Biochip Technology

Biochip Technology Biochip Technology Edited by Jing Cheng Biochip Research and Development Center State Key Laborat...

Author: Jing Cheng

279 downloads 4220 Views 10MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Biochip Technology

Biochip Technology Edited by

Jing Cheng Biochip Research and Development Center State Key Laboratory of Biomembrane and Membrane Biotechnology School of Life Sciences and Engineering Tsinghua University Beijing, The People’s Republic of China

Larry J.Kricka Department of Pathology and Laboratory Medicine University of Pennsylvania School of Medicine Philadelphia, Pennsylvania, USA

harwood academic publishers Australia • Canada • France • Germany • India • Japan • Luxembourg • Malaysia • The Netherlands • Russia • Singapore • Switzerland •

USA Publishing Office Harwood Academic Publishers A member of the Taylor & Francis Group 325 Chestnut Street, Suite 800 Philadelphia, PA 19106 Tel: (215) 6258900 Fax: (215) 625-2940 Distribution Center Harwood Academic Publishers A member of the Taylor & Francis Group 7625 Empire Drive Florence, KY 41042 Tel: (800) 634-7064 Fax: (800) 248-4724 UK Harwood Academic Publishers A member of the Taylor & Francis Group 11 New Fetter Lane London EC4P 4EE Tel: +44 (0) 171 583 9855 Fax: +44 (0) 171 842 2298 This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge's collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” Biochip Technology Copyright 2001 Harwood Academic Publishers. Printed in the United States of America. Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a database or retrieval system, without prior written permission of the publisher. First Printing, 2001. Library of Congress Cataloging-in-Publication data is available from the publisher. ISBN 0-203-30504-3 Master e-book ISBN

ISBN 0-203-33324-1 (Adobe e-Reader Format) ISBN 90-5702-613-9 (Print Edition)

CONTENTS

1

2 3

4 5 6

7 8 9 10

11 12

13 14

Preface Contributors About the Editors Microchips, Bioelectronic Chips, and Gene Chips: Microanalyzers for the Next Century Larry J.Kricka Microfabrication Processes for Silicon and Glass Chips Yuebin Ning and Glen Fitzpatrick Self-Assembled Monolayers: Applications in Surface Modification and Micropatterning Younan Xia, Byron Gates, and Yadong Yin Fabrication of Polymer Microfluidic Devices Holger Becker Non-Contact Microarraying Technologies Seth Taylor and Roeland Papen High-Throughput Arrays for Efficient Screening and Analysis Mitchell D.Eggers, Bill Balch, Stafford Brignac, James Gilmore, Michael Hogan, Terri King, Deval Lashkari, Aleksandar Milosavljevic, Tom Powdrill, and Amy Smith Electronic Manipulation of Cells on Microchip-Based Devices Xiao-Bo Wang and Jing Cheng Microfilter-Based Separation of Cells Paolo Fortina, Larry J.Kricka, and Peter Wilding Nucleic Acid Amplification in Microchips Peter Wilding Technology Options and Applications of DNA Microarrays Paolo Fortina, David Graves, Christian Stoeckert Jr., Steven McKenzie, and Saul Surrey Polypyrrole Biochip as a Versatile Tool for Biological Analyses Thierry Livache Microfabricated Devices for Integrated DNA Analysis Sundaresh N.Brahmasandra, Kalyan Handique, Madhavi Krishnan, Vijay Namasivayam, David T.Burke, Carlos H.Mastrangelo, and Mark A.Burns Plant Genome Analysis Using cDNA Microarrays Yijun Ruan, James Gilmore, and Timothy Conner Biochip-Based Portable Laboratory Jing Cheng, Lei Wu, Paul Swanson, Yarning Lai, and James O’Connell

vii ix xii

1 20

44 72 111

129 149 179 192

205 239

252 275 293

15 16 17 18

Biological Applications of Paramagnetic Particles in Chips Z.Hugh Fan and Rajan Kumar Microfabricated Biochip Platforms for Cell Analyses Jonathan Cooper, Tony Cass, Adam Steel, and Hongjun Yang Application of Enzyme Colorimetry for cDNA Microarray Detection Konan Peck and Yuh-Pyng Sher Nano-Scale Size-Based Biomolecular Separation Technology Derek Hansford, Tejal Desai, and Mauro Ferrari Index

316 336 354 370 394

PREFACE Microminiaturization is one of the fastest growing fields in the analytical sciences. Over the past ten years a diverse range of micrometer-scale devices has been fabricated in silicon and in glass, and more recently in different types of plastic. The scope of applications for the new microminiature analytical devices (biochips, microchips, “labon-a-chip”) spans analytical chemistry and the biomedical sciences. Devices for genetic testing have attracted particular attention from the microminiaturization community, and a fully integrated genetic analyzer that would accept a minute sample of whole blood and produce a result without further human intervention will soon be a reality. Lab-on-a-chip devices would act as personal laboratories that could be used for a broad range of home testing and directly contribute to health maintenance and quality of life. Today, the microchip devices are making major contributions to the drug discovery process. In this application, a capability for rapid high-throughput multiplexed analysis using low volumes of sample and reagent is paramount, and the microchip devices offer a convenient and cost-effective approach to this type of analytical process. Microarray devices (DNA chips, gene chips, microspot chips) comprising surface arrays of micrometer-sized patches of antibodies, cDNA, or oligonucleotides are also having a major impact in biomedical research; particularly in gene expression studies, mutation detection, and protein analysis. The ability of microanalyzers to accomplish complicated analytical tasks, particularly with samples containing cells, is increasing. These new devices (biochips) contain microelectrodes, microfluidic elements, and other microfabricated features that orchestrate a variety of sample manipulation and analytical steps. Against this background, the objective of this book is to provide up-to-date coverage of some of the emerging avenues of research and development in the field of microchip devices. The book contains descriptions of chip fabrication (micromachining, hotembossing, patterning), system development, microarrays (polypyrrole-based, nylon, glass), assays, cell isolation, and manipulation using microfilters and bioelectronic devices, and applications ranging from clinical testing (PCR chips, portable laboratories) to plant genome analysis to biohybrid organs. This book is intended to be a starting point for anyone interested in the possibilities and potential of the diverse opportunities afforded by microminiaturized analysis in a chip format. Jing Cheng Beijing, The People’s Republic of China Larry J.Kricka Philadelphia, Pennsylvania, USA

CONTRIBUTORS Bill Balch, Genometrix Incorporated, The Woodlands, Texas, USA Holger Becker, Jenoptik Microtechnik GmbH, Jena, Germany Sundaresh N.Brahmasandra, Department of Chemical Engineering, University of Michigan, Ann Arbor, USA Stafford Brignac, Genometrix Incorporated, The Woodlands, Texas, USA David T.Burke, Department of Human Genetics, University of Michigan, Ann Arbor, USA Mark A.Burns, Department of Chemical Engineering, University of Michigan, Ann Arbor, USA Tony Cass, Bioelectronics Division, Department of IEEE, Glasgow University, Glasgow, UK Jing Cheng, Biochip Research and Development Center, State Key Laboratory of Biomembrane and Membrane Biotechnology, School of Life Sciences and Engineering, Tsinghua University, Beijing, The People’s Republic of China, and Aviva Biosciences Corporation, San Diego, California, USA Timothy Conner, Gene Discovery and Expression Program, Agriculture Sector, Monsanto Company, St. Louis, Missouri,USA Jonathan Cooper, Bioelectronics Division, Department of IEEE, Glasgow University, Glasgow, UK Tejal Desai, Department of Bioengineering, University of Illinois at Chicago, Chicago, USA Mitchell D.Eggers, Genometrix Incorporated, The Woodlands, Texas, USA Z.Hugh Fan, ACLARA BioScience Incorporated, Mountain View, California, USA Mauro Ferrari, Biomedical Engineering Center, The Ohio State University, Columbus, USA Glen Fitzpatrick, Alberta Microelectronic Corporation, Edmonton, Alberta, Canada Paolo Fortina, Departments of Pediatrics, University of Pennsylvania School of Medicine, and Children’s Hospital of Philadelphia, 310-C Abramson Pediatric Research Center, Philadelphia, USA Byron Gates, Department of Chemistry, University of Washington, Seattle, USA James Gilmore, Genometrix Incorporated, The Woodlands, Texas, USA David Graves, Department of Chemical Engineering, School of Engineering and Applied Science, University of Pennsylvania, Philadelphia, USA Kalyan Handique, Department of Chemical Engineering, University of Michigan, Ann Arbor, USA Derek Hansford, Biomedical Engineering Center, The Ohio State University, Columbus, USA Michael Hogan, Genometrix Incorporated, The Woodlands, Texas, USA Terri King, Genometrix Incorporated, The Woodlands, Texas, USA

Larry J.Kricka, Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, USA Madhavi Krishnan, Department of Chemical Engineering, University of Michigan, Ann Arbor, USA Rajan Kumar, Sarnoff Corporation, Princeton, New Jersey, USA Yarning Lai, Biochip Research and Development Center, School of Life Sciences and Engineering, Tsinghua University, Beijing, The People’s Republic of China Deval Lashkari, Genometrix Incorporated, The Woodlands, Texas, USA Thierry Livache, CIS Bio International, DIVT, 30203 Bagnols/Ceze Cédex, France Steven McKenzie, Department of Hematology/Oncology, duPont Hospital for Children, Wilmington, Delaware, and Thomas Jefferson University School of Medicine, Philadelphia, Pennsylvania, USA Carlos H.Mastrangelo, Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA Aleksandar Milosavljevic, Genometrix Incorporated, The Woodlands, Texas, USA Vijay Namasivayam, Department of Chemical Engineering, University of Michigan, Ann Arbor, USA Yuebin Ning, Micralyne Incorporated, Edmonton, Alberta, Canada James O’Connell, Nanogen Incorporated, San Diego, California, USA Roeland Papen, Packard Instrument Company, Meriden, Connecticut, USA Konan Peck, Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan, Republic of China Tom Powdrill, Genometrix Incorporated, The Woodlands, Texas, USA Yijun Ruan, Biosource Genomics, Vacaville, California, USA Yuh-Pyng Sher, Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan, Republic of China Amy Smith, Genometrix Incorporated, The Woodlands, Texas, USA Adam Steel, Gene Logic Incorporated, Gaithersburg, Maryland, USA Christian Stoeckert Jr., Joseph Stokes Jr. Research Institute, Children’s Hospital of Philadelphia, and Center for Bioinformatics, University of Pennsylvania, Philadelphia, USA Saul Surrey, Department of Hematology/Oncology, duPont Hospital for Children, Wilmington, Delaware, and Thomas Jefferson University School of Medicine, Philadelphia, Pennsylvania, USA Paul Swanson, Nanogen Incorporated, San Diego, California, USA Seth Taylor, Packard Instrument Corporation, Meriden, Connecticut, USA Xiao-Bo Wang, Aviva Biosciences Corporation, San Diego, California, USA Peter Wilding, Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, USA Lei Wu, Nanogen Incorporated, San Diego, California, USA Younan Xia, Department of Chemistry, University of Washington, Seattle, USA Hongjun Yang, Gene Logic Incorporated, Gaithersburg, Maryland, USA Yadong Yin, Department of Chemistry, University of Washington, Seattle, USA

ABOUT THE EDITORS Jing Cheng, PhD, is the Cheung Kong Professor of Bioscience and Biotechnology at Tsinghua University (China) and Director of the Biochip Research and Development Center at Tsinghua University. Cheng received his BEng degree in Electrical Engineering from Shanghai Tiedao University (China) and his PhD degree in Forensic Sciences from the University of Strathclyde (UK). His experience includes eight years as an electrical engineer at Ziyang Internal Combustion Locomotive Factory (China) and as a lecturer in forensic sciences at Southwest University of Political Science and Law (China). He gained additional postdoctoral experience at the University of Strathclyde, the University of Aberdeen (UK) and the University of Pennsylvania (USA), where he was appointed as a research assistant professor in the School of Medicine. In 1996 he joined Nanogen Inc. in San Diego, California, as a staff scientist and engineer, where he was later promoted to principal scientist and engineer, and principal investigator. In 1999 he assumed the role of chief technology officer at Aviva Biosciences Corporation in San Diego. Cheng developed the world’s first laboratory-on-a-chip system in 1998, and the work was featured in the front cover story of the June 1998 issue of Nature Biotechnology, and also cited as the breakthrough of the year by Science that same year. He was awarded Nanogen’s most prestigious award, the NanoAward, and China’s Outstanding Young Scientist Award, both in 1999. Cheng has published over fifty peer-reviewed papers, of which over twenty are related to biochips. In addition, he holds more than twenty European and US patents and disclosures. He has presented at many international conferences. His current research interest is the development of microchip-based laboratory systems and ultra-high-throughput drug screening systems. Larry J.Kricka, DPhil, FRSC, CChem, FRC Path, is professor of Pathology and Laboratory Medicine at the University of Pennsylvania and director of the General Chemistry Laboratory at the Hospital of the University of Pennsylvania. He received his BA and DPhil degrees in chemistry from York University (UK), and after completing postdoctoral training at the University of Liverpool (UK), he joined the faculty in the Department of Clinical Chemistry and Wolfson Research Laboratories at the University of Birmingham (UK), where he was a reader in Clinical Chemistry. Kricka is a fellow of the Royal College of Pathologists and the Royal Society of Chemistry, and a member of the Association of Clinical Biochemists. Kricka is currently president elect of the American Association for Clinical Chemistry (AACC). At the international level, he serves as chair of the Working Group on Microtechnology of the International Federation of Clinical Chemistry (IFCC). His research interests include the analytical applications of bioluminescence and chemiluminescence, nonisotopic immunoassays, micromachined analytical systems, and heterophile antibodies. Kricka has lectured extensively and published over 250 papers and review articles, and authored or edited twelve books. He is editor-in-chief of the

Journal of Bioluminescence and Chemiluminescence and a member of the editorial board of Analytical Biochemistry.

1 Microchips, Bioelectronic Chips, and Gene Chips Microanalyzers for the Next Century Larry J.Kricka

INTRODUCTION An important direction in the development of analytical techniques is toward microminiaturized analyzers. Generic names for these new micrometer-featured devices include “micro-total analytical system” (µ-TAS) (Manz et al., 1990a), lab-on-a-chip (Colyer et al., 1997; Moser et al., 1995), biochip, or, simply “chip.” In some cases devices have been named based on their particular application, for example, PCR chips, gene chips, while for others the device is named for a characteristic structural feature, for example, microspot or microarray (Table 1). The common theme for all of these devices is the microminiaturization of an analytical process or part of an analytical process into a device built on a small piece of glass, plastic, or silicon (Beattie et al., 1995a; Becker and Manz, 1996, Berg and Bergveld, 1995; Berg and Lammerink, 1998; Collins and Jacobson, 1998; Hacia et al., 1998a; Kopp et al., 1997; Kricka, 1998a,b; Manz, 1998; Ramsay, 1998). Several factors can be identified as underpinning the renewed interest in microminiaturization of analyzers. First, a range of analytical problems has emerged for which microminiaturization has obvious benefits, and these include high-throughput massively parallel testing for drug discovery (Devlin, 1997), small hand-held portable analyzers for point-of-care testing (e.g., clinical testing or biowarfare monitoring) (Kost, 1995), and lightweight analyzers for use on space exploration missions where payload is limited. Second, miniaturization offers a route to cost reduction in analytical processes because the amount of reagent used per assay can be drastically reduced compared to conventional analysis. Similarly, in drug discovery, where there is often only a limited amount of candidate drug compound, a reduction in the volume of sample tested translates into a larger number of tests with that particular compound. Finally, an important advantage of microminiaturization is the ability to integrate all of the steps in a complex multistep analytical process onto a single device. This finds a natural parallel with integrated electronic circuits produced on silicon wafers for the electronics industry. In these devices, thousands to millions of individual components are integrated into a single chip (e.g., an Intel Pentium III is produced using a 0.25-µm manufac-

Biochip technology

2

Table 1 Nomenclature of Analytical Microchip Devices IVF chip Oligonucleotide chip Biochipa Biologic chip Lab-chip PCR chip b DNA chip Laboratory on a chip ProteinChip® ™ Lab-on-a-chip DNA MassArray SpectroChip™ Expression chip LifeGEM™ Sperm chip ® ™ LivingChip UniGEM™ FeverChip µ-TAS Gene chip Mesoscale devicec ™ Microarray Gene chip Genosensor Microspot® aThis term was originally used to refer to biological versions of electronic

microchips (Tucker 1984), bThis term was originally used in the context of computer-based

experiments (Steben 1987). cMesoscale refers to an intermediate scale, between that of large and small dimensions. turing process and the CPU includes over 9.5 million transistors) (http://developer.intel.com/design/PentiumIII/prodbref/). This chapter provides an introduction to microchip analyzers, their fabrication and applications, and discusses future trends in this emerging analytical science.

HISTORICAL PERSPECTIVE Current microanalyzers owe much to the early work of the micromachinists who were intrigued with the possibility of using silicon as a material for constructing different types of mechanical and microelectromechanical (MEMS) devices. They showed that it was possible to construct complex micrometer-sized devices such as cogs, movable mirrors, spanners, and more complex devices including an electric motor from micromachined silicon components (Amoto, 1989; Angell et al., 1983; Mallon, 1992; Petersen, 1982; Stix, 1992). Practical devices based on micromachined components have emerged including sensors for measuring blood pressure and fuel flow in automobile engines, and a device that triggers airbags in automobiles. The latter has enjoyed considerable success and is based on a microfabricated silicon beam that bends under acceleration forces. Deflection of the beam is detected, and this triggers release of the airbag (Bryzek et al., 1994). One of the earliest microanalyzers was fabricated by Terry and colleagues. They constructed a gas chromatograph (GC) on the surface of a 2-inch silicon wafer that was then bonded to a glass plate (Terry et al., 1979). There was then a hiatus of several years before interest was renewed in microanalytical devices. The next important landmarks in

Microanalyzers for the next century

3

microanalyzer technology were in the early 1980s, with the development of the microphysiometer and the i-STAT analyzer. The microphysiometer is based on micromachined silicon, 50 µm square × 50 µm deep wells that incorporate a lightaddressable pH sensor (McConnell et al., 1992; Owicki and Parce, 1990; Parce et al., 1989). This device was designed to assess cell metabolism for toxicity studies of new drug compounds. The i-STAT analyzer utilizes a dispos-able cartridge that contains an array of microelectrodes and immobilized enzyme electrodes on a silicon microchip for whole blood analysis (e.g., blood gases, electrolytes, glucose, hematocrit) (Lauks et al., 1992). By the end of the 1980s, research-and-development efforts directed toward microanalytical devices experienced a growth spurt. Some of the diverse range of analyzers, devices, tests, and procedures are listed in Table 2.

ADVANTAGES AND LIMITATIONS OF MICROANALYZERS There are a series of compelling reasons why microanalyzers will find widespread use for analysis (Table 3). Microminiature analyzers are small and compact and thus suitable for use in non-laboratory settings (e.g., point-of-care testing) where hand-held portable analyzers are required. Miniaturized arrays of different re-agents on planar surfaces (e.g., plastic, glass, or silicon) permit simultaneous testing of a sample for specific components. The volume of sample required for analysis is reduced in microanalyzers (e.g., nL–pL volumes), and this is beneficial in a clinical setting as it reduces the amount of blood that must be drawn from a patient. There is also a reduction in the volume of reagent required per test, and this provides an economic benefit.

Table 2 Micromachined Analyzers, Devices,and Assays Analyzers and devices Biocapusule (Desai et al., 1998) Capillary electrophoresis analyzer (Jacobson and Ramsey, 1995; Seiler et al., 1993) Controlled release system (Sheppard et al., 1996) Blood gas analyzer (Arquint et al., 1994; Shoji and Esashi, 1995) Electrochemiluminescence detector (Arora et al., 1997) Electrolyte analyzer (Moritz et al., 1993) Electroporation system (Murakami et al., 1993) Flow-injection analyzer (Manz et al., 1991; Suda et al., 1993) Gas chromatograph (Terry and Hawker 1983; Terry et al., 1979) Haemorheometer (Tracey et al., 1995) In vitro fertilization chamber (Kricka et al., 1995) Liquid chromatograph (Manz et al., 1990b; Ross et al., 1998; Xue et al., 1997a,b) Thermal cycler (Northrup et al., 1996; Wilding et al., 1994) Test or procedure Antibody analysis (Rodriguez et al., 1997)

Biochip technology

4

Cell movement and responses (Oakley and Brunette, 1995) Cell traction force (Galbraith and Sheetz, 1997) DNA analysis (Shalon et al., 1996; Sheldon et al., 1993) DNA sequencing (Drobyshev et al., 1997; Southern, 1996) Expression monitoring (Schena et al., 1995; Lockhart et al., 1996; Wodicka et al., 1997) Immunoassay (Koutny et al., 1996; von Heeren et al., 1996; Song et al., 1994) Mutation testing (Gingeras et al., 1996; Hacia, 1999; Hacia et al., 1997) Nerve regeneration (Zhao et al., 1997) Nucleic acid hybridization (Beattie et al., 1995b; Fodor, 1993; Southern et al., 1999) PCR (Belgrader et al., 1998; Cheng et al., 1996a; Kopp et al., 1998; Waters et al., 1998a,b) Semen testing (Kricka et al., 1997) Serum protein analysis (Colyer et al., 1997) Topographic guidance of cells (Oakley et al., 1997)

Table 3 Advantages and Disadvantages of Microanalyzers Disposable Advantages Portable Fast response time Low power consumption Disadvantages Low production costs Human interface Mass production Obtaining a representative sample Diverse range of applications Exceeding the analytical detection Integration of steps in an analytical limit process Fabrication of microanalyzers derives benefit from the manufacturing processes used in the microelectronics industry that are geared to high-volume production. Many different designs can be simultaneously fabricated on the same wafer and then tested. This allows rapid design cycles and the potential for more design iterations than would be normally possible for a macroscale device. Microanalyzers can improve analytical reliability through multiple test sites for simultaneous parallel assays. This degree of redundancy provides an analytical safeguard that cannot be easily achieved in macroscale analyzers, where duplicate assays represent the normal extent of repetitive assay of a specimen. Encapsulation of microscale devices provides extended operation over a wider range of environmental conditions of humidity and temperature than can be achieved with a conventional analyzer. One of key advantage of microanalyzers is the opportunity to integrate all of the steps in a complex multistep analytical process into a single device. The scale of a microchip is such that it is feasible to design structures to perform individual tasks, including sample addition, processing, analysis, and read-out of the results, all on a microchip that is 2×2


5

cm or smaller. An even greater degree of integration is achieved by further combination of analytical steps into individual microstructures on the microchip (e.g., cell separation and nucleic acid amplification). Added to this is the availability of a large number of microminiaturized components (e.g., lasers, pumps, valves) that enhance the capabilities of the device. Table 4 lists some miniaturized components that are available for incorporation into microchip analyzers. This, of course, also includes the electronic circuitry to operate and control the analytical process, which would easily fit onto the surface of the type of devices currently being developed. Microminiaturization does have some disadvantages. As the size of a sample is successively decreased, an immediate concern relates to how representative the sample is of the specimen from which it was derived. This is a problem for inhomogeneous biological specimens that contain a diversity of constituents (e.g., cells, proteins, lipids). For example, a submicroliter blood sample is unlikely to contain rare cells such as trophoblasts in maternal circulation, which may only be present at one per million or one per ten million cells. This problem can be overcome by developing flow-through sampling in which larger volumes of sample are flowed through a low volumemicrominiature device. Another issue that arises as the volume of the sample is reduced is that of detectability. If an analyte is present at only 1 femtomole/L in the original specimen, then a 1 µL sample contains 600 molecules (1×10–15×10–6×6 1023). Further reduction of the sample

Table 4 Microminiaturized Components Accelerometer Microbeam Air turbine Microbearing Anemometer Microbridge Cables Microflexible arm Cantilever Micromotor Diaphragm Microphone Flow sensor Micropipette Fuse Microturbine Gears Mirror Peltier heater/cooler Hinge Laser Pirani pressure gauge Membrane Pressure sensor

Pump Refrigerator Relay Resonator Robot Screw SFM and STM tips Sieve Solenoid Tweezer Vacuum tube Valve

size to 1 nL produces a sample that contains one thousand times fewer molecules, that is, less than one molecule, and this would not be detectable (Petersen et al., 1998).

FABRICATING MICROCHIPS Microfabrication methods used to make the different types of microanalyzers are summarized in Table 5 (Qin et al., 1998). In many cases the basic technology has been adapted from the microelectronics industry (e.g., photolithography for glass and silicon

Biochip technology

6

devices) or from the printing industry (e.g., ink jet printing). The size of the features that can be fabricated are in the micrometer range for photolithographic, molding, and printing methods and in the nanometer range for patterning. An important current direction in microfabrication is the manufacture of plastic microchips (Becker and Dietz, 1998; Ford et al., 1999; Friedrich and Vasile, 1996; Friedrich et al., 1997; McCormick et al., 1997). These may be easier to manufacture and at lower cost than glass or siliconglass chips and, additionally, may provide greater flexibility in design. Fabrication of microanalyzers also requires ancillary processes for assembling the microcomponents (e.g., anodic and thermal bonding), and methods to introduce directaccess ports into structures formed by bonding microparts together (e.g., mechanical, ultrasonic, and laser drilling) (Shoji and Esashi, 1995). Handling and manipulating very small microchips is difficult, but this can be overcome by packaging the microchip into a substantially larger holder or by mounting one or more microchips onto a platform.

ON-CHIP DETECTION METHODS Fluorescent detection methods currently dominate microchip analyses. Laser-induced fluorescence (LIF) is widely used with capillary electrophoresis chips to detect separated components (Cheng et al., 1996b; Effenhauser et al., 1993; Harrison et al., 1993). Confocal fluorescence microscopy is the most common detection

Table 5 Materials and Fabrication Processes Materials Acrylic copolymer (McCormick et al., 1997) Glass (Effenhauser et al., 1993) Photoresist (Gorowitz and Saia, 1984) Polyacrylamide (Proudnikov et al., 1998) Polycarbonate (jenoptik Mikrotechnik, Jena, Germany) Poly(dimethylsiloxane) (Qin et al., 1998) Polymethyl methacrylate (Jenoptik Mikrotechnik Jena, Germany) Polypropylene (Matson et al., 1995) Quartz (Danel and Delapierre, 1991) Silicon (Petersen, 1982) Processes Anodic bonding (Spangler and Wise, 1990) Contact printing (Jackman et al., 1995) Covalent bonding (Drobyshev et al., 1997) Deposition (Beattie et al., 1995a; Shalon et al., 1996) Electrochemical micromachining (Datta, 1995) Embossing (Becker and Dietz, 1998) Injection molding (McCormick et al., 1997) Ink jet printing (De Saizieu et al., 1998) ln-situ synthesis (Fodor et al., 1994; Southern, 1996)


7

Laser ablation (Hennink, 1997; Zimmer et al., 1996) LIGA (Lithographie, Galvanoformung, Abformung) (White et al., 1995) Microcontact printing (Kane et al., 1999) Micromilling (Friedrich et al., 1997; Friedrich and Vasile, 1996) Pattering . (Sleytr et al., 1992) Pattern transfer (Xia et al., 1996) Reactive ion etching (Gorowitz and Saia, 1984) Silicon fusion bonding (Petersen et al., 1991) Thermal bonding (Lasky, 1986) Ultrasonic impact grinding (Qin et al., 1998) Wet-etching (Petersen, 1982) method for assessing antibody-antigen binding and hybridization on microarrays. This technique is highly sensitive and can detect 5–10 fluorescein labels per µm2 (Chu et al., 1996; Sheldon et al., 1993). Fluorescence has also been used for TaqMantype assays in arrays of glass microwells in combination with a charged-coupled device (CCD) (Taylor et al., 1998). Both one- and two-color fluorescence procedures have been devised for use with microarrays. For example, in the microspot assay, the capture antibody is labeled with Texas Red and the detection antibody is labeled with fluorescein (Chu et al., 1996; Ekins, 1998), whereas in the gene expression assays the test and control are labeled with lissamine and fluorescein or with Cy 3 and fluorescein, respectively (DeRisi et al., 1996; Hacia et al., 1998b). An alternative two-color detection strategy employs a βgalactosidase label and an alkaline phosphatase label detected using X-Gal and Fast Red TR/naphthol ASMX substrates (Chen et al., 1998). Chemiluminescence methods have also been used to study reactions in microchips (Kricka et al., 1994), for example, genetic (Rajeevan et al., 1999) and immu-nological assays performed in a microarray format (Dzgoev et al., 1996). Other detection options include electrochemiluminescence (Arora et al., 1997), and electrochemical methods (Murakami et al., 1993). Qualitative methods also have a role to play in microanalytical methods. For example, simple visual inspection using a reflecting microscope has been used to assess agglutination and to monitor the progress of human sperm in glass and silicon microchannels (Kricka et al., 1993, 1997).

APPLICATIONS AND TYPES OF MICROCHIP ANALYZERS Microchip analyzers can be broadly subdivided into microfluidic devices, array-based analyzers, and bioelectronic chips. An alternative classification divides chips into active and passive types. Active chips contain components that actively manipulate the sample (e.g., electrodes). Passive chips do not contain such components, and simple microfluidic devices or a microarray would fall into this category.

Biochip technology

8

Microfluidic Devices Microfluidic devices utilize µm-sized chambers, channels, filters, and other structural features to manipulate µL and sub-µL volumes of sample and reagents. It is also possible to incorporate micromachined pumps and valves to direct and control fluid flow (Berg et al., 1998). Fluid flow can also be controlled via electroosmotic pumping using electrodes contacted with the chip (Harrison et al., 1991). Array-Based Analyzers These contain microminiature arrays of reagents (antigens, antibodies, oligonucleotides, cDNA) on the surface of a small piece of glass or other material (Beattie et al., 1995a; Brown and Botstein, 1999; Chetverin and Kramer, 1994; Cheung et al., 1999; Ekins, 1998; Fodor, 1993; Hoheisel, 1997; Lemieux et al., 1998; Livache et al., 1998a,c; Macas et al., 1998; Marshall and Hodgson, 1998; Martin et al., 1998; Matson et al., 1995; McConnell et al., 1999; Nelson, 1996; Schena et al., 1998; Shalon et al., 1996; Shoji and Esashi, 1995; Watson et al., 1998). The size of individual locations in an array can be as small as 10×10 µm, and the number of discrete locations in the array can be greater than 40,000. Individual reagents in the array react with specific molecules contained in the sample applied to the array, and the pattern of reaction provides information on the composition of the sample. For example, specific oligonucleotide probes arrayed on a glass surface (1.28×1.28 cm) are used to detect specific sequences of nucleic acids or nucleic acid fragments, for purposes of sequencing, resequencing, strain identification, or monitoring of gene expression (see Table 2). Bioelectronic Chips Various types of bioelectronic chip have been devised that combine microfluidic and electrical manipulations of sample and reagents. In some devices, µm-sized electrodes serve as discrete predetermined locations for immobilization of specific reagents, for example, vinyl pyrrole derivatized oligonucletotides (Livache et al., 1998b). In others, the electrodes facilitate nucleic acid hybridization or are used to manipulate cells (e.g., isolation, enrichment, and lysis) prior to genetic analysis (Cheng et al., 1998). A further level of sophistication is provided by a microfabricated silicon chip (250×250×250 µm) that contains a transponder that can be programmed with a unique identifier. The outside surface of these devices are coated with different molecular recognition reagents (e.g., DNA probes). Individual devices that react with specific targets in a sample can be identified by monitoring the surface of the cube and by interrogating the transponder for the unique identifier (radio frequency signal) (Mandecki, 1998). Integration of analytical functions is an important goal in all types of microchip analyzers, and various strategies have evolved (Freaney et al., 1997; Manz et al., 1991; Ocvirk et al., 1995; Raymond et al., 1994; Seiler et al., 1993; Waters et al., 1998b; Wilding et al., 1998). Individual chips can be stacked one on top of another and interconnected such that each layer performs a particular task in a sequence of analytical


9

steps (Fettinger et al., 1993). Alternatively, a microchip or groups of microchips (one per analytical operation) can be housed on a platform that provides external fluidic connections to the microchips for serial sample and reagent addition. Finally, the entire set of analytical steps can be performed on a single microchip that has an integrated structure comprising the individual structures required to perform the serial steps in the analytical process (e.g., sampling, sample preparation, analytical reaction, measurement, data processing). The scope of microchip integration is expanding and includes the combination of sample preparation (e.g., cell isolation) (Wilding et al., 1998), analytical reactions (e.g., polymerase chain reaction, DNA restriction), and detection reactions (e.g., capillary electrophoresis analysis) (Waters et al., 1998b). The ultimate goal is a single self-contained device that only requires the operator to add the sample. It would contain all of the necessary reagents and detection systems, and would perform the appropriate data analysis and data output functions.

CONCLUSIONS AND FUTURE DEVELOPMENTS The rate of progress in the design, construction and testing of microminiature analyzers has been rapid. It has spawned a large number of start-up companies (e.g., ACLARA Biosciences, Advanced BioAnalaytical Services, Affymax, Affymetrix, Caliper Technologies, Cepheid, Incyte, Clinical Microsystems, Microcosm Technologies, Micronics, Nanogen, Orchid Biocomputer, Sequenom, Synteni), a specialized journal (Biomedical Microdevices, http://www.wkap.nl), a series of websites (www.genechip.com), and a burgeoning portfolio of patents protecting this type of technology as exemplified by the selection of recent U.S. patents listed in Table 6. Microchips represent the first step in a miniaturization process that will ultimately lead to still smaller devices constructed at the nanometer-scale (1 nanometer=10–9 m) from individual atoms and molecules (Crandall and Lewis, 1992; Drexler, 1986,1991; Fahy, 1993; Kaehler, 1994; Murphy et al., 1994). As yet there are no practical examples of a nanochip, but the rapid progress that is being made in

Table 6 Recent United States Patents on Biochips, Microchips, Microfluidics, and Microanalysis 5,942,443High throughput screening assay systems in microscale fluidic devices 5,922,591Integrated nucleic acid diagnostic device 5,880,071Electropipettor and compensation means for electrophoretic bias 5,876,675Microfluidic devices and systems 5,874,214Remotely programmable matrices with memories 5,872,010Microscale fluid handling system 5,869,004Methods and apparatus for in-situ concentration and/or dilution of materials in microfluidic systems 5,866,345Apparatus for detection of an analyte utilizing mesoscale flow

Biochip technology

10

systems 5,863,708Partitioned microelectronic device array 5,858,804Immunological assay conducted in a microlaboratory array 5,858,195Apparatus and method for performing microfluidic manipulations for chemical analysis and synthesis 5,858,187Apparatus and method for performing electrodynamic focusing on a microchip 5,856,174Integrated nucleic acid diagnostic device 5,852,495Fourier detection of species migrating in a microchannel 5,849,208Making apparatus for conducting biochemical analyses 5,846,396Liquid distribution system 5,842,787Microfluidic systems incorporating varied channel dimensions 5,779,868Electropipettor and compensation means for electrophoretic bias 5,770,029Integrated electrophoretic microdevices 5,755,942Partitioned microelectronic device array 5,744,366Mesoscale devices and methods for analysis of motile cells 5,726,026Mesoscale sample preparation device and systems for determination and processing of analytes 5,716,825Integrated nucleic acid analysis system for MALDI-TOF MS 5,699,157Fourier detection of species migrating in a microchannel 5,681,484Etching to form crossover nonintersecting channel networks for use in partitioned microelectronic and fluidic device arrays for clinical diagnostics and chemical synthesis 5,661,028Large-scale DNA microsequencing device 5,645,702Low-voltage miniaturized column analytical apparatus and method 5,643,738Method for synthesis of plurality of compounds in parallel using a partitioned solid support 5,637,469Methods and apparatus for detection of an analyte utilizing mesoscale flow systems 5,635,358Fluid handling methods for use in mesoscale analytical devices 5,632,876Apparatus and methods for controlling fluid flow in microchannels 5,603,351Method and system for inhibiting cross-contamination in fluids of combinatorial chemistry device 5,593,838Partitioned microelectronic device array 5,587,128Mesoscale polynucleotide amplification devices 5,585,069Partitioned microelectronic and fluidic device array for clinical diagnostics and chemical synthesis 5,583,281Microminiature gas chromatograph 5,498,392Mesoscale polynucleotide amplification device and method 5,492,867Method for manufacturing a miniaturized solid-state mass spectrograph


11

5,486,335Analysis based on flow restriction 5,427,946Mesoscale sperm handling devices 5,304,487Fluid handling in mesoscale analytical devices 5,296,375Mesoscale sperm handling devices 5,126,978Undersea data collection, analysis, and display system 4,935,040Miniature devices useful for gas chromatography molecular self-assembly and related techniques provides the first steps toward a technological base for this futuristic technology (Mao and Richards, 1999; Stevens et al., 1997).

REFERENCES Amoto, I. 1989. Small things considered: Scientists craft machines that seem impossibly tiny. Sci. News 136:8–10. Angell, J.B., S.C.Terry and P.W.Barth. 1983. Silicon micromechanical devices. Scient. Am. 248:36–47. Arora, A., A.J.De Mello, and A.Manz. 1997. Sub-microliter electrochemiluminescence detector: A model for small volume analysis systems. Anal. Commun. 34:393–395. Arquint, P., M.Koudelka-Hep, B.H.van der Schoot, P.van der Wal, and N.E de Rooij. 1994. Micromachined analyzers on a silicon chip. Clin. Chem. 40:1805–1809. Beattie, K.L., W.G.Beattie, L.Meng, S.L.Turner, R.Coral-Vazquez, D.D.Smith, P.M.Mcintyre, and D.D.Dao. 1995a. Advances in genosensor research. Clin. Chem. 41:700–706. Beattie, W.G., L.Meng, S.L.Turner, R.S.Varma, D.D.Dao, and K.L.Beattie. 1995b. Hybridization of DNA targets to glass-tethered oligonucleotide probes. Mol. Biotechnol. 4:213–225. Becker, H., and A.Manz. 1996. Miniaturised total analysis systems: Microsystem technology entering the life sciences. Sci. Progr. 79:49–63. Becker, H., and W.Dietz. 1998. Microfluidic devices for µ-TAS applications fabricated by polymer hot embossing. Proc. SPIE Microfluid. Devices Syst. 3515:177–182. Belgrader, P., W.Benett, D.Hadley, G.Long, R.Mariella Jr., F.Milanovich, S.Nasarabadi, W. Nelson, J.Richards, and P.Stratton. 1998. Rapid pathogen detection using a microchip PCR array instrument. Clin. Chem. 44:2191–194. Berg, A.van den, and P.Bergveld. 1995. Micro Total Analytical Systems. Kluwer, Dordrecht. Berg, A.van den, and T.S.J.Lammerink. 1998. Micro total analysis systems: Microfluidic aspects, integration concept and applications. In Microsystem Technology in Chemistry and Life Sciences. A.Manz and H.Becker, eds. Springer, Berlin, pp. 21–49. Brown, P.O., and D.Botstein. 1999. Exploring the new world of the genome with DNA microarrays. Nature Genet. 21:33–37. Bryzek, J., K.Petersen, and W.McCulley. 1994. Micromachines on the march. IEEE Spectrum 31:20–31.

Biochip technology

12

Chen, J.J.W., R.Wu, P.C.Yang, J.Y.Huang, Y.P.Sher, M.H.Han, W.C.Kao, P.J.Lee, T.F.Chiu, F.Chang, Y.W.Chu, C.W.Wu, and K.Peck. 1998. Profiling expression patterns and isolating differentially expressed genes by cDNA microarray system with colorimetry detection. Genomics 51:313–324. Cheng, J., M.A.Shoffner, G.E.Hvichia, L.J.Kricka, and P.Wilding. 1996a. Chip PCR, II: Investigation of different PCR amplification systems in microfabricated silicon-glass chips. Nucleic Acids Res. 24:380–385. Cheng, J., M.A.Shoffner, K.R.Mitchelson, L.J.Kricka, and P. Wilding. 1996b. Analysis of ligase chain reaction products amplified in a silicon-glass chip using capillary electrophoresis. J. Chromatogr. A 732:151–158. Cheng, J., E.L.Sheldon, L.Wu, A.Uribe, L.O.Gerrue, J.Carrino, M.J.Heller, and J.P O’Connell. 1998. Preparation and hybridization analysis of DNA/RNA from E. coli on microfabricated bioelectronic chips. Nature Biotechnol. 16:541–546. Chetverin, A.B., and F.R.Kramer. 1994. Oligonucleotide arrays: New concepts and possibilities. Bio/Technol. 12:1093–1099. Cheung, V.G., M.Morley, F.Aguilar, A.Massimi, R.Kucherlapati, and G.Childs. 1999. Making and reading microarrays. Nature Genet. 21:15–19. Chu, F., P.Edwards, R.P.Ekins, H.Berger, and P.Finch. 1996. Microarray-based immunoassays. 211th Am. Chem. Soc. Natl. Mtg. 211:16. Collins, M., and B.Jacobson. 1998. Evaluating gene expression in SJL mice undergoing relapsing EAE using high-density oligonucleotide arrays (gene chip). 5th Int. Congr. Int. Soc. Neuroimmunol. 90:57. Colyer, C.L., T.Tang, N.Chiem, and D.J.Harrison. 1997. Clinical potential of microchip capillary electrophoresis systems. Electrophoresis 18:1733–1741. Crandall, B.C., and J.Lewis. 1992. Nanotechnology: Research and Perspectives. MIT Press, Cambridge. Danel, J.S., and G.Delapierre. 1991. Quartz: A material for microdevices. J. Micromech. Miocroeng. 1:187–198. Datta, M. 1995. Fabrication of an array of precision nozzles by through-mask electrochemical micromachining. J. Electrochem. Soc. 142:3801–3805. De Saizieu, A., U.Certa, J.Warrington, C.Gray, W.Keck, and J.Mous. 1998. Bacterial transcript imaging by hybridization of total RNA to oligonucleotide arrays. Nature Biotechnol. 16:45–48. DeRisi, J., L.Penland, P.O.Brown, M.L.Bittner, P.S.Meltzer, M.Ray, Y.Chen, Y.A.Su, and J.M. Trent. 1996. Use of a cDNA microarray to analyse gene expression patterns in human cancer. Nature Genet. 14:457–460. Desai, T.A., W.H.Chu, J.K.Tu, G.M.Beattie, A.Hayek, and M.Ferrari. 1998. Microfabricated immunoisolating biocapsules . Biotechnol. Bioeng. 57:118–120. Devlin, J.P., ed. 1997. High Throughput Screening. Dekker, New York. Drexler, K.E. 1986. Engines of Creation. Anchor Press/Doubleday, Garden City, NY. Drexler, K.E. 1991. Molecular directions in nanotechnology. Nanotechnology 2:113–118. Drobyshev, A., N.Mologina, V.Shik, D.Pobedimskaya, G.Yershov, and A.Mirzabekov. 1997. Sequence analysis by hybridization with oligonucleotide microchip: Identification of β-thalassemia mutations. Gene 188:45–52. Dzgoev, A., M.Mecklenburg, P.-O.Larsson, and B.Danielsson. 1996. Microformat


13

imaging ELISA for pesticide determination. Anal. Chem. 68:3364–3369. Effenhauser, C.S., A.Manz, and H.M.Widmer. 1993. Glass chips for high-speed capillary electrophoresis separations with submicrometer plate heights. Anal Chem. 65:2637– 2642. Ekins, R.P. 1998. Ligand assays: From electrophoresis to miniaturized microarrays. Clin. Chem. 44:2015–2030. Fahy, G.M. 1993. Molecular nanotechnology. Clin. Chem. 39:2011–2016. Fettinger, J.C., A.Manz, H.Ludi, and H.M.Widmer. 1993. Stacked modules for micro flow systems in chemical analysis: Concept and studies using an enlarged model. Sens. Actuators B17:19–25. Fodor, S.P.A., X.C.Huang, and R.J.Lipshutz. 1993. Oligonucleotide arrays and sequence analysis by hybridization. 8th San Diego Conf. Beyond DNA Probes 40:653. Fodor, S.P.A., R.J.Lipshutz, and R.P.Rava. 1994. Oligonucleotide arrays and sequence analysis by hybridization. 207th Natl. Mtg. Am. Chem. Soc. 207:44. Ford, S.M., J.Davies, B.Kar, S.D.Qi, S.McWhorter, S.A.Soper, and C.K.Malek. 1999. Micromachining in plastics using X-ray lithography for the fabrication of microelectrophoresis devices.J. Biomech. Eng. 121:13–21. Freaney, R., A.McShane, T.V.Keaveny, M.McKenna, K.Rabenstein, F.W.Scheller, D.Pfeiffer, G.Urban, I.Moser, G.Jobst, A.Manz, E.Verpoorte, M.W.Widmer, D.Diamond, E. Dempsey, F.J.Saez de Viteri, and M.Smyth. 1997. Novel instrumentation for real-time monitoring using miniaturized flow systems with integrated biosensors. Ann. Clin. Biochem. 34:291–302. Friedrich, C.R., and M.J.Vasile. 1996. Development of the micromilling process for highaspect-ratio microstructures. J. Microelectromech. Syst. 5:33–38. Friedrich, C.R., P.J.Coane, and M.J.Vasile. 1997. Micromilling development and applications for microfabrication. Microelectron. Eng. 35:367–372. Galbraith, C.G., and M.P.Sheetz. 1997. A micromachined device provides a new bend on fibroblast traction forces. Proc. Natl Acad. Sci. U.S.A. 94:9114–9118. Gingeras, T.R., A.Berno, M.Chee, and J.Drenkow. 1996. Detection of rifampin conferring mutations and mycobacteria speciation using Myco GeneChip. 11th Int. Conf. AIDS, pp. 218–219. Gorowitz, B., and R.J.Saia. 1984. Reactive ion etching. Plasma Process. VLSI 8:297– 339. Hacia, J.G. 1999. Resequencing and mutational analysis using oligonucleotide microarrays. Nature Genet. 21:42–47. Hacia, J.G., K.Edgemon, N.Hunt, S.P.A.Fodor, L.C.Brody, and F.S.Collins. 1997. Mutation screening and phylogenetic analysis of hereditary breast cancer genes using high density oligonucleotide arrays. 47th Annu. Mtg. Am. Soc. Hum. Gen. 61. Hacia, J.G., L.C.Brody, and F.S.Collins. 1998a. Applications of DNA chips for genomic analysis. Mol Psych. 3:483–492. Hacia, J.G., K.Edgemon, B.Sun, D.Stern, S.P.A.Fodor, and F.S.Collins. 1998b. Two color hybridization analysis using high density oligonucleotide arrays and energy transfer dyes. Nucleic Acids Res. 26:3865–3866. Harrison, D.J., A.Manz, and P.G.Glavina. 1991. Electroosmotic pumping within a chemical sensor system integrated on silicon. Transducers ‘91. Dig. Tech. Papers, pp.

Biochip technology

14

792–795. Harrison, D.J., K.Fluri, K.Seiler, Fan Zhonghui, C.S.Effenhauser, and A.Manz. 1993. Micromachining a miniaturized capillary electrophoresis-based chemical analysis system on a chip. Science 261:895–897. Hennink, S.C. 1997. Machining lasers. Photonics Spectra (November), pp. 116–118. Hoheisel, J.D. 1997. Oligomer-chip technology. Trends Biotechnol 15:465–469. Jackman, R.J., J.L.Wilbur, and G.M.Whitesides. 1995. Fabrication of submicrometer features on curved substrates by microcontact printing. Science 269:664–666. Jacobson, S.C., and J.M.Ramsey. 1995. Microchip electrophoresis with sample stacking. Electrophoresis 16:481–486. Kaehler, T. 1994. Nanotechnology: Basic concepts and definitions. Clin. Chem. 40:1797– 1799. Kane, R.S., S.Takayama, E.Ostuni, D.E.Ingber, and G.M.Whitesides. 1999. Patterning proteins and cells using soft lithography. Biomaterials 20:2363–2376. Kopp, M.U., H.J.Crabtree, and A.Manz. 1997. Developments in technology and applications of microsystems. Curr. Opin. Chem. Biol. 1:410–419. Kopp, M.U., A.J.Mello, and A.Manz. 1998. Chemical amplification: Continuous-flow PCR on a chip. Science 280:1046–1048. Kost, G.J. 1995. Guidelines for point-of-care testing. Improving patient outcomes. Am. J. Clin. Pathol 104(Suppl. 1):S111-S127. Koutny, L.B., D.Schmalzing, T.A.Taylor, and M.Fuchs. 1996. Microchip electrophoretic immunoassay for serum cortisol. Anal Chem. 68:18–22. Kricka, L.J. 1998a. Miniaturization of analytical systems. Clin. Chem. 44:2008–2014. Kricka, L.J. 1998b. Revolution on a square centimeter. Nature Biotechnol. 16:513–514. Kricka, L.J., O.Nozaki, S.Heyner, W.T.Garside, and P.Wilding. 1993. Applications of a microfabricated device for evaluating sperm function. Clin. Chem. 39:1944–1947. Kricka, L.J., X.Ji, O.Nozaki, and P.Wilding. 1994. Imaging of chemiluminescent reactions in mesoscale silicon-glass microstructures. J. Biolumin. Chemilumin. 9:135– 138. Kricka, L.J., I.Faro, S.Heyner, W.T.Garside, G.Fitzpatrick, and P.Wilding. 1995. Micromachined glass-glass microchips for in vitro fertilization (IVF). Clin. Chem. 41:1358–1359. Kricka, L.J., I.Faro, S.Heyner, W.T.Garside, G.Fitzpatrick, G.McKinnon, J.Ho, and P. Wilding . 1997. Micromachined analytical devices: Microchips for semen testing. J. Pharm. Biomed. Anal 15:1443–1447. Lasky, J.B. 1986. Wafer bonding for silicon-on-insulator technologies. Appl. Phys. Lett. 48:78–80. Lauks, I.R., H.J.Wieck, M.P.Zelin, and P.Blyskal. 1992. Disposable sensing device for real time fluid analysis. U.S. Patent 5,096,669. Lemieux, B., A.Aharoni, and M.Schena. 1998. Overview of DNA chip technology. Mol Breed. 4:277–289. Livache, T, H.Bazin, P.Caillat, and A.Roget. 1998a. Electroconducting polymers for the construction of DNA or peptide arrays on silicon chips. Biosens. Bioelectron. 13:629– 634. Livache, T., B.Fouque, A.Roget, J.Marchand, G.Bidan, R.Teole, and G.Mathis. 1998b.


15

Polypyrrole DNA chip on a silicon device: Example of hepatitis C virus genotyping. Anal Biochem. 255:188–194. Livache, T., B.Fouque, A.Roget, J.Marchand, G.Bidan, R.Teoule, and G.Mathis. 1998c. Polypyrrole DNA chip on a silicon device: Example of hepatitis C virus genotyping. Anal. Biochem. 255:188–194. Lockhart, D.J., H.Dong, M.C.Byrne, M.T.Follettie, M.V.Gallo, M.S.Chee, M.Mittmann, C. Wang, M.Kobayashi, H.Horton, and E.L.Brown. 1996. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nature Biotechnol. 14:1675– 1680. Macas, J., M.Nouzova, and D.W.Galbraith. 1998. Adapting the Biomek 2000 laboratory automation workstation for printing DNA microarrays. BioTechniques 25:106–110. Mallon, J. 1992. Nanotechnology from a micromachinist’s point of view. In Nanotechnology: Research and Perspectives. B.Crandall and J.Lewis, eds. MIT Press, Cambridge, pp. 215–240. Mandecki, W. 1998. Method of determining the sequence of nucleic acids employing solid-phase particles carrying transponders. U.S. Patent 5,736,332. Manz, A. 1998. Miniaturization of analytical techniques: Lab on a chip. 135th Mtg. Br. Pharmaceut. Conf. 50:42. Manz, A., N.Graber, and H.M.Widmer. 1990a. Miniaturized total chemical analysis systems: A novel concept for chemical sensing. Sens. Actuators B1:244–248. Manz, A., Y.Miyahara, J.Miura, Y.Watanabe, H.Miyagi, and K.Sato. 1990b. Design of an open-tubular column liquid chromatograph using silicon chip technology. Sens. Actuators B1:249–255. Manz, A., D.J.Harrison, J.C.Fettinger, E.Verpoorte, H.Ludi, and H.M.Widmer. 1991. Integrated electroosmotic pumps and flow manifolds for total chemical analysis systems. Transducers ‘91. Dig. Tech. Papers, pp. 939–941. Mao, C, W.Sun, Z.Shen, and N.C.Seeman. 1999. A nanomechanical device based on the B-Z transition of DNA. Nature 397:144–146. Marshall, A., and J.Hodgson. 1998. DNA chips: An array of possibilities. Nature Biotechnol 16:27–31. Martin, B.D., B.P.Gaber, C.H.Patterson, and D.C.Turner. 1998. Direct protein microarray fabrication using a hydrogel “stamper.” Langmuir 14:3971–3975. Matson, R.S., J.Rampal, S.L.Pentoney Jr., P.D.Anderson, and P.Coassin. 1995. Biopolymer synthesis on polypropylene supports: Oligonucleotide arrays. Anal Biochem. 224:116–116. McConnell, H.M., J.C.Owicki, J.W.Parce, D.L.Miller, G.T.Baxter, H.G.Wada, and S. Pitchford. 1992. The cytosensor microphysiometer: Biological applications of silicon technology. Science 257:1906–1912. McConnell, I.V., J.V.Lamont, and S.P.FitzGerald. 1999. Simultaneous multi-analyte analysis by biochip technology. Clin. Chem. Lab. Med. 37:S394. McCormick, R.M., R.J.Nelson, M.G.Alonos-Amigo, D.J.Benvegnu, and H.H.Hooper. 1997. MicroChannel electrophoretic separations of DNA in injection-molded plastic substrates. Anal Chem. 69:2626–2630. Moritz, W., F.Lisdat, B.H.van den Schoot, N.F.de Rooij, H.H. van den Vlekkert, H.C.G. Ligtenberg, and I.Grohnian. 1993. Flow injection analysis using pH/pF ISFET

Biochip technology

16

combinations for determination of very low fluoride concentrations. Sens. Actuators B15:223–227. Moser, I., G.Jobst, E.Aschauer, P.Svasek, M.Varahram, G.Urban, V.A.Zanin, G.Y.Tjoutrina, A.V.Zharikova, and T.T.Berezov. 1995. Miniaturized thin film glutamate and glutamine biosensors. Biosens. Bioelectron. 10:527–532. Murakami, Y., T.Takenchi, K.Yokoyama, E.Tamiya, I.Karube, and M.Suda. 1993. Integration of enzyme-immobilized column with electrochemical flow cell using micromachining techniques for a glucose detection system. Anal Chem. 65:2731–2735. Murphy, J., B.Carr, and T.Atkinson. 1994. Nanotechnology in medicine and the biosciences. Trends Biotechnol 12:289–90. Nelson, N. 1996. Microarrays pave the way to 21st century medicine. J. Natl. Cancer Inst. 88:1803–1805. Northrup, M.A., B.Beeman, D.Hadley, P.Landre, and S.Lehew. 1996. A miniature analytical thermal cycler based on micromachining technology for PCR. 96th Gen. Mtg. Am. Soc. Microbiol. 96:38. Oakley, C., and D.M.Brunette. 1995. Topographic compensation: Guidance and directed locomotion of fibroblasts on grooved micromachined substrata in the absence of microtubules. Cell Motil Cytoskel 31:45–58. Oakley, C., N.A.Jaeger, and D.M.Brunette. 1997. Sensitivity of fibroblasts and their cytoskeletons to substratum topographies: Topographic guidance and topographic compensation by micromachined grooves of different dimensions. Exp. Cell Res. 234:413–424. Ocvirk, G., E.Verpoorte, A.Manz, and H.M.Widmer. 1995. Integration of a micro liquid chromatograph onto a silicon chip. 8th Int. Conf. Sol. State Sens. Actuators Eurosensors, IX: Dig. Tech. Papers 1:756–759. Owicki, J.C., and J.W.Parce. 1990. Bioassays with a microphysiometer. Nature 344:271. Parce, J.W., J.C.Owicki, K.M.Kercso, G.B.Sigal, Wada H.G., V.C.Muir, L.J.Bousse, K.L. Ross, B.I.Sikic, and H.M.McConnell. 1989. Detection of cell-affecting agents with a silicon biosensor. Science 246:243–247. Petersen, K.E. 1982. Silicon as a mechanical material. Proc. IEEE 70:420–456. Petersen, K.E., W.A.McMillan, G.T.A.Kovacs, M.A.Northrup, L.A.Christel, and F. Pourahmadi. 1998. Toward next generation clinical diagnostic instruments: Scaling and new processing paradigms. J. Biomed. Microdevices 1:71–79. Petersen, K.J., D.Gee, F.Pourahmadi, R.Craddock, J.Brown, and L.Christel. 1991. Surface micromachined structures fabricated with silicon fusion bonding. Transducers ’91. Dig. Tech. Papers, pp. 397–399. Proudnikov, D., E.Timofeev, and A.Mirzabekov. 1998. Immobilization of DNA in polyacrylamide gel for the manufacture of DNA and DNA-oligonucleotide microchips. Anal Biochem. 259:34–41. Qin, D., Y.Xia, J.A.Rogers, R.J.Jackman, X.-M.Zhao, and G.M.Whitsides. 1998. Microfabrication, microstructures and microsystems. In Microsystem Technology in Chemistry and Life Sciences. A.Manz and H.Becker, eds. Springer, Berlin, pp. 1–20. Rajeevan, M.S., I.M.Dimulescu, E.R.Unger, and S.D.Vernon. 1999. Chemiluminescent analysis of gene expression on high-density filter arrays. J. Histochem. Cytochem. 47:337–342.


17

Ramsay, G. 1998. DNA chips: State-of-the art. Nature Biotechnol 16:40–44. Raymond, D.E., A.Manz, and H.M.Widmer. 1994. Continuous sample pretreatment using a free-flow electrophoresis device integrated onto a silicon chip. Anal. Chem. 66:2858– 2865. Rodriguez, I., Y.Zhang, H.K.Lee, and S.F.Y.Li. 1997. Conventional capillary electrophoresis in comparison with short-capillary electrophoresis and microfabricated glass chip capillary electrophoresis for the analysis of fluorescein isothiocyanate antihuman immunoglobulin G. J. Chromatogr. 781:287–293. Ross, P.L., P.A.Davis, and P.Belgrader. 1998. Analysis of DNA fragments from conventional and microfabricated PCR devices using delayed extraction MALDI-TOF mass spectrometry. Anal. Chem. 70:2067–2073. Schena, M., D.Shalon, R.W.Davis, and P.O.Brown. 1995. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270:467– 470. Schena, M., R.A.Heller, T.P.Theriault, K.Konrad, E.Lachenmeier, and R.W.Davis. 1998. Microassays: Biotechnology’s discovery platform for functional genomics. Trends Biotechnol. 16:301–306. Seiler, K., D.J.Harrison, and A.Manz. 1993. Planar glass chips for capillary electrophoresis repetitive sample injection quantitation and separation efficiency. Anal. Chem. 65:1481–1490. Shalon, D., S.J.Smith, and P.O.Brown. 1996. A DNA microarray system for analyzing complex DNA samples using two-color fluorescent probe hybridization. Genome Res. 6:639–45. Sheldon, E.L., J.Briggs, R.Bryan, M.Cronin, M.Oval, G.McGall, E.Gentalen, C.G.Miyada, R.Masino, D.Modlin, A.Pease, D.Solas, and S.P.A.Fodor. 1993. Matrix DNA hybridization, clin. Chem. 39:718–719. Sheppard Jr., N.F., D.J.Mears, and S.W.Straka. 1996. Micromachined silicon structures for modelling polymer matrix controlled release systems. J. Contr. Rel. 42:15–24. Shoji, S., and M.Esashi. 1995. Bonding and assembling methods for realizing a µ-TAS. In Micro Total Analytical Systems. A.van den Berg and P.Bergveld, eds. Kluwer Academic Publishers, Dordrecht, pp. 165–179. Sleytr, U.B., D.Pum, M.Sara, and P.Messner. 1992. Two-dimensional protein crystals as patterning elements in molecular nanotechnology. AIP Conf. Proc. 262:167–177. Song, M.I., K.Iwata, M.Yamada, K.Yokoyama, T.Takeuchi, E.Tamiya, and I.Karube. 1994. Multisample analysis using an array of microreactors for an alternating-current field-enhanced latex immunoassay. Anal. Chem. 66:778–781. Southern, E.M. 1996. DNA chips: Analysing sequence by hybridization to oligonucleotides on a large scale. Trends Genet. 12:110–115. Southern, E., K.Mir, and M.Shchepinov. 1999. Molecular interactions on microarrays. Nature Genet. 21:5–9. Spangler, L.J., and K.D.Wise. 1990. A bulk silicon SOI process for active integrated sensors. Sens. Actuators A24:117–122. Steber, G.R. 1987. Audio frequency DSP laboratory on a chip-TMS32010. Proc. IECON ’87: 1987 Int. Conf. Industr. Electronics, Control Instrument. (Cat. No.CH2484–4) 2:1047–1051.

Biochip technology

18

Stevens, A.M., and C.J.Richards. 1997. A metallocene molecular gear. Tetrahedron Lett. 38:7805–7808. Stix, G. 1992. Micron machinations. Scient. Am. 267:106–117. Suda, M., T.Sakuhara, Y.Murakami, and Karube I. 1993. Micromachined detectors for an enzyme-based FIA. Appl. Biochem. Biotechnol. 41:11–15. Taylor, T.B., S.E.Harvey, M.Albin, L.Lebak, N.Ning, I.Mowat, T.Scheurlein, and E. Principe.1998. process control for optimal PCR performance in glass microstructures. J. Biomed. Microdevices 1:65–70. Terry, S.C., and D.A.Hawker. 1983. Automated high speed natural gas analysis using a new microcomputer controlled, high resolution GC analyzer. Adv. lustrum. 38:387– 398. Terry, S.C., J.H.Jerman, and J.B.Angell. 1979. A gas chromatographic air analyzer fabricated on a silicon wafer. IEEE Trans. Electron. Devices ED-26:1880–1886. Tracey, M.C., R.S.Greenaway, A.Das, P.H.Kaye, and A.J.Barnes. 1995. A silicon micromachined device for use in blood cell deformability studies. IEEE Trans. Biomed. Eng. 42:751–761. Tucker, A.J. 1984. Biochips: Can molecules compute. High Tech. 79:36–47. von Heeren, F., E.Verpoorte, A.Manz, and W.Thormann. 1996. Micellar electrokinetic chromatography separations and analyses of biological samples on a cyclic planar microstructure. Anal. Chem. 68:2044–2053. Waters, L.C., S.C.Jacobson, N.Kroutchinina, J.Khandurina, R.S.Foote, and J.M.Ramsey. 1998a. Microchip device for cell lysis, multiplex PCR amplification, and electrophoretic sizing. Anal Chem. 70:158–162. Waters, L.C, S.C.Jacobson, N.Kroutchinina, J.Khandurina, R.S.Foote, and J.M.Ramsey. 1998b. Multiple sample PCR amplification and electrophoretic analysis on a microchip. Anal. Chem. 70:5172–5176. Watson, A., A.Mazumder, M.Stewart, and S.Balasubramanian. 1998. Technology for microarray analysis of gene expression . Curr. Opin. Biotechnol. 9:609–614. White, V., R.Ghodssi, C.Herdey, D.D.Denton, and L.McCaughan. 1995. Use of photosensitive polyimide for deep X-ray lithography. Appl. Phys. Lett. 66:2072–2073. Wilding, P., M.A.Shoffner, and L.J.Kricka. 1994. PCR in a silicon microstructure. Clin. Chem. 40:1815–1818. Wilding, P., L.J.Kricka, J.Cheng, G.Hvichia, M.A.Shoffner, and R.Fortina. 1998. Integrated cell isolation and polymerase chain reaction analysis using silicon microfilter chambers. Anal. Biochem. 257:95–100. Wodicka, L., H.Dong, M.Mittmann, M.H.Ho, and D.J. Lockhart. 1997. Genome-wide expression monitoring in Saccharomyces cerevisiae. Nature Biotechnol. 15:1359– 1367. Xia, Y., X.-M.Zhao, and G.M.Whitesides. 1996. Pattern transfer: Self-assembled monolayers as ultrathin resists. Microelectron. Eng. 32:255–268. Xue, Q., Y.M.Dunayevskiy, F.Foret, and B.L.Karger. 1997a. Integrated multichannel microchip electrospray ionization mass spectrometry: Analysis of peptides from onchip tryptic digestion of melittin. Rapid Commun. Mass Spectrom. 11:1253–1256. Xue, Q., F.Foret, Y.M.Dunayevskiy, P.M.Zavracky, N.E.McGruer, and B.L.Karger. 1997b. Multichannel microchip electrospray mass spectrometry. Anal Chem. 69:426–


19

430. Zhao, Q., J.Drott, T.Laurell, L.Wallman, K.Lindstrom, L.M.Bjursten, G.Lundborg, L. Montelius, and N.Danielsen. 1997. Rat sciatic nerve regeneration through a micromachined silicon chip. Biomaterials 18:75–80. Zimmer, K., D.Hirsch, and F.Bigl. 1996. Excimer laser machining for the fabrication of analogous microstructures. Appl Surf. Sci. 96:425–429.

2 Microfabrication Processes for Silicon and Glass Chips Yuebin Ning and Glen Fitzpatrick

INTRODUCTION Microfabrication technology, also known as micromachining, refers to the fabrication processes employed in the manufacture of microelectromechanical systems (MEMSs) and is the foundation of the manufacturing processes for silicon and glass-based chemical and biological microchips. Within the realm of microfabrication technology, there are a number of approaches that have been utilized in fabricating chemical and biological microchips. These approaches can be broadly divided into two categories: (1) Microfluidic technology, which uses all facets of the microfabrication processes to create 3D structures for chemical reactions and separations through the manipulation of fluid movement; and (2) Microarray technology, which uses microlithography, contact, or drop-on-demand printing to form 2D biologically active arrays on flat substrate surfaces for biological assays. Silicon and glass-based devices are presently the most developed in microfluidic technology and have found a wide range of applications in the field of chemical and biological analyses. In addition to its remarkable electronic properties, single-crystal silicon also displays high mechanical strength and strong orientation dependence of etch rate in a number of wet etch systems (Petersen, 1982). These unique properties, combined with the ability to grow silicon oxide and nitride films on the surface, have made silicon wafers an ideal choice of substrate material for fabrication of microchips. Silicon also has excellent thermal conductivity that can be exploited in device applications requiring special thermal management. Although not as versatile as silicon, glass substrates are excellent electrical insulators and are optically transparent. Both characteristics are critical for microchip applications that require a window on the reactions inside the chip—for instance, the detection of fluorescence and the application of strong electric fields required for electroosmosis and electrophoresis. More importantly, aside from the ability to fabricate reservoirs, piping, reaction chambers, and conduits through etching and bonding techniques, silicon and glass also offer a great deal of promise for future applications that require a much higher level of complexity and integration with microelectronics and microoptics. This chapter will focus on the microfabrication processes most commonly employed in the field today for the manufacture of silicon and glass-based microfluidic chips for chemical and biological analyses.

Microfabrication processes for silicon and glass chips

21

BASICS OF MICROFABRICATION PROCESSES A significant portion of the microfabrication processes employed in MEMS today is a direct adaptation of the processes developed by the semiconductor integrated-circuit (IC) industry over the years. The processes normally start with the deposition of metal or dielectric thin film onto a silicon wafer or glass substrate. A photolithography step then transfers a pattern or image from a photomask onto a photo-sensitive and etch-resistant polymer coating deposited over the film. The polymer used is commonly known as a photoresist. The patterned photoresist is used as a template and mask for the subsequent etch steps to pattern the thin film. The photoresist can then be removed, and the surface is ready for the next step, which could be an etching step to sculpt the silicon or glass substrate, another thin film deposition, further photolithography and etching, or stripping of the thin film if it has already served its purpose. Once the desired features are etched into silicon wafer or glass substrate, a bonding process is usually performed to attach a cover on top, forming the reaction chambers and fluid channels. Needless to say, these steps can be repeated, allowing more sophisticated devices to be built. A number of books and reviews have been published on various aspects of microfabrication processes (Campbell and Lewerenz, 1998; Madou, 1997; Muller et al., 1990; Rai-Choudhury, 1997; Ristic, 1994; Tong and Gösele, 1999), and the authors recommend that the readers consult these references for more in-depth coverage. In this section, we will attempt to provide a brief introduction to the basics of microfabrication processes most relevant to microchip manufacturing. Thin Film Deposition Thin films are fundamentally important and versatile structures that are employed in the fabrication of both IC and MEMS devices (Maissel and Glang, 1970). Thin films are usually in the order of a few tens of angstroms to roughly 10 microns [1 micron (µm) =104 angstroms (Å)]. Films for MEMS devices have many demands placed on them and are required to function as chemical barriers, hard coatings, sacrificial layers, flexible membranes, porous layers, heating elements, catalytic surfaces, electrode materials, and optical components, to name a few. Two basic film deposition methods (Vossen and Kern, 1978) are employed with many variations on the theme. One is chemical vapor deposition (CVD), which grows the film from gas and liquid sources, combining them at low pressure and elevated temperature, sometimes adding additional energy with a plasma (plasma-enhanced CVD or PECVD). For a typical CVD process, silane (SiH4) is usually reacted with a variety of species to make silicon, silicon dioxide, silicon carbide, or silicon nitride. These films are versatile and have been used very effectively as dielectric layers and etch masks in silicon and glass microchip fabrication processes. CVD processes offer good film coverage over surfaces, inde-pendent of shapes and shadowing, which is important when coating uneven surfaces. For silicon wafers, the thermal oxidation process is also commonly used to form SiO2

Biochip technology

22

layers. This process is easy to implement, and the oxide grown is of high quality. The process temperature for the growth of thermal oxide however is very high, typically in the range of 1000–1200°C, which limits its application to mostly the growth of etch mask layers. In addition, the thermal oxidation process is diffusion limited and the growth rate diminishes quickly as the film grows, making such a process practical only if the oxide thickness is below one micron. The other method is physical vapor deposition (PVD), which is the movement of atoms or molecules from a source to a substrate. The simplest model is that of vacuum evaporation (Glang, 1970), where within a vacuum chamber a source of desired material is heated to the evaporation point, at which stage the material is coated over the exposed parts of the substrate. Evaporation of the source can be accomplished by resistive heating, or by electron beam methods. Another film deposition method that offers a great range of control and material options is sputtering (Chapman, 1980; Vossen and Kern, 1978). This involves having a target of the desired material become the cathode in a low-pressure plasma of inert gas. The plasma erodes the target, and the ejected adatoms coat the substrate. The efficiency of the sputtering process can be enhanced by magnetic confinement of the plasma (magnetron sputtering). The density and stress of the films can be controlled by the sputtering parameters (gas pressure, power, and target/substrate spacing) as well as by bombardment of the surface of the substrate (bias sputtering). Both evaporation and sputtering methods can yield oxides or nitrides by incorporating oxygen or nitrogen in the vacuum of evaporation or the usually inert sputter gas of argon or xenon (reactive evaporation/sputtering). Most of the metal films used in silicon and glass microchip manufacturing are deposited using either the thermal evaporation or sputtering process, including chromium, gold, and platinum, which are often used as etch mask and electrode layers. Photolithography Photolithography is the photochemical process of making a template for selective masking of areas that need to be protected during the subsequent thin film etching process (Elliot, 1982; Thompson et al., 1994). Positive photoresist is composed of resin (“Novolak”), solvent (usually either an ether acetate or ethyl lactate), developer inhibitors to keep unexposed resist from dissolving in developer, the photoactive compound (PAC), a diazo quinone, as well as dyes to inhibit stray reflections from exposing the photoresist and to lend contrast to the photoresist for inspection purposes (Dammel, 1993). Catalyzed by water in the air, the areas that are exposed to UV light become soluble in a mild base. Negative photoresist employs a photo-induced crosslinking phenomenon, and the development is the dissolution of unexposed photoresist by its solvent. Exposure can take place by image projection, a proximity mask placed nearby and exposed with collimated UV light, or by a contact mask at the surface of the photoresist. Each method offers different advantages in terms of cost, throughput, reduced surface damage, and ease of use. Figure 1 is a schematic illustrating the


23

Figure 1. Schematic representation of contact printing photolithography. basic ideas of photolithography. The IC industry has pushed the limits of this technology far beyond what was first imagined possible. Smaller features are achievable by decreasing the wavelength of the UV exposure (to reduce diffraction limitations), using phase-shift masks (nonlinear masks that trim the Gaussian distribution of light through a slit), and of course by using electron beam and x-ray to further decrease exposure wavelength. The photoresist is baked prior to being used as an etch mask to increase adhesion and mechanical stability, and therefore resistance to ion bombardment in a plasma, and to decrease chemical permeability. Baking above the flow temperature of about 140°C leads to distorted features. If that is not a concern, baking at 165°C or still higher temperatures can enhance the photoresist performance in etching but is difficult to remove after. To maintain the integrity of the photoresist at these elevated temperatures, a UV exposure (shorter than a 350-nm wavelength) to induce crosslinking can be employed during the bake. Removal of UV-crosslinked photoresist can be effected by oxygen plasma stripping, followed by a “hot piranha” (3 parts 98% H2SO4 and 1 part 30% H2O2 by volume) clean for 10–15 minutes. A photoresist is made of organic compounds, and as such has little affinity for glass surfaces. When the photoresist is used on a glassy surface (or any dielectric), the adhesion can be enhanced by a surface siliation treatment, most commonly a hexamethyldisilazane (HMDS) coating, which functions as a hydrophobic surface for the photoresist to adhere well, after dehydrating the surface (Dammel, 1993). This is adequate for etches that do not attack the surfaces strongly but is insufficient for fabrication of glass channels. The authors and other groups have achieved short-duration etches of silicon and glass using photoresist, but a patterned hard mask of dielectric or metal thin film is usually necessary for etches of reasonable depth.

Biochip technology

24

Chemical Etching of Thin Films After photolithography, a chemical etching step is carried out with the silicon or glass substrates to pattern the thin-film coatings. This chemical etching step is designed to either define the electrical elements such as on-chip electrodes or heaters, or to open up the etch mask layer for subsequent silicon and glass etch processes. The removal of material by chemical etching is a long-established practice that can take place by either wet etching or reactive ion etching (RIE) methods. In simple terms, wet etching is the dissolution of material (films or substrate) by immersion in a solution of reactive species (Kern and Deckert, 1978), and RIE is the volatilization of material by reactive species introduced in a plasma (Chapman, 1980; Melliar-Smith and Mogab, 1978; Mucha et al., 1994). The simplest wet etch technique is liquid chemical immersion or dip etching, where the photoresist-masked wafers or substrates are submerged in the etch solution, usually contained in a beaker or etch tank. Mechanical agitation such as magnetic stirring should be performed to achieve good etch uniformity. Spray etching is another commonly used wet etch method, particularly when very small features are present that are difficult for the etchant to wet due to surface tension. Wet etching techniques are usually simple and effective, and can be readily applied to most of the metal and dielectric thin films used in microchip fabrication, as illustrated by the few practical examples discussed below. Chromium There are several etch systems that are effective and compatible with photoresist. The one used at Micralyne is a commercial “Chrome Etch”, which is a mixture of nitric acid, eerie ammonium nitrate, and water. The etch rate is about 800 Å/min when the etchant is fresh. Small amounts of acid-compatible surfactant can be added to this system to improve its wetting of small features without the formation of precipitates. Gold The most widely used etchant for gold is tri-iodide, which consists of a mixture of 400 g of KI, 200 g of I2, and 1000 mL of H2O. With an approximate etch rate of 3000 Å/min, this etchant is compatible with photoresist and is excellent for gold thin-film patterning. However, this etchant does not have good solubility for surfactant. To improve the wetting and etch uniformity, spray etch may be required for very small features (a few microns). Aqua Regia (3 parts 37% HC1 and 1 part 70% HNO3 by volume) is also commonly used as a general etchant for gold removal or patterning of noncritical features. Freshly mixed Aqua Regia has a very high and unstable etch rate that is typically in the range of 10–15 µm/min at room temperature.


25

Platinum There is no etchant that is effective and convenient to use at room temperature. A mixture of 70% HNO3, 37% HCl, and H2O (1:7:8 by volume) heated to 85°C can etch platinum at a rate of 400–500 Å/min (Rand, 1975; Rand and Roberts, 1974). Aqua Regia is also used as an etchant for platinum (Glang and Gregor, 1970). Both etchants, however, suffer from a lack of stability and repeatability, which makes good process control difficult. An alternative to the etching approach is to deposit Pt onto patterned photoresist and then dissolve the photoresist in acetone to “lift-off” the platinum in unwanted regions. This “lift-off” process is effective if the platinum film thickness is under about 2000 Å. SiO2 and Si3N4 Hydrofluoric acid, with or without further dilution, is an effective general etchant for SiO2. For pattern etching using photoresist as an etch mask, NH4F is usually added to HF (also known as buffered oxide etch, or BOE) to control the pH in order to minimize deterioration of the photoresist and the polymer/dielectric interface (Kern and Deckert, 1978). This is less of an issue when the interface is formed between photoresist and a metal such as gold or platinum. The etch rate in BOE is dependent on the density of SiO2. For thermal SiO2, the etch rate is about 600 Å/min in BOE with a 10:1 volumetric ratio of 40% NH4F and 49% HF. For PECVD oxide deposited at 300°C, the etch rate can be as high as 1500 Å/min. For Si3N4, however, the etch rate in BOE is too slow to be of practical use, and one usually has to use concentrated HF to achieve any significant etch rate. Phosphoric acid can also be used to etch Si3N4, and boiling 85% H3PO4 at 180°C is reported to have an etch rate of 100 Å/min for CVD Si3N4 (Kern and Deckert, 1978). In either case, wet etching of silicon nitride is not straightforward, and an RIE process is a preferred alternative. Although not as high in substrate throughput, RIE offers certain process advantages over the wet etch method (Chapman, 1980; Melliar-Smith and Mogab, 1978; Mucha et al., 1994). Compared with wet etching processes, RIE has better critical dimension control and is less demanding of photoresist adhesion. RIE also eliminates wetting and other problems associated with surface tension of the wet etchant, which is important for etching small features and releasing fragile structures. Many gas mixtures have been developed over the years for a variety of RIE processes. Å particularly useful gas mixture is the CF4/CHF3–O2 system, which, with the proper composition, pressure,, and RF power, can be very effective in etching silicon nitride, silicon oxide and polysilicon layers. Depending on the detailed process parameters, etch rates of 400–600 Å/min can be readily achieved for SiO2 and Si3N4, and rates much higher still can be obtained for polysilicon. Other important gases include SF6, Cl2, and a few other fluorine- and chlorine-containing gases. These gases are often mixed with O2 to produce etching mixtures for metal and dielectric films.

Biochip technology

26

SILICON AND GLASS ETCHING PROCESSES Silicon and glass etching techniques use chemical processes to create 3D structures in silicon and glass wafers. A number of wet chemical etching as well as RIE processes have been developed for silicon over the last few decades because of its unparalleled importance in microelectronics. As a result, a tremendous amount of work has been published in the literature, thus providing excellent coverage on this subject. In contrast, little work has been reported on glass etching processes due to limited interest outside of the biochip field. We will therefore limit our discussion on the silicon etching processes to the most commonly employed etch systems in MEMS and then turn our focus to detailed discussions of glass etching processes for microchip fabrications. Silicon Etching There are two general classes of wet chemical etching systems for silicon, namely isotropic and anisotropic etching systems. For isotropic etching of silicon, the most commonly used etchants are mixtures of nitric acid and hydrofluoric acid, with water or acetic acid as a diluent (Kern and Deckert, 1978). The etch rate of silicon in these mixtures will depend on the mixing ratios, temperature, and agitation. For etching smooth channels and reaction wells with depths under 100 microns, the etchant should be low in HF and high in HNO3 concentrations in order to obtain a slower etch rate and a smooth etch surface. Etch rates of a few microns per minute at room temperature are usually convenient for process control and efficiency. For example, a mixture of 49% HF, 70% HNO3, and 99% CH3COOH (HNA) with a volumetric mixing ratio of 8:75:17 yields a silicon etch rate of ~5 µm/min with a smooth surface (Madou, 1997). Such an etch system is ideally suited for silicon microchips with relatively simple structures. In an isotropic etching process, the etch rate is identical in every direction and does not depend on the crystal plane orientation of a silicon wafer. As a result, the etched feature size will grow with etch depth. For example, if a channel has an initial opening width of X in the etch mask, by the time the channel is etched to a depth of D, the resultant channel top width W will be

This expression provides a reasonable estimate of the resultant channel width. It is important, however, to realize that several other factors also come into play in determining the final channel width. These include photoresist overdevelopment, overetching of the etch masks, the ability to replenish etchant through the etch mask overhang, and adhesion of the etch mask layer onto the silicon wafer surface. A number of thin films can be used as an etch mask. Thermally grown SiO2 has an etch rate of 400– 800 Å/min in HNA (Madou, 1997) and is convenient to use as an etch mask for relatively shallow silicon etches. Chromium and gold can also be used as an etch mask when thermal oxide alone is insufficient. For deep etches, CVD or PECVD silicon nitride is a


27

better choice of etch mask material. Figure 2 shows an etched microchannel in silicon with the etch mask (thermal oxide with thin Cr/Au on top) remaining. One can clearly see the expected isotropic etch profile and the mask overhang. In an anisotropic etch process, the etch rate is dependent on the crystal plane orientation of the Si wafers (Bassous, 1978; Bean, 1978; Petersen, 1982). The basis of

Figure 2. SEM photo of a cross-section of a silicon microchannel etched with HNA. anisotropy is that the dissolution reaction rates in the Si(100) and (110) planes are much higher than that in the (111) plane. This anisotropy has been exploited to create various 3D structures in silicon microchip fabrication. For example, if a square or rectangular opening in the etch mask of a wafer is aligned to that plane and exposed to such an etchant, the fast dissolution of the (100) plane will result in four convergent (111) planes each at an angle of 54.74° to the surface plane of the wafer (Petersen, 1982). This characteristic is often used to etch V grooves and reaction wells (Figure 3) for various applications. Similarly, one can also exploit the high (110) etch rate to create vertical structures (Bassous, 1978; Bean, 1978; Kendall, 1979), as shown in Figure 4. A number of anisotropic etch systems have been developed over the years, with KOH, ethylenediamine pyrocatechol (EDP), and tetramethyl ammonium hydroxide (TMAH) being the most popular etch systems. The reported etch rates and etch anisotropy for each system vary greatly, possibly due to variations in wafer, and mask quality, chemical purity, alignment and measurement accuracy. Table 1 summarizes the principal etch characteristics of these three etch systems. One undesirable characteristic of an anisotropic etch is severe undercutting of convex

Biochip technology

28

corners, which makes it difficult to produce well-defined and smooth turns in a channel network. Recent studies (Merlos et al., 1993; Sekimura, 1999) show isopropyl alcohol and surfactant can be added to KOH and TMAH to reduce the undercutting of convex corners, eliminating the need for corner compensation with physical structures. From the manufacturer’s point of view, consistency in the purity of chemicals used is of paramount importance in maintaining a reproducible process. Metallic impurities in particular need to be carefully controlled, as they can influence etch rate, anisotropy, and surface roughness (Hein et al., 1997).

Figure 3. High-density PCR well array etched in (100) silicon. Device fabrication by T.Zhou and J. Broughton of AMC. Design courtesy of T Woudenberg of PE Biosystems.


29

Figure 4. Comb filter etched in (110) silicon using KOH. Table 1 Principal Etch Characteristics of the Three Anisotropic Etch Systemsa Etch rate ratio Etchant and etch Etch rate (100)/ Mask (µm/min) (111) materials Comments temperature KOH at 80°C, 30 (100): 1.1 50–400 SiO2, Simple to use, wt%, dissolved in (110): 1.6 nontoxic, higher Si3N4 water selectivity to (111) plane, improved smoothness if agitated ultrasonically during etching Low etch selectivity to SiO2 (etch rate 30 Å/min) Smooth surface, high EDP (F etch) at 115°C (100): 1.3 20–35 SiO2,

Biochip technology

750 mL (110): 1.4 ethylenediamine 240 g pyrocatechol 240 mL water 4.5 g pyrazine

30

Si3N4

etch selectivity to SiO2 Low etch selectivity to (111) plane, toxic, low stability due to aging IC compatible, TMAH at 90°C 22 (100): 1.0 50 SiO2, nontoxic, smooth wt%, dissolved in (110): 1.4 Si3N4 etch, high etch water selectivity to SiO2 Low etch selectivity to (111) plane aEtch rates adapted from Petersen (1982), Tabata et al. (1992), Reisman et al. (1979), Seidel et al. (1990), Madou (1997), Kandall and Shoultz (1997), and the authors’own work 3.2. Glass Etching The glass substrates of interest contain a significant amount of SiO2, with the rest of the constituents being various metal oxides. Naturally, an effective wet etch solution would start with hydrofluoric acid, which attacks the Si—O bond aggressively and etches SiO2 at a significant rate. Because of the presence of various metal oxides in glass, another acid such as HNO3 or HCl is usually added to the etch system to convert insoluble metal fluorides into soluble salts, thus reducing etch roughness (Kern and Deckert, 1978). For practical reasons, the etch solution is usually diluted with water to tailor the etch rate to suit the type of glass. An example of such an etch solution has a volumetric composition of 20 parts 49% HF, 14 parts 70% HNO3, and 66 parts H2O (Kern and Deckert, 1978). This etch system produces smooth etches (see Figure 5), with an etch rate of 1.6 µm/min for Corning 0211 glass, and an etch rate of 0.4 µm/min for Borofloat glass from Schott. Etch rate in this regime provides a good balance between process time and depth control for the channel etch depth ranging from a few microns to tens of microns typically encountered in glass microchip fabrication. In situations where a higher etch rate is desired for deep etches, one can simply increase the HF concentration accordingly. There are two hard etch masks that have been reported to work well for an HF-based glass etch process. One of them is a Cr–Au system used by various groups (Fan and Harrison, 1994; Jacobson et al., 1994). The other etch mask system is an amorphous silicon mask, with a typical thickness of 1000–2000 Å (Simpson et al., 1998). The authors have had the most success utilizing sputtering as the method for depositing hard coatings to serve as the mask for glass etching. A number of metal systems have been studied over the years, and the best to date has been a system of chromium (200–400 Å) as the adhesion layer and gold (1500–2000 Å) as the protection layer. Sputtered films are well suited for producing an etch mask owing to their high film density, good adhesion, and control of stress. A high film density keeps etchant diffusion from occurring during


31

the glass etch process. Adhesion is critical, because a film that is only loosely bound to the surface offers a low-energy path for the etchants between the film/glass interface. The phenomenon is known as undercut, and a 1:1 ratio of sideways etch to downward etch is the theoretical best case. Often, the worst case of undercut will leave a ragged edge due to nonuniformity of adhesion across the surface, but the authors have seen up to 4:1 undercut with absolutely perfect uniformity, resulting in a final product that appears to have followed the masked features and is only somewhat bloated. Similarly, stress control is important, so when the glass is being removed by the glass etch process the film does not peel back on itself to increase the undercut process.

Figure 5. MicroChannel etched in Corning 0211 glass. The etch process itself is diffusion limited when the temperature is unchanged. Both removal of reactants and replenishment of the etchant in the etch areas are critical to achieving a smoothly etched surface. For this reason, the etch solution is usually agitated vigorously using magnetic stir bars. Other means of agitation such as forced circulation using pumps are also effective. Even with agitation, channel bottom roughness could still occur as the etch progresses. This is particularly true for very narrow channels, which, with the presence of hard mask overhangs, will only have very narrow slits through which the etchant is replenished. On the other hand, with large open geometry and vigorous agitation, the etched surface can remain fairly smooth even at 200 µm etch depths. Figure 6 illustrates the above two scenarios, one with a narrow (10 µm) and one with a wide (600 µm) initial channel opening.

Biochip technology

32

Figure 6. Channel bottom roughness as a function of initial open width and etch depth. Roughness was measured over a distance of 400 µm using a profilometer This measurement is an indication of large scale roughness, not the microroughness we will discuss later The instrument limitation is 15 Å. 3.3. Reactive lon Etching Reactive ion etching is also a well-established alternative etch method for sculpting 3D structures in silicon. The RIE process is most useful when high-aspect-ratio (etch depth over width) features are required. In the most standard processes, a fluorine-containing gas such as CF4 and SF6 is introduced to a parallel plate etcher, where an RF plasma with a DC bias is established for directional etch of silicon. The etch rate and aspect ratio in a parallel plate etcher are relatively low, and the process has been limited to applications requiring fairly shallow etch depth and low aspect ratios. More recent development in high-density plasma sources (Mucha et al., 1994) such as electron cyclotron resonance (ECR) and inductively coupled plasma (ICP) has led to reactive ion etching systems capable of a much higher etch rate, and with very high aspect ratios. Because of the extraordinary etch depth these systems afford, they are also referred to as deep reactive ion etch (DRIE). With commercial ICP-based DRIE units, silicon etch rates as high as 10 µm/min can be achieved (Pandhumsopom et al., 1998) and aspect ratios greater than 40 have been reported (Sasserath et al., 1997). Although much less developed for glass substrates, reactive ion etching processes, especially the ICP etch process more recently developed, hold more promise for certain novel applications requiring very small features with high aspect ratios. Micralyne has developed an RIE process for quartz wafers that allows fabrication


33

Figure 7. RIE-etched features in quartz. Reprinted with permission from He et al. (1998). Copyright © 1998 American Chemical Society. of features only 1–2 µm in size with an aspect ratio of 10. Figure 7 is an example of a device that can be fabricated using such an RIE process. The etch process is based on fluorine chemistry and has an etch rate of approximately 2 µm/hr. With such a low etch rate, the process is only practical for etch depths under 10 µm. A fluorine chemistrybased process with about a 10 times higher etch rate has also been reported recently where a commercial ICP-based DRIE was used (Constantine, 1997). The same process in principle can also be applied to other types of glass substrates. However, due to the presence of metal oxides, the process may require modification to produce similar etch qualities as with quartz, because the other constituents are not volatilized in the plasma. Material Issues Single-crystal silicon wafers are of very high quality in terms of their purity, crystal structure, surface finish, and flatness due to the constant refinement demanded by the IC industry. Glass, on the other hand, encompasses a rather large spectrum of materials, ranging from the crude material used for windows to the purest form of SiO2 quartz. Table 2 summarizes the chemical compositions of three types of commonly used glass substrates and quartz.

Biochip technology

34

The glassy surface is a nonhomogeneous rolling plain of various chemical species that often presents serious challenges to achieving good etch profile, either due to poor adhesion of etch mask to the surface or preferential etch at the subsurface layer. Because of these issues, glass substrates suitable for microchip fabrication are somewhat limited. Borofloat glass from Schott is a reasonable choice

Table 2 Approximate Composition of Different Types of Glass (wt%) Other oxides SiO2 AI2O3 B2O3 Na2O K2O Borofloat Pyrex 0211 Quartz

70–87% 1–7% 7–15% 0–5% 0–5% 0–8% 80% 2.25% 13.1% 3.5% 1.1% 0.05% Fe2O3 65% 2% 9% 7% 7% 7% ZnO, 3% TiO 100%

for capillary electrophoresis devices due to its low cost, good flatness, wide range of thicknesses, and excellent etch quality. One issue with Borofloat glass is its higher content of tin on one side due to the fabrication process. The tin-rich side has a preferential etch rate within the first few microns of the surface, which can result in poor etch width control and should be avoided. Pyrex glass has similar chemical composition to Borofloat, and hence similar etch characteristics and optical properties. Although also used for microchip fabrication, Pyrex (Coming 7740) comes in the form of large ingots or relatively thick rolled plates, and is therefore more costly to turn into the thin flat plates required. Moreover, the grinding and polishing process employed to produce plates will also introduce surface scratches, resulting in etch defects. The impact of the surface scratches can be partly eliminated by thermal annealing at temperatures between the strain and annealing points (510 and 680°C, respectively), but the process will undoubtedly add to the fabrication cost. Coming 0211 glass has been a popular choice for microchip fabrication work at Micralyne for the last several years because of its low cost, consistently high etch quality, and high yield. The main drawback of Coming 0211 is that its maximum available thickness is only 0.55 mm, which can be a limitation for designs requiring high mechanical strength. When the surface is ground and polished to a high precision, such as the quartz wafers and plates supplied by Hoya, the quality of wet etched channels in quartz is usually consistent and good. However, quartz has a low etch rate in HF, and it is difficult to etch deep features (over 100 µm) as the etch mask will start to deteriorate due to prolonged exposure to HF. Quartz etches well when using an RIE process, which offers additional design capabilities. The main drawback of quartz is the high substrate cost, usually at least 10 times that of borosilicate glass. Quartz requires a significantly higher bonding temperature (~1100°C) to achieve good bonding, a process that will limit the metal systems that can be used as embedded electrodes. The presence of various metal oxides in glass makes it easier to work with in the manufacturing stage and in the microfabrication stage. They are also the main source of the glass fluorescence under UV excitation. Figure 8 is a graphical comparison of the UV-induced fluorescence of the three glass and quartz substrates discussed. Although all three glasses are considered to have low UV-induced fluorescence, Coming 0211 glass


35

does display a comparatively higher level of fluorescence, which can be a limitation when a very-low-fluorescence background is required. In contrast, quartz has a very low level of UV-induced fluorescence and is ideally suited for applications requiring a lowfluorescence background.

Figure 8. A comparison of UV-induced fluorescence for various glasses and fused silica. UV excitation is from a 248nm wavelength laser. The intensity of fluorescence is in an arbitrary unit. MECHANICAL MILLING AND DRILLING For chemical and biological applications, microchips need to have access ports for sample and reagent loading, as well as electrical connections. As such, fabrication of access ports in glass plates remains a vital part of the microchip fabrication process. Although a variety of methods exist for producing good-quality slots and holes in glass, the most practical ones are the ultrasonic and laser drilling processes. Ultrasonic Milling Ultrasonic milling removes material from defined areas of a workpiece by the abrading action of a grit-loaded liquid slurry generated by a tool vibrating at ultrasonic frequencies (Bellows and Kohls, 1982; Moreland, 1992). The process can produce good-quality holes and slots in glass without introducing slag to interfere with the bonding process (see Figure 9). The smallest holes that the ultrasonic milling process can produce practically with consistent hole quality and uniformity are about 200 µm in diameter. Since the milling process does not involve any rotary mechanism, the same process can also produce noncircular features such as slots, groves, and fluid interconnects. The milled glass plate does need to go through very vigorous cleaning protocols to remove the

Biochip technology

36

thermal wax used for mounting the substrate during milling. Depending on the type of thermal wax used, the cleaning solvent can be acetone or denatured alcohol heated to 50– 70°C and agitated using an ultrasonic cleaning bath. The solvent cleaning step may be repeated 2–3 times with a duration of 30–60 minutes each to ensure complete removal of any residual thermal wax on the glass surface. This is especially important for cover plates with small and relatively deep holes. The ultrasonic agitation

Figure 9. SEM photo of 380 µm diameter hole milled ultrasonically in 0211 glass of 0.55 mm thickness. also helps to shake off debris loosely attached to the side wall of the holes such as SiC from the abrasive slurry and glass shards. This is necessary to prevent the debris from plugging the channels during operation of the microchips. Laser Drilling The laser drilling process removes material by melting, ablating, and vaporizing the workpiece at the point of impingement of a laser beam (Bellows and Kohls, 1982). Laser drilling can be a viable alternative to ultrasonic milling, particularly for small holes (under 200 microns in diameter) when high placement accuracy is required. Potentially, laser drilling can also be more cost competitive for large-volume manufacturing. The current laser drilling technology is most developed for drilling and cutting quartz, silicon wafers, and ceramic plates. The present process available needs further development for glass substrates, such as Borofloat and Coming 0211, to become a useful alternative to the ultrasonic milling process. The laser drilling process generates aerosol particles that


37

can form deposits around the drilled holes. Ultrasonic cleaning combined with scrubbing with liquid detergent and a sponge is usually effective in removing these deposits. Care must be exercised, however, during the scrubbing process to minimize surface scratches. Figure 10 shows a CO2 laser-drilled hole in a quartz wafer, with the majority of aerosol deposits already removed.

BONDING Very few MEMS devices can be made without employing a bonding process to join silicon wafers and glass plates together. There are several bonding schemes that

Figure 10. SEM photo of 80 µm diameter hole drilled by a CO2 laser in a quartz wafer of 0.4 mm thickness (at the end of an etched channel). have been developed over the years. For the fabrication of silicon and glass microchips, two bonding methods are particularly important—namely, the anodic bonding process and the fusion bonding process. The bonding mechanisms have been the subject of a number of articles, conference proceedings, and books (Ko et al., 1985; Obermeier, 1995; Ristic, 1994; long and Gösele, 1999), and will not be discussed here. In essence, for both bonding processes, it is important that the wafers and plates be flat, and that the surfaces be smooth. To facilitate the bonding process, the silicon or glass substrate surface should be clean and free of particulate, organic, and metallic contamination. The surface also needs to be hydrophilic, so that a room-temperature bond can be formed when two wafers

Biochip technology

38

or plates are brought into contact. To fulfill these requirements, wafers and substrates for bonding need to go through carefully designed chemical cleaning steps, and the bonding step should be performed in a clean-room facility. Based on the authors’ practical experience and various literature reports, the wafer and plate flatness (defined as the total thickness variation or TTV) required for the bonding processes should be in the range of 1–3 µm, and the surface microroughness needs to be less than 10 Å (Tong and Gösele, 1999). Larger TTVs can be tolerated in the case of anodic bonding where a large local force can be exerted by the applied electrical field. Of course, thicker wafers and plates will have more stringent flatness requirements due to their reduced ability to conform to each other. The silicon wafers readily available for IC processes are usually of sufficiently high surface quality and flatness to meet the requirements for bonding. Most of the commercial glass plates and wafers commonly available can meet the flatness and smoothness requirements for bonding, though some will require grinding and polishing. In practice, a glass plate with a flatness of 3 µm and a surface quality of 20/10 (scratch/dig of MIL Spec Standards) grade optical finish will be sufficient for achieving good hermetic bonding. Surface Cleaning and Preparation An effective process for cleaning and preparing both silicon and glass surfaces for bonding is to perform a hot piranha clean for 10–20 minutes, followed by a deionized (DI) water rinse. Various mixing ratios of H2SO4 and H2O2 for the piranha solution have been reported, ranging from 2:1 to 4:1. When freshly mixed, the bath temperature can reach up to 160°C, depending on the mixing ratio. The piranha cleaning step has two primary functions. It is effective in cleaning off organic contaminants such as thermal wax or photoresist residues, resulting in a hydrophilic surface. It is also moderately effective in removing particles. Oxygen plasma and UV ozone cleanings are also effective in preparation of the silicon or glass surface for bonding. Oxygen plasma has long been used to clean off organic residues. Both cleaning processes will result in a hydrophilic surface. Oxygen plasma and UV ozone cleanings are particularly attractive when components such as metal electrodes are present. For example, both Ti and Ta, often used as an adhesion layer for noble metals, have appreciable etch rates in hot piranha solution. The standard RCA clean can also be used to effectively clean and prepare silicon wafers for bonding (Tong and Gösele, 1999). The RCA clean comprises two steps. The first (RCA1) is a mixture of 64% NH4OH, 30% H2O2, and H2O with a volumetric mixing ratio of 1:1:5 to 1:2:7 and is used in the temperature range of 75–85°C. The second (RCA2) is a mixture of 37% HCl, 30% H2O2, and H2O with a volumetric mixing ratio of 1:1:6 to 1:2:8 and a temperature range of 75–85°C. Both cleaning steps should be followed with DI water rinsing. RCA1 is effective in removing organic contamination and particles, and the resulting surface is highly hydrophilic. RCA2 is effective in removing metal contaminants and should be used for devices with applications that are sensitive to the presence of trace metals. The resulting surface, however, is less hydrophilic, unless the RCA2 temperature is kept at 100°C. It is worth noting that the RCA1 clean does not result in a hydrophilic surface on glass plates, so an oxygen plasma


39

or UV ozone cleaning step is needed to convert the surface to a hydrophilic state. Anodic and Fusion Bonding Anodic bonding is a simple and effective process to permanently bond a silicon wafer to a glass substrate (Albaugh et al., 1988; Frank et al., 1994; Ko et al., 1985; Younger, 1980). The bonding process is performed at a temperature ranging from 300 to 500°C, with an applied voltage of 500–1000 volts. The anode of the power supply is connected to the silicon wafer and the cathode to the glass wafer. The bonding duration is dependent on the temperature and the voltage but is usually on the order of 10–15 minutes. A good match of thermal expansion coefficients between the silicon and glass is necessary to avoid buckling after bonding and cooling. Pyrex glass from Coming is a suitable choice for close match of its thermal expansion coefficient to that of silicon. Hoya SD-2 glass can also be used; it has an even better match of thermal expansion coefficient to that of silicon, and is more suitable for applications requiring low warping and low stress. Because of the relatively low temperature employed in anodic bonding, the bonding interface is usually free of gas bubbles as gas entrapment does not occur readily at these temperatures. Fusion bonding, also referred to as direct bonding (d’Aragona and Ristic, 1994; Tong and Gösele, 1999), is the most commonly employed process in forming enclosures such as chambers, channels, and cavities in a glass microchip. Although the initial bond formed at room temperature is adequate for applications requiring very low operational pressure, a thermal fusion process is usually performed to enhance the bond strength of the glass microchip. The thermal fusion temperatures should be above the annealing point but well below the softening point of the glass to achieve maximum bonding strength without deforming the microchannel features. The thermal fusion duration is dependent on the temperature employed, but usually is on the order of 1–2 hours. To minimize the stress, the cooling schedule should be carefully designed so that the microchip plates go through the strain point at a gradual rate. A cooling rate of 10°C per minute is usually sufficient. The bond strength will be dependent on the type of glass used and the detailed thermal fusion parameters. When the thermal conditions are optimized, a microchip made in Coming 0211 glass with a single channel several hundred microns wide and several centimeters long can typically hold up to several hundred p.s.i. pressure without showing any sign of failure. Certainly, densely placed multichannel devices fabricated in even thicker glass plates with optimized bonding conditions are expected to hold pressures up to 1000 p.s.i. Embedding Electrodes For many applications, it is desirable to incorporate biocompatible electrodes, usually gold or platinum, on the microchips for applying separation voltage and for electrochemical sensing. However, these metals require an adhesion layer to stick well onto silicon and glass surfaces. Chromium, although a common adhesion layer, is not recommended for on-chip electrodes due to its bioincompatibility. Ni, Ti, Ta, and W are generally considered to be reasonable adhesion layer materials for gold and platinum

Biochip technology

40

electrodes in microchips. The electrodes can be placed onto either the etched channel plate or the cover plate by photolithographic and metal etching processes. The approach of placing the electrodes onto the cover plates has the advantage of simpler photolithography, but will usually require high-precision alignment accuracy during the bonding process. When the electrodes are placed onto the plate with etched channels, a thick photoresist is usually required to achieve proper step coverage over the etched features. This approach, however, is limited to designs with channel depths of less than 50 microns. Deeper channels will result in poor electrode definition and are of little practical use. The total metal thickness needs to be under 1000 Å to allow proper fusion bonding of the two plates without leaving air gaps along the electrodes. If a thicker metal layer is required for the embedded electrodes, counter-sinking becomes necessary to achieve proper bonding. As discussed in the section on “Surface Cleaning and Preparation,” the plate with the electrodes cannot go through the usual piranha or RCA cleaning steps required for bonding. Instead, oxygen plasma or UV ozone treatments should be used to clean and prepare plates for bonding. The thermal fusion temperature and duration should also be carefully controlled and kept to a minimum, as diffusion and oxidation degrade embedded electrodes, particularly ones made of gold. To access the contact pads of the on-chip electrodes, one common design practice is to have holes drilled directly above the contact pads in the opposite piece. Alternatively, one can use dissimilar plate sizes to expose the contact pads. Figure 11 illustrates the designs the authors have practiced over the last few years.

SUMMARY The most basic fabrication processes and techniques for silicon and glass microchips have been presented in this chapter, with an emphasis on the practical aspects of the technology. Despite its success in producing a wide variety of research devices and even small-scale commercial products for various applications, microfabrication technology for silicon and glass-based chemical and biological chips is still in its infancy. Many fabrication issues need to be addressed in greater detail before a higher level of integration can be achieved. One such major issue is the packaging and interface of the devices with the macro world. There is no mature and standard interface technology at present. Each device and application has generated a unique solution. It is the authors’ view that significant progress must be made in devising more standard solutions for an integrated fluid interface and delivery of sample and reagents, as well as detection schemes, before the potential of microchip technology can be fully exploited.


41

Figure 11. Schematic representation of chip designs with on-chip electrodes. REFERENCES Albaugh K.B., Cade P.E., and Rasmussen D. 1988. Mechanism of anodic bonding of silicon to Pyrex glass. Tech Dig. IEEE Solid-State Sens. Actuator Workshop, Hilton Head, SC, pp. 109–110. Bassous E. 1978. Fabrication of novel three-dimensional microstructures by anisotropic etching of (100) and (110) silicon. IEEE Trans. Electron. Devices ED-25:1178–1184. Bean K. 1978. Anisotropic etching of silicon. IEEE Trans. Electron. Devices ED25:1185–1193. Bellows G., and Kohls J.B. 1982. Drilling without drills. Am. Mach. Special Rep. 743:173–188. Campbell S.A., and Lewerenz H.J. 1998. Semiconductor Micromachining, Vol. 2. John Wiley & Sons, New York. Chapman B.N. 1980. Glow Discharge Processes: Sputtering and Plasma Etching. John Wiley & Sons, New York. Constantine C. 1997. Plasma etching of quartz, glasses holds promise for optical applications. Micromachine Devices 2:12. Dammel R. 1993. Diazonaphthoquinone-Based Resists. SPIE Tutorial Texts, Vol. TT11. SPIE-The International Society for Optical Engineering, Bellingham, Washington. d’Aragona F.S., and Ristic Lj. 1994. Silicon direct wafer bonding. In Sensor Technology and Devices. Lj. Ristic, ed. Artech House, Boston/London, pp. 157–201. Elliot D. 1982. Integrated Circuit Fabrication Technology. McGraw-Hill, New York. Fan Z.H., and Harrison D.J. 1994. Micromachining of capillary electrophoresis injectors

Biochip technology

42

and separators on glass chips and evaluation of flow at capillary intersections. Anal. Chem. 66:177–184. Frank R., Kniffin M.L., and Ristic Lj. 1994. Packaging for sensors. In Sensor Technology and Devices, Lj. Ristic, ed. Artech House, Boston/London, pp. 203–238. Glang R. 1970. Vacuum evaporation. In Handbook of Thin Film Technology. L.I.Maissel and R. Glang, eds. McGraw-Hill, New York, 1:1–130. Glang R., and Gregor L.V. 1970. Generation of patterns in thin films. In Handbook of Thin Film Technology. L.I.Maissel and R.Glang, eds. McGraw-Hill, New York, 7:1– 66. He, B., Tait N., and Regnier F. (1990). Fabrication of nanocolumns for liquid chromatography Anal. Chem. 70:3790–3797. Hein A, Dorsch O., and Obermeier E. 1997. Effects of metallic impurities on anisotropic etching of silicon in aqueous KOH solutions. Proc. 1997 Intl. Conf. Solid State Sens. Actuators, Chicago, pp. 687–690. Jacobson S.C., Hergenroeder R., Koutny L.B., Warmack R.J., and Ramsey J.M. 1994. Effects of injection schemes and column geometry on the performance of microchip electrophoresis. Anal. Chem. 66:1107–1113. Kendall D. 1979. Vertical etching of silicon at very high aspect ratios. Annu. Rev. Mater. Sci. 9:373–403. Kendall D., and Schoults R.A. 1997. Wet chemical etching of silicon and SiO2, and ten challenges for micromachiners. In Handbook of Microlithography, Micromachining, and Microfabrication. P.Rai-Choudhury, ed. SPIE Press monograph PM40 and IEE Materials and Devices Series 12B, pp. 41–97. Kern W, and Deckert C. 1978. Chemical etching. In Thin Film Processes. J.Vossen and W. Kern, eds. Academic Press, San Diego, pp. 401–481. Ko W.H., Suminto J.T., and Yeh J.G. 1985. Bonding techniques for microsensors. In Micromachining and Micropackaging of Transducers. C.D.Fung, P.W.Cheung, W.H.Ko, and D.G. Fleming, eds. Elsevier Science Publishers, Amsterdam, pp. 41–61. Madou M. 1997. Fundamentals of Microfabrication. CRC Press, Boca Raton, Florida. Maissel L.I., and Glang R. 1970. Handbook of Thin Film Technology. McGraw-Hill, New York. Melliar-Smith C.M., and Mogab C.J. 1978. Plasma assisted etching techniques for pattern delineation. In Thin Film Processes. J.Vossen and W.Kern, eds. Academic Press, San Diego, pp. 497–556. Merlos A., Acero M., Bao M.H., Bausells J., and Esteve J. 1993. TMAH/IPA anisotropic etching characteristics. Sens. Actuators A37–38:737–743. Moreland M.A. 1992. Ultrasonic machining. In Engineered Materials Handbook. S.J.Schneider, ed. ASM International, Metals Park, OH, pp. 359–362. Mucha J.A., Hess D.W., and Aydil E.S. 1994. Plasma etching. In Introduction to Microlithography, 2nd ed. L.F.Thompson, C.G.Wilson, and M.J.Bowden, eds. American Chemical Society, Washington, DC, pp. 377–507. Muller R.S., Howe R.T., Senturia S.D., Smith R.L., and White R.M. 1990. Microsensors. IEEE Press, The Institute of Electrical and Electronics Engineers, New York. Obermeier E. 1995. Anodic wafer bonding. Proc. 3rd Int. Symp. Semiconductor Wafer Bonding: Physics and Applications. C.E.Hunt, H.Baumgart, S.S.Iyer, T.Abe, and


43

U.Gösele, eds. The Electrochemical Society, New York, pp. 212–220. Pandhumsopom et al. 1998. High etch rate, deep anisotropic plasma etching of silicon for MEMS fabrication. SPIE Symp. Smart Struct. Mater., San Diego. Excerpted in Micromachine Devices 3:5–6. Petersen K. 1982. Silicon as a mechanical material. Proc. IEEE 70:420–457. Rai-Choudhury, P. 1997. Handbook of Microlithography, Micromachining, and Microfabrication, Vols. 1 and 2. SPIE Press monograph PM40 and IEE Materials and Devices Series 12B. Rand M.J. 1975. I–V characteristics of PtSi-Si contacts made from CVD platinum. J. Electrochem. Soc. 122:811–815. Rand M.J., and Roberts J.F. 1974. Formation and etching of platinum silicide. Appl. Phys. Lett. 24:49–51. Reisman A., Berkenblit M., Chan S.A., Kaufman F.B., and Green D.C. 1979. The controlled etching of silicon in catalyzed ethylenediamine-pyrocatechol-water solutions. J. Electrochem. Soc. 126:1406–1415. Ristic Lj. 1994. Sensor Technology and Devices. Artech House, Boston/London. Sasserath J., Johnson D., DeVre M., Hartwell P., and Chong J.M. 1997. DRIE profile control holds promise for varied applications. Micromachine Devices 2:1. Seidel H., Csepregi L., Heuberger A., and Baumgarten H. 1990. Anisotropic etching of silicon in alkaline solutions. J. Electrochem. Soc. 137:3612–3626. Sekimura M. 1999. Anisotropic etching of surfactant-added TMAH solution. Tech. Dig., 12th IEEE Int. Conf. Micro Electromech. Systems (MEMS’99) FL, pp. 650–655. Simpson P.C., Woolley A.T., and Mathies R.A. 1998. Microfabrication technology for the production of capillary array electrophoresis chips. Biomed. Microdevices 1:7–26. Tabata O., Asahi R., Funabashi H., Shimaoka K., and Sugiyama S. 1992. Anisotropic etching of silicon in TMAH solutions. Sens. Actuators A34:51–57. Thompson L.F., Wilson C.G., and Bowden M.J. 1994. Introduction to Microlithography, 2nd ed. American Chemical Society, Washington, DC. Tong Q.-Y, and Gösele U. 1999. Semiconductor Wafer Bonding. John Wiley & Sons, New York. Vossen J.L., and Kern W. 1978. Thin Film Processes. Academic Press, San Diego. Younger P.R. 1980. Hermetic glass sealing by electrostatic bonding. J. Non-Crystalline Solids 38/39:909–914.

3 Self-Assembled Monolayers Applications in Surface Modification and Micropatterning Younan Xia, Byron Gates, and Yadong Yin

INTRODUCTION Self-assembled monolayers (SAMs) are highly ordered two-dimensional (2D) arrays that form spontaneously by chemisorption and self-organization of functionalized long-chain organic molecules on the surfaces of appropriate solid substrates (Ulman, 1991; Whitesides and Laibinis, 1990). The formation of SAMs represents a good example of molecular self-assembly, in which molecules organize themselves into stable welldefined structures by non-covalent forces (Lehn, 1990; Whitesides et al., 1991; Whitesides, 1995). The key idea in self-assembly is that the final structure is defined and directed by the intrinsic characteristics—for example, shape, length, and functionality— of the starting molecules (Isaacs et al., 1999). Because self-assembly usually leads to an equilibrium state that is at, or close to a free energy minimum, the self-assembling structure tends to be self-healing, defect-rejecting, and capable of achieving a greater order than can be reached by non-self-assembling approaches (Whitesides, 1995). Many different types of SAMs have been demonstrated, and new systems are still being developed (Isaacs et al., 1999; Ulman, 1991). Table 1 summarizes organic molecules that have been shown to form stable SAMs on the corresponding substrates. There are also a few other systems in which only chemisorption was involved, and neither monolayers nor ordered structures were formed: these have also been called SAMs. The spontaneity of forming an ordered monolayer is driven by the thermodynamically favored segregation of molecules to the phase boundary between the solid substrate and the solution (or vapor). The principal requirement for forming an SAM is that the organic molecules bear an end group (or in the form of a precursor) that is reasonably reactive toward the surface atoms (or functional groups) of the solid substrate. In addition, the organic molecules have to be long enough to ensure a sufficiently strong van der Waals interaction among the alkyl backbones (Delamarche et al., 1996). It is an interplay of these two types of interactions that drives the organic molecules into a closely packed 2D assembly on the surface of the solid substrate. The most studied and best characterized systems of SAMs are alkanethiolates [CH3(CH2)nS–] on coinage metals (Au and Ag) (Delamarche et al., 1996; Dubois and Nuzzo, 1992; Whitesides and Laibinis, 1990), and

Applications in surface modification and micropatterning

45

Table 1 Ligands That Form Stable SAMs on the Corresponding Substrates Substrate Ligand or precursor Binding Au, Ag, Cu RSH, ArSH (thiols) RS–M (M=Au,Ag,Cu) Au RSSR′ (disulfides) RS–Au, R′S–Au Au RSR′ (sulfides) RS–Au, R′S–Au Au Pd Pt GaAs, lnP SiO2, glass

RSO2H RSH, ArSH RNC RSH RSiCl3, RSi(OR′)3

RSO2–Au RS–Pd RNC–Pt RS–GaAs, RS–lnP Siloxane

Si/Si–H Si/Si–H Si/Si–Cl Metal oxides

(RCOO)2 (neat) RCH=CH2 RLi, R–MgX RCOOH

Metal oxides ZrO2

RCONHOH RPO3H2

R–Si RCH2CH2Si R–Si RCOO– … MOn RCONHOH … MOn

ln2O3/SnO2 (lTO)

RPO3H2

RPO3 2– … Zr(IV) RPO3 2– … M(n+)

alkylsiloxanes on hydroxyl-terminated surfaces (for example, Si/SiO2, A1/AI2O3, glass, and oxygen plasma-treated polymers) (Allara et al., 1995; Ulman, 1991; Wirth et al., 1997). Other systems shown in Table 1 are relatively less established; some of them were only studied very briefly by one or two research groups. SAMs are robust and relatively stable. They also have the capability and flexibility (both at the molecular and materials levels) required to tailor the interfacial properties— for example, surface free energy, wettability, and biocompatability—of a rich variety of solid substrates (Bain and Whitesides, 1989a; Mrksich and Whitesides, 1996; Whitesides and Laibinis, 1990). They have, in the past, been extensively explored as model systems to study a wide range of interfacial phenomena, such as wetting, dewetting, spreading, adhesion, lubrication, corrosion, condensation, nucleation, and protein adsorption (Bain and Whitesides, 1989b; Mrksich and Whitesides, 1996; Prime and Whitesides, 1993; Wasserman et al., 1989). More recently, SAMs were seriously examined and evaluated as nanometerthick resists for pattern transfer in generating high-quality micro- and nanostructures (Kumar et al., 1995; Xia et al., 1996e, 1999). The use of SAMs as ultrathin resists provides several advantages over traditional materials (usually, thin films of polymers). For example, microfabrication methods involving SAMs are relatively lowcost compared with conventional microlithographic methods; the relatively low solidvapor interfacial free energies of CH3- and CF3-terminated SAMs allow them to be handled outside clean room facilities without irreversible contamination. Because SAMs are so thin, some issues—such as depth of focus, optical transparency in ultraviolet (UV)

Biochip technology

46

and vacuum UV regions, shadowing, and undercutting—that are currently influencing the performance of photoresists in high-resolution imaging processes do not apply to SAMs. The availability of nanometer-thick resists also opens the door to new microlithographic approaches. For example, in a process that uses metastable argon or cesium atoms as the pattern-generating species, the resist must be less than 2 nm thick, because the damage in the resist by contact with the metastable atoms is often limited to a surface layer of ≤0.5 nm thick (Berggren et al., 1995, 1997; Youkin et al., 1997). Because a number of reviews have already been devoted to SAMs and their applications (Delamarche et al., 1996; Dubois and Nuzzo, 1992; Ulman, 1991; Whitesides and Laibinis, 1990; Xia and Whitesides, 1998; Xu and Li, 1995), this chapter will only focus on SAMs of alkanethiolates on evaporated polycrystalline thin films of gold (or silver). We shall give a brief overview on the formation and characterization (order, molecular structure, and defects) of these SAMs, as well as their unique applications in surface modification and micropatterning.

PREPARATION AND CHARACTERIZATION OF SAMs The formation of a highly ordered SAM is a complicated multiple-stage process that involves at least three spontaneous steps: reaction, adsorption, and self-organization (Ulman, 1991; Delamarche et al., 1996). Nevertheless, SAMs are remarkably easy to prepare in an ordinary laboratory: they can be simply obtained by immersion of the substrate in a solution containing molecules reactive toward the surface, or by exposure of the substrate to the vapor of the reactive species. In most cases, the substrate has to be prepared by a specific method in order to obtain a high-quality SAM. For example, the gold (or silver) substrates are usually prepared by thermal evaporation or e-beam sputtering as thin films (20–200 nm thick) on Si/SiO2 or glass supports that have been primed with 2- to 3-nm thick layers of titanium or chromium. The Si/SiO2 and glass substrates for use with alkylsiloxane SAMs are often cleaned by treating with piranha solution (a 3:7 mixture of and H2SO4) that is heated to ~70°C. Formation of Alkanethiolate SAMs Ordered SAMs of alkanethiolates on gold (Figure 1A) are usually prepared by immersing polycrystalline thin films of gold in approximately 2 mM solutions of alkanethiols in ethanol for several minutes (Whitesides and Laibinis, 1990). It is generally accepted that the reaction between alkanethiols and gold surface occurs by loss of dihydrogen gas:

Per mechanism, the alkanethiol molecules chemisorb on the gold surface and form immobilized alkanethiolate species. The chemical bonding between the sulfur atoms of alkanethiolates and the gold surface (≈44 kcal/mol) is strong enough to anchor the alkyl chains on gold and bring them into close contact with each other. The van der Waals interactions of these contacts freeze out configurational entropy and eventually lead to the


47

formation of a closely packed 2D array of alkanethiolate molecules (Delamarche et al., 1996), up to alkyl chains of approximately 20 carbon atoms. The degree of interaction in an alkanethiolate SAM increases with the density of molecules on the surface and the number of methylene units of the alkyl backbone. It has been shown that only alkanethiolates with n>11 form highly ordered polycrystalline structures on the surface of gold (Delamarche et al., 1996).

Figure 1. (A,B) Representation of a highly ordered monolayer of alkanethiolate, X(CH2)nS–, on the surface of Au(111). The head group, X, allows the interfacial properties of the monolayer to be controlled at the molecular level. The thickness of an SAM usually increases linearly with the number (n) of methylene groups in the alkyl chain. The alkanethiolate molecules are, on average, tilted by ~30° from the normal to the surface of gold. (C) Representation of a densely packed SAM on gold that is formed from a mixture of different alkanethiols. The mechanistic details of the reaction between alkanethiols and gold surfaces are still under debate. For instance, the fate of hydrogen atoms and the exact format of the resulting species on the gold surface still need to be resolved. A recent X-ray diffraction study suggested the possibility of having disulfides (X–CnH2n–S–S–CnH2n–X) rather

Biochip technology

48

than thiolates (X–CnH2n–S–) on the surface of gold, although this suggestion requires physically unreasonable bond length (Fenter et al., 1994). Recent studies based on mass spectrometry also indicated that the SAMs prepared from solutions of alkanethiols in the ambient atmosphere of a laboratory are often made of a mixture of and because the oxidation of alkanethiolates by air is a relatively fast process in the presence of UV light (Huang and Hemminger, 1993; Li et al., 1992; McCarley and McCarley, 1997). The kinetics for the formation of alkanethiolate SAMs on gold films has been extensively studied using a range of techniques (Ulman, 1991): contact angle (Bain and Whitesides, 1989a), ellipsometry (Bain and Whitesides, 1989a), quartz crystal microbalance (QCM) (Buttry and Ward, 1992; Karpovich and Blanchard, 1994), surface acoustic wave (Buttry and Ward, 1992), and surface plasmon resonance (SPR) (Jordon and Corn, 1997). These studies suggested that the deposition rate at any moment was proportional to the number of unoccupied sites remaining on the gold surface, and could be described as a first-order Langmuir adsorption process. Although some of these techniques are capable of monitoring the formation of SAMs in situ, most of them are spatially averaging techniques, and thus left questions about the microscopic aspect of self-assembly unanswered. This situation did not change until scanning tunneling microscopy (STM) was recently employed to investigate this process (Poirier, 1997). Based on their results obtained from an ultrahigh-vacuum (UHV) STM study, Poirier et al. proposed a molecular-scale mechanism for the formation of an ordered SAM, in which the alkanethiols sequentially form the following phases with an increasingly higher coverage: a lattice-gas phase, a low-density solid phase, and a high-density solid phase. This mechanism might be a general one for the self-assembly of alkanethiolates on a gold surface. Since the tunneling current of an STM measurement decays exponentially with distance, this growth model has only been observed in systems having relatively short alkyl chains, including SAMs from the vapor phase of HSC6H13COOH or HSC9H19CH3, and SAMs from the solution phase of HSC9H19CH3 (Poirier and Pylant, 1996). Structure and Order of SAMs The structure and order of an SAM can be characterized by a wide variety of techniques (Table 2): for example, polarized infrared external reflectance spectroscopy (PIERS), transmission electron diffraction, low-energy helium diffraction, low-angle X-ray scattering, and scanning probe microscopy (Alves et al., 1992; Camillone et al., 1991, 1993a,b; Poirier, 1997; Strong and Whitesides, 1989). The results from diffraction and SPM studies suggested that the Au atoms on a polycrystalline gold film have an arrangement similar to the (111) face of a single crystal of gold, and that the sulfur atoms of the alkanethiolate molecules form a

Table 2 Methods for the Characterization of Alkanethiolate SAMs on Gold Property of SAMs Technique Structure and order Scanning probe microscopy (STM, AFM, and LFM)


Composition

Wettability Thickness Coverage or kinetics Defects and/or degree of perfection

49

Infrared spectroscopy (PIERS) Low-energy helium diffraction X-ray diffraction Transmission electron diffraction Surface Raman scattering Sum frequency spectroscopy (SFS) X-ray photoelectron spectroscopy (XPS) Temperature-programmed desorption (TPD) Mass spectrometry (MS) Contact angle, surface energy Ellipsometry, surface reflectivity Quartz crystal microbalance (QCM) Surface acoustic wave device (SAWD) STM and AFM Wet chemical etching Electrochemistry (e.g., cyclic voltammetry)

overlayer structure on this Au(111) surface (Figure 1B). PIERS measurements indicated that the alkyl chains usually tilt at an angle of ~30° from the surface normal in order to maximize the van der Waals interactions among the methylene groups of adjacent alkyl chains (~1.5 kcal/mol per CH2) (Dubois and Nuzzo, 1992). Figure 2 shows an atomic resolution STM image of the SAM of dodecanethiolate (n=12) on an Au(111) surface that was obtained by Biebuyck and coworkers (Delamarche et al., 1996). This image clearly shows a hexagonal close packing for the adsorbed dodecanethiolate molecules. According to these authors, the gold terrace shown here has five depressions (black “holes”) that are approximately one gold step (~2.4 Å) in depth. These depressions are pits in gold rather than defects in SAMs because their surfaces are still covered with ordered SAMs having a lattice characteristic of the packing of molecules in the SAM on the flat surface. The origin of these depressions in gold surface is still not completely understood: they could originate from a corrosion process or from a construction of the gold surface caused by adsorption of the thiols (McDermott et al., 1995; Schonenberger et al., 1994). Structural studies suggest that the order in the top part of an SAM is not solely determined by the sulfur atoms that bond directly to the gold surface, but also strongly depends on the intermolecular interactions among the alkyl chains. The alkyl chains may also have several different types of conformations (e.g., a mixture of cis and trans), and thus form a “superlattice” at the surface of the monolayer (Camillone et al., 1993a,b). As a result, it has been difficult to predict and determine the structures of SAMs formed from alkanethiolates that are terminated in head groups other than the methyl group (Delamarche et al., 1996). It has been demonstrated that the end group (in particular, a bulky one) may, in some cases, play the most important role in determining the packing structure and order of an SAM (Delamarche et al., 1996).

Biochip technology

50

Figure 2. The STM image of an SAM of dodecanethiolate (n=12) on the surface of Au(111) that shows a packing structure corresponding to a phase of C(4 * 2) rectangular superlattice. It also shows five pits in gold that were ~2.4 Å deep and linked by domain boundaries (Delamarche et al., 1996). Reprinted with the permission and through the courtesy of Dr Delamarche, IBM. Defects in SAMs The density of defects in SAMs is a very important issue, since it may ultimately determine the practical value of these materials in many applications, particularly in those related to micro- and nanofabrication (Xia et al., 1996a–e, 1999; Xia and Whitesides, 1998). Although SAMs are representative self-assembling systems that are supposed to be self-healing and defect-rejecting, formation of defects in SAMs seems to be inevitable in reality because the true thermodynamic equilibrium may never be achieved in the preparation of SAMs. The typical defects in an SAM are pinholes where the surface is not derivatized by the organic molecules. Current estimates for the number of pinholes in SAMs of hexadecanethiolate on evaporated thin films of gold range from two to several thousand per cm2, with the latter value being more realistic. A recent study using a twostage wet etching method to amplify the defects in SAMs gave ~90 pits /mm2 as a minimum number for the density of defects in an SAM of hexadecanethiolate on 20-nm thick gold (Zhao et al., 1996). Preparation of truly defect-free SAMs still remains a great challenge. It has been found that the formation and distribution of defects in an SAM depends on a number of factors:


51

for example, the atomic structure of the surface, the length of the alkyl chain, and the conditions under which the SAM is prepared (Zhao et al., 1996). Annealing at elevated temperatures may induce reorganization of the SAM and the surface, and thus reduces the density of pinholes in the SAM by migration and coalescence of pinholes in gold with steps (Delamarche, et al., 1996). While it has been extremely difficult to predict and control the density of defects in SAMs, a rich variety of methods (see Table 2) have been demonstrated for examining and evaluating the characteristics of these defects (Poirier, 1997; Sun and Crooks, 1993; Zhao et al., 1996). These methods are complementary: for instance, the SPM method can provide useful information about the nature and origin of defects although it can only collect information over a relatively small area; the electrochemical method can only give a statistical estimation of defects over a relatively large area; and the wet etching method may increase the density of defects due to the harsh conditions usually involved in etching. As a result, a combination of these methods seems to be the most powerful approach to an understanding of defects in SAMs. Stability of SAMs Different types of SAMs have different stabilities to heating and chemicals. In vacuum or under an inert atmosphere, the thermal stability of an SAM should be directly proportional to the strength of the covalent bonding between the solid surface and the organic molecules, and this stability can be significantly increased by creating H-bonding or chemical crosslinking among the alkyl chains (Tam-Chang et al., 1995). For SAMs of alkanethiolates on thin films of gold, the adsorbed molecules usually become disordered and/or decompose around ~100°C; oxidation of alkanethiolates to alkanesulfonates in the presence of UV light and ozone also greatly reduces stabilities of these SAMs (Delamarche et al., 1994; Dubois and Nuzzo, 1992; Huang and Hemminger, 1993; Ulman, 1991). In contrast, some SAMs of alkylsiloxanes on Si/SiO2 could be stable up to ~450°C (Fontaine, et al., 1993); part of the reason lies in the fact that the alkyl chains in siloxane SAMs are covalently crosslinked into a 2D sheet via siloxane bonds.

APPLICATION OF SAMs IN SURFACE MODIFICATION SAMs exhibit many of the attractive features that are characteristic of a self-assembling system: for example, ease of preparation, good stability in atmosphere, and relatively low density of defects. More importantly, SAMs provide a versatile means for tailoring (in a controllable way) the interfacial properties (e.g., physical, chemical, electrochemical, biochemical, biological, and tribological) of a variety of substrates (Bain and Whitesides, 1989b; Whitesides and Laibinis, 1990; Zhuk et al., 1998). For a highly ordered SAM of alkanethiolates on gold (Figure 1A), its surface properties are mainly determined by the head groups in which the alkyl chains terminate. For example, alkyl chains terminated in a –CH3 or –CF3 group are extremely hydrophobic, while those terminated in a –COOH or –OH group are highly hydrophilic (Atre et al., 1995; Bain and Whitesides, 1989b). Alkyl chains terminated in ethyleneglycol units (or oligomers) usually have the capability to resist the adsorption of certain types of proteins (Mrksich and Whitesides, 1995; Prime

Biochip technology

52

and Whitesides, 1993). Ligands specific to certain biomacromolecules can also be incorporated into SAMs to generate biosurfaces with well-controlled activities (Mrksich and Whitesides, 1996). These biosurfaces have been used as active elements to fabricate sensors, and as model systems to investigate the attachment and spreading of mammalian cells (Mrksich and Whitesides, 1996; Mrksich et al., 1996; Vigmond et al., 1994). If necessary, mixed SAMs (Figure 1C) with more than one head group and/or chain length can also be prepared to generate interfaces with more complex properties (Hickman et al., 1991; Whitesides and Laibinis, 1990). This type of surface has been actively explored as a model system to investigate the influence of heterogeneity on the wettability of a solid surface. For example, Whitesides and coworkers have studied the wetting behavior of gold surfaces that had been derivatized with mixed SAMs of alkanethiolates terminated in CH3– and COOH-groups (Bain and Whitesides, 1989a). Olbris and coworkers (1995) studied a similar system, and proposed a modified model to explain the wetting behavior of hexadecane on such a kind of surface. This type of surface has also been widely used to study the adsorption of proteins on solid substrates (Mrksich and Whitesides 1996; Prime and Whitesides, 1993; Roberts et al., 1998). In the past, the composition of a mixed SAM was usually determined as an averaged value by using XPS. With the development of SPM techniques, more and more information (at the molecular level) has been obtained on the structure and order in mixed SAMs. Different from what has been known about mixed Langmuir-Blodgett (LB) films (Overney et al., 1992), no phase segregation has been observed at scales larger than 50 nm for twocomponent SAMs on gold surfaces. Asymmetric disulfides (RS-SR′) seem to provide the most versatile system for studying phase segregation in mixed SAMs because the adsorption of disulfide on gold leads to two equal populations of end groups in the resulting SAM (Biebuyck et al., 1994; Ishida et al., 1997; Schönherr and Ringsdorf, 1996). An STM study on mixed SAMs consisting of methyl and hydroxyl head groups demonstrated that the two components are well mixed without disruption of the packing in the monolayer (Takami et al., 1995).

FORMATION OF PATTERNED SAMs Patterning of SAMs in the plane of the monolayer allows one to engineer the interfacial properties of a substrate with one more degree of freedom. The ability to form patterned SAMs, for example, offers immediate opportunities to prepare systems in which structures can be tightly controlled in the plane of the interface. These patterned SAMs serve as tunable model systems to study nucleation, growth, adsorption, wetting, and other interfacial phenomena on well-defined heterogeneous surfaces (Aizenberg et al., 1999; Gau et al., 1999; Lopez et al., 1993a,b; Singhvi et al., 1994). They can be used as templates to define and control the assembly of a variety of materials to form functional microstructures on solid substrates (Biebuyck and Whitesides, 1994a,b; Kim et al., 1996a,b; Lahiri et al., 1999). They can also be used as ultrathin resists in directing the dissolution of the underlying substrates to form patterned microstructures in various types of materials, such as Au, Ag, Cu, SiO2, Si, and GaAs (Xia and Whitesides, 1998; Xia et al., 1995a, 1996e). As a matter of fact, patterning of SAMs provides an alternative


53

approach to microfabrication, an area that has so far been dominated by photolithography (Madou, 1997). Lateral patterning of SAMs can be achieved by a variety of different techniques (Table 3). Conventional lithographic methods such as UV photolithography (Calvert, 1995; Chan et al., 1995; Huang et al., 1994; Tarlov et al., 1993), e-beam writing (Gillen et al., 1994; Lercel et al., 1993; Sondag-Huethorst, 1994), and micromachining using an STM tip (Ross et al., 1993) or sharp stylus (Abbott et al., 1992) have been demonstrated for generating patterns in SAMs of alkanethiolates on gold, silver, and gallium arsenide, and SAMs of alkylsiloxanes on silicon. More recently, new approaches such as microcontact printing (µCP) (Kumar and Whitesides, 1993;

Table 3 Methods That Have Been Demonstrated for Patterning SAMs Patterning technique SAM/substrate Resolutiona Microcontact printing (µCP) RSH/Au ~35 nm RSH/Ag ~100 nm RSH/Cu ~500 nm RSH/Pd ~500 nm Photooxidation RSH/Au ~10 µm RSH/Au ~10 µm Photo-crosslinking Photoactivation RSH/Au ~10 µm RSH/Au ~75 nm Electron beam writing RSH/Ag ~10 µm Focused ion beam writing Neutral metastable atom writing RSH/Au ~70 nm SPM lithography RSH/Au ~10 nm RSH/Au ~100 nm Micromachining RSH/Au ~10µm Micropen writing aThe lateral dimension of the smallest feature that has been fabricated. Kumar et al., 1994, 1995; Wilbur et al., 1994) and microlithography with metastable atoms (Berggren et al., 1995) have been developed for generating patterned SAMs. The patterned features of SAMs can be directly visualized using a range of techniques such as scanning electron microscopy (SEM) (Lopez et al., 1993a,b; Kumar et al., 1994), scanning probe microscopy (SPM) (Wilbur et al., 1995a), secondary ion mass spectrometry (SIMS) (Lopez et al., 1993a), condensation figures (CFs) (Lopez et al., 1993a), and surface-enhanced Raman microscopy (Yang et al., 1996). Because microcontact printing seems to offer the most versatile combination of simplicity, new capability, and convenience, we will focus on this technique in the next several sections.

MICROCONTACT PRINTING (µCP) The procedure for µCP is quiet and straightforward: the printing step involves a direct

Biochip technology

54

contact between the raised portions of an elastomeric stamp and the surface of a solid substrate. The elastomer stamp that has been commonly used in most demonstrations is prepared by casting a liquid prepolymer (e.g., Sylgard 184 from Dow Coming) of poly (dimethylsiloxane) (PDMS) against a master whose surface has been patterned with complementary relief structures using conventional microlithographic techniques such as photolithography or micromachining (Kumar and Whitesides, 1993). The stamp may also be replicated from other types of masters that are commercially available: for example, TEM grids, optical diffraction gratings, compact disks (CDs), corner cube reflectors, antireflection structures, polymer beads assembled on solid supports, and relief structures etched in metals or other kinds of solid substrates (Wilbur et al., 1995b; Xia et al., 1996d). In most cases, the master has to be silanized by exposure to the vapor of octadecyltrichlorosilane (or other alkyltrichlorosilanes terminated in hydrophobic groups) for ~0.5 hr. The liquid prepolymer of PDMS (Sylgard 184) is usually cured by heating at ~70°C for 2–5 hr.

Figure 3. Schematic procedure for carrying out µCP with a planar PDMS stamp on a planar substrate of gold. After the surface of this stamp had been inked with an ethanol solution of hexadecanethiol, it was dried under a stream of N2, and then brought into contact with the surface of gold for 5–10 seconds. A test pattern of hexadecanethiolate SAM was formed on the gold surface by direct contact with the stamp. Microcontact printing is an intrinsically parallel process. It can be performed with three different configurations: printing on a planar surface with a planar stamp (Kumar et al., 1994), printing on a curved surface with a planar stamp (Jackman et al., 1995), and printing on a planar surface over a large area with a rolling stamp that has been mounted on a cylindrical support (Xia et al., 1996b). Figure 3 shows the schematic procedure for printing on a planar surface of gold with a planar PDMS stamp: the stamp is wetted with a solution of hexadecanethiol (n=16) in ethanol (~2 mM) and then brought into contact with the surface of gold for 5–10 seconds. The thiol transfers from the raised regions of the stamp to the gold surface and generates patterned features of SAMs on the surface of


55

gold by forming hexadecanethiolate (CH3(CH2)15S–). A contact time between 5 and 10 seconds has to be used because longer contact times may result in destruction of the pattern due to transport of hexadecanethiol from the stamp to the gold surface in noncontact regions by diffusion through the vapor phase (Xia and Whitesides, 1995a). The success of µCP relies on two characteristics of the system: the rapid formation of a highly ordered monolayer, and the autophobicity of the printed monolayer that is able to block the spreading of the excess ink liquid across the surface (Biebuyck and Whitesides, 1994a). The adsorption and self-organization of alkanethiolates on gold is a relatively fast process: when a gold substrate is immersed in an approximately 2 mM solution of hexadecanethiol in ethanol, an ordered SAM of hexadecanethiolate will be formed within a few minutes. The formation of highly ordered SAMs of alkanethiolates during µCP may occur as quickly as within a few seconds. A recent STM study by Biebuyck and coworkers showed that, for an ink solution of approximately 100 mM dodecanethiol in ethanol, a contact time of >0.3 seconds was enough to generate a highly ordered SAM on gold that was indistinguishable from those formed by equilibration in solutions (Larsen et al., 1997; Delamarche et al., 1998a). Microcontact printing was first demonstrated using SAMs of hexadecanethiolate on gold as the example (Kumar and Whitesides, 1993). Since its first demonstration, this technique has now been successfully demonstrated to pattern a wide variety of materials (Xia and Whitesides, 1998; Xia et al., 1996e). For example, it has been extended to different types of SAMs, including SAMs of alkanethiolates on silver (Xia et al., 1996a, 1997; Yang et al., 1996), SAMs of alkanethiolates on copper (Moffat and Yang, 1995; Xia et al., 1996c); SAMs of alkylsiloxanes on HO-terminated surfaces (Jeon et al., 1995a,b, 1996, 1997; St. John and Craighead, 1996; Wang et al., 1997; Xia et al., 1995b); SAMs of RPO3H2 on A1/Al2O3 (Deng et al., 1999; Goetting et al., 1999); and SAMs of alkylamines on carboxylic anhydride-terminated surfaces (Yan et al., 1998, 1999). In the current stage of development, microcontact printing with hexadecanethiol on evaporated thin (20–100 nm thick) films of gold or silver appears to be the most reproducible system: both generate patterns of highly ordered SAMs with relatively low densities of defects. Microcontact printing of other systems is substantially less tractable: it usually yields disordered structures, and in some cases submonolayers or multilayers. Microcontact printing has also been applied to the patterning of colloidal Pd particles (Hidber et al., 1996a,b) or proteins (Delamarche et al., 1998b; Mrksich et al., 1996) on Si/SiO2 or polymeric substrates, though the formation of uniform monolayers of such kinds of materials by contact printing might be very difficult to achieve. The printed features of colloidal Pd have been used as catalysts to produce certain types of threedimensional structures of metals (e.g., copper or nickel) by electroless deposition (Hidber et al., 1996a,b).

PATTERN TRANSFER BY SELECTIVE ETCHING Although self-assembled monolayers are only 2–3 nm thick, they are robust enough to protect the underlying substrates from corrosion and dissolution. This protection mainly stems from the inherent structures of an SAM: a relatively low density of defects and a

Biochip technology

56

nearly crystalline packing. When an SAM-printed substrate is placed in an appropriate etching solution, a pattern develops in the surface of the underlying substrate due to the difference in etch rates between SAM-covered and bare regions (Xia et al., 1995a, 1996e). Because SAMs are extremely thin, there is little loss in edge definition due to the thickness of the resist film. The major determinants of edge resolution seem to be the fidelity of the printing process and the anisotropy in the etching of the underlying substrate. A number of wet etchants have been examined in conjunction with printed SAMs on varies types of substrates, and it has been demonstrated that aqueous solutions containing K2S2O3/K3Fe(CN)6/K4Fe(CN)6 are effective for use with SAMs of alkanethiolates on gold or silver, that aqueous solutions containing FeCl3 and HCl (or NH4Cl) are effective for use with SAMs of alkanethiolates on copper, and that aqueous solutions containing HCl/HNO3 are effective for SAMs of alkanethiolates on GaAs (Xia et al., 1996e). Although polymer structures assembled on patterned SAMs can be directly used as resists in conventional reactive ion etching (RIE), self-assembled monolayers alone do not have the durability to serve as resists for pattern transfer by RIE (Madou, 1997). For patterning of coinage metals by µCP, silver appears to be the most suitable element due to the small grain size observed in evaporated thin films of silver and its moderate reactivity toward wet etchants (Xia et al., 1996a). Figure 4 shows SEM images of several test patterns of silver that were fabricated using µCP with hexadecanethiol, followed by selective wet etching in an aqueous ferri-/ferrocyanide solution (Xia et al., 1996a). These test patterns represent the level of complexity, perfection, and feature size that can be produced routinely using this technique. These patterned structures of metals can be directly used as ordered arrays of microelectrodes or diffractive optical components (Kumar et al., 1994). The smallest features that have been fabricated to date with a combination of µCP with SAMs and wet etching are trenches etched in gold films that were approximately 35 nm in width, approximately 350 nm in separation, and over an area of approximately 10 µm2 (Biebuyck et al., 1997). The minimum feature size that can be achieved by microcontact printing still remains to be defined. Absorbates and particles on the substrate, the roughness of the surface, and materials properties (especially deformation and distortion) of the elastomeric stamp also influence the resolution and feature size of patterns that can be formed using µCP (Delamarche et al., 1997). Some tailoring of the properties of the PDMS stamp or development of optimized elastomeric materials will be useful for the regime smaller than 100 nm. Another important application of µCP is in the preparation of patterns of gold or silver to be used as secondary masks in the etching of underlying substrates such as silicon dioxide, glass, silicon, and gallium arsenide (Kim et al., 1995, 1996b; Whidden et al., 1996; Xia et al., 1996e). Figure 5 shows SEM images of microstructures of silicon that were fabricated using anisotropic etching in a hot aqueous solution containing KOH and 2-PrOH, with patterned structures of silver (~50 nm


57

Figure 4. SEM images of test patterns of silver (approximately 50 nm thick) that were fabricated by µCP with hexadecanethiol, followed by selective etching in an aqueous solution that contained K2S2O3 (0.1 M), K3Fe (CN)6 (0.01 M), and K4Fe(CN)6 (0.001 M) (Xia et al., 1996). The bright regions are silver, the dark regions are Si/SiO2, where the silver unprotected by monolayer has been removed by etching. These images were obtained by Younan Xia while he was working with Professor Whitesides at Harvard University. thick) as masks. The silver masks were, in turn, fabricated by µCP with hexadecanethiol, followed by selective chemical etching in an aqueous ferri-/ferrocyanide solution (Xia et al., 1996a). The channels fabricated by this method are directly useful in making microfluidic devices and microanalytical systems (Kovacs et al., 1996). Microcontact printing with an elastomeric stamp also provides a simple and convenient approach to forming patterned microstructures on curved surfaces. It has been demonstrated that µCP with alkanethiols on evaporated thin films of gold or silver can generate micropatterns on both planar and nonplanar substrates with approximately the same fidelity and edge resolution (Jackman et al., 1995). This technique was further developed by introducing a monitoring and registration system into the experimental procedure (Rogers et al., 1997a-d). More recently, it was demonstrated that patterned microstructures of silver could also be fabricated on the inside surface of glass capillaries

Biochip technology

58

by µCP (Xia et al., 1997). In this case, electroless deposition rather than metal evaporation has to be used to prepare the substrate, and an appropriately configured rolling stamp has to be employed. These demonstrations open the door to a wide range of new types of microstructures that are expected to find applications in a number of areas. Functional devices and systems that have already been successfully fabricated using this technique include in-fiber notch filters and Bragg gratings (Rogers et al., 1997d), intravascular stents (Rogers et al., 1997b), microsprings (Rogers et al., 1997a,b), microcoils for

Figure 5. SEM images of patterned relief structures in silicon that were fabricated by anisotropic etching in aqueous KOH/i-propanol solutions, with patterned films of silver as masks (Xia et al., 1996). The silver masks were, in turn, generated by µCP with hexadecanethiol, followed by wet etching in aqueous ferri-/ferrocyanide solutions. The silver mask in (A) still remains on the surface of the silicon substrate. These images were obtained by Younan Xia while he was working with Professor Whitesides at Harvard University. high-resolution NMR spectroscopy (Rogers et al., 1997c), and microtransformers Jackman, 1997). Microcontact printing is attractive because it is straightforward, inexpensive, flexible, and convenient. Routine access to a clean room facility is not necessary, although


59

occasional use of such a facility is convenient for making masters. The elastomeric PDMS stamp and the surface chemistry for formation of SAMs can also be manipulated in a number of ways to modify the features generated using µCP (Xia and Whitesides, 1995a,b; 1997). It is, in principle, well suited for a range of microfabrication tasks usually involved in the fabrication of sensors (Wise and Najafi, 1991; Vellekoop et al., 1994), optical and electrooptical components (Kim et al., 1996a; Rajkumar and McMullin, 1995), microelectromechanical systems (MEMS) (Bryzek et al., 1994; Fuhr et al., 1994; MacDonald, 1996), microfluidic devices (Kovacs et al., 1996), and microanalytical systems (Service, 1995; Goffeau, 1997).

PATTERN TRANSFER BY SELECTIVE DEPOSITION Although the initial products of µCP are printed features of SAMs, the materials that can be patterned using µCP are not limited to SAMs. The printed SAMs can be used as templates to pattern a variety of other materials—for example, liquid prepolymers (Biebuyck and Whitesides, 1994b; Gorman et al., 1995b; Hammond and Whitesides, 1995), conducting polymers (Gorman et al., 1995a; Huang et al., 1997), inorganic salts (Palacin et al., 1996), metals Jeon et al., 1995a, 1996), ceramic materials (Jeon et al., 1995b), proteins, and cells (Mrksich et al., 1996; Singhvi et al., 1994). Most of these processes involve the use of self-assembly at two scales: the formation of patterned SAMs at the molecular scale, and the selective deposition of other types of materials on the patterned SAMs at the mesoscopic scale. Figure 6 shows SEM images of several examples of patterned microstructures that were formed with printed SAMs as templates. Figure 6A shows the SEM image of an array of stars of polyurethane that was fabricated using a combination of µCP and selective dewetting (Biebuyck and Whitesides, 1994b; Xia et al., 1996d). When a liquid prepolymer of polyurethane was placed on a surface patterned with hexadecanethiolate SAMs, it selectively dewetted the hydrophobic (CH3-terminated) regions and formed patterned microstructures on the hydrophilic (bare) regions (Xia et al., 1996c). The liquid prepolymer selectively trapped in the hydrophilic regions was then crosslinked by exposure to UV light. Figure 6B shows the SEM image of an array of microdots of CuSO4 that was prepared by selectively wetting an SAM-patterned surface of gold with an aqueous solution containing CuSO4, followed by evaporation of water (Palacin et al., 1996). Using this simple approach, it has been possible to form regular arrays of dots of CuSO4 with lateral dimensions as small as ~50 nm. Nuzzo et al. have extensively explored the use of printed SAMs on Si/SiO2 as templates to control the nucleation and growth of metals by selective chemical vapor deposition (CVD) (Jeon et al., 1995a), and ceramic materials by selective deposition from sol-gel precursors (Jeon et al., 1996). Figure 7 shows two typical examples: selective CVD of Cu and sol-gel deposition of LiNbO3. The patterned SAMs of CH3terminated alkylsiloxanes defined and directed CVD by inhibiting nucleation. The materials to be deposited only nucleated and grew on the bare regions (SiO2) that were not covered by CH3-terminated SAMs. The nucleation and growth on the polar regions give patterned microstructures. These demonstrations clearly indicate that µCP with

Biochip technology

60

SAMs, in combination with other processes and

Figure 6. Demonstration of selective wetting/dewetting with patterned SAMs as templates. (A) An optical micrograph of water condensed on an SAM-patterned surface of gold (Lopez et al., 1993). (B) An SEM image of microstructures of UV-curable polyurethane assembled using selective dewetting (Xia et al., 1996). Only the hydrophilic bare regions of gold were covered by water or the polyurethane liquid prepolymer Reprinted with the permission and through the courtesy of Professor Whitesides, Harvard University.

Figure 7. Demonstration of selective nucleation and deposition with patterned SAMs as templates. (A) An SEM image of microstructures of LiNbO3 on Si/SiO2 produced using selective deposition with a sol-gel precursor (Jeon et al., 1995). (B) An SEM image of microstructures of copper formed in silicon microtrenches using selective CVD (Jeon et al., 1995). Copper only nucleated and grew on bare regions of SiO2 underivatized by CH3-terminated siloxane SAMs.


61

Reprinted with the permission and through the courtesy of Professor Nuzzo, University of lllinois, Urbana— Champaign. materials, can be used to generate patterned microstructures of a rich variety of materials. Patterned SAMs have also been used as templates to define and control the adsorption of extracellular matrix proteins, and consequently the attachment and spreading of mammalian cells (Mrksich and Whitesides, 1996). For example, it has been possible to place cells in spatially controlled locations in an array with well-defined shapes, sizes, and distances of separation. Figure 8 shows SEM images of cells that have been selectively attached to a patterned planar (Singhvi et al., 1994) or contoured surface (Mrksich et al., 1996), respectively. It is also possible to dictate

Figure 8. Demonstration of selective attachment of cells with patterned SAMs as templates. (A) Optical micrographs of hepatocytes placed on SAM-patterned (left) and bare (right) surfaces to show the ability to control the size and shape of cells (Singhvi et al., 1994). (B) An SEM image of mammalian cells selectively attached to the plateaus of a contoured surface (Mrksich et al., 1996). The surfaces were printed with SAMs in such a way that certain regions of the surface terminated in the methyl groups, while others terminated in the oligo (ethyleneglycol) groups. The matrix proteins (fibronectin) only adsorb to the methyl-terminated regions, and cells can only attach to those regions that have been covered by the matrix proteins. Reprinted with the permission and through the courtesy of Professor Whitesides, Harvard University. the shape assumed by a cell that attaches to a surface and thus to control certain aspects of cell growth and protein secretion. This technique allows for direct examination of cell metabolism as influenced by cell morphology. This approach should find use in areas such as biotechnology that requires biochips—2D arrays of individual cells cultured at

Biochip technology

62

relatively high densities. The cells placed in specified locations can also be addressed separately and repeatedly. The results of these studies may eventually shed light on complex phenomena such as contact inhibition of cell proliferation, or lead to new analytical systems based on arrays of cells.

CONCLUSIONS Self-assembled monolayers represent the best-developed class of nonbiological systems involving molecular self-assembly. They provide an effective and versatile strategy for interface engineering and micropatterning. For example, they offer a simple and convenient method for forming well-defined nanometer-thick “coatings” on a variety of solid substrates. Because a wide range of functional groups can be incorporated into and/or at the termini of the alkyl chain, SAMs can serve as a tunable, good model system for studying interfacial phenomena related to wetting, dewetting, spreading, adhesion, lubrication, corrosion, nucleation, protein adsorption, and cell attachment. They can also be employed as a platform to fabricate sensors that involve physical, chemical, electrochemical, biochemical, or biological interactions. The ability to pattern SAMs in the lateral dimensions provides an alternative approach to microfabrication. Among the patterning techniques, µCP is the one that seems to offer the most attractive combination of convenience, simplicity, and new capability. Microcontact printing with SAMs illustrates the largely unexplored potential of nonphotolithographic techniques for microfabrication. It provides a flexible and effective route to high-quality microstructures with remarkably little of the investment required by the more familiar clean-room methods commonly used in microfabrication. In a research setting, µCP is capable of routinely generating patterned structures with submicron feature sizes. Structures smaller than 100 nm can also be fabricated by µCP, albeit such fabrication is usually less reproducible. The limitations of µCP after serious development still remain to be defined. For example, the elastomeric character of the master seems to provide both problems and opportunities in registration. The capability for large-area patterning by µCP is substantial, but remains to be developed. At the present time, µCP can only be used to print fewer than three different types of molecules on the same substrate. New printing procedures need to be developed in order to apply µCP to the fabrication of biochips, which usually contain a huge number of different probes on the surface of a single chip.


63

ACKNOWLEDGMENTS Y.X. would like to thank Professor George M. Whitesides at Harvard University for introducing him to the areas of self-assembled monolayers, interface engineering, and microfabrication. The preparation of this manuscript has been supported in part by a New Faculty Award from the Dreyfus Foundation, a subcontract from the AFOSR MURI Center (F49620–96–1–0035) at the University of Southern California, and start-up funds from the University of Washington. B.G. thanks the Center for Nanotechnology at the University of Washington for a fellowship.

REFERENCES Abbott, N.L., J.R.Folkers, and G.M.Whitesides. 1992. Manipulation of the wettability of surfaces on the 0.1–1-micrometer scale through micromachining and molecular selfassembly, Science 257:1380–1382. Aizenberg, J., A.J.Black, and G.M.Whitesides. 1999. Oriented growth of calcite controlled by self-assembled monolayers of functionalized alkanethiols supported on gold and silver. J. Am. Chem. Soc. 121:4500–4509. Allara, D.L., A.N.Parikh, and F.Rondelez. 1995. Evidence for a unique chain organization in long chain silane monolayers deposited on two widely different solid substrates. Langmuir 11:2357–2360. Alves, C.A., E.L.Smith, and M.D.Porter. 1992. Atomic scale imaging of alkanethiolate monolayers at gold surfaces with atomic force microscopy. J. Am. Chem. Soc. 114:1222–1227. Atre, S.V., B.Liedberg, and D.L.Allara. 1995. Chain length dependence of the structure and wetting properties in binary composition monolayers of OH- and CH3-terminated alkanethiolates on gold. Langmuir 11:3882–3893. Bain, C.D., and G.M.Whitesides. 1989a. Formation of monolayers by the coadsorption of thiols on gold: Variation in the length of the alkyl chain. J. Am. Chem. Soc. 111:7164– 7175. Bain, C.D., and G.M.Whitesides. 1989b. Modeling organic surfaces with self-assembled monolayers. Angew. Chem. Int. Ed. Engl. 28:506–512. Berggren, K.K., A.Bard, J.L.Wilbur, J.D.Gillaspy, A.G.Helg, J.J.McClelland, S.L.Rolston, W.D.Phillips, M.Prentiss, and G.M.Whitesides. 1995. Microlithography by using neutral metastable atoms and self-assembled monolayers. Science 269:1255– 1257. Berggren, K.K., R.Younkin, E.Cheung, M.Prentiss, A.J.Black, G.M.Whitesides, D.C.Ralph, A.T.Black, and M.Tinkham. 1997. Demonstration of a nanolithographic technique using a self-assembled monolayer resist for neutral atomic cesium. Adv. Mater. 9:52–55. Biebuyck, H.A., and G.M.Whitesides. 1994a. Autophobic pinning of drops of alkanethiols on gold. Langmuir 10:4581–4587.

Biochip technology

64

Biebuyck, H.A., and G.M.Whitesides. 1994b. Self-organization of organic liquids on patterned self-assembled monolayers of alkanethiolates on gold. Langmuir 10:2790– 2793. Biebuyck, H.A., C.D.Bain, and G.M.Whitesides. 1994. Comparison of organic monolayers on polycrystalline gold spontaneously assembled from solutions containing dialkyl disulfides or alkanethiols . Langmuir 10:1825–1831. Biebuyck, H.A., N.B.Larsen, E.Delamarche, and B.Michel. 1997. Lithography beyond light: Microcontact printing with monolayer resists. IBM J. Res. Devel 41:159–170. Bryzek, J., K.Peterson, and W.McCulley. 1994. Micromachines on the march. IEEE Spectrum 31:20–31. Buttry, D.A., and M.D.Ward. 1992. Measurement of interfacial processes at electrode surfaces with the electrochemical quartz crystal microbalance. Chem. Rev. 92:1355– 1379. Calvert, J.M. 1995. Lithographically patterned self-assembled films. In Thin Films, Vol. 20. A. Ulman, ed. Academic Press, Boston, pp. 109–141. Camillone III, N., C.E.D.Chidsey, G.-Y.Liu, T.M.Putvinski, and G.Scoles. 1991. Surface structure and thermal motion of n-alkane thiols self-assembled on Au(111) studied by low energy helium diffraction. J. Chem. Phys. 94:8493–8502. Camillone III, N., C.E.D.Chidsey, G.-Y.Liu, and G.Scoles. 1993a. Substrate dependence of the surface structure and chain packing of docosyl mercaptan self-assembled on the (111), (110), and (100) faces of single crystal gold. J. Chem. Phys. 98:4234–4245. Camillone III, N., C.E.D.Chidsey, P.Eisenberger, P.Fenter, J.Li, K.S.Liang, G.-Y.Liu, and G. Scoles. 1993b. Structural defects in self-assembled organic monolayers via combined atomic beam and X-ray diffraction. J. Chem. Phys. 99:744–747. Chan, K.C., T.Kim, J.K.Schoer, and R.M.Crooks. 1995. Polymeric self-assembled monolayers, 3: Pattern transfer by use of photolithography, electrochemical methods, and an ultrathin, self-assembled diacetylenic resists. J. Am. Chem. Soc. 117:5875– 5876. Delamarche, E., B.Michel, H.Kang, and C.Gerber. 1994. Thermal stability of selfassembled monolayers. Langmuir 10:4103–4108. Delamarche, E., B.Michel, H.A.Biebuyck, and C.Gerber. 1996. Golden interfaces: The surface of self-assembled monolayers . Adv. Mater. 8:719–729. Delamarche, E., H.Schmid, B.Michel, and H.A.Biebuyck. 1997. Stability of molded polydimethylsiloxane microstructures. Adv. Mater. 9:741–746. Delamarche, E., H.Schmid, B.Bietsch, N.B.Larsen, H.Rothuizen, B.Michel, and H.A. Biebuyck. 1998a. Transport mechanisms of alkanthiols during microcontact printing. J. Phys. Chem.-B 102:3324–3334. Delamarche, E., H.Schmid, B.Michel, H.R.Bosshard, and H.A.Biebuyck. 1998b. Printing patterns of proteins. Langmuir 14:2225–2229. Deng, T., L.B.Goetting, J.M.Hu, and G.M.Whitesides. 1999. Microfabrication of halfwave rectifier circuits using soft lithography. Sens. Actuators A-Physical 75:60–64. Dubois, L.H., and R.G.Nuzzo. 1992. Synthesis, structure, and properties of model organic surfaces . Annu. Rev. Phys. Chem. 43:437–463. Fenter, P., A.Eberhardt, and P.Eisenberger. 1994. Self-assembly of n-alkyl thiols as disulfides on Au(111). Science 266:1216–1218.


65

Fontaine, P., D.Gogeuenheim, D.Deresnes, D.Vaillaume, M.Caret, and E.Rondelez. 1993. Octadecyltrichlorosilane monolayers as ultrathin gate insulating films in metalinsulator-semiconductor devices. Appl. Phys. Lett. 62:2256–2258. Fuhr, G., T.Muller, T.Shnelle, R.Hagedorn, A.Boigt, S.Giedler, W.A.Arnold, U.Zimmermann, B.Wagner, and A.Heuberger. 1994. Radio-frequency microtools for particle and live cell manipulation. Naturwissenschaften 81:528–535. Gau, H., S.Herminghaus, P.Lenz, and R.Lipowsky. 1999. Liquid morphologies on structured surfaces: From microchannels to microchips. Science 283:46–49. Gillen, G., S.Wight, J.Bennett, and M.J.Tarlov. 1994. Patterning of self-assembled alkanethiol monolayers on silver by microfocus ion and electron beam bombardment. Appl. Phys. Lett. 65:534–536. Goetting, L.B., T.Deng, and G.M.Whitesides. 1999. Microcontact printing of alkanephosphonic acids on aluminum: Pattern transfer by wet chemical etching. Langmuir 15:1182–1191. Goffeau, A. 1997. DNA technology: Molecular fish on chips. Nature 385:202–203. Gorman, C.B., H.A.Biebuyck, and G.M.Whitesides. 1995a. Fabrication of patterned electrically conducting polypyrrole using a self-assembled monolayer: A route to allorganic circuits. Chem. Mater. 7:526–529. Gorman, C.B., H.A.Biebuyck, and G.M.Whitesides. 1995b. Use of a patterned selfassembled monolayer to control the formation of a liquid resist pattern on a gold surface. Chem. Mater. 7:252–254. Hammond, P., and G.M.Whitesides. 1995. Formation of polymer microstructures by selective deposition on polyion multilayers using patterned self-assembled monolayers as a template. Macromolecules 28:7569–7571. Hickman, J.J., D.Ofer, P.E.Laibinis, G.M.Whitesides, and M.S.Wrighton. 1991. Molecular self-assembly of two-terminal, voltammetric microsensors with internal references. Science 252:688–691. Hidber, P.C., P.F.Neeley, W.Heldbig, and G.M.Whitesides. 1996a. New strategy for controlling the size and shape of metallic features formed by electroless deposition of copper: Microcontact printing of catalysts on oriented polymers, followed by thermal shrinkage. Langmuir 12:5209–5215. Hidber, P.C., W.Helbig, E.Kim, and G.M.Whitesides. 1996b. Microcontact printing of palladium colloids: Micron-scale patterning by electroless deposition of copper. Langmuir 12:1375–1380. Huang, J., and J.C.Hemminger. 1993. Photooxidation of thiols in self-assembled monolayers on gold. J. Am. Chem. Soc. 115:3342–3343. Huang, J., D.A.Dahlgren, and J.C.Hemminger. 1994. Photopatterning of self-assembled alkanethiolate monolayers on gold: a simple monolayer photoresist utilizing aqueous chemistry. Langmuir 10:626–628. Huang, Z., P.-C.Wang, A.G.MacDiarmid, Y.Xia, and G.M.Whitesides. 1997. Selective deposition of conducting polymers on hydroxyl-terminated surfaces with printed monolayers of alkylsiloxanes as templates. Langmuir 13:6480–6484. Isaacs, L., D.Chin, N.Bowen, Y.Xia, and G.M.Whitesides. 1999. Self-assembling systems on scales from nanometers to millimeters: design and discovery. In Supermolecular Materials and Technology. D.N.Reinhoudt, ed. John Wiley & Sons,

Biochip technology

66

New York, pp. 1–46. Ishida, T., S.Yamamoto, W.Mizutani, M.Motomatsu, H.Tokumoto, H.Hokari, H.Azehara, and M.Fujihara. 1997. Evidence for cleavage of disulfides in the self-assembled monolayer on Au (111). Langmuir 13:3261–3265. Jackman, R.J., J.L.Wilbur, and G.M.Whitesides. 1995. Fabrication of submicrometer features on curved substrates by microcontact printing. Science 269:664–666. Jackman, R.J., J.A.Rogers, and G.M.Whitesides. 1997. Fabrication and characterization of a concentric cylindrical microtransformer. IEEE Trans. Mag. 33:2501–2503. Jeon, N.L., R.G.Nuzzo, Y.Xia, M.Mrksich, and G.M.Whitesides. 1995a. Patterned selfassembled monolayers formed by microcontact printing direct selective metallization by chemical vapor deposition on planar and non-planar substrates. Langmuir 11:3024– 3026. Jeon, N.L., P.G.Clem, R.G.Nuzzo, and A.A.Payne. 1995b. Patterning of dielectric oxide thin layers by microcontact printing of self-assembled monolayers. J. Mater. Res. 10:2996–2999. Jeon, N.L., P.G.Clem, A.A.Payne, and R.G.Nuzzo. 1996. A monolayer-based lift-off process for patterning chemical vapor deposition copper thin films. Langmuir 12:5350–5355. Jeon, N.L., K. Finnie, K. Branshaw, and R.G. Nuzzo. 1997. Structure and stability of patterned self-assembled films of octadecyltrichlorosilane formed by contact printing. Langmuir 13:3382–3391. Jordon, C.E, and R.M.Corn. 1997. Surface plasmon resonance imaging measurements of electrostatic biopolymer adsorption onto chemically modified gold surfaces. Anal. Chem. 69:1449–1456. Karpovich, D.S., and G.J.Blanchard. 1994. Direct measurement of the adsorption kinetics of alkanethiolate self-assembled monolayers on a microcrystalline gold surface. Langmuir 10:3315–3322. Kim, E., A.Kumar, and G.M.Whitesides. 1995. Combining patterned self-assembled monolayers of alkanethiolates on gold with anisotropic etching of silicon to generate controlled surface morphologies. J. Electrochem. Soc. 142:628–633. Kim, E., G.M.Whitesides, L.K.Lee, S.P.Smith, and M.Prentiss. 1996a. Fabrication of arrays of channel waveguides by self-assembly using patterned organic monolayers as templates. Adv. Mater. 8:139–142. Kim, E., G.M.Whitesides, M.B.Freiler, M.Levy, J.L.Lin, and R.M.Osgood Jr. 1996b. Fabrication of micrometer-scale structures on GaAs and GaAs/AlGaAs quantum well material using microcontact printing. Nanotechnology 7:266–269. Kovacs, G.T.A., K.Petersen, and M.Albin. 1996. Silicon micromachining: Sensors to systems. Anal Chem. 68:407A-112A. Kumar, A., and G.M.Whitesides. 1993. Features of gold having micrometer to centimeter dimensions can be formed through a combination of stamping with an elastomeric stamp and an alkanethiol “ink” followed by chemical etching. Appl Phys. Lett. 63:2002–2004. Kumar, A., H.A.Biebuyck, and G.M.Whitesides. 1994. Patterning self-assembled monolayers: Applications in material science. Langmuir 10:1498–1511. Kumar, A., N.L.Abbott, E.Kim, H.A.Biebuyck, and G.M.Whitesides. 1995. Patterned


67

self-assembled monolayers and meso-scale phenomena. Acc. Chem. Res. 28:219–226. Lahiri, J., E.Ostuni, and G.M.Whitesides. 1999. Patterning ligands on reactive SAMs by microcontact printing. Langmuir 15:2055–2060. Larsen, N.B., H.Biebuyck, E.Delamarche, and B.Michel. 1997. Order in microcontact printed self-assembled monolayers. J. Am. Chem. Soc. 119:3017–3026. Lehn, J.-M. 1990. Perspectives in supermolecular chemistry: From molecular recognition towards molecular information processing and self-organization. Angew. Chem. 102: 1304–1319. Lercel, M., R.C.Tiberio, P.F.Chapman, H.G.Craighead, C.W.Sheen, A.N.Parikh, and D.L. Allara. 1993. Self-assembled monolayer electron beam resist on GaAs and SiO2. J. Vac. Sci. Technol. B 11:2823–2828. Li, Y., J.Huang, R.T.McIver Jr., and J.C.Hemminger. 1992. Characterization of thiol selfassembled films by laser deposition Fourier transform mass spectrometry. J. Am. Chem. Soc. 11:2428–2432. Lopez, G.P., H.A.Biebuyck, C.D.Frisbie, and G.M.Whitesides. 1993a. Imaging of features on surfaces by condensation figures. Science 260:647–649. Lopez, G.P., H.A.Biebuyck, R.Horter, A.Kumar, and G.M.Whitesides. 1993b. Fabrication and imaging of two-dimensional patterns of proteins adsorbed on selfassembled monolayers by scanning electron microscopy. J. Am. Chem. Soc. 115:10774–10781. MacDonald, N.C. 1996. SCREAM microelectromechanical systems. Microelectron. Eng. 32: 49–73. Madou, M. 1997. Fundamentals of Microfabrication. CRC Press, Boca Raton, FL. McCarley, T.D., and R.L.McCarley. 1997. Toward the analysis of electrochemically modified self-assembled monolayers: electrospray ionization mass spectrometry of organothiolates. Anal Chem. 69:130–136. McDermott, C.A., M.T.McDermott, J.-B.Green, and M.D.Porter. 1995. Structural origins of the surface depressions at alkane thiolate monolayers on Au(111): A scanning tunneling and atomic force microscopic investigation. J. Phys. Chem. 99:13257– 13267'. Moffat, T.P., and H.Yang. 1995. Patterned metal electodeposition using an alkanethiolate mask. J. Electrochem. Soc. 142:L220-L222. Mrksich, M., and G.M.Whitesides. 1995. Patterning self-assembled monolayers using microcontact printing: a new technology for biosensors? TIBTECH 13:228–235. Mrksich, M., and G.M.Whitesides. 1996. Using self-assembled monolayers to understand the interactions of man-made surfaces with proteins and cells. Annu. Rev. Biophys. Biomol. Struct. 25:55–78. Mrksich, M., C.S.Chen, Y.Xia, L.E.Dike, D.E.Ingber, and G.M.Whitesides. 1996. Controlling cell attachment on contoured surfaces with self-assembled monolayers of alkanethiolates on gold. Proc. Natl. Acad. Sci. U.S.A. 93:10775–10778. Olbris, D.J., A.Ulman, and Y.J.Shnidman. 1995. Interplay of wetting and adsorption at mixed self-assembled monolayers. J. Chem. Phys. 102:6865–6873. Overney, R.M., E.Meyer, J.Frommer, D.Brodbeck, R.Luthi, H.-J.Guntherodt, L.Howald, M. Fujihara, H.Takano, and Y.Gotoh. 1992. Friction measurements on phase-separated thin films with a modified atomic force microscope. Nature 359:133–135.

Biochip technology

68

Palacin, S., P.C.Hidber, J.-R.Bourgoin, C.Miramond, C.Fermon, and G.M.Whitesides. 1996. Patterning with magnetic materials at the micron scale. Chem. Mater. 8:1316– 1325. Poirier, G.E. 1997. Characterization of organosulfur molecular monolayers on Au (111) using scanning tunneling microscopy. Chem. Rev. 97:1117–1127. Poirier, G.E., and E.D.Pylant. 1996. The self-assembly mechanism of alkanethiols on Au (111). Science 272:1145–1148. Prime, K.L., and G.M.Whitesides. 1993. Adsorption of proteins onto surface containing end-attached oligo(ethylene oxide): A model system using self-assembled monolayers. J. Am. Chem. Soc. 115:10714–10721. Rajkumar, N., and J.N.McMullin. 1995. V-groove gratings on silicon for infrared beam splitting. Appl Opt. 34:2556–2559. Roberts, C., C.S.Chen, M.Mrksich, V.Martichonok, D.E.Ingber, and G.M.Whitesides. 1998. Using mixed self-assembled monolayers presenting RGD and (EG)(3)OH groups to characterize long-term attachment of bovine capillary endothelial cells to surfaces. J. Am. Chem. Soc. 120:6548–6555. Rogers, J.A., R.J.Jackman, and G.M.Whitesides. 1997a. Constructing single—and multiple—helical microcoils and characterizing their performance as components of microinductors and microelectromagnets. J. Microelec. Syst. (Sept), pp. 184–192. Rogers, J.A., R.J.Jackman, and G.M.Whitesides. 1997b. Microcontact printing and electroplating on curved substrates: production of free-standing three-dimensional metallic microstructures. Adv. Mater. 9:475–477. Rogers, J.A., R.J.Jackman, G.M.Whitesides, D.L.Olson, and J.V.Sweedler. 1997c. Using microcontact printing to fabricate microcoils on capillaries for high resolution proton nuclear magnetic resonance on nanoliter volumes. Appl. Phys. Lett. 70:2464–2466. Rogers, J.A., R.J.Jackman, G.M.Whitesides, J.L.Wagener, and A.M.Vengsarkar. 1997d. Using microcontact printing to generate amplitude photomasks on the surfaces of optical fibers: A method for producing in-fiber gratings. Appl. Phys. Lett. 70:7–9. Ross, C.B., L.Sun, and R.M.Crooks. 1993. Scanning probe lithography, 1: Scanning tunneling microscope induced lithography of self-assembled n-alkanethiol monolayer resists. Langmuir 9:632–636. St. John, P.M. and H.G.Craighead. 1996. Microcontact printing and pattern transfer using trichlorosilanes on oxide substrates. Appl Phys. Lett. 68:1022–1024. Schonenberger, C., J.A.M.Sondag-Huethorst, J.Jorritsma, and L.G.J.Fokkink. 1994. What are the “holes” in self-assembled monolayers of alkanethiols on gold? Langmuir 10:611–614. Schönherr, H., and H.Ringsdorf. 1996. Self- assembled monolayers of symmetrical and mixed alkyl fluoroalkyl disulfides on gold, 1: Synthesis of disulfides and investigation of monolayer properties. Langmuir 12:3891–3897. Service, R.E. 1995. The incredible shrinking laboratory. Science 268:26–27. Singhvi, R., A.Kumar, G.P.Lopez, G.P.Stephanopoulos, D.I.C.Wang, G.M.Whitesides, and D.E.Ingber. 1994. Engineering cell shape and function. Science 264:696–698. Sondag-Huethorst, J.A.M., H.R.J.van Helleputte, and L.G. Fokkink. 1994. Generation of electrochemically deposited metal patterns by means of electron beam (nano) lithography of self-assembled monolayer resists. Appl. Phys. Lett. 64:285–287.


69

Strong L., and G.M.Whitesides. 1989. Structures of self-assembled monolayer films of organosulfur compounds adsorbed on gold single crystals: Electron diffraction studies. Langmuir 4:546–558. Sun, L., and R.M.Crooks. 1993. Indirect visualization of defect structures contained within self-assembled organomercaptan monolayers: Combined use of electrochemistry and scanning tunneling microscopy. Langmuir 9:1951–1954. Takami, T., E.Delamarche, B.Michel, C.Gerber, H.Wolf, and H.Ringsdorf. 1995. Recognition of individual tail groups in self-assembled monolayers. Langmuir 11:3876–3881. Tam-Chang, S.-W., H.A.Biebuyck, G.M.Whitesides, N.Jeon, and R.G.Nuzzo. 1995. Selfassembled monolayers on gold generated from alkanethiols with the structure RNHCOCH2SH. Langmuir 11:4371–4382. Tarlov, M.J., D.R.F.Burgess Jr., and G.Gillen. 1993. UV photopatterning of alkanethiolate monolayers assembled on gold and silver. J. Am. Chem. Soc. 115:5305– 5306. Ulman, A. 1991. Introduction to thin organic films: From Langmuir-Blodgett to SelfAssembly. Academic Press, Boston. Vellekoop, M.J., G.W.Lubking, P.M.Sarro, and A.Venema. 1994. Integrated-circuitcompatible design and technoology of acoustic-wave-based microsensors. Sens. Actuators A 44:249–263. Vigmond, S.J., M.Iwakura, F.Mizutani, and T.Katsura. 1994. Site-specific immobilization of molecularly engineered dihydrofolate reductase to gold surfaces. Langmuir 10:2860–2862. Wang, D., S.G.Thomas, K.L.Wang, Y.Xia, and G.M.Whitesides. 1997. Nanometer scale patterning and pattern transfer on amorphous Si, crystalline Si, and SiO2 surfaces using self-assembled monolayers. Appl. Phys. Lett. 70:1593–1595. Wasserman S.R., H.A.Biebuyck, and G.M.Whitesides. 1989. Monolayers of 11trichlorosilylundecyl thioacetate: A system that promotes adhesion between silicon dioxide and evaporated gold. J. Mater. Res. 4:886–891. Whidden, T.K., D.K.Ferry, M.N.Kozicki, E.Kim, A.Kumar, J.L.Wilbur, and G.M. Whitesides. 1996. Pattern transfer to silicon by microcontact printing and RIE. Nanotechnology 7:447–451. Whitesides, G.M. 1995. Self-assemblying materials. Scient. Am. 273:146–149. Whitesides, G.M., and P.E.Laibinis. 1990. Wet chemical approaches to the characterization of organic surface: Self-assembled monolayers, wetting, and the physical-organic chemistry of the solid-liquid interface. Langmuir 6:87–96. Whitesides, G.M., J.P.Mathias, and C.T.Seto. 1991. Molecular self-assembly and nanochemistry: A chemical strategy for the synthesis of nanostructures. Science 254:1312–1319. Wilbur, J.L., A.Kumar, E.Kim, and G.M.Whitesides. 1994. Microfabrication by microcontact printing of self-assembled monolayers. Adv. Mater. 6:600–604. Wilbur, J.L., H.A.Biebuyck, J.C.MacDonald, and G.M.Whitesides. 1995a. Scanning force microscopies can image patterned self-assembled monolayers. Langmuir 11:825– 831. Wilbur, J.L., E.Kim, Y.Xia, and G.M.Whitesides. 1995b. Lithographic molding: A

Biochip technology

70

convenient route to structures with sub-micrometer dimensions. Adv. Mater. 7:649–652 Wirth, M.J., R.W.P.Fairbank, and H.O.Fatunmbi. 1997. Mixed self-assembled monolayers in chemical separations. Science 275:44–47. Wise, K.D., and K.Najafi, 1991. Microfabrication techniques for integrated sensors and microsystems. Science 254:1335–1342. Xia, Y., and G.M.Whitesides. 1995a. Use of controlled reactive spreading of liquid alkanethiol on the surface of gold to modify the size of features produced by microcontact printing. J. Am. Chem. Soc. 117:3274–3275. Xia, Y., and G.M.Whitesides. 1995b. Reduction in the size of features of patterned SAMs generated by microcontact printing with mechanical compression of the stamp. Adv. Mater. 7:471–473. Xia, Y., and G.M.Whitesides. 1997. Extending microcontact printing as a microlithographic technique. Langmuir 13:2059–2067. Xia, Y., and G.M.Whitesides. 1998. Soft lithography. Angew. Chem. Int. Ed. Engl 37:551–575. Xia, Y., X.-M.Zhao, E.Kim, and G.M.Whitesides. 1995a. A selective etching solution for use with patterned self-assembled monolayers of alkanethiolates on gold. Chem. Mater. 12: 2332–2337. Xia, Y., M.Mrksich, E.Kim, and G.M.Whitesides. 1995b. Microcontact printing of octadecylsiloxane on the surface of silicon dioxide and its application in microfabrication. J. Am. Chem. Soc. 117:9576–9577. Xia, Y., E.Kim, and G.M.Whitesides. 1996a. Microcontact printing of alkanethiols on silver and its application in microfabrication. J. Electrochem. Soc. 143:1070–1079. Xia, Y., D.Qin, and G.M.Whitesides. 1996b. Microcontact printing with a cylindrical rolling stamp: A practical step toward automatic manufacturing of patterns with submicrometer-sized features. Adv. Mater. 8:1015–1017. Xia, Y., E.Kim, M.Mrksich, and G.M.Whitesides. 1996c. Microcontact printing of alkanethiols on copper and its application in microfabrication. Chem. Mater. 8:601– 603. Xia, Y., J.Tien, D.Qin, and G.M.Whitesides. 1996d. Non-photolithographic methods for fabrication of elastomeric stamps for use in microcontact printing. Langmuir 12:4033– 4038. Xia, Y., X.-M.Zhao, and G.M.Whitesides. 1996e. Pattern transfer: Self-assembled monolayers as ultrathin resists. Microelectron. Eng. 32:255–268. Xia, Y., N.Venkateswaren, D.Qin, J.Tien, and G.M.Whitesides. 1997. The use of electroless silver as the substrate in microcontact printing of alkanethiols, and its application in microfabrication. Langmuir 14:363–371. Xia, Y., J.A.Rogers, K.Paul, and G.M.Whitesides. 1999. Unconventional methods for fabricating and patterning nanostructures. Chem. Rev. 99:1823–1848. Xu, J., and H.-L.Li. 1995. The chemistry of self-assembled long-chain alkanethiol monolayers on gold. J. Colloid Interface Sci. 176:138–149. Yan, L., X.M.Zhao, and G.M.Whitesides. 1998. Patterning of performed, reactive SAM using microcontact printing. Langmuir 120:6179–6180. Yan, L., W.T.S.Huck, X.M.Zhao, and G.M.Whitesides. 1999. Patterning thin films of poly(ethylene imine) on a reactive SAM using microcontact printing. Langmuir


71

15:1208–1214. Yang, X.M., A.A.Tryk, K.Hasimoto, and A.Fujishima. 1996. Surface enhanced Raman imaging of a patterned self-assembled monolayer formed by microcontact printing on a silver film. Appl. Phys. Lett. 69:4020–4022. Youkin, R., K.K.Berggren, K.S., Johnson, M.Prentiss, D.C.Ralph, and G.M.Whitesides. 1997. Demonstration of a nanolithographic technique using a self-assembled monolayer resist for neutral atomic cesium. Appl. Phys. Lett. 71:1261–1263. Zhao, X.-M., J.L.Wilbur, and G.M.Whitesides. 1996. Using two-stage chemical amplification to determine the density of defects in self-assembled monolayers of alkanethiolates on gold. Langmuir 12:3257–3264. Zhuk, A.V., A.G.Evans, J.W.Hutchinson, and G.M.Whitesides. 1998. The adhesion energy between polymer thin film and self-assembled monolayers. J. Mater. Res. 13:3555–3564.

4 Fabrication of Polymer Microfluidic Devices Holger Becker

INTRODUCTION What microelectronics has done for information technology is being repeated in the life sciences. The process of miniaturization and the application of microsystem technologies (MST or, in the American community MEMS, for Micro-Electromechanical Systems, or Micromachine in Japan) has had a significant impact in the life sciences. Research on the human genome, the drug discovery process in the pharmaceutical industry clinical diagnostics, and analytical chemistry are experiencing rapid change due to new tools produced through miniaturization (Manz and Becker, 1998). The concept behind this development originated in analytical chemistry and is called the miniaturized total chemical analysis system (µ-TAS). Already in the 1970s, in a remarkable effort, Stephen Terry miniaturized a gas chromatography system and integrated the complete system on a silicon wafer (Terry et al., 1979). This work, however, went unnoticed for more than a decade, until 1990 when the concept of µ-TAS was advanced by Andreas Manz and his team at Ciba-Geigy (Manz et al., 1990). This paper triggered an avalanche of developments and discoveries, which led to a truly exponential growth of this field, initially in academic research, but since the mid-1990s also on a commercial basis (Latta, 1997; Walsh, 1999). In the early years, many devices were fabricated using the techniques developed in microelectronics, for example, starting with a silicon wafer, using standard photolithography and subsequent wet etching as the method for producing microchannels on a planar substrate (Manz et al., 1991). As electrokinetic pumping was established as the method of choice for transporting liquid samples in these microchannels, the development focused on various types of glass (Harrison et al., 1993; Fan and Harrison, 1994; Effenhauser et al., 1993; Jacobson et al., 1994a,b) or quartz (Jacobson and Ramsey, 1995; Jacobson et al., 1995; Becker et al., 1998b) as a substrate, as the conductivity of silicon proved problematic for the application of the high voltage needed for electroosmotic flow (for recent reviews, see Becker and Manz, 1998; Kopp et al., 1997). The fabrication methods, however, remained essentially the same: isotropic wet etching using hydrofluoric acid (HF) or KOH as an etchant. For the ongoing commercialization of this technology, these fabrication processes have several disadvantages. As many microfluidic devices have a comparatively large footprint (typically several cm2; up to several 100 cm2), to achieve either long separation channel length or a high integration in their functions, the cost of the substrate material plays an important role in high-volume production. But while in microelectronics the size of a microchip has become smaller and smaller due to progress in circuit engineering and lithographical

Fabrication of polymer microfluidic devices

73

techniques, the area of a microfluidic chip often cannot simply be decreased by progress in fabrication methods. Instead, size is determined by functionality, and a reduction in size may result in a loss of performance (i.e., by making a curved channel instead of a straight one) or loss of compatibility with existing systems (i.e., the footprint of a microwell plate or pipetting robot). Conventional fabrication methods involve a large number of process steps (cleaning, resist coating, photolithography, development, wet etching) as well as harmful wet chemistry reagents, such as KOH or HF. Each device has to go through this fabrication process serially, which, despite the fact that these process steps are well known from the microelectronics industry and can be fully automated, increases the risk of a low yield, lengthens fabrication time, and therefore raises costs. In addition, process costs are significant due to the reagents involved as well as their waste disposal. For etching methods, the available geometry for microchannels is limited due to the isotropicity of the etching process. These etching methods allow only shallow, mainly semicircular channel cross-sections in glass substrates. For many applications, however, a range of channel cross-sections are desirable—for example, high aspect ratio square channels, channels with a defined but arbitrary wall angle or channels with different heights. These are not achievable with standard microfabrication methods in glass or quartz substrates. Advanced silicon dry etch processes for silicon can produce a larger variety of geometries, particularly vertical trenches with a high aspect ratio; however, the process and equipment costs are significant (Hansen et al., 1996); in addition, the surface chemistry of silicon substrates poses a problem, as biomolecules (oligonucleotides, DNA, proteins, etc.) tend to bind to the substrate surface. This can be prevented with a surface coating (e.g., silanization), which represents an additional process step. As microfabrication methods and materials developed, additional technologies that had not existed in the microelectronics world were introduced. In particular, the introduction of polymer microfabrication technologies has opened new possibilities for microfluidic applications. Polymers as substrate materials can avoid many of the above-mentioned fabrication challenges and lend themselves to mass fabrication of microfluidic devices. They have a wide range of material properties, and are normally low-cost. The development of suitable polymer microfabrication methods over recent years has attracted an enormous interest, particularly as this provides a route to high-volume production of disposable microfluidic devices, which allows for successful commercialization of the µ-TAS concept. This review is limited to methods for the fabrication of analytical microfluidic devices. Other aspects of microfluidics—such as micropumps, valves, mixers, chemically modified surfaces (DNA-array technologies), as well as other polymer fabrication technologies that so far have not been applied to microfluidics—are not considered.

POLYMER MATERIALS Polymer materials have historic roots in the microelectronics industry and thus entered the microfabrication arena rather late. However, they have proved to be the most promising materials for microfluidic systems since they are suitable for such mass

Biochip technology

74

replication technologies as injection molding and hot embossing as well as for rapid prototyping methods, for example, casting or laser micromachining. Polymer materials offer a wide choice of material properties: mechanical properties (e.g., stiffness, tensile strength), optical characteristics (absorption, refractive index, fluorescence), temperature stability, and resistance against such chemicals as acids, alkalis, or organic solutions, and can be biodegradable. Therefore, the material can be matched to a specific application (or microfabrication method), and this can lead to significant performance enhancement. Polymers are macromolecular substances with a relative molecular weight between 10,000 and 100,000 Da, with more than 1000 monomeric units. The polymerization process is started by an initiator or by a change in a physical parameter (light, pressure, temperature). Polymers exhibit a range of bulk properties, and can be amorphous or microcrystalline (when the length of polymer chains is larger than the size of crystallites). Generally, a polymeric material does not have an exactly defined melting temperature due to variation in the length of the polymer chains. Instead, there exists a melt interval where the viscosity changes markedly and the material turns into a highly viscous mass. The decomposition temperature is another characteristic of a polymer above which thermal cracking of the material starts and the material loses integrity. Many of these high-molecular-weight substances solidify after cooling under the socalled glass transition temperature (Tg), and the solid phase that results is hard and brittle. For fabrication processes, this is a particularly important parameter. If the temperature is increased above Tg, the material becomes plastic and viscous and can be molded. It is important for the molding process and the stability of the resulting structures to cool the material below Tg before demolding. Otherwise, the geometric stability of the molded component can suffer due to relaxation during demolding and the resulting entropy elasticity. Softeners (plasticizers) can be used to lower glass transition temperatures, but in the presence of the softeners the elasticity, impact strength, and expansion of the polymer increase and hardness decreases. Polymers can be classified into the following three categories according to their molding behavior: Thermoplastic polymers. These consist of unlinked or weakly linked polymer chains. At a temperature above the glass transition temperature, these materials become plastic and can be molded into specific shapes, which they will retain after cooling below Tg. They form the most important group of polymers used in microfabrication. Duroplastic polymers. In these materials, the polymer chains are strongly crosslinked, so that a molecular movement resulting in a change in shape is not easily possible. Therefore, these materials have to be cast into their final shape. They are harder and more brittle than thermoplastic materials and soften very little before reaching the decomposition temperature. Elastomeric polymers. These materials have very weakly crosslinked polymer chains. If an external force is applied, the molecular chains can be stretched, but they relax back to their original state (higher entropy) once the external force is removed. Elastomers do not melt before reaching decomposition temperature. A wide variety of polymer materials have been used for microfabrication processes. Standard polymer materials include polyamide (PA), polybutyleneterephthalate (PBT),

Fabrication of polymer microfluidic devices

75

polycarbonate (PC), polyethylene (PE), polymethylmethacrylate (PMMA), polyoxymethylene (POM), polypropylene (PP), polyphenylenether (PPE), polystyrene (PS), and polysulphone (PSU), engineering plastics like liquid crystal polymer (LCP), polyetheretherketone (PEEK), and polyetherimide (PEI), as well as biodegradable materials like polylactide. Up to now, PMMA and PC are the most popular polymer materials for microfabrication via hot embossing and injection molding. A cycloolefincopolymer (COC) is currently under test in the hot embossing process. This new material is extremely promising for applications in chemical engineering and molecular biotechnology since it has high chemical stability and is optically transparent (Ehrfeld et al., 1999). Tables 1 and 2 summarize the physical and chemical properties of the most commonly used thermoplastic polymers for micromolding, and Table 3 gives an overview of the typical materials used for micromolding and the details of their behavior in the injection molding process. However, the specific characteristics depend on product, polymer type, and the respective conditions in the injection molding process. For the hot embossing process, filling and separation of the mold are less critical than for injection molding. Up to now, there has been no special development of plastics for the micromolding process because mass production even at the scale of a million pieces accounts for only 1 ton of polymer material, and this is much too small an amount to warrant special manufacturing efforts. Therefore, the existing polymers are used for micromolding processes, but the current applications show that there is already a variety of materials on the market suited for microfabrication. Only thermoplastic and elastomeric materials have so far been used for fabrication of microfluidic devices. Photoresists are another class of polymers used in microsystem technology and for microfluidic systems. Irradiation—with electrons, ions, X-rays, UV, or visible light— leads to a photochemical reaction of the resist material, which is coated onto a carrier substrate (typically silicon or another polymer). In the case of the so-called positive resists, the solubility of the irradiated areas increases, while in negative resists it decreases. The irradiated or non-irradiated areas will be removed by a developer. For deep X-ray lithography, the standard resist material is PMMA, which acts as a positive resist. This can be used directly to form microchannels (see the section on “Deep X-Ray Lithography”). Polylactide or copolymers of lactide and glycolide can be used as resist materials for X-ray lithography (Ehrfeld et al., 1999). In the longer-wavelength range, SU-8 is a resist material developed by IBM for UV lithography that allows fabrication of structures with heights greater than 1000 µm (see the section on “Optical Lithography in Deep Resist (SU-8)”)

POLYMER REPLICATION TECHNOLOGIES Low-cost manufacturing processes are the most important motivation for using polymer microfabrication technologies in microfluidics. In fact, they hold the key

Biochip technology

76

Table 1 Basic Physical Properties of Molding Polymer Materialsa Thermal Linear DensityGlass conductivityexpansion Thermoplastic Permanent (≥103 temperaturetemperatureλ (W m–1 coefficien materials for kg/mb) Tg (°C) (10–6 K–1 micromolding of use (°C) K–1) Polyamide 6 (PA6) 1.13 60 80–100 0.29 80 Polyamide 66 (PA66) 1.14 70 80–120 0.23 80 Polycarbonate (PC) 1.2 150 115–130 0.21 65 Poluyoxymethylene 1.41– –60 90–110 0.23–0.31 90–110 (POM) 1.42 NA NA Cycloolefin copolymer 1.01e 138e 60e d (COC) Polymethylmethacrylate 1.18– 106 82–98 0.186 70–90 (PMMA) 1.19 0.349 140 Polyethylene low 10 kilobases) and large protein molecules (>150 kilodaltons). In addition, certain solutes interact with glass or silica to slowly form a coating on the inside or outside of the dispensing tube or capillary. Increased wetting of the dispensing tip orifice that results from solute buildup often hampers correct liquid dispensing by increasing the amount of energy necessary to eject a drop. If the energy provided by the dispensing tip is insufficient to eject and separate a drop from the orifice, then the dispensing tip may exhibit the “frog-tongue” effect, where a drop comes out and then retracts back into the orifice. Cleaning the dispensing tip end is critical for accurate drop formation with noncontact liquid dispensing robots. In addition, one must control for environmental conditions such as air drafts and static electricity to achieve high-quality performance of non-contact liquid dispensing technology. Clogging of the dispensing tip with dust, particulate matter, or aggregates of solute is another problem occasionally encountered with non-contact liquid dispensing robots. This problem can be avoided by careful attention to the environment and solutions used with non-contact liquid dispensing robots. Normal operating procedures should include filtering or centrifugation of all solutions that are to be processed by the robot, periodic cleaning of the fluid lines leading to the dispensing head, and housing the liquid dispensing robot in a clean room or hood. One should also consider the possibility of crystal or aggregate formation in certain liquids as the robots are processing them. The development of crystals and aggregates can be triggered as a result of local evaporation at

Non-contact microarraying technologies

119

the dispensing tip during long pauses in the dispensing run. Minimizing the time between washes in the liquid dispensing protocol and reducing the concentration of solutions are two ways to avoid this problem. Substrate Characteristics One critical parameter in producing high-quality arrays is controlling substrate effects. The chemical and physical properties of the substrate influence the behavior of the droplet once it reaches the substrate surface. The surface tension of the sample combined with the surface characteristics of the substrate decide the size and morphology of the final spot. A drop will flow along the substrate surface on impact and grow larger in diameter, thereby reducing the volume of solution over each point in the surface area of the spot. Drying effects, where the process of evaporation deposits solute on the substrate surface, often cause nonuniform deposition of solute within a spot on a solid substrate such as a glass slide. Hydrophobic or hydrophilic coatings are typically used on nonporous substrates to limit both fluid flow on the substrate surface and drying effects following drop impact. These coatings occasionally introduce problems at subsequent points in the experiment by limiting the fluid accessibility of molecules that are immobilized on the surface. Porous substrates overcome many of the limitations of surface coatings by channeling fluid flow to the Z-plane of the substrate interior from the X-Y plane of the substrate surface. Porous substrates with pore sizes substantially smaller than the drop diameter result in only marginal spreading of the drop as it enters the pores. This enables higher volumes of solution over the surface area of the spot than that achievable with flat, nonporous surfaces. In addition, surface tension plays less of a role in distribution of the material during evaporation from a porous substrate than in the case of nonporous substrates, thereby overcoming drying effects. For example, a 325-picoliter drop with a free-flight diameter of 85 microns often results in a spot diameter on a flat, nonporous surface of about 170 microns. This limits the maximum density of spots on the substrate to approximately 1600 spots per square centimeter. This same drop dispensed to an Anapore membrane, a porous substrate with 250-nanometer channels, results in a spot size of 105 microns and a corresponding maximum density of 4000 spots per square centimeter. Substrates such as polyacrylamide gel pads from Argonne National Laboratories (Argonne, IL), Anapore™ membrane from Whatman (Maidstone, UK), or the Flow-Through Chip™ from Gene Logic (Gaithersburg, MD) are examples of porous substrates suitable for microarray research (Guschin et al., 1997; Yershov et al., 1996). Porous substrates offer a number of advantages to experiments using microarrays. The detection and evaluation of biological molecules is enhanced by uniform distribution of material across the spot. In addition, some porous substrates such as polyacrylamide gel pads enhance the three-dimensional accessibility of immobilized molecules (Guschin et al., 1997; Yershov et al., 1996). Porous substrates often allow for greater sample loading, thereby increasing the concentration of molecules immobilized on the substrate and the sensitivity of the resulting assay without sacrificing microarray spot density. The proper selection of porous substrate also avoids problems such as nonuniform drying, quenching, and stacking of fluorescent dyes. Contact liquid dispensing techniques are often not suitable for use with these porous

Biochip technology

120

substrates. This is due to errors in the amount of deposited material resulting from capillary forces generated by the substrate and through damage to the substrate from the impact of the dispensing tip (Schober et al., 1993). Non-contact liquid dispensing overcomes these limitations. In addition, porous substrates overcome problems with drop bouncing and other movement along nonporous substrates sometimes encountered with non-contact liquid dispensing strategies. Hence, non-contact dispensing and porous substrates are an ideal combination for generating high-quality microarrays.

TECHNIQUES FOR SUCCESSFUL NON-CONTACT LIQUID DISPENSING A non-contact liquid dispensing robot requires a number of component systems for monitoring liquid dispensing and cleaning of the system between samples. In addition, effective use of these robots in high-throughput scientific applications requires substantial software and informatics support. This section will describe techniques for process monitoring, tip maintenance, and software controls that enhance the capabilities of noncontact liquid dispensing robots. Process Monitoring Effective non-contact liquid dispensing of small volumes of liquid at high throughput requires techniques to monitor the progress of the dispensing process. Monitoring often occurs following the dispensing process by examining the quality of the drops on the substrate. It is also possible to monitor system parameters during the dispensing process to allow for real-time modification of dispensing conditions. The later monitoring process is more challenging, but affords a substantial improvement in quality of the liquid dispensing process. Vision control systems are installed on most non-contact liquid dispensing robots. These systems are often composed of a strobe light and a camera to capture the drop in midflight following ejection from the dispensing head (see Figure 2). This vision system enables assessment of drop characteristics, such as satellite formation and consistency of drop diameter. Typically, this analysis is performed to adjust the robot prior to performing a dispensing run. Some robots, such as the Biochip Arrayer from Packard Instrument Company, can evaluate the performance of the run by maintaining a log of the pressure changes that occur during the liquid dispensing cycle. These pressure changes are linked to the volume of drops dispensed to a specific location, allowing the system to flag the user when a tip on the dispensing head is not performing correctly. Options in the software allow for automatic elimination of such a tip and the creation of a corresponding error log. The BioChip Arrayer automatically generates a liquid dispensing protocol from this error log that enables activation of post-run operation to dispense to locations missed during the original dispensing run. Vision systems based on video cameras with spot analysis software can also be used to ascertain the quality of the spots on the substrate. This level of vision control is especially important in large-scale manufacturing operations.


121

Tip Maintenance Carryover between dispense cycles is a problem faced by all liquid dispensing robots. Large-volume liquid dispensing robots can incorporate disposable tips with filters to eliminate most carryover. Unfortunately, disposable tips are not feasible for the very small volumes addressed by non-contact liquid dispensing robots. To keep carryover to a minimum, it is necessary to use sophisticated surface cleaning techniques. Ultrasonic washbowls are the best approach for physical cleaning of the tip surface. In addition, it is important to devise cleansing solvents that effectively remove the samples being dispensed, yet do not damage the tip or react chemically with future samples. A combination of ethanol and water is often used for cleaning tips used for dispensing DNA. Software Control Software control systems are a critical component of non-contact liquid dispensing robots, and they often impact the benefit that these robots provide to end-users. Software control systems provide the user with a software interface to manage the operation of the instrument in its minimal configuration. More robust configurations include software components that address the broader informatics and performance needs of the end-user. There are two capabilities that are especially important to those generating microarrays and miniaturizing complex assays, dispense-on-the-fly software, and sample tracking software. Dispense-on-the-fly software provides users with the capability to precisely coordinate the dispensing of liquids with the movement of the dispensing head or robot stage. Microfab Technologies (Plano, TX) uses dispense-on-the-fly software to control the picoliter dispensing of solder used for high-throughput manufacture of integrated circuit packaging and printed circuit boards (Wallace and Hayes, 1997). In this application, 200 drops per second of solder were deposited at a 250-micrometer pitch by a non-contact dispensing robot with a piezoelectric actuator. Higher deposition rates and different drop pitches can be achieved by modifying the print head movement with the actuation rate of the dispensing tip. Sample tracking software is required for applications where large numbers of samples are being dispensed by the liquid handling robot. This requirement is more acute for noncontact liquid dispensing robots where more than one solution can be dispensed to the same location. Integrating sample tracking information with a data management system enables a scientist to rapidly integrate results of microarray experiments with the liquid dispensing procedures and background information available for each element of the microarray. MolecularWare (Cambridge, MA) produces a suite of software applications that provide a unified data management solution for microarray and high-throughput screening applications. The ArrayDesigner™ module of this suite is integrated with the operating software of the BioChip Arrayer to provide end-users with a robust interface for tracking sample information (see Figure 3, see Color Plate 5.3). The software tracks both sample information and the liquid dispensing information associated with each

Biochip technology

122

element of a microarray. This interface can be integrated with a complete data management system that automatically integrates intensity data from scanned images with sample information, protocol information, and user annotations.

EMERGING BIOCHIP APPLICATIONS FOR NON-CONTACT LIQUID DISPENSING Non-contact liquid dispensing provides a number of features that are critical for life sciences applications. These features overcome limitations encountered with contactbased liquid dispensing. One key feature is the ability to dispense to multiple locations using dispense-on-the-fly. This increases throughput for screening applications involving thousands of samples. The absence of surface contact also provides a unique capability to dispense to a location multiple times. This

Figure 3. See Color Plate 5.3. A screen shot of the ArrayDesigner from MolecularWare. The ArrayDesigner provides an intuitive interface for designing the source—destination liquid dispensing procedures for manufacturing microarrays. Colorcoding and a procedure tree are used to visualize the complex source—destination sample relationships that are stored in a database for tracking sample information. capability is absolutely critical in the effort to reduce the volumes of multistep assays


123

used in the pharmaceutical screening process. Finally, non-contact liquid dispensing is much more flexible in adjusting the volume of solution dispensed to each location. For example, the volume dispense to one location can be completely different than an adjacent location in a single aspirate and dispense cycle. This feature can substantially reduce the time it takes to generate complex assays in the nanoliter to picoliter volume range. This section will examine the stability of biological samples in non-contact liquid dispensing systems and demonstrate how non-contact liquid dispensing has provided a benefit in a number life sciences applications. Sample Stability In extending the capability of microarrays and assay miniaturization, it is necessary to accurately dispense a variety of solutions without destroying or modifying the constituents of the solution. For example, liquid acceleration of 50,000–100,000×g through tips with a bore diameter of 30–150 µm is typical for piezoelectric dispensers, raising concerns that this liquid dispensing technology is too harsh for many life science applications. In one study, the effect of shear stress on the stability of lambda phage plasmid DNA was measured in a glass capillary driven by a piezoelectric actuator (Evensen et al., 1998). The investigators found that plasmid DNA at 1 and 10 nanograms per microliter can be mixed at up to 2000 Hz for 20 minutes without sustaining any damage. A more extensive study was performed on the ability of piezoelectric dispensers to pipet a large variety of compounds and reagents (Schober et al., 1993). This study found that a wide variety of compounds could be accurately dispensed intact using piezoelectric technology. Supercoiled DNA plasmids (4100 bp) and tRNA were both dispensed with a drop volume of 500 picoliters at 2000 drops per second. Agarose gels indicate that in both cases the starting material is indistinguishable from the dispensed material, with no change in the relative abundance of closed circular DNA versus supercoiled DNA in the case of plasmid DNA. This study also evaluated the activity of Taq DNA polymerase following dispensing with a piezoelectric system and found that the resulting enzyme is functionally equivalent to an untreated control in PCR assays. Finally, E. coli cells were able to stand up to the rigors of piezoelectric dispensing with a droplet volume of 500 picoliters and a dispense rate of 2000 drops per second. Electron microscopic examination of the E. coli following piezoelectric dispensing revealed that fimbrial surface appendages on the dispensed E. coli were no different than those on nondispensed controls. The conclusions of these studies suggest that non-contact dispensing of picoliter-scale volumes is applicable to a broad range of biological molecules. In the case of proteins, it is very important to choose solvent conditions that prevent the formation of protein aggregates and minimize the interaction of the protein with the glass capillary. In addition, the choice of solvent may be critical in preventing protein denaturation during the dispensing routine. Surface treatments of the dispensing nozzles and improvements in materials may go a long way to reducing the process of trial and error necessary to find the best conditions to dispense a specific molecule or cell using a system based on piezoelectric or micro-solenoid valve actuation.

Biochip technology

124

Example Applications There are a number of applications that have been successfully miniaturized using noncontact liquid dispensing. There is a great deal of interest in miniaturizing the preparation of biological assays and combinatorial libraries, with substantial effort focused on increasing the density of microplates used in these procedures (Lemmo et al., 1997). These efforts are driven by the desire to reduce the working volumes of expensive reagents and to increase the number of data points generated in each experiment. Noncontact liquid dispensing is also a critical enabling technology for a number of key biochip applications. This section will demonstrate how large-scale preparation of DNA microarrays, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS), and sample loading for capillary electrophoresis or lab-on-a-chip applications can all be improved by non-contact liquid dispensing technology. Large-Scale Preparation of DNA Microarrays Critical to the successful miniaturization of biological assays is dispensing of small volumes of reagents used in the generation of the microarray. Direct synthesis of oligonucleotides on the microarray or the attachment of pre-made oligonucleotides or DNA fragments to the microarray represent the two paths for generating a DNA microarray. The direct synthesis of nucleic acids on a glass microscope slide substrate can be performed using a suitable length linker such as hexethylene glycol followed by nucleoside addition with traditional DNA phosporamidite chemistry (Maskos and Southern, 1992; Southern et al., 1994). These procedures rely on temporary wells to confine reagents to a specific region of the microarray to generate oligonucleotides that typically range in size from 5 to 15 mers. The substitution of a polypropylene tape with hydroxyl groups for a derivatized glass support enables the synthesis of up to 200-mer oligonucleotides (Bader et al., 1997). Non-contact liquid dispensing robots have been adapted by companies such as Protogene Laboratories (Palo Alto, CA), Incyte Pharmaceuticals (Palo Alto, CA), and Rosetta Inpharmatics (Kirkland, WA) to synthesize 10- to 50-mer oligonucleotides in microarrays of up to 8000 elements (Catellino, 1997; Marshall and Hodgson, 1998). Spotting robots are also used to synthesize peptide nucleic acid (PNA) arrays of up to 1000 elements on polymer membranes using an adaptation of oligopeptide synthesis methodology (Weiler et al., 1997). The step yields achieved in the synthesis of microarrays of oligonucleotides are similar to those obtained with traditional oligonucleotide synthesis on controlled pore size glass (CPG) using a Perkin-Elmer (ABI) synthesizer (Foster City, CA). While all DNA synthesis procedures suffer from truncated oligonucleotide chains, DNA synthesis on microarrays suffers from the inability to subject the completed oligonucleotides to traditional methods of post synthesis purification. Truncated oligonucleotides can create problems with quality control and signal background due to incorrect binding of the labeled sample to truncated oligonucleotides. The combination of ellipsometric and interferometric analysis can provide scientists with information on the length of the synthesized oligonucleotides in each spot of a microarray to enhance quality control efforts (Gray et al., 1997).


125

Affymetrix is a leading commercial source of microarrays and one of the pioneers in performing non-contact DNA synthesis on a microarray. The dense microarrays available from Affymetrix typically contain 64,000 oligonucleotides tethered to a silicon wafer (Lockart et al., 1996). However, only about 3000 genes can be accurately analyzed with this microarray due to the strategy of using 20–300 oligonucleotides to perform match and mismatch analysis for each gene (Jordan, 1998). This high level of redundancy overcomes errors in DNA synthesis and hybridization to detect one copy of mRNA against a background of 300,000 copies of total RNA. The cumbersome and time-consuming procedural steps required to synthesize oligonucleotides on a microarray lead most scientists to generate microarrays of presynthesized oligonucleotides or DNA fragments. This approach has the added benefit that all components of the microarray can be purified and undergo quality control prior to being placed on the microarray. Most laboratories are dispensing PCR products of between 500 and 1000 bases onto a nonporous support such as a microscope slide. Excellent results can also be obtained using porous supports such as membranes with radioactive labels, where sensitivities of one copy of mRNA against a background of 10,000 copies of total RNA are typical (Jordan 1998). Non-contact liquid dispensing robots offer more flexibility than contact liquid dispensing robots by dispensing to delicate structures such as membranes and microgels as well as solid supports. In addition, by allowing for multiple dispense-and-evaporate cycles for each spot, noncontact liquid dispensing robots allow the DNA concentration of each spot on a microarray to be increased. Modifying the concentration of oligonucleotides at each spot of a microarray and the concomitant duplex stability can improve the consistency of hybridization across a dense microarray of oligonucleotides (Marshall and Hodgson, 1998). This approach is especially effective in polyacrylamide gel pads (Schober et al., 1993) due to retarded diffusion (Livshits and Mirzabekov, 1996). The other benefit of multiple dispenses to a single spot is that it may allow the densities of pre-synthesized DNA microarrays to approach the densities obtained when DNA is directly synthesized on the microarray. A key limitation to increased densities for microarrays of presynthesized DNA will be the minimum drop size possible with non-contact liquid dispensing robots, a limitation that will rapidly fall away with improvements in piezoelectric dispensing technology. Non-contact liquid dispensing also allows for printing of microarrays at very high throughputs as a result of dispense-on-the-fly techniques. Mutation Analysis Using MALDI-TOF A rapid scan of patient DNA for mutations or polymorphisms using microarrays has the potential to revolutionize DNA diagnostics. One of the most popular techniques for examining point mutations and single nucleotide polymorphisms is the base extension reaction with dideoxy terminator bases (Pastinen et al., 1997). However, one of the problems associated with this technique is occasional mismatch between the base extension primer and the target DNA, which can lead to false positive and false negative results (Kozal et al., 1996; Marshall and Hodgson, 1998). Mistakes of this nature can be very costly in clinical settings, especially where false-negative diagnostic results lead to a

Biochip technology

126

delay in treatment. Scientists at Sequenom Inc. (Hamburg, Germany and La Jolla, CA) recognized that MALDI-TOF has the potential to provide very accurate diagnostic analysis of mutations that may overcome the errors associated with mismatch events (Koster et al., 1996). This approach enables scientists to use the mass of the extended primers to verify the identity of the bases added in a base extension reaction. Knowing the base composition of the extended primer makes it possible to discriminate between match and mismatch primer recognition events. Critical to the success of this approach as a commercial DNA diagnostic platform is overcoming the sensitivity and reproducibility limitations associated with the preparation of MALDI-TOF samples. Traditional MALDI-TOF samples are prepared with a volume of matrix that is much larger in scale than the laser irradiation profile used in the extraction step. Scientists are required to manually find regions of the crystallized matrix that are rich in sample, a process that prevents automation of the MALDI-TOF procedure. This limitation to automation was overcome by using a piezoelectric-driven non-contact liquid dispensing robot to sequentially spot 6-nanoliter volumes of MALDI-TOF matrix and extracted base extension products (Little et al., 1997). The result is a very small and uniform spot of crystallized MALDI-TOF matrix that can be entirely covered by the laser irradiation profile. This breakthrough in sample preparation allows the entire MALDI-TOF procedure to be automated, leading to a viable platform for commercial scale DNA diagnostics. Both Sequenom Inc. and Brax (Cambridge, UK) are developing a commercial platform for DNA diagnostics based on mass spectrometry. Sample Loading for Lab-on-a-Chip Applications Lab-on-a-chip technology promises to enhance the productivity of many laboratory procedures by miniaturizing movement of liquids through capillaries (Kricka, 1998; Pfost, 1998). Caliper Technologies (Mountain View, CA), Orchid BioComputer (Princeton, NJ), and Soane BioSciences (Hayward, CA) are examples of companies building systems that use electroosmotic pressure to move molecules and solutions through a network of micro-etched channels that lead to chambers for performing mixing, separations, and chemical reactions. Applications being targeted for miniaturization by lab-on-a-chip technology include the separation and analysis of DNA sequencing ladders and the synthesis of compound libraries. Non-contact liquid dispensing robots may contribute to improvements in the loading of small volumes of sample into lab-on-a-chip cartridges. Non-contact liquid dispensing robots have been successfully adapted to load capillary electrophoresis systems (Sziele et al., 1994). The benefit of non-contact liquid dispensing robots for lab-on-a-chip applications is the ability to load numerous reagents and samples into multiple ports in manner carefully orchestrated to coincide with the reactions occurring in the lab-on-a-chip.

CONCLUSIONS Non-contact liquid dispensing technology holds great promise for enhancing the quality and reducing the cost of biochips. Critical factors affecting drop formation and drop


127

trajectory have been worked out, and commercial versions of non-contact liquid dispensing robots are now available. Advances in piezoelectric actuation will continue to drive the minimum drop volume, thereby reducing the spot size for biochip applications. Add to this the benefits of substrate flexibility and the high throughputs enabled by dispense-on-the-fly, and it becomes clear that non-contact liquid dispensing technology will soon emerge as the dominant platform for the manufacture of biochips.

REFERENCES Bader, R., M.Hinz, B.Schu, and H.Seliger. 1997. Oligonucleotide microsynthesis of a 200-mer and of one-dimensional arrays on a surface hydroxylated polypropylene tape. Nucleosides Nucleotides 16:829–333. Catellino, A.M. 1997. When the chips are down. Genome Res. 7:943–946. de Rooij, N. 1999. Fluidic microsystems for samll volume dispensing. Paper read at IBC 4th Microfabrication and Microfluidics Conference, San Francisco. Evensen, H.T., D.R.Meldrum, and D.L.Cunningham. 1998. Automated fluid mixing in glass capillaries. Rev. Sci. Instrum. 69:529–526. Graves, D.J., H.-J.Su, S.E.McKenzie, S.Surrey, and P.Fortina. 1998. System for preparing microhybridization arrays on glass slides. Anal. Chem. 70:5085–5092. Gray, D.E., S.C.Case-Green, T.S.Fell, P.J.Dobson, and E.M.Southern. 1997. Ellipsometric and interferometric characterization of DNA probes immobilized on a combinatorial array. Langmuir 13:2833–2842. Gruhler, H. 1999. A fast method for spotting microarrays. Paper read at IBC 4th Microfabrication and Microfludics Conference, San Francisco. Guschin, D., G.Yershov, A.Zaslavsky, A.Gemmell, V.Schick, D.Prudnikov, and A.Mirzabekov. 1997. Manual manufacturing of oligonucleotide, DNA, and protein microchips. Anal. Biochem. 250:203–211. Jordan, B.R. 1998. Large-scale expression measurement by hybridization methods: From high-density membranes to “DNA Chips.” J. Biochem. 124:251–258. Koster, H., K.Tang, D.-J.Fu, A.Braun, D.van den Boom, C.L.Smith, R.J.Cotter, and C.R. Cantor. 1996. A strategy for rapid and efficient DNA sequencing by mass spectrometry. Nature Biotechnol 14:1123–1128. Kozal, M.J., Nila Shah, Naiping Shen, R.Yang, R.Fucini, T.C.Merigan, D.D.Richman, D.D. Morris, E.Hubbell, M.Chee, and T.Gingeras. 1996. Extensive polymorphisms observed in HIV-1 clade B protease gene using high-density oligonucleotide arrays. Nature Med. 2:753–759. Kricka, L.J. 1998. Revolution on a square centimeter. Nature Biotechnol. 16:513. Lemmo, A.V., J.T.Fisher, H.M.Geyson, and D.J.Rose. 1997. Characterization of an inkjet chemical microdispensor for combinatorial library synthesis. Anal Chem. 69:543–551. Little, D.P., T.J.Cornish, M.J.O’Donnell, A.Braun, R.J.Cotter, and H.Koster. 1997. MALDI on a chip: Analysis of arrays of low-femtomole to sub-femtomole quantities of synthetic oligonucleotides and DNA diagnostic products dispensed by a piezoelectric pipet. Anal. Chem. 69:4540–4546. Livshits, M.A., and A.Mirzabekov. 1996. Theoretical analysis of the kinetics of DNA

Biochip technology

128

hybridization with gel-immobilized oligonucleotides. Biophys. J. 71:2795–2801. Lockart, D.J., H.Dong, M.C.Byrne, M.T.Follettie, M.V.Gallo, M.S.Chee, M.Mittmann, C. Wang, M.Kobayashi, H.Horton, and E.L.Brown. 1996. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nature Biotechnol. 14:1675– 1680. Marshall, A., and J.Hodgson. 1998. DNA: An array of possibilities. Nature Biotechnol. 16:27–31. Maskos, U., and E.M.Southern. 1992. Parallel analysis of oligodeoxyribonucleotide (oligonucleotide) interactions, I: Analysis of factors influencing oligonucleotide duplex formation. Nucleic Acids Res. 20:1675–1678. Nilsson, J. 1999. An automated microscaled system for rapid protein identification. Paper read at IBC 4th Microfabrication and Microfluidics Conference, San Francisco. Pastinen, T., A.Kurg, A.Metspalu, L.Peltonen, and A.-C.Syvanen. 1997. Minisequencing: A specific tool for DNA analysis and diagnostics on oligonucleotide arrays. Genome Res. 7:606–614. Pfost, D.R. 1998. The engineering of drug discovery. Nature Biotechnol. 16:313. Schober, A., R.Gunther, A.Schwienhorst, M.Doring, and B.F.Lindemann. 1993. Accurate high-speed liquid handling of very small biological samples. BioTechniques 15:324– 329. Southern, E.M., S.C.Case-Green, J.K.Elder, M.Johnson, K.U.Mir, L.Wang, and J.C.Williams. 1994. Arrays of complementary oligonucleotides for analysing the hybridization behavior of nucleic acids. Nucleic Acids Research 22:1368–1373. Stimpson, D.I., P.W.Cooley, S.M.Knepper, and D.B.Wallace. 1998. Parallel production of oligonucleotide arrays using membranes and reagent jet printing. BioTechniques 25:886–890. Sziele, D., O.Bruggemann, M.Doring, R.Fretag, and K.Schugerl. 1994. Adaption of a microdrop injector to sampling in capillary electrophoresis. J. Chromatogr. A 669:254– 258. Wallace, D.B., and D.J.Hayes. 1997. Solder jet technology update. Paper read at ISHM ’97 Conference. Weiler, J., H.Gausepohl, N.Hauser, O.N.Jensen, and J.D.Hoheisel. 1997. Hybridizationbased DNA screening on peptide nucleic acid (PNA) oligomer arrays. Nucleic Acids Res. 25:2792–2799. Yershov, G., V.Barsky, A.Belgovsky, E.Kirillov, E.Krendlin, I.Ivanov, S.Parinov, D.Guschin, S.Dubiley, and A.Mirzabekov. 1996. DNA analysis and diagnostics on oligonucelotide microchips. Proc. Natl. Acad. Sci. U.S.A. 93:4913–4918.

6 High-Throughput Arrays for Efficient Screening and Analysis Mitchell D.Eggers, Bill Balch, Stafford Brignac, James Gilmore, Michael Hogan, Terri King, Deval Lashkari, Aleksandar Milosavljevic, Tom Powdrill, and Amy Smith

INTRODUCTION DNA microchips are well established as the latest technology in the pharmaceutical industry’s drug development toolbox. DNA chips have already had a significant impact in the field of genomics (Iyer et al., 1999), which is described as the broad uncovering and understanding of the workings of the entire human genome. The human genome and its construction are immensely complex. Understanding the genome will provide a greater comprehension of the molecular and genetic basis of disease. The daunting task of deciphering the genome would be impossible without the development of new technologies that enable faster, better, and smarter analysis of the genome. For each of the ~100,000 genes in the human genome, researchers need to link the actual DNA text (a gene) with its meaning, or function. With the development of high-throughput detection and analysis methods, associating function to specific genes and sequences is a reality. One such method, DNA chips, permits this broad view across the entire genomic landscape. DNA microarrays blend engineering know-how with biological expertise, and their rapid development is due in part to the maturing of both microelectronics and molecular biology. The impact of DNA arrays has already been felt in drug discovery. As new genes are discovered, their functions are being better understood by the use of the chips. For example, researchers use DNA arrays to search for genes that have a particular expression pattern across different experimental conditions. This allows assessment of the behavior of hundreds to thousands of genes simultaneously and selection of those that respond in a certain way for further in-depth analysis. Arrays can be used in gene mapping, gene expression monitoring, polymorphism detection, and diagnostics (Eggers and Ehrlich, 1995).

BACKGROUND ON ARRAYS Biology is entering the industrial age. PCR amplifications, once restricted to only a small set of tubes, are now routinely performed in batches containing 384 samples. DNA sequencing, once restricted to a large gel and manual analysis that yielded a read rate of

Biochip technology

130

approximately 4000 bases in 24 hours, has matured to bundled capillary electrophoresis units capable of reading 1,000,000 bases in 24 hours (Spectrumedix). Similarly, the process of nucleotide analysis for expression and polymorphisms is undergoing a revolution due to the invention of array technology. In the early 1990s, the average researcher was limited by gene expression analyses of more than a handful of genes for any given study. With high-density microarrays, it is now possible to analyze entire genomes across the same small set of parameters (Wodicka et al., 1997). However, it is still difficult to analyze routinely any set of genes across a thousand or more parameters. High throughput is required in a multitude of applications: from polymorphism analysis (Cargill et al., 1999), which requires thousands of patient samples, to lead optimization, where the number of compounds to be screened is in the millions. A number of companies have technologies amenable to high-throughput screening applications. DNA or RNA analysis is routinely assayed using PCR (e.g., TaqMan®, Applied Biosystems). The drawback, however, is that these assays screen only a few genes. DNA microchips provide an exciting alternative, efficiently screening many genes simultaneously. Incyte, Affymetrix, and Molecular Dynamics/Amersham are industry leaders in high-density arrays (more than 1000 genes). The main barriers to utilizing these technologies for high-throughput screening have been cost and the ability to fabricate, hybridize, and image arrays in the required quantities. Other possible arraybased solutions may be provided through electronic hybridization (Nanogen) or flowthrough technology (GeneLogic). Over several years, Genometrix has developed a high-throughput microarray technology platform. Microarray production is based on the deposition of oligonucleotides onto a glass substrate using a capillary printer (Figure 1). Each array consists of 256 elements. Samples are prepared by a series of custom robots, and hybridization of the sample to the array is performed at room temperature. Proprietary room temperature chemistries are employed to maintain an open architecture following PCR, thereby facilitating full automation of all subsequent reactions. The array is analyzed by imaging via a patented CCD imaging system, and the data are interpreted using proprietary algorithms in the multi-application Vista-Logic™ software system developed by Genometrix. The Genometrix system can process more than 10,000 samples per day, yielding in excess of 250,000 data points. This rate can be sustained for both genotyping and gene expression. Because this system is so flexible, new arrays can be fabricated, validated, and assayed within 45 days. This flexibility is especially advantageous for polymorphism analysis, as it allows the researcher to quickly choose recently disclosed polymorphisms and assess their relationship to diseases or therapeutics. A further advantage of the Genometrix platform is that it can discriminate gene expression between gene families, a capability not currently available to users of cDNA arrays. It also provides a secure encrypted connection that allows clients the flexibility to design, run, and analyze experimental outcomes (expression analysis and genotyping) over the Internet in a simple point-and-click environment.

High-throughput arrays for efficient screening and analysis

131

Figure 1. Genometrix Capillary Array Printer The Genometrix capillary array printer deposits oligonucleotides onto a glass substrate. Each array consists of 256 elements. GENOMETRIX TECHNOLOGY Automation Functional applications of microarrays such as population-wide genetic screening, clinical diagnostics, and drug candidate screening often require the processing of large numbers of patient or cultured samples. Front-end processing of these samples is labor intensive, hazardous, highly repetitive, and susceptible to human error. Bottlenecks during complex sample preparation and processing are common, occurring during bar code labeling; sample tracking, storage, and archiving; and data management. Therefore, the implementation of integrated high-throughput automated systems capable of processing and managing data from multiple sources provides a logical solution. Our high-throughput production facility utilizes integrated automated workstations and laboratory management software tools. Novel high-throughput integrated systems consisting of off-the-shelf and custom components are currently being developed. Throughput, flexibility, and reliability were key factors in the design and implementation of the systems. Sample processing consists of four base workstations: (1) DNA/RNA extraction and purification, (2) blood card processing, (3) PCR assembly, and (4)

Biochip technology

132

hybridization and detection. The independent workstation platforms were designed to perform single processes in a very robust fashion, with an emphasis on throughput and reliability. All four can be configured in a variety of combinations to yield a maximum throughput in excess of 10,000 samples per day. The Genometrix workstations are based on a Beckman Coulter Orca® Robot utilizing modified off-the-shelf and custom devices to enhance throughput. Custom device drivers and control software were developed internally for seamless integration of the automated workstation’s components. The workstations are directly linked to an Oracle® database for real-time sample tracking and storage of sample processing data such as microarray raw images. The independent workstation module concept provides a flexible architecture to accommodate shifting production requirements while utilizing batch processes to maximize efficiency, throughput, and flow. The DNA/RNA workstation is currently configured to extract and purify high-quality DNA or RNA from cultured cells or homogenized lysates. Utilizing a 96-well filter plate format, bead-based assays are used with custom vacuum devices for processing large numbers of cultured samples (Figures 2 and 3, see Color Plates 6.2 and 6.3). Normalized aliquots of the resulting purified DNA/RNA samples are input to the PCR assembly workstation. The system, located in an environmentally controlled enclosure, is capable of unattended and autonomous operations. Finally, sample tracking and data sharing are accomplished via network interfaces. A single DNA/RNA workstation has a throughput of 5000 samples per day, far exceeding conventional purification systems currently available. The blood card processing workstation currently in development consists of proprietary blood card handling and punching instrumentation, which allows safe automated high-throughput blood sample analysis. Blood samples are spotted on-or offsite on a custom Genometrix blood card prior to robot processing. Bar-coded blood cards are placed in racks for automated random or sequential access. Subsequently, a single 1-mm punch is extracted from the blood cards and placed in 96-well plates for DNA extraction or PCR amplification directly from the sample punch. With a throughput of over 1000 samples per day, the first-generation blood processing workstation will be the first of its kind for high-throughput blood sample analysis on microarrays. The PCR assembly workstation is currently configured for high-throughput assembly of PCR reactions. The workstation is composed of Beckman Coulter Biomek® 2000 robots and fluorescence plate readers. Samples processed by the DNA/RNA or blood card workstations are input for the PCR assembly workstation. The workstation simply adds sample templates to pre-aliquotted master mixes, and quantifies and then normalizes the resulting offline thermal-cycled PCR products. PCR thermal cycling is one of the most difficult assays to automate, due to sealing and contamination issues, and is the least labor intensive. Automation would provide only minimal gains and is not essential. The PCR assembly workstation throughput is greater than 5000 samples per day. The hybridization and detection workstations serve as genotyping and gene expression data transducers and are the core of Genometrix’s production facility. This workstation utilizes Genometrix’s robotically controlled proprietary array imagers to combine highthroughput microarray hybridization and high-speed microarray detection. The imagers provide simultaneous high-speed, sensitive, two-color detection of 96 microarrays at a


133

rate of one array/second. This exceeds the throughput of conventional scanners by at least one order of magnitude (Brignac et al., 1999). The combination of the microtiter format arrays and standard room-temperature chemistries allows microarray hybridization assays to be easily

Figure 2. Diagram of a 96-Well Genometrix Array. See Color Plate 6.2. The 8×12 format contains 96 wells. Each well (see expanded view) contains a DNA array of up to 256 elements. The diagram demonstrates an array of 192 elements. A different experimental sample can be analyzed in each well. automated utilizing modified liquid and plate-handling components. The integrated hybridization and detection workstation autonomously hybridizes and detects 1200 microarrays per batch; the resulting raw images generated by the high-throughput imager are sent to the Genometrix database via the local network for automated bioinformatics processing. The hybridization and detection workstation generates more than 900,000 genetic analysis data points per day. As a result, robust bioinformatics processing is essential for capitalizing on the performance and data output of the workstation. Processing With the explosion of applications for microarrays, users worldwide are faced with the challenge of handling and processing sample sets of enormous size. For example, our imagers produce 48-megabyte files of quantitative data derived across 96 microarrays within a minute. To utilize microarrays in large-scale population-based genetic analysis,

Biochip technology

134

it is critical that the isolation, amplification, and hybridization steps be simple, efficient, automated, and support high-throughput processing.

Figure 3. Typical data images from hybridized arrays. See Color Plate 6.3. (A) Full unprocessed image of Genometrix 96-well array containing immobilized DNA and hybridized with labeled material. (B) Processed, closeup view of an individual well from the 96-well array.


135

To meet the need for reliable and secure high-throughput microarray processing, we implemented a system that optimizes automation and bioinformatics. Each individual processing step has been up-scaled and is designed for the standard 96-well format. Because all processing steps utilize this format, groups of 96 sam-ples travel through the entire operation as one unit. This format standardization simplifies sample tracking, data handling, and quality control (Figure 4). Genometrix’s VistaLogic software incorporates all aspects of a standard laboratory information management system (LIMS). Barcodes identify and allow samples to be tracked through all stages of processing, from patient/donor through data analysis, updating sample status at each step. Each process functions separately and is controlled by a work-order system, resulting in a steady flow of processing throughout operations. The sample-processing aspect of the system allows a manager to schedule processing of a specific group of samples and to assign work orders to operators. Work orders are generated by each individual process; completion of one process triggers the start of the next (Figure 5). By breaking down the overall operation into individual processes, we can control failures as they occur. This allows waste to be significantly reduced and increases overall system efficiency. One of the major challenges in the microarray field is keeping pace with highthroughput operations. Our approach to high-throughput sample processing capitalizes on standardized operations. Standardization, which eliminates expediting, reworking, and constant rescheduling, provides a fixed volume of production that improves scheduling accuracy and the ability to satisfy operational priorities. As most individuals in a manufacturing or production environment know, automation can be both a benefit and an impediment. Only when the automated system is designed and implemented properly do benefits begin to greatly outweigh impediments. The modular processing approach reaps its greatest benefit with automation. Independent processing modules significantly reduce instrument downtime while providing an additional level of isolation and contamination control of critical materials. Operations scheduling becomes much less complex, and the independent processing modules provide greater flexibility to accommodate changes in the production or business plan. Analysis To discover patterns in genomic data, scientists typically need to combine information about biologically relevant genomic DNA variation with information about gene expression at both mRNA and protein levels. The VistaLogic system offers high samplethroughput array-based data collection and analysis through distinct integrated modules: VistaMorph™ (SNP genotyping, pharmacogenomics, and genetic epidemiology), VistaExpress™ (gene expression analysis), and VistaPro™ (protein expression analysis; under development). Our system provides clients with access to information of specific interest via a secure encrypted connection as well as to the sample archive and database. The sample archive contains clinically and epidemiologically annotated blood samples ready for genotyping. The database contains genotyping and gene-related information that may be used immediately for discovery in conjunction with attendant medical records.

Biochip technology

136

Process Integration Most researchers require simple and instant access to experimental data. A hallmark of our process is to provide complete control and transparency of the data

Figure 4. Process used by Genometrix when receiving and storing samples. production process, from the moment samples are received and processing requests are made via the Internet, to the instant the experimental data are filed in the database and made available for analysis. Specifically, VistaLogic is deployed via a secure encrypted connection between Genometrix and client sites. Using this connection, VistaLogic seamlessly integrates the data production and discovery processes at Genometrix into client operations. Clients may initiate design and manufacture of proprietary arrays via Vista-Logic. Specific genes, polymorphisms, and alleles of interest are selected and submitted to our array design group. Once requests are received, the design and manufacture processes are initiated. Clients can request and monitor sample processing online. VistaLogic enables clients to be in full control of processing, starting from plating and normalization, through extraction of genotyping information from array images. As soon as the genotyping or gene expression information is collected and stored in the database, clients may begin performing desired analyses. Biological information may be divided into two general categories: phenome and genome. Phenome refers to the totality of information about the outward characteristics of an organism, such as anatomy, physiological characteristics, clinical information, and behavior. Unlike the genomic DNA sequence information, which is essentially text in the four-letter alphabet of nucleotides, phenomic information must frequently be accompanied by a precise semantic definition. “Smoker” in one epidemiological study may be defined differently than “smoker” in another study. Thus, medical and epidemiological data entered into VistaLogic are checked for possible errors and coded


137

using standard dictionaries with precisely defined terms.

Figure 5. Workflow used by Genometrix for genotyping services. Genotyping, Pharmacogenetics, and Genetic Epidemiology Pharmacogenetic and epidemiological data are typically analyzed using both parametric and non-parametric statistical methods, such as contingency tables and logistic regression, and estimation techniques such as maximum likelihood. Most frequently used statistical methods, such as those provided by Selvin (1996) for epidemiology, are currently being implemented in our system. Microarrays provide a multitude of allele scores per sample. In contrast, traditional epidemiological studies may score only a few polymorphisms. However, only a small fraction of the scored polymorphisms may be predictive of the variable of interest. Furthermore, the set of relevant polymorphisms may not be known in advance. To address this imbalance, a number of proprietary analysis algorithms have been implemented in VistaLogic that detect a small number of predictive polymorphisms among a potentially large set of irrelevant ones. In contrast to RFLP and STR polymorphisms that typically consist of more than two alleles, SNPs are typically biallelic. Thus, the information content of individual polymorphisms may be too small for significant predictive power. However, groups of biallelic SNP polymorphisms that are jointly predictive of the variable of interest may

Biochip technology

138

exist. Therefore, our system implements methods that discover such predictive groups of polymorphisms even in a situation where none of the polymorphisms is individually predictive. The success of pharmacogenetic and epidemiological studies critically depends on the quality of study design. To facilitate the study design process, VistaLogic offers an interactive SQL-like query builder and a set of visualization tools, including scatterplots, bar charts, line graphs, pie charts, and histograms. Gene Expression Analysis With the help of expression microarrays, the detailed state of living cells can now be observed as they react to external stimuli or transition from one physiological state to another. For example, changes in mRNA levels of the total set of 6100 yeast genes have been measured in response to the change in nutrients (DeRisi et al., 1997; Wodicka at al., 1997), in response to temperature change (Lashkari et al., 1997), and during cell division (Chu et al., 1998; DeRisi et al., 1997). Genes have been clustered based on their levels of expression across a number of consecutive time-points (Chu et al., 1998; DeRisi et al., 1997; Wen et al., 1998), revealing groups of coregulated genes. Two-way clustering of genes and cell states has resulted in the discovery of biologically significant patterns (Alon et al., 1999; Weinstein et al., 1997). Expression patterns hold promise of providing a highly informative model in screening and similar applications. Time profiles of gene expression obtained by microarrays could lead to complete “reverse engineering” of the genetic regulatory networks. In order to fully realize this promise, we have designed technologies for high sample throughputs. Specifically, the VistaLogic tools for high sample-throughput data analysis are: • Analytical: clustering algorithms and algorithms for inferring expression signatures • Query: an interactive tool for building SQL-like queries to retrieve data sets for analysis or visualization • Visualization: an interactive visualization toolbox, including scatter-plots, bar and pie charts, graphs, and histograms. Data Integration It is now widely recognized that data integration may be a limiting factor for the discovery process. Examples of successful data integration projects include Entrez (Schuler et al., 1996) and SRS (European Bioinformatics Institute), which provide crossindexing of DNA sequence, protein, and textual information. The GeneCards system (Rebhan et al., 1998) indexes information on the World Wide Web, pooling weblinks related to each named human gene into individual webpages. An integrated data view in VistaLogic enables both discovery and rapid user-driven design of custom arrays. VistaLogic currently integrates the following types of information (Figure 3, see Color Plate 6.3): • processed medical record information for the samples accessible through the Genometrix archive and database


139

• polymorphism and gene expression information obtained through the information factory • external gene-related information, including gene sequence, gene function, mapping, disease-relatedness, polymorphism, expression, and protein product information It also allows for the creation of proprietary and private experimental results and gene sequence data sets that can be viewed and used only by the designated client. Genometrix Platforms The ability to assay biological variation of information-rich macromolecules in a highthroughput fashion is critical to deriving the maximum amount of information needed to support the next generation of molecular biology developments. Information derived from the primary sequence of nucleic acids and proteins, and the corresponding structureactivity relationships derived therefrom, may be considered secondary and tertiary types of information, respectively (primary being the chemical nature of the monomeric units of such molecules). Now a quaternary type of information is needed, assimilating the effects of genetic composition and variation of a large number of genes and their gene products (mRNA and protein) with disease risk predisposition (molecular epidemiology, target lead identification) and, in a more applied format, therapeutic drug efficacy (pharmacogenetics). In order to achieve these goals, a large number of individual samples must be analyzed, requiring significant computational power and a robust assay platform. We have developed the 96-well high-throughput microarray assay platform with the ability to efficiently measure DNA sequence polymorphisms, mRNA expression levels, and protein expression levels (Figure 6). DNA Analysis At the level of genomic sequence variation, Genometrix has developed the Vista-Morph microarray technology, capable of detecting sequence polymorphisms with single-base resolution in a large number of polymorphic loci. In addition to processing large numbers of samples per day the salient features of this system include: 1. Computer algorithms developed in-house to quickly design probes and primers for each polymorphic locus. The parameters incorporated in these programs assure, to a high degree of confidence, the ability to amplify large numbers of targets in a multiplex fashion and significantly decrease the design time necessary for new arrays. Similarly, probe design processes have been streamlined so the sequence composition for each surface probe is normalized to give proper sensitivity and specificity at each polymorphic locus. In addition, probes are designed to interrogate both strands of the double-stranded PCR product (analogous to sequencing both strands) in order to give the highest confidence in the sequence assignment at each locus. 2. Hybridization protocols allowing binding and washing at room temperature.

Biochip technology

140

This feature significantly simplifies automation of the hybridization process, allowing the open architecture of the 96-well array chips. In addition, it obviates the requirement for any controlled temperature incubation apparatus. 3. Control and validation elements. For each polymorphic target locus, control elements representing the wild-type and all variants are synthesized that are equal in length to the ultimate PCR targets to be assayed. In this manner, the hybridization signature patterns for all homozygous and heterozygous allelic variants may be assayed and validated. This validation exercise is extremely important; it eliminates the need to obtain externally genotyped samples to validate the signal sensitivity and specificity for each probe/target combination. In addition, a control probe is included for every locus on the array to ensure proper amplification of the target sequence. The inclusion of this probe, designed to bind all possible sequence variants at a given locus, eliminates the need to run gels to ascertain the presence of the desired product. (Running gels would be difficult as the multiplex-generated targets are designed to be a relatively small, uniform length.) To date, we have designed and validated arrays interrogating genetic polymorphisms involved in environmental toxin clearance (N-acetyl-transferases, glutathione-Stransferases, cytochrome P450 monooxygenases, catechol-O-methyl transferases), pharmaceutical drug metabolism (predominantly the cytochrome P450 family), DNA repair (XRCC- and XPD-related series), and cardiovascular risk assessment (e.g., the Apo A,B,C,E series, LPL, β-fibrinogen, prothrombin). In addi-tion, neonatal screening arrays have been designed with collaborators (NeoGen) assaying for sickle-cell disorder, cystic fibrosis, and α-1-antitrypsin deficiency. Assays for somatic cell mutations (e.g., Kras oncogene) have also been developed. IIIustrated in the DNA portion of Figure 6 is a prototype array interrogating seven polymorphic loci of the N-acetyl-transferase 2 (NAT2) gene and one locus of the catechol-O-methyl transferase (COMT) gene. mRNA Analysis At Genometrix, VistaExpress mRNA analysis is performed using mRNA or firststrand cDNA-specific oligonucleotide surface capture probes on the same 96-well array chip platform. Automated methods are used to isolate either total or mRNA from cellular samples. In addition, several other features are important. The limited number of mRNA-specific probes per array (50–100) allows each probe to be evaluated for adequate signal production and specificity. The entire array is validated using labeled synthetic targets or biosynthetic targets derived from T7 polymerasegenerated transcripts from amplified material. Because the T7 transcript material is an RNA molecule, it also may be spiked into reactions in known molar amounts as a control. This allows the entire array to be validated in a manner not possible with existing largeelement or cDNA-based arrays. Each array is also provided with a control element, which can assay exo-genously added RNA derived from T7-derived transcription of amplicons derived from yeast intergenic regions. The use of surface probes specific for these transcripts, which are spiked into the reaction in known molar quantities, allows for control of the labeling and


141

binding reactions.

Figure 6. Flexibility of the Genometrix 96-well array platform. The Genometrix high-throughput platform may be utilized for assay of biological variation using DNA, RNA, or protein as the target macromolecule of interest. A number of labeling techniques may be employed to assay mRNA expression. cDNA labeled internally through the reverse transcription reaction may be derived from either

Biochip technology

142

oligo-dT or gene-specific priming. Gene-specific primers are favored. These allow the primer to be positioned close to the capture-probe binding site, eliminating the need for reverse transcription over long distances. In addition, the specificity afforded by genespecific priming contributes to overall lower background. Typically, label is incorporated using biotin- or digoxigenin-modified nucleotides in the reverse transcription reaction. mRNA may also be labeled directly through the use of covalent modifications introduced through the cisplatinum derivatives of digoxigenin or biotin (Dig-Chem Link, BiotinChem Link, Roche). The direct mRNA labeling reaction is faster, needs no enzymatic step, and fragments the mRNA during the process. The fragmentation of mRNA allows for more robust binding of smaller fragments to the surface-immobilized oligonucleotide probes. Of course, oligonucleotide arrays devoted to direct mRNA binding are of opposite sense to those used for first-strand cDNA binding. The dynamic range of the imaging array instrumentation, which is based on UV excitation and CCD detection, is approximately three orders of magnitude. In addition, an entire 96-well array containing 96 individual microarrays can be imaged in about one minute. The fluorescence detected is amplified through the use of enzyme-linked fluorescence (ELF, Molecular Probes). The sensitivity allows ample signal from as little as 50 nanograms of starting mRNA. This is crucial, as it allows gene expression to be monitored from starting cellular material from a single well of a 96-well plate, facilitating lead compound evaluation from large sample sets. Genometrix has developed expression array probes for approximately 200 genes of general interest to oncologists. In addition, genes involved in the metabolism of pharmaceutical compounds are being developed, and custom arrays may be fabricated and validated in a short time (approximately 45 days). Protein Arrays Although relatively early in development, the use of microarrays for the assay of protein expression is proving to be very promising. The basis of our microassays has been the well-characterized antigen-antibody reaction used for several decades in traditional platebased ELISA assays. Thus, monoclonal antibody reagents for thousands of proteins or other antigens of interest already exist and can be easily incorporated into the microELISA format. The power provided by combining this technology with the highthroughput 96-well array format comes from the ability to assay tens to hundreds of protein analytes in parallel in a single well. In addition, the methodologies developed at Genometrix for signal development and imaging are identical to those used for nucleic acid detection. Figure 6 shows a VistaPro protein expression microELISA in which capture monoclonal antibodies for five different cytokines were printed in quadruplicate in a single well. The single vertical column of signal in each is a result of capture of the IL-6 protein at a concentration of 780 pg/mL in a volume of 10 µL, followed by detection with a labeled second antibody in a traditional sandwich assay. Each of the other capture antibodies was found to be specific for its appropriate antigens (not shown). It is now possible to consider applications previously not feasible using conventional technology: parallel testing of blood-borne pathogens, sexually transmitted disease


143

pathogens, panels of tumor-specific antigen arrays, and other research applications. In addition, it now may be possible to correlate mRNA expression with protein expression explicitly

APPLICATIONS Disease-Gene Association The advent of high-throughput chip technology opened new opportunities for understanding how genetics affects the human condition. Beginning with the first identification of a gene-disease association, the interest in and understanding of the role of genetics in human health has grown exponentially. One of the most interesting applications of this is genetic epidemiology, the marriage of genetics and epidemiology. Genetic epidemiology is defined as “a science that deals with the etiology, distribution, and control of disease in groups of relatives and with inherited causes of disease in population” (Morton and Chung, 1978). During the 1970s and 1980s, genetic epidemiological research focused on the family and was analyzed by sophisticated statistical models (e.g., segregation analysis and linkage analysis). These methods very successfully identified single-gene causes of disease (e.g., Huntington’s disease, cystic fibrosis, Marfan syndrome). However, they have been less successful in understanding common diseases, such as heart disease and cancer, that may be the result of multiple genes without complete penetrance (the probability of an individual with a particular genotype having the disease). During this same period, molecular epidemiology was introduced (Perrera and Weinstein, 1982), and this subfield uses traditional association studies with biological markers as the risk factors of interest. The markers include subtle alterations in molecular processes that reflect known biological pathways. The markers could reflect alterations in metabolic processing, the presence of toxic compounds, or the effects of other pathologic processes. In recent years, genetic epidemiology and molecular epidemiology have moved closer together. Molecular epidemiologists, particularly those studying cancer, began investigating the relationship between functional allele variants in genes known to be associated with the processing of toxins. These include the cytochrome P450 family, Nacetyl transferase (NAT), and catechol-O-methyltransferase (COMT). Using these variants, a number of studies associated mutations with increased risk of lung cancer and other diseases. The potential for these studies was severely limited by the time required to obtain genotypes for inclusion in analysis. A single epidemiologic study searching for one association with a risk factor requires several hundred cases and controls. That number increases as the effects of the risk factor become more modest. This led to very large studies that examined one or two genes and required a number of years to collect samples and genotype individuals. These studies determined that some mutations predispose individuals with given risk factors to developing disease, confirming the hypothesis that high-frequency low-penetrant genes play an important role in defining individual

Biochip technology

144

susceptibility profiles. Moreover, this susceptibility is greatly altered by exposure to environmental factors such as occupational chemicals, smoking, and alcohol. The disease patterns for these complex multifactorial genes will be best explained by a profile of genes. The traditional method of genotyping does not allow timely genotyping of multiple genes on large numbers of individuals. However, with high-throughput genotyping technology, the ability to screen large numbers of individuals is now reality. Using this technology, researchers now have the ability to begin untangling the complicated relationship between disease, susceptibility, and environmental risk factors. In the future, one smoker may be profiled with genotypes that could better characterize the risk of developing lung cancer or emphysema. Additionally, by understanding the addictive pathways, information may be provided about which cessation products would be more effective for an individual, resulting in pharmaceutical rather than behavior modification. Genetic Information in Drug Discovery and Drug Development The genomics industry holds great promise to aid in the development of new health-care opportunities. Powerful tools are being developed to help scientists understand the complex biological pathways that underlie many human diseases. This knowledge has created many new targets for therapeutic intervention and an ability to determine which patients will best respond to these new treatments. The pharmaceutical industry is being pressured by managed care and healthcare reform to hold down costs. Currently, the industry spends $42 billion per year in research and development. A single drug can cost up to $500 million and take 10 years to develop, and only 1 drug out of 10 makes it through clinical trials (Wolpe Brown Whelan & Co., 1997). The pressure to contain costs has created opportunities for companies that can accelerate the production of novel drug candidates, decrease the length of development, or increase the percentage of drug candidates that make it through clinical trials. The Pharmaceutical Industry Into its Second Century: From Serendipity to Strategy (Boston Consulting Group, 1999) characterized the industry by three factors: 1. Competitive advantage has been driven by blockbuster drugs. 2. Companies have sought broadly applicable solutions to broadly defined conditions because detailed understanding of the basis and variations in disease was not scientifically possible. 3. Blockbuster drugs created temporary, but not sustainable advantage. Patent expirations often reversed the positions of even the most successful companies. Among the factors expected to characterize future successful pharmaceutical companies is the implementation of new discovery and high-throughput screening programs. These are expected to create a tenfold increase in the number of targets, and will be driven by genetic understanding and the availability of many new lead compounds. Improved understanding of the genetic basis of disease and genetic variations across populations will drive the industry toward more individualized disease solutions.


145

Pharmacogenomics/Pharmacogenetics The impact of genetic variation sparked development of single nucleotide polymorphism (SNP) analysis platforms. These platforms focus on SNP discovery or SNP screening. Both are important to understanding the role of SNPs in complex biological systems. Estimates indicate approximately 3 million genetic polymorphisms or 1 SNP per 1000 base pairs in the genome. These markers will be used to map the genes involved in such complex multigenic diseases as diabetes and hypertension. Different populations exhibit different allelic frequencies, and correlations can only be obtained by studying large diverse sample sets. Mapping and linkage studies are currently done with conventional gel-based technology. Pharmacogenomic applications focus on drug efficacy and safety. It has been estimated that one-third of the population does not respond optimally to a given drug. In some cases, only 20% of the population responds optimally. Toxic events associated with genetic variation in drug metabolism enzymes have been studied for more than 40 years, yet adverse drug interactions are estimated to be the fourth leading cause of death in the United States, with over 100,000 deaths each year (Stix, 1998). The cytochrome P450 family is responsible for significant variations in drug metabolism. The variations in response divide the population into two groups: poor metabolizers (PMs), whose enzyme activity is low or absent, and extensive metabolizers (EMs), with normal metabolic activity An example of this is the cytochrome P450 enzyme CYP2D6 (debrisoquine 4-hydroxylase). Initially discovered by debrisoquineinduced hypotension (Silas et al., 1977), CYP2D6 has been found to be a metabolic factor in multiple classes of pharmaceutical agents (e.g., tricyclic antidepressants, lipophylic β-blockers) (Evans, 1993). People who are PMs are very susceptible to express significant adverse drug effects when administered compounds processed by CYP2D6. There is wide population variation of PM. Euro-Caucasians have a frequency of PM of between 7 and 10%. This compares to Far Eastern Mongoloids and some black Africans, who have PM rates of VFFF1) and exit the chamber earlier than those at lower positions. (C) A DEP-FFF fractogram for separating MDA435 human breast cancer cells ( ) from erythrocytes (O). Figures 6B and 6C are reprinted with permission from Yang et al. (1999b).

Electronic manipulation of cells on microchip-based devices

163

Figure 7. (A) Schematic representation of a dielectrophoretic cell shift register as developed by Washizu et al. (1990). By sequentially applying electrical signals to electrode pairs (e.g., el. 1 & e1. 2; 1′ & 2′; 2 & 3; etc.), the cells experiencing positive dielectrophoresis can be shifted along the channel between the opposite electrode arrays. Finger-type structures between neighboring electrodes are made of dielectric materials. (B-D) Electrode structures developed by Müller et al. (1999) for a DEP funnel (B), aligner (C), and switch (D) for guiding the cells in a fluid flow. comprising multiple electrodes coupled with an inlet channel and two outlet ports, cells

Biochip technology

164

can be directed to either one of the two outlet ports by changing the electrical excitation pattern on the electrodes. Recently, several microchip-based devices were reported for controlling the motion of individual cells or particles in a fluid flow (Fiedler et al., 1998; Muller et al., 1999). Planar microelectrodes of various geometry are produced on silicon chips, and two identical ones are assembled facing each other and spaced several hundred micrometers apart. When the cells are carried with a fluid flow into these microstructures, 3D DEP forces are superimposed with fluid flow forces to modify cell motions and to funnel, align, trap, and switch them to different flow paths (Figure 7B-D). Such cell manipulation can be readily controlled by altering applied electrical signals, and is operative despite significant cell velocities induced by fluid flow. DEP manipulation of individual cells described here provides a homogeneous cell processing tool in which cell handling depends only on the application of electrical signals to microelectrodes, but not on the differences in cell dielectric properties. For example, the “cell deflector” may be used to sort cell mixtures into multiple fractions by synchronizing the arrival of individual cells at the deflector with the application of electrical signals. Cells in the mixtures need to be pre-identified and labeled so that the deflector can be supplied with information on the exit port to which the individual cells should be deflected. It remains to be seen whether it is possible to incorporate such functionality as cell identification into these manipulation devices. Dielectric Field Cages Over recent years, Fuhr and his co-workers have developed numerous three-dimensional multi-microelectrode devices for trapping and manipulating particles or cells (Fuhr et al., 1995a; Müller et al., 1999; Schnelle et al., 1993). A typical configuration, termed an octopole dielectric field cage, takes the form of a center-cut dielectric spacer sandwiched between two facing microchips, each having four identical microelectrodes arranged at the corners of a square. When appropriate electrical signals (e.g., phase-sequential AC voltages) are applied to the microelectrodes, negative dielectrophoretic forces are generated to drive particles toward the center region between the electrode elements. The surface of constant centrally directed DEP force depends sensitively on the applied electric signals and takes spherical, elliptical, cubic and needle-like shapes (Schnelle et al., 1993), forming a DEP trapping cage. Depending on the relative dimensions of particles with respect to electrode elements, a dielectric field cage may trap only one particle (Figure 8A; see Color Plate 7.8) or multiple particles (Figure 8B; see Color Plate 7.8). Typically, a particle (or a cell) is driven into a dielectric field cage by a fluid flow with little or no signal applied to the microelectrodes, and is then directed toward the center of the cage and trapped there by increasing signal amplitudes. A trapped particle can be released from the cage by changing electrode excitation conditions, for example, reducing the signal amplitude on one electrode element. Coupled with diffusion, sedimentation, and dipole-dipole interaction forces, a field cage can collect and trap submicroparticles (such as viruses) as aggregates, and hold them there for a prolonged time (Figure 8C; see Color Plate 7.8). The application of dielectric field cages includes focusing cells in a fluid stream, and trapping and levitating single cells for characterization.


165

Traveling-Wave DEP Manipulation The DEP manipulation techniques discussed in the previous four sections exploit nonuniform distributions of field magnitudes. On the other hand, phase distribution of field components, as generated by energizing spatially arranged microelectrode elements with electrical signals of different phases, can lead to a traveling-wave DEP force for manipulating cells (see Eq. (3)). For example, two facing linear

Figure 8. See Color Plate 7.8. (A) A 100-µm Sephadex particle, suspended in a 2-mS/m aqueous solution, is trapped in an octopole dielectric field cage under a 1-MHz, 7-V rms applied field. The 1-µm thick gold electrode arrays were fabricated on glass, and two identical ones facing each other were assembled together with a 200-µm spacing (only one electrode array is visible). (B) 3.4µm latex particles, suspended in a 1 -mS/m aqueous solution, are trapped and aggregated in two castellated interdigitated microelectrode caging arrays separated by a 48-µm spacing (only one electrode array is visible). (C) A simulation result showing submicron particles aggregated at the center of an octopole electrode cage under a rotating electrical field. Figures 8A-C are kindly provided by Drs. G. Fuhr and Th. Schnelle from Humboldt University in Berlin. microelectrode arrays on a microchip, when connected with electrical signals having

Biochip technology

166

sequential phase values, can produce a traveling electric field in the channel separating the electrode arrays (Hagedorn et al., 1992; Huang et al., 1993; Masuda et al., 1989). The field mainly lies in the direction across the channel. Field magnitude decays with the height from the electrode plane and field phase follows an approximate linear dependency with the distance along the channel. The cells in the channel will experience vertical DEP force components to trap or levitate them and horizontal components to transport them along the channel. Large area traveling-wave DEP manipulation of cells can be further achieved in the space above a linear microelectrode array (Fuhr et al., 1995b; Morgan et al., 1997; Wang et al., 1997b). The linear microelectrode arrays have been mainly used for characterizing cell responses to traveling-wave electrical fields. An improvement to the linear array is the so-called “meander” structure (Figure 9; see Color Plate 7.9) (Fuhr et al., 1994a). Cells can be transported into (and out of) the central region of the electrode array for characterization and analysis. Several approaches to traveling-wave DEP separation of cells with linear electrode arrays have been reported. In one approach, electrical signals are applied so that one cell type in a cell mixture is trapped onto electrode edges and, simultane-

Figure 9. See Color Plate 7.9. Traveling-wave DEP manipulation of a 110-µm latex particle suspended in a-10 mS/m aqueous solution, with a planar meander structure. (A) The particle is directed from left to the center of the electrode array under a 200-kHz field. (B) The particle is trapped and immobilized at the center of the electrode array by a-800 kHz field. (C) The particle is directed from the center to an exit channel in the vertical direction. Figures 9A-9C are kindly provided by Drs. G. Fuhr and Th. Schnelle from Humboldt University in Berlin.


167

ously, other cells are levitated and transported (Talary et al., 1996). Another approach involves the combined use of traveling-wave DEP forces with a fluid flow profile (De Gasperis et al., 1999). A linear array fabricated on a microchip generates a travelingwave electrical field in a thin chamber. A fluid flow profile is produced inside the chamber so that fluid velocity increases with distance from the chamber walls. Cells in the chamber are levitated to different heights in the profile under the balance of dielectrophoretic levitation and sedimentation forces, and are transported by the fluid flow at different velocities. In addition, cells experience horizontal traveling-wave DEP forces perpendicular to the fluid flow, which deflect them across the flow stream. Cell discrimination and separation can be achieved by exploiting cell velocity differences in both directions, that is, with and across the fluid flow. This approach not only provides increased flexibility and versatility for DEP discrimination of cells, but it also allows for continuous rather than batch-mode DEP separation (De Gasperis et al., 1999). A novel use of traveling-wave fields for cell manipulation involves a spiral electrode array consisting of four parallel linear spiral elements on a microchip (Wang et al., 1997b; Goater et al., 1998). When energized with phase-quadrature signals, the spiral array produces an electric field that travels along the radial direction into or out of the center. Cells subject to such an array will experience vertical DEP force components that levitate them against gravity or trap them on the electrode elements. Simultaneously, they experience radial force components that drive them toward or away from the center of the spiral electrodes. With appropriate choices of electrical signals applied to the electrodes, a spiral array can be used to trap cells on the electrode elements, or to concentrate them to the center (Figure 10), or separate cell mixtures by selectively focusing one cell type to the center of the array and simultaneously moving other types to the periphery (Wang et al., 1997c). Electrical-Field-lnduced Fluid Flow The methods to manipulate cells described in the previous five sections exploit dielectrophoretic forces that are exerted directly on the cells. Other electrical-field phenomena result in the forces acting on the fluid that suspend cells, and cell manipulation can also utilize such indirect forces. Two examples of electrical-fieldinduced fluid flow effects are traveling-wave pumping and dielectroosmosis. The traveling-wave pumping of fluids was first described and demonstrated by Melcher in his pioneering electrohydrodynamic works in the 1960s (Melcher and Firebaugh, 1967; Melcher and Taylor, 1969). In one of the approaches, a traveling electric field is produced in a chamber containing a liquid medium by applying phasesequential electrical signals to appropriately arranged electrode elements. Simultaneously, a temperature gradient is generated and maintained in the liquid along a direction normal to the field, resulting in a heterogeneous distribution of the temperaturedependent dielectric properties of the liquid medium. Thus, the applied field induces the volume charges in the medium, which interact with the field to produce volume forces acting on the medium in the direction with or against the field travel. The fluid motion resulting from such volume forces can propagate through the whole medium, leading to traveling-wave pumping of the medium.

Biochip technology

168

Microchip-based traveling-wave pumping was first demonstrated by Fuhr and his coworkers with linear microelectrode arrays fabricated on the bottom surface of a thin chamber (Fuhr et al., 1994b; Müller et al., 1993). Its operating principle for

Figure 10. (A) Human breast cancer MDA-MB-231 cells randomly distributed on a spiral electrode array prior to the application of electrical signals. (B) 20 seconds following application of voltage signals of 50 kHz and 0.7 V (rms) and outward phase sequence, the cells were collected at the central region of the spiral electrode array. Figures 10A and 10B are reprinted with permission from Wang et al. (1997c). fluid pumping is the same as that described above, except that the temperature gradient across the chamber is established and self-stabilized by the nonuniform joule heating of the fluid due to the applied electrical field. The traveling-wave fluid pump has no moving parts and can be electronically controlled. It may be coupled with dielectrophoresis so that cells or particles in a fluid suspension can be trapped and collected onto electrodes by positive DEP forces, and the fluid suspension can be continuously pumped simultaneously. The second electrical-field-induced fluid flow effect is dielectroosmosis. We shall first briefly examine electroosmosis, which refers to the electrical-field-induced motion of an electrolyte solution in the vicinity of an immobile charged wall (Ajdari, 1996). The static electric charge at the wall attracts electrolyte ions of opposite polarity in the solution. This process results in build-up of a charge double-layer structure. An applied external DC electric field will exert electrical forces on the volume charge in the solution phase of the double-layer, which causes the motion of the electrolyte solution. Electroosmosis plays an important role in capillary electrophoresis separation of particles and molecules (Bello et al., 1994; St. Claire, 1996). Dielectroosmosis refers to an AC-field-induced fluid motion phenomenon (Yeh et al., 1997). When AC electrical voltage signals are applied to solid electrodes with which an electrolyte solution is in contact, a frequency-dependent charge double-layer is induced at the solution/electrode interface. The local electrical field generates a time-dependent


169

volume force on the charges in the solution phase of the double-layer. The time-averaged volume forces drive the movement of the solution at the double layer, which is then propagated to the rest of the solution. Such an AC-field-induced fluid-flow effect may be implicated in the so-called low-frequency anomalous dielectrophoresis phenomenon (Pethig et al., 1992), and deserves further investigation and exploration.

OTHER ELECTRICAL-FIELD-BASED EFFECTS Electrical Deformation and Lysis of Cells In addition to the above electric-field-based manipulation methods for handling biological cells and other particles in suspension, field-cell interaction can lead to a number of other effects such as orientation of nonspherical cells, electroporation, and electrofusion. Here we consider two effects: electrical deformation and lysis of cells, and electrical monitoring of cell biological activities on microchips. It has long been known that electrical fields can induce stresses on cell membranes (Sale and Hamilton, 1968; Tsong, 1990). For example, pulsed electrical fields can induce membrane pores and cause cell electroporation, which can be exploited to facilitate delivery of biomolecular agents into cells or tissues (Wong and Neumann, 1982; Dev and Hofmann, 1996). It can also be used to release large molecules (e.g., DNA and mRNA) from cells including bacteria (Cheng et al., 1998c; Ohshima et al., 1998). For mammalian cells, AC electrical fields can cause cell deformation, and the frequency dependency of such cell deformation provides insights into the forces exerted on cells and can be modeled for analysis of cell dielectric and mechanical properties (Engelhardt et al., 1984; Sukhorukov et al., 1998). Electrical fields of sufficiently high strength can lead to the complete breakdown of cell membranes and result in cell lysis. Biomolecules—including DNA, RNA, and proteins—can be released from the cells for further biochemical analysis and assay. Such electrical lysis of cells, performed on microelectrode-based chips, can be integrated into a microfluidic device as the step of molecular extraction from target cells. Its advantages include that it does not require special cell lysing solutions, the process is fast, and it can be controlled electronically. Electrical Monitoring of Cell Activities Microelectrode structures have been used by a number of researchers to monitor biological activities of cells (Giaever and Keese, 1991). The general approach is to grow the cells over the microelectrodes and to monitor changes in the impedance. The time dependency of the impedance can reflect a number of cell activities such as cell spreading, cell adhesion, and cell division, and can be used to analyze and derive cell growth curves. It has been used to follow cell growth characteristics and to monitor cell responses to drugs, chemical stimulants, and toxic agents (Giaever and Keese, 1992, 1993; Lo et al., 1995).

Biochip technology

170

CONCLUSIONS AND PERSPECTIVES Dielectrophoresis and other AC-field-based effects provide effective mechanisms for manipulating and handling cells in aqueous suspensions on microelectrode devices. The electronic manipulation of cells depends sensitively on field configuration, field magnitude, and frequency, allowing for a versatile, flexible, and automatic control of the manipulation process through electronic means. Electrical field distributions can be analyzed and designed for different applications. Microelectrodes are ideally suited for the generation and application of well-structured microscopic-scale forces, and can be readily integrated with other microfluidic analytical components. Electronic manipulation of cells can be divided into two types: homogeneous manipulation and selective manipulation. In homogeneous manipulation, the forces exerted on cells are mainly dependent on the applied field conditions, and all the cells are subject the same or similar manipulation forces. Homogeneous manipulation of cells provides a tool for handling single or multiple cells in suspension, and can be applied in combination with other cell characterization and identification procedures. Examples of homogeneous manipulation include dielectric field cages, cell deflectors, and electricalfield-induced fluid motion. Selective manipulation refers to the separation, isolation, or handling of one or more target cell types from a mixture by exploiting differential manipulation forces acting on different cell types. Selective manipulation of cells is based on the observation that cell dielectric properties depend on the cell types and their biological states. Thus, by exploiting differential forces exerted on cells having different dielectric properties target cells can be separated, isolated, and concentrated from a mixture. Examples of selective manipulation include DEP migration, DEP affinity, and DEP field-flow-fractionation. A frequent question concerning electronic manipulation of cells relates to the invasiveness of the applied field. This is important not only in terms of academic interest but also from the practical consideration of the possible use of cells for analysis, growth, or even transplantation after being manipulated. It is evident from our discussion that high field strength does induce significant stresses to the cells, and can even cause lysis. Moderate field conditions have been demonstrated to lead to small or no changes in cell growth characteristics, except at low frequencies, where when electrical signals are applied certain electrochemical products may be generated to damage the cells (Wang et al., 1999a). Another recent report suggests that dielectrophoretic manipulation appears to regulate or modulate certain gene products (Archer et al., 1999), indicating that further work in fully understanding the interaction between the cells and electrical fields is required. An improved understanding and continuous development of dielectrophoresis and other electrical-field-related kinetic effects of cells has led to many important advances in electronic manipulation techniques over the last decade. Nevertheless, the majority of electronic manipulation work has, until now, been limited to demonstrations of operating principles on artificial samples. Applying dielectrophoresis to address real-world (e.g., biomedical) problems on microchips remains a significant challenge where issues such as


171

sample size, target cell number, and total number of cell types in the mixture may complicate electronic manipulation procedures.

ACKNOWLEDGMENTS We are grateful to Drs. G. Fuhr and Th. Schnelle for allowing us to use Figures 8 and 9 in this chapter, to Dr. L. Wu for his valuable comments and to Mr. J. Xu for his help with preparing Figures 1 and 7. JC acknowledges the support from the National Natural Science Foundation (Contract Nos. 39880035 and 39825108) and the National High Technology Program (863) (Contract No. 103–13–05–02). XBW acknowledges the support and help from his colleagues of many years in dielectrophoresis, particularly Drs. M. Arnold, F.F. Becker, J.P.H. Burt, G. De Gasperis, P.R.C. Gascoyne, R. Hölzel, Y. Huang, J. Noshari, R. Pethig, J. Vykoukal, X.J. Wang, M. Washizu, and J. Yang.

REFERENCES Ajdari, A. 1996. Generation of transverse fluid currents and forces by an electric field: Electro-osmosis on charge-modulated and unmodulated surfaces. Phys. Rev. E. 53:4996–5005. Archer, S.T., L.S.Britland, and H.Morgan. 1999. The influence of radio frequency electric fields as used in dielectrophoresis on gene expression in Baby Hamster Kidney fibroblasts clone 13. Biochim. Biophys. Res. Commun 257:687–698. Arnold, W.M. 1988. Analysis of optimum electro-rotation technique. Ferroelectrics 86:225–244. Arnold, W.M., and U.Zimmermann. 1982. Rotating-field-induced rotation and measurement of the membrane capacitance of single mesophyll cells of Avena sativa. Z. Naturforsch. 37c:908–915. Arnold, W.M., U.Zimmermann, W.Pauli, M.Benzing, C.Niehrs, and J.Ahlers. 1988. The comparative influence of substituted phenols (especially chlorophenols) on yeast cells assayed by electro-rotation and other methods. Biochim. Biophys. Acta 942:83–95. Arnold, W.M., R.K.Schmutzler, S.Al-Hasani, D.Krebs, and U.Zimmermann. 1989. Differences in membrane properties between unfertilised and fertilised single rabbit oocytes demonstrated by electro-rotation. Comparison with cells from early embryos. Biochim. Biophys. Acta 979:142–146. Bauer, J. 1999. Advances in cell separation: Recent developments in counterflow centrifuge elutriation and continuous flow cell separation. J. Chromatog. B 722:55–69. Becker, F.F, X.-B.Wang, Y.Huang, R.Pethig, J.Vykoukal, and P.R.C.Gascoyne. 1995. Separation of human breast cancer cells from blood by differential dielectric affinity. Proc. Nat. Acad. Sci. U.S.A. 29:860–864. Bello, M.S., L.Capelli, and P.G.Righetti. 1994. Dependence of the electroosmotic mobility on the applied electric field and its reproducibility in capillary electrophoresis. J. Chromatog. 684:311–322. Block, S.M. 1992. Making light work with optical tweezers. Nature 360:493–495.

Biochip technology

172

Carlson, R.H., C.Gabel, S.Chan, and R.H.Austin. 1998. Activation and sorting of human white blood cells. Biomed. Microdevices 1:39–48. Cheng, J., L.J.Kricka, E.L.Sheldon, and P.Wilding. 1998a. Sample preparation in microstructured devices. Topics Curr. Chem. 194:215–232. Cheng, J., E.L.Sheldon, L.Wu, M.J.Heller, and J.P.O’Connell. 1998b. Isolation of cultured cervical carcinoma cells mixed with peripheral blood cells on a bioelectronic chip. Anal. Chem. 70:2321–2326. Cheng, J., E.L.Sheldon, L.Wu, A.Uribe, L.O.Gerrue, J.Carrino, M.J.Heller, and J.P. O’Connell. 1998c. Preparation and hybridization analysis of DNA/RNA from E. coli on microfabricated bioelectronic chips. Nature Biotechnol. 16:541–546. De Gasperis, G., J.Yang, F.F.Becker, P.R.C.Gascoyne, and X.-B.Wang. 1999. Microfluidic cell separation based on generalized dielectrophoresis field flow fractionation. Biomed. Microdevices 2:41–49. Dev, S.B., and G.A.Hofmann. 1996. Clinical applications of electroporation. In Electrical Manipulation of Cells. P.T.Lynch and M.R.Davey, eds. Chapman & Hall, London, pp. 157–184. Engelhardt, H., H.Gaub, and E.Sakmann. 1984. Viscoelastic properties of erythrocyte membranes in high-frequency electric fields. Nature 307:378–380. Fiedler, S., Th. Schnelle B.Wagner, and G.Fuhr. 1995. Electrocasting: Formation and structuring of suspended microbodies using a.C. generated field cages. Microsyst. Technol. 2:1–7. Fiedler, S., S.G.Shirley, Th. Schnelle, and G.Fuhr. 1998. Dielectrophoretic sorting of particles and cells in a microsystem. Anal. Chem. 70:1909–1915. Foster, K., F.A.Sauer, and H.P.Schwan. 1992. Electrorotation and levitation of cells and colloidal particles. Biophys. J. 63:180–190. Fuhr, G., and R.Hagedorn. 1996. Cell electrorotation. In Electrical Manipulation of Cells. P.T. Lynch and M.R.Davey, eds. Chapman & Hall, London, pp. 37–70. Fuhr, G., and S.G.Shirley. 1998. Biological applications of microstructures. Topics Curr. Chem. 194:83–116. Fuhr, G., W.M.Arnold, R.Hagedorn, T.Müller, W.Benecke, B.Wagner, and U.Zimmermann. 1992. Levitation, holding and rotation of cells within traps made by high-frequency fields. Biochim. Biophys. Acta 1108:215–233. Fuhr, G., S.Fiedler, T.Müller, Th. Schnelle, H.Glasser, T.Lisec, and B.Wagner. 1994a. Particle micromanipulator consisting of two orthogonal channels with traveling-wave electrode structures. Sens. Actuators A 41:230–239. Fuhr, G., Th. Schnelle, and B.Wagner. 1994b. Traveling wave-driven microfabricated electrohydrodynamic pumps for liquids. J. Micromech. Microeng. 4:217–226. Fuhr, G., Th. Schnelle, R.Hagedorn, S.G.Shirley. 1995a. Dielectrophoretic field cages: Technique for cell, virus and macromolecule handling. Cell. Eng., Autumn, pp. 47–57. Fuhr, G., Th. Schnelle, T.Müller, and H.Glasser, T.Lisec, and B.Wagner. 1995b. Positioning and manipulation of cells and microparticles using miniaturized electric field traps and traveling waves. Sens. Materials. 7:131–146. Gascoyne, P.R.C., Y.Huang, R.Pethig, J.Vykoukal, and F.F.Becker. 1992. Dielectrophoretic separation of mammalian cells studied by computerized image analysis. Meas. Sci. Technol 3:439–445.


173

Gascoyne, P.R.C., Y.Huang, X.Wang, J.Yang, G.De Gasperis, and X.-B.Wang. 1996. Cell separation by conventional dielectrophoresis combined with field-flowfractionation Biophys. J. 70:A233. Gascoyne, P.R.C., X.-B.Wang, Y.Huang, and F.F.Becker. 1997. Dielectrophoretic separation of cancer cells from blood. IEEE Trans. Ind. Appl 33:670–678. Giaever, L., and C.R.Keese. 1991. Micromotion of mammalian cells measured electrically. Proc. Natl. Acad. Sci. U.S.A. 88:7896–7900. Giaever, L., and C.R.Keese. 1992. Toxic? Cells can tell. Chemtechnology (Feb), pp. 116– 125. Giaever, L., and C.R.Keese. 1993. A morphological biosensor for mammalian cells. Nature 366:591–592. Gimsa, J., R.Glaser, and G.Fuhr. 1991a. Theory and application of the rotation of biological cells in rotating electrical fields (electrorotation). In Physical Characterization of Biological Cells. W.Schutt, H.Klinkman, and I.Lamprecht, eds. Verlag Gesundheit, Berlin, pp. 295–323. Gimsa, J., P.Marszalek, U.Loewe, and T.Y.Tsong. 1991b. Dielectrophoresis and electrorotation of neurospora slime and murine myeloma cells. Biophys. J. 60:749– 760. Gimsa, J., T.Müller, T.Schnelle, and G.Fuhr. 1996. Dielectric spectroscopy of single human erythrocytes at physiological ionic strength: Dispersion of the cytoplasm. Biophys. J. 71:495–506. Goater, A.D., J.P.H.Burt, and R.Pethig. 1998. A combined traveling wave dielectrophoresis and electrorotation device applied to the concentration and viability determination of Cryptosporidium. J. Phys. D: Appl. Phys. 30:L65-L69. Golovanov, M.V. 1994. Electrophoresis of cells at a physiological ionic strength. In Cell Electrophoresis. J.Bauer, ed. CRC Press, Boca Raton, FL, pp. 181–196. Green, N.G., and H.Morgan. 1997. Dielectrophoretic separation of nano-particles. J. Phys. D: Appl. Phys. 30:L41-L44. Handgretinger, R., P.Lang, M.Schumm, G.Taylor, S.Neu, E.Koscielnak, D.Niethammer, and T.Klingebiel. 1998. Isolation and transplantation of autologous peripheral CD34+ progenitor cells highly purified by magnetic-activated cell sorting . Bone Marrow Transplant. 21:987–993. Hagedorn, R., G.Fuhr, T.Müller, and J.Gimsa. 1992. Traveling-wave dielectrophoresis of microparticles. Electrophoresis 13:49–54. Hawkes, J.J., G.P.Archer, and W.B.Betts. 1993. A dielectrophoretic spectrometer for characterizing microorganisms and other particles. Microbios 73:81–86. Hölzel, R., and I.Lamprecht. 1991. Dielectric properties of yeast cells as determined by electrorotation. Biochim. Biophys. Acta 1104:195–200. Huang, Y., and R.Pethig. 1991. Electrode design for negative dielectrophoresis. Meas. Sci. Technol 2:1142–1146. Huang, Y., R.Hölzel, R.Pethig, and X.-B.Wang. 1992. Differences in the AC electrodynamics of viable and non-viable yeast cells determined through combined dielectrophoresis and ROT studies. Phys. Med. Biol 37:1499–1517. Huang, Y., X.-B.Wang, J.A.Tame, and R.Pethig. 1993. Electrokinetic behaviour of colloidal particles in traveling electric fields:studies using yeast cells. J. Phys. D: Appl.

Biochip technology

174

Phys. 26:1528–1535. Huang, Y., X.-B.Wang, F.F.Becker, and P.R.C.Gascoyne. 1995. Electrorotational studies of the cytoplasmic dielectric properties of Friend murine erythroleukemia cells. Phys. Med. Biol. 40:1789–1806. Huang, Y., X.-B.Wang, F.E.Becker, and P.R.C.Gascoyne. 1996. Membrane changes associated with the temperature-sensitive P85gag-mos-dependent transformation of rat kidney cells as determined from dielectrophoresis and electrorotation. Biochim. Biophys. Acta 1282:76–84. Huang, Y., X.-B.Wang, P.R.C.Gascoyne, and F.F.Becker. 1997. Introducing dielectrophoresis as a new force field for field-flow-fractionation. Biophys. J. 73:1118– 1129. Huang, Y., X.-B.Wang, P.R.C.Gascoyne, and F.F.Becker. 1999a. Membrane dielectric responses of human T-lymphocytes following mitogenic stimulation. Biochim. Biophys. Acta 1417:51–62. Huang, Y., J.Yang, X.-B.Wang, F.E.Becker, and P.R.C.Gascoyne. 1999b. Purging breast cancer cells from hematopoietic cells by dielectrophoretic field-flow-fractionation. J. Hematotherapy 8:481–490. Hughes, M.P., H.Morgan, F.J.Rixon, J.P.H.Burt, and R.Pethig. 1998. Manipulation of herpes simplex virus type 1 by dielectrophoresis. Biochim. Biophys. Acta 1425:119– 126. Irimajiri, A., T.Hanai, and A.Inouye. 1979. A dielectric theory of “multi-stratified shell” model with its application to a lymphoma cell. J. Theor. Biol. 78:251–269. Jones, T.B. 1995. Electromechanics of Particles. Cambridge University Press, Cambridge. Jones, T.B., and G.W.Bliss. 1977. Bubble dielectrophoresis. J. Appl. Phys. 48:1412– 1417. Jones, T.B., and M.Washizu. 1996. Multipolar dielectrophoretic and electrorotation theory. J. Electrostal. 37:121–129. Kaler, K.V.I.S., and T.B.Jones. 1990. Dielectrophoretic spectra of single cells determined by feedback-controlled levitation . Biophys. J. 57:173–182. Kaler, K.V.I.S., J.Xie, T.B.Jones, and R.Paul. 1992. Dual-frequency dielectrophoretic levitation of canola protoplasts Biophys. J. 63:58–69. Lo, C.-M.,, C.R.Keese, and I.Giaever. 1995. Impedance analysis of MDCK cells measured by electric cell-substrate impedance sensing. Biophys. J. 69:2800–2807. Markx, G.H., and R.Pethig. 1995. Dielectrophoretic separation of cells: Continuous separation. Biotechnol. Bioeng. 45:337–343. Markx, G.H., Y.Huang, X.-R.Zhou, and R.Pethig. 1994. Dielectrophoretic characterization and separation of micro-organisms. Microbiology 140:585–591. Markx, G.H., J.Rousselet, and R.Pethig. 1997. DEP-FFF: Field-flow fractionations using non-uniform electric fields. J. Liq. Chromatog. Rel. Technol. 220:2857–2872. Masuda, S., M.Washizu, and M.Iwadare. 1989. Separation of small particles suspended in liquid by nonuniform traveling field. IEEE Trans. Ind. Appl. 23:474–480. Melcher, J., and M.S.Firebaugh. 1967. Traveling-wave bulk electroconvection induced across a temperature gradient. Phys. Fluids 11:1178–1185. Melcher, J., and G.Taylor. 1969. Electrohydrodynamics: A review of the role of


175

interfacial shear stresses. Annu. Rev. Fluid Mech., pp. 111–146. Morgan, H., N.G.Green, M.P.Hughes, W.Monaghan, and T.C.Tan. 1997. Large-area traveling-wave dielectrophoresis particle separator. J. Micromech. Microeng. 7:65–70. Morgan, H., M.P.Hughes, and N.G.Green. 1999. Separation of sub-micron bio-particles by dielectrophoresis. Biophys J. 77:516–525. Müller, T., W.M.Arnold, Th. Schnelle, R.Hagedorn, G.Fuhr, and U.Zimmermann. 1993. A traveling-wave micropump for aqueous solution: Comparison of 1 g and µg results. Electrophoresis 14:764–772. Müller, T., G.Gradl, S.Howitz, S.Shirley, Th. Schnelle, and G.Fuhr. 1999. A 3D microelectrode system for handling and caging single cells and particles. Biosens. Bioelectron. 14:247–256. Ohshima, T., T.Qno, and M.Sato. 1998. Digestion of nucleic acid molecules in pulsed electric field and its release from recombinant E. coli. ESA-IEJ Joint Symp. Electrostatics 1998, pp. 334–343. Pauly, H., and H.R.Schwan. 1959. Über die Impedancz einer Suspension von kugelförmigen Teilchen mit einer Schale (Ein Modell für das dielektrische Verhälten von Zellsuspensionen und von Proteinlösungen) Z. Naturforsch. 14b:125–131. Pethig, R., Y.Huang, X.-B.Wang, and J.P.H.Burt. 1992. Positive and negative dielectrophoretic collection of colloidal particles using interdigitated castellated microelectrodes J. Phys. D: Appl. Phys. 25:881–888. Pohl, H. 1978. Dielectrophoresis. Cambridge University Press, Cambridge. Sale, A.J.H., and W.A.Hamilton. 1968. Effects of high electric fields on microorganisms, III: Lysis of erythrocytes and protoplasts. Biochim. Biophys. Acta 163:37– 43. Schnelle, T., R.Hagedorn, G.Fuhr, S.Fiedler, and T.Müller. 1993. Three-dimensional electric field traps for manipulation of cells: Calculation and experimental verification. yBiochim. Biophys. Acta 1157:127–140. St. Claire III, R.L. 1996. Capillary electrophoresis. Anal. Chem. 68:569R-586R. Stephens, M., M.S.Talary, R.Pethig, A.K.Burnett, and K.I.Mills. 1996. The dielectrophoresis enrichment of CD34+ cells from peripheral blood stem cell harvests. Bone Marrow Transplant. 18:777–782. Sukhorukov, V.L., and U.Zimmermann. 1996. Electrorotation of erythrocytes treated with dipicrylamine: Mobil charges within the membrane show their signature in rotational spectra. J. Membr. Biol 151:161–169. Sukhorukov, V.L., W.M.Arnold, and U.Zimmermann. 1993. Hypotonically induced changes in the plasma membrane of cultured mammalian cells. J. Membr. Biol. 132:27–40. Sukhorukov, V.L., H.Mussauer, and U.Zimmermann. 1998. The effect of electrical deformation forces on the electropermeabilization of erythrocyte membranes in lowand high-conductivity media. J. Membr. Biol. 163:235–245. Talary, M.S., J.P.H.Burt, J.A.Tame, and R.Pethig. 1996. Electromanipulation and separation of cells using traveling electric fields. J. Phys. D: Appl. Phys. 29:2198– 2203. Tsong, T.Y. 1990. On electroporation of cell membranes and some related phenomena. Bioelectrochem. Bioenerget. 24:271–295.

Biochip technology

176

Villas, B.H. 1998. Flow cytometry: An overview. Cell Vision 5:56–61. Wang, X.-B., Y.Huang, J.P.H.Burt, G.H.Markx, and R.Pethig. 1993a. Selective dielectrophoretic confinement of bioparticles in potential energy wells. J. Phys. D: Appl. Phys. 26:1278–1285. Wang, X.-B., Y.Huang, R.Hölzel, J.P.H.Burt, and R.Pethig. 1993b. Experimental investigations of interdependence of dielectric, dielectrophoretic and electrorotational responses of polarised particles J. Phys. D: Appl. Phys. 26:312–322. Wang, X.-B., Y.Huang, F.F.Becker, and P.R.C.Gascoyne. 1994a. A unified theory of dielectrophoresis and traveling-wave dielectrophoresis J. Phys. D: Appl. Phys. 27:1571–1574. Wang, X.-B., Y.Huang, P.R.C.Gascoyne, F.F. Becker, R.Hölzel, and R.Pethig. 1994b. Changes in Friend murine erythroleukaemia cell membranes during induced differentiation determined by electrorotation. Biochim. Biophys. Acta 1193:330–344 Wang, X., X.-B.Wang, and P.R.C.Gascoyne. 1997a. General expressions for dielectrophoretic force and electrorotational torque derived using the Maxwell stress tensor method. J. Electrostal. 39:277–295. Wang, X.-B., Y.Huang, F.F.Becker, and P.R.C.Gascoyne. 1997b. Dielectrophoretic manipulation of particles. IEEE/IAS Trans. 33:660–669. Wang, X.-B., Y.Huang, X.Wang, F.F.Becker, and P.R.C.Gascoyne. 1997c. Dielectrophoretic manipulation of cells using spiral electrodes. Biophys. J. 72:1887– 1899. Wang, X., J.Yang, and P.R.C.Gascoyne. 1999a. Role of peroxide in AC electrical field exposure effects on Friend murine erythroleukemia cells during dielectrophoretic manipulations. Biochim. Biophys. Acta 1426:53–68. Wang, X.-B., J.Yang, Y.Huang, J.Vykoukal, F.F.Becker, and P.R.C.Gascoyne. 2000. Cell separation by dielectrophoretic field-flow-fractionation. Anal. Chem.. 72:832–839. Washizu, M., S.Suzuki, O.Kurosawa, T.Nishizaka, and T.Shinohara. 1994. Molecular dielectrophoresis of biopolymers. IEEE Trans. Ind. Appl 30:835–843. Washizu, M.S., T.Nanba, and S.Masuda. 1990. Handling of biological cells using fluid integrated circuit. IEEE Trans. Ind. Appl. 26:352–358. Weigl, B.H., and P.Yager. 1999. Microfluidic diffusion-based separation and detection. Science 283:346–347. Wilding, P., L.J.Kricka, J.Cheng, G.Hvichia, M.A.Shoffner, and P.Fortina. 1998. Preparation and hybridization analysis of DNA and RNA from E. coli on microfabricated bioelectronic chips. Anal Biochem. 257:95–100. Wong, T-K., and E.Neumann. 1982. Electric field mediated gene transfer. Biochem. Biophys. Res. Comm. 107:584–587. Yang, J., Y.Huang, X.Wang, X.-B.Wang, F.F.Becker, and P.R.C.Gascoyne. 1999a. Dielectric properties of human leukocyte subpopulation determined by electrorotation as a separation criteria. Biophys. J. 76:3307–3314. Yang, J., Y.Huang, X.-B.Wang, F.F.Becker, and P.R.C.Gascoyne. 1999b. Cell separation on microfabricated electrodes using dielectrophoretic/gravitational field-flowfractionation. Anal. Chem. 71:911–918. Yang, J., Y.Huang, X.-B.Wang, F.F.Becker, and P.R.C.Gascoyne. 2000. Human leukocyte differential analysis by dielectrophoretic field-flow-fractionation. Biophys. J.


177

78:2680–2689. Yeh, S.R., M.Seul, and B.I.Shraiman 1997. Assembly of ordered colloidal aggregates by electric field-induced fluid flow. Nature 386:57–59. Zubrisky, E. 1999. Sorting cells. Anal. Chem. 71:171A.

8 Microfilter-Based Separation of Cells Paolo Fortina, Larry J.Kricka, and Peter Wilding

INTRODUCTION The development of micron-sized components that can perform analytical operations including separation and isolation of specific living cell types is one of the major aims of current research in system integration of biological assays (Anderson et al., 1999; Burns et al., 1998; Burke et al., 1997; Cheng et al., 1996, 1998a; Harrison et al., 1993; He et al., 1999; Kricka, 1998a; Northrup et al., 1998; Schmalzing et al., 1997; Waters et al., 1998; Wilding et al., 1998). Microchips containing microfabricated silicon filters have been developed and shown to be effective for isolation of white blood cells (Wilding et al., 1998). The objective of this chapter is to provide the most recent results in the field of cell separation using microfilter-based devices. It is anticipated that future modifications in size and geometry of the filters will permit precise selection of different-sized cells while further development should increase cell isolation yields. Finally, selection and isolation of rare cell types such as fetal cells in maternal peripheral blood, as well as cancer cells, may be possible using specific capture agents included within microfilterbased chips. Alternative microchip-based methods in which mechanical separation is replaced with an electronically driven system will also be illustrated (Cheng et al., 1998b,c; Hughes et al., 1998; Pethig and Markx, 1997; Stephens et al., 1996; Yang et al., 1999).

THE SHRINKING LABORATORY The exponential increase in genomic sequence data has given investigators the potential for identifying genes, determining their functions, characterizing mutations, and relating these changes to disease development at an accelerated pace (Lander, 1996). As a result, this deluge of information has generated demand for novel DNA technologies that will permit high-throughput analytical systems for the analysis of thousands of biological samples simultaneously (Kricka, 1998b; Marshall and Hodgson, 1998; Ramsay, 1998; Regnier et al., 1999; Robertson, 1998; Service, 1995, 1998a,b). In addition, such socalled highly parallel processes will permit analysis of multiple loci concurrently (Brown and Botstein, 1999; Duggan et al., 1999; Graves, 1999; Hacia, 1999; Lipshutz et al., 1999; McKenzie et al., 1998). At present, this necessity is partially addressed by employing different pieces of hardware, which automate and expedite the manual steps required for genetic analysis (Winn-Deen, 1994). More recently, semiconductor manufacturing techniques have been increasingly employed to miniaturize the array of

Biochip technology

180

equipment used in conventional genetic assays, including cell sorting, nucleic acid amplification, hybridization, electrophoresis, and detection (Effenhauser, 1998; van den Berg and Lammerink, 1998; Qin et al., 1998). Subsequently, the different modules have been assembled onto a microfluidic platform in order to connect and integrate the different functions (Harrison et al., 1993; Jacobson and Ramsey, 1996; Waters et al., 1998; Woolley et al., 1996). These microstructures are expected to have a major impact in genome-based studies, clinical molecular diagnosis, and in food, water, and environmental testing. The principal benefits of providing automated, integrated, and portable analyzers are convenience and simplicity of operation due to the reduction in the number of analytical steps (Schmalzing et al., 1998). The reduced manufacturing costs possible for mass-produced microscale devices may also lead to lower assay costs. Eventually, chip-based integrated microfluidic devices will allow those involved in genome analysis to go from a drop of blood to a genetic profile, quickly and inexpensively. However, although the individual analytical steps required for routine biological procedures have been performed, tested, and demonstrated to be effective, the greatest bottleneck in DNA analysis on microchips remains sample preparation, which involves the steps of cell separation, cell manipulation, and cell isolation.

CELL SEPARATION OPTIONS A variety of techniques for cell separation are routinely employed in the macroscopic world, and these techniques can be divided into three major categories based on: (1) physical differences between cell populations, (2) cell surface properties, and (3) functional differences between cell populations. Physical properties that can be exploited include density (continuous/discontinuous density gradients), size (velocity sedimentation), size filtration chromatography, and measurement of electrical impedance. Separation methods based on differences in charge include electrophoresis, phase partition, and dielectrophoresis. Other physical methods include those that utilize differences in the optical and magnetic properties of the cells and are achieved via magnetic and/or fluorescence- activated cell sorting. Methods of cell separation based on the properties of the cell surface permit a high degree of purification and include methods that exploit adherence and antibody-defined surface markers. The extensive and growing list of monoclonal antibodies with specificity for cell surface markers provides a broad scope for immunological separation strategies. Finally, cell separation can also be achieved on the basis of functional properties; however, few of the existing methods based on differences in cell function are in routine use. For a complete review of cell separation on a macro scale, the reader is directed elsewhere (Kumar and Lykke, 1984). Microsystems Micro-total analytical systems (µ-TAS) or lab-on-a-chip devices have progressed rapidly from concept to prototype devices. Recent work has demonstrated the scope of integration possible for microfabricated devices. Volumes of sample analyzed are usually

Microfilter-based separation of cells

181

in the microliter range, but this is being successively reduced—for example, a nanoliter DNA analytical device has been described that combines fluidic channels, heaters, temperature sensors, and fluorescence detectors for amplification or digestion of DNA samples coupled with separation and detection of products (Burns et al., 1998). Other groups have produced plastic cartridges that perform a series of processes including DNA extraction, amplification, and delivery of products to a microarray (Service, 1998a,b). Effective implementation of methods by which different cell types are discriminated and selectively isolated on a microfabricated analytical module is critical for the development of effective integrated microfluidic devices for genetic analyses. Human blood contains white blood cells (WBCs) with sizes ranging from 5 to 15 µm, red blood cells (RBCs) with a size of less than 7 µm, and plasma. One microliter of human blood normally contains an average of 5×103 WBCs and 5×106 RBCs. The main objective of microfilter design is isolation of WBCs from whole blood as part of sample preparation for nucleic acid amplification. Although the nucleic acid content of a single WBC is sufficient as template for an amplification-based assay (e.g., polymerase chain reaction (PCR)) filtration of spherical WBCs and discoid RBCs is influenced by several variables. For instance, while the deformable WBCs can be retarded by filters with gaps of 7 µm, RBCs can easily align them to pass through a 3.0–3.5 µm gap in a filter bed. An interesting consideration is that WBCs vary in size, type, and frequency. If an investigator is interested in genomic DNA, the only concern is to obtain an abundance of nucleated cells in order to perform PCR amplification. However, RBC contamination may become a limiting factor in an amplification-based assay since hemoglobin inhibits polymerase activity. Therefore, in terms of WBC isolation, efficiency is essential to achieve an adequate number of cells, which translates into approximately a 10% yield or 500 WBCs isolated from a 1.0-µL sample of whole blood containing approximately 5000 WBCs and 50,000 RBCs. In other words, a reliable microfiltration system must be able to perform WBC isolation with at least 10% efficiency as well as being able to concurrently eliminate 99% of contaminating RBCs. Currently available filter chips meet these requirements (Wilding et al., 1998). Microfabrication and Microfilters A range of microfilter geometries and designs has been investigated, including simple arrays of posts (Figure 1A,B, see Color Plate 8.1), comb-shaped filters (Figure 2A,B, see Color Plate 8.2), and weir-type filters (Figure 3A-C, see Color Plate 8.3) (Kricka et al., 1993, 1997; Wilding et al., 1994, 1998). A filter chip is a miniaturized sealed filtration chamber with an inlet and outlet for filling and emptying. A typical 15×18 mm and 400 µm thick device is micromachined from silicon using conventional photolithographic techniques. The steps required include CAD generation of the photomask, which defines the structures to be etched, oxidation

Biochip technology

182

Figure 1. See Color Plates 8.1 A,B. Micropost-type silicon filters. (A) Offset array of simple microposts (13 µm×20 µm spaced 7 µm apart) set across a 500 µm wide×20 µm deep silicon channel. (B) Isolation of 5.78 µm diameter latex microspheres by a post-type filter (5µm channels between 73 µm wide posts set across a 500 µm wide 5.7 µm deep channel). Reprinted with permission from Wilding et al. (1998).

Figure 2. See Color Plates 8.2A.B. Comb-type silicon microfilters. (A) Filter formed from an array of 120 posts (175 µm long×18 µm wide) separated by 6-µm channels set across a 3 mm wide×13 µm deep silicon channel. (B) New methylene-blue-stained white blood cells isolated by a comb-type filter (cells released from the front surface (upper) of the filter by reversing the flow through the filter). Reprinted with permission from Wilding et al. (1998). of a silicon wafer to produce a surface layer of silicon dioxide (2000 Å), coating of the oxidized wafer with photoresist using a spin-coating technique, and, finally, patterning the resist using the photomask and a source of UV light. For a positive resist, areas of the resist exposed to light are solubilized and then removed by washing. The unexposed resist defines the pattern of the structures to be etched, and the exposed oxide surface is


183

etched using hot potassium hydroxide (KOH). Subsequently, the photoresist is stripped off the wafer. The whole process is then repeated to etch further features such as inlet and outlet ports as well as holes through the silicon, if required. The silicon wafer is then diced into single chips and pieces of Pyrex glass are cut to size and placed on top of the silicon chip to form a pre-sealed chamber. The assembled chip and Pyrex glass cap are placed on a hot plate connected to a power supply. An electrode is contacted with the glass cover and a bond forms between the glass and silicon surfaces to produce the readyto-use microfilter chip.

Figure 3. See Color Plates 8.3A–3C. Weir-type silicon microfilters. (A) Schematic of weir-type filter A 3.5µm gap between the top of the etched silicon dam and the Pyrex glass cover provides active filtration of cells based on size. (B) Stained white cells filtered by a weirtype filter The cells are trapped on top of the filter beneath the underside of the glass cover on the chip. (C) Integrated filter PCR chip based on linear weir-type filter in the PCR chamber. Reprinted with permission from Wilding et at. (1998). Early filter chip designs consisted of a channel or chamber containing 1–2 rows of flow deflectors upstream from different-sized filters ranging from 5 to 20 µm (Figure 4,

Biochip technology

184

see Color Plate 8.4). The role of the flow deflectors is to distribute the diluted blood sample into the entire channel that leads from the inlet to the filter bed. Larger-sized filters were designed to remove small amounts of cell debris and prevent clogging of the smaller filters. The smaller filters were designed to only allow human red blood cells and plasma to pass through while retaining WBCs. On-chip studies have demonstrated that lysing cells on microfilters to remove cell debris and isolate nucleic acids is problematic for two reasons. First, lysed cells generate a large quantity of cell debris that can easily clog microfilters, therefore making nucleic acid isolation difficult. Second, materials released from lysed red and white cells may inhibit enzymes such as Taq DNA polymerase, DNA ligase, or restriction endonuclease used in subsequent analytical reactions. Attempts to make

Figure 4. See Color Plate 8.4. Filter chip with three test channels containing different designs of flow deflector and serial filters. Reprinted with permission from Wilding et al. (1998). more complex filter designs by wet-etching narrower gaps have been problematic as the undercutting of the photoresist during the etching processes tends to compromise the integrity of the filter beds. One of the most effective filter chip designs is based on a silicon weir. In a weir-type filter, a narrow micrometer-sized gap between a silicon dam fabricated across the entire width of a channel or chamber and the glass cover acts as the cell filter (Figure 3A, see Color Plate 8.3). The active part of the microfilter is a 3.5-µm gap between the top of the silicon dam and the glass used to cap the chip. Each chip contains a series of filters (filter beds) of sufficient capacity (length or weir) (Figure 3C, see Color Plate 8.3) in order to ensure adequate yield of isolated WBCs that approximates 1200 from 2 µL of a 1:1 dilution of whole human blood. In human blood there are populations of white cells with differing sizes. In adults, approximately 34% of white cells are lymphocytes (6–15 mm in diameter), and these are smaller than the polymorphonuclear cells (e.g., neutrophils, average diameter 12 µm; eosinophils, average diameter 13 µm) and monocytes (14–20 µm) (Lentner, 1981). Assuming that smaller white cells are not retained by weir-type filters (Figure 3B, see Color Plate 8.3), then the approximate isolation yield is 6.8%. In subsequent studies, WBC isolation yields ranged


185

from 4 to 15% in a procedure that could be completed in less than 8 minutes. Separation and isolation of specific cell subsets or microorganisms using microfilterbased chips is also under extensive investigation. Important applications include bone marrow transplantation, antenatal diagnosis, and testing food and water supplies, as well as detecting and identifying biowarfare agents (Simpson and Elias, 1994). However, the current yield of nucleated cells of 15% from a 1–2-µL sample of whole blood makes isolation of rare cells problematic, at least in humans. In the past, affinity methods based on antibodies for cell-surface markers immobilized on the surface of microbeads have been employed. For instance, human sperm cells were isolated in microchannel-type filters using anti-human IgG-coated microbeads (Kricka et al., 1998b). The same principle could be applied for enrichment of other entities from bulk fluid, such as the isolation of cancer cells, trophoblast cells in peripheral maternal blood, or bacteria. The combination of microfiltration technology and immunologically based preconcentration methods such as latex beads, agarose, or magnetic beads with attached target sample-specific antibody may contribute to efficient sample preparation prior to the amplification reaction in a microchip. Cell selection could be achieved by adapting a separation protocol based on molecular recognition of the specific cell subset using monoclonal antibodies. Monoclonal antibodies for white cell subsets are well characterized and available commercially. These antibodies are also available bound to magnetic microspheres (2.8–5 µm in diameter). Magnetic beads can then be manipulated and moved around within a microstructure using a micromagnet positioned against the surface of the chip, providing an additional level of control of reactants within the microchip. Others Applications of Microfilters Solvents and reagents can sometimes contain particulate impurities or be contaminated with bacteria, and these materials can clog microstructures. To address this issue, solvent and reagent filters have been micromachined into quartz wafers using deep reactive ion etching to create a network of intersecting 1.5-by-10-micron channels. When placed at the bottom of reservoirs with a side exit, this channel network behaves as a lateral percolation filter composed of an array of cube-like structures one layer deep. Flow through these filters is driven by electroosmotic flow. Silanol groups at the walls of channels in the network provide the requisite charge to trigger electroosmotic flow when a voltage is applied laterally to the filter. Adsorption of cationic proteins in this silanolrich matrix is controlled by the application of a polyacrylamide coating prepared by bonding N-hydroxysuccinimide (NHS)-activated poly (acrylic acid) to (γ-aminopropyl) silane-derivatized filters. Subsequent reaction of residual NHS groups in the coating with 2-(2-aminoethoxy) ethanol provided channels of low charge density and adsorptivity. These lateral percolation filters have been shown to be effective in filtering solvents containing a variety of particulate materials, ranging from dust to cells (He et al., 1999). Studies of microfilter-based cell separation has produced unexpected findings on the properties of blood cells. Microfabricated channels have been used to demonstrate that the mobility of RBC under physiological flow conditions correlates with calcium concentration in the cells. Therefore, as calcium is known to be the molecular trigger of

Biochip technology

186

smooth muscle action, the finding indicates that, contrary to previous thinking, the cytoskeleton of RBC is in fact an active structure. The hydrodynamic assays created using silicon-based microchannels provided a new perspective on human RBCs (Brody et al., 1995). In a similar study, silicon-fabricated arrays of channels with dimensions similar to those of human capillaries demonstrated novel activation behavior of human leukocytes. The mobility of leukocytes is greatly reduced by passage through the channels, which eventually clogs the channel. Studies of chemotaxis and locomotion of WBCs in a controlled confined environment generated by microfabrication is just one novel approach, which may eventually lead to better means for cell isolation (Carlson et al., 1997, 1998). Electroosmotic and/or electrophoretic pumping have also been used to drive cell transport within a network of capillary channels (15×55 micron cross-sections) etched in silicon. Whole cells such as Saccharomyces cerevisiae, canine erythrocytes, and E. coli were employed in these studies. At an intersection within the chip, canine erythrocytes were mixed with the lysing agent, sodium dodecyl sulfate, to demonstrate that cell selection and subsequent reactions can be accomplished within the microchip (Li and Harrison, 1997). Microelectrode-Based Separation Alternative microchip-based methods in which mechanical separation is replaced with an electronically driven system are also gaining importance. Dielectrophoresis is the motion of particles—including cells, viruses, bacteria, proteins, and nucleic acids—determined by electrical polarization effects in nonuniform electric fields. The motion is determined by the magnitude and polarity of the charges induced by the applied field in the particle, which becomes an electrical dipole. Unlike electrophoresis, the particles do not need a net electrical charge for motion to occur, and alternating current fields of a wide range of frequencies are used instead of direct current or a low-frequency homogeneous electrical field (Pethig and Markx, 1997). Dielectrically polarizable particles are subject to the dielectrophoretic forces as long as the effective polarizability of the particles is different from that of the surrounding medium (Fuhr and Shirley, 1998). A living cell is a complex-structured particle with several dielectric interfaces, such as the external medium and the cell membrane, as well as the cell membrane itself and the cytoplasm. Forces that develop at each interface have different frequency dependencies generating a force spectrum. Semiconductor fabrication technology has successfully produced electrode systems on the micrometer scale, allowing precise cell handling system to be a constructed. Exploiting differential dielectric properties, dielectrophoresis has been used to enrich selected cell subpopulations in mixed cell populations, as well as separation of mixtures of bacteria, viable and nonviable cells, cancerous and normal cells, and RBCs and WBCs. Enrichment of stem cells expressing the CD34+ antigen has been achieved for bone marrow samples and peripheral blood. Improved methods such as dielectrophoretic field-flow fractionation have also been developed further improving the cell separation process (Hughes et al., 1998; Stephens et al., 1996; Thalary et al., 1995; Yang et al., 1999). Electronic biochips have also been developed to efficiently isolated cells such as


187

cultured cervical carcinoma cells from whole human blood, as well as a variety of microorganisms including E. coli, Micrococcus lysodeikticus, and Staphylococcus epidermidis. Bioelectronic chips are composed of an addressable array of 25 platinumbased 80-µm-diameter microelectrodes covered by an agarose permeation layer confined in a 4.84-µL flow cell (Cheng et al., 1998b,c). By means of dielectrophoresis, and therefore based on differences in charge and size, an alternating current field is set up within the chip to separate different cells within the sample, which zone to different regions of the microelectrode-based array. Although the method is proprietary, the device holds promise for successful cell enrichment, nucleic acid purification, and sequence specific amplification.

CONCLUSIONS In the area of cell separation it is clear that miniaturization will be a key technology. Extensive progress has been made using both simple microfabricated filters and microelectrodes. Examples of assay integration that include cell isolation, sample preparation, and mixing together with other analytical procedures have been demonstrated, thus laying the groundwork for a fully integrated micromachined nucleic acid analyzer.

ACKNOWLEDGMENTS This work was supported in part by NIH grants P60-HL38632, (PF, LK and PW) and NCI grant CA 78848-02 (LK and PW). We wish to thank our former colleague Dr. Jing Cheng.

REFERENCES Anderson, R.C, Lipshutz, R.J., Rava, R.P., and Fodor, S.P.A. 1999. Integrated nucleic acid diagnostic device. U.S. Patent 5,922,591. Brody, J.B., Han, Y, Austin, R.H., and Bitensky, M. 1995. Deformation and flow of red blood cells in a synthetic lattice: evidence for an active cytoskeleton. Biophys. J. 68:2224–2232. Brown, P.O., and Botstein, D. 1999. Exploring the new world of the genome with DNA microarrays. Nature Genet. 21(Suppl. l):33–37. Burke, D.T., Burns, M.A., and Mastrangelo, C. 1997. Microfabrication technologies for integrated nucleic acid analysis. Genome Res. 7:189–197. Burns, M.A., Johnson, B.N., Brahmasandra, S.N., Handique, K., Webster, J.R., Krishnan, M., Sammarco, T.S, Man, P.M., Jones, D., Heldsinger, D., Mastrangelo, C.H., and Burke, D.T. 1998. An integrated nanoliter DNA analysis device. Science 282:484–487. Carlson, R.H., Brody, J.P, Chan, S., Gabel, C., Winkleman, J., and Austin, R.H. 1997. Self-sorting of white blood cells in lattice. Phys. Rev. Lett. 79:2149–2152.

Biochip technology

188

Carlson, R.H., Gabel, C., Chan, S., and Austin, R.H. 1998. Activation and sorting of human white blood cells. J. Biomed. Microdevices 1:39–47. Cheng, J., Fortina, P., Surrey, S., Kricka, L.J., and Wilding, P. 1996. Microchip-based devices for molecular diagnosis of genetic diseases. Mol. Diagn. 1:183–200. Cheng, J., Kricka, L.J., Sheldon, E.L., and Wilding, P. 1998a. Sample preparation in microstructured devices Topics Curr. Chem. 194:216–231. Cheng, J., Sheldon, E.L., Wu, L., Heller, M.J., and O’Connell, J.P. 1998b. Isolation of cultured cervical carcinoma cells mixed with peripheral blood cells on a bioelectronic chip. Anal. Chem. 70:2321–2326. Cheng, J., Sheldon, E.L., Wu, L., Uribe, A., Gerrue, L.O., Carrino, J., Heller, M.J., and O’Connell, J.P. 1998c. Preparation and hybridization analysis of DNA/RNA from E. coli on microfabricated bioelectronic chips. Nature Biotechnol. 16:541–546. Duggan, D.J., Bittner, M., Chen, Y., Meltzer, P., and Trent, J.M. 1999. Expression profiling using cDNA microarrays . Nature Genet. 21 (Suppl. 1):10–14. Effenhauser, C.S. 1998. Integrated chip-based microcolumn separation systems Topics Curr. Chem. 194:51–82. Fuhr, G., and Shirley, S.G. 1998. Biological application of microstructures. Topics Curr. Chem. 194:83–116. Graves, D.J. 1999. Powerful tools for genetic analysis come of age. Trends Biotechnol. 17:127–134. Hacia, J.G. 1999. Resequencing and mutational analysis using oligonucleotide microarrays. Nature Genet. 21(Suppl. l):42–47. Harrison, D.J., Glavina, P.G., and Manz, A. 1993. Towards miniaturized electrophoresis and chemical analysis systems on silicon: An alternative to chemical sensors. Sens. Actuators B 10:107–116. He, B., Tan, L., and Regnier, F. 1999. Microfabricated filters for microfluidic analytical systems. Anal. Chem. 71:1464–1468. Hughes, M.P., Morgan, H., Rixon, F.J., Burt, J.P., and Pethig, R. 1998. Manipulation of herpes simplex virus type 1 by dielectrophoresis. Biochim. Biophys. Acta 1425:119– 126. Jacobson, S.C., and Ramsey, J.M. 1996. Integrated microdevice for DNA restriction fragment analysis. Anal Chem. 68:720–723. Kricka, L.J. 1998a. Revolution on a square centimeter. Nature Biotechnol. 16:513–514. Kricka, L.J. 1998b. Miniaturization of analytical systems. Clin. Chem. 44:2008–2014. Kricka, L.J., Ji, X., Nozaki, O., Heyner, S., Garside, W.T., and Wilding, P. 1993. Applications of a microfabricated device for evaluating sperm function. Clin. Chem. 39:1944–1947 Kricka, L.J., Faro, L., Heyner, S., Garside, W.T., McKinnon, G., and Wilding, P. 1997. Micromachined analytical devices: Microchips for semen testing. J. Pharm. Biomed. Anal. 15:1443–1447. Kumar, R.K., and Lykke, A.W. 1984. Cell separation: A review. Pathology 16:53–62. Lander, E.S. 1996. The new genomics: Global views of biology. Science 274:536–539. Lentner, C. 1981. Geigy Scientific Tables. S. Lentner, ed. Ciba-Geigy, Basel, Germany. Li, P.C.H., and Harrison, D.J. 1997. Transport, manipulation and reaction of biological cells on-chip using electrokinetic effects. Anal. Chem. 69:1564–1568.


189

Lipshutz, R.J., Fodor, S.P.A., Gingeras, T.R., and Lockhart, D. 1999. High density synthetic oligonucleotide arrays. Nature Genet. 21(Suppl. l):20–24. Marshall, A., and Hodgson, J. 1998. DNA chips: An array of possibilities. Nature Biotechnol. 16:27–31. McKenzie, S., Mansfield, E., Rappaport, E., Surrey, S., and Fortina, P. 1998. Parallel molecular genetic analysis. Eur. J. Hum. Genet. 6:417–429. Northrup, M.A., Bennett, B., Hadley, D., Landre, P., Lehew, S., Richards, J., and Stratton, P. 1998. A miniature analytical instrument for nucleic acids based on micromachined silicon reaction chambers. Anal. Chem. 70:918–922. Pethig, R., and Markx, G.H. 1997. Applications of dielectrophoresis in biotechnology. Trends Biotechnol. 15:426–432. Qin, D., Xia, Y., Rogers, J.A., Jackman, R.J., Zhao, X.-M., and Whitesides, G.M. 1998. Microfabrication, microstructures and microsystems. Topics Curr. Chem. 194:1–20. Ramsay, G. 1998. DNA chips: State-of-the-art. Nature Biotechnol. 16:40–44. Regnier, F.E., He, B., Lin, S., and Busse, J. 1999. Chromatography and electrophoresis on chips: Critical elements of future integrated, microfluidic analytical systems for life science. Trends Biotechnol. 17:101–106. Robertson, D. 1998. Chip advance but cost constraints remain. Nature Biotechnol. 16:509. Schmalzing, D., Koutny, L., Adourian, A., Belgrader, P., Matsudaira, P., and Ehrlich, D. 1997. DNA typing in thirty seconds with a microfabricated device. Proc. Natl. Acad. Sci. U.S.A. 94:10273–10278. Schmalzing, D., Adourian, A., Koutny, L., Ziaugra, L., Matsudaira, P., and Ehrlich, D. 1998. DNA sequencing on microfabricated electrophoretic devices. Anal. Chem. 70:2303–2310. Service, R.F. 1995. The incredible shrinking laboratory. Science 268:26–27. Service, R.F. 1998a. Microchip arrays put DNA on the spot. Science 282:396–399. Service, R.F. 1998b. Coming soon: The pocket DNA sequencer. Science 282:399–401. Simpson, J.L., and Elias, S. 1994. Isolating fetal cells in maternal circulation for prenatal diagnosis. Prenatal Diagn. 14:1229–1242. Stephens, M., Talary, M.S., Pethig, R., Burnett, A.K., and Mills, K.I. 1996. The dielectrophoresis enrichment of CD34+ cells from peripheral blood stem cell harvests. Bone Marrow Transplant. 18:777–882. Thalary, M.S., Mills, K.I., Hoy, T., Burnett, A.K., and Pethig, R. 1995. Dielectrophoretic separation and enrichment of CD34+ cell subpopulation from bone marrow and peripheral blood stem cells. Med. Biol Eng. Comput. 33:235–237. van den Berg, A., and Lammerink, T.S.J. 1998. Micro total analysis systems: Microfluidic aspects, integration concept and applications. Topics Curr. Chem. 194:21–49. Waters, L.C., Jacobson, S.C., Kroutchinina, N., Khandurina, J., Foote, R.S., and Ramsey, J.M. 1998. Microchip device for cell lysis, multiplex PCR amplification, and electrophoretic sizing. Anal. Chem. 70:158–162. Wilding, P., Pfahler, J., Bau, H.H., Zemel, J.N., and Kricka, L.J. 1994. Manipulation and flow of biological fluids in straight channels micromachined in silicon. Clin. Chem. 40:43–47.

Biochip technology

190

Wilding, P., Kricka, L.J., Cheng, J., Hvichia, G., Shoffner, M.A., and Fortina, P. 1998. Integrated cell isolation and PCR analysis using silicon microfilter-chambers. Anal. Biochem. 257:95–100. Winn-Deen, E.S. 1994. Molecular techniques and their automation in the clinical laboratory. In Molecular Pathology: Approaches to Diagnosing Human Disease in the Clinical Laboratory. R.A. Heim and L.M. Silverman, eds. Carolina Academic Press, Durham, NC, pp. 5–31. Woolley, A.T., Hadley, D., Landre, P., deMello, A.J., Mathies, R.A., and Northrup, M.A. 1996. Functional integration of PCR amplification and capillary electrophoresis in a microfabricated DNA analysis device. Anal Chem. 68:4081–4086. Yang, J., Huang, Y., Wang, X.B., Becker, F.F., and Gascoyne, P.R. 1999. Cell separation on microfabricated electrodes using dielectrophoretic/ gravitational field-flow fractionation. Anal. Chem. 71:911–918.

9 Nucleic Acid Amplification in Microchips Peter Wilding

INTRODUCTION The advent of analytical microchips (i.e., microfabricated devices that perform some form of analytical function) has provided the molecular biologist with many new opportunities for performing nucleic acid amplification. These devices provide a special type of environment for amplification that was absent in the more conventional equipment that has served this purpose over the past decade. Today, it seems relatively obvious that nucleic acid amplification would be one of the applications of the growing plethora of analytical functions carried out on microchips. In reality, the first publications in this area only date from 1994, showing how reticent developmental scientists were to investigate miniaturization of a procedure that was showing marked inconsistencies and reliability problems using conventional tubes and thermocycling equipment. In addition to this, there appeared little benefit at that time, other than potential savings in terms of reagent, to embark on development programs for nucleic acid amplification when efforts to miniaturize more common assays such as immunoassay-based tests, colorimetry, and electrophoresis were showing little progress. It was only when breakthroughs in fabrication techniques occurred that provided the ability to construct microchip-based devices incorporating channels and chambers with evidence of fluidic control that the realization occurred. The early work showed elementary devices with chambers that could be subjected to thermocycling and thus laid the foundation for microchip-based amplification. It would take several years and considerable effort to overcome many difficulties of fluidic control, efficient thermocycling, surface chemistry, specimen introduction, and quantitation of the amplicate. However, it is clear today, in 2000, that microchip-based amplification will play a major role in molecular biology and that the devices of the future will fulfill many other roles, such as sample preparation and amplicate detection.

BACKGROUND The technology that provided effective and reliable nucleic acid amplification on a microchip has its foundation in the early efforts to describe and delineate microchip fluidics (Kricka et al., 1989, 1993; Wilding et al., 1994b). The earliest reports (Wilding et al., 1994a; Northrup et al., 1994) described elementary microchips that facilitated performance of the polymerase chain reaction (PCR). These microchips contained volumes of 10–50 µL of PCR reaction mixtures (i.e., target DNA, Taq-polymerase, DNA

Nucleic acid amplification in microchips

193

primers, and the requisite nucleotides) and were heated by external thermocyclers. Early efforts were successful, but the reports lacked evidence of reliability, precision, and sensitivity (i.e., an index of amplification), and it soon became apparent that other issues such as surface chemistry (Wilding et al., 1995) and good hearing control (Northrup et al., 1995) must be solved before effective systems could be developed. It was assumed in 1994 that chip-based PCR and other nucleic acid amplification would employ very short cycling times. Earlier work using glass capillary-based devices (Wittwer et al., 1989) employed forced air for thermocycling, but these were passive systems unlikely to provide a basis for more complex devices. However, using these glass structures, it was demonstrated that very short cycling times could be achieved (Wittwer et al., 1991). Since the early demonstrations of microchip-based PCR, a large number of authors have reviewed the topic, assessed the technology or forecast the future direction of this technology Early reviews concentrated on the problems of fluidics (Ramsey et al., 1995) and the direction that gene-based diagnostics would follow (Eggers and Ehrlich, 1995). However, more recent reviews (Burns et al., 1996; Burke et al., 1997; Kricka, 1998; Peterson et al., 1998) reflect the developing technology as the integration of additional features to the microchip were successfully demonstrated. Other reviews (O’Donnell, 1996a,b; Kopp et al., 1997; Zlatanova, 1999) have described the role of microchip-based PCR in DNA sequencing and genetic analysis, the problems of process control, and the state of the art. It was projected in the mid-1990s that chip-based devices would become more complex and incorporate features that would facilitate sample preparation, amplification, and amplicate detection. As a result, the next generation of chip-based devices not only brought complexity but a clear awareness of the limitations and hurdles that microchipbased technologies present. Benefits of Microchip PCR The majority of the benefits of using microchip-based devices for PCR are still perceived, rather than realized. It has been assumed since the first illustration of microchip-based PCR in the early 1990s that features such as low reagent consumption, low-volume sample requirements, and rapid cycle times would be a consequence of this technology. However, the ability to couple the PCR process with other features such as sample preparation and amplicate detection quickly initiated a drive to the design and construction of integrated devices that will ultimately provide more convenient and cheaper methods in the many fields that molecular biology is practiced. The economy of manufacture has yet to be realized, but it is assumed that the pattern will follow that of the electronics industry, where millions of micro-devices can be produced at low cost. This is a realistic projection, as it is unlikely that the micro-devices that ultimately serve the needs of the analyst and researcher will ever incorporate the degree of complexity of electronics microchips. Furthermore, because of the relative simplicity of the biological microchips and the roles they will serve, it is probable that the manufacturing processes will be more likely based on plastic than silicon. Other benefits that are somewhat theoretical in 2000 are the processes that are permissible only in a microenvironment. These include the opportunity to manipulate

Biochip technology

194

solutes and formed elements (e.g., cells) in ways that provide improved methods of reagent and sample transfer, mixing using techniques such as electrokinesis (Ramsey et al., 1995), isolation or separation of macromolecules or cells by filtration (Wilding et al., 1998), and electrical charge or electronic stringency (Cheng et al., 1998a). Each of these techniques has been clearly demonstrated, and their incorporation into commercial devices is imminent. Opportunities Presented by Microchip PCR The opportunities presented by this technology are almost too numerous to list. The many commercial and academic centers that are exploring this technology have already demonstrated that DNA targets from several biological systems—including the human genome, viruses, and microbes—can be amplified by PCR on miniaturized devices (Kopp et al., 1998; Belgrader et al., 1998a). The fields in which the growth will first emerge probably relate to drug development in the pharmaceutical industry, where determination of inhibition, or enhancement, of nucleic acid replication is important. Other early products include devices designed for the defense industry, where the detection and identification of toxic agents is desirable—e.g., products developed by the Cepheid Corporation (Sunnyvale, CA) (Figure 1). Another key area is the provision of products that facilitate the parallel operation of PCR on micro-samples for the life sciences, where production of sufficient amplicate is a requisite for sequencing studies.

Figure 1. Hand-held device for PCR. Reprinted with permission from the Cepheid Corporation, Sunnyvale California.


195

Importance of Surface Chemistry An early discovery during the development of microchip PCR was the importance of surface chemistry in microchip operation and design. Wilding et al. (1995) and Shoffner et al. (1996) clearly showed that PCR in microchips was severely inhibited if the surface chemistry was unsuitable. These workers noted that the increase in surface area in microchips, relative to volume, could be 20-fold greater, compared to conventional tubes used for PCR (Figure 2). Moreover, if the surface was only partially coated or existed as silicon, silicon nitride, or certain other substances, then it was very difficult to achieve reproducible or adequate amplification. However, successful and reproducible amplification could be achieved if the surface was subjected to coating with a suitable passifying layer (e.g., 2000 Å of silicon oxide). Alternative Microtechnologies Alternative micro-methods for performing PCR have been reported that do not employ microchips. Conventional tube-based PCR can be carried out using reaction volumes below 5 µL. However, the difficulties of sample manipulation, product transfer, and amplicate detection limit the roles this can play. A notable technique that has provided the basis for micro-volume PCR is that using micro-capillaries (Wittwer et al., 1989). It was this technique that first allowed clear demonstration of rapid thermocycling (Wittwer et al., 1990, 1991) and facilitated numerous studies on the optimization of the PCR reaction (Taylor et al., 1997, 1998).

Figure 2. Ratio of surface area to volume in three different containers used for PCR.

Biochip technology

196

An example of the high sensitivity achievable with this type of equipment is rapid PCR of the outer surface protein A (OspA) gene fragment of Borrelia burgdorferi (the agent of Lyme disease) reported by Mouritsen and colleagues in 1996. Other workers (Swerdlow et al., 1997) have also demonstrated thermocycling in an air jacket where PCR is carried out in a capillary tube that is connected to a microchromatographic device that facilitates detection of the amplicate by laser-induced fluorescence.

INTEGRATION OF PCR WITH OTHER MICROCHIP-BASED PROCESSES As previously stated, microchip-based PCR devices hold great promise for their incorporation into integrated devices that permit both sample preparation and amplicate detection. However, for this to be achieved it has been necessary to demonstrate that each of the key elements (i.e., sample preparation, PCR, and amplicate detection) meet the rigid requirements of successful nucleic acid amplification. These include high sensitivity, reproducibility (i.e., precision), elimination of contamination, convenience of sample presentation, a detection system that meets the need in hand, and a product that can be produced economically. By 2000, all of these key features have been demonstrated. Basic PCR Devices To achieve PCR or some other form of amplification on a microchip, only an elementary chip design is required (Figure 3). Thus, several forms of amplification have been reported that mirror the techniques used throughout molecular biology. Chip-based PCR (Wilding et al., 1994b), reverse transcriptase PCR (RT-PCR) (Chang et al., 1999), the ligase chain reaction (LCR) (Cheng et al., 1996b), and degenerate oligonucleotide primed PCR (DOP-PCR) (Cheng et al., 1998b) have all been reported. Most of these have involved silicon-glass microchip structures holding volumes of 3–20 µL. One important requirement for PCR is thermocycling. While early efforts to achieve effective amplification used external heating and cooling systems (e.g., Wilding et al., 1994a), more recent developments have incorporated the heating systems into the chip using photolithographic construction techniques (Burns et al., 1996, 1998). The most likely method of heating (i.e., internal vs. external) for microchip PCR devices in the future will be influenced greatly by production costs and the efficiency of the heating systems. It is too early to predict which of these methods will dominate. Multichamber Microchip PCR With the demonstration of successful microchip-based PCR, it was inevitable that multichamber PCR would be developed on microchips (Taylor et al., 1997). This aspect of microchip-based PCR has since been further developed (Taylor et al., 1998). These


197

authors have reported on multichamber devices in silicon and glass and have optimized conditions for PCR using volumes of only 0.5 µL and with markedly reduced cycle times. More recent developments have allowed multiple-sample PCR coupled with electrophoretic analysis on a chip (Waters et al., 1998), while another has facilitated detection of microbes and bacterial spores using a 10-chamber device on a battery operated system (Belgrader et al., 1998a).

Figure 3. Basic microchips used for PCR. The microchips are 14×17 mm and the chamber within the chips hold ~12 µL. Reprinted with permission from Cheng et al. (1996c). Sample Preparation for Microchip-Based Amplification Systems Most developments in the efforts to achieve integration have concentrated on the integration of amplification and detection (Cheng et al., 1998a; Ibrahim et al., 1998). Another movement toward total integration of all the key elements of nucleic acid amplification was also demonstrated when white blood cells were isolated from whole blood samples within a microchip chamber that also served as the PCR chamber (Cheng et al., 1996a; Wilding et al., 1998). The method used to isolate the white blood cells from red blood cells and plasma inside the PCR chamber was filtration though a 3.5-µm gap (Figure 4, see Color Plate 8.3A). PCR reagents (i.e., Taq polymerase, primers, and nucleotides) were then added to the isolated cells and thermocycling commenced. Successful PCR indicates that successful lysis of the isolated white blood cells occurs

Biochip technology

198

during thermocycling and that yield of amplicate is not adversely affected. This demonstration of sample preparation within the PCR chamber of a microchip has been one of the indications that fully integrated microchip-based nucleic acid amplification will be made available. A more recent publication (Christel et al., 1999) reports on efforts to advance sample preparation involving extraction of DNA and concentration of DNA from samples using silicon fluidic microchips with high surface-area-to-volume ratios. Short (500 bp) and medium-sized (48,000 bp) DNA has been captured, washed, and eluted using the silicon dioxide surfaces of these chips. Chaotropic guanidinium hydrochloride solutions were used as binding agents. Wash and elution agents consisted of ethanol-based solutions and water, respectively.

Figure 4. See Color Plate 8.3A. Principle of the weir-type filter used to capture white blood cells in a microchip PCR chamber Reprinted with permission from Wilding et al. (1998). PCR and Quantitation of the Amplicate The longstanding method of quantifying an amplicate product from a PCR has been the use of stained electrophoresis gels. However, it is obvious that transfer of micro-volumes of less that 1.0 µL creates problems. Furthermore, it is obviously more convenient if quantitation can be achieved within the same microchip or in a micro-system (e.g., another microchip that provides capillary electrophoresis or a microarray). An early effort to achieve on-board detection (Hsueh et al., 1995) used tris(2,2'-bipyridyl) ruthenium as an electroluminescent DNA label and incorporated a microfabricated electrochemical cell


199

to monitor the PCR. Parallel to these developments, a great deal of effort was being spent in perfecting microchip-based capillary electrophoresis (CE) (Ramsey, 1998; Ramsey et al., 1995). This sensitive technique provides an ideal detection principle for amplified DNA targets. Illustration of the utility of conventional CE as a detection tool for microchip-based PCR products was demonstrated in PCR reactions using an extracted human genomic DNA and whole white blood cells directly in the microchip PCR chamber (Cheng et al., 1996a). Successful use of microchips for both PCR and CE was then demonstrated by amplification of a β-globin DNA target and a CE separation of the DNA amplicate in less than 45 minutes (Woolley et al., 1996). Another report (Fortina et al., 1997) described the use of entangled solution capillary electrophoresis (ESCE) and laser-induced fluorescence detection (LIF) for separation-based diagnostics in the quantitative analysis of multiplex PCR products for determination of carrier status of Duchenne/Becker muscular dystrophy (DMD/BMD). In this latter report, the quantitation of the digested PCR amplicate was subjected to both microchip CE and conventional CE with excellent results. Another device that was designed for rapid pathogen detection provides a result within 16 minutes and is based on a multichamber format. Using this instrument, detection of microbes has been achieved at concentrations of 102–104 organisms per mL (Belgrader et al., 1998b). These accomplishments clearly illustrate that complex and sensitive methods can be adapted to a microchip format and that they will be routinely available within the foreseeable future.

MICROFLUIDICS IN MICROCHIP-BASED PCR DEVICES One feature of microchip devices that is rarely addressed is the fluidic properties that determine the efficiency with which fluids and their contents are transferred within a microchip, or the platform arrangement on which it is mounted. Too many publications underemphasize this issue and the difficulties that exist with regard to ensuring that the fluidic properties are controllable and reproducible. In many instances, the design features (see section on “Design Features and Limitations”) will follow those applicable to macrostructures, but the increased surface area, marked changes in heat transfer, and the fluid flow in capillaries that may have internal diameters less that 10 µm will prove problematic (Wilding et al., 1994b). It is for this reason that new studies have been necessary to quantify the performance of PCR in a microenvironment and to determine the optimum conditions to apply. The author and his colleagues have designed numerous types of biological microchips and have learned that optimization of fluidics is more problematic than optimization of the PCR reaction (i.e., thermocycling times and reagent concentrations). The importance of fluidic control has been addressed by other several groups (Burns et al., 1996, 1998; Northrup et al., 1998), and it has been emphasized that the control of fluid transfer is markedly affected by demands created by on-board heaters, elimination of air bubbles, and the overall complexity of the structure. As designs become more complex, the need for a detailed understanding of microfluidics becomes

Biochip technology

200

more essential, particularly if the final product is to have micro-volume requirements. The need to transfer fluids from chamber to chamber on a microchip platform without compromising the reactions or the efficiency of fluid transfer is also important. A complex system that starts with a crude biological specimen will undoubtedly require numerous reagent additions, several transfer functions, valves, and pumps. A few clear descriptions of pumping mechanisms have been reported, and these include surface tension-based pumps that can move discrete nanoliter drops through enclosed channels (Burns et al., 1996) and electrokinetic pumping that allows precise transport and mixing in very small capillaries (Haab and Mathies, 1998; Woolley et al., 1996). Some workers have also achieved fully automated DNA reactions and analysis in a fluidic capillary instrument (Swerdlow et al., 1997).

FABRICATION AND FABRICATION MATERIALS Design Features and Limitations The ideal microchip-based PCR device would provide convenience of sample and reagent presentation, perfect mixing, efficient fluid transfer, and on-board detection. It would be capable of producing precise results at high sensitivity. Table 1 provides a brief review of some of the necessary features, with illustrations of ways in which these features have been addressed. Construction Materials The dimensions of the early microchip-based PCR devices that were constructed of silicon (Northrup et al., 1994a; Wilding et al., 1994a), with anodically bonded glass covers, approximated 1–2 cm2. Since then, chambers and channels in silicon structures have been etched to various depths depending on the size of the device concerned, and smaller chips have been fabricated. However, as glass has more favorable features for capillary electrophoresis, some workers have used small chambers in glass chips to perform PCR. Also, combinations of silicon chips for PCR and glass chips for detection have been employed (Woolley et al., 1996). More recently, workers have employed plastics (Kopp et al., 1997; Vasiliskov et al., 1998) and ceramic tapes (Zhong et al., 1999) to construct microchips. The former material has been used to construct a thermocycler incorporating a PCR device. It is highly probable that these materials will be the ones of choice when large-scale production of microchip-based devices commences.

THE FUTURE The level of investment in 2000 in microchip development indicates that this technology will dominate the life sciences in the early decades of the twenty-first century The many millions of dollars being spent to achieve a “lab-on-a-chip” (Jacobson and Ramsey, 1998)


201

by the numerous start-up and established companies

Table 1 Required Features of Microchip-Based PCR Devices Construction Silicon, plastic polymers or glass materials All body fluids, viruses, microbes, extracted DNA Sample type Sample volume 0.5–20 µL PCR volume 0.5–20 µL 10 sec to 1 min Cycling time Thermocycling On-board, external contact, air system Thermal properties that allow fast cycle times Micro-chamber Inert in PCR (e.g., silicon oxide) surface Fluidics Good fluid flow without bubbles Input/output ports to allow sample and reagent addition/disposal Amplicate detection TaqMan, CE, Microarray are likely to produce a virtual plethora of devices for many functions. As nucleic acid amplification will remain a cornerstone technique in biological research for the foreseeable future, there are likely to be many versions of PCR devices to meet the demand.

REFERENCES Belgrader, P., J.K.Smith, V.W.Weedn, and M.A.Northrup. 1998a. Rapid PCR for identity testing using a battery-powered miniature thermal cycler. J. Forensic Sci. 43:315–319. Belgrader, P., W.Benett, D.Hadley, G.Long, R.Mariella Jr., E.Milanovich, S.Nasarabadi, W. Nelson, J.Richards, and P.Stratton. 1998b. Rapid pathogen detection using a microchip PCR array instrument. Clin. Chem. 44:2191–2194. Burke, D.T., M.A.Burns, and C.Mastrangelo. 1997. Microfabrication technologies for integrated nucleic acid analysis. Genome Res. 7:189–97. Burns, M.A., C.H.Mastrangelo, T.S.Sammarco, F.P.Man, J.R.Webster, B.N.Johnson, B. Foerster, D.Jones, Y.Fields, A.R.Kaiser, and D.T.Burke. 1996. Microfabricated structures for integrated DNA analysis. Proc. Natl. Acad. Sci. USA 93:5556–5561. Burns, M.A., 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. 1998. An integrated nanoliter DNA analysis device. Science 282:484–487. Chang, C., S.Perkins, and C.Wittwer. 1999. Measurement of MDR1 gene expression by real time quantitative RT-PCR. Mod. Pathol. 12:190A. Cheng, J., M.A.Shoffner, G.E.Hvichia, L.J. Kricka, and P.Wilding. 1996a. Chip PCR, II: Investigation of different PCR amplification systems in microfabricated silicon-glass chips. Nucleic Acids Res. 24:380–385.

Biochip technology

202

Cheng, J., M.A.Shoffner, K.R.Mitchelson, L.J.Kricka, and P.Wilding. 1996b. Analysis of ligase chain reaction products amplified in a silicon-glass chip using capillary electrophoresis. J. Chromatog. A 732:151–158. Cheng, J., P.Fortina, S.Surrey, L.J.Kricka, and P.Wilding. 1996c. Microchip-based devices for molecular diagnosis of genetic diseases. Mol. Diagn. 1:183–200. Cheng, J., E.L.Sheldon, L.Wu, A.Uribe, L.O.Gerrue, J.Carrino, M.J.Heller, and J.P O’Connell. 1998a. Preparation and hybridization analysis of DNA/RNA from E. coli on microfabricated bioelectronic chips. Nature Biotechnol. 16:541–546. Cheng, J., L.C.Waters, P.Fortina, G.Hvichia, S.C.Jacobson, J.M.Ramsey, L.J.Kricka, and P. Wilding. 1998b. Degenerate oligonucleotide primed-polymerase chain reaction and capillary electrophoretic analysis of human DNA on microchip-based devices. Anal. Biochem. 257:101–106. Christel, L.A., K.Petersen, W.McMillan, and M,A, Northrup. 1999. Rapid, automated nucleic acid probe assays using silicon microstructures for nucleic acid concentration. J. Biomech. Eng. 121:22–27. Eggers, M., and D.Ehrlich. 1995. A review of microfabricated devices for gene-based diagnostics. Hematol Pathol 9:1–15. Fortina, P., J.Cheng, M.A.Shoffner, S.Surrey, W.M.Hitchcock, L.J.Kricka, and P.Wilding. 1997. Diagnosis of Duchenne/Becker muscular dystrophy and quantitative identification of carrier status by use of entangled solution capillary electrophoresis. Clin. Chem. 43:745–751. Haab, B.B., and R.A.Mathies. 1998. Single molecule DNA detection in microfabricated capillary electrophoresis chips. Biophys. J. 74.A9. Hsueh, Y-T., R.L.Smith, and M.A.Northrup. 1995. A microfabricated, electrochemiluminescence cell for the detection of amplified DNA. 8th Int. Conf. SolState Sens. Actuators Eurosens., IX: Digest of Technical Papers (IEEE Cat. No.95TH8173). Foundation Sensors Actuators Technol, Stockholm, Sweden. Vol. 1, pp. 768–771. Ibrahim, M.S., R.S.Lofts, P.B.Jahrling, E.A.Henchal, V.W.Weedn, M.A.Northrup, and P. Belgrader. 1998. Real-time microchip PCR for detecting single-base differences in viral and human DNA. Anal Chem. 70:2013–2017. Jacobson, S.C., and J.M.Ramsey. 1998. Lab-on-a-chip devices for chemical and biochemical assays. Abstr. Pap., 172nd Mtg., Am. Chem. Soc. 215(1–2):123. Kopp, M.U., H.J.Crabtree, and A.Manz. 1997. Developments in technology and applications of microsystems. Curr. Opin. Chem. Biol 1:410–419. Kopp, M.U., A.J.de Mello, and A.Manz. 1998. Chemical amplification: Continuous-flow PCR on a chip. Science 280:1046–1047. Kricka, L.J. 1998. Miniaturization of analytical systems. Clin. Chem. 44:2008–2014. Kricka, L.J., P.Wilding, J.Pfahler, J.Harley, H.Bau, and J.Zemel. 1989. Liquid transport in micron and submicron channels. J. Soc. Photo-Opt. Instr. Eng. 1167:159–168. Kricka, L.J., O.Nozaki, S.Heyner, W.T.Garside, and P.Wilding. 1993. Applications of a microfabricated device for evaluating sperm function. Clin. Chem. 39:1944–1947. Mouritsen, C.L., C.T.Wittwer, C.M.Litwin, L.Yang, J.J.Weis, T.B.Martins, T.D.Jaskowski, and H.R.Hill. 1996. Polymerase chain reaction detection of Lyme disease: Correlation with clinical manifestations and serologic responses. Amer. J.


203

Clin. Pathol 105:647–654. Northrup, M.A., M.T.Ching, R.M.White, and R.T.Watson. 1994. DNA amplification with a microfabricated reaction chamber. 7th Int. Conf. Sol-State Sens. Actuators, Yokohama, Japan. Northrup, M.A., C.Gonzalez, D.Hadley, R.F.Hills, P.Landre, S.Lehew, R.Saiki, J.J.Sninsky, R.Watson, and R.Watson Jr. 1995. A MEMS-based miniature DNA analysis system. 8th Int. Conf. Sol-State Sens. Actuators Eurosens., IX: Digest of Technical Papers (IEEE Cat. No.95TH8173). Foundation Sensors & Actuator Technol., Stockholm. Vol.1, pp. 764–767. Northrup, M.A., B.Benett, D.Hadley, P.Landre, S.Lehew, J.Richards, P.Stratton. 1998. A miniature analytical instrument for nucleic acids based on micromachined silicon reaction chambers. Anal Chem. 70:918–922. O’Donnell-Maloney, M.J., and D.P.Little. 1996a. Microfabrication and array technologies for DNA sequencing and diagnostics. Gen. Anal: Biomol Eng. 13:151– 157. O’Donnell-Maloney, M.J., C.L.Smith, and C.R.Cantor. 1996b. The development of microfabricated arrays for DNA sequencing and analysis. Trends Biotechnol 14:401– 407. Peterson, K.E., W.A.McMillan, G.T.A.Kovacs, M.A.Northrup, L.A.Christel, and F. Pourahmadi. 1998. Toward next generation clinical diagnostic instruments: Scaling and new processing paradigms. J. Biomed. Devices 1:71–79. Ramsey, J.M. 1998. Microdomain chemical analysis. Abstr. Pap., 172nd Mtg., Am. Chem. Soc. 216:190. Ramsey, J.M., S.C.Jacobson, and M.R.Knapp. 1995. Microfabricated chemical measurements systems. Nature Med. 1:1093–1096. Shoffner, M.A., J.Cheng, G.E. Hvichia, L.J. Kricka, and P.Wilding. 1996. Chip PCR, I: Surface passivation of microfabricated silicon-glass chips for PCR. Nucleic Acids Res. 24:375–379. Swerdlow, H., B.J.Jones, and C.T.Wittwer. 1997. Fully automated DNA reaction and analysis in a fluidic capillary instrument. Anal. Chem. 69:848–855. Taylor, T.B., E.S.Winn-Deen, E.Picozza, T.M.Woudenberg, and M.Albin. 1997. Optimization of the performance of the polymerase chain reaction in silicon-based microstructures. Nucleic Acids Res. 25:3164–3168. Taylor, T.B., S.E.Harvey, M.Albin, L.Lebak, Y.Ning, I.Mowat, T.Schuerlein, and E. Principe. 1998. Process control for optimal PCR performance in glass microstructures. J. Biomed. Devices 1:65–70. Vasiliskov, V.A., E.N.Timofeev, S.A.Surzhikov, A.L.Drobyshev, V.V.Shik, and A.D.Mirzabekov. 1998. A method for making microchips by copolymerization with acrylamide. Molekulyarnaya Biologiya (Moscow) 32:923–925. Waters, L.C., S.C.Jacobson, N.Kroutchina, J.Khandurina, R.S.Foote, and J.M.Ramsey. 1998. Multiple sample PCR amplification and electrophoretic analysis on a microchip. Anal. Chem. 70:5172–5176. Wilding, P., J.Pfahler, H.H.Bau, J.N.Zemel, and L.J.Kricka. 1994a. Manipulation and flow of biological fluids in straight channels micromachined in silicon. Clin. Chem. 40:43–47.

Biochip technology

204

Wilding, P., M.A.Shoffner, and L.J.Kricka. 1994b. PCR in a silicon microstructure. Clin. Chem. 40:1815–1818. Wilding, P., M.A.Shoffner, G.Hvichia, and L.J.Kricka. 1995. Thermal cycling and surface passivation of micromachined devices for PCR. Clin. Chem. 41:1367–1368. Wilding, P., L.J.Kricka, J.Cheng, G.Hvichia, M.A.Shoffner, and P.Fortina. 1998. Integrated cell isolation and polymerase chain reaction analysis using silicon microfilter chambers. Anal Biochem. 257:95–100. Wittwer, C.T., G.C.Fillmore, and D.R.Hillyard. 1989. Automated polymerase chain reaction in capillary tubes with hot air. Nucleic Acids Res. 17:4353–4357. Wittwer, C.T., G.C.Fillmore, and D.J.Garling. 1990. Minimizing the time required for DNA amplification by efficient heat transfer to small samples. Anal. Biochem. 186:328–331. Wittwer, C.T., and D.J.Garling. 1991. Rapid cycle DNA amplification: Time and temperature optimization. BioTechniques 10:76–83. Wittwer, C.T., M.G.Herrmann, A.A.Moss, and R.P.Rasmussen. 1997. Continuous fluorescence monitoring of rapid cycle DNA amplification. BioTechniques 22:130–1. Woolley, A.T., D.Hadley, P.Landre, A.J.deMello, R.A.Mathies, and M.A.Northrup. 1996. Functional integration of PCR amplification and capillary electrophoresis in a microfabricated DNA analysis device. Anal. Chem. 68:4081–4086. Zlatanova, J., S.Bavykin, and A.Mirzabekov. 1999. Biological microchips: State-of-theart. Biophys. J. 76:A456. Zhong, J., Y.Mingqiang, and H.H.Bau. 1999. A thermal cycler fabricated with low temperature, co-fired ceramic tapes. IMECE 1999 MEMS Symp. Proc., pp. 1–6.

10 Technology Options and Applications of DNA Microarrays Paolo Fortina, David Graves, Christian Stoeckert Jr., Steven McKenzie, and Saul Surrey

INTRODUCTION Early work on high-density peptide and oligonucleotide microarrays fabricated using semiconductor-based technologies (Fodor et al., 1991, 1993) has stimulated much research using these chip-based approaches to answer important biological questions. Substantial progress in the application of these devices has been made; however, opinions differ on which microarray format to use (Bowtell, 1999; Gerhold et al., 1999; Marshall and Hodgson, 1998; O’Donnell-Maloney and Little, 1996; O’Donnell-Maloney et al., 1996; Ramsay, 1998; Southern, 1996). In addition, access to these new tools has been constrained by considerable start-up costs, delayed product release dates, uncertainty about the best technology, and incompletely optimized commercial systems (Castellino, 1997). This has resulted in a relative confinement to commercial ventures and large wellfunded research laboratories. Nonetheless, microarray technology continues to contribute much to our understanding of human gene organization and expression, and a large body of research has focused on the use of DNA chips, reflecting the increasing power and availability of this technology (Aitman et al., 1999; Brown and Botstein, 1999; Debouck and Goodfellow, 1999; Duggan et al., 1999; Hacia, 1999; Lander, 1999; Lipshutz et al., 1999; Southern et al., 1999). There are a number of microarray-related websites detailing academic as well as commercial efforts in this area (i.e., http://www.gene-chips.com). Although it is clear that no single approach provides all solutions, we provide here a perspective on options available to exploit such technology and how one might build a complete system suitable for users operating in “low-tech” research or clinical laboratory settings. In addition, we provide a summary of our results from basic biophysical studies aimed at helping to define parameters for array optimization. A broad definition of a DNA microarray includes nylon-, glass-, and silicon-based surfaces holding large sets of single-stranded polymeric molecules such as oligonucleotides, peptide nucleic acids (PNA), or cDNAs at discrete locations. The immobilized molecules or probes participate in a heterogeneous hybridization with DNA or RNA molecules or targets in solution, which are usually labeled either directly with fluorescent or radioactive tags or indirectly with conjugates that subsequently bind fluorescent, chemiluminescent, or radioactive molecules. In addition, numerous reactions can be performed after probe/target capture on the array. Following hybridization or the subsequent biochemical reactions, a scan of the spots at each register results in a genetic

Biochip technology

206

profile on the target molecules. Refinement and optimization in all aspects of the technique including the biochemical processes, signal discrimination, and software analysis are still required (Adams and Kron, 1997; Bassett et al., 1999; Chan et al., 1995; Chan et al., 1997, 1998; Chrisey et al., 1996; Dubiley et al., 1997; Graves, 1999; Guo et al., 1994, 1997; Guschin et al., 1997; Kwiatkowski et al., 1999; Lamture et al., 1994; Maskos and Southern, 1992, 1993a,b; McKenzie et al., 1998; Mir and Southern, 1999; Nguyen et al., 1997; Nilsson et al., 1994; Rogers et al., 1999; Shchepinov et al., 1997; Southern et al., 1992). However, the relative simplicity, the advantage of flexibility, and the potential for high-throughput parallel molecular genetic analysis are demonstrating that DNA microarray technology is one of the most promising analytical techniques to identify single-nucleotide polymorphisms and mutations, small insertions and deletions, and to assess gene copy number and genome-wide mRNA expression patterns. Other anticipated uses of DNA microarrays include drug discovery, linkage analysis, forensic investigation, pathogen identification, predicting responsiveness to drugs, and sequence analysis by hybridization of previously characterized as well as uncharacterized targets (Cheung et al., 1999; Drmanac et al., 1993, 1998; Gentalen and Chee, 1999; Hacia et al., 1998a; Heller et al., 1997; Livache et al., 1998a; Mirzabekov, 1994). Generating useful and reliable microarray-based data appears easy and simple, but considerable engineering, biochemical, and bioinformatic expertise is required for success. In addition, the design of a DNA array is based on the amount and complexity of information to be obtained from a single experiment. Arrays may contain DNA fragments of any length and compositions ranging from short oligonucleotides to megabase clones. At one extreme, a small number of elements (e.g., a set of allele-specific oligonucleotides) can be arrayed to examine several mutations in a particular gene. At the other, hundreds to thousands of oligonucleotides or clones can be arrayed to examine gene expression patterns. Many products and services are commercially available ranging from $1000 for low-density arrays to over $250,000 for a complete platform for hybridization/washing with a scanner capable of analyzing arrays of several hundred thousand ordered sets of DNA molecules of known sequence. The number of elements arrayed on a solid support also depends on the method employed to manufacture the arrays.

PROBE SYNTHESIS/DEPOSITION ON MICROARRAYS A number of options exist for probe synthesis/deposition on slides, and the following outlines some of those options and describes new technologies currently in development. Broadly speaking, microarrays can be bonded to a glass surface either covalently or noncovalently. One form of covalent attachment is direct synthesis of oligonucleotides on solid planar platforms using photolithographic or other types of masking techniques; however, maskless methods have also been described (http://pompous.swmed.edu; Blanchard and Friend, 1999; Singh-Gasson et al., 1999). Alternatively, pre-synthesized amino-modified oligonucleotides can be at-tached covalently to derivatized glass surfaces using, for example, 1,4-phenylene diisothiocyanate (PDC) (Sanguedolce et al., 1999), or noncovalent methods (essentially by irreversible adsorption) can be used to immobilize

Technology options and applications of dna microarrays

207

DNA molecules onto polylysine-coated or amino-silanized slides (Eisen and Brown, 1999). For the most part, microarrays can be generated by one of the following three methods: (1) in-situ synthesis of oligonucleotide or PNA probes to a solid surface, (2) site-specific attachment of pre-synthesized oligonucleotides, PNAs, PCR products, and/or clones, and (3) spotting probes on positively charged coated glass slides. Several variations exist such as use of microelectrodes to immobilize presynthesized oligonucleotides under controlled electrical fields (Beattie et al., 1993, 1995; Cheng et al., 1998a,b; Edman et al., 1997; Livache et al., 1998b,c; Sosnowski et al., 1997) as well as attachment of probes within small three-dimensional spots of gel arrayed on the solid surface (Drobyshev et al., 1997; Guschin et al., 1997; Livache et al., 1994; Proudnikov et al., 1998; Rehman et al., 1999; Vasiliskov et al., 1999). Finally, cDNA/EST arrays on polymeric filter sheets are also commercially available. These are relatively inexpensive and reproducible when the amount of data is considered in comparison to Northern blotting or RT-PCR for thousands of transcripts. In-situ Synthesis Four major approaches have been developed to assemble base-by-base oligonucleotide arrays containing a variety of addressable sequences for high-throughput analyses: (1) photomask-guided photolithography, (2) maskless photolithography, (3) piezoelectric printing, and (4) surface tension-based deposition. Each one requires accurate monitoring to warrant fidelity of the growing oligonucleotide probes at a specific address on the planar solid surface containing multiple reactive sites. Oligonucleotide synthesis using a conventional CPG-based synthesizers results in a step yield of approximately 99%, but surface synthesis is considerably less efficient. Small differences in yield have a dramatic impact on the quality of the final product. For example, in order to synthesize a 24 mer oligonucleotide probe, 24 chemical steps are required; and, if the yield of each step is successful only 94% of the time, then the overall yield of full-length product is (0.94) 24=22.65%. Improvements in synthetic yield are critical since this should lead to increased target hybridization and subsequent improved detection. In-situ synthesis is currently limited to large laboratories and companies; however, recent developments may make such arrays standard tools for almost any micro-molecular array-based assay (Blanchard and Friend, 1999). Photolithographic Masks The first approach to in-situ synthesis was developed by Affymetrix for the GeneChip™ using photomask-guided photolithography combined with light-directed chemical synthesis (Fodor et al., 1991, 1993; Lipshutz et al., 1995; McGall et al., 1996; Pease et al., 1994). Briefly, the approximately 1-cm2 silicon surface is derivatized with a linker bearing a photo-protecting group. Subsequently, specific regions are deprotected by a mercury light-based beam through a chrome-glass mask. A nucleoside phosphoramidite carrying its own photolabile protecting group is then coupled to the free 5′-hydroxyl group at the exposed sites. Repeated cycles of deprotecting and coupling are used to

Biochip technology

208

synthesize all probes in parallel one base at a time, facilitating large numbers of probe combinations using a limited number of steps. For example, to make 4N-oligonucleotides or any subset of N-mer sequences of length N, only 4N steps of deprotection and coupling are required, one for each of the four photolabile phosphoramidite deoxynucleoside times N base positions (Pease et al., 1994). For example, if one wants to synthesize all possible oligonucleotides (1,048,576) of length 10, then 4×10 steps will be required. Using custom masks, arrays can be synthesized to represent any DNA sequence of interest. Each position in the target sequence can be then interrogated by sets of four probes that are identical except at the interrogation position (A, C, G, and T) and tiled across the sequence in increments of one nucleotide. This approach provides both more information than a single spot and increased confidence in base calling; however, it does necessitate use of multiple probes to make a single call, thereby decreasing the overall number of targets that can be interrogated on a single chip. In a typical experiment, after PCR amplification of the region(s) of interest, the user prepares single-stranded fluorescently labeled target, which, following fragmentation, is hybridized in the GeneChip probe array using a fluidic station where all steps are automated. An argon ion laser-based scanner is then used to excite the fluorescent reporter groups incorporated into the target, which is hybridized to the complementary probe array, and then image processing software acquires emitted signals over each feature. Proprietary algorithms use average intensity values to generate genetic information usable for gene expression monitoring, genotyping, and sequence analysis (Chee et al., 1996; Cronin et al., 1996; Hacia et al., 1996; Kozal et al., 1996; Lockhart et al., 1996; Winzeler et al., 1998; Wodicka et al., 1997). Various GeneChip™ products are available, including those for human, mouse, and yeast expression studies, as well as for analysis of specific sequences (e.g., HIV-1, human p53 tumor suppressor gene, and cytochrome P450 enzymes). A1500 human SNP GeneChip is also available for gene-mapping studies (Wang et al., 1998). One limitation is that the hardware platform to read results is specifically designed to accept only proprietary probe arrays. This limits the user’s options for making custom arrays or using alternate substrates such as microscope slides. However, the major disadvantage of this approach results from the chip production method itself, with the high costs of mask design and fabrication, which translates into high up-front and ongoing operational costs that are generally not affordable by the vast majority of academic-based diagnostic and research laboratories. Maskless Photolithography To address some of the above-mentioned limitations, research and development in hardware for in-situ maskless photolithography-based array synthesis (MAS) is underway. Combining digital light processing (DLP) technology with optical deprotection photochemistry, collectively defined as Digital Optical Chemistry (DOC) (http://pompous.swmed.edu), digital or virtual masks have been demonstrated to overcome limitations in high-density microarray fabrication (Singh-Gasson et al., 1999). DLP is a technology based on Texas Instruments’ Digital Micromirror Device (DMD) that is used in high-definition television (HDTV). Using DLP, a video signal is translated into a digital bitstream, which controls an array of hundreds of thousands to millions of


209

microscopic mirrors under computer digital control. In HDTV, the mirrors produce a gray-scale image by switching on and off more than 5000 times a second to pass, partially pass, or block transmission of light. The same DLP mirror technology can now be used to selectively focus UV light onto a glass slide. The system can manufacture oligo arrays with more than 1,000,000 spots for DNA resequencing or expression studies within hours with integration to a fluidic station delivering sequence-dependent photolabile phosphoramidite deoxynucleosides and to a computer controlling the DLP apparatus (http://pompous.swmed.edu). It is anticipated that the hardware will be able to be replicated for use in any laboratory, therefore improving the costs and efficiencies of building custom chips. Piezoelectric Deposition The third approach uses piezoelectric printing technology similar to that employed in inkjet printers. Three main types of inkjet dispensers (e.g., piezoelectric, solenoid, and thermal) differ in how the liquid is ejected through a small hole as a droplet with sufficient velocity (Lemmo et al., 1998). These means to create the driving force define the three different types of inkjet dispensers. In the thermal type, the fluid is heated, causing a vapor bubble to form and expand. The solenoid type uses gas or hydraulic pressure to compress the fluid against a valve so that when the valve is opened an acoustic or pressure wave is generated, allowing fluid dispensing. Finally, the piezoelectric type uses a piezoelectric crystal coupled to a fluid reservoir. Changing the crystal dimension causes the reservoir to compress and eject the liquid from the hole. Due to their relative simplicity, piezoelectric dispensers are widely used in microarray fabrication, offering several advantages over syringe pump-based systems. A piezoelectric dispenser can deliver small drops of fluid with volumes ranging from 30 to 500 picoliters within a 40- to 100-micrometer spot diameter. By properly selecting the drive electronics, users can vary the volume by changing voltage and pulse width as well as the frequency and number of pulses. Therefore, one drop at a time or more than 1000 drops per second can be delivered. Different array-making protocols have been developed; however, in the most recent the protecting group on the linker is removed simultaneously from all spots of the array (Matson et al., 1995; O’Donnell-Maloney, 1996). A multiple jet dispenser then delivers the correct nucleotide to each site, allowing simultaneous extension of one base in a single step. There are a few commercial inkjet dispensers now available. Surface/Tension-Based Deposition The last approach was developed by Protogene (Palo Alto, CA) and generates arrays by first placing on a glass surface a mask over the spots intended to become the reaction sites, then the rest of the glass is chemically treated to make it hydrophobic to all aqueous fluids. The mask is then removed, and a second chemical treatment makes the reaction sites hydrophilic for DNA reagents. Subsequent chemical treatments attach linking groups, capped with chemical groups ready to receive the first base of an oligonucleotide, to the surface within the individual reaction sites (Dr. M. Cronin, personal

Biochip technology

210

communication). There are several benefits in using this technology. Finally, a critical issue to obtain full-length large-scale synthesis is the ability to control each chemical reaction at the specific address on the surface. Use of an acid precursor that can be converted in situ into an acid by a solution photolytic process should help facilitate oligonucleotide synthesis with greater efficiency than direct photolysis. A photo-generated acid such as triarylsulfonium hexafluorophosphate can cleave the 4,4′-dimethoxytrityl (DMT) group on the 5′-O position of a growing oligonucleotide chain to provide free 5′-OH groups, thus permitting synthesis of oligonucleotides using, for example, phosphoramidite chemistry employing 5′-O-DMTprotected nucleophosphoramidites. Advancements in solution photo-controllable reactions should provide benefits from the sets of reagents already existing (Gao et al., 1998). Post-Synthetic Deposition Several methods have been used to attach pre-synthesized 5′-amino-modified oligonucleotides to SiO2-modified silanized glass microscope slides with homo-bifunctional crosslinkers such as glutaraldehyde, a standard protein immobilization reagent, and 1,4-phenylene diisothiocyanate (PDC) (Sanguedolce et al., 1999). Other methods add an epoxysilane moiety using 3′-glycidoxy propyltrimethyloxysilane in xylene containing a catalytic amount of diisopropylethylamine. 3′-Amino-modified oligomers are then allowed to react with the epoxide surface (Beattie et al., 1995; Guo et al., 1994; Southern et al., 1992). Alternative methods involve the electrostatic attachment of unmodified oligonucleotides to polylysinetreated surfaces and amino-silanized glass (Ganachaud et al., 1997) and the covalent and noncovalent attachment to plastic (Matson et al., 1994, 1995; Ogura et al., 1994). One of the major differences in post-synthetic deposition is in the oligonucleotide orientation. In-situ synthesis yields a 3′-to-5′ orientation out of the covalent link. Synthetic deposition permits a 5′-to-3′ orientation from the covalent linkage so that polymerase-catalyzed extension reactions can be performed. Another attachment method involves denaturing the DNA fragments in alkaline solution, then covalently attaching the DNA onto a poly-L-lysine-coated glass slide by UV irradiation and/or by overnight drying (Schena et al., 1995). All of the above methods allow attachment of a high-density of DNA fragments onto a small surface for further hybridization. Preparation of Slides Glass slides must be cleaned either with an alkaline solution as described (http://cmgn.stanford.edu/pbrown/protocols) or with acid (Sanguedolce et al., 1999). Briefly, in the former, slides are immersed in a solution made by dissolving 100 g of NaOH in 300 mL ddH20, then adding 400 mL 95% (v/v) EtOH followed by stirring until completely mixed. Glass slides are loaded into a slide holder and completely immersed in the cleaning solution for 2 hours, rinsed in ddH20 3X, and then transferred to a 1000-mL beaker containing poly-L-lysine solution for 1 hour. Poly-L-lysine solution is prepared by


211

adding 70 mL poly-L-lysine solution to 280 mL ddH20. Slides are then removed, dried at 40°C for 5 min, and stored in a closed box for at least 2 weeks before use. Surface Modification and Deposition of Oligonucleotide Probes A variety of options exist for surface modification and attachment of probes to arrays. One such option is use of PDC treatment on poly-L-lysine-coated slides to attach oligonucleotides or PCR products. Poly-L-lysine slides can be modified by treatment with 0.2% (w/v) PDC in 10% (v/v) pyridine/dimethylformamide for 2 hours at room temperature and then washed with HPLC-grade methanol and acetone. After drying at 110°C for 5 min, the PDC support is ready for amino-modified oligonucleotide attachment. The amino-modified oligonucleotides at 100 pmol/µL are mixed 1:1 with Micro-Spotting solution (Telchem, Sunnyvale, CA) and arrayed robotically onto the glass supports. The attachment oligos can be deposited in 5-nL or larger spots and dried overnight at room temperature. Remaining reactive groups on slides are blocked by incubation in 1 M Tris-HCl (pH 7.5) for 1 hour, rinsed in 1 M NaCl, and then washed 3X in ddH20, twice at room temperature followed by a final wash at 55°C for 15 min. Slides are dried by centrifugation at 500 rpm for 2 min in a tabletop centrifuge using a microtiter centrifuge plate holder (Sanguedolce et al., 1999). Probes, Targets, and Hybridization The relationships among target length and concentration, probe length and surface density, and extent of hybridization over time have been explored theoretically (Chan et al., 1996) and experimentally (Chan et al., 1997, 1998; Lockhart et al., 1996; Sanguedolce et al., 1999; Southern et al., 1999). Of the various predictions with practical implications, two stand out. First, in the absence of applied electric field, short (

Biochip Technology

Biochip Technology

Microarray Biochip Technology (MOLECULAR LABORATORY METHODS (BIOTECHNIQUES BOOKS))

Network Technology for Digital Audio (Music Technology) (Music Technology)

Network Technology for Digital Audio (Music Technology) (Music Technology)

Enzyme Technology

VLSI Technology

Space Technology

Nuclear Technology

Absorbent Technology (Textile Science and Technology)

Pet Packaging Technology (Sheffield Packaging Technology)

Object Technology

Web Technology

Magnesium Technology

Magnesium Technology

Nano-technology

Questioning Technology

Microsystem Technology

Eyewitness: Technology

Textile Technology

Concrete technology

Darkness Technology

Ancient Technology

Neurorehabilitation Technology

Reproductive technology

Space technology

Space Technology

Terahertz Technology

Microwave Technology

Endovascular Technology

Biochip Technology