11571162

1159

diamagnetic particles (WBCs). For magnetic particles

placed on the y-axis (u & 90 in Figure), sin2u & 0,
cos2u & -1, the material attracts diamagnetic particles
(WBCs) and repels paramagnetic particles (RBCs). The
first geometric configuration has been called the paramagnetic capture (PMC) mode; the latter has been called
the diamagnetic capture mode (DMC). Therefore, to
achieve a higher magnetophoretic force, the magnetophoretic microseparator is designed in the PMC mode with a
rectangular ferromagnetic structure.
To summarize, the main parameters that influence the
magnetic force applied on the particles affected by and
external magnetic field are: the intensity and orientation of
the magnetic field Ho; the permeability of the buffer solution
and the ferromagnetic material; the susceptibilities of the
cells and the buffer solution and the dimension of the device.
Can be shown (Ramadan et al. 2006a) that the Eq. 1 is
simplified for particles suspended in solution to:
Fm

1 DvV
rB2
2 l0

Equation 3 states that the magnetic force acting on an

induced magnetic dipole (in the particle) directly depends
on the energy-density gradient. Theoretically a magnetic
particle suspended in a perfectly uniform magnetic field of
density experiences no net magnetic force (Ramadan et al.
2006a), but the magnetic particle creates its own magnetic
field which generates magnetic field gradient. The equation
underlines the importance of generating a magnetic field
gradient especially when the size of the particle is microor nano-meter range. As the magnetic field gradient
increases the magnetic force increases and the particle
can be moved.
In addition to the magnetic force, the suspended magnetic particle experiences hydrodynamic drag and
gravitational forces. The effect of gravity is assumed to be
negligible due to the extremely small size of the particles.
In order to assess the feasibility of magnetic RBCs trapping by the system, it is necessary to evaluate the magnetic
force given by Eq. 3 and to compare it with the competing
forces acting on the bead during the separation process.
The hydrodynamic drag forceFhdis the most important
competing force opposing the magnetic force acting on the
bead and it is given by:
Fhd 6pgRv

1 DvR2
rB2
9 l0 g

The fabrication process of the device consists of three

important steps: fabrication of the top wafer with the inlet/
outlet holes and the microfluidic channel, fabrication of the
bottom wafer with ferromagnetic concentrators and
assembling of the wafers.
The fabrication of the top wafer was performed on a 4
glass wafer Corning 7740due to its good wet etching
properties (Iliescu et al. 2008)and is presented in Fig. 2.
For the etching of the microfluidic channelFig. 2aa Cr/
Au/photoresist mask was used (Iliescu et al. 2007a). The Cr/
Au (50 nm/500 nm) layer was deposited on a CHA e-beam
evaporator, the Au layer being deposited in three steps in
order to cover the deposition defects (Tay et al. 2006). The
patterning of the Cr/Au was performed using a 2 lm-thick
photoresist mask (AZ7220 from Clariant) and classical Au
and Cr etchants. In order to increase the hydrophobicity of
the masking layer, recommended for deep wet etching
process in high concentrated HF solution (Iliescu et al.
2006), the photoresist was hard backed at 120C for 30 min
on a hot plate. The wet etching process (Fig. 2b) was

where R is the particle radius, g is the solution viscosity,

and v is the particle velocity. Solving Eqs. 3 and 4, we
obtain the trapping condition:
v

4 Fabrication process

Fig. 2 Main steps of the fabrication process of the top wafer:

a processing of the Cr/Au/photoresist masking layer, b wet etching
of the microfluidic channel, c removing of the masking layer,
d processing of the Cr/Au/photoresist masking layer for inlet/outlet
holes, e wet etching of inlet/outlet holes, f removing of the second
masking layer

123

1160

performed in HF (49%)/HCl (37%) in ratio 10/1, optimized

by Iliescu et al. 2005. This etching solution assures a fast
etching rate (7.5 lm/min) as well as a good roughness of the
generated surface (Ra = 3 nm). HCl removes the insoluble
products resulted after the reaction of Al2O3 (present in the
glass composition) with HF. The etch depth was 60 lm
(performed during 8 min in a Teflon container using magnetic stirring). During the wet etching process the backside
of the wafer was protected by a dummy silicon wafer
bonded with wax on the glass wafer. The wax bonding and
debonding process was performed on a hot plate. After
removing of the Cr/Au/photoresist mask (Fig. 2c) using
classical resist stripper (NMP)for photoresist and residual
waxand Au and Cr etchants, a second Cr/Au/photoresist
mask was applied on the opposite surface of the glass wafer
(Fig. 2d). The thickness of the deposited Au layer was
1 lm, due to the fact that the required depth of the etching
process is increased (around 440 lm). The preparation of
the mask was similar with the process described for the
generation of the microfluidic channel. Similar, a second
wax bonding process will assure the protection of the
microfluidic channel, while the wax served also as etch-stop
layer for the deep wet etching process in same HF/HCl
(Fig. 2e). The etching time was 70 min (overetch for
compensation of the nonuniformity of the etching process).
Finally, the glass wafer was debonded from the dummy
silicon wafer and the masking layer was removed with a
similar procedure as was described before.
The fabrication of the bottom wafer consists of a simple
lift off process presented in Fig. 3. On a similar 4 glass
wafer (Corning 7740) a 5 lm-thick photoresist mask
(AZ4620 from Clariant) was deposited (Fig. 3a) followed by
a descuum process (1 min) on a RIE system using oxygen.
The metal layer Ti/Ni (50 nm/2 lm) was deposited in a CHA
e-beam evaporator (Fig. 3b). The photoresist masking layer
was finally removed in an ultrasonic bath using acetone.
For the assembling of the glass wafers a simple adhesive
bonding, using SU8-5 photoresist, was preferred to glass

Fig. 3 Main steps of the bottom wafer fabrication process: a

deposition of the photoresist mask, b deposition of Ti/Ni layer, c
removing of the photoresist mask

123

Microsyst Technol (2009) 15:11571162

Fig. 4 Assembling of the wafers: a deposition of SU8 on a dummy

wafer, b imprinting of the SU8 on a Teflon cylinder, c imprinting SU8
on the bonding area, d alignment and bonding

fusion bonding (Saarela et al. 2007) or adhesive bonding

using parylene (Poenar et al. 2007, Kim and Najafi 2005)
used in others applications. The technique consists of
imprinting of an SU8-5 layer initially spun on a dummy
silicon wafer (Fig. 4a) on a Teflon cylinder (Fig. 4b) and
transferring further the adhesive from the cylinder to the
top glass wafer (Fig. 4c). This contact imprinting technique
allowed deposition of a thin adhesive layer only on the
bonding regions. In the last step, the wafers are manually
align and bonded at 150C for 30 min with an applied
pressure of 1,000 N (Fig. 4d). Finally, the wafer is diced.
An image of the fabricated device is presented in Fig. 5.
The dimensions of the chip are 32 mm 9 9 mm.

Fig. 5 Photo of the microfluidic chip for RBCs trapping

Microsyst Technol (2009) 15:11571162

1161

Fig. 6 Image of the field

density of the cells on Neubauer
hemocytometer before and after
flowing the blood sample
through the microfluidic device

5 Results
Diluted blood (1:20 in PBS) was used for testing purpose.
The permanent magnet creates an external magnetic flux of
0.2 T and the flow rate was set in the range between 0.5 and
0.7 ml/h (using a syringe pump NE1000 from New Era
Pump Systems Inc, USA). Two connectors fabricated by
polymer printing secured the inlet and outlet connections.
The results are based on the analysis of the quantity of red
cells collected at the output of the device. Experimental
results show an average of 5% of red blood cells collected
at the output of the device. Figure 6 presents the image
with the field densities of cells before and after flowing the
sample through the microfluidic device.

6 Conclusions
The paper presents a microfluidic device for magnetophoretic trapping of red blood cells under continuous flow,
using high gradient of magnetic field. Experimental results
show a good trapping efficiency that can be further
improved.

References
Deng T, Prentiss M, Whitesides GM (2002) Fabrication of magnetic
microfiltration systems using soft lithography. Appl Phys Lett
80(3):461463. doi:10.1063/1.1436282
Gascoyne PRC, Vykoukal J (2002) Particle separation by dielectrophoresis. Electrophoresis 23(13):19731983. doi:10.1002/15222683(200207)23:13\;1973::AID-ELPS1973[;3.0.CO;2-1
Han KH, Frazier AB (2004) Continuous magnetophoretic separation
of blood cells in microdevice format. J Appl Phys 96(10):5797
5802. doi:10.1063/1.1803628
Han K, Frazier AB (2006) Paramagnetic capture mode magnetophoretic microseparator for high efficiency blood cell separations.
Lab Chip 6(2):265273. doi:10.1039/b514539b
Holmes D, Green NG, Morgan H (2003) Microdevices for dielectrophoretic flow-through cell separation. IEEE Eng Med Biol
Mag 22(6):8590. doi:10.1109/MEMB.2003.1266051

Hughes MP (2002) Strategies for dielectrophoretic separation in

laboratory-on-a-chip systems. Electrophoresis 23(16):25692582.
doi:10.1002/1522-2683(200208)23:16\;2569::AID-ELPS2569[;
3.0.CO;2-M
Iliescu C, Jing J, Tay FEH, Miao J, Sun TT (2005) Characterization of
masking layers for deep wet etching of glass in an improved HF/
HCl solution. Surf Coat Tech 198(13):314318. doi:10.1016/
j.surfcoat.2004.10.094
Iliescu C, Yu L, Xu G, Tay FEH (2006) A dielectrophoretic chip with
a 3D electric field gradient. J Microelectromech Syst 15(6):
15061513. doi:10.1109/JMEMS.2006.883567
Iliescu C, Tay FEH, Miao J (2007a) Strategies in deep wet etching of
Pyrex glass. Sens Actuators A 133(2):395400. doi:10.1016/
j.sna.2006.06.044
Iliescu C, Xu G, Loe F, Ong PL, Tay FEH (2007b) A 3 dimensional
dielectrophoretic filter chip. Electrophoresis 28(7):11071114.
doi:10.1002/elps.200600431
Iliescu C, Chen B, Miao J (2008) On the wet etching of Pyrex
glass. Sens Actuators A 143(1):154161. doi:10.1016/j.sna.2007.
11.022
Inglis DW, Riehn R, Austin RH, Sturm JC (2004) Continuous
microfluidic immunomagnetic cell separation. Appl Phys Lett
85(21):50935095. doi:10.1063/1.1823015
Inglis DW, Riehn R, Sturm JC, Austin RH (2006) Microfluidic high
gradient magnetic cell separation. J Appl Phys 99(8):08K101
Kim H, Najafi K (2005) Characterization of low-temperature wafer
bonding using thin-film parylene. J Microelectromech Syst
14(6):13471355. doi:10.1109/JMEMS.2005.859102
Kua CH, Lam YC, Rodriguez I, Yang C, Youcef-Toumi K (2007)
Dynamic cell fractionation and transportation using moving
dielectrophoresis. Anal Chem 79(18):69756987. doi:10.1021/
ac070810u
Poenar DP, Iliescu C, Carp M, Pang AJ, Leck KJ (2007) Glass-based
microfluidic device fabricated by Parylene wafer-to-wafer
bonding for impedance spectroscopy. Sens Actuators A 139(1
2):162171. doi:10.1016/j.sna.2006.10.009
Ramadan Q, Samper VD, Puiu DP, Yu C (2006a) Fabrication of threedimensional magnetic microdevices with embedded microcoils
for magnetic potential concentration. J Microelectromech Syst
15(3):624638. doi:10.1109/JMEMS.2006.876788
Ramadan Q, Yu C, Samper V, Poenar DP (2006b) Microcoils for
transport of magnetic beads. Appl Phys Lett 88(3):032501. doi:
10.1063/1.2149150
Saarela V, Haapala M, Kostiainen R, Kotiaho T, Franssila S (2007)
Glass microfabricated nebulizer chip for mass spectrometry. Lab
Chip 7(5):644646. doi:10.1039/b700101k
Svoboda J (2000) Separation of red blood cells by magnetic means.
J Magn Magn Mater 220(23):L103L105. doi:10.1016/S03048853(00)00479-0

123

1162
Tay FEH, Iliescu C, Jing J, Miao J (2006) Defect-free wet etching
through Pyrex glass using Cr/Au mask. Microsyst Technol
12(1011):935939. doi:10.1007/s00542-006-0116-0
Tay FEH, Yu L, Pang AJ, Iliescu C (2007) Electrical and thermal
characterization of a dielectrophoretic chip with 3D electrodes
for cells manipulation. Electrochim Acta 52(8):28622868. doi:
10.1016/j.electacta.2006.09.022
Tresset G, Iliescu C (2007) Electrical control of loaded biomimetic
femtoliter vesicles in microfluidic system. Appl Phys Lett
90(17):173901. doi:10.1063/1.2731517

123

Microsyst Technol (2009) 15:11571162

Wang L, Flanagan LA, Jeon NL, Monuki E, Lee AP (2007)
Dielectrophoresis switching with vertical sidewall electrodes
for microfluidic flow cytometry. Lab Chip 7(9):11141120. doi:
10.1039/b705386j
Yu L, Iliescu C, Xu G, Tay FEH (2007) Sequential field-flow cell
separation method in a dielectrophoretic chip with 3D electrodes. J Microelectromech Syst 16(5):11201129. doi:10.1109/
JMEMS.2007.901136