Microfluidics: Theory and Applications (Physics Research and Technology)

PHYSICS RESEARCH AND TECHNOLOGY

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.

PHYSICS RESEARCH AND TECHNOLOGY Additional books in this series can be found on Nova’s website under the Series tab.

Additional E-books in this series can be found on Nova’s website under the E-books tab.

PHYSICS RESEARCH AND TECHNOLOGY

MICROFLUIDICS: THEORY AND APPLICATIONS

IVAN A. KUZNETSOV EDITOR

Nova Science Publishers, Inc. New York

Copyright © 2010 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, 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 in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS.

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

Published by Nova Science Publishers, Inc. © New York

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

1

55

103

135

149

vi Chapter 6

Index

Contents Application to Adaptive Optics and Laser Microfluidics Jean-Pierre Delville, Alexis Casner, Régis 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 (μ-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

This book chapter consists of two parts. In Part 1, electroosmotic flow of power-law fluids in a slit channel is analyzed. The governing equations including the linearized Poisson–Boltzmann equation, the Cauchy momentum equation and the continuity equation are solved to seek analytical expressions for the shear stress, dynamic viscosity and velocity distribution. Particularly, a counterpart for the classic Smoluchowski velocity is introduced by taking into account contributions due to the finite thickness of the electric double layer (EDL) and the flow behavior index of power-law fluids. In Part 2, the pressure driven flow of power-law fluids in microchannels subject to electrokinetic effects is addressed. The Cauchy momentum equation together with the power-law fluid constitutive equation is used to describe the power-law fluid flow in a slit microchannel with consideration of a body force resulting from the interaction of the charge density in the electrical double layer of the channel and the flow-induced electrokinetic potential. The velocity profile, volumetric flow rate, apparent viscosity and friction coefficient are analytically evaluated, and the influencing factors including ionic concentration, wall zeta potential, flow behavior index and pressure difference are investigated. It is found that the pseudoplastic fluids are more susceptible to electrokinetic effects than the dilatant fluids, and thus flow characteristics of the pseudoplastic fluids are found to deviate drastically from those of Newtonian fluids. Chapter 3 - The purpose of this chapter is to describe the applications of microfluidic principles for creation of capillary electrophoresis (CE) chip devices with electrochemiluminescent (ECL) detection and to discuss the problems associated with their interfacing and the approaches that have been developed to surmount them. Basic methodology, instrumentation, unique features, and specific futures of microfluidic ECL detection with CE are discussed. ECL assay detection types, such as direct and indirect analysis, coreactant use including are considered. Publications data of scientific centers, involved in development of CE chips with ECL detection from the start of microfluidic CE/ECL joint technique are summarized. Basic tendencies of this direction future developments are covered. Chapter 4 - Microfluidic networks consisting of more than one channel must often be controlled by valves. External valves are sturdy, but low dead volumes require internal valves. Many present-day valves contain membranes (that are, e.g., driven by air pressure) that become leaky by overpressure or particles, and simple microfluidic valves control only a single channel at a

x

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".

In: Microfluidics: Theory and Applications ISBN 978-1-61668-570-6 Editor: I. A. Kuznetsov, pp. 1-54 © 2010 Nova Science Publishers, Inc.

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

State Key Laboratory of High Field Laser Physics, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, P.O. Box 800-211, Shanghai 201800, China 2 RIKEN - Advanced Science Institute, Hirosawa 2-1, Wako, Saitama 351-0198, Japan

ABSTRACT Microbiochips such as lab-on-a-chip (LOC) devices and micro total analysis systems (μ-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

2

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al. 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 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.

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

Microbiochips Monolithically Integrated with Microfluidics …

3

fabricating LOC. For example, enclosed microfluidic structures such as microfluidic channels and chambers are key elements in almost all LOC devices; however, because of their inherently 3D nature, they cannot be formed directly inside transparent substrates by 2D microfabrication techniques. To overcome this problem, open microchannels and/or chambers must be initially formed on the surfaces of substrates by photolithography or soft lithography [3−5], and then stacking and bonding procedures must be used. This leads to additional cost and complexity. On the other hand, unlike semiconductor chips, which mostly consist of just integrated microelectronics, LOC chips are hybrid integrated systems that incorporate elements having different functions. Currently, a common way of achieving monolithic integration of multifunctional components is to first separately fabricate the different types of components and then assemble them onto a substrate using microassembly techniques. However, because of the rapid development of LOC technology, LOC systems are now becoming increasingly complex, making assembly and packaging very difficult, if not impossible. Thus, new fabrication processes need to be developed to tackle these problems. In this chapter, we demonstrate that 3D femtosecond laser direct writing has great potential for fabricating LOC devices [6,7]. On the one hand, it allows for direct formation of hollow 3D microstructures embedded in transparent materials without employing any stacking or bonding procedures; such microstructures can serve as microfluidic, microoptical, and micromechanical elements, etc. On the other hand, with this technique, monolithic integration of multifunctional components into a single chip can be realized after a single continuous laser writing process followed by chemical treatments. Since the chemical treatments are batch processes, the increase in the cost of individual LOC chips is insignificant and will not pose a major problem [8]. This chapter is organized as follows: we begin by providing a brief introduction to the basic concepts of 3D femtosecond laser direct writing in Section 2. In Section 3, we show how to fabricate multifunctional structures embedded in transparent materials with true 3D configurations. In Section 4, we present a few examples of monolithically integrated, fully functional microdevices. In Section 5, we demonstrate the use of microfluidic chips fabricated by femtosecond laser micromachining for biological applications such as dynamic observation of microorganisms. Finally, we summarize the current major challenges of this technique and discuss possible solutions to some of these problems in Section 6.

4

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

2. CONCEPT OF 3D DIRECT WRITING INSIDE TRANSPARENT MATERIALS BY FEMTOSECOND LASER Due to its ultrashort pulse duration, peak intensities of over 1014 W/cm2 can easily be attained by a tightly focused femtosecond laser beam even if the pulse energy is only of the order of microjoules or millijoules. At such high intensities, the electric field of the electromagnetic wave can significantly distort the Coulomb potential of most neutral atoms, making the interaction between the light and matter extremely nonlinear [9,10]. For this reason, femtosecond laser microfabrication has several distinct advantages over conventional laser microfabrication techniques that usually employ nanosecond or continuous-wave lasers. The first advantage is that the interaction between a femtosecond laser beam and a transparent material can occur only in the vicinity of the focal point where the peak intensity is sufficiently high to initiate the multiphoton process. This important property lays the foundation for 3D direct writing inside transparent materials with femtosecond laser pulses, because there is virtually no out-of-focus absorption when a femtosecond laser beam is focused into a bulk material. The second advantage is related to the fact that conventional laser microfabrication relies on linear absorption of light. Because materials have complex electronic band structures, lasers that operate at different wavelengths are required for different materials. This is completely unnecessary for femtosecond laser microfabrication because by gradually increasing the peak intensity, the electric field of a femtosecond laser can always be increased to a level that is not negligible compared with the binding field experienced by electrons in transparent materials. The materials are then forced to absorb photons via nonlinear absorption processes. Thus, a variety of materials can be processed using a single femtosecond laser system by tuning the peak intensity. The third advantage is also a result of the ultrashort pulse duration, namely thermal effects in femtosecond laser fabrication are significantly suppressed, particularly when the peak intensity of the laser is controlled to be near the threshold of multiphoton absorption and the repetition rate is sufficiently low (e.g., a few kilohertz or a few tens of kilohertz). Effective suppression of the heat-affected zone permits fine structures with micrometer- or nanometer-scale features to be fabricated by femtosecond lasers. The fourth advantage is that the physical and chemical properties of materials can be finely tuned or even completely altered in a spatially selective manner using femtosecond laser

Microbiochips Monolithically Integrated with Microfluidics …

5

pulses. Mainly as a result of this unique characteristic, 3D femtosecond laser direct writing can integrate multiple functions on a single substrate. Figure 1 shows a typical layout of the 3D femtosecond laser direct writing systems used in our experiments [11]. In most of our experiments, the laser wavelength is ~800 nm, the pulse width is ~150 fs, and the repetition rate is 1 kHz, although sometimes we used a laser with a wavelength of 800 nm, a pulse width of ~50 fs, and a repetition rate of 1 kHz. To ensure a high beam quality, the original size of the output laser beam is usually reduced to 3 mm using a circular aperture placed before the focusing system. Microscope objectives with numerical apertures (NA) in the range 0.2−0.8 are selected for experiments requiring different fabrication resolutions and working distances, and we typically use a long-working-distance, ×20 objective lens with an NA of 0.46, which can offer a spatial resolution of ~1 μm and a working distance of 1.8 mm. With this working distance, 3D microstructures can be fabricated in a transparent sample at depths greater than 2 mm due to the refractive index of the material being greater than unity. 3D microstructures are generally formed by the direct writing technique by keeping the laser focal spot fixed while translating the sample in 3D space using a precision XYZ stage. The fabrication process is monitored using a charge coupled device (CCD) camera, the output of which is displayed on a PC monitor.

Figure 1. Schematic of a typical system for 3D femtosecond laser direct writing.

In principle, 3D femtosecond laser direct writing can be performed inside any material that is transparent to the incident beam [6]. However, to realize multiple functions in a single substrate, most of our experiments were performed with a photosensitive glass manufactured by Schott Corporation

6

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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].

Microbiochips Monolithically Integrated with Microfluidics …

7

3. FEMTOSECOND LASER FABRICATION OF 3D FUNCTIONAL COMPONENTS 3.1. Microfluidic Components and Controlling the Aspect Ratio of Microchannels Initially, our research was mainly focused on establishing a technique for directly forming a variety of microfluidic components within glass [14,23,24]. The fabrication process consists of three main steps: (1) 3D direct writing of latent images in the sample using a femtosecond laser beam tightly focused by an objective lens, (2) baking the sample in a programmable furnace for the formation of modified regions, and (3) etching the sample in a 10% aqueous solution of hydrofluoric (HF) acid in an ultrasonic bath to selectively remove the modified regions. Figure 2(a) shows X-shaped microfluidic channels formed at a depth of 300 μm below the sample surface with a channel length of ~2800 μm and a width of ~45 μm. The channels appear to be uniform, which is probably a result of using a relatively high concentration of HF acid (10% in our case compared with 5% HF for UV exposure by H. Helvajian et al. [8]) and a high-intensity ultrasonic bath. We find that using an ultrasonic bath is critical for fabricating uniform structures by chemical etching because it greatly enhances the supply of fresh HF acid and reactant in the narrow channels and tiny chambers during etching. Moreover, a chemical microreactor was fabricated by forming microchannels and microchambers with true 3D configurations inside Foturan glass, as shown in Figure 3 [23]. It is noteworthy that all these structures are directly formed in a glass coupon without using any multistep procedures such as stacking and bonding, greatly reducing the cost by enabling high-throughput manufacturing. A frequent problem in 3D laser microprocessing is that the vertical resolution (i.e., parallel to the laser beam) is always inferior to the lateral resolution due to the focal spot being elongated in the propagation direction of the laser beam (i.e., the axial direction). Consequently, the cross-sections of fabricated microfluidic channels have high aspect ratios (i.e., the height to width ratio). The cross-section of the Y-branched microfluidic channel in Figure 4 has an aspect ratio of ~4.2 [14]. To overcome this problem, we inserted a narrow slit above the objective lens to tailor the focal spot shape, as illustrated in Figure 5(a). The slit functions as a diffraction aperture and allows the shape of the focal spot and hence the aspect ratio of fabricated channels to be controlled [17]. This is because that the slits causes the laser beam to be

8

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

loosely focused in the direction perpendicular to the slit, which laterally expands the focal spot; meanwhile, the laser beam will still be tightly focused in the direction parallel to the slit to ensure a high axial resolution. Figure 5(c) shows the cross-section of a microfluidic channel fabricated with a combination of an objective lens with an NA of 0.46 and a 500 μm (width) × 3000 μm (length) slit; it shows that the slit reduces the aspect ratio of the cross-section from ~3 to ~1.6. This technique is now also widely used for writing optical waveguides in glasses [25−27]. Using an objective lens with an NA of 0.8 and a slit with dimensions 200 μm (width) × 3000 μm (length) produced waveguides with a completely circular cross-sections (diameter: ~9 μm) that can be used as single-mode waveguides [28].

Figure 2. A large-scale crossed-channel buried in Foturan glass.

Microbiochips Monolithically Integrated with Microfluidics …

Figure 3. A chemical microreactor with a vertical configuration.

Figure 4. (a) A horizontal Y-branched microchannel structure embedded 300 μm below the sample surface. (b) Cross-section of the microchannel. The cutting point is indicated in Figure 4(a).

9

10

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

Microbiochips Monolithically Integrated with Microfluidics …

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 α-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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

Microbiochips Monolithically Integrated with Microfluidics …

13

To demonstrate the effectiveness of the crossed-beam irradiation geometry, we fabricated several microfluidic channels by both single-beam and crossed-beam irradiation methods, and scanned the samples in different directions. Figures 7(a) and (b) clearly show that with single-beam irradiation, the fabricated channel has an elliptical cross-section. In contrast, when the crossed-beam irradiation method was adopted, the cross-sections of the fabricated microchannels were circular regardless of the scanning direction (see Figures 7(c)−(f)). The circular cross-section generated at arbitrary scanning directions is clear evidence that a nearly spherical focal spot was generated inside the Foturan glass sample. We believe that this will be very useful in a broad range of applications, including 3D data storage and photonic crystal fabrication by two-photon polymerization [30].

Figure 7. Optical micrographs of the cross-sectional shapes of the microfluidic hollow channels fabricated by (a) single-beam irradiation, (b) crossed-beam irradiation with the laser scanning direction perpendicular to the plane defined by the two beams, and (c) crossed-beam irradiation with the laser scanning direction along the axis of one of the two crossed beams. The scanning geometries for fabricating the structures in (a)−(c) are presented in (d)−(f), respectively.

14

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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

Microbiochips Monolithically Integrated with Microfluidics …

15

(Figure 9(b)). Thus, this microplate can switch the flow direction of the reagent like a microvalve.

Figure 8. Schematic illustration of the femtosecond laser exposure for fabricating a freely movable microplate embedded in glass. (a) The dark regions are scanned by the focused femtosecond laser beam. (b) The dark regions are completely removed after postannealing and subsequent chemical etching in an HF solution. The movable glass plate is left in the hollow chamber buried in the glass (the white regions).

Figure 9. Schematic illustration of the microplate functioning as a microvalve in a microreactor.

16

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Figure 10 shows two photographs of the fabricated sample. It is clear that the microplate moves from one side to another when the infusion direction of the compressed air is switched. We confirmed that the microvalve can switch the flow direction of microfluids by injecting a liquid into the microdevice. More recently, a freely movable micromechanical needle was integrated into a microfluidic biochip to elucidate the information transmission process in Pleurosira laevis [32].

Figure 10. Photographs of a fabricated microreactor in which a freely movable microplate is embedded. (a) When compressed air was infused from the left opening in the top layer, the microplate moved to the right side. (b) When the infusion of the compressed air was changed to the right opening, the microplate moved to the left side.

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 μm. Thus, since the total sample thickness was 2000 μm, we

Microbiochips Monolithically Integrated with Microfluidics …

17

scanned 140 parallel lines to form a plane structure vertically embedded in glass. The laser pulse energy was 700 nJ and the scanning speed was 500 μm/s. After laser irradiation, the sample was subjected to heat treatment and then to chemical etching. In this case, chemical etching was performed for about ~1 h. Finally, we rinsed the sample in distilled water and dried it with nitrogen gas flow. To examine the optical properties of the micromirror, we polished the four sidewalls of the glass sample. We then examined the beam spot produced by reflecting a helium−neon (He−Ne) laser beam from the etched internal surface. An incident angle of 45° resulted in total reflection. The two arrows in Figure 11(a) indicate the optical path. A receiving screen was placed 10 mm from the end of the Foturan glass. The arrow head in Figure 11(b) indicates the reflected beam spot on this screen. The beam spot was significantly larger than the incident laser beam, indicating that the reflected beam is highly divergent. Since the lithium metasilicate crystallites developed by postannealing must grow to a certain size (a few microns) to form an etchable network, etching of the crystallites naturally leaves a rough surface. This high roughness causes strong scattering and consequently the reflected light beam is highly divergent and has a high loss. Therefore, it is essential to reduce the surface roughness for microoptical applications.

18

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

We overcame this problem by including an additional annealing process. After chemical etching, we baked the Foturan glass sample again. The

Microbiochips Monolithically Integrated with Microfluidics …

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 = (n−1)(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 He−Ne laser beam such that it passed through the microlens inside the glass, and the focused laser beam was projected onto a

20

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

CCD camera by a ×20 objective lens with an NA of 0.35. The focal spot size was approximately 30 μm in diameter for an incident He−Ne laser beam of Ф1.5−1.8 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 air−glass 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 air−glass 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-μm-pitch micrograting together with its diffraction pattern obtained using a He−Ne 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

Microbiochips Monolithically Integrated with Microfluidics …

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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Figure 13. SEM images of microoptical (a) cylindrical and (b) hemispherical lenses. The focal spots produced both lenses are presented in the right panel.

Microbiochips Monolithically Integrated with Microfluidics …

23

Figure 14. Schematic illustration and optical microscope images of a plano-convex microlens buried in Foturan glass.

Figure 15. A 5-μm-pitch microoptical grating (upper) embedded in Foturan glass and its diffraction pattern (lower).

24

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Even femtosecond laser direct writing without postannealing can increase the refractive index in Foturan glass [38,39]. This process is based on photophysical and/or photothermal reactions such as compaction and direct bond breaking, so it requires much higher pulse energies (1−2 μJ/pulse) are required than femtosecond laser direct writing in Foturan followed by annealing. However, optical waveguides written by this process have higher transmission efficiencies, especially in the visible region, since silver nanoparticles are not precipitated. The propagation loss at 632.8 nm was evaluated by fabricating three waveguides of different lengths under the same conditions in coupons cut out from the same substrate. Figure 16 shows the optical loss as a function of waveguide length for three different scanning speeds. The data were fit with linear equations; the slope of the linear fit represents the propagation loss in decibels per centimeter and the y-axis intercept gives the coupling loss from both facets of the waveguide. The coupling loss is clearly caused by a size mismatch or misalignment between the waveguide and the focused He–Ne laser beam. However, the values obtained under the different conditions in this experiment are quite similar to each other, being in the range 1.5 to 1.6 dB. The propagation loss is around 0.5 dB/cm for all samples under the conditions investigated. This propagation loss is within acceptable limits for biophotonic microchip applications. Since the physical properties of the photosensitive glass in unirradiated regions did not change markedly even after multiple thermal treatments, we were able to write optical waveguides inside the glass by femtosecond-laserinduced refractive index modification after fabricating 3D hollow structures. Thus, 3D integration of waveguides with microoptics, such as micromirrors and microlenses with hollow structures, can be realized in a single glass chip. Figure 17 shows an optical microscope image of a microoptical circuit in which two waveguides were integrated by a micromirror and a microlens in a single glass chip. In the fabricated structures, a 5-mm-long waveguide (waveguide I) is connected to a micromirror at an angle of 45° and it is also connected to a 4-mm-long waveguide (waveguide II) at an angle of 90°. The writing of waveguide II was terminated 2 mm before the plano-convex microlens so as to obtain a focused beam output. In order to characterize the 3D integrated microoptical device, we coupled a He−Ne laser beam into waveguide I and placed a ×20 objective lens outside the glass chip to capture images of the focused beam spot (indicated by the red arrow in Figure 17) using a CCD camera. In this manner, the near-field focal spot profile was observed. As seen in Figure 17, the output spot size (dark blue spot) was about 7 μm in diameter. Thus, the combination of a microlens with a waveguide is

Microbiochips Monolithically Integrated with Microfluidics …

25

highly effective for laser beam transmission. As mentioned above, the propagation loss of the optical waveguide was 0.5 dB/cm and the net loss of the microlens was ~6%. In addition, the bending loss at the micromirror was evaluated to be less than 0.3 dB at a wavelength of 632.8 nm [38]. One of the biggest advantages of this structure is that it can bend a light beam in a small area with a small bending loss. From a practical point of view, a light beam guided by a waveguide often needs to be bent when it is integrated into optical and microfluidic devices. A curved waveguide is commonly used to bend a beam, but its bending loss is significant and cannot be disregarded. To minimize the bending loss, the curvature of a curved waveguide should be greater than several millimeters (e.g., 5−6 mm), but this causes an undesirable increase in the device size. Thus, 3D integration of waveguides with a microlens and a micromirror should enable efficient transmission of a light beam in biophotonic microchip devices.

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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Figure 17. Optical microscope image and characterization of 3D integration of two waveguides with a micromirror and an optical plano-convex microlens in a single glass chip. Solid gray lines indicate the invisible waveguides in the glass.

3.4. Microelectronics Fabrication of microelectronic components by femtosecond laser direct writing is certainly desirable for developing monolithic and compact LOC systems [40−45]. 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 μm/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].

Microbiochips Monolithically Integrated with Microfluidics …

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 μm/s.

With the purpose of realizing selective metallization on surfaces of dielectric materials other than Foturan and microscope slide glass, a modified femtosecond laser direct writing was developed [42−45]. As illustrated in Figure 20, the fabrication process consists of four main steps: (1) formation of silver nitrate thin films on insulator substrates by dip coating, (2) selective modification of insulator surfaces by femtosecond laser direct writing, (3) removal of unirradiated silver nitrate film by acetone, and (4) selective copper coating by electroless plating.

28

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

Microbiochips Monolithically Integrated with Microfluidics …

29

In our experiment, a 3-mm-thick lithium niobate (LiNbO3) crystal was used as the insulator substrate because of its potential application to electrooptic integration [42]. Figures 21(a) and (b) show top and end view images of electrodes embedded in a crystal, respectively. Two 10-μm-wide grooves with a 16 μm center-to-center interval between them were ablated using a tightly focused femtosecond laser beam with an average power of 10 mW and a scanning speed of 200 μm/s. The micrograph of the V-shaped cross-section of the electrodes shown in Figure 21(b) reveals that the two ~10-μm-deep grooves have been filled with deposited copper after electroless plating for 120 min. Since optical waveguides can also be directly written inside a LiNbO3 crystal using femtosecond laser pulses [46], the embedded microelectrodes can thus be easily integrated with the buried optical waveguides, opening up the potential to fabricate 3D micro-electro-optical devices such as optical switches and optical modulators.

Figure 21. Optical micrographs of microelectrodes embedded in a LiNbO3 crystal: (a) top view and (b) end view.

30

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Although selective metallization of insulators using nanosecond lasers has been reported [47,48], the use of femtosecond lasers has several critical advantages. One advantage is that, as Figure 21(b) clearly shows, the microelectrodes are embedded deep within the insulator, which is a desirable characteristic for many novel microdevices requiring 3D configurations. Another advantage is that when the femtosecond laser intensity is well controlled near the ablation threshold, selective metallization of glass can be realized with a nanometer-scale spatial resolution, as demonstrated by the fact that tiny nanostructures (i.e., nanometer-sized holes and grooves) have been reliably produced in dielectrics using a femtosecond laser [49]. A further advantage of this technique is that it offers the ability to directly incorporate microelectrical elements into microoptical and/or microfluidic circuits in a single glass chip. Future development of this technique holds great promise for rapid, flexible, and cost-effective fabrication of monolithic 3D micro-electroopto-fluidic devices.

4. MONOLITHIC INTEGRATION OF MICROFLUIDICS, PHOTONICS, AND ELECTRONICS 4.1. Microfluidic Dye Laser With the establishment of the above-mentioned capabilities, microoptics and microfluidics can be readily integrated into a single glass chip by 3D femtosecond laser direct writing for fabricating hybrid devices. In particular, we are interested in a microfluidic dye laser which is a useful light source for optical analyses such as fluorescence detection or photoabsorption spectroscopy in LOC systems [22]. Figure 22(a) shows a micrograph of the top view of a fabricated microfluidic laser that has an optical microcavity composed of four 45° micromirrors vertically buried in glass, a horizontal microfluidic chamber embedded ~400 μm beneath the glass surface, and a microfluidic through channel that passes through the center of the microchamber. Figure 22(b) shows a micrograph of the side view of a fabricated microfluidic laser, showing the microchannel that has an average diameter of ~80 μm and the microchamber that is ~200 μm thick. Figure 22(c) depicts the optical path of the microfluidic laser. The optical cavity is composed of a pair of corner mirrors, which are formed by two micromirrors on the left-hand side and two micromirrors on the right-hand side. Light

Microbiochips Monolithically Integrated with Microfluidics …

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.

We carried out lasing experiments to demonstrate the function of the microfluidic laser. A syringe needle was used to fill the microfluidic chamber with laser dye Rh6G dissolved in ethanol (~0.02 mol/L). The microfluidic laser was then attached to an optical alignment stage and pumped by a pulsed, frequency-doubled Nd:YAG laser with a pulse duration of 5 ns and a

32

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

We measured the emission spectra of the microfluidic laser at different pumping energies. The detector head of the spectrometer (USB2000, Ocean Optics, Inc.) was placed near the output end of the microfluidic laser to collect the light from the beam that propagated downward. After measuring each spectrum, a power meter (Lasermate, Coherent, Inc.) was used to measure the average power of the pumping laser. When a low pumping pulse fluence of 0.46 mJ/cm2 was applied, only spontaneous emission with a broad spectrum was observed. Lasing commenced near a pumping fluence of 1.66 mJ/cm2. Further increasing the pumping fluence greatly increased the output power of the microfluidic laser and narrowed its bandwidth. A typical emission spectrum of the microfluidic laser with a center wavelength of ~578 nm under a high pumping energy of 4.49 mJ/cm2 was obtained with a bandwidth of approximately 5 nm. The output power of the microfluidic laser was evaluated at this pumping energy; the measured average power of one beam reached ~10

Microbiochips Monolithically Integrated with Microfluidics …

33

μW at a repetition rate of 15 Hz. Thus, a conservative estimate of the total pulse energy of both beams emitted from the microfluidic laser is approximately 1 μJ. Using the same technique, a dual-wavelength microfluidic dye laser was fabricated in Foturan in which two microfluidic chambers were embedded at different depths in the glass [52]. By designing two microfluidic chambers serially stacked in the glass, we built a dual-wavelength microfluidic laser that produces two laser emissions with different wavelengths using a single pumping laser. Figures 24(a) and (b) show that the fabricated microfluidic laser has two horizontal microchambers, which share a common optical ring cavity. The dual-wavelength laser spectrum is presented in Figure 25; one peak is centered at ~568 nm and the other at ~618 nm, corresponding respectively to the stimulated emissions of Rh6G and Rh640 dyes used in this experiment. A different approach for achieving multiwavelength output from microfluidic dyes is to produce a mixed gain medium by dissolving several different kinds of laser dyes into the solution; a dual-wavelength dye laser produced in this manner has been reported by Q. Kou et al. [53].

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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

4.2. Optofluidic Integration In fact, the above-mentioned microfluidic dye laser is just one example of the general concept of optofluidic integration [54,55]. Toward this end, optical waveguides have been buried in Foturan glass by modifying its refractive index using femtosecond laser pulses, as described in Section 3.3 [38,39]. Additionally, optical waveguides and other microoptics such as microlenses and microfluidics can be easily integrated in a single glass chip by a single ultrafast laser system for manufacturing microchips for biochemical analysis and medical inspection. Such integrated microchips are often called optofluidics. Figure 26 shows a schematic illustration of a 3D integrated optofluidic system for photonic biosensing, in which one 6-mm-long optical waveguide is connected to a microfluidic chamber with dimensions of 1.0 mm × 1.0 mm × 1.0 mm, and two microlenses with a radius of curvature of 0.75 mm are separately arranged on the left side of the microchamber for fluorescence measurements and on the opposite side from the optical waveguide across the microchamber for absorption measurements with a distance of 300 μm. The inset shows an optical microscope image of the fabricated microchip. Experimental demonstration of photonic biosensing using the integrated microchip revealed that fluorescence analysis and

Microbiochips Monolithically Integrated with Microfluidics …

35

absorption measurement of liquid samples can be performed with efficiencies enhanced by factors of 8 and 3, respectively [56].

(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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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 μm (width) × 80 μm (height). Light was coupled into the fiber by focusing a He−Ne 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 He−Ne 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

Microbiochips Monolithically Integrated with Microfluidics …

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 He−Ne laser beam into the freestanding fiber through the micromirror.

38

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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 He−Ne 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.

Besides integrating individual fluidic and optical components in a single microchip, optofluidic chips can also be produced by employing microfluids as tunable optical media. One example of this is a microfluidic waveguide uses a liquid as the waveguide cores [59−62]. Microfluidic waveguides can easily be fabricated by filling a microfluidic channel formed on a glass surface by femtosecond laser ablation with a liquid that has an adjustable refractive index [59]. Figure 29(a) shows a digital camera image of a 90° arc-shaped microchannel with a radius of curvature of 5 mm fabricated on a glass plate. A cross-sectional view of the waveguide is shown in the inset; it shows a Vshaped channel with a width of ~7 μm at the opening and a depth of ~10 μm. Figure 29(b) shows a top view of the microfluidic optical waveguide carrying a He–Ne laser beam. Figure 30 shows end views of microfluidic waveguides carrying a He−Ne laser beam. It can be seen that using the same microfluidic channel, the waveguide can be switched between multimode (Figure 30(a)) and single-mode operation (Figure 30(b)). This was achieved by using a mixture of two liquids (paraffin, n=1.474, and α-bromnaphtalene, n=1.658) with different refractive indices so that the refractive index can be continuously tuned by varying the mixing ratio of the two liquids. In Figure 30(a), the refractive index of the liquid core was set to 1.658, which resulted in a multimode waveguide. When the refractive index was reduced to 1.527,

Microbiochips Monolithically Integrated with Microfluidics …

39

which led to the formation of a single-mode waveguide as shown in Figure 30(b). Figure 30(c) shows that the single-mode beam has a diameter of ~5 μm.

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 He–Ne laser beam.

40

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Figure 30. Near-field patterns of (a) multimode and (b) single-mode beams. (c) Beam profile of the single-mode beam.

4.3. Electro-Optic Integration Although optofluidic integration by femtosecond laser direct writing is currently under intense investigation [63−65], electro-optic (EO) integration

Microbiochips Monolithically Integrated with Microfluidics …

41

by the same approach has not been widely demonstrated to date. EO integration in transparent materials can be achieved by combining selective metallization and waveguide writing by focused femtosecond laser beam irradiation [45]. An excellent material for demonstrating EO integration is a lithium niobate (LiNbO3) crystal, which is one of the most widely used nonlinear optical materials in integrated optics. Figure 31 shows a schematic diagram of a Mach−Zehnder interferometer (MZI) EO modulator. Commercially available MgO-doped x-cut LiNbO3 crystals were used in the experiment. To produce thermally stable waveguides in the low repetition rate regime, we wrote two parallel lines with a close separation, which produced a guiding region between the two lines [66]. Unlike waveguides formed in the focal volume, a waveguide between the two tracks of the laser modified areas can preserve the nonlinearity of the bulk crystal [67]. Three embedded electrodes were integrated into the LiNbO3 crystal, as illustrated in Figure 31. In addition, three metallic pads and connecting lines, which allow the modulator to be connected to an external electrical source, were fabricated together with the microelectrodes by the same technique.

Figure 31. Schematic layout of the EO modulator.

The device was examined using a 633-nm He−Ne 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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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].

Figure 32. Near-field intensity distribution measured at the exit of EO modulator at dc voltages of (a) 0 V and (b) 19 V.

The examples listed in this section demonstrate the potential of 3D femtosecond laser direct writing to integrate several fundamentally different functions in a single substrate. To our knowledge, there is currently no other continuous processing technique that offers the same capability.

5. NANOAQUARIUM FOR DYNAMIC OBSERVATION OF MICROORGANISMS 5.1. Concept of Nanoaquarium One important application of microfluidic structures embedded in Foturan glass fabricated by the present technique is the dynamic observation of microorganisms. A large variety of microorganisms live on the earth. Some of them move extremely rapidly, which is unusual in the macro-scale world in which we live, and exhibit unique 3D movement that goes against gravity. Most of them are single cells. It is very important to investigate their dynamic movement and physiological energy generation mechanisms to gain a better understanding of the potential ability and function of single cells that make up more complex organisms such as humans. The results will also be useful for

Microbiochips Monolithically Integrated with Microfluidics …

43

developing biomotors. Consequently, observing microorganisms is currently important for cell biologists. Optical microscopes equipped with high-speed cameras are commonly used by cell biologists to observe microorganisms [69−72]. In conventional observation systems, a glass slide with a coverslip or a Petri dish is generally used. However, the high-NA objective lens typically used for these observations limits the field of view to several hundred micrometers and it also restricts the depth of focus to several hundred micrometers, making it difficult to capture images of rapidly moving microorganisms. Consequently, it takes a very long time to obtain clear images of moving microorganisms. Reduced observation times are urgently needed by biologists not only in the interests of cost and time effectiveness, but also because of limited computer memory, which becomes problematic when acquiring movies using high-speed cameras. To solve these problems, we propose using microchips with 3D microfluidic structures for observing microorganisms [32]. The microchip reduces the size of observation area; that is, it can three-dimensionally encapsulate microorganisms in a limited area. But it still provides sufficient space for them to move, making it much easier to capture images of their movement, as shown in Figure 33. Using a microchip in which microfluidics are threedimensionally confined in glass has an additional advantage of enabling microorganisms to remain highly active for a long time, since such a structure prevents evaporation and leakage of water.

Figure 33. Concept of nanoaquarium for dynamic observation of microorganisms.

44

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

One advantage of our technique for fabricating 3D hollow microstructures is that a wide variety of structures can easily be fabricated by the same procedure, because of its ability to manufacture rapidly. The ability to rapidly manufacture various microchips with different structures and functions is needed by biologists to enable them to observe different kinds of microorganisms [73−76]. In addition, micromechanical components such as microvalves can be integrated in the microfluidics, as described in Section 3.2. Such functional microcomponents need to be integrated into microchips for some kinds of microorganisms (e.g., to stimulate a cell using a micromechanical component). Such functional microcomponents can be easily integrated by our technique. We refer to such microchips for observing microorganisms as nanoaquariums, because they are far smaller than conventional aquariums (although the microchannel widths in the microchips are of the order of several tens of microns). In addition, the volume of liquids used in such microchips is of the order of a nanoliter.

5.2. Nanoaquarium for Observing the Motion of Euglena gracilis Euglena gracilis is a single-celled organism that lives in fresh water. It has a flagellum that emerges from the anterior end of the cell and it whips its flagellum rapidly to swim in water. Many biologists have used microscopes to investigate the continuous flagellum movement for both biomotor applications and to determine the origins of this functionality. However, only the thrusting movement of the flagellum has been investigated in previous research [77,78] and its detailed mechanism still remains unknown due to difficulties in capturing continuous high-speed images of the flagellum. To enable clear and efficient observation of flagellum movement, we fabricated a microchannel in photosensitive glass, as depicted schematically in Figure 33. One Euglena gracilis is about 100 µm long and 40 µm wide and it generally propels itself forward by whipping its flagellum around its body. Therefore, we fabricated a microstructure with a 1-mm-long channel and with a cross-section of 150 µm × 150 µm embedded 150 µm below the glass surface. At both ends of the channel, two open reservoirs with dimensions of 500 µm × 500 µm were connected to allow the introduction of Euglena gracilis in water. The most important constraint in fabricating this microchip is that the distance between the glass surface and the upper wall of the microchannel needs to be between 130 and 170 µm, as determined by the working distance of the objective lens used for the observation. In this

Microbiochips Monolithically Integrated with Microfluidics …

45

experiment, this distance was designed to be 150 µm. Furthermore, the etched glass surface must be smooth and the upper wall of the channel must be flat and parallel to the glass surface to obtain clear images. For observations, a Euglena gracilis is introduced into one of the reservoirs using an injection syringe filled with water. The microchannel is immediately filled with water and the Euglena gracilis swims into the microchannel by itself since there is no water flow in the microchannel. In the microchannel, the Euglena gracilis is confined in a limited area enabling dynamic observations of the flagellum movement to be performed easily using a microscope above the glass surface, as shown in Figure 33. For microchip fabrication, the laser beam was translated in the glass sample line-by-line with a pitch of 5 µm and layer-by-layer with a pitch of 10 µm. After etching for 30 min in a 10% HF solution, additional heat treatment was performed to smooth the etched surface. This smoothing process is essential for obtaining clear images of Euglena gracilis in microscopic observations. Figure 34(a) shows optical micrographs of the top view of the microchip, and Figure 34(b) shows the side view of the microchannel when cutting the microchip along the dashed line indicated in Figure 34(a). As Figure 34(a) shows, a 1-mm-long microchannel with an almost constant width of 150 µm was fabricated. Figure 34(b) reveals that the microchannel has a rectangular cross-section and is embedded 150 µm below the glass surface, which satisfies the requirements for observations. In addition, the top internal wall of the microchannel is flat and smooth and is parallel to the glass surface. Such an internal wall was fabricated by multiple scanning of the laser beam with lateral shifts and additional annealing.

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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Figure 35(a) shows an optical microscope image of Euglena gracilis swimming in the embedded microchannel using the scheme shown in Figure 33. We also succeeded in recording movies, and Figure 35(b) shows enlarged sequential images of an advancing Euglena gracilis obtained from a movie. It shows that the Euglena gracilis coils its flagellum around its body and rotates very rapidly to swim in a straight line.

Figure 35. Microscope images of (a) encapsulated Euglena gracilis swimming in the microchannel and (b) sequential pictures of advancing Euglena gracilis.

Using this microchannel, we could stimulate the Euglena gracilis by irradiating light from any direction, and by this means we could easily control its motion. Figure 36(a) shows sequential pictures of a rotating Euglena gracilis when white light was irradiated from the bottom of the microchip, as shown in Figure 36(b). The Euglena gracilis turns its body to get away from the light since it is very sensitive to strong light. Interestingly, the Euglena gracilis thrust its flagellum forward to turn its body, unlike when it swims in a straight line (see Figure 36(b)). Furthermore, it has been impossible to observe Euglena gracilis from the front using conventional methods, although such observations are highly desirable for biologists to enable more detailed analysis of the flagellum movement. The microchip we developed enables such observations using the scheme shown in Figure 37(a). Figure 37(b) shows, for the first time, a front view of Euglena gracilis swimming upward in the reservoir, which is a part of a movie. Using the microchip, the observation time can be reduced by a factor of more than 10 compared to the conventional method that uses a Petri dish. In

Microbiochips Monolithically Integrated with Microfluidics …

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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

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.

6. CONCLUSIONS AND OUTLOOK As we have demonstrated above, we are currently able to obtain functions such as microfluidics, microoptics, photonics, micromechanics, and microelectronics in various transparent materials by 3D femtosecond laser direct writing. These functions can also be integrated into microchips to construct hybrid microdevices. A few integrated microfluidic biochips have been successfully used in biological research and they have important advantages over conventional techniques. All these facts clearly indicate that this technique has great potential to become an important tool for manufacturing LOC devices. Although the examples given above make it appear straightforward to integrate functional components by 3D femtosecond laser direct writing, this is often not the case. The major difficulties that we currently face are as follows: 1. Creation of embedded microfluidic channels and chambers inevitably requires chemical wet etching, while a limited contrast ratio in etching

Microbiochips Monolithically Integrated with Microfluidics …

49

rates between modified and unmodified regions results in inhomogeneity in the microstructures (they are usually wider at openings and narrower in the middle). This problem becomes particularly severe when the channels are long. Thus, the lengths of most microfluidic structures produced by femtosecond laser direct writing followed by chemical etching are limited to a few millimeters or less, which is not sufficiently long for some LOC applications. 2. The optical properties of microoptical components are still relatively poor compared to those of finely polished optical components fabricated by conventional techniques. For example, although over a small area (i.e., 20 μm × 20 μm), the average roughness on an optical surface created by femtosecond laser direct writing can be as low as ~0.8 nm, the surface is still slightly wavy over a larger scale (e.g., a few hundred microns). As a result, a microlens fabricated by femtosecond laser direct writing followed by chemical etching and postannealing creates a significantly larger focal spot than the diffraction-limited one. This aspect certainly needs to be improved. 3. The compatibility of the fabrication procedures for different functional components requires further optimization, because it determines the ability to integrate certain elements. For example, in the fabrication of Foturan glass, postannealing after laser irradiation is required to develop a chemically etchable phase in the glass matrix. This will be unacceptable if it is desired to write an optical waveguide in glass by inducing refractive index changes in glass while simultaneously writing hollow structures. To overcome this problem, a waveguide can be written after forming all the hollow structures, but this solution means that fabrication is no longer a single continuous process. 4. Because of the chemical etching procedure involved, the spatial resolution of this technique is of the order of micrometers. This urgently requires improvement. This is particularly important for microdevices integrated with microoptics. As is well known, for optical applications such as imaging or beam focusing, the positioning precision of lens should be on the wavelength scale or even better. It is very difficult to meet this requirement because of the uncertainty in removing glass material in the chemical etching process. Fortunately, optofluidic components are usually tunable so that this problem can be overcome by exploiting tunability.

50

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

Although the above-mentioned difficulties exist, we are optimistic that they can all be overcome by technical advances in the near future. At this stage, only microdevices integrated with a small number of elements have been successfully fabricated, such as the microfluidic laser, the nanoaquarium, and the microoptical modulator. However, we have started to fabricate more complex microdevices by femtosecond laser direct writing, which may eventually replace expensive and bulky equipment such as optical microscopes. The integration of microfluidics, microoptics, and microelectronics in a single glass chip may lead to active devices for applications in both biotechnology and information technology. Finally, we believe that close collaboration between physicists and biologists will lead to innovative devices that have never been previously fabricated being produced by 3D femtosecond laser direct writing.

REFERENCES [1]

M. A. Burns, B. N. Johnson, S. N. Brahmasandra, K. Handique, J. R. Webster, M. Krishnan, T. S. Sammarco, P. M. Man, D. Jones, D. Heldsinger, C. H. Mastrangelo, and D. T. Burke, Science, 282, 484 (1998). [2] P. Yager, T. Edwards, E. Fu, K. Helton, K. Nelson, M. R. Tam and B. H. Weigl, Nature, 442, 412 (2006). [3] R. J. Jackman, T. M. Floyd, R. Ghodssi, M. A. Schmidt and K. F. Jensen, J. Micromech. Microeng. 11, 263 (2001) [4] Y. Xia, and G. M. Whitesides, Annu. Rev. Mater. Sci. 28, 153 (1998) [5] M. A. Unger, H. P. Chou, T. Thorsen, A. Scherer, and S. R. Quake, Science, 288, 113 (2000). [6] R. Gattass, and E. Mazur, Nat. Photon. 2, 219 (2008) [7] K. Sugioka, Y. Cheng, and K. Midorikawa, Appl. Phys. A, 81, 1 (2005). [8] H. Helvajian, P. D. Fuqua, W. W. Hansen, and S. Janson, RIKEN Rev. 32, 57 (2001). [9] Brabec, T. & Krausz, Rev. Mod. Phys. 72, 545 (2000). [10] M. Gertsvolf, H. Jean-Ruel, P. P. Rajeev, D. D. Klug, D. M. Rayner, and P. B. Corkum, Phys. Rev. Lett. 101, 243001 (2008). [11] Y. Cheng, K. Sugioka, M. Masuda, K. Toyoda, M. Kawachi, K. Shihoyama, and K. Midorikawa, RIKEN Rev. 50, 101 (2003). [12] See http://www.mikroglas.com/foturane.htm

Microbiochips Monolithically Integrated with Microfluidics …

51

[13] Stooky, “Photosensitively opacifiable glass,” U.S. Patent 2684911 (1954). [14] M. Masuda, K. Sugioka, Y. Cheng, N. Aoki, M. Kawachi, K. Shihoyama, K. Toyoda, H. Helvajian, and K. Midorikawa, Appl. Phys. A, 76, 857, (2003). [15] Y. Kondo, J. Qiu, T. Mistsuyu, K. Hirao, and T. Yoko, Jpn. J. Appl. Phys. 38, L1146-1148 (1999). [16] T. Hongo, K. Sugioka, H. Niino, Y. Cheng, M. Masuda, I. Miyamoto, H. Takai, K. Midorikawa, J. Appl. Phys. 97, 063517 (2005). [17] Y. Cheng, K. Sugioka, K. Midorikawa, M. Masuda, K. Toyoda, M. Kawachi, and K. Shihoyama, Opt. Lett. 28, 1144 (2003). [18] Y. Cheng, H. L. Tsai, K. Sugioka, and K. Midorikawa, Appl. Phys. A, 85, 11 (2006). [19] J. Qiu, M. Shirai, T. Nakaya, J. Shi, X. Jiang, and C. Zhu, Appl. Phys. Lett. 81, 3040 (2002). [20] Y. Cheng, K. Sugioka, M. Masuda, K. Shihoyama, K. Toyoda, and K. Midorikawa, Opt. Express, 11, 1809 (2003). [21] C. Monat, P. Domachuk, and B. J. Eggleton, Nat. Photon. 1, 106 (2007). [22] Y. Cheng, K. Sugioka, and K. Midorikawa, Opt. Lett. 29, 2007 (2004). [23] K. Sugioka, M. Masuda, T. Hongo, Y. Cheng, K. Shihoyama, and K. Midorikawa, Appl. Phys. A, 78, 815 (2004). [24] M. Masuda, K. Sugioka, Y. Cheng, T. Hongo, K. Shihoyama, H. Takai, I. Miyamoto, and K. Midorikawa, Appl. Phys. A, 78, 1029 (2004). [25] M. Ams, G. Marshall, D. Spence, and M. Withford, .Opt. Express, 13, 5676 (2005). [26] K. Moh, Y. Tan, X.-C. Yuan, D. Low, and Z. Li, Opt. Express, 13, 7288 (2005) [27] S. Sowa, W. Watanabe, T. Tamaki, J. Nishii, and K. Itoh, Opt. Express, 14, 291 (2006). [28] S. Ho, P. R. Herman, Y. Cheng, K. Sugioka, and K. Midorikawa, “Direct ultrafast laser writing of buried waveguides in Foturan glass,” in Conference on Lasers and Electro-Optics (CLEO), Vol. 96 of OSA Trends in Optics and Photonics Series, (Optical Society of America, Washington, D.C., 2004), paper CThD6. [29] K. Sugioka, Y. Cheng, K. Midorikawa, F. Takase, and H. Takai, Opt. Lett. 31, 208 (2006). [30] H. B. Sun, Y. Xu, S. Juodkazis, K. Sun, M. Watanabe, S. Matsuo, H. Misawa, and J. Nishii, Opt. Lett. 26, 325 (2001).

52

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

[31] S. Matsuo, S. Kiyama, Y. Shichijo, T. Tomita, S. Hashimoto, Y. Hosokawa and H. Masuhara, Appl. Phys. Lett. 93, 051107 (2008). [32] Y. Hanada, K Sugioka, H. Kawano, I. S. Ishikawa, A. Miyawaki and K. Midorikawa, Biomed. Microdevices, 10, 403 (2008). [33] E. Verpoorte, Lab. Chip, 3, 42N (2003). [34] K.W. Ho, K. Lim, B.C. Shim, J.H. Hahn, Anal. Chem. 77, 5160 (2005). [35] Z. Wang, K. Sugioka, Y. Hanada, and K. Midorikawa, Appl. Phys. A, 89, 951 (2008). [36] K. M. Davis, K. Miura, N. Sugimoto, and K. Hirao, Opt. Lett. 21, 1729 (1996). [37] F. He, H. Sun, M. Huang, J. Xu, Y. Liao, Z. Zhou, Y. Cheng, Z. Xu, K. Sugioka and K. Midorikawa, Appl. Phys. A, online first, DOI: 10.1007/s00339-009-5338-4 [38] Z. Wang, K. Sugioka, Y. Hanada, and K. Midorikawa, Appl. Phys. A, 88, 699 (2007). [39] V.R. Bhardwaj, E. Simova, P. B. Corkum, D.M. Rayner, C. Hantovsky, R.S. Taylor, B. Schreder, M. Kluge, J. Zimmer, J. Appl. Phys. 97, 0 831 021 (2005). [40] Y. Hanada, K. Sugioka, and K. Midorikawa, Appl. Phys. A, 90, 603 (2008). [41] K. Sugioka, T. Hongo, and H. Takai, Appl. Phys. Lett. 86, 171901 (2005). [42] J. Xu, Y. Liao, H. Zeng, Z. Zhou, H. Sun, J. Song, X. Wang, Y. Cheng, Z. Xu, K. Sugioka, and K. Midorikawa, Opt. Express, 15, 12743 (2007). [43] J. Xu, Y. Liao, H. Zeng, Y. Cheng, Zhizhan Xu, Koji Sugioka and Katsumi Midorikawa, Opt. Commun. 281, 3505 (2008). [44] Y. Liao, J. Xu, H. Sun, J. Song, X. Wang, and Y. Cheng, Appl. Surf. Sci. 254, 7018 (2008). [45] Y. Liao, J. Xu, Y. Cheng, Z. Zhou, F. He, H. Sun, J. Song, X. Wang, Z. Xu, K. Sugioka, and K. Midorikawa, Opt. Lett. 33, 2281 (2008). [46] L. Gui, B. Xi, and T.C. Chong, IEEE Photonics Technol. Lett. 16, 1337 (2004). [47] G. A. Shafeev, Appl. Phys. A 67, 303 (1998). [48] T. J. Hirsch, R. F. Miracky, and C. Lin, Appl. Phys. Lett. 57, 1357 (1990). [49] P. Joglekar, H. Liu, E. Meyhöfer, G. Mourou, and A. J. Hunt, Proc. Natl. Acad. Sci. USA, 101, 5856 (2004). [50] W. Poon, F. Courvoisier, and R. K. Chang, Opt. Lett. 26, 632 (2001). [51] H. J. Moon, Y. T. Chough, and K. An, Phys. Rev. Lett. 85, 3161 (2000).

Microbiochips Monolithically Integrated with Microfluidics …

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. Tünnermann, Appl. Phys. Lett. 89, 081108 (2006) [67] J. Thomas, M. Heinrich, J. Burghoff, S. Nolte, A. Ancona, A. Tünnermann, 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

Ya Cheng, Koji Sugioka, Katsumi Midorikawa et al.

[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. Brösing, M. Brand, and J. M. Köhler, 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).

In: Microfluidics: Theory and Applications ISBN 978-1-61668-570-6 Editor: I. A. Kuznetsov, pp. 55-101 © 2010 Nova Science Publishers, Inc.

Chapter 2

ELECTROKINETIC FLOWS OF NONNEWTONIAN FLUIDS IN MICROFLUIDIC CHANNELS Cunlu Zhao and Chun Yang* School of Mechanical and Aerospace Engineering Nanyang Technological University, Singapore 639798, Republic of Singapore

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: [email protected].

56

Cunlu Zhao and Chun Yang 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 NavierStokes equation should be used to describe the flow characteristics of non-Newtonian fluids. This book chapter consists of two parts. In Part 1, electroosmotic flow of power-law fluids in a slit channel is analyzed. The governing equations including the linearized Poisson–Boltzmann equation, the Cauchy momentum equation and the continuity equation are solved to seek analytical expressions for the shear stress, dynamic viscosity and velocity distribution. Particularly, a counterpart for the classic Smoluchowski velocity is introduced by taking into account contributions due to the finite thickness of the electric double layer (EDL) and the flow behavior index of power-law fluids. In Part 2, the pressure driven flow of power-law fluids in microchannels subject to electrokinetic effects is addressed. The Cauchy momentum equation together with the power-law fluid constitutive equation is used to describe the power-law fluid flow in a slit microchannel with consideration of a body force resulting from the interaction of the charge density in the electrical double layer of the channel and the flow-induced electrokinetic potential. The velocity profile, volumetric flow rate, apparent viscosity and friction coefficient are analytically evaluated, and the influencing factors including ionic concentration, wall zeta potential, flow behavior index and pressure difference are investigated. It is found that the pseudoplastic fluids are more susceptible to electrokinetic effects than the dilatant fluids, and thus flow characteristics of the pseudoplastic fluids are found to deviate drastically from those of Newtonian fluids.

Electrokinetic Flows of Non-Newtonian Fluids …

57

1. ELECTROOSMOTIC FLOW OF POWER-LAW FLUIDS IN A SLIT MICROCHANNEL 1.1. Introduction Advanced microfluidic devices can perform complete biochemical analysis in a single fabricated chip. The generic microfluidic systems involve buffer fluid and sample manipulations such as pumping, valving, mixing, injection, dispensing, etc [1-3]. 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 respective pressure-driven flow or electrokinetically-driven flow. 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, electroosmotic 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 electroosmotic flow can result in reduced dispersion of sample species, making capillary electrophoresis become one of most successful technologies for chemical and biomedical analyses [4] . Extensive studies of electroosmotic flow in microcapillaries have been reported in the literature. Rice and Whitehead [5] analyzed electroosmotic flow in a narrow cylindrical capillary within the framework of the Debye– Hückel approximation. Following the method proposed by Philip and Wooding [6] for solving the nonlinear Poisson-Boltzmann equation , Levine et al. [7]extended Rice and Whitehead’s model to high zeta potentials for the electrokinetic flow in the same geometry. Using the slip velocity approach, Minor et al. [8] performed an analysis of dynamic aspects of electroosmosis and electrophoresis, and Santiago [9] studied the effects of fluid inertia and pressure on transient electroosmotic flows in a two-parallel plate. The slip velocity model was first proposed by Overbeek [10] who showed that for

58

Cunlu Zhao and Chun Yang

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 Poisson–Boltzmann equation for arbitrary zetapotentials, and analyzed the dynamic electroosmotic flow in a cylindrical capillary. In Xuan and Li’s 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. Otevřel and Klepárník [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 Chakraborty’s analyses were based on an approximation of the hyperbolic sine function, and such approximation can lead to large errors when the electrokinetic parameter κ H is small, e.g., the case of overlapped double layers. Moreover, their velocity derivation

Electrokinetic Flows of Non-Newtonian Fluids …

59

cannot be reduced to the well-known velocity expression for the Newtonian fluids when the flow behavior index is equal to unit one. This study reports analyses of the electroosmotic flow of power-law fluids in a slit microchannel. Exact solutions of the velocity distributions are found for several special values of the flow behavior index. For arbitrary values of the flow behavior index, approximate solutions of the velocity distributions are obtained by using an approximate scheme for the hyperbolic sine function. The approximate solutions are found to be in good agreement with the exact solutions and the numerical solutions. In addition, a general Smoluchowski velocity is proposed for power-law fluids. Parametric studies are performed to examine effects of the dimensionless electrokinetic parameter, κ H , flow behavior index, double layer thickness, and applied electric field on the shear stress, dynamic viscosity and velocity distributions of the electroosmotic flow of power-law fluids.

1.2. Power-Law Fluids and Governing Equations The difference between non-Newtonian and Newtonian fluids lies in that the viscous stress is not a linear function of the rate of strain tensor. A number of empirical expressions have been used to describe variations in the apparent viscosity with the rate of strain. A scalar measure of the rate of strain suitable for such expression, is the magnitude of the rate of strain tensor, which is defined as [25]

⎡1 ⎤ Γ ≡ ⎢ ( Γ : Γ )⎥ ⎣2 ⎦

1/ 2

(1)

where Γ is the rate of strain tensor and Γ is its magnitude. The fluid viscosity then can be expressed as a function of Γ , namely

μ ( Γ ) . In the present work,

a special non-Newtonian fluid termed as the power-law fluid is assumed, and its dynamic viscosity, μ is given by [25]

μ = m ( 2Γ )

n −1

(2)

Cunlu Zhao and Chun Yang

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)

where v is the velocity vector. Using a general relationship between the viscous stress tensor and the rate of strain tensor, given by Eq. (4), T τ = 2 μ ( Γ ) Γ = μ ( Γ ) ⎡∇ v + ( ∇ v ) ⎤

⎣

(4)

⎦

One can readily show that the Cauchy monmentum equation is expressed as [25]

ρ

{

}

Dv T = −∇ P + F + ∇ ⋅ μ ( Γ ) ⎡ ∇ v + ( ∇ v ) ⎤ ⎣ ⎦ Dt

(5)

where ρ is the density, P is the pressure, F is the body force vector, ∇v is the velocity gradient tensor and ( ∇ v )

T

gradient tensor.

is the transpose of the velocity

Electrokinetic Flows of Non-Newtonian Fluids …

61

1.3. Exact Solutions of Electroosmosis of Power-Law Fluids in a Slit Microchannel Figure 1 shows a two-dimensional slit microchannel of height 2H. The channel is filled with a liquid electrolyte of dielectric constant, ε . It is assumed that the slit wall is uniformly charged with a zeta-potential, ψ w , and the liquid solution behaves as power-law fluid with a flow consistency index m, and a flow behavior index n. When an external electric field E0 is imposed along x direction, the liquid sets in motion due to electroosmosis, and the flow field is governed by Eq. (5). Because of symmetry, the analysis is restricted in the upper half domain of the slit microchannel.

Figure 1. The sketch of a two-dimensional slit channel.

For a steady, fully-developed flow, the velocity components satisfy

vx = vx ( y ) and v y = vz = 0 . Therefore, the material derivative of v with respect to time vanishes and the continuity equation is automatically satisfied. Furthermore, for electroosmotic flow, no pressure is applied and gravitational body force is negligible, and thus the only driving force is due to the interaction of the applied electrical field E0 and the net charge density in the electric double layer (EDL) of the channel wall. Such force acts only along x direction, and is expressed by [4]

Fx = E0 ρ e

(6)

Cunlu Zhao and Chun Yang

62

According to the theory of electrostatics, the net charge density ρ e in the diffuse layer of the wall EDL is given by the Poisson equation, which takes the form [4]

ε

d 2ψ = − ρe dy 2

(7)

With the assumptions of the Boltzmann distribution and small zeta potentials, the electrical potential profile in the EDL is governed by the linearized Poisson–Boltzmann equation expressed by [4]

d 2ψ = κ 2ψ dy 2

(8)

which is subject to the following boundary conditions:

ψ

y=H

dψ dy

=ψ w

(9a)

=0

(9b)

y =0

κ −1 is called the Debye length, and is defined as κ −1 = ( ε k BT / 2e 2 z 2 n∞ ) , 1/ 2

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

ψ ( y) =ψ w

cosh (κ y )

cosh (κ H )

(10)

Electrokinetic Flows of Non-Newtonian Fluids …

63

Then the net charge density ρ e can be expressed as a function of the EDL potential,

ρe ( y ) = −κ 2εψ

(11)

Recalling that the magnitude of the rate of strain tensor defined in Eq. (1),

in this case is given by Γ = (1/ 2 ) dvx / dy , the viscosity defined in Eq. (2) can be expressed in terms of the velocity gradient as follows

μ = m ( 2Γ )

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 ⎤ ⎥ − κ 2ε E0ψ = 0 dy ⎥ ⎦

(13)

This equation constrained by the following boundary conditions:

vx

y=H

dvx dy

=0

=0

(14a)

(14b)

y =0

gives the electroosmotic flow field of power-law fluids in a slit microchannel. Equation (13) is equivalent to

Cunlu Zhao and Chun Yang

64

n κ 2ε E0 d ⎡⎛ dvx ⎞ ⎤ ψ ⎢⎜ − ⎟ ⎥=− dy ⎢⎝ dy ⎠ ⎥ m ⎣ ⎦

(15)

Integrating this equation from 0 to y with consideration of the boundary condition given by Eq. (14b) and the electrical potential distribution expressed by Eq. (10) leads to n

⎛ dvx ⎞ κε E0ψ w sinh(κ y ) ⎜− ⎟ =− m cosh (κ H ) ⎝ dy ⎠

(16)

Consequently, the shear stress distribution can be obtained as

τ yx = κε E0ψ w

sinh(κ y ) cosh (κ H )

(17)

From Eq. (16), the velocity gradient can be expressed as 1

dvx ⎛ κε E0ψ w ⎞ n = −⎜− ⎟ dy m ⎠ ⎝

1

⎡ sinh(κ y ) ⎤ n ⎢ ⎥ ⎢⎣ cosh (κ H ) ⎥⎦

(18)

Substituting Eq. (18) into Eq. (12) gives the expression for the viscosity of the power-law fluids,

⎛ dvx ⎞ ⎟ ⎝ dy ⎠

μ = m⎜ −

n −1

=m

1 n

( −κε E0ψ w )

n −1 n

⎡ sinh(κ y ) ⎤ ⎢ ⎥ ⎣⎢ cosh (κ H ) ⎦⎥

n −1 n

(19)

Introducing

τ ws = κε Eψ w 1 n

μ ws = m ( −κε E0ψ w )

(20a) n −1 n

(20b)

Electrokinetic Flows of Non-Newtonian Fluids …

65

which under the liming case of κ 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 τ yx and the viscosity of the power-law

μ , respectively given by Eq. (17) and Eq. (19), can be rewritten in terms of τ ws and μ w s as the following forms,

fluids

τ yx = τ ws

sinh(κ y ) cosh (κ H )

⎡ sinh(κ y ) ⎤ ⎥ ⎣⎢ cosh (κ H ) ⎦⎥

μ = μws ⎢

(21)

n −1 n

(22)

Finally, integrating Eq. (18) from y to H with boundary condition given by Eq. (14a) taken into account can lead to the velocity distribution,

vx ( y ) = κ

1− n n

⎛ εψ w E0 ⎞ ⎜− ⎟ m ⎠ ⎝

1 n

κH

∫κ

y

1

( )

sinh n (κ y ' )d κ y ' cosh

1 n

(23)

(κ H )

Examination of Eq. (23) suggests that the integration can be analytically carried out only for specific values of the flow behavior index, n such as 1, and

1 etc. 3 The case of n = 1 corresponds to a Newtonian fluid where

1 2

μ = m , and

thus Eq. (23) can be evaluated as

vx ( y ) = −

cosh (κ y ) ⎤ ⎢1 − ⎥ ⎣⎢ cosh (κ H ) ⎦⎥

εψ w E0 ⎡ m

(24)

Cunlu Zhao and Chun Yang

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 (κ H ) 1 and show that ⎟ ⎝ κε E0ψ w ⎠ sinh(κ y ) ⎛

μ = m2 ⎜ −

1 ⎛ εψ E ⎞ [sinh(2kH ) − sinh(2ky ) − 2(kH − ky ) ] vx ( y ) = κ ⎜ − w 0 ⎟ m ⎠ 2 ⎝ 2 cosh 2 (kH ) 2

(25)

⎞ cosh 2 (κ H ) 1 1 3⎛ Similarly, for n = , one can get μ = m ⎜ − ⎟ 2 3 ⎝ κε E0ψ w ⎠ sinh (κ y ) 2

and show that 3 1 ⎛ εψ E ⎞ ⎡sinh(3κ H ) − sin(3κ y ) + 9sinh ( kH ) − 9sinh ( ky ) ⎤⎦ (26) vx ( y ) = κ 2 ⎜ − w 0 ⎟ ⎣ 3 ⎝ 4cosh 3 ( kH ) m ⎠

1.4. Approximate Analytical Solutions of Electroosmosis of Power-Law Fluids in a Slit Microchannel As the integral in Eq. (23) can be analytically evaluated only under certain circumstances, in the following we will present an approximate approach to obtain the velocity distributions from Eq. (23). Mathematically, the hyperbolic sine function can be approximated as

⎧κ y ⎪ sinh (κ y ) ≈ ⎨ 1 κ y ⎪⎩ 2 e

0 < κ y ≤1

κ y >1

(27)

Such approximation was successfully used by Philip and Wooding [6] in their study to obtain the electrical potential distribution outside a charged cylindrical particle immersed in an electrolyte. Substituting Eq. (27) into Eq. (23), one can analytically obtain the velocity distributions with consideration of two different cases, κ H > 1 and 0 < κ H ≤ 1 .

Electrokinetic Flows of Non-Newtonian Fluids …

67

Case I, κ H > 1 (a) In this case, if 0 ≤ κ 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 κ H ⎡ 1 (κ y ' ) ⎤ n ⎪ ' n ' κ y d κ y + ∫ ⎢ e ⎥ d κ y' 1 ⎨ ∫ky 1 1− n ⎣2 ⎦ ⎛ εψ E ⎞ n ⎪ vx ( y ) = κ n ⎜ − w 0 ⎟ ⎩ 1 m ⎠ ⎝ cosh n (κ H )

( ) ( )

( )

⎫ ⎪ ⎬ ⎭⎪

κH 1 ⎫ ⎧ 1 1+ n ⎞ ⎡ ⎤ 1 ⎛ n ⎪ n ⎪ n − + − κ y e e 1 ( ) ⎥ 1⎜ ⎟⎬ 1 ⎨ ⎢ 1− n n +1 ⎣ ⎦ ⎠ ⎪⎭ n ⎪ n ⎝ E εψ ⎛ ⎞ 2 (28) = nκ n ⎜ − w 0 ⎟ ⎩ 1 m ⎠ ⎝ n cosh (κ H )

If κ y > 1 , Eq. (23) can be evaluated as

(b)

1

vx ( y ) = κ

1− n n

1

⎛ εψ w E0 ⎞ n ⎜− ⎟ m ⎠ ⎝

κH

∫κ

y

⎡ 1 (κ y ' ) ⎤ n ' ⎢⎣ 2 e ⎥⎦ d κ y

( )

1

cosh n (κ H )

κH κy ⎞ 1 ⎛ n e −e n ⎟ 1 1 ⎜ 1− n ⎠ ⎛ εψ E ⎞ n n ⎝ = nκ n ⎜ − w 0 ⎟ 2 1 m ⎠ ⎝ cosh n (κ H )

(29)

Average velocity can be obtained from Eqs. (28) and (29) using the following definition

V=

1 H

∫

H

0

vx dy

Therefore, the average velocity is

(30)

Cunlu Zhao and Chun Yang

68

1 ⎡⎛ n 1− 1 1 ⎢⎜ 1− n κ H ⎝ ⎛ εψ E ⎞ n n ⎣ V = nκ n ⎜ − w 0 ⎟ 2 m ⎠ ⎝

⎞ ⎟e ⎠

κH n

+

cosh

1 n −1 n ⎤ 1 e ⎥+ κ H ⎦ κ H (1 + 2n ) 1 n

(31)

(κ H )

The volumetric flow rate can also be found as

Q = 2V H = nk

1− n n

Furthermore, if

⎛ εψ w E0 ⎞ ⎜− m ⎟ ⎝ ⎠

1 n

2

n −1 n

κH 1 ⎡⎛ n ⎞ n n −1 n ⎤ 2 e ⎥+ ⎢⎜ H − ⎟ e + 1 2n ) + κ κ κ ( ⎠ ⎣⎝ ⎦

cosh

1 n

(32)

(κ H )

κ H >> 1 it can be shown that 1

⎛ e kH ⎞ n cosh (κ H ) ≈ ⎜ ⎟ ⎝ 2 ⎠ 1 n

(33)

In this case, Eq. (31) is thus reduced to

V = Vs = nκ

1− n n

1

⎛ εψ w E0 ⎞ n ⎜− ⎟ m ⎠ ⎝

(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.

Electrokinetic Flows of Non-Newtonian Fluids …

69

The velocity distribution hence can be further rewritten in terms of the average velocity as ⎧ ⎪ ⎪ ⎪V ⎪ ⎪ ⎪ vx ( y ) = ⎨ ⎪ ⎪ ⎪V ⎪ ⎪ ⎪ ⎩

1

kH 1 1+ n ⎞ 2n ⎡ ⎤ ⎛ n n n ky e e 1 − + − ( ) ⎜ ⎟ ⎢ ⎥ n +1 ⎣ ⎦ ⎝ ⎠

⎡⎛ n ⎢⎜ 1 − ⎣⎝ κ H

⎞ ⎟e ⎠

κH n

1

κy ⎛ κnH ⎞ ⎜e −e n ⎟ ⎝ ⎠

⎡⎛ n ⎢⎜ 1 − ⎣⎝ κ H

⎞ ⎟e ⎠

κH n

0 ≤ ky ≤ 1

1 n −1 n ⎤ 2n e ⎥+ + κ H ⎦ κ H (1 + 2n )

1

(35)

ky > 1

2n n −1 n ⎤ e ⎥+ + κ H ⎦ κ H (1 + 2n ) 1

It should be pointed out here that although the approximation given in Eq. (27) leads to discontinuity at κy=1, the velocity distribution expressed in Eq. (35) is continuous at κy=1.

Case II, 0 < κ H ≤ 1 The integration of Eq. (23) gives 1+ n 1+ n 1 ⎡ ⎤ n − (κ y ) n κ H ( ) 1− n ⎢ ⎥ ⎛ εψ E ⎞ n + 1 ⎣ ⎦ vx ( y ) = nκ n ⎜ − w 0 ⎟ 1 m ⎠ ⎝ cosh n (κ 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 κ H) n ( ⎛ εψ w E0 ⎞ 1 + 2n ⎜− m ⎟ 1 ⎝ ⎠ cosh n (κ H ) 1 n

(37)

70

Cunlu Zhao and Chun Yang 1+ n

1+ 2 n

2 κ nH n 1− n εψ E ⎛ ⎞ n + 1 2 w 0 n Q = 2V H = nk ⎜ − 1 m ⎟⎠ ⎝ cosh n (κ H )

(38)

1+ n ⎡ ⎤ 2n + 1 ⎢ ⎛ κ y ⎞ n ⎥ vx ( y ) = V 1− ⎜ ⎟ n +1 ⎢ ⎝ κ H ⎠ ⎥ ⎣ ⎦

(39)

1 n

1.5. Results and Discussion In calculations, the following parameters and constants are used: the relative permittivity, ε r =80, the vacuum, ε 0 = 8.85 × 10 −12 F/m , the absolute temperature, T=300K, the valence of ions, z=1, the wall zeta potential,l

ψ w = −50 mV , and the flow consistency index, m = 0.90 × 10−3 Pa ⋅ s n . 1.5.1. Comparison of the Exact and Approximate Solutions with the Numerical Simulations In order to verify the approximate solutions presented in Section 1.4, we consider a special case of Newtonian fluids, namely n=1; in this case the exact solution for the velocity distribution is given by Eq. (24), numerical integration of Eq. (23) can be performed using the Romberg method with a specified accuracy of 10-6, and the approximate solution for the velocity distribution is either given by Eqs. (28) and (29) when κ H >1 or by Eq. (36) when κ H ≤ 1 . Comparison of the Newtonian fluid velocity distributions obtained from the approximate solutions, the exact solution, and the numerical solution is shown in Figure 2A (when κ H > 1 ) and in Figure 2B (when κ H < 1 ). It is observed that the flow exhibits a plug like profile when κ H is large, say κ H ≥ 50 . Decrease in κ H (e.g., κ H ≤ 3 ) causes the overlapped wall EDLs, which in turn not only reduces the maximum velocity at the channel centerline, but also makes the velocity distribution exhibit a parabolic like profile. It also can be seen that the velocity distributions predicted by the exact solution and the numerical solution are indistinguishable. The approximate solutions can provide a good estimation of the velocity field for either

Electrokinetic Flows of Non-Newtonian Fluids …

71

relatively large κ H values (e.g., κ H ≥ 3 ) or relatively small κ H values (e.g., κ H ≤ 0.7 ). Since the cause of the deviation of the approximation solutions from the exact and the numerical solutions results from the approximate scheme in Eq. (27), it is expected that the maximum error induced by the approximate solution occurs when κ H = 1 . Therefore, as for microfluidic applications concerned where κ H is usually very large, Eqs. (28) and (29) are more appropriate. 1.0

vx/Vs

0.8

κH=10

0.6

κH=50

κH=2 κH=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

κH=0.7

vx/Vs

0.16

κH=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 κ 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 κ H < 1

Cunlu Zhao and Chun Yang

72

1.5.2. Characteristics of Electroosmotic Flow of Power-Law Fluids Figure 3 shows the normalized shear stress distributions (evaluated using Eq. (21) with the shear stress on the channel wall given by Eq. (20a) as the reference shear stress) for different values of κ H . The shear stress is independent of the fluid behavior index, n, and is zero at the channel centerline due to the symmetry. While for κ H > 1 , the shear stress distribution exhibits nonlinear. As κ H increases, the magnitude of the dimensionless shear stress on the channel wall approaches to unit one. In the limiting case where κ H → ∞ , the shear stress remains zero almost in the entire portion of the channel except for the channel wall where the shear stress rapidly jumps to unit one. The features shown by Figure 3 can be understood as follows: when κH is small (e.g., κH<1), the EDL occupies the entire channel, which induces a relatively uniform body force to drive liquid fluid. It can be expected that the flow pattern resembles that of the pressure driven flow. Hence, the shear stress linearly grows along y direction, but it cannot reach the magnitude of the wall shear stress,τ ws (= κε E0ψ w ) which is defined in Eq. (20a) for large values of κH. However, for the large κH, the EDL only exists near the channel wall region, and so does the driving force. Consequently the shear stress changes slowly in the middle portion of the channel (i.e., outside the EDL region), but quickly surpasses the shear stress for the small κH and finally reaches the magnitude of the wall shear stress, τ ws on the channel wall. 1.0

τyx/τws

0.8 0.6

κH=0.7 κH=1.0 κH=3.0 κH=10.0

0.4 0.2

κH→∞

0.0 0.0

0.2

0.4

y/H

0.6

0.8

1.0

Figure 3. Normalized shear stress distributions τ yx / τ 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 κH

Electrokinetic Flows of Non-Newtonian Fluids …

73

Figure 4 shows the dimensionless dynamic viscosities μ / μ ws calculated from Eq. (22) for different values of the fluid behavior index, n while keeping κH=10. For pseudoelastic fluids, namely n < 1 , the dimensionless dynamic viscosity monotonically decreases from the channel centerline to the channel wall. It should be pointed out that Eq. (22) fails to predict the dynamic viscosity at the channel centerline, because it produces an infinite value there. For dilatant fluids, namely n > 1 , the fluids exhibit inviscid at the channel centerline, and thus the viscosity increases gradually. At the channel wall, the dimensionless viscosity approaches to unit one regardless of the values of the fluid behavior index, n. For Newtonian fluids, namely n = 1 , the dismensionless viscosity is equal to unit one. 20 18

n=0.8 n=0.9 n=1.1 n=1.2

16

μ/μws

14 12 10 8 6

n=1.0

4 2 0 0

2

4

κy

6

8

10

Figure 4. Dimensionless dynamic viscosity μ / μ 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 κH = 10

The dynamic viscosity on the channel wall, μ ws calculated using Eq. (20b) is shown in Figure 5A (for various Debye lengths/EDL thicknesses) and in Figure 5B (for various applied electric field strengths). In Figures 5A and 5B, the Newtonian fluid viscosity μ0 is chosen as the reference viscosity, and is indicated at point (1.0, 1.0). It can be noted that different from Newtonian fluids, whose dynamic viscosity is constant, the wall dynamic viscosity of power-law fluids not only depends on the flow behavior index, n, but also depends on both the EDL thickness and the applied external electric field. For

Cunlu Zhao and Chun Yang

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

−1

-3

−1

-4

μws/μ0

κ =1nm (C=9.5x10 M) 10

κ =5nm (C=3.8x10 M)

8

κ =10nm (C=9.5x10 M)

6 4 2 0 0.80

0.85

0.90

0.95

1.00

n

1.05

1.10

1.15

1.20

1.05

1.10

1.15

1.20

(A) 14 12

μws/μ0

10

−1

-3

κ =8nm(C=1.5x10 M) 5

E0=1.0× 10 V/m 5

8

E0=5.0× 10 V/m

6

E0=1.0× 10 V/m

6

4 2 0 0.80

0.85

0.90

0.95

1.00

n

(B) Figure 5. (A) Effect of the EDL thickness on the dynamic viscosity at the channel wall, μ ws (calculated using Eq. (20b)) normalized with the Newtonian fluid viscosity μ0 . (B) Effect of the applied electric field on the dynamic viscosity at the channel wall, μ ws (calculated using Eq. (20b)) normalized with the Newtonian fluid viscosity μ0

Electrokinetic Flows of Non-Newtonian Fluids …

75

Figure 6 shows the dimensionless velocity distributions vx / Vs (evaluated from Eqs. (28) and (29)) for various values of the fluid behavior index, n while keeping κH=10. The dimensionless velocity was obtained using the generalized Smoluchowski velocity given by Eq. (34) as the reference velocity. It can be seen that irrespective of the value of the fluid behavior index, the velocity near the center portion of the channel approaches to the generalized Smoluchowski velocity. The velocity profile becomes more pluglike as the fluid behavior index decreases, thereby selecting Smoluchowski velocity as the slip velocity is more appropriate for characterizing electroosmotic flow of power fluids with lower flow behavior index.

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

2

4

κy

6

8

10

Figure 6. Dimensionless velocity distributions vx / Vs (evaluated from Eqs. (28) and (29)) normalized with the generalized Smoluchowski velocity (given by Eq. (34)) for various values of the fluid behavior index, n while keeping κ H = 10

In certain practical applications, constant flow rate/average velocity is purposefully maintained. Figure 7 shows the velocity distributions normalized by the average velocity given by Eq. (31) for various values of the fluid behavior index, n while keeping κ H = 10 . In the channel center region, the velocity increases with increasing flow behavior index; while near the channel wall, it shows an opposite trend. However, the geometric area under each curve is same so as to ensure the same flow rate.

Cunlu Zhao and Chun Yang

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

2

4

κy

6

8

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 κ 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 κ H . The difference between the average velocity and the Smoluchowski velocity reduces with increasing κ H or decreasing the flow behavior index. As κ 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 = −ε E0ψ w / μ0 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

Electrokinetic Flows of Non-Newtonian Fluids …

77

parameters unchanged such as applied electric field, zeta potential, dielectric constant etc. 1.0

κH→∞

V/Vs

0.9 0.8 0.7

κH=3 κH=5 κH=10 κH=50

0.6 0.5 0.4 0.0

0.4

0.8

n

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 κ H

40 5

E0=2.0x10 V/m

Vs/Vs0

−1

-2

−1

-3

κ =1nm (C=9.5x10 M)

30

κ =5nm (C=3.8x10 M) −1

-4

κ =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

n

Figure 9. The generalized Smoluchowski velocity Vs (for power-law fluids) normalized with the conventional Smoluchowski velocity ( Vs 0 = −ε E0ψ w / μ0 for Newtonian fluids) versus the flow behavior index, n for various EDL thicknesses

Figure 10 shows the normalized volumetric flow rate versus the flow behavior index, n, for various electric field strengths while keeping a constant

Cunlu Zhao and Chun Yang

78

microchannel height of H = 5 μ m . The volumetric flow rate, Q0 is evaluated from Eq. (32) and the reference flow rate is chosen as

Q0 = −

ε E0ψ w 2H μ0

(40)

which is corresponding to the flow rate of Newtonian fluids when κ H >> 1 . The volumetric flow rate is reduced as increasing the flow behavior index. Under the conditions considered, for certain dilatant liquids, for example n ≥ 1.3 , imposing high electric field does not necessarily increase the electrokinetic flow rate. However, for pseudoplastic liquids, n < 1 , the electrokinetic flow rate can be significantly higher than that for Newtonian fluids. 36 30

H=5μm −1 -6 κ =250nm (C=1.5x10 M)

Q/Q0

24

5

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

n

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 μ m . The volumetric flow rate Q is evaluated from Eq. (32) and the reference flow rate is chosen as Q0 = −2 H ε E0ψ w / μ0 .

1.6. Summary for Electroosmotic Flow of Power-Law Fluids This study presents a mathematical model for describing the electroosmotic flow of power-law fluids in a slit microchannel. Within the

Electrokinetic Flows of Non-Newtonian Fluids …

79

framework of the linearized Poisson-Boltzmann theory, exact solutions of the velocity distribution are found for several special values of the flow behavior index. Furthermore, by utilizing an approximate scheme for the hyperbolic cosine function, approximate solutions of the velocity distribution are also obtained. The approximate solutions are validated by comparing with numerical solutions. A generalized Smoluchowski velocity is introduced by taking into account contributions due to the finite EDL thickness and the flow behavior index of power-law fluids. The calculations show that the shear stress is independent of the fluid behavior index. For pseudoplastic fluids, n < 1 , the dimensionless dynamic viscosity monotonically decreases from the channel centerline to the channel wall. For dilatant fluids, n > 1 , the fluids exhibit inviscid at the channel centerline, and then the viscosity increases gradually. At the channel wall, the dimensionless viscosity approaches to unit one regardless of the values of the fluid behavior index, n. Irrespective of the value of the fluid behavior index, the velocity near the center portion of the channel approaches to the generalized Smoluchowski velocity. The velocity profile becomes more plug like as the fluid behavior index decreases, suggesting that the Smoluchowski velocity is more appropriate as the slip velocity for electroosmotic flow of power fluids with low flow behavior index. The generalized 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 . It is also found that for pseudoplastic liquids, n < 1 , the generalized Smoluchowski velocity can be several times of the conventional Smoluchowski velocity, and thus the electrokinetic flow rate can be significantly higher than that for Newtonian fluids.

2. PRESSURE DRIVEN FLOW OF POWER-LAW FLUIDS IN A MICROCHANNEL WITH ELECTROKINETIC EFFECTS 2.1. Introduction Liquid flow through microchannels has found its applications in microfluidic devices, ranging from pH and temperature sensors, to fluid actuators, such as pumps, mixers, and valves [27], as well as Lab-on-a-Chip

80

Cunlu Zhao and Chun Yang

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

Electrokinetic Flows of Non-Newtonian Fluids …

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.

2.2. Pressure Driven Flow Field of Power-Law Fluids in a Slit Microchannel with Electrokinetic Effects Consider a slit microchannel of height 2H and length L as illustrated in Figure 11. The channel is filled with an incompressible, power-law electrolyte of constant dielectric constant ε , flow consistency index m, and flow behavior index n. The slit wall is assumed to be uniformly charged with a zeta potential

ψ w . Because of geometric symmetry, the analysis is restricted in the upper half domain of the slit microchannel. When a pressure difference is applied along the microchannel, the liquid flow is governed by the Cauchy momentum Eq. (5). For a steady, fully developed flow, the components of v satisfy vx = vx ( y ) and v y = vz = 0 . The hydraulic pressure gradient is a constant. Therefore, the material derivative of v with respect to time vanishes and the continuity equation is automatically satisfied.

Cunlu Zhao and Chun Yang

82

Figure 11. Schematic configuration of a microchannel slit with height of 2H and length of L, and with a uniformly zeta potential of ψw.

Furthermore, with negligible gravitational force, the only body force considered here is due to the interaction of the net charge density in the channel EDL

ρe and the induced streaming potential Ex . Such force acts only

along x direction, and is given by

Fx = Ex ρe

(41)

When the wall zeta potential ψ w is small, the net charge density

ρe can

be expressed as a function of the EDL potential [4]

ρe ( y ) = −κ 2εψ

(42)

κ −1 is termed as the Debye length, and is defined as κ −1 = (ε k BT / 2e 2 z 2 n∞ )1/ 2 (here n∞ and z are the bulk number

where

concentration and the valence of ions, respectively, e is the fundamental charge, kB is the Boltzmann constant, and T is the absolute temperature). An expression for the EDL potential distribution is of the following form [4]

ψ ( y) = ψ w

cosh(κ y ) cosh(κ H )

(43)

Defining the dimensionless groups: K=κH, Y=y/H and Ψ=zeψ/kbT, we can nondimensionalize Eq. (43) as

Electrokinetic Flows of Non-Newtonian Fluids …

Ψ (Y ) = Ψ w

83

cosh( KY ) cosh ( K )

(44)

Recalling that the magnitude of the rate of strain tensor in this case is expressed as Γ = (1/ 2) dvx / dy , we can obtain an expression for the viscosity using Eq. (2),

μ = 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 ] − κ 2ε Exψ = 0 dx dy dy dy

(46)

By introducing the following dimensionless parameters,

v=

vx V

Px =

H dp ρV 2 dx

Ex =

Ex H

ζ0

G1 =

2zen∞ζ 0 ρV 2

1

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 Ψ = 0 n dY dY n

(48)

where V is the centerline velocity without consideration of the EDL effect and ζ0 is a reference electrical potential. Eq. (48) can be solved using the following boundary conditions

Cunlu Zhao and Chun Yang

84

v Y =1 = 0

dv dY

=0

(49)

Y =0

An analytical solution of Eq. (48) can be obtained as

v(Y ) =

1 − 1 G E Ψ 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 = μ , and Eq. (50) can be evaluated as

v (Y ) = (1 − Y 2 ) − 2( − Px ) −1 G1Ψ w Ex

cosh ( K ) − cosh ( KY ) K 2 cosh ( K )

(51)

When n = 1/ 2 , we can show that the dimensionless velocity can be expressed as [sinh(2 K ) − sinh(2 KY )] − 2( K − KY ) 4 K 3 cosh 2 ( K ) 2{[ K cosh( K ) − KY cosh( KY )] − [sinh( K ) − sinh( KY )]} − 3(− Px ) −1 G1Ψ w E x K 3 cosh ( K ) (52)

v (Y ) = (1 − Y 3 ) + 3(− Px ) −2 (G1Ψ w Ex ) 2

Approximate Analytical Solution As the integral in Eq. (50) can be analytically evaluated only under certain circumstances, in the following we will present an approximate approach to obtain the velocity distributions from Eq. (50). It is assumed that in Eq. (50) the actuating pressure force term is much larger than the induced electrostatic body force term due to electrokinetic effects, and thus we have the following assumption

Electrokinetic Flows of Non-Newtonian Fluids …

G1Ψ w Ex sinh( KY ' ) K cosh( K ) 1 (− Px )Y '

85

(53)

Using Taylor’s series for |x| 1 and an arbitrary real number η , we have

(1 + x )

η

≈ 1+η x +

η (η − 1) 2

x 2 + ......

Therefore, Eq. (50) can be approximations in Eqs. (53) and (54), v (Y) =

analytically

(54) integrated

using

the

1 1 1 1 2 2 2 2 − −1 sinh(KY ') 1− n G E Ψ −2 sinh (KY ') 1 n +1 1 GE Ψ + 2 1 x2 w (−PY (−Px ) n ∫ [(−PY ')n − 1 x w (−PY ')n ')n ]dY' x x x Y n n K K cosh(K) 2n cosh2 (K) n+1

= (1−Y n ) −

n +1 F (n, K) − F(n, KY) 1− n2 H(n, K) − H(n, KY) + 3 (−Px )−2 G12Ψ2wEx2 (−Px )−1G1ΨwEx n+1 n+1 2 n n 2 K n cosh(K) K n cosh2 (K)

(55) where the two auxiliary functions, i.e., F(n, x) and H(n, x), are defined in Appendix. Likewise, all other auxiliary functions, including F1 (n, x), F2 (n, x), F3 (n, x), H1 (n, x) and H2 (n, x), used in the following are also defined in Appendix without otherwise specified, and the detailed formulation of these functions can also be found in [41]. In case no electrokinetic effects, the second term and third term on the right hand side of Eq. (55) vanish. Eq. (55) reduces to

v0 (Y ) = 1 − Y

n +1 n

(56)

which is the well-known pressure-driven flow velocity profile of power-law fluids through a parallel-plate channel. Using Eqs. (55) and (56), we can show that the mean velocity with and without the consideration of the electrokinetic effects respectively are

Cunlu Zhao and Chun Yang

86 1

Vav = ∫ v ( Y ) dY = 0

KF ( n, K ) − ⎡⎣ F1 ( n, K ) − F1 ( n, 0 ) ⎤⎦ −1 n +1 n +1 − 2 ( − Px ) G1Ψ w Ex 2 n +1 2n + 1 n K n cosh( K )

+

KH ( n, K ) − ⎡⎣ H1 ( n, K ) − H1 ( n, 0 ) ⎤⎦ −2 1 − n2 − Px ) G12 Ψ 2w Ex2 2 n +1 3 ( 2n K n cosh 2 ( K )

(57) and 1

Vav 0 = ∫ v0 (Y ) dY = 0

n +1 2n + 1

(58)

Hence, the non-dimensional volumetric flow rate through the slit microchannel, defined by Q =Q/(2HV0), is given by

Q=

V Vav V0

(59)

Here another constant reference velocity V0 is adopted instead of using the reference velocity V introduced earlier. The reason is that we usually want to examine the effects of pressure gradient and flow behavior index on the flow rate, but the reference velocity V already includes the pressure gradient and flow behavior index. Correspondingly, in the absence of the electrokinetic effects, the non-dimensional volumetric flow rate is expressed as

Q0 =

V Vav 0 V0

(60)

2.3. Streaming Potential As seen from Eqs. (55) and (57), the local and mean velocity can be evaluated only when the induced streaming potential Ex is known. As explained previously, under a steady-state condition, the conduction current Ic is equal to the streaming current Is, and the net electrical current I should be zero

Electrokinetic Flows of Non-Newtonian Fluids …

I = I s + Ic = 0

87 (61)

Due to symmetry of the microchannel, the electrical streaming current Is is defined as H

1

0

0

I s = 2∫ vx ( y ) ρe dy = −4ezn∞ HV ∫ v (Y )ΨdY

(62)

The electrical conduction current Ic in the microchannel consist of two parts: one is due to the conductance of the bulk liquid; the other is due to the surface conductance of the compact layer of the EDL. The electrical conductance current can be expressed as

I c = I bc + I sc = λt Ex Ac = 2λtζ 0 Ex

(63)

where I bc , I sc represent the bulk and surface conductance current respectively.

λt is the total electrical conductivity and it can be calculated by λt=λb+ λsPs/Ac. Here Ps and Ac are the wetting perimeter and the cross-sectional area of the channel, respectively. λb is the bulk conductivity of the solution, and λs is the surface conductivity, which may be determined by experiment. From Eqs. (61), (62) and (63), we can obtain an expression for the streaming potential 1

Ex = G2 ∫ v (Y )ΨdY 0

(64)

Here we introduce a dimensionless parameter, G2 = 2 zen∞ HV /(λtζ 0 ) . Using the velocity distribution (i.e., Eq. (55)) and the EDL potential profile (i.e., Eq. (44)), we can show that the streaming potential satisfies the following quadratic equation

an Ex2 − (1 + bn ) Ex + cn = 0 where three constant coefficients an , bn , and cn are given by

(65)

Cunlu Zhao and Chun Yang

88 an =

sinh( K ) H ( K , n) − [ H 2 (n, K ) − H 2 (n, 0)] 1 − n2 G1G2 ( − Px ) −2 Ψ 3w 2 n +1 3 2n K n cosh 3 ( K )

bn =

sinh( K ) F ( K , n) − [ F2 (n, K ) − F2 (n, 0)] n +1 G1G2 (− Px )−1 Ψ 2w 2 n +1 n2 K n cosh 2 ( K )

cn = G2 Ψ w

K

n +1 n

sinh( K ) − [ F3 ( n, K ) − F3 ( n, 0)] K

2 n +1 n

(66a)

(66b)

(66c)

cosh ( K )

Then the dimensionless streaming potential can be determined by using Eq. (67)

⎧ 1+ b − n) ⎪( ⎪ Ex = ⎨ ⎪ cn ⎪1 + b n ⎩

(1 + bn ) 2an

2

− 4an cn

n ≠1 (67)

n =1

Recall that in the previous section, the exact solutions of the velocity distributions are already obtained for the flow index, n=1 and 1/ 2 . Therefore, the exact solutions of the streaming potential for these two cases can also be found. For Newtonian fluids when n=1, substituting Eqs. (51) and (44) into Eq. (64), we can show that the streaming potential satisfies a linear equation as follow

−(b1 + 1) Ex + c1 = 0

(68)

where

b1 = G1G2 (− Px ) −1 Ψ 2w

cosh ( K ) sinh( K ) − K K 3 cosh 2 ( K )

(69a)

Electrokinetic Flows of Non-Newtonian Fluids …

c1 = 2G2 Ψ w

K cosh ( K ) − sinh ( K ) K 3 cosh ( K )

89

(69b)

The exact solution of the streaming potential can be readily evaluated from Eq. (68)

Ex =

c1 b1 + 1

(70)

Likewise, for n =

1 , it can be shown that by substituting Eqs. (52) and 2

(44) into Eq. (64), we can have the following quadric equation,

a1/ 2 Ex2 − (b1/ 2 + 1) Ex + c1/ 2 = 0

(71)

where

a1/ 2

1 5 8 sinh (2 K ) sinh ( K ) − cosh (3K ) − cosh ( K ) + 6 2 3 = 3G1G2 (− Px ) Ψ 4 K 4 cosh 3 ( K ) −2

3 w

b1/ 2 = 3G1G2 ( − Px ) −1 Ψ 2w

c1/ 2 = 3G2 Ψ w

K sinh (2 K ) − sinh 2 ( K ) − K 2 2( K ) 4 cosh 2 ( K )

K 2 cosh ( K ) − 2 K sinh ( K ) + 2 cosh ( K ) − 2 K 4 cosh ( K )

(72a)

(72b)

(72c)

From Eq. (71), the exact solution of the streaming potential is determined from Eq. (73),

Ex =

(1 + b1/ 2 ) − (1 + b1/ 2 ) 2 − 4a1/ 2 c1/ 2 2a1/ 2

(73)

Cunlu Zhao and Chun Yang

90

One can readily verify that the two exact results given by Eqs. (70) and (73) can be recovered from the general solution (Eq. (67)) as special cases when n respectively equals 1 and 1 / 2 .

2.4. Apparent Viscosity and Electroviscous Effects In analogy to the expression for the volumetric flow rate of the classical Poiseuille flow, we define an apparent viscosity μa to express the volumetric flow rate as

Qp =

2 H 3 dp (− ) 3μ a dx

(74)

Eq. (74) can be nondimensionalized to

Qp =

1 V ρVH (− Px ) 3 V0 μa

(75)

Substituting Eq. (59) and Eq. (60) respectively into Eq. (75), we can obtain the apparent viscosities of power-law fluids μa =

ρVH V − Px with consideration of the electrokinetic effects

μa 0 =

ρVH V − Px without consideration of the electrokinetic effects (77)

3 V0 Q

(76)

3 V0 Q0

Then the ratio of the apparent viscosity with electrokinetic effects to that without electrokinetic effects is

μa Q0 = μa 0 Q

(78)

Eq. (78) can be used to characterize the electroviscous effect. From a physics viewpoint, this ratio should be always larger than unit one.

Electrokinetic Flows of Non-Newtonian Fluids …

91

2.5. Friction Coefficient The friction factor for the flow through a channel is defined as

f =

−

dP Dh / 4 −P dx = 2 2x 2 Vav ρVav / 2

(79)

where Dh is the hydrodynamic diameter and Dh =4H for the present microchannel slit. Therefore, the friction coefficient, i.e., the product of the friction factor f and Reynolds number, is given by

C f = f Re = 2(

n +1 4 n ) n Vav

(80)

where the Reynolds number is defined as Re = ρVav Dh / m . 2− n

n

2.6. Results and Discussion Examination of the afore-derived analytical expressions reveals that the characteristics of power-law fluids flow in a microchannel slit are determined by the four dimensionless parameters: K, Px , G1 and G2 . Physically, the nondimensional electrokinetic diameter, K=κH, represents the ratio of half channel height to the thickness of the EDL. By definition, the non-dimensional H dp 1 pressure gradient, Px = / ρV 2 , can be interpreted as the ratio of the 2 dx 2 pressure energy to the kinetic energy. G1 = 2 zen∞ / ρV 2 characterizes the ratio of the electrical energy of the solution to the mechanical kinetic energy. G2 = 2 zen∞ HV / λtζ 0 represents the ratio of the streaming current to the conduction current[36-38]. In calculation, without other specifications the following parameters and constants are used: the slit channel height 2H=20 μm and length L=2 cm, the relative permittivity ε r = 80 , the absolute temperature T=300 K, the valence

Cunlu Zhao and Chun Yang

92

of ions z+ = − z− = z = 1 , the wall zeta potential ψ w = −70 mV, the ionic number concentration, 6.022×1020/m3, the fluid consistency index m=0.90×10-3 Pa·sn, and the pressure difference ∆p = 20 kPa. It should be pointed out that from the definitions of all the auxiliary functions in Appendix, they all have a singular point at x = 0 . Therefore, in the calculations the limit values when x approaches to zero are used to evaluate their values at x = 0 .

2.6.1. Velocity Distribution Figure 12 shows the dimensionless velocity distributions (valuated using Eq. (55) with the centerline velocity in the absence of the EDL effects as the reference velocity given in Eq. (47)) for three different flow behavior index of power-law fluids. In Figures 12(A)-(C), the dimensionless velocity distributions of power-law fluids without the electrokinetic effects are also plotted in dotted lines for comparison. As seen from the figures, the EDL exhibits stronger effects on the velocity distributions with lower flow behavior index than that with higher fluid behavior index. For a small fluid behavior index in Figure 12(A), the velocity distribution is distorted and the flow velocity approaches zero in the neighborhood of the channel wall region due to the action of the EDL field and the induced streaming potential. 1.0 0.8

v

0.6 0.4

n=0.8

With EDL effects Without EDL effects

0.2 0.0 0.0

0.2

0.4

Y

(A)

0.6

0.8

1.0

Electrokinetic Flows of Non-Newtonian Fluids …

93

1.0 0.8

v

0.6 0.4

n=1.0 With EDL effects Without EDL effects

0.2 0.0 0.0

0.2

0.4

Y

0.6

0.8

1.0

(B) 1.0 0.8

v

0.6 0.4 0.2

n=1.2

With EDL effects Without EDL effects

0.0 0.0

0.2

0.4

Y

0.6

0.8

1.0

(C) Figure 12. Non-dimensional Velocity distributions for three different flow behavior index (A) n=0.8; (B) n=1.0; and (C) n=1.2. The solid lines represent with EDL effects and the dotted line denote without EDL effects. Other parameters are the wall zeta potential ψw=-70mV, the ionic concentration n∞= 6.022 ×1020 / m3 , and the applied pressure difference Δp=10kPa.

Meanwhile, the velocity at the channel centerline is significantly reduced. As the fluid behavior index is increased (e.g., Figure 12(B)), the distortion becomes smaller. In Figure 12(C) where the fluid behavior index is n=1.2, the difference of the velocity distributions with and without consideration of the electrokinetic effects become indistinguishable, indicating negligible electrokinetic effects in this case. In addition, as revealed by Figure 12(A) and

Cunlu Zhao and Chun Yang

94

12(B), changing flow behavior index from 1 to 0.8 can increase one order of magnitude of the velocity, suggesting that the magnitude of velocity for pseudoplastic fluids (i.e., n<1) is very sensitive to the flow behavior index. Hence this feature may be able to be used as an effective way to adjust the flow rate in practical applications.

2.6.2. Non-Dimensional Induced Streaming Potential Figure 13 which shows the non-dimensional induced streaming potential (calculated from Eq. (67)) versus the flow behavior index for three different pressure differences. As elaborated earlier, in pressure-driven flow a streaming potential is induced along the channel axial direction due to the presence of the channel EDL. Therefore, a larger pressure difference can cause a larger amount of fluid transport and hence more ions are carried to the downstream end of the channel, giving rise to a stronger (more negative) induced electrokinetic potential. Also, it is shown that under the same pressure difference, the streaming potential is larger for the pseudoplastic fluids than for the dilatants fluids (i.e., n>1). Once the fluid behavior index is very large, say n>1.3 in this case, no streaming potential is generated regardless of the difference in applied pressure. 0.0

1 -0.5

2

1 Δ p=10kPa 2 Δ p=20kPa 3 Δ p=40kPa

Ex

-1.0 -1.5

3

-2.0 -2.5 0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

1.5

n

Figure 13. Non-dimensional streaming potential versus flow behavior index for three different pressure differences Δp=10kPa, 20kPa, and 40kPa. Other parameters are the wall zeta potential ψw=-70mV and the ionic concentration n∞= 6.022 ×1020 / m3 .

2.6.3. Volumetric Flow Rate In Figure 14, the non-dimensional volumetric flow rate is plotted as a function of the flow behavior index for two different wall zeta potentials and

Electrokinetic Flows of Non-Newtonian Fluids …

95

pressure differences. As expected, the volumetric flow rate decreases with increasing the flow behavior index. For large flow behavior index of dilatants fluids, e.g., n>1.3, the fluids become so viscous so that no flow occurs under such applied pressures. Also, the electrokinetic effects are observed only in the pseudoplastic fluids, i.e., n<1. The consequence of electrokinetic effects is a reduction of flow rate, and such electrokinetic effects are stronger for a larger applied pressure difference or/and higher channel zeta potential.

2.6.4. Apparent Viscosity (Electroviscous Effects) The ratio of the apparent viscosity (defined by Eq. (78)) with the EDL effects to that without consideration of the EDL effects is presented in Figure 15 which shows this ratio versus the flow behavior index for four different electrokinetic parameters. For a small electrokinetic parameters of K=10, it is noticeable that the electroviscous effect is present in almost the entire range of the flow behavior index range studied here. Specifically, the apparent viscosity ratio can be 2.5 times for a pseudoplastic fluid of n=0.6. As K increases, the range of fluids (characterized by the flow behavior index) where electroviscous effects are present shrinks significantly. Since increasing K either means an increase of the EDL thinness or a decrease of the channel height, both reduce the predominance of the EDL in the flow domain, resulting in weaker electroviscous effects. In a limiting case of K=100, the ratio μa/ μa0 becomes unity one, indicating that no electroviscous effects can be observed irrespective of the flow behavior index. 0 .8 0 .7

1 Δ p=40kPa 2 Δ p=10kPa

0 .6 0 .5

Q, Q

0

W ith o u t E D L e ffe c ts ψ w = -3 0 m V ψ w = -7 0 m V

0 .4 0 .3

1

0 .2 0 .1 0 .0 0 .8

2 0 .9

1 .0

n

1 .1

1 .2

1 .3

Figure 14. Non-dimensional volumetric flow rate versus flow behavior index for two different pressure differences (Δp=10kPa and 40kPa) and two zeta potentials (ψw=-30mV and -70mV).

Cunlu Zhao and Chun Yang

96 2.6 2.4 2.2

μa/μa0

2.0

K=10

1.8 1.6 1.4 1.2 1.0 0.8 0.6

K=20 K=50 K=100 0.7

0.8

0.9

n

1.0

1.1

1.2

1.3

Figure 15. Variation of μa/μa0 with flow behavior index for four different electrokinetic parameters K with an applied pressure difference Δp =20kP and a wall potential ψw=70mV.

2.6.5. Friction Coefficient Figure 16 depicts variation of the friction coefficient, expressed by Eq. (80) with the flow behavior index for three different bulk ionic number concentrations. The general trend is that the friction coefficient increases with increasing the flow behavior index. Also, a well-known friction coefficient of 24 for Newtonian fluids without electrokinetic effects is retrieved in this figure. All these coincide with our expectations. Figure 16 shows that the friction coefficient with EDL effects is always larger than that without EDL effects. The electrokinetic effects intensify as the ionic concentration decreases. However, for a concentrated solution of n∞=6.022×1023/m3, the electrokinetic effects vanish. As shown in the definition of κ −1 = (ε k BT / 2e 2 z 2 n∞ )1/ 2 , a decrease of ionic concentration leads to a thicker EDL, and thus the EDL exhibits stronger effects. Moreover, decreasing ionic concentration elevates the friction coefficient in the pseudoplastic domain more remarkably than that in the dilatant domain. This feature can also be ascribed to the fact that the pseudoplastic fluids are more sensitive to the hindrance of electroviscous effects.

Electrokinetic Flows of Non-Newtonian Fluids …

97

50

40

Without EDL effects 20 3 n∞=6.022×10 /m 21

3

23

3

Cf

n∞=6.022×10 /m

n∞=6.022×10 /m

30

20

10 0.7

0.8

0.9

1.0

1.1

1.2

1.3

n

Figure 16. Variation of friction coefficient with flow behavior index for three different bulk ionic concentrations n∞ = 6.022 ×1020 / m3 , 6.022 × 1021 / m3 and 6.022 × 10 23 / m3 with an applied pressure difference Δp=20kPa and a wall zeta potential ψw=-70mV.

2.7. Summary for Pressure Driven Flow of Power-Law Fluids with Electrokinetic Effects The electrokinetic effects on the liquid flow of power-law fluids through a microchannel slit are studied analytically. An electrostatic body force is considered in the Cauchy momentum equation governing the flow behavior of power-law fluids to account for the electrokinetic effects caused by the interaction of the channel wall EDL field and the induced streaming potential. The analytical solutions to the Cauchy momentum equation is obtained by using an approximate scheme. The expressions for the streaming potential, velocity distribution, volumetric flow rate, apparent viscosity and friction coefficient are derived. The computational results show that the electrokinetic effects result in the velocity distribution distortion, and thus cause stronger retarded flow of power-law fluids with smaller behavior index. Small dimensionless electrokinetic diameters or dilute ionic concentrations can leads to stronger electrokinetic effects, giving rise to larger apparent viscosity ratio and higher friction coefficient. Overall, the electrokinetic effects in the pseudoplastic domain are more remarkable than in the dilatant domain.

98

Cunlu Zhao and Chun Yang

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

D. J. Harrison et al., Science 261, 895 (1993). A. Manz et al., Journal of Micromechanics and Microengineering 4, 257 (1994). L. Bousse et al., Annual Review of Biophysics and Biomolecular Structure 29, 155 (2000). J. H. Masliyah, and S. Bhattacharjee, Electrokinetic and Colloid Transport Phenomena (Wiley-Interscience, Hoboken, N.J., 2006). C. L. Rice, and R. Whitehead, Journal of Physical Chemistry 69, 4017 (1965). J. R. Philip, and R. A. Wooding, Journal of Chemical Physics 52, 953 (1970). S. Levine et al., Journal of Colloid and Interface Science 52, 136 (1975). M. Minor et al., Journal of Colloid and Interface Science 189, 370 (1997). J. G. Santiago, Analytical Chemistry 73, 2353 (2001). J. T. G. Overbeek, Phenomenology of Lyophobic (Chapter II) (Colloid Science, Amsterdam, 1952). Y. Kang, C. Yang, and X. Huang, International Journal of Engineering Science 40, 2203 (2002). X. Xuan, and D. Li, Journal of Colloid and Interface Science 289, 291 (2005). D. G. Yan, C. Yang, and X. Y. Huang, Microfluidics and Nanofluidics 3, 333 (2007). F. Bianchi, R. Ferrigno, and H. H. Girault, Analytical Chemistry 72, 1987 (2000). N. A. Patankar, and H. H. Hu, Analytical Chemistry 70, 1870 (1998). C. L. Ren, and D. Li, Journal of Colloid and Interface Science 294, 482 (2006). G. Y. Tang et al., Analytica Chimica Acta 507, 27 (2004). Y. Q. Zu, and Y. Y. Yan, Journal of Bionic Engineering 3, 179 (2006). F. Kamişli, International Journal of Engineering Science 41, 1059 (2003). W. B. Zimmerman, J. M. Rees, and T. J. Craven, Microfluidics and Nanofluidics 2, 481 (2006). Y. H. Koh et al., International Communications in Heat and Mass Transfer 31, 1005 (2004).

Electrokinetic Flows of Non-Newtonian Fluids …

99

[22] M. Das, V. K. Jain, and P. S. Ghoshdastidar, International Journal of Machine Tools and Manufacture 48, 415 (2008). [23] M. Otevřel, and K. Klepárník, Electrophoresis 23, 3574 (2002). [24] S. Das, and S. Chakraborty, Analytica Chimica Acta 559, 15 (2006). [25] W. M. Deen, Analysis of Transport Phenomena (Oxford University Press, New York, 1998). [26] C. Zhao, and C. Yang, Electrophoresis 31, 973 (2010). [27] C. M. Ho, and Y. C. Tai, 1998), pp. 579. [28] A. C. Eringen, Journal of Mathematics and Mechanics 16, 1 (1966). [29] N. P. Migun, and P. P. Prokhorenko, Colloid journal of the USSR 49, 894 (1987). [30] G. M. Mala et al., International Journal of Heat and Fluid Flow 18, 489 (1997). [31] L. Ren, D. Li, and W. Qu, Journal of Colloid and Interface Science 233, 12 (2001). [32] L. Kulinsky, Y. Wang, and M. Ferrari, Proceedings of SPIE - The International Society for Optical Engineering 3606, 158 (1999). [33] D. Brutin, and L. Tadrist, Physics of Fluids 15, 653 (2003). [34] G. M. Mala, D. Li, and J. D. Dale, International Journal of Heat and Mass Transfer 40, 3079 (1997). [35] M. S. Chun, and H. W. Kwak, Korea-Australia Rheology J. 15, 83 (2003). [36] C. Yang, and D. Li, Journal of Colloid and Interface Science 194, 95 (1997). [37] C. Yang, and D. Li, Colloids and Surfaces A: Physicochemical and Engineering Aspects 143, 339 (1998). [38] C. Yang, D. Li, and J. H. Masliyah, International Journal of Heat and Mass Transfer 41, 4229 (1998). [39] C. L. A. Berli, and M. L. Olivares, Journal of Colloid and Interface Science 320, 582 (2008). [40] C. Zhao et al., Journal of Colloid and Interface Science 326, 503 (2008). [41] C. Zhao, and C. Yang, International Journal of Emerging Multidisciplinary Fluid Sciences 1, 37 (2009).

APPENDIX In the following, we will define several auxiliary functions which facilitate the analytical evaluation of the pertinent expressions in the second

Cunlu Zhao and Chun Yang

100

part of present work. All these functions are obtained through integrations and are found to have a combination of the incomplete Gamma function.

F ( n, x ) = ∫ x

−1+

1 n

sinh ( x ) dx

1 1 − 1 1 1 = [Γ ( , x) − (− x) n x n Γ ( , − x)] 2 n n

(A1)

F1 ( n, x) = ∫ F ( n, x) dx 1 −1− 1 1+ 1 1 1 1 1 = [ x n Γ(1 + , − x)( − x ) n − Γ(1 + , x) + xΓ( , x) − ( − x) −1/ n x1+1/ n Γ( , − x)] 2 n n n n

(A2) F2 ( n, x) = ∫ F ( n, x) cosh ( x) dx n +1 n

1 1 (− x) −1/ n {x n Γ( , −2 x) + n 1 1 1 1 1 1+ 1 1 1 + (− x ) n Γ( , 2 x)2 n [ n(− x 2 ) n − x n Γ( , − x) sinh ( x) + (− x) n Γ ( , x) sinh ( x)]} n n n

=2

−1−

(A3) 1+

F3 (n, x) = ∫ x

1 n

cosh ( x) dx

1 1 − 1 1 1 = − {(− x) n x n Γ(2 + , − x) + Γ(2 + , x)} 2 n n

H ( n, x) = ∫ x

−2 +

1 n

(A4)

sinh 2 ( x) dx

−1+

1

1 nx n 1 1 = [ + 2−1/ n ( − x) −1/ n x1/ n Γ( −1 + , −2 x) − 2−1/ n Γ( −1 + , 2 x)] 2 −1 + n n n (A5)

Electrokinetic Flows of Non-Newtonian Fluids …

101

H1 (n, x) = ∫ H (n, x) dx 1 1 1 1 = {2−1/ n (− x) −1/ n x1/ n [ xΓ(−1 + , −2 x) + Γ( , −2 x)] − n 2 2 n n 2 x1/ n 1 1 1 − 2−1/ n [ xΓ(−1 + , 2 x) − Γ( , 2 x)] + } n n −1 2 n

(A6)

H 2 (n, x ) = ∫ H (n, x ) cosh ( x )dx 1+ n 1 − − +2 1 1 1 1 = {(− x )−1/ n x1/ n [Γ(−1 + , − x ) − 3 n Γ(−1 + , −3x ) + 2 n Γ (−1 + , −2 x ) sinh ( x )] − 8 n n n 1 −1+ n − +2 1 1 1 − [−Γ (−1 + , x ) + 3 n Γ (−1 + ,3x ) + 2 n Γ (−1 + , 2 x )sinh ( x )] + n n n 1 2n 1 1 [−(− x )−1/ n x n Γ( , − x ) − Γ ( , x )]} + ( −1 + n ) n n

(A7) where Γ (α , ξ ) is the incomplete Gamma function.

In: Microfluidics: Theory and Applications ISBN 978-1-61668-570-6 Editor: I. A. Kuznetsov, pp. 103-134 © 2010 Nova Science Publishers, Inc.

Chapter 3

MICROFLUIDIC ELECTROCHEMILUMINESCENT DETECTION DEVICES WITH CAPILLARY ELECTROPHORESIS K. M. Muzyka1 and M. M. Rozhitskii2 Kharkiv National University of Raio Electronics, Laboratory of Analytical Optochemotronics, Departament of Biomedical Electronic Apparatus and Systems, Kharkiv, Ukraine

ABSTRACT The purpose of this chapter is to describe the applications of microfluidic principles for creation of capillary electrophoresis (CE) chip devices with electrochemiluminescent (ECL) detection and to discuss the problems associated with their interfacing and the approaches that have been developed to surmount them. Basic methodology, instrumentation, unique features, and specific futures of microfluidic ECL detection with CE are discussed. ECL assay detection types, such as direct and indirect analysis, coreactant use including are considered. Publications data of scientific centers, involved in development of CE chips with ECL detection from the start of microfluidic CE/ECL joint

1 2

Email: [email protected] Email: [email protected]

104

K. M. Muzyka and M. M. Rozhitskii technique are summarized. Basic tendencies of this direction future developments are covered.

ABBREVIATION Auxiliary electrode 2-(2-aminoethyl)-1methylpyrrolidine capillary electrophoresis chemiluminescence electrochemiluminescence electrochemistry electroosmotic flow fluorescence indium tin oxide limit of detection microfluidic chip nanoparticles printed circuit board photomultiplier tube polydimethylsiloxane pulse electrolysis quantum dots reference electrode signal-to-noise ratio spectroscopic tri-n-propylamin tris(bipyridil)ruthenium (II) working electrode

(AE) (AEMP) (CE) (CL) (ECL) (EC) (EOF) (FL) (ITO) (LOD) (MFC) (NPs) (PCB) (PMT) (PDMS) (PE) (QDs) (RE) (S/N) (SP) (TPrA) (ТBR) (WE)

INTRODUCTION The important features in many areas of human activity are definition of substances with low and trace concentration (medicine, ecology, food industry etc.). Successful execution of such assays is possible using novel methods and technologies, which have very low limit of detection (LOD), high selectivity,

Microfluidic Electro-chemiluminescent Detection Devices …

105

very small dimensions and price such as lab-on-a-chip technologies and devices. Capillary electrophoresis (CE) technologies [1] in microfluidic chips (MFC) have received considerable interest in analytical and bioanalytical applications recently due to their promising features such as efficient separation capabilities, high selectivity, short analysis time, very small consumption of samples and reagents, portability, low assay and instruments cost. On the other hand, the low sample consumption requires very sensitive detection systems to be used. In general CE with fluorescent (FL), electrochemical (EC), spectroscopic (SP) and chemiluminescent (CL) detectors is well-known. However from the middle of this decade the efforts of many researches are directed to implementation of electrochemiluminescent (ECL) techniques in microchip CE devices [2-7]. ECL is a type of CL in which the light-emitting process is initiated by electrochemical reactions. The electrochemically generated reactants undergo subsequent electron transfer reactions with production of excited molecules, which emit light. In comparison with the most widespread EC, luminescent (especially FL) and spectroscopic analytical methods ECL have very low LOD [8]. Interest to combine ECL analysis with separation methods appeared lately. Notwithstanding that the term “electrochemiluminescence” was born in 1929, the first papers devotes to ECL phenomena in CE devices in chip format have appeared at the beginning of 2000. In contrast to microanalytical systems with FL, CL, EC, SP detection [9-13], microfluidic ECL devices are on a stage of development (Figure 1.).

Figure 1. The diagram of microfluidic devices with FL, SP, EC, CL and ECL detectors since 2000 till 2010.

106

K. M. Muzyka and M. M. Rozhitskii

UNIQUE FEATURES OF ECL The analytical characterization of the detection methods included the evaluation of the following parameters: sensitivity, linear concentration range, LOD, specificity and selectivity, response time, stability and reproducibility. Among the top analytical features sensitivity, specificity and selectivity are the most important in microfluidic applications. That is caused by a very low amount at analyte concentration in submililiters MFD sample volume. Sensitivity as analytical property can be ascribed as “the ability to detect (qualitative analysis) or determine (quantitative analysis) of small amounts of an analyte in a sample”. Sensitivity is defined by several parameters that establish the minimum concentration that can be detect or determinated. Thus, the LOD is the analyte concentration which produces a signal that can be statistically distinguished from the blank signal. Analytical method is “selective” if it can determine simultaneously several components independently from each other. On the other hand, an analytical method is “specific” if only one component (“species”) can be determined independently from all the other components which give no analytical signal in this case. As an analytical technique, ECL possesses several advantages over other most widespread methods (see Table 1). The ECL technique is very sensitive, since very low light levels can be measured (e.g., by single photon counting methods). ECL in comparison with photoexcitation has the advantage that a excitation light source is not used, so scattered excitation light and interferences by luminescent impurities emission are not the problems. This results in low limits of detection in ECL assays as well as excellent selectivity in comparison with FL detection. ECL prosesses are much selective and possibilitics to computerization (automation) than other chemiluminescent methods, since the electrochemical excitation allows temporal and spatial control over light-emitting reactions. ECL can be used to detect either the emitting species (which often serve as labels) or a coreactant (see later) that ameliorate specificity [8]. Notice that selectivity and spesificity in various detecting methods can be improved by different ways. So, the basic manner of selection in EC analysis are electrolysis mode and potential change, in FL and spectroscopy – change of optical radiation wave length; in CL – change of reagents and media.

Table 1. Competitive matrix of ECL and widespread detection methods on a microfluidic chip

TYPES OF TRANSDUSERS

ATTENDANT PROBLEMS AND CHARACTERISTICS OF DETECTIONS METHODS

ECL CL FL SP EC

Analytical signal time control

Varying electrode potentials possibility

Background currents of electrolysis

Scattered excitation light and luminescent impurities

Analytical signal position control

Light source

Optical filters

Photodetector sensitivity

Improving sensitivity

Selectivity increasing

Sensitivity decreasing

Sensitivity decreasing

Multianalyteanalysis

Cost and dimensions increasing

Cost and dimensions increasing

Necessity of registration of faint and superfaint light flux

Yes Difficult Yes Yes Yes

Yes No No No Yes

No No No No Yes

No No Yes Yes No

Yes No Yes Yes Yes

No No Yes Yes No

No No Yes Yes No

Yes Yes No No No

Disadvantages Advantages

Tables color scheme (dark corresponds to disadvantage, light – advantage) indicated unique characteristics of ECL analysis in comparison with related detecting methods

K. M. Muzyka and M. M. Rozhitskii

108

ECL ANALYSIS TYPES The Direct and Indirect ECL Analysis In the first case the objects of detection are substances (analytes), which possess its own ECL, in the second one – all substances that can influence ECL of reactants substances [14-17]. Direct analysis has some specific issues, such us: • • •

own ECL (sufficient quantum efficiency of luminescence); EC activity of the determined components; presence in solution or on the electrode surface electrone donors or acceptors which in electron transfer reactions with the electrooxidized or electroreduced analyte create electron-excited molecules.

These requirements satisfy significant numbers of electrochemilumiphores of various chemical classes [8, 18, 19]. Tris(bipuridil)ruthenium(II) (Ru(bpy)32+) (ТBR) complex and its derivatives [20] are the most used due to rather high (in comparison with luminol) selectivity with a high ECL emitters output in reactions with determined components.

ECL with Cоreactant Due to high energy of annihilation, narrow working potentials window for conventional metal electrodes and poor solubility in water of organic compounds method of classical ECL excitation known for aprotic solutions can not in general be utilized in water and bioliquids assays due to narrow working potential range in water. Among possible solution of mentioned problem is using so called coreactants. They can produce highly reducing or oxidizing radicals species that can react with an oxidized or reduced ECL luminophore to generate excited states. Unlike ion annihilation ECL, in which electrolytic generation of both the oxidized and reduced ECL precursors is required, ECL with coreactants can be generated with one direction potential scan of an electrode in a solution containing luminophore species (“emitter”). To be a good ECL coreactant, a number of criteria need to be met, namely [21]:

Microfluidic Electro-chemiluminescent Detection Devices … • • • •

109

solubility, electrochemical activity, reactions participation, absence of background ECL, etc.

Among those the reduction/oxidation properties of the coreactant play important role. The coreactant should be easily oxidized or reduced either at electrode processes or by the luminophore species near the electrode and undergo a rapid chemical reactions to form an intermediate with subsequent productions luminophore excited state [14]. The most widespread соreactant for ECL emission are oxalates and tertiary amines [14, 21, 22]. The TBR/tri-n-propylamin (TPrA) ECL system, a good example of an oxidative–reductive system, has been extensively studied [23].

BASIC METHODOLOGY OF CE/ECL ANALYSIS IN MICROFLUIDICS The majority of CE/ECL MFC embodies two essential stages. On the first one the substances of interest are isolated from the matrix of the sample by separation; on the second one identification and quantitatively determination of analytes are taking place. Whereas CE realized in central body structure of MFC with the microscale channels which define the selectivity of analytes determination, detection element of MFC characterizes sensitivity of analysis. Capillary electrophoresis is the motion of charged particles relative to the surrounding liquid due to an imposed external electric field. In contrast, electroosmosis is the electrically driven motion of a liquid relative to the walls of the solid borders. Separation in electrophoresis is based on differences in solute mobility that is the cause of probes dividing on individual zones. The efficiency of this separation process can be described, by analogy with chromatography, in terms of the number of theoretical plates. The difference necessary to resolve two zones is dependent on the zone length; the latter is strongly dependent on the dispersive processes. Dispersion should be controlled because it increases zone length and the mobility difference necessary to achieve separation. Dispersion, or spreading of the solute zone, result from differences in solute velocity within that zone, and can be defined as the baseline peak width.

K. M. Muzyka and M. M. Rozhitskii

110

The velocity of migration, u, of particular ionic species in buffer solution is given by:

u = μa E ,

(1)

where

μa = μe + μeof – the apparent solute mobility, (m2.V-1.s-1); μe, μeof – electrophoretic and electrosmotic mobilities accordingly; E – applied electric field, V/m.

Thus, due to difference in electrophoretic particles mobility probes zones reach detection zone at different time. The received peaks sequence presents so called electrophoregramme. The typical MFC contains two microchannels that crossed at an intersection while the ends of the microchannels are connected with reservoirs (Figure2). Chemical analysis can be divided into steps of sampling, sample preparation, separation, detection, and data processing. Because of the minute sample volumes and masses used in microfluidic CE, LOD are often relatively poor. However, by using preconcentration techniques, improved concentration LOD and separation properties can be realized for biological analytes. Generally, MFC operations can be divided onto the following stages. During the first stages the sample is injected by applying high voltage between reservoirs 3 and 4 (Figure 2, a). A very small plug (of pl volumes) can be introduced, with dimensions approximately the same as the channel width (Figure 2, b). For electrophoresis, the second stage, the reactant is injected electrophoretically into the intersection resulting in a sample plug shiffing. The separation is performed by applying a voltage between reservoirs 1 and 2. The sample plug then moved into another reservoir 2 through detection zone (Figure 2, c). At this zone, ECL is emitted and is detection is performed as the third stages. There are analytical characteristics of ECL analysis such as spectral and integral intensity linked by certain dependence to analyte concentration, providing possibility to define it by measurement of light intensity.

Microfluidicc Electro-chem miluminescent Detection Deevices …

1 111

(A)

(B)

(C) Figure 2. MFC design (Reservoirs: ( 1 – buffer solutionn, 2 – sample inj njection waste, 3 – sample, and 4 – waste) and electrophorresis procedures, (b) schematicc representationn of sample injection (doublle-T structure foor sample injecttion), (c) pullingg back to form a p and (d) sepparation. The caationic analyte (1) has the highhest mobility sample plug, followedd by the neutralss (n) and two annionic analytes (2, 3).

112

K. M. Muzyka and M. M. Rozhitskii

ECL INSTRUMENTATION An ECL system often includes two sub-units [8]: optical part for luminescence detection and electrochemical one that initiates the ECL reaction.

Optical Part of ECL Mode An efficient and accurate light detection is a common need to all analytical luminescence methods. The size of the required photosensitive area of the detector varies depending on the application. Usually high detection sensitivity of photodetector is required, but the detector must also be able to register many orders of magnitude higher light fluxes than those corresponding to LOD of the analytes in question [24].

Photomultiplier Tube The photomultiplier tube (PMT) is the most popular device adopted for collection of the luminescence signal for its high sensitivity. Channel Photomultipliers Channel photomultipliers (or microchannel plate photomultipliers) are performing better than ordinary PMTs. In these devices, electrons from the photocathode pass through a narrow semi-conductive channel. Multiple secondary electrons are emitted each time the electrons on their way to the anode hit the inner wall of the curved channel. This effect occurs multiple times along the path, leading to an avalanche effect with a gain exceeding 108. These detectors also have extremely low dark current (much lower than the traditional PMTs), but they require special high voltage sources due to the needed very high operating voltage of 2400–3000V. Diode array Relatively large and rather costly PMT detection system which also requires high voltage to achieve maximum sensitivity is mainly suitable for laboratory analysis. To overcome the size and cost limitations of PMTs, in some of the ECL set-ups PMT are replaced with small device, such as a silicon PIN diode that could be operated at low voltages. Semiconductor

Microfluidic Electro-chemiluminescent Detection Devices …

113

photosensitive devices can occasionally be used when sensitivity is not the first consideration [25].

Electrochemical Part of ECL Mode Having the character of an EC method, ECL detection has great potential for miniaturization — the electrode can be directly formed on the chip by a thin-film photolithographic procedures. The well developed micro-electronics industry makes it possible to effectively produce integrated microfluidic-ECL systems with high reproducibility at rather low cost. ECL detector cell typically consists of three electrodes – the working, auxiliary, and reference – placed in electrolyte solution. ECL assay is controlled by application of the desired potential to the working electrode. Types and characteristics of ECL electrodes are shown in table 2. Table 2. Types and characteristics of ECL electrodes TYPE OF ELECTRODES

Working (WE)

Reference (RE)

Auxiliary (counter) (AE)

CHARACTERISTICS

WE is the electrode in an EC/ECL system on which the reaction of interest is occurring, therefore electric current flow through this electrode and subsequent ECL emission constitute the analytical signals. The minimum requirements are reasonable degrees of electrochemical inertness and electrical conductivity, optical transparency and minimal rate quenching of excited states near electrode surface RE is an electrode which has a stable and well-known electrode potential in a given electrolyzed solution. The high stability of the electrode potential is usually reached by employing a redox system with constant (buffered or saturated, e.g. saturated calomel, Ag/AgCl, etc.) concentrations of each participants of the redox reactions AE is needed to provide a current path that closes the electrical circuit

114

K. M. Muzyka and M. M. Rozhitskii

The chemical composition/structure of the WE can vary considerably, and a proper selection is often critical to achieving success in a given application. Wide-spread WE include metals, such as Pt, Au, or Hg, semiconductors (indium tin oxide (ITO)), carbon. Notice that the optical transparency property makes ITO electrodes an ideal electrode material for ECL analysis. In addition, the surface of these electrode materials can also be modified chemically by a number of different approaches in order to optimize performance for a specific EC and ECL process [26]. For example, the charge, polarity, porosity, and specific chemical and biochemical reactivity can be adjusted by the addition of appropriate functional groups, self-assembled mono- and multilayers, and polymer coatings [27]. A representative three-electrode ECL detection configuration and a simplified potentiostat circuit are shown in Figure 3. The WE and RE are placed in the amplifier feedback loops so that the voltage difference between them must match an adjustable external voltage source. Because the composition of the RE is designed to maintain its potential at a constant value, the WE potential is just the external voltage compared to or “versus” the specific RE. The RE is connected to a high input impedance buffer amplifier, which serves to limit the current flow through it to a negligible level, which in turn serves to keep RE composition unchanged and its potential at its starting value. The oxidation or reduction reactions occurred when an electroactive analyte at the WE produces current flow that is then converted into ECL signal trough appropriate sequence of reactions.

Figure 3. Scheme of EC part of microfluidic ECL mode.

Microfluidic Electro-chemiluminescent Detection Devices …

115

The three-electrode systems are generally used, but for simple applications the two-electrode system can be utilize also. At that AE function is combine with RE and lead to design simplicity. However two-electrode systems have not full potential control possibility. In the simple EC system it is not a problem, but for high precision analysis three-electrodes systems will be needed. If two-electrodes systems is been used then several criteria ought to be accomplished, such as electrodes surfaces ratio, its geometry etc. [8].

Working Electrode Placement in CE/ECL Detection There are many diverse separating procedures and many different identifying instruments, and each particular combination demands a unique interface. CE/ECL combination is difficult but, nevertheless, can be successfully achieved by the use of some cleverly designed interfaces. In the microfluidic system, the separation electric field has profound effects on EC detection as well as the conventional CE. Three different approaches have been developed for this purpose and are depicted in Figure 4. These approaches are termed end-channel, in-channel, and off-channel detection [28, 29].

Figure 4. Four configurations for aligning the WE in EC as well as ECL detection that facilitate isolation of the detector from the separation voltage. Reprint from ref. [28] with modification.

As well as in electrochemical mode, the placement of the WE in CE/ECL detection systems can be divided into three different categories, off-channel, end-channel and in-channel detection (see table 3).

116

K. M. Muzyka and M. M. Rozhitskii

Table 3. Special set-up and characteristics of three modes of CE/ECL detection systems Set-up

Characteristics END-CHANNEL Potential shifts at the WE due to the Set-up with precise electrode remaining separation field; positioning (just beyond the end of band-broadening and decreasing the separation channel in the CE detector response due to diffusion; buffer reservoir) is needed; higher background currents and less the WE is positioned tens of sensitivity than the other detection micrometers from the exit of the modes; separation channel; decreasing separation efficiency due to the separation voltage could be grounded through the potentiostat to the analyte diffusion that occurs in the area between the exit of the separation prevent electronics damaging channel and the WE; suitable for single-use, disposable microchips. OFF-CHANNEL The decoupler effectively shunts the Placing the WE directly within the separation voltage to ground; separation channel; Electroosmotic flow (EOF)generated an additional decoupler is needed, before the decoupler; e.g. electrically conducting decoupler can absorb hydrogen evolved “fracture” and thin crack, opening on the capillary electrophoretic ground or hole created in the capillary, electrode; channel wall is in a short distance technical difficulties to implement. from the exit. IN-CHANNEL Limitation due to bubble generation in Placing the WE directly within the the channel; separation channel; decreasing the electrochemical signalan electrically isolated potentiostat to-noise (S/N) ratio; is needed. background current generated at the surface of the WE; elimination of the band-broadening.

Microfluidic Electro-chemiluminescent Detection Devices …

117

MICROFLUIDIC DEVICES PROBLEMS It should be notice that LOD reduction in joint CE/ECL devices can be reached not only by improving detection, but also by peaks broadening reduction.

The Major Factors Loss of Separation Efficiency The major factors bringing loss of separation efficiency (peaks broadening) as follows [30]). (1) Radial diffusion. In contrast to high-pressure liquid chromatography where diffusion has three components, in CE only longitudinal diffusion is taking place. (2) Dispersion due to Joule heating, At worst Joule heating could result in convective overturning that obliterates all bands or production of microbubbles, resulting in complete blockage of EOF and current (“vapor lock”). In modern CE systems with capillary diameters less than 100 μm, such failures are rare; however, Joule heating can be a significant source of band broadening [31]. Elimination of these effects is possible by changing of a capillary diameter and concentration of the buffer reducing and/or applying the buffer with low ionic conductivity and, also, capillary cooling. (3) An electric dispersion (electric field local infringement). The phenomenon is caused by the big distinction of conductivity in the buffer and in a zone of test, heterogeneity of zeta-potential [31, 32] (because of a roughness of a surface of the channel, heterogeneity of the channel material, absorption on walls, a variation of рН buffers values) [33]. Undesirable effects are eliminated, using special coverings and high concentrated buffers and also providing the minimal roughness and the maximal uniformity of the channel [34]. Electric field local infringement is possible due to bubbles formation (H2 and O2) as a result of the electrolysis. This effect can be suppressed partly by all solution filtration, WE positioning concerning the separation channel end, change of рН [35]. (4) Dispersion due to channel curvature [30]. This can be observed, for example, using of the serpentine channel (this is frequently done for the chip contraction).

118

K. M. Muzyka and M. M. Rozhitskii (5) Probes zone passive transport due to great EOF values [30]. Adjusting buffer pH and channel surface properties leads to EOF suppressing.

Precise design and operational control of the liquid matrix separation process are keys to performance of microfluidic CE analysis. In a typical separation experiment there are a number of parameters affecting the system performance. These can be broadly categorized as parameters determined by the chemical peculiarities such as mobility of ions, diffusion coefficients, sample concentration and EOF mobility; detector controlled parameters such as width of the detector, sensitivity and noise level; and some other parameters such as width of the sample plug, distance to the detector, electric field intensity. In general the designer has little control over the first set of these and the second are typically limited by the equipment available. One example from the engineering point of view the fluid flow mechanism occurring in CE separation channel was analyzed in [36]. Onedimensional model of electrokinetic sample motion was developed to simulate the separation process of sample with aminoacids (tryptophan, tyrosine, proline, methionine) that migrate in buffer solution through straight polymethylmethacrylate separation channel within microfluidic chip under different conditions [36]. On this basis optimal channel size, chip material, applied voltage and pH dependencies on separation efficiency were predicted. This is important to know before chip fabrications and assay conducting.

CE/ECL Key Specific Futures In CE/ECL technology combination of two complex measuring systems must be achieved with neither influencing upon the other. For CE as well as ECL detection some problems that have affected analytical mode are peculiar. The main problem for CE is the large numbers effect that results in enhanced dispersion, which limits the efficiency of separation. There are also hardships for ECL with respect to right choose of ECL excitation modes. Linear sweep voltammetry is a voltammetric method where the current at a WE is measured while the potential between the WE and a RE is swept linearly in time. Oxidation or reduction of species is registered as a peak or trough in the current signal at the potential at which the species begins to be

Microfluidic Electro-chemiluminescent Detection Devices …

119

oxidized or reduced. The main disadvantage of this technique is the rapid electrode fouling. For example, after the oxidation process, some analytes will form a thin layer on the electrode, decreasing the “effective area” and therefore, the signal. Coating or mechanical treatment the electrodes can solve this problem, however, they require disassembly of flow cells and lengthen time between runs. Therefore, the use of pulse electrolysis (PE) is a way to overcome problems associated with electrode fouling (see Table 4). PE can be considered as a derivative of linear sweep voltammetry, with a series of regular voltage pulses superimposed on the potential linear sweep or stair steps. The system for this measurement is usually the same as that for the standard voltammetry. The potential of WE is repeated for about 5 to 100 milliseconds, changing the final potential, and a constant difference is kept between the initial and the interlevel potential. PE is an excellent method for detection of analytes that can foul electrodes, espessially in microfluidic applications at water and bioliquids assay. WE cleaning and reactivation potentials can be identified by performing cyclic voltammogram and looking for the electrode oxidation and reduction potentials. It is worth noting that the electrode requires the possibility of being oxidized and reduced without significant loss of its material. For this reason, gold and platinum have been extensively used in a variety of media [37], unlike electrodes with diamond-like films (see later). In PE, the WE electrode is first cleaned at a high potential, then reactivated at a opposite potential dissolving the surface contaminant, and finally used to determinate the analyte at a moderate positive potential. Moreover, in MFD with ECL detection PE using is very promosing because of the following two features: (1) in these measurements, the effect of the charging (non faradic) current can be minimized, so higher assay efficiency can be achieved; (2) faradaic current can be extracted, so electrode reactions can be studied more precisely. Due to the advantages and versatility of PE, many different variations of the detection waveform is be described [37]. However, two problems are still present. First by, specific (and sometimes more expensive) instrumentation is required. Second by, there are more variables to adjust than using linear sweep voltammetry, so longer optimization time than simple electrochemical techniques is required. For CE/ECL systems, however, there are key specific issues that must be taking into account. The main features are presented in Table 4 [21].

K. M. Muzyka and M. M. Rozhitskii

120

Table 4. Specific futures of ECL microfluidic detection PROBLEMS

Redox potential change of the analyte species due to its affect on electrical current under high electric field Poor sensitivity and reproducibility (e.g. too short distance would block the ECL reactant diffusion and too long one would lower the analyte concentration at the WE) Band broadening at short length (e.g., a few centimeters) of detection capillary ECL excitation mode right choose

LOD LOWERING FACTORS The elimination or reduction of the CE current from the Faradic current of analyte/reactant species

DECISION

Use of CE decoupler

The distance between the end of the capillary and the WE needs to be optimized

Necessary high precision of the alignment or arrangement of the WE against the detection zone and the photodetector should be rached

Rising pressure in the capillary due to electric field gradient absence in the detection zone

Length of the detection zone should be optimized

WE fouling during electrolysis

PE mode or disposable electrodes should be used

KEY RESEARCH FINDINGS CE/ECL microdevice concept was first demonstrated by Manz et al. in 2001 [38]. In their article, a microfabricated glass device was characterized for micellar electrokinetic chromatography of Ru(bpy)33+ and Ru(phen)33+. In this design the legs of a ‘U’-shaped Pt electrode placed across the separation channel act as the WE and AE. The required potential difference for the ECL reaction is supplied by the electrophoretic electric field. Indirect detection of three amino acids (proline, valine, and phenylalanine) was also reported. Later, Crooks’ group reported one- [39], two- [40], and three-channel microfluidic sensors [41] that could detect redox reactions indirectly using anodic coreactant ECL [i.e., TBR/TPrA system].

Microfluidic Electro-chemiluminescent Detection Devices …

121

As shown on Figure 5, a [41], a one-channel microfluidic device, incorporating either one or two electrodes, is able to detect electrochemical processes at the cathode and provide information about them via light emission at the anode.

(A)

(B)

(C) Reprinted from [41]. Figure 5. Scheme of microfluidic devices.

Similarly, a two-electrode, two-channel microfluidic device (Figure 5, b) can be used in the same manner, but with complete chemical separation of the detection and reporting (light-emission) functions. The one- and two-channel methods, however, permit detection only of analytes that can be reduced when the Ru(bpy)32+/TPrA system is used. This problem was overcome by using a multichannel approach, such as the three-channel configuration (Figure 5, c). In this case, channel 1 houses the cathode and a flowing solution of an electroactive molecules [e.g., Ru(NH3)63+] that can be easily reduced. A solution containing Ru(bpy)32+/TPrA flows in channel 2 and the oxidizable analyte of interest is present in channel 3. Both of these channels share a common anode. When a sufficiently large potential is applied between the

122

K. M. Muzyka and M. M. Rozhitskii

cathode and anode, Ru(NH3)63+ is reduced to Ru(NH3)62+, whereas Ru(bpy)32+ and TPrA are oxidized. In the absence of a redox-active analyte in channel 3, maximum light emission is observed at the anode. However, when an oxidizable analyte is present in channel 3, it competes with the ECL reactions in channel 2 to provide electrons for the cathodic reduction in channel 1. This results in a decrease in light intensity from the electrode in channel 2, indicating the presence of the target analyte. Reducible analytes can be detected also with this three-channel method. For example, if only buffer is present in channel 3 and the analyte is present in channel 1, then the device is essentially identical to the two-channel device shown in Figure 5, b. All three proposed schemes are based on charge balance between the anode and cathode. In other words, the cathode current must be equal to the anode one. Accordingly, there is a correspondence between the number of electrons consumed at the cathode and the ECL photons emitting at the anode. An integrated ITO electrode-based Ru(bpy)32+ ECL detector for a polydimethylsiloxane (PDMS) microchip CE device (Figure 6) was first reported by Wang’s group in 2003 [42]. The microchip CE-ECL system described in this article consists of the PDMS layer containing separation and injection channels and an electrode plate with an ITO electrode fabricated by a photolithographic method. The PDMS layer was reversibly bound to the ITO electrode plate, which greatly simplified the alignment of the separation channel with the WE and enhanced the photon-capturing efficiency. Moreover, the high separation electric field had no significant influence on the ECL detector, and decouplers for isolating the separation electric field in the microchip CE-ECL system were not needed. Proline was selected to test the microchip device with a LOD of 1.2 μM (S/N = 3) and a linear range from 5 to 600 μM.

Reprinted from ref [43]. Figure 6. Schematic of the detection region enlargement in microchip CE-ECL device. Distance between the separation channel outlet and the ITO electrode is 30 μm.

Microfluidic Electro-chemiluminescent Detection Devices …

123

The transparency of the ITO material makes these electrodes superior to the more widely used platinum-based electrodes, because of sufficient diminishing of ECL quanta losses. This microchip CE-ECL system can be used for the rapid analysis of lincomycin. Under the optimized conditions, the linear range was obtained from 5 to 100 μM with correlation coefficient of 0.998. The LOD of 3.1 μM was obtained for lincomycin in the standard solution. This method was used for determination of lincomycin in the urine matrix without pretreatment. The LOD of 9.0 μM was obtained. In dual detection scheme TBR was used as an ECL reagent as well as a precursor (in the formation of Ru(bpy)33+) for the EC detection [4]. In the Ru(bpy)32+-ECL process, Ru(bpy)33+ was generated and then reacted with analytes resulting in an ECL emission and a great current enhancement in EC detection due to the catalysis of Ru(bpy)33+. In the experiments, dopamine and three kinds of pharmaceuticals, anisodamine, ofloxacin, and lidocaine, were selected to validate this dual detection strategy. Typically, for the EC detection of dopamine with the presence of Ru(bpy)32+, an approximately 5 times higher S/N can be achieved than that without Ru(bpy)32+ during the simultaneous EC and ECL detection of a mixture of dopamine and lidocaine using CE separation. The results indicated that this dual EC and ECL detection strategy could provide a simple and convenient detection method for analysis of more types of analytes in CE separation than in the sole EC or ECL detection alone. Additionally more information about detected analytes could be achieved using this assay mode. The solid-state ECL detector was fabricated by immobilizing TBR into an Eastman AQ55D-silica-carbon nanotube composite thin film on an ITO electrode [44]. After being made by a photolithographic method, the surface of the ITO electrode was coated with a thin composite film through a micromolding in capillary technique using a PDMS microchannel with the same pattern as an ITO electrode. Then the TBR was immobilized via ion exchange by immersing the ITO electrode containing the thin film in TBR aqueous solution. The whole system was built by reversibly sealing the TBRmodified ITO electrode plate with a PDMS layer containing electrophoresis microchannels. The results indicated that the present solid-state ECL detector displayed good durability and stability in the microchip CE-ECL system. Proline was selected to test the microchip device with a LOD of 2 μM (S/N=3) and a linear range from 25 to 1000 μM. In comparison with the CE-ECL of TBR in aqueous solution, while the CE microchip with solid-state ECL

124

K. M. Muzyka and M. M. Rozhitskii

detector system had the same assay sensitivity, a much lower TBR consumption and a high integration of the whole system were obtained. The microchip CE/ECL system utilizes tertiary amine derivative, 2-(2aminoethyl)-1-methylpyrrolidine (AEMP) determination [45]. The system was characterized by the interaction between biotin and avidin. A 4.5 cm microchannel was used to separate the mixture of AEMP and biotinylated AEMP. The results indicated that AEMP has a good reactivity to the analytes containing carboxyl group with a similar ECL efficiency to TPA. Under optimal condition, the LOD (based on 3 S/N) of AEMP was 2.7 μM. The system was also validated by the reaction between biotin and avidin. The calculated binding ratio between avidin and biotin based on the present method was 4.4 [45]. Printed circuit board (PCB) based technology to integrate ECL analysis in microfluidic systems was used in [46]. PCB gold macro- (10 mm2) and micro(0.09 mm2) electrodes and two ECL microfluidic devices were designed, fabricated and tested with luminol ECL detection. Potential modulation was performed between 0.7 and 0 V vs. Ag/AgCl for luminol oxidation, thus giving rise to on/off ECL responses in the presence of hydrogen peroxide. Synchronous detection was adopted to allow weak ECL signal recovery at a very low S/N ratio. The LOD obtained with the two ECL microfluidic devices was 50 nM and 100 nM H2O2 for macroelectrodes and microelectrodes, respectively. Similar results were published in [47, 48]. The Japanese researchers have developed ECL microfluidic system for amino acids determination [49, 50]. The ECL chip for mercury ions definition with LOD of 66 nМ at a current 1.85 pА is created [51]. There are also other publications on microfluidics ECL systems [52, 53]. A microfluidic cell designed to transport and mix two different solutions on the chip and generate ECL using TBR as luminophor and amino acids as coreactants was presented recently by Hosono et al. [54]. The fabricated system consisted of the PDMS substrate with a flow channel structure and three-electrode systems formed on a glass substrate that resulted in the microfluidic transport and the ECL emission. An amino acid and a reagent solution containing TBR were transported by electrowetting by applying a negative potential to gold WEs formed along the flow channels. The two solutions were then mixed in the mixing channel using an additional gold WE formed between the electrodes for transport. ECL was generated by applying a potential to a platinum WE formed in the mixing channel. The ECL from lproline, l-lysine, l-leucine, l-valine, and l-histidine was registered.

Microfluidic Electro-chemiluminescent Detection Devices …

125

Solid state ECL detectors coupled with microchip CE by immobilizing TBR into either Eastman AQ55D–silica– carbon nanotube composite thin film on a patterned ITO electrode or zirconia-Nafion composite on a glassy carbone disk electrode have been constructed and used for the detection of proline and pharmaceuticals of tramadol, lidocaine, and ofloxacin [55]. Detailed fabrication of the other solid state ECL detector can be found in ref [56].

GEOGRAPHY OF SCIENTIFIC CENTRES Scientific centers engaged in research and application of ECL phenomenon in a combination with CE are concentrated in East Asia, basically, in China. In the first turn it is Prof. E. Wangs group (Lab. of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry of the Chinese Academy of Sciences), but development and applications of microfluidic ECL devices are done in France, Great Britain, Canada, USA and some other countries. The basic centers on these subjects are summarized in table 5.

POSSIBLE FUTURE DEVELOPMENTS (BASIC TENDENCIES) At present CE/ECL systems have been developed. Unification of separation and ECL detecting techniques are obvious aspects of this phenomenon. Further improving of CE/ECL system and its miniaturization are possible by using of micro- and nanotechnologies opportunities. ECL analytical characteristics can be improved by development and use of new electrode materials. Electrodes covered by diamond-like films exhibit enhanced sensitivity compared to a commonly used thick-film carbon detector. Their advantages in the ECL analysis in comparison with electrodes from carbon or platinum are caused by stability, reproducibility of surface condition (frequently without regular recycling processing), wide area of ideal polarization, low size of electrolysis background currents. Electrode stability is caused by inertness to adsorption of reagents and products of reaction, and also by tolerance to dissolved oxygen in water solutions (both alkaline and acid). The first articles on the given subjects have appeared in 2003 [57-59] though electrochemical research of doped diamond began about twenty years ago by Yu. Pleskov [60]. Doped diamond electrodes are promising candidates

126

K. M. Muzyka and M. M. Rozhitskii

as the electrode material for ECL chip devices with a wide ECL potential region and high stability [60-63]. Table 5. Scientific centers involved in development of CE chips with ECL detection Scientific centers Dalian University of Technology: Laboratory of Precision and Non-Traditional Machining Technology. Key Laboratory for Micro/Nano Technology and System of Liaoning Province. Chinese Academy of Sciences State (Changchun): Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry Korea University (Seoul): Departments of Biomicrosystem Technology. South Korea University of Tsukuba: Graduate School of Pure and Applied Sciences Université Claude Bernard (Lyon): Institut des Nanotechnologies de Lyon (INL), CNRS; Institut de Chimie et Biochimie Moléculaires et Supramoléculaires (ICBMS), Laboratoire de Génie Enzymatique et Biomoléculaire (LGEB), Institut des Nanotechnologies de Lyon (INL), CNRS. Univ. Sci./Technol. (Lille), CNRS Université de Montréal Institute for Microstructural Sciences, National Research Council Canada, Département de Chimie University of Manchester School of Chemical Engineering and Analytical Science. Institute of Research in the Applied Natural Sciences (Luton) University of Neuchatel: Institute of Microtechnology, Sensors, Actuators and Microsystems Laboratory Petroleum University of Technology University of Athens: Laboratory of Analytical Chemistry, Department of Chemistry Cornell University (Ithaca): Department of Biological and Environmental Engineering. Innovative Biotechnologies International, Inc., Grand Island

Countries

China

Korea Japan

France

Canada Great Britain Switzerland Iran Greece USA

Microfluidic Electro-chemiluminescent Detection Devices …

127

It is necessary to remember that new electrode materials use will demand new fabrication technique development, resolving the problems of studying materials properties and their application. Another possible tendency is electrochemilluminophore immobilization on electrodes surface (creation of solid-state devices). It allows to save expensive reagents, to use electrodes repeatedly, to simplify the system design and miniaturized it, to expand analytes range due to possibility of use insoluble in water electrochemiluminophores [27]. The first article on TBR immobilization was published in 1980 [64], works in this direction continue to be published until now [65-67], and development of micro- and nanotechnologies contributes to "solid-state" detectors tendency strengthening. Scientific and technology progress of this direction can be achieved in matrixes, new electrochemiluminophore-reagents, and also immobilization techniques development [68-72]. Nanodimension materials, as alternative to traditional electrochemiluminophores – one more step in improvement of analytical characteristics of chip-using ECL detectiong. Semiconductor nanocrystals or quantum dots (QDs) are bright and photostable materials with broad excitation spectra but narrow Gaussion type emission spectra superior in comparison to conventional organic fluorescent dyes. Furthermore, their emission positions are tunable in a wide emission range from ultraviolet to near infrared depending on QDs radius due to the quantum confined effect. ECL study of semiconductor nanoparticles (NPs) (also known as nanocrystals, QDs) was first reported in 2002 for Si NPs, where ECL was generated from both annihilation and coreactant oxalate and persulfate systems in MeCN [73]. Moreover, the increasing numbers of publications indicate a growing interest to this direction, and its wide analytical perspectives.

CONCLUSION In spite of the fact that microfluidic principles with ECL detection yet have not so widespread in comparison, for example, with fluorescent methods (see Figure 1), their numerous positive characteristics allow to solve difficult analytical tasks quite simple and effectively. The opportunity of decrease in LOD, increase in sensitivity of the analysis with the help of new types of electrodes and luminophores also raises interest to this area [74]. Despite the obvious CE/ECL potential advantages, presence of experimental samples yet has not led to commercial products creation.

128

K. M. Muzyka and M. M. Rozhitskii

However it is connected most probably not to the analytical reasons, but to technical, organizational and information one. It is reasonable to suggest, that the stage of industrial samples creation and their active use in practice now begins; there is no doubts in the broadest opportunities of ECL applications. That is why in the nearest future the commercialization of ECL microfluidic-chips and their wide utilization will be more and more obvious.

ACKNOWLEDGMENT We express our sincere appreciation for financial support from Scientific and Technology Centre in Ukraine (Projects ## GE 77, 4180, 4495).

REFERENCES [1]

[2]

[3]

[4]

[5]

[6]

Kraly, J.; Fazal, M.A.; Schoenherr, R.M.; Bonn, R.; Harwood, M.M.; Turner, E.; Jones, M.; Dovichi, N.J. (2006) Bioanalytical Applications of Capillary Electrophoresis. Anal. Chem. 78. 4097-4110. Redha Z.M., Baldock S.J., Fielden P.R., Goddard N.J., Brown B.J., Haggett B.G., Andres R., Birch B.J. (2009) Hybrid Microfluidic Sensors Fabricated by Screen Printing and Injection Molding for Electrochemical and Electrochemiluminescence Detection. Electroanalysis. 21. 422 – 430. Yin X.-B., Wang E. (2005) Capillary Electrophoresis Coupling with Electrochemiluminescence Detection: a review. Analytica Chimica Acta. 533. 113-120. Qiu, H.; Yin, X.B.; Yan, J.; Zhao, X.; Yang, X.; Wang, E. (2005) Simultaneous Electrochemical and Electrochemiluminescence Detection for Microchip and Conventional Capillary Electrophoresis. Electrophoresis. 26. 687-93. Yan, J.; Yang, X.; Wang, E. (2005) Electrogenerated Chemiluminescence on Microfluidic Chips. Anal. Bioanal. Chem. 381. 48–50. Du, Y. & Wang, E. (2007) Capillary Electrophoresis and Micro chip Capillary Electrophoresis with Electrochemical and Electrochemiluminescence Detection. J. of Separation Science. 30. 875890.

Microfluidic Electro-chemiluminescent Detection Devices … [7]

[8] [9]

[10]

[11]

[12]

[13]

[14] [15]

[16]

[17]

[18]

[19]

129

Yin, X.; Yang, X.; Wang, E. (2005) Some Aspects in Capillary Electrophoresis-Electrochemical/Electrochemiluminescence with Microchip Technique. Progress in Chemistry. 17. 181-185. Rozhitskii, N.N. (1992) Electrochemiluminescence Analysis of Solutions. J. Anal. Chem. V.47. P. 1765-1793. Kuswandi, B.; Nuriman, Huskens, J., Verboom, W. (2007) Optical Sensing Systems for Microfluidic Devices: A review. Analytica Chimica Acta. 601. 141-155. Walter, R.; Vandaveer, IV; Pasas-Farmer, S.; Fischer D.J., Frankenfeld C.N., Lunte S.M. (2004) Recent Developments in Electrochemical Detection for Microchip Capillary Electrophoresis. Review. Electrophoresis. 25. 3528–3549. Matysik F.-M. (2008) Review. Advances in Amperometric and Conductometric Detection in Capillary and Chip-based Electrophoresis. Microchim Acta. 160. 1–14. Xu, J.-J.; Wang A.-J.; Chen, H.-Y. (2007) Electrochemical Detection Modes for Microchip Capillary Electrophoresis. TrAC - Trends in Analytical Chemistry. 26. 125-132. Sassa, F.; Morimoto, K.; Satoh W.; Suzuki, H. (2008) Electrochemical Techniques for Microfluidic Applications. Electrophoresis. 29. 1787– 1800. Bard A.J. (2004). Electrogenerated Chemiluminescense. N.-Y.: Marcel Dekker, 342 p. Rozhitskii N.N., Bykh A.I., Krasnogolovetz M.A. (2000) Electrochemical Luminescence.-Kharkiv, Kharkiv Technical University of Radio Electronics, 2000.-320 p. (In Russian). Bard, A. J., Faulkner, L.R. (2001) Electrochemical Methods: Fundamentals and Applications: John Willy & Sons, INC, 2nd ed. NY, P.833. Kukoba, A.V.; Bykh, A.I.; Svir, I.B. (2000) Analytical Applications of Electrochemiluminescence: An overview. Fresenius' Journal of Analytical Chemistry. 368. 439-442. Haapakka, K.E.; Kankare, J.J. (1982) The Mechanism of the Electrogenerated Chemiluminescence of Luminol in Aqueous Alkaline Solution. Analytica Chimica Acta. 138. 263-275. Wilson, R.; Akhavan-Tafti, H.; DeSilva, R.; Schaap, A.P. (2001) Comparison Between Acridan Ester, Luminol, and Ruthenium chelate Electrochemiluminescence. Electroanalysis. 13. 1083-1092.

130

K. M. Muzyka and M. M. Rozhitskii

[20] Dickson, J.A., Ferris, M.M.; Milofsky, R.E. (1997) Tris(2,2′bipyridyl)ruthenium(III) as a Chemiluminescent Reagent for Detection in Capillary Electrophoresis. HRC Journal of High Resolution Chromatography. 20. 643-646. [21] Miao, W. (2008) Electrogenerated Chemiluminescence and its Biorelated Applications. Chemical Reviews. 108. 2506-2553. [22] Forster, R.J.; Bertoncello, P.; Keyes, T.E. (2009) Electrogenerated Chemiluminescence. Ann. Rev. Anal. Chem. 2. 359–385. [23] Miao, W.J.; Choi J.P.; Bard A.J. (2002) Electrogenerated Chemiluminescence 69: The Tris(2,2_-bipyridine)ruthenium(II), (Ru(bpy)2+3)/Tri-n-propylamine (TPrA) System Revisited — A New Route Involving TPrA Cation Radicals. J Am Chem Soc. 124. 1447814485. [24] Kulmala, S.; Suomi, J. (2003) Current Status of Modern Analytical Luminescence Methods. Analytica Chimica Acta. 500. 21-69. [25] Pal, S.; Kim, M.J.; Tak Y.K.; Kwon H.T.; Song J.M. (2009) Interdigitated Microelectrode Array-coupled Bipolar Semiconductor Photodiode Array (IMEA-PDA) Microchip for On-chip Electrochemiluminescence Detection. Biomedical Microdevices. 11. 971-980. [26] Auroux P.A.; Iossifidis D.; Reyes D.R.; Manz A. (2002) Micro Total Analysis Systems. 2. Analytical Standard Operations and Applications. Anal. Chem. 74. 2637-2652. [27] Zholudov, Y.; Snizhko, D.; Kukoba, A.; Bilash, H.; Rozhitskii, M. (2008). Aqueous Electrochemiluminescence of Polycyclic Aromatic Hydrocarbons Immobilized into Langmuir-Blodgett Film at the Electrode. Electrochimica Acta. 54. 360-363. [28] Walter, R.; Vandaveer, V.; Pasas-Farmer, S.; Fischer, D. J.; Frankenfeld C.N., Lunte S.M. (2004) Recent Developments in Electrochemical Detection for Microchip Capillary Electrophoresis. Review. Electrophoresis. 25. 3528–3549. [29] Xu J.-J.; Wang, A.-J.; Chen H.-Y. (2007) Electrochemical Detection Modes for Microchip Capillary Electrophoresis. Trends in Analytical Chemistry. 26. 125-132. [30] Ghosal, S. (2006) Electrokinetic Flow and Dispersion in Capillary Electrophoresis. Annu. Rev. Fluid Mech. 38. 309–338. [31] Xuan, X.; Xu, B.; Li D (2004), Electroosmotic flow with Joule heating effects. Lab on a Chip. 4. 230-236.

Microfluidic Electro-chemiluminescent Detection Devices …

131

[32] Hu, Y; Werner, C; Li, D. (2003) Electrokinetic Transport through Rough Microchannels. Anal. Chem. 75. 5747-5758. [33] Schure, M; Lenhoff, A. (1993). Consequences of wall adsorption in capillary electrophoresis: Theory & simulation. Anal. Chem. 65. 3024– 3037. [34] Herr, E.; Molho, J.I.; Santiago J.G.; Mungal, M.G.; Kenny, T.W. (2000) Electroosmotic Capillary Flow with Nonuniform Zeta Potential. Anal. Chem. 72. 1053-1057. [35] Go, R.S.; Nnemeyerb J.K.; Karst U. (2008) Quantitative analysis by microchip capillary electrophoresis – current limitations and problemsolving strategies. Analyst. 133. 167–174. [36] Muzуka K.; Rozhitskii M. (2009) An Approach to Optimize the Design of Microfluidic Chip for Electrophoretic Separations. Microchemica Acta. 164. 257-262. [37] Garcia, C.D.; Henry, C.S. (2005) Coupling Capillary Electrophoresis and Pulsed Electrochemical Detection. Electroanalysis. 17. 1125-1131. [38] Arora, A.; Eijkel, J.C.T.; Morf, W.E.; Manz, A. (2001) A Wireless Electrochemiluminescence Detector Applied to Direct and Indirect Detection for Electrophoresis on a Microfabricated Glass Device. Anal. Chem. 73. 3282-3288. [39] Zhan, W.; Alvarez, J.; Crooks, R.M. (2002) Electrochemical Sensing in Microfluidic Systems Using Electrogenerated Chemiluminescence as a Photonic Reporter of Redox Reactions. J. of the Amer. Chem. Soc. 124. 13265-13270. [40] Zhan, W.; Alvarez, J.; Crooks, R.M. (2003) A Two-channel Microfluidic Sensor that Uses Anodic Electrogenerated Chemiluminescence as a Photonic Reporter of Cathodic Redox Reactions. Anal. Chem. 75. 313318. [41] Zhan, W.; Alvarez, J.; Sun, L.; Crooks, R.M. (2003) A Multichannel Microfluidic Sensor that Detects Anodic Redox Reactions Indirectly Using Anodic Electrogenerated Chemiluminescence. Anal. Chem. 75. 1233-1238. [42] Qiu, H.; Yan, J.; Sun, X.; Liu, J.; Cao, W.; Yang, X.; Wang, E. (2003) Microchip Capillary Electrophoresis with an Integrated Indium Tin Oxide Electrode-Based Electrochemiluminescence Detector. Anal. Chem. 75. 5435-5440. [43] Zhao, X.; You, T.; Qiu, H.; Yan, J.; Yang, X.; Wang, E. (2004) Electrochemiluminescence Detection with Integrated Indium Tin Oxide Electrode on Electrophoretic Microchip for Direct Bioanalysis of

132

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

K. M. Muzyka and M. M. Rozhitskii Lincomycin in the Urine. J. of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences. 810. 137-142. Du, Y.; Wei, H.; Kang, J.; Yan, J.; Yin, X.B.; Yang, X.; Wang, E. (2005) Microchip Capillary Electrophoresis with Solid-state Electrochemiluminescence Detector. Anal. Chem. 15. 7993-7997. Yin, X.-B.; Du, Y.; Yang, X.; Wang, E. (2005) Microfluidic Chip with Electrochemiluminescence Detection Using 2-(2-aminoethyl)-1methylpyrrolidine Labeling. J. of Chromatography A. 1091. 158-162. Pittet, P,; Lu, G.N.; Galvan, J.M.; Ferrigno, R.; Blum, L.J.; LecaBouvier B. (2007) PCB-based Integration of Electrochemiluminescence Detection for Microfluidic Systems. Analyst. 132. 409-411. Pittet, P.; Lu, G.-N.; Galvan, J.-M.; Ferrigno, R.; Stephan, K.; Blum, L.J.; Leca-Bouvier B. (2008) A Novel Low-cost Approach of Implementing Electrochemiluminescence Detection for Microfluidic Analytical Systems. Mater. Science and Engineering C. 2008. 28. 891895. Pittet, P.; Lu, G.-N.; Galvan, J.-M.,; Ferrigno, R.; Blum, L.J.; Leca-Bouvier, B.D. (2008) PCB Technology-Based Electrochemiluminescence Microfluidic Device for Low-Cost Portable Analytical systems. IEEE Sensors Journal.8. 565-571. Hosono, H.; Satoh, W.; Fukuda, J.; Suzuki, H. (2007) Microanalysis System Based on Electrochemiluminescence Detection. Sensors and Materials. 19. 191-201. Hosono, H.; Satoh, W.; Suzuki, H. (2006) Integrated Microfluidic System Based on Electrowetting and Its Application to Amino Acid Sensing Based on Electrochemiluminescence. IEEJ Transactions on Sensors and Micromachines. 126. 200-201. Schäfer, H.; Lin, H.; Schmittel, M.; Böhm, M. (2008) A Labchip for Highly Selective and Sensitive Electrochemiluminescence Detection of Hg2+ Ions in Aqueous Solution Employing Integrated Amorphous Thin Film Diodes. Microsystem Technologies. 14. 589-599. Ding, S.-N.; Xu, J.-J.; Chen, H.-Y. (2006) Microchip Capillary Electrophoresis Coupled with an End-column Electro chemiluminescence Detection. Talanta. 70. 403-407. Qi, N.; Luo, Y.; Yang, F. (2007) Low-temperature Bonding for Electrochemiluminescence Chemiluminescence Microfluidic Chip with Glass. Nanotechnology and Precision Engineering. 5. 331-334.

Microfluidic Electro-chemiluminescent Detection Devices …

133

[54] Hosono, H.; Satoh, W.; Fukuda, J.; Suzuki, H. (2007) On-chip Handling of Solutions and Electrochemiluminescence Detection of Amino Acids. Sensors and Actuators, B: Chemical. 122. 542-548. [55] Ding, S.-N.; Xu, J.-J.; Zhang, W.-J.; Chen, H.-Y. (2006) Tris(2,2′bipyridyl)ruthenium(II)-Zirconia-Nafion Composite Modified Electrode Applied as Solid-state Electrochemiluminescence Detector on Electrophoretic Microchip for Detection of Pharmaceuticals of Tramadol, Lidocaine and Ofloxacin. Talanta. 70. 572-577. [56] Ding, S.-N.; Xu, J.-J.; Chen, H.-Y. (2005) Tris(2,2′-bipyridyl) ruthenium(II)-zirconia-Nafion Composite Films Applied as Solid-state Electrochemiluminescence Detector for Capillary Electrophoresis. Electrophoresis. 26. 1737-1744. [57] Honda K., Yoshimura M., Rao T.N., Fujishima A. Electrogenerated Chemiluminescence of the Ruthenium Tris(2,2′)bipyridyl/amines System on a Boron-doped Diamond Electrode // J. of Physical Chem. B. 2003. 107. 1653-1663. [58] Marselli, B.; Garcia-Gomez, J.; Michaud, P.-A.; Rodrigo, M.A.; Comninellis, Ch. (2003). Electrogeneration of Hydroxyl Radicals on Boron-doped Diamond Electrodes. J.of the Electrochem. Soc. 150 D79D83. [59] Honda, K.; Yamaguchi, Y.; Yamanaka, Y.; Yoshimatsu, M.; Fukuda, Y.; Fujishima, A. (2005) Hydroxyl Radical-related Electrogenerated Chemiluminescence Reaction for a Ruthenium Tris(2,2)bipyridyl/coreactants System at Boron-doped Diamond Electrodes. Electrochimica Acta. 51. 588-597. [60] Pleskov, Yu.V.; Sakharova A.Ya.; Krotova, M.D.; Bouilov, L.L.; Spitsyn, B.V. (1987). Photoelectrochemical Properties of Semiconductor Diamond. J. Electroanal. Chem. 228. 19-27. [61] Fujishima A., Einaga Y., Rao T.N. and Tryk D.A. (2005), Editors, Diamond Electrochemistry, Elsevier-BKC, Tokyo, Japan. P. 586. [62] Wang J.; Gang C.; Madhu P.C.; Fujishima A.; Tryk D.A.; Shin D. (2003) Microchip Capillary Electrophoresis Coupled with a BoronDoped Diamond Electrode-Based Electrochemical Detector. Anal. Chem. 75. 935-939. [63] Cvacka J.; Quaiserova V.; Jinwoo P.; Show Y.; Muck A.; Swain G.M. (2003) Boron-Doped Diamond Microelectrodes for Use in Capillary Electrophoresis with Electrochemical Detection. Anal. Chem. 75. 26782687.

134

K. M. Muzyka and M. M. Rozhitskii

[64] Rubinstein, I.; Bard, A.J. (1980) Polymer films on electrodes. 4. Nafioncoated Electrodes and Electrogenerated Chemiluminescence of Surfaceattached Ru(bpy). J. of the Amer. Chem. Soc. 102. 6641-6642. [65] Wei, H.; Wang, E. (2008) Solid-state Electrochemiluminescence of tris(2,2′-bipyridyl) Ruthenium. TrAC - Trends in Anal. Chem. 27. 447459. [66] Tao, Y.; Lin, Z.; Chen, X.; Chen, X. (2008) Solid-State Electrochemiluminescence on Tris(2,2′-bipyridine) Ruthenium(II) Modified Electrode. Progress in Chemistry. 20. 362-367. [67] Qian, L.; Yang, X. (2008) One-step Immobilization of Tris(2,2′bipyridyl)ruthenium(II) via Vapor-surface Sol-Gel Deposition Towards Solid-state Electrochemiluminescence Detection. Anal. Chim. Acta. 609. 210-214. [68] Ide, K.; Fujimoto, M.; Kado, T.; Hayase, S. (2008) Increase in Intensity of Electrochemiluminescence from Cell Consisting of TiO2 Nanohole Array Film. J. of the Electrochem. Soc. 155. B645-B649. [69] Lin, Z.; Chen, Z.; Liu, Y.; Wang, J.; Chen, G. (2008) An Electrochemiluminescent Biosensor for Uric Acid Based on the Electrochemiluminescence of Bis-[3,4,6-trichloro-2-(pentyloxycarbonyl) -phenyl] Oxalate on an ITO Electrode Modified by an Electropolymerized Nickel Phthalocyanine Film. Analyst, 133. 797-801. [70] Wang, X.; Yun, W.; Dong, P.; Zhou, J.; He, P.; Fang, Y. (2008) A Controllable Solid-state Ru(bpy)32+ Electrochemiluminescence Film Based on Conformation Change of Ferrocene-labeled DNA Molecular Beacon. Langmuir. 24. 2200-2205. [71] Dai, H.; Wu, X.; Wang, Y.; Zhou, W.; Chen, G. (2008) An Electrochemiluminescent Biosensor for Vitamin C Based on Inhibition of Luminol Electrochemiluminescence on Graphite/poly (methylmethacrylate) Composite Electrode. Electrochimica Acta. 53. 5113-5117. [72] Qi, B.; Yin, X.-B.; Du, Y.; Yang, X. (2008) Unique Electro chemiluminescence Behavior of Ru(bpy)32+ in a Gold/Nafion/Ru(bpy)32+ Composite. Materials Letters. 62. 458-461. [73] Ding, Z.; Quinn, B.M.; Haram, S.K.; Pell, L.E.; Korgel, B.A.; Bard, A.J. (2002) Electrochemistry and Electrogenerated Chemiluminescence from Silicon Nanocrystal Quantum Dots. Science. 296. 1293-1297. [74] Muzyka K.M.; Rozhitskii M.M. (2010) Capillary Electrophoretic Systems in Electrochemiluminescent Analysis. J. of Anal. Chem. 6. in print.

In: Microfluidics: Theory and Applications ISBN 978-1-61668-570-6 Editor: I. A. Kuznetsov, pp. 135-148 © 2010 Nova Science Publishers, Inc.

Chapter 4

MICROFLUIDIC VALVES WITHOUT DIAPHRAGMS: HYDROGEL VALVES AND PDMS-BASED ROTARY SELECTION VALVES Steffen Howitz1, Frank Baudisch1, Frank-Ulrich Gast1, Andreas Richter2, Andreas Grodrian3, Gunter Gastrock3, and Josef Metze3 1

GeSiM mbH, 01454 Grosserkmannsdorf, Germany Technische Universität Dresden, Institute of Semiconductors and Microsystems, 01069 Dresden, Germany 3 Institut für Bioprozess- und Analysenmesstechnik e. V., 37308 Heilbad Heiligenstadt, Germany 2

ABSTRACT Microfluidic networks consisting of more than one channel must often be controlled by valves. External valves are sturdy, but low dead volumes require internal valves. Many present-day valves contain membranes (that are, e.g., driven by air pressure) that become leaky by overpressure or particles, and simple microfluidic valves control only a single channel at a time. We describe here two different new approaches: a hydrogel valve, which is robust, and a rotary valve, which is versatile. Both valves are pressure-stable. 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

136

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al. 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.

INTRODUCTION Around 1990, a novel concept, micro total analysis system or µTAS, emerged [1-3], in which chemical sensors were placed in small channels that were fabricated with methods used in microelectronics. Since then, this concept has matured into systems suited for such diverse tasks as chemical synthesis [4], chemical sensing, or processing and analysis of biological material [1,2]. Whereas the first systems were controlled by constantly driven syringe pumps and so contained mostly simple branches to mix and separate fluids, the growing complexity, e.g. for the preparation of biological samples, required the integration of actuator systems such as mechanical microvalves. Today a multitude of valve designs exists [5]. Many valves are based on a membrane or diaphragm [5-10, further citations in 11 and 17] that is either actively actuated using different methods or passively driven by the fluid, e.g. in a check valve. (Such microvalves that are not actively driven, e.g. those on gyrating discs, shall not be discussed here.) The idea of a diaphragm was also applied in the field of soft lithography that came up in the late 1990s [11-16]. Here, a channel structure is replicated in an elastomeric material such as PDMS by casting the gelous material on a negatively structured master, followed by crosslinking of the gel. Its charm is the easy handling and its cost-effectiveness, making this technology available to standard laboratories. One of the first attempts to create a monolithic PDMS valve was taken by the Quake group in 2000 [14], which led to diverse variants and their integration into complicated microreactor networks [examples: 11,15,17].

Microfluidic Valves without Diaphragms …

137

Generally, membrane-based valves suffer from a complicated design and the problem of leakage, especially when particles are present in the fluid. PDMS channels, on the other hand, can collapse when the geometry is not optimized [11], and their pneumatic control is complex. Moreover, membrane valves can only be opened or closed and so represent the equivalent of a 2/2 valve in the macroscopic world, “2/2” meaning two inlets/outlets and two valve positions (i.e. open and closed). More complicated switching regimes, like in valves with single input and multiple output (distribution or selection valves) or in HPLC injection valves, are unattainable by this design or can only be realized by combining several 2/2 valves, resulting in a larger size and more control inputs. We looked for alternative switching modes and came up with two novel concepts, both of which allow small dead volumes: a sturdy, normally closed, 2/2-type hydrogel valve that, due to its soft actuator material, has excellent sealing properties, and a rotary valve with PDMS rotor disc that is flexible and uses marginal space.

HYDROGEL VALVE Originally, hydrogel actuators were used in chemomechanical devices that are sensitive to pH, temperature, and alcohol concentration [18-21]. The functional principle of pH- and temperature-dependent hydrogels and their use in chemical sensors and valves are discussed in [22,23]. We were among the first to employ the thermally induced swelling of a hydrogel for displacementfree valve actuation, practically without moving parts [24-26]. Others, applying the same idea, photopolymerized a monolithic block of poly(Nisopropylacrylamide) (PNIPAAm) inside a microchannel to create a hydrogel microvalve in situ [27]. We, on the other hand, have devised a particle actuator in an etched silicon chamber. A block of PNIPAAm is polymerized from Nisopropylacrylamide in the presence of N,N’-methylenebisacrylamide as crosslinker [25]. The chamber space is then filled to about 50% with hydrogel particles that were ground from the dry PNIPAAm block and sieved. Use of particles significantly reduces switching time, as the volume change is diffusion-limited [26]. The volume phase transition temperature (Tt) of PNIPAAm, at which the volume increases or decreases by about one order of magnitude, is approximately 34 °C and thus biocompatible. At room temperature, the hydrogel is completely hydrated and the valve thus normally closed [23,25,26] whereas at higher temperature the weak hydrogen bonds

138

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al.

between polar solvent (water) and slightly hydrophobic hydrogel break down, leading to a collapse of the polymer chains [22]. Addition of polar solvents (methanol, ethanol, 1- and 2-propanol, acetone) or salt leads to a decrease of Tt [28], which allows measurement of their concentrations, higher solvent concentrations even to complete loss of the volume increase [28], but this is generally reversible. Interestingly, hydrogel swelling returns to normal for short-chain alcohols above concentrations of 80% (v/v) (data not shown). The actuator chamber is manufactured by methods of microsystem technology. Swelling of the hydrogel particles is induced by filling the valve with a pH-neutral, water-based liquid; the valve is ready for operation after a few swelling/shrinking cycles [22,26] and very durable as long as no salts are trapped within [23,28]. Manufacturing of a separate device (as opposed to photopolymerization of a gel plug in a microchannel [25,27]) has the advantage that it can be individually controlled and tested before using it in a complicated channel system. In the first designs, the actuator chamber (version number PV5: 500 µm x 500 µm x 200 µm) was placed in a horizontal flow channel [25,26] and closed by a glued-on Pyrex glass cover containing a microstructured thin-film heater; anodic bonding of glass and silicon would destroy the hydrogel. Loss of hydrogel particles was avoided by barriers surrounding the actuator chamber so that the space between silicon and glass became very narrow. Inlets and outlets of the valve channel were through-etched on the bottom layer, so to mount tubes to the front side, a second, anodically bonded silicon/glass manifold was glued onto the rear of the chip, which was therefore not autoclavable. This valve was placed in a PEEK housing that allowed the connection of tubes via UNF (Unified National Fine Thread) fittings (data not shown). Its larger size limits its use in small microsystems, but it can be also directly attached to microfluidic manifolds. A more compact design (PV6) was realized by directing the flow vertically through the chip [28]; this is currently the system of choice and commercially available (Figures 1-2). Direct photopatterning of hydrogel pads in microchannels [24,25,27] is currently not offered to customers. To manufacture the 500 µm x 500 µm x 300 µm large actuator chamber, bottom and the cover are thinned by a two-side process on 4-inch silicon wafers: wet etching with KOH followed by through-etching of 25 to 40 µm wide pores via deep reactive ion etching (RIE) using the classical ASE-Bosch process [29]. Platinum heater and sensor (Figure 1A,C) are structured via liftoff technology [25]. Filling of the chamber with dry hydrogel powder (particle size 125±25 µm; Figure 1A) is done manually; an automatic method using an

Microfluidic Valves without Diaphragms …

139

XYZ robotics (GeSiM Nano-Plotter 2.1) with a newly developed small powder dispenser and camera-aided target finding is under development. The chamber is closed by flip-chip gluing of top and bottom silicon layers, followed by wire bonding of heater and temperature sensor to a printed circuit board (PCB) as shown in Figure 1B. A scanning electron micrograph of the actuator chamber is presented in Figure 1C. Loss of hydrogel material is prevented by making the pore size about three times smaller than the particle size [28] and avoiding pressures larger than 4·105 Pa (4 bar), although the valve can withstand pressures in water of at least 6·105 Pa. If the pressure limit is not exceeded, no leakage occurs, even if dust particles are trapped in the valve, due to the softness of the hydrogel [23,26].

Figure 1. Small hydrogel microvalve, PV6, with 500 µm x 500 µm x 300 µm large actuator chamber and an overall chip size of 5 mm x 5 mm. (A) The picture shows the top sieve plate with meandering temperature sensor (left) and the bottom sieve plate filled with hydrogel particles and Pt heater (right). (B) PV6, glued together and wirebonded to a PCB. The socket to connect the device to external control electronics is not yet soldered on the chip. (C) Scanning electron micrograph of a cut hydrogel valve showing the KOH-etched actuator chamber, RIE-etched vertical sieve pores (diameter 35 µm, length 80 µm), and dehydrated hydrogel particles. (D) Eight PV6 valves glued to a microfluidic manifold with PCB. The variation of the direction of the flow through the central cuvette over 360° by combining different inlets and outlets is used for hydrodynamic experiments in which elongated molecules like DNA are aligned in the flow field. The hydrogel valves are cooled by a heat sink placed at their top (not shown).

140

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al.

The microvalve is connected to an external microcontroller-based electronics (Figure 2A) with I2C bus via a small socket soldered on the PCB. To heat and thus to open the valve, a voltage of 3.5 V DC at 150 mW is applied. To avoid overheating and hence long closing times, the temperature is continuously measured by the integrated T-sensor and the heating power is adjusted via pulse-width modulation to values down to 50 mW. Closing of the valve is passive by switching off the heater and thus depends on the ambient temperature. Such valves exhibit opening times of around one second and closing times of a few seconds [26]. Active cooling by a Peltier element can further decrease the shut-off time, but the effect is too small to justify the effort. Direct attachment of hydrogel microvalves to a microfluidic system is shown in Figure 1D. The small size of the hydrogel valve makes it possible to place it inside a UNF fitting as a microfluidic control element, e.g. in GeSiM’s modular PDMS microflow cell for microscopes, MicCell (Figure 2A). The rationale behind the MicCell [30] is a reusable microscope chip holder with constant, softwarecontrolled fluidic periphery and standardized chip-to-world interface, while the microfluidic channel system can be arbitrary: it is molded in a special casting station on the surface of a customized silicon master chip. The microchannel layer is then taken out, capped with a coverslip and immediately ready to use. A hydrogel valve in a fitting screwed into an opening of a Tbranch system controls the loading of test substances into the main channel. Normally, buffer or medium is aspirated from a reservoir through the other branch of the T-junction. Test compounds are filled into the small cavity at the top of the valve fitting (Figure 2A), and to inject them, the flow is stopped, the main inlet blocked by an external valve, and the substance sucked into the channel after opening the hydrogel valve. To clean the dead volume in the bore downstream of the valve, a cleaning solution can be applied to the opening of the valve and sucked through an extra channel that bypasses the main microfluidic channel (K-type channel setup) [30]. Injection of fluids can also be achieved by connecting the hydrogel valve with a tube (Figure 2B). Due to the attachment to a PCB and socket, PV6 cannot be autoclaved. We have therefore devised PV7 whose platinum heater and sensor films are directly contacted without intermediary PCB (Figure 2B). Six of these valves were built into a six-well plate in which hybridoma cells were grown on the milliliter scale to overexpress monoclonal antibodies. The valves controlled the automatic withdrawal of microliter amounts of sample to measure protein expression and hence optimize growth conditions [31], which requires an autoclavable system. As the cells were maintained at 37 °C, the hydrogel valve

Microfluidic Valves without Diaphragms …

141

had to be cooled by a heat sink chilled by either a Peltier element or cooling water to maintain it in its normally closed position.

A

B

Figure 2. Small hydrogel microvalve in a UNF 1-4/28 fitting. Sealing is achieved by an O-ring. (A) PV6 assembled in a MicCell [30] with 22 mm x 22 mm large fluid system, resting in a blue metal support that is placed in an inverted microscope. The valve is marked by the arrow. The flow control unit is in the background; it contains a syringe pump operating in suction mode (left), hydrogel control electronics and external 2/2 valve (middle), and 4 in / 1 out selection (turning) valve to switch between fluids and stopping the flow (right). (B) PV7 in fitting, disassembled. The four electrical pins contact the actuator chip without intermediary PCB, hence this microvalve is autoclavable (> 10 times for 20 min at 125 °C / 2 bar).

142

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al.

ROTARY VALVE More complicated valves with several inlets and outlets exist in the macroworld. The hydrogel valve is a simple on/off (2/2) valve, so such a valve must be built into each fluidic channel to be controlled, much like the many valves in a soft lithography rotary micropump [32]. A 2/3 valve, with one inlet and two outlets between which the flow switches, already needs two hydrogel valves; if all two-out-of-three combinations should be available, all three channels must be controlled. A rotary (turning) valve would be such a solution, as not only many pipes can be hooked up, but inlet and outlet channels can also be interconnected in a flexible way, e.g. as in a distribution/selection valve (one input, many outputs or vice versa). Generally, such a valve consists of a stator containing all fluidic inlets and outlets and a rotor containing one or more channels that bridge individual channels in different ways upon rotation (Figure 3). The rotor face must be in sealing contact with the stator surface. Like the hydrogel valve, rotary valves such as HPLC injection valves can switch flows with very little fluid displacement. A microfluidic rotary valve consisting of micromachined parts has been proposed, but its design is complicated [33]. Our experience in silicon master micromachining and PDMS molding [30] led us to propose that the rotor material should be PDMS [11-17], as this material is easy to form by simple gel casting and curing, durable, biocompatible, and elastomeric, i.e. it can be used not only as rotor, but also as sealing surface. Our first attempt to turn the valve used a pneumatic drive (Figure 4A), but better control is obtained with a encoder-controlled, brushless electrical gear micromotor (Dr. Fritz Faulhaber GmbH & Co. KG, Schönaich, Germany) with an outer diameter of 6.5 mm, a switching power of 1 W, and a switching time of 250 milliseconds (Figure 4D). The PDMS rotor has a diameter of 3.5 mm and is mounted on a shaped piece (Figure 4B) for force transmission. Due to outside contacts, the overall space required for the valve is about 10 mm x 13 mm. We soon found out that the rotor material becomes swollen (see also [34]), thus reducing its durability due to friction between PDMS and silicon. Sputtering with Cr 5 nm/Au 20 nm (Figure 4C) increased the lifetime of the rotor from about 1000 to 3000 cycles. If the rotor is worn out, it can be replaced by casting a new one and coating it with metal, all other parts are reused. The stator can be of any material, as long as it is plane; we have used stators of PMMA (Figure 4A) and silicon (Figure 5).

Microfluidic Valves without Diaphragms …

143

Figure 3. Examples of a rotary valve layout, with channels in black, through-etched holes in grey, and rotor in orange. The rotor channel in the first motor position is depicted as solid red line, alternative positions after turning the valve are shown as dashed red lines.

Figure 4. PDMS rotary valve. (A) Pressure-driven valve in a PMMA manifold. (B) Tip of the actuator mechanics in a fitting that is screwed into a tapped hole with O-ring sealing. (C) PDMS rotor, metal-coated with Cr/Au, with single channel (length 1400 µm, width 500 µm, depth 300 µm), used. (D) CAD sketch of the electric motor drive.

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al.

144

A patent has been applied for this new valve type, which is rather new so that only few applications exist. As an example, Figure 5 shows a chip to automatically culture cells in a microsystem. Cells are grown on a silicon sieve plate with ~ 25 µm through-etched holes through which medium is passed; the sieve is placed in a holder attached to the microchannel via two holes (Figure 5A, arrow). The rotary valves (Figure 5B) allow to change between different media and to clean the channel, in the forward and backward direction.

A

B

Figure 5. Experimental setup with three rotary valves for cell culturing in a flowthrough system. (A) Silicon/glass fluidic channels, seen through the glass cover. Each of the through-etched triple holes in the stator regions is contacted by a PDMS valve rotor on the rear. Single holes on the chip margins are inlets and outlets, connected to reservoirs containing medium. (B) Ready to use setup, with servo motors at the bottom and microfluidic system at the top.

Microfluidic Valves without Diaphragms …

145

CONCLUSION We have outlined here two different concepts for microfluidic valves. Which valve to use depends on the application: if a sturdy, pressure-resistant and particle-tolerant 2/2 valve is required that is practically displacement-free (as the gel material simply takes up the surrounding process medium) and may even be autoclaved, use the hydrogel valve. Its drawback is its somewhat slow switching time and that the hydrogel comes in contact with the liquid, thus possibly adsorbing solutes. Nevertheless, blocking of live cell suspensions without harming the cells has been successfully demonstrated (data not shown). If complicated switching processes (such as a selector or an HPCL injection valve), high flexibility, and short opening and closing times need to be realized on a small footprint of approx. two square centimeters, use the PDMS rotary valve. We expect both valves to be valuable for microfluidics, biotechnology, and chemical and medical engineering. It has come to our attention that turning valves similar to ours, but using different material, have also been developed at the Institut für Mikrotechnik Mainz (IMM), Germany.

ACKNOWLEGDMENTS We are grateful to Prof. Dirk Kuckling, Universität Paderborn, who was involved in the initial experiments with the hydrogel valve. This work has been supported by grants from the German Ministry of Education, Science and Technology (BMBF), projects “Protein Processing Platform (3P)” (FKZ 16SV1832, Förderkonzept Mikrosystemtechnik 2000+, 2003 - 2006), and “µCryoLab” (FKZ 16SV2382, Rahmenprogramm Mikrosysteme, 2006 2010).

REFERENCES [1]

[2]

Manz, A; Graber, N; Widmer, HM. Miniaturized total chemical analysis systems: A novel concept for chemical sensing. Sens Actuators B Chem, 1990, 1, 244-248. Manz, A; Fettinger, JC; Verpoorte, E; Lüdi, H; Widmer, HM; Harrison, DJ. Micromachining of monocrystralline silicon and glass for chemical

146

[3]

[4] [5] [6]

[7]

[8]

[9]

[10]

[11]

[12] [13]

[14]

[15] [16]

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al. analysis systems: a look into next century’s technology or just a fashionable craze? Trends Anal. Chem., 1991, 10, 144-149. Metze, J; Gastrock, G; Howitz, S; Wegener, T; Bürger, M. Utilization of microsystem-parts in sampling system of bioreactors. Anal. Meth. Instr. Spec. Issue µTAS’96, 1996, 234. Ehrfeld, W; Hessel, V; Löwe, H. Microreactors. New Technology for Modern Chemistry. Weinheim: Wiley-VCH Verlag GmbH; 2000. Oh, KW; Ahn, CH. A review of microvalves. J. Micromech. Microeng., 2006, 16, R13-R39. Shoji, S; van der Schoot, B; de Rooij, N; Esashi, M. Smallest dead volume microvalve for integrated chemical analyzing systems. Proc. Transducers’91, San Francicso 1991, 1052-1055. Français, O; Dufour, I. Enhancement of elementary displaced volume with electrostatically actuated diaphragms: applications to electrostatic micropumps. J. Micromech. Microeng., 2000, 10, 282-286. Böhm, S; Burger, GJ; Korthorst, MT; Roseboom, F. A micromachined silicon valve driven by a miniature bi-stable electromagnetic actuator. Sens. Actuators A, 2000, 80, 77-83. Bergstrom, PL; Ji, J; Liu, Y-N; Kaviany, M; Wise, KD. Thermally driven phase-change microactuation. J. Microelectromech. Syst., 1995, 4, 10-17. Kohl, M; Dittmann, D; Quandt, E; Winzek, B. Thin film shape memory microvalves with adjustable operation temperature. Sens. Acutators A, 2000, 83, 214-219. Melin, J; Quake, SR. Microfluidic large-scale integration: the evolution of design rules for biological automation. Annu. Rev. Biophys. Biomol. Struct., 2007, 36, 213-231. Xia, Y; Whitesides, GM. Soft lithography. Angew. Chem. Int. Ed. Engl., 1998, 37, 550-575. Wolfe, DB; Qin, D; Whitesides, GM. Rapid prototyping of microstructures by soft lithography for biotechnology. Methods Mol. Biol., 2010, 583, 81-107. Unger, MA; Chou, H-P; Thorsen, T; Scherer, A; Quake, SR. Monolithic microfabricated valves and pumps by multilayer soft lithography. Science, 2000, 288, 113-116. Quake, SR; Scherer A. From micro- to nanofabrication with soft materials. Science, 2000, 290, 1536-1540. Thorsen, T; Maerkl, SJ; Quake, SR. Microfluidic large-scale integration. Science, 2002, 298, 580-584.

Microfluidic Valves without Diaphragms …

147

[17] Grover, WH; Ivester, RHC; Jensen, EC; Mathies, RA. Development and multiplexed control of latching pneumatic valves using microfluidic logical structures. Lab Chip, 2006, 6, 623-631. [18] Arndt, K-F; Kuckling, D; Richter, A. Application of sensitive hydrogels in flow control. Polym. Adv. Technol., 2000, 11, 496-505. [19] Beebe, DJ; Moore, JS; Bauer, JM; Yu, Q; Liu, RH; Devadoss, C; Jo, BH. Functional hydrogel structures for autonomous flow control inside microfluidic channels. Nature, 2000, 404, 588-590. [20] Eddington, DT; Beebe, DJ. Flow control with hydrogels. Adv. Drug Del. Rev., 2004, 56, 199-210. [21] Park, JY; Oh, HJ; Kim, DJ; Baek, JY; Lee, SH. A polymeric microfluidic valve employing a pH-responsive hydrogel microsphere as an actuating source. J. Micromech. Microeng., 2006, 16, 656-663. [22] Richter, A; Paschew, G; Klatt, S; Lienig, J; Arndt, K-F; Adler, H-JP. Review on hydrogel-based pH sensors and microsensors. Sensors, 2008, 8, 561-581. [23] Gerlach, G; Arndt, K-F (Eds.). Hydrogel Sensors and Actuators. Engineering and Technology. Berlin, Heidelberg: Springer Series on Chemical Sensors and Biosensors Vol. 6; 2009. [24] Hoffmann, J; Plötner, M; Kuckling, D; Fischer, W-J. Photopatterning of thermally sensitive hydrogels useful for microactuators. Sens. Actuators A, 1999, 77, 139-144. [25] Richter, A; Kuckling, D; Howitz, S; Gehring, T; Arndt, K-F. Electronically controllable microvalves based on smart hydrogels: magnitudes and potential applications. J. Microelectromech. Syst., 2003, 12, 748-753. [26] Richter, A; Howitz, S; Kuckling, D; Arndt, K-F. Influence of phenomena of volume phase transition at the behavior of hydrogel based valves. Sens. Actuators B, 2004, 99, 451-458. [27] Yu, C; Mutlu, S; Selvaganapathy, P; Mastrangelo, CH; Svec, F; Fréchet, JMJ. Flow control valves for analytical microfluidic chips without mechanical parts based on thermally responsive monolithic polymers. Anal. Chem., 2003, 75, 1958-1961. [28] Richter, A; Türke, A; Pich, A. Controlled double-sensitivity of microgels applied to electronically adjustable chemostats. Adv. Mater., 2007, 19, 1109-1112. [29] Lärmer, F; Schilp, A. Method of anisotropically etching silicon. US Patent No. 5501893, German Patent DE 4241045, 1994.

148

Steffen Howitz, Frank Baudisch, Frank-Ulrich Gast et al.

[30] Gast, F-U; Dittrich, PS; Schwille P; Weigel, M; Mertig, M; Opitz, J; Queitsch, U; Diez, S; Lincoln, B; Wottawah, F; Schinkinger, S; Guck, J; Käs, J; Smolinski, J; Salchert, K; Werner, C; Duschl, C; Jäger, MS; Uhlig, K; Geggier, P; Howitz, S. The microscopy cell (MicCell), a versatile modular flowthrough system for cell biology, biomaterial research, and nanotechnology. Microfluid. Nanofluid., 2006, 2, 21-36. [31] Gastrock, G; Lemke, K; Römer, R; Howitz, S; Bertram, J; Hottenrott, M; Metze, J. Protein-Processing Platform (3P) - a new concept for the characterization of cell cultures on a mL-scale using microfluidic components. Eng. Life Sci., 2008, 8, 73-80. [32] Chou, H-P; Unger, MA; Quake, SR. A microfabricated rotary pump. Biomed. Microdev., 2001, 4, 323-330. [33] Hartshorne, HA; McKinon, GH; Yu, P; Lewis, DF. Microfluidic valve and method of manufacturing same. US Patent No. 6748975 B2, Jun. 15, 2004. [34] Lee, JN; Park, C; Whitesides, GM. Solvent compatibility of poly(dimethylsiloxane)-based microfluidic devices. Anal. Chem., 2003, 75, 6544-6554.

In: Microfluidics: Theory and Applications ISBN 978-1-61668-570-6 Editor: I. A. Kuznetsov, pp. 149-165 © 2010 Nova Science Publishers, Inc.

Chapter 5

BIOCOMPATIBLE AND MASS PRODUCTIVE MEMS DEVICE FOR LOCALIZED SURFACE PLASMON RESONANCE Xiaodong Zhou and Hong Liu Institute of Materials Research and Engineering, A*STAR (Agency for Science, Technology and Research), 3 Research Link, Singapore 117602

ABSTRACT 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

150

Xiaodong Zhou and Hong Liu 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 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.

1. INTRODUCTION Biosensors have been exploited by researchers for many years in the areas of diagnosis and monitoring of diseases, drug discovery, proteomics and environmental monitoring. Fundamentally, biosensor is an analytical device which converts biological response into a detectable electrical or optical signal. Much biosensor research effort has been devoted to the evaluation of the relative merits of various signal transduction methods in optics, radiochemistry, electrochemistry, piezoelectricity, magnetoelectricity and micromechanics, and two or more mechanisms might be combined or multiplexed in a sensor or sensor array. Recently, biological applications upon the localized surface plasmon resonance (LSPR) spectroscopy have been demonstrated by several research groups. Noble metal nanoparticles exhibit a strong UV-visible absorption band that is absent in the spectrum of the bulk noble metal. This absorption results when the incident photon frequency resonates with the collective oscillation of the conduction electrons in the metal nanoparticles, which is known as localized surface plasmon resonance (LSPR) [1-7]. Since the LSPR extinction wavelength depends on the refractive index of the media adjacent to the metal nanoparticles, LSPR is employed as a sensitive, real-time and label-free sensing mechanism revealing the surface interactions of analytes. Detecting a biomarker by LSPR starts from bonding a ligand/anchor of the biomarker onto

Biocompatible and Mass Productive MEMS Device …

151

the gold nanoparticles; subsequently these functionalized nanoparticles are rinsed (sometimes incubated) with the serum or buffer with biomarkers; and the existence, concentration, or the binding dynamics of the biomarkers can be extrapolated from the wavelength shift of the LSPR’s extinction peak. Similar to surface plasmon resonance (SPR) [8, 9], which is a well established technology and has been applied for the detections of hundreds of analytes, LSPR is also an effective tool to characterize the biological interfaces in realtime, and it has ten times higher sensitivity than SPR within its electromagnetic decay length of 5 – 15 nm, which is very suitable for monolayer molecular or short-chain DNA detections. Compared with SPR which has a noble metal film and needs some optical structures (either prism, or grating, or waveguide) to enhance the momentum of the light so as to excite SPR, the occurrence condition of LSPR is much relaxed, LSPR happens whenever the light irradiates onto the metal nanostructures, thus LSPR eliminates the bulky optical setup of SPR, and it can be miniaturized into a glass chip and easily be measured through a conventional UV-vis-near-IR spectroscopy, by taking the absorption spectrum of a glass substrate fabricated with noble metal (silver or gold) nanostructures. The peak of the LSPR spectrum is also tunable according to the size, shape, interval, material and surrounding medium of the metal nanostructures, which provides the convenience of enhancing the signals through the surface plasmon polariton wavelength matching. Moreover, when LSPR chip is fabricated into a microfluidics device or microarray, it can characterize the biological samples with an ultra-small amount, renders much shorter measurement time and multiple sample measurements. It has been demonstrated that a bio-chip with up to 300 of sensing spots can be realized by using the LSPR chip combined with microfluidic channels fabricated by polydimethylsiloxane (PDMS) [6, 7], a polymer that can form microfluidic channels quickly by molding and can be covered on a glass chip to form microfluidics. However, PDMS has the problems of swelling and dissolution of PDMS oligomers in solvent, it is preferred that silicon or glass materials, which are more stable and durable in chemicals, can be used for LSPR micro-electromechanical systems (MEMS). LSPR MEMS is actually a challenging task to fulfill. In order to realize it, the MEMS fabrication must be nanofabrication compatible and biocompatible, and it should be robust, cost-effective, transparent with minimum light scatterings, and preferably can be fabricated through mass production. These requirements eliminate a lot of conventional MEMS fabrication process, for instance, anodic bonding at high temperature

152

Xiaodong Zhou and Hong Liu

will damage the gold nanostructures, glass etching will roughen the surface of the glass wafers and cause optical scatterings in LSPR. Combining all of the above requirements in our design, we have developed a cost-effective LSPR MEMS prototype that consists of a glass-silicon-glass sandwich structure, with successful design and fabrication, the application of this LSPR MEMS on bovine serum albumin (BSA) test is demonstrated. We also prove that 3 mm thick PDMS layer obviously widens the LSPR spectrum, so it should be avoided in LSPR MEMS device in terms of both spectral broadening and biological absorptions.

2. DEVICE DESIGN AND FABRICATION The proposed microfluidic biosensor is made of commonly-used semiconductor materials, i.e., silicon and Pyrex 7740 glass through MEMS fabrication technology, as illustrated in Figure 1. In Figure 1, the bottom glass layer is with gold nanostructures fabricated for LSPR sensing, the middle layer is with silicon to form the microfluidic channels, and the top layer is glass with holes drilled for inlet and outlet of the device.

Figure 1. The cross-section view of the fabricated microfluidic LSPR biosensor.

Prior to MEMS fabrication, a layer of gold nanostructures was fabricated on the bottom Pyrex 7740 glass chip through dispersed nanosphere lithography (NSL). NSL is a powerful technique to inexpensively fabricate gold nanostructures on the glass substrate with controlled shape, size and

Biocompatible and Mass Productive MEMS Device …

153

interparticle spacing, and its fabricated nanostructures are suitable for LSPR sensing because of their tunability and varieties. In our sensing device, the gold nanostructures on the glass substrate were fabricated as following. First a 4” Pyrex 7740 glass wafer was implanted with silicon ions in a Varian EHP200 ion implanter at an energy of 5 keV and a dose of 1E14/cm2, this step is to modify the surface of the glass in order to disperse the nanospheres. Then the glass wafer was diced into a 20 mm × 10 mm chip. After cleaning, the glass substrate was dip coated with a 1:5 diluted poly(diallyldimethyl ammonium chloride) (PDDA) solution for 30 sec, PDDA was purchased from Sigma Aldrich as a 20 wt% solution in water. After rinsing and drying, the glass substrate was drop coated with 1 mL 10 times diluted solution of polystyrene nanospheres bought from Duke Scientific Ltd. The positive charges in PDDA cancelled out the negative charges of the nanospheres as well as the charges of the silicon ion implanted glass surface, thus the nanospheres were dispersed on the glass surface. Finally, 50 nm of gold was evaporated onto the nanospheres at 70˚ (see Figure 5). Under these fabrication conditions, the gold on the nanospheres attaches to the gold on the glass substrate and forms nonconformal 3D gold nanostructures [10], which yield a LSPR extinction peak when light passing through these nanostructures. The wavelength redshift of the LSPR peak reflects the reflective index increment of adjacent medium, or the bonding of analyte onto the gold nanostructures. The fabrication process for the silicon layer started from a standard RCA cleaning of a 4” silicon wafer (100) with a thickness of 400 µm. Then, a 1 µm thick silicon dioxide layer was deposited by PECVD on both sides of the wafer. After that, the silicon wafer was spin coated with a layer of AZ4620 photoresist on both sides. The silicon wafer was subsequently exposed at 7 mW/cm2 for 13 s and developed with AZ developer for 45 s to open the windows of inlet, outlet and light path. The silicon dioxide was etched by buffered oxide etch solution (7:1) at an etch rate of 850 ⊕/min, as shown in Figure 2(a). A microchannel of 100 µm wide was patterned on the other side of the wafer by photoresist coating, exposure, developing and oxide removal as shown in Figure 2(b). Figure 2(c) presents that a channel depth of 50 µm was achieved using DRIE. Then the inlet, outlet and light path were etched through by DRIE to connect with the fluid path of the microchannel as shown in Figure 2(d). This silicon wafer was diced into 20 mm × 10 mm chips for the following bonding process. A Pyrex 7740 glass wafer was diced into chips with the same size of silicon chips. These glass chips serve as the top covers of the microfluidics and are regarded as top glass chips. Inlet and outlet were drilled in the top glass

154

Xiaodong Zhou and Hong Liu

chip and Loctitte 3301 UV curable epoxy with a refractive index of 1.48 was spin coated on the top glass chip. After aligning properly, the silicon chip and top glass chip were bonded together and exposed under UV light at 7 mW/cm2 (365 nm wavelength) for 5 min to cure the epoxy and form the bonding strongly, which is shown in Figure 2(e). A layer of Loctite 3301 UV curable epoxy was spin coated on the silicon surface of the previously bonded Si-glass structure, which was then carefully aligned to the bottom glass chip with gold nanostructures fabricated to bond them together. After UV exposure, the glasssilicon-glass sandwich structure device was completed as shown in Figure 2(f).

Figure 2. The MEMS fabrication process of the microfluidic LSPR biosensor.

Biocompatible and Mass Productive MEMS Device …

155

Figure 3. The schematic of the assembly for LSPR measurements (a) and the photo of the packaged device (b).

156

Xiaodong Zhou and Hong Liu

Figure 4. 3D demonstration for the fabrication and assembly of the LSPR MEMS. (a) shows the fabricated gold nanostructures on glass substrate, (b) shows the silicon microfluidic layer on the substrate, (c) shows the glass-silicon-glass structured LSPR chip, (d) shows the fabricated chip in the bottom part of the package, with gaskets and ferrules affixed, (e) and (f) are respectively the top and side views of the packaged device.

For demonstration and testing purpose, a plastic jig was fabricated to assemble the prototype with commercially available microfluidic connections including a pair of tubing, fittings, ferrules and gaskets. Figure 3(a) illustrates the schematic of the assembly for testing, and Figure 3(b) is the photo of the packaged device in the size of 36 mm × 20 mm × 10 mm. The fabrication and assembly process shown in Figure 4 gives a more authentic relative size comparison of the sensing area, the chip and the package. In order to make sure that the MEMS fabrication process did not affect the fabricated gold nanostructures on the bottom of the device, one of the devices was pried apart, and the discrete nanostructures on the bottom layer of the

Biocompatible and Mass Productive MEMS Device …

157

glass substrate were observed under a scanning electron microscope (SEM). As presented in Figure 5, the structural integrity of the gold nanostructures did not adversely change after carrying out the fabrication steps to make the microfluidic sensor chip, which indicates that the performance of the LSPR will not degrade when the fabricated microfluidic chamber is used.

Figure 5. SEM pictures of the gold nanostructures in the fabricated device, which proves that the gold nanostructures kept intact at the bottom glass substrate. (a) was observed in an area of 46.7 μm × 35 μm, and (b) is with 9.2× magnification of (a).

158

Xiaodong Zhou and Hong Liu

3. UNIQUENESS OF THE DEVICE The uniqueness of the device is that it is mass productive, because prior to the epoxy bonding, the process is in wafer level. At the same time, it gets rid of PDMS, and the light blockage of the silicon layer increases the signal-tonoise ratio of the device.

1. Comparison to the Device With PDMS The LSPR spectra of the device were measured with an Ocean Optics spectrometer at the wavelength range of 400 - 900 nm and an optical resolution of 0.35 - 0.36 nm, as plotted in Figure 6. Figure 6a shows the LSPR spectrum for a glass-silicon-glass structure when the adjacent media of the gold nanostructures were air and water. It can be seen that when the adjacent medium changed from air to water (i.e., the refractive index changes from 1 to 1.33), the extinction peak of the LSPR spectra shifted from 642.16 nm to 677.35 nm, which is equivalent to a LSPR sensitivity of 106 nm/RIU, RIU is the unit of the refractive index. It should be noted that this sensitivity depends on the size, shape and material of the nanostructures, and is not related with the microfluidic design. Gold nanostructures used in the experiments were fabricated as described in Section 2, and they looked similar to the ones photographed in Figure 5. Since the angle and thickness of the gold deposition might be different for each sample, the LSPR sensitivity also differs a little. Figure 6(b) compares the LSPR spectra when a layer of 3 mm thick PDMS was added on the top of the LSPR MEMS device. PDMS is transparent and allows light to be transmitted to the nanostructures. There is a slight spectrum broadening due to the addition of PDMS. From the experimental results provided in Figure 6(b), it can be seen that PDMS may still be used for LSPR MEMS, however, the LSPR signal is more superior when glass is being used as the transparent layer. Moreover, PDMS is a porous material and it is not suitable for long term use in detecting biological samples because it tends to absorb some biological samples. For multiple use of a device, it is believed that the residue in PDMS may also influence the accuracy of LSPR measurements adversely. In conclusion, silica and glass based microfluidic device for LSPR application is most desirable in terms of durability, stability, reliability, repeatability and high optical transparency.

Biocompatible and Mass Productive MEMS Device …

159

(a)

(b) Figure 6. LSPR spectra of the LSPR MEMS device measured with an Ocean Optics spectrometer. (a) shows the LSPR spectrum for a glass-silicon-glass structure when the adjacent media of the gold nanostructures were air and water, (b) shows the LSPR spectrum for the same structure with a 3 mm thick layer of PDMS added, when the nanostructures’ adjacent medium was water.

2. Signal Enhancement by Light Blockage Our experiments also prove that the silicon layer in the device can effectively block the light and increase the signal-to-noise ratio of the device.

160

Xiaodong Zhou and Hong Liu

The experimental process was as following: the gold nanostructures were fabricated on the substrate of the LSPR MEMS device, and the LSPR spectrum was taken (marked as “without block in air” in Figure 8, with the diagram drawn in Figure 7(a)); then this chip was bonded with the silicon layer with microfluidics and the top glass layer to form the MEMS device to measure the LSPR spectra (marked as “with block in air” in Figure 8, with the diagram drawn in Figure 7(b)). Four samples were measured in total as presented in Figure 8, and the results show that the light blockage improved the shapes of LSPR spectra, due to the fact that the gold nanostructures were more evenly distributed in a smaller MEMS sensing area (less than 1 mm in diameter) than a large area of the whole light spot (about 5 mm in diameter).

(a)

(b) Figure 7. The diagram of the LSPR measurement for comparing the effect of the light blockage in LSPR MEMS device. Pink rectangle is the glass substrate, the red circle indicates the size of the light spot of our spectrometer (about 5 mm in diameter), and the yellow spots represent gold nanostructures. (a) is without light blockage before MEMS fabrication; (b) is with light blockage in the MEMS structure.

Biocompatible and Mass Productive MEMS Device … 0.7

without block in air with block in air

Extinction

0.6

0.5

0.4

0.3

0.2 400

500

600

700

800

900

800

900

Wavelength / nm

(a) 0.38

without block in air with block in air

0.36 0.34

Extinction

0.32 0.30 0.28 0.26 0.24 0.22 0.20 0.18 400

500

600

700

Wavelength / nm

(b)

161

Xiaodong Zhou and Hong Liu

162

0.60

without block in air with block in air

0.55

Extinction

0.50 0.45 0.40 0.35 0.30 0.25 0.20 400

500

600

700

800

900

Wavelength/ nm

(c) 0.40

without block in air with block in air

0.36

Extinction

0.32 0.28 0.24 0.20 0.16 0.12 0.08 400

500

600

700

800

900

Wavelength / nm

(d) Figure 8. (a)-(d) LSPR spectra taken when with and without light blockage for 4 LSPR MEMS devices, the adjacent medium of the nanostructures was air.

Biocompatible and Mass Productive MEMS Device …

163

4. LSPR SENSING To demonstrate that such a device is suitable for biological test, a LSPR chip with the gold nanostructures fabricated as in Figure 5 was used to measure LSPR spectra for bovine serum albumin (BSA), a kind of serum albumin protein that has many biochemical applications in immunoassays. PerkinElmer spectrometer was used to detect the LSPR spectra at the wavelength range of 400 - 1000 nm. First a LSPR spectrum was taken in air, subsequently 1 µg/ml phosphate buffered saline (PBS) buffer was added and the LSPR spectra was taken, and finally BSA was injected and incubated at room temperature for 1 hour, rinsed with PBS and the LSPR spectrum was measured.

Figure 9. LSPR spectra of a glass chip with gold nanostructures fabricated for BSA measurement.

The spectra of the experiment are shown in Figure 9. The extinction peak wavelength of LSPR in air is 789 nm. BSA adsorption redshifted the spectrum of PBS buffer obviously but slightly to the right, so it has to be discerned carefully. Since even in the insert the peaks of the two LSPR spectra look close, the wavelength shift is better to be identified by averaging the two wavelengths at the extinction intensity of 1.35, because at 1.35 this shift is discernable and is very close to the peak intensity of 1.37. For the PBS buffer,

164

Xiaodong Zhou and Hong Liu

the left and right wavelengths at the intensity of 1.35 are respectively 798 and 873.5 nm, thus renders a central wavelength of 835.75 nm. After BSA adsorption, the left and right wavelengths at the intensity of 1.35 are respectively 802.5 and 881.5 nm, thus gives a central wavelength of 842 nm. So the adsorption of BSA caused the wavelength to red shift 6.25 nm. According to the central wavelength in air and water, the sensitivity of the LSPR chip was calculated to be 139.69 nm/RIU. Gao et al. [11] also measured the wavelength shift of a LSPR chip with BSA on nanoholes. But in that work, the chip was first immersed in BSA and rinsed by PBS, then dried and measured in air. With the chip sensitivity of 110 nm/RIU, they detected a 7 nm shift caused by BSA in air. Our chip has higher sensitivity of 139.69 nm/RIU, for BSA test, our sensitivity is also higher, because the wavelength shift of an analyte is proportional to the refractive index difference between the analyte and adjacent medium. If we also dried the chip in air, a wavelength shift caused by BSA adsorption should be much higher than Ref. [11]. Our biological test results manifest that the BSA adsorption and the function of our LSPR chip are well and expectable.

5. CONCLUSION This chapter investigated a method to fabricate LSPR MEMS in a glasssilicon-glass sandwich structure. The fabrication processes of the gold nanostructures and the microfluidics as well as the package of the device were described in detail. Both the nanofabrication and MEMS fabrication were costeffective and suitable for mass production. The microfluidics fabrication was carried out at room temperature without introducing any stress, thus it does not affect the nanostructures on the bottom layer of the device, which was revealed by the SEM picture of the nanostructures on the bottom glass chip after prying the device apart. Besides mass productive and highly transparent, this device has two major advantages: PDMS free and the light blockage of the non-sensing area. Our MEMS fabricated with glass and silicon avoided the optical and biological signal degradation might be caused by PDMS formed microfluidics. Our experiments indicate that a 3 mm thick layer of PDMS obviously broadens and deforms the LSPR spectra. Silicon layer blocks the light out of the sensing area and increases the signal-to-noise ratio of the device, because this guarantees that the LSPR signal only generates from a smaller area where the gold nanostructures are relatively even distributed.

Biocompatible and Mass Productive MEMS Device …

165

The LSPR sensing was conducted by measuring the BSA protein. In this experiment, the gold nanostructures fabricated by evaporating 50 nm of gold onto the 170 nm diameter polystyrene nanospheres at the angle of 70º rendered a LSPR sensitivity of 139.69 nm/RIU, and the LSPR spectrum redshifted 6.25 nm after BSA adsorption. This result is more sensitive than previously reported BSA sensing with gold nanoholes. Overall, due to its low-cost, high signal-to-noise ration, biocompatibility, highly transparency and mass production, the glass and silicon based LSPR MEMS we proposed is ideal for LSPR based biosensing, and it can also be fabricated into a LSPR microarray or LSPR/SPR combined microfluidics, which can be used in a point-of-care medical diagnostic system by just integrating with a miniature spectrometer.

REFERENCES Haes, J.; Stuart, D. A.; Nie, S.; Van Duyne, R. P. J. Fluoresc. 2004, 14, 355-367. [2] Willets, K. A.; Van Duyne, R. P. Ann. Rev. Phys. Chem. 2007, 58, 267297. [3] Zhao, J.; Zhang, X.; Yonzon, C.; Haes, A. J.; Van Duyne, R. P. Nanomedicine 2006, 1, 219-228. [4] Stewart, M. E.; Anderton, C. R.; Thompson, L. B.; Maria, J.; Gray, S. K.; Rogers, J. A.; Nuzzo, R. G. Chem. Rev. 2008, 108, 494-521. [5] Anker, J. N.; Hall, W. P.; Lyandres, O.; Shah, N. C.; Zhao, J.; Van Duyne, R. P. Nat. Mater. 2008, 7, 442-453. [6] Endo, T.; Kerman, K.; Nagatani, N.; Hiepa, H. M.; Kim, D-K.; Yonezawa, Y.; Nakano, K.; Tamiya, E. Anal. Chem. 2006, 78, 64656475. [7] Hiep, H. M.; Nakayama, T.; Saito, M.; Yamamura, S.; Takamura, Y.; Tamiya, E. Jpn. J. Appl. Phys. 2008, 47, 1337-1341. [8] Homola, J. Anal. Bioanal. Chem. 2003, 377, 528-539. [9] Homoal, J.; Yee, S. S.; Gauglitz, G. Sensor. Actuat. B-Chem. 1999, 54, 3-15. [10] Zhou, X.; Knoll, W.; Liu, K. Y.; Tse, M. S.; Oh, S. R.; Zhang, N J. Nanophotonics 2008, 2, 023502. [11] Gao, D.; Chen, W.; Mulchandani, A.; Schultz, J. S. Appl. Phys. Lett. 2007, 90, 073901. [1]

In: Microfluidics: Theory and Applications ISBN 978-1-61668-570-6 Editor: I. A. Kuznetsov, pp. 167-174 © 2010 Nova Science Publishers, Inc.

Chapter 6

APPLICATION TO ADAPTIVE OPTICS AND LASER MICROFLUIDICS Jean-Pierre Delville, Alexis Casner, Régis Wunenburger and Iver Brevik Universite Bordeaux, France

Linear interface deformations can 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".

1. ADAPTIVE OPTICS Beyond soft lensing, the liquid columns stabilized by radiation pressure can be viewed as appealing structures for optical guiding. Indeed, as the index of refraction of the columns is necessarily larger than that of the surrounding fluid, laser beam is automatically captured by total reflection and guided within the liquid medium. This light confinement is dynamically illustrated in Figure 1 over its entire development, i.e. from soft lensing ( t = 1 − 1.5 s ) to

168

Jean-Pierre Delville, Alexis Casner, Régis Wunenburger et al.

jetting ( t = 2 − 16 s ) and finally laser guiding ( t = 19 − 22 s ). The final liquid fiber length is 455 µm .

Figure 1. Dynamical behavior of soft focusing, laser self-trapping by the induced liquid jet and final optical guiding by the stationary liquid column formed. Parameters are

ω0 = 3.47 µm , P = 470 mW , and (T − TC ) = 4 K . The liquid fiber length

is

455 µm .

Application to Adaptive Optics and Laser Microfluidics

169

This laser self-guiding is also efficient in smaller liquid bridges as illustrated in Figure 2a, where a liquid column of aspect ratio Λ = 14 is stabilized with a glass capillary of height 200 µm . One can therefore put forward the concept of liquid step-index optical fibre, which provides a very new approach towards self-induced waveguiding. Indeed, contrarily to optical adaptive waveguides written by photopolymerisation [1] or laser damage in glasses [2], liquid fibres are nonpermanent. They are thus completely reconfigurable. The laser creates its own channel that is automatically optimised to its waist and power as illustrated in Figure 3, where adaptation to incident beam power is clearly observed.

Figure 2. a) Liquid bridge with aspect ratio

Λ = 14

induced and stabilized in a

200 µm glass capillary by a laser beam of beam waist ω0 = 3.2 µm and power

P = 1400 mW . The temperature gap is (T − TC ) = 6 K . The brightness of the

column evidences laser self-guiding. b) Same experimental conditions as in a) but with a tilted exciting beam. c) Liquid elbow created in a two-laser configuration (upward/downward) at

(T − TC ) = 6 K . Both wetting films are destabilized,

intercept and form a wedge. The directions of propagation of each beam (

ω0 = 3.5 µm , P = 800 mW ) are indicated by the arrows.

These tuneable optical fibres can furthermore be oriented in any direction by tilting the exciting beam. Figure 2b gives an example of such an inclined liquid bridge stabilized by a laser beam propagating obliquely. Even more surprising are artificial structures such as the stable liquid elbow created in a two-laser beams configuration and presented in Figure 2c. Optically induced

170

Jean-Pierre Delville, Alexis Casner, Régis Wunenburger et al.

liquid columns are thus particularly efficient to control beam propagation or to optimise light coupling devices because self-adaptation considerably reduces the sensitivity to precise mechanical alignments of the optical components used.

Figure 3. Adaptation of the liquid fiber diameter to the input power. Parameters are

ω0 = 3.47 µm

and

(T − TC ) = 4 K . The liquid fiber length is 334 µm . As

demonstrated by the emission of droplets from liquid jets, note that hydrodynamic flow still persists within the liquid columns as illustrated by the fluid accumulation at their feet.

2. LASER MICROFLUIDICS Aside optical guiding, the liquid jets and columns stabilized by radiation pressure can be viewed as three dimensional microfluidic devices with "soft" wall because liquid is transported from the upper liquid phase when P ≥ P↑ , i.e. above interface instability onset. Two types of micro-flow can be generated: droplet flow and pipe flow. As illustrated dynamically in Figure 4, droplets are continuously generated at the tip of a stationary jet. Moreover, since the index of refraction of the droplet is larger than that of the surrounding fluid, the beam automatically traps them. This brings directionality in droplet emission and transport that can

Application to Adaptive Optics and Laser Microfluidics

171

be actuated by tilting the beam as illustrated in Figure 2b. Finally, at fixed

(T − TC ) and ω0 ,

the droplet flow rate can be controlled by the incident

beam power. Consequently, contrary to other methods [3], this optical approach of microfluidics provides droplets that are directly produced within the chosen micrometric size in a contact less way without further processing. Also, as droplets are produced in three dimensions from a channel that can be controlled in size by the beam, no particular microfluidic device is required to manage the droplet flow.

Figure 4. Continuous droplet emission and trapping from a jet of stationary length. Note the regularity in flow rate and the relative monodispersity of the induced droplets. The time delay between two pictures is 1 s and the jet length is 330 µm . Control parameters are

(T − TC ) = 6 K , ω0 = 3.5 µm , and P = P↑ = 490 mW .

On the other hand, liquid flow within the jet persists when it reaches the bottom face of the cell and forms a liquid bridge. This is illustrated in Figure 3, were accumulation of liquid on the bottom window of the cell is clearly observed. This means that laser-sustained liquid bridges behave as micro-pipes that can be used to transfer fluid from one reservoir to another. One even more surprising fact is that the micro-pipe section can be actuated continuously by varying the incident beam power (see Figure 3). Finally, let us note that liquid columns are unstable in classical conditions. As column breaking generally leads to at least bimodal droplet distributions, often called droplets and satellites [4] (see the left picture of Figure 5, the

172

Jean-Pierre Delville, Alexis Casner, Régis Wunenburger et al.

important drawback for microfluidic applications is control over micro-jet fission to form a regular assembly of micro-droplets [5, 6]. On the other hand, for applications motivated by fluid transport alone annular flows are much more efficient than droplet flows, even if the first situation is difficult to implement due to the Rayleigh-Plateau instability. Since radiation pressure of laser waves is able to prevent column breaking, it becomes possible to reverse the microfluidic approach in order to transform a bubbly flow into an annular flow. This process is dynamically illustrated in Figure 5. Starting from a linear assembly of well-separated droplets, the radiation pressure deforms droplets and forces them to coalesce transforming the initial droplet assembly into a liquid column.

Figure 5. Transition from droplet to annular flow resulting from droplet deformation and coalescences driven by the optical radiation pressure. The acquisition frequency is

f = 20 Hz . Control parameters are ( T − TC ) = 5 K , ω0 = 3.47 µm , and

P = 630 mW > P↑ .

CONCLUSION AND PROSPECTS The main purpose of this review was to theoretically and experimentally explore fluid interface deformation driven by the optical radiation pressure of a continuous laser wave. Using near-critical liquid-liquid interfaces, to strongly reduce surface tension and enhance radiation pressure, we validated a universal description of the process. We also investigated nonlinear regimes in deformation. Asides surprising stable tether shapes, we have presented a new electromagnetic instability

Application to Adaptive Optics and Laser Microfluidics

173

mechanism of fluid interfaces driven by continuous laser wave. The very good agreement observed between measurements and expectations demonstrates the universality of this nonlinear process. The resulting jet is also expected to occur quite universally and its regularity, as well as that of the produced micro-droplets, should be promising in microfluidics and electromagnetic spraying techniques. We finally extended this investigation on nonlinear behaviors to the formation of stable liquid columns. While the results presented here were obtained with a single laser beam, the method could easily be extended to a parallel approach by forming liquid-bridge patterns or adaptive liquid gratings by tailoring the intensity distribution with interfering pump beams. Even if we try to give an extended view of laser-induced fluid interface deformation, most of its theoretical description, particularly in the nonlinear regime, is still missing. For example, we did not explain the observed tether shape of the deformation. An investigation of liquid jet properties is also missing. The main reason is linked to the complex nonlinear coupling between a mobile and deformable interface and a laser beam as the interface adapts its position to optical excitation. Also, we did not discuss dynamical behaviors, even in the linear regime in deformation. Even if we do have some results on this important aspect, particularly for applications, they are too incomplete to be developed within a general scheme. Nevertheless, we hope that our exhaustive presentation of the "static" manifestations of a liquid interface under the radiation pressure of a laser wave will promote an optical approach to interface actuation and, to build microfluidic devices with optical forces [7] or, conversely, to anticipate new optical micro-systems based on microfluidics [8] for flow guiding and light coupling applications.

REFERENCES [1] [2] [3] [4]

S. Shoji, S. Kawata, A. A. Sukhorukov, and Y. S. Kivshar, Optics Lett. 27, 185 (2002). K. Miura, J. Qiu, H. Inouye, T. Mitsuyu, and K. Hirao, Appl. Phys. Lett. 71, 3329 (1997). D. R. Link, S. L. Anna, D. A. Weitz, and H. A. Stone, Phys. Rev. Lett. 92, 054503 (2004). J. Eggers, Rev. Mod. Phys. 69, 865 (1997).

174 [5] [6] [7] [8]

Jean-Pierre Delville, Alexis Casner, Régis Wunenburger et al. R. Dreyfus, P. Tabeling, and H. Willaime, Phys. Rev. Lett. 90, 144505 (2003). S. L. Anna, N. Bontoux, and H. A. Stone, Appl. Phys. Lett. 82, 364 (2003). A. Terray, J. Oakey, and D. W. M. Marr, Science 296, 1841 (2002). P. Mach, M. Dolinski, K. W. Baldwin, J. A. Rogers, C. Kerbage, R. S. Windeler, and B. J. Eggleton, Appl. Phys. Lett. 80, 4294 (2002).

INDEX A  absorption, viii, 2, 4, 6, 34, 117, 150, 151  acceptors, 108  accuracy, 70, 158  acetone, 27, 28, 138  acid, 6, 7, 124, 125  actuation, 137, 173  actuators, 79, 137  adaptation, 169, 170  adhesion, 26  adsorption, 125, 131, 163, 164, 165  AFM, 21  air, ix, 11, 14, 16, 20, 37, 135, 158, 159, 160,  162, 163, 164  albumin, xi, 150, 152, 163  alcohol, 137  alkaline, 125  alternative, 127, 137, 143  aluminosilicate, 6  aluminum, 31  amine, 124  amino, 120, 124  amino acid, 120, 124  ammonium, 153  ammonium chloride, 153  amorphous, 6  annealing, 18, 20, 21, 24, 45  annihilation, 108, 127 

anode, 112, 121  application, xi, 167  appropriate technology, 2  aqueous solution, 7, 123  Asia, 125  aspect ratio, 7, 10, 169  assumptions, 62  atoms, 4, 6  attachment, 140  automation, 2, 106, 146  averaging, 163 

B  back, 31, 111  background, xi, 109, 116, 125, 141, 150  baking, 7, 36  bandwidth, 32  barriers, 138  beams, 11, 12, 13, 20, 32, 33, 36, 40, 170,  173  behavior, vii, ix, 56, 59, 60, 61, 65, 68, 72,  73, 75, 76, 77, 78, 79, 81, 84, 86, 92, 93,  94, 95, 96, 97, 147, 168  bending, 25  binding, 4, 124, 151  bioanalytical, 128  biocompatible, x, 137, 142, 149, 151  biological behavior, viii, 2 

Index

176

biomarker, 150  biomaterial, 148  bioreactors, 146  biosensor, 134  biotechnology, vii, 50, 145, 146  biotin, 124  blocks, x, 150, 164  Boltzmann constant, 62, 82  Boltzmann distribution, 62  bonding, xi, 3, 7, 138, 139, 150, 151, 153,  154, 158  boundary conditions, 62, 63, 83  bovine, xi, 150, 152, 163  broad spectrum, 32  bubble, 116  buffer, viii, 55, 57, 110, 111, 114, 116, 117,  118, 122, 140, 151, 163  bulk crystal, 41 

C  CAD, 143  candidates, 125  capillary, vii, viii, ix, 56, 57, 58, 80, 103, 104,  116, 117, 120, 123, 131, 169  carbon, 114, 123, 125  carboxyl, 124  casting, 136, 140, 142  catalysis, 123  cathode, 121  cell, 11, 14, 43, 44, 113, 124, 140, 144, 145,  148, 171  cell culture, 148  cerium, 6  challenges, viii, 2, 3  channels, viii, x, 3, 7, 11, 13, 49, 55, 57, 80,  109, 121, 122, 124, 136, 137, 142, 143,  144, 147, 151, 152  character, 113  charge coupled device, 5  charge density, ix, 56, 61, 62, 63, 80, 81, 82  charged particle, 109 

charm, 136  chemical etching, 6, 7, 15, 17, 18, 21, 49  chemical properties, 4  chemical reactions, 109  chemical sensing, 136, 145  chemicals, x, 149, 151  chemiluminescence, 104, 132, 134  chloride, 153  chromatography, 109, 117, 120  classes, 108  classical, 80, 90, 108, 138, 171  cleaning, 119, 140, 153  clusters, 6  coatings, 114  collaboration, 50  color, iv, 107  commercialization, 128  compaction, 24  compatibility, 49, 148  complexity, 3, 136  components, vii, viii, 2, 3, 6, 7, 16, 20, 26,  38, 44, 48, 49, 56, 57, 61, 81, 106, 108,  117, 148, 170  composition, 114  compounds, 108, 140  computerization, 106  concentration, viii, ix, 7, 55, 56, 62, 81, 82,  92, 93, 94, 96, 104, 106, 110, 117, 118,  120, 137, 151  conduction, 80, 86, 87, 91, 150  conductivity, 87, 113, 117  configuration, 9, 35, 82, 114, 121, 169, 170  confinement, 167  consumption, 105, 124  contaminant, 119  continuity, ix, 56, 60, 61, 81  control, vii, viii, ix, 46, 55, 57, 80, 106, 107,  115, 118, 135, 137, 139, 140, 141, 142,  147, 170, 172  convective, 117  convex, 19, 23, 24, 26, 35  cooling, 117, 140, 141  copper, 26, 27, 28, 29 

Index correlation, 123  correlation coefficient, 123  cosine, 79  cost, x, xi, 3, 7, 16, 30, 43, 105, 112, 113,  132, 136, 149, 150, 151, 164, 165  cost‐effective, 30, 136, 151, 164  coupling, 24, 25, 36, 37, 170, 173  crack, 116  crosslinking, 136  cross‐sectional, 13, 36, 38, 39, 87  crystalline, 6  crystallites, 14, 17  crystals, 41  culture, 144  curing, 142  current limit, 131  customers, 138  cycles, x, 136, 138, 142 

D  data processing, 110  decay, 151  defects, viii, 56, 57  definition, 67, 91, 96, 104, 124  deformation, xi, 167, 172, 173  degradation, 164  delivery, 80  density, ix, 56, 60, 61, 62, 63, 80, 81, 82  deposition, 158  derivatives, 108  detection, vii, ix, 16, 30, 103, 104, 105, 106,  107, 108, 109, 110, 112, 113, 114, 115,  116, 117, 118, 119, 120, 121, 122, 123,  124, 125, 126, 127  detection system, 105, 112, 115, 116  detection techniques, vii  deviation, 71, 80  diamond, 119, 125  diaphragm, 136  dielectric constant, 61, 77, 81  dielectric materials, 27 

177

dielectrics, 30  diffraction, 7, 20, 23, 49  diffusion, 116, 117, 118, 120, 137  directionality, 171  discontinuity, 69  discs, 136  dispersion, viii, 56, 57, 117, 118  displacement, 137, 142, 145  dissolved oxygen, 125  distilled water, 17, 80  distribution, ix, 11, 12, 42, 56, 62, 64, 65,  66, 67, 69, 70, 72, 79, 80, 81, 82, 87, 92,  97, 137, 142, 173  divergence, 19, 36  DNA, 134, 139, 151  donors, 108  dopamine, 123  doped, 6, 41, 125, 133  drawing, 37  drug delivery, 80  drug discovery, 150  drying, 153  durability, 123, 142, 158  duration, 4, 31  dust, 139  dyes, 33, 127  dynamic viscosity, ix, 56, 58, 59, 65, 73, 74,  79  dynamics, viii, 2, 151 

E  earth, 42  East Asia, 125  ecology, 104  education, 145  election, 142  electric current, 113  electric field, viii, 4, 55, 57, 59, 61, 68, 73,  74, 77, 78, 80, 109, 110, 115, 117, 118,  120, 122  electric power, 26 

Index

178

electrical conductivity, 87, 113  electrical properties, 80  electroanalysis, 128, 129, 131  electrochemical reaction, 105  electrochemistry, 104, 150  electrodes, 29, 41, 108, 113, 114, 115, 119,  120, 121, 123, 124, 125, 127, 134  electro‐hydrodynamics, xi, 167  electrolysis, 104, 106, 107, 117, 119, 120,  125  electrolyte, 61, 66, 81, 113  electromagnetic, 4, 146, 151, 173  electromagnetic wave, 4  electron, 6, 105, 108, 139, 157  electroosmosis, 57, 61, 109  electrophoresis, vii, viii, ix, 56, 57, 103, 104,  105, 109, 110, 111, 123, 131  emission, 31, 32, 106, 109, 113, 121, 123,  124, 127, 170, 171  emitters, 108  encapsulated, 14, 46  energy, 4, 11, 12, 17, 20, 32, 42, 91, 108,  153  engineering, vii, 118, 145  enlargement, 122  epoxy, x, 150, 154, 158  equipment, 50, 118  etching, 6, 7, 14, 15, 17, 18, 21, 36, 45, 48,  49, 138, 147, 152  ethanol, 31, 138  evaporation, 43  evolution, 146  excitation, 6, 106, 107, 108, 118, 120, 127,  173  experimental condition, 169  exposure, 6, 7, 14, 15, 153, 154  extinction, x, 42, 149, 150, 153, 158, 163 

F  fabrication, viii, x, 2, 3, 4, 5, 6, 7, 13, 14, 20,  27, 28, 30, 31, 36, 45, 49, 56, 57, 125, 

127, 149, 151, 152, 153, 154, 156, 160,  164  failure, viii, 56, 57  fatigue, viii, 56, 57  feedback, 114  fiber, 36, 37, 38, 168, 170  film, 26, 27, 113, 123, 125, 138, 146, 151  filters, 107  filtration, 117  financial support, 128  fission, 172  flagellum, 44, 46, 47  flexibility, 145  flow field, 58, 60, 61, 63, 81, 139  flow rate, ix, 56, 68, 69, 75, 77, 78, 79, 80,  81, 86, 90, 94, 95, 97, 171  fluid, ix, 56, 57, 58, 59, 60, 61, 65, 70, 72,  73, 74, 75, 76, 79, 80, 92, 93, 94, 95, 118,  136, 137, 141, 142, 153, 167, 170, 171,  172, 173  fluid interfaces, 173  fluid transport, 94, 172  fluorescence, 11, 30, 34, 104  focusing, 5, 10, 11, 19, 20, 36, 49, 168  food, 104  food industry, 104  formula, 19  fouling, 119, 120  fracture, 116  fresh water, 44  friction, ix, 56, 80, 81, 91, 96, 97, 142  futures, ix, 103, 120 

G  gas, 17  gel, 136, 138, 142, 145  generation, 42, 108, 116  gold, x, 119, 124, 149, 151, 152, 154, 156,  157, 158, 159, 160, 163, 164, 165  gold nanoparticles, 151  gracilis, 44, 45, 46, 47, 48 

Index grants, 145  gratings, 173  gravitational force, 82  gravity, 42  groups, 82, 114, 150  growth, 2, 47, 140 

H  handling, 136  hardships, 118  healthcare, 2  heat, 4, 6, 17, 45, 58, 80, 139, 140, 141  heat transfer, 80  height, 7, 36, 61, 78, 80, 81, 82, 91, 95, 169  helium, 17  heterogeneity, 117  high pressure, x, 150  high temperature, x, 36, 150, 151  high‐speed, 43, 44, 47  hip, vii, 26, 35, 36, 37, 38, 50, 140, 151, 152,  163  histidine, 124  housing, 138  HPLC, 137, 142  human, 104  human activity, 104  hybrid, viii, 2, 3, 30, 48  hybridoma, 140  hydrodynamic, 91, 139, 170  hydrofluoric acid, 6  hydrogels, 137, 147  hydrogen, 116, 124, 137  hydrogen bonds, 137  hydrogen peroxide, 124  hydrophobic, 138  hydrostatic pressure, 80  hyperbolic, 58, 59, 66, 79  hypothesis, 66 

179 I 

ideal, vii, 1, 6, 114, 125, 165  identification, 109  illumination, 11  image, 24, 26, 28, 34, 35, 37, 38, 39, 46, 48  immobilization, 127  immunoassays, 163  implementation, 105  impurities, 106, 107  in situ, 137  incompressible, 60, 81  indices, 11, 20, 38  indium, 104, 114  indium tin oxide (ITO), 104, 114, 122, 123,  125, 134  industry, viii, 2, 104, 113  inertia, 57  inertness, 113, 125  infinite, 73  information technology, 50  infrared, 127  infringement, 117  inhomogeneity, 49  injection, viii, 45, 55, 57, 111, 122, 137, 142,  145  inspection, 34  instability, 170, 172, 173  instruments, 105, 115  insulators, 28, 30  integrated circuits, vii, 1  integrated optics, 41  integration, vii, 1, 3, 16, 24, 26, 29, 34, 40,  50, 65, 69, 70, 124, 136, 146  integrity, 157  intensity, 173  interaction, viii, ix, x, 2, 4, 6, 56, 61, 81, 82,  97, 124, 149  interface, xi, 20, 115, 140, 167, 170, 172,  173  interval, 16, 29, 151  ionic, ix, 56, 81, 91, 93, 94, 96, 97, 110, 117 

Index

180

ions, 6, 62, 70, 80, 82, 91, 94, 118, 124, 153  IR spectroscopy, 151  irradiation, 6, 11, 12, 13, 14, 17, 20, 21, 36,  41, 49  isolation, 115  isotropic, 11  issues, 108, 119 

J  Japanese, 124  Joule heating, 117, 130 

K  kinetic energy, 91  KOH, 138, 139 

L  label‐free, 150  Langmuir‐Blodgett, 130  laser ablation, 26, 38  lasing threshold, 32  law, vii, ix, 56, 58, 59, 60, 61, 63, 64, 65, 68,  73, 76, 77, 79, 81, 85, 90, 91, 92, 97  leakage, 43, 137, 139  lens, 5, 7, 19, 20, 24, 36, 43, 44, 49  leucine, 124  lifetime, 142  lift, 138  ligand, 150  light beam, 17, 25, 36  light scattering, 151  limitations, 112, 131  linear, xi, 4, 6, 24, 26, 59, 88, 106, 119, 122,  123, 167, 172, 173  linear function, 26, 59  liquid chromatography, 117  liquid interfaces, 172  liquid phase, 170  liquids, 11, 38, 44, 74, 76, 78, 79 

lithium, 6, 14, 17, 19, 29, 41  lithography, vii, x, 2, 3, 136, 142, 146, 150,  152  loading, 140  LOD, 104, 105, 106, 110, 112, 117, 120,  122, 123, 124, 127  losses, 37, 123  luminescence, 108, 112  lysine, 124 

M  mainstream, vii, 2  majority, 109  manifold, 138, 139, 143  manufacturing, viii, 2, 7, 34, 48, 148  mass transfer, 58  matrix, 6, 19, 49, 107, 109, 118, 123  measurement, 35, 110, 119, 138, 151, 160,  163  media, 38, 106, 119, 144, 150, 158, 159  membranes, ix, 135  memory, 43, 146  MEMS, v, x, 149, 151, 152, 154, 156, 158,  159, 160, 162, 164, 165  mercury, 124  metal nanoparticles, x, 149, 150  metals, 114  meter, 32  methanol, 138  methionine, 118  methodology, ix, 103  MFC, 104, 105, 109, 110, 111  microarray, 151, 165  microcavity, 30  microchip, 2, 24, 25, 34, 35, 36, 38, 43, 44,  45, 46, 47, 105, 122, 123, 124, 125, 131  microelectrodes, 29, 30, 41, 124  microelectronics, vii, 1, 3, 48, 50, 136  microfabrication, vii, 2, 3, 4, 6  microfluidic channels, 3, 7, 11, 13, 48, 147,  151, 152 

Index microfluidic devices, viii, 25, 55, 57, 58, 76,  79, 105, 121, 124, 148, 170, 173  microgels, 147  micrometer, 4  microorganisms, viii, 2, 3, 42, 43, 44  microscope, 19, 23, 24, 26, 27, 28, 34, 35,  45, 46, 140, 141, 157  microstructure, 10, 36, 44  microsystem, 138, 144, 146  microvoids, 11  migration, 110  miniature, 146, 165  Ministry of Education, 145  mixing, viii, 11, 38, 55, 57, 124  mobility, 109, 110, 111, 118  modulation, 124, 140  molecules, 105, 108, 121, 139  momentum, viii, ix, 56, 58, 60, 63, 81, 83,  97, 151  monoclonal, 140  monoclonal antibodies, 140  monolayer, 151  motion, viii, 14, 45, 46, 47, 55, 57, 61, 109,  118  motors, 144  movement, 42, 43, 44, 46, 47 

N  nanoclusters, 6  nanocrystal, 134  nanodots, 19  nanofabrication, x, 146, 149, 150, 151, 164  nanometer, 4, 30  nanoparticles, x, 6, 20, 24, 104, 127, 149,  150  nanostructures, x, 30, 149, 151, 152, 154,  156, 157, 158, 159, 160, 162, 163, 164,  165  nanotechnology, 148  nanotube, 123, 125  National Research Council, 126 

181

national security, 2  natural, xi, 167  Navier‐Stokes equation, viii, 56, 58  neon, 17  network, 14, 17  Newtonian, v, viii, ix, 55, 56, 58, 59, 60, 65,  66, 68, 70, 71, 73, 74, 76, 77, 78, 79, 81,  84, 88, 96  nitrate, 27, 28  nitrogen, 17  nitrogen gas, 17  noise, viii, xi, 56, 57, 104, 116, 118, 150,  158, 159, 164, 165  nonlinear, xi, 167, 173  non‐Newtonian, viii, 56, 58, 59, 81  non‐Newtonian fluid, viii, 56, 58, 59, 81  normal, 138  nuclei, 6  numerical aperture, 5 

O  observations, 43, 45, 46  ofloxacin, 123, 125  oligomers, 151  one dimension, 20  online, 52  opportunities, 125, 128  optical fiber, 36, 37, 38  optical micrographs, 45  optical properties, 6, 17, 19, 49  optics, 41, 150  optimization, 49, 119  organic, 108, 127  organic compounds, 108  organism, 44  OSA, 51  oscillation, 150  overlap, 11, 42  oxalate, 127  oxidation, 109, 114, 119, 124  oxygen, 125 

Index

182 P  packaging, 3  parabolic, 70  parallel, 7, 11, 16, 21, 41, 45, 57, 85, 173  parameter, 58, 59, 71, 87  particles, ix, 109, 110, 135, 137, 138, 139  passive, 118, 140  PDMS, v, x, xi, 104, 122, 123, 124, 135,  136, 137, 140, 142, 143, 144, 145, 150,  151, 158, 159, 164  PEEK, 138  performance, 19, 114, 118, 157  permit, 121  permittivity, 70, 91  peroxide, 124  petri dish, 43  petroleum, 126  pharmaceuticals, 123, 125  phenylalanine, 120  phosphate, 163  photoabsorption, 30, 37  photochemical, 20  photoexcitation, 106  photographs, 16  photolithography, 3  photon, 6, 13, 106, 122, 150  photonic, 13, 20, 34  photopolymerization, 138  physical properties, 24  physicists, 50  physics, vii, 80, 90  physiological, 42  piezoelectricity, 150  pitch, 20, 23, 45  planar, vii, 1  plastic, 156  platinum, 119, 123, 124, 125, 140  play, 109  PMMA, 142, 143  Poisson equation, 62  Poisson‐Boltzmann equation, 57 

polarity, 114  polarization, 125  polydimethylsiloxane, xi, 104, 122, 150, 151  polymer, x, 47, 58, 114, 138, 149, 151  polymer chains, 138  polymer electrolytes, 58  polymethylmethacrylate, 118  polystyrene, 153, 165  poor, 6, 49, 108, 110  pore, 139  portability, 105  powder, 138  power, vii, ix, 26, 27, 29, 32, 56, 59, 60, 61,  63, 64, 65, 68, 73, 75, 76, 77, 78, 79, 81,  85, 90, 91, 92, 97, 140, 142, 169, 170,  171  power‐law, vii, ix, 56, 59, 60, 61, 63, 64, 65,  68, 73, 76, 77, 78, 81, 85, 90, 91, 92, 97  pressure, viii, ix, x, 55, 56, 57, 60, 61, 72, 80,  81, 84, 85, 86, 91, 92, 93, 94, 95, 96, 97,  117, 120, 135, 139, 145, 150, 167, 170,  172, 173  problem‐solving, 131  problem‐solving strategies, 131  production, xi, 2, 105, 117, 150, 151, 164,  165  promote, 173  propagation, xi, 6, 7, 24, 25, 167, 169, 170  properties, 4, 6, 17, 19, 24, 49, 80, 109,  110, 118, 127, 137, 173  protein, 140, 163, 165  proteomics, 150  prototype, 152, 156  pulse, 4, 5, 17, 20, 24, 31, 32, 104, 119, 140  pumps, viii, 55, 57, 79, 136, 146 

Q  quanta, 123  quantum, 104, 108, 127  quantum dot, 104, 127 

Index

R  radiation, 106, 167, 170, 172, 173  radicals, 108  radius, 19, 34, 38, 39, 127  range, viii, xi, 2, 5, 13, 24, 56, 57, 58, 95,  106, 108, 122, 123, 127, 158, 163, 167  reactant, 7, 110, 120  reactions, 24, 105, 106, 108, 109, 113, 114,  119, 120, 122  reactive ion, 138  reactivity, 114, 124  reagent, 2, 14, 123, 124  recovery, 124  recycling, 125  red shift, 164  redistribution, 80  redox, 113, 120, 122  redox‐active, 122  redshift, 153  reflection, 17, 31, 167  refractive index, 5, 6, 11, 19, 20, 24, 34, 36,  37, 38, 49, 150, 154, 158, 164  refractive indices, 11, 20, 38  regular, 119, 125, 172  relationship, 60  relative size, 156  reliability, viii, 56, 57, 158  reparation, 110  repeatability, 158  requirements, 45, 108, 113, 151  reservoir, 46, 48, 58, 110, 116, 140, 171  resolution, 5, 6, 7, 12, 14, 20, 30, 49, 158  respect, 61, 81, 118  response time, 106  returns, 138  Reynolds number, 91  RIE, 138, 139  robotics, 139  robustness, 16  room temperature, x, 19, 135, 137, 163,  164 

183

roughness, 17, 19, 26, 49, 117  routing, 14  ruthenium, 104, 108, 130, 133, 134 

S  saline, 163  salt, 138  sample, viii, 2, 5, 6, 7, 9, 11, 13, 16, 17, 18,  20, 32, 36, 37, 45, 56, 57, 105, 106, 109,  110, 111, 118, 140, 151, 158  satellites, 172  scalar, 59  Scanning electron, 139  scattered light, 34  scattering, 17, 36  security, 2  selectivity, 104, 105, 106, 108, 109  self‐adaptation, 170  SEM, 22, 157, 164  semiconductor, 3, 127, 152  sensing, x, 136, 145, 149, 150, 151, 152,  153, 156, 160, 164, 165  sensitivity, 37, 106, 107, 109, 112, 116, 118,  120, 124, 125, 127, 147, 151, 158, 164,  165, 170  sensors, 79, 120, 136, 137, 147  separation, 41, 105, 109, 110, 111, 115,  116, 117, 118, 120, 121, 122, 123, 125  series, 36, 85, 119  serum, xi, 150, 151, 152, 163  serum albumin, xi, 150, 152, 163  shape, 7, 11, 19, 36, 146, 151, 152, 158, 173  shear, ix, 56, 58, 59, 60, 64, 65, 72, 79  shunts, 116  sieve plate, 139, 144  sign, 63, 83  signal transduction, 150  signals, 113, 151  signal‐to‐noise ratio, 104, 158, 159, 164,  165  silica, 26, 123, 125, 158 

184

Index

silicon, vii, x, 1, 14, 112, 137, 138, 140, 142,  144, 145, 146, 147, 149, 151, 152, 153,  156, 158, 159, 164, 165  silicon dioxide, 153  silver, 6, 20, 24, 27, 28, 151  simulation, 58, 131  sine, 58, 59, 66  singular, 92  smoothing, 45  smoothness, 19  soft lithography, 3, 136, 142, 146  software, 140  solid state, 125  solid surfaces, 68, 80  solubility, 108, 109  solvent, 138, 151  space, 5, 11, 43, 137, 138, 142  spatial, 5, 14, 20, 30, 49, 106  species, viii, 56, 57, 106, 108, 109, 110, 118,  120  specifications, 91  spectroscopy, xi, 30, 106, 150, 151  spectrum, x, 32, 33, 34, 149, 150, 151, 152,  158, 159, 160, 163, 165  speed, 11, 17, 20, 26, 27, 29, 43, 44, 47  spin, 153, 154  SPR, 151, 164, 165  stability, 16, 106, 113, 123, 125, 158  stages, 109, 110  standardization, 2  steady state, 80  storage, 11, 13  strain, 59, 60, 63, 83  strategy, 123  strength, 68, 74  stress, ix, x, 56, 59, 60, 64, 65, 72, 79, 150,  164  structuring, vii, 2  substances, 104, 108, 109, 140  substrates, 3, 14, 27, 28  supply, 7  suppression, 4  surface properties, 118 

surface roughness, 17, 19, 26  surface tension, 172  suspensions, 145  swelling, x, 136, 137, 138, 151  switching, 14, 137, 140, 142, 145  symmetry, 61, 72, 81, 87  synthesis, 136  systems, 173 

T  temperature, x, 19, 26, 28, 36, 62, 70, 79,  82, 91, 132, 135, 137, 139, 140, 146,  150, 151, 163, 164, 169  temporal, 106  tension, 172  testing, 156  thermal treatment, 24  thin film, 27, 28, 123, 125  thinning, 60  three‐dimensional, 11, 43  threshold, viii, 2, 4, 14, 30, 32  time, 171  tin, 104, 114  tin oxide, 104, 114  tolerance, 125  total internal reflection, 31  TPA, 124  tracks, 41  trade, 6  trans, 11  transduction, 150  transfer, 58, 80, 105, 108, 171  transition, 137, 147  transition temperature, 137  translation, 11, 42  transmission, x, 16, 24, 25, 47, 142, 149  transparent, vii, x, 2, 3, 4, 5, 20, 41, 48, 149,  150, 151, 158, 164  transport, 94, 118, 124, 171, 172  transpose, 60  traps, 171 

Index tryptophan, 118  two‐dimensional, 58, 61  tyrosine, 118 

U  ultraviolet, 127  uncertainty, 49  uniform, 7, 72  unique features, ix, 103  universality, 173  urine, 123  UV exposure, 6, 7, 154  UV light, 154 

V  vacuum, 70  valence, 62, 70, 82, 91  valine, 120, 124  values, 24, 59, 65, 71, 72, 73, 75, 76, 77, 79,  84, 92, 117, 118, 140  vapor, 117  variables, 119  variation, 96, 117, 139  vector, 60  velocity, viii, ix, 14, 56, 57, 58, 59, 60, 61,  63, 64, 65, 66, 67, 68, 69, 70, 71, 75, 76,  77, 79, 80, 81, 83, 84, 85, 86, 87, 88, 92,  93, 97, 109, 110  versatility, 119 

185

viscosity, ix, 56, 58, 59, 60, 63, 64, 65, 73,  74, 79, 80, 81, 83, 90, 95, 97  visible, 19, 24, 150  Vitamin C, 134  voltammetric, 118 

W  waste, 2, 111  water, 17, 43, 44, 47, 60, 80, 108, 119, 125,  127, 138, 139, 141, 153, 158, 159, 164  waveguide, 20, 24, 34, 35, 38, 39, 41, 49,  151  wavelengths, 4, 33, 163  wetting, 80, 87, 169  windows, 153  withdrawal, 140  writing, vii, 2, 3, 4, 5, 7, 8, 14, 19, 20, 24, 26,  27, 28, 30, 40, 42, 48, 49, 50, 51  writing process, 3 

Y  yield, 153  yttrium, 31 

Z  zeta potential, ix, 56, 57, 62, 68, 70, 77, 81,  82, 91, 93, 94, 95, 97  zirconia, 125, 133