Professional Documents
Culture Documents
Microfluidics PDF
Microfluidics PDF
MICROFLUIDICS:
THEORY AND APPLICATIONS
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or
by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no
expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No
liability is assumed for incidental or consequential damages in connection with or arising out of information
contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in
rendering legal, medical or any other professional services.
MICROFLUIDICS:
THEORY AND APPLICATIONS
IVAN A. KUZNETSOV
EDITOR
Additional color graphics may be available in the e-book version of this book.
LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA
Microfluidics : theory and applications / editor, Ivan A. Kuznetsov.
p. cm.
Includes index.
ISBN 978-1-62100-034-1 (eBook)
1. Microfluidics. I. Kuznetsov, I. A. (Ivan Aleksandrovich), 1953TJ853.4.M53M545 2009
629.8'042--dc22
2010012217
CONTENTS
Preface
Chapter 1
Chapter 2
Chapter 3
Chapter 4
Chapter 5
vii
Microbiochips Monolithically Integrated
with Microfluidics, Micromechanics, Photonics, and
Electronics by 3D Femtosecond Laser Direct Writing
Ya Cheng, Koji Sugioka,
Katsumi Midorikawa and Zhizhan Xu
Electrokinetic Flows of Non-Newtonian
Fluids in Microfluidic Channels
Cunlu Zhao and Chun Yang
Microfluidic Electro- chemiluminescent Detection
Devices with Capillary Electrophoresis
K. M. Muzyka and M. M. Rozhitskii
Microfluidic Valves without Diaphragms:
Hydrogel Valves and PDMS-Based
Rotary Selection Valves
Steffen Howitz, Frank Baudisch,
Frank-Ulrich Gast, Andreas Richter,
Andreas Grodrian, Gunter Gastrock
and Josef Metze
Biocompatible and Mass Productive MEMS
Device for Localized Surface Plasmon Resonance
Xiaodong Zhou and Hong Liu
55
103
135
149
vi
Chapter 6
Index
Contents
Application to Adaptive Optics
and Laser Microfluidics
Jean-Pierre Delville, Alexis Casner,
Rgis Wunenburger and Iver Brevik
167
175
PREFACE
Microfluidics deals with the behavior, precise control and manipulation of
fluids that are geometrically constrained to a small, typically sub-millimeter
scale. It is a multidisciplinary field intersecting engineering, physics,
chemistry, microtechnology and biotechnology, with practical applications to
the design of systems in which such small volumes of fluids will be used. This
book presents topical data on microfluidics such as the application of
microfluidic principles for the creation of tandem capillary electrophoresis
chip devices with electrochemiluminscent detection techniques; microbiochips
monolithically integrated with microfluidics; analysis of the electroosmotic
flow of power-law fluids in a slit channel and microfluidic valves without
diaphragms.
Chapter 1 - Microbiochips such as lab-on-a-chip (LOC) devices and micro
total analysis systems (-TAS) can be regarded conceptually as a biological
equivalent of conventional silicon integrated circuits, which involve
miniaturization and integration of electronics. As its name implies, an ideal
LOC should have a very small footprint and yet still be capable of functioning
as a laboratory in which partial or complete chemical or biological analysis
can be performed automatically. A planar silicon chip that integrates only
microelectronics is clearly far from adequate for such a purpose, and other
functional components such as microfluidics, microoptics, and
micromechanics need to be incorporated into a chip with 3D configurations.
This poses a formidable challenge, because current mainstream
microfabrication technologies mostly rely on optical lithography, which is
essentially a surface structuring technology. Recently, femtosecond laser direct
writing has exhibited great potential for producing a variety of true 3D
microstructures in various transparent materials. This is enabled by the
viii
Ivan A. Kuznetsov
nonlinear interaction between the tightly focused femtosecond laser beam and
the material that is transparent to the laser beam, since the interaction can be
effectively confined to the vicinity of focal point where the laser intensity
exceeds the threshold for multiphoton absorption. In this chapter, we provide
an overview of 3D femtosecond laser direct writing technology and highlight
its potential for fabrication of complex smart microsystems by furnishing some
examples of fabrication and hybrid integration of microfluidics,
micromechanics, photonics, and electronics. In addition, application of some
of our microfluidic chips to biological research has resulted in new insights
into various biological behaviors such the dynamics and functions of
microorganisms. Possible solutions for overcoming remaining challenges are
discussed in the hope that this technology will provide a manufacturing
solution for the emerging LOC industry.
Chapter 2 - Advanced microfluidic devices can perform complete
biochemical analysis in a single fabricated chip. The generic microfluidic
systems involve buffer solutions and samples manipulations such as pumping,
valving, mixing, injection, dispensing, concentration etc. Fundamental
understanding of the liquid flow characteristics in microchannels is thus
essential to optimum design and precise control of microfluidic devices. In
general, liquid motion can be generated by either applying a pressure gradient
or imposing an electric field, leading to pressure-driven flow or
electrokinetically-driven flow, respectively. Traditionally, in large-sized
channels flow is often driven by pressure that is usually generated by
mechanical pumps. In microchannels however it becomes increasingly
difficult to utilize pressure-driven flow mode as the channel size shrinks,
especially down to micro-and submicron range. Moreover, some parts like
microvalves and micropumps with moving components are difficult to
fabricate, and they are prone to mechanical failure due to fatigue and
fabrication defects. Alternatively, electrokinetic flow enjoys numerous
advantages (over pressure-driven flow), including ease of fabrication and
control, no need for moving parts, high reliability, no noise etc. Specifically, a
plug-like velocity profile in electrokinetic flow can result in reduced
dispersion of sample species, making capillary electrophoresis become one of
most successful technologies for chemical and biomedical analyses. Most of
existing studies regarding electrokinetics flows focus on Newtonian fluids.
However, microfluidic devices are usually used to analyze biofluids which
may not be treated as Newtonian fluids. Thus, the more general Cauchy
momentum equation, instead of the Navier-Stokes equation should be used to
describe the flow characteristics of non-Newtonian fluids.
Preface
ix
Ivan A. Kuznetsov
time. We describe here two different new approaches: a hydrogel valve, which
is robust, and a rotary valve, which is versatile. Both valves are pressurestable. The hydrogel valve contains a microfluidic chamber filled with
hydrogel particles that are swollen and thus close the valve at room
temperature (i.e. normally closed 2/2-way valve). Upon heating, the hydrogel
becomes dehydrated and opens reversibly in two to four seconds. This valve is
robust, autoclavable, and can retain cells without stressing them. The rotary
valve contains a valve seal cast from poly(dimethylsiloxane) (PDMS) with one
or more micromolded channels of extremely low dead volume; it can be
turned, in just 250 milliseconds, to a new position by air pressure or a
micromotor, thereby connecting new inlet and outlet holes. Many different
setups of a multiple-way selector/injector valves can be realized, and so it is
probably the most versatile and quick microfluidic valve available today. To
reduce swelling and abrasion and to allow thousands of cycles, the PDMS
rotor is metal-coated. These two valves represent attractive new ways to
control flows in microfluidic applications.
Chapter 5 - Localized surface plasmon resonance (LSPR) is the interaction
between the light and collective electrons of noble metal nanoparticles, which
exhibits as an extinction peak shift in the transmission spectrum of the
nanoparticles. LSPR is utilized to detect label free chemical or biological
samples, such as finding the kinetic coefficients between two kinds of
chemicals on the surface. To incorporate the noble metal nanoparticle chip into
a compact MEMS device will minimize the expensive biological samples
required, and will be able to make the LSPR biosensors into a large sensing
array for quick detections. The desired LSPR MEMS device should possess
the characteristics of nanofabrication compatible, biocompatible, highly
transparent, cost effective and mass productive.
This chapter introduces the design, fabrication and test of such a LSPR
MEMS device. Our MEMS device is based on a glass-silicon-glass sandwich
structure, it is biocompatible due to the stable nature of the glass and silicon
compared with other polymer based disposable MEMS devices. On the bottom
glass, gold nanostructures for LSPR are fabricated by nanosphere lithography.
The middle silicon layer forms the microfluidics channel and chamber of the
device, and also blocks the light from shedding onto the non-sensing areas.
The top glass is drilled with inlet and outlet for microfluidics. The three layers
are bonded together by UV cure epoxy. As the whole fabrication process
avoids the high temperature, high pressure and high stress, it is
nanofabrication compatible, that the gold nanostructures are intact even when
after the MEMS device is pried apart. The device is highly transparent with
Preface
xi
high signal-to-noise, because the glass surface was not damaged during the
process, and the light blockage of silicon greatly reduced the background of
the light. As the MEMS fabrication for theses three layers are prepared in
wafer level prior to epoxy bonding, the device is suitable for mass production
with low cost. The fabricated LSPR MEMS device is tested by UV-Vis
spectroscopy with bovine serum albumin sensing, and it demonstrates
satisfactory LSPR sensing function. Our experiments prove that such a design
has prominent advantages over the one without light blockage or with
polydimethylsiloxane (PDMS) polymer.
Chapter 6 - We already showed in Section V that linear interface
deformations could be used for soft lensing with large variations in accessible
focal distances. The nonlinear regime in deformation, particularly optically
driven liquid jetting, offers an even wider range of application since
hydrodynamics starts to couple with light propagation. One major point here is
that, contrary to electro-hydrodynamics where micrometric features are
difficult to implement, these are natural length scales in "optohydrodynamics".
Chapter 1
MICROBIOCHIPS MONOLITHICALLY
INTEGRATED WITH MICROFLUIDICS,
MICROMECHANICS, PHOTONICS, AND
ELECTRONICS BY 3D FEMTOSECOND LASER
DIRECT WRITING
Ya Cheng1, Koji Sugioka2, Katsumi Midorikawa2, and
Zhizhan Xu1
1
ABSTRACT
Microbiochips such as lab-on-a-chip (LOC) devices and micro total
analysis systems (-TAS) can be regarded conceptually as a biological
equivalent of conventional silicon integrated circuits, which involve
miniaturization and integration of electronics. As its name implies, an
ideal LOC should have a very small footprint and yet still be capable of
functioning as a laboratory in which partial or complete chemical or
biological analysis can be performed automatically. A planar silicon chip
that integrates only microelectronics is clearly far from adequate for such
Keywords: femtosecond laser direct writing, 3D microfabrication, lab-on-achip, microfluidics, micromechanics, photonics, electronics, integrated
microchip.
1. INTRODUCTION
The field of lab-on-a-chip (LOC) technology has experienced tremendous
growth over the last few years. A LOC device is a miniaturized system that
integrates one or more laboratory functions for chemical and biological
analyses. Because of its compact dimensions and high functionality, a LOC
device allows chemical and biological analyses to be performed easily with
low sample and reagent consumptions, low waste production, rapid analysis,
and high reproducibility due to standardization and automation [1,2]. Although
LOC technology has been used for a broad range of applications in a variety of
fields including medicine, healthcare, pharmaceutics, environmental
monitoring, and national security, it is not yet mature and it is still under active
development. One major problem is the lack of appropriate technology for
and sold under the trade name Foturan [8,12]. The history of photosensitive
glass and its fabrication commenced half a century ago. Stooky at Corning
conducted the earliest work in developing this material in the 1950s [13]. The
photosensitive glass Foturan is lithium aluminosilicate doped with trace
amounts of silver and cerium. The cerium (Ce3+) ion functions as a
photosensitizer by absorbing a UV photon at a wavelength near 315 nm and
releasing an electron to become Ce4+. Some silver ions then capture some of
the free electrons and become silver atoms. In a subsequent heat treatment, the
silver atoms diffuse and agglomerate to form nanoclusters at about 500 C,
and then at about 600 C the crystalline phase of lithium metasilicate grows
around the silver clusters (which act as nuclei) in the amorphous glass matrix.
This crystalline phase of lithium metasilicate can be preferentially etched since
it has a much higher etching rate in dilute hydrofluoric acid than the glass
matrix. The conventional exposure procedure for 2D microfabrication on the
surface of Foturan glass is a UV lamp photography step. However, to fabricate
3D hollow microstructures in Foturan, it is necessary to focus the laser beam
into the sample. Since UV exposure is essentially a single-photon process, the
UV beam undergoes linear absorption as it propagates from the glass surface,
making internal modification very difficult. In addition, single-photon
processes intrinsically have poor fabrication resolutions in the laser beam
propagation direction. In contrast, an exposure procedure employing a
femtosecond laser can confine the absorption region to an area near the focal
spot by using a multiphoton process, improving the axial resolution, which is
vital for achieving true 3D microprocessing [14,15]. Our early investigations
revealed the photoreaction initiated by multiphoton excitation has a different
mechanism from that of single-photon excitation, because the interaction of
the high-intensity femtosecond laser with the glass matrix of Foturan generates
a large amount of free electrons. Consequently, it is not necessary to dope
Foturan glass with cerium for multiphoton processing [16].
In addition, using Foturan glass permits both the chemical etching rate and
the optical properties to be controlled. The hollow structures formed inside
Foturan usually have rough surfaces, but they can be greatly smoothened for
optical components (e.g., micromirrors and microoptical lenses) by applying a
postannealing process [17,18]. Furthermore, since the silver nanoparticles
produced by femtosecond laser irradiation exhibit the plasmonic effect, they
can be used to tune both the absorption and refractive index of Foturan
[19,20]. Thus, Foturan glass is an ideal material for fabricating optofluidic
devices, which integrate microfluidics and microoptics in a common substrate
[21,22].
10
Figure 5. (a) Close-up view of a focusing system with a slit for controlling the
microfluidic channel aspect ratio. (b) Side view and cross-section of a bridge-like
microstructure fabricated without a slit. (c) Side view and cross-section of the same
structure fabricated with a 0.5-mm-wide slit.
11
Although using a slit to shape the beam can achieve symmetrical crosssections when fabricating microfluidic channels, it is not the only solution to
the problem. In fact, after being shaped by the narrow slit, the focal spot will
not be truly spherical in 3D space, but instead it be compressed in the direction
parallel to the slit. Compression of the focal spot in the direction parallel to the
slit should not be a problem when fabricating 1D microstructures, such as
microfluidic channels or optical waveguides, because it can be compensated
by reducing the laser scanning speed in the sample; however, an anisotropic
focal spot is undesirable in many applications (e.g., 3D data storage in glass
using femtosecond laser generated microvoids). In such cases, a threedimensionally isotropic focal spot is preferable. This can be achieved by using
a novel focusing geometry in which two focused femtosecond laser beams are
perpendicular to each other [29].
Figure 6(a) shows a schematic view of the crossed-beam irradiation
system; the combined optical fields of the two beams results in an isotropic
energy distribution in the central region of the focal spot (the area enclosed by
the dashed red circle in Figure 6(b)). To ensure an isotropic illumination
volume, the foci of the two beams in the crossed-beam system must
spatiotemporally overlap throughout the entire period of laser scanning. This is
impossible to fulfill if we scan the glass in air due to the mismatch in the
refractive indices of Foturan glass (n~1.52) and air (n~1). To overcome this
problem, we designed a special XYZ stage which translates the glass sample
in a refractive index matching liquid, as shown in Figure 6(a). The liquid is a
mixture of -bromnaphtalene (n~1.66) and paraffin (n~1.48); thus, an
identical refractive index to that of Foturan glass can be achieved by mixing
the two liquids in the appropriate ratio. Since translating the glass sample in a
fixed glass cell containing the refractive index matching liquid will not alter
the optical paths of the two orthogonal beams, two foci that are initially
aligned will continue to spatiotemporally overlap each other over the entire
scanning step. Before we started to scan the sample, the two objective lenses
were aligned by observing femtosecond-laser-induced fluorescence in the
glass sample. That is, when two high-intensity femtosecond laser beams were
simultaneously sent into the two objective lenses with a crossed-beam
focusing system, two lines of fluorescence were simultaneously observed on
the PC monitor connected to a CCD camera. The CCD camera was installed
with its optical axis perpendicular to the plane defined by the two crossed
beams. We carefully adjusted the positions of the objective lenses using a
computer-controlled translation stage to make the centers of the two
fluorescence lines spatially overlap. We then adjusted the time delay between
12
the two beams by varying the optical path length using another translation
stage until the fluorescence in the overlapping region of the two fluorescence
lines was maximized. When this alignment procedure was completed, the
crossed-beam irradiation system was able to create microstructures inside
glass with isotropic 3D resolution.
Figure 6. (a) Schematic view of crossed-beam irradiation system. (b) The energy
distribution in the focal spot synthesized by the two perpendicular beams.
13
14
3.2. Micromechanics
Micromechanical structures such as micropumps and active microvalves
are important elements for realizing microfluidic routing and switching in
LOC applications [14,31,32]. The 3D femtosecond laser direct writing
technique allows direct fabrication of mechanical elements encapsulated in a
microfluidic cavity. As an example, we first demonstrated the fabrication of a
freely movable microplate in a microfluidic chamber that serves as a
microvalve (see Figure 8) [14].
Figure 8 shows the exposure scheme in Foturan glass used to fabricate the
freely movable microplate. After scanning the femtosecond laser beam at a
velocity of 2 mm/s and a laser fluence of 170 mJ/cm2 in the dark regions in
Figure 8(a), the above-mentioned postannealing was performed to form
modified regions of lithium metasilicate crystallites. To obtain a high spatial
resolution, the irradiation conditions were set slightly above the threshold
fluence above which the photosensitive glass was modified but below which it
was not modified at all [14]. The modified regions were completely removed
in the subsequent chemical wet etching so that hollow structures (indicated by
the white regions in Figure 8(b)) were formed in the glass. A movable glass
plate was left in the hollow structure, which can serve as a microfluidic
chamber. The fabricated movable microplate can function as a microvalve in
the microfluidic network, as illustrated in Figure 9. This device was
manufactured by stacking three Foturan glass substrates, each with a thickness
of 2 mm (the thickest substrate available). A syringe was used as an air
compressor to drive the microplate motion. The top layer contains three inlet
cells; the center cell is an opening for adding the reagents while the left and
right cells are openings attached to silicon tubes for infusing compressed air
from the syringes. In the middle layer, a movable microplate was embedded in
a rectangular hollow chamber connected to five microchannels to the cells in
the top and bottom plates. The microplate was fabricated by the exposure
scheme shown in Figure 8. In the bottom layer, two cells were installed in the
glass as outlets for the reagents. Each structure was fabricated by the same
procedure using the femtosecond laser. The microplate was driven to the right
when compressed air was infused from the left opening in the top layer. In this
case, the flow channel of the reagents to the right outlet was switched off so
that the reagents could flow only to the left outlet (Figure 9(a)). When the
compressed air infusion was changed to the right opening, the microplate was
driven to the left. As a result, the reagent flow was switched to the right outlet
15
(Figure 9(b)). Thus, this microplate can switch the flow direction of the
reagent like a microvalve.
16
3.3. Microoptics
Since chemical and biological analyses frequently employ optical
detection methods, integration of microoptics into LOC chips has advantages
such as low cost, robustness, stability, and ease of operation [33]. The
microstructuring of Foturan glass by a femtosecond laser is essentially nonablative processing that results in smooth and debris-free internal surfaces;
thus, both 3D microfluidic components and 3D microoptics can be
simultaneously fabricated inside the glass. For example, Figure 11(a) shows a
45 micromirror embedded in Foturan glass [17]. To fabricate this structure,
we scanned parallel lines layer by layer from the top surface to the bottom of
the sample. The interval between two adjacent lines in the Foturan glass was
set at ~15 m. Thus, since the total sample thickness was 2000 m, we
17
18
Figure 11. (a) Top view of the 3D micromirror fabricated inside Foturan glass with its
optical path indicated by arrows. Beam spots reflected from the sample (b) before and
(c) after the additional annealing.
19
temperature of this additional annealing was lower than that used to crystallize
lithium metasilicate. In this annealing process, the sample was heated to 570
C at 5 C/min, it was held at this temperature for 5 h, and then cooled to 370
C at 1 C/min. After the sample cooled to room temperature, we reexamined
its optical properties. The results are shown in Figure 11(c). The reflected
beam spot is clearly much smaller than that from the sample that was not
subjected to the additional annealing, which demonstrates that the divergence
of reflected beam has been reduced. Figures 12(a) and (b) show the
morphologies of etched surfaces that had not and had been subjected to the
additional annealing, respectively. Figure 12(b) reveals that after the additional
annealing, the smoothness of surface is comparable to that of a polished glass
surface (although there are still a few irregular nanodots on the surface). The
average surface roughness before the additional annealing was measured to be
~81 nm, whereas after the annealing it was only ~0.8 nm. Consequently, both
the divergence angle and the optical loss of the reflected beam can be reduced.
In addition to microoptical structures with plane surfaces, microlenses
with curved surfaces can be produced using this technique [18]. Microlenses
are important elements for optical biosensors, and they are used as collimators,
focusers, and imaging elements [34]. Figures 13(a) and (b) show respectively a
microoptical cylindrical lens and a hemispherical lens fabricated on Foturan
glass by 3D femtosecond laser direct writing; the focal spots produced by
these lenses are shown on the right. These microlenses were fabricated by
gouging them out from the Foturan matrix, but they can be also formed inside
the glass chip by the present technique for fabricating hollow microstructures
[35]. The fabricated hollow structures have openings at one or both ends in a
Foturan glass chip and one of the internal sidewalls of the hollow structure is
spherical in shape, (i.e., it has a plano-concave hollow structure), as shown in
Figure 14. The curvature of the spherical surface corresponds exactly to the
designed optical microlens curvature so that the curved surface can function as
a plano-convex lens. Thus, a microoptical lens can be fabricated inside a glass
chip. Figure 14 also shows optical microscope images of a plano-convex
microlens fabricated inside a glass chip with a thickness of 2 mm, a radius of
curvature R of 0.75 mm (designed value), and a focal length of 1.5 mm
(calculated using the formula: 1/f = (n1)(1/R), n=1.5 (refractive index of
photosensitive glass)). The figure shows that one of the sidewalls of the buried
hollow structure is spherical in shape with a smooth surface, although a small
bump is visible on the lens surface. To examine the focusing performance of
the microlens, we irradiated a HeNe laser beam such that it passed through
the microlens inside the glass, and the focused laser beam was projected onto a
20
CCD camera by a 20 objective lens with an NA of 0.35. The focal spot size
was approximately 30 m in diameter for an incident HeNe laser beam of
1.51.8 mm. Furthermore, we roughly estimated the focal length
experimentally and found that it was about 1.7 mm. This focal length is
slightly longer than the designed value of 1.5 mm. Reflection at the airglass
interface is probably responsible for the longer focal length; in other words,
the laser beam is reflected by the flat internal wall of the hollow structure after
the lens. Another cause for the longer focal length may be somewhat imperfect
fabrication precision. The efficiency of the fabricated microlens was evaluated
to be more than 80%. The optical loss is considered to be mainly due to four
Fresnel reflections at the airglass interfaces (two glass chip sidewalls and two
lens surfaces). Subtracting the loss of the four Fresnel reflections (4% for
each), the net loss is estimated to be ~6%. However, the focusing results and
the efficiency evaluation indicate that the smooth spherical surface of the
buried hollow structure is suitable for optical applications.
By using 3D femtosecond laser direct writing, microoptical and photonic
components have now been fabricated in various transparent materials by
modifying their refractive indices [6,20,36]. Changes in the refractive index
can be induced in Foturan glass by femtosecond laser irradiation followed by
annealing at 520 C [20]. This results in silver nanoparticles being selectively
precipitated in the irradiated region, which changes the refractive index by as
much as ~0.4%. Since the process involves only a photochemical reaction, the
pulse energy for such processing was as low as 70 nJ/pulse for an objective
lens with an NA of 0.8. Such a low pulse energy induced little thermal effects,
resulting in a high spatial resolution. Figure 15 shows a 5-m-pitch
micrograting together with its diffraction pattern obtained using a HeNe laser
beam. Using the same principle, optical waveguides can be fabricated inside
Foturan glass; guided beams in these waveguides have a single-mode profile
because the beam was shaped using a slit during waveguide writing [28]. Very
recently, a volume optical grating has been formed inside Foturan glass by
focusing a femtosecond laser beam using a pair of cylindrical lenses [37].
Each cylindrical lens focuses the beam in one dimension, so that each
femtosecond laser beam pulse forms a vertical plane at the focus. This enables
the volume grating to be fabricated by simply translating the sample
perpendicular to the cylindrical lens at a constant speed while the laser beam is
periodically switched on and off using a programmable shutter. This technique
significantly reduces the time required to fabricate a volume grating because
21
there is no need to scan the laser beam along the direction parallel to the
cylindrical lens.
Figure 12. AFM images of the surface of Foturan glass after the irradiation by the
femtosecond laser beam and chemical etching (a) before and (b) after the additional
annealing.
22
Figure 13. SEM images of microoptical (a) cylindrical and (b) hemispherical lenses.
The focal spots produced both lenses are presented in the right panel.
23
Figure 15. A 5-m-pitch microoptical grating (upper) embedded in Foturan glass and
its diffraction pattern (lower).
24
25
Figure 16. Dependence of optical loss on the length of waveguides written at different
laser energies and scanning speeds. Linear fit equations give the propagation loss in
dB/cm (slope) and the coupling loss in dB (y-axis intercept).
26
3.4. Microelectronics
Fabrication of microelectronic components by femtosecond laser direct
writing is certainly desirable for developing monolithic and compact LOC
systems [4045]. Selective metallization of the surfaces of some glass
materials (e.g., Foturan and microscope slide glass) has been achieved by
direct femtosecond laser ablation followed by electroless copper plating, but
this technique cannot be applied to other glass materials such as fused silica
[40]. The process requires a certain degree of surface roughness for the socalled anchor effect to operate, which enables the deposited metal film to
adhesion strongly to the glass surface. Figure 18 shows optical microscope
images of fine metal lines deposited on a Foturan surface after electroless
plating. The laser power was varied from 1.2 to 3 mW and the laser scanning
speed was 50 m/s. As an application of this selective metallization technique,
Figure 19 shows a microheater that was fabricated on a Foturan glass chip.
The microheater temperature is a linear function of the electric power applied
to it [40].
27
Figure 18. Optical microscope images of (a) as ablated and (b) copper-plated glass
after ablation. The laser power is varied from 1.2 to 3 mW. The laser scanning speed is
50 m/s.
28
Figure 19. (a) Schematic diagram of a microheater; (b) microscope image of the
microheater; (c) enlarged image of the left-hand end of the microheater; (d)
experimental setup for measuring the temperature of the microheater.
Figure 20. Schematic illustration of the fabrication process for the selective
metallization of insulators: (1) formation of silver nitrate thin films on insulator
substrates; (2) modification of insulator surfaces by femtosecond laser direct writing;
(3) removal of unirradiated silver nitrate films by acetone; (4) copper coating by
selective electroless plating.
29
30
31
bounces back and forth in the optical cavity by total internal reflection. Lasing
can occur if the microchamber is filled with a gain medium laser dye
rhodamine 6G (Rh6G) and then pumped by a frequency-doubled Nd:yttrium
aluminum garnet (Nd:YAG) laser. Since the surfaces of the micromirrors
could not be fabricated perfectly smooth, and also the angle between any two
micromirrors could not be exactly 90 due to the limited fabrication precision,
a small amount of light will eventually leak out of the optical cavity and be
emitted tangentially from the internal surfaces of the micromirrors. Light
emission from the side of the cavity has been frequently observed in many
experiments with microcavity lasers [50,51].
Figure 22. (a) Optical micrographs of the top-view of the microfluidic laser and (b) of
the side view of the microfluidic chamber and through channel. (c) The light path in
the microfluidic laser.
32
repetition rate of 15 Hz. When the pumping power was increased above the
lasing threshold, light emission could be clearly observed from the output end
of the dye microlaser, as shown in Figure 23(a). We then placed a receiving
screen approximately 8 mm from the output end of the structure to take a
photograph of the far-field pattern of the emission, as shown in Figure 23(b).
Two laser beams emitted tangentially from the internal surfaces of the two 45
mirrors were simultaneously observed on the screen, with one propagating
upward and the other downward. The laser beams were well confined in the
direction perpendicular to the plane of the optical cavity.
Figure 23. Digital camera images of (a) laser emission from glass sample and (b) farfield pattern of the laser beam on a receiving screen.
33
Figure 24. Optical micrographs of (a) top view of the dual-wavelength microfluidic
laser and of (b) the side view of the two microfluidic chambers serially embedded in
glass with a center-to-center distance of ~1000 mm.
34
Figure 25. The spectrum of the dual-wavelength microfluidic laser. The peaks centered
at 532, 568, and 618 nm correspond to scattered light from the pump laser, the Rh6G
dye laser, and the rhodamine 640 dye laser, respectively.
35
(a)
(b)
Figure 26. Schematic configuration of optofluidics in which microoptics, such as
microoptical plano-convex lenses and an optical waveguide, are integrated with a
microfluidic chamber in a single glass chip. An optical microscope image of the top
view of the fabricated microchip is shown in the upper left corner.
36
In the above examples, the waveguides have to be written into the Foturan
glass after fabricating micromirrors and microlenses, because the refractive
index changes induced by femtosecond laser irradiation cannot withstand the
high temperature applied to the Foturan sample during postannealing. This
problem can easily be resolved by fabricating freestanding optical fibers rather
than buried optical waveguides [57].
A freestanding fiber was fabricated by scanning the area surrounding the
fiber in a glass chip and then baking and etching the glass sample as described
above. Figure 27(a) shows a structure composed of a freestanding fiber
integrated with a 45 micromirror at the entrance of the fiber in the glass. The
arrows in Figure 27(a) indicate the optical path of the coupling scheme. The
45 micromirror allows us to couple the light into the fiber from the side of the
sample. Figure 27(b) shows the optical micrograph of the micromirror and
entrance of the fiber. The inset of Figure 27(b) (upper right corner) shows the
cross-sectional shape of the fabricated fiber, which has approximate
dimensions of 100 m (width) 80 m (height). Light was coupled into the
fiber by focusing a HeNe laser beam on the micromirror using an objective
lens with an NA of 0.46, as shown in Figure 27(c). The guided light was
clearly observed at exit of the fiber. The total length of the fiber in Figure
27(c) is 8 mm, which is sufficiently long for many microchip applications.
Fabricating longer fibers is not technically challenging, but it requires longer
fabrication times. The measured loss for the freestanding fibers was
approximately 1.3 dB/cm.
The freestanding fibers were then incorporated into a microfluidic circuit
for on-chip biophotonic applications. Figure 28(a) illustrates a 3D schematic
view of the integrated structure, which is a biosensor composed of two series
of freestanding fibers intercepted by a microwell fabricated on a glass chip.
Figure 28(b) shows an optical micrograph of part of the fabricated
microstructure, showing the fibers and the microwell on the glass surface. To
demonstrate that the light exiting from the first fiber can be effectively coupled
into the second fiber, we focused the HeNe laser beam onto the entrance
facet of the first fiber using a 20 objective lens. As shown in Figure 28(c),
both scattering light at the microwell and the exiting light at the end of the
second fiber can clearly be observed. The coupling loss between the two fibers
intercepted by the microwell was approximately 1 dB. We ascribe such a low
coupling loss between the two fibers to two main factors: (1) the relatively
large diameter of the fiber permits light beams with large mode areas to
propagate in the fiber, thereby reducing the divergence angle of the exiting
37
light, and (2) the inner walls of the microchannel between the two fibers are
smooth and fabricated nearly perpendicular to the fibers, preventing the
wavefront for the beam exiting the first fiber and entering the second fiber
from tilting.
A freestanding fiber fabricated on a Foturan glass chip has many potential
applications besides functioning as a waveguiding element. For example, the
fiber could be fabricated inside a microfluidic channel and immersed in the
sample solution. By measuring the optical loss of the beam propagating in the
fiber, high-sensitivity photoabsorption measurements could be conducted.
Moreover, because of the large refractive index difference between the
freestanding fiber and the surrounding air, curved fibers can be fabricated with
tight bends without significant optical losses. Recently, it has been pointed out
that the curved freestanding fibers may find applications in astrophotonics
(e.g., 3D mode converters in giant telescopes) [58].
Figure 27. (a) 3D schematic drawing of a freestanding optical fiber integrated with a
micromirror fabricated on a glass chip. Red arrows indicate the optical path of the
coupling scheme. (b) Optical micrograph of the top view of the freestanding fiber and
the micromirror. The inset (upper right corner) shows the cross-section of the
fabricated fiber. (c) Digital-camera image of the side coupling of a HeNe laser beam
into the freestanding fiber through the micromirror.
38
Figure 28. (a) Schematic 3D diagram of two freestanding optical fibers intercepted by
a microwell fabricated on a glass chip. (b) Optical micrograph of the structure. (c)
Image of the guiding HeNe light through the entire structure. Scattering light at
microwell and the guided light at the exit facet of the second fiber are indicated by
arrows.
39
Figure 29. (a) Digital camera captured image of a 90 arc-shaped microchannel with a
radius of curvature of 5 mm fabricated on a glass plate. Inset: cross-sectional view of
the waveguide. (b) CCD camera image of the microfluidic optical waveguide carrying
a HeNe laser beam.
40
Figure 30. Near-field patterns of (a) multimode and (b) single-mode beams. (c) Beam
profile of the single-mode beam.
41
The device was examined using a 633-nm HeNe laser polarized in the
extraordinary (Z) direction by applying a varying dc voltage to the electrodes,
and we simultaneously captured images of the near-field mode at the output
42
end of the MZI using a CCD camera. The results are shown in Figure 32. The
measured extinction ratio was ~9.2 dB. A higher extinction ratio is expected if
the two arms of the MZI can be fabricated more symmetrically by improving
the translation precision of the XYZ stage. The voltage required to completely
switch the modulator on and off was measured to be ~19 V, indicating an
excellent EO overlap integral of ~0.95 [68].
43
44
45
Figure 34. Optical microscope images of (a) the top view and (b) the side view of the
microchip used for observing the motion of Euglena gracilis.
46
Figure 35. Microscope images of (a) encapsulated Euglena gracilis swimming in the
microchannel and (b) sequential pictures of advancing Euglena gracilis.
47
addition, water in the embedded channel does not evaporate or leak, unlike
when using a Petri dish, bonded glass, or polymer microchips. Consequently,
quantitative analysis of the flagellum movement could be easily carried out
using the fabricated microchip. We have also succeeded in fabricating some
other kinds of nanoaquariums with different structures and functionalities for
various applications, including determining the information transmission
process in Pleurosira laevis, observing the high-speed motion of
Cryptomonas, and attaching Phormidium to seedling roots to promote the
growth of Komatsuna.
Figure 36. (a) Sequential pictures of rotating Euglena gracilis stimulated by light and
(b) schematic illustration of the light stimulation of Euglena gracilis in the microchip.
48
Figure 37. (a) Schematic illustration of the scheme for observing Euglena gracilis from
the front and (b) microscopic image of the front view of Euglena gracilis swimming
upward in the reservoir.
49
50
REFERENCES
[1]
51
52
53
[52] Y. Cheng, K. Sugioka, and K. Midorikawa, Appl. Surf. Sci. 248, 172
(2005).
[53] Q. Kou, I. Yesilyurt, and Y. Chen, Appl. Phys. Lett. 88, 091101 (2005).
[54] D. Psaltis, S. R. Quake, C. H. Yang, Nature, 442, 381 (2006).
[55] Y. Cheng, K. Sugioka, and K. Midorikawa, Proc. SPIE, 5662, 209
(2004).
[56] Z. Wang, K. Sugioka, and K. Midorikawa, Appl. Phys. A, 93, 225
(2008).
[57] Y. Cheng, K. Sugioka, K. Midorikawa, Opt. Express, 13, 7225 (2005).
[58] R. R. Thomson, A. K. Kar, and J. Allington-Smith, .Opt. Express, 17,
1963 (2009).
[59] H. Sun, F. He, Z. Zhou, Y. Cheng, Z. Xu, K. Sugioka, and K.
Midorikawa, Opt. Lett. 32, 1536 (2007).
[60] D. B. Wolfe, R. S. Conroy, P. Garstecki, B. T. Mayers, M. A. Fischbach,
K. E. Paul, M. Prentiss, and G. M. Whitesides, Proc. Natl. Acad. Sci.
U.S.A. 101, 12434 (2004).
[61] P. Dumais, C. L. Callender, J. P. Noad, and C. J. Ledderhof, IEEE
Photon. Technol. Lett. 18, 746 (2006).
[62] M. Brown, T. Vestad, J. Oakey, and D. W. M. Marret, Appl. Phys. Lett.
88, 134109 (2006).
[63] R. Osellame, V. Maselli, R. M. Vazquez, R. Ramponi, and G. Cerullo,
Appl. Phys. Lett. 90, 231118 (2007)
[64] V. Maselli, J. R. Grenier, S. Ho, and P. R. Herman, Opt. Express, 17,
11719 (2009).
[65] M. Kim, D. J. Hwang, H. Jeon, K. Hiromatsu, and C. P. Grigoropoulos,
Lab Chip, 9, 311 (2009).
[66] J. Burghoff, C. Grebing, S. Nolte, and A. Tnnermann, Appl. Phys. Lett.
89, 081108 (2006)
[67] J. Thomas, M. Heinrich, J. Burghoff, S. Nolte, A. Ancona, A.
Tnnermann, Appl. Phys. Lett. 91, 151108 (2007)
[68] L. N. Binh, J. Cryst. Growth, 288, 180 (2006).
[69] K. Okano, E. Hunter, and N. Fusetani, J. Exp. Zool. 276, 138 (1996).
[70] K. Yoshimura, C. Shingyoji, and K. Takahashi, Cell Motil.
Cytoskeleton, 36, 236 (1997).
[71] S.L. Fleming and C. L. Rieder, Cell Motil. Cytoskeleton 56, 141 (2003).
[72] D. J. Stephens, and V. J. Allan, Science, 300, 82 (2003)
[73] Yamahata, C. Vandevyver, F. Lacharme, P. Izewska, H. Vogel, R.
Freitag, and M.A.M. Gijs, Lab Chip, 5, 1083 (2005).
54
[74] Y. Choi, M.A. McClain, M.C. LaPlaca, A.B. Frazier, and M.G. Allen,
Biomed. Microdevices, 9, 7 (2007).
[75] S. Haeberle and R. Zengerle, Lab Chip. 7, 1094 (2007).
[76] Funfak, A. Brsing, M. Brand, and J. M. Khler, Lab Chip. 7, 1132
(2007).
[77] K. M. Nichols, and R. Rikmenspoel, J. Cell Sci. 23, 211 (1977).
[78] K. M. Nichols, and R. Rikmenspoel, J. Cell Sci. 29, 233 (1978).
Chapter 2
ABSTRACT
Advanced microfluidic devices can perform complete biochemical
analysis in a single fabricated chip. The generic microfluidic systems
involve buffer solutions and samples manipulations such as pumping,
valving, mixing, injection, dispensing, concentration etc. Fundamental
understanding of the liquid flow characteristics in microchannels is thus
essential to optimum design and precise control of microfluidic devices.
In general, liquid motion can be generated by either applying a pressure
gradient or imposing an electric field, leading to pressure-driven flow or
electrokinetically-driven flow, respectively. Traditionally, in large-sized
channels flow is often driven by pressure that is usually generated by
mechanical pumps. In microchannels however it becomes increasingly
difficult to utilize pressure-driven flow mode as the channel size shrinks,
* To whom the correspondence should be addressed.
Tel: (+65) 6790-4883; Fax: (65) 6792-4062
E-mail: mcyang@ntu.edu.sg.
56
57
58
microchannels with relatively thin EDL thickness, the flow field outside the
EDL is an irrotational flow with a slip velocity boundary condition determined
by the well-known Smoluchowski equation. Kang et al. [11] presented an
analytical scheme to solve the PoissonBoltzmann equation for arbitrary zetapotentials, and analyzed the dynamic electroosmotic flow in a cylindrical
capillary. In Xuan and Lis work [12], the electroosmotic flow was analyzed
for microchannels with arbitrary cross-section and heterogeneous potential.
Yan et al. [13] presented a model to analyze the finite reservoir effects on EOF
in a rectangular microchannel, and performed a micro-PIV experiment to
validate the proposed model. Other than analytical and experimental
investigations, numerical simulations of the electroosmotic flow in complex
geometry of microchannel networks have been reported [14-18].
All the aforementioned studies are concerned with Newtonian fluids.
However, microfluidic devices are usually used to analyze biofluids which
may not be treated as Newtonian fluids. Thus, the more general Cauchy
momentum equation, instead of the Navier-Stokes equation should be used to
describe the flow characteristics of non-Newtonian fluids provided that proper
constitutive equations are available. The aim of constructing constitutive
equations for non-Newtonian fluids is to find correlations between dynamic
viscosity and shear rate. The Power-law model [19], Carreau model [20],
Moldflow first-order model [21], and Bingham model [22] have been
successfully developed to analyze non-Newtonian fluid flow and heat/mass
transfer. As for electroosmotic flow of non-Newtonian fluids, however there
are few studies reported in the literature. Zimmerman et al. [20] presented a
two-dimensional finite element simulation of electrokinetic flow in a
microchannel T-junction for fluid with a Carreau-type nonlinear viscosity.
They claimed that the fluid experiences a range of shear rate as it turns around
the corner, and the flow field is shown to be sensitive to the non-Newtonian
characteristics of the Carreau-type. Otevel and Kleprnk [23] derived the
exact solutions for electroosmotic flow in a capillary channel filled with
polymer electrolytes having a varying viscosity, and the steady-state velocity
profile was calculated by assuming that the viscosity decreases exponentially
in the radial direction (i.e., from the capillary wall to the centerline). To our
best knowledge, the only relevant study that we could locate is by Das and
Chakraborty [24] who analyzed the electroosmotic flow of a Non-Newtonian
fluid in microchannels. However, Das and Chakrabortys analyses were based
on an approximation of the hyperbolic sine function, and such approximation
can lead to large errors when the electrokinetic parameter H is small, e.g.,
the case of overlapped double layers. Moreover, their velocity derivation
59
( : )
2
1/ 2
(1)
where is the rate of strain tensor and is its magnitude. The fluid viscosity
then can be expressed as a function of , namely
= m ( 2 )
n 1
(2)
60
where m is the flow consistency index, and n is the flow behavior index which
represents an apparent or effective viscosity being a function of the shear rate.
Shear-thinning (also termed as pseudoplastic) behavior is obtained for n < 1,
and it indicates that the fluid viscosity decreases with increasing the rate of
shear. Pseudoplasticity can be demonstrated by shaking a bottle of ketchup
such that the ketchup undergoes an unpredictable change in its viscosity.
Newtonian behavior is obtained for n = 1 . Shear-thickening (also termed as
dilatant) behavior is obtained for n > 1 , and it shows that the fluid viscosity
increases with the rate of shear. The dilatant effect can readily be seen with a
mixture of cornstarch and water, which acts in counter-intuitive ways when
struck or thrown against a surface.
The flow field of the power-law fluids is governed by the continuity
equation and the Cauthy momentum equation. For an incompressible fluid, the
continuity equation can be written as [25]
v = 0
(3)
(4)
One can readily show that the Cauchy monmentum equation is expressed
as [25]
Dv
T
= P + F + ( ) v + ( v )
Dt
(5)
gradient tensor.
61
Fx = E0 e
(6)
62
d 2
= e
dy 2
(7)
d 2
= 2
dy 2
(8)
y=H
d
dy
= w
(9a)
=0
(9b)
y =0
where n and z are the bulk number concentration and the valence of ions,
respectively, e is the fundamental charge, kB is the Boltzmann constant, and T
is the absolute temperature.
The solution for the electrical potential distribution is of the following
form
( y) = w
cosh ( y )
cosh ( H )
(10)
63
Then the net charge density e can be expressed as a function of the EDL
potential,
e ( y ) = 2
(11)
Recalling that the magnitude of the rate of strain tensor defined in Eq. (1),
in this case is given by = (1/ 2 ) dvx / dy , the viscosity defined in Eq. (2)
can be expressed in terms of the velocity gradient as follows
= m ( 2 )
n 1
dv
=m x
dy
n 1
dv
= m x
dy
n 1
(12)
Here the negative sign is chosen because the velocity decreases with
increasing y.
Therefore, one can show that the Cauchy momentum equation (i.e., Eq.
(5)) can be considerably simplified to
d dvx
m
dy dy
n 1
dvx
2 E0 = 0
dy
(13)
vx
y=H
dvx
dy
=0
=0
(14a)
(14b)
y =0
64
n
2 E0
d dvx
=
dy dy
m
(15)
dvx
E0 w sinh( y )
=
m cosh ( H )
dy
(16)
yx = E0 w
sinh( y )
cosh ( H )
(17)
dvx
E0 w n
=
dy
m
sinh( y ) n
cosh ( H )
(18)
Substituting Eq. (18) into Eq. (12) gives the expression for the viscosity of
the power-law fluids,
dvx
dy
= m
n 1
=m
1
n
( E0 w )
n 1
n
sinh( y )
cosh ( H )
n 1
n
(19)
Introducing
ws = E w
1
n
ws = m ( E0 w )
(20a)
n 1
n
(20b)
65
which under the liming case of H are respectively the shear stress
and the dynamic viscosity on the channel wall according to Eqs.(17) and (19).
Therefore, the shear stress distribution yx and the viscosity of the power-law
fluids
yx = ws
sinh( y )
cosh ( H )
sinh( y )
cosh ( H )
= ws
(21)
n 1
n
(22)
vx ( y ) =
1 n
n
w E0
1
n
( )
1
n
(23)
( H )
1
etc.
3
The case of n = 1 corresponds to a Newtonian fluid where
1
2
= m , and
vx ( y ) =
cosh ( y )
1
cosh ( H )
w E0
m
(24)
66
which is the same as the velocity distribution shown in [4] using the
Newtonian fluids hypothesis. When n = 1/ 2 , one can have that
cosh ( H )
1
and show that
E0 w sinh( y )
= m2
(25)
cosh 2 ( H )
1
1
3
Similarly, for n = , one can get = m
2
3
E0 w sinh ( y )
2
sinh ( y ) 1 y
2 e
0 < y 1
y >1
(27)
67
Case I, H > 1
(a) In this case, if 0 y 1 , Eq. (23) can be integrated piecewise with
the above approximation adopted, and thus an analytical expression
for the velocity distribution can be obtained as,
1
1
1
H 1 ( y ' ) n
' n
'
y d y + e d y'
1 ky
1
1 n
2
E n
vx ( y ) = n w 0
1
m
cosh n ( H )
( ) ( )
( )
H
1
1
1+ n
1 n
n
n
y
e
e
1
( ) 1
1 n
n +1
n
n
E
2
(28)
= n n w 0
1
m
n
cosh ( H )
(b)
vx ( y ) =
1 n
n
w E0 n
1 ( y ' ) n
'
2 e d y
( )
cosh n ( H )
H
y
1 n
e e n
1
1
1 n
E n n
= n n w 0 2
1
m
cosh n ( H )
(29)
Average velocity can be obtained from Eqs. (28) and (29) using the
following definition
V=
1
H
vx dy
(30)
68
1
n
1
1
1
1 n
E n n
V = n n w 0 2
m
H
n
cosh
1
n 1 n
1
e +
H H (1 + 2n )
1
n
(31)
( H )
Q = 2V H = nk
1 n
n
Furthermore, if
w E0
m
1
n
n 1
n
H
1
n n n 1 n
2
e +
H e +
1
2n )
+
cosh
1
n
(32)
( H )
e kH n
cosh ( H )
2
1
n
(33)
V = Vs = n
1 n
n
w E0 n
(34)
Equation (34) shows that for Newtonian fluids the average velocity is linearly
proportional to external electric field strength and wall zeta potential, but
independent of EDL thickness. However, for the power-law fluids whose flow
behavior index is not equal to unit one, the dependence of the velocity on
external electric field strength, wall potential, and EDL thickness becomes
nonlinear. In addition, it is interesting to note that the Smoluchowski velocity
can be recovered from Eq. (34) when n=1. We thus can term Vs as the
generalized Smoluchowski velocity for power-law fluids. In Section 1.5,
without otherwise specified, the generalized Smoluchowski velocity, Vs , will
be used as a reference velocity to obtain dimensionless velocity. It is noted that
this generalized Smoluchowski velocity is only valide for solid surfaces with
small zeta potentials.In the more recent work [26], a more general
Smoluchowski velocity for arbitrary zeta potentials is presented.
69
vx ( y ) =
kH
1
1+ n
2n
n
n
n
ky
e
e
1
(
)
n +1
n
1
H
H
n
y
nH
e e n
n
1
H
H
n
0 ky 1
1
n 1 n
2n
e +
+
H H (1 + 2n )
(35)
ky > 1
2n
n 1 n
e +
+
H H (1 + 2n )
1
It should be pointed out here that although the approximation given in Eq.
(27) leads to discontinuity at y=1, the velocity distribution expressed in Eq.
(35) is continuous at y=1.
n ( y ) n
H
(
)
1 n
E n + 1
vx ( y ) = n n w 0
1
m
cosh n ( H )
1
n
(36)
Likewise, the average velocity, the volumetric flow rate, and the velocity
distribution can be obtained respectively as
V = nk
1 n
n
1+ n
1
H) n
(
w E0 1 + 2n
m
1
cosh n ( H )
1
n
(37)
70
1+ 2 n
2
nH n
1 n
n
+
1
2
w 0
n
Q = 2V H = nk
1
m
cosh n ( H )
(38)
1+ n
2n + 1 y n
vx ( y ) =
V 1
n +1 H
(39)
1
n
71
vx/Vs
0.8
H=10
0.6
H=50
H=2
H=3
0.4
0.2
0.0
0.0
exact solution
approximate solution
numerical solution
0.2
0.4
y/H
0.6
0.8
1.0
0.8
1.0
(A)
0.20
H=0.7
vx/Vs
0.16
H=0.5
0.12
0.08
0.04
0.00
0.0
exact solution
approximate solution
numerical solution
0.2
0.4
y/H
0.6
(B)
Figure 2. (A) Comparison of the exact solution, approximate and numerical solutions
for the velocity distributions of Newtonian fluids (i.e., n=1) when the electrokinetic
parameter H > 1 . (B) Comparison of the exact, approximate and numerical solutions
for the velocity distributions of Newtonian fluids (i.e., n=1) when the electrokinetic
parameter H < 1
72
yx/ws
0.8
0.6
H=0.7
H=1.0
H=3.0
H=10.0
0.4
0.2
0.0
0.0
0.2
0.4
y/H
0.6
0.8
1.0
Figure 3. Normalized shear stress distributions yx / ws (evaluated using Eq. (21) with
the shear stress on the channel wall given by Eq. (20a) as the reference shear stress) for
various values of H
73
n=0.8
n=0.9
n=1.1
n=1.2
16
/ws
14
12
10
8
6
n=1.0
4
2
0
0
10
Figure 4. Dimensionless dynamic viscosity / ws (calculated from Eq. (22) with the
dynamic viscosity on the channel wall given by Eq. (20b) as the reference viscosity)
for various values of the fluid behavior index, n while keeping H = 10
74
dilatant liquids, n > 1 , thinner EDL thickness or higher electric field strength
will result in higher wall dynamic viscosity. Whereas, for pseudoplastic
liquids, n < 1 , the dependence of the wall dynamic viscosity on the EDL
thickness and the external electric field is insignificant.
14
5
12
E0=2x10 V/m
1
-2
-3
-4
ws/0
=1nm (C=9.5x10 M)
10
=5nm (C=3.8x10 M)
=10nm (C=9.5x10 M)
6
4
2
0
0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
1.05
1.10
1.15
1.20
(A)
14
12
ws/0
10
-3
=8nm(C=1.5x10 M)
5
E0=1.0 10 V/m
5
E0=5.0 10 V/m
E0=1.0 10 V/m
4
2
0
0.80
0.85
0.90
0.95
1.00
(B)
Figure 5. (A) Effect of the EDL thickness on the dynamic viscosity at the channel wall,
ws (calculated using Eq. (20b)) normalized with the Newtonian fluid viscosity 0 . (B)
Effect of the applied electric field on the dynamic viscosity at the channel wall, ws
(calculated using Eq. (20b)) normalized with the Newtonian fluid viscosity 0
75
1.0
vx/Vs
0.8
n=1.5
n=1.2
n=1.0
n=0.8
n=0.5
0.6
0.4
0.2
0.0
0
10
76
1.2
1.0
vx/V
0.8
0.6
n=1.5
n=1.2
n=1.0
n=0.8
n=0.5
0.4
0.2
0.0
0
10
Figure 7. Dimensionless velocity distributions (evaluated from Eqs. (28) and (29))
normalized by the average velocity (given by Eq. (31)) for various values of the fluid
behavior index, n while keeping H = 10
Figure 8 depicts the ratio of the average velocity (calculated using Eq.
(31)) to the generalized Smoluchowski velocity versus the flow behavior
index, n for various values of H . The difference between the average
velocity and the Smoluchowski velocity reduces with increasing H or
decreasing the flow behavior index. As H , the ratio becomes unit one,
indicating that in this case the average velocity is the same as the
Smoluchowski velocity regardless of change of the flow behavior index.
Figure 9 shows the generalized Smoluchowski velocity Vs (for power-law
fluids) normalized by the conventional Smoluchowski velocity
( Vs 0 = E0 w / 0 for Newtonian fluids) versus the flow behavior index, n for
various EDL thicknesses. It is noted that all the three curves pass through a
unique point (1, 1). This indicates at such point, the Smoluchowski velocity is
independent of the EDL thickness, and is corresponding to the Newtonian
fluids. In general, the Smoluchowski velocity monotonically decreases when
increasing the flow behavior index. For pseudoplastic liquids, n < 1 , the effect
of the EDL thickness on the Smoluchowski velocity is more pronounced than
that for dilatant liquids, n > 1 . Interestingly, it is found that for small flow
behavior index, say n = 0.9 , the generalized Smoluchowski velocity can be
several times larger than the conventional Smoluchowski velocity. This
suggests a possible method of increasing the electrokinetic flow velocity in
microfluidic devices by adjusting the flow behavior index while keeping other
77
V/Vs
0.9
0.8
0.7
H=3
H=5
H=10
H=50
0.6
0.5
0.4
0.0
0.4
0.8
1.2
1.6
2.0
Figure 8. The ratio of the average velocity (calculated using Eq. (31)) to the
generalized Smoluchowski velocity (given by Eq. (34)) versus the flow behavior
index, n for various values of H
40
5
E0=2.0x10 V/m
Vs/Vs0
-2
-3
=1nm (C=9.5x10 M)
30
=5nm (C=3.8x10 M)
1
-4
=10nm (C=9.5x10 M)
20
10
0
0.80
0.85
0.90
0.95
1.00
1.05
1.10
1.15
1.20
Figure 10 shows the normalized volumetric flow rate versus the flow
behavior index, n, for various electric field strengths while keeping a constant
78
Q0 =
E0 w
2H
0
(40)
H=5m
1
-6
=250nm (C=1.5x10 M)
Q/Q0
24
E0=1.0x10 V/m
5
18
E0=5.0x10 V/m
6
E0=1.0x10 V/m
12
6
0
0.9
1.0
1.1
1.2
1.3
1.4
1.5
Figure 10. Dimensionless volumetric flow rate versus the flow behavior index, n, for
various electric field strengths while keeping a constant microchannel height of
H = 5 m . The volumetric flow rate Q is evaluated from Eq. (32) and the reference
flow rate is chosen as Q0 = 2 H E0 w / 0 .
79
80
systems for drug delivery, chemical analysis, and biomedical diagnosis [3].
Understanding of flow physics in microchannels is of great importance to the
successful and optimal design and precise control of microfluidic devices.
However, the existing theories cannot be scaled down to describe completely
the flow in microchannels, where some surface phenomena such as capillary,
wetting, electrokinetic effects, can cause the flow characteristics to deviate
from those in large-sized channels.
In the literature, numerous theoretical studies were reported to explain the
deviation of microscale flow characteristics; the micro-polar fluid theory [28],
the micro-moment theory [29], and the electrokinetics [5] are a few to name.
In this study, the electrokinetic effects are considered. It is known that most
solid surfaces acquire electrostatic charges, i.e., an electrical surface potential.
The presence of such charges would cause the redistribution ions in the
neighborhood of the charged surface, leading to the development of a so-called
electrical double layer (EDL). An EDL consists of an immobile compact layer
and a mobile diffuse layer where there are more counter-ions than co-ions and
hence the net charge density is not zero. When a liquid is forced through a
microchannel under an applied hydrostatic pressure, more counter-ions in the
diffuse layer are carried towards the downstream to form a streaming current,
along the direction of the liquid flow. Meanwhile, the accumulation of
counter-ions in the downstream end builds up an electric field with a streaming
potential which in turn generates a conduction current, in the opposite
direction of the flow. When the conduction current equals the streaming
current, a steady state is reached. It is easy to comprehend that the streaming
potential would exert electrostatic resistant force on the net charge density in
the diffuse layer, thereby hinder the pressure-driven flow, which is also termed
as the electroviscous effects. The electrokinetic electroviscous effects become
significant for liquid flow in a microchannel where the thickness of the EDL is
often comparable with the channel dimension.
The electrokinetic effects on microchannel flow have been
experimentally studied by Mala et al.[30], Ren et al.[31], Kulinsky et al.
[32] and Brutin and Tadrist [33]. Their results showed that depending on
the channel height and the electrical properties of the channel surface, the
measured flow rate of the distilled water can be 80% lower than that predicted
from the classical Poiseuille flow equation. The electroviscous effects have
also been theoretically studied for slit-like channels (Mala et al.[34], Chun and
Kwak [35]) and for rectangular channels (Yang and Li [36, 37], Yang et
al.[38] ). In these studies, the electrokinetic effects on velocity distribution,
friction coefficient, apparent viscosity, and heat transfer were examined. Their
81
analyses predicted that the electrokinetic effects can result in a higher friction
coefficient, a larger apparent viscosity, and a reduced Nusselt number.
However few studies have been reported for the flow of non-Newtonian fluids
in microchannels. Das and Chakraborty [24] considered the electroosmotic
flow of power-law fluids in a slit. Zimmerman et al.[20] studied the
electrokinetic flow of Carreau fluids in a T-shaped microchannel. Berli and
Olivares [39] analyzed the wall depletion effect on flow of non-Newtonian
fluids by extending the general force-flux relations for simple fluids to nonNewtonian fluids. More recently, Zhao et al. [40] derived a generalized
Smoluchowski slip velocity for electroosmotic flow of power-law fluids.
In this work, the electrokinetic effects on pressure driven flow of powerlaw fluids in a microchannel are studied. The flow field of power-law fluids is
governed by the general Cauchy momentum equation with consideration of a
body force originating from the interaction of the net charge density in the
channel EDL and the induced electrokinetic streaming potential. Analytical
expressions are obtained for the velocity distribution, volumetric flow rate,
apparent viscosity and friction coefficient. Parametric studies of the
electrokinetic effects on flow of power-law fluids in a microchannel under the
influence of the ionic concentration, wall zeta potential, flow behavior
index and pressure difference are performed.
82
Figure 11. Schematic configuration of a microchannel slit with height of 2H and length
of L, and with a uniformly zeta potential of w.
Fx = Ex e
(41)
When the wall zeta potential w is small, the net charge density
e can
e ( y ) = 2
(42)
where
( y) = w
cosh( y )
cosh( H )
(43)
(Y ) = w
83
cosh( KY )
cosh ( K )
(44)
Recalling that the magnitude of the rate of strain tensor in this case is
expressed as = (1/ 2) dvx / dy , we can obtain an expression for the viscosity
using Eq. (2),
= m(
d vx n 1
)
dy
(45)
where the negative sign is chosen because the velocity decreases with
increasing y in the channel.
Therefore, we can show that the Cauchy momentum Eq. (5) can be
simplified to
dv
dv
dp d
+
[m ( x ) n 1 x ] 2 Ex = 0
dx dy
dy
dy
(46)
v=
vx
V
Px =
H dp
V 2 dx
Ex =
Ex H
G1 =
2zen 0
V 2
n 1 dp n nn+1
V=
H
n + 1 m dx
(47)
we can obtain Eq. (48) that gives the dimensionless form of Eq. (46),
n +1 n d
dv n
n +1 n
)
[(
) ] (
) ( Px ) 1 G1 E x = 0
n
dY
dY
n
(48)
where V is the centerline velocity without consideration of the EDL effect and
0 is a reference electrical potential. Eq. (48) can be solved using the following
boundary conditions
84
v Y =1 = 0
dv
dY
=0
(49)
Y =0
v(Y ) =
1
G E sinh( KY ' ) 1n '
n +1
( P ) n [( Px )Y ' 1 x w
] dY
Y
cosh( K )
n
K
(50)
Eq. (50) shows that the flow is retarded due to the induced streaming
potential. The integral can be carried out analytically only for specific values
of the flow behavior index n, such as 1, 1/ 2 and 1/ 3 etc.
Specific Cases
The case of n = 1 corresponds to Newtonian fluids where m = , and
Eq. (50) can be evaluated as
v (Y ) = (1 Y 2 ) 2( Px ) 1 G1 w Ex
cosh ( K ) cosh ( KY )
K 2 cosh ( K )
(51)
v (Y ) = (1 Y 3 ) + 3( Px ) 2 (G1 w Ex ) 2
G1 w Ex sinh( KY ' )
K
cosh( K )