Unconventional Nanopatterning Techniques and Applications

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

UNCONVENTIONAL NANOPATTERNING TECHNIQUES AND APPLICATIONS

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

UNCONVENTIONAL NANOPATTERNING TECHNIQUES AND APPLICATIONS John A. Rogers Hong H. Lee

A John Wiley & Sons, Inc., Publication

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

C 2009 by John Wiley & Sons, Inc. All rights reserved. Copyright


Published by John Wiley & Sons, Inc., Hoboken, New Jersey. Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at www.wiley.com/go/permissions. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com. Library of Congress Cataloging-in-Publication Data:

ISBN: 978-0-470-09957-5 Printed in the United States of America 10 9 8 7 6 5 4 3 2 1

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

CONTENTS PREFACE I NANOPATTERNING TECHNIQUES

xv 1

1

INTRODUCTION

3

2

MATERIALS 2.1 Introduction / 7 2.2 Mold Materials and Mold Preparation / 8 2.2.1 Soft Molds / 8 2.2.2 Hard Molds / 19 2.2.3 Rigiflex Molds / 19 2.3 Surface Treatment and Modification / 21 References / 23

7

3

PATTERNING BASED ON NATURAL FORCE 3.1 Introduction / 27 3.2 Capillary Force / 28 3.2.1 Open-Ended Capillary / 29 3.2.2 Closed Permeable Capillary / 31 3.2.3 Completely Closed Capillary / 40 3.2.4 Fast Patterning / 43 3.2.5 Capillary Kinetics / 45 3.3 London Force and Liquid Filament Stability / 48 3.3.1 Patterning by Selective Dewetting / 49 3.3.2 Liquid Filament Stability: Filling and Patterning / 51 3.4 Mechanical Stress: Patterning of A Metal Surface / 56 References / 63

27

4

PATTERNING BASED ON WORK OF ADHESION 4.1 Introduction / 67 4.2 Work of Adhesion / 68

67

v

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

vi

21:8

Printer Name: Yet to Come

CONTENTS

4.3 Kinetic Effects / 71 4.4 Transfer Patterning / 74 4.5 Subtractive Transfer Patterning / 79 4.6 Transfer Printing / 82 References / 91 5 PATTERNING BASED ON LIGHT: OPTICAL SOFT LITHOGRAPHY 5.1 Introduction / 95 5.2 System Elements / 96 5.2.1 Overview / 96 5.2.2 Elastomeric Photomasks / 96 5.2.3 Photosensitive Materials / 99 5.3 Two-Dimensional Optical Soft Lithography (OSL) / 100 5.3.1 Two-Dimensional OSL with Phase Masks / 100 5.3.2 Two-Dimensional OSL with Embossed Masks / 104 5.3.3 Two-Dimensional OSL with Amplitude Masks / 105 5.3.4 Two-Dimensional OSL with Amplitude/Phase Masks / 109 5.4 Three-Dimensional Optical Soft Lithography / 110 5.4.1 Optics / 111 5.4.2 Patterning Results / 112 5.5 Applications / 117 5.5.1 Low-Voltage Organic Electronics / 117 5.5.2 Filters and Mixers for Microfluidics / 118 5.5.3 High Energy Fusion Targets and Media for Chemical Release / 118 5.5.4 Photonic Bandgap Materials / 120 References / 122 6 PATTERNING BASED ON EXTERNAL FORCE: NANOIMPRINT LITHOGRAPHY L. Jay Guo

6.1 Introduction / 129 6.2 NIL MOLD / 133 6.2.1 Mold Fabrication / 133 6.2.2 Mold Surface Preparation / 137 6.2.3 Flexible Fluoropolymer Mold / 137 6.3 NIL Resist / 138 6.3.1 Thermoplastic Resist / 139 6.3.2 Copolymer Thermoplastic Resists / 141 6.3.3 Thermal-Curable Resists / 142 6.3.4 UV-Curable Resist / 146 6.3.5 Other Imprintable Materials / 148 6.4 The Nanoimprint Process / 149 6.4.1 Cavity Fill Process / 149

95

129

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

CONTENTS

vii

6.5 Variations of NIL Processes / 152 6.5.1 Reverse Nanoimprint / 152 6.5.2 Combined Nanoimprint and Photolithography / 155 6.5.3 Roll-to-Roll Nanoimprint Lithography (R2RNIL) / 156 6.6 Conclusion / 159 References / 160 7

PATTERNING BASED ON EDGE EFFECTS: EDGE LITHOGRAPHY

167

Matthias Geissler, Joseph M. McLellan, Eric P. Lee and Younan Xia

7.1 Introduction / 167 7.2 Topography-Directed Pattern Transfer / 169 7.2.1 Photolithography with Phase-Shifting Masks / 170 7.2.2 Use of Edge-Defined Defects in SAMs / 172 7.2.3 Controlled Undercutting / 175 7.2.4 Edge-Spreading Lithography / 176 7.2.5 Edge Transfer Lithography / 178 7.2.6 Step-Edge Decoration / 180 7.3 Exposure of Nanoscale Edges / 181 7.3.1 Fracturing of Thin Films / 182 7.3.2 Sectioning of Encapsulated Thin Films / 182 7.3.3 Thin Metallic Films along Sidewalls of Patterned Stamps / 184 7.3.4 Topographic Reorientation / 186 7.4 Conclusion and Outlook / 187 References / 188 8

PATTERNING WITH ELECTROLYTE: SOLID-STATE SUPERIONIC STAMPING

195

Keng H. Hsu, Peter L. Schultz, Nicholas X. Fang, and Placid M. Ferreira

8.1 8.2 8.3 8.4 8.5

Introduction / 195 Solid-State Superionic Stamping / 197 Process Technology / 199 Process Capabilities / 203 Examples of Electrochemically Imprinted Nanostructures Using the S4 Process / 208 Acknowledgments / 211 References / 211 9

PATTERNING WITH GELS: LATTICE-GAS MODELS Paul J. Wesson and Bartosz A. Grzybowski

9.1 Introduction / 215 9.2 The RDF Method / 218

215

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

viii

21:8

Printer Name: Yet to Come

CONTENTS

9.3 Microlenses: Fabrication / 218 9.4 Microlenses: Modeling Aspects / 220 9.4.1 Modeling Using PDEs / 220 9.4.2 Modeling Using Lattice-Gas Method / 221 9.5 RDF at the Nanoscale / 222 9.5.1 Nanoscopic Features from Counter-Propagating RD Fronts / 222 9.5.2 Failure of Continuum Description / 225 9.5.3 Lattice-Gas Models at the Nanoscale / 227 9.6 Summary and Outlook / 229 References / 230 10 PATTERNING WITH BLOCK COPOLYMERS

233

Jia-Yu Wang, Wei Chen, and Thomas P. Russell

10.1 Introduction / 233 10.2 Orientation / 235 10.2.1 Self-Assembling / 235 10.2.2 Self-Directing / 247 10.3 Long-Range / 254 10.3.1 Solvent Annealing / 254 10.3.2 Graphoepitaxy / 256 10.3.3 Sequential, Orthogonal Fields / 260 10.4 Nanoporous BCP Films / 262 10.4.1 Ozonolysis / 264 10.4.2 Thermal Degradation / 264 10.4.3 UV Degradation / 267 10.4.4 Selective Extraction / 271 10.4.5 “Soft” Chemical Etch / 272 10.4.6 Cleavable Junction / 272 10.4.7 Solvent-Induced Film Reconstruction / 274 References / 276 11 PERSPECTIVE ON APPLICATIONS

II APPLICATIONS 12 SOFT LITHOGRAPHY FOR MICROFLUIDIC MICROELECTROMECHANICAL SYSTEMS (MEMS) AND OPTICAL DEVICES Svetlana M. Mitrovski, Shraddha Avasthy, Evan M. Erickson, Matthew E. Stewart, John A. Rogers, and Ralph G. Nuzzo

12.1 Introduction / 295 12.2 Microfluidic Devices for Concentration Gradients / 297

291

293

295

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

CONTENTS

ix

12.3 12.4 12.5 12.6

Electrochemistry and Microfluidics / 300 PDMS and Electrochemistry / 302 Optics and Microfluidics / 306 Unconventional Soft Lithographic Fabrication of Optical Sensors / 314 Acknowledgments / 317 References / 318 13

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

325

Pilnam Kim, Yanan Du, Ali Khademhosseini, Robert Langer, and Kahp Y. Suh

13.1 Introduction / 325 13.2 Fabrication of Nanofluidic System for Biological Applications / 326 13.2.1 Unconventional Methods for Fabrication of Nanochannel / 326 13.2.2 Application of Nanofluidic System / 332 13.3 Fabrication of Biomolecular Nanoarrays for Biological Applications / 338 13.3.1 DNA Nanoarray / 338 13.3.2 Protein Arrays / 340 13.3.3 Lipid Array / 345 13.4 Fabrication of Nanoscale Topographies for Tissue Engineering Applications / 347 13.4.1 Nanotopography-Induced Changes in Cell Adhesion / 347 13.4.2 Nanotopography-Induced Changes in Cell Morphology / 348 References / 349 14

MICRO TOTAL ANALYSIS SYSTEM Yuki Tanaka and Takehiko Kitamori

14.1 Introduction / 359 14.1.1 Historical Backgrounds / 359 14.2 Fundamentals on Microchip Chemistry / 361 14.2.1 Characteristics of Liquid Microspace / 361 14.2.2 Liquid Handling / 362 14.2.3 Concepts of Micro Unit Operation and Continuous-Flow Chemical Processing / 362 14.3 Key Technologies / 365 14.3.1 Fabrication of Microchips / 365 14.3.2 Patterning for Fluid Control / 366 14.3.3 Detection / 366 14.4 Applications / 368 14.4.1 Synthesis / 368

359

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

x

21:8

Printer Name: Yet to Come

CONTENTS

14.4.2 Cell Adhesion Control / 369 14.4.3 Liquid Handling: Valve Using Wettability / 370 References / 372

15 COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES AND THEIR APPLICATION TO CREATE FUNCTIONAL DEVICES

379

Pascale Maury, David N. Reinhoudt, and Jurriaan Huskens

15.1 Introduction / 379 15.2 Top-Down and Bottom-Up Techniques / 380 15.2.1 Top-Down Techniques / 380 15.2.2 Bottom-Up Techniques / 383 15.2.3 Mixed Techniques / 384 15.3 Combining Top-Down and Bottom-Up Techniques for High Resolution Patterning / 385 15.3.1 Top-Down Nanofabrication and Polymerization / 386 15.3.2 Top-Down Nanofabrication and Micelles / 387 15.3.3 Top-Down Nanofabrication and Block Copolymer Assembly / 387 15.3.4 Top-Down Nanofabrication and NP Assembly / 389 15.3.5 Top-Down Nanofabrication and Layer-by-Layer Assembly / 392 15.4 Applicaion of Combined Top-Down and Bottom-Up Nanofabrication for Creating Functional Devices / 397 15.4.1 Photonic Crystal Devices / 397 15.4.2 Protein Assays / 400 References / 406

16 ORGANIC ELECTRONIC DEVICES 16.1 Introduction / 419 16.2 Organic Light-Emitting Diodes / 420 16.3 Organic Thin Film Transistors / 429 References / 439

419

17 INORGANIC ELECTRONIC DEVICES 17.1 Introduction / 445 17.2 Inorganic Semiconductor Materials for Flexible Electronics / 446 17.2.1 “Bottom-Up” Approaches / 447 17.2.2 “Top-Down” Approaches / 449

445

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

CONTENTS

xi

17.3 Soft Lithography Techniques for Generating Inorganic Electronic Systems / 452 17.3.1 Micromolding in Capillaries / 453 17.3.2 Imprint Lithography / 454 17.3.3 Dry Transfer Printing / 454 17.4 Fabrication of Electronic Devices / 459 17.4.1 Transistors on Rigid Substrates via MIMIC Processing / 459 17.4.2 Flexible Inorganic Transistors / 459 17.4.3 Flexible Integrated Circuits / 463 17.4.4 Heterogeneous Electronics / 466 17.4.5 Stretchable Electronics / 469 References / 475 18

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

483

Hanqing Jiang, Jizhou Song, Yonggang Huang, and John A. Rogers

18.1 Introduction / 483 18.2 Buckling Analysis of Stiff Thin Ribbons on Compliant Substrates / 484 18.3 Finite-Deformation Buckling Analysis of Stiff Thin Ribbons on Compliant Substrates / 488 18.4 Edge Effects / 495 18.5 Effect of Ribbon Width and Spacing / 498 18.6 Buckling Analysis of Stiff Thin Membranes on Compliant Substrates / 502 18.6.1 One-Dimensional Buckling Mode / 504 18.6.2 Checkerboard Buckling Mode / 506 18.6.3 Herrington Buckling Mode / 506 18.7 Precisely Controlled Buckling of Stiff Thin Ribbons on Compliant Substrates / 507 18.8 Concluding Remarks / 512 Acknowledgments / 512 References / 512 19

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES Joel Henzie, Min H. Lee, and Teri W. Odom

19.1 Introduction / 515 19.1.1 Brief Primer on Surface Plasmons / 517 19.1.2 Conventional Methods to Plasmonic Structures / 518 19.2 Soft Lithography and Metal Nanostructures / 518 19.3 A Platform for Multiscale Patterning / 520

515

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

xii

21:8

Printer Name: Yet to Come

CONTENTS

19.3.1 Soft Interference Lithography: Patterns on a Nanoscale Pitch / 520 19.3.2 Phase-Shifting Photolithography: Patterns on a Microscale Pitch / 520 19.3.3 PEEL: Transferring Photoresist Patterns to Plasmonic Materials / 521 19.4 Subwavelength Arrays of Nanoholes: Plasmonic Materials / 522 19.4.1 Infinite Arrays of Nanoholes / 523 19.4.2 Finite Arrays (Patches) of Nanoholes / 525 19.5 Microscale Arrays of Nanoscale Holes / 526 19.6 Plasmonic Particle Arrays / 528 19.6.1 Metal and Dielectric Nanoparticles / 528 19.6.2 Anisotropic Nanoparticles / 531 19.6.3 Pyramidal Nanostructures / 531 Acknowledgments / 533 References / 533 20 A RIGIFLEX MOLD AND ITS APPLICATIONS

539

Se-Jin Choi, Tae-Wan Kim, and Seung-Jun Baek

20.1 Introduction / 539 20.2 Modulus-Tunable Rigiflex Mold / 540 20.3 Applications of Rigiflex Mold / 544 20.3.1 From Nanoimprint to Microcontact Printing / 544 20.3.2 Rapid Flash Patterning for Residue-Free Patterning / 547 20.3.3 Continuous Rigiflex Imprinting / 549 20.3.4 Soft Molding Application / 553 20.3.5 Capillary Force Lithography Applications / 556 20.3.6 Transfer Fabrication Technique / 558 References / 561 21 NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY Jong M. Kim, Hwan Y. Choi, Moon-G. Lee, Seungho Nam, Jin H. Kim, Seongmo Whang, Soo M. Lee, Byoung H. Cheong, Hyuk Kim, Ji M. Lee, and In T. Han

21.1 Introduction / 565 21.2 Holographic LGP / 569 21.2.1 Design and Properties of Holographic LGP / 570 21.2.2 NI Technology for the Holographic LGP / 572 21.3 Polarized LGP / 573 21.3.1 Design and Properties of Polarized LGP / 574

565

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

xiii

CONTENTS

21.3.2 Fabrication of the Polarized LGP / 575 21.3.3 Optical Performance of the Polarized LGP / 576 21.4 Reflective Polarizer: Wire Grid Polarizer / 579 21.4.1 Design and Properies of WGP / 580 21.4.2 Fabrication and Applications / 581 21.5 Transflective Display / 585 21.5.1 Design and Optical Properties of Reflecting Pattern / 587 21.5.2 Fabrication of the Reflecting Pattern / 588 References / 592

INDEX

595

P1: OTA/XYZ P2: ABC fm JWBK252/Lee September 26, 2008

21:8

Printer Name: Yet to Come

PREFACE

The area of nanofabrication is a dynamic and rapidly growing field that is sometimes dominated by activity focused on the development of systems for fabrication in microelectronics. In the mid-1990s a growing appreciation of the value of alternative methods, often driven primarily by materials and chemistry rather than by optics and physics, led to the formation of a separate field of study on unconventional or alternative techniques for nanofabrication. Early demonstrations of various forms of soft lithography by George Whitesides (Harvard University), nanoimprint lithography by Stephen Chou (then at University of Minnesota, and presently at Princeton University), and polymer phase separation by Richard Register and Paul Chaikin (Princeton University) were among the important events that catalyzed these developments. One of us (John A. Rogers) was in the Whitesides group as a Harvard Junior Fellow during this time, and has remained active in the field ever since. The interest of the other (Hong H. Lee) derived from a nanoproject on tera bit level memory device. Taken together, we have published more than 250 papers on various aspects of unconventional nanofabrication and its application to diverse device classes, and they have trained more than 80 students in these areas. The gradual maturing of the field and the emergence of meaningful applications provide the motivation for assembling a book at this time. We hope that the outcome will be useful as a reference text for practitioners and developers alike. HHL is thankful to Audrey Lee for assistance. We both are very grateful to Ms Hyewon Kang, who took care of the details of editing for the whole book. John A. Rogers Champaign, USA Hong H. Lee Seoul, Korea

xv

P1: OTA/XYZ P2: ABC c01 JWBK252/Lee September 18, 2008

13:47

Printer Name: Yet to Come

I NANOPATTERNING TECHNIQUES

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


1

P1: OTA/XYZ P2: ABC c01 JWBK252/Lee September 18, 2008

13:47

Printer Name: Yet to Come

1 INTRODUCTION

Tools for nanofabrication are central to every field of nanoscience and nanotechnology. For research and initial development purposes, nanofabrication typically involves the use of specialized techniques to fabricate small collections of nanoscale devices, in processes that resemble a form of craftsmanship. Discoveries that emerge from such work will only yield valuable nanotechnologies, however, when they can be implemented with techniques that can be scaled for cost-effective manufacturing. As a result, the success of nanotechnology depends not only on versatile nanofabrication techniques for discovery in nanoscience, but also on approaches that offer low cost operation and high throughputs, suitable for mass production. In some cases, the techniques might rely on adapted versions of methods whose origins are in the microelectronics industry, such as photolithography and electron-beam lithography. In many others, including certain areas of photonics, microfluidics, biotechnology, and flexible electronics, new approaches are required, either to facilitate commercialization, to allow manipulation of unusual materials, or to enable challenging features sizes and structure geometries. The need for unconventional nanofabrication techniques was recognized broadly in the early 1990s, even before nanotechnology was generally recognized as a separate field of study. During this time, new areas of research emerged around soft lithography, imprint lithography, and various types of self-assembly and scanning probe based patterning methods. The interest in these approaches is driven by their diverse, underlying scientific content, their conceptual novelty, and their technical capabilities for nanofabrication. Their ultimate success, however, is measured first by the extent of their adoption for research purposes and then by their use in manufacturing. Self-assembly and scanning probe techniques will be useful for some applications, but their limited patterning versatility (i.e., materials and geometries) and modest throughput, represent significant disadvantages. Soft lithography and

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


3

P1: OTA/XYZ P2: ABC c01 JWBK252/Lee September 18, 2008

4

13:47

Printer Name: Yet to Come

INTRODUCTION

1994

1996

1998

2000

2002

2004

2006

2002

2004

2006

Number of citations

Number of citations

8000

6000

4000

2000

Number of publications

0 500

Number of publications

400 300 200 100 0 1994

1996

1998

2000

Publication Year Figure 1.1. Numbers of scientific publications (bottom) and citations (top) for the fields of soft lithography and imprint lithography, since 1993.

imprint lithography avoid these problems and, in our view, have significant potential both for research and for realistic implementations in wide-ranging classes of applications. The growth of research in these areas has been explosive, starting with the introduction of microcontact printing, the first form of soft lithography, in 1993 and then imprint lithography in 1995. Figure 1.1 shows the numbers of papers in soft and imprint lithography, and citations of these papers as a function of time. These data indicate clearly the value of these methods for laboratory scale applications and research. This growth and the substantial development work on these methods at small and large companies also suggest an expansion of their use to prototyping and manufacturing. This book covers unconventional methods for nanofabrication, with a focus on soft lithographic and related imprint lithographic methods, but also with a summary of some of the most promising self-assembly methods. The content is organized in two separate parts. The first deals with the principles and underlying science

P1: OTA/XYZ P2: ABC c01 JWBK252/Lee September 18, 2008

13:47

Printer Name: Yet to Come

INTRODUCTION

5

associated with a range of different techniques. In particular, the first chapter covers the classes of materials and surface chemistries that are most commonly used for the stamps, molds, and conformable photomasks of soft lithography. The next several chapters review established and new strategies for using these and analogous “hard” elements in procedures that range from transfer of solid materials to control of the flow of photons to molding of liquid or softened polymers to control of diffusion of chemicals or ions into and out of the substrate. The final chapter in this section demonstrates the power of self-assembly in procedures that rely on polymer phase separation. The second part of the book focuses on applications of these techniques in some of the most promising areas, as outlined in more detail in Chapter 11. The content is intended for practitioners, for researchers developing new methods, and for students in specialized courses in chemistry, physics, biology, chemical engineering, materials science, electrical engineering, or mechanical engineering. The sizes of these communities are growing rapidly, due to the high level of importance of the methods to broad areas of nanotechnology, information technology, biotechnology, and related fields. We hope that this book will help expand even further the reach of these methods and that this expansion will facilitate their further development, potentially culminating with their broad-based use in bridging the gap between nanoscience and nanotechnology.

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2 MATERIALS

2.1 INTRODUCTION The unconventional nanopatterning techniques treated in this book are based primarily on the use of molds or stamps. The features of these techniques are largely determined by the properties of the materials used. These properties include surface energy, Young’s modulus, transparency to light, and compliance or flexibility. For example, the mold material should have a low surface energy for the mold to be released cleanly and easily from the material being patterned. Similarly, if the material to be patterned is an ultraviolet (UV) curable prepolymer, the mold should be transparent to the light. Feature resolution and large area applicability are two main items of interest in any patterning technique. The smallest feature size that can be patterned with a mold is largely determined by the rigidity of the mold, a measure of which is its Young’s modulus, E. Generally, a more rigid mold, or a mold with a higher Young’s modulus, permits a better resolution. The large area applicability is mainly dependent on the extent of compliance of the mold to substrate surface, or mold flexibility. If the extent to which a sheet deflects when it is subjected to a load is used as a measure of flexibility, then the flexibility is determined by Et3 , where t is the sheet thickness (see Section 4.4). Any rigid mold can be made flexible by making the sheet sufficiently thin. In terms of Young’s modulus, there are two extremes for techniques that are in widespread use: molds made from poly(dimethylsiloxane) (PDMS) and those made from silicon. The Young’s modulus of typical PDMS is less than 2 MPa, whereas that of silicon is around 130 GPa. Soft molds, such as those made with PDMS, are used for soft lithography [1] and hard molds, such as those made with silicon, are used for imprint lithography [2]. On the other hand, a mold can have a Young’s modulus

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


7

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

8

13:54

Printer Name: Yet to Come

MATERIALS

between the two extremes, which is more than tens of MPa but less than a few GPa. A mold with a Young’s modulus in this range is rigid enough for nanoscale patterning but at the same time flexible enough in its film form for large area applications. As opposed to soft and hard molds, these molds that are rigid yet flexible are often referred to as “rigiflex” molds [3]. Polymers are typically used for soft and rigiflex molds, whereas silicon, quartz or metals are used for hard molds. Polymer molds are most often prepared by casting a liquid prepolymer onto a master or template and then curing it either photochemically or thermally. The master is typically fabricated by photolithography or electron-beam lithography. While polymer molds can be produced from the master, as many times as desired, the master itself typically becomes the mold for patterning when a hard mold is used. This chapter reviews materials used for molds, beginning with soft molds, and then following with hard molds and rigiflex molds. The surface of a mold is often treated with a material of low surface energy to ensure clean and easy release of the mold from the material being patterned. The need for surface treatment becomes more acute for smaller feature size and more densely populated pattern features. The latter part of the chapter covers this subject of surface treatment.

2.2 MOLD MATERIALS AND MOLD PREPARATION Nanopatterning can be accomplished with hard, soft or rigiflex molds. The choice of mold depends on the requirements of the application. A soft mold is typically used for soft lithography, whereas a hard mold is generally used for imprint lithography. A rigiflex mold can be used in place of a hard or soft mold in most cases. This section covers materials and preparation methods for these three types of molds. 2.2.1 Soft Molds The most representative of soft molds are made from Sylgard 184 (Dow Corning). Molds of this type date back to the first reports of microcontact printing, the first type of soft lithographic technique, in 1993 [4]. Such molds are fabricated by casting a mixture of prepolymer and cross-linker at a recommended ratio of 10:1 against a master with relief structures that correspond to the desired pattern. A curing time of 4–6 h and a curing temperature of 60◦ C should be used. A lower ratio of cross-linker leads to a stickier mold surface. The Sylgard PDMS, sometimes referred to as soft PDMS (s-PDMS), has a number of characteristics and physical properties that are well suited for soft lithography. Its surface energy is low at 21 dyn cm−1 and it is transparent in the UV and visible regions. In addition, it has high gas permeability. For example, the permeability for O2 is 10−11 cm3 cm (cm2 s Pa)−1 . Its flexibility and tackiness allow conformal contact of the mold with the underlying surface. The thermal expansion coefficient is, however, relatively high at 310 µm (m ◦ C)−1 such that a linear shrinkage on the order of 1.5% occurs when cooled after curing at 60◦ C [5]. The mold also has the disadvantage that it swells in many organic solvents as summarized in Table 2.1 [6].

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.2 MOLD MATERIALS AND MOLD PREPARATION

9

Table 2.1. Solubility Parameters, Swelling Ratios, and Dipole Moments of Various Solvents Used in Organic Synthesis

Solvent Perfluorotributylamine Perfluorodecalin Pentane Poly(dimethylsiloxane) Diisopropylamine Hexanes n-Heptane Triethylamine Ether Cyclohexane Trichloroethylene Dimethylethoxyethane (DME) Xylenes Toluene Ethyl acetate Benzene Chloroform 2-Butanone Tetrahydrofurane (THF) Dimethyl carbonate Chlorobenzene Methylene chloride Acetone Dioxane Pyridine N-Methylpyrrolidone (NMP) tert-Butyl alcohol Acetonitrile 1-Propanol Phenol Dimethylformamide (DMF) Nitromethane Ethyl alcohol Dimethyl sulfoxide (DMSO) Propylene carbonate Methanol Ethylene glycol Glycerol Water

δa

Sb

µ(D)

5.6 6.6 7.1 7.3 7.3 7.3 7.4 7.5 7.5 8.2 9.2 8.8 8.9 8.9 9.0 9.2 9.2 9.3 9.3 9.5 9.5 9.9 9.9 10.0 10.6 11.1 10.6 11.9 11.9 12.0 12.1 12.6 12.7 13.0 13.3 14.5 14.6 21.1 23.4

1.00 1.00 1.44 ∞ 2.13 1.35 1.34 1.58 1.38 1.33 1.34 1.32 1.41 1.31 1.18 2.28 1.39 1.21 1.38 1.03 1.22 1.22 1.06 1.16 1.06 1.03 1.21 1.01 1.09 1.01 1.02 1.00 1.04 1.00 1.01 1.02 1.00 1.00 1.00

0.0 0.0 0.0 0.6–0.9 1.2 0.0 0.0 0.7 1.1 0.0 0.9 1.6 0.3 0.4 1.8 0.0 1.0 2.8 1.7 0.9 1.7 1.6 2.9 0.5 2.2 3.8 1.6 4.0 1.6 1.2 3.8 3.5 1.7 4.0 4.8 1.7 2.3 2.6 1.9

Source: Reprinted with permission from [6]. Copyright 2003 American Chemical Society. parameter δ in units of cal1/2 cm−3/2 . b S denotes the swelling ratio that was measured experimentally; S = D/D , where D is the length of 0 PDMS in the solvent and D0 is the length of the dry PDMS, and µ denotes the dipole moment.

a Solubility

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

10

13:54

Printer Name: Yet to Come

MATERIALS

A larger swelling ratio indicates more solvent-induced swelling. A solvent with a solubility parameter close to that of PDMS is a good solvent. For example, triethylamine with the solubility parameter of 7.5 is a better solvent than benzene for which the parameter is 9.2 since the solubility parameter of PDMS is 7.3 cal1/2 cm−3/2 . The modulus of elasticity or Young’s modulus of s-PDMS (Sylgard 184) is around 2 MPa, depending on the mixing method, curing time, and temperature. This low modulus limits the fabrication of features with high aspect ratios due to collapse, merging, and buckling of the structures of relief [7–9]. These deformation modes have been examined both theoretically [8, 9] and experimentally [10]. The theoretical criteria [8–10] that can be used for dimensional stability and conformity of a mold are summarized in Table 2.2. Conformity here means full contact of the mold feature with the underlying substrate surface. The criteria given in the table are such that if they are satisfied then the mold avoids the associated deformation. For instance, if the equation in the first entry of the table is satisfied, then the mold will not undergo roof collapse. The results suggest that roof collapse, buckling, and lateral collapse (merging) can be avoided by increasing E∗ or Young’s modulus. On the other hand, the conformity decreases when Young’s modulus increases. Deformations other than those related to conformity can prevent accurate patterning. It is instructive to examine the criteria for an equal line and space pattern, for which a = w or a/w = 1. For a given load, the roof collapse is determined by 1/(EA), where A is the aspect ratio given by h/a or h/w. Roof collapse, therefore, is less likely to occur for a mold with a higher aspect ratio and higher Young’s modulus. On the other hand, buckling is more likely to occur for a mold with a higher aspect ratio but less likely for a mold with a higher Young’s modulus because buckling is determined by A/E. Lateral collapse is determined by A/(Ea)1/4 . Therefore, lateral collapse is more likely to occur if the aspect ratio is higher, the Young’s modulus is lower, or the feature size is smaller. The conformity is determined by EA, meaning that conformal contact improves as Young’s modulus and the aspect ratio decrease. As the feature size is reduced, the lateral collapse problem becomes more acute. This is one of the reasons why a s-PDMS mold cannot be used for feature sizes smaller than several hundred nanometers for equal line and space patterns with an aspect ratio larger than unity. An obvious way to overcome unwanted deformations is to use a mold with a higher Young’s modulus. Therefore, there have been attempts to use materials of high Young’s modulus for soft molds. The earliest example used alternative siloxane polymers [5] having a Young’s modulus of around 9 MPa, known as hard PDMS (h-PDMS). To overcome the shortcomings of h-PDMS, such as its brittleness, and the need to apply pressure to achieve conformal contact with a substrate, a composite mold of PDMS was introduced [7], in which a thin h-PDMS layer with relief structure is supported by a thick layer of s-PDMS. This composite design combines the advantages of both a more rigid layer (to achieve high resolution patterning) and a more flexible support (to facilitate handling and the establishment of conformal contact) [7]. Figure 2.1 illustrates the procedure for preparing such composite elements. The h-PDMS is formed [7] by mixing and degassing for 1–2 min 3.4 g of a vinyl PDMS prepolymer (VDT-731, Gelest Corp.), 18 µL of a Pt catalyst (platinum

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.2 MOLD MATERIALS AND MOLD PREPARATION

11

Table 2.2. Dimensional Stability and Conformity Criteria

σ∞

Roof collapse: unwanted contact

−

a 4σ∞ w
1 + cosh−1 π E ∗h w −1   wπ <1 − cos 2(w + a)

σ∞ 12σ∞ h 2 1 < π 2 E ∗a2 1 + (w/a)

Buckling

−

Lateral collapse

h 2a

Smooth surface asperities (conformity)

w −π E ∗ h <1+ 2σ∞ L a

Radius of an edge rounded by surface tension

R



R∼

2γs 3E ∗ a



1/4 <

w a

γ 2E

Source: Reprinted with permission from [8]. Copyright 2002 American Chemical Society. E ∗ = E/(1/ν 2 ): E—Young’s modulus; ν—Poisson ratio; γ —surface tension; L—period (2(a+w)); σ —remote stress (load).

divinyltetramethyldisiloxane, SIP 6831.1, Gelest Corp.), and one drop (approximately 0.1 wt%) of a moderator (2,4,6,8-tetramethyl-tetravinylcyclotetrasiloxane, 87927, Sigma-Aldrich). Then, 1 g of a hydrosilane prepolymer (HMS-301, Gelest Corp.) is stirred into this mixture. Within 3 min, a thin layer (30–40 µm) of h-PDMS is spin coated onto a master and cured for 30 min at 60◦ C. Then, a liquid polymer layer (∼3 mm) of Sylgard 184 PDMS is poured onto the h-PDMS layer and cured for at least an hour at 60◦ C. The composite mold is released from the surface by cutting and peeling the mold from the surface while warm.

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

12

13:54

Printer Name: Yet to Come

MATERIALS

Master

Si Si

500 nm

Spin coat h-PDMS h-PDMS

40 µm

Si Si Cast 184 PDMS

3 mm

184 PDMS

h-PDMS

Si Si Release from master Composite stamp

184 PDMS

h-PDMS

Figure 2.1. Procedure for fabricating a two-layer composite stamp. h-PDMS is spin coated onto a master and cured at 60◦ C for 30 min (after this process, the layer is still tacky). PDMS 184 prepolymer is poured on top of this h-PDMS layer and cured at 60◦ C for at least 1 h. The composite stamp is released from the master by (i) cutting around the patterned areas with a razor blade and (ii) removing the stamp from the surface using tweezers. (Reprinted with permission from [7]. Copyright 2002 American Chemical Society.)

In preparing molds with casting and curing procedures such as those in Figure 2.1, the mode of interaction between the mold material and the master is extremely important. The role of fluid dynamics and wettability in preparing the PDMS mold are usually ignored. One of the most common reasons for the unsuccessful fabrication of a mold, particularly at small feature sizes, is the inability of the liquid prepolymer to penetrate into and fill up the holes and grooves in the master template [11]. The main factor that governs flow in this process is viscosity. The viscosity can be lowered by simply mixing the prepolymer with a solvent. It has been shown for a polymer solution that the viscosity at a polymer volume fraction α, η(α), is related to that at another volume fraction β, η(β), as follows [12]: η(α) = η(β)

 5.1 α β

(2.1)

The relationship indicates that a considerable reduction in viscosity can be realized by introducing a solvent. When PDMS oligomers are diluted with toluene to 69% by weight, the measured viscosity decreases from 362 to 50 cP. Although the weight fraction is different from the volume fraction unless the density is unity, a reduction to 69 wt% should lead to a viscosity of approximately 55 cP according to equation 2.1, which compares well with the measured value of 50 cP [11]. The rate of curing of the PDMS is also important. For example, the curing process should be delayed until after the solvent is completely volatilized and only the prepolymer fills up the void space in the master. Otherwise, incomplete filling and/or pattern shrinking will result. The moderator used in preparing the PDMS mold

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.2 MOLD MATERIALS AND MOLD PREPARATION

13

lowers the rate of cross-linking but does not take part in the cross-linking reactions. Therefore, an excessive amount of moderator can be used to give more time for the solvent to evaporate. The method of preparing composite PDMS mold by this improved method [11] is the same as that given above except that 0.1 g of moderator is used. This amounts to 2.2% by weight compared with 0.1 wt% used in the original h-PDMS preparation [5]. The mixture is then blended with 2 g of a chosen solvent, making the total weight 6.5 g. After the mixture is spin coated onto a master, the spin-coated h-PDMS is left undisturbed for an hour to provide enough time for the penetration of the solution into the voids of the master and for solvent evaporation. The material is then cured for 30 min at 60◦ C. A layer of liquid prepolymer of Sylgard 184 PDMS is cast on this h-PDMS, and the entire system is capped with a glass backplane. The hard backplane prevents distortion of the PDMS mold. After degassing, it is cured for an hour at 60◦ C. Figure 2.2 presents cross-sectional scanning electron microscope (SEM) images of representative patterning results achieved using such a composite element to mold

(a)

(b)

(c)

Figure 2.2. Comparison between soft lithographic molding using unoptimized and optimized procedures for making the molds. SEM images of (a) a 100 nm line and space master pattern, (b) pattern of polymer created using a composite PDMS stamp prepared in the usual way, and (c ) in an improved manner. The scale bar is 500 nm. (Reprinted with permission from [11]. Copyright 2006 Institute of Physics (the “Institute”) and IOP Publishing Limited (“IOPP”).)

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

14

13:54

Printer Name: Yet to Come

MATERIALS

(a)

(b)

(c)

(d)

Figure 2.3. SEM images of (a) the master pattern (75-nm lines separated by 75 nm) and layers of polymer patterned with composite molds prepared using solvents of (b) triethylamine, (c ) toluene, and (d ) hexane. (Reprinted with permission from [11]. Copyright 2006 Institute of Physics (the “Institute”) and IOP Publishing Limited (“IOPP”).)

a liquid photopolymer (see section 3.2.2 for the patterning method (soft molding) used). The results compare the master pattern shown in Figure 2.2a, which consists of 100-nm-wide lines spaced by 100 nm, and the polymer pattern formed by soft molding [13]. In this case, only a small fraction of the relief on the master is replicated with composite molds created in the usual way without the moderator, as shown in Figure 2.2b. With the improved method of preparing the composite mold, the master pattern can be replicated as shown in Figure 2.2c. Published results [11] indicate that even 40-nm-wide lines separated by 50 nm and with heights greater than 60 nm can be replicated. It is instructive to see how the choice of a solvent can affect the fidelity of the pattern formed, due to the solvent’s wettability. The importance of the wettability as affected by the choice of a solvent is illustrated in Figure 2.3. Shown in Figure 2.3a is a master pattern consisting of 75 nm wide aluminum lines separated by 75 nm on a glass substrate. The line height is 150 nm, corresponding to an aspect ratio of 2. Because of the native aluminum oxide surface and glass substrate, the surface is hydrophilic. Therefore, a polar solvent should result in a better wettability and thus a PDMS mold with improved pattern fidelity. Three solvents were chosen from the list of good solvents of PDMS (Table 2.1): triethylamine, toluene, and hexane. Of these, triethylamine is the most hydrophilic and hexane is the least. These solvents

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.2 MOLD MATERIALS AND MOLD PREPARATION

15

were used in preparing PDMS molds by the improved method and these molds were in turn used to pattern poly(vinylpyridine) layers coated onto silicon substrates. Patterning results are shown in Figures 2.3b, c, and d for PDMS molds prepared with triethylamine, toluene, and hexane, respectively. It is clear from the cross-sectional SEM micrographs that the PDMS mold prepared with the most hydrophilic solvent triethylamine yields the best pattern fidelity (Figure 2.3b) and that the mold prepared with the nonpolar solvent hexane does not produce the desired pattern at all, despite the lowered viscosity. Although this example corresponds to the case of a hydrophilic surface, the wettability of a liquid on any surface can be determined from the spreading coefficient (see Section 3.3.2). A solvent with the largest spreading coefficient should provide the best pattern fidelity. Recent studies on microcontact printing with PDMS have shown that cured s-PDMS and h-PDMS contain low molecular weight components, either volatile or non-cross-linked, that can migrate to the surface and cause contamination during printing or molding operations [14–18]. The degree of contamination depends on the formulation of the mold material, the curing conditions of the mold (i.e., curing time and temperature), the nature of the contact surface, as well as the stamping conditions (i.e., inking, stamping time, pressure) [14, 19]. The nature of the ink, in the case of printing, also affects the degree of contamination [15]: polar inks show higher contamination. Several cleaning processes have been demonstrated that minimize the degree of contamination, including long (e.g., 1 week) extraction cleaning [16], oxygen plasma treatment [20], UV/ozone treatment [17], and use of a Soxhlet extractor [14]. Without cleaning, the levels of contamination can be sufficiently high, in fact, to be used in an “inkless” printing process [21], or in a strategy to create hydrophobic regions on biochips [22]. The presence of the low molecular weight PDMS fragments can also be advantageous in the printing of certain biomolecules such as DNA [14]. To further improve the resolution in patterning with a soft mold and to overcome some of the shortcomings of PDMS such as swelling, a commercially available form of perfluoropolyether (PFPE) (CN 4000, Sartomer Co., MW = 1000 g mol−1 ) has been explored as a mold material. Unlike certain PFPE formulations described earlier [23], which have a Young’s modulus of around 4 MPa and have been demonstrated for microfluidic device fabrication [24] and imprint lithography [23, 25], the Young’s modulus of acryloxy PFPE, denoted as a-PFPE, is 10.5 MPa [26]. The low surface energy (∼18.5 mN m−1 ) as well as the chemical inertness of a-PFPE eliminates the necessity of treating the surfaces of the masters with fluorinated silanes to avoid sticking during casting and curing of the mold. Figure 2.4 illustrates the fabrication of composite a-PFPE molds using backing layers of s-PDMS and polyethylene terephthalate (PET). The photocurable fluorinated acrylate oligomer CN 4000 is mixed with 0.5 wt% of a photoinitiator (Darocurr 4265, Ciba Specialty Chemicals) and filtered through a 0.22 µm syringe filter. A thin layer (∼2 µm) of the a-PFPE resin is spin coated (4000 rpm, 30 s) on the master and cured in UV light (350–380 nm, 4 mW cm−2 ) under nitrogen purge for 2 h. Then a layer (∼4 mm) of Sylgard 184 PDMS is poured onto the a-PFPE layer and cured at room temperature for ∼48 h or at 65◦ C for 2 h. The a-PFPE/s-PDMS composite mold is peeled off from

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

16

13:54

Printer Name: Yet to Come

MATERIALS

1 nm–5 µm

Substrate

1 nm–1.5 µm

Master

~1–5 µm

a-PFPE Substrate

Backing Support

0.3–5 mm

s-PDMS or PET film

Substrate

Back support a-PFPE

a-PFPE composite stamp

Figure 2.4. Fabrication of composite a-PFPE mold with either s-PDMS or PET films as backing supports. (Reprinted with permission from [26]. Copyright 2007 American Chemical Society.)

the master. When the master consists of molecular scale patterns such as a network of single-walled carbon nanotubes (SWNTs), the composite mold benefits from the use of a comparatively high modulus backing support of PET (film, ∼0.3 mm thick) to minimize distortions. The fabrication in this case begins with placing a few drops of a-PFPE onto the master. A PET film with an adhesion layer of polyurethane (NOA-73, Norland Optical Adhesives, 1:2 diluted with acetone, and spin coated with 3000 rpm for 30 s) is placed on top of the a-PFPE, spreading the liquid resin over the master. The a-PFPE is then cured with UV light (350–380 nm, 4 mW cm−2 ) for 2 h in nitrogen purge. The a-PFPE/PET composite mold is released from the surface by peeling the mold from the master [26]. Figure 2.5 shows SEM images of h-PDMS/s-PDMS molds (parts a, c, e) and of a-PFPE/s-PDMS molds (parts b, d, f ). Patterns consisting of arrays of cylindrical holes with diameters of 450 nm, periodicities of 750 nm, and depths of 300 nm could be replicated accurately with both a-PFPE and h-PDMS as shown in Figures 2.5a and b. For smaller features, h-PDMS exhibits defects in the form of merged posts in Figure 2.5c or missing posts in Figure 2.5e, whereas those defects are not present in the case of a-PFPE (Figures 2.5d and f ). These improvements result primarily from the more efficient release of the a-PFPE from the master compared to the case of h-PDMS [26]. The resolution capabilities of both a-PFPE and h-PDMS are remarkable. In recent experiments, they have been used for molecular scale molding where SWNTs with diameters of 1–3 nm act as the templates [26–28]. Systematic studies show that the

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.2 MOLD MATERIALS AND MOLD PREPARATION

17

(b)

(a)

1 µm

1 µm

5 µm

5 µm (d )

(c)

2 µm

2 µm (f )

(e)

1 µm

1 µm

Figure 2.5. SEM images of composite molds of h-PDMS/s-PDMS (a, c , e) and a-PFPE/s-PDMS (b, d , f ). Elements shown in (a) and (b) consist of square arrays of posts (diameter = 450 nm, periodicity = 750 nm, and relief depth = 300 nm). Elements in (c ) and (d ) consist of hexagonal arrays of posts (diameter = 350 nm, periodicity = 500 nm, and relief depth = 420 nm). Elements in (e) and (f ) consist of hexagonal arrays of posts (diameter = 130 nm, periodicity = 800 nm, and relief depth = 420 nm). (Reprinted with permission from [26]. Copyright 2007 American Chemical Society.)

resolution for these materials extends nearly to the ∼1-nm regime, where the ultimate resolution tends to be limited by the surface roughness of the molded patterns. This roughness derives partly from the molds and partly from the molded polymers themselves. Experiments illustrate a correlation between the resolution limit and the average distance between cross-links in the PDMS. As demonstrations, Figures 2.6a and b show the atomic force microscopic (AFM) images of a network of SWNTs on an Si/SiO2 substrate and the molded structure on polyurethane (PU), respectively. Figure 2.6c compares the replication fidelity between a-PFPE and h-PDMS [26].

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

18

13:54

Printer Name: Yet to Come

MATERIALS

(a)

(b) 10 nm

1.37 nm

1.07 nm

500 nm

(c)

0.63 nm

0

500 nm

0.39 nm

4 Replica from a-PFPE Replica from h-PDMS *

Height on replica (nm)

3

2

1

0 0

1

2

3

4

Height on template (nm) Figure 2.6. (a) Atomic force microscopic (AFM) image of a network of individual SWNTs that serve as a master for molecular scale imprinting. (b) AFM image of a structure molded onto the surface of a layer of polyurethane (PU) by use of an a-PFPE mold created from the master shown in (a). (c ) Comparison of the heights of features formed by molding with the heights of the corresponding SWNTs on the master as obtained with a-PFPE and h-PDMS molds. (Reprinted with permission from [26]. Copyright 2007 American Chemical Society.)

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.2 MOLD MATERIALS AND MOLD PREPARATION

19

The high resolution capabilities of a-PFPE, together with other properties such as resistance to swelling, chemical inertness, and photocurability make it a promising alternative to PDMS for certain soft lithographic techniques [26]. 2.2.2 Hard Molds Hot embossing has traditionally been used for various purposes perhaps most prominently in the production of compact discs. This procedure as applied to sub-100 nm and dimensions implemented with an etching procedure to define isolated features is known as nano imprint lithography (NIL) [2]. In addition to NIL, there are two other closely related lithographic techniques: step and flash imprint lithography (SFIL) [29] and room temperature imprint lithography (RTIL) [30]. Hard molds such as those prepared by photolithography or electron-beam lithography on silicon wafers are used for NIL and RTIL. The Young’s modulus of these hard molds is high enough to withstand the high applied pressures that are often required for patterning. Patterning by NIL is based on pressure-driven flow of a polymer melt; patterning by RTIL relies on the plastic deformation of polymer. High pressures are required in both cases. By contrast, low pressures are sufficient in SFIL due the use of low viscosity UV curable liquid prepolymers in this technique. Quartz molds are typically used for SFIL, due to their high modulus, dimensional stability and transparency in the UV range. Unlike soft molds, which can be replicated many times easily and cheaply from a single master, the master template itself is usually used for hard molds. For NIL and RTIL, hard molds in the form of nickel stamps can be replicated from a master. If the master is a silicon mold, a thin metal layer can be coated and nickel electroplating can be carried out to replicate the master [31]. Another method is to deposit a seed layer by electroless deposition, followed by nickel electroplating in a nickel sulfamate solution [32]. For quartz molds used for SFIL, the transparency requirement precludes the method of nickel electroplating for replication. 2.2.3 Rigiflex Molds A rigiflex mold is one with a Young’s modulus between several tens of MPa and a few GPa. This type of mold is rigid enough for high resolution patterning and yet flexible enough in its film form for conformal contact with a surface. It is noted in this regard that the flexibility is inversely proportional to the product of the Young’s modules and the thickness cubed. Therefore, a 1-µm-thick sheet with the modulus of 1 GPa is as flexible as a 10-µm-thick sheet with the modulus of 1 MPa. As an example, a 1-mm-thick Teflon mold, for which the modulus is 1.6 GPa, is more flexible than a 1-cm-thick s-PDMS mold, for which the modulus is around 2.0 MPa. Rigiflex molds based on Teflon and UV curable poly(urethaneacrylate) (PUA) have been introduced recently [3]. The PUA mold can be prepared [33] by pouring liquid prepolymer (MINS, Minuta Inc.) onto a master template. A flexible and transparent backing layer of PU elastomer or soft epoxy resin is brought into contact with the liquid prepolymer. It

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

20

13:54

Printer Name: Yet to Come

MATERIALS

is then exposed to UV (250–400 nm) for 2 min at room temperature through the transparent backside. After the curing, it is peeled off from the master. The Young’s modulus of MINS ERM is around 100 MPa and that of MINS 311RM is around 400 MPa. The mold can be self-replicated and its surface energy is around 23 dyn cm−1 , which is comparable to that of PDMS. It is almost impermeable to gases, inert to chemicals and solvents such that there is no swelling problem, and it is transparent to light in the UV and visible regions. Because of the rigiflex nature of the mold and the tunability of the modulus, it can be used [34] not only for microcontact printing but also for imprinting. More details can be found in Chapter 20. Another type of rigiflex mold is one made of amorphous Teflon (Dupont, AF 2400), for which the Young’s modulus is 1.6 GPa. This material has a low surface energy, 15.6 dyn cm−1 such that no surface treatment of the mold is needed in patterning. It is inert to all solvents except for perfluorinated solvents (3M, FC-77) such that there is no swelling problem. It is also inert to all chemicals. It is transparent to light in the region between deep UV and near infrared. Therefore, it can be used for SFIL in place of a quartz mold. In addition, it has a large gas permeability. For example, the permeability for O2 is 10−7 cm3 cm (cm2 s Pa)−1 , which is much larger than that for PDMS (10−11 cm3 cm (cm2 s Pa)−1 ). Therefore, it can be used for any lithography involving a solvent such as soft molding [13] and micromolding in capillaries (MIMIC) [35]. There are two ways of preparing the Teflon mold: solvent-casting method [36] and compression molding [37]. The AF 2400 fluoropolymer powder can be dissolved in a perfluorinated solvent (3M, FC-77) up to about 2 wt%. The fluoropolymer solution is poured into a Petri dish that contains a master template for replication. The solvent does not dissolve other materials, such as inorganics, metals, and even organics except for fluorinated materials. After the solvent evaporates completely, the dried film is peeled off from the master. The boiling point of the solvent is 97◦ C. In this solventcasting method, it is desirable to have a film Teflon mold that is more than 100 µm thick for ease of handling. In the compression molding, a jig containing the master is filled with the fluoropolymer powder. This jig is placed between the platens of a hydraulic press with heating elements. It is then heated to 350◦ C for 10–20 min at a pressure of 150 MPa. After compression molding, the sample is cooled to room temperature while maintaining the pressure to prevent mold deformation. The compression-molded stamp can be separated manually from the master mold. The thin film mold prepared by the solvent-casting method requires careful handling and it is difficult to reuse repeatedly. Although it can be made thick by repetitive solvent casting and evaporation, this procedure is cumbersome because the maximum solubility in the solvent is only 2% by weight. The block mold resulting from the compression molding is not transparent unless prepared in a vacuum because of air bubbles trapped between the master mold and Teflon powders. One disadvantage of compression molding is that only masters made of hard materials such as silicon can be used to prepare the molds. This feature precludes the use of masters made of organics, such as patterns of photoresist, due to the use of high pressures and temperatures needed for compression molding. A “solutal” method [38] can be used to prepare a Teflon mold that is thick enough for easy handling and with good transparency. In this method, a flat, unpatterned

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

2.3 SURFACE TREATMENT AND MODIFICATION

(a)

(b) Stainless steel

21

Teflon sheet Teflon solution Master

AF 2400 Teflon powder Polyimide film

Jig

Pressing

Heating

Teflon mold

Figure 2.7. (a) Schematic illustration of the procedure for preparing a flat Teflon sheet. (b) Schematic illustration of the procedure for making a Teflon mold. (Reprinted with permission from [38]. Copyright 2004 American Institute of Physics.)

Teflon sheet is prepared by compression molding as shown in Figure 2.7a, which is used as a backing layer. For the preparation, Teflon powders are placed in a jig on a hot plate. The jig surface is covered with a stainless steel sheet and heated to 300◦ C and held there for a few minutes. Pressure is then applied in steps of 1 MPa for 1 min in raising the pressure from 1 to 7 MPa at 300◦ C. When the final pressure is reached, the pressure is held for 2–5 min, after which the heater is turned off, allowing for it to cool naturally to room temperature. Two polyimide sheets are placed above and below the jig for easy release of the compression-molded Teflon sheet. This flat Teflon sheet is transparent. With this support sheet, the Teflon sheet mold is prepared as shown in Figure 2.7b. The Teflon solution (0.5 wt% fluoropolymer in FC-77 solvent) is dispensed onto the master having the desired pattern. The sheet made in Figure 2.7a is then placed on the solution-coated master on a hot plate at 100◦ C. A pressure of 0.43 MPa is applied on the Teflon sheet for 20 min. After cooling to room temperature, the pressure is relieved and the prepared Teflon mold is peeled off. The bonding between the sheet and the thin Teflon film on the master occurs through the slight dissolution of the sheet surface upon contacting the solution on the master at the applied pressure. The heating is only for removing the solvent.

2.3 SURFACE TREATMENT AND MODIFICATION When necessary, the surface of a mold is treated to have a surface energy low enough that the material being patterned does not stick to the mold surface when the mold is removed after patterning. The molds made of fluoro-compounds such as Teflon and a-PFPE do not require any surface treatment because their surface energy is very

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

22

13:54

Printer Name: Yet to Come

MATERIALS

low at around 15 dyn cm−1 . In fact, surface treatments of molds made from other materials usually involve the use of fluorinated compounds. The surface energies of PDMS and PUA are approximately 21 and 23 dyn cm−1 , respectively. Molds made of these polymers are, in general, used without any surface treatment except in transfer patterning and printing. However, the hard molds that are typically used for imprint lithography, such as those made with silicon or quartz, do require surface treatment because the surface energies are high due to the hydroxyl groups present on the surfaces. For mold materials with suitable chemistry, self-assembled monolayers (SAMs) are used for the surface treatment. The SAM material most frequently used is tridecafluoro-1,1,2,2,-tetrahydrooctyltrichlorosilane (CF3 -(CF2 )5 (CH2 )2 SiCl3 ) (FOTCS). The chlorine in the trichlorosilane reacts in the presence of moisture with the hydroxyl groups on the mold surface to form Si–O–Si linkages with the fluorinated alkyl chain intact. Before treating with the SAM, the wafer mold is dipped in a piranha solution or treated with oxygen plasma to create surface hydroxyl groups. Two different methods can be used to self-assemble the monolayer: liquid- or vapor-phase treatment. Of the two methods, the vapor-phase treatment is known to lead to a lower surface energy [39, 40]. In a simple liquid-phase treatment [41] without any pretreatment, the mold can be dipped into a 6 × 10−4 M FOTCS solution in anhydrous heptane for a period of time, rinsed with heptane after removing it from the solution, and then dried at 80◦ C for 20 min to drive off any unbonded FOTCS. An optimum dipping time of 10 min has been found to yield the lowest surface energy at approximately 15 dyn cm−1 for the case of silicon molds [41]. Elaborate schemes [39, 40] can be used for the vapor-phase treatment. However, a simple method usually suffices. A container, such as a Chalet filled with a FOTCS solution, can be placed in an enclosure in which the patterned silicon wafer or mold is also placed. The enclosure becomes saturated with FOTCS vapor and the chemical adsorption of FOTCS takes place. After an hour, the mold is removed and then heated at 120◦ C for 5 min to complete the treatment. As indicated earlier, PDMS and PUA molds do not require surface treatment for patterning except for transfer patterning or transfer printing (Chapter 4). In the case of the rigiflex mold of PUA, the surface is treated to facilitate the transfer of the layer deposited on the mold to a substrate surface. The same surface treatments discussed for the hard mold can be used for the rigiflex mold prior to depositing a material on the mold. In these treatments, a surface carboxyl functional group (COOH) is involved instead of a hydroxyl (OH) group in the case of silicon wafers. Vapor deposition of a fluoro-compound can also be used to lower the surface energy on PUA. An example is the vapor deposition of fluorinated ethylene propylene (FEP). The film thickness should be at least 5 nm for the film to be effective, and annealing at 200◦ C helps improve the effectiveness of the film [42]. No annealing is needed when the film thickness is thicker than 10 nm. The surfaces of PDMS molds are treated when it is used for transfer printing or lamination (Chapter 4) to make the surface more or less adhesive to the material transferred or laminated. In addition to chemistries related to those described above, oxygen plasma treatment can also be used. The surface of PDMS is terminated with

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

REFERENCES

23

methyl groups, making the surface hydrophobic and a low energy surface. Upon exposing to oxygen plasma [43], these groups turn into silanol groups. The surface hydroxyl groups make the surface hydrophilic and a high energy surface. However, the surface does not remain OH terminated because uncross-linked PDMS chains tend to migrate to the surface, making it methyl terminated again, within 1 day if the surface is exposed to air. Keeping the surface exposed to water prevents this process. If desired, low molecular weight residues can be extracted by putting PDMS in various organic solvents [16]. The oxygen plasma treatment is often used for the fabrication of microfluidic channels. A PDMS mold with protruding lines and recessed spaces can be sealed with a flat PDMS sheet to yield microfluidic channels. The sealing is irreversible when both surfaces are treated with oxygen plasma. This irreversible sealing of the patterned PDMS mold is possible with a flat sheet of glass, silicon, polystyrene, polyethylene, or silicon nitride [44].

REFERENCES 1. Xia, Y. and Whitesides, G. M. (1998) Soft lithography. Angew. Chem. 37, 550–575. 2. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1995) Imprint of sub-25 nm vias and trenches in polymers. Appl. Phys. Lett. 67, 3114–3116. 3. Suh, D., Choi, S. J., and Lee, H. H. (2005) Rigiflex lithography for nanostructure transfer. Adv. Mater. 17, 1554–1560. 4. Kumar, A. and Whitesides, G. M. (1993) Features of gold having micrometer to centimeter dimension 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. 5. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high accuracy soft lithography. Macromolecules 33, 3042–3049. 6. Lee, J. N., Park, C., and Whitesides, G. M. (2003) Solvent compatibility of poly(dimethylsiloxane)-based microfluidic devices. Anal. Chem. 75, 6544–6554. 7. Odom, T. W., Love, J. C., Wolfe, D. B., Paul, K. E., and Whitesides, G. M. (2002) Improved pattern transfer in soft lithography using composite stamps. Langmuir 18, 5314–5320. 8. Hui, C. Y., Jagota, A., Lin, Y. Y., and Kramer, E. J. (2002) Constraints on microcontact printing imposed by stamp deformation. Langmuir 18, 1394–1407. 9. Huang, Y. Y., Zhou, W., Hsia, K. J., Menard, E., Park, J.-U., Rogers, J. A., and Alleyne, A. G. (2005) Stamp collapse in soft lithography. Langmuir 21, 8058–8068. 10. Sharp, K. G., Blackman, G. S., Glassmaker, N. J., Jagota, A., and Hui, C. Y. (2004) Effect of stamp deformation on the quality of microcontact printing: theory and experiment. Langmuir 20, 6430–6438. 11. Kang, H., Lee, J., Park, J., and Lee, H. H. (2006) An improved method of preparing composite poly(dimethylsiloxane) moulds. Nanotechnology 17, 197–200. 12. Khang, D. Y. and Lee, H. H. (2000) Room-temperature imprint lithography by solvent vapor treatment. Appl. Phys. Lett. 76, 870–872.

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

24

13:54

Printer Name: Yet to Come

MATERIALS

13. Kim, Y. S., Suh, K. Y., and Lee, H. H. (2001) Fabrication of three-dimensional microstructures by soft molding. Appl. Phys. Lett. 79, 2285–2287. 14. Thibault, C., Severac, C., Mingotaud, A. F., Vieu, C., and Mauzac, M.,(2007) Poly(dimethylsiloxane) contamination in microcontact printing and its influence on patterning oligonucleotides. Langmuir 23, 10706–10714. 15. Sharpe, R. B. A., Burdinski, D., van der Marel, C., Jansen, J. A. J., Huskens, J., Zandvliet, H. J. W., Reinhoudt, D. N., and Poelsema, B. (2006) Ink dependence of poly(dimethylsiloxane) contamination in microcontact printing. Langmuir 22, 5945–5951. 16. Graham, D. J., Price, D. D., and Ratner, B. (2002) Solution assembled and microcontact printed monolayers of dodecanethiol on gold: a multivariate exploration of chemistry and contamination. Langmuir 18, 1518–1527. 17. Glasmastar, K., Gold, J., Andersson, A. S., Sutherland, D. S., and Kasemo, B. (2003) Silicone transfer during microcontact printing. Langmuir 19, 5475–5483. 18. Andersson, L. H. U. and Hjertberg, T. (2003) Silicone elastomers for electronic applications. I. Analyses of the noncrosslinked fractions. J. Appl. Polym. Sci. 88, 2073–2081. 19. Hale, P. S., Kappen, P., Prissanaroon, W., Brack, N., Pigram, P. J., and Liesegang, J. (2007) Minimizing silicone transfer during micro-contact printing. Appl. Surf. Sci. 253, 3746–3750. 20. Langowski, B. A. and Uhrich, K. E. (2005) Oxygen plasma-treatment effects on Si transfer. Langmuir 21, 6366–6372. 21. Wang, X., Ostblom, M., Johansson, T., and Inganas, O. (2004) PEDOT surface energy pattern controls fluorescent polymer deposition by dewetting. Thin Solid Films 449, 125–132. 22. Asberg, P., Nilsson, K. P. R., and Inganas, O. (2006) Surface energy modified chips for detection of conformational states and enzymatic activity in biomolecules. Langmuir 22, 2205–2211. 23. Rolland, J. P., Hagberg, E. C., Denison, G. M., Cater, K. R., and De Simone, J. M. (2004) High-resolution soft lithography: enabling materials for nanotechnologies. Angew. Chem. Int. Ed. 43, 5796–5799. 24. Rolland, J. P., Van Dam, R. M., Schorzman, D. A., Quake, S. R., and De Simone, J. M. (2004) Solvent-resistant photocurable “liquid teflon” for microfluidic device fabrication. J. Am. Chem. Soc. 126, 2322–2323. 25. Rothrock, G. D., Maynor, B., Rolland, J. P., and De Simone, J. M. (2006) HighPerformance Imprint Lithography and Novel Metrology Methods Using Multifunctional Perfluoropolyethers, Archie, C. N., ed., SPIE, Bellingham, WA, p. 61523F. 26. Truong, T. T., Lin, R., Jeon, S., Lee, H. H., Maria, J., Gaur, A., Hua, F., Meinel, I., et al. (2007) Soft lithography using acryloxy perfluoropolyether composite stamp. Langmuir 23, 2898–2905. 27. Hua, F., Sun, Y., Gaur, A., Meitl, M. A., Bilhaut, L., Rotkina, L., Wang, J., Geil, P., et al. (2004) Polymer imprint lithography with molecular-scale resolution. Nano Lett. 4, 2467–2471. 28. Hua, F., Gaur, A., Sun, Y., Word, M., Jin, N., Adesida, I., Shim, M., Shim, A., et al. (2006) Processing dependent behavior of soft imprint lithography on the 1–10-nm scale. IEEE Trans. Nanotechnol. 5, 301–308.

P1: OTA/XYZ P2: ABC c02 JWBK252/Lee September 18, 2008

13:54

Printer Name: Yet to Come

REFERENCES

25

29. Ruchhoeft, P., Colburn, M., Choi, B., Nounu, H., Johnson, S., Bailey, T., Damle, S., Stewart, M., et al. (1999) Patterning curved surfaces: template generation by ion beam proximity lithography and relief transfer by step and flash imprint lithography. J. Vac. Sci. Technol. B 17, 2965–2969. 30. Khang, D. Y., Yoon, H., and Lee, H. H. (2001) Room temperature imprint lithography. Adv. Mater. 13, 749–752. 31. Ansari, K., Anton van Kan, J., Bettiol, A. A., and Watt, F. (2006) Stamps for nanoimprint lithography fabricated by proton beam writing and nickel electroplating. J. Micromech. Microeng. 16, 1967–1974. 32. Hirai, Y., Harada, S., Kikuta, H., Tanaka, Y., Okano, M., Isaka, S., and Kobayasi, M. (2002) Imprint lithography for curved cross-sectional structure using replicated Ni mold. J. Vac. Sci. Technol. B 20, 2867–2871. 33. Choi, S.-J., Yoo, P. J., Baek, S. J., Kim, T. W., and Lee, H. H. (2004) An ultraviolet-curable mold for sub-100-nm lithography. J. Am. Chem. Soc. 126, 7744–7745. 34. Yoo, P. J., Choi, S.-J., Kim, J. H., Suh, D., Baek, S. J., Kim, T. W., and Lee, H. H. (2004) Unconventional patterning with a modulus-tunable mold: from imprinting to microcontact printing. Chem. Mater. 16, 5000–5005. 35. Kim, E., Xia, Y., and Whitesides, G. M. (1996) Micromolding in capillaries: applications in materials science. J. Am. Chem. Soc. 118, 5722–5731. 36. Khang, D. Y. and Lee, H. H. (2004) Sub-100 nm patterning with an amorphous fluoropolymer mold. Langmuir 20, 2445–2448. 37. Khang, D. Y., Kang, H., Kim, T., and Lee, H. H. (2004) Low pressure nanoimprint lithography. Nano Lett. 4, 633–637. 38. Kim, M. J., Park, J.-E., Song, S., and Lee, H. H. (2007) Simple “solutal” method for preparing teflon nanostructures and molds. J. Vac. Sci. Technol. B 25, 1412–1415. 39. Jung, G.-Y., Li, Z., Wu, W., Chen, Y., Olynick, D. L., Wang, S.-Y., Tong, W. M., and Williams, R. S. (2005) Vapor-phase self-assembled monolayer for improved mold release in nanoimprint lithography. Langmuir 21, 1158–1161. 40. Schift, H., Saxer, S., Park, S., Padeste, C., Pieles, U., and Gobrecht, J. (2005) Controlled co-evaporation of silanes for nanoimprint stamps. Nanotechnology 16, S171–S175. 41. Hong, P. S. (2006) Improved Imprint Lithography for Nanofabrication. Doctoral Dissertation, Seoul National University, Seoul, Korea, pp. 49–51. 42. Suh, D. (2005) Metal/Polymer Bilayer Transfer by Rigiflex Lithography. Doctoral Dissertation, Seoul National University, Seoul, Korea, p. 55. 43. Chaudhury, M. K. and Whitesides, G. M. (1991) Direct measurement of interfacial interactions between semispherical lenses and flat sheets of poly(dimethylsiloxane) and their chemical derivatives. Langmuir 7, 1013–1025. 44. McDonald, J. C., Duffy, D. C., Anderson, J. R., Chiu, D. T., Wu, H., Schueller, O. J. A., and Whitesides, G. M. (2000) Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis 21, 27–40.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3 PATTERNING BASED ON NATURAL FORCE

3.1 INTRODUCTION Patterning based on natural force is attractive because it occurs spontaneously, in a natural manner, without any intervention. Of course, a suitable environment must be established for the action of such forces. When capillary force is utilized, a capillary structure must be formed and located in intimate contact with fluid for the wetting to take place. In the case of London force, which involves Van der Waals’ interactions, the film thickness should be sufficiently small for the force to be effective. When heat-generated mechanical stress is used, the force must be guided for it to be useful. This chapter describes the use of these three forces for patterning. When a mold with relief features is placed on a fluid or solid surface, the space between the mold and the surface creates a network of capillaries. If the ends of all the capillaries are open to air, the capillary is referred to as open-ended. In most applications, the ends of some or all of the capillaries are closed. In such cases, the air in the capillaries gets compressed as the capillary filling takes place thereby building up pressure that can impede flow. If the mold is permeable such that the air permeates out of the mold, then the capillary can be filled. This type of capillary is referred to as a closed permeable capillary. The mold, on the other hand, can be impermeable, in which case we have a completely closed capillary. The material to be patterned on a substrate is either a liquid or solid. While the liquid can be used as such, the solid, which is typically a polymeric material, has to be made mobile or fluidlike for the capillary rise to take place. A polymeric material becomes fluidlike when it is heated above its glass transition temperature (T g ), which is typically around 100◦ C. A polymer can also be made fluidlike through the action

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright 

27

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

28

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

of a solvent. After the liquid fills the capillaries, it is solidified either thermally or by exposure to ultraviolet (UV) light before the mold is removed to complete the patterning. In the case of a fluidlike polymer, either the temperature is lowered or the solvent is evaporated before the mold is removed. In patterning, a thin film on a substrate is often involved. Van der Waals’ force, sometimes called London force, becomes large when the film thickness is small, say less than 100 nm, since the force is inversely proportional to the third power of the film thickness. When the film thickness is small enough, the film can break into islands by dewetting, exposing the substrate surface between the islands. Under the confinement imposed by a mold, an ordered island structure results. For relatively thick films, a selective dewetting can be accomplished for patterning by making the film thickness thin enough only at the desired locations with the aid of a mold with protruding features. The dewetting phenomenon can also be used to fill nanowells and nanochannels selectively with a liquid without leaving any residue. It is in general difficult to use heat-generated mechanical stress for generalpurpose patterning. It is, however, possible to bring order to the pattern that results when a film is subjected to a mechanical stress with the aid of a mold. In fact, repeated pattern of any desired shape can be generated by guiding the mechanical stress with a mold. Of the three forces, the most useful for patterning is capillary force, followed by London force and mechanical stress. In what follows, these three forces are treated in succession.

3.2 CAPILLARY FORCE The first use of capillary force for patterning was described by Kim et al. [1] where open-ended capillaries were used for micromolding in capillaries (MIMIC). As a natural extension, closed but permeable capillaries were utilized for patterning in many different forms for which the permeability of the mold plays a critical role. This permeability places a constraint on the choice of mold material. The mold material most commonly used for patterning based on capillary force is poly(dimethylsiloxane) (PDMS). The mechanical strength of PDMS is relatively low such that problems of lateral and/or roof collapse of mold features arise as the feature sizes decrease, as described in Chapter 2. To extend the patterning capability to nanometer scale, a Teflon mold was introduced, which has a permeability higher than that of PDMS and a Young’s modulus higher by more than two orders of magnitude. In this case, unlike for PDMS, slight pressure is needed to make sure that the mold surface makes conformal contact with the underlying surface so that wetting of the capillary can take place. Capillary force is inversely proportional to the capillary radius or feature size. This capillary force can exceed 10 times the atmospheric pressure if the feature size is in the sub-100-nm range. Therefore, an impermeable mold or completely closed capillary can be used for patterning even if the pressure in the capillary builds up due to the capillary rise. The three cases of open-ended capillary, closed permeable capillary, and completely closed capillary will be treated separately in what follows.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

29

The speed at which a capillary is filled can be quite high. When a polymer is rendered fluidlike by heating, however, the time it takes for the patterning can be relatively long. A rapid heating method can be used to remedy the problem. The filling speed has to do with capillary kinetics. The kinetics depends on the type of capillary involved. These two subjects are treated in the following. 3.2.1 Open-Ended Capillary The most representative method involving an open-ended capillary is MIMIC, illustrated schematically in Figure 3.1. In this method, a PDMS mold is placed on the surface of a substrate. One of the desirable properties of PDMS is that it readily makes reversible conformal contact with most surfaces. In general, a mold surface makes conformal contact with the underlying substrate surface if it has a surface energy lower than that of the substrate and the mold is sufficiently flexible, as described in Chapter 2. The surface energy of PDMS is 21.6 dyn cm–2 , which is lower than that of most surfaces. The low modulus of s-PDMS makes it well suited for these procedures. Unless otherwise noted, therefore, the PDMS mold referred to in this chapter is s-PDMS. Because of the open-ended relief structure in the mold, the PDMS mold placed on a substrate creates an open-ended network of capillaries. When a liquid is placed at one end of the open capillaries, the liquid fills the capillaries by capillary

PDMS mold Support

Place a drop of prepolymer at one end

Fill channels by capillary action

Cure, remove mold

Figure 3.1. Schematic illustration of the procedure for MIMIC. (Reprinted with permission from [1]. Copyright 1995, Macmillan Publishers Ltd.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

30

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

force. The liquid is typically a prepolymer that can be cured or solidified either by exposure to UV light or by heating. After the capillaries are filled, the prepolymer is cured and the mold is removed, leaving the solidified polymer structure thus patterned on the substrate surface. A variety of materials [2] can be used for the purpose, including precursors to carbon or ceramics, polymer beads, colloids, inorganic salts, sol–gel materials. Polymer dissolved in solvent can also be used provided the solvent does not swell the PDMS mold [3]. Because of the nature of the process, MIMIC is applicable only to a network of capillaries that are interconnected and surfaces that are flat, and without contours. Although the capillary filling over a short distance (ca. 1 cm) is relatively fast and efficient, the filling over a large distance can be quite slow due to viscous drag of the fluid. A relatively simple formulation based on Poiseuille flow [4] leads to Washburn kinetics [5] for the rate of capillary filling πr 4 P dV = dt 8ηz

(3.1)

where V is the filled liquid volume, η is its viscosity, t is the time, r is the capillary radius, and z is the distance to which the liquid fills the capillary. The driving force for the filling is the Laplace pressure PL given by PL =

2γ cos θ r

(3.2)

where γ is the surface tension for the liquid–vapor interface and θ is the contact angle the liquid makes with the capillary wall. Use of equation 3.2 in equation 3.1 with P replaced by PL leads to r γ cos θ dz = dt 4ηz

(3.3)

since V = πr 2 z. The solution gives the Washburn kinetics:


z = κt

1/2

r γ cos θ κ≡ 2η

1/2 (3.4)

For a given time, the distance to which the capillary filling occurs becomes smaller with smaller capillary size, surface tension, and contact angle but with increasing viscosity since a single parameter κ determines the distance. While equilibrium contact angle is used here, the dynamic contact angle, which in turn depends on time, should be used for a more accurate description.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

31

3.2.2 Closed Permeable Capillary General-purpose patterning involves not just interconnected features as in openended capillaries but a combination of isolated and connected pattern features. As in photolithography, where a polymer or photoresist (PR) film is used for patterning, a polymer film on a substrate is typically patterned, and then used as the etch mask for the pattern transfer to the substrate. Fabrication of any type of device can be achieved by repetitive patterning followed by etching. This general-purpose patterning based on capillarity was first accomplished by solvent-assisted micromolding or SAMIM [6]. As shown in Figure 3.2a, a PDMS mold is wetted with a solvent that is a good solvent for a polymer to be molded, and is then brought into contact with the surface of the polymer. While in contact, the solvent dissolves a thin layer of the polymer and this polymer solution conforms to the surface topology of the mold by capillarity. The solvent permeates out of the mold, leading to the solidification of polymeric relief structures with a pattern complimentary to that on the mold surface. An example of a line pattern fabricated in PR by SAMIM is given in Figure 3.2b. As the atomic force microscope (AFM) images show, the lines are about 60 nm wide and 50 nm high with a spacing of about 1.2 µm. When the mold is wetted with a solvent, the liquid solvent tends to fill the recessed regions of the mold to maximize the area of the solid–liquid interface at the expense of the liquid–vapor interface. Therefore, mainly the area of the polymer corresponding to the recessed regions gets dissolved. In this example, the initial PR thickness was about 0.4 µm.

(a)

PDMS mold

(b) 14 µm

Wetting liquid

53 nm

Place the mold on a support Polymer film

0 nm

7 µm

Support Evaporate solvent

7 µm

0

14 µm ∼50 nm

4 µm

4 µm

2 µm Remove mold

0

2 µm ∼60 nm

Support Residual film Figure 3.2. (a) Schematic procedure for SAMIM. (Reprinted with permission from [2]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.) (b) AFM images of a polymeric line pattern fabricated by SAMIM. (Reprinted with permission from [6]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

32

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

The solvent used for SAMIM should not swell the PDMS mold, because swelling can destroy the conformal contact between the polymer and the mold. Many nonpolar solvents such as hexane, toluene, and methylene chloride cannot be used because of the swelling problem. The extent of swelling for a large number of solvents can be found in [7] (see Table 2.1). In general, the solvent should have a relatively high vapor pressure and a moderately high surface tension such as methanol, ethanol, and acetone to ensure rapid evaporation of the excess solvent and minimal swelling of the mold. Adding a solvent to a polymer brings about two effects [8]: plasticization and dilution. The plasticization leads to a decrease in the glass transition temperature, thereby lowering the viscosity of the polymer. This decrease in T g can be inferred from the following approximate relationship: Tg ≈

Tgp 1 + (X − 1)(1 − φp )

X≈

(3.5)

Tgp 2 Tgs ≈ Tms Tgs 3

where T gp is T g for the pure polymer, T ms is the melting temperature of the solvent, which is 200 K for trichloroethylene (TCE), for example, and φp is the volume fraction of the polymer. For polystyrene (PS) dissolved in TCE, for example, the glass transition temperature decreases from 378 to 294 K (a 21◦ C decrease) when the polymer volume fraction decreases from unity to 85%. A solid polymer starts flowing viscously when the temperature is higher than 1.2 T g . Therefore, an 85% PS solution behaves like a viscous fluid around room temperature. The dilution effect also leads to a decrease in viscosity. When a polymer is diluted, there is a corresponding increase in the critical molecular weight at which entanglements of polymer chains become important in hindering viscous flow. The following relationship applies [9]: η(φp = a)  a 5.1 ∝ η(φp = b) b

(3.6)

For the example just considered, the viscosity decreases by about 60% when the polymer volume decreases from unity to 85%. The pattern fabricated by SAMIM is often textured and distorted. Typically, the base of a rectangular stripe becomes wider than the top and the structure formed contracts as the solvent evaporates from the molded polymer. The solvent in the PDMS mold is localized to the recessed regions, thereby creating a nonuniform solvent distribution in the polymer layer when the PDMS mold makes contact with the polymer surface. Furthermore, often there is insufficient amount of mobile polymer solution to fill the capillaries completely. These solvent related problems can be overcome by utilizing a solvent-laden polymer film instead of a wetted mold, leading to a technique known as soft molding [10].

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

33

Elastomeric mold Resin solution

Substrate (a)

Slight pressing Solvent

Resin solution

(b)

Molding Solvent absorption into the mold and solidification

Removing the mold (c)

Figure 3.3. Schematic illustration of procedure for SOMO. (Reprinted with permission from [10]. Copyright 2001, American Institute of Physics.)

The procedure involved in soft molding (SOMO) is schematically illustrated in Figure 3.3. An elastomeric mold with the desired pattern on its surface is placed onto a polymer film, without drying, immediately after the film is spin coated onto a substrate, and then pressed slightly at a pressure less than 1 N cm−2 to ensure conformal contact of the mold with the solvent-laden polymer surface. After releasing the pressure, the mold and the substrate are allowed to remain undisturbed for a period of time, say 10 min. During this process, the solvent in the polymer film diffuses toward the interface between the molded polymer and the mold due to the concentration gradient and is absorbed into the mold. The solvent then permeates through the mold and finally evaporates into the air. After the evaporation, the mold is removed, thus completing the patterning. Figure 3.4 shows images by scanning electron microscopy (SEM) of a threedimensional structure that was fabricated by SOMO. A commercial novolac resin dissolved in propylene glycol mono ether acetate (PGMEA), which is a typical solvent for PRs, was spin coated onto a wafer for SOMO. The master from which the PDMS mold was made is shown in Figure 3.4a and the patterned polymer film in Figure 3.4b. As apparent from the figures, the polymer structures fabricated by SOMO do not have defects or distortion. The reason for these favorable outcomes is that the solvent in the molded polymer is continuously removed by absorption into the mold

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

34

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

(a)

Tilted image

(b)

Tilted image

Figure 3.4. SEM images of (a) master from which a PDMS mold is made and (b) polymer film patterned by SOMO. (Reprinted with permission from [10]. Copyright 2001, American Institute of Physics.)

at the mold–polymer interface and at the same time continuously replenished by solvent diffusion to the interface. The gradual and continuous replenishment of solvent and accompanying polymer to the interface assure the fidelity of the molded polymer structure. The mold could be used 20 times without any treatment. Since the mold is initially free of solvent and the absorbed solvent is continuously removed from the mold by evaporation into the air, the solvent retained in the mold during SOMO is negligibly small, alleviating to a large extent the swelling problem. There are two rate processes governing the fabrication by SOMO: solvent removal or evaporation from the polymer solution surface and subsequent absorption into the mold. If it is assumed that the rate of solvent evaporation is proportional to the solvent concentration, it follows that dM = −k M dt

or

M = M0 e−kt

(3.7)

where k is the proportionality constant, M is the solvent concentration in terms of weight per unit area, and M 0 is its initial concentration. Diffusion effects can be neglected because the film of interest is quite thin. The solvent concentration can be determined experimentally by dividing the weight of solvent evaporated by its surface area. Shown in Figure 3.5a are experimental data fitted with the first-order

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

35

(a)

Solvent weight (mg cm−2)

1

K = 0.00278 s−1

0.1

0

100

200

300

400

Time (s) (b)

Absorption weight (mg cm−2)

10

8 α = 0.204 mg s−1

6

4

2

0 0

5

10

15

20

25

30

35

40

45

Time (s)

Figure 3.5. Experimental data on (a) solvent weight change with time and (b) solvent absorption weight with time. (Reprinted with permission from [10]. Copyright 2001, American Institute of Physics.)

kinetics with the novolac resin in PGMEA. The rate of absorption can be determined by having the PDMS mold float in a bath filled with the solvent. Since the solvent concentration is constant, the rate of absorption per exposed surface area is independent of the concentration such that dMa = −α (3.8) dt

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

36

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

where α is a constant and M a is the concentration of the solvent per unit area in the mold. The experimental data plotted in Figure 3.5b show that the concentration M a increases linearly with time. For patterning by SOMO to be relatively fast, the rate of absorption should be larger than the rate of evaporation such that the solvent reaching the interface between the film and the mold is immediately absorbed into the mold. Since the evaporation is exponential whereas the absorption is linear with time, the initial rates should be used for the condition that the absorption rate is greater than the evaporation rate. It follows from equations 3.7 to 3.8 that α > k M0 . In SOMO, the solvent is absorbed into the mold at the interface between the polymer film and the mold. Therefore, the value of α in SOMO would be smaller than that of α determined in a bath of solvent. Therefore, the condition should be modified to α 1 k M0

(3.9)

Some solvent–polymer pairs satisfying this condition are PS in toluene and polymethylmethacrylate in toluene. Patterning by SOMO leaves behind a residual layer such that the substrate surface is not exposed. Therefore, a two-step etching process is needed to transfer the desired pattern into the substrate: first to remove the residual layer and second to etch into the substrate with the remaining polymer pattern as the etch mask. In a modified version of SOMO [11], the residual layer is made thin enough (ca. 10 nm) for one-step etching for the pattern transfer into the substrate. Here, a more diluted polymer solution is used than in the original SOMO to ensure continuous movement of polymer from the contact regions into the voids of the mold. A longer molding time is also often needed. Dip coating or drop casting of dilute polymer solution is typically used for the patterning purpose. Shown in Figure 3.6 are AFM images of 60-nm-wide polymer lines patterned by modified SOMO (Figure 3.6a) and those of the lines transferred into the underlying SiO2 by one-step reactive ion etching. The SiO2 lines (Figure 3.6b) are about 45 nm high. The polymer lines were patterned with 1 wt% 2-poly(vinylpyridine) in methanol. In this modified process, the solvent should be present in the polymer in an amount sufficient enough for continuous movement of polymer into the void space by capillarity. Therefore, unlike the original SOMO, the solvent and polymer pair should be chosen in such a way that the solvent can be removed slowly by absorption into the mold. Alternatively, the rate of absorption should be less than the rate of removal of solvent from the film or the rate of solvent evaporation from the polymer surface. Therefore, the following expression represents a requirement for this process: α <1 k M0

(3.10)

A solid polymer can be made to behave like a polymer solution by heating. A solid polymer starts flowing viscously when the temperature is higher than 1.2 T g .

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

37

3.2 CAPILLARY FORCE

(a)

−45

Polymer pattern

nm

45

0

1 µm

2

Horiz distance (L) 58.6 nm 0

1 µm

2

(b) ∼ 60 nm 45

−45 0

1 µm

SiO2 pattern

nm

0

2

1 µm Horiz distance (L) 58.6 nm

2

3D image

∼ 45 nm 2 µm 0.5

1.0

1.5

2.0

µm

Figure 3.6. AFM images of (a) polymer lines formed by modified SOMO and (b) line pattern transferred into the SiO2 substrate. The three-dimensional image shows 60-nm SiO2 lines fabricated by one-step reactive ion etching. (Reprinted with permission from [11]. Copyright 2002, American Institute of Physics.)

This thermally induced fluidity is utilized in capillary force lithography (CFL) [12], which is schematically illustrated in Figure 3.7. In this method, a polymer solution is spin coated onto a substrate and then the polymer film is dried. A PDMS mold made from a master with the desired pattern is placed on the polymer film, taking care to avoid small gaps or bubbles between the mold and the polymer surface. The film is then annealed at a temperature above T g . The void between the mold and the film is filled up by capillary action. After a period of time, the temperature is lowered to room temperature and the mold is removed, thereby generating the negative

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

38

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

(a)

(b)

Mold

Polymer Substrate Place the mold on the polymer surface

Heating (T > Ts, ∼130°C)

Cooling and mold removal

Figure 3.7. Schematic procedure for CFL. (a) When the film is relatively thick with respect to the mold step height, a residual layer remains after the patterning. (b) When it is sufficiently thin, no residual layer remains and substrate surface is exposed. (Reprinted with permission from [12]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

replica of the mold pattern, which is the master pattern. As shown in Figure 3.7, a residual layer remains after the patterning if the film is thick relative to the mold step height. However, no residual layer remains and the substrate surface gets exposed if the film is relatively thin. In this regime, initial wetting conditions are critical. If the mold leans a little toward one side, an asymmetric structure results. Moreover, the annealing time should not be too long to avoid dewetting. When the thermal expansion coefficient of substrate is considerably smaller than that of PDMS, which is typically the case for most inorganic materials, the mold tends to separate from the polymer surface when the annealing temperature is high.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

39

To prevent separation, the mold can be made soft (curing agent ∼6%) and annealing can be carried out with low heating rates. The maximum height hmax to which the polymer melt can rise through a channel of width L by capillarity can be obtained by equating the capillary force to the resistant gravitational force, which gives [4] h max =

2γ cos θ ρgL

(3.11)

where ρ is the polymer density and g is the gravity. While the pattern depth of interest in most applications is of micrometer range, the maximum height, as estimated with (3.11), can be quite large, reaching as high as a few meters for a typical polymer [12]. While Figures 3.2 and 3.6 show patterning results of 60-nm-wide lines, the spaces between the lines is quite large, say about 1 µm. For sparse features, results shown in Chapter 2 indicate the resolution at the single nanometer level is possible. For dense structures such as equal line and space patterns, the resolution that can be attained with PDMS mold is rather limited, particularly when the aspect ratio, defined as the height divided by the line width, is larger than unity. With soft PDMS, the resolution limit is about 500 nm; with hard PDMS it is about 100 nm. Even with an improved method of the mold fabrication [13], the resolution limit is 50 nm. The modest resolution stems from the low modulus of PDMS (modulus of about 1.8 MPa for soft PDMS and 8.2 MPa for hard PDMS) that leads to buckling, lateral collapse or pairing, roof sagging, and rounding of fine pattern features of the mold [14]. To extend patterning based on capillary force to the sub-100-nm range, a mold material is needed that is stiffer than PDMS and at the same time permeable to gases. The Teflon mold material discussed in Chapter 2 has a number of attractive material properties [15]. The tensile modulus is about 1.6 GPa, which is almost three orders of magnitude higher than that of PDMS, making it stiff and strong enough to preserve the mechanical integrity of fine features of the mold. It has a higher gas permeability than PDMS (for O2 , ∼10−7 cm3 cm (cm2 s Pa)–1 vs. ∼10−11 for PDMS). It is also nonreactive and inert. It has a low surface energy of 15.6 dyn cm–1 (PDMS ∼ 19.6 dyn cm–1 ) such that the Teflon mold can be cleanly removed from the patterned polymer after molding. For the solvent-based methods such as SAMIM and SOMO, Teflon is an ideal mold material since it is neither soluble nor swellable by a solvent except for fluorinated ones. It is also transparent to light ranging from UV to infrared. Because of the stiff nature of a Teflon mold, however, slight pressure (2–3 bar) must be applied for conformal contact of the mold with the polymer layer for CFL [16]. For SAMIM and SOMO, the pressure required is low because the solvent helps make good contact between the mold and the polymer film surface. Because of the optical transparency of Teflon, it can also be used for patterning UV-curable prepolymers. Figure 3.8 shows an SEM image of a line and space pattern formed with a Teflon mold and a UV-curable prepolymer. The dense 80-nm lines thus

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

40

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

(a)

(b)

Figure 3.8. (a) SEM image of dense 80-nm lines patterned with a Teflon mold and a UV-curable prepolymer. (b) Magnified version of (a). The lines are 500 nm high, corresponding to an aspect ratio larger than 6. (Reprinted with permission from [15]. Copyright 2004, American Chemical Society.)

patterned have step heights of about 500 nm, giving an aspect ratio larger than 6. With Teflon molds, patterning based on capillary force can be routinely carried out for nanometer scale patterning of structures with high aspect ratios.

3.2.3 Completely Closed Capillary A completely closed capillary corresponds to the case when an impermeable mold is placed on a polymer surface. With such an impermeable mold, the solvent-based methods such as SAMIM and SOMO cannot be used since the polymer cannot be solidified without the removal of solvent. Therefore, only CFL is applicable, as described in this section [17]. When an impermeable mold is used, air pressure simply builds up as the closed capillary is filled with polymer by capillary action. The trapped air is compressed, which impedes the capillary rise. If the capillary rise takes place at atmospheric pressure to a height z in a capillary of length Z (Figure 3.9a), the pressure of the trapped

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

8

(b) Air pressure Z z

Pressure (bar)

(a)

41

Capillary pressure

6

1 bar

4

2

Laplace pressure

0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

Figure 3.9. (a) Capillary rise in a completely closed capillary. (b) Laplace pressure as a function of feature size. (Reprinted with permission from [17]. Copyright 2006, American Institute of Physics.)

air in the capillary, Pa , is given by P0 Z /(Z − z) or P0 /(1 − z m ) where zm is z/Z and P0 is the initial air pressure (1 bar). The capillary rise will cease when Pa equals the Laplace pressure PL given by equation 3.2. Therefore, the normalized height zm to which the material of interest rises at atmospheric pressure (1 bar) is given by zm = 1 −

1 PL

(3.12)

where PL is in bar unit. As the relationship shows, no capillary rise takes place if the Laplace pressure is less than 1 bar. On the other hand, equation 3.2 tells us that the Laplace pressure increases with decreasing capillary radius or equivalently decreasing feature size of a pattern. Therefore, the pressure can exceed 1 bar, sufficient for capillary rise, at sufficiently small feature size. The geometric factor r/2 in equation 3.2 is the ratio of the exposed surface area to the circumference that is in contact with the wall. For an infinitely long channel, the geometric factor becomes L/2, where L is the channel width. Therefore, equation 3.2 applies to the channel if r is replaced by L. The Laplace pressure calculated for a PS with poly(urethaneacrylate) (PUA) mold is shown in Figure 3.9b, for which γ = 40 mN m–1 and θ = 56.6◦ . This result shows that the Laplace pressure becomes larger than the atmospheric pressure (1-bar line in the figure) corresponding to the onset of capillary rise for feature sizes smaller than 440 nm. While this result for the critical size Lc was obtained for a particular set of materials, it can be determined in general from the relationship L c = 2γ cos θ , which follows from equation 3.2 by setting PL equal to 1 bar. For the example just considered, the Laplace pressure is 5.5 bar for a 80-nm line and space pattern or L = 80 nm. According to equation 3.12, zm is about 0.82, meaning that the channel will be filled to 82% of the channel height. If the channel depth is 170 nm, the capillary rise will be to 140 nm.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

42

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

In general, CFL is ineffective with a hard mold on a hard substrate. The reason lies in the difficulty in making intimate contact between the mold and the substrate over a large area. Without the intimate contact, the material of interest simply cannot wet the capillary wall and hence no capillary rise can take place. When a hard mold is pressed against a hard substrate as in nanoimprint lithography [18], intimate contact cannot be made uniformly over large areas unless high pressure is applied because of slight roughness or uneven layouts that can be present on the mold as well as on the substrate. The pressure needed in this case is so high that it far exceeds the Laplace pressure and squeezed flow ensues. Therefore, the capillary force is inconsequential. CFL is most commonly and conveniently carried out with a flexible mold since the pressure needed for the intimate contact is of the order of 0.1 bar. The flexibility enables conformal contact with the underlying layer at a low pressure even if the surface is uneven and rough. At the same time, the surface of the mold should be sufficiently rigid for nanopatterning. The PUA molds described in Chapter 2 satisfy these criteria [19]. This almost impermeable mold is rigid enough to be used for nanoimprinting [20] and yet flexible in a film form. For the same reason, a hard mold such as one made with a silicon wafer can also be used if the substrate is flexible. Given in Figure 3.10 are SEM micrographs of PS patterns fabricated by CFL with impermeable and almost impermeable molds. The PS patterning result in Figure 3.10a obtained with a PUA mold on a silicon substrate corresponds to 80-nm-wide lines separated by 60 nm. The PS patterning result in Figure 3.10b obtained with a silicon wafer mold on a polyethylene tetraphthalate (PET) film corresponds to 80-nm-wide lines separated by 30 nm between lines. In both cases, the applied pressure is on the order of 100 g cm–2 or 0.1 bar [17]. Nanopatterning by CFL is effective over large areas with impermeable molds if either the mold or the substrate is flexible enough for intimate contact between the mold and the substrate. For the intimate contact to occur, either the mold should deflect enough to touch the underlying layer or the film on a substrate should touch

(a)

(b)

500 nm

500 nm

Figure 3.10. SEM images of a film of PS patterned by CFL with impermeable molds. (a) 80-nm line/60-nm space patterned with a flexible PUA film mold. (b) 80-nm line/30-nm space patterned with rigid silicon mold wafer with PS coated on flexible PET film. (Reprinted with permission from [17]. Copyright 2006, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

43

the mold above when a load is applied. For a simply supported thin plate at both edges, but free at the other edges of the plate, the maximum deflection ωm that occurs when a uniformly distributed load P is applied, is given [21] by ωm =

12(1 − ν 2 )P f (α, β) Et 3

(3.13)

where f depends on the dimensions of the plate, α and β, E is Young’s modulus, t is the plate thickness, and ν is the Poisson ratio. For a given mold or substrate, the deflection needed for the intimate contact is proportional to P/(Et 3 ). Therefore, the pressure needed for the intimate contact decreases with decreasing value of Et3 . The smaller the Young’s modulus of the mold material and the thinner the mold, the lower the required pressure becomes for intimate contact. A measure of the flexibility required can be established from experimental results with the silicon mold as illustrated in Figure 3.10b. For the 0.1 bar of pressure needed for patterning, the thickness of PET film is 100 µm, a typical value of the Young’s modulus of PET is 2500 MPa. Therefore, an Et3 value of 2.5 × 10−3 MPa cm3 is a measure of flexibility for intimate contact at a pressure of 0.1 bar. For any pair of impermeable mold and substrate, therefore, the flexibility needed for CFL is defined. For instance, if the Et3 value of the flexible substrate or mold is less than 2.5 × 10−3 MPa cm3 , the pressure required would be 0.1 bar. If 1 bar of pressure is allowed, the value becomes 2.5 × 10−2 MPa cm3 . The capability of nanopatterning by CFL with impermeable molds extends to the sub-10-nm range since silicon wafer molds or any other type of hard mold can be used with a flexible substrate. Furthermore, the driving force for the process increases with decreasing size. In the nanometer regime, the Laplace pressure is extremely high. For the example considered earlier, for instance, the Laplace pressure approaches 88 bar for 5-nm-wide channel. Furthermore, the compressed air pressure can be eliminated by carrying out the experiment in vacuum.

3.2.4 Fast Patterning The time for patterning by solutal or thermal capillarity is relatively long. When thermal capillarity is utilized as in CFL, patterning is completed in 30 min under typical conditions. The temperature used in CFL is typically less than 150◦ C, where the viscosity of many polymers of interest is relatively high, leading to relatively slow filling of the capillaries. Furthermore, it takes time for the system to reach the desired final temperature by heating. The time required for the patterning by solutal capillarity is largely determined by the time it takes for the vaporized solvent to permeate out of the mold. As a result, the concentration of the polymer in the polymer solution affects the patterning time. While heating should help, often the rate-limiting step is the diffusion of the solvent vapor into and then out of the mold, as shown shortly. Despite these effects, the patterning time for CFL can be shortened considerably.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

44

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

While thermal heating is typically used for CFL, infrared (IR) light can also be used. Typical molds made of PDMS, PUA, and Teflon are all transparent to light such that only the polymer is primarily heated. In the rapid flash patterning method (RFP) [22], a halogen lamp with emission in the IR range is flashed for a short period of time, typically 60 s with a 500-W lamp, but 10 s with a 10-kW lamp array. The temperature reached is typically greater than 250◦ C. In this way, the time needed for patterning or capillary rise can be less than 10 s, which compares with 30 min typically needed with thermal heating. The cooling time is also shortened, requiring only a few minutes without any active cooling. The step and repeat procedure for patterning that is often used for large area can be accomplished with the use of a light shield. Shown in Figure 3.11 is a patterning result obtained by RFP with an impermeable PUA mold in the form of a thin film. In this case, the mold was placed on a film of

(a)

(b)

Figure 3.11. PVA pattern formed by RFP in which the substrate surface is exposed in the regions between the features. (a) SEM image of 80-nm line/270-nm space pattern. (b) Cross-sectional SEM image of the line and space pattern in (a). The line height is 440 nm. (Reprinted with permission from [22]. Copyright 2004, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

45

poly(vinylalcohol) (PVA) spin coated on a silicon wafer and then dried. Onto this PUA film mold, a flat PDMS block was placed to distribute uniformly 3–4 bar of pressure applied to the system with a glass plate. A 500-W IR lamp was then flashed for 10 s. The resulting PVA pattern in Figure 3.11a is a combination of a lines and spaces (line = 80 nm, space = 270 nm) and a meandering line pattern. The magnified SEM image in Figure 3.11b shows that the pattern height is about 440 nm, giving an aspect ratio larger than 5. Another advantage of the RFP method is that it can yield thick patterned features that are completely isolated with exposed substrate surface in between. For example, the silicon wafer surface in Figure 3.11b is exposed except where the PVA lines are present. The process leading to the exposure of the substrate surface can be considered to occur in two steps. In the initial phase of RFP, the polymer melt is pushed into the capillaries by capillary action and the applied pressure, thinning the polymer layer between the protruding parts of the mold and the substrate. When this layer becomes sufficiently thin, the second step of dewetting takes over, leaving behind dry spots. While dewetting is known to take place for a thin, free liquid film on a solid substrate [23], spontaneous dewetting [24, 25] occurs when a liquid film is squeezed between a hard substrate and rubbery plate. When the protruding part of the rubbery plate approaches the substrate surface close enough that the radius of this thin layer exceeds a critical value, the liquid drains out and a dry spot appears [24, 25]. Once the dry spot forms, it expands at the velocity Vd (= k1 S 2 /(ηEh)) [24, 25] where k1 is a prefactor, η is the liquid viscosity, E is the elastic modulus of the rubber, h is the film thickness, and the spreading coefficient S is given by equation 3.23. Therefore, the essence of RFP lies in the use of a flexible, rubbery mold since the use of a hard mold will not lead to the surface exposure over a large area. The additional pressure (3–4 bar) in excess of that needed for conformal contact, which is typically on the order of 0.1 bar with a flexible mold, is utilized to make the residual layer thin enough for the spontaneous dewetting to take place.

3.2.5 Capillary Kinetics The capillary kinetics given by equation 3.3 or 3.4 is sufficient for open-ended horizontal capillary flow as in MIMIC provided that the capillary is not too long. Additional factors come into play when vertical capillary movement is involved in a closed capillary, as is the case for all the methods discussed in this chapter except for MIMIC. Consider first the kinetics for a closed permeable capillary. In such a capillary, the Laplace pressure cannot be available entirely for the capillary rise. There is a certain pressure drop across the permeable mold since either air or solvent vapor permeates through the mold into the air by diffusion. Therefore, the net driving force for the capillary rise is the Laplace pressure minus the pressure drop across the mold. For the one-dimensional flux being considered, the definition of permeability Pe yields

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

46

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

the following hydrostatic pressure difference, PH [26], PH =

U dz Pe dt

(3.14)

where U is the permeation length of the mold or the mold thickness. In CFL, there are two Poiseuille flows in series as illustrated in Figure 3.12. The first is the flow in the mass reservoir confined by the mold and the substrate (region I

Mold

z

h

Polymer Substrate

∆P = PL−PH

d PH

Air PL

h(z) = Lo − nz, n ≡ L/λ

h Mold

Substrate

L

Region II Region I

Figure 3.12. Schematic illustration of capillary rise into permeable cavities, showing the geometry and two channel resistances in series. (Reprinted with permission from [27]. Copyright 2004, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.2 CAPILLARY FORCE

47

in Figure 3.12c) and the second is that within the channels (region II). The Poiseuille flow in region I is that between two parallel plates as is that in region II for the channels formed by a line and space pattern. The Poiseuille flow between two plates leads to a constant of 12 instead of 8 in equation 3.1, which corresponds to flow in a cylinder. The resistance to flow in region I is proportional to 1/(h/2)2 , but there are two paths leading to the channel such that it is proportional to 2/h2 while that in the channel is proportional to 1/(L/2)2 . With these two resistances in series, the equation equivalent to equation 3.3 [27] is P dz = dt 3ηz(4/L 2 + 2/ h 2 )

P ≡ PL − PH

(3.15)

The film thickness h depends on z. The mass transport is almost entirely from the regions adjacent to the channels and not from outside the mold. Therefore, for a line and space pattern, n = L/λ and h(z) is given by h(z) = h 0 − nz

(3.16)

where h0 is the initial film thickness and λ is the length of the space between channels. For equal line and space pattern, for example, n = 1. The solution to equation 3.15, when solved with the aid of equations 3.14 and 3.16, is n 2 L 2U L 2 nz h0 n 2 Lγ cos θ z + n2 z2 + − L 2 ln = t 6η Pe h 0 − nz h 0 − nz 3

(3.17)

It can be shown [27] that the contribution of gas permeation to the overall kinetics is negligible for PDMS molds. Since Teflon has a permeability higher than PDMS, the same applies to Teflon mold. For these molds, therefore, equation 3.17 reduces to  nz 2 L

+

  n 2 γ cos θ h 0 − nz nz = + ln t h 0 − nz h0 3L

(3.18)

When the permeability is low, on the other hand, the first term dominates in the lefthand side of equation 3.17 and z is given by (2γ Pe cos θ t/(LU )), which applies to the capillary rise in a nearly impermeable mold. It can be seen that the distance z is proportional to the time t in this case. While the capillary kinetics governing CFL is given by equation 3.17, the kinetics underlying SOMO is different because the solvent evaporates in the course of the patterning and therefore the viscosity changes with time. Equation 3.15 applies here too, but the time dependence of η must be specified. If we neglect the difference in density between solvent and polymer, the ratio of the polymer volume fraction in equation 3.6 can be written as follows with the aid of equation 3.7: φ0 = φ0 + (1 − φ0 )e−kt φ

(3.19)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

48

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

where φ0 is the initial volume fraction and φ the volume fraction at time t. Equation 3.6 then becomes η = η0 [φ0 + (1 − φ0 )e−kt ]−5.1

η0 ≡ η(φ0 )

(3.20)

where η0 is the viscosity of the initial polymer solution or the prepared polymer solution for SOMO; when spin coating is used in place of dip coating, however, it is the viscosity of the coated polymer film. This expression for η can be used in equation 3.15 for the capillary kinetics. The resulting kinetic equation does not render itself to an analytical solution for SOMO and therefore the solution can only be obtained numerically. For a completely closed capillary, the kinetics are also governed by equation 3.15. However, the driving force P must be modified for CFL with an impermeable mold. The mold is impermeable and therefore PH = 0. However, the air pressure in the capillary builds up as the capillary rise takes place. This compressed air pressure was earlier given by 1/(1 − z/Z ), where Z is the depth of capillary or the pattern depth and the initial air pressure of 1 bar was used. The actual driving force for the capillary rise is the Laplace pressure minus the compressed air pressure in excess of 1 bar such that P in equation 3.15 becomes  1 −1 P = PL − 1 − z/Z


(3.21)

where PL is in bar unit. The capillary kinetics then are determined by equation 3.15 with P given by equation 3.21. No analytical solution is attainable in this case and therefore one has to resort to numerical methods. When the film thickness is sufficiently large, the capillary kinetics are governed by L 2 P dz = dt 12ηz

(3.22)

which results from equation 3.15 for (h/L)2  1/2. In this case, the analytical expression for capillary kinetics can be obtained.

3.3 LONDON FORCE AND LIQUID FILAMENT STABILITY Thin polymer films (<100 nm) on solid substrates undergo dewetting when annealed above the glass transition temperature [28]. Initially, cylindrical holes are created in the film and as these holes grow in size, the mass removed from the substrate for the growth of dry holes creates rims around the holes. When these outwardly expanding holes contact one another, the rims break up into droplets, creating cellular structures. The driving force or the disjoining pressure for this spinodal dewetting [23] is of the order of A/h3 , where A is the Hamaker constant [29] and h is the film thickness. As apparent from the relationship, this disjoining pressure increases very rapidly with

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.3 LONDON FORCE AND LIQUID FILAMENT STABILITY

49

decreasing film thickness and is very substantial for thin films. This opening up of holes in thin polymer films can be utilized for patterning. General dewetting phenomenon can also be used to fill holes and/or channels present on a mold or substrate without leaving any residue on the rest of the surface. This selective filling of micro/nanoholes or channels is not only important in biomedical area but can also be used for patterning by transferring the filled content to a substrate or utilizing a sacrificial layer. These two topics are treated in this section.

3.3.1 Patterning by Selective Dewetting The basic concept of patterning based on dewetting is to make the polymer film thin enough only at the intended places so that upon heating above the glass transition temperature, these places open up, exposing the substrate surface. This process [30] is schematically illustrated in Figure 3.13. An elastomeric mold is prepared that has either pointed top edges for lines or points for dots. These wedge-shaped features serve two purposes. When the mold is placed on a dry polymer film coated on a substrate and the temperature is raised above the glass transition temperature, the wedge helps the mold sink into the polymer melt by the mold weight itself. The width of the pointed edges for lines or the diameter of the points for dots, as discussed shortly, determines the smallest feature size that can be obtained for line or dot pattern, respectively. The mold is then placed on a polymer layer coated on a substrate and pressed slightly for conformal contact (step (a)). The mold and substrate are heated to a temperature higher than T g of the polymer and allowed to remain undisturbed for a period of time (step (b)). During this heating, the mold sinks into the polymer melt. When the sharp edges of the mold approach the substrate surface close enough that London force comes into play, dewetting begins, which in turn breaks up the film, opening holes or lines depending on whether the pattern on the mold consists of sharp points or edges. After a specified time, the polymer is cooled with the mold still in place and then the mold is simply removed (step (c)). The thermal instability of a thin polymer film leading to film breakup occurs at the location where the thickness is the smallest [31]. This allows selective dewetting and hole formation. Once dry spots (holes) form, the hole radius grows with time, which in turn allows the hole size to be controlled by simply manipulating the heating time. The time progression of hole formation and subsequent growth is shown in Figure 3.14. As the AFM micrographs show, the hole does not yet extend to the substrate surface after 1 min. A flat portion starts appearing after 1.5 min, corresponding to hole formation. The hole size, initially at 10 nm, grows rather rapidly with time, reaching 40 nm in 2 min and then 360 nm in 7 min. In general, the radius r for free film increases in proportion to the time t raised to a power q, i.e., r ∝ t q , where q is in turn a function of time [31]. The result in Figure 3.14 shows that the radius increases nonlinearly with time. It is notable that the feature size can be controlled down to the 10-nm level. More importantly, hole sizes ranging from 10 nm to in excess of 360 nm can be formed with one and the same mold by controlling the heating time.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

50

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

Elastomeric mold Polymer layer

Substrate

(a)

Annealing (T > TS) (b)

Dewetting of thin polymer layer

Removing the mold (c)

Figure 3.13. Illustration of the procedure for generating a desired pattern by selective dewetting. (a) Placing a mold on the polymer film coated on a substrate. (b) Making conformal contact by slight pressing and then leaving it undisturbed at T > T g for a period of time. (c ) Removing the mold after cooling. (Reprinted with permission from [30]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

As discussed earlier, line patterns can be obtained using a mold with sharp edged lines. Shown in Figures 3.15a are copper dots (180 nm) and lines (200 nm wide) formed by electroless plating. The dot and line patterns were obtained by selective dewetting, electroless deposition of copper, followed by dissolution of the PS film in toluene. Also shown in Figure 3.15c is a fine-line (50 nm) pattern obtained by dissolving the oxide layer on a Si wafer in HF through lines formed by selective dewetting. In general, a low molecular weight polymer is more conductive to spinodal dewetting. A drawback in using such a polymer is that it is difficult to obtain feature sizes in the micrometer range since the time for the hole to reach that size is long. During this time, spinodal dewetting can take place all over the film, not just in the pinched locations. For micrometer-sized features, therefore, a higher molecular weight or a thicker layer of polymer should be used such that spinodal dewetting is suppressed

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

51

3.3 LONDON FORCE AND LIQUID FILAMENT STABILITY

1.5 min

1 min

0

2 µm

0 nm 75

−75

0 nm 75

0

∼10 nm

2 µm

0

2 µm

−75

7 min 2

2

0

nm

0

0 nm 75

2 µm

0

4 min

0

2

2 µm

−75

2 µm

0

2 min 2

2

2 µm

75

0

0 nm

2 µm

75

−75

Time

Flat size

1 min

–

1.5 min

10 nm

2 min

40 nm

4 min

220 nm

7 min

360 nm

−75 0

2 µm

0

2 µm

Figure 3.14. Temporal progression of hole opening in a 75-nm-thick PS (M W = 10,800) film on a Si wafer at 130◦ C. (Reprinted with permission from [30]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

but at the expense of a longer dewetting time. By the nature of hole formation, the pattern formed by the selective dewetting involves slanted sidewalls.

3.3.2 Liquid Filament Stability: Filling and Patterning Filling minute holes or lines (channels) with a liquid without leaving any material in other regions is becoming important for the development of new analytical methods in both chemistry and biology. This selective filling of microwells with a liquid can be accomplished under certain conditions on the contact angle and the viscosity, by

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

52

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

(a)

(b)

8.00 (c) 6.00

10.0 µm 5.0 µm 0.0 µm

4.00

2.00

0

2.00

4.00

6.00

0 8.00 µm

Figure 3.15. SEM images of Cu (a) dots and (b) lines. Holes and lines through PS film are first obtained by selective dewetting, followed by electroless Cu deposition. (c ) AFM image of a fine-line (50 nm) pattern obtained by opening lines in PS (M W = 42,000) film (100 nm thick) by selective dewetting, followed by etching the oxide layer on a supporting Si wafer through the open lines by dipping in HF. (Reprinted with permission from [30]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.3 LONDON FORCE AND LIQUID FILAMENT STABILITY

53

simply dipping a substrate with microwells into a liquid reservoir and then drawing it upward [32]. The wettability of a solid surface with a liquid is determined by spreading coefficient [4] S, which is given by S = σs + σl − σsl

(3.23)

where σs and σl are the surface energies of the solid and the liquid, respectively, and σsl is the interfacial surface energy. If the spreading coefficient is positive, the liquid forms a continuous film on the solid; if negative, the liquid dewets and beads up on the solid surface. In general, holes or channels on a substrate surface can be filled selectively if the spreading coefficient is negative whether the substrate is dip coated or spin coated, provided the coating speed, i.e., drawing speed in dip coating and spinning rate in spin coating, is high enough. In some cases, the coating speed needed for selective filling can lead to partial filling, which may represent a disadvantage for certain applications as described in the following. A patterning method that exploits selective filling involving first filling a recessed pattern on a mold with a UV-curable polymer precursor liquid, transferring the liquid to a substrate by contacting the mold with the substrate, curing the polymer, and then removing the mold. For effective transfer, the holes or channels must be completely filled. When the selective filling is used for patterning, on the other hand, there is a conflict between the requirement of complete filling and that of residue free, selective filling. Coating speeds that are sufficiently high for selective filling can cause the liquid to drain away from the holes or channels leading to incomplete filling; lower speeds can lead to residual drops on the background surface. The coating speed necessary for the complete filling becomes smaller for more negative spreading coefficient, or for larger contact angle as the Young–Dupree relationship [4] reveals. The ability to accomplish residue-free and complete filling by utilizing a high contact angle is limited by the instability of the liquid filament that can arise with high contact angle. The stability of a liquid filament in a channel is determined by the contact angle and the aspect ratio of the channel [33], as defined by the ratio of the channel depth to its width. When it is unstable, the liquid breaks into droplets. The window of stability can be considerably widened in terms of the contact angle θ and the aspect ratio X (= h/L), where h is the depth and L is the width of the channel, by making the cross-section of the channel trapezoidal. The stability condition is given [34] by  R cos θ = P P ≡ 1 − 2X/ tan φ

cos θ θ + 2 sin θ 2



Q ≡ X − X 2 / tan φ

−

2Q sin θ P

(3.24)

R ≡ 1 + 2X/ sin φ

where θ is in radian and φ is the corner angle of the trapezoid. If a point given by X and θ for a selected φ lies above the curve given by equation 3.24, the liquid filament

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

54

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

is unstable; below the line, it is stable. By widening the stability region, even a liquid with a high contact angle can be used for complete, selective filling for patterning by transfer. The shape of the meniscus of the completely filled liquid is determined by R cos θ = P

(3.25)

If the point defined by X and θ lies above the curve given by equation 3.25, a convex shape results; a concave shape if it lies below the line. As apparent from the discussion so far, a liquid that readily dewets on a solid surface must be used for selective filling or at least the spreading coefficient should be negative. A natural question is whether a readily wetting liquid can also be used for filling only holes or channels. For such a liquid, a filament-merging method [35] can be used. Consider a mold with a channel pattern where the ridges of the channels are contoured as shown in Figure 3.16. A continuous liquid film forms over the ridges and grooves when a readily wetting liquid is spin coated on the mold. When a flat rubbery plate is placed onto the liquid-covered ridges, an interfacial meniscus forms between the mold and the plate. The stability of this liquid filament determines the fate of the filament. If it is unstable, liquid droplets form and therefore residue-free filling cannot occur. If it is stable, then the filament simply merges into the liquid in the groove leaving no residue on the ridge since the filament meets the liquid at the edge of the groove. This draining of the filament into the liquid in the groove is similar to the situation where a liquid drop is absorbed into a liquid reservoir when it

(b)

(a)

Stable liquid filament merged

Place plate on the surface Flat plate

into the liquid in the grooves Flat plate

Thin liquid film

Contact Corrugated pattern and hemispherical ridges

(c)

(d ) Remove the plate

Etching Sol−gel

Flat plate

Polymer Figure 3.16. Schematic illustration of the procedure involved in filament-merging method for filling only channels with a liquid that readily wets the surface. (Reprinted with permission from [35]. Copyright 2006, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

55

3.3 LONDON FORCE AND LIQUID FILAMENT STABILITY

comes into contact with the reservoir. The stability condition [35] is simply that the average contact angle θa (= (θf + θm )/2) is less than or equal to π /4 or 45◦ , where θ f is the contact angle the liquid makes with the flat plate and θ m the contact angle with the mold. The average contact angle being less than 45◦ implies that any liquid that readily wets the surface can be used or for that matter a liquid with a positive spreading coefficient can be used. In this filament-merging method, a liquid that satisfies the contact angle requirement is spin coated on a substrate surface having a channel pattern with contoured ridges. This corrugated pattern can be obtained by a suitable method such as CFL (Figure 3.16a). A flat rubbery plate such as a PDMS plate is placed on the liquidcovered ridges with a slight pressure for intimate physical contact with the ridges (Figure 3.16b). After a sufficient time has elapsed for solvent evaporation, if the liquid is a polymer or sol–gel solution, the flat plate is simply removed (Figure 3.16c). As shown in Figure 3.16d, if the filled material is a sol–gel solution, the solidified sol–gel can be used as an etch mask for etching the underlying substrate. Shown in Figure 3.17 is a line (70 nm wide) and space (70 nm) pattern of PS filled with a SiOx sol–gel solution (spin-on-glass, Honeywell). For this system, the

(a)

0

1.00

(b) 2.00

2.00

1.00

1.00

0 2.00

0

1.00

0 2.00

(d )

(c)

200 nm

200 nm

1 µm

1 µm

Figure 3.17. Selective filling of a line (70 nm) and space (70 nm) pattern of PS with SiOx sol–gel solution for which the spreading coefficient is positive. (a) AFM image of the surface coated with the sol–gel. (b) AFM image of the surface where only the channels are filled with the sol–gel. (c ) Surface corresponding to the one in Figure 3.16d after dry etching PS. (d ) Aluminum pattern obtained after the metal etching with the patterned sol–gel as the etch mask. (Reprinted with permission from [35]. Copyright 2006, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

56

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

average contact angle is 28◦ , which satisfies the requirement on the contact angle, and the spreading coefficient is positive. Figure 3.17a shows AFM images of a line and space pattern coated with a sol–gel solution. When the spreading coefficient is positive, the solvent evaporates upon spin coating. Therefore, the flat mold must be soaked with solvent before contacting the ridges to endow fluidity to the solventevaporated sol–gel [35]. This soaking is not necessary when the spreading coefficient is negative. Upon contacting the surface with the solvent-soaked flat PDMS plate, the sol–gel solution on the ridges merges into the grooves, filling the grooves without leaving residue on the ridges, as the AFM image in Figure 3.17b shows. When the surface is dry etched to remove the polymer between the grooves that are filled with the by-now solidified sol–gel, the patterning result such as the one in Figure 3.17c is obtained. With this patterned sol–gel as the etch mask, the underlying substrate, which supports an aluminum film in this case, can be etched as shown in Figure 3.17d. It should be noted in this regard that the sol–gel is a much better etch mask or has a better etch selectivity than PS polymer with respect to aluminum.

3.4 MECHANICAL STRESS: PATTERNING OF A METAL SURFACE A mechanical stress is generated when two bonded layers are heated or cooled because of a difference in the thermal expansion coefficients of the two materials. This mechanical stress has been utilized [36] to bring about ordering with a metal deposited on prepatterned PDMS molds. To relieve the stress generated, the bilayer buckles, leading to the formation of wrinkles. A typical example is a bilayer of metal deposited on polymer. Typically, the wrinkles are random and isotropic although they do have a certain periodicity. These isotropic wrinkles can be made to self-organize themselves with the aid of a mold to order the structures. When a bilayer of metal on polymer is used, in effect the metal surface is patterned with a mold. This patterning of a metal surface is limited in that arbitrary patterns cannot be generated. On the other hand, it is a versatile method in that any surface shape that is represented by a Fourier series, albeit limited in the number of terms, can be generated with a single mold. Similar mechanics concepts can be used in stretchable electronics, as discussed theoretically in Chapter 18 and experimentally in Chapter 17. This process is most effectively accomplished with low modulus molds such as those made with s-PDMS. Typically, the molds are prepared by mixing a siloxane base oligomer and a curing agent (Sylgard 184, Dow Corning) in the ratio of 15 to 1 by weight, curing at 50◦ C for 12 h. The polymer for the metal should be of high molecular weight type. For the PS used in the experiments, the number averaged molecular weight is 4.04 × 106 with a polydispersity of 1.05 (Polymer Sources, Inc.). The PDMS mold and the bilayer are separately heated to a temperature slightly below T g to avoid the undue stress generated during heating. The mold is then placed on the sample and the temperature is raised to 130◦ C, which is well above the T g of PS (105◦ C). Typical annealing time is 2–12 h. The procedure is schematically illustrated in Figure 3.18. A substrate is first prepared by spin coating a polymer and then depositing a metal (Figure 3.18a). A PDMS mold with a desired pattern is placed on the metal surface and then heated to a

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.4 MECHANICAL STRESS: PATTERNING OF A METAL SURFACE

(a)

57

PDMS mold Metal layer Polymer layer Substrate

(b)

Mold contact and heating

(c) Buckling in confined geometry

(d)

Mold removal

Figure 3.18. Schematic illustration of the procedure for patterning a layer of metal on a polymer film. (a) A patterned PDMS mold is placed on a bilayer of metal and polymer coated on a substrate. (b) After ensuring conformal contact, the system is heated above the glass transition temperature of polymer. (c ) During heating, buckling takes place. (d ) After cooling, the mold is removed. (Reprinted with permission from [37]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

temperature above the glass transition temperature (Figure 3.18b), which causes the layer to buckle leading to the formation of wrinkles. Because of the presence of the mold, the wrinkles self-organize to ordered structures as dictated by the mold pattern. The mold is then removed after cooling, completing the patterning. In this process, the relative magnitude of the intrinsic buckling wavelength with respect to the period of the mold pattern has a direct bearing on the type of ordered

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

58

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

structure resulting from the patterning. The intrinsic wavelength of the wrinkles, which is the wavelength in the absence of the mold, is given [37] by  L= L ≡ λ/2tm

1+



Y

1/3 (3.26)

1 + 12YH3

Y ≡ (E m /E p )/[2(1 − νm2 )]

H ≡ tm /tp

where λ is the intrinsic wavelength, E is Young’s modulus, ν is the Poisson ratio, t is the film thickness, and the subscripts m and p are for the metal and the polymer, respectively. The Young’s modulus for the polymer Ep decreases drastically above the glass transition temperature and this effect can be described by Williams–Landel–Ferry equation [38], log E p /E p0 =

−C1 (T − T0 ) C2 + (T − T0 )

(3.27)

where C1 and C2 are fitting parameters, T is the temperature, T 0 is the glass transition temperature, and E p0 is Ep at T 0 . Experimentally determined values of these constants for PS, for example, are 8.4 and 2.5 K for C1 and C2 , respectively. Figure 3.19a shows wrinkles generated by buckling when a bilayer of aluminum on a layer of PS spin-coated on a silicon substrate was heated to 120◦ C [37]. When this bilayer is buckled by heating with a PDMS mold on the metal surface, ordered structures emerge as shown in Figures 3.19b–e. A sinusoidal-line structure forms when the PDMS mold has a line and space pattern. The structure is exactly the negative replica of the PDMS mold pattern except that the surface is sinusoidal. The period of the PDMS mold pattern is 2 µm and so is the period of the wave resulting from the patterning in Figure 3.19b. The intrinsic wavelength in this case is 2.6 µm. For the patterning result in Figure 3.19c, the period of the mold pattern and that of the patterned metal surface are both 4 µm. The intrinsic wavelength is 4.2 µm. As discussed later, the metal surface pattern complies with the period of the mold pattern as long as the ratio of the mold wavelength (λm ) to the intrinsic wavelength (λi ) is larger than 0.69, or for λm /λi > 0.69. When the PDMS mold presents a cylindershaped dot pattern, a two-directional sinusoidal surface results as shown in Figure 3.19d for which the mold and intrinsic wavelengths are 2.4 and 2.6 µm, respectively. On the other hand, the metal surface changes to a nonsinusoidal periodic structure, which is a hole pattern shown in Figure 3.19e, if the mold period is much larger than the intrinsic wavelength or if the wavelength ratio λm /λi becomes larger than 1.61. For the example in Figure 3.19e, the mold period is 6.3 µm and the intrinsic period is 3.4 µm. When the ratio λm /λi is between 0.69 and 1.61, the shape of the patterned metal surface corresponds to a sinusoidal negative replica of the mold pattern as in the usual patterning of a polymer film with a mold. The change in the shape of the patterned metal surface from the expected shape as in Figure 3.19d to another that is not expected as in Figure 3.19e is the result of the externally imposed periodic nodes provided by the mold pattern, or the wave of

(d)

Period : 2.4 µm

Period : 2.0 µm

82 nm

89 nm

(e)

(c)

106 nm

Period : 6.3 µm

160 nm

Period : 4.0 µm

13:57

Figure 3.19. Patterned metal surfaces. (a) Wrinkles generated by buckling without PDMS mold. (b) Patterning result from applying an equal line and space mold with a period of 2 µm (intrinsic wavelength = 2.6 µm). (c ) Same as in (b) except that the period is 4 µm (intrinsic wavelength = 4.2 µm). (d ) Patterning result from applying a cylindrical dot mold with a period of 2.4 µm (intrinsic wavelength = 2.6 µm). (e) Same as in (d ) except that the period is 6.3 µm (intrinsic wavelength = 3.4 µm). (Reprinted with permission from [37]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.)

(a)

(b)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008 Printer Name: Yet to Come

59

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

60

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

the mold dictating the number of harmonic modes allowed within the node interval [39]. In this wave-directed self-organization, the internal wave in the bilayer, which is the intrinsic wave, selects the type, number of harmonic modes, and the fractional magnitude that each allowed harmonic mode contributes to the overall surface shape. Therefore, shape engineering of the metal surface is entirely possible by selecting properly the period of the mold and the intrinsic wavelength. It is therefore possible to generate different shapes of the metal surface with a single mold by simply changing the intrinsic wavelength by manipulating the thicknesses of the metal and polymer layers. An example is given in Figure 3.20 where the same mold with a equal line and space pattern (period: 3 µm) is used to generate four different metal surface shapes. Without the mold, an isotropic pattern in Figure 3.20a results upon heating the bilayer. As the intrinsic wavelength of the bilayer is changed by varying the layer thicknesses from 4.68 µm, to 3.92 µm, to 2.96 µm, and to 1.68 µm, the patterned metal surface changes to those in

(b)

(c)

20 µm

20 10

10 Symmetric single

Symmetric double (a)

X : 5.0 µm/div Z : 200 nm/div

20 Isotropic wave

10 (d)

(e)

20

20 10 Asymmetric single (Echellettelike shape)

10 Asymmetric double (Cascadelike shape)

Figure 3.20. Different metal surface patterning results obtained from the same PDMS mold (3-µm line and space). (a) Isotropic waves in absence of the mold. (b) Symmetric single mode or single sinusoidal surface (intrinsic wavelength (λi ) = 4.68 µm). (c ) Asymmetric single mode (λi = 3.92 µm). (d ) Asymmetric double mode (λi = 2.96 µm). (e) Symmetric double mode (λi = 1.68 µm). (Reprinted with permission from [39]. Copyright 2003, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

3.4 MECHANICAL STRESS: PATTERNING OF A METAL SURFACE

61

Figures 3.20b–e. When a patterned PDMS mold is placed on the bilayer, it makes a strong conformal contact with the underlying metal surface, such that the edges of the pattern act as nodes. These nodes dictate the nodal condition to creating waves within the period of the mold pattern. These nodal waves created by the action of the mold interact with the internal, intrinsic wave. The shaping of the metal surface is the result of a linear combination of the nodal waves. If we normalize the metal wave profile with respect to the amplitude ε and utilize the symmetric nature of the mold, the normalized profile ω(x) can be written as ω(x) = ε

∞

n=1

fn sin[(2n − 1)km x] (2n − 1)

n = 1, 2, 3, . . .

(3.28)

where x is the axis of the wave, fn is the fractional contribution from the nth harmonic, and km is the mold wave number given by 2π /λm . Odd harmonics were chosen to satisfy the symmetry condition in the repeating unit structure consisting of one void and one contact for the equal line and space pattern being considered. The number of harmonics and their types in the series can be determined from a freeenergy approach. The free energy of the bilayer [37, 40] F is given by the sum of the bending energy of the metal and the elastic deformation energy of the underlying polymer layer,   2E p (εk)2 Ep E m k 2 tm3

+ 4 3+ (3.29) F(k) = k tp 3k 4 12 1 − νm2 where k is the wave number. Although the wave given by equation 3.28 represents a symmetric harmonic series that consists of odd harmonic waves, asymmetric waves do form as shown in Figures 3.20d and e. This symmetry breaking originates from the system’s desire to minimize its free energy at the risk of an uneven contact of PDMS mold’s pattern, allowing even harmonic waves, in effect selecting nodes from within two periods rather than one period. It can be shown [39] that this minimization of the free energy leads to the range of λm /λi ratio that defines the metal surface shape. Theoretical and experimental values of this range for the type of the surface shape are given in Table 3.1. Table 3.1. Comparison between theory and experiment for the range of the ratio of mold wavelength (λm ) to intrinsic wavelength (λi ) that defines the metal surface shape

Theoretical range (λm /λi )

Shape of harmonic wave

Experimental range (λm /λi )

0.69–1.61 1.61–2.43 2.43–3.82 3.82–4.46 4.46–5.88 5.88–7.93 7.93–

Symmetric single mode Asymmetric double mode Symmetric double mode Asymmetric triple mode Symmetric triple mode Symmetric quadruple mode Symmetric quintuple mode

0.64–1.59 1.50–2.56 2.38–4.23 4.23–4.76 4.55–6.35 6.36–8.33 8.47–

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

62

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

(a) λi = 1.68 µm

(b) λi = 2.96 µm

(c) λi = 4.68 µm

λm= 4 µm

λm= 6 µm

λm= 10 µm

λm= 14 µm

Scan size: 20 µm

Scan size: 30 µm

Scan size: 40 µm

Figure 3.21. Shape engineering of a metal surface. Two-dimensional AFM micrographs and the corresponding surface profiles (insets) of the self-organized metal surface. (a) Column for λi of 1.68 µm. The shape changes from asymmetric double mode (λm = 4 µm), to symmetric double mode (λm = 6 µm), to symmetric triple mode (λm = 10 µm), and to symmetric quadruple mode (λm = 14 µm). (b) Column for λi of 2.96 µm. The shape changes from symmetric single mode (λm = 4 µm), to asymmetric double (λm = 6 µm), to symmetric double (λm = 10 µm), and to symmetric triple (λm = 14 µm). (c ) Column for λi of 4.68 µm. The shape changes from symmetric single mode (λm = 4 and 6 µm), to asymmetric double (λm = 10 µm), and to symmetric double (λm = 14 µm). (Reprinted with permission from [39]. Copyright 2003, American Institute of Physics.)

Shown in Figure 3.21 are AFM images of surface profiles that were generated by wave-directed self-organization. For a given bilayer, the surface shapes are entirely determined by the ratio of the mold and intrinsic wavelengths. From the top down in any given column in the figure, one can see the shape change with the increase in the mold wavelength (λm ) for a given intrinsic wavelength (λi ). Conversely, from left

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

REFERENCES

63

to right in any given row, one can also observe the change with increasing intrinsic wavelength for a given mold. For the case of λm =6 µm, for example, one can generate symmetric double mode, asymmetric double mode, and symmetric single mode wave simply by varying the film thickness. With the number and types of harmonics determined, only the fractional weightings, fn , must be resolved. The expression for the free energy can be derived in a straightforward manner, although cumbersome [41, 42]. The expression is Fn =

cn =

n (ε0 ki )2 2 f cn 4 n=1 n

(3.30)

2E p Ep (2n − 1)3 E m qm2 tm3

+ + 3 4 3 2 (2n − 1) km tp 3km 12 1 − νm

where ε0 is the average amplitude of the intrinsic buckling wave. Minimization of the free energy Fn with respect to fn yields the relative fraction. While patterning a metal surface in this manner is simple and versatile and shape engineering allows one to pattern almost any geometry, it has some limitations. For the method to be more useful, the wave period and amplitude should be extended below 1 µm and above 0.5 µm, respectively.

REFERENCES 1. Kim, E., Xia, Y., and Whiteside, G. M. (1995) Polymer microstructures formed by moulding in capillaries. Nature 376, 581–584. 2. Xia, Y. and Whiteside, G. M. (1998) Soft lithography. Angew. Chem. Int. Ed. 37, 550–575. 3. Trau, M., Yao, N., Kim, E., Xia, Y., and Whitesides, G. M. (1997) Microscopic patterning of oriented mesoscopic silica through guided growth. Nature 390, 674–676. 4. Myers, D. (1991) Surface, Interfaces, and Colloids, VCH, New York, pp. 97–118. 5. Washburn, E. W. (1921) The dynamics of capillary flow. Phys. Rev. 17, 273–283. 6. King, E., Xia, Y., Zhao, X. M., and Whitesides, G. M. (1997) Solvent-assisted microcontact molding: a convenient method for fabricating three-dimensional structures on surface of polymers. Adv. Mater. 9, 651–654. 7. Lee, J. N., Park, C., and Whitesides, G. M. (2003) Solvent compatibility of poly(dimethylsiloxane)-based microfluidic devices. Anal. Chem. 75, 6544–6554. 8. Van Krevelen, D. W. (1990) Properties of Polymers, 3rd edn., Elsevier, Amsterdam. 9. Khang, D. Y. and Lee, H. H. (2000) Room-temperature imprint lithography by solvent vapor treatment. Appl. Phys. Lett. 76, 870–873. 10. Kim, Y. S., Suh, K. Y., and Lee, H. H. (2001) Fabrication of three-dimensional microstructures by soft molding. Appl. Phys. Lett. 79, 2285–2287. 11. Kim, Y. S., Park, J., and Lee, H. H. (2002) Three-dimensional pattern transfer and nanolithography: modified soft molding. Appl. Phys. Lett. 81, 1011–1013.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

64

13:57

Printer Name: Yet to Come

PATTERNING BASED ON NATURAL FORCE

12. Suh, K. Y., Kim, Y. S., and Lee, H. H. (2001) Capillary force lithography. Adv. Mater. 13, 1386–1389. 13. Kang, H., Lee, J., Park, J., and Lee, H. H. (2006) An improved method of preparing composite poly(dimethylsiloxane) molds. Nanotechnology 17, 197–200. 14. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. 15. Khang, D. Y. and Lee, H. H. (2004) Sub-100 nm patterning with an amorphous fluoropolymer mold. Langmuir 20, 2445–2448. 16. Khang, D. Y. and Lee, H. H. (2004) Pressure-assisted capillary force lithography. Adv. Mater. 16, 176–179. 17. Yoon, H., Kim, T., Choi, S., Suh, K. Y., Kim, M. J., and Lee, H. H. (2006) Capillary force lithography with impermeable molds. Appl. Phys. Lett. 88, 254104. 18. Chou, S. Y., Krauss, P. R., and Renstrom, P. T. (1996) Imprint lithography with 25nanometer resolution. Science 272, 85–87. 19. Choi, S., Yoo, P. J., Beak, S. J., Kim, T. W., and Lee, H. H. (2004) An ultraviolet-curable mold for sub-100-nm lithography. J. Am. Chem. Soc. 126, 7744–7745. 20. Yoo, P. J., Choi, S., Kim, J. H., Suh, D., Beak, S. J., Kim, T. W., and Lee, H. H. (2004) Unconventional patterning with a modulus-tunable mold: from imprinting to microcontact printing. Chem. Mater. 16, 5000–5005. 21. Timoshenko, S. and Woinowsky-Krieger, S. (1959) Theory of Plates and Shells, 2nd edn., McGraw-Hill, New York. 22. Yoon, H., Lee, K. M., Khang, D. Y., and Lee, H. H. (2004) Rapid flash patterning of nanostructures. Appl. Phys. Lett. 85, 1793–1795. 23. Xie, R., Karim, A., Douglas, J. F., Han, C. C., and Weiss, R. A. (1998) Spinodal dewetting of thin polymer films. Phys. Rev. Lett. 81, 1251–1254. 24. Martin, P. and Brochard-Wyart, F. (1998) Dewetting at soft surface. Phys. Rev. Lett. 80, 3296–3299. 25. Martin, A., Clain, J., Buguin, A., and Brochard-Wyart, F. (2002) Wetting transition at soft, sliding interfaces. Phys. Rev. E 65, 031605. 26. Merkel, T. C., Bondar, V. I., Nagai, K., Freeman, B. D., and Pinnau, I. (2000) Gas sorption, diffusion, and permeation in poly(dimethylsiloxane). J. Polym. Sci., Part B: Polym. Phys. 38, 415–434. 27. Suh, K. Y., Kim, P., and Lee, H. H. (2004) Capillary kinetics of thin polymer films in permeable microcavities. Appl. Phys. Lett. 85, 4019–4021. 28. Reiter, G. (1991) Dewetting of thin polymer films. Phys. Rev. Lett. 68, 75–78. 29. Israelachvili, J. N. (1991) Intermolecular and Surface Forces, 2nd edn., Academic Press, London, pp. 176–180. 30. Kim, Y. S. and Lee, H. H. (2003) Selective dewetting for general purpose patterning. Adv. Mater. 15, 332–334. 31. Masson, J. L. and Green, P. F. (2002) Hole formation in thin polymer films: a two-stage process. Phys. Rev. Lett. 88, 205504. 32. Jackman, R. J., Duffy, D. C., Ostuni, E., Willmore, N. D., and Whitesides, G. M. (1998) Fabricating large arrays of microwells with arbitrary dimensions and filling them using discontinuous dewetting. Anal. Chem. 70, 2280–2287.

P1: OTA/XYZ P2: ABC c03 JWBK252/Lee September 18, 2008

13:57

Printer Name: Yet to Come

REFERENCES

65

33. Seemann, R., Brinkmann, M., Kramer, E. J., Lange, F. F., and Lipowsky, R. (2005) Wetting morphologies at microstructured surfaces. Proc. Natl Acad. Sci. USA 102, 1848–1852. 34. Kim, M. J., Song, S., Kwon, S. J., and Lee, H. H. (2006) Trapezoidal structure for residual-free filling and patterning. J. Phys. Chem. C 111, 1140–1145. 35. Kim, T., Kwon, S. J., Lee, J., and Lee, H. H. (2006) Residual-free nanofilling with wetting solutions. Appl. Phys. Lett. 89, 173115. 36. Bowden, N., Brittain, S., Evans, A. G., Hutchinson, J. W., and Whitesides, G. M. (1998) Spontaneous formation of ordered structures in thin film of metals supported on an elastomeric polymer. Nature 393, 146–149. 37. Yoo, P. J., Suh, K. Y., Park, S. Y., and Lee, H. H. (2002) Physical self-assembly of microstructures by anisotropic buckling. Adv. Mater. 14, 1383–1387. 38. Ferry, J. D. (1980) Viscoelastic Properties of Polymers, John Wiley & Sons, New York. 39. Yoo, P. J., Park, S. Y., Kwon, S. J., Suh, K. Y., and Lee, H. H. (2003) Microshaping metal surfaces by wave-directed self-organization. Appl. Phys. Lett. 83, 4444–4446. 40. Groenewold, J. (2001) Wrinkling of plates coupled with soft elastic media. Physica A 298, 32–45. 41. Landau, L. D. and Lifshitz, E. M. (1970) Theory of Elasticity, Pergamon, Oxford. 42. Fredrickson, G. H., Ajdari, A., Leibler, L., and Carton, J. P. (1992) Surface modes and deformation energy of a molten polymer brush. Macromolecules 25, 2882–2889.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4 PATTERNING BASED ON WORK OF ADHESION

4.1 INTRODUCTION Patterning based on work of adhesion typically involves the transfer of a film on a patterned mold to a substrate. The resulting pattern corresponds to the raised regions on the mold or stamp since typically only the parts of the film on these regions are transferred. The film may consist of only one layer or a number of layers of different materials. For the transfer to be successful, the work of adhesion at the mold–film interface has to be smaller than that at the film–substrate interface. This type of approach has advantages compared to imprint lithography and other methods such as those described in Chapter 3. First, the substrate surface is exposed except where the film is transferred. Therefore, there is no need for subsequent reactive ion etching, which is typically needed to remove the residual layer that can remain after patterning as in imprint lithography, for example. Another, more subtle, advantage is that the patterning does not involve any solvent or liquid making it well suited for the patterning of organic devices, where exposure to solvents often leads to degradation or damage of the active layers. While any type of mold can be used for transfer patterning, it is extremely difficult to use a hard mold, such as one made with a silicon wafer, due to practical challenges in establishing intimate physical contact. When a soft mold such as poly(dimethylsiloxane) (PDMS) mold [1] is used, the flexibility of the material allows for conformal contact with the substrate surface, even in the presence of surface roughness and topography. Because the applied pressure is felt sequentially, first by the greatest protrusion and then by smaller protrusions, a much lower pressure than with a hard mold can be used for the transfer patterning. When pattern registration

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


67

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

68

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

and fine features are important, soft molds are inadequate because of mold deformation under pressure. In such cases, rigid backing supports or rigiflex molds [2], as discussed in Chapter 2, can be used. In this chapter, the work of adhesion, which has a significant role in transfer patterning, is discussed first. This quantity can be determined from contact angle measurements of two probe liquids on the two surfaces that constitute an interface. The work of adhesion can be manipulated by treating the mold surface or substrate surface with a film coating or chemical modification, as described in Chapter 2 and subsequently in the present chapter. This discussion is followed by a section on the kinetic effects that are important in the transfer process. With this background information, we then discuss patterning based on well engineered work of adhesion. Transfer patterning involving a single layer is treated first, followed by a treatment on multilayer transfer. This latter process takes on a meaning that goes beyond simple patterning when the multilayer represents a whole device structure because the transfer itself signifies the fabrication of a device. The transfer of a film pattern from a mold onto a substrate is referred to as additive patterning. When the work of adhesion at the mold–film interface is larger than that at the film–substrate interface, the parts of a film coated on a substrate that are in contact with the patterned mold can be removed from the substrate, which corresponds to subtractive patterning. A final type of process involves the use of a mold to pick up a pattern from a substrate and then to transfer it to another surface. This form of transfer printing is treated last, after discussion of the additive and subtractive modes.

4.2 WORK OF ADHESION The basic concept behind patterning based on work of adhesion is illustrated in Figure 4.1. A mold prepared for the desired pattern is coated with a film to be transferred. This mold is brought into contact with a substrate and pressure is applied typically at a mild temperature to facilitate the transfer. After a certain period of time and cooling, the mold is removed. In this removal step, the film on the protruding parts of the mold gets transferred to the substrate. For this transfer to take place, the work of adhesion at the mold–film interface Wmf has to be smaller than that at the film–substrate interface Wfs , i.e., Wfs > Wmf . For the case of intimate contact and absence of specific chemical interactions, this work of adhesion can be determined from contact angle measurements of two probe liquids on the surfaces of interest [3, 4]. Based on the linearly additive nature of molecular forces, the work of adhesion Wij between two surfaces, i and j, is written as follows, p

Wij = Wijd + Wij

(4.1)

where the dispersion (nonpolar) and polar components are denoted by the superscripts d and p, respectively. For a liquid on a solid surface [3], the geometric

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.2 WORK OF ADHESION

69

Mold Wmf

Film to be transferred

Substrate

Mold Substrate

Contact of the mold with the substrate while applying pressure and mild temperature

Mold Removal of the mold

Wfs Substrate

Figure 4.1. Schematic illustration of the procedure for patterning based on work of adhesion (transfer patterning).

mean is used for each of the components such that the interfacial work of adhesion becomes

p 1/2 1/2 p Wls = Wlsd + Wls = 2 γld γsd + 2 γl γsp

(4.2)

where γ is the surface tension and the subscripts l and s are for the liquid and solid, respectively. From the Young–Dupree equation, however, the work of adhesion is given by Wls = γl (1 + cos θ )

(4.3)

p

where θ is the contact angle and γl = γld + γl . Combining equations 4.2 and 4.3 yields 1/2

p 1/2 γl (1 + cos θ ) = 2 γld γsd + 2 γl γsp

(4.4)

Two probe liquids, typically water and ethylene glycol, for which the dispersion and p p polar components γld and γl are known, are used for the determination of γsd and γs . Take as an example the transfer of a polystyrene (PS) film on a poly (urethaneacrylate) (PUA) mold to a silicon wafer surface with an oxide layer (S). The contact angles determined experimentally with water and ethylene glycol θ w and θ E , respectively, on PUA, PS, and S are given in Table 4.1a. Also given in the table are the dispersion and polar components of the two liquids. To determine the values of γsd

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

70

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

Table 4.1. Contact angles and surface tensions

(a) Contact angles of water (θ W ) and ethylene glycol (θ E ) on PUA, PS, and S

θW θE

Water Ethylene glycol

PUA

PS

S (SiO2 )

60.3◦ 16.7◦

96.2◦ 63.7◦

54.1◦ 16.2◦

γld (mJ m−2 )

γl (mJ m−2 )

γl (mJ m−2 )

23.9 29.2

48.8 18.3

72.2 47.5

p

(b) Dispersion and polar components of surface tensions of PUA, PS, and S

PUA PS S

γ d (mJ m−2 )

γ p (mJ m−2 )

γ (mJ m−2 )

30.5 35.9 22.1

15.4 0.2 24.6

45.9 36.1 46.7

p

and γs for PUA, for example, equation 4.4 is written twice for the two different liquids, utilizing the measured contact angles:
1/2
1/2 72.7(1 + cos θw ) = 2 γsd · 23.9 + 2 γsp · 48.8 for water
d 1/2
p 1/2 47.5(1 + cos θE ) = 2 γs · 29.2 + 2 γs · 18.3 for ethylene glycol Since θw = 60.3◦ and θE = 16.7◦ on PUA, these values can be used in the p d = above equations to solve for γsd and γs . The values thus determined are γPUA p −2 −2 −2 30.5 mJ m , γPUA = 15.4 mJ m , and γPUA = 45.9 mJ m . These values, determined for all three surfaces of interest, are summarized in Table 4.1b. For the interfacial work of adhesion between two solid materials [3, 4], the harmonic mean approximation is used for each of the components in equation 4.1, yielding Wij =

4γid γ jd γid + γ jd

p

+

p

4γi γ j p

p

γi + γ j

(4.5)

For the example just considered, the information that is necessary to determine Wij is given in Table 4.1. The interfacial work of adhesion calculated from equation 4.5 with the information in Table 4.1 is WPUA/PS = 66.8 mJ m−2 , WPS/S = 55.5 mJ m−2 . These values suggest that a film of PS on a PUA mold cannot be transferred to a silicon wafer surface since WPUA/PS > WPS/S . These quantities do indicate, on the other hand, the possibility for transferring PS from a silicon mold to PUA.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.3 KINETIC EFFECTS

71

This example raises another possibility. If the work of adhesion between a PUA mold and PS is larger than that between PS and a substrate, then a PS film coated on that substrate could be removed, or selectively lifted off, by the mold. This is the case for subtractive patterning as opposed to the additive patterning, which we have discussed so far. This mode of subtractive transfer is different in nature from a similar process based on cold welding [5], where a metal on a substrate is selectively lifted off by the same metal deposited on a mold. When two atomically flat metallic surfaces are brought into contact, an abrupt transition, termed an adhesive avalanche, takes place transforming the geometry from two spatially distinct surfaces to a single homogeneous structure [6], behaving like a single bulk metal. In the patterning based on work of adhesion, there exist two spatially distinct surfaces and no single homogeneous structure results. Many additional details are involved in subtractive transfer. Although not shown in Figure 4.1, the material coated on the mold could cover the entire contour of the mold. When pressure is applied, the mechanical stress diverges at the edges of the raised parts of the mold [7], leading to film fracture at these locations. In the subtractive transfer, however, fracture is not necessarily caused by the applied pressure. When a patterned mold is brought into contact with a film coated on a substrate and pressed under moderate temperature, the raised parts of the mold get pushed into the film. Upon lifting the stamp, separation of the film from the substrate surface takes place first and the film is lifted up since the work of adhesion at the mold–film interface is larger than that at the film–substrate interface. Once the film is separated or lifted from the substrate, fracture occurs along the edges of the raised parts. Therefore, the work of adhesion at the stamp–film interface has to be larger than the fracture energy. This requirement places a constraint on the choice of film material. If a polymer is used, for instance, the molecular weight of the polymer has to be well below a critical value. In general, the fracture energy increases with molecular weight [8]. An abrupt decrease in the fracture energy occurs at critical molecular weight, which, for example, is 33,000 for PS. Below the critical molecular weight, the fracture energy is proportional to the square root of the molecular weight.

4.3 KINETIC EFFECTS Kinetically controlled adhesion to stamps and molds offers a velocity-dependent approach for manipulation and patterning of films or discrete objects on a substrate [9, 10]. Meitl et al. [9] have shown that when an elastomeric mold is placed in contact with a substrate supporting a thin film or an organized array of solid objects, fast separation of the stamp from the substrate with peeling velocities of ∼10 cm s−1 or faster can greatly increase the tendency of elements to adhere preferentially to the stamp. Conversely, slow delamination with slow peel velocities of ∼1 mm s−1 or less can result in the transfer of these elements from the mold to a target substrate. This technique provides an important combination of capabilities that are not offered

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

72

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

by other assembly methods such as wafer bonding or “pick-and-place” technologies. Applications of this technique in relation to transfer printing of device components and materials elements are described in Section 4.6. To explain the adhesive strength rate dependence, separation at the film–stamp and film–substrate interfaces has been modeled as a propagating crack [11]. In this analysis, film or object retrieval and printing serve as two competing fracture pathways, each with a characteristic energy release rate G defined as G=

F w

(4.6)

where F is the peel force applied to the stamp and w is the out-of-plane width. Here, G accounts for both the interfacial bond breaking and viscoelastic energy dissipation around the “crack” tip [11, 12], and is thus different from the work of adhesion introduced in Section 4.2. The action of pickup and printing of objects correspond to critical values of the energy release rate GCRIT at the stamp–film and film–substrate interfaces. To determine which mode is favored, comparison of these critical values (and hence comparison of critical peel forces) establishes a criterion for retrieval or printing: film

stamp

film substrate

stamp film

substrate film < G CRIT G CRIT

G CRIT > G CRIT

for pickup

(4.7a)

for printing

(4.7b)

Due to the elastic nature of both the film and the substrate, there is no dependency stamp

film

substrate film on peel velocity for G CRIT . The mold, however, is viscoelastic and as a result G CRIT stamp

film is velocity dependent, i.e., G CRIT (ν) [12]. This implies that for a critical velocity νc the two energy release rates are equal,

film

stamp

substrate film G CRIT = G CRIT (νc )

(4.8)

marking the transition from the pickup to printing regime. Using a power law relation that is often found to be useful in solving problems involving soft adhesion [10, 12], stamp film

G CRIT

  n  ν = G0 1 + ν0

(4.9)

the critical velocity can be determined by rearrangement, 

1/n stamp film G CRIT − G0  νc = ν0  G0

(4.10)

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.3 KINETIC EFFECTS

73

where G0 is the critical energy release rate as indicated by a peeling velocity v approaching zero, ν0 is a reference peeling velocity related to G0 , and n is a scaling parameter. Figure 4.2a illustrates graphically the pickup to printing transition as a film substrate function of velocity. Several values of G CRIT are plotted as horizontal lines (velocity independent) corresponding to critical energy release rates on surfaces with various stamp film (ν) determines the relaadhesive strengths. Intersection of these curves with G CRIT tive νc [10]. For the mid-level curve in Figure 4.2a, the critical velocity is marked,

G

Strong film−substrate interface

Film Substrate (J-m−2)

(a)

G

Stamp Film crit

(v)

Pickup

Gcrit

Pnnting

G0

Weak film−substrate interface

0

Vmax

Vc

V

20 (b)

4°C 24°C

G (J m−2)

15

10 37°C 5 Experimental data Stamp Film

Gcrit 0 0

10

20

30 40 V (cm s−1)

50

60

Figure 4.2. Schematic diagram of critical energy release rates. (a) Intersection of the horizontal line in the middle with the curve represents the critical peel velocity for the kinetically controlled transfer printing. The horizontal lines at the bottom and top represent very weak and very strong film–substrate interface, respectively, corresponding to pickup only and printing only. (b) Temperature dependence of critical energy release rates. Increasing temperature leads to larger critical peel velocity. (Reprinted with permission from [10]. Copyright 2007, American Chemical Society.)

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

74

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

but when no intersections are present, as is the case for the top and bottom horizontal lines, no practical νc can be realized for the system. In this example, the lower horizontal line indicates a weak film–substrate interface, which prevents printing of objects or films using only kinetic control. The upper horizontal curve meanwhile represents a strong film–substrate interface, which prohibits film retrieval. In such cases, transfer requires chemical or physical mediation, such as glue layers or surface treatments. Chapter 18 discusses in more detail the mechanics of soft lithography. Temperature can also significantly alter the transfer characteristics of film pickup and printing by changing the values of the relative critical velocity. Figure 4.2b illustrates the critical energy release rate for different velocities across several ambient temperatures ranging from 4◦ C to 37◦ C. In the experiment, flat PDMS molds were brought into contact with thin films of Ti/Au (1 nm/100 nm thick) supported on glass slides. Controlled loads were then applied to the stamp to obtain well-defined sepfilm

substrate aration velocities. As evident from the plot, for a constant value of G CRIT , νc increases with increasing temperature [10]. Similar trends are found analytically when full consideration is given to temperature influences on the mechanics of viscoelastic materials, which is beyond the scope of this text [12–14]. The results, however, indicate that for kinetically controlled adhesion, low temperatures are preferable for film or object retrieval while elevated temperatures favor printing.

4.4 TRANSFER PATTERNING In contrast to microcontact printing [15], in which molecular materials are patterned onto a substrate, transfer patterning involves manipulation of layers of solid materials. Transfer patterning is attractive for several applications, one of which is patterning of thin metal films to act as etch resists for deep etching processes. This technique can also be used to pattern device components during fabrication of organic and inorganic electronics, topics that will be discussed in detail in Chapters 16 and 17, respectively. Transfer patterning is particularly appealing to these systems since solvents used in traditional photolithographic processing can damage or degrade an organic layer or substrate. Shadow mask methods and other alternatives have limited resolution and are difficult to use over large areas. Although Figure 4.1 shows only the layer to be transferred on the stamp, typically a release layer of low surface energy material is inserted between the transfer layer and the mold to facilitate the process. For example, in the formation of an aluminum cathode for an organic light-emitting diode (OLED) [16], a self-assembled monolayer (SAM) of (tridecafluoro-1,1,2,2,-tetrahydrooctyl)-trichlorosilane is first deposited on the mold followed by the deposition of the aluminum for transfer. In surface chemistry mediated transfer processes [17], chemical bonding involving hydroxyl groups can be utilized for the transfer as an example. For this purpose, either the transfer layer or the substrate or both are treated by plasma oxidation to generate surface hydroxyl groups. This unique way of pattern transfer was later extended

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.4 TRANSFER PATTERNING

75

to more general cases, known as nanotransfer printing [19], which is equivalent to transfer patterning. An essential aspect of transfer patterning is the nature of the mold material. When a hard mold such as a silicon wafer or quartz plate is used, the pressure needed for transfer can become extremely high. The necessity of intimate physical contact over the entire contact area requires application of an excessive load which increases with increasing contact area. This makes large area applications difficult to address with a hard mold. A soft mold such as PDMS, on the other hand, easily deforms when a pressure is applied. However, there are inherent mechanical problems such as lateral deformation and roof collapse associated with PDMS stamps making them inadequate for patterning dense, sub-micrometer features [20]. A mold that is suitable for transfer patterning should be rigid enough so that it has the mechanical stability to withstand the applied pressures without deforming small mold features while simultaneously being flexible enough to provide intimate physical contact over large areas. Examples of such a rigiflex mold are a PUA mold [2] and a Teflon mold [21]. The tensile, or Young’s modulus, of these materials are 0.4 GPa for PUA and 1.6 GPa for Teflon, values between the modulus of PDMS (10 MPa at most) and silicon hard molds (100 GPa). These rigiflex molds are rigid enough to provide mechanical stability and yet flexible in their film form. The maximum deflection of a plate W m when subjected to a pressure P, for which the thickness is t and Young’s modulus is E, is given [22] by Wm =

12(1 − ν 2 )P g(α, β) Et 3

(4.11)

where ν is the Poisson ratio and g depends on the dimensions of the plate α and β. The deflection or flexibility is inversely proportional to Et3 . By making the film mold thin or (making t small), the flexibility can be greatly enhanced. When used to transfer a metal pattern, it can be advantageous to transfer a bilayer of metal on a polymer such that the polymer serves as a binding layer to the substrate. This bilayer transfer (BLT), [23], is illustrated in Figure 4.3. In this example, a rigiflex film mold of PUA with a poly(ethylene terephthalate) (PET) sheet as a backing layer is utilized. The molds were replicated from the master patterns formed by electron-beam lithography on a silicon wafer. Onto this mold, perfluoroethylenepropylene (FEP) is thermally deposited as an anti-adhesion layer designed to lower the interfacial work of adhesion. The mold is then coated by aluminum via thermal deposition. Poly(vinylacetate) (PVAc) is then spin coated onto the metal surface, providing an adhesion layer. The stamp with the bilayer is then brought into contact with a silicon wafer surface with a pressure of 0.8 MPa at 50◦ C (second frame of Figure 4.3a). With the pressure maintained, the temperature is then lowered to room temperature and the pressure is relieved, thus completing the BLT to the substrate. The PVAc layer ruptures cleanly along the edges of the contact area and with the metal pattern transferred as an etch mask, the substrate is then etched by reactive ion etching (RIE). After etching, the bilayer is removed by immersing in polypylene glycol monomethyl ether acetate, a typical developer, which dissolves the polymer

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

76

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

(a)

FEP AI PVAc Substrate Pressing

Detaching

RIE

Dissolving

Pattern (b)

Mo

ld

Im p cy res lin si de on r

Substrate

Figure 4.3. Schematic illustration of bilayer transfer. (a) Bilayer transfer (BLT) with a flat mold. (b) Roller BLT with a cylindrical roller. PUA, poly(urethane-acrylate); PET, poly(ethylene terephthalate); FEP, perfluoroethylenepropylene; PVAc, poly(vinyl acetate); RIE, reactive ion etching. (Reprinted with permission from [23]. Copyright 2005, Wiley-VCH Verlag GmbH & Co.)

selectively, thereby completing the pattern formation on the substrate. The fact that the mold is flexible implies that it is possible to perform BLT in a continuous process, which is illustrated in Figure 4.3b. A film PUA mold with deposited bilayer is inserted with a hard or soft substrate between the two rollers as shown in the figure. In this roller BLT mode BLT, the PUA film mold is pressed into the substrate, and the rotation of the rollers pushes both the mold and substrate forward to generate a patterned bilayer on the substrate.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.4 TRANSFER PATTERNING

(a)

(b)

(c)

(d )

77

Figure 4.4. SEM images of master and transferred patterns. (a) Master with 80-nm line width and 270-nm spacing. (b) Bilayer pattern transferred with the film PUA mold made from the master. (c ) Bilayer transfer result of a combination of small and large features (bilayer: dark; SiO2 surface: bright). (d ) 1.2 µm diameter dots and holes transferred with negative and positive molds. (Reprinted with permission from [23]. Copyright 2005, Wiley-VCH Verlag GmbH & Co.)

Shown in Figure 4.4 are some results of the BLT. The master in Figure 4.4a is a line (80 nm) and space (270 nm) pattern, which was used to produce the BLT result in Figure 4.4b. The PUA mold pattern is the negative of the master pattern. Therefore, the transferred bilayer in Figure 4.4b has a line width of 270 nm and a spacing of 80 nm. As is apparent in the first two panels of Figure 4.4, the rounded edges of the master are replicated in the PUA mold (roughness ∼10 nm), which results in the roughness of the edges of the transferred lines in Figure 4.4b. The results in Figures 4.4c and d show that the BLT method is effective for small and large features (Figure 4.4c) and negative and positive features (Figure 4.4d), which in this case represents a combination of raised dots and recessed holes. One major purpose of patterning in device fabrication is to use the patterned layer as an etch mask when transferring the pattern to the underlying substrate via etching. It is often very difficult to produce a thick enough resist layer for small features, resulting in shallow etching of the underlying substrate. Metals, however, are usually good etch resists, yielding a large etch selectivity. For example, an SiO2 layer can be etched to 900 nm deep with a 100-nm-thick aluminum bilayer, giving an etch selectivity of 9. An example of the results obtainable by roller BLT [23] is shown in Figure 4.5. A complex pattern transferred to a SiO2 substrate is given in Figure 4.5a. The substrate etched by a CF4 /CHF3 RIE with the transferred bilayer as the etch mask is shown

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

78

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

(a)

(b)

(c)

Figure 4.5. SEM images of a complex pattern transferred by roller bilayer transfer. (a) Bilayer pattern transferred to SiO2 substrate. (b) SiO2 substrate after etching with CF4 /CHF3 plasma RIE. (c ) Tilted view of (b). (Reprinted with permission from [23]. Copyright 2005, Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.5 SUBTRACTIVE TRANSFER PATTERNING

79

in Figure 4.5b, and its tilted view in Figure 4.5c. The smallest feature size is about 200 nm. The excellent fidelity of the etched pattern is self-evident. In this example of roller BLT, the bottom roller in Figure 4.3b was heated to 50◦ C while the upper roller was not. The fidelity of the transferred pattern was poor when the temperature was too low, the rolling speed was too high, or the pressure was not high enough. The results shown in Figure 4.5 were obtained at a roller speed of 700 µm s−1 , a temperature of 50◦ C, and a pressure of about 0.4 MPa. It is noted in this regard that the transferred pattern does not have the deformation problem that has been found in other forms of roller imprinting [24]. The thickness of the transferred bilayer is uniform provided that the layers are deposited uniformly on the mold surface. This fact can be utilized to fabricate multilayer structures by simply carrying out the transfer repeatedly. Various multilayer structures can be formed with different step heights or configurations and by changing the direction of the pattern transfer. Because of the relatively low temperature used in the transfer, step and repeat, which is typically used for patterning fine feature sizes over large areas, can be carried out. As discussed earlier, a promising area for the application of transfer patterning is in the fabrication of organic devices. A metal, organic, or polymer layer that constitutes an element of an organic device, or multiple layers of these materials, can be transferred for device fabrication. These applications of the transfer patterning are detailed in Chapter 16. When multiple layers deposited on a mold are transferred to a substrate, additional consideration must be given to the work of adhesion between the various layers. For an N-layer system, let W 12 be the work of adhesion between the stamps anti-adhesion layer and the layer deposited on top of it. Similarly, let Wij be the interfacial work of adhesion between intermediate layers i and j (i, j = 2, . . . , N − 1) and W (N − 1)N be the adhesive work between the (N − 1) and Nth layers. In order to support multilayer transfer in this system, the following condition must be satisfied: W12 < Wij < W(N −1)N

(4.12)

This condition applies not only to transferring layers used to build a single device, but also for the patterning of completed device layers. One example of a multilayer device structure utilizing this transfer technique is the formation of red, green, and blue OLED pixel clusters [25].

4.5 SUBTRACTIVE TRANSFER PATTERNING As mentioned previously, patterning of a film coated on a substrate by removing regions of the film via mold relief features is termed subtractive transfer patterning. As opposed to the additive nature of transfer patterning, where a film on a mold is selectively added to a substrate, subtractive patterning effectively removes film from the substrate in the desired geometry. The process of subtractive transfer patterning is illustrated in Figure 4.6a [26]. In this example, the substrate, a metal film deposited on glass, has a polymer layer spin

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

80

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

(a) PUA mold

Polymer Metal

Glass Contact

PUA mold

Polymer Metal

Glass Heat (T >Tg) and press

Polymer Metal

PUA mold

Glass Detach

Polymer

PUA mold

Polymer Metal Glass

(b)

L

2L

L

PUA mold H

Polymer

h

k

Substrate

Figure 4.6. Schematic illustration of subtractive transfer patterning. (Reprinted with permission from [26]. Copyright 2005, American Institute of Physics.)

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.5 SUBTRACTIVE TRANSFER PATTERNING

81

coated on it. A patterned PUA mold is placed on the polymer film and an external pressure is applied while heating to a temperature higher than the glass transition temperature (T g ) of the polymer. After a period of contact (∼10 s), the system is cooled to room temperature and the mold removed. The parts of the polymer film that were in contact with the raised regions of the stamp adhere to the mold and are subsequently removed, leaving behind the desired polymer pattern on the metal surface. Subtractive transfer patterning of this type has found some applications in patterning of conducting polymers [27]. While many of the same mechanical and geometrical considerations that apply to transfer patterning also apply to subtractive patterning, there are also additional factors that have to be taken into account. In transfer patterning, the applied mechanical stresses are often high enough to overcome the fracture energy needed to separate the film along the edges of the protruding features of the mold. Hence, the pressure is responsible for film fracture. In subtractive transfer patterning however, the flat polymer film on a substrate fractures along the edges of the protruding mold features during liftoff and removal. Therefore, pressure plays no role in the actual fracturing of the film. The difference in the interfacial work of adhesion, W, which is the work at the mold–polymer interface minus that at the polymer–substrate interface, must than be larger than the fracture energy for the subtractive pattern transfer to be effective, or W (2L) > K h

(4.13)

where 2L is the pitch of an equal line and space pattern, h is the polymer film thickness, and K is the fracture energy. An implicit assumption is that L  H (see Figure 4.6a). The fracture energy of a polymer increases slowly with its molecular weight up to a critical value, M c , and then increases dramatically with increasing molecular weight [8]. Therefore, the molecular weight of the polymer has to be smaller 1/2 than M c . In this range, the fracture energy is given by K = 0.0065Mn (J m–2 ) [8], where M n is the number-averaged molecular weight. Equation 4.13 can be rewritten as follows: 1/2

W >

0.0065 Mn h , 2L

Mn < Mc

(4.14)

There is an additional constraint on the use of a polymer film for subtractive transfer patterning. In a polymer film, propagation of a craze/crack in the film leads to the fracture. The microstructure of a craze in a polymer film strongly depends on the film thickness [28]. When the film thickness is on the order of 100 nm or larger, the craze structure is a network of fibrils, but below this critical thickness, the craze structure is essentially two dimensional, across which the film fractures readily. Because of this structural advantage, the fracture energy for polymer film with its thickness less than 100 nm is much smaller than that given by equation 4.14. For the subtractive transfer patterning to work in an optimal way with a polymer film, therefore, the film

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

82

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

thickness should be less than 100 nm and the molecular weight should be well below the critical molecular weight. While the fracture energy plays an essential role in patterning with an organic polymer film, it plays a negligible role with a small molecule organic film since fracture energies involved are much less than the interfacial work of adhesion. Therefore, similar principles for transfer patterning of organic films also apply to subtractive transfer patterning. Subtractive patterning of organic films is simple and requires only small applied pressures that ensure intimate physical contact between the mold and substrate. In fact, any pressure applied is just to ensure the intimate physical contact of the mold with the substrate whether an organic or polymeric film is involved. Due to the negligible fracture energy of small molecule organics, the simple application of subtractive transfer patterning results in patterned organic layers. One application utilizing this approach is the patterning of organic layers in OLEDs [29] with PDMS molds. Typically, the patterning temperature is raised to just below T g of the oligomeric organic, below ∼90◦ C, but no pressure is required because of the good conformal contact between the PDMS mold and the organic layer. The duration of the contact at a given temperature determines whether a residual layer remains after the subtractive transfer; i.e., a short contact leads to some residual layer remaining after the transfer. This mode of patterning of organic layers has been extended [30] to nanometer range with the use of a rigiflex mold and with application of low pressure (1–2 bar) at ambient conditions.

4.6 TRANSFER PRINTING “Printing,” as the term is most commonly used in the literature, refers to the process of retrieving a pattern or object from a donor substrate via an elastomeric stamp or mold followed by transfer to a target or acceptor substrate [31–34]. In contrast to the types of patterning techniques described previously in Sections 4.4 and 4.5, the transfer printing methods introduced in this section can be used to print objects that are complex, fully formed structural units such as device components (semiconductor ribbons, carbon nanotubes) [34–36], materials elements (microspheres, graphite sheets) [9], or biological entities rather than chemical or physical films intended for use in subsequent processing steps. As a consequence, it is possible that these objects have undergone extensive prefabrication (doping, etching, etc.) prior to printing. Figure 4.7 illustrates schematically the printing process. The sequence begins with the preparation of a patterned array of objects on a donor substrate. These objects are brought into conformal contact with an elastomeric stamp whose subsequent removal retrieves the structures from surface. The “inked” stamp is next brought into contact with a receiving substrate and then peeled away, during which the objects can transfer from mold to target. Factors such as peel velocity [9], chemical treatment [37], surface energy [10], mold topography, and substrate geometry play critical roles in mediating the transfer as well as determining the final placement and assembly of the transferred objects. These and other effects will be addressed throughout this section.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.6 TRANSFER PRINTING

83

Stamp

(i) Prepare donor substrate apply rubber stamp

Donor

(ii) Quickly peel-back stamp; grab object off of donor

(iii) Apply inked stamp to receiving substrate

Receiver

(iv) Slowly peel-bake stamp; print object onto receiver

Figure 4.7. Overview of the transfer printing process. Objects from a donor substrate are retrieved using a flexible, elastomeric stamp. This stamp is then placed in contact with an acceptor substrate and by controlling the peel velocity the objects can be transferred to the acceptor. (Reprinted with permission from [9]. Copyright 2006, Nature Publishing Group.)

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

84

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

Stamps used for transfer printing, and soft lithography in general, can be made from a wide variety of materials, ranging from rigid solids such as glass or silicon as discussed in earlier sections to flexible plastics or viscoelastic elastomers such as PDMS [38, 39]. The choice of mold material, as in the cases of transfer and subtractive patterning, depends on a variety of factors including applied stamp pressure and substrate material and is discussed in more detail in Chapter 2. As outlined in Chapter 2, the high resolution molding capabilities, chemical stability, and low mechanical stiffness (∼1–10 MPa) afforded by PDMS represent key attractive features for printing applications, making it one of the most commonly used stamp materials [39, 40]. Additionally, in the absence of surface treatments, PDMS does not exhibit excessive tackiness, providing reversible, nondestructive conformal interface contacts [38] (i.e. stamps, objects, substrates) and minimal distortion during both the pickup and printing phases, when the stamps are coupled with hard backings such as glass slides [39]. Transfer printing using viscoelastic molds also enables unique microfabrication capabilities including sub-micrometer, and even molecular scale [40], patterning resolution and large area selective printing [41, 42]. The stamp geometry can take the form of flat slabs used for uniform transfer as in Figure 4.7 [41], or can incorporate complex surface features for selective contact with a substrate [43], similar to the molds used in transfer patterning. Contoured stamps are often fabricated by casting uncured liquid PDMS against a rigid substrate that has been patterned to have a specific surface structure. Curing and removal of the PDMS results in molded regions of raised surface relief that when placed in contact with a substrate will preferentially retrieve or deposit objects. This type of selective printing has the attractive application of “area multiplication” in which selected sets of large numbers of components fabricated on a bulk substrate can be transferred, in a step-and-repeat fashion, to cover a larger substrate or multiple targets at comparatively low fill fractions [41]. In this process, repetitive stamping distributes organized collections of objects in a well-controlled manner across a large target area. Figure 4.8 illustrates some examples of selective transfer and area multiplication using microribbons formed via lithography and etching [41, 45, 46]. After processing, the ribbons are transferred over several printing steps from the donor wafer onto large, flexible plastic sheets [47]. With these procedures, ribbon fabrication and printing can be carried out multiple times to consume the thickness of the source wafer [44]. Successful implementation of transfer printing requires a sequential process, similar in many regards to the patterning process, in which printed elements first adhere preferentially to the stamp (rather than the donor substrate) and then preferentially to the target substrate (rather than the stamp). Such a situation can be realized in several ways, including chemical treatment of the stamp and/or target surface [31, 35, 37, 39, 42], kinetic manipulation of adhesion at the ink–stamp interface [9, 10], or by a combination of both [41]. As mentioned in Section 4.4, surface treatment of PDMS stamps and inorganic layers can, for example, lead to the formation of strong, bridging Si–O–Si bonds between the mold and object, such as in the case of oxidized PDMS and SiO2 [48]. These covalent bonds result from condensation reactions between silanol groups on the stamp and –OH groups on the SiO2

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.6 TRANSFER PRINTING

(a)

PDMS Si donor substrate

85

(b) 6 mm

(i)

Retrieval (ii) Repeat (iii)

Multiple transfer print

(iv)

20 mm

(c)

Regeneration

Receiving substrate

Regenerated substrate 2 cm

Figure 4.8. Large area selective printing of Si ribbons. (a) Schematic for area multiplication of Si ribbons. PDMS stamp retrieves ribbons from templated substrate (i)–(ii) and prints them onto acceptor (iii). Process continues until donor is depleted of ribbons at which point chemical and/or mechanical polishing regenerates a flat Si surface available for subsequent patterning. (b) Optical image of a PDMS-coated glass substrate with arrays of printed Si ribbons. Fabrication involved repeated transfer printing from a 12 mm × 14 mm donor (inset) onto a 40 mm × 48 mm piece of glass. (Reprinted with permission from [44]. Copyright 2007, Wiley-VCH Verlag GmbH & Co.) (c ) Si structures distributed across a large PET substrate. (Reprinted with permission from [41]. Copyright 2005, Wiley-VCH Verlag GmbH & Co.)

[37]. The resulting strong bonds can, if necessary, provide the adhesion necessary to remove objects from a source substrate. Alternatively, chemistries between noble metals and thiol surface groups [15, 42, 49, 50], usually in the form of alkanethiol SAM’s, or cold welding [5, 33, 51] between thin metal layers on the stamp and object can produce significant adhesion. In the absence of chemical or physical bonding mechanisms, generalized short range van der Waals’ forces can often be sufficient for pickup of objects [52, 53]. Release of objects from the stamp to a target substrate can be enhanced through the use of thin adhesive layers, especially if the target exhibits even modest surface roughness. During printing, thin films of soft or liquid adhesives flow and conform to the printable elements upon contact [43, 54]. The adhesive is then thermally or photocured and the mold removed, completing the transfer. Typical adhesive

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

86

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

O2 plasma-treated PDMS PDMS

Si–O–Si bond Anisotropically etched GaAs

Picking up in inked state

Retrieving GaAs wires PDMS

Printing on PU-coated PET film PU PDMS PET Printing

UV curving Embeded GaAs Wet etching

Post-treatement

Figure 4.9. Stamp and acceptor substrate chemical treatments as used to promote retrieval and transfer of GaAs nanowires.

materials include polyimide [31], epoxies [43], polyurethanes (PU) [31], and thin layers of PDMS [54]. Figure 4.9 shows an example of chemically mediated printing that utilizes both a stamp surface treatment and a target adhesive layer. In this sequence, GaAs nanowires are fabricated from a bulk wafer with a residual SiO2 etch mask. Contact is established between the wires and a flat PDMS stamp that has been exposed to an oxygen plasma, allowing strong covalent siloxane bonds to form at the wire-PDMS interface. Removal of the stamp then disengages the wires from the donor wafer. Printing involves a target substrate of PET coated with a liquid layer of PU. After placing the inked stamp against this layer, the PU is cured into a solid form by exposure to ultraviolet light. Removing the PDMS leaves the GaAs wires embedded in the matrix of cured PU with an exposed SiO2 surface layer. In a final step, the SiO2 mask can be removed by a wet etchant. Kinetic effects, rather than adhesives, can also be used effectively in transfer printing, in a manner that is particularly attractive for applications that involve intimate contact between the printed objects and the target substrate [9, 10]. As discussed in Section 4.3, velocity plays a critical role in retrieving or printing objects using an elastomeric stamp. Removing objects from a donor substrate requires high peel velocities, whereas transfer to a target substrate necessitates much lower velocities,

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.6 TRANSFER PRINTING

87

often slower by several orders of magnitude [9]. In analogy to the arguments of Section 4.3 and equation 4.7, comparison of the critical energy release rate determines whether retrieval or printing will be preferred. For retrieval, object

stamp

object substrate < G CRIT G CRIT

(4.15a)

while for printing, object

stamp

object substrate G CRIT > G CRIT

(4.15b) stamp

object Critical peel velocities can be calculated via equation 4.10 by substituting in G CRIT [12] or by graphical methods illustrated in Figure 4.2. Additionally, cooling the system can facilitate easier retrieval while heating aids in printing. Such dynamic variability in the transfer mechanism illustrates that kinetic techniques can be used in cooperation with adhesive layers or on surfaces in which chemical treatment would prove impractical. The inherently additive nature of transfer printing and its ability to print on a variety of substrates [54] makes it attractive for the patterning of inorganic semiconductor components having special form factors [45]. Though not limited to inorganics in general, such components are appealing since they can be fabricated prior to transfer using traditional microfabrication techniques [31, 34, 45, 48] (lithography, etching, etc.) that are incompatible with exotic device substrates such as glass or plastic [38, 46, 47]. Additionally, preprocessing allows for control over the organization and layout of components on the donor wafer, which, in conjunction with stamp design, can determine specific device geometries to be printed on a final substrate [54]. These features facilitate the integration of a diversity of material elements [31, 34, 35, 43, 46] and structures into heterogeneous systems on a single substrate [41, 47, 54–59]. Examples of some printable forms of inorganic semiconductors are provided in Figure 4.10 for the case of nanowires made of GaAs [45, 56] and Si [46, 54] and for carbon nanotubes [57, 58]. In the case of GaAs, Figures 4.10a–c, preparation of the donor substrate includes patterning SiO2 stripes as etch masks on a source wafer. Subsequent anisotropic etching results in a reverse mesa shape such as that shown in the SEM image of Figure 4.10b [45]. Complete etching releases the GaAs nanowires at the ends and along the mesa apex to produce the free bundles shown in Figure 4.10c. Conceptually similar processing approaches with Si wafers result in the nanoribbons of Figure 4.10d [35, 47]. Other types of semiconducting components can also be used in transfer printing. As an example, Figure 4.10e shows a dense network of single wall carbon nanotubes (SWCNT) grown via chemical vapor deposition. Selective transfer of these multiple different inks, as depicted in Figure 4.10f , provides a deterministic mechanism for assembling dissimilar classes of materials into single, integrated systems. Detailed examples and applications involving printed inorganic (ribbons, wires) and organic (SWCNT) structures of this type appear in Chapters 16 and 17, respectively.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

88

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

SiO2 mask (a) Preparation of donor substrate

(b)

GaAs wafer Wet etching

500 nm

(c)

(d)

60 µm

60 µm Source wafer

(e)

(f )

Stamp Print

Device substrate Figure 4.10. Printable components for heterogeneous devices. (a) Generation of micro/ nanowires using established lithography and etching techniques. (b) Image of single GaAs wire with inverse mesa geometry. (c ) Released, flexible GaAs microwires. (d ) Si ribbons released under similar processing. (Reprinted with permission from [44]. Copyright 2005, Wiley-VCH Verlag GmbH & Co.) (e) Dense arrays of SWCNT. (f ) The printing technique to transfer the different components. (Reprinted with permission from [54]. Copyright 2006, AAAS.)

The conformable nature of PDMS stamps and the nondestructive contacts that can be established even with fragile nanomaterials facilitates printing on a wide variety of target substrates [41] and the fabrication of multilayer, quasi-three-dimensional structures [9, 33, 60]. Complex, stacked structures can be generated in a straightforward manner by taking advantage of the additive nature of repetitive printing [33, 60]. Figure 4.11 illustrates several such constructions, including multilayer assemblies that exhibit device-scale sub-micrometer pattern features. Figure 4.11a shows n-Si chiplets (100 µm × 100 µm × 2.5 µm) capable of supporting complex

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

4.6 TRANSFER PRINTING

89

(a)

(b)

(c)

Figure 4.11. Examples of transfer-printed objects including multilayer structures. (a) Single crystal n-Si chiplets printed onto p-Si without adhesives. (b) Crossed ribbons of Si printed on (1 0 0) Si. (Reprinted with permission from [9]. Copyright 2006, Nature Publishing Group.) (c ) Crosssection image of 10 consecutively printed layers of 100-nm gold channels. At each print, stamps were rotated 90◦ with respect to the direction of the channels of the underlying layer. The first layer adheres to a GaAs substrate through chemical treatment while cold welding bonds the subsequent Au layers to each other. (Reprinted with permission from [33]. Copyright 2003, American Chemical Society.)

circuitry, printed onto a p-Si wafer using the kinetically controlled approach [9]. Figures 4.11b and c illustrate multilevel stacks of components such as Si ribbons (b) [60] and contoured Au stripes (c) [33] with each level rotated 90◦ with respect to the underlying layer. Fabrication of these structures depends critically on the soft van der Waals contact of the PDMS stamps, the structural integrity of the printed layers, and

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

90

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

3 (a)

IDS (mA)

2

1

0 0

1

2

3

VDS (V) (b)

(c)

Figure 4.12. Printed structures on flexible and nonplanar substrates. (a) Arrays of printed Si TFTs on a thin polyimide sheet with the inset showing an optical image of the top view of the device. Typical I–V characteristics of a single device are shown in the side panel. (Reprinted with permission from [47]. Copyright 2006, IEEE.) Si structures can also be printed onto a cylindrical lens (b) and spherical polycarbonate lens (c ). (Reprinted with permission from [9]. Copyright 2006, Nature Publishing Group.)

the relatively strong bonds formed between the layers and with the target substrate as discussed earlier in this section. Transfer printing can also deliver objects onto nonplanar [9] and mechanically flexible substrates [34, 35, 43, 46, 47, 54, 55, 59]. The latter capability is important for flexible inorganic electronics, as discussed in Chapter 17 and as illustrated in Figure 4.12. Figure 4.12a shows an array of Si thinfilm transistors (TFTs) printed onto a 25-µm-thick polyimide sheet. Typical device current–voltage characteristics are plotted in the side panel [47]. Figures 4.12b and c provide examples of nonplanar printing onto cylindrical and spherical lenses, respectively. Printing onto curved surfaces such as these is accomplished by inking a mold in the usual fashion and then rolling or pressing the curved surface over the stamp to transfer the ink [9]. Such nonplanar microfabrication techniques have potential applications in optics, light detection systems, and photonics. The patterning and printing methods developed in this chapter are versatile techniques at the heart of soft lithography. Whether patterning a polymeric or metallic

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

REFERENCES

91

film as part of a processing sequence or assembling device components on nonnative substrates, the variety of printing mechanisms presented here offer advantages over traditional lithography. These techniques give rise to novel fabrication methods and allow the realization of unusual classes of materials with unique properties. Such features suggest wide-ranging areas of application, including plastic electronics, inexpensive organic/inorganic devices, unusual optoelectronic systems, and others.

REFERENCES 1. Kumar, A., Biebuyck, H. A., and Whiteside, G. M. (1994) Patterning self-assembled monolayers—applications in materials science. Langmuir 10, 1498–1511. 2. Choi, S., Yoo, P. J., Baek, S. J., Kim, T. W., and Lee, H. H. (2004) An ultra-violet curable mold for sub-100 nm lithography. J. Am. Chem. Soc. 126, 7744–7745. 3. Wu, S. (1982) Polymer Interface and Adhesion, Marcel Dekker, New York, pp. 96–104. 4. Wang, Z., Zhang, J., Xing, R., Yuan, J., Yan, D., and Han, Y. (2003) Micropatterning of organic semiconductor microcrystalline materials and OFET fabrication by “hot lift off”. J. Am. Chem. Soc. 125, 15278–15279. 5. Kim, S., Burrows, P. E., and Forrest, S. R. (2000) Micropatterning of organic electronic devices by cold-welding. Science 288, 831–833. 6. Taylor, P. A., Nelson, J. S., and Dodson, B. W. (1991) Adhesion between atomically flat metallic surfaces. Phys. Rev. B 44, 5834–5841. 7. Barber, J. R. (1966) Elasticity, Kluwer, Boston. 8. Richard, P. W. (1997) Polymer Interfaces: Structure and Strength, Hausser, New York. 9. Meitl, M. A., Zhu, Z.-T., Kumar, V., Lee, K. J., Feng, X., Huang, Y. Y., Adesida, I., Nuzzo, R. G., et al. (2006) Transfer printing by kinetic control of adhesion to an elastomeric stamp. Nat. Mater. 5, 33–38. 10. Feng, X., Meitl, M. A., Bowen, A., Huang, Y., Nuzzo, R. G., and Rogers, J. A. (2007) Competing fracture in kinetically controlled transfer printing. Langmuir 23, 12555– 12560. 11. Anderson, T. L. (1995) Fracture Mechanics: Fundamentals and Applications, 2nd edn., CRC Press, Boca Raton, FL. 12. Tsai, K. H. and Kim, K. S. (1993) Stick-slip in the thin film peel test-I. The 90◦ peel test. Int. J. Solids Struct. 30, 1789–1806. 13. Gent, A. N. (1996) Adhesion and strength of viscoelastic solids. Is there a relationship between adhesion and bulk properties? Langmuir 12, 4492–4496. 14. Gent, A. N. and Lai, S. M. (1994) Interfacial bonding, energy dissipation, and adhesion. J. Polym. Sci. Polym. Phys. 32, 1543–1555. 15. Wilbur, J. L., Kumar, A., Kim, E., and Whitesides, G. M. (1994) Microfabrication by microcontact printing of self-assembled monolayers. Adv. Mater. 6, 600–604. 16. Rhee, J. and Lee, H. H. (2002) Patterning organic light-emitting diodes by cathode transfer. Appl. Phys. Lett. 81, 4165–4167. 17. Loo, Y.-L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Additive, nanoscale patterning of metal films with a stamp and a surface chemistry mediated transfer process: applications in plastic electronics. Appl. Phys. Lett. 81, 562–564.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

92

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

18. Huang, X. D., Bao, L.-R., Cheng, X., Guo, L. J., Pang, S. W., and Yee, A. F. (2002) Reversal imprinting by transferring polymer from mold to substrate. J. Vac. Sci. Technol. B 20, 2872–2876. 19. Hur, S.-L., Khang, D.-Y., Kocabas, C., and Rogers, J. A. (2004) Nanotransfer printing by use of noncovalent surface forces: application to thin-film transistors that use singlewalled carbon nanotube networks and semiconducting polymers. Appl. Phys. Lett. 85, 5730–5732. 20. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. 21. Khang, D.-Y., Kang, H., Kim, T.-I., and Lee, H. H. (2004) Low-pressure nanoimprint lithography. Nano Lett. 4, 633–637. 22. Ugural, A. C. (1999) Stresses in Plates and Shells, 2nd edn., McGraw-Hill, Boston. 23. Suh, D., Choi, S.-J., and Lee, H. H. (2005) Rigiflex lithography for nanostructure transfer. Adv. Mater. 17, 1554–1560. 24. Tan, H., Gilbertson, A., and Chou, S. Y. (1998) Roller nanoimprint lithography. J. Vac. Sci. Technol. B 16, 3926–3928. 25. Choi, J., Kim, K.-H., Choi, S.-J., and Lee, H. H. (2006) Whole device printing for full color displays with organic light emitting diodes. Nanotechnology 17, 2246–2249. 26. Seo, S., Park, J., and Lee, H. H. (2005) Micropatterning of metal substrate by adhesive force lithography. Appl. Phys. Lett. 86, 133114. 27. Granlund, T., Nyberg, T., Roman, L. S., Svensson, M., and Ingan¨as, O. (2000) Patterning of polymer light emitting diodes with soft lithography. Adv. Mater. 12, 269–273. 28. Krupenkin, T. N. and Fredrickson, G. H. (1999) Crazing in two and three dimensions. 1. Two-dimensional crazing. Macromolecules 32, 5024–5035. 29. Choi, J., Kim, D., Yoo, P. J., and Lee, H. H. (2005) Simple detachment patterning of organic layers and its application to organic light-emitting diodes. Adv. Mater. 17, 166–171. 30. Kim, J. K., Park, J. W., Yang, H., Choi, M., Choi, J., and Suh, K. Y. (2006) Low pressure detachment nanolithography. Nanotechnology 17, 940–946. 31. Sun, Y. G. and Rogers, J. A. (2004) Fabricating semiconductor nano/microwires and transfer printing ordered arrays of them onto plastic substrates. Nano Lett. 4, 1953–1959. 32. Loo, Y.-L., Lang, D. V., Rogers, J. A., and Hsu, J. W. P. (2003) Electrical contacts to molecular layers by nanotransfer printing. Nano Lett. 3, 913–991. 33. Zaumseil, J., Meitl, M. A., Hsu, J. W. P., Acharya, B., Baldwin, K. W., Loo, Y.-L., and Rogers, J. A. (2003) Three-dimensional and multilayer nanostructures formed by nanotransfer printing. Nano Lett. 3, 1223–1227. 34. Sun, Y. G., Kim, H.-S., Menard, E., Kim, S., Adesida, I., and Rogers, J. A. (2006) Printed arrays of aligned GaAs wires for flexible transistors, diodes, and circuits on plastic substrates. Small 2, 1330–1334. 35. Menard, E., Lee, K. J., Khang, D.-Y., Nuzzo, R. G., and Rogers, J. A. (2004) A printable form of silicon for high performance thin film transistors on plastic substrates. Appl. Phys. Lett. 84, 5398–5400. 36. Meitl, M. A., Zhou, Y., Gaur, A., Jeon, S., Usrey, M. L., Strano, M. S., and Rogers, J. A. (2004) Solution casting and transfer printing single-walled carbon nanotube films. Nano Lett. 4, 1643–1647.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

13:58

Printer Name: Yet to Come

REFERENCES

93

37. Loo, Y. L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Interfacial chemistries for nanoscale transfer printing. J. Am. Chem. Soc. 124, 7654–7655. 38. Menard, E., Park, J., Jeon, S., Shir, D., Nam, Y., Meitl, M., and Rogers, J. A. (2007) Micro and nanopatterning techniques for organic electronic and optoelectronic systems. Chem. Rev. 107, 1117–1160. 39. Xia, Y. N. and Whitesides, G. M. (1998) Soft lithography. Annu. Rev. Mater. Sci. 28, 153–184. 40. Hua, F., Sun, Y., Gaur, A., Meitl, M. A., Bihaut, L., Rotkina, L., Wang, J., Geil, P., et al. (2004) Polymer imprint lithography with molecular-scale resolution. Nano Lett. 4, 2467–2471. 41. Lee, K. J., Motala, M. J., Meitl, M. A., Childs, W. R., Menard, E., Shim, A. K., Rogers, J. A., and Nuzzo, R. G. (2005) Large area, selective transfer of microstructured Si (µs-Si): a printing-based approach to high performance thin film transistors supported on flexible substrates. Adv. Mater. 17, 2332. 42. Loo, Y. L., Hsu, J. W. P., Willett, R. L., Baldwin, K. W., West, K. W., and Rogers, J. A. (2002) High-resolution transfer printing on GaAs surfaces using alkane dithiol monolayers. J. Vac. Sci. Technol. B 20, 2853–2856. 43. Menard, E., Nuzzo, R. G., and Rogers, J. A. (2005) Bendable single crystal silicon thin film transistors formed by printing on plastic substrates. Appl. Phys. Lett. 86, 093507-1–093507-3. 44. Ko, H. C., Baca, A., and Rogers, J. A. (2006) Bulk quantities of single-crystal silicon micro/nanoribbons generated from bulk wafers. Nano Lett. 6, 2318–2324. 45. Sun, Y. G., Khang, D.-Y., Hurley, K., Nuzzo, R. G., and Rogers, J. A. (2005) Photolithographic route to the fabrication of micro/nanowires of III-V semiconductors. Adv. Funct. Mat. 15, 30–40. 46. Mack, S., Meitl, M. A., Baca, A. J., Zhu, Z. T., and Rogers, J. A. (2006) Mechanically flexible thin-film transistors that use ultrathin ribbons of silicon derived from bulk wafers. Appl. Phys. Lett. 88, 213101. 47. Ahn, J. H., Kim, H.-S., Lee, K. J., Zhu, Z., Menard, E., Nuzzo, R. G., and Rogers, J. A. (2006) High speed, mechanically flexible single-crystal silicon thin-film transistors on plastic substrates. IEEE Electron Device Lett. 27, 460–462. 48. Schuelller, O. J. A., Duffy, D., Rogers, J. A., Brittain, S. T., and Whitesides, G. M. (1998) Reconfigurable diffraction gratings based on elastomeric microfluidic devices. Sensors Actuators A 78, 149–159. 49. Menard, E., Bilhaut, L., Zaumseil, J., and Rogers, J. A. (2004) Improved surface chemistries, thin film deposition techniques, and stamp designs for nanotransfer printing. Langmuir 20, 6871–6878. 50. Hsu, J. W. P., Loo, Y.-L., Lang, D. V., and Rogers, J. A. (2003) Nature of electrical contacts in a metal-molecule-semiconductor system. J. Vac. Sci. Technol. B 21, 1928– 1935. 51. Ferguson, G. S., Chaudhury, M. K., Sigal, G., and Whitesides, G. M. (1991) Contact adhesion of thin gold-films on elastomeric supports—cold welding under ambient conditions. Science 253, 776–778. 52. Hsia, K. J., Huang, Y., Menard, E., Park, J.-U., Zhou, W., Rogers, J. A., and Fulton, J. M. (2005) Collapse of stamps for soft lithography due to interfacial adhesion. Appl. Phys. Lett., 86, 154106-1–154106-3.

P1: OTA/XYZ P2: ABC c04 JWBK252/Lee September 18, 2008

94

13:58

Printer Name: Yet to Come

PATTERNING BASED ON WORK OF ADHESION

53. Huang, Y. G. Y., Zhou, W., Hisa, K. J., Menard, E., Park, J.-U., Rogers, J. A., and Alleyne, A. G. (2005) Stamp collapse in soft lithography. Langmuir 21, 8058–8068. 54. Ahn, J. H., Kim, H.-S., Lee, K. J., Jeon, S., Kang, S. J., Sun, Y. G., Nuzzo, R. G., and Rogers, J. A. (2006) Heterogeneous three dimensional electronics using printed semiconductor nanomaterials. Science 314, 1754–1757. 55. Ahn, J.-H., Kim, H.-S., Menard, E., Lee, K. J., Zhu, Z., Kim, D.-H., Nuzzo, R. G., Rogers, J. A., et al. (2007) Bendable integrated circuits on plastic substrates by use of printed ribbons of single-crystalline silicon. Appl. Phys. Lett. 90, 213501. 56. Lee, K. J., Meitl, M. A., Ahn, J.-H., Rogers, J. A., Nuzzo, R. G., Kumar, V., and Adesida, I. (2006) Bendable GaN high electron mobility transistors on plastic substrates. J. Appl. Phys. 100, 124507. 57. Hur, S.-H., Park, O. O., and Rogers, J. A. (2005) Extreme bendability in single walled carbon nanotube networks transferred from high temperature substrates to plastic and their us in thin film transistors. Appl. Phys. Lett. 86, 243502. 58. Cao, Q., Xia, M.-G., Shim, M., and Rogers, J. A. (2006) Bilayer organic/inorganic gate dielectrics for high performance, low-voltage single walled carbon nanotube thin-film transistors, complementary logic gates and p–n diodes on plastic substrates. Adv. Funct. Mater. 16, 2355–2362. 59. Sun, Y., Menard, E., Rogers, J. A., Kim, H.-S., Kim, S., Chen, G., Adesida, I., Dettmer, R., et al. (2006) Gigahertz operation in mechanically flexible transistors on plastic substrates. Appl. Phys. Lett. 88, 183509. 60. Jeon, S., Menard, E., Park, S., Maria, L., Meitl, M., Zaumseil, J., and Rogers, J. A. (2004) Three dimensional nanofabrication with rubber stamps and conformable photomasks. Adv. Mater. 16, 1369–1373.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5 PATTERNING BASED ON LIGHT: OPTICAL SOFT LITHOGRAPHY

5.1 INTRODUCTION Capabilities for generating two-dimensional (2D) or three-dimensional (3D) nanostructures are extremely important for many areas of research. As argued in Chapter 1, manufacturing approaches that offer such capabilities are, in fact, critical to efforts to convert nanoscience into nanotechnology. Among various techniques, optical lithography (i.e., photolithography, in contact, proximity, or projection modes) is the most widely used, due primarily to its implementation in semiconductor device manufacturing. Conventional photolithography systems can define patterns with high resolution (tens of nanometers) and with high precision, throughput, and yield. These tools are, however, very expensive and generally incapable of patterning nonflat surfaces or complex 3D structures. Optical soft lithography refers collectively to a class of optical patterning methods that use the soft, elastomeric stamps and molds of soft lithography, or variants of these kinds of elements, as “conformal” photomasks. These masks provide experimentally simple and nondestructive ways to perform a type of “intimate contact” mode photolithography that provides fast, low cost, and large area patterning capabilities, not limited to ultraflat surfaces or planar geometries. The following sections present overviews of these methods, many of which can be used for either 2D or 3D patterning. Diverse classes of nanostructures demonstrate some of the patterning capabilities, with application examples in microelectronics, microfluidics, photonics, high energy density science, chemical release, and others.

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


95

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

96

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

5.2 SYSTEM ELEMENTS 5.2.1 Overview Optical soft lithography involves three main elements: (i) a light source for photoexposure, (ii) a soft, elastomeric photomask, and (iii) a photosensitive material. The process begins with a layer of photosensitive material cast onto a substrate. This layer is typically a solid, although liquids can be used with appropriate modification to the exposure geometry and/or photomask. Placing the elastomeric mask against the photosensitive layer leads to atomic scale, conformal contact due to the action of generalized adhesion forces [1–3] without application of pressure. When light passes through the photomask, an intensity distribution forms in the photosensitive material that patterns its exposure. This distribution extends from the near-surface region of the mask (i.e., within a wavelength; the near-field region) to distances of many wavelengths (i.e., the region in proximity to the surface of the mask, which we refer to loosely as the “proximity field” region) and in geometries that can be useful for patterning up to several millimeters away, limited only by the size of the mask and the optics of the exposure source. The near-field region can be exploited to generate nanoscale 2D features in thin photosensitive layers. The proximity field provides a route to creating 3D structures in a single exposure step in thicker layers. Removing the mask following this exposure and then developing away the regions of photosensitive material that were (or were not, depending on the chemistry) exposed completes the fabrication. Structures formed in this fashion can be used directly in devices or they can template the deposition or growth of other materials to create replicas. Figure 5.1 illustrates various types of masks that have been used in optical soft lithography, ranging from purely phase modulating masks (Figure 5.1a), to setups in which molded relief structures in the photosensitive material provide the phase shifting (Figure 5.1b), to reflective masks (Figure 5.1c), to masks that incorporate phase modulating and absorbing/reflecting solid films (Figure 5.1d) or inks (Figure 5.1e). Figure 5.2 outlines the process flow in the case of 3D optical soft lithography, in which a mask of the type of Figure 5.1a is used.

5.2.2 Elastomeric Photomasks The soft, elastomeric photomasks are central to optical soft lithography. These masks can be formed easily by casting and curing liquid prepolymers against structures of relief, known as “masters,” using the well-established methods of soft lithography [5, 6], as described in Chapter 2. This process can be repeated many times with a single “master” to create many high quality photomasks, each of which can be used multiple times. The “masters” can take many forms, but they most commonly consist of patterns of resist or etched structures defined by conventional lithographic methods. The masks are typically made from commercially available elastomers based on poly(dimethylsiloxanes) (PDMS; Sylgard 184, Dow Corning), [5–8] or perfluoropolyethers (PFPE; CN4000, Sartomer Company, Inc.) [9, 10]. These materials,

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.2 SYSTEM ELEMENTS

(a)

97

(b) UV light

UV light

Phase mask Photoresist Photoresist

Si wafer

Si wafer

(d)

(c)

UV light

UV light Photomask

Photomask

Attenuator Photoresist Photoresist Si wafer

(e)

Si wafer

UV light Photomask UV ink Photoresist Si wafer

Figure 5.1. Schematic illustration of the use of various kinds of photomasks for optical soft lithography. (a) Elastomeric phase mask. (b) Embossed relief as a phase mask. (c ) Elastomeric reflective mask. (d ) Elastomeric phase mask with optically attenuating films in the recessed regions. (e) Elastomeric phase mask with absorbing UV ink in the recessed regions.

which are described in detail in Chapter 2, are useful due to their optical transparency down to wavelengths of ∼250–300 nm and their ability to replicate accurately features of relief in the “masters” with dimensions down to ∼1 nm [11–13]. Also, their moderate to low Young’s moduli (i.e., between 1 and 10 MPa) [5–10] enables soft, conformal contact of the mask with the photosensitive layer and, therefore, repeatable alignment of the optical element with respect to the photopolymer surface to nanometer precision. This type of reversible and nondestructive physical contact provides (i) reliable, gap-free optical coupling between the mask and photosensitive layer, thereby enabling the regions near the surface of the mask to be exploited for optical patterning, (ii) simple setups in which position control systems for the outof-plane direction and optical elements (i.e., imaging lenses, etc.) are not required, (iii) an insensitivity to mechanical vibrations during exposure, due to the conformal

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

98 (a)

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(b)

Rubber phase mask

Initiate conformal contact

Thick photopolymer 5–15 µm Glass substrate (c)

(d ) Expose UV or visible light Remove mask, develop

3D nanostructure

Figure 5.2. Schematic illustration of a 3D optical soft lithography method, referred to as proximity field nanopatterning (PnP). (a) Placement of an elastomeric phase mask (SEM image in the inset) on a transparent, photosensitive film (thick photopolymer). (b) Mask and film come into conformal contact (optical image in the inset). (c ) Passing light through the mask while it is in contact with the film generates a 3D intensity distribution throughout the film (calculated intensity in the inset). (d ) Removing the unexposed (or exposed, depending on the chemistry) regions yields 3D nanostructures with geometries defined by the intensity distributions (SEM image in the inset). (Reprinted from [4]. Copyright 2004, National Academy of Sciences, USA.)

contact, (iv) relaxed requirements on the coherence of the exposure source due to the proximity geometry, thereby enabling patterning even with incoherent light from a lamp, and (v) the ability to exploit single or multiphoton effects, using the same setups with suitable light sources. These elements can be used directly as phase masks, or they can be integrated with other materials to yield more complex designs, as shown in Figure 5.1. For example, the addition of metal layers to the recessed region of a binary phase mask (Figure 5.1d) introduces amplitude-modulating elements. In this case, light is blocked in the recessed regions such that the pattern in the photosensitive material replicates in a one-to-one fashion the geometry of relief (i.e., metal) on the mask. Alternatively, absorbing liquid inks can be used to fill the recessed regions to yield similar results (Figure 5.1e). The inks, unlike the metal layers, can move with the mask without cracking, thereby retaining the mechanical toughness and full conformal nature of the mask. The liquids also provide the opportunity for gray-scale amplitude

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.2 SYSTEM ELEMENTS

99

modulation and tuning through microfluidic pumping mechanisms, as described in a subsequent section. Alternatively, the masks can be used as molds to create surface relief on the photosensitive layer (Figure 5.1b). This relief can provide the phase and/or amplitude modulation to pattern the exposure light, according to optical effects that are similar to those generated by a separate mask. The masks of Figure 5.1 are equally powerful for 3D optical soft lithography, and several can be used with short-pulsed lasers for multiphoton exposure effects, using the same basic setups. 5.2.3 Photosensitive Materials The photosensitive materials capture, in the form of 3D solid structures or 2D patterned films, the optical intensity distributions near the surface of the photomasks. The photosensitivity, the contrast, the index of refraction, the diffusion of photocatalysts, and other properties associated with these materials affect the resolution. The mechanical and chemical properties determine their ability to serve as resists in etching processes, as well as the robustness and strength of the resulting 2D and 3D structures. Other properties such as absorption length, wavelength sensitivity, and two-photon or three-photon absorption cross sections all can be exploited to achieve various capabilities, as described in subsequent sections. Two-dimensional optical soft lithography can be used with a wide variety of photosensitive materials, most commonly the types of positive and negative tone photoresists that are already well developed for conventional photolithography. Positive tone photoresist interacts with light to become soluble to photoresist developers, yielding the structures that have geometries defined by the positions of low light intensities. Negative tone resists produce geometries with the opposite tone, and most commonly use epoxy or novolac-based resin. Chemical amplification is generally employed to enhance the quantum efficiency of the photoresist. Chemically amplified negative tone resist was first developed in a three-component system including novolac as resin, photoacid generator, and melamine cross-linker. In this system, the photoacid generator produces acid upon light exposure and the acid cross-links the phenolic resin through acid-catalyzed condensation [14]. The crosslinked photopolymer becomes insoluble while the unexposed part remains soluble, thus allowing development of a patterned structure with the use of an appropriate solvent. Positive photoresist generally offers higher resolution capabilities and thus is often exploited for nanoscale patterning. Most of the 2D structures reported in this article use commercial positive photosensitive materials. Some popular choices include AZ6612 (Hoechst), AZ 5200 series (Hoechst), or Shipley 1805. These materials consist of a base resin, typically novolac, and photoactive component (PAC) such as diazoquinone [15]. The PAC is usually sensitive to ultraviolet (UV) or deep UV light. Upon light exposure, the PAC undergoes a Wolff rearrangement that generates an acid that makes the photosensitive materials locally soluble in a basic developer. The unexposed regions, by contrast, remain insoluble. The development process therefore produces a pattern in which the exposed regions are removed. Image reversal resist has the ability to serve in both positive and negative modes. Such resist contains a PAC, novolac and/or

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

100

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

polyvinyl phenol resin, and a cross-linking compound. Upon light exposure, the image reversal resist acts just like a normal positive resist. However, the cross-linking compound in the image reversal resist provides the ability to initiate cross-linking in the presence of photoacid and thermal treatment. Baking after exposure initiates the cross-linking, thereby making the exposed region insoluble. A second exposure converts the uncross-linked regions into a soluble form that can be removed with proper solvents. In this way, the image reversal resist can act as a negative resist [16]. In 3D photo patterning, substantial optical transparency in the photosensitive material is typically required. In most reports, an epoxy-based photosensitive material (SU-8, MicroChem, Inc.) has been used, due partly by its commercial availability and its good properties [17, 18]. SU-8 is a negative tone, cationic photoresist that offers high sensitivity, high resolution, and good transparency from the near infrared (IR) to the visible and the near-UV range. SU-8 also has good structural properties and levels of two-photon sensitivity [18–20] that are sufficiently high to allow patterning with common femtosecond-pulsed laser sources. The material consists R SU-8 resin (Shell Chemical), a photosensitizer (propylene carbonate), of EPON
and a photoacid generator (onium salt) in a solvent (γ -butyrolactone). In the case of epoxy-based negative tone photoresists, the photochemical process involves optical absorption by the photosensitizer and transfer of an electron to the photoacid generator [18, 21, 22]. This acid promotes a thermally initiated cationic chain reaction that opens one of eight ring structures in the epoxy groups of the monomer for cross-linking. Due to absorption and sensitivity of the photosensitizer, the range of exposure wavelengths for implementation with one-photon effects is between 400 and 350 nm. The amplification provided by the chain reaction enables sufficiently low concentrations of acid that the optical properties are unaffected, thereby causing negligible impact on the 3D patterning process. Other photosensitive materials such as poly(methyl silsesquioxanes) (PMSSQs) [23, 24] and certain chalcogenides [25, 26] can also be used in 3D or 2D optical soft lithography. PMSSQ is attractive in part because its high thermal stability allows high temperature processing; thereby enabling its use as a template, for example, in deposition processes at up to ∼500◦ C [23]. Chalcogenide glasses are interesting, in part, because their high refractive index (2.35–3.5) and high densities (3.2–6.35 g cm–3 , depending on materials composition) enable certain optical and mechanical applications, as discussed in following sections. 5.3 TWO-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY (OSL) 5.3.1 Two-Dimensional OSL with Phase Masks The first reports of OSL involved binary elastomeric phase masks to pattern, via phase-shifting effects, lines, dots, and related structures as small as ∼50 nm in conventional photoresists [27–29]. The method involves bringing a PDMS phase mask into conformal contact with a thin layer of photoresist. Passing light from a mercury lamp through the mask exposes the resist to the intensity distribution that forms at the mask surface. Figure 5.3 shows a schematic illustration of a PDMS

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.3 TWO-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY (OSL)

101

UV Light Phase mask

Phtoresist

Develop

100 nm

Figure 5.3. Schematic illustration of the exposure and development process for optical soft lithography with a phase mask. A step edge of relief with a depth suitable to modulate the phase of light by π leads to local minima in intensity with widths of ∼100 nm for 365-nm exposure light. Removing the exposed parts of a positive photoresist yields a line with similar width. (Reprinted from [29]. Copyright 2004, with permission from Elsevier.)

phase mask with a depth that modulates the phase of the transmitted light by π . Local minima in the transmitted intensity appear at the step edges of relief in the mask. After development, structures of lines (trenches) of positive (negative or image reversal) photoresist can be produced from these intensity distributions. Figure 5.4 shows scanning electron micrographs (SEMs) of structures in photoresist formed by this technique. Figure 5.4a shows lines of photoresist that form at the step edges of the phase mask (the relief structure of the mask is shown on the top). Lines with curved shapes are also possible. Figure 5.4b shows arrays of 100-nm-diameter posts obtained by two exposures with a 90◦ rotation of the mask in between. This method can also generate ∼100-nm features on curved substrates. In this case, exposure is performed through a thin phase mask draped over the surface of a curved object coated with photoresist. Figure 5.4c shows lines of photoresist formed on a cylindrical lens with a 15-cm radius of curvature. The details of the optics associated with this process can be revealed through near-field optical measurements of light passing through phase masks. Figure 5.5 shows representative results collected from masks with periodicities of 600 nm (Figure 5.5a) and 10 µm (Figure 5.5b) [30]. Finite element modeling (FEM) computations appear in the left insets. Figure 5.5c and d show measured and simulated linecuts corresponding to the depth locations illustrated by lines in the top frames. The modeling results compare well with the experimental results. For periodicities larger than the optical wavelength (Figure 5.5b), the computations exhibit “nulls” in intensity appearing at the step edges of relief in the phase mask, whose centers are slightly shifted toward the recessed regions of the mask. They also capture the

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

102

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(a)

(c)

1 cm 100 µm (b)

(d)

100 nm

100 µm

5 µm

Figure 5.4. Scanning electron micrographs (SEMs) of structures of photoresist formed using phase shift masks in optical soft lithography. (a) Lines ∼50 nm wide. (b) Arrays of dots with diameter ∼100 nm formed by exposing the resist through the same mask twice with a 90◦ rotation in between exposures. (Reprinted from [29]. Copyright 2004, with permission from Elsevier.) (c ) Photograph of a cylindrical lens patterned with lines of photoresist defined with a thin elastomeric phase mask. (d ) Uniform lines with widths ∼100 nm. Dark and light regions correspond to glass and photoresist, respectively. (Reprinted with permission from [27]. Copyright 1997, American Institute of Physics.)

(a)

(b)

10 µm

2 µm

1 0.8 0.6 0.4 0.2 0.0 0

(d ) Intensity

Intensity

(c)

0.5

1 1.5 Distance (µm)

2

1 0.8 0.6 0.4 0.2 0.0 0

2

4 Distance (µm)

6

Figure 5.5. (a) and (b) Near-field optical measurements and simulations (regions on the left) of the propagation of laser light (442 nm, TM polarization) through line grating, surface relief phase masks. The mask in part (a) has line widths and spacings of 300 nm. The mask in part (b) has line widths and spacings of 4.4 and 5.6 µm. The tops of the graphs show schematic layouts of the masks. (c ), (d ) Linecuts of the simulations (solid line) and measurements (dashed line), evaluated at positions indicated by the dashed lines in (a) and (b), respectively. (Reprinted with permission from [30]. Copyright 2006, AVS The Science and Technology Society.)

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.3 TWO-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY (OSL)

nm 500

(a)

(b)

103 nm

550

250 2 µm

0

2 µm

0

Figure 5.6. Photoresist structures defined with a grating mask with line widths and spacings of 5.6 and 4.4 µm. (a) AFM image of the structure. (b) Computed profile. (Reprinted with permission from [30]. Copyright 2006, AVS The Science and Technology Society.)

lower amplitude and frequency of intensity oscillations that appear in the recessed compared to the raised regions of relief. For periods comparable to or less than the optical wavelength (Figure 5.5a), the modeling reveals an effective type of subwavelength focusing effect that creates high and low intensity regions near the surface of the mask at the raised and recessed features of relief, respectively. With low exposure doses and short development times, it is possible to produce relief structures in positive or image reversal photoresists that follow the intensity distributions near the surface of the PDMS as illustrated in Figure 5.5 [30, 31]. Figure 5.6 compares these structures to computations that include the PDMS mask in contact with resist on a silicon wafer. The mask in this case has line widths and spacings of 5.6 and 5.4 µm, respectively, and a relief depth of 1.42 µm. The computed relief profiles were obtained by applying a cutoff filter to the intensity distribution evaluated at a depth of 50 nm into the photoresist, which simulates, in a simple way, the exposure and development processes [30]. The results compare well with the experimental observations and reveal even subtle features of the system such as the fact that the positive peaks of intensity are narrower (by a factor of ∼0.4) than the “nulls” (as measured at 0.7 of the maximum; peak width ∼225 nm, null width ∼575 nm). Figure 5.7 shows a range of structures that can be produced with this technique, beyond the lines and dots of Figure 5.4a and b [27, 28], by exploiting other features of the distributions of intensity revealed by the computations. Related results, without the detailed measurements or computations were reported previously [32]. For example, in the regions where the mask contacts the resist, positive peaks in the distribution of intensity appear adjacent to the nulls that occur at the edges of relief. These peaks provide sufficient contrast to allow the formation of trenches in the positive resist. The computed intensity distributions also show that this contrast increases with the relief depth; experimentally this effect was demonstrated by patterning different numbers of trenches by changing the mask relief depth [30]. The left frame of Figure 5.7a shows patterning of one trench in the resist using a phase mask with a relief height of 420 nm and periodicity of 4 µm (line width = 2 µm), while Figure 5.7b shows patterning of two trenches using a phase mask with a relief height of 1.42 µm and periodicity of 10 µm (line width = 5.6 µm). Also illustrated in Figure 5.7 is the capability of the PDMS phase mask to form resist features in exactly the geometry of relief on the mask.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

104

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(a)

2 µm

4 µm

2 µm

(b)

5 µm

5 µm

5 µm

Figure 5.7. Patterned metal structures (bottom frames in (a) and (b)) formed using patterns of photoresist (top frames in (a) and (b)) defined by phase mask optical soft lithography. The examples, as obtained with two different phase masks, with different exposure and development conditions, illustrate the variety of patterns that can be formed. The results of part (a) used a mask with 2-µm lines and spaces, and a relief of 420 nm. The exposure times were 2.5, 4, and 5 s, from left to right; the development time (6 s) was constant. Part (b) used a phase mask with 5.6-µm lines spaced by 4.4 µm and with a relief depth of 1.42 µm. The exposure times were 1.5, 1.5, and 3 s and development times were 5, 20, and 5 s, from left to right, respectively. (Reprinted with permission from [30]. Copyright 2006, AVS The Science and Technology Society.)

5.3.2 Two-Dimensional OSL with Embossed Masks Features of relief molded with a phase mask into a layer of photoresist can itself act as an optical element for patterning the exposure light. These relief features focus/ disperse and phase shift the incident light in the optical near field inside the resist layer. This technique is called topographically directed photolithography (TOP) [33] and it involves three steps: (i) embossing the surface of a layer of photoresist with features that act as lenses and phase shift elements by solvent-assisted embossing

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.3 TWO-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY (OSL)

(a)

(b)

105

(c)

Figure 5.8. SEMs of photoresist structures generated using topographically directed photolithography (TOP). (a) Approximately 50 nm wide features generated using a rectangular relief structure, with line width of 150 nm and periodicity of 800 nm embossed on an ∼200 nm wide layer of photoresist. (b) Approximately 150 nm wide curved features generated using rectangular relief with diamond geometries. The black arrows in (a) and (b) indicate the photoresist–substrate interface. (c ) Features generated using square pyramidal patterns of relief. (Reprinted with permission from [33]. Copyright 1998, American Institute of Physics.)

[5, 34], (ii) exposing this topographically patterned photoresist layer to flood illumination, and (iii) developing the exposed photoresist. In this method, the embossed surface of the photoresist acts as its own optical element. Features ranging in size from 70 to 200 nm can be produced [35] with light of wavelength between 365 and 436 nm. Thin molds can conform to both flat and cylindrically or spherically curved surfaces [36, 37]. Figure 5.8a shows that the steps at the edges of embossed rectangular gratings generate patterns that are similar to those formed by masks of the type in Figure 5.1a [28, 30]. Lines fabricated using rectangular gratings with an embossed periodicity of 800 nm are ∼50 nm in width appear in Figure 5.8a. Figure 5.8b shows embossed patterns of diamonds with flat top surfaces and the resulting photoresist structures. Figure 5.8c shows 3D embossed pyramids and the resulting photoresist structures. The patterns formed in this manner depend not only on the intensity distributions and the optics but also on the kinetics of development and the shape of the initial relief [28]. 5.3.3 Two-Dimensional OSL with Amplitude Masks

5.3.3.1 Transparent Reflective Masks. Reflective masks are similar to the phase masks of the previous section, in the sense that they are formed by the casting and curing processes of soft lithography and they do not incorporate any materials other than the elastomer of the mask. Instead of designs that alter primarily only

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

106

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

the phase of the transmitted light, however, transparent reflective masks modulate the amplitude of the light via large reflections at the interfaces between certain recessed regions of the mask and the surrounding air. In a simple example, PDMS reflective masks can be fabricated from a Si (1 0 0) master whose surface has been patterned with V-shaped trenches or pyramidal pits using anisotropic etching [38]. The sidewalls of the trenches and pits in the silicon master meet with the plateaus in dihedral angles of 54◦ . When normally incident light (350–400 nm) passes through the regions with sloping features, the incident angle is 54◦ at the PDMS–air interface, which exceeds the critical angle of 47◦ (determined by the refractive indices of PDMS (n ≈ 1.43) and air (n ≈ 1.00)) for total internal reflection. This light is therefore totally reflected from the PDMS–air interface and eventually refracted out of the mask. As a result, light does not pass through these regions, thereby leading to dark areas in the corresponding locations in the photoresist. Figure 5.9a shows the use of such a reflective photomask for generating patterns in a positive photoresist. An interesting feature of this type of mask is its mechanical tunability. In particular, under the weight of the stamp, the deformation at the tips of the V-shaped grooves in the mask is not sufficient to allow the light to pass through. On the other hand, upon sufficient vertical pressure applied to the backside of the mask (Figure 5.9a right), the mask deforms and light can pass through regions that were previously reflective. As a result, patterns produced using these elastomeric photomasks can be modified in shape and size by changing the applied pressure [38]. Modifications in the geometry of the masks can also be achieved through the use of masters generated with different etching conditions (e.g., time and temperature) [38]. The profiles and dimensions of grooves into the Si can be controlled, to some extent, in this manner. Figure 5.9b shows two typical examples–-underetching (Figure 5.9b left) and overetching (Figure 5.9b right)—of Si (1 0 0) starting from a master with lines and spaces of 2 µm. The top frames show cross-sectional SEM images of the Si (1 0 0) substrates and the bottom frames show SEM images of the corresponding patterns generated in thin films of photoresist using photolithography. Underetching (Figure 5.9b left) allows the generation of patterns with smaller feature sizes and doubled density compared to the master. Overetching (Figure 5.9b right) enables the reduction of the lateral dimension of the plateau from ∼2 to ∼0.2 µm.

5.3.3.2 Ink Lithography. When phase masks are combined with liquid UVabsorbing inks, they effectively work as fluidic types of masks in which the phase and the amplitude of the transmitted light are modulated according to the relief structures and the optical properties of the inks. In the simplest process, an absorbing ink is first placed (by drop- or spin-casting) on a layer of photoresist on a substrate (Figure 5.10a) [39]. The elastomeric mask is then pushed against this liquid ink layer such that the ink becomes localized to the recessed regions of relief (Figure 5.10b). The intensity of UV light passing through the stamp, the ink, and into the underlying resist is spatially modulated in intensity according to thickness variations in the ink (which have the same geometry as the relief structure in the stamp). Removing the mask, rinsing away the ink, and then developing the photoresist yields patterns of resist (Figure 5.10d) that can then be used in conventional processing sequences to

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.3 TWO-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY (OSL)

(a)

No vertical pressure

hv

hv

PDMS

107

With vertical pressure hv

hv

hv

hv

PDMS

Reflective mask photoresist

Si

Si Develop photoresist

Develop photoresist

Si

Si

(b)

500 nm

1 µm

200 nm

Figure 5.9. (a) Schematic illustration of the use of a reflective mask in optical soft lithography. (b) Top frames show cross-sectional SEMs of patterns of masters used to form the masks. Bottom frames show SEMs of patterns of photoresist on Si/SiO2 fabricated using photolithography with reflective photomasks molded from the masters shown in the top frames. (Reprinted with permission from [38]. Copyright 1998, AVS The Science and Technology Society.)

pattern other materials. The fluidic mask can be implemented in this type of fashion, or it can be formed by sealing the mask against the resist and then filling the channels that form as a result of this contact [40]. In this geometry, pumping different fluids into and out of the channels provides a kind of fluidic tuning of the optical properties that can be useful in various contexts.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

108

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

UV light (a)

PDMS stamp

Ink

PR Si wafer (b)

Stamp floating on ink

Final pattern

(d )

(c) Without UV ink

Stamp pressed against ink

2.3 µm

With UV ink

2.32 µm 1.68 µm

1.7 µm

5 µm

5 µm

1 µm

Figure 5.10. (a) Schematic illustration of the ink lithography method: a conformable phase mask is brought into contact with a photoresist layer with a few drops of ink on its surface; the ink flows to the recessed regions of the mask. Exposing with UV light modulates the distribution intensity in a fashion defined by the optical thickness of the ink present between the phase mask and substrate. (b) Photomicrograph images showing the interaction of the phase mask with an ink: PDMS mask floating on ink (left; no external force is applied to generate conformal contact between mask and photoresist, only parts of the mask-recessed regions are filled with ink and there is excess ink in the nonrecessed regions), PDMS mask pressed onto the ink and photoresist (middle; recessed regions filled with ink and ink squeezed out of the nonrecessed regions), resultant pattern after exposure and development (right; features in the photoresist correspond to regions of the mask filled with ink). SEMs of a photoresist structure patterned using phase shift lithography: (c ) without a UV absorber ink, (d ) with a UV absorber ink.

One advantage of fluidic masks is that the optical density of the fluids can be adjusted over a large range. In this manner, gray-scale functionality can be achieved, in which different parts of a photosensitive material experience different exposure doses over some well-controlled range. Developing the material yields patterns that can have both diverse layouts in the plane of the substrate as well as various heights, associated with the different exposure doses. Figure 5.11 illustrates a type of photoresist pattern that can be achieved with a microfluidic phase mask consisting of five microchannels, each filled with dye at a different concentration. A fluorescence image provides a semiquantitative map of the topographical height of the resist that compares well to the quantitative height measurements obtained with a profilometer

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.3 TWO-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY (OSL)

109

(b)

(a)

20 10 0 0

4

3

2

200 300 100 Distance (µm)

1

250 150 50

Brightness

Height (µm)

5

Top view

Cross-section

Figure 5.11. (a) Top frame shows a microfluidic photomask with relief depth of 38.9 µm filled with different dye concentrations: 5–167 mg cm L–1 , 4–76.6 mg cm L–1 , 3–62.7 mg cm L–1 , 2–41.7 mg cm L–1 , and 1–27.9 mg cm L–1 . Bottom frame shows a fluorescence micrograph of the photoresist pattern (large fluorescence intensities correspond to large photoresist heights). The line shows the photoresist profile at the location indicated by a dashed line between two arrows. (b) Top frame shows a microfluidic device made using the photoresist pattern of part (a) consisting of five channels (filled with the same blue dye solution) of different height (left to right): 5.4, 7.7, 11.1, 13.4, and 18.1 µm. Bottom frame shows a cross-section of the device at the location indicated by the dashed line between two arrows. (Reprinted with permission from [40]. Copyright 2003, National Academy of Sciences, USA.)

(blue traces in graphs). The arrows indicate the direction of filling of the channels. Increasing dye concentrations correspond to arrows of darker tones and higher numbers (0–5). Fabrication of this same pattern by conventional photolithography would have required multiple separate exposures. The microchannels shown in Figure 5.11b are made using PDMS elements formed with the photoresist structures of Figure 5.11a. Cross-sectional micrographs of the channels along the dashed lines are shown. All five channels were filled with the same blue-dye solution, but they appear different due to differences in channel height. 5.3.4 Two-Dimensional OSL with Amplitude/Phase Masks Phase masks like those of Figure 5.1a, but modified with the addition of thin metal films to the flat recessed regions, are sometimes referred to as light-coupling masks (LCMs). These absorbing/reflecting metal features give optical properties that have some similarities to those of masks used in the ink lithography process. Figure 5.1c

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

110

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(b)

Resist

Gap

Mask

(a)

100 nm

Figure 5.12. (a) Simulation of 248-nm light propagating through a 100-nm-wide relief feature on a light-coupling mask (LCM) with 60-nm air gaps. Contours are isointensity lines; the arrows are the time-averaged Poynting vectors. (b) SEM of positive resist (UV5, Shipley) exposed with an LCM having a mixture of small (120 nm, see inset) and large features and using 256-nm light. The relief height on the mask was 100 nm and it did not include absorbers. (Reprinted with permission from [42]. Copyright 1998, AVS The Science and Technology Society.)

illustrates the use of LCMs [41, 42]. When in contact with a layer of resist, the LCMs create regions of strong lateral confinement and amplification of light within the mask (Figure 5.12a). These effects arise because the field prefers to propagate along paths of higher index (i.e., through the mask protrusions) and because the light is blocked by the metal. The result is a high contrast pattern of intensity at the surface of the mask [43] that is particularly pronounced at feature sizes comparable to the effective wavelength in the LCM medium. Figure 5.12b shows photoresist patterns generated by this technique; structures with a resolution of ∼100 nm at a 200-nm pitch can be produced using a wavelength of light of 248 nm [44].

5.4 THREE-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY The 2D patterning capabilities provided by several of the OSL methods described in the previous sections can be extended to 3D geometries through the use of transparent photosensitive materials and appropriate modifications to the experimental setups. The sorts of 3D structures that can be fabricated have the potential for application in microfluidics [45–47], sensors [48], photonics [49–51], electrodes in fuel cells [52], catalyst carriers [53, 54], data storage [55], and many other systems. Approaches that have been used in the past include colloidal self-assembly [53, 56–59], phase separation of polymers [60–62], template-controlled growth [63, 64], self-assembly in fluids [65, 66], holographic-based lithography [67–69], controlled chemical etching [70, 71], direct-write techniques based on multiphoton exposures [55, 72, 73] and filamentary inks [51, 74], and many others. Each is capable of fabricating certain

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.4 THREE-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY

111

classes of 3D structures that would be difficult or impossible to produce otherwise, but none provides a complete solution to the challenge of 3D nanofabrication. Threedimensional OSL provides an alternative approach that can complement these other methods. Here, instead of only intensities next to the surface of the mask, 3D OSL uses distributions through a much larger depth, in the region that is in proximity to the mask. For this reason, the 3D OSL technique is referred to as proximity field nanopatterning (PnP). The following section reviews the optics associated with 3D OSL and some of its patterning capabilities. 5.4.1 Optics The optical effects used in PnP are the same as those that are used in the 2D OSL methods described previously, although the near-field optical aspects play an even less prominent role. In particular, the approach exploits the intensity distributions that develop as a function of distance away from the surface of the mask, as illustrated in Figure 5.5. The nature of these distributions can be understood as manifestations of the well-known Talbot effect, or self-imaging effect [75–77]. Here, an image (i.e., distribution of intensity) that forms at any given distance from the surface of the mask repeats at integer multiples of the Talbot distance Z T , which is given by [4, 77, 78] ZT =

λ 2 p2 ∼ , = 1 − (1 − λ2 / p 2 )0.5 λ

(when λ/ p is small)

(5.1)

where λ is the wavelength of light in the media it propagates and p is the periodicity of the grating. This simple effect, and related phenomena such as the fractional Talbot effect, explain the periodicity and basic features of the variations in intensity with distance from the surface of the mask, illustrated most clearly in Figure 5.5a. The case of Figure 5.5b is less obvious because Z T is much larger than the scan distance in this image. As described previously, FEM techniques can be used to calculate accurately these effects. An alternative view of the optics is that the masks give rise to multiple diffracted beams in the far field. These beams overlap and interfere near the mask to produce the distributions of intensity that are involved in the exposure. For the case of the short period mask of Figure 5.5a, only three diffracted beams are generated; their interference creates a relatively simple pattern. The long period mask of Figure 5.5b, by contrast, generates 45 diffracted beams whose overlap creates a complex distribution with a large number of spatial Fourier components. The computational approach that emerges from this picture involves determining the angles, intensities, phases, and polarization states of beams produced by far field diffraction and then computing the interference patterns that result from overlap of these beams. Such calculations reproduce the observed patterns, consistent with FEM [4, 30, 80]. Disadvantages include a neglect of near-field effects (rarely important for PnP), and a limited ability to compute cases of complex, aperiodic masks.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

112

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

5.4.2 Patterning Results The following sections provide examples of 3D structures obtained via PnP with different phase masks, using one- and two-photon processes, controlled levels of angular and spectral bandwidth for the exposure light, and engineered absorption in the photosensitive materials. We also describe a PnP TOP process that uses surface relief embossed into the photosensitive layer either alone or with a separate phase mask.

5.4.2.1 Periodic Structures by One-Photon PnP. In one-photon PnP, each incident photon has sufficient energy to initiate the chemical processes that expose the photosensitive material. Figure 5.13 shows some structures formed in this manner in SU-8 with various phase masks and 355- and 514-nm laser light and 365 nm light from a mercury lamp [4]. The phase mask defines the dominant spatial Fourier component in the in-plane direction, while the Talbot and related effects determine the variations in out-of-plane direction. As the mask periodicity decreases relative to the optical wavelength, the structures exhibit fewer spatial Fourier components. As an example, Figures 5.13c and d show structures formed with the same phase mask at exposure wavelengths of 355 and 514 nm, respectively. The exposures can be implemented with incoherent light from UV lamps, enabled by the proximity mode geometry. This feature provides a level of experimental simplicity that might be useful for manufacturing or large-scale patterning. Figures 5.13e and f provide

(a)

(d)

(b)

(e)

(c)

(f )

Figure 5.13. SEMs of 3D nanostructures formed by one-photon PnP: (a) and (b) with 355-nm light (tripled Nd:YAG) and a phase mask that consists of square array of posts (dot diameter d = 375 nm, relief depth rd = 420 nm, periodicity p = 566 nm) (side view and top view shown in (a) and (b), respectively; inset shows modeling results); (c ) and (d ) with 355- and 514-nm light (Ar ion) laser light and a phase mask (d = 570 nm, rd = 420 nm, p = 1140 nm), respectively (inset in (d ) shows modeling results); (e) and (f ) with a mercury lamp and same phase mask as the one used in (a) and (b) (inset in (f ) shows modeling results). (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.4 THREE-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY

113

examples using 365-nm light from a mercury lamp. In these and all other cases, modeling agrees well with the observed shapes. Any type of mask geometry can be implemented in this process; this design flexibility represents an important feature of PnP.

5.4.2.2 Periodic Structures by Two-Photon PnP. The same proximity geometry that enables incoherent light from a lamp to be used for exposures also allows for the use of short-pulsed lasers [79, 80]. Such lasers are useful because they can offer sufficiently high powers that two-photon interactions in commercially available photosensitive materials such as SU-8 can be exploited in patterning. For SU-8, in particular, the minimum peak power required for practical use of such a process is ∼1 TW cm–2 at wavelengths of ∼800 nm. The output of an amplified Ti:Sapphire laser (Spectra-Physics, Spitfire Pro) provides a convenient light source for this purpose, because the required peak powers can be obtained with beam diameters of >1 mm. For distances of a few tens of micrometers from the surface of a mask, spectral or temporal walk-off associated with diffraction can be neglected [79, 80]. This type of two-photon lithography offers advantages over the generally slow, serial operation of traditional methods [55, 72]. Compared to one-photon PnP, the two-photon process provides enhanced contrast ratios and fewer numbers of spatial Fourier components in the 3D structures. Two-photon PnP therefore provides access to lattice geometries that are not easily achieved with one-photon interactions; certain of these structures could be useful for photonic bandgap applications [79, 80]. Figure 5.14 illustrates some two-photon patterning results. The high contrast provided by the two-photon interaction can be exploited to form colloidal particles, wires, and other objects [79], as illustrated in Figure 5.14f . These and other particles can be used for fundamental studies of assembly [81] as well as applications in photonics and chemical sensing [48, 82]. For Figure 5.14a, the mask consisted of a square array of circular posts (diameter d = 570 nm, relief depth rd = 510 nm, and periodicity p = 710 nm). The periodicity of this mask is less than the wavelength of the exposure light (∼800 nm). In this regime, the polarization can have pronounced effects on the structure geometries [79]. In particular, structures with some directionality result from the use of linearly polarized light (Figure 5.14d), while circular polarization produces isotropic structures (Figure 5.14e). Polarization can, in this manner, be used to advantage to control the geometries. 5.4.2.3 Density-Graded Structures. The angular and spectral bandwidth of the exposure light can also influence the structure geometries. These properties, as well as absorption, can be used, for example, to control variations in the average pore sizes in the 3D structures through their thicknesses, to provide density gradient structures (DGS) [83, 84]. In a simple case, sequential exposures at angles of 0◦ (i.e., normal to the surface of the mask), −θ and θ lead to changes in pore sizes that are periodic with depth, yielding nonmonotonic variations in density through the thickness [83]. Monotonic DGS can be created by introducing absorption in the photosensitive polymer through the addition of dyes [84]. The absorption causes the average exposure dose to decrease with depth into the material, leading to associated increases in the level of porosity in the developed structures. Figure 5.15 shows an

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

114

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(a)

(b)

(c)

(d)

(e)

(f )

Figure 5.14. SEMs and modeling results for 3D structures patterned with two-photon PnP. (a) Low magnification angled and top (inset) views of a structure formed using a mask with a square array of posts (diameter d = 570 nm, relief depth rd = 510 nm, periodicity p = 710 nm) and exposure light from the amplified output of a Ti:sapphire laser operating at 800 nm. (b), (c ) SEM views of structures made using this laser and a phase mask with a triangular array of posts (diameter d = 1120 nm, relief depth rd = 420 nm, periodicity p = 1500 nm (inset in (c ) presents modeling results). (d ), (e) Images of structures generated with the mask used for structure (a) with circular and linearly (along the [0,1] direction of the mask) polarized light from the Ti:sapphire laser, respectively (insets present modeling results). (f ) Image of ellipsoidal particles and modeling (inset) generated with the mask in (a). (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.4 THREE-DIMENSIONAL OPTICAL SOFT LITHOGRAPHY

(b)

Density (g cm−3)

(a)

1% Br/CH

115

Solid/graded density SU8

1.0

0.5

0.0 −100

−50 Distance to back of reservior (µm)

0

Figure 5.15. Density gradient structure (DGS) generated by PnP using a phase mask (a-PFPE) with relief in the form of a hexagonal array of cylindrical posts (diameter d = 460 nm, relief depth rd = 420 nm, periodicity p = 600 nm). (a) Cross-sectional image of this sample. (b) Xray radiographs showing the effective density as a function of position through the thickness. (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

example of a thick DGS formed in this way, using a backside exposure geometry [84]. Figure 5.15b shows an electron micrograph indicating that the density ranges from 19% to 100% of full density over a thickness of 60 µm distance. The magnitude of the gradient can be adjusted by controlling the strength of absorption.

5.4.2.4 Structures Formed with Quasicrystal Phase Masks. An advanced type of phase mask that has been the topic of recent work consists of relief features in quasicrystalline layouts. One promise of such masks in PnP is that they might enable the formation of full 3D quasicrystals or related structures, which might be useful in photonic bandgap (PBG) applications [85–87]. The fabrication of 3D quasicrystals with dimensions in a range suitable for operation at visible or near-IR wavelengths is extremely challenging. Quasicrystals can be generated by two-photon direct-write [87], but the fabrication process is slow and incompatible with realistic manufacturing. Figure 5.16 presents structures formed by PnP with masks that have Penrose quasicrystal layouts of holes with diameters of 400 nm and average hole-tohole spacings of 800 nm. The pattern generates hundreds of diffracted beams. These beams interfere to form a complex distribution of intensity near the surface of the mask, as illustrated in the near-field scanning optical microscope (NSOM) images (Figures 5.16c–e). Exposing a layer of SU-8 using a TOP, two-photon procedure leads to 3D structures like the one shown in Figure 5.16f . Measurements indicate an absence of a simple unit cell in these systems, consistent with a quasicrystal-like geometry. These results illustrate the feasibility of generating well-defined 3D structures with Penrose quasicrystal masks. Additional study is needed to define the optics associated with the fabrication process, as well as the geometries of the 3D structures and their optical properties. 5.4.2.5 Structures by Molding and PnP. Molded relief in the photosensitive material, as in the TOP method, can be combined with PnP concepts to yield

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

116

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(a)

(c)

(b)

(d)

(e)

(f )

Figure 5.16. Images of structures formed with one- and two-photon PnP using phase masks with Penrose quasicrystal layouts. (a), (b) Penrose quasicrystal “master” and corresponding phase mask, respectively. (c ), (d ) Near-field scanning optical microscope (NSOM) images measured near the surface of the Penrose quasicrystal phase mask in (b) at z (i.e., distance from the surface of the mask) = 0 and 2.4 µm, respectively. (e) NSOM images in the x –z plane at z = 0 µm and z = 1.50 µm, respectively. (f ) Three-dimensional structure generated by two-photon PnP with the phase mask in (b). The surface of the photosensitive material was molded with the mask and the resulting structure provided the phase modulation for exposure. The insets in (f ) show cross-sectional (right) and top view (left) images. (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

structures with geometries that are inaccessible with PnP alone. A combination of coarse (several micrometers) and fine molded features can be used with (or without) PnP masks to create elaborate structures of various types. For example, molded rib waveguides with internal 3D nanostructures, of the sort that might be useful for applications in microfluidics or optical waveguides, can be generated in this

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.5 APPLICATIONS

(a)

117

(b)

Figure 5.17. (a) Top view and (b) cross-sectional SEMs of 3D structures formed by molding fine features (equally spaced lines and spaces with relief depth rd = 220 nm and periodicity p = 400 nm) into a layer of SU-8 and then performing PnP on the molded structure with the same element used for the molding but with a 90◦ rotation. (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

fashion [88]. The use of fine molded structures with phase masks can create multiple levels of phase modulation to achieve, for example, woodpile-type 3D structures, as illustrated in Figure 5.17.

5.5 APPLICATIONS The wide range of 2D and 3D structures that can be generated by OSL, as illustrated in the previous sections, create many application possibilities. The following provides some examples. 5.5.1 Low-Voltage Organic Electronics The basic near-field phase shift lithography technique, as described in Section 5.3.1, can form small gaps between metal electrodes, in geometries that can be useful in transistor applications. Here, the widths of the narrow lines of photoresist define channel lengths with similar dimensions, i.e., ∼0.1 µm [89]. Short channels are important because they enable high current outputs and fast switching. The fabrication sequence for a complementary inverter formed in this manner consists in generating photoresist lines of 100-nm width by near-field phase shift lithography. Next, ∼20 nm of gold is deposited onto the patterned photoresist and the resist is then removed with acetone. This process yields ∼0.1-µm slits in the gold (Figure 5.18a). The lateral dimensions of source/drain electrodes that are separated by the ∼0.1-µm slits can be defined by microcontact printing, shadow masking, or other methods. In the results shown here [89], the lines were ∼200 µm wide, and designed to produce arrays of interconnected transistors for inverter circuits. Deposition of semiconductors onto these electrodes by sublimation through a shadow mask defines n- and p-channel transistors. The underlying doped silicon substrate serves as the gate for both types of devices. A thin (∼20-nm) layer of SiO2 or SiNx substrate insulates the gate from the source/drain electrodes.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

(a)

Current (µA)

(b) 4

100 µm 100 nm

(c) Output voltage (V)

118

13:59

3 2 1 0

0

1 2 3 Drain/source voltage (V)

4

6 4 2 0

0

1

2 3 4 5 Input voltage (V)

6

7

Figure 5.18. (a) SEM of an ∼0.1-µm-wide slit in a film of gold formed by deposition of gold onto an ∼0.1-µm-wide and several millimeters long line of photoresist patterned by near-field photolithography, followed by removal of the photoresist with acetone. (b) Current–voltage characteristics of a n-channel transistor formed with the organic semiconductor F16CuPc and source/drain electrodes ∼200 µm wide and separated by ∼0.1 µm. (c ) Transfer characteristics of a simple complementary inverter circuit formed with the organic semiconductors α-6T (p-channel) and F16CuPc (n-channel). The characteristics of this inverter are consistent with expectation based on the behavior of long channel devices; the slightly nonideal behavior is likely due to leakage currents that occur in devices with unpatterned gates and semiconductors and/or a mismatch between the mobilities of the α-6T and F16CuPc. (Reprinted with permission from [89]. Copyright 1999, American Institute of Physics.)

Figures 5.18b illustrates the performance of an n-channel transistor that uses copper hexadecafluorophthalocyanine (F16 CuPc) and electrodes patterned using the procedures described above. The data indicate good performance at low voltages, consistent with the scaled behavior of large devices with similar designs. Figure 5.18c shows a schematic diagram of a circuit and the transfer curve measured for the case of an α-6T based p-channel transistor and an F16 CuPc n-channel transistor. The inverter, which uses the p- and n-channel devices as the drivers and the loads, respectively, shows good characteristics. 5.5.2 Filters and Mixers for Microfluidics The sorts of 3D nanostructures that can be produced easily by PnP [4, 80, 90] have many possible applications in microfluidics. Passive mixers represent one example, in which the 3D structures create multiple substream flow paths to provide chaotic mixing and/or laminating flows that can reduce the distances for diffusive mixing [91–93]. Figure 5.19 shows an example of such a mixer built by PnP and integrated into a serpentine microfluidic channel. These mixers exhibit high efficiencies even under small Reynolds numbers, likely due to multiple laminating flows forced by the 3D structure [90]. Similar structures can be used as filters and separation membranes. Figure 5.19d provides an image of an example. 5.5.3 High Energy Fusion Targets and Media for Chemical Release DGS formed by PnP have many potential uses, including time-dependent release of chemicals. Experiments show that the release of riboflavin filled into a DGS has kinetics that are influenced by the structure geometry [84]. In a different

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.5 APPLICATIONS

(a)

119

(b)

(c)

(d)

Figure 5.19. (a) and (b) Images of a microfluidic device that contains an integrated 3D nanostructure formed by PnP. (c ) A confocal micrograph of a microfluidic mixer that consists of such a nanostructure integrated in a serpentine channel, collected at a flow velocity (right to left) of 6.67 mm s–1 . The white corresponds to fluorescence from Rhodamine dye in water, excited with a laser at 514 nm. The uniform intensity at the output (left) indicates good mixing of initially unmixed, laminar streams at the input (right). (d ) An image of a nanostructured microfluidic filter (flow from left to right). The beads are stopped at the edge of the structure. (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

application, thick DGS can serve as reservoir targets for high energy density science [94]. Here, appropriately designed DGS can shape the pressure profiles formed during shockless, laser-induced compression by slowing the transfer of momentum into the target. Figure 5.20 shows a time history of the ramp pressure profile resulting from a DGS (inset) designed for this purpose. The data indicate a 30% increase

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

120

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

Normalized pressure (Mbar)

0.25

0.20

0.15

0.10

Typical shockup region

Solid density Solid density (calc) Graded density Graded density (calc)

0.05

0.00 25

30

35

40

45

Normalized time (ns) Figure 5.20. Normalized pressure profile created by shockless wave compression as a function of time with and without a polymer density gradient produced by PnP in a layer of SU-8. The dashed lines are calculated pressure profiles. The pressure and time axes have been normalized. (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

in time required to achieve the similar levels of stress in the DGS compared to the solid structure, making it possible to achieve higher peak stress without shocking the sample. 5.5.4 Photonic Bandgap Materials PnP has features that might enable the fabrication of 3D PBG materials over large areas and with simple processing. With TOP PnP, it is possible to fabricate 3D structures with symmetries approaching face center cubic, similar to those in Figure 5.21, which could be useful for PBG applications. High normal incidence reflection (∼60%) was measured from these structures, as illustrated in Figure 5.21d [88]. The measurements show a peak position that agrees well with theory, but with smaller reflectance [88]. Structural disorder and shrinkage upon developing might cause this discrepancy. Improved photosensitive materials, together with optimized PnP approaches have the potential to overcome these challenges. OSL provides very simple routes to a wide range of 2D and 3D structures, useful for applications. The ability to produce uniform large area coverage in thin or thick geometries (up to ∼100 µm) in a single exposure step, and with low cost light sources (i.e., lamps), make these methods attractive compared to other technologies. Nearfield and Talbot effect optics, elastomeric photomasks, and suitable photosensitive materials are key aspects of the approach. This chapter summarizes these aspects,

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

5.5 APPLICATIONS

121

(a)

(b)

(d)

(c)

1.0 Reflection intensity (100%)

(e)

Experiment 0.8

Modeling

0.6

0.4 0.2

0.0 0.8

1.0

1.2

1.4

1.6

1.8

2.0

Wavelength (µm) Figure 5.21. Images of 3D nanostructures formed by two-photon PnP by exposure through molded relief in a layer of photopolymer (SU-8) formed with a mold consisting of a square array of posts (d = 375 nm, rd = 420 nm, and p = 566 nm). (a), (b) Cross-section and angled view images, respectively, of the resulting 3D structure. (c ), (d ) Intensity distributions along the (0 0 1) plane and the (1 1 0) plane, respectively, of a calculation of the intensity distribution, as rendered in 3D through application of a cutoff filter. (e) Experimental and theoretical reflection spectra for these structures. (Reprinted with permission from [80]. Copyright 2007, American Chemical Society.)

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

122

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

in each of several different types of OSL methods, as well as the main patterning capabilities. Several application examples in electronics, microfluidics, chemical release, high energy density science, colloidal particle fabrication and photonic structures were described. Future opportunities exist in these and other applications, as well as in further development of the patterning approach. For example, the development of inverse computational algorithms that allow masks to be designed to achieve desired patterning requirements should be possible. These and other topics represent promising directions for future research. REFERENCES 1. Zhou, W., Huang, Y., Menard, E., Aluru, N. R., Rogers, J. A., and Alleyne, A. G. (2005) Mechanism for stamp collapse in soft lithography. Appl. Phys. Lett. 87, 251925–251927. 2. Huang, Y. G. Y., Zhou, W. X., Hsia, K. J., Menard, E., Park, J. U., Rogers, J. A., and Alleyne, A. G. (2005) Stamp collapse in soft lithography. Langmuir 21, 8058–8068. 3. Hsia, K. J., Huang, Y., Menard, E., Park, J. U., Zhou, W., Rogers, J. A., and Fulton, J. M. (2005) Collapse of stamps for soft lithography due to interfacial adhesion. Appl. Phys. Lett. 86, 154106. 4. Jeon, S., Park, J. U., Cirelli, R., Yang, S., Heitzman, C. E., Braun, P. V., Kenis, P. J., and Rogers, J. A. (2004) Fabricating complex three-dimensional nanostructures with highresolution conformable phase masks. Proc. Natl Acad. Sci. USA 101, 12428–12433. 5. Xia, Y. and Whitesides, G. M. (1998) Soft lithography. Angew. Chem. Int. Ed. Engl. 37, 551–575. 6. Xia, Y., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1999) Unconventional methods for fabricating and patterning nanostructures. Chem. Rev. 99, 1823–1848. 7. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. 8. Odom, T. W., Love, J. C., Wolfe, D. B., Paul, K. E., and Whitesides, G. M. (2002) Improved pattern transfer in soft lithography using composite stamps. Langmuir 18, 5314–5320. 9. Rolland, J. P., Hagberg, E. C., Denison, G. M., Cater, K. R., and Desimone, J. M. (2004) High-resolution soft lithography: enabling materials for nanotechnologies. Angew. Chem. Int. Ed. 43, 5796–5799. 10. Truong, T. T., Lin, R., Jeon, S., Lee, H. H., Maria, J., Gaur, A., Hua, F., Meinel, I., et al. (2007) Soft lithography using acryloxy perfluoropolyether composite stamps. Langmuir 23, 2898–2905. 11. Hua, F., Gaur, A., Sun, Y., Word, M., Niu, J., Adesida, I., Shim, M., and Rogers, J. A. (2006) Processing dependent behavior of soft imprint lithography on the 1–10 nm scale. IEEE Trans. Nanotechnol. 5, 301–308. 12. Hua, F., Sun, Y. G., Gaur, A., Meitl, M. A., Bilhaut, L., Rotkina, L., Wang, J. F., Geil, P., et al. (2004) Polymer imprint lithography with molecular-scale resolution. Nano Lett. 4, 2467–2471. 13. Gates, B. D. and Whitesides, G. M. (2003) Replication of vertical features smaller than 2 nm by soft lithography. J. Am. Chem. Soc. 125, 14986–14987. 14. Imhof, J. C. and Stein, C. M. (1986) The role of the latent image in a new dual image, aqueous developable, thermally stable photoresist. Polym. Eng. Sci. 26, 1101–1104.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

REFERENCES

123

15. Vollenbroek, F. A., Nijssen, W. P. M., Mutsaers, C. M. J., Geomini, M. J. H. J., Reuhman, M. E., and Visser, R. J. (1989) The chemistry of G-line photoresist processes. Polym. Eng. Sci. 29, 928–936. 16. Klose, H., Sigush, R., and Arden, W. (1985) Image reversal of positive photoresist – characterization and modeling. IEEE Trans. Electron Devices 32, 1654–1661. 17. Labianca, N. and Gelorme, J. D. (1995) High-aspect-ratio resist for thick-film applications. Proc. SPIE 2438, 846–852. 18. Teh, W. H., Durig, U., Drechsler, U., Smith, C. G., and Guntherodt, H. J. (2005) Effect of low numerical-aperture femtosecond two-photon absorption on (SU-8) resist for ultrahigh-aspect-ratio microstereolithography. J. Appl. Phys. 97, 054907. 19. Kuebler, S. M., Braun, K. L., Zhou, W. H., Cammack, J. K., Yu, T. Y., Ober, C. K., Marder, S. R., and Perry, J. W. (2003) Design and application of high-sensitivity twophoton initiators for three-dimensional microfabrication. J. Photochem. Photobiol. A 158, 163–170. 20. Witzgall, G., Vrijen, R., Yablonovitch, E., Doan, V., and Schwartz, B. J. (1998) Single-shot two-photon exposure of commercial photoresist for the production of threedimensional structures. Opt. Lett. 23, 1745–1747. 21. Shaw, J. M., Gelorme, J. D., Labianca, N. C., Conley, W. E., and Holmes, S. J. (1997) Negative photoresists for optical lithography. IBM J. Res. Dev. 41, 81–94. 22. Crivello, J. V. (1999) The discovery and development of onium salt cationic photoinitiators. J. Polym. Sci. A1 37, 4241–4254. 23. Jun, Y. H., Leatherdale, C. A., and Norris, D. J. (2005) Tailoring air defects in selfassembled photonic bandgap crystals. Adv. Mater. 17, 1908–1911. 24. Kim, H. J., Lee, J. K., Kim, J. B., Park, E. S., Park, S. J., Yoo, D. Y., and Yoon, D. Y. (2001) Substituent effects on microstructure and polymerization of polyalkylsilsesquioxanes. J. Am. Chem. Soc. 123, 12121–12122. 25. Dror, R., Feigel, A. I., Veinguer, M., Sfez, B. G., Klebanov, M., Arsh, A., and Lyubin, V. (2005) Sub-micron sculpturing on chalcogenide films. Proc. SPIE— Int. Soc. Opt. Eng. 5720, 56–67. 26. Wong, S., Deubel, M., Perez-Willard, F., John, S., Ozin, G. A., Wegener, M., and Freymann, G. V. (2006) Direct laser writing of three-dimensional photonic crystals with complete a photonic bandgap in chalcogenide glasses. Adv. Mater. 18, 265–269. 27. Rogers, J. A., Paul, K. E., Jackman, R. J., and Whitesides, G. M. (1997) Using an elastomeric phase mask for sub-100 nm photolithography in the optical near field. Appl. Phys. Lett. 70, 2658–2660. 28. Rogers, J. A., Paul, K. E., Jackman, R. J., and Whitesides, G. M. (1998) Generating similar to 90 nanometer features using near-field contact-mode photolithography with an elastomeric phase mask. J. Vac. Sci. Technol. B 16, 59–68. 29. Maria, J., Jeon, S., and Rogers, J. A. (2004) Nanopatterning with conformable phase masks. J. Photochem. Photobiol. A 166, 149–154. 30. Maria, J., Malyarchuk, V., White, J., and Rogers, J. A. (2006) Experimental and computational studies of phase shift lithography with binary elastomeric masks. J. Vac. Sci. Technol. B. 24, 828–835. 31. Aizenberg, J., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1997) Imaging the irradiance distribution in the optical near field. Appl. Phys. Lett. 71, 3773–3775. 32. Li, Z.-Y., Yin, Y., and Xia, Y. (2001) Optimization of elastomeric phase masks for nearfield photolithography. Appl. Phys. Lett. 78, 2431–2433.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

124

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

33. Paul, K. E., Breen, T. L., Aizenberg, J., and Whitesides, G. M. (1998) Maskless photolithography: embossed photoresist as its own optical element. Appl. Phys. Lett. 73, 2893–2895. 34. Kim, E., Xia, Y., Zhao, X. M., and Whitesides, G. M. (1997) Solvent-assisted microcontact molding: a convenient method for fabricating three-dimensional structures on surfaces of polymers. Adv. Mater. 9, 651–654. 35. Paul, K. E., Breen, T. L., Hadzik, T., Whitesides, G. M., Smith, S. P., and Prentiss, M. (2005) Imaging patterns of intensity in topographically directed photolithography. J. Vac. Sci. Technol. B 23, 918–925. 36. Black, A. J., Paul, K. E., Aizenberg, J., and Whitesides, G. M. (1999) Patterning disorder in monolayer resists for the fabrication of sub-100-nm structures in silver, gold, silicon, and aluminum. J. Am. Chem. Soc. 121, 8356–8365. 37. Paul, K. E., Prentiss, M., and Whitesides, G. M. (2003) Patterning spherical surfaces at the two-hundred-nanometer scale using soft lithography. Adv. Funct. Mater. 13, 259–262. 38. Qin, D., Xia, Y., Black, A. J., and Whitesides, G. M. (1998) Photolithography with transparent reflective photomasks. J. Vac. Sci. Technol. B 16, 98–103. 39. Rogers, J. A. and Menard, E. (Unpublished) Devices and methods for pattern generation by ink lithography. 40. Chen, C., Hirdes, D., and Folch, A. (2003) Gray-scale photolithography using microfluidic photomasks. Proc. Natl Acad. Sci. U S A 100, 1499–1504. 41. Schmid, H., Biebuyck, H., Michel, B., and Martin, O. J. F. (1998) Light-coupling masks for lensless, sub-wavelength optical lithography. Appl. Phys. Lett. 72, 2379–2381. 42. Schmid, H., Biebuyck, H., Michel, B., Martin, O. J. F., and Piller, N. B. (1998) Lightcoupling masks: an alternative, lensless approach to high-resolution optical contact lithography. J. Vac. Sci. Technol. B 16, 3422–3425. 43. Martin, O. J. F., Piller, N. B., Schmid, H., Biebuyck, H., and Michel, B. (1998) Energy flow in light-coupling masks for lensless optical lithography. Opt. Express 3, 280– 285. 44. Michel, B., Bernard, A., Bietsch, A., Delamarche, E., Geissler, M., Juncker, D., Kind, H., Renault, J. P., et al. (2001) Printing meets lithography: soft approaches to high-resolution printing. IBM J. Res. Dev. 45, 697–719. 45. Moorthy, J., Mensing, G. A., Kim, D., Mohanty, S., Eddington, D. T., Tepp, W. H., Johnson, E. A., and Beebe, D. J. (2004) Microfluidic tectonics platform: a colorimetric, disposable botulinum toxin enzyme-linked immunosorbent assay system. Electrophoresis 25, 1705–1713. 46. Vijayendran, R. A., Motsegood, K. M., Beebe, D. J., and Leckband, D. E. (2003) Evaluation of a three-dimensional micromixer in a surface-based biosensor. Langmuir 19, 1824–1828. 47. Pregibon, D. C., Toner, M., and Doyle, P. S. (2006) Magnetically and biologically active bead-patterned hydrogels. Langmuir 22, 5122–5128. 48. Holtz, J. H. and Asher, S. A. (1997) Polymerized colloidal crystal hydrogel films as intelligent chemical sensing materials. Nature 389, 829–832. 49. Grier, D. G. (2003) A revolution in optical manipulation. Nature 424, 810–816. 50. Christodoulides, D. N., Lederer, F., and Silberberg, Y. (2003) Discretizing light behaviour in linear and nonlinear waveguide lattices. Nature 424, 817–823.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

REFERENCES

125

51. Braun, P. V., Rinne, S. A., and Garcia-Santamaria, F. (2006) Introducing defects in 3D photonic crystals: state of the art. Adv. Mater. 18, 2665–2677. 52. Long, J. W., Dunn, B., Rolison, D. R., and White, H. S. (2004) Three-dimensional battery architectures. Chem. Rev. 104, 4463–4492. 53. Holland, B. T. and Blanford, C. (1998) Synthesis of macroporous minerals with highly ordered three-dimensional arrays of spheroidal voids. Science 281, 538–540. 54. Chai, G. S., Shin, I. S., and Yu, J. S. (2004) Synthesis of ordered, uniform, macroporous carbons with mesoporous walls templated by aggregates of polystyrene spheres and silica particles for use as catalyst supports in direct methanol fuel cells. Adv. Mater. 16, 2057–2061. 55. Parthenopoulos, D. A. and Rentzepis, P. M. (1989) Three-dimensional optical storage memory. Science 245, 843–845. 56. Dinsmore, A. D., Hsu, M. F., Nikolaides, M. G., Marquez, M., Bausch, A. R., and Weitz, D. A. (2002) Colloidosomes: selectively permeable capsules composed of colloidal particles. Science 298, 1006–1009. 57. Jiang, P., Cizeron, J., Bertone, J. F., and Colvin, V. L. (1999) Preparation of macroporous metal films from colloidal crystals. J. Am. Chem. Soc. 121, 7957–7958. 58. Velev, O. D., Jede, T. A., Lobo, R. F., and Lenhoff, A. M. (1997) Porous silica via colloidal crystallization. Nature 389, 447–448. 59. Vlasov, Y. A., Bo, X. Z., Sturm, J. C., and Norris, D. J. (2001) On-chip natural assembly of silicon photonic bandgap crystals. Nature 414, 289–293. 60. Bates, F. S. (1991) Polymer-polymer phase-behavior. Science 251, 898–905. 61. Boltau, M., Walheim, S., Mlynek, J., Krausch, G., and Steiner, U. (1998) Surfaceinduced structure formation of polymer blends on patterned substrates. Nature 391, 877– 879. 62. Kim, S. O., Solak, H. H., Stoykovich, M. P., Ferrier, N. J., Depablo, J. J., and Nealey, P. F. (2003) Epitaxial self-assembly of block copolymers on lithographically defined nanopatterned substrates. Nature 424, 411–414. 63. Furneaux, R. C., Rigby, W. R., and Davidson, A. P. (1989) The formation of controlledporosity membranes from anodically oxidized aluminum. Nature 337, 147–149. 64. Trau, M., Yao, N., Kim, E., Xia, Y., and Whitesides, G. M. (1997) Microscopic patterning of orientated mesoscopic silica through guided growth. Nature 390, 674–676. 65. Jacobs, H. O., Tao, A. R., Schwartz, A., Gracias, D. H., and Whitesides, G. M. (2002) Fabrication of a cylindrical display by patterned assembly. Science 296, 323–325. 66. Yeh, H. J. J. and Smith, J. S. (1994) Fluidic self-assembly for the integration of GaAs light-emitting-diodes on Si substrates. IEEE Photon. Technol. Lett. 6, 706–708. 67. Campbell, M., Sharp, D. N., Harrison, M. T., Denning, R. G., and Tuberfield, A. J. (2000) Fabrication of photonic crystals for the visible spectrum by holographic lithography. Nature 404, 53–56. 68. Divliansky, I., Mayer, T. S., Holliday, K. S., and Crespi, V. H. (2003) Fabrication of three-dimensional polymer photonic crystal structures using single diffraction element interference lithography. Appl. Phys. Lett. 82, 1667–1669. 69. Ullal, C. K., Maldovan, M., Thomas, E. L., Chen, G., Han, Y. J., and Yang, S. (2004) Photonic crystals through holographic lithography: simple cubic, diamond-like, and gyroidlike structures. Appl. Phys. Lett. 84, 5434–5436.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

126

13:59

Printer Name: Yet to Come

PATTERNING BASED ON LIGHT

70. Matthias, S. and Muller, F. (2003) Asymmetric pores in a silicon membrane acting as massively parallel brownian ratchets. Nature 424, 53–57. 71. Matthias, S., Muller, F., and Gosele, U. (2005) Simple cubic three-dimensional photonic crystals based on macroporous silicon and anisotropic posttreatment. J. Appl. Phys. 98, 023524. 72. Kawata, S., Sun, H. B., Tanaka, T., and Takada, K. (2001) Finer features for functional microdevices—micromachines can be created with higher resolution using two-photon absorption. Nature 412, 697–698. 73. Galajda, P. and Ormos, P. (2001) Complex micromachines produced and driven by light. Appl. Phys. Lett. 78, 249–251. 74. Gratson, G. M., Xu, M. J., and Lewis, J. A. (2004) Microperiodic structures—direct writing of three-dimensional webs. Nature 428, 386–386. 75. Talbot, H. F. (1836) Facts relating to optical science, no. IV. Phil. Mag. 9, 401–407. 76. Rayleigh, L. (1881) On copying diffraction gratings, and on some phenomenon. Connected therewith. Phil. Mag. 11, 196. 77. Berry, M., Marzoli, I., and Schleich W. (1992) Quantum carpets, carpets of light. Phys. World 14, 39–44. 78. Latimer, P. and Crouse, R. F. (1992) Talbot effect reinterpreted. Appl. Opt. 31, 80–89. 79. Jeon, S., Malyarchuck, V., Rogers, J. A., and Wiederrecht, G. P. (2006) Fabricating three dimensional nanostructures using two photon lithography in a single exposure step. Opt. Express 14, 2300–2308. 80. Shir, D. J., Jeon, S., Liao, H., Highland, M., Cahill, D. G., Su, M. F., El-Kady, I. F., Christodoulou, C. G., et al. (2007) Three-dimensional nanofabrication with elastomeric phase masks. J. Phys. Chem. B 111, 12945–12958. 81. Shevchenko, E. V., Talapin, D. V., Kotov, N. A., O’Brien, S., and Murray, C. B. (2006) Structural diversity in binary nanoparticle superlattices. Nature 439, 55–59. 82. Park, S. H. and Xia, Y. N. (1999) Assembly of mesoscale particles over large areas and its application in fabricating tunable optical filters. Langmuir 15, 266–273. 83. Jeon, S., Nam, Y. S., Shir, D., and Rogers, J. A. (2006) Three dimensional nanoporous density graded materials formed by optical exposures through conformable phase masks. Appl. Phys. Lett. 89, 253101. 84. Nam, Y. S., Jeon, S., Shir, D., and Rogers, J. A. (2007) Thick, three-dimensional nanoporous density-graded materials formed by optical exposures of photopolymers with controlled levels of absorption. Appl. Opt. 46, 6350–6354. 85. Shir, D., Liao, H., Jeon, S., Xiao, D., Johnson, H. T., Bogart, G. R., Bogart K. H. A., Rogers, J. A. (2008) Three-dimensional nanostructures formed by single step, two-photon exposures through elastomeric Penrose quasicrystal phase masks. Nano Lett. In press. 86. Zoorob, M. E., Charlton, M. D. B., Parker, G. J., Baumberg, J. J., Netti, M. C. (2000) Complete photonic bandgaps in 12-fold symmetric quasicrystals. Nature 404, 740–743. 87. Ledermann, A., Cademartiri, L., Hermatschweiler, M., Toninelli, C., Ozin, G. A., Wiersma, D. S., Wegener, M., and Freymann, G. V. (2006) Three-dimensional silicon inverse photonic quasicrystals for infrared wavelengths. Nature Mater. 5, 942–945. 88. Jeon, S., Shir, D., Nam, Y. S., and Rogers, J. A. (2007) Molded transparent photopolymers and phase shift optics for fabricating three dimensional nanostructures. Opt. Express 15, 6358–6366.

P1: OTA/XYZ P2: ABC c05 JWBK252/Lee September 18, 2008

13:59

Printer Name: Yet to Come

REFERENCES

127

89. Rogers, J. A., Dodabalapur, A., Bao, Z., and Katz, H. E. (1999) Low-voltage 0.1 µm organic transistors and complementary inverter circuits fabricated with a low-cost form of near-field photolithography. Appl. Phys. Lett. 75, 1010–1012. 90. Jeon, S., Malyarchuk, V., White, J. O., and Rogers, J. A. (2005) Optically fabricated three dimensional nanofluidic mixers for microfluidic devices. Nano Lett., 5, 1351–1356. 91. Kakuta, M., Bessoth, F. G., and Manz, A. (2001) Microfabricated devices for fluid mixing and their application for chemical synthesis. Chem. Rec. 1, 395–405. 92. Stroock, A. D., Dertinger, S. K. W., Ajdari, A., Mezit, I., Stone, H. A., and Whitesides, G. M. (2002) Chaotic mixer for microchannels. Science 295, 647–651. 93. Kenis, P. J. A., Ismagilov, R. F., and Whitesides, G. M. (1999) Microfabrication inside capillaries using multiphase laminar flow patterning. Science 285, 83–85. 94. Smith, R. F., Lorenz, K. T., Ho, D., Remington, B. A., Hamza, A., Rogers, J. A., Pollaine, S., Jeon, S., et al. (2007) Graded-density reservoirs for accessing high stress low temperature material states. Astrophys. Space Sci. 307, 269–272.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6 PATTERNING BASED ON EXTERNAL FORCE: NANOIMPRINT LITHOGRAPHY L. Jay Guo

6.1 INTRODUCTION The ability to fabricate structures from micro- to nanoscale in a wide variety of materials with high precision is important to the advancement of nanotechnologies. One of the primary drivers in developing precision nanoscale lithography is the desire to make ever-shrinking transistors on IC chips. Considerable industrial effort has been devoted to the leading-edge optical method and the so-called next generation lithography (NGL) techniques by exposing resist material with energetic beams from extreme ultraviolet (UV), electron-beam, ion-beam, or x-ray sources. Not only are there a lot of issues that are yet to be solved in deep ultra-violet (DUV) photolithography and various NGL methods, but also the cost for a single NGL tool could exceed $50 million in the next few years—a formidable price tag for most potential users. Critical issues such as resolution, reliability, speed, and overlay accuracy all need to be addressed in developing new lithography methodologies for such stringent industrial processes. On the other hand, less demanding conditions are required for many other areas of development, such as photonics, micro- and nanofluidics, chip-based sensors, and most of the biological applications. Several alternative approaches have been exploited in the past decade for nanostructure fabrication without

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


129

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

130

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

resorting to expensive tools such as those used in deep-UV projection lithography and electron-beam lithography. These techniques include microcontact printing, nanoimprint lithography, scanning probe-based techniques such as atomic force microscopic (AFM) lithography and dip-pen lithography. An overview of some of these techniques were presented in a recent paper [1]. Since mid-1990s, nanoimprint lithography (NIL), initially proposed and developed by the Chou group [2, 3], has emerged as one of the most promising technologies for high throughput nanoscale patterning. In this method and its variants, such as step-and-flash imprint lithography (S-FIL) that was developed by Willson’s group, pattern replication is done nontraditionally by mechanically deforming the resist materials, which completely free itself from the resolution-limiting factors such as light diffraction and beam scattering that are often inherent with the more traditional approaches. We will discuss the basic principle, material requirement, and progress being made in this field in recent years. The principle of nanoimprint is very simple. Figure 6.1a shows the schematic of the originally proposed NIL process. A hard mold that contains nanoscale surface relief features is pressed into a polymer material cast on a substrate under controlled temperature and pressure, which creates thickness contrast in the polymer material. A thin residual layer is left intentionally underneath the mold protrusions acting as a soft cushion layer that prevents direct impact of the hard mold onto the substrate, which effectively protects the delicate nanoscale features on the mold surface. In most of the applications this residual layer needs to be removed by an anisotropic

(a)

(b)

Figure 6.1. Schematic of (a) nanoimprint lithography and (b) step-and-flash imprint lithography. (Reproduced with permission from [4] and the Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.1 INTRODUCTION

131

10 nm

10 nm

60 nm

(a)

(b)

Figure 6.2. SEM micrographs of (a) fabricated mold of with 10-nm-diameter pillar array and (b) the imprinted hole arrays in PMMA by using such a mold. (Reproduced with permission from [5] and the American Institute of Physics.)

O2 plasma etching process to complete the pattern definition. What makes NIL an attractive and widely studied technology is that it has demonstrated the ultra-high resolution soon after its inception. Figures 6.2a and b show the scanning electron micrographs (SEMs) of a mold with 10-nm-diameter pillar array and the imprinted 10 nm hole arrays in poly(methylmethacrylate) (PMMA) that was obtained a decade ago [5]. NIL is inherently high throughput due to parallel printing; and it only requires simple equipment setup that leads to low cost processing. A variation of the NIL technique, S-FIL [6, 7] was developed soon after, which uses a transparent mold and UV-curable precursor liquid to define the pattern (Figure 6.1b), allowing the process to be carried out at room temperature and making it very attractive to the IC semiconductor device manufacturing [8]. In S-FIL process, the substrate is first coated with an organic transfer layer; then a surface-treated, transparent template with surface relief patterns is brought close and aligned to the coated substrate. Once in proximity, a drop of low viscosity, photopolymerizable organosilicon solution is introduced into the gap between the template and the substrate. The organosilicon fluid spreads out and fills the gap by capillary action. Next, the template is pressed against the substrate to close the gap and the assembly is irradiated with UV light, which cures the photopolymer to make it a solidified and a silicon-rich replica of the template. The rest of the process is very similar to that in thermal NIL. Due to its fast development in the past decade and its potential for sub-50-nm lithography, MIT’s Technology Review listed NIL as one of 10 emerging technologies that will strongly impact the world [9], and in 2003 the International Technology Roadmap for Semiconductors (ITRS) also announced the inclusion of NIL onto their roadmap as a candidate technology for future IC production. Significant efforts from both academia and industry have been put in S-FIL research and development [6], template fabrication methods [10, 11], and defect analysis [11, 12]. Figure 6.3a shows 20-nm lines printed by S-FIL and Figure 6.3b shows 40-nm lines printed with a template used over 1500 times. Using interferometric

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

132

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

(c) (a)

20-nm lines

(b)

(d )

Figure 6.3. (a) Twenty-nanometer lines patterned by step-and-flash imprint lithography (S-FIL). (b) Forty-nanometer line pattern printed with template used over 1500 times. (c ), (d ) Multitiered S-FIL template and the resulting imprint in resist material. (Reprinted with permission from [7].)

in situ alignment techniques, overlay accuracy of 50 nm have been achieved [13]. Higher degree of accuracy can also be anticipated. Although NIL-based approach proved excellent resolution, there are still significant challenges in meeting the stringent requirement in semiconductor IC manufacturing, especially in terms of defect control and throughput, which requires printing of 60–80 wafers per hour in production. On the other hand, the simplicity of this technique have found numerous applications in electronics such as hybrid plastic electronics [14], organic electronics and photonics [15, 16], and nanoelectronic devices in Si [17, 18] and in GaAs [19]; in photonics such as organic lasers [20], solar cells [21], conjugate [22] and nonlinear optical polymer nanostructures [23], high resolution organic light-emitting diode (OLED) pixels [24, 25], diffractive optical elements [26], and broadband polarizers [27–29]; in magnetic devices such as single-domain magnetic structures [30, 31], high density patterned magnetic media and high capacity disks [32, 33], and patterned magnetic media [34]; nanoscale control of polymer crystallization [35]; as well as in biological applications such as manipulating DNA in nanofluidic channels [36, 37], nanoscale protein patterning [38, 39], and the effect of imprinted nanostructures on cell culture [40]. Because NIL is based on mechanical molding of the polymer materials, which is very different from traditional lithographic techniques, it has to deal with a new set of

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.2 NIL MOLD

133

issues and challenges. In what follows, we will discuss in detail the mold and resist requirement and identify critical issues in the imprinting, separation, and etching steps and describe solutions to overcome the various technical challenges.

6.2 NIL MOLD The elements required for NIL are as follows: (1) a mold with predefined surface relief nanostructures and (2) a suitable resist material that can be deformed and hardened to preserve the shape of the impression. Usually the resist material is applied on top of a substrate. The molds used in NIL can essentially be any type of solid materials that have high strength and durability. The resist material can be thermoplastic, thermosetting, or low viscosity precursors that can cross-link either by thermal curing or by UV light curing. The molds are normally made in silicon, dielectric materials such as silicon dioxide or silicon nitride, metals such as Ni, or polymeric materials. The common feature of different types of molds used in NIL is that they are hard and have high mechanical strength. The hard molds used in NIL contrast the soft elastomeric stamps used in soft lithography, and is required for producing nanoscale features because the protrusion patterns on the mold do not deform, buckle, or collapse during imprinting even at elevated temperature, thereby preserving the shape and aspect ratio, and guarantee faithful pattern definition down to sub-10-nm scale. This requirement does not preclude the use of a flexible backplane supporting the hard surface relief structures. For certain applications concerning very large area imprinting, it is advantageous for the mold to have global flexibility and local rigidity especially when the substrate used is not flat, because a flexible mold can provide large-scale conformal contact with the substrate without resorting to high pressure. Rigiflex lithography explicitly exploit this feature for NIL patterning [41]. 6.2.1 Mold Fabrication Considerations for selecting mold materials should also include compatibility with traditional microfabrication processing and thermal expansion coefficient. Li has examined the relative hardness of various materials suitable for imprinting mold, including Si, SiO2 , SiC, silicon nitride, and sapphire [42]. Taniguchi et al. [43] have investigated diamond as a potential mold material for NIL. On the other hand, work from many groups have shown that Si and SiO2 have sufficient hardness and durability for the nanoimprint application. Thermal expansion coefficient is especially important in the thermal NIL process where a temperature over 100◦ C is typically required in the imprinting step. Thermal mismatch between the mold and the substrate could result in pattern distortions or stress buildup during the cooling cycle, which would affect the pattern fidelity and registration accuracy. In this regard, Si mold and Si substrate make a very good pair for the NIL process for applications with very precise critical dimension controls. The thermal expansion mismatch can be ignored if the NIL process is carried at room temperature, like in UV-assisted NIL or S-FIL.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

134

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

The fabrication of Si mold is straightforward if one has access to other lithographic processing and reactive ion etching (RIE) facilities. The commonly used processing steps are illustrated in Figure 6.4a. First, a resist material is spin coated on the mold surface (either Si substrate or Si with a thermally grown oxide), followed by lithography to define the desired mold patterns. One can choose from UV lithography for relatively large features, electron-beam lithography for very small features, interference lithography for large-area periodic features, or NIL itself. A hard masking layer such as metal can be deposited over the patterned resist template followed by a liftoff process that removes the resist template and the material on top, leaving a patterned mask layer on the Si substrate. Next, an anisotropic RIE process is used to selectively etch away the Si material in the unmasked region, producing the surface relief structures required in NIL. Figure 6.4b shows a scanning electron micrograph of a fabricated Si mold with protrusion features etched into a thermal oxide layer on top of the Si substrate. Figure 6.5 shows another example where the mold having different periodic features are fabricated by using with a grating mold and perform NIL twice at either 90◦ or 60◦ orientations with respect to each other, followed by a metal deposition and liftoff, and finally RIE to produce the protrusion features. Since the original grating mold can be fabricated by using holographic interference lithography, the molds created by double imprinting and RIE can cover very large area (approximately inches). For example, these molds have been used to fabricate large-area uniform and oriented metal nanoparticle arrays for studying localized surface plasmon resonances [44]. Mold or template used in S-FIL can be fabricated in similar fashions. But because S-FIL requires transparent template (typically made in quartz or silica), charging effect could occur during the electron-beam lithography process due to the nonconductive nature of the dielectric materials. Charging effect could severely distort the electron beam and affect the pattern resolution and fidelity. A couple of methods have been used to eliminate such effect. One is to coat a thin metal layer (e.g., 15 nm Cr) on top of the quartz substrate before spin-coating the electron-beam resist to help discharge [10]. Another advantage of the Cr layer is that it can be used directly as a etch mask in the subsequent RIE of quartz to form mold features. The second method is to deposit a transparent conductive oxide layer such as indium tin-oxide (ITO) to function both as a discharge layer and as an integral part of the mold structure itself [11]. The charging effect is suppressed not only during electron-beam lithography of making the mold, but also during final inspection stage, which greatly facilitates the mold manufacturing. These two methods have been used to fabricate S-FIL templates on standard photomask plate. Another advantage of the second method is that multiple layers of ITO can be deposited and a three-dimensional (3D) tiered structures can be fabricated. An example of this is shown in Figure 6.3c, showing a threelayer mold structure. Figure 6.3d is the imprinted pattern using such a mold. This approach can truly take the advantage of the mechanical molding nature of the NIL process and produce 3D patterns in polymers in a single step. It is also possible to use focused ion beam to directly “carve out” the desired mold features, especially 3D or gray scale patterns that is very difficult to obtain if using other lithographic techniques.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.2 NIL MOLD

135

Figure 6.4. (a) Schematic to show the Si mold fabrication process. (b) SEM picture of a fabricated grating mold on Si wafer. (The roughness seen on top of the etched oxide ridges is due to the metal mask used in RIE process, which has not been removed.) (Reproduced with permission from [4] and the Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

136

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

(a)

(b)

(c)

Figure 6.5. SEM pictures of large-area molds of various feature arrays. (a) Nanopillar array is produced by imprinting twice with the same grating mold but orthogonal direction. (b) Nanobar array produced by two grating molds with different period. (c ) Diamond-shaped dot array by two imprints using the same grating mold but are oriented 60◦ with respect to each other.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.2 NIL MOLD

137

6.2.2 Mold Surface Preparation Mold used for imprint lithography typically has high density of nanoscale protrusion features on its surface, which effectively increases the total surface area that contacts the imprinted polymer, leading to strong adhesion of imprinted polymer to the mold. This effect can be easily seen by the sticking of resist material on a mold without any special treatment. The solution to this problem is as follows: (1) incorporate internal release agent to the resist formulation (as what is done in the precursor mixer used for S-FIL); (2) apply a low surface tension coating to the mold to reduce its surface energy; or a combination of both approaches; and (3) choose a mold material that has low surface energy. The most widely adopted approach is to form a self-assembled monolayer of release agent to the mold surface (e.g., 1H,1H,2H,2Hperfluorodecyl-trichlorosilane), either in the solution phase or vapor phase [45]. This approach can be readily applied to oxide surfaces or to Si surfaces that have been oxidized with a piranha soak to generate the needed hydroxyl terminal groups. Silanization of oxidized silicon with a RSiCl3 precursor begins with the hydrolysis of the polar headgroup, which converts the Si–Cl bonds to Si–OH (silanol) groups. The generated silanol groups, which are strongly attracted to the hydrophilic surface of oxidized silicon, condense and react with the hydroxyl group on the surface as well as other monomer silanol group to form networks of covalent siloxane bonds, Si–O–Si (Figure 6.6). Such covalent bonding makes the surfactant coating layer chemically and thermally stable. By comparing the substrates treated by two different processes using AFM, ellipsometry, infrared spectroscopy, and contact angle measurement, Jung et al. [46] showed that the vapor phase coating provide superior surface releasing property. Schift et al. [47] found that the antiadhesion properties for mold surfaces coated with fluorinated trichlorosilane can be further improved by the codeposition of monochlorosilanes. This is because the introduction of monochlorosilanes helps to reduce the steric hindrance between the trichlorosilane molecules bond to the mold surface, resulting in a better molecular packing than that of the individual silane coatings. It has been shown experimentally that fluorosilane-coated molds can be used to imprint hundreds to thousand times before its antistiction property degrades.

6.2.3 Flexible Fluoropolymer Mold The durability issue of the surface coating on molds can be alleviated if the mold itself is made of a material that has low surface energy and sufficient mechanical strength. Lee et al. [48] have demonstrated that amorphous fluoropolymers, such as Teflon AF (T g = 240◦ C), can be used as an imprinting mold without any surface treatment. The polymer has a tensile modulus of ∼1.6 GPa, which is stiff enough for patterning small features without mold deformation. Also, the inert nature with a low surface energy of 15.6 dyn cm–1 makes it easy to demold after the imprinting process without additional mold surface treatment. Mold fabrication can also be simplified by casting of the fluoropolymer solutions over a prefabricated template and drying off the solvent [49] or by direct molding or imprinting of this fluoropolymer at 350◦ C under a high pressure [48]. Choi et al. [50] demonstrated a fluorinated

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

138

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

Figure 6.6. Formation of a monolayer of perfluorodecyl-trichlorosilane (FDTS) molecules on SiO2 to create a low energy surface. (Reprinted with permission from [45] and the Institute of Physics.)

organic–inorganic hybrid mold that has a thermal stability over 350◦ C by using a nonhydrolytic sol–gel process. We have utilized another fluoropolymer, ethylenetetrafluoroethylene (ETFE), which is copolymer of ethylene and tetrafluoroethylene, as a flexible mold in NIL. This will be discussed in more detail in later session on roller nanoimprint. These materials can be used to make inexpensive copies of the original mold that may be difficult or expensive to make. Additionally, a flexible film can provide better conformal contact with the substrate to be patterned and reduce the pressure needed during the imprinting step. A further benefit is that demolding can be achieved by pealing off the mold from the imprinted substrate with a small demolding area.

6.3 NIL RESIST Because imprint lithography makes a conformal replica of the surface relief patterns by mechanical embossing, the resist materials used in imprinting should be

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.3 NIL RESIST

139

deformable easily under an applied pressure, have sufficient mechanical strength and good mold-releasing property to maintain structural integrity during the demolding process, and for some applications possess good etching properties in RIE processes. Therefore, nanoimprint resist systems with a combined mold release and etch resistance properties that allow fast and precise nanopatterning are highly desired. The rapid development of NIL in recent years has also stimulated the research for new materials that are better suited as the nanoimprint resist [51–53]. There are two material properties that are important for the imprinting process. The resist material should have a modulus lower than that of the mold during imprinting. The minimal pressure required to perform the imprint should be higher than the sheer modulus of the polymer [54]. On the other hand, the low modulus of the resist material is only a necessary condition for it to be used in NIL. In order to complete the imprinting process with practical speed, the resist material should also have sufficient low viscosity [55]. This requirement can be understood by considering a simplified model for NIL, namely, squeezed flow of a Newtonian fluid (resist material) between two plates that represent the mold and the substrate surface. Suppose the plates have a radius R and an initial gap distance of d, which can approximate the process where a mold protrusion with size R is imprinted into a resist of an initial thickness d on a substrate. One can obtain a simple solution that expresses the pressure (P) needed to obtain certain imprinting speed (dh/dt) in terms of these dimensions and the fluid viscosity: P = (3 • R 3 /4d 3 )(dh/dt). Numerical simulation based on the finite difference method using a non-steady-state Navier–Stokes equation has also given such relationship for simple periodic mold features [56]. Integrating this equation can give the time required to reduce the thickness of the fluidic layer by half (i.e., to imprint half way through the feature height): t = 9h R 2 /16Pd 2 . This simple analysis says that under a fixed pressure the imprinting time scales linearly with viscosity and quadratically with the pattern size. Therefore, the time required for imprinting large-size patterns will be significantly longer than that for nanopatterns. 6.3.1 Thermoplastic Resist For the thermoplastic materials used in the thermal NIL process, the two requirements can be satisfied simultaneously by raising the temperature to above the glass transition temperature (T g ) of the polymer, such that both the modulus and viscosity will drop by several orders of magnitude as compared with their respective values at room temperature (the Young’s modulus for glassy polymers just below T g is approximately constant over a wide range of polymers ∼3 × 109 Pa). In thermal NIL, the mold and imprinted polymer must be cooled to below T g of the polymer in order to preserve the pattern after mold is removed. In practice the temperature chosen for NIL is about 70–90◦ C above the T g , so the polymer material could reach viscous flow state [45]. In this temperature range, the viscosity of the polymer can be described by the following equation by Williams, Landel, and Ferry, log η(T ) = log η(Tg ) −

C1 (T − Tg ) C2 + (T − Tg )

(6.1)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

140

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

where the values for C1 and C2 are 17.44 and 51.6, respectively. This equation describes the effect of temperature on viscosity for a large number of polymers. Since T g is the onset temperature for molecular motion in polymers, factors that increase the energy required for molecular motion (e.g., intermolecular forces, intrachain steric hindrance such as branching or cross-linking, and bulky and stiff side groups) also increase T g ; those that decrease the energy requirement (e.g., flexible bonds, flexible side groups) lower T g . These considerations can be exploited for choosing the desired T g for the imprint resist. The viscosity of a polymer material not only depends on the temperature, but also strongly depends on the polymer’s molecular weight M w , relative to the so-called critical molecular weight of a given polymer M c . M c can be interpreted as the molecular weight at which a temporary network of entanglements spans over macroscopic dimensions. In practice, low molecular weight polymers with M < M c can be imprinted at lower temperatures, lower pressures, or within shorter times. However, the absence of a network of entanglements may lead to more brittle behavior and could result in the fracture of the imprinted polymer features during the mold separation step. Therefore, the choice of T g and M w is important in pattern structural stability. In addition, the stress buildup due to applied pressure at a temperature below T g after the imprinting strongly affect the polymer pattern integrity during mold separation. Hirai et al. [57] have investigated this problem in detail. For more discussions on the material rheology issues related to nanoimprint, the readers are referred to Chapters 3 and 4 in [58]. For many applications, it is desirable to use lower temperature in the processing. But there is a trade-off between the imprinting temperature and the thermal stability, which is illustrated in the following example for polymers of different T g . Figure 6.7a shows the imprinted patterns in poly(benzyl methacrylate) (M w ∼ 70,000, T g = 54◦ C) at temperature of 134◦ C; and Figure 6.7b the patterns imprinted in poly(cyclohexyl acrylate) (M w ∼150,000; T g = 19◦ C) at 99◦ C. In both cases, the imprinting temperatures were chosen to be T g + 80◦ C and the applied pressure was 50 kg cm–2 , and very good pattern definition were obtained. Although low T g material can be used for NIL for the sake of reducing processing temperature,

(a)

(b )

(c )

Figure 6.7. SEM micrograph of patterns imprinted in (a) poly(benzyl methacrylate) and (b) poly(cyclohexyl acrylate); and (c ) relaxation of poly(cyclohexyl acrylate) patterns 10 days after the imprinting. (Reprinted with permission from [4] and the Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.3 NIL RESIST

141

the imprinted patterns are also unstable and tend to deform at temperatures close to the imprinting temperature. Figure 6.7c shows the pattern relaxation at room temperature that was observed 10 days after the structures shown in Figure 6.7b were imprinted (in this case the resist has a T g of 19◦ C, less than room temperature). The patterned structures will relax faster at elevated temperature, such as in a RIE chamber during subsequent residual resist removal or pattern transfer into substrate by RIE. Reduction in imprinting temperature has to be compensated by increase in pressure and time to obtain satisfactory results [59]. Various alternatives have been exploited to make the thermoplastics imprintable at a temperature close to room temperature, including dissolving the polymer into its monomer [60] or to incorporate a certain amount of solvent in the polymer. Lee et al. developed a room temperature nanoimprint technique based on the latter strategy and using poly(dimethylsiloxane) (PDMS) as stamp material. By choosing a solvent that can be readily absorbed into the PDMS mold and evaporates through the mold into air, pattern replication with high fidelity can be obtained at room temperature and a pressure of less than 1 N cm–2 [61]. The drawback of this technique is the relatively long time (∼10 min) required for solvent to completely evaporate through the elastomeric mold. 6.3.2 Copolymer Thermoplastic Resists Selection of a polymer system for use as the NIL resist should consider critical aspects of correct pattern replication, modest imprint temperature and pressure, proper mold release, and etch selectivity. Much of the thermal NIL uses homopolymer resists such as PMMA and polystyrene (PS), but they are susceptible to mold-sticking and fracture defects during mold release that are intolerable for many device applications. In this respect, the polymer should satisfy seemingly contradictory requirements of having a low surface energy and high fracture strength, yet maintaining sufficient adhesion to the substrate. Although the mold surface is normally treated with low surface energy surfactants [62], when imprinting high density nanoscale structures or high aspect ratio patterns, the imprinted polymer still tends to adhere to the mold. To address these critical needs, materials that possess dual surface properties are needed. One attractive system is the PDMS–organic block or graft copolymers. In contrast to PMMA and organic polymers in general, siloxane copolymers exhibit significant differences by virtue of the highly open, flexible, and mobile Si–O–Si backbone. These qualities include low surface energy, low glass transition, and high thermal stability. Furthermore, it is known that these copolymers undergo microphase segregation above their glass transition temperature (T g ) due to the unfavorable enthalpy of mixing. When cast or hot pressed on a high surface energy substrate such as silicon, glass, or metal, the copolymer film forms a lower surface energy component (PDMS)-enriched air–polymer interface, and a higher surface energy component (organic block) dominated polymer–substrate interface [63]. The dual surface character makes these copolymers excellent candidates as NIL resists, because they allow easy mold resist separation (a lower surface energy), and at the same time good adhesion to the substrate (a higher energy surface). This duality is not possible with homopolymers. Siloxane copolymers offer another advantage over

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

142

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

homopolymers as NIL resist in that they can offer much improved etching resistance due to high Si content and high Si–O bond strength. Choi et al. [64] have investigated a number of siloxane block and graft copolymers for NIL application, including poly(dimethylsiloxane)–polystyrene diblock copolymer (PDMS-b-PS) and poly(dimethylsiloxane)-graft-poly(methylmethacrylate) (PDMS-g-PMMA). Figure 6.8a shows a 250-nm linewidth gratings imprinted in PDMS-b-PS, and Figures 6.8b and c 70-nm-linewidth grating silicon nitride mold and the imprinted poly(dimethylsiloxane)-graft-poly(methyl acrylate)-co-poly(isobornyl acrylate) (PDMS-g-PMA-co-PIA) grating pattern. These siloxane copolymers all show excellent mold-releasing properties, though one of the trade-offs for having high silicon content is increased roughness due to phase separation. To demonstrate the largearea NIL performance of the copolymers, a full 4-in. wafer with 100-nm linewidth and spacing grating features was used to do imprint. Such large area and dense features made it impossible to achieve mold–resist separation with homopolymers such as PMMA without breaking the mold or the substrate due to exceedingly strong adhesion force even with a fluorosilane-treated mold. With PDMS-g-PMA-co-PIA graft copolymer, excellent mold releasing was obtained. Figure 6.8d is a photograph showing strong blue light diffraction from the periodic gratings imprinted into the PDMS-g-PMA-co-PIA when the sample is immersed in water (diffraction in air is in the UV range and invisible to eye). 6.3.3 Thermal-Curable Resists Thermal-curable or thermosetting polymers can be excellent resist systems for NIL due to the possibility of low pressure imprinting and good mechanical integrity after cross-linking by thermal treatment. A very attractive method for achieving room temperature and low pressure imprinting is to utilize a liquid precursor that can be cured and solidified by heating or UV light. For such materials, the requirements for low modulus and low viscosity during NIL are naturally satisfied. Because of the low viscosity of the monomer fluid, these imprinting processes are less sensitive to the effects of pattern density reported for NIL. Another advantage is that both processes allow the use of a small-area mold to pattern a large-area substrate by a step-and-repeat process. This is because UV-cured process can normally take place at room temperature and the thermally cured resist become cross-linked and therefore thermally stable. The shrinkage of the resist after cross-linking must be very small to ensure the fidelity of pattern transfer and to prevent possible film delamination. These liquid materials can be imprinted in a short period of time under low pressure and temperature. Among many candidate material systems, PDMS is a class of thermal-curable material widely used by many research groups, mainly in the context of soft lithography. In addition to its well-known transparency to UV and visible light along with its high biocompatibility, PDMS is a low surface energy (20 mN m–1 ) material, which allows an easy mold releasing without causing any structural damage to the imprinted structures. Moreover, it posses a very high resistance to oxygen plasma. However, its low modulus after curing (∼2 MPa for commercial Sylgard 184) impedes a good pattern definition at nanoscale. This can be understood by considering the radius

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.3 NIL RESIST

143

Figure 6.8. Nanoimprint lithography results using (a) 250-nm-linewidth PDMS-b-PS grating. (b) Silicon nitride mold with 70-nm-wide trenches. (c ) Seventy-nanometer linewdith grating in PDMSg -PMIA imprinted using the mold shown in (b). (d ) Strong light diffraction of imprinted 200-nm period grating in PDMS-g -PMIA copolymer on a 4-in. wafer immersed in water. (Reprinted with permission from [4] and the Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

144

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

of curvature of a cured PDMS with a modulus E and surface energy (γ ), which is r = γ /E [65]; too small a modulus will result in a large radius of curvature, which will limit the ability to produce sharply defined nanoscale features. The low modulus also causes the lateral collapsing of the imprinted structures when replicating patterns smaller than 500 nm. A higher modulus (∼8 MPa) hard PDMS was developed by Schmid et al. [66] to achieve the patterning of nanoscale structures for soft lithography. Unfortunately, the imprinting cycle (heating–cooling time) for this hard PDMS was too long (about 2 h) for it to be used as a NIL resist. Following a similar strategy, Malaquin et al. [67] used a reformulated PDMS (prepolymer with a smaller chain length) as a thermal-curable resist for NIL. The shorter chain length of the prepolymer utilized in this formulation provided the stiffness necessary to replicate gratings with linewidths as small as 200 nm. However, the cross-linking time is about 60 min at 100◦ C and 15 min at 150◦ C using a printing pressure of 10 bar; and still structures with aspect ratio greater than 1 could not be obtained when replicating the 200-nm linewidth structures [68]. Recently Pina et al. [69] reported a fast, thermal-curable liquid resist that can be imprinted under a low pressure with a very high precision and speed. This system is based on the same hydrosilylation chemistry as used in commercial Sylgard material and consists of four basic chemical components: a vinyl-terminated PDMS polymer, a silyl-hydride (SiH) based dimethylsiloxane cross-linker, a platinum catalyst, and an inhibitor. The high Si content in this polymer system guarantees that the resist has high etching resistance in RIE processes. The modulus of the cross-linked material is a critical factor for NIL. Ideal imprinting requires a careful balance between the ease of mold separation and the mechanical integrity of the imprinted structures, especially those with high aspect ratios. A resist with a high modulus allows high aspect ratio patterns to be replicated without lateral collapsing resulted from the capillary force, but they are also brittle and may crack and break during mold separation. A resist with a low modulus allows for a clean mold separation, but the replicated structures may suffer from collapsing. In general the modulus of a cured polymer material depends on the cross-linking density. The cross-linking density and therefore the modulus of the cured PDMS can be adjusted by tuning the molecular mass as well as the SiH concentration of the crosslinker. Figure 6.9 (data obtained by Dynamical Mechanical Analysis, DMA) shows that a high modulus material can be achieved by employing either a prepolymer with smaller degree of polymerization (DP) (e.g., DP 10 vs. 25) or a cross-linker with a higher amount of SiH functional groups (e.g., 1.56% vs. 1.05%). Both strategies increase the cross-linking density per unit volume of the cured material. The liquid resist allows a uniform film formation on a silicon wafer by simple spin-coating. The imprinting pressures is in the range 20–100 psi at room temperature and the resist is cross-linked after being heated above 80◦ C within 1 min, and 120◦ C within 10 s. The short time for the curing the material is in high contrast with that of commercial PDMS, which requires curing for at least tens of minutes and sometimes hours. With the increased modulus, structures with feature size range from micrometers down to 70 nm have been successfully replicated (Figure 6.10). Due to the low surface tension of the patterned film, the demolding process is

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.3 NIL RESIST

145

13 Cross-linker 1: 1.05 % H of SiH

12

Cross-linker 2: 1.56 % H of SiH

Modulus (MPa)

11 10 9 8 7 6 5 4 0

1000 2000 3000 4000 5000 Molecular weigth of vinyl prepolymer (g mol−1)

Figure 6.9. Modulus of the cured material as a function of molecular weight of vinyl-terminated polydimethylsiloxane polymers. (Reprinted with permission from [69] and the Wiley-VCH Verlag GmbH & Co.)

(a)

(c)

(b)

(d )

Figure 6.10. SEM micrographs of imprinted and thermal-cured polydimethylsiloxane structures. (a) Microscale patterns using resist with DP 65. (b) Patterns with a 350-nm linewidth and spacing, using resist with DP 25. (c ) One-hundred-twenty-nanometer linewidth patterns, using a resist with DP 10. (d ) Seventy-nanometer linewidth patterns using a resist with DP 10 and higher SiH cross-linker. (Reprinted with permission from [69] and the Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

146

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

quite easy even for high aspect ratio features and the imprinted resist surface is very smooth. Importantly, the mold separation does not require cooling to room temperature, thereby increasing the process throughput. 6.3.4 UV-Curable Resist The high temperature and pressure required for the nanoimprinting of thermoplastic materials could limit the throughput and the application scope of the NIL technique. In addition, the thermal expansion mismatch between the mold and the substrate incurred in a thermal cycle often presents an obstacle for pattern alignment over large substrate. An alternative is to use liquid precursors that can be UV cured. The liquid resist developed for S-FIL process is a multicomponent solution containing a photoinitiator, a monomer with high Si content to provide O2 RIE etch resistance, a difunctional monomer to allow cross-linking, and a low molecular weight monomer to reduce solution viscosity [70]. This resist formulation is referred as a etch barrier layer, which is the layer to be imprinted in S-FIL process. After patterning, this etch barrier layer is used as a mask to transfer the pattern into a high aspect ratio pattern into an undercoating polymer resist layer. The organic monomer, n-butyl acrylate, serves as a solvent and a mass-persistent component in the etch barrier formulation. This assists in minimizing shrinkage during polymerization. The silylated monomer provides etch resistance in the O2 RIE etch that transfers the low aspect ratio, high resolution relief structures into high aspect ratio features in the undercoating resist layer. The cross-linker provides mechanical stability to the cured etch barrier relief structure and also serves to improve the cohesive strength of the etch barrier, which are both necessary for clean and reproducible mold separation. The resist formulation does not use metal catalyst for curing, which makes it very attractive for integrated circuit applications where metal contaminations could affect or degrade the performance of semiconductor devices. The free radical polymerization of acrylic and methacrylic monomers used in S-FIL liquid resist can suffer from oxygen sensitivity issue: dissolved oxygen scavenges free radical species and thus inhibits polymerization process at the resist surface at the onset of exposure that prolongs the required exposure time. Oxygen diffusion from ambient causes a thin perimeter of undercured material surrounding the mold. Detailed kinetics study has been carried out to evaluate the impact of oxygen in S-FIL [71, 72]. Furthermore, the acrylate-based UV-imprint resist has a large shrinkage upon curing (∼10%), which may affect the pattern definition or resist adhesion on certain substrates, especially metals and plastics. An alternative formulation using vinyl ether has been investigated because it is based on cationic polymerization and is less sensitive to oxygen [73]. For many applications, the droplet dispensing method used in S-FIL to apply the liquid resist on a substrate could significantly limit the throughput of the nanopatterning process. It is highly desirable to be able to spin coat a uniform liquid resist on a large-area substrate. Cheng et al. [74] have reported a UV-curable epoxysilicone material based on cationic cross-linking of cycloaliphatic epoxies. This resist combines a number of desired features for nanoimprinting. Since cationic polymerization is not prone to

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

147

6.3 NIL RESIST

Me (a)

O Me

2

Si o Me

(c)

(b)

10µm

Me O

Si

200nm

O

Si 4

(d)

100nm

Figure 6.11. (a) Structures of monomer and cross-linker molecules. SEM micrographs of imprinted and UV-cured resist patterns: (b) 20-µm-diameter recessed patterns. (c ) Sub-100-nmtrench patterns and (d ) 20-nm trenches. (Reproduced with permission from [74] and the WileyVCH Verlag GmbH & Co.)

oxygen inhibition as compared to the free radical polymerization of the acrylate monomers, fewer defects are expected. The resist exhibits very good dry etching resistance due to high silicon content. Furthermore, its very low shrinkage after curing, only a fraction of the acrylate system, allows a reliable patterning. The liquid UVcurable resist consists of a silicone-diepoxy monomer, a silicone cross-linking agent, and a photoacid generator. Organic solvents, such as propylene glycol monomethyl ether acetate, can be used to adjust the viscosity of the resist so that film thickness ranging from 1 µm down to sub-50 nm can be readily obtained. Although direct spin-coating of this resist on Si or oxide substrate causes dewetting of the thin liquid film, as one would expect for a very low viscosity liquid, it was found that by using a suitable undercoating polymer layer, such as baked PMMA or photoresist such as SU-8 resist to provide better matching of interfacial energies, stable and uniform liquid thin films can be formed on Si or any other substrates. This capability is critical to imprinting large-area sample (e.g., wafer size) in a single step. Due to its low viscosity, both micro- and nanoscale patterns (Figures 6.11b–d) can be imprinted at room temperature and at a pressure of less than 0.1 MPa using a conventional contact exposure tool. The UV-cured resist has very desirable plasma etching characteristics, i.e., a very high resistance for O2 plasma etching, making it suitable as etch mask for pattern transferring into any underlying organic layers. The UV-curable resists are also commercially available from companies such as Molecular Imprints, Nanonex, and Microresist. For example mr-L 6000 resist from Microresist is essentially a chemically amplified negative tone photoresist sensitive to near-UV exposure, composing of a multifunctional epoxidized novolak resin and a photoacid generator. Upon UV exposure, the acids initiates the cationic polymerization of the epoxy resin forming a tightly cross-linked network [75].

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

148

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

6.3.5 Other Imprintable Materials It should be recognized that NIL cannot only be used to form patterns in a polymer resist, it can also be extended to create desired structures in many polymer systems, especially those having special functionalities or can be used to form functional device structures directly. Guo et al. [23] demonstrated imprinting of 2D photonic crystal nanostructures in nonlinear optical polymers. Pisignano et al. [22, 76] showed that conjugated polymers and oligomers can be imprinted at room temperature and the resulting nanostructures maintained active optical properties. Room temperature processes are preferred in patterning organic semiconductors to avoid oxygen incorporation and substitution into conjugated polymers that occur at high temperature, which can cause irreversible degradation of their optical and electrical properties. Kim et al. [21] imprinted a conjugated polymer material in a solar cell device. The imprinting was performed in vacuum and the process created welldefined nanoscale interface between an electron donor and acceptor layers, which resulted in enhanced power conversion efficiency. Li et al. [16] patterned conductive polymer poly(3,4-ethylenedioxythiophene):poly(4-styrenesulphonate) (PEDOT: PSS) to use as electrodes for organic thin film transistors with high resolutions. . The process is based on a reverse-imprinting principle (to be discussed in later) and is also carried at room temperature to guarantee the material’s conductive property after the patterning process. Nielsen et al. [77] demonstrated thermal NIL of cyclic olefin R , a thermoplastic material that is highly UV transparent and chemcopolymer Topas
ically resistant to hydrolysis, acids, and organic polar solvents, making it suitable R ,a for lab-on-a-chip applications. Cheng et al. imprinted a fluoropolymer CYTOP
low-κ dielectric material but also possesses interesting property to prevent protein absorption. The authors utilized this property to pattern the motor proteins and uses the imprinted CYTOP nanostructures as physically barriers to guide the motion of microtubules with extremely high efficiency [78]. Biodegradable polymers are attractive for many biomedical applications, such as DNA and protein analysis chips, as well as for supporting structures in tissue engineering. One popular example is Poly(L-lactic acid) (PLA), which has a glass transition temperature of ∼60◦ C. Hirai et al. [54] has demonstrated imprinting of commercial PLA plastics at 75◦ C to produce nanochannels and nanoscale hole arrays. His group has also shown imprinting of spin on glass for optical applications. Matsui et al. [79, 80] have demonstrated room temperature NIL by using another type of spin on glass, hydrogen silsequioxane (HSQ), as the imprint resist material. The pressure needed for these inorganic materials was very high because of their large modulus and high viscosity. On the other hand, its hydrophilic surface property (like glass) and the ability to be nanopatterned at room temperature, makes it an ideal candidate material for creating nanofluidic channels. Cheng et al. developed a technique to form such channels by a direct imprinting of hydrophilic HSQ material and sealing with another HSQ thin film, with all processing steps performed near room temperature (Figure 6.12). The width of the channel is determined by the mold feature and the depth can be controlled by imprinting pressure and imprinting time [81]. The capability of creating nanoscale fluidic structures allows fundamental studies of ion or molecule transport in confined spaces [82]. The studies of nanofluidics could result in some practical tools for the analysis of biomolecules, such as DNAs [36, 37] and proteins.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.4 THE NANOIMPRINT PROCESS

(b)

(a) (1) Spin coat HSQ on substrate HSQ

149

(3) Transfer an HSQ film from an O2 plasma treated PDMS stamp

Si substrate PDMS HSQ

(2) Direct imprinting of HSQ film with a pressure of 1000 psi at 25°C

Si substrate Mold

PDMS

Si substrate HSQ Si substrate

(c)

HSQ Si substrate

Figure 6.12. (a) Schematics of fabrication of nanofluidic channels by direct imprinting of HSQ followed by bonding another HSQ layer. (b), (c ) SEM pictures of all-HSQ nanofluidic channels of different cross-sections. (Reprinted with permission from [81].)

6.4 THE NANOIMPRINT PROCESS 6.4.1 Cavity Fill Process Now we take an in-depth look at the imprinting step itself, which is a critical step in the NIL process. In order to form the desired pattern that is conformal to the nanoscale protrusion features on the mold, the resist material underneath the protrusions must be displaced and transported to nearby trenches or cavities. An intuitive way to think about the NIL process is to imagine pushing very sharp pins (with diameter down to 10 nm) into a polymeric film. It is easy to understand that in NIL small-scale features are much easier to form than large ones, which strongly contrasts other lithographic techniques. Also from the simple squeezed flow model introduced earlier, it can be seen that the time required for imprinting scales quadratically with of the pattern size. As a result it takes significantly longer time to imprint large size patterns than nanopatterns. Therefore, the name “nanoimprint” is well deserved and very accurately illustrates the process. The resist’s flow ability is perhaps the most important property in determining the imprinting condition and the time required to complete the pattern replication. As an excellent demonstration, Xia et al. showed that by using rapid heating from a pulsed UV laser, the polymer resist can be imprinted within hundreds of nanoseconds. This ultrafast NIL process is achieved because heat is absorbed by the polymer and a very thin Si surface layer only, and the efficient heating increases the polymer temperature well beyond the T g of the material. As a result, drastically reduced viscosity facilitates the molten polymer to fill in the mold cavities very quickly. Yu et al. developed a technique based on

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

150

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

860 nm

time-resolved scatterometry to perform in situ real-time monitoring of the cavity filling process in NIL [83]. The authors used a surface relief grating structure as the imprinting mold and monitored the diffracted light continuously as the mold penetrates into the polymer resist. The results match the simulation very well, indicating this technique a powerful tool for in situ NIL process control. Hirai et al. [84] was the first to investigate the polymer deformation process by numerical simulations . The simulation results were compared with the those obtained from the NIL experiments and the findings summarized in [85]. They have studied the pressures required for successful imprinting and the filling rate into the mold grooves as a function of the aspect ratio of the pattern, the initial thickness of the polymer, and the duty cycle of the structure for a periodic pattern. They found that the required pressure increases not only for high aspect ratio patterns but for low aspect ratio pattern as well. This is because for wide trenches, the pressure is not evenly distributed due to the polymer flow resistance, and the polymers fill in from the edges with a slower filling rate than at the center of the cavity. The pressure also increases when the initial thickness of the polymer film decreases to less than twice of the mold depth. This again can be attributed to the increase in the resist flow resistance in the confined “nanofluidic channels” formed between the mold protrusions and the substrate surface. These results agree very well with the experimental observations. Based on these theoretical and experimental studies, the group has successfully demonstrated the imprinting of a high aspect ratio pattern having 100 nm width and 860 nm height by using a thick polymer resist layer (Figure 6.13) Several studies by Rowland et al. [86–88] have investigated the impact of polymer material properties, embossing template geometry, and process conditions on polymer deformation and further investigated the impact of polymer deformation mode on replication time. They also found that when polymer flows vertically into an open

100 nm

Figure 6.13. High aspect ratio grating pattern imprinted into a thick polymer (4.0-µm-thick PMMA-based polymer on Si substrate, M w = 100, 000) at 170◦ C and with a pressure of 20 MPa. (Reprinted with permission from [85].)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.4 THE NANOIMPRINT PROCESS

151

cavity during embossing, the polymer can deform as a single peak centered in the cavity or as a dual peak where each peak remains close to the vertical sidewalls, depending upon geometry. The ratio of cavity width to film thickness modulates singleversus dual-peak cavity filling regardless of the absolute size of the features and also regardless of the pressure or temperature applied during embossing. The ability to displace viscous polymer and fill in cavities can be demonstrated in another example, where NIL is used to form optical waveguide directly from a very thin polymer layer. The device is a polymer-based optical microring resonator [89], which can find applications in optical communication, label-free biosensing [90], and high frequency ultrasound detection [91, 92]. In this process, an inverse pattern of the waveguide and the microring is fabricated as mold, having a depth of ∼ 2 µm (Figure 6.14a). For the NIL process, only a thin polymer layer with an initial thickness of ∼200 nm is spin-coated, but the final waveguide thickness reaches ∼2 µm (Figure 6.14b). This implies that the amount of polymers being displaced from

(a)

1 µm

(b)

Figure 6.14. SEM micrographs showing perspective view of (a) a mold used to imprint the coupled polymer waveguides, and (b) the imprinted polymer microring resonator device (insert shows the cross-section of the waveguide after a undercutting wet etch). (Reprint with permission from [4] and the Wiley-VCH Verlag GmbH & Co.)

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

152

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

nearby region and transported to the mold trench is quite significant. Such a mode of imprinting is very different from what is typically used in NIL where nanoscale mold protrusions are imprinted into a polymer resist. Consequently, the conditions for imprinting need to be modified accordingly by using higher pressure and longer imprinting time to assist the polymer flow [89].

6.5 VARIATIONS OF NIL PROCESSES In recent years various new approaches have been studied in order to enhance the capability of NIL. In S-FIL, a reverse-tone bilayer approach was developed to enhance the reliability in fabricating high aspect ratio structures [93]. Borzenko et al. [94] developed a polymer-bonding technique to reduce the temperature and pressure used in NIL and to reduce the impact of limited polymer transport on the pattern formation. Roller NIL was proposed to provide better uniformity and decrease the printing force, and demonstrated sub-100-nm feature transfer by using a rod roller (rather than a platen) to apply pressure to a piece of Si mold [95]. Rigiflex mold was developed to provide better conformal contact with the substrate and to reduce the pressure required during imprinting [41]. In principle, this approach can be extended for roll-to-roll imprinting of nanostructures. Later in this section we will describe a recently demonstrated true roll-to-roll nanoimprint process.

6.5.1 Reverse Nanoimprint A reverse nanoimprinting technique was developed to address the issue of patterning on topographies and on flexible substrate. Reverse nanoimprint is based on the following consideration. If a polymer film is spin coated onto a mold, the polymer will fill up the trench regions of the surface relief patterns. This means that a replica of the mold pattern is formed in the polymer film simply by spin-coating. Now if this film can be transferred from the mold to a substrate, patterned structures are obtained (Figure 6.15a). Figure 6.15b shows the imprinted 350-nm line-spacing PMMA grating obtained by reverse imprint process at a temperature of 105◦ C, i.e., at the T g of PMMA. The key to the successful film transfer lies in the fact that the mold has a lower surface energy than that of the substrate, so the polymer film has stronger adhesion to the substrate and therefore can be detached from the mold. Because in reverse nanoimprinting coating of polymer resist on the substrate is not required, it is therefore possible to use this technique to transfer patterns onto substrates that are not suitable for spin-coating or have surface topographies (Figure 6.16a). The reverse imprinting technique also offers a simple method to fabricate 3D polymer nanostructures by simply repeating the reverse imprinting process to build up the structure in a layer-by-layer fashion (Figure 6.16b) [96, 97]. By using two molds, Kong et al. [98] showed another way of constructing simple 3D structures using the reverse imprinting principle. Figure 6.17a shows the reverse imprinted polycarbonate grating structure that is suspended over etched features on Si substrate;

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.5 VARIATIONS OF NIL PROCESSES

(a)

153

Mold Polymer Imprint

Substrate

Separate

(b)

Figure 6.15. (a) Schematic of reverse nanoimprint.(b) SEM of reverse imprinted PMMA gratings with 350-nm linewidth/spacing. (Reprint with permission from [4] and the Wiley-VCH Verlag GmbH & Co.)

Figure 6.17b shows a stacked polymer grating structures created by three successive reverse imprinting processes; and Figure 6.17c shows a large-area (10 cm size) PMMA grating reverse imprinted over a PVC thin film. The results showed high yield and little defects. Strong rainbow colors can be observed from the flexible film due to the diffraction grating structures on its surface. The reverse nanoimprinting method can be operated in another mode when the depth of the mold feature is large as compared with the coated polymer film. In this case, the coated polymer cannot planarize the relief features on the mold surface, and

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

154

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

Surfactanttreated mold

(a)

Film thickness

Suspended polymer structure

Prepatterned substrate

(b)

Repeat

Reversal imprint O2 RIE to remove residue

First polymer Second layer layer

Third layer

Figure 6.16. Schematics of (a) reverse imprinting polymer nanostructures over topography and (b) creating stacked 3D polymer structures . (Reprinted with permission from [96].)

when pressed against a substrate only the material on top of the mold protrusion features are transferred to the substrate [99, 100]. It is similar to the inking process but the stamp has nanoscale surface relief features. Polymer inking method is essentially an additive patterning technique, which does not require further processing steps. This feature is very attractive for patterning polymers with special functionalities

(a)

(b)

(c)

Figure 6.17. SEM micrograph of (a) 700-nm period grating in polycarbonate reverse imprinted on topographies and (b) reverse-imprinted three-layer polymer grating structure. (c ) Sevenhundred-nanometer period grating in PMMA reverse imprinted over a 10-cm size flexible PVC thin film, showing strong and uniform light diffraction across the whole imprinted area.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.5 VARIATIONS OF NIL PROCESSES

155

that are sensitive to chemical or plasma processing. Such examples include conductive polymer and conjugated organic materials. Recently, conductive polymer PEDOT has been patterned using the polymer inking technique to function as electrodes for organic electronic applications [16]. Lee et al. [101] used a similar technique to pattern bilayer of metal and polymer. 6.5.2 Combined Nanoimprint and Photolithography NIL works effectively for nanoscale features but experiences difficulty when replicating large features. On the other hand, a general lithographic technique should be capable of producing both large and small features with various combinations and pattern densities. Previous studies have shown that defects and even pattern failures in the form of incomplete pattern transfer can occur due to the high viscosity of the polymer melt and the varied pattern densities on the mold [102]. A mix-andmatch lithography approach based on two separate lithography steps have been proposed to address this problem [103], with the expense of increased processing steps and complexity and the requirement of pattern registration between the two separate steps. To solve the problem, Cheng et al. [104] developed a technique by introducing a hybrid mask concept and combining nanoimprint and photolithography (CNP). As shown in the Figure 6.18a, the hybrid mold is made of UV-transparent material and act both as a NIL mold and as a photolithography mask. Protrusions are made on the mold for imprinting nanoscale features, while metal pads are embedded into the mold that serve as metal mask for photolithography to replicate the large patterns. Detailed fabrication procedures can be found in [106]. In the CNP process, the hybrid mold is first imprinted into the resist layer by pressure, then the whole mold–substrate assembly is exposed by UV radiation. After the hybrid mold and the substrate are separated, the substrate is immersed in a developer solution to remove the unexposed resist that are blocked by the metal pads. After developing, both large and nanoscale patterns are created in the polymer resist in one step. The authors have demonstrated the effectiveness of this technique by using a negative tone photoresist and have also fabricated nanoelectrode structures with drastically different length scales from 150 µm down to 10 nm (Figure 6.18b). Such structures are used in making nanoscale organic thin film transistor [106]. The CNP process offers several advantages. First, the nanoscale protrusion features on the hybrid mold only need to displace a very small amount of polymer, which ensures a low pressure process. Second, by forming the large patterns as a photomask (i.e., making them as metal pads), the residue layer thickness distribution is simplified, which can significantly ease residue removal step. Third, as a further improvement, if the metal mask used for etching the mold is left on top of the mold protrusions, it can prevent the exposure of the resist layer underneath. The unexposed residual layer can be removed in a developer solution without additional O2 dry etching step [105]. Therefore, it is possible to eliminate the separate residual removal step in NIL completely, and could simultaneously solve the nonuniform residual layer thickness problem altogether. In addition, the metal layer on the hybrid mold used in CNP serves as better registration marks as compared with the only relief structures used in S-FIL or NIL. Because the

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

156

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

(a)

(b)

UV exposure

After developing

Figure 6.18. (a) Schematics of the CNP technique by using a hybrid mask mold for one-step lithography of both large- and nanopatterns. (Reproduced with permission from [105].) (b) SEM micrographs of a finger-shaped nanoelectrode with large metal pads patterned by the CNP approach. (Reproduced with permission from [106].)

latter cannot create sufficient refractive index contrast upon intimate contact with a polymer layer that has an index close to that of the mold, and therefore registration between different lithographic layers is difficult. To take the advantage of the high contrast alignment marks in CNP, one can design grating patterns on the mold and on substrate and use Moir´e interference technique to achieve high accuracy alignment. 6.5.3 Roll-to-Roll Nanoimprint Lithography (R2RNIL) Though NIL is considered as one of the most promising and competitive technologies for high throughput and low cost nanopatterning, the current process and throughput (approximately several minutes or longer per Si wafer) is still far from meeting the demands of many practical applications, especially in the area of organic electronics and biotechnologies. To meet these demands, faster and more economical fabrication

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.5 VARIATIONS OF NIL PROCESSES

157

method is necessary. A continuous roll-to-roll nanoimprint technique can provide a solution for high speed large-area nanoscale patterning with greatly improved throughput; furthermore, it can overcome the challenges faced by conventional NIL in maintaining pressure uniformity and successful demolding in large-area printing. In the conventional approach, embossing a large area requires very large force. Huge contact area between the mold surface and the imprinted nanostructures also produce significant adhesion force, making the mold sample separation difficult or even impossible without damaging the substrate. In thermal NIL, if the mold and substrate are made of materials with different thermal expansion coefficients such as Si mold and polymer substrate, stress can build up during a thermal cycle with such a magnitude that even destroys the Si mold during mold releasing. R2RNIL provides a unique solution to these challenges encountered in the conventional wafer scale NIL process, because imprinting in R2RNIL proceeds in a narrow region transverse to the web moving direction, thus requires much smaller force to replicate the patterns. Also, since the mold used in R2RNIL is in the form of a roller, the mold sample separation proceeds in a “peeling” fashion, which requires much less force and reduces the probability of defect generation. Figure 6.19 shows the overall configuration of a continuous R2RNIL process, which consists of three separate processing steps: (1) coating process, (2) imprinting and separating process, and (3) any of the subsequent processes. As an example, the last step in this schematic represents a continuous metal coating process for making structures such as metal wire-grid polarizers [28, 29] or transparent wire-grid electrode [107].

Figure 6.19. Schematic of a R2RNIL apparatus consisting of (1) coating unit, (2) imprinting unit, and (3) subsequent processing unit .

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

158

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

True roll-to-roll nanoimprinting has been a challenge to the community because it requires a complete set of solutions to address a number of interrelated material issues. Firstly, a special roller mold is required for continuous roll-to-roll imprinting of nanostructures. Mold used in R2RNIL should be flexible enough to be wrapped on the roller surface, and it should also have sufficient modulus and strength to be able to imprint other materials. Secondly, liquid resists having good coating property and low viscosity should be used to ensure fast imprinting; and they should be cured quickly with minimal shrinkage. Therefore, conventional resist materials dissolved in solvents and requiring additional baking process are not suitable for the R2RNIL process. Two types of liquid resists introduced earlier were exploited for the R2RNIL application. For thermal R2RNIL, the fast thermal-curable liquid PDMS resist [69] was used. For even higher speed R2RNIL, the UV-curable low viscosity epoxysilicone [74] was used as imprint resist. Also two types of coating methods, reverse and forward web coating, were used to provide a continuous and uniform resist coating on the plastic web from a resist container by using a coating roller. The resist thickness is controlled by using a doctor blade. The imprint module is composed of an imprint roller, two backup rollers, a release roller, and a curing section. A flexible copolymer of ethylene and tetrafluoroethylene (ETFE), is used as mold material and wrapped around a stainless steel roller. ETFE has high modulus (∼1.2 GPa) at room temperature but can be softened at elevated temperature. An ETFE mold can be easily replicated from an original Si mold by a thermal NIL process at 200◦ C, where the original Si mold was fabricated by laser interference lithography. Moreover, the exceptional antisticking property of ETFE (surface energy of 15.6 dyn cm–1 , cf. PDMS ∼19.6 dyn cm–1 ) makes it easy to demold after imprinting without any mold surface treatment and without deterioration in surface properties over long imprinting cycles. In the demonstrated roller imprinting process the pressure is applied by means of web tension and forces from the backup rollers. Moreover, the two backup rollers with spring system also guarantee nonslip motion in the rolling process, which is very important for successful pattern replication. In the next step, resist precursor is cured either by convection heating or UV irradiation. Finally, poly(ethylene terephthalate) (PET) substrate with roller-imprinted nanostructures continuously separates from the roller mold via the release roller. For easy visualization of the imprinting results, a grating pattern of 700-nm period was chosen because well-replicated grating structure should show strong light diffraction and the pattern quality can be easily examined by eye. Three-hundrednanometer linewidth and 700-nm period gratings imprinted using the thermally cured PDMS and UV-cured epoxysilicone on PET substrate are shown in Figures 6.20a and b, respectively. Scanning electron microscopy (SEM) shows that the UV-cured resist pattern has higher quality than the thermal-cured PDMS, likely due to the lower viscosity of the material that facilitates the fast filling of the mold cavity. Printing speed can be adjusted depending on the period of grating pattern and its aspect ratio. The fast curing of the resist material enabled a web speed of ∼3.5 ft. min–1 . Figure 6.20c shows UV R2RNIL result of a 570-mm-long, 700-nm period grating structure created on PET substrate with bright light diffraction. High aspect ratio or denser gratings results in larger contact area and therefore stronger adhesion force between the EFTE mold and the imprinted resist pattern.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

6.6 CONCLUSION

(a)

159

(c)

(b)

Figure 6.20. Nanogratings fabricated by R2RNIL. SEM micrographs of 700-nm period, 300-nm linewidth gratings imprinted on PET flexible substrate by using (a) thermally cured PDMS, and (b) UV-cured epoxysilicone. (c ) Photograph showing bright light diffraction from a section of a 570-mm PET strip imprinted with 700-nm grating pattern.

In order to produce these patterns successfully, the PET surface needs to be plasma and chemically treated to increase its adhesion to the cured epoxysilicone. Figures 6.21a and b show the 700-nm period grating with aspect ratio of 5.4 on the original Si mold and the epoxysilicone grating pattern replicated from the ETFE mold. Very faithful pattern transfer can be observed, even down to the fine details at the bottom of the trenches Figures 6.21c and d show the replicated 70-nm linewidth grating in epoxysilicone resist having 200-nm and 100-nm period, respectively. Such a high speed nanoimprint capability could lead to many interesting applications of largearea polymer nanostructures.

6.6 CONCLUSION Nanoimprint technique has enabled parallel nanoscale patterning with ultrahigh resolution. The simplicity of this method has made it appealing to researchers in various fields. We provided an overview of some aspects of the rapid development in the field of nanoimprint technology and especially its material development in recent years. New techniques based on the concept of mechanical printing or embossing are appearing very rapidly. New variations of the technique aimed at different applications have also been developed at a fast pace. It is certain that advances in new materials for imprinting will fuel the development in this field. One exciting opportunity

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

160

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

(a)

(c)

(b)

(d )

Figure 6.21. High aspect ratio grating structures fabricated by UV R2RNIL. SEM micrographs of (a) the original Si mold, (b) epoxysilicone gratings replicated from the ETFE mold, (c ) 200-nm period, 70-nm linewidth epoxysilicone pattern, and (d ) 100-nm period, 70-nm linewidth epoxysilicone pattern fabricated by UV R2RNIL.

is to develop roll-to-roll nanoimprinting of nanostrucutures, which provides an unprecedented throughput for many practical applications. We wish to inspire more researchers to explore this technique and to exploit many new possibilities in the future.

REFERENCES 1. Gates, B. D., Xu, Q. B., Stewart, M., Ryan, D., Willson, C. G., and Whitesides, G. M. (2005) New approaches to nanofabrication: molding, printing, and other techniques. Chem. Rev. 105, 1171–1196. 2. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1995) Imprint of sub-25 nm vias and trenches in polymers. Appl. Phys. Lett. 67, 3114–3116. 3. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1996) Imprint lithography with 25nanometer resolution. Science 272, 85–87. 4. Guo, L. J. (2007) Nanoimprint lithography: methods and material requirement. Adv. Mater. 19, 495–513. 5. Chou, S. Y., Krauss, P. R., Zhang, W., Guo, L. J., and Zhuang, L. (1997) Sub-10 nm imprint lithography and applications. J. Vac. Sci. Technol. B 15, 2897–2904. 6. Ruchhoeft, P., Colburn, M., Choi, B., Nounu, H., Johnson, S., Bailey, T., Damle, S., Stewart, M., et al. (1999) Patterning curved surfaces: template generation by ion beam

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

REFERENCES

7.

8.

9. 10.

11.

12.

13.

14. 15.

16. 17. 18. 19.

20.

21.

22.

161

proximity lithography and relief transfer by step and flash imprint lithography. J. Vac. Sci. Technol. B 17, 2965–2969. Stewart, M. D., Johnson, S. C., Sreenivasan, S. V., Resnick, D. J., and Willson, C. G. (2005) Nanofabrication with step and flash imprint lithography. J. Microlithogr. Microfab. Microsyst. 4, 011002. Resnick, D. J., Dauksher, W. J., Mancini, D., Nordquist, K. J., Bailey, T. C., Johnson, S., Stacey, N., Ekerdt, J. G., et al. (2003) Imprint lithography for integrated circuit fabrication. J. Vac. Sci. Technol. B 21, 2624–2631. Special Report (2003) Technol. Rev. 106, 36. Bailey, T. C., Resnick, D. J., Mancini, D., Nordquist, K. J., Dauksher, W. J., Ainley, E., Talin, A., Gehoski, K., et al. (2002) Template fabrication schemes for step and flash imprint lithography. Microelectron. Eng. 61–62, 461–467. Dauksher, W. J., Nordquist, K. J., Mancini, D. P., Resnick, D. J., Baker, J. H., Hooper, A. E., Talin, A. A., Bailey, T. C., et al. (2002) Characterization of and imprint results using indium tin oxide-based step and flash imprint lithography templates. J. Vac. Sci. Technol. B 20, 2857–2861. Bailey, T., Choi, B. J., Colburn, M., Meissl, M., Shaya, S., Ekerdt, J. G., Sreenivasan, S. V., and Willson, C. G. (2000) Step and flash imprint lithography: template surface treatment and defect analysis. J. Vac. Sci. Technol. B 18, 3572–3577. Fuchs, A., Vratzov, B., Wahlbrink, T., Georgiev, Y., and Kurz, H. (2004) Interferometric in situ alignment for UV-based nanoimprint. J. Vac. Sci. Technol. B 22, 3242– 3245. McAlpine, M. C., Friedman, R. S., and Lieber, D. M. (2003) Nanoimprint lithography for hybrid plastic electronics. Nano Lett. 3, 443–445. Cedeno, C. C., Seekamp, J., Kam, A. P., Hoffmann, T., Zankovych, S., Torres, C. M. S., Menozzi, C., Cavallini, M., et al. (2002) Nanoimprint lithography for organic electronics. Microelectron. Eng. 61–62, 25–31. Li, D. and Guo, L. J. (2006) Micron-scale organic thin film transistors with conducting polymer electrodes patterned by polymer inking. Appl. Phys. Lett. 88, 63513. Guo, L. J., Krauss, P. R., and Chou, S. Y. (1997) Nanoscale silicon field effect transistors fabricated using imprint lithography. Appl. Phys. Lett. 71, 1881–1883. Zhang, W. and Chou, S. Y. (2003) Fabrication of 60-nm transistors on 4-in. wafer using nanoimprint at all lithography levels. Appl. Phys. Lett. 83, 1632–1634. Macintyre, D. S., Chen, Y., Gourlay, D., Boyd, E., Moran, D., Cao, X., Elgaid, K., Stanley, C. R., et al. (2003) Nanoimprint lithography process optimization for the fabrication of high electron mobility transistors. J. Vac. Sci. Technol. B 21, 2783–2787. Pisignano, D., Persano, L., Mele, E., Visconti, P., Anni, M., Gigli, G., Cingolani, R., Favaretto, L., et al. (2005) First-order imprinted organic distributed feedback lasers. Synth. Met. 153, 237–240. Kim, M.-S., Kim, J.-S., Cho, J., Shtein, M., Guo, L. J., and Kim, J. (2007) Flexible conjugated polymer photovoltaic cells with controlled heterojunctions fabricated using nanoimprint lithography. Appl. Phys. Lett. 90, 123113. Pisignano, D., Persano, L., Raganato, M. F., Visconti, P., Cingolani, R., Barbarella, G., Favaretto, L., and Gigli, G. (2004) Room-temperature nanoimprint lithography of nonthermoplastic organic films. Adv. Mater. 16, 525.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

162

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

23. Guo, L. J., Cheng, X., and Chao, C. Y. (2002) Fabrication of photonic nanostructures in nonlinear optical polymers. J. Mod. Opt. 49, 663–673. 24. Cheng, X., Hong, Y. T., Kanicki, J., and Guo, L. J. (2002) High-resolution organic polymer light-emitting pixels fabricated by imprinting technique. J. Vac. Sci. Technol. B 20, 2877–2880. 25. Kao, P. C., Chu, S. Y., Chen, T. Y., Zhan, C. Y., Hong, F. C., Chang, C. Y., Hsu, L. C., Liao, W. C., et al. (2005) Fabrication of large-scaled organic light emitting devices on the flexible substrates using low-pressure imprinting lithography. IEEE Trans. Electron Devices 52, 1722–1726. 26. Wang, J. and Kostal, H. (2005) Diffractive optics—nanoimprint lithography enables fabrication of subwavelength optics. Laser Focus World 41, 76. 27. Wang, J. J., Zhang, W., Deng, X. G., Deng, J. D., Liu, F., Sciortino, P., and Chen, L. (2005) High-performance nanowire-grid polarizers. Opt. Lett. 30, 195–197. 28. Ahn, S. W., Lee, K. D., Kim, J. S., Kim, S. H., Park, J. D., Lee, S. H., and Yoon, P. W. (2005) Fabrication of a 50 nm half-pitch wire grid polarizer using nanoimprint lithography. Nanotechnology 16, 1874–1877. 29. Ekinci, Y., Solak, H. H., David, C., and Sigg, H. (2006) Bilayer Al wire-grids as broadband and high-performance polarizers. Opt. Express 14, 2323–2334. 30. Chou, S. Y. (1997) Patterned magnetic nanostructures and quantized magnetic disks. Proc. IEEE 85, 652–671. 31. Wu, W., Cui, B., Sun, X. Y., Zhang, W., Zhuang, L., Kong, L. S., and Chou, S. Y. (1998) Large area high density quantized magnetic disks fabricated using nanoimprint lithography. J. Vac. Sci. Technol. B 16, 3825–3829. 32. Martin, J. I., Nogues, J., Liu, K., Vicent, J. L., and Schuller, I. K. (2003) Ordered magnetic nanostructures: fabrication and properties. J. Magn. Magn. Mater. 256, 449–501. 33. Glinsner, T., Hangweier, P., Luesebrink, H., Dorsey, P., Homola, A., and Wachenschwanz, D. (2005) Nanoimprint lithography enables patterned tracks for high-capacity hard disks. Solid State Technol. 48, 51. 34. McClelland, G. M., Hart, M. W., Rettner, C. T., Best, M. E., Carter, K. R., and Terris, B. D. (2002) Nanoscale patterning of magnetic islands by imprint lithography using a flexible mold. Appl. Phys. Lett. 81, 1483–1485. 35. Hu, Z. J., Baralia, G., Bayot, V., Gohy, J. F., and Jonas, A. M. (2005) Nanoscale control of polymer crystallization by nanoimprint lithography. Nano Lett. 5, 1738–1743. 36. Cao, H., Yu, Z. N., Wang, J., Tegenfeldt, J. O., Austin, R. H., Chen, E., Wu, W., and Chou, S. Y. (2002) Fabrication of 10 nm enclosed nanofluidic channels. Appl. Phys. Lett. 81, 174–176. 37. Guo, L. J., Cheng, X., and Chou, C. F. (2004) Fabrication of size-controllable nanofluidic channels by nanoimprinting and its application for DNA stretching. Nano Lett. 4, 69–73. 38. Hoff, J. D., Cheng, L. J., Meyhofer, E., Guo, L. J., and Hunt, A. J. (2004) Nanoscale protein patterning by imprint lithography. Nano Lett. 4, 853–857. 39. Falconnet, D., Pasqui, D., Park, S., Eckert, R., Schift, H., Gobrecht, J., Barbucci, R., and Textor, M. (2004) A novel approach to produce protein nanopatterns by combining nanoimprint lithography and molecular self-assembly. Nano Lett. 4, 1909–1914. 40. Hu, W., Yim, E. K. F., Reano, R. M., Leong, K. W., and Pang, S. W. (2005) Effects of nanoimprinted patterns in tissue-culture polystyrene on cell behavior. J. Vac. Sci. Technol. B 23, 2984–2989.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

REFERENCES

163

41. Suh, D., Choi, S. J., and Lee, H. H. (2005) Rigiflex lithography for nanostructure transfer. Adv. Mater. 17, 1554–1560. 42. Li, M. (2003) PhD Thesis, Princeton University, Princeton, USA. 43. Taniguchi, J., Tokano, Y., Miyamoto, I., Komuro, M., and Hiroshima, H. (2002) Diamond nanoimprint lithography. Nanotechnology 13, 592–596. 44. Lucas, B. D., Kim, J., Chin, C., and Guo, L. J. (2008) Nanoimprint lithography based approach for the fabrication of large-area, uniformly-oriented plasmonic arrays. Adv. Mater., 20, 1129–1134. 45. Guo, L. J. (2004) Recent progress in nanoimprint technology and its applications. J. Phys. D:Appl. Phys. 37, R123–R141. 46. Jung, G. Y., Li, Z. Y., Wu, W., Chen, Y., Olynick, D. L., Wang, S. Y., Tong, W. M., and Williams, R. S. (2005) Vapor-phase self-assembled monolayer for improved mold release in nanoimprint lithography. Langmuir 21, 1158–1161. 47. Schift, H., Saxer, S., Park, S., Padeste, C., Pieles, U., and Gobrecht, J. (2005) Controlled co-evaporation of silanes for nanoimprint stamps. Nanotechnology 16, S171–S175. 48. Khang, D. Y. and Lee, H. H. (2004) Sub-100 nm patterning with an amorphous fluoropolymer mold. Langmuir 20, 2445–2448. 49. Khang, D. Y., Kang, H., Kim, T., and Lee, H. H. (2004) Low-pressure nanoimprint lithography. Nano Lett. 4, 633–637. 50. Choi, D. G., Jeong, J. H., Sim, Y. S., Lee, E. S., Kim, W. S., and Bae, B. S. (2005) Fluorinated organic–inorganic hybrid mold as a new stamp for nanoimprint and soft lithography. Langmuir 21, 9390–9392. 51. Stewart, M. D. and Willson, C. G. (2005) Imprint materials for nanoscale devices. MRS Bull. 30, 947–951. 52. Reuther, F. (2005) Advanced polymers and resists—a key to the development of nanoimprint lithography. J. Photopolym. Sci. Technol. 18, 525–530. 53. Sakai, N., Taniguchi, J., Kawaguchi, K., Ohtaguchi, M., and Hirasawa, T. (2005) Investigation of application availability of UV-NIL by using several types of photo-curable resin. J. Photopolym. Sci. Technol. 18, 531–536. 54. Hirai, Y. and Tanaka, Y. (2002) Application of nano-imprint lithography. J. Photopolym. Sci. Technol. 15, 475–480. 55. Hirai, Y. (2005) Polymer science in nanoimprint lithography. J. Photopolym. Sci. Technol. 18, 551–558. 56. Wu, L. and Chou, S. Y. (2003) Technical Digest of 47th EIPBN Conference, pp. 293–294. 57. Hirai, Y., Yoshida, S., and Takagi, N. (2003) Defect analysis in thermal nanoimprint lithography. J. Vac. Sci. Technol. B 21, 2765–2770. 58. Sotomayer-Torres, C. M., ed. (2003) Alternative Lithography: Unleashing the Potentials of Nanotechnology. Kluwer, New York. 59. Scheer, H. C., Bogdanski, N., Wissen, M., Konishi, T., and Hirai, Y. (2005) Polymer time constants during low temperature nanoimprint lithography. J. Vac. Sci. Technol. B 23, 2963–2966. 60. Jung, G. Y., Ganapathiappan, S., Li, X., Ohlberg, D. A. A., Olynick, D. L., Chen, Y., Tong, W. M., and Williams, R. S. (2004) Fabrication of molecular-electronic circuits by nanoimprint lithography at low temperatures and pressures. Appl. Phys. A 78, 1169–1173.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

164

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

61. Khang, D. Y. and Lee, H. H. (2000) Room-temperature imprint lithography by solvent vapor treatment. Appl. Phys. Lett. 76, 870–872. 62. Beck, M., Graczyk, M., Maximov, I., Sarwe, E. L., Ling, T. G. I., Keil, M., and Montelius, L. (2002) Improving stamps for 10 nm level wafer scale nanoimprint lithography. Microelectron. Eng. 61–62, 441–448. 63. Fu, P.-F. and Tomalia, M. K. (2004) Carboxyl-terminated isotactic polypropylene. Preparation, characterization, kinetics, and reactivities. Macromolecules 37, 267. 64. Choi, P., Fu, P.-F., and Guo, L. J. (2006) Adv. Funct. Mater., in press. 65. Hui, C. Y., Jagota, A., Lin, Y. Y., and Kramer, E. J. (2002) Constraints on microcontact printing imposed by stamp deformation. Langmuir 18, 1394–1407. 66. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. 67. Malaquin, L., Carcenac, F., Vieu, C., and Mauzac, M. (2002) Using polydimethylsiloxane as a thermocurable resist for a soft imprint lithography process. Microelectron. Eng. 61-2, 379–384. 68. Malaquin, L. and Vieu, C. (2003) Using PDMS as a thermocurable resist for a mold assisted imprint process. In: Alternative Lithography, Sotomayer-Torres, C. M., ed., Kluwer, Boston, pp. 169–203. 69. Pina, C., Fu, P.-F., Kim, J., and Guo, L. J. (2007) High throughput and etch selective nanoimprinting and stamping based on fast thermal-curable polydimethysiloxanes. Adv. Mater. 19, 1222–1227. 70. Colburn, M., Johnson, S., Stewart, M., Damle, S., Bailey, T., Choi, B., Wedlake, M., Michaelson, T., et al. (1999) Step and flash imprint lithography: a new approach to highresolution patterning. Proc. SPIE 3676, 379. 71. Dickey, M. D., Burns, R. L., Kim, E. K., Johnson, S. C., Stacey, N. A., and Willson, C. G. (2005) Study of the kinetics of step and flash imprint lithography photopolymerization. AIChE J. 51, 2547–2555. 72. Dickey, M. D. and Willson, C. G. (2006) Kinetic parameters for step and flash imprint lithography photopolymerization. AIChE J. 52, 777–784. 73. Kim, E. K., Stewart, M. D., Wu, K., Palmieri, F. L., Dickey, M. D., Ekerdt, J. G., and Willson, C. G. (2005) Vinyl ether formulations for step and flash imprint lithography. J. Vac. Sci. Technol. B 23, 2967–2971. 74. Cheng, X., Guo, L. J., and Fu, P. F. (2005) Room-temperature, low-pressure nanoimprinting based on cationic photopolymerization of novel epoxysilicone monomers. Adv. Mater. 17, 1419–1424. 75. Pfeiffer, K., Reuther, F., Fink, M., Gruetzner, G., Carlberg, P., Maximov, I., Montelius, L., Seekamp, J., et al. (2003) A Comparison of thermally and photochemically cross-linked polymers for nanoimprinting. Microelectron. Eng. 67-8, 266– 273. 76. Mele, E., Di Benedetto, F., Persano, L., Cingolani, R., and Pisignano, D. (2005) Multilevel, room-temperature nanoimprint lithography for conjugated polymer-based photonics. Nano Lett. 5, 1915–1919. 77. Nielsen, T., Nilsson, D., Bundgaard, F., Shi, P., Szabo, P., Geschke, O., and Kristensen, R , a highly A. (2004) Nanoimprint lithography in the cyclic olefin copolymer, Topas
ultraviolet-transparent and chemically resistant thermoplast. J. Vac. Sci. Technol. B 22, 1770–1775.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

14:0

Printer Name: Yet to Come

REFERENCES

165

78. Cheng, L. J., Kao, M. T., Meyhc¸fer, E., and Guo, L. J. (2005) Biomolecular transport highly efficient guiding of microtubule transport with imprinted CYTOP nanotracks. Small 1, 409–414. 79. Igaku, Y., Matsui, S., Ishigaki, H., Fujita, J., Ishida, M., Ochiai, Y., Namatsu, H., Komuro, M., et al. (2002) Room temperature nanoimprint technology using hydrogen silsesquioxane (HSQ). Japan. J. Appl. Phys. 41, 4198–4202 (Part 1: Regular Papers Short Notes and Review Papers). 80. Matsui, S., Igaku, Y., Ishigaki, H., Fujita, J., Ishida, M., Ochiai, Y., Namatsu, H., and Komuro, M. (2003) Room-temperature nanoimprint and nanotransfer printing using hydrogen silsequioxane. J. Vac. Sci. Technol. B 21, 688–692. 81. Cheng, L.-J., Chang, S.-T., and Guo, L. J. (2005) Nanoimprint of nanofluidic channels by using hydrophilic hydrogen silsesquioxane (HSQ). Proceedings of microTAS 2005, pp. 518–520. 82. Karnik, R., Castelino, K., Fan, R., Yang, P., and Majumdar, A. (2005) Effects of biological reactions and modifications on conductance of nanofluidic channels. Nano Lett. 5, 1638–1642. 83. Yu, Z. N., Gao, H., and Chou, S. Y. (2004) In situ real time process characterization in nanoimprint lithography using time-resolved diffractive scatterometry. Appl. Phys. Lett. 85, 4166–4168. 84. Hirai, Y., Fujiwara, M., Okuno, T., Tanaka, Y., Endo, M., Irie, S., Nakagawa, K., and Sasago, M. (2001) Study of the resist deformation in nanoimprint lithography. J. Vac. Sci. Technol. B 19, 2811–2815. 85. Hirai, Y., Konishi, T., Yoshikawa, T., and Yoshida, S. (2004) Simulation and experimental study of polymer deformation in nanoimprint lithography. J. Vac. Sci. Technol. B 22, 3288–3293. 86. Rowland, H. D. and King, W. P. (2004) Polymer deformation and filling modes during microembossing. J. Micromech. Microeng. 14, 1625–1632. 87. Rowland, H. D., Sun, A. C., Schunk, P. R., and King, W. P. (2005) Impact of polymer film thickness and cavity size on polymer flow during embossing: toward process design rules for nanoimprint lithography. J. Micromech. Microeng. 15, 2414–2425. 88. Rowland, H. D., King, W. P., Sun, A. C., and Schunk, P. R. (2005) Simulations of nonuniform embossing: the effect of asymmetric neighbor cavities on polymer flow during nanoimprint lithography. J. Vac. Sci. Technol. B 23, 2958–2962. 89. Chao, C. Y. and Guo, L. J. (2002) Polymer microring resonators fabricated by nanoimprint technique. J. Vac. Sci. Technol. B 20, 2862–2866. 90. Chao, C. Y., Fung, W., and Guo, L. J. (2006) Polymer microring resonators for biochemical sensing applications. IEEE J. Sel. Top. Quantum Electron. 12, 134–142. 91. Ashkenazi, S., Chao, C. Y., Guo, L. J., and O’Donnell, M. (2004) Ultrasound detection using polymer microring optical resonator. Appl. Phys. Lett. 85, 5418. 92. Chao, C. Y., Ashkenazi, S., Huang, S.-W., O’Donnell, M., and Guo, L. J. (2007) Highfrequency ultrasound sensors using polymer microring resonators. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 54, 957–965. 93. Sreenivasan, S. V., McMackin, A., Xu, F., Wang, D., and Stacey, N. (2005) Using reverse-tone bilayer etch in ultraviolet nanoimprint lithography. MICRO 23, 37–. 94. Borzenko, T., Tormen, M., Schmidt, G., Molenkamp, L. W., and Janssen, H. (2001) Polymer bonding process for nanolithography. Appl. Phys. Lett. 79, 2246–2248.

P1: OTA/XYZ P2: ABC c06 JWBK252/Lee September 18, 2008

166

14:0

Printer Name: Yet to Come

PATTERNING BASED ON EXTERNAL FORCE

95. Tan, H., Gilbertson, A., and Chou, S. Y. (1998) Roller nanoimprint lithography. J. Vac. Sci. Technol. B 16, 3926–3928. 96. Bao, L. R., Tan, L., Huang, X. D., Kong, Y. P., Guo, L. J., Pang, S. W., and Yee, A. F. (2003) Polymer inking as a micro- and nanopatterning technique. J. Vac. Sci. Technol. B 21, 2749–2754. 97. Kehagias, N., Zelsmann, M., Torres, C. M. S., Pfeiffer, K., Ahrens, G., and Gruetzner, G. (2005) Three-dimensional polymer structures fabricated by reversal ultraviolet-curing imprint lithography. J. Vac. Sci. Technol. B 23, 2954–2957. 98. Kong, Y. P., Low, H. Y., Pang, S. W., and Yee, A. F. (2004) Duo-mold imprinting of three-dimensional polymeric structures. J. Vac. Sci. Technol. B 22, 3251–3256. 99. Huang, X. D., Bao, L. R., Cheng, X., Guo, L. J., Pang, S. W., and Yee, A. F. (2002) Reversal imprinting by transferring polymer from mold to substrate. J. Vac. Sci. Technol. B 20, 2872–2876. 100. Tan, L., Kong, Y. P., Pang, S. W., and Yee, A. F. (2004) Imprinting of polymer at low temperature and pressure. J. Vac. Sci. Technol. B 22, 2486–2492. 101. Suh, D., Rhee, J., and Lee, H. H. (2004) Bilayer reversal imprint lithography: direct metal–polymer transfer. Nanotechnology 15, 1103–1107. 102. Scheer, H. C. and Schulz, H. (2001) A contribution to the flow behaviour of thin polymer films during hot embossing lithography. Microelectron. Eng. 56, 311–332. 103. Pfeiffer, K., Fink, A., Gruetzner, G., Bleidiessel, G., Schulz, H., and Scheer, H. (2001) Multistep profiles by mix and match of nanoimprint and UV lithography. Microelectron. Eng. 57–8, 381–387. 104. Cheng, X. and Guo, L. J. (2004) One-step lithography for various size patterns with a hybrid mask-mold. Microelectron. Eng. 71, 288–293. 105. Cheng, X. and Guo, L. J. (2004) A combined-nanoimprint-and-photolithography patterning technique. Microelectron. Eng. 71, 277–282. 106. Cheng, X., Li, D. W., and Guo, L. J. (2006) A hybrid mask-mould lithography scheme and its application in nanoscale organic thin film transistors. Nanotechnology 17, 927–932. 107. Kang, M.-G. and Guo, L. J. (2007) Nanoimprinted semi-transparent metal electrode and its application in OLED. Adv. Mater. 19, 1391–1396.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7 PATTERNING BASED ON EDGE EFFECTS: EDGE LITHOGRAPHY Matthias Geissler, Joseph M. McLellan, Eric P. Lee and Younan Xia

7.1 INTRODUCTION The ability to fabricate functional nanostructures has become indispensable to progress in many areas of research and technology ranging from materials science, to microelectronics, catalysis, and biomedical diagnostics. Hence, the development of nanolithographic techniques capable of controlling material properties with a spatial resolution of 1–100 nm is currently the focus of considerable research and development efforts. Semiconductor industry has responded to this challenge by investing in sophisticated, high precision technologies that allow for mass-producing integrated circuits at ever-increasing density, complexity, and performance [1–5]. The enabling technology is optical projection lithography which, up to this date, has prevailed as the dominant method for large-scale production purposes [4]. Over the past 50 years since its introduction, limits in resolution set by optical diffraction have been pushed back continuously, mainly through the use of shorter radiation wavelengths, improved lens materials, advanced projection systems, and optimized resist formulations [3–5]. For example, it is possible to achieve a minimum transistor gate length of 25 nm and a dynamic random access memory periodicity (half-pitch) of 65 nm using current state-of-the art step-and-repeat exposure tools [5]. At the same time, scanning beams of high energy particles comprising atoms, ions, and electrons have evolved into robust and mature technologies, which equally can be used for

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


167

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

168

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

micro- and nanofabrication purposes [6, 7]. Each of these techniques can create arbitrary features at very high resolution (e.g., 5 nm and below), but their serial nature limits the scope of applications to selected, low volume fabrication tasks. Complexity at the equipment level combined with elevated operational cost renders these well-established techniques unpractical for research purposes, especially in common laboratory environments. Alternative (or unconventional) concepts of micro- and nanofabrication [8–15] have become attractive for applications where conventional techniques fail to be effective. These are processes that involve sensitive materials being incompatible with resist and development procedures, curved or uneven substrates, and large areas; rapid turnover times and a reasonably low cost are often additional concerns. Alternative techniques can be subdivided into four categories. One category includes prototyping methods based on various forms of embossing, molding, and printing [8–15]. The virtue of these techniques is to replicate and transfer multiple copies of a master pattern into a variety of functional materials with high fidelity and at a broad range of length scales (e.g., from several millimeters to below 50 nm). Techniques such as nanoimprint lithography (NIL) [16] and replica molding [17] are the methods of choice when nanometer features in polymeric materials are the fabrication target. Microcontact printing (µCP) [18], on the other hand, is more versatile; it mediates pattern transfer through contact with an elastomeric stamp and can achieve sub-micrometer resolution for a variety of ink/substrate systems including self-assembled monolayers (SAMs) of alkanethiols on gold [19, 20], inorganic catalysts on pretreated substrates [21], and proteins on silicon or glass [22]. Scanning probe lithography constitutes another category [23]. Methods that belong to this category take advantage of the high resolution capabilities provided by a sharp tip mounted on a cantilever; they can be used to write arbitrary patterns by moving the tip across the surface of a responsive medium, thereby inducing changes either mechanically (e.g., by displacement), electrically (e.g., through oxidation), or via deposition of a proper material in an add-on process as done in dip pen nanolithography [24]. Depending on the sharpness of the tip, the modification process can be accomplished in a highly localized manner approaching the level of individual molecules and atoms [25]. However, this process typically comes at the expense of limited robustness and relatively low throughput, although efforts toward parallelization are underway [26, 27]. A third category comprises techniques that involve self-assembly. Conceptually, self-assembly borrows from principles of synthetic chemistry: proper building blocks that can interact with each other organize into stable aggregates of desired size, shape, and functionality [28]. Examples of this approach can be found in the field of phase-separated block copolymers [29] and in the colloidal domain [30] where self-assembled structures can serve as masks for etching or deposition processes. In practical terms, however, self-assembly approaches are still at the stage of infancy, and their full potential for lithographic applications remains to be revealed. The fourth and last category in this context is edge lithography, which is the subject of this chapter. Edge lithography is, in contrast to other lithographic methods, not a single technique that follows a routine and preset patterning scheme. Instead, it is a synonym for a multitude of approaches that connect to the idea of involving the edges of a

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER

169

structure to define the features of the final pattern. The origins of edge lithography are in microelectronics when phase-shift technology (see Section 7.2.1) was first explored in the early 1980s as a means of enhancing resolution for contact-mode optical lithography [31]. Yet, it took until the mid 1990s before edge lithography evolved into a broader strategy for nanofabrication. Its evolution was promoted in part by the advent of soft lithography [8, 11] and the concept of rapid prototyping, making flexible, micropatterned masters, molds, and stamps available to a larger scientific community. As of today, there are at least 10 different approaches that are centered on the use of edges for patterning purposes. From a methodological point of view, these are replication techniques that emphasize the virtue of size reduction—i.e., a miniaturization process during which a larger structure is translated into a smaller one. Edge lithography therefore is essentially a top-down approach, even though a number of techniques comprise elements of bottom-up fabrication to an almost equal extent. This chapter is intended to provide a short, yet comprehensive, overview of edge lithography. In a set of dedicated sections, we outline patterning principles, emphasize suitable materials, and highlight relevant demonstrations for each edge-lithographic approach that is known from the open literature. An excellent survey of this field has previously been published by Whitesides and co-workers [15]; overviews of edge lithography can also be found in a number of review articles devoted to the fields of micro- and nanofabrication [9, 13, 14]. Herein, we complement preceding work with recent developments to sketch a portrait of the field that is accurate as of August 2007. Previously, Whitesides and co-workers have discriminated two distinct strategies for edge lithography, which create nanostructures either at the edges of a topographic pattern, or by exposing the cross-section of a thin film [15]. For the educational purpose of this chapter we adapted the same classification scheme, and organized the forthcoming sections in the following order. Section 7.2 deals with the concept of using topographic edges for pattern transfer. Strategies that belong to this category include (i) photolithography using phase-shifting masks, (ii) use of edge-defined defects in SAMs to direct deposition or removal processes, (iii) undercutting at topographic step edges via etching, (iv) reactive spreading of SAMs guided by topographic features, (v) edge-controlled release of SAM-forming molecules from elastomeric stamps, and (vi) selective decoration of step edges using add-on processes. Section 7.3 discusses approaches that involve the deposition of metallic thin films and subsequent exposure of their cross-sections as a means of producing nanostructures. Demonstrated strategies developed to this end are (i) fracturing of multilayer deposits, (ii) sectioning of encapsulated thin films, (iii) selective removal of thin films from a patterned elastomeric stamp, and (iv) reorientation of topographic features decorated with a thin film. Section 7.4 concludes this chapter with a personal remark on the perspective of edge lithography in the context of nanofabrication.

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER The approaches described in this section employ changes at the edges of a topographic feature to yield structures at nanometer length scales. In a sense, they are

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

170

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

similar to the process of reproducing shape with pen and paper by using a ruler or any other suitable object, which is a basic scheme familiar to most people from early childhood on. The mark of the pen—be it wide or narrow—eventually represents an accurate replica of the ruler’s edge on the paper. While typically employed at much smaller length scales (e.g., in the micrometer regime), the role played by the ruler is essentially the same for edge lithography and any of its technological applications. Here, pattern replication is, by contrast, a largely parallel process that differs by the choice of materials involved and the principles that govern structure formation, making pen and paper become abstract rather than literate constituents. One approach, photolithography with phase-shifting masks (see Section 7.2.1), uses light to locally induce changes in a radiation-sensitive material (commonly referred to as photoresist) at the edges of a relief pattern provided by a photomask. Other techniques involve selective placement of a material at the edges of topographic features on a proper surface (see Sections 7.2.2 and 7.2.4–7.2.6), which can be mediated either by diffusion, nucleation, or electrochemical reduction. Finally, material can also be removed at an edge through wet-chemical etching (see Section 7.2.3). Several approaches described in this section rely on SAMs, especially those formed from alkanethiols on Au or on other coinage metals [32] (see Sections 7.2.2–7.2.4). These ultra-thin (typically ∼1–3 nm) organic layers are suitable for a number of lithographic applications since they can alter effectively surface properties for interfacial phenomena such as wetting, nucleation, and corrosion. In addition, SAMs allow for introducing functionality to a surface in a reliable and chemically welldefined manner. 7.2.1 Photolithography with Phase-Shifting Masks Phase-shifting photolithography is a technique that enables the fabrication of nanostructures beyond the optical diffraction limit [31, 33–49]. It was developed as a resolution-enhancement technique for optical contact printing to be used in semiconductor industries [31]. The key element of phase-shift photolithography is a binary phase mask comprising relief structures that have dimensions comparable to the radiation wavelength (Figure 7.1a). The depth D of the features on the mask relates to the wavelength of the illuminating light λ with D = λ/2(n – 1), where n is the index of refraction of the material forming the mask. Light passing through the mask experiences modulation of its intensity at the vertical edges of the pattern that induces local variations in the exposure of the resist film. This allows one to produce ridges in a positive-tone photoresist, whereas trenches would be obtained in the same areas when a negative-tone resist is used. Features generated by this technique typically have a width of 30–100 nm [34, 42], although smaller structures have also been reported [45]. The near-field effect is maximized when the transmitted light is shifted by odd numbers of π radians, which reduces the intensity near the surface to zero at the edges of relief, causing the formation of dark-line features in the resist (Figure 7.1b). Moreover, scattering of light at phase boundaries yields a modulated intensity distribution with coexisting bright- and dark-line features of varying amplitude and periodicity. As the transmitted light directly affects the thickness of the photoresist

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

171

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER

(a)

Total energy

(b)

1.5

PDMS stamp D

y (µm)

(i) Place mask onto positive photoresist (ii) Expose to UV light

2

1

1.5

0.5

1

0

hν

0.5 −0.5 −10

−5

0 x (µm)

5

10

−5

0 x (µm)

5

10

0

Develop photoresist 30–100 nm

Positive photoresist pattern

(c)

Normalized total energy

Silicon substrate

1

0.5

0 −10

(d)

350 nm

2 µm 0

2

4

6

µm

Figure 7.1. Near-field optical lithography with conformal phase-shifting masks. (a) A binary relief mask made of PDMS is placed on a thin layer of positive-tone photoresist supported on a solid substrate. Upon exposure and subsequent development, a pattern of resist lines is obtained that correspond to the edges of relief. The width of these lines can be in the range of 30–100 nm. (b) Simulation of light polarized in the transverse magnetic mode (electric field perpendicular to the grating wave vector) propagating through a topographic phase mask comprising a set of parallel lines that have a width of 10 µm, a depth of 1.4 µm, and a spacing of 10 µm. The thickness of the resist layer is 500 nm. The linecut in the lower panel shows the intensity of light in the resist layer at a depth of 50 nm. Both images reveal a small difference in intensity between contact and no contact regions of the phase mask. (Reprinted with permission from [49]. Copyright 2006, American Vacuum Society.) (c ) Three-dimensional AFM images of a positive photoresist layer that was patterned using a binary phase mask with a period of 4 µm and exposure times of 1 s (upper panel) and 5 s (lower panel). (Reprinted with permission from [40]. Copyright 2001, American Institute of Physics.) (d ) Scanning electron micrograph of silicon NWs that were produced by near-field optical lithography followed by pattern transfer using RIE, oxidation of silicon in air at 850◦ C for 1 h, and liftoff in a solution of HF. Each wire has a length of ∼2 cm and a width of 80 nm. (Reprinted with permission from [39]. Copyright 2000, Wiley-VCH.)

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

172

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

film after development, it is even possible to obtain trench or line patterns in a positive resist by changing the exposure time [40], which is exemplified by the images shown in Figure 7.1c. In order for a mask to be useful in phase-shift optical lithography, it has to be largely transparent and must be brought as close as possible to the surface of the resist layer. Both rigid and elastomeric materials have been employed for mask fabrication. Elastomers are advantageous due to their relatively low cost and inherent ability to reversibly conform to a photoresist layer on both even and nonplanar substrates [34, 46]. On the other hand, rigid masks (made from quartz or glass) are more straightforward to handle and generally provide higher accuracy of features. Most elastomeric phase masks have been fabricated from poly(dimethylsiloxane) (PDMS), although composite stamps comprising tailored siloxanes or nonsiloxane-based materials have also been employed for a number of applications [50, 51]. Whitesides and co-workers have shown that embossed photoresist can equally serve as its own optical element for near-field photolithography [52]. Resist patterns produced by phase-shift photolithography are commonly transferred into the underlying substrate using reactive ion etching (RIE). For example, Rogers and co-workers have fabricated metal features that can serve as optical polarizer and gates for organic transistors using soft polymer masks [34, 38]. Fritze and co-workers [45] have demonstrated the fabrication of transistors with a gate length of 9 nm, as well as complex patterns of interconnecting features using multiple exposures with specially designed glass-based phase-shift masks. Odom and co-workers relied on near-field optical lithography to fabricate patterns that were used to direct the assembly of semiconductor quantum dots [53], and the growth of ZnO nanowires (NWs) into ordered arrays [54]. Xia and co-workers have shown that it is feasible to generate freestanding single-crystal Si rings, rods, and wires with lateral dimensions as small as 40 nm using this approach [39]. These authors employed a silicon-oninsulator (SOI) wafer for pattern transfer via RIE, which was followed by oxidation of the Si, and liftoff using HF (Figure 7.1d). Dissolution of the oxide layer around the Si further reduced the dimensions of the features that were formed initially. More recently, Rogers and co-workers have demonstrated the fabrication of 3D nanostructures using phase-shift photolithography [47, 48] (Figure 7.2). Here, structure formation is the result of a complex intensity distribution that exposes certain areas within the bulk of a thicker photoresist layer. The authors have implemented these structures into channels of microfluidic devices where they can efficiently serve as passive, nanoporous filters [47], or as mixing elements [48]. 7.2.2 Use of Edge-Defined Defects in SAMs This approach was pioneered by Whitesides and co-workers who have demonstrated that disorder in SAMs of alkanethiols can be controlled through variations in topography and composition of the supporting metal substrate [55–57]. As illustrated in Figure 7.3, the transition zones between two metal features represent labile, disordered regions in which molecules are prone to displacement when exposed to a high concentration of another thiol. The binary monolayer that forms as a result of this

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER

(a)

(b)

(c)

(d )

173

Figure 7.2. Scanning electron micrographs of 3D nanostructures fabricated in SU-8 using a conformal phase-shifting mask. The mask was fabricated in PDMS and had a square lattice of isolated relief features (rounded square) with a diameter of 1 µm, a relief depth of 420 nm, and a duty cycle of 40%. Insets show the corresponding computed optical intensity distributions. (a) Large-scale image of the structure. (b) Close-up view of the tilted (100) facet with modeling inset that corresponds to a cross-section of the pillars. (c , d ) Magnified views of top and bottom surfaces, respectively. The thickness of the SU-8 layer was ∼10 µm, and the tripled output (355 nm) of a Nd:YAG laser was used for exposure. (Reprinted with permission from [47]. Copyright 2004, National Academy of Sciences of the USA.)

Region of disorder

SAM of alkanethiol 1 Metal 2

Metal 2

Metal 1

Metal 1

Exchange with alkanethiol 2 in ethanol

Alkanethiol 2 Metal 2 Metal 2 Metal 1

Metal 1

Figure 7.3. Control of local disorder in SAMs. Molecules at the edges of their patterned metal supports are prone to displacement when exposed to a solution of another alkanethiol. The shape of the metal structures has an influence on the degree of disorder and the lateral dimension of the area in which exchange of molecules takes place. (Reprinted with permission from [55]. Copyright 1998, Macmillan Publishers.)

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

174

(a)

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

(c)

(b)

~50 nm

Figure 7.4. Patterns generated by using edge-defined defects in SAMs. (a, b) Optical micrographs of condensation figures formed on SAMs supported on micropatterned silver substrates prepared by deposition of 50 nm of Ag through a stencil mask. Substrates were exposed to a 10 mM solution of the first thiol in ethanol for 30 s, and then incubated with a solution of the second thiol for 10 h. The SAM in (a) comprises HDT exchanged with MHA; the one in (b) consists of MHA exchanged with HDT. (c ) Scanning electron micrographs of circular trenches etched into an Ag substrate that was fabricated by deposition of 50 nm of Ag through a patterned photoresist layer followed by liftoff. The substrate was incubated with a 10 mM solution of HDT in ethanol for 1 h, rinsed and etched using an aqueous solution of ferrocyanide, ferricyanide, and sodium thiosulphate. (Reprinted with permission from [55]. Copyright 1998, Macmillan Publishers.)

localized exchange can then be used in a variety of surface-sensitive patterning processes. One demonstration to this end is the formation of condensation patterns using water droplets. For example, when a SAM formed from n-hexadecanethiol (HDT) was allowed to exchange with 16-mercaptohexadecanoic acid (MHA), elongated drops were observed to form in the hydrophilic transition zones (Figure 7.4a) [55]. In contrast, the condensation pattern was reversed when molecules in a carboxylterminated SAM were displaced with a methyl-terminated thiol (Figure 7.4b). Topology of the transition zone between two metals seems to affect both size and shape of the regions in which exchange of molecules occurs. Selective nucleation and crystal growth constitute another area of application. For example, Aizenberg and coworkers have shown that well-defined patterns of calcite crystallites can be obtained on a variety of binary SAM systems [57, 58].

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER

175

Disordered regions in a SAM are sensitive to selective etching, leading to the development of a technique that became known as topographically directed etching (TODE) [55, 56, 59]. TODE enables a variety of fine features including lines, trenches, and dots to be generated with high fidelity over large areas. Figure 7.4c shows a trench pattern that was etched into a thin Ag film. Here, the lateral dimensions of the trenches were on the order of 50 nm or less. The width of features generated by TODE is potentially as small as the region of disorder, but is further limited by morphology and grain size of the metal layer that is used for experiments. An elegant extension of this approach has been shown by Whitesides and co-workers who have produced Ag features on an Ag film with a thin (5 nm) Ti layer sandwiched in between the two [60]. The Ag surfaces were then coated with a thiol monolayer yielding a pattern of nanometer-sized gaps in the SAM induced by the presence of the Ti layer. In a process similar to step-edge decoration (see Section 7.2.6), the uncovered Ti edges were used as nanoelectrodes to produce Cu wires with diameters as small as 70 nm via electrodeposition. These Cu wires could then be lifted off with Scotch tape and used as transparent, flexible, and self-adhesive optical polarizer.

7.2.3 Controlled Undercutting Controlled undercutting [61] is a method that can be used to produce trenches as narrow as 50 nm in a variety of materials. Originally demonstrated by Whitesides and co-workers, the process starts by patterning a resist layer on a thin metal film supported on a planar, solid substrate using photolithography (Figure 7.5a). The unprotected regions of the metal are then dissolved in an aqueous etch solution. The key to this approach is keeping the substrate in the etch bath for a longer time than necessary to completely remove metal from the exposed regions. In doing so, a slight undercut is induced in the regions covered by the resist. Upon collimated deposition of another metal layer onto the substrate, the resist film is removed in a liftoff process, yielding a groove or trench pattern in the metal film that corresponds to the undercut regions. An example of a pattern that has been fabricated using this approach is shown in Figure 7.5b [61]. The trenches can further be transferred into the underlying substrate by using the patterned metal layer as a mask in a subsequent etching step (Figure 7.5c). While this approach has been demonstrated for trenches that are 50 nm wide, it should be possible to tune the degree of undercutting by changing the time of etching and/or the strength of the etch bath. Controlled undercutting has been employed for reducing the dimensions of patterns generated by near-field photolithography. For example, periodic arrays of Au dots as small as 40 nm were patterned over areas as large as 1 cm2 , and then used as a catalyst for the growth of ZnO NWs [54]. The technique has further been applied to the patterning of Al on optical grade Si/SiO2 substrates to produce frequencyselective devices. Here, square loops were etched into Al that exhibited a strong resonance band in the infrared (IR) region at ∼7.5 µm, which could be utilized as band-pass filters for IR applications [61].

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

176

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

Metal layer

(a)

(b)

Substrate Photolithography Photoresist

Overetch

(c) Metal deposition Metal layer

Liftoff 50–200 nm Test optically Etch

Cr–Si interface

Figure 7.5. Formation of nanoscale features using controlled undercutting. (a) Schematic illustration of a typical process flow using patterned photoresist features. The metal substrate beneath the resist is etched unisotropically that results in a slight undercut of the base layer. Evaporation of a second metal layer followed by liftoff results in 50–200 nm wide gaps in the metal film. The patterned layer can then be used either as an optical filter or as a physical mask for pattern transfer via etching. (b) Scanning electron micrograph of trenches in the form of a star pattern in a thin Cr layer. (c ) Scanning electron micrographs of linear trenches produced by controlled undercutting of 100-nm-wide lines transferred into a Si(100) substrate. Trenches are 75 nm wide and 250 nm deep. (Reprinted with permission from [61]. Copyright 2001, Wiley-VCH.)

7.2.4 Edge-Spreading Lithography Edge-spreading lithography (ESL) [62–66] has been developed by Xia and coworkers for the patterning of alkanethiol monolayers on coinage metals at submicrometer length scales. This technique relies on two key features: a mesoscopic relief structure on the metal substrate and reactive spreading. The relief structure has dual functions: it mediates the transport of alkanethiol molecules from a planar PDMS stamp to a gold surface and determines the shape of the emerging SAM. For example, when a 2D array of silica beads is used as the guide, the circular footprint of each bead produces a pattern of monolayer rings with hexagonal arrangement (Figure 7.6a). Reactive spreading [67, 68] is the extension of a SAM across a metal surface; it occurs when an excess of unbound thiol molecules and an unoccupied area on the surface into which the SAM can expand are both available.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER

Silica beads

(a)

177

PDMS stamp

Au Silicon substrate

Print ODT Remove stamp

ODT molecules

Top view

Remove beads

Cross-section

ODT SAM

(b)

(c)

(d)

(e)

Figure 7.6. Formation of monolayer ring patterns on Au substrates using ESL. (a) Schematic illustration of ESL using an array of SiO2 beads on a thin Au film. Each bead guides the delivery of ODT molecules from a planar PDMS stamp to the surface of the metal film where they form a SAM. (b–e) Examples of ring patterns generated with silica beads being 1.6 µm in diameter. (b) LFM image of ODT monolayer rings on an Au surface. (c ) Scanning electron micrograph of Au rings formed by selective wet etching using an ODT monolayer as resist. (d ) LFM image of concentric SAM rings comprising MHA (bright), HDDT (gray), and ODT (dark) patterned by three successive ESL steps. (Reprinted with permission from [64]. Copyright 2005, Wiley-VCH.) (e) LFM image of an array of SAM rings each containing a gradient of MHA and ODT (from bright to dark, respectively). (Reprinted with permission from [66]. Copyright 2006, Wiley-VCH.) Scale bars in (b), (d ), and (e) denote 500 nm.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

178

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

Lateral force microscopy (LFM) [69, 70] revealed that monolayer rings formed by ESL show high contrast and accuracy for a number of thiols. When n-octadecanethiol (ODT) was used, resultant monolayer rings appeared dark as a result of the relatively weak interaction between the nonpolar molecules and the polar surface of the silicon nitride tip that was used for the measurement (Figure 7.6b). Figure 7.6c illustrates that this array could be successfully transferred into the supporting Au substrate via selective wet etching using the SAM as a resist [62]. ESL further allows for the patterning of multiple SAMs terminated with a variety of functional groups in the form of concentric rings using successive printing steps on the same substrate [64]. Figure 7.6d shows the result of such a process with MHA, 12-hydroxyundecanethiol (HDDT), and ODT. It should be noted that the width of each segment in these rings can be tuned by variation of ink concentration and printing time, and that these thiols could be printed in any order. Concentric rings comprising distinct portions as small as 30 nm have been fabricated using this approach. Xia and co-workers have further demonstrated that concentric gold rings can be fabricated via selective etching when a less protective amine monolayer is sandwiched between two concentric rings of thiolate SAMs [64]. Rings formed from a 1:1 mixture of MHA and ODT showed a gradual change in composition from the inside to the outside (Figure 7.6e) [66]. This gradient develops due to differences in the velocity of the two components to reach the gold surface, and its steepness can be tuned to some extent by varying the ratio of the two components in the solution used to ink the stamp. Patterns comprising structures other than a hexagonal array of rings can be generated if silica beads are replaced with polymeric resist features of appropriate shape and geometry [63]. Polymers being compatible with this approach include AZ 1512 photoresist, polyimide (PI), and poly(methylmethacrylate) (PMMA). Examples of Au pattern that were produced this way are shown in Figure 7.7.

7.2.5 Edge Transfer Lithography Edge-transfer lithography (ETL) [71, 72]. can be viewed as an extended variant of µCP but also shares similarities with ESL (see Section 7.2.4). In µCP, a topographically patterned PDMS stamp is inked and then brought into contact with a substrate resulting in ink transfer at the raised portions of the stamp [18–20]. ETL differs from this patterning scheme in that ink transfer only occurs at the edge of features on the stamp rather than over the entire area that is in contact with the substrate. This method was originally demonstrated for silane compounds as well as Ti nanoparticles that are prone to dewetting the surface of PDMS when applied in polar solvents, leading to accumulation of ink in the recessed regions of the stamp [71]. Upon contact with a substrate, ink is transferred to the surface from the sidewalls of the topographic features. The inverse strategy uses exposure of PDMS to oxygen plasma to render the stamp impermeable and repellent to hydrophobic ink molecules. This way, Huskens and co-workers have confined the formation of an ODT monolayer to the edges of features that come in contact with a gold substrate [72].

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.2 TOPOGRAPHY-DIRECTED PATTERN TRANSFER

(a)

(b)

(c)

(d)

179

Figure 7.7. Scanning electron micrographs of Au test patterns on a silicon wafer prepared by ESL and selective wet etching. All patterns represent accurate outlines of the corresponding photoresist features that are shown by the optical microscope images in the insets. Planar PDMS stamps were inked with a 2.0-mM solution of ODT in ethanol, and placed on the 4.0-µm-thick PI resist features for 15 and 4 min to achieve the patterns depicted in (a) and (b), respectively. For (c ) and (d ), the thickness of the AZ 1512/PMMA layer was 1.6 µm, and a contact time of 15 min was used for both samples. Unprotected Au was etched in aqueous solution of Fe(NO3 )3 and thiourea. (Reprinted with permission from [63]. Copyright 2005, American Chemical Society.)

In another variant of ETL, these authors have covered certain areas of the stamp with a blocking layer that prevents the transfer of ink to the substrate, but does not prevent the stamp from spontaneously making conformal contact when placed on a surface [72]. This approach is schematically illustrated in Figure 7.8a. Here, a thin layer of Ti was deposited onto the stamp in a collimated manner followed by exposure to oxygen plasma to provide a suitable blocking layer on the planar portions of the stamp. Thiol molecules that accumulate in the bulk of the PDMS stamp during inking can leave the stamp only through the uncoated regions (e.g., the sidewalls of the raised features) while being in contact with a gold substrate. Hence, a SAM forms predominantly at the periphery of the topographic portions of the stamp. The Au features shown in Figure 7.8b serve as an example to this approach. Changing both printing time and ink concentration can be used to dynamically control the width of the emerging SAM while maintaining lateral dimensions substantially smaller than those provided by the stamp.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

180 (a)

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

Polymer Ti TiOx

(b)

Figure 7.8. Pattern formation by ETL using an anisotropically modified stamp and wet etching. (a) A thin layer of Ti/TiOx deposited onto the surface of a patterned stamp does not inhibit the formation of conformal contact but prevents direct transfer of ink from the features of relief. Ink molecules reach the Au surface via diffusion from the unprotected side walls, thus forming a SAM at the periphery of the features being in contact with the substrate. (b) Optical micrograph of an Au pattern fabricated by ETL using a Ti-covered PDMS stamp and ODT molecules as the ink, followed by selective wet etching of the unprotected Au regions. (Reprinted with permission from [72]. Copyright 2006, American Chemical Society.)

7.2.6 Step-Edge Decoration Step-edge decoration is an approach to nanofabrication that uses selective deposition of metals or oxides at the step edges of a single-crystal surface [73–84]. Lattice step edges are typically more reactive than other lattice planes on a surface, and therefore can serve as preferential nucleation sites for atoms in deposition processes [85, 86]. For example, nanoparticles of Cu, Co, and Ag have been deposited from the vapor phase at the step edges of single-crystal metal substrates including Mo(1 1 0), Ag(1 1 1), and Cu(1 1 1) [73, 75, 77]. More recently, Penner and co-workers have used atomic step edges in highly oriented pyrolytic graphite (HOPG) to selectively deposit one-dimensional nanostructures in a diversified range of materials including metals (e.g., Ag, Pd, Cu, and Au) [78, 80], oxides (e.g., MoOx ) [76, 83], and semiconductors (e.g., MoS2 and Bi2 Te3 ) [82, 84] by electroplating, as schematically illustrated in Figures 7.9a and b. These authors have shown that the diameter of MoO2 NWs, for example, can be tuned between 40 and 700 nm by varying the applied potential and the time used for electroplating (Figure 7.9c) [81]. One remarkable feature of these wires is their high degree of uniformity suggesting convergent growth during the electrodeposition process. Pd NWs produced by direct electrochemical reduction of Pd2+ at the step edges of HOPG have been transferred onto a cyanoacrylate film and used as a sensor for hydrogen gas [78]. It is believed that this strategy has potential in generating functional building blocks for device fabrication. To this end, step-edge decoration has mainly been shown for HOPG templates on which structural information was naturally present rather than being fabricated intentionally, and the approach therefore remained constrained by random orientation of wires and distribution of step-edge defects. In addition, there can be variations in

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.3 EXPOSURE OF NANOSCALE EDGES

(a)

Step edges

181

(c)

Metal oxide electrodeposition Metal oxide nanowires

128 s

Wire diameter (nm)

Metal nanowires

800

600

400

200

0

(b)

Step edges

256 s

1000

H2, 500°C 1 hr

0

8s

5

10

15

20

Time1/2, s1/2

32 s

Mn+ + ne− → M° Metal nanowires

Figure 7.9. Formation of metal NWs via electrochemical decoration of atomic step edges on HOPG substrates. (a) Deposition of a metal oxide by electroplating followed by reduction with hydrogen. (b) Direct electrochemical reduction of metal ions. (c ) Plot of diameter for MoO2 NWs as a function of deposition time. Deposition was done at a potential of −0.90 V (SCE) using a solution of 0.16 mM Na2 MoO4 , 1.0 mM NaCl, 1.0 M NH4 Cl at pH 8.5. The scanning electron micrographs show HOPG surfaces after deposition of MoO2 for the durations that are indicated. (Reprinted with permission from [81]. Copyright 2003, Wiley-VCH.)

the diameter of wires across a substrate, as well as nanoparticle nucleation at lattice defects other than step edges.

7.3 EXPOSURE OF NANOSCALE EDGES Exposing the edge of a thin film constitutes another approach to producing nanostructures, which is conceptually different from the methods described in the previous section. Thin films having a thickness of 1–100 nm on solid supports can be prepared on a routine basis using a number of techniques. Metal films, for example, are

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

182

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

usually fabricated by physical vapor deposition techniques such as thermal evaporation or sputtering [87], whereas oxides and semiconductor layers are typically grown by molecular beam epitaxy (MBE) [88]. These films can be converted into nanoscale features by taking advantage of their small vertical dimensions. Strategies that have been pursued to this end involve fracturing (see Sections 7.3.1 and 7.3.3), cutting (see Section 7.3.2), and topographic reorientation (see Section 7.3.4)—basic mechanical operations that had, until recently, no apparent connection to nanofabrication. 7.3.1 Fracturing of Thin Films The edge of a thin film deposited on a crystalline semiconductor substrate such as silicon can be exposed by fracturing the substrate. Pfeiffer and co-workers have used fractured multilayered MBE-grown substrates to direct the nucleation of quantum wires and dots [89–91]. These authors also have used an MBE-grown substrate consisting of alternating layers of AlGaAs and GaAs to fabricate an array of field-effect transistors (FETs) with a gate length of ∼20 nm [90]. More recently, Heath and coworkers have shown that the edges of fractured multilayer substrates can be used as templates for the formation of NWs by deposition from the vapor phase (Figure 7.10a) [92–95]. This method has been called superlattice nanowire pattern transfer (SNAP); it uses selective removal of the AlGaAs component in a AlGaAs/GaAs superlattice to generate an array of nanometer-wide grooves. These grooves are then selectively decorated with a metal (e.g., Pt) by tilting the substrate with respect to the deposition direction to produce an array of metal NWs. The width of each wire and the spacing between them can both be controlled by the thickness of the layers that form the superlattice. The array of metal NWs can be transferred onto another substrate coated with an adhesive layer. For example, arrays of Pt NWs as small as 8 nm in width spaced by 16 nm have been printed successfully onto a polymer-coated Si wafer [92]. Cross-bar arrays comprising two sets of parallel wires that overlap perpendicular to each other have been obtained by performing the SNAP process twice on the same substrate. It is possible to further transfer the printed NW array into the underlying Si using RIE (Figure 7.10b). In conjunction with appropriate doping procedures, arrays comprising SNAP Si NWs have been used, for example, in demultiplexing circuits [93], or in FETs where they showed remarkable performance as chemical sensor elements [95]. A narrow edge can also serve as a template to direct assembly processes. For example, Artemyev and co-workers have shown that CdSe nanorods (7 nm in diameter and 35 nm in length) can be oriented on the exposed edge of a ZnS film (5 nm in width and functionalized with a SAM of 1,6-hexanedithiol) supported on a BaF2 substrate [96]. 7.3.2 Sectioning of Encapsulated Thin Films In this approach, nanoscale edges are produced by taking the cross-section of a thin film (or another suitable feature) embedded in a proper matrix. Soft polymers are generally preferable as embodiment because these materials are easier to be processed than hard and brittle ones, especially when using conventional cutting and

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.3 EXPOSURE OF NANOSCALE EDGES

(a)

183

GaAs

BOE Etch

AI0.8Ga0.2As

O2 Plasma

(b)

Figure 7.10. Formation of aligned NW arrays using the SNAP technique. (a) Etching the AlGaAs component of a fractured AlGaAs/GaAs superlattice results in a topographic nanostructure that is selectively decorated with a thin metal film. Resultant wires are then transferred to a silicon substrate coated with an adhesive layer, and subsequently released by etching of the support structures. O2 plasma treatment finally removes the adhesive layer in between wires. (Reprinted with permission from [92]. Copyright 2003, American Association for the Advancement of Science.) (b) Scanning electron micrograph of an array of silicon NWs produced by the SNAP technique. A templated array of Pt NWs was transferred onto an SOI wafer coated with PMMA/epoxy serving as an adhesion layer. Upon drying, the superlattice was released by selective etching and the pattern was transferred into the underlying Si by using the Pt NWs as a mask for RIE. The width of each NW is 17 nm; the pitch in the array is 34 nm. (Reprinted with permission from [94]. Copyright 2006, American Chemical Society.)

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

184

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

abrasion tools. White and co-workers have applied this approach to the fabrication of encapsulated, ultrasmall band electrodes [97]. These authors deposited a thin layer of Pt on cleaved mica, followed by embedding the deposit in epoxy resin. Upon curing, the electrode was exposed by ablation of the epoxy matrix perpendicular to the deposited Pt film using silicon carbide paper. Later, Crooks and co-workers have fabricated even smaller electrodes by exposing the cross-section of a single carbon nanotube embedded in polyphenol [98]. More recently, Whitesides and co-workers have demonstrated sectioning of a topographically patterned epoxy mold that supports a thin, evaporated film of Au by using a microtome [99–102]. This process, called nanoskiving, results in encapsulated arrays of fine gold features. Depending on the design of the mold and the cutting direction, a number of linear, closed, and openloop metallic nanostructures can be fabricated using this technique. Figure 7.11a illustrates the procedure for making arrays of free-standing Au NWs on a solid support [101]. The process involves photolithography, deposition of gold from the vapor phase, and sectioning. The thickness of the sectioned epoxy slabs can be as small as 30–100 nm. Upon collection on a solid substrate, the epoxy matrix can be removed by exposure to oxygen plasma yielding an array of oriented, free-standing NWs on the substrate (Figure 7.11b). Wires with different cross-sectional dimensions have been obtained by changing both thickness of the metal film and width of sectioning, as exemplified by the images in Figure 7.11c. These wires have been characterized by means of optical scattering, which revealed a shift of the peak maxima to longer wavelengths with increase in height (Figure 7.11d). Moreover, L- and U-shaped Au nanostructures fabricated in a similar manner have been shown to serve as mid-IR band-stop filters [102]. The authors also have demonstrated that it is possible to section multilayer sandwich structures composed of Ni and SiO2 [99]. Preventing delamination between encapsulated thin films and the surrounding polymer during the cutting process seems to be a major challenge to this approach. One prerequisite to this end is modification of the surfaces (e.g., via exposure to oxygen plasma) to enhance adhesion with the epoxy resin [99]. Delamination further can be limited by sectioning the matrix at lower temperatures (e.g., −120◦ C) and optimal orientation of the sample with respect to the cutting direction [99]. However, maintaining the integrity of an original sample is by no means straightforward. In fact, distortion of microtomed tissue, for example, is a widely recognized phenomenon that demands for applying appropriate correction algorithms to accurately reconstruct a histological specimen from its corresponding sections [103]. 7.3.3 Thin Metallic Films along Sidewalls of Patterned Stamps This approach was demonstrated by Whitesides and co-workers who collimated deposition of a metal onto a topographically patterned PDMS stamp followed by selective removal of the metal film from the top regions of the raised features via transfer to a proper substrate (Figure 7.12a) [104]. Conceptually, this method borrows from principles of nanotransfer printing—a variant of µCP that was earlier developed by Rogers and co-workers [105–107]. These authors have demonstrated that thin metallic films can be transferred from an elastomeric stamp via conformal contact provided

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

185

7.3 EXPOSURE OF NANOSCALE EDGES

(a) Photoresist

(b) Embed in epoxy prepolymer and cure Epoxy

Si Deposit Au

Section with microtome

2 µm

45° 10 nm × 100 nm

(c)

Si

10 nm × 70 nm

Lift off the photoresist

10 nm × 50 nm 10 nm × 30 nm

Si Coat epoxy prepolymer

Substrate

Si Cure and peel off the epoxy slab

Au nanowire z

Epoxy y x

(d) Intensity (a.u.)

Place section on silicon and oxygen plasma remove epoxy

10 nm × 50 nm 10 nm × 70 nm 10 nm × 30 nm 10 nm × 100 nm

1

0.5

0 500

600 700 800 Wavelength (nm)

Figure 7.11. Fabrication of metal NWs through sectioning with a microtome. (a) The fabrication process illustrated here involves the deposition of a thin Au film onto patterned photoresist followed by liftoff. The remaining Au pattern is then embedded in epoxy resin and sectioned with a microtome. The length (x ) of resultant wires is defined by photolithography, the width (y ) by the thickness of the deposited Au film, and the height (z ) by the microtome. (b) Scanning electron micrograph of free-standing Au NWs viewed at an angle of 45◦ . The wires have dimensions of ∼2 µm × 10 nm × 100 nm. The inset shows a high magnification image of the same sample. (c ) Dark-field optical microscope images of Au NWs having different cross-sectional dimensions (x = 2 µm, y = 10 nm, and z = 30–100 nm). The scale bar corresponds to 20 µm. (d ) SPR scattering spectra of the Au NWs shown in (c ). (Reprinted with permission from [101]. Copyright 2006, Wiley-VCH.)

that the metal film adheres more strongly to the substrate than to the surface of the stamp. Interfacial chemistries that promote transfer of metal films use either covalent bonding of coinage metals via thiol-terminated organic monolayers [105, 107, 108], or condensation of hydroxyl-terminated oxides [105, 106] as employed for the process shown in Figure 7.12a. While being connected to a conductive metal film, resultant edges can be used as nanoscale electrodes to generate patterns of charge in thin dielectric films supported on a solid substrate using electrical microcontact printing (e-µCP) [109] (Figure 7.12b). It has been shown that patterns of both positive and negative charge can be generated in a number of polymeric and inorganic dielectric materials using this approach. Characterization of embedded potentials using Kelvin probe force microscopy (KFM) [110] revealed a width that was typically on the order of 300 nm for metal films that were 10–40 nm thick, as exemplified by the image in Figure 7.12c.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

186

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

(a)

(b) Au/Ti

Au/Ti PDMS

PDMS V

Adhere substrate to an oxidized PDMS slab

PMMA

Si Remove PDMS stamp

Plasma-treated PDMS

Pattern of charge

Au/Ti

PMMA

Si

PDMS

∼300 nm (c)

Remove PDMS slab Metal edge

PDMS

Freshly exposed PDMS Au/Ti

0 µm

(V)

10 0

20 10

20

30 30

Figure 7.12. Formation and use of a patterned PDMS stamp comprising thin metallic edges. (a) Schematic illustration of the fabrication process. A patterned PDMS stamp is first coated with a thin layer of Au and Ti, respectively, and then brought into contact with a planar, oxidized PDMS slab to remove the metal layer form the raised features of the stamp. (b) Schematic illustration of patterning charge in a dielectric thin film using e-µCP. The shape of the pattern corresponds to the outline of the relief features on the stamp. (c ) KFM image of a charge pattern (positive surface potential) embedded in a layer of PMMA using e-µCP. (Reprinted with permission from [104]. Copyright 2005, Wiley-VCH.)

7.3.4 Topographic Reorientation This approach has been introduced by Whitesides and co-workers who have shown that ordered arrays of nanoscale edges can be produced by directing the collapse of metal-capped microposts on the surface of a solid support [111]. The authors used epoxy resin to first create an array of posts on a silicon substrate (Figure 7.13a). Each post was capped selectively with a thin layer of Pd through electron-beam (e-beam)mediated deposition of the metal at a tilt angle of 20◦ with respect to the surface plane. These structures (called microdominos) fracture from the supporting substrate when a shear force is applied using a planar PDMS slab, thereby exposing nanoscale Pd edges. The process can generate uniform patterns of collapsed microdominos over an area as large as ∼1 cm2 . Interestingly, the reoriented microdominos adhere to the PDMS slab used to apply a shear force. It is possible to electrically connect each of the arrayed edges by transferring the collapsed microdominos to a PDMS slab that was previously coated with a layer of conductive polymer such as polyaniline (PANI) (Figure 7.13b). Such an array of linear nanoscale electrodes can be used to

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

7.4 CONCLUSIONS AND OUTLOOK

(a)

(b)

Horizontal shear

PDMS

187

Pd

θ = 20°

Si

Silicon wafer

Deposit metal

Separate

PDMS

15 µm PDMS SU-8

(c)

∼600 nm

500 mV Pd edge Rotate PDMS SU-8

Pd edge SU-8

PDMS

20 µm

PDMS 0 µm

Figure 7.13. Exposure of Pd edges through collapse of microdominos. (a) Schematic illustration of shear-induced reorientation of photolithographically prepared SU-8 posts using a planar PDMS slab. An array of exposed Pd edges can be obtained when the top part of each post was decorated selectively with a thin layer of Pd prior to collapse. The reoriented features adhere to the PDMS slab when separated from the surface. (Reprinted with permission from [13]. Copyright 2004, Annual Reviews.) (b) Scanning electron micrograph of collapsed microdominos on a PANI-coated PDMS substrate exposing a Pd edge of ∼15 nm in thickness. (Reprinted with permission from [111]. Copyright 2004, Wiley-VCH.) (c ) KFM image revealing the surface potential of a PMMA thin film with 600-nm wide regions of embedded charge printed with an array of exposed Pd edges that were supported on a conductive PDMS substrate using e-µCP. (Reprinted with permission from [111]. Copyright 2004, Wiley-VCH.)

print a pattern of charge into a thin film of PMMA (Figure 7.13c) [111]. The authors further have demonstrated that arrays comprising features of different geometry and orientation can be fabricated by changing both the shape of the posts and the direction at which shear force is applied.

7.4 CONCLUSIONS AND OUTLOOK Among current patterning techniques, edge lithography stands out as a striking example of how relatively simple and experimentally unsophisticated operations can be turned into effective tools for nanofabrication. Translating the edge of a topographic structure or that of a thin film into a nanoscale feature represents a concept to miniaturization that is both ingenious and truly unconventional. Over the course of the past 10 years, this concept has been diversified resulting in a variety of approaches that

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

188

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

allow one to access nanostructures without the need of resorting to elaborative and cost-intensive equipment. The minimum feature size that ultimately can be achieved may vary for each edge-lithographic technique, but proof-of-principle demonstrations for a number of approaches suggest capabilities that are comparable to those of other nanofabrication methods such as NIL, µCP, and, to some extent, e-beam methods. As illustrated by several examples, edge lithography is capable of producing functional nanostructures that can serve, depending on their size, morphology, and composition, as nanoelectrodes, optical or conductive elements in sensors, or catalytic sites for chemical reactions. Some of the features are unique in the sense that they would be difficult if not impossible to be realized using other techniques, which further accounts for the current popularity of edge lithography in (mostly) academic environments concerned with the study of nanoscale systems. Although a number of approaches hold promise for applications outside a research laboratory, it is not always obvious whether they can evolve toward robust and efficient fabrication schemes. A notable exception to this end is phase-shift photolithography, which has reached a high level of maturity compared to most other techniques; it is being actively pursued by several semiconductor and microelectronic manufacturers for various fabrication or prototyping purposes. It is evident however, that edge lithography is not a universal approach to nanopatterning. In fact, most of its techniques are, at present, suitable only for producing regular and relatively simple patterns as their abilities to create features of arbitrary shape and geometry are limited. Also, the fabrication of larger and smaller structures in a parallel fashion seems to be nontrivial in many cases. Moreover, edge lithography replicates structural information provided by a mask, mold, or master, and therefore remains inherently complementary to standard manufacturing schemes capable of generating an original pattern. Nevertheless, it is likely that ongoing research efforts will continue fostering novel, unprecedented avenues that may overcome some of the current constraints and broaden the range of possibilities to produce, modify, and assemble nanoscale structures using edgerelated fabrication techniques.

REFERENCES 1. Madou, M. J. (2002) Fundamentals of Microfabrication: The Science of Miniaturization, 2nd edn., CRC Press, Boca Raton. 2. Wallraff, G. M. and Hinsberg, W. D. (1999) Lithographic imaging techniques for the formation of nanoscopic features. Chem. Rev. 99, 1801–1821. 3. Ito, T. and Okazaki, S. (2000) Pushing the limits of lithography. Nature 406, 1027–1031. 4. Chiu, G. L.-T. and Shaw, J. M., eds. (1997) Optical lithography. IBM J. Res. Dev. 41, 3–158. 5. Arden, W., Cogez, P., Graef, M., Ishiuchi, H., Osada, T., Moon, J.-T., Sohn, H.-C., Liang, M.-S., et al., eds. (2006) International Technology Roadmap for Semiconductors, Semiconductor Industry Association, San Jos´e. 6. Orloff, J., Utlaut, M., and Swanson, L. (2003) High Resolution Focused Ion Beams: FIB and Its Applications, Kluwer, New York.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

REFERENCES

189

7. Dhaliwal, R. S., Enichen, W. A., Golladay, S. D., Gordon, M. S., Kendall, R. A., Lieberman, J. E., Pfeiffer, H. C., Pinckney, D. J., et al. (2001) PREVAIL—electron projection technology for next-generation lithography. IBM J. Res. Dev. 45, 615–638. 8. Xia, Y. and Whitesides, G. M. (1998) Soft lithography. Angew. Chem. Int. Ed. 37, 550–575. 9. Xia, Y., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1999) Unconventional methods for fabricating and patterning nanostructures. Chem. Rev. 99, 1823–1848. 10. Mirkin, C. A. and Rogers, J. A., eds. (2001) Emerging methods for micro- and nanofabrication. MRS Bull. 26, 506–546. 11. Michel, B., Bernard, A., Bietsch, A., Delamarche, E., Geissler, M., Juncker, D., Kind, H., Renault, J.-P., et al. (2001) Printing meets lithography: soft approaches to highresolution patterning. IBM J. Res. Dev. 45, 697–719. 12. Sotomayor Torres, C. M., ed. (2003) Alternative Lithography: Unleashing the Potentials of Nanotechnology, Kluwer, New York. 13. Gates, B. D., Xu, Q., Love, J. C., Wolfe, D. B., and Whitesides, G. M. (2004) Unconventional nanofabrication. Annu. Rev. Mater. Res. 34, 339–372. 14. Geissler, M. and Xia, Y. (2004) Patterning: principles and some new developments. Adv. Mater. 16, 1249–1269. 15. Gates, B. D., Xu, Q., Stewart, M., Ryan, D., Willson, C. G., and Whitesides, G. M. (2005) New approaches to nanofabrication: molding, printing, and other techniques. Chem. Rev. 105, 1171–1196. 16. Guo, L. J. (2007) Nanoimprint lithography: methods and material requirements. Adv. Mater. 19, 495–513. 17. Xia, Y., McClelland, J. J., Gupta, R., Qin, D., Zhao, X.-M., Sohn, L. L., Celotta, R. J., and Whitesides, G. M. (1997) Replica molding using polymeric materials: a practical step toward nanomanufacturing. Adv. Mater. 9, 147–149. 18. Kumar, A., Biebuyck, H. A., and Whitesides, G. M. (1994) Patterning self-assembled monolayers: applications in materials science. Langmuir 10, 1498–1511. 19. Delamarche, E., Schmid, H., Bietsch, A., Larsen, N. B., Rothuizen, H., Michel, B., and Biebuyck, H. (1998) Transport mechanisms of alkanethiols during microcontact printing on gold. J. Phys. Chem. B 102, 3324–3334. 20. Geissler, M., Wolf, H., Stutz, R., Delamarche, E., Grummt, U.-W., Michel, B., and Bietsch, A. (2003) Fabrication of metal nanowires using microcontact printing. Langmuir 19, 6301–6311. 21. Kind, H., Geissler, M., Schmid, H., Michel, B., Kern, K., and Delamarche, E. (2000) Patterned electroless deposition of copper by microcontact printing palladium(II) complexes on titanium-covered surfaces. Langmuir 16, 6367–6373. 22. Renault, J.-P., Bernard, A., Bietsch, A., Michel, B., Bosshard, H. R., Delamarche, E., Kreiter, M., Hecht, B., et al. (2003) Fabricating arrays of single protein molecules on glass using microcontact printing. J. Phys. Chem. B 107, 703–711. 23. Kr¨amer, S., Fuierer, R. R., and Gorman, C. B. (2003) Scanning probe lithography using self-assembled monolayers. Chem. Rev. 103, 4367–4418. 24. Ginger, D. S., Zhang, H., and Mirkin, C. A. (2004) The evolution of dip-pen nanolithography. Angew. Chem. Int. Ed. 43, 30–45. 25. Eigler, D. M. and Schweizer, E. K. (1990) Positioning single atoms with a scanning tunnelling microscope. Nature 344, 524–526.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

190

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

26. Vettiger, P., Despont, M., Drechsler, U., D¨urig, U., H¨aberle, W., Lutwyche, M. I., Rothuizen, H. E., Stutz, R., et al. (2000) The millipede—more than one thousand tips for future AFM data storage. IBM J. Res. Dev. 44, 323–340. 27. Salaita, K., Wang, Y., Fragala, J., Vega, R. A., Liu, C., and Mirkin, C. A. (2006) Massively parallel dip-pen nanolithography with 55 000-Pen two-dimensional arrays. Angew. Chem. Int. Ed. 45, 7220–7223. 28. Whitesides, G. M. and Grzybowski, B. A. (2002) Self-assembly at all scales. Science 295, 2418–2421. 29. Park, M., Harrison, C., Chaikin, P. M., Register, R. A., and Adamson, D. H. (1997) Block copolymer lithography: periodic arrays of ∼1011 holes in 1 square centimeter. Science 276, 1401–1404. 30. Haynes, C. L. and Van Duyne, R. P. (2001) Nanosphere lithography: a versatile nanofabrication tool for studies of size-dependent nanoparticle optics. J. Phys. Chem. B 105, 5599–5611. 31. Levenson, M. D., Viswanathan, N. S., and Simpson, R. A. (1982) Improving resolution in photolithography with a phase-shifting mask. IEEE Trans. Electron Devices 29, 1828–1836. 32. Love, J. C., Estroff, L. A., Kriebel, J. K., Nuzzo, R. G., and Whitesides, G. M. (2005) Self-assembled monolayers of thiolates on metals as a form of nanotechnology. Chem. Rev. 105, 1103–1169. 33. Toh, K. K. H., Dao, G., Singh, R., and Gaw, H. (1990) Chromeless phase-shifted masks: a new approach to phase-shifting masks. Proc. SPIE, 10th Annual Symposium on Microlithography 1496, 27–53. 34. Rogers, J. A., Paul, K. E., Jackman, R. J., and Whitesides, G. M. (1997) Using an elastomeric phase mask for sub-100 nm photolithography in the optical near field. Appl. Phys. Lett. 70, 2658–2660. 35. Aizenberg, J., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1997) Imaging the irradiance distribution in the optical near field. Appl. Phys. Lett. 71, 3773–3775. 36. Aizenberg, J., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1998) Imaging profiles of light intensity in the near field: applications to phase-shift photolithography. Appl. Opt. 37, 2145–2152. 37. Rogers, J. A., Paul, K. E., Jackman, R. J., and Whitesides, G. M. (1998) Generating ∼90 nanometer features using near-field contact-mode photolithography with an elastomeric phase mask. J. Vac. Sci. Technol. B 16, 59–68. 38. Rogers, J. A., Dodabalapur, A., Bao, Z., and Katz, H. E. (1999) Low-voltage 0.1 µm organic transistors and complementary inverter circuits fabricated with a low-cost form of near-field photolithography. Appl. Phys. Lett. 75, 1010–1012. 39. Yin, Y., Gates, B., and Xia, Y. (2000) A soft lithography approach to the fabrication of nanostructures of single crystalline silicon with well-defined dimensions and shapes. Adv. Mater. 12, 1426–1430. 40. Li, Z.-Y., Yin, Y., and Xia, Y. (2001) Optimization of elastomeric phase masks for nearfield photolithography. Appl. Phys. Lett. 78, 2431–2433. 41. Fritze, M., Tyrrell, B., Astolfi, D., Yost, D., Davis, P., Wheeler, B., Mallen, R., Jarmolowicz, J., et al. (2001) Gratings of regular arrays and trim exposures for ultralarge scale integrated circuit phase-shift lithography. J. Vac. Sci. Technol. B 19, 2366– 2370.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

REFERENCES

191

42. Odom, T. W., Thalladi, V. R., Love, J. C., and Whitesides, G. M. (2002) Generation of 30–50 nm structures using easily fabricated, composite PDMS masks. J. Am. Chem. Soc. 124, 12112–12113. 43. Kunz, R. R., Rothschild, M., and Yeung, M. S. (2003) Large-area patterning of ∼50 nm structures on flexible substrates using near-field 193 nm radiation. J. Vac. Sci. Technol. B 21, 78–81. 44. Dang, H., Tan, J. L.-P., and Horn, M. W. (2003) Sub-150 nm, high-aspect-ratio features using near-field phase-shifting contact lithography. J. Vac. Sci. Technol. B 21, 1143–1148. 45. Fritze, M., Tyrrell, B. M., Astolfi, D. K., Lambert, R. D., Yost, D.-R. W., Forte, A. R., Cann, S. G., and Wheeler, B. D. (2003) Subwavelength optical lithography with phaseshift photomasks. Linc. Lab. J. 14, 237–250. 46. Maria, J., Jeon, S., and Rogers, J. A. (2004) Nanopatterning with conformable phase masks. J. Photochem. Photobiol. A 166, 149–154. 47. Jeon, S., Park, J.-U., Cirelli, R., Yang, S., Heitzman, C. E., Braun, P. V., Kenis, P. J. A., and Rogers, J. A. (2004) Fabricating complex three-dimensional nanostructures with high-resolution conformable phase masks. Proc. Natl. Acad. Sci. USA 101, 12428–12433. 48. Jeon, S., Malyarchuk, V., White, J. O., and Rogers, J. A. (2005) Optically fabricated three dimensional nanofluidic mixers for microfluidic devices. Nano Lett. 5, 1351–1356. 49. Maria, J., Malyarchuk, V., White, J., and Rogers, J. A. (2006) Experimental and computational studies of phase shift lithography with binary elastomeric masks. J. Vac. Sci. Technol. B 24, 828–835. 50. Truong, T. T., Lin, R., Jeon, S., Lee, H. H., Maria, J., Gaur, A., Hua, F., Meinel, I., et al. (2007) Soft lithography using acryloxy perfluoropolyether composite stamps. Langmuir 23, 2898–2905. 51. Odom, T. W., Love, J. C., Wolfe, D. B., Paul, K. E., and Whitesides, G. M. (2002) Improved pattern transfer in soft lithography using composite stamps. Langmuir 18, 5314–5320. 52. Paul, K. E., Breen, T. L., Aizenberg, J., and Whitesides, G. M. (1998) Maskless photolithography: embossed photoresist as its own optical element. Appl. Phys. Lett. 73, 2893–2895. 53. Babayan, Y., Barton, J. E., Greyson, E. C., and Odom, T. W. (2004) Templated and hierarchical assembly of CdSe/ZnS quantum dots. Adv. Mater. 16, 1341–1345. 54. Greyson, E. C., Babayan, Y., and Odom, T. W. (2004) Directed growth of ordered arrays of small-diameter ZnO nanowires. Adv. Mater. 16, 1348–1352. 55. Aizenberg, J., Black, A. J., and Whitesides, G. M. (1998) Controlling local disorder in self-assembled monolayers by patterning the topography of their metallic supports. Nature 394, 868–871. 56. Black, A. J., Paul, K. E., Aizenberg, J., and Whitesides, G. M. (1999) Patterning disorder in monolayer resists for the fabrication of sub-100-nm structures in silver, gold, silicon, and aluminum. J. Am. Chem. Soc. 121, 8356–8365. 57. Aizenberg, J. (2000) Patterned crystallisation on self-assembled monolayers with integrated regions of disorder. J. Chem. Soc. Dalton Trans. 21, 3963–3968. 58. Aizenberg, J. (2004) Crystallization in patterns: a bio-inspired approach. Adv. Mater. 16, 1295–1302.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

192

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

59. Sundar, V. C. and Aizenberg, J. (2003) Narrow features in metals at the interfaces between different etch resists. Appl. Phys. Lett. 83, 2259–2261. 60. Yang, H., Love, J. C., Arias, F., and Whitesides, G. M. (2002) Electroforming of copper structures at nanometer-sized gaps of self-assembled monolayers on silver. Chem. Mater. 14, 1385–1390. 61. Love, J. C., Paul, K. E., and Whitesides, G. M. (2001) Fabrication of nanometer-scale features by controlled isotropic wet chemical etching. Adv. Mater. 13, 604–607. 62. McLellan, J. M., Geissler, M., and Xia, Y. (2004) Edge spreading lithography and its application to the fabrication of mesoscopic gold and silver rings. J. Am. Chem. Soc. 126, 10830–10831. 63. Geissler, M., McLellan, J. M., and Xia, Y. (2005) Edge-spreading lithography: use of patterned photoresist structures to direct the spreading of alkanethiols on gold. Nano Lett. 5, 31–36. 64. Geissler, M., McLellan, J. M., Chen, J., and Xia, Y. (2005) Side-by-side patterning of multiple alkanethiolate monolayers on gold by edge-spreading lithography. Angew. Chem. Int. Ed. 44, 3596–3600. 65. McLellan, J. M., Geissler, M., and Xia, Y. (2005) Self-assembly of hexadecanethiol molecules on gold from the vapour phase as directed by a two-dimensional array of silica beads. Chem. Phys. Lett. 408, 80–83. 66. Geissler, M., Chalsani, P., Cameron, N. S., and Veres, T. (2006) Patterning of chemical gradients with submicrometer resolution using edge-spreading lithography. Small 2, 760–765. 67. Biebuyck, H. A. and Whitesides, G. M. (1994) Autophobic pinning of drops of alkanethiols on gold. Langmuir 10, 4581–4587. 68. Xia, Y. and Whitesides, G. M. (1995) 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. 69. Noy, A., Frisbie, C. D., Rozsnyai, L. F., Wrighton, M. S., and Lieber, C. M. (1995) Chemical force microscopy: exploiting chemically-modified tips to quantify adhesion, friction, and functional group distributions in molecular assemblies. J. Am. Chem. Soc. 117, 7943–7951. 70. Carpick, R. W. and Salmeron, M. (1997) Scratching the surface: fundamental investigations of tribology with atomic force microscopy. Chem. Rev. 97, 1163–1194. 71. Cherniavskaya, O., Adzic, A., Knutson, C., Gross, B. J., Zang, L., Liu, R., and Adams, D. M. (2002) Edge transfer lithography of molecular and nanoparticle materials. Langmuir 18, 7029–7034. 72. Sharpe, R. B. A., Titulaer, B. J. F., Peeters, E., Burdinski, D., Huskens, J., Zandvliet, H. J. W., Reinhoudt, D. N., and Poelsema, B. (2006) Edge transfer lithography using alkanethiol inks. Nano Lett. 6, 1235–1239. 73. de la Figuera, J., Huerta-Garnica, M. A., Prieto, J. E., Ocal, C., and Miranda, R. (1995) Fabrication of magnetic quantum wires by step-flow growth of cobalt on copper surfaces. Appl. Phys. Lett. 66, 1006–1008. 74. Petrovykh, D. Y., Himpsel, F. J., and Jung, T. (1998) Width distribution of nanowires grown by step decoration. Surf. Sci. 407, 189–199. 75. Morin, S., Lachenwitzer, A., Magnussen, O. M., and Behm, R. J. (1999) Potentialcontrolled step flow to 3D step decoration transition: Ni electrodeposition on Ag(111). Phys. Rev. Lett. 83, 5066–5069.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

14:1

Printer Name: Yet to Come

REFERENCES

193

76. Zach, M. P., Ng, K. H., and Penner, R. M. (2000) Molybdenum nanowires by electrodeposition. Science 290, 2120–2123. 77. Gambardella, P., Blanc, M., Brune, H., Kuhnke, K., and Kern, K. (2000) Onedimensional metal chains on Pt vicinal surfaces. Phys. Rev. B 61, 2254–2262. 78. Favier, F., Walter, E. C., Zach, M. P., Benter, T., and Penner, R. M. (2001) Hydrogen sensors and switches from electrodeposited palladium mesowire arrays. Science 293, 2227–2231. 79. Penner, R. M. (2002) Mesoscopic metal particles and wires by electrodeposition. J. Phys. Chem. B 106, 3339–3353. 80. Walter, E. C., Murray, B. J., Favier, F., Kaltenpoth, G., Grunze, M., and Penner, R. M. (2002) Noble and coinage metal nanowires by electrochemical step edge decoration. J. Phys. Chem. B 106, 11407–11411. 81. Walter, E. C., Zach, M. P., Favier, F., Murray, B. J., Inazu, K., Hemminger, J. C., and Penner, R. M. (2003) Metal nanowire arrays by electrodeposition. Chem. Phys. Chem. 4, 131–138. 82. Li, Q., Newberg, J. T., Walter, E. C., Hemminger, J. C., and Penner, R. M. (2004) Polycrystalline molybdenum disulfide (2H-MoS2 ) nano- and microribbons by electrochemical/chemical synthesis. Nano Lett. 4, 277–281. 83. Li, Q., Olson, J. B., and Penner, R. M. (2004) Nanocrystalline α-MnO2 nanowires by electrochemical step-edge decoration. Chem. Mater. 16, 3402–3405. 84. Menke, E. J., Li, Q., and Penner, R. M. (2004) Bismuth telluride (Bi2 Te3 ) nanowires synthesized by cyclic electrodeposition/stripping coupled with step edge decoration. Nano Lett. 4, 2009–2014. 85. Himpsel, F. J. and Ortega, J. E. (1994) Edge state and terrace state for Cu on W(331) and W(110). Phys. Rev. B 50, 4992–4995. 86. Jung, T., Mo, Y. W., and Himpsel, F. J. (1995) Identification of metals in scanning tunneling microscopy via image states. Phys. Rev. Lett. 74, 1641–1644. 87. Harsha, K. S. S. (2006) Principles of Physical Vapor Deposition of Thin Films, Elsevier, Oxford. 88. Herman, M. A., Richter, W., and Sitter, H. (2004) Epitaxy—Physical Principles and Technical Implementations, Springer, Berlin. 89. Pfeiffer, L., West, K. W., Stormer, H. L., Eisenstein, J. P., Baldwin, K. W., Gershoni, D., and Spector, J. (1990) Formation of a high quality two-dimensional electron gas on cleaved GaAs. Appl. Phys. Lett. 56, 1697–1699. 90. Stormer, H. L., Baldwin, K. W., Pfeiffer, L. N., and West, K. W. (1991) GaAs field-effect transistor with an atomically precise ultrashort gate. Appl. Phys. Lett. 59, 1111–1113. 91. Natelson, D., Willett, R. L., West, K. W., and Pfeiffer, L. N. (2000) Fabrication of extremely narrow metal wires. Appl. Phys. Lett. 77, 1991–1993. 92. Melosh, N. A., Boukai, A., Diana, F., Gerardot, B., Badolato, A., Petroff, P. M., and Heath, J. R. (2003) Ultrahigh-density nanowire lattices and circuits. Science 300, 112–115. 93. Beckman, R., Johnston-Halperin, E., Luo, Y., Green, J. E., and Heath, J. R. (2005) Bridging dimensions: demultiplexing ultrahigh-density nanowire circuits. Science 310, 465–468. 94. Wang, D., Sheriff, B. A., and Heath, J. R. (2006) Silicon p-FETs from ultrahigh density nanowire arrays. Nano Lett. 6, 1096–1100.

P1: OTA/XYZ P2: ABC c07 JWBK252/Lee September 18, 2008

194

14:1

Printer Name: Yet to Come

PATTERNING BASED ON EDGE EFFECTS

95. McAlpine, M. C., Ahmad, H., Wang, D., and Heath, J. R. (2007) Highly ordered nanowire arrays on plastic substrates for ultrasensitive flexible chemical sensors. Nat. Mater. 6, 379–384. 96. Artemyev, M., M¨oller, B., and Woggon, U. (2003) Unidirectional alignment of CdSe nanorods. Nano Lett. 3, 509–512. 97. Morris, R. B., Franta, D. J., and White, H. S. (1987) Electrochemistry at Pt band electrodes of width approaching molecular dimensions. Breakdown of transport equations at very small electrodes. J. Phys. Chem. 91, 3559–3564. 98. Campbell, J. K., Sun, L., and Crooks, R. M. (1999) Electrochemistry using single carbon nanotubes. J. Am. Chem. Soc. 121, 3779–3780. 99. Xu, Q., Gates, B. D., and Whitesides, G. M. (2004) Fabrication of metal structures with nanometer-scale lateral dimensions by sectioning using a microtome. J. Am. Chem. Soc. 126, 1332–1333. 100. Xu, Q., Perez-Castillejos, R., Li, Z., and Whitesides, G. M. (2006) Fabrication of highaspect-ratio metallic nanostructures using nanoskiving. Nano Lett. 6, 2163–2165. 101. Xu, Q., Bao, J., Capasso, F., and Whitesides, G. M. (2006) Surface plasmon resonances of free-standing gold nanowires fabricated by nanoskiving. Angew. Chem. Int. Ed. 45, 3631–3635. 102. Xu, Q., Bao, J., Rioux, R. M., Perez-Castillejos, R., Capasso, F., and Whitesides, G. M. (2007) Fabrication of large-area patterned nanostructures for optical applications by nanoskiving. Nano Lett. 7, 2800–2805. 103. Jones, A. S., Milthorpe, B. K., and Howlett, C. R. (1994) Measurement of microtomy induced section distortion and its correction for 3-dimensional histological reconstructions. Cytometry 15, 95–105. 104. Cao, T., Xu, Q., Winkleman, A., and Whitesides, G. M. (2005) Fabrication of thin, metallic films along the sidewalls of a topographically patterned stamp and their application in charge printing. Small 1, 1191–1195. 105. Loo, Y.-L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Interfacial chemistries for nanoscale transfer printing. J. Am. Chem. Soc. 124, 7654–7655. 106. Loo, Y.-L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Additive, nanoscale patterning of metal films with a stamp and a surface chemistry mediated transfer process: applications in plastic electronics. Appl. Phys. Lett. 81, 562–564. 107. Loo, Y.-L., Lang, D. V., Rogers, J. A., and Hsu, J. W. P. (2003) Electrical contacts to molecular layers by nanotransfer printing. Nano Lett. 3, 913–917. 108. Felmet, K., Loo, Y.-L., and Sun, Y. (2004) Patterning conductive copper by nanotransfer printing. Appl. Phys. Lett. 85, 3316–3318. 109. Jacobs, H. O. and Whitesides, G. M. (2001) Submicrometer patterning of charge in thinfilm electrets. Science 291, 1763–1766. 110. Jacobs, H. O., Knapp, H. F., and Stemmer, A. (1999) Practical aspects of Kelvin probe force microscopy. Rev. Sci. Instrum. 70, 1756–1760. 111. Gates, B. D., Xu, Q., Thalladi, V. R., Cao, T., Knickerbocker, T., and Whitesides, G. M. (2004) Shear patterning of microdominos: a new class of procedures for making microand nanostructures. Angew. Chem. Int. Ed. 43, 2780–2783.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8 PATTERNING WITH ELECTROLYTE: SOLID-STATE SUPERIONIC STAMPING Keng H. Hsu, Peter L. Schultz, Nicholas X. Fang, and Placid M. Ferreira

8.1 INTRODUCTION Metallic structures are ubiquitous in micro- and nanotechnology. From interconnects in electronics to electrodes in sensors, batteries, and fuel cells, they play a pivotal role in the performance of devices [1–4]. Such metallic structures are also becoming increasingly important in emerging fields related to subwavelength optics, such as surface plasmons and plasmonic waveguides [5, 6]. Because metallic structures are such an integral part of ever-shrinking micro- and nanoscale devices and systems, it is of critical importance to be able to economically manufacture them at these length scales. However, the common practice for generating metallic patterns has relied on an indirect approach. For example, micro- or nanoscale patterns are first lithographically created in photoresist that is used as a sacrificial mold. Metal is deposited by evaporation or sputtering and the subsequent liftoff of the polymer and excess metal leaves behind the metal pattern [1]. Similarly, in the damascene process [7], pursued by the semiconductor industry, copper interconnects are created by electrochemically depositing copper into trenches patterned in a dielectric film. The inlaid metallic interconnect patterns are left behind after chemical–mechanical polishing is used to remove excess metal. Such indirect processes tend to be

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


195

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

196

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

expensive and complex, and require multiple process steps (around 20 per layer [8]) with stringent process environment control and very costly equipment. There has been a renewed interest in direct patterning methods for metals in the electronics industry because of the high resistivity encountered in conductors with shrinking lateral dimensions. As the lateral dimensions approach the mean free electron path, roughly around 40 nm at 25◦ C, the electrical resistivity increases rapidly [9] (e.g., a 50-nm wire has twice the bulk resistivity of copper [10]). Grain boundary and sidewall scattering are thought to be the primary reasons. In the damascene process, the narrow trench geometries and impurities are thought to impede the grain growth rendering annealing and other grain growth techniques ineffective. As a result, process sequences in which blanket films are presented for patterning are thought to be advantageous [9]. Direct patterning such as metallic, primarily copper films, using reactive ion etching (RIE) or dry-etch processes, has been attempted by a number of researchers. Steinbruchel [11] gives a comprehensive discussion of the issues involved in RIE patterning of copper. Schwartz and Schaible [12] have used a CCl4 while Ohno et al. [13, 14] have used a SiCl4 mixture with N2 , Cl2 , and NH3 and others [15] have used with BCl3 . In these processes, the high substrate temperatures (typically in the range of 250◦ C) make finding suitable mask materials a challenge. Low temperature etching of such metals is difficult because of the low volatility of the reaction products. Lee and Kuo [16] report the removal of CuCl2 after a Cl2 plasma etch with a dilute solution of HCl. Recent work by Tamirisa et al. [9] reports successful etching of copper at room temperatures by using alternating cycles of H2 and Cl2 plasmas; however, patterning results are yet to be reported. In summary, in spite of a need for direct, dry patterning of metal films such as Cu, the use of RIE processes has proved to be challenging. Less conventional methods, such as nanotransfer printing, use a poly(dimethylsiloxane) (PDMS) stamp to directly place the pattern on the substrate and are suitable for micrometer-sized features; but often require pretreatment of the substrate [3]. Still other attempts have been made at using PDMS-based microcontact printing to deposit an etch resist prior to electrochemical etching of unwanted portions of film [17]. While economical, such techniques are primarily used for copper patterns with critical features between 1 and 500 µm. Nanosphere lithography [18] can be used for creating periodic patterns of metallic structures by using closely packed polystyrene microspheres as a mask. Another method, electrochemical micromachining, using a liquid electrolyte, has been proposed to directly produce submicrometer metallic features [19, 20]. Variations in resistance to flows and the high diffusion lengths of reacting species pose significant limitations on lateral extension of the features associated with the process. Accelerated etching at sharp edges and corners also leads to low geometrical fidelity in the pattern transferred from the electrochemical tool to substrate surface [19]. In addition, the use of liquids might contaminate both the tool and the substrate. Given the widespread use and need for metallic nanostructures, there is a general death of process technology for producing them directly. Therefore, it is of considerable interest to explore approaches that add to our repertoire of metallization

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.2 SOLID-STATE SUPERIONIC STAMPING

197

processes that are capable of directly and efficiently creating metallic structures with general geometries and nanometer-scale resolution.

8.2 SOLID-STATE SUPERIONIC STAMPING Solid-State Superionic Stamping (S4) [21, 22] is a solid-state electrochemical imprinting process that directly creates high resolution metallic nanopatterns in a single step. Similar to imprint lithography [23, 24], but different in that it does not pattern a polymer or use mechanical forces to squeeze it into a pattern, S4 directly imprints metals with an electrochemical reaction. Conceptually, it combines the large area, in-parallel, single-step process economics of nanoimprint lithography with the efficiency, precision, and low mechanical forces of electrochemical machining, as shown in Figure 8.1. At the center of this process is a solid electrolyte or superionic conductor [21]. Widely used in battery and fuel cell applications, due to their excellent ionic conductivity at room and relatively low temperatures, such materials

Techniques

Rate

Resolution

Damascene

∼ 1 nm

Micro-EDM

> 1 µm

Electrochem. micro-machining

< 1 µm

nanoimprint lithography

Limitations

< 100 nm Mold

ECM Electrolyte flow

Tool

PR Metal substrate

e−

− e− e

Solid electrolyte stamp

Metal

Figure 8.1. The solid-state superionic stamping processes combine the ideas of electrochemical machining and nanoimprint lithography, substituting photoresist (PR) for a metal film and liquid electrolyte by a nanopatterned solid electrolyte to produce a method for directly nanoimprinting metals.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

198

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

offer the possibility of precise and efficient control of mass (ion) transport from or to a substrate to be patterned. A stamp made of a superionic conductor with a mobile cation (silver or copper sulfide, for example, in which the silver or copper ions are mobile) is prepatterned with fine features and brought into contact with the metallic substrate to be patterned. On the application of an electrical bias with the substrate as anode and a metallic electrode at the back of the stamp as cathode, a solidstate electrochemical reaction resulting in anodic dissolution of the metal begins at the contact interface with the stamp. At the anode–electrolyte interface, an appreciable potential drop causes the oxidation of metal atoms on the substrate to produce mobile cations. These mobilized ions migrate across the interface and through the interstitial channels and defect network in the lattice of the superionic conductor toward the cathode, until they recombine with electrons. The anodic dissolution progressively removes a metallic layer of the substrate at the contact area with the stamp. Assisted by a nominal pressure to maintain electrical contact, the stamp progresses into the substrate and generates a shape in the metallic substrate complementary to the prepatterned features on it. This idea is shown schematically in Figure 8.2. The advantage of using solid-state superionic conductors is that mass transport is restricted to the physical contact interface between the patterned electrolyte and the substrate (the anode), making it an ideal tool for nanoscale pattern transfer with high fidelity.

Figure 8.2. Schematic of the solid-state superionic stamping process [22]. (a) A prepatterned superionic conductor. In this case, Cu2 S is placed in contact with the substrate to be patterned. (b) An applied voltage with substrate as anode serves to initiate and propagate the anodic dissolution at the stamp–work piece interface. (c ) At the interface, copper atoms are split into ions and electrons, which migrate through the crystal lattice and external circuit, respectively. (d ) Removal of the stamp reveals the complementary pattern left behind. (Reproduced with permission from [22]. Copyright 2007 American Vacuum Society.)

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.3 PROCESS TECHNOLOGY

199

The mobility of ions in an array of solid electrolytes, such as Ag2 S and RbAg4 I5 , has been exploited to create nanostructures in “direct-write” like processes. Sub100-nm line and dot patterns have been written using an STM (scanning tunneling microscope) or AFM (atomic force microscope) tip [25, 26]. These techniques use an electric potential applied across a scanning probe and a solid electrolyte substrate surface to induce the migration of metal ions from within the substrate to the vicinity of the probe to form metallic clusters that create lines or dots on the electrolyte surface [27]. The practicality of this direct pattern writing as a manufacturing process is limited because of low throughput, difficulties in dimensional control of the structures formed, and processing parameters (the standoff distance of the probe must be precisely regulated and its travel speed must be coordinated to the growth of the structure. Most importantly, the metal structures are embedded on the electrolyte surface layer, making their subsequent use in any applications difficult. The S4 process, unlike these processes, is an area-patterning process. It is a subtractive process and produces the metallic pattern removing material from the substrate. Finally, the resulting nanopattern may reside on any substrate on which a metallic film can be deposited. The introduction of soft or imprint lithography techniques [23, 24] for nanopatterning has attracted widespread research and commercial attention because of its ability to pattern large areas at resolutions reaching the single-digit nanometer range. The S4 process is an electrochemical imprinting process and like other imprint lithography processes uses a stamp or mold that is brought into contact with a substrate on which an imprint is to be generated. As a result, it is very compatible with existing imprint equipment and adds to the repertoire of imprint lithography technology by providing a means of directly patterning metals without the need for imprinting polymer masks or molds. To date, we have reported and demonstrated the use of this technique to create copper and silver nanostructures, reaching resolution as low as 30 nm with patterning rates (rate at which the stamp travels into the substrate) in the range of 0.1–5 nm s–1 . Figure 8.3 shows a figure of a stamp as well as the results of the stamping process. The figure also draws correspondence with the process schematic. In the sections that follow, we will describe some of the salient features of this technology; manufacture of S4 stamps; the experimental stamping setup; monitoring the progress of the imprinting process; typical results obtained in using this process. We conclude with applications and directions for further development.

8.3 PROCESS TECHNOLOGY Central to the S4 process is the stamp. As previously mentioned, the stamp material is a superionic conductor in which the metal ions are the mobile species. Generally, one can find candidates for electrochemical stamps from four categories of superionic solids: crystalline, amorphous glassy, polymeric, and composite [28]. Crystalline ionic conductors are further subdivided into soft- and hard-framework crystals, differing in their bond type (ionic or covalent), polarizability (high or low), and

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

200

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

Figure 8.3. (a) A stamp made of silver sulfide and patterned using FIB with nanoscale alphabet and a series of nanoantennae patterns. (b) The results of superionic stamping. (c ), (d ) Isometric closeups of some of the patterning results. (Reproduced with permission from [21]. Copyright 2007 American Chemical Society.)

presence of a sharp order–disorder phase transition. While the amorphous-glassy type has high, isotropic conductivity and the absence of grain boundaries, polymeric electrolytes are characterized by combining polar polymers with ionic salts to achieve reduced weight and flexibility at the cost of lower ionic conductivity. Composite electrolytes result from the combination of several materials to improve the ionic conductivity at room temperature. For the S4 process, it is necessary to choose stamp materials that are chemically stable and display high ionic conductivity at room or reasonably low (around 100◦ C) temperatures. Further, the stamp materials should be dense rather than porous and when polycrystalline should be relatively isotropic in terms of ionic conductivity. Finally, when possible, a ductile and malleable material facilitates ease of stamp preparation. The solid electrolyte system for silver is very rich, while that for copper, especially for room temperature ionic conductors, is quite restrictive. We chose Ag2 S and Cu2 S because of their stability, reasonably good mechanical and electrical

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.3 PROCESS TECHNOLOGY

201

properties, and the relative ease of synthesis. Copper sulfide (Cu2 S), chalcocite, is a known mixed conductor wherein both electron holes and copper ions can carry charge. While its electronic conduction resembles that of a p-type semiconductor, its phase-dependent ionic conduction changes dramatically with temperature [29, 30]. In the γ -phase (orthorhombic) below 105◦ C, Cu2 S exhibits little or no ionic activity. At a transition temperature of around 105◦ C, its crystal structure changes from orthorhombic to hexagonal (β-phase), along with a significant drop in electronic conductivity and a dramatic increase, by a factor of 104 , in ionic conductivity (10−5 −1 cm−1 at 15◦ C to 10−1 −1 cm−1 at 105◦ C) [30]. Finally, at 470◦ C it undergoes another transition from β to α (cubic) in which the ionic conductivity drops dramatically again [31]. This variability of ionic conductivity can be exploited to tune the behavior of the S4 process by changing the operating temperature. Similarly, Ag2 S is a mixed ionic conductor with Ag+ ions carrying charge. In the low temperature β-phase (acanthite) below 177◦ C, the ionic conductivity is a significant portion of the total conductivity and follows the Arrhenius type of relation with temperature, with a hopping activation energy of 0.2 eV [32, 33]. Following a transition to α-phase (argentite) above 177◦ C, the conductivity is predominantly electronic with a smaller dependence on temperature. In both phases, the conductivity is strongly influenced by composition, increasing with higher silver concentration. A large collection of ionic conductors with their ionic conductivities has been compiled by Agrawal and Gupta [28], should the reader wish to explore other stamp materials. The silver or copper sulfide stamps are prepared by first synthesizing a dense silver or copper sulfide pellet in an improvized furnace with programmable temperature control. Sulfur powder with 99.999% purity (from Fisher Scientific Company) is hand pressed into a pellet of diameter 3 mm and inserted into a glass tube (3 mm internal diameter) with a silver or copper pellet (from Kurt J. Lesker Company). The two are then held together with a small constant force (required for obtaining a dense pellet) provided by a spring. The assembly is placed in the furnace at 400◦ C for 10 h to allow formation and annealing of sliver/copper sulfide. The pellets created in this manner are characterized with x-ray diffraction (Rigaku D-Max System with a scanning range (2θ ) from 0◦ to 60◦ and a scan rate of 0.8◦ min–1 ) and compared with standard peaks for the powder forms of β-silver or copper sulfide. Characterization of a number of pellets has confirmed the composition of the silver and copper sulfide pellets and the consistent output of the above process. Following growth, the pellets are machined to produce a conical end with a flat stamping surface (600 µm in diameter) that are either polished using lapping films of 1, 0.3, and 0.01 µm particle sizes, or trimmed with an ultramicrotome and a Diatome diamond knife (Leica EM UC6). Patterning of the stamp is accomplished by both, focused ion beam (FIB) milling as well as embossing (in case of silver sulfide stamps). The patterns on the stamp, such as those shown in Figure 8.4a, were produced by FIB milling (FEI Dual-Beam DB-235) with a 50-pA aperture at a milling rate of about 50 nm min–1 . The deepest trench on the stamp in Figure 8.4a was about 250 nm. We have demonstrated direct embossing of the silver sulfide stamps against silicon masters, such as calibration grids as shown in Figure 8.4b.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

202

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

Figure 8.4. (a) A silver sulfide stamp with a pattern of 70-nm triangles made by FIB along with the results stamping shown below it. (b) A silver sulfide stamp made by embossing against a silicon master along with sub-100-nm stamping results below it.

The setup we have used to date is relatively simple and attests to the robust nature of the process. Shown in Figure 8.5, is the setup used for stamping. The stamp is mounted on a single-axis stage that is used to feed the stamp to the substrate. The stamp is attached to this assembly via an elastomer that provides 48 MPa of pressure at 20% compressive strain uniformly across the actual contact area between the stamp and the substrate to be imprinted. This substrate is mounted on a second stage that allows transverse or lateral positioning relative to the stamp. The nominal pressure (well below the yield stress of silver or copper sulfide) between the stamp and the substrate ensures a consistent contact between them during the progress of electrochemical imprinting. The electrical potential for electrochemical imprinting is controlled by a digital potentiostat (Gamry Instruments Model Reference 600) with a blocking electrode attached to the stamp as the cathode and the substrate being patterned as the anode. The process is performed in the chronoamperometry mode of the potentiostat, keeping the potential between anode and cathode constant, while monitoring the current flowing across them. To maintain the elevated temperature required for enhanced ionic conductivity in Cu2 S when performing electrochemical imprinting of copper, the stamp is enclosed in a circular heater set at 150 ◦ C, while a second heater maintains the substrate at the same temperature.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.4 PROCESS CAPABILITIES

203

b

L c G

A A A

a

V

e

I d

A

Figure 8.5. The S4 process setup. (a) X–Y stage with substrate holder, (b) dual stage for positioning and feeding stamp, (c ) visco-elastic material and stamp holder, (d ) silver sulfide stamp, (e) substrate with silver film, (f ) potentiostat, (g ) objective for observation. (Reproduced with permission from [21]. Copyright 2007 American Chemical Society.)

Our work has thus far concentrated on patterning of silver and copper films. These films are prepared by electron-beam evaporation of silver onto a 300-µm-thick glass cover slip or silicon wafer, cleaned using RCA1 solution. The films are deposited over a 10-nm Cr seed layer at a chamber pressure of 5 × 10−6 Torr and a stable rate of around 1 nm s–1 . Typical film thickness used by us ranges from about 50 to 500 nm, with 100-nm films being used for most of our experiments. For the copper films, prior to use, the substrates are cleaned using acetic acid (99.97+ ) at room temperature for 1 min to remove any oxidation layers on the copper surface according to the technique proposed by Chavez et al. [34]. This step is necessary because the presence of a copper oxide layer was found to inhibit anodic dissolution at the interface and impede the S4 process.

8.4 PROCESS CAPABILITIES An electrochemical imprinting process may be characterized by a number of parameters such as imprinting speed, resolution (finest feature transferred), stamping area, and stamp life. In addition, one may consider how these parameters change with process settings (e.g., dependence of imprinting speed on voltage or temperature, stamp life on imprinting force or voltage). Because this process is relatively new, such dependencies of process output on operating conditions are still being studied. This discussion will therefore compare the relative performance of the process for the two materials, copper and silver. Further, we will not comment on stamping area because, to date, our stamping experiments are conducted on an improvized setup. The stamping area and, to a large extent, stamp life depend on the alignment and control of imprinting force between the stamp and the imprinted substrate.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

204

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

(d)

(a)

(b)

14:2

(f)

(c) (g) (e)

Figure 8.6. Line calibration results. (a), (b) SEM (scanning electron microscope) images of the Cu2 S stamp and resulting pattern imprinted in a 70-nm-thick Cu film. (c )–(e) AFM scan of pattern illustrating spacing and linewidths of 80 nm. (f ), (g ) SEM images of line patterns imprinted in a 100-nm thick Ag film. (h) AFM images and profiles illustrating spacing and linewidth of 50 nm. (Parts a–e are reproduced with permission from [22]. Copyright 2007 American Vacuum Society. Parts (f ) and (g ) are reproduced with permission from [21]. Copyright 2007 American Chemical Society.)

With both materials, silver and copper, we have been able to achieve sub-100-nm resolutions. Figure 8.6 shows bar-code patterns that allow us to measure both positive and negative features imprinted by the process. For silver, we observe good transfer of 50-nm patterns, while for copper, resolutions up to around 80 nm are observed. This also reflects our experience with the process. We have been able to routinely imprint sub-50-nm patterns into silver films, while imprinting of copper requires much greater care in the preparation of the stamp and during the process. This is due, in part, to the mechanical properties of the Cu2 S stamp. Cu2 S is brittle and hard. It tends to chip easily during the making or handling of the stamp. Not being malleable, it requires a higher contact force to elastically deform and make uniform electrical contact with the substrate to be patterned, making the finer features on the stamp susceptible to damage. With respect to stamping rates, imprinting silver with silver sulfide at room temperature is faster than imprinting copper at elevated temperatures. Figure 8.7 shows the imprinting speeds, i.e., speed of travel of the stamp into the film being patterned, for silver and copper. For imprinting of silver, imprinting speeds tend to be high (typically greater than 1 nm s–1 and as high as 4 nm s–1 ) and constant (see Figure 8.7). The average rates for copper are high enough to make the process economically competitive with other patterning process, but they are lower than those observed for silver. As shown in Figure 8.7 for patterning of 100-nm copper films at 150◦ C, we observed an average of 0.295 nm s–1 at 300 mV, 0.583 nm s–1 at 600 mV, and 0.694 nm s–1 at

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.4 PROCESS CAPABILITIES

205

0.6 (a)

Etch depth

8

0.3

4

Etch rate (nm s−1)

Etch depth (µm)

12

Etch rate 0

0.0 0 (b) 160

50

100 150 Time steps (s)

200

250

0.3 V 0.6 V 0.9 V

140

Depth (nm)

120 100 80 60 40 20 50

100

150

200

250

300

Time (s) Figure 8.7. Travel of the stamp into the film as a function of time for (a) silver and (b) copper. Silver has a constant imprint speed or etch rate while copper has a high initial speed before settling to a constant lower imprint speed. (Part (a) is reproduced with permission from [21]. Copyright 2007 American Chemical Society. Part (b) is reproduced with permission from [22]. Copyright 2007 American Vacuum Society.)

900 mV. The etch rate is considerably higher at the onset, but appears to slow down significantly as the etching progresses to a constant value of about 0.2 nm s–1 , as shown in the figure. The high etching rates at the start are attributable to the high copper concentration gradient and the thermal gradient at the contact interface at the onset of the process. During this time, the patterning rate is determined by the supply of mobile ions that, in turn, is determined by the rate of anodic dissolution.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

206

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

0.10

B 0.9 V

0.06 0.7 V 0.04 0.02 0.00

0

0.5 V 0.3 V 0.1 V 100 200 300 400 500 600

0.10 0.08 0.06 0.04 0.02 0.00 0.0

0.2

0.4

0.6

0.8

1.0

Applied potential (V)

Time (s) Current density (mA cm−1)

Current (A)

Current (A)

0.08

(b)

(a) 4 200 mV 400 mV 600 mV 800 mV

2

0

0

100

200

Time (s) (c) Figure 8.8. (a) The influence of increasing voltage bias on the current during imprinting of copper with copper sulfide. (b) Plot of initial (at 1 s) electronic current versus applied potential for the Cu2 S/Cu process. (Parts (a) and (b) are reproduced with permission from [22]. Copyright 2007 American Vacuum Society.) (c ) Current density profiles while etching Ag with Ag2 S. In this case, the depletion of the film is clearly evident because of the high transference number for Ag2 S, indicating higher patterning speeds with increased voltages. (Part (c ) is reproduced with permission from [21]. Copyright 2007 American Chemical Society.)

Subsequently, the imprinting rate is limited by the diffusion rate of the mobile Cu+2 ions in Cu2 S. This low etching rate compared to the Ag2 S/Ag system can, in part, be attributed to its low transference number at 150◦ C, 0.00003 as opposed to 0.11 in Ag2 S [35]. The dependence of imprinting speed on applied voltage is shown in Figure 8.8. In both cases, the increased current indicates an increased etch rate (though only a fraction of increase in current can be attributed to increased imprinting rates as the stamp materials are mixed, not pure, ionic conductors). The S4 process is demonstrated to be repeatable from one imprint cycle to the next. Figure 8.9 shows the current profiles for imprinting of silver and copper. In both cases the current profile quickly settles to a steady state profile that displays three stages: a sharp ramp-up stage during which the anodic dissolution accelerates with activation of the electrochemical reaction at the anode, a second stage during which a slow decrease in current is observed as the stamp gets polarized, and a sharp decline when the supply of cations is depleted after the metallic film at the contact interface is completely etched away. Figure 8.10 shows a comparison of imprinting

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.4 PROCESS CAPABILITIES

Run1 Run2 Run3 Run4 Run5 Run6 Run7 Run8 Run9

(a)

1.2 1.4 0.8 0.6 0.4

(b) 0.12 0.10

Current (A)

Current density (ma cm−1)

1.4

207

0.08 0.06 0.04 0.02

0.2 0.00 0.0 0

50

100

150

380 40X

0

100

Time (s)

200

300

400

500

Time (s)

Figure 8.9. Current density and current profiles for repeated imprinting operations with the same stamp. (a) For Ag with Ag2 S, where after around two or three initial imprinting repetitions, the curve settles to a steady-state profile. (Reproduced with permission from [21]. Copyright 2007 American Chemical Society.) (b) For Cu2 S, where the current profile for etching 150-nm Cu films. (Reproduced with permission from [22]. Copyright 2007 American Vacuum Society.)

200 µm

5 µm

5 µm (a)

(b)

Figure 8.10. Imprints produced with multiple operations of the same stamp. (a) SEMs for imprinting Ag with an Ag2 S stamp show virtually no degradation of the imprint after nine repetitions. The two silver imprints correspond to the 1st and 9th curve of Figure 8.9a. (Reproduced with permission from [21]. Copyright 2007 American Chemical Society.) (b) SEMs of the 1st and 20th Cu imprint produced by the same Cu2 S stamp. Some of the smaller negative features vanish. The stamp was also damaged during the operation. (Reproduced with permission from [22]. Copyright 2007 American Vacuum Society.)

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

208

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

results as the imprinting stamp is used multiple times. With imprinting of silver, we have successfully used the same stamp tens of times, without seeing much degradation in stamp performance. Similarly, results were obtained with copper. However, because of working in a laboratory (not a clean room) environment with improvized equipment that does not allow us to fully control the process parameters such as imprinting force and to properly align the stamp with respect to the substrate, we see that some features on the stamp degrade and dust particles become embedded in the stamp. We expect that this will improve as we develop better imprinting equipment and stamp materials.

8.5 EXAMPLES OF ELECTROCHEMICALLY IMPRINTED NANOSTRUCTURES USING THE S4 PROCESS We have successfully used the S4 process to make a number of structures. The barcodes of Figure 8.6 as well as the seals shown in Figure 8.11 have features in the sub-100-nm range and can be used for tagging and protection against counterfeiting. Figure 8.12 shows different metallic devices including electrodes for chemical sensors, such as meandering nanowires and interdigitated electrodes. Figure 8.13 shows experimental antennae construction for THz-frequency ranges. Figure 8.14 shows a silicon nitride substrate with a repetitive pattern of 70-nm silver triangles and the scheme for using it to create a surface-enhance raman spectroscopy (SERS) or localized surface plasmon resonance (LSPR) substrate for chemical sensing. Electrochemical imprinting of nanoscale patterns using S4 is relatively new technology, first reported in February of 2007 [21]. In spite of limited investment in the process technology, for example, stamp materials, or electrochemical imprinting equipment, resolutions in the range of 50–100 nm have been routinely achieved, suggesting good potential for industrial application. Our continuing work includes the exploration of new stamp materials that simultaneously give us favorable electrochemical and mechanical properties to facilitate efficient, large-area imprinting capabilities. Further, this exploration will allow us to enlarge the set of materials the S4 process is capable of addressing.

(a)

(b)

(c)

Figure 8.11. Examples of S4 stampings. (a)–(c ) A high resolution silver stamping with feature definitions better than 50 nm can be used for nanoscale tagging of devices.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

8.5 EXAMPLES OF ELECTROCHEMICALLY IMPRINTED NANOSTRUCTURES

(a)

209

(b)

(c)

Figure 8.12. Different devices. (a) A 40-nm wire-based FET (field-effect transistor). (b) A meandering copper nanowire 70 nm wide and 1.35 mm long in a 20-micrometer square. (c ) A set of interdigitated electrodes, 200 nm apart. The isolation between the two electrodes is evident by charging up of only the floating electrode in the SEM.

Fundamental to the process are the ionic transport phenomena at the contact interface and in the bulk of the stamp. The exploration of anodic dissolution at solid–solid interfaces has received relatively little attention (in comparison to fluid–solid) interfaces. This is an area we are currently exploring to gain a better understanding of the process at the interface influence, the process dynamics, and the stamp surface (and hence, overall stamp life). Understanding the bulk transport of ions within the stamp plays an important part in understanding the process and use of the stamp. We have been developing models for ion transport based on the Nernst–Planck equation to predict (cat)ion concentration in the stamp at the end of each imprinting cycle. This can then be used to gauge the polarization of the stamp material and develop an idea of how often and for how long the stamp needs to be depolarized by using a blocking electrode as the anode. Research is also continuing into experimental studies of the role of imprinting pressure, voltage, and temperature on the speed of imprinting process, surface quality and fidelity of the imprinted structures, and life of the stamp. In addition, we are developing an instrumented, precision imprinting press for conducting such process characterization experiments.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

210

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

(a)

(b)

Figure 8.13. (a) A Pythagoras tree antenna with 70-nm features. (b) A 2D Menger sponge antenna for frequencies up to 40 THz.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

REFERENCES

211

LSPR

Extinction

1.5

1.0

0.5

0.0 0

40

0

45

50

0

0 0 0 0 0 0 55 60 75 70 75 80 Lamda (nm)

Figure 8.14. Large areas of repetitive patterns made by the S4 process can be used for SERS or LSPR-based sensing of chemicals. The middle shows silver triangles of silver patterned on glass. The image to the right shows dark-field images of patterned areas illuminated with white light. The figure to the right schematically shows how such substrates excite surface plasmons that extinguish certain frequencies from the transmitted spectrum.

ACKNOWLEDGMENTS This research was supported by NSF through the Center for Chemical-ElectricalMechanical Manufacturing Systems (Nano-CEMMS) under Grant DMI-0312862, the Office of Naval Research under grant N00173-07-G013, and the University of Illinois through the Grainger Foundation grant. We are grateful that part of this work was carried out in the Center for Microanalysis of Materials, University of Illinois, which is partially supported by the US Department of Energy under grant DEFG02ER45439.

REFERENCES 1. Madou, M. (2002) Fundamentals of Microfabrication, 2nd edn., CRC Press, New York. 2. Rickerby, J. and Steinke, J. (2002) Current trends in patterning with copper. Chem. Rev. 102, 1525–1549. 3. Felmet, K., Loo, Y., and Sun, Y. (2004) Patterning conductive copper by nanotransfer printing. Appl. Phys. Lett. 85, 3316–3318. 4. Ruska, W. S. (1987) Microelectronic Processing, McGraw-Hill, New York. 5. Ebbesen, T. W., Lezec, H. J., Ghaemi, H. F., Thio, T., and Wolff, P. A. (1998) Extraordinary optical transmission through sub-wavelength hole arrays. Nature 391, 667–669.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

212

14:2

Printer Name: Yet to Come

PATTERNING WITH ELECTROLYTE

6. Maiera, S. A and Atwater, H. A. (2005) Plasmonics: localization and guiding of electromagnetic energy in metal/dielectric structures. J. Appl. Phys. 98, 011101. 7. Andricacos, P. C., Uzoh, C., Dukovic, J. O., Horkans, J., and Deligianni, H. (1998) Damascene copper electroplating for chip interconnections. IBM J. Res. Dev. 42, 567–574. 8. Schmid, G. M., Stewart, M. D., Wetzel, J., Palmieri, F., Hao, J., Nishimura, Y., Jen, K., Kim, E. K., et al. (2006) Implementation of an imprint damascene process for interconnect fabrication. J. Vac. Sci. Technol. B 24, 1283. 9. Tamirisa, P. A., Levitin, G., Kulkarni, N. S., and Hess, D. W. (2007) Plasma etching of copper films at low temperature. Microelectron. Eng. 84, 105–108. 10. Steinh¨ogl, W., Schindler, G., Steinlesberger, G., Traving, M., and Engelhardt, M. (2005) Comprehensive study of the resistivity of copper wires with lateral dimensions of 100 nm and smaller. J. Appl. Phys. 97, 023706. 11. Steinbruchel, C. (1995) Patterning of copper for multilevel metallization: reactive ion etching and chemical-mechanical polishing. Appl. Surf. Sci. 91, 139–146. 12. Schwartz, G. C. and Schaible, P. M. (1983) Reactive ion etching of copper films. J. Electrochem. Soc. 130, 1777. 13. Ohno, K., Sato, M., and Arita, Y. (1996) Reactive ion etching of copper films in a SiCl4 , N2 , Cl2 , and NH3 mixture. J. Electrochem. Soc. 143, 4089. 14. Howard, B. J. and Steinbruchel, C. (1991) Reactive ion etching of copper in SCl4 -based plasmas. Appl. Phys. Lett. 59, 19. 15. Howard, B. J. and Steinbruchel, C. (1994) Reactive ion etching of copper with BCl3 and SiCl4 : plasma diagnostics and patterning. J. Vac. Sci. Technol. A 12, 1259–1264. 16. Lee, S. and Kuo, Y. (2001) Chlorine plasma/copper reaction in a new copper dry etching process. J. Electrochem. Soc. 148, G524–G529. 17. Lim, J., Kim, N., and Chang, E. (2004) Electrochemical patterning of copper using microcontact printing. J. Electrochem. Soc. 151, C455–C458. 18. Hulteen, J. C. and Van Duyne, R. P. (1995) Nanosphere lithography: a materials general fabrication process for periodic particle array surfaces. J. Vac. Sci. Technol. A 13, 1553. 19. Trimmer, A. L., Hudson, J. L., Kock, M., and Schuster, R. (2003) Single-step electrochemical machining of complex nanostructures with ultrashort voltage pulses. Appl. Phys. Lett. 82, 3327–3329. 20. Bhattacharyya, B., Doloi, B., and Sridhar, P. S. (2001) Electrochemical micro-machining: new possibilities for micro-manufacturing. J. Mater. Process. Technol. 113, 301–305. 21. Hsu, K. H., Schultz, P. L., Ferreira, P. M., and Fang, N. X. (2007) Electrochemical nanoimprinting with solid-state superionic stamps. Nano Lett. 7 (2), 446–451. 22. Schultz, P. L., Hsu, K. H., Fang, N. X., and Ferreira, P. M. (2007) Solid-state electrochemical nanoimprinting of copper. J. Vac. Sci. Technol. B 25 (6), 2419–2424. 23. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1996) Nanoimprint lithography. J. Vac. Sci. Technol. B 14, 4129–4133. 24. Chou, S. Y., Krauss, P. R., Zhang, W., Guo, L., and Zhuang, L. (1997) Sub-10 nm imprint lithography and applications. J. Vac. Sci. Technol. B 15, 2897–2904. 25. Terabe, K., Nakayama, T., Hasegawa, T., and Aono, M. (2002) Ionic/electronic mixed conductor tip of a scanning tunneling microscope as a metal atom source for nanostructuring. Appl. Phys. Lett. 80, 4009–4011.

P1: OTA/XYZ P2: ABC c08 JWBK252/Lee September 18, 2008

14:2

Printer Name: Yet to Come

REFERENCES

213

26. Terabe, K., Nakayama, T., Hasegawa, T., and Aono, M. (2002) Formation and disappearance of a nanoscale silver cluster realized by solid electrochemical reaction. J. Appl. Phys. 91, 10110–10114. 27. Lee, M., O’Hayre, R., Prinze, F. B., and Gur, T. M. (2004) Electrochemical nanopatterning of Ag on solid-state ionic conductor RbAg4 I5 using atomic force microscopy. Appl. Phys. Lett. 85, 3552–3554. 28. Agrawal, R. and Gupta, R. (1999) Superionic solids: composite electrolyte phase—an overview. J. Mater. Sci. 34, 1131–1162. 29. Miyatani, S. (1956) Point contact of Pt and γ -Cu2S. J. Phys. Soc. Japan 11, 1059–1063. 30. Allen, L. and Buhks, E. (1984) Copper electromigration in polycrystalline copper sulfide. J. Appl. Phys. 56, 327–335. 31. Hirahara, E. (1951) The physical properties of cuprous sulfides-semiconductors. J. Phys. Soc. Japan 6, 422–427. 32. Hebb, M. H. (1952) Electrical conductivity of silver sulfide. J. Chem. Phys. 20 (1), 185–190. 33. Wagner, C. (1953) Investigations on silver sulfide. J. Chem. Phys. 21, 1819–1827. 34. Chavez, K. and Hess, D. (2001) A novel method of etching copper oxide using acetic acid. J. Electrochem. Soc. 148, 640–643. 35. Pauporte, T. and Vedel, J. (1998) Electrical properties of a non-stoichiometric copper sulfide/solid electrolyte interface. Solid State Ion. 109, 125–134.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9 PATTERNING WITH GELS: LATTICE-GAS MODELS Paul J. Wesson and Bartosz A. Grzybowski

9.1 INTRODUCTION Despite enormous progress in micro- and nanofabrication technology [1–3], some types of small-scale structures remain challenging targets for conventional techniques. For instance, fabrication of multilevel/curvilinear architectures (e.g., threedimensional (3D) microfluidic systems [4] and microlenses [5]) typically involves multiple rounds of tedious photolithography/development/registration cycles [6, 7] or the use of slow serial techniques [8–14] with precise control of etching/milling depth [9, 13]. Similarly, structures with characteristic feature dimensions in tens of nanometers (e.g., for ultrahigh density electronic circuits [15], plasmonic devices [16], and nanoelectromechanical system sensors [17]) are below the limit of conventional photolithography and require the use of specialized near-field or phase-shift optics [18] or serial writing [19]. Recognizing these difficulties, scientists and engineers seek more focused ion beams, smaller microchisels, sharper AFM (atomic force microscopy) tips, and photolithographic masks with ever decreasing feature sizes. The logic behind this miniaturization effort is that smaller “tools” should translate into more precise fabrication protocols and should ultimately yield smaller and better-defined end structures. While this reasoning is certainly sound, a provocative scientist might ask whether it is the only paradigm available to us. After all, Nature—the ultimate builder—fabricates many of its intricate creations without any “imprinting” templates or writing tools; instead, skin patterns in fish [20], zebras, and tigers [21], compositional zones in seashells [22] and agates [23], or the fantastic

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


215

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

216

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

(a)

(c)

(b)

(d)

Figure 9.1. Examples of structures formed by reaction–diffusion. (a) Skin pattern on a zebra. (Courtesy of Malene Thyssen.) (b) Compositional banding in agates. (Courtesy of Lawrence Conklin.) (c ) Seashell Conus Marmoreus Linnaeus. (Courtesy of Bill Frank.) (d ) A radiolarian. (Courtesy of Dr. Peter Baumgartner.)

3D shells of radiolarians [24] and diatoms [25] (Figure 9.1) emerge spontaneously from chemical self-organization processes that an organism as a whole carries out. On the most abstract (and simplistic) level, such processes can be viewed as a chemical “programs” comprising the details of the chemical reactions involved (e.g., rate constants of mineralization of diatoms’ siliceous skeletons), information about the migration/diffusion of different substrates (e.g., diffusion constants and concentrations of silica to be mineralized), and the initial/boundary conditions (e.g., positions of silica deposition vesicles). The execution of these reaction–diffusion (RD) “instructions” is synonymous with the task of fabrication. Over the last several years, our group has applied the bio-inspired idea of “chemical programming” to small-scale fabrication [26–45]. The main objective of this effort has been to develop artificial chemical systems that would build desired micro- and nanostructures via self-organization mediated by RD. This effort has met with considerable success and we have been able to demonstrate RD systems (Figure 9.2), which spontaneously fabricate micro-optical components [34, 39],

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.1 INTRODUCTION

(a)

(b)

(c)

(d)

(e)

(f )

217

Figure 9.2. (a) Reaction–diffusion (RD) transforms a micropattern of circles into a pattern of submicron lines; interestingly, in doing so, it solves a mathematical problem of surface tiling [36]. (b) RD deposits three metal salts onto different locations of a substrate starting from the same initial geometry. Multicomponent patterns such as this one serve as wavelength-specific diffraction gratings [37]. (c ) RD fabrication builds an array of microlenses [34, 39] and (d ) multiple periodic precipitation microstructures [30, 38]. (e) RD is used to microetch Escher’s lizards into a crystal of K3 Fe(CN)6 [33]. (f ) RD detects a helix-to-coil phase transition in a gel film and reports it by switching between two different types of color patterns [40]. Scale bars are 50 µm in (a) and (c ) and 250 µm in other pictures.

multilevel microfluidic systems [35], diffraction gratings [36–38], substrates for controlled cell adhesion [29–31], chemical amplifiers [40–42], and unusual nanostructures [30, 32, 37]. Since the experimental aspects of these systems have been discussed extensively in two recent reviews [26, 27], this chapter will review them only relatively briefly and will then discuss some more fundamental issues underlying RD fabrication—in particular, implementation and applicability of theoretical models that can reproduce the emerging micro and nanostructures and assist in the rational design of RD fabrication systems. For conventional patterning/microfabrication techniques, modeling is often of only ornamental value, as these methods can be successfully practiced without any theoretical background. For RD fabrication (RDF), however, the participating chemicals evolve into final structures in nontrivial and sometimes counterintuitive ways, and simple “what you pattern is what you get” heuristics do not apply. Although this evolution can be described exactly by sets of coupled partial differential equations (PDEs), analytical solutions of these equations are often impossible and even numerical solvers might not offer adequate precision and/or solution stability to capture scale-sensitive aspects of the underlying physical/chemical processes. What is needed for practical applications of RDF are theoretical models that would be applicable to a wide variety of systems, easy to setup by experimentalists, and rapid to execute in order to guide ongoing fabrication tasks. Lattice-gas methods [35, 39, 46, 47] with appropriate reaction terms meet all these criteria and, as we

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

218

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

illustrate in this chapter, are well suited to describe complex RDF modalities at various length scales.

9.2 THE RDF METHOD Fabrication by RD relies on the migration of reacting chemicals inside of the material to be structured. Common hydrogels such as gelatin or agarose allow for such mobility, are inexpensive and easy to process. In a typical procedure [33–39], a gel containing one of the reagents (say, inorganic salt A) is cast as a thin film and dried. The RD process involving a complimentary chemical (say, B) is then initiated by wet stamping (WETS) [33–39] from a micropatterned hydrogel stamp soaked in the solution of B. When this stamp is placed onto the dry gel, water, W, starts flowing from its microfeatures into the substrate-–this flow is driven by the difference in the osmotic pressures between the two gel media and its rate decreases exponentially with time [35, 39], dW (t)/dt = W0 exp(− λt), as pressures gradually equalize. In parallel, B diffuses into the wetted thin-film substrate (also with exponentially decreasing rate, dB(t)/dt = B0 exp(− λt)), migrates within this film, and reacts with its “partner” A contained therein. The structures that ultimately emerge depend on the particular features of the A + B reaction, on the diffusivities of these chemicals, and on the boundary/initial conditions delineated by the wet-stamped micropattern.

9.3 MICROLENSES: FABRICATION WETS methodology is compatible with a range of organic and inorganic chemistries. In particular, the ionic reactions are well suited for fabrication tasks, since they can yield precipitates that deform/buckle the gel substrate and can therefore translate the extent of reaction at a given location into surface micro- or nanotopology. As an example [35, 39], consider a reaction of silver nitrate delivered from the stamp (B = AgNO3 ) with potassium hexacyanoferrate (A = [Fe(CN)6 ]4− ) uniformly dispersed in a thin layer of dry gelatin. To make microlenses, RD is initiated from a stamp patterned with an array of depressions in bas relief (Figure 9.3a). The precipitation reaction between silver cations (diffusing inward from the contours of these depressions into gelatin) and [Fe(CN)6 ]4− anions contained therein, 4Ag+ + [Fe(CN)6 ]4− → Ag4 [Fe(CN)6 ](↓), results in a pronounced expansion of the gel. The degree of this expansion is (i) proportional to the amount of precipitate formed at a given location and (ii) monotonically decreasing with the distance from the stamped features. The latter is due to the fact that as the Ag+ /precipitation front propagates inward from the feature’s edge, the unreacted [Fe(CN)6 ]4− experiences a sharp concentration gradient at this front and diffuses in its direction (i.e., outward)—as a result, the extent of reaction and the degree of swelling decrease with the distance from the edges of the stamped features, and give rise to curvilinear surface profiles. Depending on the applied pattern, these profiles can be sections of a sphere (when RD was initiated from circles) or pyramidal (when initiated from

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.3 MICROLENSES: FABRICATION

219

(a) Agarose/AgNO3

D

∼50 µm Gelatin/K2CrO7

z

Ld

r D'

r = D/2

(b)

Figure 9.3. (a) Fabrication of microlenses by wet stamping and RD. The top picture illustrates the experimental arrangement. Middle picture illustrates diffusion of Ag+ cations from the stamp into gelatin (black arrows) and the diffusion of [Fe(CN)6 ]4− ions toward the incoming reaction front (gray arrows). As a result of this process, the gel swells to give an array of hemispherical microlenses, whose characteristic dimensions (D’ and Ld ) depend on the dimensions of the features in the stamp (D ) and on the concentration of the chemicals used. For further details, see [39]. (b) Poly(dimethylsiloxane) replicas of arrays of microlenses with circular, triangular, and square bases. The lower-right picture is an SEM (scanning electron microscopy) image of an array of triangular lenses. Scale bars = 150 µm. (Reproduced from [26], with permission from Elsevier).

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

220

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

polygonal contours, Figure 9.3b). In both cases, the heights and curvatures of these microstructures can be controlled by the concentrations of the chemicals used and/or by the dimensions of the patterned features. Importantly, the depressions can easily be replicated into optically transparent polymers to give large (up to 3 cm × 3 cm) arrays of regular microlenses of excellent focusing properties.

9.4 MICROLENSES: MODELING ASPECTS Although the example of microlens fabrication might appear quite simple, several of its aspects would be hard to capture if we were to model the system by the pertinent RD PDEs.

9.4.1 Modeling Using PDEs (i) First, the ionic reaction A + 4B → C (↓) occurs very rapidly and only above a certain solubility product of the salts K sp = [A][B]4 . Consequently, to account for its threshold nature, one has to separate the time scales for reaction and diffusion and use the so-called Heaviside step function (θ (x) = 0 for x < 0 and θ (x) = 1 for x ≥ 0) in the reaction terms to get the set of PDEs of the following nondimensionalized form (for details, see [36]), ∂A 4 = D −1 A ∇ D A ∇ A − δ A θ (AB − K sp ) ∂t ∂B 4 = D −1 A ∇ D B ∇ B − δ B θ (AB − K sp ) ∂t ∂C = δC θ (AB 4 − K sp ) ∂t where K sp = (A − δ A )(B − 4δ A )4 , δ B = δ A /4, and δC = δ A . These equations are stiff and accurate integration schemes (e.g., Crank–Nicholson [48]) need to be used. (ii) Second, the boundary conditions are nontrivial given the fact that the rates of delivery of B and water change with time and are discontinuous. While the time dependencies of the concentrations and nonlinearities in concentration profiles can be separately handled by a fast Fourier transform method, accounting for both these effects simultaneously is complicated. (iii) Third, the degree of precipitation is coupled to the deformation/swelling of the gel. This swelling not only changes the diffusion coefficients (this is relatively easy to account for [36]) but also introduces an additional energetic constraint on the system’s evolution. Indeed, swelling increases the elastic potential energy ES of the gel and counters the energetic gain EW of gel wetting. As a result, the RD process occurs only as long as the overall energy is favorable—that is, if ES + EW < 0.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.4 MICROLENSES: MODELING ASPECTS

(a)

(b)

221

(c)

Figure 9.4. A lattice-gas model of the A + B → C reaction used to fabricate microlenses and microfluidic devices. The A particles (dark gray) are initially randomly and uniformly distributed in the gel, while B particles (light gray) diffuse across a constant-concentration boundary (lightgray edge), creating C particles (large gray circles) as reaction products. The three remaining boundaries are reflecting (solid black lines). During a diffusion step, (a), particles randomly explore the lattice by moving along available paths. The individual particles may (i) remain in place if the path crosses a reflecting boundary or (ii) move to an adjacent site. The subsequent reaction step, (b), identifies sites with sufficient particles to react, then, at each reacting site, removes stoichiometric amounts of A and B , and inserts a corresponding number of C. After the insertion of C , the next diffusion step, (c ), occurs. These diffusion and reaction steps are repeated until the system reaches a specified condition (e.g., no A remains in the system).

9.4.2 Modeling Using Lattice-Gas Method Lattice-gas method provides a simple and surprisingly accurate alternative to the full-fledged PDE approach. In a lattice-gas model, the space is discretized into a grid and the molecules/ions placed onto it are subjected to several basic rules describing reaction and diffusion events (Figure 9.4). In the case of our microlenses, each node on the lattice has the same initial concentration of hexacyanoferrate ions (A) while silver cations (B) and water (W) are delivered to nodes beneath the features of the stamp. The rates of their delivery decay exponentially with time, which here is equiv˙ = W0 exp(− λ n), alent to the simulation step number, n: B˙ = B0 exp(− λ n) and W where λ is a constant. In each simulation step: (i) B and W are added to the gel; (ii) A and B are allowed to perform m diffusion steps on the square lattice (relative diffusion coefficients DA /DB = 0.4 are taken from experiment [35] with equal probability along all available directions (e.g., 1/4 ; if not next to a boundary). Parameter m relates speeds of diffusion and delivery; (iii) In cells where A and B are present in sufficient quantities, they react according to A + 4B → A4 B (↓). The energy of the system is then calculated as a sum of swelling and wetting energies. Following previous works, the swelling energy is approximated as proportional to the
sum of squares of the vertical deformations of the gel over all cells [49, 50], E S ∼ i j (h i j )2 . Because, as verified experimentally [35, 39], these deformations

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

222

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS


are proportional to the amount of precipitate, we can write E S = e i j (Ci j )2 , where e is the (positive) energetic cost of swelling per unit concentration of precipitate C and the summation is over all nodes (i, j). On the other hand, wetting energy is favorable—that is, it lowers the energy of the system-–and is linearly proportional to the negative of the amount of water transferred into the gel, E W = −W0 (1 − exp(−λ n))/λ. We note that because of this linear proportionality, water molecules do not have to be explicitly accounted for in the simulations. With these assumptions and with the values of initial concentrations proportional to the experimental ones (cf. caption to Figure 9.5 and [40]), the basic moves (i)–(iii) are repeated until either all A molecules run out or the overall energy of the system becomes non-negative, ES + EW ≥ 0. As illustrated in Figure 9.5a, the simulations agree with experimental profiles of the microlenses fabricated using features of different dimensions/shapes and different concentrations of participating chemicals. The same set of parameters (e, λ, W 0 ) is used to reproduce all these structures and the execution of the program (available at http://dysa.northwestern.edu/Research/Progreactions.dwt) takes only tens of seconds on a standard desktop PC. This speed combined with the ease of coding initial conditions (a bitmap picture is sufficient as input) makes this program helpful in the design process not only of lenses but also more complex architectures for uses in microfluidics. This last capability is illustrated in Figure 9.6 where the same Ag+ /[Fe(CN)6 ]4− reaction is propagated from spatially extended contours (light gray) to fabricate long microfluidic channels with smaller buckles inside these channels serving as passive fluid-mixing elements. As expected, the locations of these ridges in actual devices fabricated by RD agree with the predictions of lattice-gas simulations.

9.5 RDF AT THE NANOSCALE In the case of microscopic RD structures, the lattice-gas models offer a simple and rapid alternative to solving RD PDEs. At the nanoscale, when the numbers of the participating molecules might be very low, PDEs become meaningless and the latticegas approach is not only simpler but also physically more realistic.

9.5.1 Nanoscopic Features from Counter-Propagating RD Fronts Let us first consider a striking example of a chemical system (Figure 9.7) that fabricates nanoscopic features by RD initiated from an array of wet-stamped parallel lines [36]. The lines deliver a solution of FeCl3 to a thin (∼100 nm) layer of dry gelatin uniformly doped with potassium hexacyanoferrate, K4 Fe(CN)6 . As the Fe3+ cations (henceforth, B) constantly delivered from the stamp migrate into gelatin, they precipitate all [Fe(CN)6 ]4− (henceforth, A) they encounter and give a deeply colored Fe4 [Fe(CN)6 ]3 precipitate known as Prussian blue (here, denoted as C∗ ; Figure 9.7b).

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.5 RDF AT THE NANOSCALE

223

(a) Experimental

Scan depth (µm)

0 −25 −50 0 −25 −50 0 −25 −50

Theoretical

10% AgNO3 0

50 100 150 200 250 300

15% AgNO3 0

50 100 150 200 250 300

20% AgNO3 0

50 100 150 200 250 300

0 −25 −50 0 −25 −50 0 −25 −50

10% AgNO3 0

50

100 150 200 250 300

0

50

100 150 200 250 300

0

50

100 150 200 250 300

15% AgNO3

20% AgNO3

Scan length (µm) (b) (i)

(ii)

(iii)

(iv)

Figure 9.5. (a) Experimental (left) and modeled (right) surface profiles for 75-µm (solid line) and 150-µm (dashed line) circles lenses fabricated with different concentrations of silver nitrate delivered to 1% w/w K4 Fe(CN)6 /gelatin. (b) Experimental and modeled microlenses with noncircular base shapes: (i) triangular, (ii) square, (iii) hexagonal, (iv) pentagonal star. The inset to (i) shows long-range order in the stamped array of triangles. Insets to (ii) and (iii) are the images taken at the microlenses’ focal planes. Scale bars in the main pictures correspond to 150 µm, in the inset to (i) to 1 µm, and in the other insets to 2 µm. For more information, see [39]. (Reproduced with permission from [39]. American Institute of Physics).

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

224

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

10 0

0

−50 15

200

0

0 −15 200 Figure 9.6. Reaction–diffusion fabricates a parallel buckle (top row) and a “caterpillar” (bottom row) passive microfluidic mixers. The left column shows results of simulations. Gray delineates stamp’s geometry; dark gray corresponds to precipitate; light-gray lines give the directions of profilometric scans shown next to the pictures. All dimensions are in micrometers. The right column gives optical micrographs of channel geometries replicated into poly(dimethylsiloxane). Scale bars are 500 µm in large-magnification images and 1 mm in the insets. For details, see [35].

The unreacted A anions experience a sharp concentration gradient at the reaction front and diffuse in its direction. In case of two reaction fronts counter-propagating from nearby features of the stamp, A between them diffuses outward—that is, toward the incoming fronts (black arrows in Figure 9.7b). This outflow continues until, ultimately, there is no more A left near the centerline between the features. Although the B cations continue to diffuse into this region, there is no more A left to precipitate, and the region remains a thin, uncolored line of welldefined boundaries and the thickness d is roughly two orders of magnitude smaller (down to ∼ 35 nm) than the spacing between the neighboring features in the stamp (L = 2 − 250 µm, Figure 9.8). Overall, the RD process transforms the original

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.5 RDF AT THE NANOSCALE

225

(a) Agarose/B L

Gelatin/A (b) B/C

A t

(c) d Figure 9.7. Fabrication of nanoscopic lines. (a) An agarose stamp with features spaced by L delivers B to a thin layer of gelatin containing A. (b) A and B react to form C , which then precipitates as C∗ , creating a sharp gradient in A (dashed line). (c ) The presence of C∗ slows the diffusion of A, ultimately allowing for the formation of a narrow, clear band containing no C∗ and having width d down to tens of nanometers.

micropattern in the stamp into its Voronoi tessellation of nanometer-thick lines (Figures 9.8a and b). The stopping of RD fronts so close to one another and with such optically sharp boundaries is a result of a nonlinear dependence of the diffusion coefficients of the reacting species on the spatial location and on the amount of precipitate present at this location. Specifically, the diffusion coefficient of large A anions depends strongly on the amount of the precipitate C∗ , while that of the smaller metal cations B does so to a much lesser extent. Consequently, as the reaction (color) fronts approach one another, it becomes increasingly difficult for A to diffuse outward—in other words, A is “trapped” between the fronts and cannot “escape” until they are very close to one another. 9.5.2 Failure of Continuum Description While continuous, PDE-based description of this system works reasonably well for the thicknesses of clear lines d in the hundreds of nanometers, it becomes physically questionable when d approaches the 35 nm limit we realized experimentally (Figures 9.8c and d). First and foremost, the numbers of molecules become too small to apply continuum approximation on which PDEs rely. For example, for L = 2 µm spacing of the stamped lines and the experimental concentrations that yield d = 35 nm features (1.6% w/w K4 Fe(CN)6 in a 100-nm-thick gelatin film; 2.5% w/w solution of FeCl3 in an 8% w/w agarose stamp); there is, on average, less than 2.5 reacting molecules of any type per 1 nm3 of the gel. Furthermore, to model 35-nm features,

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

226

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

(a)

(b)

(c)

(d)

Figure 9.8. Fabrication by counter-propagating RD fronts. (a) An array of circles of two different diameters (50 and 100 µm) delivers a solution of CuCl2 to a thin layer of dry gelatin doped with K4 Fe(CN)6 . Reaction between these two salts produces dark-brown precipitate in the gel with the exception of clear lines delineating a Voronoi tessellation of the stamped pattern. The lines are several micrometers thick. Scale bar = 250 µm. (b) Reaction between K4 Fe(CN)6 and FeCl3 initiated from an array of interdigitated lines yields “snaky” clear lines. Scale bar = 250 µm in the main picture and 100 µm in the inset. (Reproduced with permission from [36]. Copyright WileyVCH Verlag GmbH & Co. KGaA). (c ), (d ) Reaction between K4 Fe(CN)6 and FeCl3 initiated from an array of 2 µm parallel lines spaced by 2 µm gives clear lines of nanoscopic dimensions. (c ) An AFM image of the pattern. (d ) SEM image indicating the absence of the precipitate in the line region. Scale bar in (d ) = 150 nm.

the resolution of the grid on which the PDEs are solved should be few nanometers at most, which translates into only tens of molecules per grid point—that is, not enough to take meaningful first and second derivatives of concentrations. Finally, the continuum description does not take into account concentration fluctuations, which at the nanoscale are large enough to influence not only feature dimensions but also their spatial regularity.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.5 RDF AT THE NANOSCALE

227

9.5.3 Lattice-Gas Models at the Nanoscale The level of description accounting for individual molecules and for the stochastic nature of their motions (and of reactive collisions) is possible with a latticegas model. Here, the area of gel between stamped lines is represented as a twodimensional (by virtue of problem’s symmetry) square grid of X by Y nodes (typically, 1 nm × 1 nm each). To account for the initial concentration of A in the gel layer, each node is assigned probability pA of having an A anion on it, such that pA XY = [A]. The B cations are then added onto the grid from the initial locations corresponding to the stamped features, and the C adducts are created by the reaction between A and B. These adducts subsequently nucleate into the C∗ precipitate (see discussion of reaction events below). The simulation uses standard periodic-boundary conditions and alternates diffusion and reaction steps: (i) Diffusion. During a diffusion step each ion/molecule in the system is chosen to move with a certain probability p inversely proportional to its mass so that if pB is set to unity then pA = mB /mA , and pC = mB /mC . This scaling is based on the experimental observation that DA /DB = 0.3 ≈ mB /mA = 0.265 and is in line with other theories of diffusion in gels for which D ∝ (m−u ) with exponent u determined by the surrounding matrix and ranging from 0.5 to 2.1 depending on a specific system [51–53]. If a particular particle is chosen to move, it has 1/4 chance of migrating to any of the four neighboring sites. In addition, to account for the relative sizes (smaller B ≈ 0.2 nm, larger A ≈ 0.8 nm, and C ≈ 1 nm) of the species, there are two additional rules that prescribe that not more than two A molecules can fit onto one node and that C adducts (or aggregates of C, see below) cannot move onto sites occupied by either A or C. (ii) Reaction. There are two main issues involved in the description of the precipitation reaction at the nanoscale. Firstly, the use of “macroscopic” reaction stoichiometry 3A + 4B → A3 B4 (i.e., 3[Fe(CN)6 ]4− + 4Fe3+ → Fe4 [Fe(CN)6 ]3 ) cannot be simply translated into the presence of 3A and 4B molecules on one lattice node, as this event is highly improbable and does not reflect the mechanism of precipitate formation. This mechanism has been studied by others [54, 55] and is better described as a process in which A and B first react to give an adduct C, and then the adducts aggregate into crystalline particles of precipitate, C∗ . In our simulation, the A + B → C step occurs with a certain probability of a reactive collision pr when both reaction partners are present on the same node. The C adducts thus created can migrate through the lattice and aggregate with one another. As the adducts aggregate and become more massive, their diffusivity decreases inversely proportionally to mass and upon reaching a certain threshold (typically, ca. 10 molecules) is set to zero, corresponding to immobile precipitate C∗ . The sequence of the diffusion and reaction moves is repeated sequentially until all A anions are used and until all C’s are aggregated into C∗ . The simulation is then repeated multiple times (tens to hundreds) to average over the stochastic effects and the average values thus obtained are compared to the experimental results.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

228

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

Despite the simplicity of this algorithm, it reproduces the experimental observations well and gives dtheory ≈ 32.76 nm for the experimental concentrations A = 1.6% w/w in dry gelatin and B = 2.5% w/w in 8% agarose stamp. In addition, the lattice-gas model can help answer some fundamental questions related to the minimal dimensions of features that can be resolved by RDF and the effects of noise at this level (these results will be discussed in more detail in an upcoming publication).

(c)

30 10

1

20 0 0.8

1.9 2

10

Frequency

(b)

C per column

(a)

C per column

9.5.3.1 Up to What Point Is RDF Deterministic? When approaching the nanoscale, the question of resolution is no longer purely deterministic. In fact, with the same initial conditions, some lattice-gas runs can produce clean, well-defined lines, whereas others might not resolve any lines at all (Figures 9.9a and b). The question of fabricating a structure should then be considered in probabilistic terms—for example, if 995 simulation runs produce lines 30 nm wide whereas 5 runs resolve no runs, we can say that the lines appear with f = 99.5% frequency. Interestingly, when varying various system parameters (e.g., pA , B), the values of f usually change in an almost step-wise fashion-–for example, as pA decreases from 0.23 to 0.05, f drops from 99.2% to 11%, and decreasing the gelatin thickness from 28 to 12 nm causes f to drop from 99.6% to 2.4%. These “thresholds” can be used to estimate the smallest feature sizes that can be fabricated by RDF with a given level of “fidelity” (expressed by f , see Figure 9.9c) We also note that the stochastic nature of the nanoscale RDF is reflected in the “wiggliness” of the emerging lines (see Figure 9.8d), which is not due to imperfect stamping but reflects concentration fluctuations at various locations along the propagating RD front.

0 30 20

0.6 0.4 0.2

10 0

0 0

1

2 Distance (µm)

3

4

0

25

50 d (nm)

75

100

Figure 9.9. Concentration profiles of the colored precipitate C ∗ for two typical realizations of the LG simulation. (a) In the majority of cases, the counter-propagating fronts give sharp lines devoid of colored precipitate. (b) In some cases, however, the same initial conditions do not resolve clear lines at all. (c ) The frequency of observing a clear line of thickness d . Dashed lines indicate the cutoff frequency (97%) and the corresponding linewidth, ∼20 nm. Individual points were calculated using the LG model by varying pA while holding all other parameters constant.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

9.6 SUMMARY AND OUTLOOK

229

9.5.3.2 Scaling Relationships. The simulations reveal that the thickness of the clear lines scales linearly with the logarithm of the pA parameter, which, in turn, is proportional (cf. Section 9.5.3 (i) above) to the diffusivity of [Fe(CN)6 ]4– anions, d α ln(pA ) α ln(DA ). While this scaling can also be obtained from a PDE solution to a moving-front problem, the lattice-gas model can estimate (see Section 9.5.3.1) the minimal linewidth at d ≈ 19.2 nm (with f = 96.9%). Similarly, the lattice-gas model gives a power-law dependence of d on the concentration of B, d α Bα α [Fe3+ ]α , where α is a constant. The same functional form can be derived by solving appropriate PDEs, but it is insensitive to scale; in contrast, lattice-gas can determine absolute value of the smallest feature, d ≈19.6 nm (with f = 95.1%). 9.5.3.3 Does Thickness Matter? Finally, the lattice-gas model shows that linewidth decreases with increasing gelatin thickness, H, as d α c1 − c2 ln(H), where c1 and c2 are constants. This trend, however, is accompanied by the decrease in the frequency f with which an optically clean line can be resolved—in other words, the thicker the gel, the more pronounced are the concentration fluctuations along RD fronts, and the worse the resolution. The lattice-gas model indicates that the smallest linewidth that can be obtained with frequency above 95% is d ≈ 23 nm for H ≈ 160 nm.

9.6 SUMMARY AND OUTLOOK While theoretical models might not be considered an integral part of most patterning/fabrication techniques, they appear necessary when faced with often counterintuitive outcomes of RDF. Indeed, it would be rather hard to predict a priori that, for instance, a combination of two counter-propagating reaction fronts propelled by diffusion can yield 30 nm features. If we were to trust our intuition only, we would probably expect diffusive transport to smear things up, not make them sharp at the scale of nanometers! In cases like this one, lattice-gas models become really handy. As we tried to illustrate in this chapter, they are easy to implement and rapid to execute. The examples of microlenses and nanolines illustrate that lattice-gas methods can capture the essence of physical and chemical phenomena underlying RDF, and are applicable to different length scales and different types of fabricated structures. Once a lattice-gas code is in place, it can be used to test various chemistries and spatial configurations rapidly, and can guide the design of an RDF processes “in real time.” Of course, neither lattice-gas methods nor any other ones allow one to automatically reverse-engineer a RD process that would produce a desired micro/nanostructure. While a unique solution to this problem is most likely impossible, rapid lattice-gas methods combined with optimization schemes could provide useful approximate solutions. Development of reverse-engineering approximations for RDF remains a difficult but fascinating challenge for future research on this bioinspired philosophy of making small-scale structures.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

230

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

REFERENCES 1. Bratton, D., Yang, D., Dai, J. Y., and Ober, C. K. (2006) Recent progress in high resolution lithography. Polym. Adv. Technol. 17, 94–103. 2. Geissler, M. and Xia, Y. N. (2004) Patterning: principles and some new developments. Adv. Mater. 16, 1249–1269. 3. Xia, Y. N., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1999) Unconventional methods for fabricating and patterning nanostructures. Chem. Rev. 99, 1823–1848. 4. Abgrall, P. and Gue, A. M. (2007) Lab-on-chip technologies: making a microfluidic network and coupling it into a complete microsystem—a review. J. Micromech. Microeng. 17, R15–R49. 5. Ottevaere, H., Cox, R., Herzig, H. P., Miyashita, T., Naessens, K., Taghizadeh, M., Volkel, R., Woo, H. J., and Thienpont, H. (2006) Comparing glass and plastic refractive microlenses fabricated with different technologies. J. Opt. A 8, S407–S429. 6. Boche, B. (1994) A novel technique for multiple-level dry-etching (mulde) featuring selfaligned etch masks. Microelectron. Eng. 26, 63–69. 7. Walsby, E. D., Wang, S., Xu, J., Yuan, T., Blaikie, R., Durbin, S. M., Zhang, X. C., and Cumming, D. R. S. (2002) Multilevel silicon diffractive optics for terahertz waves. J. Vac. Sci. Technol. B 20, 2780–2783. 8. Belloy, E., Sayah, A., and Gijs, M. A. M. (2002) Micromachining of glass inertial sensors. J. Microelectromech. Syst. 11, 85–90. 9. Fu, Y. Q., Kok, N., Bryan, A., Hung, N. P., and Shing, O. N. (2000) Experimental study of three-dimensional microfabrication by focused ion beam technology. Rev. Sci. Instrum. 71, 1006–1008. 10. Madden, J. D. and Hunter, I. W. (1996) Three-dimensional microfabrication by localized electrochemical deposition. J. Microelectromech. Syst. 5, 24–32. 11. Rossnagel, S. M., Cuomo, J. J., and Westwood, W. D. (1990) Handbook of Plasma Processing Technology: Fundamentals, Etching, Deposition, and Surface Interactions, Noyes Publications, Park Ridge, NJ. 12. Schaffer, C. B., Brodeur, A., Garcia, J. F., and Mazur, E. (2001) Micromachining bulk glass by use of femtosecond laser pulses with nanojoule energy. Opt. Lett. 26, 93–95. 13. Schuster, R., Kirchner, V., Allongue, P., and Ertl, G. (2000) Electrochemical micromachining. Science 289, 98–101. 14. Watt, F., van Kan, J. A., and Osipowicz, T. (2000) Three dimensional microfabrication using maskless irradiation with MeV ion beams: proton-beam micromachining. MRS Bull. 25, 33–38. 15. Beckman, R., Johnston-Halperin, E., Luo, Y., Green, J. E., and Heath, J. R. (2005) Bridging dimensions: demultiplexing ultrahigh-density nanowire circuits. Science 310, 465–468. 16. Zia, R., Schuller, J. A., Chandran, A., and Brongersma, M. L. (2006) Plasmonics: the next chip-scale technology. Mater. Today 9, 20–27. 17. Ekinci, K. L. and Roukes, M. L. (2005) Nanoelectromechanical systems. Rev. Sci. Instrum. 76, 1–12. 18. Rogers, J. A., Paul, K. E., Jackman, R. J., and Whitesides, G. M. (1997) Using an elastomeric phase mask for sub-100 nm photolithography in the optical near field. Appl. Phys. Lett. 70, 2658–2660.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

14:45

Printer Name: Yet to Come

REFERENCES

231

19. Cumpston, B. H., Ananthavel, S. P., Barlow, S., Dyer, D. L., Ehrlich, J. E., Erskine, L. L., Heikal, A. A., Kuebler, S. M., et al. (1999) Two-photon polymerization initiators for three-dimensional optical data storage and microfabrication. Nature 398, 51–54. 20. Kondo, S. and Asai, R. (1995) A reaction-diffusion wave on the skin of the marine angelfish pomacanthus. Nature 376, 765–768. 21. Kondo, S. (2002) The reaction-diffusion system: a mechanism for autonomous pattern formation in the animal skin. Genes to Cells 7, 535–541. 22. Fowler, D. R., Meinhardt, H., and Prusinkiewicz, P. (1992) Modeling seashells. Comput. Graph. 26, 379–387. 23. Heaney, P. J. and Davis, A. M. (1995) Observation and origin of self-organized textures in agates. Science 269, 1562–1565. 24. Anderson, O. R. (1983) Radiolaria, Springer, New York. 25. Mann, S. and Ozin, G. A. (1996) Synthesis of inorganic materials with complex form. Nature 382, 313–318. 26. Grzybowski, B. A. and Campbell, C. J. (2007) Fabrication using ‘programmed’ reactions. Mater. Today 10, 38–46. 27. Grzybowski, B. A., Bishop, K. J. M., Campbell, C. J., Fialkowski, M., and Smoukov, S. K. (2005) Micro- and nanotechnology via reaction-diffusion. Soft Matter 1, 114–128. 28. Campbell, C. J., Smoukov, S. K., Bishop, K. J. M., Baker, E., and Grzybowski, B. A. (2006) Direct printing of 3D and curvilinear micrometer-sized architectures into solid substrates with sub-micrometer resolution. Adv. Mater. 18, 2004–2008. 29. Kandere-Grzybowska, K., Campbell, C. J., Mahmud, G., Komarova, Y., Soh, S., and Grzybowski, B. A. (2007) Cell motility on micropatterned treadmills and tracks. Soft Matter 3, 672–679. 30. Smoukov, S. K., Bitner, A., Campbell, C. J., Kandere-Grzybowska, K., and Grzybowski, B. A. (2005) Nano- and microscopic surface wrinkles of linearly increasing heights prepared by periodic precipitation. J. Am. Chem. Soc. 127, 17803–17807. 31. Kandere-Grzybowska, K., Campbell, C., Komarova, Y., Grzybowski, B. A., and Borisy, G. G. (2005) Molecular dynamics imaging in micropatterned living cells. Nat. Methods 2, 739–741. 32. Campbell, C. J., Smoukov, S. K., Bishop, K. J. M., and Grzybowski, B. A. (2005) Reactive surface micropatterning by wet stamping. Langmuir 21, 2637–2640. 33. Smoukov, S. K., Bishop, K. J. M., Klajn, R., Campbell, C. J., and Grzybowski, B. A. (2005) Cutting into solids with micropatterned gels. Adv. Mater. 17, 1361–1365. 34. Campbell, C. J., Baker, E., Fialkowski, M., Bitner, A., Smoukov, S. K., and Grzybowski, B. A. (2005) Self-organization of planar microlenses by periodic precipitation. J. Appl. Phys. 97, 1–3. 35. Campbell, C. J., Klajn, R., Fialkowski, M., and Grzybowski, B. A. (2005) One-step multilevel microfabrication by reaction-diffusion. Langmuir 21, 418–423. 36. Campbell, C. J., Fialkowski, M., Klajn, R., Bensemann, I. T., and Grzybowski, B. A. (2004) Color micro- and nanopatterning with counter-propagating reaction-diffusion fronts. Adv. Mater. 16, 1912–1917. 37. Klajn, R., Fialkowski, M., Bensemann, I. T., Bitner, A., Campbell, C. J., Bishop, K., Smoukov, S., and Grzybowski, B. A. (2004) Multicolour micropatterning of thin films of dry gels. Nat. Mater. 3, 729–735.

P1: OTA/XYZ P2: ABC c09 JWBK252/Lee September 22, 2008

232

14:45

Printer Name: Yet to Come

PATTERNING WITH GELS

38. Bensemann, I. T., Fialkowski, M., and Grzybowski, B. A. (2005) Wet stamping of microscale periodic precipitation patterns. J. Phys. Chem. B 109, 2774–2778. 39. Campbell, C. J., Baker, E., Fialkowski, M., and Grzybowski, B. A. (2004) Arrays of microlenses of complex shapes prepared by reaction-diffusion in thin films of ionically doped gels. Appl. Phys. Lett. 85, 1871–1873. 40. Bitner, A., Fialkowski, M., Smoukov, S. K., Campbell, C. J., and Grzybowski, B. A. (2005) Amplification of changes of a thin film’s macromolecular structure into macroscopic reaction-diffusion patterns. J. Am. Chem. Soc. 127, 6936–6937. 41. Bishop, K. J. M., Gray, T. P., Fialkowski, M., and Grzybowski, B. A. (2006) Microchameleons: nonlinear chemical microsystems for amplification and sensing. Chaos 16, 1–8. 42. Bishop, K. J. M. and Grzybowski, B. A. (2006) Localized chemical wave emission and mode switching in a patterned excitable medium. Phys. Rev. Lett. 97, 1–4. 43. Bishop, K. J. M., Fialkowski, M., and Grzybowski, B. A. (2005) Micropatterning chemical oscillations: waves, autofocusing, and symmetry breaking. J. Am. Chem. Soc. 127, 15943–15948. 44. Fialkowski, M., Bitner, A., and Grzybowski, B. A. (2005) Wave optics of liesegang rings. Phys. Rev. Lett. 94, 1–3. 45. Bitner, A., Fialkowski, M., and Grzybowski, B. A. (2004) Color micropatterning with reconfigurable stamps. J. Phys. Chem. B 108, 19904–19907. 46. Boon, J. P., Dab, D., Kapral, R., and Lawniczak, A. (1996) Lattice gas automata for reactive systems. Phys. Rep. 273, 55–147. 47. Chopard, B., Luthi, P., and Droz, M. (1994) Reaction-diffusion cellular-automata model for the formation of Liesegang patterns. Phys. Rev. Lett. 72, 1384–1387. 48. Crank, J. and Nicholson, P. (1947) A practical method for numerical evaluation of solutions of partial differential equations of the heat conduction type. Proc. Camb. Phil. Soc. 43, 50–67. 49. Ilmain, F., Tanaka, T., and Kokufuta, E. (1991) Volume transition in a gel driven by hydrogen-bonding. Nature 349, 400–401. 50. Sheppard, S. E., Houck, R. C., and Dittmar, C. (1942) The sorption of soluble dyes by gelatin. J. Phys. Chem. 46, 158–176. 51. Griffiths, M. C., Strauch, J., Monteiro, M. J., and Gilbert, R. G. (1998) Measurement of diffusion coefficients of oligomeric penetrants in rubbery polymer matrixes. Macromolecules 31, 7835–7844. 52. Herman, M. F., Panajotova, B., and Lorenz, K. T. (1996) A quantitative theory of linear chain polymer dynamics in the melt. General scaling behavior. J. Chem. Phys. 105, 1153–1161. 53. Waggoner, R. A., Blum, F. D., and Lang, J. C. (1995) Diffusion in aqueous-solutions of poly(ethylene glycol) at low concentrations. Macromolecules 28, 2658–2664. 54. Kuhnhardt, C. (1994) Nucleation and growth of Prussian blue films on glassy-carbon electrodes. J. Electroanal. Chem. 369, 71–78. 55. Jin, W. Q., Toutianoush, A., Pyrasch, M., Schnepf, J., Gottschalk, H., Rammensee, W., and Tieke, B. (2003) Self-assembled films of Prussian blue and analogues: structure and morphology, elemental composition, film growth, and nanosieving of ions. J. Phys. Chem. B 107, 12062–12070.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10 PATTERNING WITH BLOCK COPOLYMERS Jia-Yu Wang, Wei Chen, and Thomas P. Russell

10.1 INTRODUCTION With the miniaturization of functional devices, the ability to fabricate nanosized periodic structures is essential to technological advances. The broad scope of “topdown” process, including conventional 193-nm immersion lithography, extreme ultraviolet lithography, x-ray and electron-beam lithography, soft lithography (nanoimprint lithography), as well as multibeam-interference lithography and stepand-flash lithography, are being aggressively pushed to their limits to reduce feature size for subsequent pattern transfer [1–6]. It is not clear whether these “top-down” approaches will be able to keep up with Moore’s Law [7] and be commercially viable. The drive to smaller feature sizes with inherently decreased diffusion distances and increased surface/volume ratios may be approaching technological or cost limitations. In the past decade, nanopatterning based on block copolymers (“bottom-up” process) provides a means of breaking the barrier to nanosized features. Block copolymers (BCPs) are two or more chemically different polymer chains joined together at one end with a covalent bond [8–13]. Owing to the positive mixing enthalpy and low mixing entropy of the component segments, dissimilar blocks tend to phase segregate into ordered arrays of microdomains, tens of nanometers in size. The size of the microdomains is constrained by the chemical connectivity of the blocks and is limited to spatial dimensions of BCP chain. Depending on the volume fraction of one block, f , and the degree of microphase separation, χ N, where χ is the Flory–Huggins segmental interaction parameter and N is the number of segments

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


233

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

234

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a) 80

(b) 80 S L

C

60

C' χN

χN

60

S'

40 G 20

S

Dicordered

Dicordered 00

0.4

0.2

(c)

B

C

fA

0.6

G

0.8

C'

PL G

20

CPB'

CPB

40

PL L

C

0 0

1.0

L fA

0.2

G'

0.4

fA

0.6

C'

0.8

1.0

S'

Figure 10.1. Diblock copolymers are predicted to self-assemble according to a phase map by self-consistent mean field theory (a) and proven experimentally (b). A variety of constant-radius geometries are observed as a function of relative lengths of the two blocks (c ). (Reproduced with permission from [11]. Copyright 1999, American Institute of Physics.)

in BCP, morphologies including ordered arrays of spherical, cylindrical, or lamellar microdomains or bicontinuous gyroid morphologies are observed (Figure 10.1). For nanotechnological applications, the use of thin films of materials is highly preferred [7, 14–16]. In contrast to bulk, thin films involve two additional factors: surface/interface energies and the commensurability between the film thickness and the natural period of the structures in bulk [17]. In general, preferential interactions of one block with the substrate or a lower surface energy of one component causes the preferential segregation of one block to either the surface or the substrate [18– 21]. As a consequence of the connectivity of the blocks, the microdomains of BCP orient parallel to the substrate. Controlling interfacial interactions then is key in controlling the orientation of BCP microdomains. Alternatively, applying external forces to BCPs can also achieve the desired orientation. For thin films, the relationship between film thickness and period of the copolymer is important. It is, of course, most difficult to prepare thin films where the thickness is exactly integral or half-integral number of periods in thickness. When this is not the case, there is an additional contribution to the free energy associated with stretching or compressing the copolymer chains to accommodate this incommensurability. While BCPs can form highly ordered structures over a short distance, some applications demand long-range ordering at a specific orientation. For example, cylinders lying parallel to the surface and lamellae oriented normal to the film surface are of interest for generating templates for patterning of nanowires, whereas cylinders oriented normal to the surface or spheres may be of interest for patterning substrates with hexagonal arrays

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

235

of nanoscopic elements for data storage [22, 23]. Thus, key to the use of BCPs for the fabrication of nanostructured materials is to control the orientation and lateral ordering of the microdomains in thin films. In this chapter, we will begin with the strategies to control the orientation and long-range ordering of BCP microdomains, and then discuss the generation and application of nanoporous BCP thin films as templates in nanotechnology.

10.2 ORIENTATION 10.2.1 Self-Assembling Controlling the orientation of BCP microdomains is crucial for the optimal utilization of these self-assembling arrays of nanoscopic elements. Experiments and simulations indicate that BCP microdomains orient parallel to the substrate are thermodynamically favored when the substrate is preferentially wet by either block. However, an orientation of microdomains normal to the substrates may be more desirable or essential for some of the applications [17, 21, 24–29]. To achieve such an orientation, external fields, such as mechanical flow and electric fields, or simply controlling interfacial interactions are required. In this section, several approaches toward orienting BCP microdomains in thin films are outlined.

10.2.1.1 Thermal Annealing. Thermal annealing, i.e., heating BCP thin films to temperatures above the glass transition temperatures of all blocks, has been frequently used. For BCPs with spherical microdomains, a monolayer of spheres forms a hexagonal lattice adjacent to the substrate where each spherical microdomain has six nearest neighbors. In thicker films, layers of spheres packed in a body-centered cubic lattice [30] are observed with the (1 1 0) plane oriented parallel to the substrate [31]. This change in packing was quantitatively addressed by Kramer and co-workers where the transition point between the two different types of packing was found to depend upon the interactions of the blocks with the substrate, the segmental interactions between the blocks, and the surface energies of the blocks. By controlling the film thickness, which in the case of spin coating amounts to controlling the concentration of the solution being coated or varing the spinning speed, a single layer of spherical microdomains can be obtained by thermal annealing a thin BCP film [32–34]. The situation becomes more complex when the microdomains are anisotropic in shape, as is the case with cylindrical or lamellar microdomains. For a BCP with cylindrical microdomain morphology, a single layer of cylinders oriented parallel to the surface can easily be obtained by spin-coating a film with a specific thickness, followed by thermal annealing, and allowing interfacial interactions to produce an in-plane orientation of the microdomains [26, 29, 35, 36]. 10.2.1.2 Control of Interfacial Interactions. As mentioned before, the interfacial interactions dictate the wetting layer at both the polymer–substrate and

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

236

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

polymer–air interfaces and, consequently, the orientation of microdomains in the films [17, 19, 21, 24–29, 37–43]. If the interactions of the two blocks with the substrate are balanced, either equally attractive or repulsive, then for films ∼L0 in thickness, the microdomains will align normal to the surface. It is essential, though that the film thickness be ∼L0 , but not exactly L0 . If surface energies and interfacial interactions are balanced, then there is no energetic penalty for having either or both blocks at the surface or substrate interface. Energetically there is no difference. However, if the thickness is not exactly equal to and integra or half-integral number of L0 , which is hard to achieve, then to maintain the orientation parallel to the surface would require some stretching or compression of the copolymer chains. This additional contribution to the energy results in an orientation of the microdomains normal to the surface [44, 45]. In the case of polystyrene-block-poly(methylmethacrylate) (PS-bPMMA) on a silicon substrate, balanced interfacial interactions can be achieved by removing the oxide layer with buffered HF, leaving a silanized silicon surface with Si–H bonds at the surface. With such a surface, the interactions of PS and PMMA with the Si–H are equally nonfavorable, and upon thermal annealing, the cylindrical microdomains orient normal to the surface [29]. An alternative approach is to tune interfacial interactions by use of a self-assembled monolayer (SAM) [46–51]. Symmetric, neutral, and asymmetric wetting conditions for BCP can be achieved by controlling the surface energies and chemical composition of the SAM on the surface. In most cases, a mixed assembly of functional groups is used to control the interfacial interactions. The same end was achieved by Kellogg et al. [27] by placing random copolymers on the substrate where the random copolymer contained the same monomeric units as the diblock copolymer. Here, the fraction of monomers in the random copolymer can be varied to control the interactions of the substrate with both blocks. Even with high molecular weight random copolymers, though the diblock copolymer and the random copolymer can interdiffuse and produce a complex morphology, far from the desired result. To avoid this, Russell, Hawker and co-workers [35, 44, 52–55] anchored an end-functionalized random copolymer to the surface. After removal of the nonattached chains, the surface was essentially a random copolymer brush, where the composition of the random copolymer could be varied in the synthesis to tune the interfacial interactions. It should be noted that narrow molecular weight distributions of the random copolymer or brush length could be obtained by using nitroxide-mediated synthetic routes. Simply by varying the composition of the random copolymer, the interfacial interactions of the blocks with the modified substrate can be controlled. If we take a “random” copolymer that is pure A, then the A-block of the copolymer would preferentially wet the substrate interface and a parallel alignment of the microdomains would result. Similarly, if the “random” copolymer is pure B, then the B-block of the copolymer would wet the substrate and alignment of the microdomains parallel to the substrate would result. Consequently, there must be a small range of composition of the random copolymer where the interactions are balanced and the block copolymer microdomains would orient normal to the surface (Figure 10.2). Although this surface modification has proven to be exceptionally robust and easily applied to very large-area surfaces, this surface-modification process was restricted to homogeneous oxide surfaces. Recently, these researchers further

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

(a )

237

(b)

Figure 10.2. (a) Scanning force micrograph (SFM) of a spin-coated film of PS–PMMA diblock copolymer having ∼20-nm cylindrical microdomains of PMMA oriented normal to the film surface. The film, ∼40 nm thick, was coated onto a surface where the interfacial interactions were balanced. (b) The same film after being exposed to UV radiation, followed by rinsing with alcohol. The nanoporous film of cross-linked PS has pores that are of the same size as the original PMMA microdomains. The inset in (b) shows a field-emission electron micrograph of the film, demonstrating the uniformity in the size of the pores. The images are 2 µm × 2 µm. (Reproduced with permission from [15]. Copyright 2005, Materials Research Society.)

developed a method in which the interfacial interactions of any surface could be manipulated in a rapid, robust manner through the use of a thin, cross-linked random copolymer film [56]. Like with the anchored random copolymer, interfacial interactions are dictated by the average composition of the cross-linked random copolymer film, and the degree of cross-linking can be altered by changing the number of crosslinkable units incorporated into the copolymer. Not only does cross-linking impart insolubility to the film placed on the substrate but it also retards diffusion of BCPs through the cross-linked films, preventing the BCP from “sensing” the substrate. It was found that a 5-nm-thick film was suitably thick to fully shield the substrate. Since the requirement of chemical attachment to the underlying substrate is removed, then this process can be used with any substrate, whether it is flexible or rigid, homogeneous or heterogeneous. The effectiveness in controlling the interfacial interactions was demonstrated with thin diblock copolymer films on a wide range of substrates on which the orientation of the microdomains was perpendicular to the surface (Figure 10.3). In addition, a rough surface can also produce the same effect as discussed by Sivaniah et al. [57].

10.2.1.3 Addition of Homopolymer. Controlling the interfacial interactions provides a simple, passive route to manipulate the orientation of copolymer microdomains in thin films. However, the thickness of BCP films over which the orientation of the microdomains persists is limited to ∼L0 . For thicker films, the orientation normal to the substrate is lost and the microdomains adopt a random orientation. The addition of a small amount of homopolymer that is miscible with the

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

238

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

Thermal evaporation of Au

(a)

Patterned media Au

19 µm 35 nm

Si 6 µm PS-b-PMMA

(b)

PS-b-PMMA on Cross-linked P(S-r-BCB-r-MMA)

(c) On Au On Si

(d)

(e)

(f)

(g)

200 nm

Figure 10.3. (a) Diagram of the evaporation process used to generate gold squares in a grid of silicon oxide, a chemically heterogeneous surface. In the images on the left ((b), (d ), (f )), a diblock copolymer was directly spin coated onto this heterogeneous surface. The reflection optical micrograph (b) reveals the grid underneath the copolymer; the scanning force micrograph of the copolymer on gold (d ) shows that the cylindrical microdomains of the copolymer have a random orientation, while on silicon oxide (f ), the microdomains orient parallel to the surface. The same surface is shown in the images on the right ((c ), (e), (g )), where a cross-linked random copolymer film was used to balance interfacial interactions. The optical micrograph (c ) shows the grid under the film, and (e), (g ) scanning force micrographs show a BCP film on the gold (e) and silicon oxide (g ) covered portions of the surface. In both cases, the cylindrical microdomains are oriented normal to the surface. (Reproduced with permission from Science [56]. Copyright 2005, American Association for the Advancement of Science.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

239

Figure 10.4. Field-emission scanning electron microscopy (FE-SEM) image of the cross-section of a nanoporous film prepared from a mixture of PS–PMMA, having cylindrical microdomains, with a small amount of higher molecular weight PMMA homopolymer. The PMMA stretched along the axes of the cylindrical microdomains, has propagated the orientation of the microdomains over large distances from the surface. (Reproduced with permission from [58]. Copyright 2004, John Wiley & Sons, Inc.)

minor component block, has interactions with the major component block that are more nonfavorable than the interactions between the minor and major component blocks, or has a molecular weight higher than the minor component, has been shown to produce a marked increase in the persistence of the orientation of microdomains normal to the surface, enabling a directed self-assembly of the copolymers into arrays of highly oriented, high aspect ratio cylindrical microdomains over large areas [58– 63]. The confinement of the homopolymer to the minor component microdomain and the interactions of the added homopolymer with the two blocks cause the added homopolymer to assume an extended configuration along the axis of the cylindrical microdomain which, in turn, biases the self-assembly, causing the orientation of the cylindrical microdomain to persist along the direction of the cylindrical microdomain axis. Shown in Figure 10.4 is an example of PS-b-PMMA where a small amount of higher molecular weight PMMA was added. Here, the persistence of the orientation over large distances is evident [58]. Similar results can be achieved by the addition of poly(ethylene oxide) (PEO) to PS-b-PMMA. Here, PEO is miscible with PMMA but χ PS–PEO  χ PS–PMMA . The number of PS–PEO contacts is minimized by forcing the PEO to the center of the cylindrical PMMA microdomains [59]. Removal of both of the homopolymer and minor component block of the copolymer can be used to produce pores that are much smaller than that from the original copolymer (by removing the homopolymer with a selective solvent) or larger by removing both the homopolymer and the minor component block. Thus, in one film, two distinct sizes of pores can be accessed [59].

10.2.1.4 Application of an External Field. External fields can be used to induce the orientation of the microdomains in a specific direction. These approaches

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

240

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

rely on the ability of BCPs to couple with the external applied fields, so that directional control over the microdomains can be achieved. External fields, like electric fields [64–83], solvent evaporation [84–91], mechanical shearing [92, 93], and light [94–96] have been utilized to control the orientation of the microdomains in BCPs films. An external electric field is an effective means of aligning the microdomains along the field direction. Pioneering experiments by Amundson and co-workers [66–68] showed that the alignment of the microdomains arises from the energetic cost associated with the misalignment of the interface between two media with the direction of the applied field. The transition from parallel to perpendicular orientation is first order [70] and generally occurs when the electric field can overcome the parallel orientation induced by interfacial interactions [65, 70, 80]. The difference in the dielectric constants of the microdomains is the driving force for the alignment. In thin films, the electric field may be applied to orient the microdomains either in the plane of the substrate or normal to it. In-plane electrodes can be fabricated by the standard optical lithographic processes along with electron-beam evaporation of a metal. By applying an in-plane field, the long axes of cylinders in a PS-b-PMMA thin film is aligned along the line of the field [69]. Russell and co-workers [64, 65, 76, 83] studied this system in detail and demonstrated that a vertically ordered cylindrical PMMA microdomains could be achieved when the electric field was applied across to the film (Figure 10.5). In addition, under a sufficiently high electric field, spherical microdomains could also be deformed into ellipsoids. For a thin BCP film with multiple layers of spheres, the ellipsoids can be sufficiently stretched that eventually they interconnect to form cylinders that penetrate through the film [79]. Conservation of volume requires that the diameters of the cylinders be smaller than those of spheres. Such a sphere-to-cylinder transition may offer a simple route to generate cylinders with a high aspect ratio from the spherical microdomains [79]. The preparation of BCP thin films under various solvent evaporation conditions turns out to be a very effective route to manipulate the microdomains. The solvent evaporation rate is one of the key factors that dictate these kinetically arrested microstructures. Libera and co-workers discovered that solvent evaporation can be used to induce the ordering and orientation of BCP microdomains [85–87]. Vertically aligned cylindrical PS microdomains could be obtained in a polystyreneblock-polybutadiene-block-polystyrene (PS-b-PBD-b-PS) triblock copolymer thin film with a thickness of ∼100 nm [86]. The same effect was also found in a poly(styrene-block-ethylene oxide) (PS-b-PEO) BCP thin films due to the concentration gradient along the direction normal to the film surface during solvent evaporation [84]. However, it should be noted that in both studies, the lateral registry of the cylinders is poor. More recently, Russell and co-workers [84, 88, 89] and Krausch and coworkers [85, 90] refined this approach and exquisite control over the orientation and ordering of microdomains in BCP thin films were achieved. Shown in Figure 10.6 is a highly ordered hexagonally perforated lamella structure based on the thin film of polystyrene-block-poly(2-vinylpyridine)-block-poly(t-butylmethacrylate) (PS-bP2VP-b-PtBMA) triblock copolymer [90]. The bicontinuous nature of the perforated

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

241

(a)

(b)

Figure 10.5. (a) Small angle neutron scattering (SANS) pattern obtained from a PS–PMMA (f PMMA = 30%) film (800 nm) annealed in an electric field of 25 V mm−1 . The two equatorial reflections indicate a strong orientation of the microphase structure normal to the substrate surface. (b) Cross-sectional transmission electron micrograph obtained from a PS–MMA film (800 nm) annealed in an electric field of 25V mm−1 . The BCP film is lying on top of a dark Aufilm, which was used as the lower electrode. The upper electrode has been removed. Cylinders oriented normal to the substrate go all the way through the sample. (Reproduced with permission from [64]. Copyright 2000, John Wiley & Sons, Inc.)

lamella phase is essential for the long-range order in this system. Whenever the PSb-P2VP-b-PtBMA films do not show the perforated lamella structure (e.g., owing to different thickness), they have a poor long-range order. The long-range ordered cylindrical microdomains [88, 89] in BCP films induced by solvent evaporation will be addressed in detail in Section 10.3. BCPs in bulk can be aligned by mechanical shear. This technique has been widely used to align lamellae, cylinders, spheres, and bicontinuous microdomains, but is limited to relatively thick films. More recently, Chaikin, Register and co-workers [92, 93] have demonstrated that a single layer of cylinders [92] and bilayers of spheres

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

242 (a)

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(c)

(b)

Figure 10.6. FE-SEM images of a 37-nm-thick PS-P2VP-PtBMA film after solvent vapor treatment (a) and subsequent drying (b). (c ) Fourier transform of an FE-SEM image of size 5 µm × 5 µm. (Reproduced with permission from [90]. Copyright 2003, Nature Publishing Group)

[93] in BCP thin films can be aligned by applying shear. In their experiment, an unpatterned PDMS stamp is placed on a heated film. When subjected to a lateral force, the stamp elastically distorts and shears the microdomains in the underlying BCP film to reorient in the direction of shear (Figure 10.7). Optical-alignment methods at a molecular level have been well established in liquid crystal and the relevant systems [97, 98]. Directionally selective light excitation using linearly polarized light of photoisomerizable molecules on a liquid crystalline polymer film produces patterned, oriented microdomains in the films. Recent investigations have revealed that such photoexcited collective molecular motions can lead to lateral mass transport over distances of micrometers [99, 100]. Seki and co-workers [94] proposed a new optical (three dimensional) 3D (both out-of-plane and in-plane) alignment of nanocylinders of a BCP comprising liquid crystalline photoresponsive block chains and PEO by applying the process of photoinduced mass migration. The key for the out-of-plane alignment (whether the cylinders are oriented normal or parallel to the substrate surface) is the control of the film thickness, while that for the in-plane alignment is the direction of the linear polarization during the illumination (Figure 10.8). Moreover, Ikeda and co-workers [95] addressed a noncontacted optical method by using a polarized laser beam to control a parallel patterning of PEO nanocylinders in an amphiphilic liquid crystalline BCP film (Figure 10.9).

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

(a)

(d )

(b)

(e)

(c)

(f )

243

Figure 10.7. Left: Microscopic orientation of polystyrene-block –poly(ethylene-alt -propylene) (PS–PEP) BCP cylinders in angle-layer films by shearing. (a) An atomic force microscopy (AFM) micrograph acquired on a copolymer sample aligned by the shearing method. The cylindrical microdomains are in perfect registry, without any visible defect. The arrow indicates the shearing direction. (b) Another micrograph, acquired 8 mm away from the one shown in (a). A single dislocation is visible (these are particularly well-aligned regions). (c ) Fingerprint-like pattern characteristic of samples annealed without shearing. Scale bar = 250 nm. Right: Tapping-mode atomic force microscopy (TM-AFM) images showing shear-induced ordering of hexagonal bilayer PS–PEP BCP arrays (PS spheres are white). (d ), (e) Sheared bilayer films of sphere-forming BCP show excellent alignment, with one of the lattice vectors oriented along the (horizontal) shear direction. The two micrographs were acquired at different places, with identical orientations, on the same square centimeter of the sample. (f ) Slightly thinner or thicker films show a completely disordered lattice. Here, the micrograph was obtained on a sample incorporating a thickness gradient, in a region where the film was 1 nm (or 2%) thinner than a bilayer. Each micrograph is 1 µm2 . All inset represent the Fourier transforms of the respective AFM micrographs. (Reproduced with permission from [92, 93]. Copyright 2004 & 2005, John Wiley & Sons, Inc.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

244

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

O

H 3C

O 114

X 67

O

O l R

O

R=

(CH2)11 O

N

CH3

N

(d )

(c)

(b)

0.2 µm

2

0.4

(f ) 1 2

0 600 200

0 200 Wavelength (nm) 100 nm (h)

Absorbance

1

Absorbance

0.4

(e)

Absorbance

0.4

600 Wavelength (nm)

(g)

0 200

1

2 600 Wavelength (nm)

PEO cylinder

0 nm Figure 10.8. Thickness dependence of the azobenzene (Az) and cylinder alignment. (a) Chemical structure of the diblock copolymer consists of PEO and poly(methacrylate) containing an Az unit, with the degree of polymerization of the PEO and the azopolymer being 114 and 67, respectively (denoted as p(EO114 –Az67 )). (b)–(d ) Phase-mode AFM images (1 µm × 1 µm) of the p(EO114 –Az67 ) film with different film thickness: (b) 20 nm, (c ) 30 nm, and (d ) 70 nm after annealing and exposure to hexane vapor. (e)–(g ) UV-vis absorption spectra of the corresponding p(EO114 –Az67 ) films are shown: (e) 20 nm, (f ) 30 nm, and (g ) 70 nm thickness for as-cast (1) films and after exposure to hexane vapor (2). (h) Schematic illustration of thickness dependence on cylinder alignment. (Reproduced with permission from [94]. Copyright 2006, John Wiley & Sons, Inc.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

(a)

488 nm

245

Polarization

(b)

100 nm

100 nm Azobenzene

PEO

0

15° 30°

Substrate

Figure 10.9. Scheme of LC alignment and microphase-separated structures in the irradiated and unirradiated area of the BCP films. Inset: AFM phase images of the annealed BCP films in unirradiated (a) and irradiated (b) area. (Reproduced with permission from [95]. Copyright 2006, American Chemical Society.)

10.2.1.5 Epitaxy. Epitaxy describes an ordered crystalline growth on a monocrystalline substrate [7, 22, 101–103]. In 2000, epitaxy was first introduced by Thomas and co-workers [101] to control the spatial and orientational order of BCP microdomains. They employed epitaxy to control the molecular and microdomain orientation of a poly(ethylene-block-ethylenepropylene-block-ethylene) (PE-b-PEPb-PE) semicrystalline BCP thin film. A crystallizable organic solvent serves as a solvent for the semicrystalline BCP at temperatures above the solvent melting temperature. While the block is cooled below the melting point of the solvent, the crystallizable solvent becomes a substrate. Epitaxy usually occurs between a semicrystalline BCP block and a crystalline substrate. On such a crystalline substrate, the semicrystalline block can be oriented by the unit cell of the substrate. This shows an excellent way to form highly aligned edge-on crystalline lamellae in both lamellar and cylindrical microdomains formed from semicrystalline-amorphous BCPs. Epitaxial crystallization of the crystalline block on an organic crystalline substrate such as benzoic acid (BA) [101, 102], or anthracene (AN) [103] can result in precise control of the molecular orientation of the crystalline block and subsequent overall long-range order of the BCP microdomains. In the case of an amorphous BCP, the fast-directional solidification of the solvent during microphase separation leads to an alignment of the BCP interface parallel to the fast growth direction of the solvent crystals [102, 104–106]. Both lamellar and cylindrical microdomains in a symmetric PS-b-PMMA and in an asymmetric polystyrene-block-polyisoprene (PSb-PI) copolymer, respectively, are globally aligned by using either BA or AN as the crystallizable solvent [104, 105]. Typical examples of directionally solidified BCP microstructures are shown in Figure 10.10 [105]. 10.2.1.6 Special BCP: Liquid Crystalline and Rod-Coil BCP. Liquid crystalline (LC) BCPs offer unique interactions for controlling the order of microdomains [107, 108]. Wong et al. [109] observed lamellar microdomains vertically oriented on a glass substrate in a thin film of symmetric amorphous-smectic BCP. In general, the lamellar microdomains in thin films are oriented parallel to the substrate due to interfacial interactions. However, in the films with the mesogenic units in a side-chain LC

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

246

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

(b) PI

BA bBA

500 nm

(c)

(d )

PI

BA 500 nm

bBA

Figure 10.10. (a) TEM bright-field image of a thin film of PS/PI (45/12) BCP, directionally solidified with BA, and stained with OsO4 . The dark regions correspond to the stained PI microdomains. The cylindrical PI microdomains are well aligned along the fast growth direction of the BA crystals (crystallographic b-axis). Inset shows the FFT power spectrum of the TEM micrograph. Spot-like first reflection located on the meridian shows the nearly single crystal-like microstructure. (b) Schematic model of the microstructure of PS/PI processed with BA. Cylindrical PI microdomains are aligned along the b-axis of BA crystal. (c ) Due to very thin film thickness, vertically undulated PI cylinders transform into hexagonally packed cylinders oriented perpendicular to the BA substrate. Inset shows the Fourier transform power spectrum of the TEM micrograph. Spot-like first reflections with sixfold symmetry show the nearly hexagonally packed microstructure. (d ) Schematic model of the microstructure of PS/PI processed with BA. Cylindrical PI microdomains are oriented vertically to the substrate. (Reproduced with permission from [22]. Copyright 2003, Elsevier and [105]. Copyright 2001, American Chemical Society.)

polymer, the mesogens usually prefer homeotropic anchoring to both the substrate and the air interface (except for strongly polar end groups) [108–110]. For a smectic LC phase this will bring the smectic layers parallel to the substrate. Thus, a conflict arises between the directions of the smectic layering and the lamellar ordering that cannot both adjust to the same preferred boundary orientation and remain orthogonal to each other. The resulting frustration can be used advantageously for the control purposes. Recently, a simple rubbing technique was used to align cylindrical

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

247

microdomains in an amphiphilic diblock LC copolymer [111]. Under the influence of the novel supramolecular cooperative motions between the mesogens and the microdomains, a 3D macroscopic array of the nanocylinders oriented parallel to the rubbing direction over a large area was achieved (Figure 10.11). In the rod-coil BCP consisting of polystyrene (PS) and poly(3-(triethoxysily) propylisocyanate) (PIC) blocks, polyisocyanates are stiff, helical, and extended and form a lyotropic nematic liquid crystalline solution. Thomas et al. [112] reported a hierarchical microstructure spanning three orders of size scale consisting of crystals of the rod block (unit cell ∼1 nm), BCP microdomains (layer periodicity ∼50 nm), and N´eel domain walls (wall periodicity ∼1000 nm). This was produced via directional solvent evaporation in a lyotropic nematic solution of a rod-coil PS–PIC. Such hierarchical patterns demonstrate the feasibility of simultaneous organization at both small and large length scales (Figure 10.12). 10.2.2 Self-Directing The combination of self-assembly at different length scales leads to structural hierarchies. This offers rich possibilities to construct nanostructured materials, and switching (responsive) properties based on the phase transitions of the self-assembled structures. In this section, we will review some results of the cooperatively coupling of inorganic particles, salts and low molecular weight organic molecules, with diblock copolymers at nanoscale.

10.2.2.1 Nanoparticles. The organization of inorganic nanocomponents into self-assembled organic or biological materials has been of interest for decades to make functional hybrid materials [23, 113–116]. Theoretical arguments suggested that synergistic interactions between self-organized particles and a self-assembled matrix material can lead to hierarchically ordered structures [117–119]. Recently, Russell and co-workers [120, 121] show that mixtures of polystyrene-block-poly(2vinylpyridine) (PS-b-P2VP) BCP and cadmium selenide (CdSe) nanoparticles exhibit cooperative, coupled self-assembly at nanoscale. In thin films, the copolymers assemble into cylindrical microdomains, which dictate the spatial distribution of the nanoparticles; segregation of the particles to the interfaces of the film mediates interfacial interactions and orients the copolymer domains normal to the surface, even when one of the blocks is strongly attracted to the substrate (Figure 10.13). Experiments on different substrates and film thicknesses showed that the perpendicular orientation is observed regardless of the nature of the substrate or the film thickness. This interplay between assembly processes was also applied to a wide variety of other systems, such as a blend of poly(ethylene glycol) (PEG)-tagged ferritin bionanoparticles, denoted ferritin-PEG, and a lamella-forming BCP of P2VP and PEO, denoted P2VP-b-PEO. The ferritin-PEG bio-nanoparticles are incorporated into PEO microdomains, suppress crystallization, mediate interfacial interactions, and reorient the microdomains normal to the surface. This is an example of synthetic and biologically inspired systems, where a one-step hierarchical self-organization occurs via the interplay between distinct self-assembling processes, producing spatially ordered,

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

248

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

O 51

CH2(OCH2CH2)114 O O Mn = 30,000 Mw/Mn = 1.10

PEO 36.7 SmX 57.7 SmC 96.3 SmA 118.61

O

N N

CH 2(CH2) 2CH 2O

CH 2CH2CH 2CH 3

12 µm

PEO114-b-PMA(Az) 51

(d )

9

(b)

Rubbing direction

6

50 nm

Polyimide

3

(c)

Glass

Rubbing direction

0 1

2

3 µm

Figure 10.11. (a) The chemical structure and properties of the copolymer, PEO114 -bPMA(Az)51 , used in this study. (b) The TEM image of microphase-separated structures in the copolymer film without rubbing treatment. (c ) A schematic illustration of the rubbing technique. The PEO blocks appear as dark dots because they have been selectively stained by exposing the annealed film in RuO4 vapor for 1 min. (d ) Macroscopic regular array of PEO nanocylinders in an area of 12.4 µm × 3.1 µm. The PEO nanocylinders are in a long-range order parallel to the rubbing direction. (Reprinted with permission from [111]. Copyright 2006, John Wiley & Sons, Inc.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

(b)

(a)

40 µm

0

249

(c)

4 µm

0

(d)

PS

PIC

52 nm 0

400 nm

Figure 10.12. Tapping mode AFM images of the hierarchical structure in a PS-IC(39/23) film (thickness: 5 mm) cast on a glass slide from 1% toluene solution. Insets are the Fourier power spectra. Solvent evaporation direction is from right to left. (a) A low magnification phase image. The contrast arises from different orientations of microdomains relative to the AFM scanning direction. (b) A height image obtained after the film was immersed in water overnight to enhance contrast. Brighter regions are the higher PS domains. Arrows indicate the director orientations across the Neel ´ wall. (c ) A high magnification phase image of (a) showing the director patterns of the rods. Note tapering of rod domain at an edge dislocation (arrow). (d ) The chemical structure of PS–PIC and interdigitated smectic rod-coil packing model. (Reprinted with permission from [112]. Copyright 2003, John Wiley & Sons, Inc.)

organic–inorganic and organic–bioparticle hybrid materials. This synergy represents a significant advance over other processes that rely on sequential fabrication steps to incorporate functionality into preorganized templates.

10.2.2.2 Inorganic Salts. Recent studies by Russell et al. [122–125] found that the orientation behavior was highly influenced by the amount of inorganic salts

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

250

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

(b)

(c)

(d )

(e)

(f )

Figure 10.13. SFM topography (a) and phase images (b) of a PS-P2VP BCP film taken after spin-coating and after thermal annealing at 170◦ C for 2 days; (c ) its corresponding crosssectional TEM image. SFM topography (d ) and phase images (e) of films prepared from a mixture of PS-P2VP BCP and CdSe nanoparticles after thermal annealing at 170◦ C for 2 days; (f ) its corresponding cross-sectional TEM image. (Reprinted with permission from [120]. Copyright 2005, Nature Publishing Group.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.2 ORIENTATION

(a)

(c)

251

(b)

(d )

Figure 10.14. Cross-sectional TEM images of pure PS–PMMA thin films after thermal annealing at 175 ± 5◦ C under N2 for 24 h (a) and after applying a ∼40 V µm−1 DC electric field at 175 ± 5◦ C under N2 for 24 h (b). Cross-sectional TEM images of PS–PMMA thin films with lithiumPMMA complexes after thermal annealing at 175 ± 5◦ C under N2 for 24 h (c ) and after applying a ∼40 V µm−1 dc electric field at 175 ± 5 ◦ C under N2 for 24 h (d ). Scale bar: 200 nm. (Reprinted figure with permission from [123]. Copyright 2006, American Physical Society).

that was added into BCPs and the synergistic interactions between inorganic salts and BCPs were similar with those between BCPs and nanoparticles. Figure 10.14 shows the influence of inorganic salts on the orientation of lamellar microdomains in PS-b-PMMA thin films. For pure copolymers, after thermal annealing, several PS-b-PMMA lamellar layers adjacent to the interfaces remain oriented parallel to the interface, even under an external electric field (Figure 10.14b), due to the preferential interactions of PMMA with the substrate. In contrast to the films with lithium chloride (LiCl), after thermal annealing, lamellar microdomains do not completely cover the whole interfaces although some lamellae still orient parallel to the interfaces, indicating that the surface interaction in the system became weak (Figure 10.14c). Under an applied electric field, the complete alignment of lamellar microdomains normal to the surface can be achieved (Figure 10.14d). They showed that the formation of lithium-PMMA complexes in PMMA microdomains, after introducing LiCl into BCPs, resulted in a significantly increased dielectric contrast [123] and χ [124] between two blocks, as well as the modified surface interactions [126]. Similar phenomena were also observed by Russell and co-workers in PS-b-PEO [125] and PS-b-P2VP [127] thin films. They found the orientation of the cylindrical microdomains strongly depended on the salt concentration and the ability of the ions to complex with PEO. The addition of salt gave rise to a change in the orientation of PEO cylinders, from parallel to perpendicular, in the solvent-annealed films (Figure 10.15). The process shows large flexibility in the choice of salts used, including gold or cobalt salts, whereby well-organized patterns of nanoparticles can be generated inside the copolymer microdomains. Consequently, the added salt serves a dual role.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

252

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

(b)

(c)

(d)

Figure 10.15. SFM images of PS–PEO thin films containing KI (a) as spun and (b) after solvent annealing and pure PS–PEO thin films (c ) as spun and (d ) after solvent annealing. (Reprinted with permission from [125]. Copyright 2006, American Chemical Society.)

The first is to orient and order the copolymer microdomains while the second is to serve as a precursor to the fabrication of metal nanoparticles.

10.2.2.3 Organic Molecules. Complexation of amphiphilic organic molecules to polymers by ionic interactions, coordination, or hydrogen bonding leads to polymeric comb-shaped supramolecules (complexes), which self-assemble at a length scale of a few nanometers. Simultaneously, BCPs provide self-assembly at an order of magnitude larger length scale. Thus, directed assembly of these polymeric supramolecules leads to the control of structures at several length scales and anisotropic properties [128–130]. The physical bonds within the supramolecules allow the control of the cleavage of selected constituents. They provide templates for mesoporous materials as well as nano-objects and allow switching conductive or optical properties. Ikkala et al. [130, 131] used hydrogen bonding between 4vinylpyridine monomer units and 3-pentadecyl phenol (PDP) to modify the morphology of poly(styrene-block-4-vinylpyridine) (PS-b-P4VP). The investigation of PS-P4VP and PDP in the bulk showed that the supramolecular assembly of P4VP and PDP changed the BCP morphology from a spherical to cylindrical structures [131]. By dipping the complexes into a selective solvent, PDP can be easily removed, providing nanochannels in the matrix [132]. Figure 10.16 summarizes the recognition-driven supramolecular assemblies with comb-shaped architectures in polymers, the subsequent self-organization and preparation of functional materials as

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

Components

15:28

Printer Name: Yet to Come

Self-organization (hierarchy)

Superamolecule due to recognition

Amphiphiles cleaved (optionaly)

(a)

(b) (d )

(c)

(e)

(f )

Functionalizable nanoporous materials

(g)

Individual nanoobjects

(h)

(i)

Figure 10.16. Comb-shaped supramolecules and their hierarchical self-organization, showing primary and secondary structures. Similar schemes can, in principle, be used both for flexible and rod-like polymers. In the first case, simple hydrogen bonds can be sufficient, but in the latter case a synergistic combination of bondings (recognition) is generally required to oppose macrophase separation tendency. In (a–c ), the self-organized structures allow enhanced processibility due to plastization, and solid films can be obtained after the side chains are cleaved (d ). Selforganization of supramolecules obtained by connecting amphiphiles to one of the blocks of a diblock copolymer (e) results in hierarchically structured materials. Functionalizable nanoporous materials (g ) are obtained by cleaving the side chains from a lamellae-within-cylinders structure (f ). Disk-like objects (h) may be prepared from the same structure by cross-linking slices within the cylinders, whereas nanorods (i ) result from cleaving the side chains from a cylinders-withinlamellae structure. Without loss of generality, (a) is shown as a flexible polymer, whereas (b) and (c ) are shown as rod-like chains. Electron micrographs of the structures have been represented in [131, 133–135]. (Reprinted with permission from [128]. Copyright 2002, American Association for the Advancement of Science.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

254

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

well as nano-objects [130, 131, 133–135]. Recently, perpendicular-oriented cylindrical microdomains were observed in a system where a PS-b-P4VP BCP supramolecular assembled with 2-(4 -hydroxybenzeneazo) benzoic acid (HABA) molecules [136]. Furthermore, it was found that the film can be reversibly switched from the perpendicular to parallel orientation, and vice versa, upon exposure to 1,4-dioxane and chloroform vapor, respectively. The alignment is insensitive to the composition of the confined surface. The modes of hydrogen bonding between the pyridine repeating units and HABA dictated the switch of the orientation.

10.3 LONG-RANGE For some applications, such as separation and display, the long-range ordering of nanoscopic elements or pores is not essential, and the control over the orientation is sufficient. However, for applications requiring addressability, as in magnetic storage, long-range lateral ordering and orientation of elements are important. This section will discuss long-range ordered nanostructures of BCP thin films induced by solvent annealing, graphoepitaxy, and orthogonal fields. 10.3.1 Solvent Annealing Sibener [137] and Russell [138] et al. have shown that evaporation-induced flow in solvent-cast BCP films can produce arrays of nanoscopic cylinders oriented normal to the surface with a high degree of ordering. Recently, Krausch and co-workers [85, 139] showed that solvent annealing could markedly enhance the ordering of BCP microdomains in thin films. Russell and co-workers [88, 89] showed that, by controlling the rate of solvent evaporation and solvent annealing, nearly defect-free arrays of cylindrical microdomains normal to the surface in PS-b-PEO copolymer thin films are produced that span the entire films thickness and have a high degree of longrange lateral ordering, as shown in Figure 10.17. Moreover, the use of a co-solvent enables control over the characteristic length scales in the BCP structures even further. Recent results [140] showed that perpendicular cylindrical microdomains in thin films can be directly obtained by spin-coating PS-b-P4VP diblock copolymers from mixed solvents of toluene and tetrahydrofuran (THF) (Figure 10.18a) and highly ordered cylindrical microdomains over very large areas (Figure 10.18b) formed after annealing the films in THF. This process is independent of the substrate, but strongly dependent on the quality of the solvent and solvent evaporation rate. Solvent evaporation is a strong, highly directional field. Strong repulsion between BCP blocks combined with the directionality of solvent evaporation, where ordering is initiated at the surface of the film and propagates through the entire film, leading to a high degree of long-range lateral ordering with few defects. The power of this technique can be better appreciated if a BCP film highly swollen with a good solvent for both blocks is considered. When the concentration of the solvent in the film is high enough, microphase separation is lost and the solubilized copolymer segments mix throughout the film. As the solvent evaporates, a gradient in the solvent concentration

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.3 LONG-RANGE

(a) 2.00

255

(b)

1.00

0

1.00

0 2.00 µm

(c)

500 nm Figure 10.17. (a) SFM phase image of PS–PEO films 260 nm in thickness, spin cast from benzene, and then annealed benzene/water atmosphere for 48 h. (b) The corresponding triangulation maps. (c ) SFM phase image of thin film (∼0.1 µm) of PS–PEO/4.6k-PEO blend with 5 wt% of homopolymer obtained by spin-coating after annealing in a benzene vapor for 48 h. The images show long-range ordering of the hexagonally packed cylindrical microdomains oriented normal to the surface. (Reproduced with permission from [88, 89]. Copyright 2004, John Wiley & Sons, Inc.)

into the film is obtained, with the concentration of the solvent at the air surface being lowest. As time increases, the concentration of the solvent at the surface decreases to a point where the BCP microphase separates. If the blocks are highly immiscible, as is the case of PS-b-PEO and PS-b-P4VP, microphase separation occurs at a relatively high solvent concentration at the surface and there is significant mobility of the segments due to the presence of the solvent. The solvent also mediates the surface energies of the blocks, and as such, both blocks are located at the surface; a lateral

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

256

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

(b)

(c)

(d )

Figure 10.18. A PS-b-P4VP thin film obtained by spin-coating: (a) SFM height images of an as-spun film; (c ) SFM height images of a highly ordered and oriented array of cylindrical microdomains after solvent annealing. Insets are the corresponding Fourier transform spectra. (b), (d ) the corresponding Voronoi diagrams.

microphase separation occurs only at the surface, and the remainder of the film is disordered. Defects in the lateral ordering of the microdomains, which are energetically costly, are rapidly removed, since the chains are mobile. This produces an array of microdomains at the surface having a long-range order. Further solvent evaporation causes an ordering front to move into the film, and BCP microphase separates. This microphase separation is templated on the existing microphase-separated morphology at the surface. Finally, the front propagates through the entire film, producing highly oriented and highly ordered nanostructures.

10.3.2 Graphoepitaxy Graphoepitaxy is a process in which an artificial topographic surface pattern is employed to control orientation of crystal growth in thin films [22, 141]. It was first applied in sphere-forming BCP by Kramer and co-workers [7, 22, 142–145]. For BCPs, graphoepitaxy is achieved through patterning a hard substrate via standard

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.3 LONG-RANGE

257

photolithography, and then transferring the pattern by either chemical or physical etching. Graphoepitaxy uses these topographic features on the surfaces, simple edges or well-defined troughs, to bias the lattice orientation of BCP microdomains, providing a way for controlling multilevel ordering where a “bottom-up” method, like self-assembly of BCPs, is combined with a “top-down” lithographic method. The primary purpose of graphoepitaxy is to enhance the resolution of the conventional lithographic process by subdividing the patterned features to improve the perfection of ordering of the dense, periodic arrays of nanostructures naturally formed by BCPs.

10.3.2.1 Topographically Patterned Substrate. Kramer and co-workers [143–146] demonstrated the formation of long-range-ordered, in-plane spherical arrays on a patterned substrate. A monolayer of P2VP spherical microdomains in a PS-b-P2VP thin film was placed on the photolithographically patterned substrate, and then annealed to generate ordered structures that propagated several micrometers from the sidewalls of the grooves and the edges of the mesas (Figure 10.19). Besides

(a)

(b) 15 µm

27 nm 20 nm

30 nm

Figure 10.19. (a) AFM image (left) of a single grain of 2D periodic spheres in a PS-b-PVP film on top of a mesa. Voroni tessellation of the sphere array (right), demonstrating absence of defects over the approximately 40 × 40 domain array. Inset: The sharp high order diffraction peaks in the associated fast Fourier transform indicate that the entire mesa region is well ordered with hexagonal symmetry. (b) Perspective schematic of template and cross-sectional schematic of a polyvinylpyridine brush on the SiO2 substrate surface and a monolayer of PVP spheres encased in a styrene matrix. (Reprinted with permission from [144]. Copyright 2003, American Chemical Society.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

258

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

Figure 10.20. SEM images of ordered arrays of spherical PFS domains with number of rows in the groove n = 2–12 within 1D templates of varying width. (Reproduced with permission from [150]. Copyright 2004, Nature Publishing Group.)

one-dimensional (1D) confinement, Kramer and co-workers created a well-ordered array of PS-b-P2VP spheres in a 2D hexagonal-shaped topographical well with a diagonal width of a few micrometers [147]. Long-range ordered nanostructures were also developed through the combination of BCP lithography and graphoepitaxy [148–150]. Cheng et al. demonstrated highly ordered domains of sphere-forming polystyrene-block-polyferrocenyldimethylsilane (PS-b-PFS) in patterned silica templates with various groove widths (Figure 10.20). A nearly perfect alignment of spheres in a large area was formed when the groove width was comparable to the polymer grain size. Studies of patterned templates with various topographical confinements, in particular with the incommensurability of grooves with a bulk copolymer period, have shown that BCPs behave elastically and can conform to various groove widths, thus leading to arrays with tunable row spacings [150]. Graphoepitaxy has also been shown to order cylinder-forming BCPs with the microdomains oriented either perpendicular [151, 152] or parallel [153, 154] to the substrate, and in geometries more complex than parallel grooves, like circles [155] and bends [153].

10.3.2.2 Chemically Patterned Substrate. Chemically patterned substrates with length scales comparable to BCP periodicity can be used to control the orientation and to register the BCP microdomains. Theoretical [156–162] and experimental [33, 43, 163–168] results have indicated that under the appropriate surface patterning and boundary conditions, lateral control over nanostructures can propagate microns away from the surface (deep into the film), thus providing a

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.3 LONG-RANGE

259

true 3D control of the self-assembly process. Russell and co-workers [43, 164] and Krausch and co-workers [33, 163] introduced this concept by placing gold stripes on cleaved silicon substrates to precisely guide the orientation of symmetric PS–PMMA BCPs. Both the commensurability and chemical affinity of the microdomains to the patterned substrate are important. When the commensurate condition is fulfilled, the PS and PMMA blocks are attracted to the gold and silicon oxide regions, respectively. Lamellar microdomains orient themselves perpendicular to the substrate plane and parallel to the striping. However, a mismatch in length scale of only ±10% is sufficient to cause the loss of domain orientation. Nealey and co-workers [166–178] extended this concept by using soft x-rays to chemically pattern surfaces with very precise dimensional control over large areas (on the order of cm2 ). When the attractive energy for each block to a chemically heterogeneous substrate is sufficient, and when the BCP period L0 matches the period of the surface pattern LS , symmetric lamellar BCPs microphase segregate into perpendicular lamellae that are ordered and directed with perfection over arbitrarily large areas (Figure 10.21a). In this directed assembly, the additional interfacial interactions imparted via the surface pattern can stabilize the BCP morphologies and induce the stretching or compression of the individual polymer chains [171]. Figure 10.21b demonstrated that the final BCP structures were influenced by both LS and the surface energy. A well-registered surface-directed BCP film can be formed if the interfacial energy gained from preferential wetting of each block is sufficient to compensate the strain energy that results from deviation between LS and L0 . The perfect ordering on patterned PS brushes was achieved for 42.5 nm < LS < 52.5 nm or LS ≈ L0 ± 10% L0 . In comparison, an imaging layer of poly(styrene-random-methylmethacrylate) (PS-r-PMMA) copolymer with 50 vol% PS produces fairly low interfacial energy contrast between the chemically modified and unmodified stripes. The lower interfacial energy contrast provides less of a driving force for chain stretching or compression, so that defect-free assembly can be achieved only over a narrow range of LS (only LS = L0 ). For cylinder-forming PS–PMMA thin films [170], the cylindrical microdomains oriented in the plane of the film and formed defect-free periodic arrays over large areas in registration with the underlying chemical surface pattern if three constraints were satisfied: the substrate pattern period (LS ) was commensurate with the intercylinder period (L0 ) in BCP bulk, the initial film thickness was quantized with respect to the thickness of a half layer (L/2) or single layer of cylinders (L), and the widths of adjacent stripes of the chemical surface pattern were nearly equal. In working circuits, the final chip architecture requires a variety of designs from parallel lines to right angles, which cannot be fulfilled by BCPs with classic morphologies. Such sharp angles introduce severe curvature constraints on the copolymers. Nealey and co-workers introduced a ternary blend to further extend their surface-directed method to this commercial process. Experimentally 45◦ , 90◦ , and 135◦ bends with a surface periodicity of 65 nm < LS < 80 nm were patterned (Figure 10.22). A BCP-homopolymer blend was found to assemble with perfection in both the linear and sharp corner sections, providing the first evidence that directed assembly could fabricate nonregular geometries [174]. The simulations indicate that the bend corners have a higher homopolymer concentration, by 6–7 vol%, than the

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

260

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(a)

Ls = 42.5 nm

(b)

Ls = 45 nm

1 µm

Ls = Lo = 47.5 nm

Ls = 50 nm

Ls = 52.5 nm

PS-r-PMMA brush (50 vol.% PS and 50 vol.% PMMA)

PS brush 250 nm

Figure 10.21. (a) Top-down SEM images of photoresist and PS–PMMA copolymer (L0 = 48 nm, film thickness of 60 nm) patterns. In the SEM images of the PS–PMMA copolymer, the light and dark regions were PS and PMMA domains, respectively. The perfect epitaxial ordering of a block-copolymer pattern extended over a 5 µm by 5 µm area. In this case, the SAM surface was chemically patterned with a period of LS = 47.5 nm that matched L0 . (Reproduced with permission from [166]). (b) Directed assembly of lamellar PS-b-PMMA (L0 = 48 nm) on surfaces chemically patterned with periodicities 42.5 nm < LS < 52.5 nm. The degree of interfacial energy contrast between the chemically modified and unmodified regions of the surface pattern and the corresponding BCP domains plays a significant role in directed self-assembly. The low contrast PS-r -PMMA 50:50 brush (top row) fails to provide perfect ordering for any LS , but the high contrast PS brush (bottom row) directs the assembly of well-ordered BCP domains for the entire range of LS . (Reproduced with permission from [171]. Copyright 2004, John Wiley & Sons, Inc.)

linear sections of the lamellae. The localized redistribution of homopolymer in the film swelled the domains in the corners in order to accommodate the dimensional differences between the linear period and the corner-to-corner period in this geometry [174–179]. 10.3.3 Sequential, Orthogonal Fields The true 3D control of ordering frequently requires that more than one direction of organization can be controlled. Xu et al. [77] successfully achieved a 3D control over the orientation of lamellar microdomains in PS-b-PMMA thin films by the

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.3 LONG-RANGE

261

65 nm Homopolymer enriched 0.6 0.5

45°

0.4 0.3 500 nm

0.2 Homopolymer depleted

90°

135°

Figure 10.22. Left: Top-down SEM images of angled lamellae in a ternary PS–PMMA/PS/PMMA blend (LB = 70 nm). The chemical surface patterns are fabricated with LS values of 65 nm, and the lamellar domains of the BCP blend are self-assembled and registered around 45◦ , 90◦ , and 135◦ bends. The micrographs each depict a 2 µm by 2 µm area. Right: Redistribution of homopolymer facilitates assembly: concentration map of the homopolymers on the surface, where it is seen that the homopolymer is concentrated at the sharp edges to alleviate curvature constraints arising from the patterning. (Reprinted with permission from [174]. Copyright 2005, American Association for the Advancement of Science.)

combination of an orthogonal elongational flow field with an applied electric field. In their experiment, roll pressing was run at temperature below the order–disorder transition but above the glass transition temperatures of both blocks where the oriented nuclei of the microdomains formed. Subsequent annealing under an applied electric field across the film surface produced long-range ordered and highly oriented lamellar microdomains. As seen in Figure 10.23, there was near perfect alignment parallel

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

262

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

X

Film surface

Z

X Z

200 nm

200 nm (a)

(b)

Figure 10.23. Cross-sectional TEM images of (a) the copolymer–substrate interface and (b) the interior of the film. Insets are the Fourier transform spectra of the TEM micrograph. (Reprinted with permission from [77]. Copyright 2003, American Chemical Society.)

to the flow direction and normal to the surface. The lamellar microdomains near the substrate, however, still exhibited parallel orientations due to surface interactions. Alternatively, one could consider using a field alignment technique, such as an electric field or a surface modification to force cylinders or lamellae to orient perpendicular to a substrate in combination with topographic ordering techniques, such as graphoepitaxy, to control order within the plane of the film. Kim et al. combined solvent annealing and graphoepitaxy to allow the creation of long-range ordered cylindrical microdomains perpendicular to the substrate and in hexagonal arrays [88]. Nealey et al. [151] also used a surface neutralizing random copolymer in combination with topographic substrates to demonstrate the perpendicularly oriented cylindrical microdomains in useful architectures with a long-range ordering and regular geometry in thin films of BCPs (Figure 10.24).

10.4 NANOPOROUS BCP FILMS Porosity control of inorganic and organic nanoporous materials is increasingly critical for high technology applications, such as filtration membranes, patterned templates, and photonic materials. Owing to their ability to form periodically ordered nanostructures, BCPs have received much attention as templates, scaffolds, and masks for nanoporous and mesoporous materials that have potential applications from optics to microelectronics. Several routes to generate nanoporous films via

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.4 NANOPOROUS BCP FILMS

263

4 (a)

(b)

6

14

8

10

Number of rows

12 10 8 6 4 4

6

8

10

12

14

16

Groove width

12 (c)

14

250 nm

Transition

150 nm

Figure 10.24. PS–PMMA diblock copolymer thin films on topographic patterns. (a) Plan-view SEM images of long-range ordered PMMA cylinder arrays with row numbers of N = 4–14. (b) A graph of the number of rows of PMMA cylinders in the groove versus the groove widths expressed in units of the natural row spacing (L0 = 37 nm). (c ) Coexistence of 13 and 14 rows of PMMA cylinders in a groove with a 536 nm width, which corresponds to 14.5 times the natural row spacing (white dotted line: transition line). (Reprinted with permission from [151]. Copyright 2005, Institute of Physics Publishing Inc.)

BCP self-assembly have been reported since Nakahama and co-workers first demonstrated the formation of nanoporous polymer films from a siloxane-functionalized PS–PI system [180]. The most common strategy to generate nanopores in polymer matrices is selective chemical or physical degradation and removal of minor microdomains. Numerous chemical strategies, such as ozonolysis [34, 148, 181– 184], thermal degradation [185–188], ultraviolet (UV) degradation [52, 64, 83, 189– 193], “soft” chemical etch [194–198], as well as cleavable junction point [199–202], and physical routes, such as volume contraction triggered by cross-linking [203],

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

264

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

solvent-induced surface reconstruction [140, 204], and extraction of homopolymer or organic molecules in a blend with BCP matrix [59, 136], have been employed to create nanoporous polymer films. Because of the uniform domain size and mild removal conditions, the precise control can be achieved to yield well-defined monolithic materials. 10.4.1 Ozonolysis In 1992, Smith and Meier [181] investigated the degradation of the polydiene block in polystyrene-block-polybutadiene (PS-b-PBD) or PS-b-PI diblock copolymers by ozonolysis. This work demonstrated that the ozonolysis can effectively cleave the double bonds and remove the polydiene component without negatively affecting the un-cross-linked PS domains. In 1996, Mansky et al. [34] reported the preparation of porous thin films for use as nanolithography masks by ozonolysis of oriented PS-bPBD thin films. Subsequently, Park et al. [32] described tactics for preparing either pits or posts of SiN using BCP masks and a combination of chemical modification and reactive ion etching (RIE). In the protocol for preparing holes (pits) in a SiN substrate, a monolayer of spherical inclusions of polydiene was formed by annealing a PS-b-PI or PS-b-PBD BCP on a SiN substrate. As shown schematically in Figure 10.25, this simple array of spherical PI microdomains was used as both a positive and negative resist. Ozonolysis of the film degrades the PI and cross-links the PS matrix, leaving a thin, cross-linked film of PS containing nanoscopic spherical cavities. With RIE, the film is uniformly etched until the ion beam encounters the spherical cavities (Figure 10.25a). At the location of the cavities, the etching front is advanced by an amount that is commensurate with the size of the spherical cavities and as the etching proceeds to the substrate, a pattern of holes that replicates the pattern in the film is transferred to the layer of silicon nitride with high fidelity (Figure 10.25b). If, on the other hand, the original copolymer film is stained with osmium tetroxide, a selective heavy-element stain for PI, then upon RIE, the etch front is retarded by the osmium. As a result, the etch front in the matrix encounters the substrate first and the matrix pattern is transferred, so an array of dots is formed. Consequently, by using one copolymer, both a hole and dot pattern can be generated on a substrate with an areal density of 1.3 × 1011 elements per square centimeter. Cheng et al. [148, 182] improved the etching sensitivity of the BCPs by incorporating organometallic PFS into the copolymer architecture, while Spatz et al. [183] and Haupt et al. [184] achieved the same goal by loading Au into the P2VP core of PS-b-P2VP micelles. 10.4.2 Thermal Degradation Hedrick et al. [185–187] elegantly demonstrated an approach to generating porous materials, termed nanofoams, using polyimide-based materials. With the aim of preparing interlayer dielectrics with low dielectric constants to prevent “cross talk” in microelectronic devices, polyimide containing thermally labile blocks such as PMMA or poly(propylene oxide) (PPO) was used. Upon heating these materials above the decomposition temperatures of the minor blocks, the volatile byproducts

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

PS

PB

Silicon nitride Stained sample

Ozonated sample

Silicon nitride

Silicon nitride RIE (CF4 /O2)

RIE (CF4 or CF4/O2)

Silicon nitride

Silicon nitride

RIE

RIE

Silicon nitride

Silicon nitride

Silicon nitride

Silicon nitride

Silicon nitride

Silicon nitride

Dots

Holes (a)

(b)

500 nm

Figure 10.25. Diagram of a PS-b-PBD BCP film having a single layer of spherical microdomains of PBD that can be used as a positive resist to transfer a pattern of holes in the underlying substrate or a negative resist to produce nanoscopic dots in the underlying substrate. (a) An SEM micrograph of a partially etched, ozonated monolayer film of spherical microdomains. After the continuous PS matrix at top was taken off, the empty PI domains were exposed (as holes) and appear darker in the micrograph. (b) An SEM micrograph of hexagonally ordered arrays of holes in silicon nitride on a thick silicon wafer. The pattern was transferred from a copolymer film such as that in (a). The darker regions are ∼20-nm-deep holes in silicon nitride, which have been etched out. (Reprinted with permission from [32]. Copyright 1997, American Association for the Advancement of Science.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

266

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

are removed, leaving behind voids in the polyimide with inherently high glass transition temperature. Watkins and co-workers [197, 198] used supercritical fluids (SCFs) to swell all of the microdomains in a triblock copolymer of PEO-b-PPO-b-PEO. A BCP template containing hydrophobic PPO and hydrophilic PEO segments is prepared by spin-coating from a solution containing p-toluene sulfonic acid as an acid catalyst. Tetraethyl orthosilicate (TEOS) was selectively infused into the hydrophilic SCFswollen PEO matrix at 60◦ C and, by calcination in air at 400◦ C, TEOS was converted to silicon oxide and PEO-b-PPO-b-PEO was fully degraded. This leaves behind a mesoporous silica replica of the original microdomain morphology (Figure 10.26). These materials are being considered for low dielectric constant applications in advanced microelectronics. Self-assembled hierarchical porosity in organic polymers can be obtained in a facile manner based on pyrolyzed nanocomposites where phenolic resin is templated

Figure 10.26. SEM micrographs showing the cross-section of a highly ordered mesoporous silicate film prepared on a Si wafer in scCO2 . The film was prepared by infusion and condensation of TEOS within a preorganized PEO127 -b-PPO48 -b-PEO127 BCP film dilated with scCO2 at 60◦ C and a pressure of 123 bar followed by removal of the template by calcination. Inset: The images and their Fourier transform suggest a cubic structure, which is confirmed by XRD data. (Reprinted with permission from [197]. Copyright 2004, American Association for the Advancement of Science.)

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.4 NANOPOROUS BCP FILMS

267

by PS-b-P4VP BCPs [205, 206]. Mild pyrolysis conditions lead only to monomodal mesoscale porosity, as essentially only the PS block is removed (length scale of tens of nanometers), whereas during more severe conditions under prolonged isothermal pyrolysis at 420◦ C the P4VP chains within the phenolic matrix are also removed, leading to additional microporosity (subnanometer length scale). These thermoset resin-based materials can be practically molded into any desired shape and used for sensors, separation materials, filters, and templates for catalysis due to their high density of pores, large surface area per volume unit as well as phenolic hydroxyl groups at the pore walls. Using PS-b-PFS BCPs, ceramic nanodots by pyrolysis has also been outlined [188]. 10.4.3 UV Degradation In the work described above, many materials formed spherical microdomains in the thin film. A top layer of low surface energy block was always present and the spherical inclusions were buried in the film. Thus, the masks were generated in an anisotropic manner under etching conditions. An ideal mask, however, should have channels that were open on the surface of the film and extended through the entire film. Porous materials of this nature would allow for direct deposition (or growth) of materials in the holes, and following liftoff, leave the corresponding patterned surfaces. One way to achieve this porous mask using BCPs is to employ cylinderforming materials that could be manipulated in such a way as to allow for the cylinder orientation perpendicular to the substrate. Subsequent removal of the component that formed the cylinders would leave the nanolithographic masks. The most effective, well-studied system is the PS-b-PMMA system by Russell and co-workers [64]. The random copolymer film was used to render the surface ambivalent toward adsorption of either the PS or PMMA block and thus resulted in a thermodynamic preference for the perpendicular orientation. Exposure to UV radiation cross-links PS block, degrades PMMA, and produces a PS film with an array of nanopores that penetrates through the film via acetic acid rinsing, which is a selective solvent for PMMA [64] (as shown in Figure 10.2). By etching the film with an SF6 ion beam, the pattern of the BCP template was transferred to the underlying silicon. After removal of the copolymer film on top, amorphous silicon was evaporated onto the surface and milled to produce silicon islands, 20 nm in diameter with a center-to-center distance of 30 nm, separated by silicon oxide. Consequently, the pattern of the copolymer is replicated by the silicon and silicon oxide [207]. The conditions for optimization of order in the starting porous PS films have been detailed—the relevant variables being film thickness and annealing time and temperature [208]. This templating process developed by Hawker and Russell has been used in a flash memory application by Black and Guarini [207–210]. Pattern transfer into silicon dioxide by RIE creates a dielectric hard mask. The pattern in the oxide layer can then be transferred into silicon by chemical etching (using HBr gas) to produce silicon channels, as illustrated in Figure 10.27. This enables a higher storage capacity in the semiconductor capacitors without increase in the device area. Using this copolymer mask, the pattern can be transferred into various substrate materials even antiferromagnet–ferromagnet bilayers, such as a FeF2 –Fe bilayer [211]. Furthermore, gold, chromium, and multilayer

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

268

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

HBr

SiO2

Si

(b)

(a)

(c) (e)

(d )

100 nm

ALD TaN gate

100 nm

Silicon substrate

Figure 10.27. High aspect ratio nanopore formation in silicon using a sacrifical silicon oxide mask (itself patterned from a BCP film nanostructure). (a) The initial structure, (b) etching using HBr RIE, (c ) oxide removal by wet chemical etch, (d ) etched silicon pores, (e) atomic layer deposition of tantalum nitride, which acts as the metal gate electrode in the final step to a metal–oxide–semiconductor capacitor. (Reprinted with permission from [209]. Copyright 2002, American Institute of Physics.)

(Cr/Au) nanodot arrays were also made by this process [192]. Notably, nanoporous metal films were prepared using a BCP with the majority of the PMMA phase thus generating an array of PS pillars upon UV exposure and acetic acid rinsing (Figure 10.28). An advantage of the cylindrical nanostructure in BCP is the high aspect ratio of the domains. For a 20-µm-thick film containing cylindrical microdomains that are 20 nm in diameter, an aspect ratio of 1000:1 is obtained. This advantage has been used where, under an electric field, the cylindrical microdomains of PMMA were oriented normal to the surface in films with a thickness of ∼30 µm. Thurn-Albrecht et al. [83] oriented the PMMA cylinders in a PS-b-PMMA copolymer by applying an electric field across the polymer film (Figure 10.29). Upon the removal of PMMA microdomains, metals, including cobalt, lead, and iron, were deposited within the nanopores by electrochemical deposition. The cobalt nanowires showed single magnetic domain behaviors [83] and enhanced coercivities [212] as compared to continuous cobalt films, showing promise for use of these arrays in ultrahigh-density magnetic storage (approaching 1 terabit per square inch). As suggested in their

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.4 NANOPOROUS BCP FILMS

(a)

(b)

500 nm

269

(c)

500 nm

500 nm

Figure 10.28. Schematic diagram of the fabrication process for Cr dot arrays (upper pictures) and height images of tapping mode AFM of each step (lower pictures). (a) Nanoscopic holes in cross-linked PS matrix, (b) evaporated Cr onto the PS template, and (c ) Cr nanodot arrays. The height range of the AFM images is 10 nm. (Reprinted with permission from [192]. Copyright 2002, American Chemical Society.)

report [83], thin porous templates prepared from a PS-b-PMMA material were used by Kim et al. [63, 213] as “nanoscopic reaction vessels.” They exposed nanoporous PS films on a silicon oxide substrate to SiCl4 and later to TiCl4 . In the presence of a small amount of water, silicon oxide and titanium oxide nanodots were grown within the cylindrical pores. After the removal of the template, arrays of nanodots on the surface were produced. Moreover, the high aspect ratio holes that are generated can be used for other applications, for example, elastomer molding. Russell and co-workers [193] have explored the concept of patterning elastomers using porous BCP substrates as templates. A PDMS elastomer precursor was placed onto the nanoporous arrays and cross-linked. Since PDMS does not wet PS and air is trapped in the nanopores, a negative replica of the substrate (i.e., an array of pits in the elastomer) was produced. To obtain a positive replica, it is necessary to fill the nanopores. Arrays of hemispherical caps could be fabricated in the elastomer by removing the air in the pores or dissolving PDMS in solvent or by external forces. However, it was not possible to precisely replicate the nanoporous structure by completely filling the nanopores with precursor polymer solution due to the difficulty in fully removing the trapped air. The effectiveness of nanoporous PS-b-PMMA thin films motivated in-depth work on optimizing the fidelity of the structures and controlling the feature sizes. In 2001, Xu et al. [55] demonstrated that by balancing the interfacial interactions, perpendicular cylinders with ordering in PS–PMMA thin films can be achieved for the copolymers with molecular weights up to 103 kg mol−1 upon thermal annealing. However, a higher molecular weight PS–PMMA copolymer, such as 295 kg mol−1 only showed

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

270 (a) V

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

AI Kapton PS PMMA

(d ) Unfilled polymer terrplate

AU Kapton

(b) Air

(e)

(f)

Electrodeposited cobalt nanowires Gold film Silicon substrate

(c) Nanowires

Figure 10.29. Left: A schematic representation of high density nanowire fabrication in a polymer matrix. (a) An asymmetric diblock copolymer annealed above the glass transition temperature of the copolymer between two electrodes under an applied electric field, forming a hexagonal array of cylinders oriented normal to the film surface. (b) After removal of the minor component, a nanoporous film is formed. (c ) By electrodeposition, nanowires can be grown in the porous template, forming an array of nanowires in a polymer matrix. Right: (d ) FESEM image of a crosssection of a nanoporous BCP film from PS–PMMA. The electrochemically deposited cobalt nanowires are seen growing in the copolymer template; above the wires is an unfilled portion of the copolymer template. The gold strip under the film was used as an electrode for the deposition. (e), (f ) Magnified images of the wires where the uniformity and discreteness of the wires are evident. Label shows width of a single wire is 14.19 nm. (Reprinted with permission from [83]. Copyright 2000, American Association for the Advancement of Science.)

very poor order under the same conditions due to the slow collective motion of the polymer chains required to achieve well-ordered arrays. By applying an electric field, all PS–PMMA films studied showed a perpendicular microdomain orientation. The authors suggested that the forces presented as a result of the external field overwhelming the kinetic limitations even in the highest molecular weight PS–PMMA examined. Thus, by varying the molecular weight, the range of pore sizes could be tuned from 14 to 50 nm. Ober et al. [214, 215] described an interesting combination of “phase-selective” cross-linking chemistry and polymer degradation to generate nanoporous thin films from patternable BCP systems. They have developed a novel copolymer system using poly(α-methylstyrene-block-4-hydroxystyrene) (PαMS-b-PHOST) to achieve the spatial control through a high resolution deep UV lithographic process. Cylindrical microdomains of PαMS oriented perpendicular to the substrate were observed. Because of the high T g of PHOST and the low ceiling temperature of PαMS, UV irradiation of these films to generate free radicals at 80◦ C led to depolymerization of the PαMS. By combining the lithographic characteristics of PHOST with the degradation of PαMS, the micrometer-sized patterns of PαMS-b-PHOST thin films could be generated by negative-tone photoresist technology. This combination enables hierarchical structures formed in thin films from the macroscale down to the nanoscale.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.4 NANOPOROUS BCP FILMS

271

These materials are presently employed as a supported porous thin film to separate proteins in a selected molecular weight range. 10.4.4 Selective Extraction Although the use of PS-b-PMMA BCPs as precursors to nanoporous materials is quite well established, reducing the number of chemicals and thermal processing steps required to generate nanoporosity in BCP thin films is desirable. Recently, several variations on this theme have been reported to utilize other strategies to generate the porous materials from the same precursor blocks. Jeong et al. [59, 61, 203] described the generation of nanostructured PS-b-PMMA films containing up to 10 wt% PMMA homopolymer. The PMMA homopolymer preferentially segregated to the center of the cylindrical PMMA microdomains. Simple dissolution of the PMMA homopolymer in a nonsolvent for PS (acetic acid/water mixture) leads to the generation of voids (6 nm in diameter) in the film. By this, the limitation of the porous structure made at high temperature can be overcome if PS was first cross-linked by ozone at room temperature [203]. The addition of homopolymer to PS–PMMA films showed a quite simple technique to generating nanopores and to controlling pore size. In particular, this method can be used to generate pores that are much smaller than one can be used in the pure PS-b-PMMA copolymers. The ozone-induced cross-linking of PS in a PS-b-PMMA thin film was further used to produce nanoporous films based on the principle of the cross-linking/volume reduction [203]. Simple exposure of an oriented PS-b-PMMA thin film to ozone for a controlled time at room temperature followed by heating to 170◦ C generates nanopores in the film. Increasing the ozone exposure time leads to an increase in pore size (from 3 to 8 nm). The results present a unique method by which the size limit of the pores in BCP thin films can be extended well beyond that neat BCP itself can be reached. As discussed before, solvent annealing of PS-b-PEO thin films generated arrays of highly ordered PEO cylindrical microdomains in a PS matrix [88]. But the removal of PEO block without disturbing the ordering of the arrays would be extremely difficult. To solve the problem, Kim et al. extended the solvent annealing process to PS-b-PEO/PEO and PS-b-PEO/PMMA diblock/homopolymer blends and successfully achieved the same results as that in PS–PEO system [89]. By removal of the homopolymer, the nanoporous structure was generated. Maki-Ontto [132] and Valkama et al. [216] reported the formation of nanoporous membranes by selective extraction of low molecular weight organic molecules and alkylsulfonate Zn(II) complexes from the supramolecular assembles. By mixing PDP and PS-b-P4VP BCPs, a hierarchical structure consisting of the PS matrix with the hexagonally arranged cylinders, which is composed of PDP complexed to the nitrogen donor on P4VP was obtained. A “comb-like” layered structure was formed inside the cylinders with the long alkyl chains from the PDP layered with the PVP chains. By simply subjecting these structures to methanol (a nonsolvent for PS), a significant fraction of PDP could be removed. Building on the self-assembled structures described above, Sidorenko et al. [136] developed a related system to making porous structures (Figure 10.30).

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

272

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

(b)

(a) PS block

PVP block

CH2 CH

CH

m

N SMA

PS

PS

CH2

n

Chloroform

1,4-dioxane

PVP Solvent + HABA vapors

O

PVP

Rinse

NI

C OH N

OH

Electrodeposition

N

HABA

(c)

(d )

(e)

Figure 10.30. Preparation of nanoporous membranes from PS–PVP:HABA BCP composites. (Reproduced with permission from [136]. Copyright 2003, American Chemical Society.)

10.4.5 “Soft” Chemical Etch Recently, Hawker and co-workers [217] prepared BCPs with a cross-linkable monomer intentionally incorporated into one of the blocks. As shown in Figure 10.31, copolymers containing thermally cross-linkable groups, such as benzocyclobutene (BCB), can be annealed to induce ordered microdomains and then, subsequently, heated to cross-link the system, locking in the structure and eliminating the need for a multistep process. Thin films (∼30 nm) of poly((styrene-random-BCB)block-lactic acid) (PSBCB–b–PLA) annealed at 170◦ C, followed by cross-linking at 200◦ C, produced the perpendicular cylindrical microdomains with PLA as the minor component. By washing these films with a weak base, the PLA block was degraded and removed, producing a nanoporous cross-linked template. Nanoporous poly(4fluorostyrene) templates on gold-coated silicon/silicon oxide substrates were prepared by the electric field alignment of poly(4-fluorostyrene)-block-poly(d,l-lactide) (PFSt-b-PLA) thin films followed by mild degradation of the PLA phase via a dilute aqueous base [196]. Such “soft” chemical etch method is generally useful for the preparation of templates and nanostructures that are sensitive to more aggressive removal processes.

10.4.6 Cleavable Junction Penelle et al. [199, 200] prepared PS-b-PMMA copolymers containing (4π + 4π ) anthracene (AA) photodimer at the junction point between two blocks (PS-AAPMMA) that can be cleaved to the parent homopolymer blocks upon heating above 130◦ C or by UV irradiation at 280 nm. Microphase-separated cylindrical morphologies with the cylinders oriented normal to the surface in thin films were initially achieved by annealing spin-coated films in supercritical carbon dioxide at 80◦ C. Heating and/or UV irradiation can cleave the copolymer junction point, affecting an in situ conversion of the copolymer to its parent homopolymers. Subsequent washing with selective solvents removed one homopolymer, producing nanoporous templates. Venkataraman and co-workers [201] and Zhang [202] et al. also placed a cleavable

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

273

10.4 NANOPOROUS BCP FILMS

(a) O N

(1) Bulk, 120°C

+

+

84%

P(S-co-BCB)

16%

(1) A|Et3, benzene 24 h, R.T. (2) D, L-lactide, 80°C

HO

O O

HO O

OH

O

N

O n m 84%

16%

(b)

Base

OH OH

Figure 10.31. (a) Chemical schematic of PS-b-PLA where, in the PS block, there is a random incorporation of vinylbenzocyclobutene (BCB) to cross-link the PS block upon heating. (b) Diagram of the film and reactions. (Reprinted with permission from [217], Copyright 2005, American Chemical Society.)

juncture, a triphenylmethyl (trityl) ether linkage, between PS and PEO, to make a long-range ordered nanoporous thin film. Lee et al. [218] reported the preparation of nanoporous crystalline sheets of penta-p-phenylene (PPP) by using the rod-coil PPP and PPO BCPs with a cleavable juncture. The copolymers can self-assemble into layered microdomains that contain sheets of perforated crystalline PPP in which the perforations filled with PPO [219]. The PPO segment is covalently bonded to PPP through an ester linkage. Hydrolysis of this ester linkage with aqueous KOH and subsequent removal of the coil segments from the ordered structure resulted in a hexagonally ordered perforated layered crystalline structure (Figure 10.32). Recently, Hawker and Russell [220] further extended the cleavable-junction-point method and used PMMA as a photodegradable mid-block of triblock copolymers of poly(ethylene oxide-block-methylmethacrylate-block-styrene)(PEO-b-PMMA-bPS), where PS is the major component and PMMA and PEO are minor components.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

274

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

9 nm

12 nm

COO

O O33 CH3

(KOH)

12 nm

Figure 10.32. Preparation of supramolecular porous crystalline sheets by selective removal of coil segments from a perforated layer structure. (Reproduced with permission from [218]. Copyright 2006, American Chemical Society.)

After solvent annealing, highly ordered, nanoporous arrays with cylindrical pores of 10–15 nm were obtained (Figure 10.33). The power of using a triblock copolymer when compared to a traditional diblock copolymer is evidenced by the ability to exploit and combine the advantages of two separate diblock copolymer systems, the high degree of lateral ordering inherent in PS-b-PEO diblock copolymers plus the facile degradability of PS-b-PMMA diblock copolymer systems, while negating the corresponding disadvantages, poor degradability in PS-b-PEO systems and no long-range order for PS-b-PMMA diblock copolymers. 10.4.7 Solvent-Induced Film Reconstruction Xu et al. group [204] reported solvent-induced film reconstruction, a very simple technique for generating nanoporosity in thin films of PS-b-PMMA. By modifying

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

10.4 NANOPOROUS BCP FILMS

(a)

(b)

275

(c)

Figure 10.33. (a) SFM phase image for PEO-b-PMMA-b-PS triblock copolymer thin films after annealing for 12 h in a benzene vapor with controlled humidity, (b) correspond to triangulation images of SFM image, (c ) TEM images (top view) for PEO-b-PMMA-b-PS triblock copolymer thin films after solvent annealing, followed by UV irradiation. The bright circles correspond to nanopores where the cylindrical microdomains are removed after UV irradiation. (Reproduced with permission from [220]. Copyright 2004, John Wiley & Sons, Inc.)

the interfacial interactions, arrays of PMMA cylinders oriented normal to the surface in PS-b-PMMA thin films could be produced. By exposing the films to acetic acid, PMMA was solvated, while the glassy PS matrix remained intact. Upon drying, a film reconstruction was observed where pores were opened in the positions of the original PMMA cylinders as the PMMA within the pores was transferred to the surface. The pore generation is reversible (and can be cycled); heating the film regenerates the dense film. Extending this concept to PS-b-P4VP copolymers, Park et al. [140] also made a highly ordered nanoporous template with areal density of ∼1011 cm−2 holes (Figure 10.34). They evaporated a thin layer of gold (∼2 nm) on the surface of the reconstructed films to prevent thermal-induced reversible process and the porous structure could be stabilized up to 200◦ C. In addition, this thin gold layer acted as an etching mask, so that an enhanced etching depth with an aspect ratio of ∼3:1 during pattern transfer could be achieved. This combination of solvent annealing and surface reconstruction elegantly showed that by simple manipulation of BCP thin films, nanoporous materials can be generated with narrow pore size distributions and remarkable long-range ordering. The use of BCPs to form a variety of different nanosized periodic patterns continues to be a very active area of research. This chapter summarized efforts to control the orientation and long-range ordering of BCP microdomains in thin films. However, the precise control over the nanopattern formation has proven to be challenging and successful in opening routes to new applications. For example, external fields, solvent annealing, and graphoepitaxy are effective methods to achieve the large-area orientation and ordering, but the processes themselves are not trivial and the patterns obtained are still intact. As is obvious throughout this chapter, countless new strategies and interactions, which could be used to gain greater levels of control are still not thoroughly investigated. A more complete understanding of the potential for interactions between different controlling methods is necessary for BCPs to fulfill their technical potential for nanofabrication. Furthermore, although several different degradation techniques have been shown to be effective for preparing nanoporous

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

276

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

Figure 10.34. TEM image of highly ordered nanoporous PS-b-P4VP copolymer thin film after surface reconstruction and gold evaporation (scale bar: 100 nm). Inset is the corresponding Fourier transform spectra.

materials from ordered BCP precursors, simple and novel degradation techniques should be explored. A number of potential applications have been discussed within this chapter including nanolithographic masks for magnetic data storages and nanowires, the templates of nanoparticles, nanotubes, and nanoporous materials. To achieve enhanced nanofabrication performance, novel functional BCP systems are strongly desired. The combination of “top-down” microscale patterns with “bottom-up” nanopatterns is intriguing for integrating functional hierarchical nanostructures into multipurpose on-chip devices. If the nanopatterned materials are used in real-time applications, multifunctional elements should be incorporated into one-ordered BCP matrix. Functional BCPs are being explored for use in fuel cells, photonics, photovoltaics, multiferroics and biomaterials. REFERENCES 1. Bratton, D., Yang, D., Dai, J., and Ober, C. K. (2006) Recent progress in high resolution lithography. Polym. Adv. Technol. 17, 94–103. 2. Gates, B. D., Xu, Q., Stewart, M., Ryan, D., Willson, C. G., and Whitesides, G. M. (2005) New approaches to nanofabrication: molding, printing, and other techniques. Chem. Rev. 105, 1171–1196. 3. Guo, L. J. (2007) Nanoimprint lithography: methods and material requirements. Adv. Mater. 19, 495–513.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

277

4. Salaita, K., Wang, Y., and Mirkin, C. A. (2007) Applications of dip-pen nanolithography. Nat. Nanotechnol. 2, 145–155. 5. Smith, H. I., Schattenburg, M. L., Hector, S. D., Ferrera, J., Moon, E. E., Yang, I. Y., and Burkhardt, M. (1996) X-Ray nanolithography: extension to the limits of the lithographic process. Microelectron. Eng. 32, 143–158. 6. Campbell, M., Sharp, D. N., Harrison, M. T., Denning, R. G., and Turberfield, A. J. (2000) Fabrication of photonic crystals for the visible spectrum by holographic lithography. Nature 404, 53–56. 7. Segalman, R. A. (2005) Patterning with block copolymer thin films. Mater. Sci. Eng. R48, 191–226. 8. Hamley, I. W. (1998) The Physics of Block Copolymers, Oxford University Press, New York. 9. Hadjichristidis, N., Pispas, S., and Floudas, G. (2002) Block Copolymers: Synthetic Strategies, Physical Properties, and Applications, John Wiley & Sons, New York. 10. Bates, F. S. and Fredrickson, G. H. (1990) Block copolymer thermodynamics: theory and experiment. Annu. Rev. Phys. Chem. 41, 525–557. 11. Bates, F. S. and Fredrickson, G. H. (1999) Block copolymers—designer soft materials. Phyis. Today 52, 32–38. 12. Lodge, T. P. (2003) Block copolymers: past successes and future challenges. Macromol. Chem. Phys. 204, 265–273. 13. Ruzette, A.-V. and Leibler, L. (2005) Block copolymers in tomorrow’s plastics. Nat. Mater. 4, 19–31. 14. Mingqi Li, C. K. O. (2006) Block copolymer patterns and templates. Mater. Today 9, 30–39. 15. Hawker, C. J. and Russell, T. P. (2005) Block copolymer lithography: merging “bottomup” with “top-down” processes. MRS Bull. 30, 952–966. 16. Li, M., Coenjarts, C. A., and Ober, C. K. (2005) Patternable block copolymers. Adv. Polym. Sci. 190, 183–226. 17. Fasolka, M. J. and Mayes, A. M. (2001) Block copolymer thin films: physics and applications. Annu. Rev. Mater. Res. 31, 323–355. 18. Green, P. F., Christensen, T. M., Russell, T. P., and Jerome, R. (1989) Surface interaction in solvent-cast polystyrene-poly(methyl methacrylate) diblock copolymers. Macromolecules 22, 2189–2194. 19. Anastasiadis, S. H., Russell, T. P., Satija, S. K., and Majkrzak, C. F. (1989) Neutron reflectivity studies of the surface-induced ordering of diblock copolymer films. Phys. Rev. Lett. 62, 1852–1855. 20. Coulon, G., Russell, T. P., Deline, V. R., and Green, P. F. (1989) Surface-induced orientation of symmetric, diblock copolymers: a secondary ion mass-spectrometry study. Macromolecules 22, 2581–2589. 21. Russell, T. P., Coulon, G., Deline, V. R., and Miller, D. C. (1989) Characteristics of the surface-induced orientation for symmetric diblock PS/PMMA copolymers. Macromolecules 22, 4600–4606. 22. Park, C., Yoon, J., and Thomas, E. L. (2003) Enabling nanotechnology with self assembled block copolymer patterns. Polymer 44, 6725–6760.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

278

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

23. Hamley, I. W. (2003) Nanotechnology with soft materials. Angew. Chem. Int. Ed. 42, 1692–1712. 24. Huinink, H. P., Brokken-Zijp, J. C. M., van Dijk, M. A., and Sevink. G. J. A. (2000) Asymmetric block copolymers confined in a thin film. J. Chem. Phys. 112, 2452–2462. 25. van Dijk, M. A. and van den Berg, R. (1995) Ordering phenomena in thin block copolymer films studied using atomic force microscopy. Macromolecules 28, 6773–6778. 26. Harrison, C., Adamson, D. H., Cheng, Z., Sebastian, J. M., Sethuraman, S., Huse, D. A., Register, R. A., and Chaikin, P. M. (2000) Mechanisms of ordering in striped patterns. Science 290, 1558–1560. 27. Kellogg, G. J., Walton, D. G., Mayes, A. M., Lambooy, P., Russell, T. P., Gallagher, P. D., and Satija, S. K. (1996) Observed surface energy effects in confined diblock copolymers. Phys. Rev. Lett. 76, 2503–2506. 28. Walton, D. G., Kellogg, G. J., Mayes, A. M., Lambooy, P., and Russell, T. P. (1994) A free energy model for confined diblock copolymers. Macromolecules 27, 6225–6228. 29. Kim, H.-C. and Russell, T. P. (2001) Ordering in thin films of asymmetric diblock copolymers. J. Polym. Sci., Part B: Polym. Phys. 39, 663–668. 30. Thomas, E. L., Kinning, D. J., Alward, D. B., and Henkee, C. S. (1987) Ordered packing arrangements of spherical micelles of diblock copolymers in two and three dimensions. Macromolecules 20, 2934–2939. 31. Yokoyama, H., Mates, T. E., and Kramer, E. J. (2000) Structure of asymmetric diblock copolymers in thin films. Macromolecules 33, 1888–1898. 32. Park, M., Harrison, C., Chaikin, P. M., Register, R. A., and Adamson, D. H. (1997) Block copolymer lithography: periodic arrays of ∼1011 holes in 1 square centimeter. Science 276, 1401–1404. 33. Heier, J., Kramer, E. J., Walheim, S., and Krausch, G. (1997) Thin diblock copolymer films on chemically heterogeneous surfaces. Macromolecules 30, 6610–6614. 34. Mansky, P., Harrison, C. K., Chaikin, P. M., Register, R. A., and Yao, N. (1996) Nanolithographic templates from diblock copolymer thin films. Appl. Phys. Lett. 68, 2586–2588. 35. Huang, E., Russell, T. P., Harrison, C., Chaikin, P. M., Register, R. A., Hawker, C. J., and Mays, J. (1998) Using surface active random copolymers to control the domain orientation in diblock copolymer thin films. Macromolecules 31, 7641–7650. 36. Russell, T. P., Huang, E., and Rockford, L. (2001) Encyclopedia of Materials: Science and Technology, Elsevier Science, London, p. 676. 37. Menelle, A., Russell, T. P., Anastasiadis, S. H., Satija, S. K., and Majkrzak, C. F. (1992) Ordering of thin diblock copolymer films. Phys. Rev. Lett. 68, 67–70. 38. Bassereau, P., Brodbreck, D., Russell, T. P., Brown, H. R., and Shull, K. R. (1993) Topological coarsening of symmetric diblock copolymer films: model 2D systems. Phys. Rev. Lett. 71, 1716–1719. 39. Factor, B. J., Russell, T. P., and Toney, M. F. (1991) Surface-induced ordering of an aromatic polyimide. Phys. Rev. Lett. 66, 1181–1184. 40. Russell, T. P., Menelle, A., Anastasiadis, S. H., Satija, S. K., and Majkrzak, C. F. (1993) Reply to “comment. Fluctuation effects in the ordering of thin diblock copolymer films”. Phys. Rev. Lett. 70, 1352. 41. Green, P. F., Christensen, T. M., Russell, T. P., and Jerome, R. (1990) Equilibrium surface composition of diblock copolymers. J. Chem. Phys. 92, 1478–1482.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

279

42. Mansky, P., Russell, T. P., Hawker, C. J., Mays, J., Cook, D. C., and Satija, S. K. (1997) Interfacial segregation in disordered block copolymers: effect of tunable surface potentials. Phys. Rev. Lett. 79, 237–240. 43. Rockford, L., Liu, Y., Mansky, P., Russell, T. P., Yoon, M., and Mochrie, S. G. J. (1999) Polymers on nanoperiodic, heterogeneous surfaces. Phys. Rev. Lett. 82, 2602–2605. 44. Huang, E., Rockford, L., Russell, T. P., and Hawker, C. J. (1998) Nanodomain control in copolymer thin films. Nature 395, 757–758. 45. Pickett, G. T., Witten, T. A., and Nagel, S. R. (1993) Equilibrium surface orientation of lamellae. Macromolecules 26, 3194–3199. 46. Peters, R. D., Yang, X. M., Kim, T. K., Sohn, B. H., and Nealey, P. F. (2000) Using selfassembled monolayers exposed to X-rays to control the wetting behavior of thin films of diblock copolymers. Langmuir 16, 4625–4631. 47. Laibinis, P. E., Whitesides, G. M., Allara, D. L., Tao, Y. T., Parikh, A. N., and Nuzzo, R. G. (1991) Comparison of the structures and wetting properties of self-assembled monolayers of N-alkanethiols on the coinage metal surfaces, copper, silver, and gold. J. Am. Chem. Soc. 113, 7152–7167. 48. Netzer, L. and Sagiv, J. (1983) A new approach to construction of artificial monolayer assemblies. J. Am. Chem. Soc. 105, 674–676. 49. Troughton, E. B., Bain, C. D., Whitesides, G. M., Nuzzo, R. G.., Allara, D. L., and Porter, M. D. (1988) Monolayer films prepared by the spontaneous self-assembly of symmetrical and unsymmetrical dialkyl sulfides from solution onto gold substrates: structure, properties, and reactivity of constituent functional groups. Langmuir 4, 365–385. 50. Fadeev, A. Y. and McCarthy, T. J. (2000) Self-assembly is not the only reaction possible between alkyltrichlorosilanes and surfaces: monomolecular and oligomeric covalently attached layers of dichloro- and trichloroalkylsilanes on silicon. Langmuir 16, 7268–7274. 51. Genzer, J. and Kramer, E. J. (1997) Wetting of substrates with phase-separated binary polymer mixtures. Phys. Rev. Lett. 78, 4946–4949. 52. Mansky, P., Liu, Y., Huang, E., Russell, T. P., and Hawker, C. (1997) Controlling polymer-surface interactions with random copolymer brushes. Science 275, 1458–1460. 53. Mansky, P., Russell, T. P., Hawker, C. J., Pitsikalis, M., and Mays, J. (1997) Ordered diblock copolymer films on random copolymer brushes. Macromolecules 30, 6810–6813. 54. Harrison, C., Chaikin, P. M., Huse, D. A., Register, R. A., Adamson, D. H., Daniel, A., Huang, E., Mansky, P., et al. (2000) Reducing substrate pinning of block copolymer microdomains with a buffer layer of polymer brushes. Macromolecules 33, 857–865. 55. Xu, T., Kim, H.-C., DeRouchey, J., Seney, C., Levesque, C., Martin, P., Stafford, C. M., and Russell, T. P. (2001) The influence of molecular weight on nanoporous polymer films. Polymer 42, 9091–9095. 56. Ryu, D. Y., Shin, K., Drockenmuller, E., Hawker, C. J., and Russell, T. P. (2005) A generalized approach to the modification of solid surfaces. Science 308, 236–239. 57. Sivaniah, E., Hayashi, Y., Matsubara, S., Kiyono, S., Hashimoto, T., Fukunaga, K., Kramer, E. J., and Mates, T. (2005) Symmetric diblock copolymer thin films on rough substrates: kinetics and structure formation in pure block copolymer thin films. Macromolecules 38, 1837–1849.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

280

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

58. Jeong, U., Ryu, D. Y., Kho, D. H., Kim, J. K., Goldbach, J. T., Kim, D. H., and Russell, T. P. (2004) Enhancement in the orientation of the microdomain in block copolymer thin films upon the addition of homopolymer. Adv. Mater. 16, 533–536. 59. Jeong, U., Kim, H.-C., Rodriguez, R. L., Tsai, I. Y., Stafford, C. M., Kim, J. K., Hawker, C. J., and Russell, T. P. (2002) Asymmetric block copolymers with homopolymers: routes to multiple length scale nanostructures. Adv. Mater. 14, 274–276. 60. Jeong, U., Ryu, D. Y., Kho, D. H., Lee, D. H., Kim, J. K., and Russell, T. P. (2003) Phase behavior of mixtures of block copolymer and homopolymers in thin films and bulk. Macromolecules 36, 3626–3634. 61. Jeong, U., Ryu, D. Y., Kim, J. K., Kim, D. H., Wu, X., and Russell, T. P. (2003) Precise control of nanopore size in thin film using mixtures of asymmetric block copolymer and homopolymer. Macromolecules 36, 10126–10129. 62. Ahn, D. U. and Sancaktar, E. (2006) Perpendicularly aligned, size-and spacingcontrolled nanocylinders by molecular-weight adjustment of a homopolymer blended in an asymmetric triblock copolymer. Adv. Funct. Mater. 16, 1950–1958. 63. Kim, D. H., Lau, K. H. A., Joo, W., Peng, J., Jeong, U., Hawker, C. J., Kim, J. K., Russell, T. P., et al. (2006) An optical waveguide study on the nanopore formation in block copolymer/homopolymer thin films by selective solvent swelling. J. Phys. Chem. B 110, 15381–15388. 64. Thurn-Albrecht, T., Steiner, R., DeRouchey, J., Stafford, C. M., Huang, E., Bal, M., Tuominen, M., Hawker, C. J., et al. (2000) Nanoscopic templates from oriented block copolymer films. Adv. Mater. 12, 787–791. 65. Thurn-Albrecht, T., DeRouchey, J., Russell, T. P., and Jaeger, H. M. (2000) Overcoming interfacial interactions with electric fields. Macromolecules 33, 3250–3253. 66. Amundson, K. and Helfand, E. (1993) Quasi-static mechanical properties of lamellar block copolymer microstructure. Macromolecules 26, 1324–1332. 67. Amundson, K., Helfand, E., Quan, X., and Smith, S. D. (1993) Alignment of lamellar block copolymer microstructure in an electric field. 1. Alignment kinetics. Macromolecules 26, 2698–2703. 68. Amundson, K., Helfand, E., Quan, X., Hudson, S. D., and Smith, S. D. (1994) Alignment of lamellar block copolymer microstructure in an electric field. 2. Mechanisms of alignment. Macromolecules 27, 6559–6570. 69. Morkved, T. L., Lu, M., Urbas, A. M., Ehrichs, E. E., Jaeger, H. M., Mansky, P., and Russell, T. P. (1996) Local control of microdomain orientation in diblock copolymer thin films with electric fields. Science 273, 931–933. 70. Tsori, Y. and Andelman, D. (2002) Thin film diblock copolymers in electric field: transition from perpendicular to parallel lamellae. Macromolecules 35, 5161– 5170. 71. Pereira, G. G. and Williams, D. R. M. (1999) Diblock copolymer melts in electric fields: the transition from parallel to perpendicular alignment using a capacitor analogy. Macromolecules 32, 8115–8120. 72. Tsori, Y., Andelman, D., Lin, C.-Y., and Schick, M. (2006) Block copolymers in electric fields: a comparison of single-mode and self-consistent-field approximations. Macromolecules 39, 289–293. 73. Ashok, B., Muthukumar, M., and Russell, T. P. (2001) Confined thin film diblock copolymer in the presence of an electric field. J. Chem. Phys. 115, 1559–1564.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

281

74. Matsen, M. W. (2005) Stability of a block-copolymer lamella in a strong electric field. Phys. Rev. Lett. 95, 258302. 75. Thurn-Albrecht, T., DeRouchey, J., Russell, T. P., and Kolb, R. (2002) Pathways toward electric field induced alignment of block copolymers. Macromolecules 35, 8106– 8110. 76. Xu, T., Hawker, C. J., and Russell, T. P. (2003) Interfacial energy effects on the electric field alignment of symmetric diblock copolymers. Macromolecules 36, 6178–6182. 77. Xu, T., Goldbach, J. T., and Russell, T. P. (2003) Sequential, orthogonal fields: a path to long-range, 3-D order in block copolymer thin films. Macromolecules 36, 7296–7300. 78. Xu, T., Zhu, Y., Gido, S. P., and Russell, T. P. (2004) Electric field alignment of symmetric diblock copolymer thin films. Macromolecules 37, 2625–2629. 79. Xu, T., Zvelindovsky, A. V., Sevink, G. J. A., Gang, O., Ocko, B., Zhu, Y., Gido, S. P., and Russell, T. P. (2004) Electric field induced sphere-to-cylinder transition in diblock copolymer thin films. Macromolecules 37, 6980–6984. 80. Xu, T., Hawker, C. J., and Russell, T. P. (2005) Interfacial interaction dependence of microdomain orientation in diblock copolymer thin films. Macromolecules 38, 2802–2805. 81. Mansky, P., DeRouchey, J., Russell, T. P., Mays, J., Pitsikalis, M., Morkved, T., and Jaeger, H. (1998) Large-area domain alignment in block copolymer thin films using electric fields. Macromolecules 31, 4399–4401. 82. Xu, T., Zvelindovsky, A. V., Sevink, G. J. A., Lyakhova, K. S., Jinnai, H., and Russell, T. P. (2005) Electric field alignment of asymmetric diblock copolymer thin films. Macromolecules 38, 10788–10798. 83. Thurn-Albrecht, T., Schotter, J., Kastle, G. A., Emley, N., Shibauchi, T., Krusin-Elbaum, L., Guarini, K., Black, C. T., et al. (2000) Ultrahigh-density nanowire arrays grown in self-assembled diblock copolymer templates. Science 290, 2126–2129. 84. Lin, Z. Q., Kim, D. H., Wu, X. D., Boosahda, L., Stone, D., LaRose, L., and Russell, T. P. (2002) A rapid route to arrays of nanostructures in thin films. Adv. Mater. 14, 1373–1376. 85. Fukunaga, K., Elbs, H., Magerle, R., and Krausch, G. (2000) Large-scale alignment of ABC block copolymer microdomains via solvent vapor treatment. Macromolecules 33, 947–953. 86. Kim, G. and Libera, M. (1998) Morphological development in solvent-cast polystyrenepolybutadiene-polystyrene (SBS) triblock copolymer thin films. Macromolecules 31, 2569–2577. 87. Kim, G. and Libera, M. (1998) Kinetic constraints on the development of surface microstructure in SBS thin films. Macromolecules 31, 2670–2672. 88. Kim, S. H., Misner, M. J., Xu, T., Kimura, M., and Russell, T. P. (2004) Highly oriented and ordered arrays from block copolymers via solvent evaporation. Adv. Mater. 16, 226–231. 89. Kim, S. H., Misner, M. J., and Russell, T. P. (2004) Solvent-induced ordering in thin film diblock copolymer/homopolymer mixtures. Adv. Mater. 16, 2119–2123. 90. Ludwigs, S., B¨oker, A., Voronov, A., Rehse, N., Magerle, R., and Krausch, G. (2003) Self-assembly of functional nanostructures from ABC triblock copolymers. Nat. Mater. 2, 744–747.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

282

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

91. Cavicchi, K. A., Berthiaume, K. J., and Russell, T. P. (2007) Solvent annealed thin films of asymmetric polyisoprene-polylactide diblock copolymers. Macromolecules 40, 1181–1186. 92. Angelescu, D. E., Waller, J. H., Adamson, D. H., Deshpande, P., Chou, S. Y., Register, R. A., and Chaikin, P. M. (2004) Macroscopic orientation of block copolymer cylinders in single-layer films by shearing. Adv. Mater. 16, 1736–1740. 93. Angelescu, D. E., Waller, J. H., Register, R. A., and Chaikin, P. M. (2005) Shear-induced alignment in thin films of spherical nanodomains. Adv. Mater. 17, 1878–1881. 94. Morikawa, Y., Nagano, S., Watanabe, K., Kamata, K., Iyoda, T., and Seki, T. (2006) Optical alignment and patterning of nanoscale microdomains in a block copolymer thin film. Adv. Mater. 18, 883–886. 95. Yu, H., Iyoda, T., and Ikeda, T. (2006) Photoinduced alignment of nanocylinders by supramolecular cooperative motions. J. Am. Chem. Soc. 128, 11010–11011. 96. Morikawa, Y., Kondo, T., Nagano, S., and Seki, T. (2007) Photoinduced 3D ordering and patterning of microphase-separated nanostructure in polystyrene-based block copolymer. Chem. Mater. 19, 1540–1542. 97. Ichimura, K. (2000) Photoalignment of liquid-crystal systems. Chem. Rev. 100, 1847–1873. 98. Ikeda, T. (2003) Photomodulation of liquid crystal orientations for photonic applications. J. Mater. Chem. 13, 2037–2057. 99. Natansohn, A. and Rochon, P. (2002) Photoinduced motions in azo-containing polymers. Chem. Rev. 102, 4139–4175. 100. Berg, R. H., Hvilsted, S., and Ramanujam, P. S. (1996) Peptide oligomers for holographic data storage. Nature 383, 505–508. 101. De Rosa, C., Park, C., Lotz, B., Wittmann, J.-C., Fetters, L. J., and Thomas, E. L. (2000) Control of molecular and microdomain orientation in a semicrystalline block copolymer thin film by epitaxy. Macromolecules 33, 4871–4876. 102. De Rosa, C., Park, C., Thomas, E. L., and Lotz, B. (2000) Microdomain patterns from directional eutectic solidification and epitaxy. Nature 405, 433–437. 103. Park, C., De Rosa, C., Fetters, L. J., Lotz, B., and Thomas, E. L. (2001) Alteration of classical microdomain patterns of block copolymers by degenerate epitaxy. Adv. Mater. 13, 724–728. 104. Park, C., De Rosa, C., Lotz, B., Fetters, L. J., and Thomas, E. L. (2003) Molecular and microdomain orientation in semicrystalline block copolymer thin films by directional crystallization of the solvent and epitaxy. Macromol. Chem. Phys. 204, 1514–1523. 105. Park, C., De Rosa, C., and Thomas, E. L. (2001) Large area orientation of block copolymer microdomains in thin films via directional crystallization of a solvent. Macromolecules 34, 2602–2606. 106. Park, C., Cheng, J. Y., Fasolka, M. J., Mayes, A. M., Ross, C. A., Thomas, E. L., and Rosa, C. D. (2001) Double textured cylindrical block copolymer domains via directional solidification on a topographically patterned substrate. Appl. Phys. Lett. 79, 848–850. 107. Muthukumar, M., Ober, C. K., Thomas, E. L. (1997) Competing interactions and levels of ordering in self-organizing polymeric materials. Science 277, 1225–1232. 108. de Jeu, W. H., S´er´ero, Y., and Al-Hussein, M. (2006) Liquid crystallinity in block copolymer films for controlling polymeric nanopatterns. Adv. Polym. Sci. 200, 71–90.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

283

109. Wong, G. C. L., Commandeur, J., Fischer, H., and de Jeu, W. H. (1996) Orientational wetting in hybrid liquid crystalline block copolymers. Phys. Rev. Lett. 77, 5221–5224. 110. Jerome, B., Commandeur, J., and De Jeu, W. H. (1997) Backbone effects on the anchoring of side-chain polymer liquid crystals. Liq. Cryst. 22, 685–692. 111. Yu, H., Li, J., Ikeda, T., and Iyoda, T. (2006) Macroscopic parallel nanocylinder array fabrication using a simple rubbing technique. Adv. Mater. 18, 2213–2215. 112. Park, J.-W. and Thomas, E. L. (2003) Multiple ordering transitions: hierarchical selfassembly of rod-coil block copolymers. Adv. Mater. 15, 585–588. 113. Haryono, A. and Binder, W. H. (2006) Controlled arrangement of nanoparticle arrays in block-copolymer domains. Small 2, 600–611. 114. Hamley, I. W. (2003) Nanostructure fabrication using block copolymers. Nanotechnology 14, R39–R54. 115. Bockstaller, M. R., Mickiewicz, R. A., and Thomas, E. L. (2005) Block copolymer nanocomposites: perspectives for tailored functional materials. Adv. Mater. 17, 1331–1349. 116. Balazs, A. C., Emrick, T., and Russell, T. P. (2006) Nanoparticle polymer composites: where two small worlds meet. Science 314, 1107–1110. 117. Balazs, A. C. (2000) Interactions of nanoscopic particles with phase-separating polymeric mixtures. Curr. Opin. Colloid Interface Sci. 4, 443–448. 118. Lee, J. Y., Shou, Z., and Balazs, A. C. (2003) Modeling the self-assembly of copolymernanoparticle mixtures confined between solid surfaces. Phys. Rev. Lett. 91, 136103. 119. Lee, J. Y., Shou, Z., and Balazs, A. C. (2003) Predicting the morphologies of confined copolymer/nanoparticle mixtures. Macromolecules 36, 7730–7739. 120. Lin, Y., Boeker, A., He, J., Sill, K., Xiang, H., Abetz, C., Li, X., Wang, J., et al. (2005) Self-directed self-assembly of nanoparticle/copolymer mixtures. Nature 434, 55–59. 121. He, J., Tangirala, R., Emrick, T., Russell, T. P., Boker, A., Li, X., and Wang, J. (2007) Self-assembly of nanoparticle-copolymer mixtures: a kinetic point of view. Adv. Mater. 19, 381–385. 122. Xu, T., Goldbach, J. T., Leiston-Belanger, J., and Russell, T. P. (2004) Effect of ionic impurities on the electric field alignment of diblock copolymer thin films. Colloid Polym. Sci. 282, 927–931. 123. Wang, J.-Y., Xu, T., Leiston-Belanger J. M., Gupta, S., and Russell, T. P. (2006) Ion complexation: a route to enhanced block copolymer alignment with electric fields. Phys. Rev. Lett. 96, 128301. 124. Wang, J.-Y., Leiston-Belanger, J. M., Sievert, J. D., and Russell, T. P. (2006) Grain rotation in ion-complexed symmetric diblock copolymer thin films under an electric field. Macromolecules 39, 8487–8491. 125. Kim, S. H., Misner, M. J., Yang, L., Gang, O., Ocko, B. M., and Russell, T. P. (2006) Salt complexation in block copolymer thin films. Macromolecules 39, 8473–8479. 126. Wang, J.-Y., Chen, W., Roy, C., Sievert, J. D., and Russell, T. P. (2008) Influence of ionic complexes on phase behavior of polystyrene-b-poly(methyl methacrylate) copolymers. Macromolecules 41, 963–969. 127. He, J., Wang, J.-Y., Xu, J., Tangirala, R., Shin, D., Russell, T. P., Li, X., and Wang, J. (2007) On the influence of ion incorporation in thin films of block copolymers. Adv. Mater. 19, 4370-4374.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

284

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

128. Ikkala, O. and ten Brinke, G. (2002) Functional materials based on self-assembly of polymeric supramolecules. Science 295, 2407–2409. 129. Ikkala, O. and ten Brinke, G. (2004) Hierarchical self-assembly in polymeric complexes: towards functional materials. Chem. Commun. 16, 2131–2137. 130. ten Brinke, G., Ruokolainen, J., and Ikkala, O. (2007) Supramolecular materials based on hydrogen-bonded polymers. Adv. Polym. Sci. 207, 113–177. 131. Ruokolainen, J., Makinen, R., Torkkeli, M., Makela, T., Serimaa, R., Ten Brinke, G., and Ikkala, O. (1998) Switching supramolecular polymeric materials with multiple length scales. Science 280, 557–560. 132. Maki-Ontto, R., de Moel, K., de Odorico, W., Ruokolainen, J., Stamm, M., ten Brinke, G., and Ikkala, O. (2001) “ Hairy tubes”: mesoporous materials containing hollow selforganized cylinders with polymer brushes at the walls. Adv. Mater. 13, 117–121. 133. de Moel, K., Alberda van Ekenstein, G. O. R., Nijland, H., Polushkin, E., ten Brinke, G., Maeki-Ontto, R., and Ikkala, O. (2001) Polymeric nanofibers prepared from selforganized supramolecules. Chem. Mater. 13, 4580–4583. 134. Ruokolainen, J., Ten Brinke, G., and Ikkala, O. (1999) Supramolecular polymeric materials with hierarchical structure-within-structure morphologies. Adv. Mater. 11, 777–780. 135. Saito, R. (2001) Synthesis of discotic microgels by cross-linking of poly(styreneblock-4-vinylpyridine)/3-n-pentadecylphenol blend film. Macromolecules 34, 4299– 4301. 136. Sidorenko, A., Tokarev, I., Minko, S., and Stamm, M. (2003) Ordered reactive nanomembranes/nanotemplates from thin films of block copolymer supramolecular assembly. J. Am. Chem. Soc. 125, 12211–12216. 137. Hahm, J. and Sibener, S. J. (2000) Cylinder alignment in annular structures of microphase-separated polystyrene-b-poly(methyl methacrylate). Langmuir 16, 4766–4769. 138. Kimura, M., Misner, M. J., Xu, T., Kim, S. H., and Russell, T. P. (2003) Long-range ordering of diblock copolymers induced by droplet pinning. Langmuir 19, 9910–9913. 139. Fukunaga, K., Hashimoto, T., Elbs, H., and Krausch, G. (2002) Self-assembly of a lamellar ABC triblock copolymer thin film. Macromolecules 35, 4406–4413. 140. Park, S., Wang, J.-Y., Kim, B., Xu, J., and Russell, T. P. (2007) Unpublished result. 141. Smith, H. I. and Flanders, D. C. (1978) Oriented crystal growth on amorphous substrates using artificial surface-relief gratings. Appl. Phys. Lett. 32, 349–350. 142. Segalman, R. A., Schaefer, K. E., Fredrickson, G. H., Kramer, E. J., and Magonov, S. (2003) Topographic templating of islands and holes in highly asymmetric block copolymer films. Macromolecules 36, 4498–4506. 143. Segalman, R. A., Yokoyama, H., and Kramer, E. J. (2001) Graphoepitaxy of spherical domain block copolymer films. Adv. Mater. 13, 1152–1155. 144. Segalman, R. A., Hexemer, A., Hayward, R. C., and Kramer, E. J. (2003) Ordering and melting of block copolymer spherical domains in 2 and 3 dimensions. Macromolecules 36, 3272–3288. 145. Segalman, R. A., Hexemer, A., and Kramer, E. J. (2003) Effects of lateral confinement on order in spherical domain block copolymer thin films. Macromolecules 36, 6831–6839.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

285

146. Cheng, J. Y., Ross, C. A., Smith, H. I., and Thomas, E. L. (2006) Templated selfassembly of block copolymers: top-down helps bottom-up. Adv. Mater. 18, 2505–2521. 147. Hexemer, A., Stein, G. E., Kramer, E. J., and Magonov, S. (2005) Block copolymer monolayer structure measured with scanning force microscopy moire patterns. Macromolecules 38, 7083–7089. 148. Cheng, J. Y., Ross, C. A., Thomas, E. L., Smith, H. I., and Vancso, G. J. (2002) Fabrication of nanostructures with long-range order using block copolymer lithography. Appl. Phys. Lett. 81, 3657–3659. 149. Cheng, J. Y., Ross, C. A., Thomas, E. L., Smith, H. I., and Vancso, G. J. (2003) Templated self-assembly of block copolymers: effect of substrate topography. Adv. Mater. 15, 1599–1602. 150. Cheng, J. Y., Mayes, A. M., and Ross, C. A. (2004) Nanostructure engineering by templated self-assembly of block copolymers. Nat. Mater. 3, 823–828. 151. Xiao, S., Yang, X., Edwards, E., La, Y.-H., and Nealey, P. F. (2005) Graphoepitaxy of cylinder-forming block copolymers for use as templates to pattern magnetic metal dot arrays. Nanotechnology 16, S324–S329. 152. Li, H.-W. and Huck, W. T. S. (2004) Ordered block-copolymer assembly using nanoimprint lithography. Nano Lett. 4, 1633–1636. 153. Sundrani, D., Darling, S. B., and Sibener, S. J. (2004) Guiding polymers to perfection: macroscopic alignment of nanoscale domains. Nano Lett. 4, 273–276. 154. Sundrani, D., Darling, S. B., and Sibener, S. J. (2004) Hierarchical assembly and compliance of aligned nanoscale polymer cylinders in confinement. Langmuir 20, 5091–5099. 155. Black, C. T. and Bezencenet, O. (2004) Nanometer-scale pattern registration and alignment by directed diblock copolymer self-assembly. IEEE Transactions on Nanotechnology 3, 412–415. 156. Huang, K. and Balazs, A. C. (1991) Modeling copolymer adsorption on laterally heterogeneous surfaces. Phys. Rev. Lett. 66, 620–623. 157. Pereira, G. G. and Williams, D. R. M. (1998) Diblock copolymer thin films on heterogeneous striped surfaces: commensurate, incommensurate and inverted lamellae. Phys. Rev. Lett. 80, 2849–2852. 158. Pereira, G. G. and Williams, D. R. M. (1998) Equilibrium properties of diblock copolymer thin films on a heterogeneous, striped surface. Macromolecules 31, 5904–5915. 159. Pereira, G. G. and Williams, D. R. M. (1999) Diblock copolymer thin film melts on striped, heterogeneous surfaces: parallel, perpendicular and mixed lamellar morphologies. Macromolecules 32, 758–764. 160. Wang, Q., Nealey, P. F., and de Pablo, J. J. (2003) Simulations of the morphology of cylinder-forming asymmetric diblock copolymer thin films on nanopatterned substrates. Macromolecules 36, 1731–1740. 161. Balazs, A. C., Huang, K., McElwain, P., and Brady, J. E. (1991) Polymer adsorption on laterally heterogeneous surfaces: a monte carlo computer model. Macromolecules 24, 714–717. 162. Petera, D. and Muthukumar, M. (1998) Self-consistent field theory of diblock copolymer melts at patterned surfaces. J. Chem. Phys. 109, 5101–5107. 163. B¨oltau, M., Walheim, S., Mlynek, J., Krausch, G., and Steiner, U. (1998) Surfaceinduced structure formation of polymer blends on patterned substrates. Nature 391, 877–879.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

286

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

164. Rockford, L., Mochrie, S. G. J., and Russell, T. P. (2001) Propagation of nanopatterned substrate templated ordering of block copolymers in thick films. Macromolecules 34, 1487–1492. 165. Heier, J., Genzer, J., Kramer, E. J., Bates, F. S., Walheim, S., and Krausch, G. (1999) Transfer of a chemical substrate pattern into an island-forming diblock copolymer film. J. Chem. Phys. 111, 11101–11110. 166. Kim, S. O., Solak, H. H., Stoykovich, M. P., Ferrier, N. J., de Pablo, J. J., and Nealey, P. F. (2003) Epitaxial self-assembly of block copolymers on lithographically defined nanopatterned substrates. Nature 424, 411–414. 167. Yang, X. M., Peters, R. D., Nealey, P. F., Solak, H. H., and Cerrina, F. (2000) Guided selfassembly of symmetric diblock copolymer films on chemically nanopatterned substrates. Macromolecules 33, 9575–9582. 168. Peters, R. D., Yang, X. M., Wang, Q., de Pablo, J. J., and Nealey, P. F. (2000) Combining advanced lithographic techniques and self-assembly of thin films of diblock copolymers to produce templates for nanofabrication. J. Vac. Sci. Technol. B 18, 3530–3534. 169. Kim, S. O., Kim, B. H., Kim, K., Koo, C. M., Stoykovich, M. P., Nealey, P. F., and Solak, H. H. (2006) Defect structure in thin films of a lamellar block copolymer self-assembled on neutral homogeneous and chemically nanopatterned surfaces. Macromolecules 39, 5466–5470. 170. Edwards, E. W., Stoykovich, M. P., Solak, H. H., and Nealey, P. F. (2006) Long-range order and orientation of cylinderforming block copolymers on chemically nanopatterned striped surfaces. Macromolecules 39, 3598–3607. 171. Edwards, E. W., Montague, M. F., Solak, H. H., Hawker, C. J., and Nealey, P. F. (2004) Precise control over molecular dimensions of block-copolymer domains using the interfacial energy of chemically nanopatterned substrates. Adv. Mater. 16, 1315–1319. 172. Stoykovich, M. P. and Nealey, P. F. (2006) Block copolymers and conventional lithography. Mater. Today 9, 20–29. 173. Register, R. A. (2003) On the straight and narrow. Nature 424, 378–379. 174. Stoykovich, M. P., Mueller, M., Kim, S. O., Solak, H. H., Edwards, E. W., de Pablo, J. J., and Nealey, P. F. (2005) Directed assembly of block copolymer blends into nonregular device-oriented structures. Science 308, 1442–1446. 175. Edwards, E. W., Mueller, M., Stoykovich, M. P., Solak, H. H., de Pablo, J. J., and Nealey, P. F. (2007) Dimensions and shapes of block copolymer domains assembled on lithographically defined chemically patterned substrates. Macromolecules 40, 90–96. 176. Park, S.-M., Stoykovich, M. P., Ruiz, R., Zhang, Y., Black, C. T., and Nealey, P. F. (2007) Directed assembly of lamellae-forming block copolymers by using chemically and topographically patterned substrates. Adv. Mater. 19, 607–611. 177. Stoykovich, M. P., Edwards, E. W., Solak, H. H., and Nealey, P. F. (2006) Phase behavior of symmetric ternary block copolymer-homopolymer blends in thin films and on chemically patterned surfaces. Phys. Rev. Lett. 97, 147802. 178. In, I., La, Y.-H., Park, S.-M., Nealey, P. F., and Gopalan, P. (2006) Side-chain-grafted random copolymer brushes as neutral surfaces for controlling the orientation of block copolymer microdomains in thin films. Langmuir 22, 7855–7860. 179. Daoulas, K. C., Mueller, M., de Pablo, J. J., Nealey, P. F., and Smith, G. D. (2006) Morphology of multi-component polymer systems: single chain in mean field simulation studies. Soft Matter 2, 573–583.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

287

180. Lee, J. S., Hirao, A., and Nakahama, S. (1988) Polymerization of monomers containing functional silyl groups. 5. Synthesis of new porous membranes with functional groups. Macromolecules 21, 274–276. 181. Smith, D. R. and Meier, D. J. (1992) New techniques for determining domain morphologies in block copolymers. Polymer 33, 3777–3782. 182. Cheng, J. Y., Ross, C. A., Chan, V. Z. H., Thomas, E. L., Lammertink, R. G. H., and Vancso, G. J. (2001) Formation of a cobalt magnetic dot array via block copolymer lithography. Adv. Mater. 13, 1174–1178. 183. Spatz, J. P., Herzog, T., M¨oller, S., Ziemann, P., and M¨oller, M. (1999) Micellar inorganic-polymer hybrid systems—a tool for nanolithography. Adv. Mater. 11, 149–153. 184. Haupt, M., Miller, S., Glass, R., Arnold, M., Sauer, R., Thonke, K., M¨oller, M., and Spatz, J. P. (2003) Nanoporous gold films created using templates formed from self-assembled structures of inorganic-block copolymer micelles. Adv. Mater. 15, 829–831. 185. Hedrick, J. L., Carter, K. R., Labadie, J. W., Miller, R. D., Volksen, W., Hawker, C. J., Yoon, D. Y., Russell, T. P., et al. (1999) Nanoporous polyimides. Adv. Polym. Sci. 141, 1–43. 186. Hedrick, J. L., Labadie, J. W., Volksen, W., and Hilborn, J. G. (1999) Nanoscopically engineered polyimides. Adv. Polym. Sci. 147, 61–111. 187. Hedrick, J., Labadie, J., Russell, T., Hofer, D., and Wakharker, V. (1993) High temperature polymer foams. Polymer 34, 4717–4726. 188. Temple, K., Kulbaba, K., Power-Billard, K. N., Manners, I., Leach, K. A., Xu, T., Russell, T. P., and Hawker, C. J. (2003) Spontaneous vertical ordering and pyrolytic formation of nanoscopic ceramic patterns from poly(styrene-b-ferrocenylsilane). Adv. Mater. 15, 297–300. 189. Russell, T. P., Thurn-Albrecht, T., Tuominen, M., Huang, E., and Hawker, C. J. (2000) Block copolymers as nanoscopic templates. Macromol. Symp. 159, 77–88. 190. Kim, H.-C., Jia, X., Stafford, C. M., Kim, D. H., McCarthy, T. J., Tuominen, M., Hawker, C. J., and Russell, T. P. (2001) A route to nanoscopic SiO2 posts via block copolymer templates. Adv. Mater. 13, 795–797. 191. Jeoung, E., Galow, T. H., Schotter, J., Bal, M., Ursache, A., Tuominen, M. T., Stafford, C. M., Russell, T. P., et al. (2001) Fabrication and characterization of nanoelectrode arrays formed via block copolymer self-assembly. Langmuir 17, 6396–6398. 192. Shin, K., Leach, K. A., Goldbach, J. T., Kim, D. H., Jho, J. Y., Tuominen, M., Hawker, C. J., and Russell, T. P. (2002) A simple route to metal nanodots and nanoporous metal films. Nano Lett. 2, 933–936. 193. Kim, D. H., Lin, Z., Kim, H.-C., Jeong, U., and Russell, T. P. (2003) On the replication of block copolymer templates by poly(dimethylsiloxane) elastomers. Adv. Mater. 15, 811–814. 194. Hillmyer, M. A. (2005) Nanoporous materials from block copolymer precursors. Adv. Polym. Sci. 190, 137–181. 195. Zalusky, A. S., Olayo-Valles, R., Wolf, J. H., and Hillmyer, M. A. (2002) Ordered nanoporous polymers from polystyrene-polylactide block copolymers. J. Am. Chem. Soc. 124, 12761–12773. 196. Crossland, E. J. W., Ludwigs, S., Hillmyer, M. A., and Steiner, U. (2007) Freestanding nanowire arrays from soft-etch block copolymer templates. Soft Matter 3, 94–98.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

288

15:28

Printer Name: Yet to Come

PATTERNING WITH BLOCK COPOLYMERS

197. Pai, R. A., Humayun, R., Schulberg, M. T., Sengupta, A., Sun, J.-N., and Watkins, J. J. (2004) Mesoporous silicates prepared using preorganized templates in supercritical fluids. Science 303, 507–510. 198. Vogt, B. D., Pai, R. A., Lee, H.-J., Hedden, R. C., Soles, C. L., Wu, W.-L., Lin, E. K., Bauer, B. J., et al. (2005) Characterization of ordered mesoporous ssilica films using small-angle neutron scattering and X-ray porosimetry. Chem. Mater. 17, 1398–1408. 199. Goldbach, J. T., Russell, T. P., and Penelle, J. (2002) Synthesis and thin film characterization of poly(styrene-block-methyl methacrylate) containing an anthracene dimer photocleavable junction point. Macromolecules 35, 4271–4276. 200. Goldbach, J. T., Lavery, K. A., Penelle, J., and Russell, T. P. (2004) Nano- to macro-sized heterogeneities using cleavable diblock copolymers. Macromolecules 37, 9639–9645. 201. Yurt, S., Anyanwu, U. K., Scheintaub, J. R., Coughlin, E. B., and Venkataraman, D. (2006) Scission of diblock copolymers into their constituent blocks. Macromolecules 39, 1670–1672. 202. Zhang, M., Yang, L., and Russell, T. P. (2007) Highly ordered nanoporous thin films from cleavable polystyrene-block-poly(ethylene oxide). Adv. Mater. 19, 1571–1576. 203. Jeong, U., Ryu, D. Y., Kim, J. K., Kim, D. H., Russell, T. P., and Hawker, C. J. (2003) Volume contractions induced by crosslinking: a novel route to nanoporous polymer films. Adv. Mater. 15, 1247–1250. 204. Xu, T., Stevens, J., Villa, J. A., Goldbach, J. T., Guarini, K. W., Black, C. T., Hawker, C. J., and Russell, T. P. (2003) Block copolymer surface reconstruction: a reversible route to nanoporous films. Adv. Funct. Mater. 13, 698–702. 205. Kosonen, H., Valkama, S., Nykanen, A., Toivanen, M., ten Brinke, G., Ruokolainen, J., and Ikkala, O. (2006) Functional porous structures based on the pyrolysis of cured templates of block copolymer and phenolic resin. Adv. Mater. 18, 201–205. 206. Valkama, S., Nyk¨anen, A., Kosonen, H., Ramani, R., Tuomisto, F., Engelhardt, P., ten Brinke, G., Ikkala, O., et al. (2007) Hierarchical porosity in self-assembled polymers: post-modification of block copolymer–phenolic resin complexes by pyrolysis allows the control of micro- and mesoporosity. Adv. Funct. Mater. 17, 183–190. 207. Black, C. T., Guarini, K. W., Milkove, K. R., Baker, S. M., Russell, T. P., and Tuominen, M. T. (2001) Integration of self-assembled diblock copolymers for semiconductor capacitor fabrication. Appl. Phys. Lett. 79, 409–411. 208. Guarini, K. W., Black, C. T., and Yeung, S. H. I. (2002) Optimization of diblock copolymer thin film self assembly. Adv. Mater. 14, 1290–1294. 209. Guarini, K. W., Black, C. T., Zhang, Y., Kim, H., Sikorski, E. M., and Babich, I. V. (2002) Process integration of self-assembled polymer templates into silicon nanofabrication. J. Vac. Sci. Technol. B 20, 2788–2792. 210. Guarini, K. W., Black, C. T., Milkove, K. R., and Sandstrom, R. L. (2001) Nanoscale patterning using self-assembled polymers for semiconductor applications. J. Vac. Sci. Technol. B 19, 2784. 211. Liu, K., Baker, S. M., Tuominen, M., Russell, T. P., and Schuller, I. K. (2001) Tailoring exchange bias with magnetic nanostructures. Phys. Rev. B 63, 060403. 212. Shibauchi, T., Krusin-Elbaum, L., Gignac, L., Black, C. T., Thurn-Albrecht, T., Russell, T. P., Schotter, J., Kastle, G. A., et al. (2001) High coercivity of ultra-high-density ordered Co nanorod arrays. J. Magn. Magn. Mater. 226–230, 1553–1554.

P1: OTA/XYZ P2: ABC c10 JWBK252/Lee September 20, 2008

15:28

Printer Name: Yet to Come

REFERENCES

289

213. Kim, D. H., Jia, X., Lin, Z., Guarini, K. W., and Russell, T. P. (2004) Growth of silicon oxide in thin film block copolymer scaffolds. Adv. Mater. 16, 702–706. 214. Du, P., Li, M., Douki, K., Li, X., Garcia, C. B. W., Jain, A., Smilgies, D.-M., Fetters, L. J., et al. (2004) Additive-driven phase-selective chemistry in block copolymer thin films: the convergence of top-down and bottom-up approaches. Adv. Mater. 16, 953–957. 215. Li, M., Douki, K., Goto, K., Li, X., Coenjarts, C., Smilgies, D. M., and Ober, C. K. (2004) Spatially controlled fabrication of nanoporous block copolymers. Chem. Mater. 16, 3800–3808. 216. Valkama, S., Ruotsalainen, T., Kosonen, H., Ruokolainen, J., Torkkeli, M., Serimaa, R., ten Brinke, G., and Ikkala, O. (2003) Amphiphiles coordinated to block copolymers as a template for mesoporous materials. Macromolecules 36, 3986–3991. 217. Leiston-Belanger, J. M., Russell, T. P., Drockenmuller, E., and Hawker, C. J. (2005) A thermal and manufacturable approach to stabilized diblock copolymer templates. Macromolecules 38, 7676–7683. 218. Lee, M., Park, M.-H., Oh, N.-K., Zin, W.-C., Jung, H.-T., and Yoon, D. K. (2004) Supramolecular crystalline sheets with ordered nanopore arrays from self-assembly of rigid-rod building blocks. Angew. Chem. Int. Ed. 43, 6465–6468. 219. Lee, M., Cho, B.-K., and Zin, W.-C. (2001) Supramolecular structures from rod-coil block copolymers. Chem. Rev. 101, 3869–3892. 220. Bang, J., Kim, S. H., Drockenmuller, E., Misner, M. J., Russell, T. P., and Hawker, C. J. (2006) Defect-free nanoporous thin films from ABC triblock copolymers. J. Am. Chem. Soc. 128, 7622–7629.

P1: OTA/XYZ P2: ABC c11 JWBK252/Lee September 18, 2008

14:3

Printer Name: Yet to Come

11 PERSPECTIVE ON APPLICATIONS

The future of nanotechnology depends on the successful development of versatile, cost-effective techniques for micro- and nanofabrication. Historically, much of the work in this area has been driven by the needs of the microelectronics industry. The resulting techniques—photolithography, electron-beam lithography, etc.—are extremely well suited to the requirements of this industry, and closely related ones (e.g., certain areas of integrated optics). Such methods involve the formation of structures of radiation-sensitive materials (e.g., photoresists or electron-beam resists) on ultraflat glass or semiconductor surfaces. These template resist structures then direct subsequent processing, such as physical vapor deposition or reactive ion etching, to yield patterns of functional materials. Structures in the third (i.e., out-of-plane) dimension are built sequentially, in a layer-by-layer fashion, through repeated application of these patterning steps, often with intermediate procedures to planarize underlying patterns. Significant challenges exist in adapting these methods for manufacturing in areas of application that require unconventional systems and materials (e.g., those in biotechnology, plastic electronics, etc.), structures with nanometer dimensions (i.e., below 30–50 nm), large patterned areas (i.e., larger than a few square centimeters), nonplanar (i.e., rough or curved) surfaces, three-dimensional layouts (i.e., corresponding to more than 5–6 two dimensionally patterned layers) or demanding operating conditions (i.e., outside of controlled clean-room environments). These established techniques also suffer from capital and operational costs that are far too high for many applications. This situation creates opportunities for the insertion of unconventional approaches to nanofabrication into manufacturing flows for important devices. The most compelling of these opportunities are, of course, in areas where existing methods have significant disadvantages, as outlined above. This book summarized in its first part Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


291

P1: OTA/XYZ P2: ABC c11 JWBK252/Lee September 18, 2008

292

14:3

Printer Name: Yet to Come

PERSPECTIVE ON APPLICATIONS

some of the methods that have, in our opinion, capabilities that offer outstanding promise in this sense. The second part consists of chapters that delineate areas of application where these methods have achieved levels of technical sophistication and competitive positions that, taken together, indicate good potential for realistic implementation. These include wide-ranging possibilities and various techniques, from microfluidics to displays to flexible electronics and from soft lithography to selfassembly, respectively. In microfluidics (Chapters 12–14), for example, newly demonstrated abilities to form closed channels in single or multilayer configurations and to embed soft, deformable valves and other components derive directly from the fabrication procedures. Similar techniques provide advantages in large-area patterning for systems that do not demand accurate multilevel registration, such as liquid crystal alignment layers and other components for flat panel displays (Chapters 20 and 21) as well as newer classes of plasmonic-based optical elements (Chapter 19). Optical as well as biological devices can be explored by combining top-down approaches with bottom-up approaches (Chapter 15). In the fields of flexible and stretchable electronics (Chapters 16–18), the fabrication procedures offer not only cost-effective implementation for large areas but also the opportunity to manipulate materials (e.g., organic semiconductors and inorganic nanomaterials) that are chemically or structurally incompatible with established techniques. Through these examples, this collection of chapters covers many of the areas of opportunity expected on the basis of the operational features and expected cost structures of the patterning methods. These cases do not, however, comprise an exhaustive list. Many additional important possibilities exist, such as in subwavelength optical elements for polarizers, waveplates, and other components. An even more exciting aspect is that still others, as yet unknown, will likely be uncovered in the coming years.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

II APPLICATIONS

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


293

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12 SOFT LITHOGRAPHY FOR MICROFLUIDIC MICROELECTROMECHANICAL SYSTEMS (MEMS) AND OPTICAL DEVICES Svetlana M. Mitrovski, Shraddha Avasthy, Evan M. Erickson, Matthew E. Stewart, John A. Rogers, and Ralph G. Nuzzo

12.1 INTRODUCTION Interest in microfluidic devices has grown enormously over the past decade, with applications being found in such disparate areas as bioanalytical chemistry [1, 2] and high speed optical data switching [3, 4]. Progress in this area has followed as a direct consequence of the advent of advanced etching and lithographic techniques capable of creating fluidic circuits [5, 6]. The earliest chip-based microfluidic devices appeared as an answer to a continuously growing need to perform more rapid chemical analyses, while using less reagent, and generating less waste. These first so-called lab-on-a-chip devices [7] were relatively simple, fabricated in silicon and glass using production techniques such as photolithography and etching that were inherited from the microelectronics industry. These simple designs were rapidly supplemented with more complex microfluidic structures, resulting in longer device processing times as well as higher costs of fabrication [8–10].

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


295

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

296

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

Master Silanize

Pour PDMS and cure

Peel-off Stamp Adhere to surface

Microchannel Figure 12.1. A photoresist-derived master fabricated by photolithography is exposed to tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane (“no stick”). PDMS prepolymer is poured over the master, after which the entire assembly is cured at 70◦ C for 2 h. A mold containing the exact replica of the master is peeled off and subsequently pressed onto a surface to which it conformally seals to form microchannels.

In order to produce low cost and disposable microfluidic devices that would, in turn, allow automation and massive parallel integration of analytical systems for studies in chemistry and biology, new fabrication methods were needed. One approach that is showing significant potential in this regard is a method of fabrication known as soft lithography [11]. Soft lithography refers to a series of patterning techniques that use molded polymeric stamps (usually a heat-curable elastomer such as poly(dimethylsiloxane) (PDMS)) as both a patterning tool [12, 13] and as the physical structures of the microfluidic devices themselves [14, 15]. Figure 12.1 schematically depicts the major steps involved in the fabrication of a microfluidic device of the latter type using a rapid prototyping protocol. The process involves the replica molding of a structure known as the master, which is typically fabricated in an organic resist material supported on a Si wafer by photolithography. The height and width of the features on the master are determined by the nature of the photoresist used and the design rules of the photomask with typical dimensions residing in the micrometer size-range. A polymeric mold is then generated by casting and curing a PDMS prepolymer against the master followed by a subsequent peeling step to remove the cured PDMS form. The “stamp” obtained in this way consists of a negative replica of the master against which it was molded. A single master can be used numerous times and, unlike photolithography, does not require the use of a clean room. The PDMS stamp can also be used as a patterning tool. Soft lithographic methods such as microcontact printing (µCP) [11, 16–18] nanotransfer printing (nTP) [19–21], and soft-UV (ultraviolet) nanoimprint lithography [22, 23] are routinely

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.2 MICROFLUIDIC DEVICES FOR CONCENTRATION GRADIENTS

297

used to pattern nanoscale features of a variety of materials over large areas in parallel. The first two methods involve the transfer of an ”ink” from a structured PDMS stamp to a receiving substrate, whereas the latter method involves physically molding a material against a PDMS stamp. Soft-UV nanoimprint lithography, which can achieve replication fidelity down to the molecular regime [24], has been used to form highly ordered nanostructured surfaces for plasmonic sensing [25–27]. The advantages of soft lithography versus its more conventional counterpart— photolithography—have been discussed extensively in the literature [10, 28, 29]. Factors that appear most intriguing to workers in the field are its inherent flexibility, simple patterning protocols, fast cycle times, and relatively low cost. Another attractive aspect of PDMS is its elastomeric nature, which allows it to be conformally sealed to essentially any surface including Si [30], glass [31], and coinage metals [32]. This capacity for heterogeneous integration is a feature that facilitates the coupling of various PDMS-based components to provide complex three-dimensional (3D) microfluidic structures [33–35]. Some of these assemblies may have as many as several hundred integrated components—channels, valves, mixers, and/or pumps as examples—organizations resembling in many ways the architectures of electronic integrated circuits [5]. The photograph in Figure 12.2a powerfully illustrates the growing complexity and functionality of architectures fabricated using soft lithography that have been described in the literature. The device shown in the figure [6, 36], a system for studying protein interactions, consists of several registered layers of microfluidic channel structures that are separated by thin PDMS membranes. The membranes serve as a form of actuator that expand or contract by varying the hydrostatic pressure supplied to the upper channels (Figure 12.2a) which, in turn, drive motion in the fluid contained in the channels lying underneath. The membrane thus functions as an actuatable PDMS-based valve. Three such valves can be positioned along a channel to open and close sequentially and effectively function as a peristaltic pump, the performance of which is schematically illustrated in Figure 12.2b [35]. A number of other ways to manipulate fluids in microfluidic channels (both passive and active) have been described in the literature, including physical flow mediated by microgrooved surfaces [37], electrowetting phenomena [38], actuation via electrochemical potential [39, 40], and methods that exploit gradients in temperature [41], surface tension [42], and surface charge [43]. A review on the physics underlying these phenomena has been published recently by Squires et al. [36].

12.2 MICROFLUIDIC DEVICES FOR CONCENTRATION GRADIENTS Microfluidic devices have also proven exceptionally useful in applications that involve the formation of concentration gradients needed to study the behavior of living cells [13, 44, 45]. These studies involve cell adhesion [46–48], cell movement [49], as well as chemotaxis and axon guidance [45]. Gradient patterning on surfaces using microfluidic methods has been described by Delamarche and co-workers [50]. Whitesides and co-workers [51] described an approach that utilizes the laminar flow

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

298

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

(a)

(b) Fluid out

Vertical gap: 30 µm

Air in/out

Fluid in Figure 12.2. (a) An integrated multilayered microfluidic chip. The colored control lines manipulate valves that function as gates and pumps, which drive reagent solutions through the uncolored channels to be stored in circular chambers for subsequent reaction [36]. (Reprinted with permission from [36]. Copyright 2005 by the American Physical Society.) (b) Three valves connected in series forming a peristaltic pump [35]. (From [35]. Reprinted with permission from AAAS.)

of multiple streams flowing side by side to generate gradients within a channel that spans its width. A representative example of a microfluidic device of this type is shown in Figure 12.3a, where three dyes are driven through a series of mixing trees by the application of hydrostatic pressure to generate solution phase concentration gradients at specific locations downstream. The properties of these gradients are tunable by both flow velocities and variation of the design rules of the microfluidic device. The composition gradients created in solution can be translated onto a solid substrate to yield concentration gradients immobilized on a surface. An approach developed by our group uses microfluidic depletion to produce gradients of proteins immobilized on a functionalized glass surface [13]. The gradients that can be

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.2 MICROFLUIDIC DEVICES FOR CONCENTRATION GRADIENTS

(a)

299

(b)

Direction of flow 200 µm

Red Intensity (a.u)

50

(c)

40 30 20 10 0 0

200 400 600 800 100 1200 Distance (µm)

(d )

2 mm

(e) PDMS stamp Diffusive flux Au Si

Figure 12.3. (a) A photograph showing a microfluidic device for generating gradients. The three incoming channels (top part of the photograph) are connected to syringes via tubing (not shown). Diffusive mixing occurs in the multiple channels, which converge into a single, wide channel (bottom of the photograph shown in (a)) where a gradient is formed perpendicular to the direction of flow (across the channel) [51, 52]. (Reprinted with permission from [51].) (b) Fluorescent image of a multiple-gradient array of TRITC-labeled bovine serum albumin formed on glass via laminar flow microfluidic depletion. Channel width is 70 µm. (c ) Intensity profiles of (b) for each vertical line gradient [13]. (Reprinted with permission from [13]. Copyright 2003 American Chemical Society.) (d ) Aplysia californica bag cell neurons grown on patterned polylysine exhibiting pattern defined morphology [48]. (Reprinted with permission from [48]. Copyright 2004 FASEB Journal.) (e) A PDMS stamp is inked by placing the stamp in contact with a thiol-containing ink pad. The thiol diffuses into the bulk of the stamp and subsequently leaves the material when the stamp is placed in contact with a gold surface due to adsorption of thiol at the Au/PDMS interface. The thickness contour of the stamp produces a spatially defined concentration gradient on the surface due to depletion effects in the stamp. Unoccupied surface sites can be backfilled with other thiols to form a two-component surface composition gradient [53]. (Reprinted with permission from [53]. Copyright 2005 American Chemical Society.)

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

300

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

obtained in this way are illustrated by the fluorescence micrograph shown in Figure 12.3b and the corresponding intensity profiles in Figure 12.3c. These data demonstrate that, as a consequence of different flow rates, immobilized protein gradients of varying length and slope can be formed. Patterned surfaces of this type are proving to be especially useful as functional substrates for controlling and studying the growth of neuronal cells and circuits [48]. A representative example of cells grown on surfaces patterned by microfluidic depletion—here bag cell neurons taken from Aplysia californica—is illustrated by the micrograph shown in Figure 12.3d [48]. A recent report described an intriguing approach to forming gradients on surfaces that exploit the capabilities of PDMS to absorb certain alkanethiols, leaving them free to diffuse in the bulk material. A wedge-shaped PDMS stamp is inked with an alkanethiol by placing it in contact with a slab of PDMS containing a defined thiol concentration (the ink pad). After equilibration, the wedge-shaped stamp is placed in contact with a gold surface (Figure 12.3d). The thickness contour of the stamp leads to a spatially defined complement in the depletion gradient of the printed alkanethiol self-assembled monolayer (SAM) [53]. This latter example demonstrates the considerable capacities that soft lithography holds—here in the form of a unique application of µCP [18]—to prepare complex surface and interfacial structures.

12.3 ELECTROCHEMISTRY AND MICROFLUIDICS Electrochemistry has traditionally served as a popular detection method for analytical separations [54, 55], but it can also offer alternative ways to controllably release acids, bases, radicals, reactive gases or ions, metals, and many types of reducing and oxidizing species. Microfluidic architectures provide an empowering platform for harnessing electrochemically generated potentials and fluxes for applications in sensing and chemical actuation. Exemplary applications have been reported that include analytical separations [56], chemical titrations [57] and synthesis [58], fluid actuation [42], and surface patterning [59]. The images shown in Figure 12.4 highlight three representative examples of microfluidic devices driven by electrochemical reactions. Figure 12.4a shows an isoelectric focusing device in which the pH gradient necessary for a separation is created by the electrochemical splitting of water. The sidewalls of the microfluidic channel in this case are coated with metals that act as the electrodes of the electrochemical cell. A pH gradient, which provides the basis for this form of chemo-responsive actuation, is created as a result of the difference in the local pH that develops in close proximity to the anode and cathode of the cell [60] and elicit a separation along the transverse cell direction—across the microfluidic channel. Figure 12.4b shows an electrochemically driven microfluidic actuator that operates using gradients in surface tension created by an electrochemical reaction of a redox compound that takes place when an appropriate potential is applied to electrodes at the ends of the microfluidic channels. These “free-energy” gradients are quite powerful and have been shown to cause droplets to move with linear velocities as high as several centimeters per second [42]. Figure 12.4c illustrates yet another interesting application. In this example, microelectrodes printed as an array

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.3 ELECTROCHEMISTRY AND MICROFLUIDICS

301

(a) 1

2

3

4

5

(c)

(b)

2

100 µm

1

3

Figure 12.4. (a) Transverse isoelectric focusing (IEF) of bovine serum albumin (BSA)–Bodipy fluorescent conjugate in a pH gradient developed between two gold electrodes by applying a voltage of 2.3 V. Fluorescent images of the results of IEF of the BSA conjugate with different initial pH values: (1) 3.56, (2) 4.35, (3) 4.86, (4) 5.75, (5) 6.25 [56]. (Reprinted with permission from [56]. Copyright 2001 American Chemical Society.) (b) Time-lapse images of the pumping of liquid crystal (LC) droplets across the surface of an aqueous solution in a simple fluidic network. Platinum electrodes protrude through the surface of the solution at the ends of the left, right, and lower channels (widths ∼4 mm). The end of the top channel contains a saturated calomel electrode (SCE) and Pt counter electrode. (1) The droplet of LC dispensed into the bottom channel of the intersection is pumped into the right channel by application of a potential of

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

302

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

on the surface of a silicon chip generate chemically active species in a solution of electrolyte held between the electrode array and a glass plate. The active species induce chemical change in the molecules (dimethoxytrityl) coupled to the surface of the glass plate, which generates a surface pattern as illustrated by the image shown in Figure 12.4c [59].

12.4 PDMS AND ELECTROCHEMISTRY PDMS is electrically insulating (breakdown voltage ∼2 × 107 V m−1 ) [61] and therefore well suited for the fabrication of devices with embedded electrical circuits. The literature, however, describes very few reports of microfluidic PDMS devices in which electrochemical reactions take place [62–64]. One mitigating factor is the noted ability of this polymer to nonselectively absorb–permeate electroactive gases (e.g., oxygen and carbon dioxide), which for many applications would adversely impact the accuracy of an electrochemical measurement. Our group, nonetheless, has used the exceptional permeation properties of PDMS [65] to our advantage, by exploiting this property to design a number of PDMS-based microelectrochemical reactors, those for which the electroactive species are supplied to the electrode(s) via permeation. These microelectrochemical reactors can produce orders of magnitude higher current densities than conventional electrochemical cells in several diffusion controlled electrochemical reactions (most notably in the oxygen reduction and the hydrogen oxidation reactions) [39, 66, 67]. Figure 12.5 schematically depicts the steps involved in the soft lithographic fabrication of one such microelectrochemical device. The point of importance here is the fact that the working electrode (WE) is embedded in a PDMS-based microchannel network and is fully immersed in the liquid electrolyte confined in the channels. A Pt-wire counter electrode (CE) and an Ag-wire reference electrode (RE) are added to the assembly to yield the miniaturized electrochemical cell illustrated ←--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Figure 12.4. −0.3 V to the bottom Pt electrode and +0.3 V to the right Pt electrode. (2) The droplet of LC in the right channel of the intersection is pumped into the left channel by application of a potential of −0.3 V to the right Pt electrode and +0.3 V to the left Pt electrode. (3) A droplet of LC dispensed into the bottom channel of the intersection is pumped into the left channel by application of a potential of −0.3 V to the bottom Pt electrode and +0.3 V to the left Pt electrode [42]. (From [42]. Reprinted with permission from AAAS.) (c ) Electrochemical oxidation of hydroquinone at the anodes (+) on an array of microelectrodes delivers acid to localized regions on a surface. The counter electrode process at the adjacent cathodes (−) is the reduction of benzoquinone, which yields a radical anion reactive with protons, thus acting to deplete the acid in the cathode region and regenerate hydroquinone. The figure shows a fluorescent image of a surface after patterning with electrochemically generated acid. Fluorescence (white areas) occurs where the surface was not changed by the acid generated at the anodes whereas the dark areas represent places where an ester bond was formed after removal of the dimethoxytrityl from the surface. The active patterning reagents are generated at the anodes (+) and confined by the cathodes (−). The figure shows the ability of the cathodes to confine the acid [59]. (Reprinted with permission from [59]. Copyright 2002 American Chemical Society.)

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.4 PDMS AND ELECTROCHEMISTRY

(a)

IDA PDMS

Silanize 20 mm

3 mm

14 mm

(b)

Pour PDMS and cure

Flip over and peel-off

5 mm

O2

(c)

WE CE RE

(d)

20

Argon Air Oxygen

Current density/mA cm–2

0 –20 –40 –60 –80 –100 –120 0

50

100

150

Time / s

200

250

300

303

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

304

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

schematically in Figure 12.5c. The current densities that can be obtained in the oxygen reduction reaction (ORR) using these cells are extraordinarily high. As seen from the chronoamperometric plots in Figure 12.5d, these values exceed those obtained from conventional electrochemical cells by two orders of magnitude. There are two important consequences of these surprisingly large ORR currents: (i) the pH in the vicinity of the WE is quite high; and (ii) the bulk fluid in the microchannels can be set into motion in a direction opposite of the electro-osmotic flow normally seen in glass-based devices. Fluorescent micrographs taken during device operation (shown in Figure 12.6a and b) illustrate this point more explicitly. One clearly notes from these data that the microfluidic cell exhibits a laminar flow profile running from the direction of the WE-cathode toward the CE-anode. This unique fluidic actuation mechanism has been discussed recently in the literature [39]. The most interesting feature to note here is that the Faradaic currents sustained in this cell can produce marked displacements of fluid with linear velocities as high as 50 µm s−1 . A second feature of the microreactor is that it produces a considerable quantity of hydroxide ion during the ORR, which allows it to function as a chemical actuator. The scaling of the current densities with the distance of the electrode segments from the CE leads very naturally to the generation of a pH gradient along the intersecting channel, as evidenced from the correlated fluorescence intensity profile obtained by adding the indicating dye fluorescein to the electrolyte (Figure 12.6c). This pH gradient is stable and long lived even when the device is turned off and can attain values as high as 4 pH units over distances as large as ∼1 cm (projecting in the direction toward the CE). This pH gradient generating system obviates significant disadvantages encountered with other electrochemical means for generating pH gradients in microfluidic channels. First, the problems commonly encountered with the adsorption of analytes at the electrodes are avoided (the gradient is generated outside the area of the WE array). Second, the continuous fluid flow toward the CE enables a simultaneous elution of the species separated in the gradient. Third, the ORR is gas consuming and does not generate the bubbles that block fluid motion, as is encountered commonly in water electrolysis designs [56, 68]. ←--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Figure 12.5. (a) Top view of a Pt-interdigitated array electrode (Pt-IDA) used as a working electrode (WE) in a microfluidic electrochemical cell. (b) Fabrication of the miniaturized electrochemical cell. The WE is first pressed onto a PDMS flat, Pt side down. The PDMS-electrode assembly is silanized by exposing to tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane. PDMS prepolymer is poured over the assembly and cured at 70◦ C for 2 h, leaving the electrode sealed between the two PDMS layers. The “sandwich” is flipped over and mounted on a glass microscope slide, Pt side up. The PDMS layer covering the Pt is then peeled off, leaving the IDA embedded in PDMS. (c ) A 3D view of an assembled microfluidic electrochemical cell using a Pt counter (CE) and an Ag reference electrode (RE). (d ) Chronoamperometric plots obtained with the miniaturized electrochemical cell in an argon (black, top curve), air (black dashed line, middle curve), and oxygen (gray, bottom curve) atmosphere. The data were recorded at −0.3 V versus Ag in a 5 M H2 SO4 electrolyte solution. The microfluidic network comprises channels of width × separation × height = (10 × 10 × 2) µm.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.4 PDMS AND ELECTROCHEMISTRY

305

(a)

WE

O2 + 2 H2O + 4 e− = 4 OH−

1 mm

CE

RE

(b)

WE

flow

1 mm

CE

(c)

RE

180

Green intensity

160 140 120

**

100

*

80

After 7 min at −0.2 V After 10 min at −0.2 V After 10 min at OCP

60 40 0

2000

4000

6000

Distance from top of channel (µm) Figure 12.6. (a) Fluorescence micrograph obtained during operation of the device in Figure 12.5c . The device had been cathodically polarized at 1 mV s−1 from open circuit potential (OCP) to −0.6 V and back to −0.2 V at which it was held for 10 min. The image shown was taken immediately before the device was shut off. (b) Fluorescence micrograph taken 10 min after

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

306

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

The fabrication procedure shown in Figure 12.5b is quite general and can be applied to embed two or more electrodes in the channels. A remarkable example of a functional device that consists of two electrodes embedded in a PDMS-based microchannel network is provided by the microfluidic fuel cell illustrated schematically in Figure 12.7a—a passive design that operates with high efficiencies using the ORR in conjunction with the anodic conversion of a suitable fuel. The photograph shown in Figure 12.7b illustrates the manner in which the fuels are introduced to the electrodes by permeation. Here, the two different reagents are localized as point sources and supplied to the electrodes via direct permeation through the PDMS channel walls: oxygen to the cathode; and hydrogen, methanol, or formic acid to the anode. Typical polarization characteristics that can be obtained with these devices are shown in Figure 12.7c. These data reveal that the polarization characteristics are dependent on the mass transport of reactant at the anode. In a recent publication [66], it was shown that cells with appropriate design rules can operate stably for exceptionally long periods (>100 days in a laboratory ambient) without the requirement for a separator membrane between the anodic and the cathodic compartments. The transport enabled by the localization of permeant fluxes—in conjunction with specific alloy designs for the electrocatalysts—can fully obviate the most serious negative impacts of crossover. This latter device illustrates the empowering advantages that soft lithography brings to bear in studies concerned with the use and exploitation of microelectrochemical devices.

12.5 OPTICS AND MICROFLUIDICS Optical detection is the most common nondestructive detection method used in microfluidic systems. In these applications, the alignment of the optical components of the system is critical since it is necessary to properly focus the light onto the small volume being examined. One useful way to achieve such alignment is to directly integrate the optical devices with the microfluidic components, a method that increases the flexibility and performance of detection. A variety of optical elements such as organic light emitting diodes [69], optical fibers [70], optical waveguides [71], vertical cavity surface emitting lasers [72], and ←--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Figure 12.6. the cell had been switched off showing fluid flow in the direction of the CE. (c ) Fluorescence intensity profile taken along the single intersecting channel at the left side of parts a and b. The analysis was performed along a line placed in the center of the channel from its top, along the eight T-intersections and further toward the CE reservoir (entire length is ∼6.5 mm from the top of the channel). The intensity profiles were taken at 7 and 10 min during the potential hold at −0.2 V, and 10 min after the cell had been switched off (OCP). Solution: 0.1 M KCl + 0.1 mM fluorescein, initial pH = 4. Channel dimensions: width × separation × height = 200 µm × 200 µm × 50 µm. Magnification: 2×. Exposure time: 400 ms. (∗ ) The “spike” in the curves is caused by an imperfection in the intersecting channel. (∗∗ ) The area of the local dip corresponds to the T-intersection region where the intersecting channel meets the first horizontal channel.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.5 OPTICS AND MICROFLUIDICS

(a)

C

307

A

(b)

(c) 1000 H –O 2 2 HCOOH – O 2 CH OH – O 3 2

Cell voltage/mV

800 600 400 200 0 0.0

0.5

1.0

1.5

2.0

2.5

Current density/mA cm–2 Figure 12.7. (a) Schematic illustration of a PDMS-based microfluidic fuel cell with two embedded electrodes. (b) A photograph of a PDMS-fabricated microfluidic fuel cell during its operation when liquid fuels are used (HCOOH or CH3 OH). The arrows depict the location of the fuel reservoir and the oxygen gas inlet. (c ) Polarization curves of a microfluidic fuel cell using hydrogen (circles, top curve), formic acid (squares, middle curve), and methanol (triangles, bottom curve) fuel measured in 1 M NaOH in the direction of increasing (full symbols) and decreasing (clear symbols) current densities. The figure shows that the polarization characteristics are dependent on the mass transport of reactant at the anode.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

308

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

microlenses [73] have been integrated with microfluidic systems. The general principle that underlies the operation of these devices is that the fluids interacting with the light induce changes in its properties (e.g., in its intensity and/or phase) through changes in the absorbance [74], fluorescence [75], or refractive index [70] of the fluid. Sensing is usually carried out by measurements of reflection [76], transmission [75], or evanescent wave coupling [70]. Our group has developed microfluidic devices integrated with optical waveguides [77] for sensing fluids. These waveguides are fabricated using conventional soft lithographic methods from PDMS and high refractive index-PDMS (Hi-PDMS), which have different refractive indexes (1.4 [78, 79] and 1.46 [80], respectively). Each waveguide is 40 µm wide and 20 µm deep and is therefore, a multimode waveguide. Figure 12.8a gives a schematic of the fabrication procedure for an array of such waveguides. It should be noted that the top surface of the stamp carrying the waveguides is nearly flat, which enables complex 3D integrated structures through easy stacking of subsequent PDMS layers, such as the one shown in Figure 12.8b. The device in Figure 12.8b consists of three layers: the bottom layer with the optical waveguides, the middle layer with the fluidics network, and the top layer with another network of channels to control fluids in the middle fluidic layer (Figure 12.8b (1)). The core of the waveguide is exposed to the microfluidic channel and interacts with the fluid passing over it. Figure 12.8b (2) shows an optical image of the latter device. The PDMS surfaces are irreversibly bonded together by pretreating them with UV radiation and ozone (UVO) [82, 83]. The elasticity of the PDMS (elastic modulus 1.5–2 MPa) [84] makes the integrated device more flexible and less prone to damage as compared to devices made from inorganic semiconductor materials. The optical transparency of the PDMS and Hi-PDMS further prevents loss of guided light by absorption in the waveguide materials, which enables more efficient sensing. Shown in Figure 12.8c is a He-Ne laser (632.8 nm) butt coupled into a single waveguide in an array of waveguides with a 90◦ bent shape and a large bend radius (∼4 mm). The higher index of the Hi-PDMS core compared to the PDMS clad allows the light to be guided around the 90◦ bend. The background light, which has not been coupled, scatters out at the bend. The large bend radius reduces excessive scattering of light by reflections at the bend, preventing loss in sensing efficiency. The device shown in Figure 12.8 can be used for optical sensing of fluids flowing through the microfluidic channels using the optical setup shown in Figure 12.9a. Figure 12.9b is an optical image of the plug flow [85, 86] used to periodically vary the optical properties of the fluid flowing over the waveguides. In this experiment, two fluids were periodically passed over the waveguides: (i) water mixed with blue gel paste food dye and (ii) an immiscible carrier fluid (FC-40). Stable plug flow was formed and controlled by adjusting the capillary number and water fraction [86]. Figure 12.9c depicts the three levels of light intensity corresponding to the three different fluids that sequentially came in contact with the waveguide (i.e., air, FC-40, and aqueous dye). The air plugs in FC-40 form in the channel during the start up of the experiment from the initially air-filled channels. The device shown in Figure 12.9 operates using the principles of evanescent field sensing. The evanescent field is the

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.5 OPTICS AND MICROFLUIDICS

309

(a) Hi-PDMS

(1)

Silanized oxide wafer

(2)

Vacuum outlet (3) nhi-indus-1.46

nPDMS-1.4 (b) (1)

(2) Control layer Flow layer

Embedded wavoguide layer

600 µm

(e)

Figure 12.8. (a) Fabrication procedure for diffused waveguides. First, a PDMS stamp with a parallel array of microfluidic channels in the shape of the final waveguide is replicated using soft lithography. The channels are ∼40 µm wide and ∼20 µm deep and undergo a 90◦ bend with a bend radius of ∼4 mm. The microfluidic channels are filled with a high refractive index PDMS (Hi-PDMS) [80] using vacuum-assisted micromolding in capillaries (MIMIC) [81] (steps (1) and (2)) and cured at 80◦ C for 12 h. Hi-PDMS has a refractive index of ∼1.46 [80] as compared to that of PDMS which is ∼1.4 [78, 79]. After curing, the composite stamp is peeled from the silicon oxide substrate to give an array of diffused waveguides with Hi-PDMS cores and a PDMS clad (step (3)). (b) Schematic of the procedure for integrating multiple layers on top of the diffused waveguides (1) and an optical image of an integrated device (2). The layers are irreversibly sealed by exposing the surfaces to UVO and bringing them in conformal contact and curing at high temperatures. The inset gives a close up view of two layers of channels (flow layer and control layer) integrated on top of the layer with the diffused waveguides. (c) Top view and edge view of light coupled into one waveguide of the array. A He–Ne laser is used as a light source and the light is butt coupled into one waveguide in the array using a microscope objective (40×) to reduce the spot size of the light. An edge view of the light coming out of the waveguide captured using a black and white CCD camera.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

310

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

(a) Chopper Mirror Microscope objective

He-Ne laser

Mirror Reference signal

Silicon detector

Fluidics supply Device Microscope objective Lock-inamplifier

(b)

FC-40

Computer (labview program)

Dye water

(c) 7.00E-05

Intensity (V)

6.60E-05

Air FC-40

6.20E-05 Dye water

5.80E-05 5.40E-05 5.00E-05 160

200

240

280

300

Time (s) Figure 12.9. (a) Schematic of the optical setup used for measuring the intensity of light exiting the waveguide device in Figure 12.8. Light from a He-Ne laser is directed into a microscope objective (40×), which reduces the spot size of the light. The light coming out of the objective is butt coupled into one of the waveguides in the array. Liquids are supplied to the microfluidic channels using two syringe pumps. The light coming out of the waveguide is collected by another microscope objective and directed onto a large area silicon detector. The detector converts the light signal into an electrical signal, which is collected by a lock-in-amplifier. The lock-in-amplifier receives a reference signal from the chopper placed in the path of the laser beam before it enters the waveguide device. (b) An optical image of the plug flow used to periodically vary the fluids in the microfluidic channel that come into contact with the waveguide core. The plugs can be formed by controlling the flow rates of two fluids merging into the single serpentine channel. The fluids used are FC-40, a fully fluorinated and hydrophobic fluid, and water mixed with blue dye. (c ) A graph of the evanescent sensing results obtained by varying the optical properties of the

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.5 OPTICS AND MICROFLUIDICS

311

light that penetrates into the optically less dense clad of the waveguide. The fields of the different light modes interact with the different plugs of fluids (or changing clad) in the channel, which, in turn, affect the intensity of light traveling in the waveguide core [87]. The integration of optical elements with microfluidics is not limited to sensing. Microfluidic devices have recently found applications as tools in optical communication systems [88–92]. One such application is in tunable optical fiber grating devices [88, 93, 94]. These devices are used for dynamic chromatic dispersion compensation, programmable adding and dropping of wavelength channels, and dynamic gain equalization [93]. The concept behind such microfluidic fiber (µFF) devices is to use pumped fluids to dynamically tune the optical transmission characteristics of an optical fiber. It consists of two layers, a bottom layer supporting fluidics and a top layer supporting electrodes for electrowetting pumps. Figure 12.10a and b give the schematic of the steps involved in the fabrication of the bottom and top layers of the device, respectively. A representative example of a digitally tuned µFF device is shown in Figure 12.11. Figure 12.11a gives a schematic of the cross-sectional view of the device and of the angle views of the top and bottom substrates. The top substrate electrodes act as electrowetting pumps [95] to manipulate fluid droplets. The angle view of the top substrate shows a common ground electrode and a set of eight electrodes on either side of it corresponding to each of the eight microfluidic recirculating channels. An optical fiber is placed in a slot running perpendicular to the channels, as shown in the angle view of the bottom substrate. Conducting fluid plugs of an aqueous solution of Na2 Cr2 O7 .2H2 O (58% by weight) are inserted into the wide portion of the channels using micropipettes. The rest of the channel is backfilled with a lubricating, low viscosity, and low surface energy silicone fluid (DMSS-T00; Gelest) to lubricate the flow of the conducting plugs. By applying suitable voltages to the various electrodes, it is possible to control the position of each of the eight conducting plugs. Figure 12.11b shows optical images of the various digital states achieved by driving a certain number of conducting fluid plugs into contact with the fiber grating. It is important to note that as long as the voltages applied to various electrodes are kept constant, each digital state is exceptionally stable, requiring no active control once it has been achieved. The latter fluidic system can be used for tuning the resonance feature of a long period fiber grating (LPG) [96]. An LPG couples light between copropagating modes and exhibits high sensitivity toward the refractive index of the material ←--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Figure 12.9. fluids coming in contact with the core of the waveguide. The dashed lines indicate the measured intensity levels corresponding to the different fluid plugs of air, FC-40, and blue dye water as each one passes over the waveguide. Air bubbles enter during the start of the experiment where air in the microfluidic channels is driven out as bubbles in FC-40. The air bubbles can be observed as periodic increases in the intensity of light exiting the waveguide. The intensity of light decreases when aqueous plugs of blue dye come in contact with the waveguide due to absorption of the evanescent waves.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

312

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

(a) (1)

Resist Silicon

(2) PDMS

(3)

UV (4) DER Glass (5)

Cytop

(b) (1)

ITO Glass

(2)

(3)

SiNx

(4)

Cytop

(5)

Figure 12.10. (a) Schematic of the fabrication procedure used for making the bottom substrate of the microfluidic fiber (µFF) device. In steps (1), (2), and (3) a PDMS stamp is prepared by soft lithography. In step (4), an 800–1000-µm-thick layer of photocurable epoxy (DER; Dow Chemical) is cast on a glass slide and the PDMS stamp is brought in contact with the epoxy. The epoxy flows to conform to the shape of the microfluidic network on the PDMS stamp surface and is cured with ultraviolet light, after which the stamp is peeled away. In step (5), the epoxy is dip coated with a fluoropolymer solution (Cytop; Asahi Glass) to make the surface of the microfluidic channels hydrophobic, which avoids pinning of fluids to the channel walls as fluids flow through the channels. (b) Schematic of the procedure used for fabricating the top substrate of a µFF device. An indium tin oxide (ITO) coated 100-nm-thick glass slide (1) is modified using photolithography and etching to pattern electrodes on the glass slide (2). A uniform 1-µm-thick film of SiNx dielectric is grown on the patterned electrodes by plasma enhanced chemical vapor deposition. The dielectric is subsequently removed from the common ground electrode using patterning and reactive ion etching. A 1-µm-thick layer of Cytop is then spin cast in step (4) and mechanically scratched in step (5) to ensure good electrical contact with the common ground electrode [93]. (Reprinted with permission from [93]. © IEEE.)

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.5 OPTICS AND MICROFLUIDICS

(a)

V1

SiNx

V2

Conducting fluid

Lubricating Fiber fluid Cytop

Glass Bottom substrate

313

Channels

Top substrate

Fiber

−4 mm

Electrodes

−1 mm

(b)

− 4 mm

Fiber

− 2 mm

Transmission (dB)

(c)

0 7.8 6 5 4

−4 −8

−12

Fiber channel completely filled

3 Increasing overlap

2 1

−16

0 1500

1520

1540

1560

Wavelength (nm) Figure 12.11. (a) Side view of the integrated microfluidic fiber device and angled views of the top and bottom substrates. The top substrate supports the electrodes. Each channel has electrodes at three different voltages, V1 , ground, and V2 . A via hole connects the fluid in the channels to ground voltage. The voltage difference between V1 and V2 is used to drive the conductive fluid around or away from the optical fiber resulting from the electrowetting phenomenon. The bottom substrate supports the channels that define the microfluidic network. Each conducting fluid plug

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

314

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

surrounding the fiber. A uniform 1.5-cm long LPG with a period of 362 µm (O/E Land, Inc.), which leads to coupling between the fundamental mode LP0, 1 and the LP0, 7 mode, is used here. Figure 12.11c shows the transmission spectra of the LPG as different numbers of conducting plugs overlap it. The resonance coupling monotonically decreases due to increase in the loss of the cladding mode, frustrating resonance coupling between the two modes. The speed of switching between the different resonance states depends on the speed of moving the conducting fluid plugs into and out of contact with the LPG, which is fast because the length the plugs need to move is comparable to the diameter of the fiber. The results are fully reversible, i.e., once the plugs are removed the original spectra are regained. The µFF device described above offers significant advantages over its earlier versions in that it allows for faster speed of switching as well as greater flexibility by digital control or tunability of the eight fluid plugs. Such an approach can be useful for adding tunability to other types of fiber structures as well as planar waveguides, photonic crystals, and other components.

12.6 UNCONVENTIONAL SOFT LITHOGRAPHIC FABRICATION OF OPTICAL SENSORS Soft lithographic patterning techniques such as replica molding [11, 97] and soft nanoimprint lithography (soft-NIL) [24, 27, 97]. [24, 25] provide versatile and unconventional methods for forming optically active structures and devices. These techniques have recently been used to fabricate noble metal nanostructures that support surface plasmons for label-free refractive index sensing and surface-enhanced spectroscopies [25–27, 97]. Surface plasmon polaritons (SPPs) are collective oscillations of the conduction electrons on a metal surface that have an associated electromagnetic field that decays exponentially from the metal-dielectric interface. These plasmons can be excited by electromagnetic radiation using grating or prism couplers [98]. Changes in refractive index above the metal surface shifts the plasmon resonance condition, which can be detected as wavelength, intensity, or angle shifts [99]. This obviates the need for labels or tags, such as fluorophores, to detect surface binding events. ←--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Figure 12.11. is placed in a recirculating channel to avoid back pressure. The bottom substrate also supports an optical fiber that runs across the fluidic channels holding the conducting fluid plugs. (b) Some top views of various digital states that the device can be in by moving the conductive plugs, which appear as dark ovals, in contact or out of contact with the optical fiber. The position of the fiber has been shown with a dotted line. The transparent plastic that forms the recirculating microchannels is also visible. (c ) Transmission spectrum of a long period fiber grating placed across the fluid channels with the conductive and index matched fluid plugs. The numbers to the right indicate the number of plugs in overlap with the fiber grating. The plugs are made to overlap the fiber grating sequentially starting with the plug at the edge of the fiber grating [93]. (Reprinted with permission from [93]. © IEEE.)

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

12.6 UNCONVENTIONAL SOFT LITHOGRAPHIC FABRICATION OF OPTICAL SENSORS

315

(a)

1

2

3

4

(b)

Figure 12.12. (a) Plasmonic crystal fabrication procedure: (1) imprint, (2) UV-cure, (3) PDMS removal, and (4) gold deposition [25]. (Reprinted with permission from [25]. Copyright 2005 Optics Express.) (b) Optical and SEM images of a patterned plasmonic crystal fabricated by a form of soft nanoimprint lithography [25, 27]. (Reprinted with permission from [27]. Copyright 2006 National Academy of Sciences, USA.)

Surface plasmons can be excited on nanohole arrays via grating coupling [99, 100]. Large area, spatially coherent arrays of highly uniform nanoholes can be expensive and time consuming to fabricate using conventional lithographic methods. Our group has pioneered work using a soft lithographic imprint technique for rapidly and inexpensively forming ordered plasmonic crystals over large areas (Figure 12.12) [25, 27]. A composite hard PDMS–soft PDMS stamp presenting a square array of cylindrical posts is formed by replica molding against a master of patterned photoresist on a silicon wafer prepared by traditional photolithography. The composite PDMS stamp is used to emboss a photocurable prepolymer cast on a glass slide (Figure 12.12a (1)), which is subsequently cured by exposure to UV light (Figure 12.12a (2)). After curing, the composite stamp is removed to yield a polyurethane surface presenting an array of nanoscale holes, a polymeric replica of the master (Figure 12.12a (3)). A thin, uniform layer of Au is then deposited on the raised and recessed features of the polyurethane replica via electron-beam evaporation, yielding a quasi-3D plasmonic crystal consisting of an array of nanoscale holes (top surface) that is physically separated from a second level of gold nanodisks at the bottom of the embossed wells (Figure 12.12a (4)). Figure 12.12b shows optical and SEM images of a patterned crystal. This simple, robust, and inexpensive procedure allows the preparation of high quality, large area crystals for devices with uniform optical responses. Our results suggest that these systems are promising platforms for performing multiplexed imaging assays [26, 27]. The normal incidence transmission spectrum of these structures is complex and exhibits multiple features that are thought to arise from combinations of localized surface plasmon resonances (LSPRs), Bloch wave surface plasmon polaritons (BWSPPs), and Wood’s anomalies [27, 101]. The involvement of LSPRs, BW-SPPs, and Wood’s anomalies in the optical response of the periodic arrays suggests that the

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

316

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

(a)

z y x

Remove PDMS

(b)

710 0.5

∇

720 730

0

740 −0.5

750 (c)

760

8 4 0 2000

25

2200

10

Dist.

20 15 10 0

5

0

Thickness (nm)

Thick

Int. Resp. (∆%Tnm)

%T

Wavelength (nm)

700

0 0

500 1000 1500 Distance (µm)

2000

Figure 12.13. One dimensional spatial imaging of fibrinogen nonspecifically adsorbed to the surface of a plasmonic crystal. (a) A schematic illustrating the use of a multichannel PDMS microfluidic device to pattern the surface of a crystal with a nonspecifically adsorbed layer of fibrinogen. (b) Spectroscopic difference image of fibrinogen lines patterned on a crystal. (c ) Spatially resolved integrated response and corresponding effective thickness showing binding events in the geometry of the microfluidic channels. (Inset) A measured step edge between a fibrinogen line and bare area of the crystal (circular symbols) and a fitted step edge (curve fit to the symbols) with a Gaussian width of 20 µm [27]. (Reprinted with permission from [27]. Copyright 2006 National Academy of Sciences, USA.)

transmission features should be sensitive to changes in the adjacent dielectric environment. Our group has conducted a variety of sensitivity analyses and proof-ofconcept experiments using these plasmonic crystals for biosensing [25–27]. A full angle-dependent study of the analytical sensitivity of the crystals to surface binding events was performed using the well-studied model system of

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

ACKNOWLEDGMENTS

317

alkanethiolates on Au [25]. This work showed that these crystals are capable of single monolayer detection and that the sensitivity reaches a maximum near regions of the plasmonic Brillouin zone where the dispersion curves of multiple SPP modes converge. In particular, high sensitivity to surface binding events was observed at the
-point (0◦ from normal) where 5 substrate–metal and 1 air–metal SPRs converged, demonstrating that a simple normal incidence transmission can be used in analytical sensing applications. The multiple plasmonic resonances can be exploited for sensing applications by monitoring peak shifts and transmission changes across multiple wavelengths. This was demonstrated in recent work that showed that normal incidence transmissionbased multiwavelength spectroscopy and one-dimensional (1D) imaging could be used to detect molecular binding events at the surface of a plasmonic crystal [27]. The complex optical response resulting from changes in refractive index at the surface of the crystal was captured using a single metric by integrating the change in transmission measured over multiple wavelengths. This integrated response was used to perform quantitative analysis of the biomolecular binding events in real time with submonolayer sensitivity [27]. Spectroscopic 1D imaging was also performed on a plasmonic crystal that was patterned with fibrinogen using a multichannel, microfluidic PDMS device (Figure 12.13a and b) [27]. The integrated response of the 1D image was used in conjunction with a truncated bulk refractive index sensitivity to calculate a protein thickness that correlated well with literature values (Figure 12.13c). In another experiment, the plasmonic crystal was also imaged in a simple collinear transmission configuration using a near infrared camera to generate two-dimensional (2D) images of the fibrinogen patterns [26]. This setup demonstrated that both single wavelength and white light measurements could be used to image binding events over large areas with high sensitivities using these spatially uniform crystals. The examples given above illustrate the effectiveness of soft lithography for rapidly producing plasmonic crystals at low cost to yield highly sensitive compact form-factor sensors requiring only simple optics. This work is evolving toward disposable, integrated microfluidic devices and sensors for performing diagnostic assays.

ACKNOWLEDGMENTS Various aspects of this work are supported by the Department of Energy (DEFG0291ER45439 and DOE DE-FG02-05ER46260) and the National Science Foundation (CHE-0402420 and DMI-0355532). Research for this publication was carried out in the Center for Microanalysis of Materials and the Laser and Spectroscopy Facility in the Frederick Seitz Materials Research Laboratory at the University of Illinois at Urbana-Champaign, which is partially supported by the US Department of Energy under grant DEFG02-91-ER45439.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

318

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

REFERENCES 1. Sanders, G. H. W. and Manz, A. (2000) Chip-based microsystems for genomic and proteomic analysis. Trends Anal. Chem. 19, 364–378. 2. Zhang, C. X. and Manz, A. (2003) Trends in microfluidic devices for analytical chemistry. In: Biochips: Technology and Applications, Xing, W.-L. and Cheng, J. eds., Springer, Berlin. 3. Itoh, H., Mukai, S., and Yajima, H. (1994) Interconnection architecture based on beemsteering devices. IEICE Trans. Electron. E77C, 15–22. 4. Pilipovich, V. A., Esman, A. K., Goncharenko, I. A., and Kuleshov, V. K. (2004) Optical data switching in information processing systems. Meas. Tech. 47, 879–883. 5. Groisman, A., Enzelberger, M., and Quake, S. R. (2003) Microfluidic memory and control devices. Science 300, 955–958. 6. Thorsen, T., Maerkl, S. J., and Quake, S. R. (2002) Microfluidic large-scale integration. Science 298, 580–584. 7. Harrison, J. D., Fluri, K., Seiler, K., Fan, H. Z., Effenhauser, C. S., and Manz, A. (1993) Micromachining a miniaturized capillary-electrophoresis-based chemical-analysis system on a chip. Science 261, 895–897. 8. Zhang, X. and Haswell, S. J. (2006) Materials matter in microfluidic devices. MRS Bull. 31, 95–99. 9. Reyes, D. R., Iossifidis, D., Auroux, P. A., and Manz, A. (2002) Micro total analysis systems. 1. Introduction, theory, and technology. Anal. Chem. 74, 2623–2636. 10. Quake, S. R. and Scherer, A. (2000) From micro- to nanofabrication with soft materials. Science 290, 1536–1540. 11. Xia, Y. and Whitesides, G. M. (1998) Soft lithography. Angew. Chem. Int. Ed. 37, 550–575. 12. Biebuyck, H. A., Larsen, N. B., Delamarche, E., and Michel, B. (1997) Lithography beyond light: microcontact printing with monolayer resists. IBM J. Res. Dev. 41, 159–171. 13. Fosser, K. A. and Nuzzo, R. G. (2003) Fabrication of patterned multicomponent protein gradients and gradient arrays using microfluidic depletion. Anal. Chem. 75, 5775– 5782. 14. Whitesides, G. M. and Stroock, A. D. (2001) Flexible methods for microfluidics. Phys. Today 54, 42–49. 15. Zhao, X.-M., Xia, Y., and Whitesides, G. M. (1997) Soft lithographic methods for nanofabrication. J. Mater. Chem. 7, 1069–1074. 16. Eberhardt, A. S., Nyquist, R. M., Parikh, A. N., Zawodzinski, T., and Swanson, B. I. (1999) Defects in microcontact-printed and solution-grown self-assembled monolayers. Langmuir 15, 1595–1598. 17. Jackman, R. J., Wilbur, J. L., and Whitesides, G. M. (1995) Fabrication of submicrometer features on curved substrates by microcontact printing. Science 269, 664–666. 18. Wilbur, J. L., Kumar, A., Kim, E., and Whitesides, G. M. (1994) Microfabrication by microcontact printing of self-assembled monolayers. Adv. Mater. 6, 600–604. 19. Lee, B. H., Cho, Y. H., Lee, H., Lee, K.-D., Kim, S. H., and Sung, M. M. (2007) Highresolution patterning of aluminum thin films with a water-mediated transfer process. Adv. Mater. 19, 1714–1718.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

REFERENCES

319

20. Loo, Y.-L., Lang, D. V., Rogers, J. A., and Hsu, J. W. P. (2003) Electrical contacts to molecular layers by nanotransfer printing. Nano Lett. 3, 913–917. 21. Loo, Y.-L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Additive, nanoscale patterning of metal films with a stamp and a surface chemistry mediated transfer process: applications in plastic electronics. Appl. Phys. Lett. 81, 562–564. 22. Otto, M., Bender, M., Hadam, B., Spangenberg, B., and Kurz, H. (2001) Characterization and application of a UV-based imprint technique. Microelectron. Eng. 57–58, 361–366. 23. Plachetka, U., Bender, M., Fuchs, A., Vratzov, B., Glisner, T., Lindner, F., and Kurz, H. (2004) Wafer scale patterning by soft UV-nanoimprint lithography. Microelectron. Eng. 73–74, 167–171. 24. Hua, F., Sun, Y., Gaur, A., Meitl, M. A., Bilhaut, L., Rotkina, L., Wang, J., Geil, P., et al. (2004) Polymer imprint lithography with molecular-scale resolution. Nano Lett. 4, 2467–2471. 25. Malyarchuk, V., Hua, F., Mack, N. H., Velasquez, V. T., White, J. O., Nuzzo, R. G., and Rogers, J. A. (2005) High performance plasmonic crystal sensor formed by soft nanoimprint lithography. Opt. Express 13, 5669–5675. 26. Malyarchuk, V., Stewart, M. E., Nuzzo, R. G., and Rogers, J. A. (2007) Spatially resolved biosensing with a molded plasmonic crystal. Appl. Phys. Lett. 90, 203113. 27. Stewart, M. E., Mack, N. H., Malyarchuk, V., Soares, J. A. N. T., Lee, T.-W., Gray, S. K., Nuzzo, R. G., and Rogers, J. A. (2006) Quantitative multispectral biosensing and 1D imaging using quasi-3D plasmonic crystals. Proc. Natl. Acad. Sci. 103, 17143–17148. 28. Sia, S. K. and Whitesides, G. M. (2003) Microfluidic devices fabricated in poly(dimethylsiloxane) for biological studies. Electrophoresis 24, 3563–3576. 29. McDonald, J. C. and Whitesides, G. M. (2002) Poly(dimethylsiloxane) as a material for fabricating microfluidic devices. Acc. Chem. Res. 35, 491–499. 30. Childs, W. R., Motala, M. J., Lee, K. J., and Nuzzo, R. G. (2005) Masterless soft lithography: patterning UV/ozone-induced adhesion on poly(dimethylsiloxane) surfaces. Langmuir 21, 10096–10105. 31. Chaudhury, M. K. and Whitesides, G. M. (1991) Direct measurement of interfacial interactions between semispherical lenses and flat sheets of poly(dimethylsiloxane) and their chemical derivatives. Langmuir 7, 1013–1025. 32. Childs, W. R. and Nuzzo, R. G. (2005) Large-area patterning of coinage-metal thin films using decal transfer lithography. Langmuir 21, 195–202. 33. Yasukawa, T., Glidle, A., Nomura, M., and Cooper, J. M. (2005) Fabrication of robust 2-D and 3-D microfluidic networks for lab-on-a-chip bioassays. J. Microelectromech. Syst. 14, 839–846. 34. Wu, H. K., Thalladi, V. R., Whitesides, S., and Whitesides, G. M. (2002) Using hierarchical self-assembly to form three-dimensional lattices of spheres. J. Am. Chem. Soc. 124, 14495–14502. 35. Unger, M. A., Chou, H.-P., Thorsen, T., Scherer, A., and Quake, S. R. (2000) Monolithic microfabricated valves and pumps by multilayer soft lithography. Science 288, 113–116. 36. Squires, T. M. and Quake, S. R. (2005) Microfluidics: fluid physics at the nanoliter scale. Rev. Mod. Phys. 77, 997–1026 37. Stroock, A. D., Dertinger, S. K. W., Whitesides, G. M., and Ajdari, A. (2002) Patterning flows using grooved surfaces. Anal. Chem. 74, 5306–5312.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

320

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

38. Pollack, M. G., Shenderov, A. D., and Fair, R. B. (2002) Electrowetting-based actuation of droplets for integrated microfluidics. Lab Chip 2, 96–101. 39. Mitrovski, S. M. and Nuzzo, R. G. (2005) An electrochemically driven poly(dimethylsiloxane) microfluidic actuator: oxygen sensing and programmable flows and pH gradients. Lab Chip 5, 634–645. 40. Suzuki, H. and Yoneyama, R. (2003) Integrated microfluidic system with electrochemically actuated on-chip pumps and valves. Sensors Actuators B 96, 38–45. 41. Griss, P., Andersson, H., and Stemme, G. (2002) Expandable microspheres for the handling of liquids. Lab Chip 2, 117–120. 42. Gallardo, B. S., Gupta, V. K., Eagerton, F. D., Jong, L. I., Craig, T. V. S., Shah, R. R., and Abbott, N. L. (1999) Electrochemical principles for active control of liquids on submillimeter scales. Science 283, 57–61. 43. Stroock, A. D., Weck, M., Chiu, D. T., Huck, W. T. S., Kenis, P. J. A., Ismagilov, R. F., and Whitesides, G. M. (2000) Patterning electro-osmotic flow with patterned surface charge. Phys. Rev. Lett. 84, 3314–3317. 44. Venkateswar, R., Branch, D. W., and Wheeler, B. C. (2000) An electrophoretic method for microstamping biomolecule gradients. Biomed. Microdevices 2, 255–264. 45. Jeon, N. L., Dertinger, S. K. W., Chiu, D. T., Choi, I. S., Stroock, A. D., and Whitesides, G. M. (2000) Generation of solution and surface gradients using microfluidic systems. Langmuir 16, 8311–8316. 46. Amirpour, M. L., Ghosh, P., Lackowski, W. M., Crooks, R. M., and Pishko, M. V. (2001) Mammalian cell cultures on micropatterned surfaces of weak-acid, polyelectrolyte hyperbranched thin films on gold. Anal. Chem. 73, 1560–1566. 47. Mrksich, M., Dike, L. E., Tien, J., Ingber, D. E., and Whitesides, G. M. (1997) Using microcontact printing to pattern the attachment of mammalian cells to self-assembled monolayers of alkanethiolates on transparent films of gold and silver. Exp. Cell Res. 235, 305–313. 48. Romanova, E. V., Fosser, K. A., Rubakhin, S. S., Nuzzo, R. G., and Sweedler, J. V. (2004) Engineering the morphology and electrophysiological parameters of cultured neurons by microfluidic surface patterning. FASEB J. 18, 1267–1269. 49. McClain, M. A., Culbertson, C. T., Jacobson, S. C., and Ramsey, J. M. (2001) Flow cytometry of Escherichia coli on microfluidic devices. Anal. Chem. 73, 5334– 5338. 50. Caelen, I., Bernard, A., Juncker, D., Michel, B., Heinzelmann, H., and Delamarche, E. (2000) Formation of gradients of proteins on surfaces with microfluidic networks. Langmuir 16, 9125–9130. 51. Dertinger, S. K. W., Chiu, D. T., Jeon, N. L., and Whitesides, G. M. (2001) Generation of gradients having complex shapes using microfluidic networks. Anal. Chem. 73, 1240–1246. 52. Ismagilov, R. F., Rosmarin, D., Kenis, P. J. A., Chiu, D. T., Zhang, W., Stone, H. A., and Whitesides, G. M. (2001) Pressure-driven laminar flow in tangential microchannels: an elastomeric microfluidic switch. Anal. Chem. 73, 4682–4687. 53. Kraus, T., Stutz, R., Balmer, T. E., Schmid, H., Malaquin, L., Spencer, N. D., and Wolf, H. (2005) Printing chemical gradients. Langmuir 21, 7796–7804. 54. Baldwin, R. P., Roussel, T. J., Crain, M. M., Bathlagunda, V., Jackson, D. J., Gullapalli, J., Conklin, J. A., Pai, R., et al. (2002) Fully integrated on-chip electrochemical

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

REFERENCES

55. 56.

57. 58. 59.

60.

61. 62.

63.

64.

65.

66. 67.

68.

69.

70.

71.

321

detection for capillary electrophoresis in a microfabricated device. Anal. Chem. 74, 3690–3697. Woolley, A. T., Lao, K., Glazer, A. N., and Mathies, R. A. (1998) Capillary electrophoresis chips with integrated electrochemical detection. Anal. Chem. 70, 684–688. Macounova, K., Cabrera, C. R., and Yager, P. (2001) Concentration and separation of proteins in microfluidic channels on the basis of transverse IEF. Anal. Chem. 73, 1627–1633. Guenat, O. T., Morf, W. E., Van der Schoot, B. H., de Rooij, N. F. (1998) Universal coulometric nanotitrators with potentiometric detection. Anal. Chim. Acta 361, 261–272. Zhan, W., Seong, G. H., and Crooks, R. M. (2002) Hydrogel-based microreactors as a functional component of microfluidic systems. Anal. Chem. 74, 4647–4652. Egeland, R. D., Marken, F., and Southern, E. M. (2002) An electrochemical redox couple activitated by microelectrodes for confined chemical patterning of surfaces. Anal. Chem. 74, 1590–1596. Macounova, K., Cabrera, C. R., Holl, M. R., and Yager, P. (2000) Generation of natural pH gradients in microfluidic channels for use in isoelectric focusing. Anal. Chem. 72, 3745–3751. http://www.dowcorning.com. Lammertink, R. G. H., Schlautmann, S., Besselink, G. A. J., and Schasfoort, R. B. M. (2004) Recirculation of nanoliter volumes within microfluidic channels. Anal. Chem. 76, 3018–3022. Lee, K. J., Fosser, K. A., and Nuzzo, R. G. (2005) Fabrication of stable metallic patterns embedded in poly(dimethylsiloxane) and model applications in non-planar electronic and lab-on-a-chip device patterning. Adv. Funct. Mater. 15, 557–566. Yamaguchi, A., Jin, P., Tsuchiyama, H., Masuda, T., Sun, K., Matsuo, S., and Hiraoki, M. (2002) Rapid fabrication of electrochemical enzyme sensor chip using polydimethysiloxane microfluidic channel. Anal. Chim. Acta 468, 143–152. Merkel, T. C., Bondar, V. I., Nagai, K., Freeman, B. D., and Pinnau, I. (2000) Gas sorption, diffusion, and permeation in poly(dimethylsiloxane). J. Polym. Sci., Part B: Polymer Physics 38, 415–434. Mitrovski, S. M. and Nuzzo, R. G. (2006) A passive microfluidic hydrogen-air fuel cell with exceptional stability and high performance. Lab Chip 6, 353–361. Mitrovski, S. M., Elliott, L. C. C., and Nuzzo, R. G. (2004) Microfluidic devices for energy conversion: planar integration and performance of a passive, fully immersed H2 -O2 fuel cell. Langmuir 20, 6974–6976. McKnight, T. E., Culbertson, C. T., Jacobson, S. C., and Ramsey, J. M. (2001) Electroosmotically induced hydraulic pumping with integrated electrodes on microfluidic devices. Anal. Chem. 73, 4045–4049. Yao, B., Luo, G., Wang, L., Gao, Y., Lei, G., Ren, K., Chen, L., Wang, Y., et al. (2005) A microfluidic device using a green organic light emitting diode as an integrated excitation source. Lab Chip 5, 1041–1047. Polynkin, P., Polynkin, A., Peyghambarian, N., and Mansuripur, M. (2005) Evanescent field-based optical fiber sensing device for measuring the refractive index of liquids in microfluidic channels. Opt. Lett. 30, 1273–1275. Hofmann, O., Voirin, G., Niedermann, P., and Manz, A. (2002) Three-dimensional microfluidic confinement for efficient sample delivery to biosensor surfaces.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

322

72.

73.

74.

75. 76.

77. 78. 79. 80. 81. 82.

83. 84.

85.

86.

87.

88.

14:4

Printer Name: Yet to Come

SOFT LITHOGRAPHY FOR MEMS AND OPTICAL DEVICES

Application to immunoassays on planar optical waveguides. Anal. Chem. 74, 5243– 5250. Kemme, S. A., Warren, M. E., Sweatt, W. C., Wendt, J. R., Bailey, C. G., Matzke, C. M., Allerman, A. A., Arnold, D. W., et al. (2000) Integrated optical systems for excitation delivery and broadband detection in microfluidic electrochromatography. Proc. SPIE Int. Ed. Opt. Eng. 3952, 375–383. Jang, J. M., Shin, H. J., Hwang, S. W., Yang, E. G., Yoon, D. S., Kim, T. S., and Kang, J. Y. (2005) Miniaturized fluorescence detection system to remove background noise of the incident light using micro mirror and lens. Sensors Actuators B 108, 993–1000. Pandraud, G., Koster, T. M., Gui, C., Dijkstra, M., Van Den Berg, A., and Lambeck, P. V. (2000) Evanescent wave sensing: new features for detection in small volumes. Sensors and Actuators, A 85, 158–162. Lien, V., Berdichevsky, Y., and Lo, Y.-H. (2004) A prealigned process of integrating optical waveguides with microfluidic devices. IEEE Photonics Technol. Lett. 16, 1525. Verpoorte, E., Manz, A., Ludi, H., Bruno, A. E., Maystre, F., Krattiger, B., Widmer, H. M., Van Der Schoot, B. H., et al. (1992) A silicon flow cell for optical detection in miniaturized total chemical analysis systems. Sensors Actuators B 6, 66–70. Vukovic, A. (1999) Fourier Transformation Analysis of Optoelectronic Components and Circuits, The University of Nottingham, Nottingham. Vezenov, D. V., Mayers, B. T., Wolfe, D. B., and Whitesides, G. M. (2005) Integrated fluorescent light source for optofluidic applications. Appl. Phys. Lett. 86, 041104. Whang, C. M., Yeo, C. S., and Kim, Y. H. (2001) Preparation and characterization of solgel derived SiO2 -TiO2 -PDMS composite films. Bull. Korean Chem. Soc. 22, 1366–1370. Shim, A. (2005) Higher Refractive Index PDMS by Dow Corning, Rogers, J. A., ed. (personal communication). Jeon, N. L., Choi, I. S., Xu, B., and Whitesides, G. M. (1999) Large-area patterning by vacuum-assisted micromolding. Adv. Mater. 11, 946–950. Genzer, J., Fischer, D. A., and Efimenko, K. (2003) Fabricating two-dimensional molecular gradients via asymmetric deformation of uniformly-coated elastomer sheets. Adv. Mater. 15, 1545–1547. Hillborgl, H. and Gedde, U. W. (1999) Hydrophobicity changes in silicone rubbers. IEEE Trans. Dielectr. Electr. Insul. 6, 703–717. Stafford, C. M., Guo, S., Harrison, C., and Chiang, M. Y. M. (2005) Combinatorial and high-throughput measurements of the modulus of thin polymer films. Rev. Sci. Instrum. 76, 062207. Tice, J. D., Song, H., Lyon, A. D., and Ismagilov, R. F. (2003) Formation of droplets and mixing in multiphase microfluidics at low values of the reynolds and the capillary numbers. Langmuir 19, 9127–9133. Zheng, B., Tice, J. D., and Ismagilov, R. F. (2004) Formation of droplets of alternating composition in microfluidic channels and applications to indexing of concentrations in droplet-based assays. Anal. Chem. 76, 4977–4982. Valentino, J. P., Troian, S. M., and Wagner, S. (2005) Microfluidic detection and analysis by integration of evanescent wave sensing with thermocapillary actuation. MEMS 2005, 18th IEEE International Conference, pp. 730–733. Acharya, B. R., Krupenkin, T., Huang, C. C., and Rogers, J. A. (2003) Tunable optical fiber devices based on broadband long-period gratings and pumped microfluidics. Appl. Phys. Lett. 83, 4912–4914.

P1: OTA/XYZ P2: ABC c12 JWBK252/Lee September 18, 2008

14:4

Printer Name: Yet to Come

REFERENCES

323

89. Chronis, N., Liu, G. L., Jeong, K.-H., and Lee, L. P. (2003) Tunable liquid-filled microlens array integrated with microfluidic network. Opt. Express 11, 2370–2378. 90. Kerbage, C., Sumetsky, M., and Eggleton, B. J. (2002) Polarisation tuning by microfluidic motion in air-silica microstructured optical fibre. Electron. Lett. 38, 1015–1017. 91. Kwon, S. and Lee, L. P. (2001) Focal length control by microfabricated planar electrodes-based liquid lens (µPELL). Transducers ‘01 Eurosensors XV: The 11th International Conference on Solid-State Sensors and Actuators, Munich, Germany. 92. Liao, W., Chen, X., Chen, Y., Pu, S., Xia, Y., and Li, Q. (2005) Tunable optical fiber filters with magnetic fluids. Appl. Phys. Lett. 87, 151122. 93. Cattaneo, F., Baldwin, K., Yang, S., Krupenkine, T., Ramachandran, S., and Rogers, J. A. (2003) Digitally tunable microfluidic optical fiber devices. J. Microelectromech. Syst. 12, 907–912. 94. Hsieh, J., Mach, P., Cattaneo, F., Yang, S., Krupenkine, T., Baldwin, K., and Rogers, J. A. (2003) Tunable microfluidic optical-fiber devices based on electrowetting pumps and plastic microchannels. IEEE Photonics Technol. lett. 15, 81–83. 95. Prins, M. W. J., Welters, W. J. J., and Weekamp, J. W. (2001) Fluid control in multichannel structures by electrocapillary pressure. Science 291, 277–280. 96. Vengsarkar, A. M., Lemaire, P. J., Judkins, J. B., Bhatia, V., Ergodan, T., and Sipe, J. E. (1996) Long period fiber gratings as band rejection filters. J. Lightwave Technol. 14, 58–65. 97. Liu, G. L. and Lee, L. P. (2005) Nanowell surface enhanced Raman scattering arrays fabricated by soft-lithography for label-free biomolecular detections in integrated microfluidics. Appl. Phys. Lett. 87, 074101. 98. Raether, H. (1988) Surface Plasmons on Smooth and Rough Surfaces and on Gratings, Springer, Berlin, Vol. 111. 99. Knoll, W. (1998) Interfaces and thin films as seen by bound electromagnetic waves. Annu. Rev. Phys. Chem. 49, 569–638. 100. Homola, J., Yee, S., and Gauglitz, G. (1999) Surface plasmon resonance sensors. Sensors Actuators B 54, 3–15. 101. Chang, S.-H., Gray, S. K., and Schatz, G. (2005) Surface plasmon generation and light transmission by isolated nanoholes and arrays of nanoholes in thin metal films. Opt. Express 13, 3150–3165.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13 UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS∗ Pilnam Kim, Yanan Du, Ali Khademhosseini, Robert Langer, and Kahp Y. Suh

13.1 INTRODUCTION Unconventional nanopatterning methods are emerging as powerful tools for biological studies and tissue engineering [1–5]. In biological studies, there is a great need to detect or separate small biological species (DNA or proteins) in a rapid and high throughput manner. Most biological events occur at the nanometer scale and thus control of the phenomena at this characteristic length scale could lead to new devices with improved properties such as increased speed and sensitivity. Nanoscale patterns or channels integrated with bio-nanoelectromechanical systems (BioNEMS) enable miniaturization of biomolecular arrays or detection elements, offering a potential tool for screening libraries of small molecules or detection/separation at a single molecule level [6]. In tissue engineering, it is important to control cellular microenvironments in vitro to create the cellular niche during embryonic development. For this purpose, unconventional patterning methods can offer biomimetic nanoscale topographical features on a solid substrate similar to tissue environments, providing a route to manipulate cell functions with desired phenotypic responses. In a typical biological laboratory setup, it is beneficial to create nanoscale patterns or nanochannels without resorting to sophisticated equipments such in ∗ The

first two authors contributed equally to this chapter.

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright 

325

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

326

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

electron-beam, x-ray, and ultraviolet (UV) photolithography. Nanoimprinting (including hot embossing and UV-assisted imprinting) or soft lithographies are low cost and flexible patterning techniques, capable of fabricating polymeric structures for biological applications. Currently, established silicon and glass processes show limitations due to complex fabrication procedures, geometrical design restrictions, and high costs involved. The need for high aspect ratio features for various biological and biomedical applications also posed a serious problem to the existing nanofabrication technologies in silicon and glass. Polymers offer a solution to these challenges and enable mass fabrication of biomolecular arrays or nanofluidic devices. In addition, polymers offer several advantages that are not readily available in silicon and glass, including a wide range of material characteristics, biological compatibility, ease of processing and prototyping, and lower costs. In this chapter, we focus on simple replication technologies for polymers (contact printing, nanoimprint lithography, nanomolding, soft lithography, etc.) that have been used to create nanoscale channels, patterns, or topographies for biological studies and tissue engineering. 13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS Microfluidic devices have been a well-established microscale technology in biological applications for manipulating small quantities of sample in a fast, high resolution, and low cost manner. As devices continue to scale down to the nanometer dimension, there has been a growing interest in nanofluidics, which offers the possibility for single-molecule manipulation [7]. Nanochannel, as the most commonly used configuration for nanofluidic systems, is defined as nanofabricated channels with at least one cross-section dimension on the nanometer scale (one dimensional (1D) and two dimensional (2D) nanochannels) [8]. The nanoscale dimensions of the nanochannel can be comparable to the size of the fluid macromolecules such as proteins and DNA, and a significant amount of the fluid in the nanochannel is in contact with the channel surface leading to significant increase in the effects of surface forces [9]. These unique fluidic properties offer new opportunities to investigate new fundamental phenomena (i.e., fluidic transport and molecular behavior) and develop tools for biomedical applications [10–12]. Here, we introduce several fabrication strategies of nanochannels with the emphasis on unconventional fabrication methods, including nanoimprinting, nanomolding, and soft lithography. We then discuss several exemplified biomedical applications of the nanochannels system including DNA sequencing/stretching, protein separation, and drug delivery. 13.2.1 Unconventional Methods for Fabrication of Nanochannel Conventional microfabrication methods have been used to fabricate nanochannels in substrate (wafer or film) by standard photolithography patterning followed by wet–dry substrate etching. For standard photolithography patterning, a beam of UV

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS

327

light passes through a mask and lens, which can create nanochannel patterns with the resolution of sub-100 nm [13, 14]. Higher resolution of the patterning can be achieved by using extreme UV light [15, 16] to create narrower nanochannels. Also, lithography technologies based on focused beams such as electron-beam lithography (EBL) [7, 17] and focused ion beam (FIB) lithography [18, 19] are attractive alternatives to create highly precise and reliable nanochannel patterns with features without using the mask. Other maskless patterning methods for nanochannels fabrication include interferometric lithography (IL) (based on the interference of two and more coherent beams) [20], laser patterning [21, 22] and surface machining (by etching of the nanometer height sacrificial layer) [23]. Here, several exemplary works of applying unconventional patterning methods are listed for nanochannel fabrication. All these unconventional patterning methods can be used to fabricate nanochannels using functional materials other than photoresist with sub-100-nm resolution and the fabrication processes are fast and low cost without extensive use of “clean rooms” and photolithographic equipments [24].

13.2.1.1 Nanochannel Fabrication by Nanoimprint Lithography (NIL). NIL and NIL-related techniques, such as step-and-flash imprint lithography (S-FIL) can be used as a low cost, high throughput method for the fabrication of nanochannels. Abgrall et al. [25] applied NIL to fabricate planar low aspect ratio (AR) nanochannels with width in micrometer scale and depth under 100 nm. (Figure 13.1a). First, a silicon mold with the impression of nanochannel was fabricated using standard photolithography and reactive ion etching (RIE). Next, a layer of poly(methylmethacrylate) (PMMA) was hot embossed by this silicon mold to create the nanochannels in PMMA. Subsequently, a second layer of PMMA was bonded to the first sheet by thermal bonding for sealing of the nanochannels. They have successfully fabricated arrays of sealed planar nanochannels in PMMA with a depth of 80 nm and low AR ranging from 0.008 to 0.05. Cao et al. [26] made uniform arrays of nanochannels over large areas of ∼100-mm silicon wafer using NIL (Figure 13.1b). The NIL mold was generated by IL. The nanochannels were further narrowed and sealed by techniques that are based on nonuniform deposition (electron-beam deposition and sputter deposition). The resulting sealed channels had a cross-section as small as 10 nm by 50 nm. The same NIL technique has also been applied to build a sealed 100-nm wide, 200-nm deep nanochannel array on fused silica wafers (Figure 13.1c) [27]. Wang et al. [28] present results on fabrication of high density nanochannels in SU8 resist, based on nanoimprinting combined with UV curing (S-FIL) (Figure 13.1d). Silicon template with nanopatterns was fabricated by EBL and RIE. A thick layer of SU-8 was spin coated onto a quartz wafer, imprinted by the template, and exposed to UV light through the quartz substrate for curing. Finally, the template was removed to obtain the SU-8 nanochannel. Due to the low viscosity of the SU-8, large-area patterns with high resolution, high density, high uniformity, and high aspect ratio could be replicated under low pressure and low temperature.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

328

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

(1)

(a)

(c)

(1)

(b)

(d)

Figure 13.1. Nanochannel fabrication by NIL. (a) Planar PMMA nanochannels with a width of 10 µm and a depth of 80 nm filled with fluorescent dye. (Reprinted with permission from [25] and the Royal Society of Chemistry.) (b) Nanochannels in silicon wafer with a trench width of 85 nm. (1) The dark area is the trench. (2) The channels were narrowed down to less than 20 nm by controlled electron-beam deposition. The inset shows the channel width at a higher magnification. The scale bars are all 500 nm. (Reprinted with permission from [26] and the American Institute of Physics.) (c ) The assembly of a sealed 100-nm-wide, 200-nm-deep nanochannel array with a microfabricated coverslip. The nanoimprinted chips were made in fused silica (thickness, 1 mm). (Reprinted with permission from [27] and the National Academy of Sciences.) (d ) High density nanochannels in SU-8 fabricated by S-FIL, the pitch, the height, and the linewidth of nanochannels were 300, 850, and 150 nm, respectively with high aspect ratios of 5–6. (Reprinted with permission from [28] and the Elsevier B.V.)

13.2.1.2 Nanochannel Fabrication by Nanomolding. In principle, nanomolding techniques use a mold in the inverse shape of the desired nanostructures, which is filled with a structural material and then the mold can be etched or removed leaving the desired structure behind [13]. Common nanomolding techniques are based on micromolding in capillaries (MIMIC), solvent-assisted micromolding (SAMIM) [24], and capillary force lithography (CFL). In MIMIC, structures are filled with a thermal or photocurable, low viscosity liquid (i.e., prepolymer or hydrogel) by capillary force. When the stamp is removed, the patterned structure is left on the substrate. In SAMIM, the surface of the mold is wetted with a solvent and pressed against the polymer film at ambient conditions. The solvent softens the polymer and as the solvent evaporates, the polymer conforms to the nanostructure of mold. SAMIM enables molding of polymers that cannot be

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS

329

implemented with S-FIL or NIL. In CFL, a patterned elastomeric mold is placed on a spin coated polymer film (thermoplastic or UV-curable resin), followed either by raising temperature above the polymer’s glass transition temperature (T g ) or by direct molding prior to solvent evaporation, or by exposing UV light to solidity the precursor film. Kim et al. [29] presented a simple, yet robust method for fabricating PEG-based nanochannels by CFL. As illustrated in Figure 13.2a, a UV-curable poly(ethylene glycol) (PEG) polymer such as PEG dimethacrylate (PEG-DMA) or PEG diacrylate (PEG-DA) was drop dispensed on photolithography-fabricated silicon master and a supporting poly(ethylene terephthalate) film was carefully placed on top of the polymer surface to make conformal contact. To cure, the sample was exposed to UV for a few seconds (PEG-DA) to a few tens of seconds (PEG-DMA). A slight physical pressure was applied to make conformal contact. With additional UV exposure, irreversible bonding occurred through photo-induced cross-linking at the interface. This method was used to form channels as small as 50 nm in diameter without using a sophisticated experimental setup. Recently, the same group used the UV-curable PEG-DA as a mold to form reversibly bonded nanocapillaries (with a resolution of 50 nm on a gold or silicon substrate [33]. Lensen et al. [30] built nanochannel patterns in the bulk of the star PEG hydrogel by a UV-based imprinting with a perfluorinated soft mold. As illustrated in Figure 13.2b, a primary (hard) master was fabricated by photo- or electron-beam lithography. This primary master was replicated by the perfluorinated polyether (PFPE) material by means of UV curing of the prepolymer against the mold, resulting in an elastic, secondary master. The elasticity and the low surface energy of this PFPE material was easily peeled off mechanically and then used as a mold to imprint the structure into the UV-curable star PEG material. This method is a fast and simple technique with sub-100-nm resolution in dimension, which can be carried out on the bench top. Lee et al. fabricated nanochannel patterns in poly(4-vinylpyridine) (P4VP) via SAMIM using a blending of two UV-curable materials, Norland Optical Adhesives (NOA) 63 and PEG-DA, in an appropriate ratio. Physical nanopatterning via the SAMIM process was carried out as shown in Figure 13.2c. A layer of P4VP solution, dissolved in ethanol, was spin coated on a Si wafer treated with oxygen plasma and the P4VP film was heated on a hot plate to evaporate the residual ethanol in the coated P4VP film. The composite mold was soaked in ethanol followed by drying with air blowing. The composite mold was brought into conformal contact with the P4VP-coated Si wafer under mild heating to complete the SAMIM [31]. Another type of mold machining for fabricating nanochannels was reported by Ilic et al. [32]. The authors described a fabrication method for forming self-sealing parylene polymer tubes by depositing parylene polymer on the silicon wafer mold etched with nanochannel patterns (Figure 13.2d). The parylene polymer is chemically inert and biocompatible. Tubes were self-sealed as the parylene polymer “pinches off” during the deposition process to leave closed tubes with sub-micrometer lateral dimensions.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

330

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

(a)

UV

D

PEG mold UV

PET film

Si master (1) (b)

PEG channel

PEG

hν

PFPE DMA (2° master)

hν

Acr-star PEG

1 µm

(2)

(3)

Ethanol-soaked NOA 63-PEGDA mold (c) Si wafer

Spin coated P4VP and heated at 80°C

Heating to 65°C

(d)

500 nm

SiO2 Si

Channels etched in Si wafer

Parylene Parylene depositon

Parylene tube formation

Figure 13.2. Nanochannel fabrication by nanomolding. (a) Left: schematic illustration of the experimental procedure for nanochannel fabrication; right: cross-sectional SEM images of various PEG channels with width of 50 nm (Reprinted with permission from [29] and the Royal Society of Chemistry.) (b) Left: schematic representation of the three-step process to structure bulk star PEG material; right: optical micrographs of replicas formed by Acr-star PEG using a PFPE secondary master, scale bar represents 100 µm. Insets (scale bar represents 10 µm). (Reprinted with permission from [30] and the American Chemical Society.) (c ) Left: schematic illustrations for physical patterning of P4VP via SAMIM; right: SEM images of P4VP nanochannel on a Si wafer. (Reprinted with permission from [31] and the American Chemical Society.) (d ) Left: schematic illustration of the self-sealed parylene tube formation (Reprinted with permission from [13] and the Royal Society of Chemistry); right: array of freely suspended parylene tubes. Scale bar corresponds to 5 mm. (Reprinted with permission from [32] and the AVS Science and Technology of Materials, Interfaces, and Processing.)

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS

331

13.2.1.3 Nanochannel Fabrication by Soft Lithography. Soft lithography broadly refers to molding, embossing, and imprinting methods exclusively using an elastomeric mold and/or stamp such as poly(dimethylsiloxane) (PDMS) [24]. PDMS has been widely applied for nanochannel fabrication. Kovarik et al. fabricated PDMS nanochannel with 130 nm deep and 580 nm wide (Figure 13.3a) [9]. The authors used EBL to form nanochannel masters in SU-8. (b) (1)

(a) (1)

(2) (2)

(3)

(4)

(4)

(3)

(5)

(c) Cure, peel off PDMS Place on the substrate 2

Release

• A 40.0 0.0

PDMS patterns

µm 1 2

µm

1

0

0

Figure 13.3. Nanochannel fabrication by soft lithography. (a): (1) SEM image of a cross-channel master fabricated in SU-8, (2) image of the same master taken at an 81◦ tilt, (3) enlarged view of the intersection of the channel mold, (4) image of the cross-intersection on a PDMS replica of the master seen in parts (1)–(3). Replica channels were 130 ± 10 nm deep and had an average width of 580 ± 40 nm. (Reprinted with permission from [9] the American Chemical Society.) (b) Fabrication of structurally stable elastomeric nanochannels. (1) A PDMS slab exposed to oxygen plasma is stretched to generate linear nanoscale cracks. The nanocracks are replicated onto ultraviolet-curable epoxy. (2) PDMS prepolymer is cast against the epoxy mold to generate negative relief patterns of nanochannels. The PDMS substrate is then briefly oxidized and sealed against an oxidized PDMS slab to form an array of enclosed nanochannel. (3)–(5) The mode of transport with different size selectivity can be switched reversibly by changing the magnitude of applied force (3): the larger nanochannels allows both larger and smaller particles to pass through simultaneously; (4) low levels of compressive stress constrict the channel such that only the smaller particles pass through; (5) nanochannels loaded with larger stresses become extremely small, excluding sample particles regardless of their size. (Reprinted with permission from [34] and the Nature Publishing Group.) (c ) Left: schematic illustration of fabricating patterned PDMS by LSL; right: atomic force microscopy images of the nanochannel patterns of PDMS fabricated by LSL in silicon substrate. (Reprinted with permission from [35] and the American Chemical Society.)

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

332

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

The mixture of PDMS prepolymer and curing agent was then poured over the SU-8 master for molding and released to form the nanochannel. A tunable-oxidized PDMS nanochannel was fabricated by Huh et al. [34], which could actively manipulate nanofluidic transport through dynamic modulation of the channel cross-section using remarkably small forces. As illustrated in Figure 13.3b, the PDMS nanochannels could be easily fabricated and provide a remarkably versatile example of an active nanostructure that can change its architecture during operation to create, control, and manipulate various types of nanofluidic transport. Park et al. [35] reported a new soft lithography patterning method, called lightstamping lithography (LSL), that uses UV-induced adhesion of PDMS to fabricate PDMS nanochannels (Figure 13.3c). First, a patterned PDMS stamp was fabricated and brought into contact with a substrate surface. Second, the substrate was exposed through the PDMS stamp to a 254-nm UV lamp for 2 min. The UV light induced the formation of chemical bonds between the PDMS stamp and the substrate underneath. Finally, the PDMS stamp was physically peeled off from the substrate, with remaining torn pieces thereon. In this section, we have listed examples of the use of unconventional methods for nanochannel fabrication. Several points need to be taken into consideration when designing a nanochannel. First, there exists a critical dimension of nanochannels before collapse due to the competition between van der Waals forces and the stiffness of the nanochannel material. The maximal width of a channel with a depth of 100 nm is around 70 nm PDMS and 170 nm for PMMA, indicating that the fabrication of high AR nanostructures in soft materials is not achievable [25]. Second, the bonding of the nanochannels is challenging due to the low AR. Great care should be taken when choosing the bonding methods, such as adhesive bonding, microwave bonding, solvent-assisted bonding, oxygen plasma-assisted bonding, and thermal bonding. 13.2.2 Application of Nanofluidic System Nanochannels, having dimensions comparable to the size of biological macromolecules such as proteins and DNA, have attracted attention in biological applications. New phenomena arise with transition from microfluidics to nanofluidics as the typical dimension is reduced to below 100 nm. As the volume of fluid inside the channel becomes smaller, the quantity of surface charge becomes comparable to the quantity of bulk charge leading to co-ion exclusion and a surface-charge-governed regime of ion transport [25]. The confinement of biomolecules such as DNA or proteins in these structures allows applications, such as single-molecule manipulation/detection, bio-separation, and controlled drug delivery.

13.2.2.1 DNA Stretching and Sequencing. The ability to stretch single DNA molecules can be used for many biological applications, i.e., in DNA sequencing and in the study of DNA–protein interaction. In shotgun DNA sequencing, the location of landmark restriction sites on chromosomal length DNA molecules is a powerful method to ensure the faithful representation of the assembled DNA sequences to the native genome [36]. The restriction sites can be determined by gel electrophoresis

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS

333

[37] or alternatively by optical mapping of the stretched DNA molecules [27]. DNA stretching is imperative to guarantee one-to-one mapping between the spatial position along the DNA molecules and position within the genome by using optical techniques directly [38]. Studies of DNA–protein interaction at the single molecule level are increasingly used to quantify distributions of molecular mechanical properties, transient intermediates, and reaction pathways [39], which require stretching and immobilization of DNA molecules to suppress Brownian motion, so that protein motion along the DNA contour can be followed. Stretching of DNA trapped on a surface has been realized by several strategies, for example: (i) stretching by force exerted by a receding air–water interface on a hydrophobic surface (molecular combing) [40], (ii) stretching with optical or magnetic tweezers, laminar flows, atomic force cantilevers, or electric fields [41], or (iii) confinement stretching in the nanochannels [27]. Stretching DNA in the nanochannels has several advantages over other techniques: (i) external force is not required, because a long DNA molecule tends to spontaneously elongate and enter nanochannels directly from the environment due to the large free energy needed to reduce entropy, (ii) continuous measurement of the entire length of the DNA can be achieved due to the alignment of the elongated DNA molecular confined in the nanochannels. Wang et al. [18] have fabricated open nanofluidic channel arrays in a Si3N4 membrane surface using the FIB milling technique for DNA stretching. Fluorescently stained λ-DNA molecules were put inside the nanochannels by capillary force and were stretched and transferred along these nanochannels (Figure 13.4a). Mannion et al. [42] have used the interface between a nanochannel and a microchannel as a tool for applying controlled forces on a DNA molecule. The entropic force was used to stretch molecules, to retract molecules from the nanochannels, and to straighten folded strands. Tegenfeldt et al. [27] used a 100-nm nanochannel array for DNA stretching and measured directly the contour lengths of a single DNA molecule confined in the channels. Dukkipati et al. have developed a novel DNA stretching method named “protein-assisted DNA immobilization,” which utilizes the specific interactions between the DNA and DNA-binding proteins and evaporation-driven flow inside the microchannel. In this approach, the DNA–protein complex was first stretched out when subjected to an evaporation-driven hydrodynamic flow inside a microchannel, and then was immobilized onto the surface of the microchannel by the physical absorption of the protein part of the DNA–protein complex (Figure 13.4b) [39]. Cao et al. [26] have fabricated arrays of millions of nanochannels over a 100-mm wafer using nanoimprinting lithography to stretch, align, and analyze long genomic DNA in a highly parallel fashion. The same group also attempted to overcome the difficulties to introduce long genomic DNA molecules into nanometer scale fluidic channels directly from the macroscale world due to the steep entropic barrier. They designed continuous spatial gradient structures, which smoothly narrow the crosssection of a volume from the micro to the nanometer length scale and greatly reduce the local entropic barrier to the nanochannel entry (Figure 13.5) [43]. Several concerns may be considered when applying nanochannel for DNA stretching: (1) the dimensions of nanofluidic structures are critical for uniformly

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

334

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

(a) DNA molecules Reservoir Nanoconduits

(b) Silicon

Outlet

Glass

Initial drop of DNA-protein solution

Inlet

PMMA coating

10 µm

Figure 13.4. Examples of DNA stretching in the nanochannels. (a) Left: stained DNA molecules were put inside sub-100-nm conduits by capillary force and were stretched and transferred along these conduits; right: fluorescent images of the stretched DNA. (Reprinted with permission from [18] and the IEEE.) (b) Left: schematic diagram of the nanofluidic channels to immobilize and stretch DNA; right: fluorescent image of stretched DNA. (Reprinted with permission from [39] and the American Chemical Society.)

stretching long DNA; usually the width of the nanochannels should be smaller than the persistence length of double stranded DNA (∼50–70 nm) [44]. (2) Besides introducing the DNA molecules into the nanochannels spontaneously by passive transport (i.e., capillary force), positive transport of the DNA molecule into the nanochannels can be achieved by electrokinetic-driven or pressure-driven transport [45].

13.2.2.2 Protein Separation. Micro/nanofluidic separation systems have generated much interest as an enabling technology for processing complex protein samples. Decreasing the sample complexity by separation is crucial for increasing sensitivity of downstream detection tools for proteomic studies. This is required since most biomolecule detection tools (such as mass spectrometry and antibody-based biosensors) have a limited dynamic range of detection, while complex samples such as blood usually contains more than 10,000 different protein species with concentrations that vary up to nine orders of magnitude. Compared with conventional protein separation tools, such as nanoporous gels or membranes, micro/nano fluidic separation systems do not use any buffer additive or matrix, which facilitates their integration into the standard MEMS processes. These devices are mechanically and chemically more robust and could be precisely engineered to have better separation efficiencies.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS

335

(a) Microfulidic area Nanofluidic area

-TS (b)

Steep entropy barrier Gradient interfacing area

Microfulidic area

Nanofluidic area

-TS Gradual entropy slope Figure 13.5. An improved nanochannels system for DNA stretching. (a) Left: schematic diagram showing regular nanofluidic channels; right: fluorescent images of long DNA molecules congregating at the edge of the regular nanofluidic chip without entering the normal nanochannels. (b) Nanochannels with continuous spatial gradient structures. Partially stretched long DNA molecules in the micro-post array and the gradient zone continuously entering the nanochannels in the left and being fully stretched. (Reprinted with permission from [43] the American Institute of Physics.)

Various attempts have been made at fabricating nanofluidic-based protein separation devices. For example, Han et al. separated proteins that were smaller than the gap size of the nanofluidic channel using steric hindrance effect of the biomolecules (Figure 13.6) [46]. They fabricated nanofluidic filter devices (nanochannels with gaps) on silicon substrate by standard photolithography and etching techniques. Protein separation was achieved by the free energy barrier introduced by the nanofluidic filter, since the entropy of protein has to be decreased in order to enter the nanofluidic filter. Sodium dodecyl sulfate (SDS) coated proteins have been successfully separated in a nanofluidic channel that has the filter gap thickness between 60 and 120 nm. Wang et al. [47] have developed a micro/nano fluidic sample concentration system based on electrokinetic trapping and nonlinear electro-osmotic flow. The separation device generated ion depletion (resulted by the net transfer of the charges between the anode and cathode), which induced electrical double layer within micro/nano channels. The electrical double layer was used to both collect and trap the molecules efficiently. A rapid preconcentration of proteins and peptides was achieved up to 106 –108 fold in concentration, without using any physical barriers or reagents.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

336

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

Sample

Buffer

Nanofluidic filters Waste

Sample waste

d = 80–120 nm o

Figure 13.6. Schematic diagram of the nanochannel device for protein separation. (Reprinted with permission from [46] and the IEEE.)

Beside nanochannels, other nanostructures have also been used as “nanosieve” for protein separations, such as “nanofluidic interconnects” in a multilayered microfluidic chip [48] and self-assembly colloidal arrays with ordered nonporous structure embedded inside a microchannel [49]. The two main factors affecting the separation efficiency are solute valence and mobility ratio (virtually the solute diffusivity). Most reports of the protein separation in nanofluidic systems have been achieved by the electric field-driven separation (i.e., electrophoretic separation) that works well for proteins differing in mobility ratio. Alternatively, pressure-driven separation by nanochannels can also be considered to separate proteins with variable valences due to “solute-wall interactions” [50]. The combination of the pressure- and electric field-driven separations in nanochannels has been proposed for more effective separation [50]. In addition, befouling caused by unspecific absorption of the protein onto the surface of the nanofluidic devices should be minimized by approaches such as surface modification with hydrophilic chemical groups or using antibiofouling materials, like PEG, for nanochannel fabrication. Multidimensional protein separation is an attractive direction to integrate two different separation steps into a single device. It greatly increases the resolution and offers versatility with different separation mechanisms that may be paired in various combinations to enhance selectivity and sensitivity. Currently, multidimensional protein separation has been realized in microfluidic system [51, 52] and there is no multidimensional protein separation system based on nanochannels at present, which may require future efforts.

13.2.2.3 Drug Delivery System Using Nanochannels. Drug delivery devices have been miniaturized from the macroscale (>1 mm) to the microscale

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.2 FABRICATION OF NANOFLUIDIC SYSTEM FOR BIOLOGICAL APPLICATIONS

Entry hole

Channel Cap wafer

Exit flow chamber

Entry flow chamber Structural wafer

Anchor point

Exit hole

0.04 0.035 0.03 0.025 0.02 0.015 0.01 0.005 0 0

Entry flow chamber (d)

Entry flow chamber

4 mm

Nano channel Spacer layer

Anchor region Anchor point Output finger

Input finger

Released quality (%)

(b)

Released quality (%)

(c) (a)

2

4

6

8

6

8

Time (days) 0.0007 0.0006 0.0005 0.0004 0.0003 0.0002 0.0001 0

3 mm

337

0

2

4

Time (days)

Figure 13.7. (a) A cross-sectional view of the nDS. (b) A front view of the bottom substrate in the nDS. Release profiles of (c ) Glucose and (d ) IFN-α from nDS mounted in Costar diffusion chambers mounted on the wells of a transwell plate. (Reprinted with permission from [55] and the Ovid Technologies, Inc.)

(100–0.1 µm) or nanoscale (100–1 nm) [53]. Nanochannel delivery systems (nDSs) have demonstrated its unique potential to deliver a variety of bioactive molecules at zero-order rates [54]. For example, glucose and IFN-α (an antitumor compound) were delivered directly to the tumor microenvironment at zero-order kinetics by a nDS [55] (Figure 13.7). The compound released from the nanochannels was functionally active on both host immune cells and a human melanoma cell line in vitro. The nanochannel drug delivery system could be potentially implanted near the lesion with minimum invasion to provide local, sustained release of antitumor compounds. In another example, in vivo drug delivery has been achieved by using a device containing nanochannels [54]. Top-down nanofabrication techniques were used to create silicon-based nanopores membrane consisting of arrays of uniform nanochannels having a width as small as 7 nm. Using this system, a zero-order release of IFN-α or bovine serum albumin (BSA) diffusing through the nanopore was achieved. Following subcutaneous implantation in rats, slow release of protein was demonstrated [56]. This device has good control of channel size and pore distribution, which makes it possible to control the release rate. The ultimate goal of drug delivery research is to develop implantable drug delivery devices to improve therapeutic efficacy in clinic. Therefore, the future advancement of the nanochannel drug delivery device should have the capacity for drug release patterns other than linear release, which permit studies to determine the optimal release profile for specific therapeutic drugs [55]. Biocompatibility of the device should also be taken into consideration for the successful implantation [57].

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

338

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

Furthermore, future nanochannel drug delivery system should be amenable to external control by a programmable electronic device.

13.3 FABRICATION OF BIOMOLECULAR NANOARRAYS FOR BIOLOGICAL APPLICATIONS A variety of techniques have been developed to immobilize biomolecules such as DNA, protein, and lipid membrane onto surface with resolutions greater than 1 µm. The selective, precise immobilization of biomolecules with high density possesses high impact in many biological studies. For example, nanoscale arrays of proteins could be used to develop a novel diagnostic method in a rapid and high throughput manner, while patterning peptides could lead to greater control over cell-to-surface interactions. Also, nanoscale arrays can lead to high sensitivity and selectivity of BioNEMS devices with an extremely small amount of reagent. Although, dip-pen lithography [58, 59], photolithography [60], and electron-beam lithography [61–63] have been used to form nanoarray of biomolecules on the surface, these methods have limitations due to complicated fabrication procedures, low speed, or high costs involved. In this section, we focus on several examples of unconventional patterning methods such as nanoimprinting, nanomolding, and contact printing for the formation of biomolecule nanoarrays. These methods can achieve biomolecular arrays either by directly transferring biomolecules, or by modifying surfaces with a functional chemical or by fabricating templates of adhesion-resistant materials including PEG. 13.3.1 DNA Nanoarray The controlled placement of DNA molecules onto surfaces is an important process in the fabrication of DNA arrays. A common way to fabricate DNA arrays is to spot fluids containing the desired DNA fragment onto a microscope slide using metal pins [64] or microactuated nozzles [65]. An alternative way was demonstrated by in situ synthesis of oligonucleotides (up to 25 bases) using light-activated chemistry combined with photolithographic techniques [66, 67]. A major drawback of these techniques is the inherent sequential, low speed deposition. Therefore, techniques of fabricating arrays on predefined platforms in a parallel fashion would be useful from a production point of view. In this regard, simple printing or molding techniques have been applied to the construction of site-selective template for DNA array [68–72]. For example, Lange et al. used microcontact printing (µCP) with a modified PDMS stamp with positive charges on the surface (Figure 13.8a). The stamp was incubated with target DNA molecules in a solution of low pH. The stamp was then rinsed, blown dry, and printed to deliver the DNA to the target surface [69]. Guan et al. fabricated highly ordered arrays of stretched DNA molecules on a large area (up to a millimeter) by using a modified molecular combing method and soft lithography (Figure 13.8b) [73]. In this technique, topological micropatterning on PDMS stamps was used to mediate dynamic assembly of DNA molecules into arranged nanostrand

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

drop of DNA

(a)

(1) incubate (2) after rinsing (3) stamp (4) 500 nm

printed DNA pattern

(b)

20 bp oligo

500 nm

500 bp PCR fragment

PDMS stamp

Glass

Coiled DNA

Short DNA nanostrand

(3)

(2)

(4)

Fast peeling

Slow peeling Streched DNA

(1)

Streched DNA

Glsass or mica

Long DNA nanostrand

(c) (2)

(1)

20 nm 10 0 0

2

4 0

(3)

2

40

(4)

2

40

2

4 µm

Figure 13.8. Soft-lithographical approach for fabrication of DNA arrays. (a) Left: schematic diagram of microcontact printing of DNA molecules; right: fluorescence images of patterned FITClabeled oligonucleotides on a glass surface after printing and AFM images revealing the printed DNA molecules deposited on mica substrate. AFM images (tapping mode in air) of stamped 1-µm lines of oligonucleotides (left, 20-bp oligos; right, 500-bp PCR fragments) (Reprinted with permission from [69] and the American Chemical Society.) (b) Left: schematic of generating and transferring DNA nanostrand arrays; right: fluorescence micrographs of (1) DNA molecules combed on a flat PDMS stamp, (2) vertically and (3) diagonally aligned DNA strands on the PDMS stamps with microwells, and (4) long DNA strands on the PDMS stamp with microwells. (Scale bar, 10 µm.) (Reprinted with permission from [73] and the National Academy of Sciences.) (c ) Left: deposition of DNA from solution by MIMIC and scale reduction by pinning during the last stages of solvent evaporation; right: AFM images (above) and height profiles (below) of DNA molecules (100–8000 bp) patterned on a mica surface at different concentrations: (1) 1.25 µg mL−1 , (2) 2.0 µg mL−1 , (3) 2.8 µg mL−1 , and (4) 5.0 µg mL−1 . (Reprinted with permission from [70] and the Royal Society of Chemistry.)

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

340

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

arrays. The nanostrand arrays were transferred onto flat solid surfaces by contact printing, allowing for the creation of more complex patterns. As shown, stretched DNA molecules on a flat PDMS stamp were generated at a low peeling speed. Depending on the peeling speed, either short or long nanostrands formed. It was also reported that the stretched DNA molecules broke for short nanostrands at the edges of the microwells, resulting in portions in the microwells being untransferred. For long nanostrands, the segments suspended over the microwells were also transferred as those on the top surface of the stamp. Therefore, the authors obtained DNA nanostrands with the monodispersed length determined by the geometry of the micro features on the stamp. Bystrenova et al. [70] fabricated DNA array by MIMIC where dewetting occurred at the last stages of solvent evaporation (Figure 13.8c). When the solution was placed at an open end of the cavity, the solution flowed inside driven by capillary forces and surface tension with the boundary walls. Self-organization of the molecular solute occured at the later stages of solvent evaporation. As the solution was pinned to the edges of the channel, the fluid section profile resulted in an inhomogeneous rate of evaporation of the solvent. The convective flow of the solute toward the pinning sites resulted in the precipitation of split structures in the channel. Depending on the concentration, the solute precipitated when the critical concentration was reached. When the concentration of the DNA solution reached 1.25 µg mL−1 , the resulting pattern consisted of homogeneous lines. For DNA concentrations between 2 and 4 µg mL−1 , the dots were roughly aligned but without well-defined spacing. When the concentration exceeded 5 µg mL−1 , spatial correlations emerged among the dots, and the dot size increased. In this case, the dots were perfectly aligned along the edges of the channel as a result of the pinning of the solution and the convective flow from the center toward the edges. The major advantage of soft-lithographical approach is the capability of printing multiple arrays from one loaded stamp. This method could be developed to a potentially cost- and time-saving process, particularly for gene expression studies, where the ratio of bound to labeled molecules matters, but not the total amount of material. Alternatively, NIL and NIL-related techniques have been used to fabricate arrays of DNA molecules. Ohtake et al. [74, 75] performed DNA nanopatterning on a substrate immobilized with poly-l-lysine by using a nanoimprint process (Figure 13.9). First, poly-l-lysine-coated slide glass was used to immobilize the DNA molecules on a substrate. After imprinting of a poly(vinylalcohol) (PVA) layer and subsequent dry etching, the PVA layer was removed by water. Multiple lines of DNA were patterned and visualized by this approach using a fluorescent-labeled DNA. It was reported that the immobilized DNA pattern was stable due to strong Columbic interactions between positive charges of NH4 + in the poly-l-lysine layer and negative charges of PO4 − in the DNA molecules. 13.3.2 Protein Arrays The roles of proteins are enormously diverse and include mechanical support, signaling, and sensing. Beyond their central importance to biology, proteins are of great interest because these sub-microscale molecules have the potential to be integrated

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.3 FABRICATION OF BIOMOLECULAR NANOARRAYS FOR BIOLOGICAL APPLICATIONS

(b)

(a) Mold PVA Substrate

341

(c) Ploy-L-lysine Blass substrate

DNA layer Heating

Imprint

UV exposure

(1)

Cooling Polyvinyl alcohol Release

Dry etching

(2)

Figure 13.9. NIL and NIL-related techniques for fabrication of DNA arrays. (a) Schematic of the nanoimprint process for DNA array. The substrate coated with the PVA was imprinted using a SiO2 /Si mold. (b) A preparation process of a DNA substrate by using poly-L-lysine and PVA layers. (c ) Images of DNA nanopatterns observed with fluorescence microscopy. Single or multiple lines of DNA were seen where the linewidth and spacing were 0.7 and 3 µm, respectively. (Reprinted with permission from [74] and the AVS Science & Technology of Materials, Interfaces, and Processing.)

into BioNEMS devices. For aiming at this application, a nanopatterning technique is necessary, capable of accurately depositing proteins in predefined locations while retaining their native functionality. Other than spotting [76] or dip-pen nanolithography [59], printing or molding approaches have been proven to be successful in fabricating protein arrays. Renault et al. used µCP of single protein molecules on surfaces with the aid of mechanically improved structural features. They have created arrays of posts, ridges, lines, and mesh structures with critical dimensions ranging from 40 to 600 nm (Figure 13.10a) [77]. The inking and printing conditions were identical but the patterns printed with a smaller mesh rendered irregular aggregates of proteins that were dispersed on a width of ∼100 nm. Ross et al. [78] also used µCP with the polymerized lipid bilayers as substrates (Figure 13.10b). They demonstrated that printing of BSA onto a dried poly(bisSorbPC) planar supported lipid bilayer (PSLB) from a PDMS stamp produced a layer of strongly adsorbed protein, comparable in surface coverage to films printed on glass surfaces. The dried poly(bisSorbPC) bilayers were contacted with PDMS stamps inked with BSA, rinsed with buffer, and examined for evidence of protein transfer. AFM (atomic force microscope) scans taken over a large area (greater than 100 µm2 ) showed that the transferred protein films were very uniform with few defective areas.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

342

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

protein solution

(a)

Stuctured PDMS stamp Inked stamp Hard stbstrate Patterned surface

(b)

(1)

(3)

(c)

(2)

4 µm

4 µm

(4)

400 nm

400 nm

(d) Press, then remove stamp

Sylgard 184 h-PDMS

3 mm AFM test grating Embosss into PS

Produces protein pattern

2 nm Ink with dendrimers & point

1 nm

PS master

0 nm

Mold multilayer stamp Si substrate

0

2.5

5.0

7.5 10.0

Figure 13.10. Soft-lithographical methods for protein arrays. (a) Left: microcontact printing of protein; right: AFM images of antibodies printed on glass using high resolution PDMS stamps. The patterns in (1) and (2) were produced using a grid of 100-nm wide lines that were separated by 2 µm, and the patterns in (3) and (4) used 40-nm wide lines separated by 800 nm. (Reprinted with permission from [77] the American Chemical Society.) (b) Left: schematic of contact printing of BSA on a poly(bis-SorbPC) PSLB; right: a representative AFM image and line scan of BSA printed on a poly(bisSorbPC) PSLB from a PDMS stamp. The image size is 40 µm × 40 µm and the height scale is 10 nm. The dark line across the image indicates the position of the line scan. (Reprinted with permission from [78] and the American Chemical Society.) (c ) Left: schematic of an edge transfer lithography process that employs a surface-modified elastomeric stamp (gray) rendering impermeable and repellent to the applied ink by an applied surface barrier layer (black); right: optical and AFM images and related height profiles (inset) of gold substrates (10-nm-thick gold layer) after printing. (Reprinted with permission from [79] and the American Chemical Society.) (d ) (left): schematic of nanocontact printing using h-PDMS; (right): AFM image of printed titin multimer protein lines on silicon surface. (Reprinted with permission from [80] and the American Chemical Society.)

Specifically, this procedure could generate an array of fluorescent stripes corresponding to specific binding of avidin to the line pattern of biotinylated BSA. Sharpe et al. [79] introduced the edge transfer lithography for protein patterning (Figure 13.10c). This method is based on the use of the edges of micrometer-sized template features for the reproduction of submicrometer structures. This technique allows for local surface modification in a single step by depositing self-assembled monolayers onto a metal substrate selectively along the feature edges of an elastomeric stamp. Key parameters of a patterning scheme that employs PDMS stamps bearing an isotropic silicon oxide blocking layer are the impermeability of this barrier, and the surface energy compatibility between the stamp surface and the ink. With this approach, large areas could be patterned in a single-step process with dynamic control over feature sizes. Li et al. introduced the patterning of surfaces via nanocontact printing with chemically distinct features using h-PDMS (hard PDMS) (Figure 13.10d) [80]. The extension of contact printing to produce structures in the sub-100-nm scale has been

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.3 FABRICATION OF BIOMOLECULAR NANOARRAYS FOR BIOLOGICAL APPLICATIONS

343

hampered by two factors. The first factor is the low elastic modulus of Sylgard 184 PDMS stamp, which does not allow the replication of sub-100-nm features. An increase in the Young’s modulus of the PDMS improves the mechanical stability of the features on the stamp during printing [81]. The second problem is that the low molecular weight inks (alkanethiols and silanes) lead to diffusion during the printing step, resulting in a limited feature size. The authors used a V-shaped apex whose radius of curvature was approximately 30 nm. With this improved geometry of tip regions, various protein nanoarrays were fabricated by simple contact of inked hPDMS mold. Alternatively, NIL was used for protein patterning of high density with feature size below 100 nm, while retaining high throughput and reproducibility (Figure 13.11a) [82, 83]. In one example, a (CFx)n polymer (x = 1 or 2, n = number of (a)

(1)

(3)

(1)

(3)

(3)

(b) (1)

(5)

(2)

(6)

(3)

(7)

(4)

(8)

(3)

Figure 13.11. Examples of the nanoimprint process for fabrication of protein array. (a) Left: process flow diagram of substrate patterning and protein immobilization; right: epi-fluorescence image of rhodamine-labeled streptavidin bound to sharp uniform microscale dots (1) and lines (2) of biotinylated BSA protein on oxidized Si substrates. (3) Rhodamine-labeled streptavidin bound to patterns of immobilized biotinylated BSA on cover glass. (Reprinted with permission from [83] and the American Chemical Society.) (b) Left: schematic of combining NIL and molecular assembly patterning; (1) AFM scans of PLL-g -PEG/PEG-biotin stripes in an oxide background and (2) After PLL-g -PEG backfill the pattern is still visible due to the longer PEG chains supporting the biotin molecules (3) confocal laser scanning microscope (CLSM) image. (Reprinted with permission from [82] and the American Chemical Society.)

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

344

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

monomer subunits, monomer M W = 31 or 50) was deposited during a CHF3 RIE procedure for shielding adsorption of protein. The exposed oxide pattern selectively reacts with an aminosilane to form a covalently bound monolayer. The target protein was bound by flushing the flowcell with a 50 µg/mL biotinylated BSA solution in blocking buffer and incubating for 10 min. This technique achieved high throughput, reproducible nanoscale protein patterns with high selectivity and functionality, as measured by covalent bonding between patterned antibodies and their antigen. Falconnet et al. [82] developed a protein patterning method by combining NIL and molecular assembly patterning by liftoff (Figure 13.11b). A heated PMMA film was imprinted by a mold, followed by a dry etching step that converted the topography into a PMMA/Nb2O5 contrast. A biotin functionalized copolymer, poly(llysine)-graft-poly(ethylene glycol)-biotin (PLL-g-PEG/PEG-biotin), spontaneously adsorbed on the oxide surfaces. To inhibit nonspecific protein adsorption in the background, the Nb2O5 areas were rendered nonfouling by spontaneous adsorption of the nonfunctionalized PLL-g-PEG from an aqueous solution. After PMMA liftoff, the background was backfilled with protein-resistant PLL-g-PEG. They demonstrated that the streptavidin lines displayed an excellent contrast between the biotinylated areas and the PLL-g-PEG background. No fluorescent signal could be detected on the PEG background. Zhu et al. [84] described a rapid chemomechanical technique to fabricate widthadjustable extracellular matrices (ECM) protein for the study of cell adhesion (Figure 13.12a). The fabrication technique created nano- to microscale patterns starting from no pattern at all using simple procedures and equipment. The widths of the cracks (120–3200 nm) were similar in size to individual adhesion complexes (typically 500–3000 nm2 ) and can be modulated by adjusting the mechanical strain applied to the substrate. The cracks expose underlying material—the sidewalls of the cracks—onto which proteins were adsorbed. When the adsorbed proteins are cell adhesive and cell accessible, the patterned substrate supports cell attachment and spreading. Using this approach, they patterned a variety of proteins and demonstrated patterned attachment, spreading, and retraction of various types of cells along ECM protein-adsorbed cracks. Suh et al. demonstrated fabrication of protein patterns by using CFL technique with a nonbiofouling polymer material as the template layer against protein adsorption, resulting in patterned cell arrays [86–89]. They used nanowell patterns of a PEG-based random copolymer, poly(3-trimethoxysilyl)propyl methacrylate-rpoly(ethylene glycol) methyl ether) [poly(TMSMA-r-PEGMA)] with good physical integrity (Figure 13.12b) [85]. The height of patterns ranged from a few nanometers to a few hundred nanometers, serving as a migration barrier to maintain a sharp boundary between patterned and nonpatterned regions. This PEG copolymer has an anchoring group and thus can bind covalently to silicon surfaces, allowing for excellent water stability for at least 2 weeks. Using PDMS molds, 700-nm nanowells were fabricated on silicon wafer over a large area with a barrier height of ∼280 nm. The incubation of an antibody onto the PEG template resulted in a selective adsorption of the antibody onto the exposed substrate because the PEG copolymer surface is highly resistant to protein adhesion.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.3 FABRICATION OF BIOMOLECULAR NANOARRAYS FOR BIOLOGICAL APPLICATIONS

345

(b)

(a)

Transparent mold (PDMS) Poly(TMSMA-r-PEGMA) Substrate (glass or silicon) Step 1. Nanomolding

Step 2. Peeling off the stamp

(1)

Step 3. Treating with antibody

PLURONIC F108 (2)

(1) (3)

(2)

Figure 13.12. (a) Schematic illustration of crack patterning; (1): parallel cracks formed by the procedure using fluorescent protein (FITC-BSA); (2): a micrograph of C2C12 myoblasts spread on wide cracks; and (3) a C2C12 myoblast cell stained with phalloidin-TRITC for actin, Hoechst 33342 for the nucleus. The cell is attached to a crack coated with collagen. (Reprinted with permission from [84] and the Nature Publishing Group.) (b) Schematic illustration of nanomolding for protein patterning; (1) AFM images of 700-nm PEG nanowells in deflection mode; (2) the corresponding fluorescent image treated with the P3 antibody and a FITC-labeled secondary antibody image. (Reprinted with permission from [85] and the American Chemical Society.)

13.3.3 Lipid Array Cellular membranes contain many proteins that regulate molecular interactions. Of these, receptor proteins play a central role in cell signaling. Because lipid bilayers are held together by hydrophilic and hydrophobic interactions, biomolecules embedded in these structures are free to diffuse laterally. Preparation of arrays of lipid bilayer or membrane is important since the binding event is highly specific and varies depending on the targeting strategy employed (e.g., chemical linking, physical binding, or biospecific recognition). Lipid membranes are a highly efficient biosensing platform for electrochemical detection of ligand–receptor interactions, cellular attack, or signal transduction [90–92]. Furthermore, the inclusion of multiple functional groups on the surface is beneficial in the 2D geometry of the lipid membranes [93].

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

346

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

In 1997, Groves et al. [94] first introduced a microfabrication method for patterning surfaces with solid supported phospholipid bilayers. The method involved the patterning of photoresist on fused quartz wafers by means of standard photolithographic techniques. Since then, a number of methods have been developed for microarrays of lipid membrane, such as deep-UV illumination through a photomask under aqueous conditions [95, 96] and polymer liftoff method [97–100]. µCP of composition arrays of phospholipids bilayers was first accomplished by printing different-sized bilayers of the same composition into surface patterned corrals [101–103]. Also, MIMIC technique was used to pattern lipid membranes by utilizing a laminarly flowing stream [104–106]. The use of laminar flow inside microfluidic channels is also an effective means of producing composition arrays of supported phospholipid bilayers in which two distinct chemical components can be varied simultaneously along a one-dimensional gradient [104, 107]. Recently, CFL was used to create patterns of supported lipid bilayer (SLB) membranes onto a surface (Figure 13.13). Micro- or nanopatterns of a PEG random copolymer were fabricated on glass substrates by CFL to form a template layer against adsorption of lipid membranes [108]. As compared to µCP, the molded structures provided a clean interface at the patterned boundary and the adhesion on the PEG surface was strongly restricted. The functionality of the patterned SLBs was tested by measuring the binding interactions between biotinlabeled lipid bilayer and streptavidin. SLB arrays were fabricated with spatial resolution down to ∼500 nm on flat substrate and ∼1 µm inside microfluidic channels, respectively. (b) (1)

(a) Mold PEG copolymer Glass

(2)

Biotinylated POPC liposome Streptavidin

Step 1. Polymer coating Seeding of biotin-liposome vesicle

(3)

(4)

Step 2. Contact and Press Selective immobilization of vesicle

(5)

Step 3. Solvent evaporation Vesicle fusion and binding of Biotinstreptavidin

Step 4. Mold removal and aging

Figure 13.13. (a) A schematic diagram for patterning supported bilayer membranes (SBMs) by using CFL of a PEG-based copolymer. (b) Resulting micro- and nanopatterns of SBMs. (1), (2) Optical images of the PEG patterns and fluorescent images of the biotinylated lipid layers (inset) using microcontact printing (1) and CFL (2). The 10 µm box pattern was used for both methods. (3), (4) Fluorescent micrographs of the sub-micrometer patterns of biotinylated SBMs along with intensity profiles using CFL: (3) 500 nm lanes and (4) 500 nm grids. (5) A simple Y-shaped channel combined with nanoarrays of biotin and streptavidin after binding reaction. (Reprinted with permission from [108] and the Royal Society of Chemistry.)

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

13.4 FABRICATION OF NANOSCALE TOPOGRAPHIES

347

13.4 FABRICATION OF NANOSCALE TOPOGRAPHIES FOR TISSUE ENGINEERING APPLICATIONS Living tissues are ensembles of different cell types embedded in complex and welldefined structures of ECM with nanoscale topographical features. Frequently, the organization of ECM is hierarchical in that large-scale protein structures up to several hundred micrometers are found, which are covered with intricate nanoscale features. For example, hierarchical assembly of collagen molecules can lead to formation of fibrils, long (tens of microns) cylindrical structures, whose diameter may vary in 20–200 nm range. In connective tissues, it is also common to find bundles of collagen microfibrils running in parallel to each other, with cells of various origins attached to them. These cells might both affect and be affected by the super-structures of collagen and other ECM components. Thus, it could greatly improve our understanding on how cell–ECM interactions affect cellular processes if one could build up ECM nanopatterns on various 2D and three-dimensional (3D) substrates. A few nanopatterning approaches have been proposed to produce ECM protein patterns at the nanoscale and to control the spatial distribution of ECM proteins. For example, scanning probe lithography (SPL)-based methods were used to print collagens and collagen-like peptides (30–50 nm linewidth) without disrupting the triple-helical structures and biological activities of collagens [109, 110]. Nanopatterning of self-assembled monolayers by EBL has been developed to produce organic templates for creating high density protein nanoarrays [111, 112]. Selective molecular assembly patterning in combination with lithographically prepatterned substrate also has been proven as a fast and reliable method to create protein arrays over large areas with feature sizes comparable to SPL-based techniques [113–118]. Here, we focus on simple printing or molding techniques for incorporating complex nanostructures into an underlying substratum to mimic various in vivo 3D ECM environments with structural and mechanical similarity. These methods are of high significance due to its potential to yield both fundamental knowledge of mechanisms governing cell motility and the resulting control of cellular function that can be used in advancing tissue engineering. 13.4.1 Nanotopography-Induced Changes in Cell Adhesion Recently, CFL has been used to create various nanotopographies of UV-curable PEG hydrogel on glass substrate by using a new UV-curable mold made from polyurethane functionalized with acrylate groups (PUA) (Figure 13.14) [114]. It was found that proteins (collagen, fibronectin, immunoglobulin, and albumin) and cells (CHO, fibroblasts, and P19 EC cells) preferred to adhere on nanostructured PEG surfaces in comparison to unpatterned PEG films. However, the level of adhesion was significantly lower than that of glass controls. These results agree with other studies such that surface nanotopography enhances the cell-substratum adhesion of human corneal epithelial cells [119] and primary rat cardiomyocytes [120]. Other studies using CFL revealed that the wetting property of the nanostructured

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

348

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

Transparent mold

Step 2. Contacting the mold Active energy blanket

Step 5. Treatment of ECM protein

(b) PEG Nano Glass control

Step 1. Orienting the mold and Coating the polymer

(1)

(2)

(3)

(4)

Bare PEG

UV curable hydrogel layer

(5)

(6)

Step 3. Curing the structure Step 6. Study of cell adhesion Step 4. Separating the mold

Figure 13.14. (a) A schematic procedure of UV-assisted nanomolding of PEG hydrogel using CFL technique. (b) Optical micrographs of the stained P19 EC cells adhered on glass, bare PEG, and a PEC nanostructured surface in the presence of collagen (1), (3), (5) and in the absence of collagen (2), (4), (6). The cells were cultured for 2 days.

surface of a given chemical composition can be systematically controlled by rendering nanoscale roughness [121]. These findings suggest that the control over wettability by surface nanotopography can also modulate spatial distribution of absorbed proteins and cell adhesion properties. However, the role played by the adhesive ECM proteins coated on the nanostructured polymeric surface on cell adhesion has not been extensively addressed. Further studies by combining nanoscale mechanical topography and adhesive ECM nanopatterns on polymer cell culture substrates possibly provide more insights on how cells interact with surface nanotopography and surface adhesive nanopattern, respectively [122]. 13.4.2 Nanotopography-Induced Changes in Cell Morphology It has been found that nanotopography induces change in morphology and motility of many different cell types, which is also called “contact guidance.” Contact guidance refers to the reactions of cells with the topography of their substratum [123, 124]. Arrays of parallel nanogrooves (alternatively nanoridges) have been fabricated by nanoimprint or UV photolithography to provide an in vitro experimental model for investigating how nanotopography of the in vivo ECM can affect cell motility. Further quantitative analysis with corneal epithelial cells showed that the percentage of aligned cells was constant on the substrate topographies with lateral dimensions ranging from 70 nm to 2 µm, and increased with groove depth ranging from 150 to 600 nm [125]. Interestingly, recent ultrastructural analysis showed that primary human corneal epithelial cells oriented perpendicular to 400-nm pitch patterns (70 nm ridges) extended filopodia in a crown-like fashion with individual filopodium having orientations that were perpendicular, oblique or parallel to the underlying pattern (Figure 13.15) [126]. In contrast, in cells that were aligned parallel to larger scale

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

REFERENCES

(a)

(c)

(d)

349

(e)

(b)

Figure 13.15. Nanotopography-induced change in morphology and motility. (a), (b) SEM images of perpendicularly aligned cell on 70-nm wide ridges on a 400 nm pitch. (Reprinted with permission from [126] and the Elsevier B.V.). Immunofluorescent micrographs of cells on (c ) a substrate with 600-nm deep grooves and 70-nm wide ridges on a 400 nm pitch, (d ) a substrate with 600-nm deep grooves and 1900 nm ridges on a 4000 nm pitch, and (e) a smooth silicon oxide substrate, respectively, stained for actin, vinculin, and the nucleus. Insets: a reflection image of the substrates. (Reprinted with permission from [125] and the Company of Biologists Ltd.)

patterns (4000 nm pitch, 1900 nm ridges), filopodia were generally more guided by the substrate topographies resulting in most filopodia extending parallel to the underlying features [126]. Details on cell–substrate interactions in cell biology and tissue engineering applications can be found in an extensive review paper [4]. It is envisioned that better understanding of signaling and cellular functions through the nanoscale control of cell-ECM interaction could open up novel strategies to achieve targeted cell functions in tissue engineering applications.

REFERENCES 1. Mrksich, M., Chen, C. S., Xia, Y. N., Dike, L. E., Ingber, D. E., and Whitesides, G. M. (1996) Controlling cell attachment on contoured surfaces with self-assembled monolayers of alkanethiolates on gold. Proc. Natl. Acad. Sci. USA 93, 10775– 10778. 2. Whitesides, G. M., Ostuni, E., Takayama, S., Jiang, X. Y., and Ingber, D. E. (2001) Soft lithography in biology and biochemistry. Annu. Rev. Biomed. Eng. 3, 335–373. 3. Yim, E. K. F., Reano, R. M., Pang, S. W., Yee, A. F., Chen, C. S., and Leong, K. W. (2005) Nanopattern-induced changes in morphology and motility of smooth muscle cells. Biomaterials 26, 5405–5413. 4. Sniadecki, N., Desai, R. A., Ruiz, S. A., and Chen, C. S. (2006) Nanotechnology for cell-substrate interactions. Ann. Biomed. Eng. 34, 59–74. 5. Shin, H. (2007) Fabrication methods of an engineered microenvironment for analysis of cell-biomaterial interactions. Biomaterials 28, 126–133. 6. Wikswo, J. P., Prokop, A., Baudenbacher, F., Cliffel, D., Csukas, B., and Velkovsky, M. (2006) Engineering challenges of BioNEMS: the integration of microfluidics, micro- and

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

350

7. 8. 9. 10.

11.

12.

13. 14. 15.

16.

17.

18.

19.

20.

21. 22.

23.

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

nanodevices, models and external control for systems biology. IEE Proc., Nanobiotechnol. 153, 81–101. Pearson, D. R. S. C. J. L. (2005) A single-step process for making nanofluidic channels using electron beam lithography. Microelectron. Eng. 78–79, 343–348. Liao, K., Yao, N., and Kuo, T. (2006) Sub-60 nm nanofluidic channels fabricated by glass-glass bonding. Proceedings of the 28th IEEE, pp. 2832–2835. Kovarik, M. L. and Jacobson, S. C. (2007) Attoliter-scale dispensing in nanofluidic channels. Anal. Chem. 79, 1655–1660. Karnik, R., Castelino, K., Fan, R., Yang, P., and Majumdar, A. (2005) Effects of biological reactions and modifications on conductance of nanofluidic channels. Nano Lett. 5, 1638–1642. Turner, S. W., Cabodi, M., and Craighead, H. G. (2002) Confinement-induced entropic recoil of single DNA molecules in a nanofluidic structure. Phys. Rev. Lett. 88, 128103. Reccius, C. H., Mannion, J. T., Cross, J. D., and Craighead, H. G. (2005) Compression and free expansion of single DNA molecules in nanochannels. Phys. Rev. Lett. 95, 268101. Mijatovic, D., Eijkel, J. C., and Van Den Berg, A. (2005) Technologies for nanofluidic systems: top-down vs. bottom-up—a review. Lab Chip 5, 492–500. Mao, P. and Han, J. (2005) Fabrication and characterization of 20 nm planar nanofluidic channels by glass-glass and glass-silicon bonding. Lab Chip 5, 837–844. Chapman, H. N., Ray-Chaudhuri, A.K., Tichenor, D.A., Replogle, W.C., Stulen, R.H., Kubiak, G.D., Rockett, P.D., Klebanoff, L.E., et. al. (2001) First lithographic results from the extreme ultraviolet engineering test stand. J. Vac. Sci. Technol. B 19, 2389–2395. Naulleau, P. (2002) Sub-70 nm extreme ultraviolet lithography at the advanced light source static microfield exposure station using the engineering test stand set-2 optic. J. Vac. Sci. Technol. B 20, 2829–2833. Danelon, C., Santschi, C., Brugger, J., and Vogel, H. (2006) Fabrication and functionalization of nanochannels by electron-beam-induced silicon oxide deposition. Langmuir 22, 10711–10715. Wang, K., Yue, S., Wang, L., Jin, A., Gu, C., Wang, P., Wang, H., Xu, X., et al. (2006) Nanofluidic channels fabrication and manipulation of DNA molecules. IEE Proc. Nanobiotechnol. 153, 11–15. Kaige, W., Pengye, W., Hong, W., Changzhi, G., Shuanglin, Y., Aizi, J., Wenqing, L., and Hanben, N. (2005) Fabricating nanofluidic channels and applying it for single biomolecule study. Conf. Proc. IEEE Eng. Med. Biol. Soc. 2, 1278–1281. OBrien, M. J., II, Bisong, P., Ista, L. K., Rabinovich, E. M., Garcia, A. L., Sibett, S. S., Lopez, G. P., and Brueck, R. (2003) Fabrication of an integrated nanofluidic chip using interferometric lithography. J. Vac. Sci. Technol. B 21, 2941. M¨ullenborn, M., Dirac, H., and Petersen, J. W. (1995) Three-dimensional nanostructures by direct laser etching of Si. Appl. Surf. Sci. 86, 568–576. Hui, A. P., Shui-jie, Q., Li, W. J., and Wang, M. Y. (2002) High aspect ratio nano fluidic channels by laser controlled fracturing. The 15th IEEE International Conference on Micro Electro Mechanical Systems, pp. 156–159. Anpan, H., Nicolaas, F. D. R., and Urs, S. (2006) Design and fabrication of nanofluidic devices by surface micromachining. Nanotechnology 86, 2498–2503.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

REFERENCES

351

24. Truskett, V. N. and Watts, M. P. (2006) Trends in imprint lithography for biological applications. Trends Biotechnol. 24, 312–317. 25. Abgrall, P., Low, L. N., and Nguyen, N. T. (2007) Fabrication of planar nanofluidic channels in a thermoplastic by hot-embossing and thermal bonding. Lab Chip 7, 520–522. 26. Cao, H., Yu, Z., Wang, J., and Zhou, S. Y. (2002) Fabrication of 10 nm enclosed nanofluidic channels. Appl. Phys. Lett. 81, 174–176. 27. Tegenfeldt, J. O., Prinz, C., Cao, H., Chou, S., Reisner, W. W., Riehn, R., Wang, Y. M., Cox, E. C., et al. (2004) From the cover: the dynamics of genomic-length DNA molecules in 100-nm channels. Proc. Natl. Acad. Sci. USA 101, 10979–10983. 28. Wang, X., Chen, Y., Banu, S., Morgan, H., Fu, S., and Cui, Z. (2007) High density patterns fabricated in SU-8 by UV curing nanoimprint. Microelectron. Eng. 84, 872–876. 29. Kim, P., Jeong, H. E., Khademhosseini, A. and Suh, K. Y. (2006) Fabrication of nonbiofouling polyethylene glycol micro- and nanochannels by ultraviolet-assisted irreversible sealing. Lab Chip 6, 1432–1437. 30. Lensen, M. C., Mela, P., Mourran, A., Groll, J., Heuts, J., Rong, H., and Moller, M. (2007) Micro- and nanopatterned star poly(ethylene glycol) (PEG) materials prepared by UV-based imprint lithography. Langmuir 23, 7841–7846. 31. Lee, N. Y., Lim, J. R., Lee, M. J., Kim, J. B., Jo, S. J., Baik, H. K., and Kim, Y. S. (2006) Hydrophilic composite elastomeric mold for high-resolution soft lithography. Langmuir 22, 9018–9022. 32. Ilic, B., Czaplewski, D., Zalalutdinov, M., Schmidt, B., and Craighead, H. G. (2002) Fabrication of flexible polymer tubes for micro and nanofluidic applications. J. Vac. Sci. Technol. B 20, 2459–2465. 33. Kim P. and Suh, K. Y. (2007) Rigiflex, spontaneously wettable polymeric mold for forming reversibly bonded nanocapillaries. Langmuir 23, 4549–4553. 34. Huh, D., Mills, K. L., Zhu, X., Burns, M. A., Thouless, M. D., and Takayama, S. (2007) Tuneable elastomeric nanochannels for nanofluidic manipulation. Nat. Mater. 6, 424–428. 35. Park, K. S., Seo, E. K., Do, Y. R., Kim, K., and Sung, M. M. (2006) Light stamping lithography: microcontact printing without Inks. J. Am. Chem. Soc. 128, 858–865. 36. Slater, G. W., Desruisseaux, C., Hubert, S. J., Mercier, J. F., Labrie, J., Boileau, J., Tessier, F., and Pepin, M. P. (2000) Theory of DNA electrophoresis: a look at some current challenges. Electrophoresis 21, 3873–3887. 37. Lin, J., Qi, R., Aston, C., Jing, J., Anantharaman, T. S., Mishra, B., White, O., Daly, M. J., et al. (1999) Whole-genome shotgun optical mapping of deinococcus radiodurans. Science 285, 1558–1562. 38. Riehn, R., Lu, M., Wang, Y. M., Lim, S. F., Cox, E. C., and Austin, R. H. (2005) Restriction mapping in nanofluidic devices. Proc. Natl. Acad. Sci. USA 102, 10012– 10016. 39. Dukkipati, V. R., Kim, J. H., Pang, S. W., and Larson, R. G. (2006) Protein-assisted stretching and immobilization of DNA molecules in a microchannel. Nano Lett. 6, 2499–2504. 40. Herrick, J. and Bensimon, A. (1999) Single molecule analysis of DNA replication. Biochimie 81, 859–871.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

352

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

41. Strick, T., Allemand, J., Croquette, V., and Bensimon, D. (2000) Twisting and stretching single DNA molecules. Prog. Biophys. Mol. Biol. 74, 115–140. 42. Mannion, J. T., Reccius, C. H., Cross, J. D., and Craighead, H. G. (2006) Conformational analysis of single DNA molecules undergoing entropically induced motion in nanochannels. Biophys. J. 90, 4538–4545. 43. Cao, H., Tegenfeldt, J. O., Austin, R. H., and Chou, S. Y. (2002) Gradient nanostructures for interfacing microfluidics and nanofluidics. Appl. Phys. Lett. 81, 3058–3060. 44. Bakajin, O. B., Duke, T. A. J., Chou, C. F., Chan, S. S., Austin, R. H., and Cox, E. C. (1998) Electrohydrodynamic stretching of DNA in confined environments. Phys. Rev. Lett. 80, 2737. 45. Stein, D., Van Der Heyden, F. H., Koopmans, W. J., and Dekker, C. (2006) Pressuredriven transport of confined DNA polymers in fluidic channels. Proc. Natl. Acad. Sci. USA 103, 15853–15858. 46. Han, J. and Fu, J. (2004) Biomolecule separation by steric hindrance using nanofluidic filters. Conf. Proc. IEEE Eng. Med. Biol. Soc. 4, 2611–2614. 47. Wang, Y. C., Stevens, A. L., and Han, J. (2005) Million-fold preconcentration of proteins and peptides by nanofluidic filter. Anal. Chem. 77, 4293–4299. 48. Flachsbart, B. R., Wong, K., Iannacone, J. M., Abante, E. N., Vlach, R. L., Rauchfuss, P. A., Bohn, P. W., Sweedler, J. V., et al. (2006) Design and fabrication of a multilayered polymer microfluidic chip with nanofluidic interconnects via adhesive contact printing. Lab Chip 6, 667–674. 49. Zeng, Y. and Harrison, D. J. (2007) Self-assembled colloidal arrays as three-dimensional nanofluidic sieves for separation of biomolecules on microchips. Anal. Chem. 79, 2289–2295. 50. Xuan, X. and Li, D. (2007) Solute separation in nanofluidic channels: pressure-driven or electric field-driven?. Electrophoresis 28, 627–634. 51. Wang, Y. C., Choi, M. H., and Han, J. (2004) Two-dimensional protein separation with advanced sample and buffer isolation using microfluidic valves. Anal. Chem. 76, 4426–4431. 52. Li, Y., Buch, J. S., Rosenberger, F., DeVoe, D. L., and Lee, C. S. (2004) Integration of isoelectric focusing with parallel sodium dodecyl sulfate gel electrophoresis for multidimensional protein separations in a plastic microfluidic [Correction of microfludic] network. Anal. Chem. 76, 742–748. 53. LaVan, D. A., McGuire, T., and Langer, R. (2003) Small-scale systems for in vivo drug delivery. Nat. Biotechnol. 21, 1184–1191. 54. Martin, F., Walczak, R., Boiarski, A., Cohen, M., West, T., Cosentino, C., Shapiro, J., and Ferrari, M. (2005) Tailoring width of microfabricated nanochannels to solute size can be used to control diffusion kinetics. J. Control Release 102, 123–133. 55. Lesinski, G. B., Sharma, S., Varker, K. A., Sinha, P., Ferrari, M., and Carson, W. E. (2005) Release of biologically functional interferon-alpha from a nanochannel delivery system. Biomed. Microdevices 7, 71–79. 56. Staples, M., Daniel, K., Cima, M. J., and Langer, R. (2006) Application of micro- and nano-electromechanical devices to drug delivery. Pharm. Res. 23, 847–863. 57. Voskerician, G., Shive, M. S., Shawgo, R. S., von Recum, H., Anderson, J. M., Cima, M. J., and Langer, R. (2003) Biocompatibility and biofouling of MEMS drug delivery devices. Biomaterials 24, 1959–1967.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

REFERENCES

353

58. Piner, R. D., Zhu, J., Xu, F., Hong, S. H., and Mirkin, C. A. (1999) “Dip-Pen” nanolithography. Science 283, 661–663. 59. Lee, K. B., Park, S. J., Mirkin, C. A., Smith, J. C., and Mrksich, M. (2002) Protein nanoarrays generated by dip-pen nanolithography. Science 295, 1702–1705. 60. Schena, M., Shalon, D., Heller, R., Chai, A., Brown, P. O., and Davis, R. W. (1996) Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proc. Natl. Acad. Sci. USA 93, 10614–10619. 61. Pallandre, A., Glinel, K., Jonas, A. M., and Nysten, B. (2004) Binary nanopatterned surfaces prepared from silane monolayers. Nano Lett. 4, 365–371. 62. Harnett, C. K., Satyalakshmi, K. M., and Craighead, H. G. (2001) Bioactive templates fabricated by low-energy electron beam lithography of self-assembled monolayers. Langmuir 17, 178–182. 63. Borini, S., D’Auria, S., Rossi, M., and Rossi, A. M. (2005) Writing 3D protein nanopatterns onto a silicon nanosponge. Lab Chip 5, 1048-1052. 64. Shalon, D., Smith, S. J., and Brown, P. O. (1996) A DNA microarray system for analyzing complex DNA samples using two-color fluorescent probe hybridization. Genome Res. 6, 639–645. 65. Schober, A., Gunther, R., Schwienhorst, A., Doring, M., and Lindemann, B. F. (1993) Accurate high-speed liquid handling of very small biological samples. Biotechniques 15, 324–329. 66. Lipshutz, R. J., Fodor, S. P. A., Gingeras, T. R., and Lockhart, D. J. (1999) High density synthetic oligonucleotide arrays. Nat. Genet. 21, 20–24. 67. Fodor, S. P. A., Read, J. L., Pirrung, M. C., Stryer, L., Lu, A. T., and Solas, D. (1991) Light-directed, spatially addressable parallel chemical synthesis. Science 251, 767–773. 68. Matsumoto, F., Kamiyama, M., Nishio, K., and Masuda, H. (2005) Highly ordered nanopatterning of DNA with 40 nm diameter using anodic porous alumina substrate. Japan. J. Appl. Phys. Part 2: Lett. Express Lett. 44, L355–L358. 69. Lange, S. A., Benes, V., Kern, D. P., Horber, J. K. H., and Bernard, A. (2004) Microcontact printing of DNA molecules. Anal. Chem. 76, 1641–1647. 70. Bystrenova, E., Facchini, M., Cavallini, M., Cacace, M. G., and Biscarini, F. (2006) Multiple length-scale patterning of DNA by stamp-assisted deposition. Angew. Chem.Int. Ed. 45, 4779–4782. 71. Zhang, G. J., Tanii, T., Funatsu, T., and Ohdomari, I. (2004) Patterning of DNA nanostructures on silicon surface by electron beam lithography of self-assembled monolayer. Chem. Commun. 786–787. 72. Shin, J. S. (2004) Rewritable memory by controllable nanopatterning of DNA. Abstr. Papers Am. Chem. Soc. 227, U241–U241. 73. Guan, J. J. and Lee, J. (2005) Generating highly ordered DNA nanostrand arrays. Proc. Natl. Acad. Sci. USA 102, 18321–18325. 74. Ohtake, T., Nakamatsu, K., Matsui, S., Tabata, H., and Kawai, T. (2004) DNA nanopatterning with self-organization by using nanoimprint. J. Vac. Sci. Technol. B 22, 3275–3278. 75. Ohtake, T., Nakamatsu, K., Matsui, S., Tabata, H., and Kawai, T. (2006) Novel DNA nano-patterning design method utilizing poly-l-lysine patterning by nanoimprint lithography. J. Nanosci. Nanotechnol. 6, 2187–2190.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

354

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

76. Bergman, A. A., Buijs, J., Herbig, J., Mathes, D. T., Demarest, J. J., Wilson, C. D., Reimann, C. T., Baragiola, R. A., et al. (1998) Nanometer-scale arrangement of human serum albumin by adsorption on defect arrays created with a finely focused ion beam. Langmuir 14, 6785–6788. 77. Renaultt, J. P., Bernard, A., Bietsch, A., Michel, B., Bosshard, H. R., Delamarche, E., Kreiter, M., Hecht, B., et al. (2003) Fabricating arrays of single protein molecules on glass using microcontact printing. J. Phys. Chem. B 107, 703–711. 78. Ross, E. E., Joubert, J. R., Wysocki, R. J., Nebesny, K., Spratt, T., O’Brien, D. F., and Saavedra, S. S. (2006) Patterned protein films on poly(lipid) bilayers by microcontact printing. Biomacromolecules 7, 1393–1398. 79. Sharpe, R. B. A., Titulaer, B. J. F., Peeters, E., Burdinski, D., Huskens, J., Zandvliet, H. J. W., Reinhoudt, D. N., and Poelsema, B. (2006) Edge transfer lithography using alkanethiol inks. Nano Lett. 6, 1235–1239. 80. Li, H. W., Muir, B. V. O., Fichet, G., and Huck, W. T. S. (2003) Nanocontact printing: a route to sub-50-nm-scale chemical and biological patterning. Langmuir 19, 1963–1965. 81. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. 82. Falconnet, D., Pasqui, D., Park, S., Eckert, R., Schift, H., Gobrecht, J., Barbucci, R., and Textor, M. (2004) A novel approach to produce protein nanopatterns by combining nanoimprint lithography and molecular self-assembly. Nano Lett. 4, 1909–1914. 83. Hoff, J. D., Cheng, L. J., Meyhofer, E., Guo, L. J., and Hunt, A. J. (2004) Nanoscale protein patterning by imprint lithography. Nano Lett. 4, 853–857. 84. Zhu, X. Y., Mills, K. L., Peters, P. R., Bahng, J. H., Liu, E. H., Shim, J., Naruse, K., Csete, M. E., et al. (2005) Fabrication of reconfigurable protein matrices by cracking. Nat. Mater. 4, 403–406. 85. Suh, K. Y., Khademhosseini, A., Jon, S., and Langer, R. (2006) Direct confinement of individual viruses within polyethylene glycol (PEG) nanowells. Nano Lett. 6, 1196–1201. 86. Khademhosseini, A., Jon, S., Suh, K. Y., Tran, T. N. T., Eng, G., Yeh, J., Seong, J., and Langer, R. (2003) Direct patterning of protein- and cell-resistant polymeric monolayers and microstructures. Adv. Mater. 15, 1995–2000. 87. Khademhosseini, A., Suh, K. Y., Yang, J. M., Eng, G., Yeh, J., Levenberg, S., and Langer, R. (2004) Layer-by-layer deposition of hyaluronic acid and poly-l-lysine for patterned cell co-cultures. Biomaterials 25, 3583–3592. 88. Suh, K. Y., Khademhosseini, A., Yang, J. M., Eng, G., and Langer, R. (2004) Soft lithographic patterning of hyaluronic acid on hydrophilic substrates using molding and printing. Adv. Mater. 16, 584–588. 89. Suh, K. Y., Seong, J., Khademhosseini, A., Laibinis, P. E., and Langer, R. (2004) A simple soft lithographic route to fabrication of poly(ethylene glycol) microstructures for protein and cell patterning. Biomaterials 25, 557–563. 90. Yang, T. L., Baryshnikova, O. K., Mao, H. B., Holden, M. A., and Cremer, P. S. (2003) Investigations of bivalent antibody binding on fluid-supported phospholipid membranes: the effect of hapten density. J. Am. Chem. Soc. 125, 4779–4784. 91. Groves, J. T. and Dustin, M. L. (2003) Supported planar bilayers in studies on immune cell adhesion and communication. J. Immunol. Methods 278, 19–32. 92. Sackmann, E. and Tanaka, M. (2000) Supported membranes on soft polymer cushions: fabrication, characterization and applications. Trends Biotechnol. 18, 58–64.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

REFERENCES

355

93. Christensen, S. M. and Stamou, D. (2007) Surface-based lipid vesicle reactor systems: fabrication and applications. Soft Matter 3, 828–836. 94. Groves, J. T., Ulman, N. and Boxer, S. G. (1997) Micropatterning fluid lipid bilayers on solid supports. Science 275, 651–653. 95. Yee, C. K., Amweg, M. L., and Parikh, A. N. (2004) Membrane photolithography: direct micropatterning and manipulation of fluid phospholipid membranes in the aqueous phase using deep-UV light. Adv. Mater. 16, 1184–1189. 96. Yu, C. H., Parikh, A. N., and Groves, J. T. (2005) Direct patterning of membranederivatized colloids using in-situ UV-ozone photolithography. Adv. Mater. 17, 1477–1480. 97. Orth, R. N., Kameoka, J., Zipfel, W. R., Ilic, B., Webb, W. W., Clark, T. G., and Craighead, H. G. (2003) Creating biological membranes on the micron scale: forming patterned lipid bilayers using a polymer lift-off technique. Biophys. J. 85, 3066– 3073. 98. Orth, R. N., Wu, M., Holowka, D. A., Craighead, H. G., and Baird, B. A. (2003) Mast cell activation on patterned lipid bilayers of subcellular dimensions. Langmuir 19, 1599–1605. 99. Wu, M., Holowka, D., Craighead, H. G., and Baird, B. (2004) Visualization of plasma membrane compartmentalization with patterned lipid bilayers. Proc. Natl. Acad. Sci. USA 101, 13798–13803. 100. Moran-Mirabal, J. M., Edel, J. B., Meyer, G. D., Throckmorton, D., Singh, A. K., and Craighead, H. G. (2005) Micrometer-sized supported lipid bilayer arrays for bacterial toxin binding studies through total internal reflection fluorescence microscopy. Biophys. J. 89, 296–305. 101. Groves, J. T. and Boxer, S. G. (2002) Micropattern formation in supported lipid membranes. Acc. Chem. Res. 35, 149–157. 102. Lenz, P., Ajo-Franklin, C. M., and Boxer, S. G. (2004) Patterned supported lipid bilayers and monolayers on poly(dimethylsiloxane). Langmuir 20, 11092–11099. 103. Vidic, J., Pla-Roca, M., Grosclaude, J., Persuy, M. A., Monnerie, R., Caballero, D., Errachid, A., Hou, Y. X., et al. (2007) Gold surface functionalization and patterning for specific immobilization of olfactory receptors carried by nanosomes. Anal. Chem. 79, 3280–3290. 104. Kam, L. and Boxer, S. G. (2000) Formation of supported lipid bilayer composition arrays by controlled mixing and surface capture. J. Am. Chem. Soc. 122, 12901– 12902. 105. Schuy, S. and Janshoff, A. (2006) Structuring of phospholipid bilayers on gold surfaces by micromolding in capillaries. J. Colloid Interface Sci. 295, 93–99. 106. Yang, T. L., Jung, S. Y., Mao, H. B., and Cremer, P. S. (2001) Fabrication of phospholipid bilayer-coated microchannels for on-chip immunoassays. Anal. Chem. 73, 165–169. 107. Kam, L. and Boxer, S. G. (2003) Spatially selective manipulation of supported lipid bilayers by laminar flow: steps toward biomembrane microfluidics. Langmuir 19, 1624–1631. 108. Kim, P., Lee, S. E., Jung, H. S., Lee, H. Y., Kawai, T., and Suh, K. Y. (2006) Soft lithographic patterning of supported lipid bilayers onto a surface and inside microfluidic channels. Lab Chip 6, 54–59.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

356

15:30

Printer Name: Yet to Come

UNCONVENTIONAL PATTERNING METHODS FOR BIONEMS

109. Jiang, F. Z., Khairy, K., Poole, K., Howard, J., and Muller, D. J. (2004) Creating nanoscopic collagen matrices using atomic force microscopy. Microsc. Res. Tech. 64, 435–440. 110. Wilson, D. L., Martin, R., Hong, S., Cronin-Golomb, M., Mirkin, C. A., and Kaplan, D. L. (2001) Surface organization and nanopatterning of collagen by dip-pen nanolithography. Proc. Natl. Acad. Sci. USA 98, 13660–13664. 111. Zhang, G. J., Tanii, T., Zako, T., Hosaka, T., Miyake, T., Kanari, Y., Funatsu, T. W., and Ohdomari, I. (2005) Nanoscale patterning of protein using electron beam lithography of organosilane self-assembled monolayers. Small 1, 833–837. 112. Cherniavskaya, O., Chen, C. J., Heller, E., Sun, E., Provezano, J., Kam, L., Hone, J., Sheetz, M. P., et al. (2005) Fabrication and surface chemistry of nanoscale bioarrays designed for the study of cytoskeletal protein binding interactions and their effect on cell motility. J. Vac. Sci. Technol. B 23, 2972–2978. 113. Lussi, J. W., Tang, C., Kuenzi, P. A., Staufer, U., Csucs, G., Voros, J., Danuser, G., Hubbell, J. A., et al. (2005) Selective molecular assembly patterning at the nanoscale: a novel platform for producing protein patterns by electron-beam lithography on SiO2/indium tin oxide-coated glass substrates. Nanotechnology 16, 1781–1786. 114. Kim, P., Kim, D. H., Kim, B., Choi, S. K., Lee, S. H., Khademhosseini, A., Langer, R., and Suh, K. Y. (2005) Fabrication of nanostructures of polyethylene glycol for applications to protein adsorption and cell adhesion. Nanotechnology 16, 2420–2426. 115. Cheng, L. J., Kao, M. T., Meyhofer, E., and Guo, L. J. (2005) Highly efficient guiding of microtubule transport with imprinted CYTOP nanotracks. Small 1, 409–414. 116. Cai, Y. G. and Ocko, B. M. (2005) Large-scale fabrication of protein nanoarrays based on nanosphere lithography. Langmuir 21, 9274–9279. 117. Michel, R., Reviakine, I., Sutherland, D., Fokas, C., Csucs, G., Danuser, G., Spencer, N. D., and Textor, M. (2002) A novel approach to produce biologically relevant chemical patterns at the nanometer scale: selective molecular assembly patterning combined with colloidal lithography. Langmuir 18, 8580–8586. 118. Michel, R., Lussi, J. W., Csucs, G., Reviakine, I., Danuser, G., Ketterer, B., Hubbell, J. A., Textor, M., et al. (2002) Selective molecular assembly patterning: a new approach to micro- and nanochemical patterning of surfaces for biological applications. Langmuir 18, 3281–3287. 119. Karuri, N., Liliensiek, S., Teixeira, A., Abrams, G., Campbell, S., Nealey, P., and Murphy, C. (2004) Biological length scale topography enhances cell-substratum adhesion of human corneal epithelial cells. J. Cell Sci. 117, 3153–3164. 120. Kim, D., Kim, P., Song, I., Cha, J., Lee, S., Kim, B., and Suh, K. Y. (2006) Guided three-dimensional growth of functional cardiomyocytes on polyethylene glycol nanostructures. Langmuir 22, 5419–5426. 121. Suh, K. Y. and Jon, S. (2005) Control over wettability of polyethylene glycol surfaces using capillary lithography. Langmuir 21, 6836–6841. 122. Charest, J., Eliason, M., Garcia, A., and King, W. (2006) Combined microscale mechanical ttopography and chemical patterns on polymer cell culture substrates. Biomaterials 27, 2487–2494. 123. Rajnicek, A. M., Britland, S., and McCaig, C. D. (1997) Contact guidance of CNS neurites on grooved quartz: influence of groove dimensions, neuronal age and cell type. J. Cell Sci. 110, 2905–2913.

P1: OTA/XYZ P2: ABC c13 JWBK252/Lee September 20, 2008

15:30

Printer Name: Yet to Come

REFERENCES

357

124. Oakley, C. and Brunette, D. M. (1993) The sequence of alignment of microtubules, focal contacts and actin-filaments in fibroblasts spreading on smooth and grooved titanium substrata. J. Cell Sci. 106, 343–354. 125. Teixeira, A. I., Abrams, G. A., Bertics, P. J., Murphy, C. J., and Nealey, P. F. (2003) Epithelial contact guidance on well-defined micro- and nanostructured substrates. J. Cell Sci. 116, 1881–1892. 126. Teixeira, A. I., McKie, G. A., Foley, J. D., Bertics, P. J., Nealey, P. F., and Murphy, C. J. (2006) The effect of environmental factors on the response of human corneal epithelial cells to nanoscale substrate topography. Biomaterials 27, 3945–3954.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14 MICRO TOTAL ANALYSIS SYSTEM Yuki Tanaka and Takehiko Kitamori

14.1 INTRODUCTION Microchips (Figure 14.1) or microfluidic devices for chemical and biochemical analyses have been greatly developed owing to the progress of microfabrication techniques. Microchemical systems using these devices have attracted much attention of scientists and engineers. This new field of chemistry is known by the name of micro total analysis systems (µ-TAS), labs-on-a-chip, or integrated chemistry lab. As expressed by the name, the concept of these microchip-based systems proposes the integration of various chemical operations involved in conventional analytical processes done in a laboratory, such as mixing, reaction, and separation, into a miniaturized flow system. In this chapter, we describe brief backgrounds, fundamentals of microchip chemistry, and key technologies and then show some major applications and also recent applications focused on surface modification and patterning.

14.1.1 Historical Backgrounds Historically, in 1979, Terry et al. [1] at Stanford University reported an integrated gas chromatography system on a silicon wafer. Their gas chromatography system was fabricated utilizing microfabrication techniques developed in the semiconductor industry. The system has a sample injector, separation column, and thermal conductivity detector, which can be regarded as the first µ-TAS. However, the fabrication Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


359

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

360

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

Figure 14.1. Microchip, glass made, with three layers.

technique was too refined to utilize in chemistry readily. In 1993, A. Manz and his co-workers reported a simple capillary electrophoresis system on a glass microchip [2]. Their microchip was fabricated on a glass substrate, which is easier to use than silicon wafers, utilizing photolithography and wet etching and they demonstrated electrophoresis of amino acids in this device. After this, microchip electrophoresis system has been investigated by many research groups and some systems have been commercially marketed. For example, J. M. Ramsey’s group in Oak Ridge National Laboratory published many papers on microchip electrophoresis. They demonstrated the injection scheme utilizing a channel pattern [3], an optimization method of separation [4], open-channel electrochromatography [5], postcolumn reactions [6], precolumn reactions [7], sample stacking [8], micellar electrokinetic chromatography [9], DNA fragment analysis [10], connection of a microchip and electrospray [11], and so on. D. J. Harrison’s group in University of Alberta also investigated microchip electrophoresis and demonstrated a microchip system connected with mass spectroscopy instrumentation [12, 13]. For some years, the trend of studies describing microchip-based analytical systems had concerned DNA analysis by microchip electrophoresis with laser-induced fluorescence detection. These microchip-based electrophoretic systems have great advantages in some applications, especially in clinical diagnosis and molecular biology fields. Although they are very useful in some specific fields, other analytical methods are required for various applications, which involve several chemical processes. The potential of miniaturized systems is not restricted for microchip electrophoresis. By utilizing simple pressure-driven flow, important chemical processes can be integrated onto the microchip. Manz et al. [14] integrated the polymerase chain

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14.2 FUNDAMENTALS ON MICROCHIP CHEMISTRY

361

reaction, which amplifies DNA fragments 2n times by n times temperature control. Separation based on diffusion constant difference was demonstrated by Yager et al. [15] and chemical reactions under pressure-driven flows were also demonstrated by Whitesides et al. [16]. Only aqueous solutions were used in these applications; however, an important advantage of pressure-driven flow is that solvent is not restricted to water alone, organic solvents can also be applied. Researchers in µ-TAS have focused mainly on applications of micro systems to high performance analytical devices, or sensors or actuators, which have been valid and effective from an industrial viewpoint. Nowadays, techniques utilizing the microspace have been applied more widely in chemistry, to include organic chemistry, physical chemistry, and cell biochemistry.

14.2 FUNDAMENTALS ON MICROCHIP CHEMISTRY Miniaturization of analytical instruments has been attracting much attention. In a common approach, downscaling of individual components comprising an instrument is performed at first, followed by reconstruction of the whole instrument from these components. However, in most cases, this approach leads to poorer instrument performance because it is generally very difficult to realize miniaturization and high performance simultaneously just by simple downscaling. One of the conditions of success in this field of research is based on knowledge about chemistry and biochemistry in a microspace, especially particular properties of microspace. By utilizing such properties, microfluidic chemistry systems would have higher performance compared with macrosystems. For example, chip-based capillary electrophoresis (CE) offers excellent analysis speed compared to conventional CE because of small quantity of sample injection and heat transfer efficiency [17–19]. Here are some fundamentals should be counted. 14.2.1 Characteristics of Liquid Microspace A liquid microspace has several characteristic features different from the bulk scale, for example, short diffusion distances, high interface-to-volume ratio (specific interface area; solid–liquid or liquid–liquid), and small heat capacity. These characteristics in the microspace are key to controlling chemical unit operations, such as mixing, reaction, extraction and separation, and constructing the integrated chemical systems. Especially, to control molecular transport in a microspace, such as microchips, the molecular transportation time and the specific interface area must be considered. Figure 14.2 shows the scale dependence of the molecular transport time and the specific interface area. The transport time is proportional to the square of the scale. Therefore, the transport time takes from several hours to 1 day when the diffusion distance is 1 cm, since the diffusion coefficient of typical molecular ions is on the order of 10−5 cm2 s–1 . In contrast with that case, it takes only several tens of seconds when the diffusion distance is 100 µm. The specific interface area of the 100-µm-scale microspace is

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

3 S V

L−1 2

2 t=

L2 D

1

1

0 0

100

200

300

400

Transportation time (100 s)

Specific interface area (100 cm−1)

362

14:8

0 500

Scale L (µm) Figure 14.2. Specific interface area and molecular transportation time in microspace.

equivalent to that provided by using a separatory funnel with rather vigorous mechanical shaking. These kinds of scale merits become remarkable below a scale of about 250 µm. 14.2.2 Liquid Handling Not only molecular behavior, characteristics of fluids, such as liquids or gases are also different in microspace. The Reynolds number, the ratio of inertial forces to viscous forces, becomes extremely low inside microfluidic devices. Indeed, the laminar flows can be easily formed inside microfluidic device along the flow allowing to achieve multilayer flow. Moreover, completely novel applications without a macroscopic equivalent have recently been developed. The initial microfluidic chip has microchannel sized approximately 100 µm. The total volume of the microchannel network is below 1 µL; thus all the fluids will be manipulated with precise syringe pumps and on chip valves. 14.2.3 Concepts of Micro Unit Operation and Continuous-Flow Chemical Processing For conventional scale chemical systems, such as analysis or synthesis, a system is designed or constructed by combining unit operations. In addition to downscaling of individual components, design and construction of a microfluidic chemical device is also done by properly combining unit operations, such as mixers, reactors. [20]. The difference is that unit operations in microfluidic devices are not mere miniatures of large-scale ones. Simple miniaturization of a conventional unit operation is not effective and sometimes does not work at all, because many physical parameters, for example, heat and mass transfer efficiencies or size of the specific interfacial area,

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14.2 FUNDAMENTALS ON MICROCHIP CHEMISTRY

Molecular transport solvent extraction

Mixing and reaction

363

Molecular capture solid state extraction

Aq. Or.

Phase separation

Heating

Phase contact

Aq. Or.

Concentration

Cell culture

Gas

Figure 14.3. Examples of micro unit operations (MUOs).

are significantly different in microspaces [20–22]. It is necessary to create new operations, namely, micro unit operations (MUOs), considering these characteristics of microspaces to realize equivalent functions in microspaces. As described above, in CE, efficient heat release through capillary walls allows application of a higher voltage without unwanted temperature raise, leading to rapid separation of samples [17]. In similar ways, characteristics of microspaces make chemical operations more efficient than conventional scale ones by using MUOs. A variety of fundamental MUOs, such as mixing and reaction [23–25], two- and three-phase formations [20, 26–34], solvent extraction [20, 26–33], solid-phase extraction, and reaction on surfaces [35–37], heating [38, 39], and cell culture [40] have been successfully developed (Figure 14.3). These MUOs were based on pressure-driven flow control, which allows a wider choice of fluids than electroosmotic flow (EOF). Many organic solvents and even gaseous samples can be treated in the pressure-driven approach [21]. Low Reynolds number is one of the characteristics of microchannels, resulting in the tendency for flow in a microchannel to be laminar flow rather than turbulent flow [15, 16, 21, 41–43]. In addition, a stabilization technique of laminar flows has been demonstrated by Kitamori by using microchannels with special cross-sections [20]. In this technique, ridge-like structures at the bottom of the channel along streamlines were fabricated (Figure 14.4) and the ridges acted as guides for the stream. With the aid of stabilized multiphase laminar flows (MPLFs), elemental unit operations that could hardly be carried out in EOF-driven systems, such as solvent extraction from aqueous to organic phase (Figure 14.5) could be realized. MPLF is also useful for system construction [20]. MUOs can be connected to a continuously flowing MPLF network. Molecules are transported by spontaneous

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

364

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

Figure 14.4. Microchannel with ridge-like structures.

motion, namely, diffusion and distribution among different liquids. It is not necessary to use electric field forces to control motion of the molecules; thus a variety of chemical species can be handled regardless of their charge. Once the channel circuits were properly designed, chemical species will be conveyed from one MUO to another, and sequential chemical processes can be automatically carried out. This

Sample

Metal ion

Chelating reagents

Metal chelate Organic solvent Aqueous phase Aqueous phase Organic phase

100 µm

Figure 14.5. Solvent extraction system of metal chelate in microscale.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14.3 KEY TECHNOLOGIES

365

system construction methodology is called continuous flow chemical processing and integration of relatively complex chemical systems has been accomplished with this methodology.

14.3 KEY TECHNOLOGIES 14.3.1 Fabrication of Microchips In general, a microchip is made from a glass plate, a silicon wafer, poly (dimethilsiloxane) (PDMS), poly(methylmethacrylate), other polymers, or their combination. They all have cons and pros, so it is necessary to choose a suitable material. More recently, polymer microchips have been increasing popularity due to the convenience of fabrication processes. PDMS has become the cold standard for rapid prototyping of microfluidic systems in academia since it is inexpensive, easy to replicate by molding, and optically transparent between 240 and 1100 nm [44]. Other attractiveness of PDMS is to be easy to seal by irreversible bonding, usually with plasma treatment, to other materials made of, for example, PDMS, glass, or silicon [45]. The elastic nature of PDMS is also advantageous for the production of pumps and valve functions [46], and it has allowed fabrication of some of the most complex microfluidic systems realized to date [47, 48]. Because of the chemical and physical stability and optical transparency for detection, a glass microchip has been a mainstream from early days. For glass microchips, a variety of microfabrication methods, such as photolithography/wet etching, laser fabrication, sandblasting, reactive ion etching, and fast atomic-beam fabrication were applied for preparing microchips depending on the microchannel sizes. However, the most commonly used fabrication method for making glass microchips is the photolithography/wet etching method. An outline of a protocol from our group is shown in Figure 14.6. A mechanically polished 0.7-mmthick Pyrex glass plate was the substrate (bottom plate). The substrate plates were annealed at 570◦ C for 5 h before use. For good contact between the substrates and the photoresist and protection of the substrates during glass etching, 20-nm-thick Cr and 100-nm-thick Au layers were evaporatively deposited on the substrates under a vacuum. A 2-µm-thick positive photoresist was spin coated on the metal and baked at 90◦ C for 30 min. Ultraviolet (UV) light was exposed through a photomask by using a mask aligner to transfer the microchannel pattern onto the photoresist. The photoresist was developed and a pattern with 10-µm-wide lines was obtained. The Au and Cr layers were etched with I2 /NH4 I and Ce(NH4 )2 (NO3 )6 solutions. The bare glass surface with the microchannel pattern was etched with a 50% HF solution at an etching rate of 13 µm min–1 . After glass etching, the remaining photoresist was removed in acetone and metals were removed in I2 /NH4 I and Ce(NH4 )2 (NO3 )6 solutions. Inlet and outlet holes were drilled by ultrasonic sandblasting on another Pyrex glass substrate (cover plate). The cover and etched plates were thermally laminated in a furnace at 650◦ C for 5 h after washing in an ethanol and NaOH solution.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

366

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

Glass substrate

Develop

Deposit metal and photoresist

Etch metal layer

UV irradiation with photomask

Etch glass

Remove metal and photoresist Figure 14.6. Fabrication method for glass microchips.

14.3.2 Patterning for Fluid Control Hydrophilic/hydrophobic patterning of microchannel surface is a very effective way to stabilize an aqueous–organic interface [29, 49–52]. We have recently developed the capillarity restricted modification method in which a microchannel structure with a hydrophilic deep lane and a hydrophobic shallow lane could be made easily [49]. A microchannel with deep and shallow lanes was fabricated using the two-step photolithographic wet etching technique illustrated in Figure 14.6. There are some methods to make microchannel wall hydrophobic (Figure 14.7), like physical adsorption of fluoropolymer or silane coupling. In this application, a surface-modification solution (octadecyltrichlorosilane (ODS)/toluene) was introduced in a restricted way to the shallow lane of a microchannel by utilizing capillarity. After flushing out the modification solution with air pressure and rinsing with pure solvent, a shallow lane modified with hydrophobic ODS groups was obtained, while the deep lane remained as a bare glass surface. This microchannel structure was successfully applied to stabilization of gas–water two-phase flow and it was also applied to removal of bubbles. The capability of the surface-modified microchannels to keep the organic–aqueous interfaces stable against a pressure difference was sufficient to realize counterflow [53]. Countercurrent flow is efficient for solvent extractions that have small distribution coefficients and it provides high throughput processing (Figure 14.8). 14.3.3 Detection Detection is one of the most difficult problems in microchemical processes. Since sample volume becomes extremely small in such systems, an ultrasensitive detection method is indispensable. For example, limited sample thickness causes a very

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14.3 KEY TECHNOLOGIES

SiO2

367

OH OH OH

Bare glass wall of microchannel

OH F F OH OH F F OH OH F F OH F F OH OH F OH F

O O Si O O O Si O O O Si O

Physical adsorption

Silane coupling

Figure 14.7. Hydrophobic surface modification on microchannel surface.

small signal-to-background ratio in absorption spectroscopy, and only very concentrated samples can be analyzed. In laser-induced fluorescence (LIF), the problem is solved by using high intensity laser light and fluorimetry [19, 54, 55]. Since the wavelength of fluorescent light is slightly longer than that of excitation light, the signal intensity can be measured directly rather than calculating from a small difference between incident and transmitted beams. Due to its high sensitivity, LIF can be used even for single-molecule detection [55]. However, fluorescence spectroscopy has a serious drawback, i.e., almost all molecules would release the energy upon excitation as heat rather than fluorescent light and LIF is only applicable to fluorescent molecules. Therefore, a powerful detection scheme having both high sensitivity, like LIF, and high applicability, like UV/Vis, is desirable. Thermal lens microscope seems to be suitable for the detection scheme. Sato et al. [56] demonstrated thermal lens detection of Fe2+ chelating reaction with o-phenanthroline in a microchip and measured a 1 µM Fe2+ solution, which corresponded to an absolute amount of 6 zmol. Although excellent detection ability was demonstrated, actual flow analysis in the microchip was not discussed. In order to establish the thermal lens detection

Cocurrent extraction

Countercurrent extraction

Organic phase

Organic phase High recovery

Aqueous phase

Aqueous phase

Figure 14.8. Cocurrent and countercurrent extraction.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

368

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

method for the microchip analysis, chemical analysis processes including chemical kinetics and stoichiometry should be discussed quantitatively.

14.4 APPLICATIONS 14.4.1 Synthesis Chemical syntheses using microchemical chips have been intensively investigated in recent years. The prospect that microchemical chips or microflow reactors might bring benefits hardly obtainable using conventional scale reactors has been extensively argued in the last decade [22, 57–64]; even text books focusing solely on this subject have been published [22, 64]. A microchemical chip has microchannels with dimensions of typically ten to a few hundreds micrometers, and chemical reactions are carried out in the microchannels. When characteristics such as a large surface-to-volume ratio, short diffusion length, and small heat capacity of liquids in the microspaces are properly utilized, process intensification in chemical syntheses can be realized. Formation of hot spots and occurrence of runaway reactions can be avoided by the good heat exchange of the microchannel with a high surfaceto-volume ratio. With the small heat capacity of microchannels, reaction temperature can be controlled rapidly and precisely [22, 39, 64]. Mixing of reagents in microchannels is quite different from mixing in larger reaction vessels because flow with low Reynolds number tends to be laminar rather than turbulent [16, 22, 42, 64, 65], and often more controlled mixing can be carried out than in large reaction vessels. Reaction conditions in microreactors that are more rigorously controlled than those in conventional reactors, or in some cases, not accessible using conventional reactors, lead to higher reaction yields and selectivity in many cases. Furthermore, safe running of some reactions that had been considered too dangerous to be carried out has become possible using microreactors [22, 64, 66–70]. On-site production of highly toxic chemicals using microchemical chips is expected to be useful in avoiding the risks inherent in transportation. Whereas production adjustment of a chemical plant has been normally done by scaling-up, production adjustment of a microchemical plant may possibly be carried out by numbering-up. Scaling-up is a time- and cost-consuming process in which heat- and mass-transfer problems have to be considered repeatedly for all testing plants. Flexible adjustment of production would become possible with the numbering-up concept, by simply increasing and decreasing the number of reactors [22, 57–59, 64, 71, 72]. Multiphase catalytic reactions are also very useful application of microchip synthesis. Multiphase catalytic reactions play important roles not only in the research laboratory but also in the chemical and pharmaceutical industries [73]. They are classified according to the phases involved, such as gas–liquid, gas–liquid–liquid, or gas–liquid–solid reactions. Although numerous multiphase catalytic reactions are known and many are used in industry, these reactions are still difficult to conduct when compared to homogeneous reactions, because the efficiency of interaction and mass transfer between

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14.4 APPLICATIONS

369

SiO2

Gas phase

Solid catalyst

Liquid phase

Figure 14.9. Cross-sectional view of a mirochannel for multiphase reactions.

different phases is extremely low, and thus in most cases the reaction rates are slow. In general, to accelerate multiphase catalytic reactions, some treatment producing high interfacial area between the two or three reacting phases, such as vigorous stirring or additional equipment, is needed and the development of more effective, simple devices that can produce such a high interfacial area between different phases is a much sought after goal. As mentioned previously, specific interfacial area per volume in a microchip is extremely large. In concrete figures, this area rises to 10,000–50,000 m2 m–3 while the value is only 100 m2 m–3 in conventional reactors. By utilizing this feature, triphase reaction system, which is very difficult to control in conventional method, can be developed. Provided that the flow is well controlled, it should be possible to pass the gas through the center of the channel and the liquid along the inner surface of the channel at a particular gas pressure (Figure 14.9) [68, 74]. If the wall of the microchannel is covered with a solid catalyst, efficient gas–liquid–solid reactions might occur, because effective interaction of the three phases is expected because of the extremely large interfacial areas and the short path required for molecular diffusion in the very narrow channel space. An efficient microchannel triphase reactor has been shown by Kobayashi group at the University of Tokyo. A glass microchip has been selected and Pd catalyst was immobilized on the microchannel wall. This system has been used for hydrogenation reactions and deprotection reactions and resulted in high efficiency [75]. 14.4.2 Cell Adhesion Control Handling cells in microchips is of great interest. They are suited for cell experiments since the physical size of channels is commensurate with that of the cells. Microchips have good potential to serve as new devices for cell handling and cellular analysis. For instance, in a microchamber fabricated on a microchip, exchange of the medium or reagents can be achieved by simple operations under continuous medium flows, and mechanical operations or handling procedures necessary for bioassays are simplified by integration into a microchip. Some papers have reported cell analyses using microchips [76–79]. Moreover, rapid analyses of cellular released compounds were realized by use of flow-based analysis chips [80–82].

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

370

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

For cell-based studies, the culture environment is very important to maintain cell conditions. One of the most important factors is the surface for cell adhesion. Cell adhesion plays a critical role in adherent cells because they trigger signal transduction inside the cells, while affecting cellular growth, proliferation, and differentiation. The interactions between cells and surfaces have been the focus of studies in biological and medical research fields and many biocompatible materials having cell attachable or detachable properties have been developed. The control of cell adhesion and micropatterning of cells or proteins using cell–surface interactions has been applied to cell array and coculture systems in order to advance such fields as drug discovery and tissue engineering [83, 84]. Because cells can be attached to the hydrophobic surface via extracellular matrix proteins, hydrophobic modification is a popular method for cell patterning. Many studies on the interactions between cells and surfaces have been reported. Some studies concentrate on surfaces patterned at the subcellular-scale (10–1000 nm), the size reaching that of a focal adhesion. Focal adhesion plays a pivotal role in cell adhesion and signal transduction, and subcellular-scale patterning should allow control of the morphology, proliferation, and differentiation of a cell [85]. In order to investigate cell–surface interactions precisely and to control cell adhesion, it is necessary to develop surfaces with patterns much smaller than an individual cell size, i.e., nanopatterned surfaces. Nanopatterned surfaces for cell attachment have been fabricated by colloidal lithography [86], polymer demixing [87], and copolymer formation [88]. These methods provide surfaces that have nanometer-scale topography, but precise control of the scale and the shape of the patterns are very difficult. Electron-beam lithography and a dry etching process can control the scale and the shape of the patterns precisely on silicon substrates [89, 90]. Cells on the surface can be stimulated by the nanometerscale topography and cells can be aligned along line-and-space patterns. However, it is difficult to make a pattern that can provide selected chemical properties on the surface. Thus, a microchip with cell adhesion surface, controlled by both nanometer-scale topography and chemical patterning, has been shown using semiconductor fabrication methods and the formation of self-assembled monolayers [91, 92]. The patterned surface had a sharp contrast between the adsorption and nonadsorption of proteins and cells, and the contrast could be maintained for more than 10 days. The patterning method could easily realize a single cell array and control of the cell morphology. The nanometer-scale patterned surface could control cell adhesion and proliferation. The combination of the patterned surface and the microchip will contribute to studies about cell–surface interactions. 14.4.3 Liquid Handling: Valve Using Wettability Microfluidic control is quite important technique, which is still in developing phase, particularly that provided by valves in microchannels. A number of miniaturized mechanical valves fabricated by micromachining technologies have been developed. Pneumatic-controlled valves, using soft material, have also been reported [46]. All of these valves require fabrication and construction of mechanical moving parts or

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

14.4 APPLICATIONS

371

external control parts in the channels. Alternatively, reports of nonmechanical methods utilizing phase transition of solution [93], a thermoresponsive polymer [94], or a hydrogel [95] have been made. Using surface properties, particularly wettability, is another approach to microfluid control. In a microchannel, a microfluid is strongly affected by the surface wettability due to the large specific interface. For example, it is easy to introduce water into a glass tube having a hydrophilic surface while introduction of water into a Teflon tube requires additional pressure due to its hydrophobicity. Many research groups have reported fluid control methods using wettabilitypatterned surfaces in microchannels [29, 49–51, 96–106]. The most common way to control fluid with wettability is to make stop valve. The pressure barriers originating in the Laplace pressure can be generated at the boundary of the wettability change in the channel. These pressure barriers work as stop valves. A variety of papers using wettability-based Laplace valves have already been reported. For example, electrowetting devices in which wettability can be controlled by an applied voltage and thus they can act as an active valve, can be applied [98–100]. However, the wettability-controlled surfaces can only be fabricated on electrode surfaces. By using an appropriate wettability patterning method, on the other hand, the wettability-based valve can be constructed at a desired position and the magnitude of the pressure barrier can be tuned, even after channel fabrication. Unlike a flat substrate, however, the wettability patterning and the tuning inside the microchannel are limited. Photopatterning using photoresponsive molecules is one promising method. As examples of wettability-based Laplace valves, some groups reported pressure-controllable microfluidic gates using photocleavable hydrophobic molecules [50, 51, 107] or the reversible polarity change of photochromic molecules [108]. In all cases, the channel surfaces were modified with the photoresponsive molecules and then wettability was patterned in situ by photoirradiation through appropriate masks. The position of the valve is determined by the mask design. Since the degree of the photoreaction can be controlled by light intensity or irradiation time, the wettability can be tuned with controlled photoirradiation. The wettability changing range is wide enough to control, indeed, further improvement is required. Takei et al. [106] have shown passive stop valves by using characteristics of titania nanoparticles. Titania modification of a microchannel provided a nanometersized surface roughness and the subsequent hydrophobic treatment made the surface superhydrophobic [109]. Photocatalytic decomposition of the coated hydrophobic molecules was used to pattern the surface wettability, which was tuned in the range from superhydrophobic to superhydrophilic under controlled photoirradiation [110]. This method provides flexible patterning in a wide range of tuned wettability surfaces in microchannels even after channel fabrication and it can be applied to various two- or multiphase microfluidic systems. We described brief backgrounds, fundamentals of microchip chemistry, and key technologies and then show some recent applications focused on surface modification and patterning. This µ-TAS field is ongoing so actively that the applications shown above are only a drop in bucket. For more applications, there are many reviews on µ-TAS; for example, an overview review on µ-TAS is published every 2 years from Analytical Chemistry—the most recent one is [111].

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

372

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

As mentioned many times in this chapter, the surface conditions or modifications of microchip affect the system inside much more than those of conventional scale containers. By combining surface modification, including nanopatterning, micro systems are promising as flexible, smart, and mobile advanced chemical systems.

REFERENCES 1. Terry, S. C., Jerman, J. H., and Angell, J. B. (1979) A gas chromatographic air analyzer fabricated on a silicon wafer. IEEE Trans. Electron Devices ED-26, 1880–1886. 2. Harrison, D. J., Fluri, K., Seiler, K., Fan, Z. H., Effenhauser, C. S., and Manz, A. (1993) Micromachining a miniaturized capillary electrophoresis-based chemical-analysis system on a chip. Science 261, 895–897. 3. Jacobson, S. C., Hergenroder, 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 devices. Anal. Chem. 66, 1107–1113. 4. Jacobson, S. C., Hergenroder, R., Koutny, L. B., and Ramsey, J. M. (1994) High-speed separations on a microchip. Anal. Chem. 66, 1114–1118. 5. Jacobson, S. C., Hergenroder, R., Koutny, L. B., and Ramsey, J. M. (1994) Open-channel electrochromatography on a microchip. Anal. Chem. 66, 2369–2373. 6. Jacobson, S. C., Koutny, L. B., Hergenroder, R., Moore, A. W., and Ramsey, J. M. (1994) Microchip capillary electrophoresis with an integrated postcolumn reactor. Anal. Chem. 66, 3472–3476. 7. Jacobson, S. C., Hergenroder, R., Moore, A. W., and Ramsey, J. M. (1994) Precolumn reactions with electrophoretic analysis integrated on a microchip. Anal. Chem. 66, 4127–4132. 8. Jacobson, S. C. and Ramsey, J. M. (1995) Microchip electrophoresis with sample stacking. Electrophoresis 16, 481–486. 9. Moore, A. W., Jacobson, S. C., and Ramsey, J. M. (1995) Microchip separations of neutral species via micellar electrokinetic capillary chromatography. Anal. Chem. 67, 4184–4189. 10. Jacobson, S. C. and Ramsey, J. M. (1996) Integrated microdevice for DNA restriction fragment analysis. Anal. Chem. 68, 720–723. 11. Ramsey, R. S. and Ramsey, J. M. (1997) Generating electrospray from microchip devices using electroosmotic pumping. Anal. Chem. 69, 1174–1178. 12. Bings, N. H., Wang, C., Skinner, C. D., Colyer, C. L., Thibault, P., and Harrison, D. J. (1999) Microfluidic devises connected to fused-silica capillaries with minimal dead volume. Anal. Chem. 71, 3292–3296. 13. Oleschuk, R. D. and Harrison, D. J. (2000) Analytical microdevices for mass spectrometry. TRAC Trends Anal. Chem. 19, 379–388. 14. Kopp, M. U., de Mello, A. J., and Manz, A. (1998) Chemical amplification: continuousflow PCR on a chip. Science 280, 1046–1048. 15. Weigl, B. H. and Yager, P. (1999) Microfluidics-microfluidic diffusion-based separation and detection. Science 283, 346–347.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

REFERENCES

373

16. Kenis, P. J. A., Ismagilov, R. F., and Whitesides, G. M. (1999) Microfabrication inside capillaries using multiphase laminar flow patterning. Science 285, 83–85. 17. Landers, J. P., ed. (1996) Handbook of Capillary Electrophoresis, CRC Press, Boca Raton, FL. 18. Bruin, G. J. M. (2000) Recent developments in electrokinetically driven analysis on microfabricated devices. Electrophoresis 21, 3931–3951. 19. Ambrose, W. P., Goodwin, P. M., Jett, J. H., Orden, A. V., Werner, J. H., and Keller, R. A. (1999) Single molecule fluorescence spectroscopy at ambient temperature. Chem. Rev. 99, 2929–2956. 20. Tokeshi, M., Minagawa, T., Uchiyama, K., Hibara, A., Sato, K., Hisamoto, H., and Kitamori, T. (2002) Continuous-flow chemical processing on a microchip by combining microunit operations and a multiphase flow network. Anal. Chem. 74, 1565–1571. 21. Hibara, A., Tokeshi, M., Uchiyama, K., Hisamoto, H., and Kitamori, T. (2001) Integrated multilayer flow system on a microchip. Anal. Sci. 17, 89–93. 22. Ehfeld, W., Hessel, V., and Loewe, H. (2000) Microreactors: New Technology for Modern Chemistry, Wiley-VCH, Weinheim. 23. Sato, K., Tokeshi, M., Kitamori, T., and Sawada, T. (1999) Integration of flow injection analysis and zeptomole-level detection of the Fe(II)–o–phenanthroline complex. Anal. Sci. 15, 641–645. 24. Sorouraddin, H. M., Hibara, A., Proskurnin, M. A., and Kitamori, T. (2000) Integrated FIA for the determination of ascorbic acid and dehydroascorbic acid in a microfabricated glass-channel by thermal-lens microscopy. Anal. Sci. 16, 1033–1037. 25. Sorouraddin, H. M., Hibara, A., and Kitamori, T. (2001) Use of a thermal lens microscope in integrated catecholamine determination on a microchip. Fresenius’ J. Anal. Chem. 371, 91–96. 26. Surmeian, M., Hibara, A., Slyadnev, M., Uchiyama, K., Hisamoto, H., and Kitamori, T. (2001) Distribution of methyl red on the water-organic liquid interface in a microchannel. Anal. Lett. 34, 1421–1429. 27. Hisamoto, H., Horiuchi, T., Tokeshi, M., Hibara, A., and Kitamori, T. (2001) On-chip integration of neutral ionophore-based ion pair extraction reaction. Anal. Chem. 73, 1382–1386. 28. Hisamoto, H., Horiuchi, T., Uchiyama, K., Tokeshi, M., Hibara, A., and Kitamori, T. (2001) On-chip integration of sequential ion-sensing system based on intermittent reagent pumping and formation of two-layer flow. Anal. Chem. 73, 5551– 5556. 29. Hibara, A., Nonaka, M., Hisamoto, H., Uchiyama, K., Kikutani, Y., Tokeshi, M., and Kitamori, T. (2002) Stabilization of liquid interface and control of two-phase confluence and separation in glass microchips by utilizing octadecylsilane modification of microchannels. Anal. Chem. 74, 1724–1728. 30. Tokeshi, M., Minagawa, T., and Kitamori, T. (2000) Integration of a microextraction system on a glass chip: ion-pair solvent extraction of Fe(II) with 4,7-diphenyl-1,10phenanthrolinedisulfonic acid and tri-n-octylmethylammonium chloride. Anal. Chem. 72, 1711–1714. 31. Sato, K., Tokeshi, M., Sawada, T., and Kitamori, T. (2000) Molecular transport between two phases in a microchannel. Anal. Sci. 16, 455–456.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

374

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

32. Tokeshi, M., Minagawa, T., and Kitamori, T. (2000) Integration of a microextraction system solvent extraction of a co–2-nitroso-5-dimethylaminophenol complex on a microchip. J. Chromatogr. A 894, 19–23. 33. Minagawa, T., Tokeshi, M., and Kitamori, T. (2001) Integration of a wet analysis system on a glass chip: determination of Co(II) as 2-nitroso-1-naphthol chelates by solvent extraction and thermal lens microscopy. Lab Chip 1, 72–75. 34. Hisamoto, H., Shimizu, Y., Uchiyama, K., Tokeshi, M., Kikutani, Y., Hibara, A., and Kitamori, T. (2003) Chemicofunctional membrane for integrated chemical processes on a microchip. Anal. Chem. 75, 350–354. 35. Sato, K., Tokeshi, M., Odake, T., Kimura, H., Ooi, T., Nakao, M., and Kitamori, T. (2000) Integration of an immunosorbent assay system: analysis of secretory human immunoglobulin a on polystyrene beads in a microchip. Anal. Chem. 72, 1144–1147. 36. Sato, K., Tokeshi, M., Kimura, H., and Kitamori, T. (2001) Determination of carcinoembryonic antigen in human sera by integrated bead-bed immunoassay in a microchip for cancer diagnosis. Anal. Chem. 73, 1213–1218. 37. Sato, K., Yamanaka, M., Takahashi, H., Tokeshi, M., Kimura, H., and Kitamori, T. (2002) Microchip-based immunoassay system with branching multichannels for simultaneous determination of interferon-gamma. Electrophoresis 23, 734–739. 38. Tanaka, Y., Slyadnev, M. N., Hibara, A., Tokeshi, M., and Kitamori, T. (2000) Noncontact photothermal control of enzyme reactions on a microchip by using a compact diode laser. J. Chromatogr. A 894, 45–51. 39. Slyadnev, M. N., Tanaka, Y., Tokeshi, M., and Kitamori, T. (2001) Photothermal temperature control of a chemical reaction on a microchip using an infrared diode laser. Anal. Chem. 73, 4037–4044. 40. Tamaki, E., Sato, K., Tokeshi, M., Sato, K., Aihara, M., and Kitamori, T. (2002) Singlecell analysis by a scanning thermal lens microscope with a microchip: direct monitoring of cytochrome c distribution during apoptosis process. Anal. Chem. 74, 1560–1564. 41. Kim, H.-B., Ueno, K., Chiba, M., Kogi, O., and Kitamura, N. (2000) Spatially-resolved fluorescence spectroscopic study on liquid/liquid extraction processes in polymer microchannels. Anal. Sci. 16, 871–876. 42. Kamholz, A. E., Weigl, B. H., Finleyson, B. A., and Yager, P. (1999) Quantitative analysis of molecular interaction in a microfluidic channel: the T-sensor. Anal. Chem. 71, 5340–5347. 43. Brody, J. P. and Yager, P. (1997) Diffusion-based extraction in a microfabricated device. Sensors Actuators A 58, 13–18. 44. McDonald, J. C. and Whitesides, G. M. (2002) Poly(dimethylsiloxane) as a material for fabricating microfluidic devices. Acc. Chem. Res. 35, 491–499. 45. Duffy, D. C., McDonald, J. C., Schueller, O. J. A., and Whitesides, G. M. (1998) Rapid prototyping of microfluidic systems in poly(dimethylsiloxane). Anal. Chem. 70, 4974–4984. 46. Unger, M. A., Chou, H. P., Thorsen, T., Scherer, A., and Quake, S. R. (2000) Monolithic microfabricated valves and pumps by multilayer soft lithography. Science 288, 113–116. 47. Thorsen, T., Maerkl, S. J., and Quake, S. R. (2002) Microfluidic large-scale integration. Science 298, 580–584. 48. Fu, A. Y., Chou, H. P., Spence, C., Arnold, F. H., and Quake, S. R. (2002) An integrated microfabricated cell sorter. Anal. Chem. 74, 2451–2457.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

REFERENCES

375

49. Hibara, A., Iwayama, S., Matsuoka, S., Ueno, M., Kikutani, Y., Tokeshi, M., and Kitamori, T. (2005) Surface modification method of microchannels for gas–liquid twophase flow in microchips. Anal. Chem. 77, 943–947. 50. Zhao, B., Moore, J. S., and Beebe, D. J. (2001) Surface-directed liquid flow inside microchannels. Science 291, 1023–1026. 51. Zhao, B., Viernes, N. O. L., Moore, J. S., and Beebe, D. J. (2002) Control and applications of immiscible liquids in microchannels. J. Am. Chem. Soc. 124, 5284– 5285. 52. Peterson, D. S., Rohr, T., Svec, F., and Frechet, J. (2002) Enzymatic microreactor-on-achip: protein mapping using trypsin immobilized on porous polymer monoliths molded in channels of microfluidic devices. Anal. Chem. 74, 4081–4088. 53. Aota, A., Nonaka, M., Hibara, A., and Kitamori, T. (2003) Final Program and Book of Abstracts of 16th International Symposium on Microscale Separations and Analysis, HPCE2003, San Diego, California, p. 192. 54. Ocvirk, G., Tang, T., and Harrison, D. J. (1998) Optimization of confocal epifluorescence microscopy for microchip-based miniaturized total analysis systems. Analyst 123, 1429–1434. 55. Hill, E. K. and de Mello, A. (2000) Single-molecule detection using confocal fluorescence detection: assessment of optical probe volumes. Analyst 125, 1033–1036. 56. Sato, K., Kawanishi, H., Tokeshi, M., Kitamori, T., and Sawada, T. (1999) Subzeptomole detection in a microfabricated glass channel by thermal-lens microscopy. Anal. Sci. 15, 525–529. 57. Haswell, S. J., Middleton, R. J., O’Sullivan, B., Skelton, V., Watts, P. and Styring, P. (2001) The application of micro reactors to synthetic chemistry. Chem. Commun. 391–398. 58. Jensen, K. F. (2001) Microreaction engineering—is small better?. Chem. Eng. Sci. 56, 293–303. 59. de Mello, A. and Wootton, R. (2002) But what is it good for? Applications of microreactor technology for the fine chemical industry. Lab Chip 2, 7N. 60. Fletcher, P. D. I., Haswell, S. J., Pombo-Villar, E., Warrington, B. H., Watts, P., and Wong, S. Y. F. (2002) Micro reactors: principles and applications in organic synthesis. Tetrahedron 58, 4735–4757. 61. Watts, P. and Haswell, S. J. (2003) Microfluidic combinatorial chemistry. Curr. Opin. Chem. Biol. 7, 380–387. 62. Hisamoto, H., Saito, T., Tokeshi, M., Hibara, A., and Kitamori, T. (2001) Fast and high conversion phase-transfer synthesis exploiting the liquid–liquid interface formed in a microchannel chip. Chem. Commun. 2662–2663. 63. Ueno, M., Hisamoto, H., Kitamori, T., and Kobayashi, S. (2003) Phase-transfer alkylation reactions using microreactors. Chem. Commun. 8, 936. 64. Hessel, V., Hardt, S., and Loewe, H. (2004) Chemical Micro Process Engineering, Wiley-VCH, Weinheim. 65. Nguyen, N.-T. and Wereley, S. T. (2002) Fundamentals and Applications of Microfluidics MEMS Microelectromechanical Systems Series, Artech House, Boston. 66. Chambers, R. D., Holling, D., Spink, R. C. H., and Sandford, G. (2001) Elemental fluorine part 13. Gas–liquid thin film microreactors for selective direct fluorination. Lab Chip 1, 132–137.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

376

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

67. Chambers, R. D., Hutchinson, J., and Sandford, G.J. (1999) Recent studies at Durham on direct fluorination. Fluorine Chem. 100, 63–73. 68. Chambers, R. D. and Spink, R. C. H. (1999) Microreactors for elemental fluorine. Chem. Commun. 10, 883–884. 69. Jahnisch, K., Baerns, M., Hessel, V., Ehrfeld, W., Haverkamp, V., Loewe, H., Wille, C., and Guber, A. (2000) Direct fluorination of toluene using elemental fluorine in gas/liquid microreactors. J. Fluorine Chem. 105, 117–128. 70. Ajmera, S. K., Losey, M. W., and Jensen, K. F. (2001) Reactors, kinetics, and catalysis: microfabricated packed-bed reactor for phosgene synthesis. AIChE J. 47, 1639–1647. 71. Kikutani, Y., Hibara, A., Uchiyama, K., Hisamoto, H., Tokeshi, M., and Kitamori, T. (2002) Pile-up glass microreactor. Lab Chip 2, 193–196. 72. Ehrfeld, W., Golbig, K., Hessel, V., Loewe, H., and Richter, T. (1999) Characterization of mixing in micromixers by a test reaction: single mixing units and mixer arrays. Ind. Eng. Chem. Res. 38, 1075–1082. 73. Mills, P. L. and Chaudhari, R. V. (1997) Multiphase catalytic reactor engineering and design for pharmaceuticals and fine chemicals. Catal. Today 37, 367–404. 74. Serizawa, A., Feng, Z., and Kawara, Z. (2002) Two-phase flow in microchannels. Exp. Therm. Fluid Sci. 26, 703–714. 75. Kobayashi, J., Mori, Y., Okamoto, K., Akiyama, R., Ueno, M., Kitamori, T., and Kobayashi, S. (2004) A microfluidic device for conducting gas-liquid-solid hydrogenation reactions. Science 304, 1305–1308. 76. Takayama, S., Ostuni, E., LeDuc, P., Naruse, K., Ingber, D. E., and Whitesides, G. M. (2001) Subcellular positioning of small molecules. Nature 411, 1016. 77. McClain, M. A., Culbertson, C. T., Jacobson, S. C., Allbritton, N. L., Sims, C. E., and Ramsey, J. M. (2003) Microfluidic devices for the high-throughput chemical analysis of cells. Anal. Chem. 75, 5646–5655. 78. Fujii, S., Uematsu, M., Yabuki, S., Abo, M., Yoshimura, E., and Sato, K. (2006) Microbioassay system for an anti-cancer agent test using animal cells on a microfluidic gradient mixer. Anal. Sci. 22, 87–90. 79. Tanaka, Y., Sato, K., Yamato, M., Okano, T., and Kitamori, T. (2004) Drug response assay system in a microchip using human hepatoma cells. Anal. Sci. 20, 411–423. 80. Goto, M., Sato, K., Murakami, A., Tokeshi, M., and Kitamori, T. (2005) Development of a microchip-based bioassay system using cultured cells. Anal. Chem. 77, 2125–2131. 81. Tokuyama, T., Fujii, S., Sato, K., Abo, M., and Okubo, A. (2005) Microbioassay system for antiallergic drug screening using suspension cells retaining in a poly(dimethylsiloxane) microfluidic device. Anal. Chem. 77, 3309–3314. 82. Sato, K., Egami, A., Odake, T., Tokeshi, M., Aihara, M., and Kitamori, T. (2006) Monitoring of intercellular messengers released from neuron networks cultured in a microchip. J. Chromatogr. A 1111, 228–232. 83. Luk, Y. Y., Kato, M., and Mrksich, M. (2000) Self-assembled monolayers of alkanethiolates presenting mannitol groups are inert to protein adsorption and cell attachment. Langmuir 16, 9604–9608. 84. Liu, V. A., Jastromb, W. E., and Bhatia, S. N. (2002) Engineering protein and cell adhesivity using PEO-terminated triblock polymers. J. Biomed. Mater. Res. 60, 126–134. 85. Curtis, A. and Wilkinson, C. (2001) Nanotechniques and approaches in biotechnology. Trends Biotechnol. 19, 97–101.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

14:8

Printer Name: Yet to Come

REFERENCES

377

86. Dalby, M. J., Berry, C. C., Riehle, M. O., Sutherland, D. S., Agheli, H., and Curtis, A. S. G. (2004) Attempted endocytosis of nano-environment produced by colloidal lithography by human fibroblasts. Exp. Cell. Res. 295, 387–394. 87. Dalby, M. J., Riehle, M. O., Johnstone, H. J. H., Affrossman, S., and Curtis, A. S. G. (2003) Nonadhesive nanotopography: fibroblast response to poly(n-butyl methacrylate)-poly(styrene) demixed surface features. J. Biomed. Mater. Res. A 67A, 1025–1032. 88. Long, S. F., Clarke, S., Davies, M. C., Lewis, A. L., Hanlon, G. W., and Lloyd, A. W. (2003) Controlled biological response on blends of a phosphorylcholine-based copolymer with poly(butyl methacrylate). Biomaterials 24, 4115–4121. 89. Teixeira, A. I., Abrams, G. A., Bertics, P. J., Murphy, C. J., and Nealey, P. F. (2003) Epithelial contact guidance on well-defined micro- and nanostructured substrates. J. Cell Sci. 116, 1881–1892. 90. Teixeira, A. I., Nealey, P. F., and Murphy, C. J. (2004) Responses of human keratocytes to micro- and nanostructured substrates. J. Biomed. Mater. Res. A 71A, 369–376. 91. Goto, M., Tsukahara, T., Sato, K., Konno, T., Ishihara, K., Sato, K., and Kitamori, T. (2007) Nanometer-scale patterned surfaces for control of cell adhesion. Anal. Sci. 23, 245–247. 92. Goto, M., Tsukahara, T., Sato, K., and Kitamori, T. (2008) Micro- and nanometerscale patterned surface in a microchannel for cell culture in microfluidic devices. Anal. Bioanal. Chem. 390, 817–823. 93. Gui, L. and Liu, J. (2004) Ice valve for a mini/micro flow channel. J. Micromech. Microeng. 14, 242–246. 94. Yu, C., Mutlu, S., Selvaganapathy, P., Mastrangelo, C. H., Svec, F., and Frechett, J. M. J. (2003) Flow control valves for analytical microfluidic chips without mechanical parts based on thermally responsive monolithic polymers. Anal. Chem. 75, 1958–1961. 95. Beebe, D. J., Moore, J. S., Bauer, J. M., Yu, Q., Liu, R. H., Devadoss, C., and Jo, B. H. (2000) Functional hydrogel structures for autonomous flow control inside microfluidic channels. Nature 404, 588–590. 96. Huang, T. T., Mosier, N. S., and Ladisch, M. R. (2006) Surface engineering of microchannel walls for protein separation and directed microfluidic flow. J. Sep. Sci. 29, 1733–1742. 97. Huang, T. T., Taylor, D. G., Sedlak, M., Mosier, N. S., and Ladisch M. R. (2005) Microfiber-directed boundary flow in press-fit microdevices fabricated from selfadhesive hydrophobic surfaces. Anal. Chem. 77, 3671–3675. 98. Lee, J. and Kim, C. J. (2000) Surface-tension-driven microactuation based on continuous electrowetting. J. Microelectromech. Syst. 9, 171–180. 99. Huh, D., Tkaczyk, A. H., Bahng, J. H., Chang, Y., Wei, H. H., Grotberg, J. B., Kim, C. J., Kurabayashi, K., et al. (2003) Reversible switching of high-speed air-liquid twophase flows using electrowetting-assisted flow-pattern change. J. Am. Chem. Soc. 125, 14678–14679. 100. Srinivasan, V., Pamula, V. K., and Fair, R. B. (2004) Droplet-based microfluidic lab-ona-chip for glucose detection. Anal. Chim. Acta 507, 145–150. 101. Handique, K., Burke, D. T., Mastrangelo, C. H., and Burns, M. A. (2000) Nanoliter liquid metering in microchannels using hydrophobic patterns. Anal. Chem. 72, 4100–4109.

P1: OTA/XYZ P2: ABC c14 JWBK252/Lee September 18, 2008

378

14:8

Printer Name: Yet to Come

MICRO TOTAL ANALYSIS SYSTEM

102. Andersson, H., Van Der Wijngaart, W., Griss, P., Niklaus, F., and Stemme, G. (2001) Hydrophobic valves of plasma deposited octafluorocyclobutane in DRIE channels. Sensors Actuators B 75, 136–141. 103. Andersson, H., Van Der Wijngaart, W., and Stemme, G. (2001) Micromachined filterchamber array with passive valves for biochemical assays on beads. Electrophoresis 22, 249–257. 104. Lee, S. H., Lee, C. S., Kim, B. G., and Kim, Y. K. (2003) Quantitatively controlled nanoliter liquid manipulation using hydrophobic valving and control of surface wettability. J. Micromech. Microeng. 13, 89–97. 105. Feng, Y. Y., Zhou, Z. Y., Ye, X. Y., and Xiong, H. J. (2003) Passive valves based on hydrophobic microfluidics. Sensors Actuators A 108, 138–143. 106. Takei, G., Nonogi M., Hibara, A., Kitamori, T., and Kim, H.-B. (2007) Tuning microchannel wettability and fabrication of multiple-step laplace valves. Lab Chip 7, 596–602. 107. Besson, E., Gue, A. M., Sudor, J., Korri-Youssoufi, H., Jaffrezic, N., and Tardy, J. (2006) A novel and simplified procedure for patterning hydrophobic and hydrophilic SAMs for microfluidic devices by using UV photolithography. Langmuir 22, 8346–8352. 108. Koide, T., Takei, G., Kitamori, T., and Kim, H.-B. (2003) Proceedings of Micro Total Analysis Systems 2003. Transducers Research Foundation, San Diego, CA, p. 769. 109. Miwa, M., Nakajima, A., Fujishima, A., Hashimoto, K., and Watanabe, T. (2000) Effects of the surface roughness on sliding angles of water droplets on superhydrophobic surfaces. Langmuir 16, 5754–5760. 110. Tadanaga, K., Morinaga, J., Matsuda, A., and Minami, T. (2000) Superhydrophobicsuperhydrophilic micropatterning on flowerlike alumina coating film by the sol–gel method. Chem. Mater. 12, 590–592. 111. Dittrich, P. S., Tachikawa, K., and Manz, A. (2006) Micro total analysis systems. Latest advancements and trends. Anal. Chem. 78, 3887–3908.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15 COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES AND THEIR APPLICATION TO CREATE FUNCTIONAL DEVICES Pascale Maury, David N. Reinhoudt, and Jurriaan Huskens

15.1 INTRODUCTION Lithography techniques are used to create patterns on substrates. Thanks to lithography, a wide variety of devices can be made. Nanotechnology requires lithography techniques with improved resolution down to the sub-100-nm scale. In microelectronics, lithography techniques have improved over time in such a way that the transistor density of integrated circuits doubles every 1.5 years, following what is called Moore’s law. On the other hand, in a new trend of microelectronic applications, integration of components that classically have not been integrated, instead of miniaturizations, is of key importance. These types of devices are comprised in what is called “more than Moore” applications and include, for instance, sensing, actuating, and processing for wireless device applications.

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


379

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

380

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

Two approaches exist for the creation of structures: (i) the top-down approach, where a bulk material is directly patterned using irradiation or molding [1], and (ii) the bottom-up techniques using self-assembly of molecules or compounds on a surface. The top-down approach is currently used in microelectronics while the bottomup approach has been developed in chemistry and biology. The main top-down and bottom-up techniques that are used in nanofabrication will be presented in Section 15.2.2. Since the emphasis lies here on the combined methodologies, these descriptions will only be brief. Mixed techniques that make use of a combination of both approaches will constitute the core of this chapter. First, confining a bottom-up assembly in a pattern created by a top-down technique will be shown as a strategy to obtain high resolution features. Then examples of devices created by the combination of top-down and bottom-up techniques will be shown. Photonic devices fabricated by the patterned assembly of particles and biological assays formed by controlling the protein position will be demonstrated, as they are also topics described in other chapters of this thesis.

15.2 TOP-DOWN AND BOTTOM-UP TECHNIQUES 15.2.1 Top-Down Techniques Decreasingly smaller patterns are created by photolithography following Moore’s law. The latest microelectronic production equipment has a resolution of 45 nm with a throughput of around 130 (300 mm diameter) wafers per hour and development of 32-nm resolution equipment is on the way. In addition to high resolution, other important parameters for silicon technologies are overlay and ability of multistep and multilayer patterning, needed to create integrated circuits.

15.2.1.1 Patterning a Polymer Template. Lithography techniques based on patterning a polymer spin-coated on a substrate are presented first. The resulting polymer template is used as a mask to transfer the pattern onto the substrate, by additive techniques such as liftoff or by subtractive techniques such as etching or by modifying the substrate properties by ion implantation or diffusion. Photolithography makes use of a light-sensitive resist that is exposed with a light source through a mask applied on the substrate. After development, the irradiated photoresist is removed when using a positive-tone resist, or the nonirradiated photoresist is removed when a negative-tone resist is used. The resolution (d) is related to the wavelength (λ) by d = k1 λ/NA (Rayleigh criterion), where NA is the numerical aperture of the projection lens system and k1 the process factor that depends on the process. Photolithography is a parallel technique with high throughput that allows overlay for multistep processes. A limitation at high resolution is the proximity effect that consists of degradation of the shape of the pattern features because of the unintentional exposure of the resist in high density patterns. This effect can be reduced by using resolution enhancement techniques that consist of compensating the effect by modifying the mask design.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.2 TOP-DOWN AND BOTTOM-UP TECHNIQUES

381

The concept of photolithography is used with some changes in order to achieve a higher resolution. In a technique called immersion lithography, a liquid is used between the lens and the substrate in order to reduce the NA and increase the depth of focus. In this way, a resolution down to 45 nm is obtained with production class equipment. Another technique to obtain higher resolution than photolithography makes use of extreme ultraviolet (EUV) radiation [2]. Smaller features can be obtained by using shorter wavelengths. Features down to 32 nm have been obtained with this method. This technique has not yet reached production level and one of the challenges is the low power of the EUV sources and the need of vacuum chambers. In laser interference lithography (LIL) or holographic lithography, the photoresist is irradiated directly by a laser beam without using a mask. The pattern is given by an interference pattern created by splitting and recombining a laser beam. The advantage of this technique is that high resolution can be obtained on large areas. Resolution is on the order of 100 nm. The disadvantage is that the pattern choice is limited to periodic patterns obtained by interference. Electron beam lithography (EBL) makes use of an electron beam to expose an electron sensitive resist. The resist is usually a polymer, the molecules of which are broken or cross-linked upon electron irradiation. The resolution of EBL goes down to 1 nm. The throughput of this technique is relatively low because of its serial nature. In industry, EBL is used, for instance, for the fabrication of masks for photolithography. Nanoimprint lithography (NIL) is a nanofabrication technique that employs physical deformation of a polymer film by using a hard stamp. Chou discovered NIL in 1995 obtaining for the first time sub-micrometer features (25 nm) on poly(methylmethacrylate) (PMMA) polymer films using a SiO2 stamp [3]. NIL has had a growing success in research laboratories for its simplicity and low cost [4]. Its application requires only a press with heating capability. Stamps are elaborated by traditional lithography techniques (photolithography or EBL lithography, followed by reactive ion etching (RIE)). Stamps can be made of silicon, silica, metal, or glass. The sample to be patterned is composed of a substrate covered with a uniform polymer thin film. The stamp is applied on the sample and both are placed in the press. The temperature is increased and pressure is applied for a certain time (Figure 15.1). Then the pressure is released and the sample is cooled down to below the glass transition temperature of the polymer. At that point, stamp and imprint are separated. UV-NIL and step-and-flash lithography have been developed because of the need for overlay. In that case, a transparent mold is used to pattern a monomer that is polymerized after irradiation.

15.2.1.2 Direct Patterning. Nanostencil lithography can be used to produce patterns by means of a shadow mask and the evaporation of material in a vacuum [5]. The method consists of placing a mask between a source and a substrate, the evaporated material being deposited only in the nonshadowed area as can be seen in Figure 15.2. It is a one-step technique because a liftoff step is not necessary. The advantages of this method can be summarized as follows: (a) it can be used on fragile substrates (like sol–gels, hydrogels, biological macromolecules, and organometallic molecules) because it does not need chemical or thermal treatment, (b) it can

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

382

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

Stamp

T >Tg

PMMA Imprint Stamp

Heating and pressure

Stamp removal T = Tg Residual layer removal

Figure 15.1. The nanoimprint lithography process.

be used in situ for high vacuum experiments, (c) it allows patterning on nonflat substrates such as microelectromechanical systems and atomic force microscopy (AFM) cantilevers. Focused ion beam (FIB) lithography consists of direct writing with an ion beam on a substrate such as Si [6, 7]. Direct writing on silicon is possible because ions are able to etch silicon. The resolution limit of the method is about 20 nm. It is a low throughput technique because the features are written or milled directly by the ion beam. It is used mainly in industry for mask repair and sample preparation for characterization.

Nanostencil Photoresist ring Substrate Metal evaporation

Figure 15.2. Shadow mask evaporation using a nanostencil. The nanostencil behaves as a mask for patterning a metal on a substrate.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.2 TOP-DOWN AND BOTTOM-UP TECHNIQUES

383

15.2.2 Bottom-Up Techniques Self-assembly is the spontaneous organization of molecules or objects into stable, well-defined structures by noncovalent forces. The final structure is close to or at thermodynamic equilibrium; it therefore tends to form spontaneously and to selfrepair defects. Bottom-up nanofabrication makes use of self-assembly of components to create structures on the nanometer scale. The advantage of bottom-up techniques is the straightforward application of the associated self-assembly processes such as immersion, drop-casting, spin-coating, dip-coating. It also allows large area processing but the choice of patterns is limited to periodic structures with homogeneous sizes. A self-assembled monolayer (SAM) is an ordered molecular assembly that is formed spontaneously by adsorption of a surfactant with a specific affinity of its headgroup to the surface. It consists of a two-dimensional (2D) film, one-molecule thick, covalently organized, or noncovalently assembled at an interface. Molecules used to form SAMs have an anchoring part that attaches to the substrate, an alkyl chain and a functional end-group. These molecules have the property to selfassemble on the surface with high order. The most widely used types of SAMs are thiols on Au and silanes on SiO2 surfaces. SAMs can be used to anchor functional components or as etch barriers. Block copolymers consist of molecules containing two polymer chains of different nature. The chains have the property to rearrange in order to form spheres, cylinders (orthogonal or parallel), and lamellae spontaneously after heating. The ability to rearrange is controlled by the nature and the length of the two polymer chains and can be tuned easily [8]. These blocks can be either periodic, alternating, or random [9]. The advantage of using block copolymers is that the range of feature sizes that can be created is between 1 and 100 nm, making them particularly interesting for high resolution patterning [10]. In order to transfer the patterns to the substrate, one of the polymer chains can be modified to behave as an etch barrier [11]. Colloidal lithography consists of using nanoparticles arranged in a hexagonal fashion on a substrate. The highly ordered colloidal arrangement is usually obtained using immersion in a suspension or procedures based on capillary forces exerted on the nanoparticles [4]. The nanoparticles patterns are then used as a mask for liftoff or etching. The inter-nanoparticle distance dictates the resolution of the pattern that is related to the size of the nanoparticles. A resolution of 50 nm has been obtained [12]. Layer-by-layer (LBL) assembly consists of building up multilayered films by alternatingly depositing components held together owing to attractive forces such as electrostatic interactions (Figure 15.3) [13]. The advantage of this technique is that the height of the LBL assemblies is accurately controlled by the total number of deposited layers. Moreover, LBL assembly allows to create composite films by applying layers with different properties. The properties of the LBL assembly can be modified by changing pH and ionic strength, for instance.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

384

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

Polyanion

Polycation

Figure 15.3. Representation of layer-by-layer assembly.

15.2.3 Mixed Techniques Many research groups do not have access to clean-room facilities and many university clean rooms lack lithographic equipment capable of high resolution patterning. Therefore, several alternative and low cost methods have been developed that give access to high resolution patterning while keeping the research interests in mind, i.e., with a focus on versatility and fundamental science rather than mass-scale production. Additionally, there are special applications where alternative lithography techniques are more efficient or the only possibility. Microcontact printing (µCP) consists of depositing an ink that forms a SAM on a substrate with the help of a soft stamp by simple contact [14]. It is a direct patterning method that permits working on nonflat surfaces. As the deposited ink molecules are highly ordered, they allow controlling the surface chemistry of substrates, and they can be used to form resist masks for etching. Using this technique, surfaces can be functionalized with small molecules (thiols on gold), precursors (catalyst, functionalized polymers) as well as biological molecules (proteins, DNA, spacers) [15]. An advantage of µCP is that it is possible to print on flexible substrates. The resolution is limited by the size of the stamp features, the stability of the flexible stamp structures, and ink diffusion. The usual feature sizes produced by µCP are on the micrometer scale, but sub-micrometer features, down to about 100 nm, have been created by using fine-tuned processes [16]. Printing with poly(dimethylsiloxane) (PDMS) stamps is not limited to thiol inks. The low surface energy of PDMS stamps, which makes them inert toward chemical and metallic compounds, allows the printing of elements such as metals, for example, Au, or molecules used for LBL assembly. After the application of the stamp to the substrate, only the material in contact with the substrate is transferred, creating a pattern (Figure 15.4). This application, known as nanotransfer printing (nTP), is important for processes where solvents need to be avoided [17]. For example, in

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES

385

Assembly on PDMS

PDMS Transfer on Si

Figure 15.4. Representation of nanotransfer printing.

nanoelectronics, nTP of Au was used to create Au pads on a material that would be damaged by the usual Au liftoff technique normally used to create electrodes. Dip-pen nanolithography (DPN) is a direct-write lithographic technique that uses an AFM to build a pattern on the substrate material [18]. The tip is inked with molecules, and then used to write on the substrate. The resolution is related to the size of the AFM tip and to the diffusion of the molecule used as an ink, typically down to 40 nm [19].

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES FOR HIGH RESOLUTION PATTERNING In recent years, there has been an increasing number of reports on techniques to pattern bottom-up assemblies based on the use of physical or chemical patterns produced by means of top-down techniques. In the physical patterning techniques, topographical patterns are created on Si or polymer substrates using top-down lithography techniques such as photolithography or EBL (Figure 15.5a). The pattern defines the lateral resolution and provides the confinement needed to obtain the desired structures and ordering. In addition, the polymer pattern acts as a mask to prevent unwanted adsorption. After liftoff of the polymer mask the pattern is revealed. The disadvantage of this technique is that the solvent needed to remove the polymer during the liftoff step may induce disorder into the attached nanostructures. On the other hand, chemical patterning consists of using substrates having regions of different properties or composition. Examples of chemical patterns include molecular patterns of dissimilar nature, such as SAMs with different functionalities, alternating charges on the substrate surface, and metallic patterns on Si (Figure 15.5b). The functionalization can lead to differences in wettability, hydrophobicity, charge, or surface composition (e.g., Si/Au). A usual way of creating chemical patterns is to use mixed techniques such as µCP or DPN. Patterning methods for SAMs on Au [16] or SiO2 [20] have been reviewed elsewhere. The elements to be patterned

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

386

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

Building block

After rinsing

Polymer removal

(b)

(a) Figure 15.5. Utilization of (a) topographical and (b) chemical templates to pattern building blocks.

attach selectively to one component of the pattern. The advantage of this approach is that the liftoff step is avoided because the compound is deposited directly on the pattern and adsorbs only on the desired areas, without further processing steps. The disadvantage is the risk of nonspecific adsorption due to the lack of physical barriers and reduced ordering due to the absence of physical confinement. Another way to create a pattern is to use irradiation, for example, photolithography after the bottom-up assembly of components such as block copolymer [21, 22], nanoparticles (NPs) [23, 24], and LBL assembly [25]. The irradiated part is removed in solvent. The use of topographical and chemical patterns for the confinement of bottom-up assembly in order to increase the pattern resolution is treated in this section. Confinement of surface-initiated polymerization, micelles, block copolymers, NPs, and LBL assemblies will be introduced. 15.3.1 Top-Down Nanofabrication and Polymerization Surface-initiated polymerization is a technique that makes use of a surface initiator to graft polymers. When the surface initiator is patterned, polymerization is induced locally on the predefined sites. This technique can be used to amplify a pattern. The advantage is that the thickness and polymer graft density can be controlled, in contrast to other methods to produce grafted polymers. Surface initiators can be selectively attached on chemical patterns created by µCP [26–29], irradiation [30], and by metal evaporation by nanoshaving [31, 32]. The polymer is grown from the surface initiator by immersion in a monomer solution. The polymer brushes that have been used are polystyrene (PS), other olefins, and caprolactone. The typical height obtained after polymerization varied from 6 to 30 nm, which is high enough to be useful as a pattern enhancement technique. Additionally, polymers can be patterned at high resolution by coating an AFM tip

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES

387

with monomers. An applied voltage polymerizes the monomers, and polymer lines down to 30 nm in width have been created in this way [33]. STM (scanning tunneling microscope) tips can also be used to control chain polymerization [34, 35]. Zauscher studied the formation of long polymer sizes up to 200 nm. Under the same conditions as for the micrometer features, a smaller height of 170 nm was measured for sub-micrometer features. They explained this lower height by considering the effective initiator surface density and chain crowding. Polymerization occurred vertically and also laterally on the Au features, and appears to be anisotropic (lateral 220 nm/vertical 250 nm). In an experiment inspired by the molecular ruler, which is treated in Section 15.3.5, the controlled lateral growth of polymer brushes was used as a way to reduce the interfeature distance on a polymer substrate. The measured (isotropic) height increase of 15 nm on the polymer structure caused the reduction of the interfeature distance to 30 nm [36]. 15.3.2 Top-Down Nanofabrication and Micelles A beautiful integration of top-down and bottom-up techniques using micelles has been shown by Spatz [37, 38]. EBL and attachment of 7-nm Au clusters loaded in 300-nm micelles were combined in this approach toward ultrahigh resolution. Micelles are spherical aggregates formed dynamically from surfactants. Au NPs were homogeneously confined in the polymer pattern because the micelles behave as spacers controlling the spatial orientation and avoiding aggregation of the Au NPs. Oxygen plasma etching was performed to remove the micelle envelopes after polymer liftoff. Single Au clusters of 7 nm with a periodicity of 2 µm as well as 60-nm lines were obtained. 15.3.3 Top-Down Nanofabrication and Block Copolymer Assembly When nonpatterned, block copolymers have no preferred orientation and films formed from them contain defects in their arrangement on large areas. Confinement in a physical pattern directs the orientation of the block copolymers and diminishes the defects in self-assembly by introducing geometric substrate anisotropy. Additionally, this technique allows integrating block copolymers in device applications. As a first example of this technique, one-dimensional (1D) designs or lines will be discussed [39]. PS-b-poly(ethylenepropylene) (PS-b-PEP) block copolymers, the chosen material, form hexagonally packed cylinders in bulk (L = 23 nm, p = 26.6 nm, where L is the diameter of the cylinder and p its period). When the block copolymer is spin-coated into grooves made in Si, PS chains wetted the sidewalls preferentially and aligned along them, forming cylinders. The linewidth and the periodicity were controlled by the size of the chains of the block copolymer. The usual range of the chains is between 10 and 50 nm with a filling factor of 50%. The width of the grooves determined the number of lines, while the edge roughness of the sidewalls influenced the rectilinearity of the adjacent PS cylinders as they closely followed the geometry of the edges. Lines in the middle of the pattern were not influenced. Defect-free areas were larger than those on nonconfined block copolymers and corresponded to regions of about 100 µm2 . The thickness of the spin-coated

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

388

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

film determined the number of layers. For a thickness smaller than one cylinder, no patterning was obtained. For a thickness corresponding to three layers, a multilayer of ordered cylinders was obtained. In a second example, 2D designs such as dots were targeted [40]. In this case, 2D PS-block-poly(ferrocenyldimethylsilane) (PS-b-PFS) block copolymers with a spherical morphology were used. Polymer patterns were made using EBL on hydrogen silsesquioxane resist. PS wetted the sidewalls while, after curing, the PFS phase was arranged as spheres. The spheres aligned inside the grooves in a hexagonal configuration. In this example, two degrees of freedom are involved, which caused the assemblies to have more defects than the cylinders of PS-b-PEP. Similar to 1D patterning, edge roughness influenced the homogeneity and arrangement on the adjacent spheres. Domains in the middle of the templates were not influenced by edge roughness. Nevertheless, block copolymers could be confined successfully in a cavity formed by walls at a 60◦ angle and the dots were aligned according to this angle. Confinement of a single line of spheres of block copolymers (of a diameter d) was obtained when the linewidth (W) was narrowed [41]. For W/d < 1, spheres were elongated parallel to the sidewalls in order to fit inside the line while for W/d > 1, they are elongated perpendicular to them. For W/d > 1.5, two lines of spheres were obtained in a zigzag configuration. Two layers of spheres were also prepared with block copolymers. The phases of the block copolymers are deformable in contrast to the rigid colloids or metallic NPs. Block copolymers adapt their configuration to the shape of the confining features following the roughness of the sidewalls or assuming elongated shapes if necessary. The attractiveness of this approach is the homogeneity of the obtained high resolution structures, the usual range being 10–50 nm. In contrast, this high resolution and homogeneity can be difficult to obtain with NPs because of their polydispersity at this scale. A new and interesting approach has been developed by Huck who uses NIL to pattern block copolymers [42]. In this method, patterning and annealing (assembly) are performed simultaneously. The mold features provided the physical confinement that allowed the block copolymers to rearrange following the borders of the sidewalls while being annealed at the same time. At the end of the process, the mold was removed. It is important to note that the resulting patterned block copolymer structures that consist of substrate areas covered with polymer and others free of polymer can be used directly as a resist. Even though the results are still preliminary, the concept seems very promising. In addition to physical templates, chemical patterns have also been used to align and orient lamellar domains of symmetric PS-b-PMMA block copolymer films perpendicularly to the surface [43, 44]. For monodomain ordering, the chemical pattern needs to have the same dimensions as the block copolymer patterns, which are on the 50-nm scale. EBL or extreme ultraviolet lithography were used to create a polymer pattern that was used as a mask for SAM formation in order to create chemical patterns, for example, OTS/SiO2 [44] or PS brush/SiO2 [45, 46]. The aim was to create a hydrophobic/hydrophilic pattern that is wetted preferentially by one of the chains of the block copolymer. PMMA wets SiO2 and PS wets OTS or PS brush. Ls represents the periodicity of the chemical pattern and Lo the lamellar period of the block

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES

389

copolymer. When Ls = Lo , defect-free lines were obtained because each part of the block copolymer is commensurate to the corresponding chemical area (e.g., Ls = 47.5 nm, Lo = 48 nm). For Ls < Lo , (Ls = 45 nm, Lo = 48 nm), dislocations appeared and for Ls > Lo (Ls = 50 nm, Lo = 48 nm) lamellae were undulated and tilted out of plane. The thickness appeared to be important as well. Above a certain thickness (60 nm), lamellae went out of plane with a tilt angle θ , for which cosθ = Lo /Ls [47]. In order to increase the range of defect-free areas, a significant improvement was made by increasing the contrast in interfacial energy or wetting behavior between adjacent chemically patterned regions [48]. Defect-free areas were obtained for 40 nm < Ls < 52.5 nm. Owing to this improved process, continuous patterns of nonregular structures could be created with angles from 30◦ to 135◦ [46]. 15.3.4 Top-Down Nanofabrication and NP Assembly Interesting effects can be observed when creating NP patterns depending on the relation between particle size and the pattern feature size, shape of the confining features, and type of confinement. Unlike block copolymers, NPs can be confined on any shape of pattern. When using topographical templates, capillary forces play an important role in the ordering process especially with respect to pattern design and the continuity of the features. In a striking example, Xia used templates made of Si (photolithography and RIE) to confine particles inside the patterns by means of the meniscus method [49, 50]. In the meniscus method, capillary forces push the particles inside the pattern. Xia studied the different configurations obtained with particles confined in holes of various shapes such as circles, squares, rectangles (lines), triangles, and rings. Expressions for the geometrical requirements in order to obtain a given cluster configuration taking into account the diameter of the particle d, and the dimension of the pattern, a hole diameter D or width W of a line were deducted and explained the observed results. Moreover, zigzag confinement in helicoidal shape was shown by using V grooves [50]. Kumacheva studied NPs confined in micrometer grooves in which, by tuning the width of the grooves, perfect ordering or defect zones were seen, showing the importance of the d/W ratio on ordering [51, 52]. Two states of the arrays of particles could be distinguished: highly ordered hexagonal packing and random dense packing corresponding to the configuration where NPs have too much free space inside the pattern. It was predicted that hexagonal packing of NPs could be obtained when the width of the grooves (Dc ) and the radius of the NPs used (R) was given by the expression: Dc = 2R[(n − 1) cos 30◦ + 1]. Following these examples, intense research efforts were devoted to confine NPs in nanopatterns on SiO2 using different patterning techniques such as FIB [53] and EBL [54], and on other substrates [55–57]. Ozin made use of microfluidics to pattern particles inside V grooves closed by a PDMS flat stamp [58]. Colloids were patterned in diverse, nonregular and nonrectilinear, patterns [54]. Kim showed 1D groove patterns with single particle attachment in the shape of a necklace [50]. At the nanometer scale, Alivisatos noticed that particles smaller than 8 nm could not be physically confined [54].

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

390

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

After using a topographical pattern for confining particles, the physical barrier can be removed. Xia demonstrated confinement on the sub-200-nm scale by spin-coating NPs into polymer grooves made by LIL [1]. The polymer pattern was removed afterward. Capillary forces pushed NPs inside the grooves, but confining particles in closed patterns such as holes were not so efficient. Using the same process, Koo demonstrated the confinement of 8- and 20-nm NPs in a polymer pattern created by EBL [59]. In both cases, NPs were annealed in order to improve their attachment to the substrate and to remove the polymer. Pattern confinement has been used to control lattice and superlattice symmetry by modeling the periodicity (p) and height (h) of the pattern. This technique is called colloidal epitaxy. Van Blaaderen was the first to demonstrate epitaxial assembly of NPs using topographically patterned square arrays of holes, creating a 2D square microarray [60]. The directing pattern is underneath and the size of the features is smaller than the NPs in order to direct their ordering [52, 61–63]. Kim showed that for h/D (D being the diameter of the colloid) > 0.35, multilayer square array patterns were obtained [50]. For 0.28 < h/D < 0.35, pseudo-{1 1 0} structures are obtained and for h/D < 0.28 no three-dimensional (3D) crystallization was observed, but tetragonal or hexagonal clusters. Manipulation of individual NPs to create a pattern has also been demonstrated. Van Blaaderen made use of optical tweezers in order to arrange particles one by one in the desired shape [64–66]. With the use of a nanorobot [67], Lopez constructed a multilayer of particles with a diamond-type structure on a patterned silicon substrate that directed the lattice formation. Chemical templates can be used to pattern particles usually composed of polymer (PS or PMMA), SiO2 , or magnetic material [68]. The generally functionalized particles are attached selectively to one area of the pattern. One of the advantages of this process is that the substrate is not modified topographically and can be used after particle attachment for further processing. The most popular chemical pattern for NP attachment is a hydrophobic/hydrophilic contrast made in SAMs patterned by, for example, photolithography [69, 70] or µCP [71, 72]. NP suspensions do not wet the hydrophobic areas and NPs remain on the hydrophilic areas. Various shapes, periodic and nonperiodic, have been made using NPs commensurate to the chemical pattern [73]. Jonas and Koumoto studied the effect of chemical confinement while using the meniscus method [73, 74]. The attachment of the particles was governed by wetting properties and the shape of the meniscus. The discontinuity of the wetting properties at the interface on the chemical pattern made well-defined edges difficult to obtain. Moreover, the shape of the meniscus, which was higher on the center of the pattern than on the edges, led to the formation of a multilayer in the center [75]. Ordered single particle attachment along a continuous line was obtained by Koumoto using a chemical pattern made by a liquid mold and its drying process [73]. These studies were made on the micrometer scale. The process of attachment on chemical templates is more sensitive to wetting than on physical templates because attachment occurs along the meniscus (the meniscus needs to be homogeneous across the sample in order for the method to produce homogeneous linewidths).

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES

391

Another popular technique for patterning NPs consists of making chemical patterns with positively/negatively charged and positively/uncharged areas by using patterned SAMs [63, 71], LBL of polyelectrolytes [76], and xenography [77]. Charged particles, usually negatively charged by means of carboxylate functionalization, are used. This method presents the advantage that the negatively charged NPs slightly repel each other, which avoids particle aggregation and improves ordering. The NPs are attached using the meniscus method or immersion. Various geometries have been obtained and NPs attach with high selectivity and commensurate to the appropriate chemical pattern. NPs can be patterned with hexagonal packing, except on the edges of the pattern. Confinement in clusters where the NPs need to take a certain configuration are more difficult to obtain with chemical patterns than with physical patterns because of the lack of a physical confinement effect that would add capillary forces. Nevertheless, confinement has been shown in 1D and 2D on the micrometer scale. The quality of the features was comparable to the one obtained with physical confinement. Particles were shown to attach with a zigzag configuration [72, 73, 78] or on a single line [79]. Confinement in 2D features is more difficult to obtain than in 1D as reported in several papers [71, 79–81]. It should be noted that most of these studies deal with electrostatic attachment. Hammond produced confined clusters of NPs with shapes similar to the ones obtained with physical confinement with the difference that the D/d value is slightly higher for the chemical pattern [76, 81]. This is due to the fact that for chemical patterns, particles can adsorb on patterns smaller than their size because it is the contact area of the colloids that needs to fit the pattern. In contrast, in the case of physical patterns, the whole NP has to fit inside the pattern in order to be attached. Therefore, physical patterns need to be slightly larger than the NP size. Single particle attachment on a chemical array was obtained by Shinya [53]. Less effort has been devoted to study chemical confinement on the sub-200-nm scale as compared to physical confinement, because chemical templates in that range are difficult to prepare and the resulting templates are difficult to image. Nevertheless, NPs have been selectively attached on chemical patterns down to 200 nm made using LIL [71], 90 nm using edge lithography [82], and 40 nm [83, 84] and 30 nm [85] using DPN. Sagiv showed the attachment of 17-nm Au NPs densely packed and with a degree of order similar to the one obtained by physical confinement [83, 84]. A new approach consists of using ordered NPs to be transported to another substrate using a PDMS stamp, which might be patterned or not. The key to this process is to play with the surface energy of the particles on the surface. When the surface energy is low, particles can be picked up from the substrate provided their interaction with the stamp is higher than with the source substrate. They are then deposited on a target substrate with an even higher interaction with the colloids. This is a clear example of how the integration of top-down and bottom-up technique to control the surface energy allows control of building blocks such as NPs. In a first example of this approach, the meniscus method was used to attach and pattern NPs by means of physical confinement [86]. The height of the template was smaller than the size of the NPs in order to have the NPs protruding from the pattern. The substrate was

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

392

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

treated in advance with a fluoroalkyl SAM to minimize the adhesion between the surface and the NPs. A flat PDMS stamp was used to pick up the patterned NPs to subsequently deposit them on a Si substrate, having a higher adhesion with the NPs than PDMS. In this way, patterns of single NPs could be obtained at 5-µm period and multilayer colloidal slabs could be transported. Additionally, NPs could also be applied on top of another layer of NPs. In another experiment, a patterned PDMS stamp was applied on ordered NPs from a substrate to pick them up [87, 88]. By deposition on another substrate, the NPs were patterned and remained ordered. In this way, multilayers and patterned multilayers could be fabricated by printing one layer on top of another keeping the original orientation or orthogonal to the previous one. Yang took advantage of the swelling (using solvent) and stretching properties of PDMS used as the transporting material [89]. In this way, the PDMS stamp can be deformed after picking up the NPs thus changing the periodicity of the NPs that can be deposited on the substrate (enlarged in general). These experiments were done on the micrometer scale. Choi used ordered NPs in order to pattern PDMS stamps on the sub-micrometer scale [90]. NPs were first ordered in the grooves of a PDMS stamp using the meniscus method. These colloids arranged in typical shapes given by the physical confinement and the design were made such that they protrude from the PDMS stamp. Then this hybrid stamp was used to emboss a substrate coated with polyurethane polymer. Therefore, a mold with features in two size ranges was obtained. Similar experiments were performed using two types of NPs with different sizes [91–93]. We have recently used NIL to create sub-micrometer-sized patterned monolayers of (functionalized) silanes on silicon oxide that were used as topographical and chemical templates for NP assembly [94, 95]. Densely packed, hexagonally ordered NP structures were obtained only when combining the templating with capillaryassisted assembly (Figure 15.6) [95]. Downscaling allowed the creation of single NP lines of 350-nm PS and even 60-nm silica NPs. A supramolecular attachment approach was followed in the use of patterned molecular printboards as templates for directed NP assembly [96, 97]. Molecular printboards are well-characterized and ordered monolayers of cyclodextrin (CD) receptors to which molecules, NPs, and biomolecules have been attached via multivalent supramolecular host–guest interactions with control over: binding strength, orientation, reversibility, and specificity [98–100]. Specificity for anchoring CD-coated silica NPs was proven to occur on molecular printboards microcontact printed with guest-modified dendrimers [96]. Such NPs went selectively onto the printed areas in which complementary sites were thus present for linking the NPs to the underlying substrates. In a later study, NIL-patterned printboards were created to attach such NPs with higher resolution, again down to single NP dots smaller than 100 nm (Figure 15.7) [97]. 15.3.5 Top-Down Nanofabrication and Layer-by-Layer Assembly LBL assembly is a technique that allows controlling the thickness and composition of multilayered films. LBL assemblies are usually formed by alternating immersion of a

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES

500 nm (a)

393

100 nm (b)

Figure 15.6. Left: SEM image of 350-nm carboxylate-functionalized PS particles selectively attached to aminoalkyl SAMs deposited by capillary-assisted assembly on NIL-imprinted substrates. NIL was performed at 180◦ C and 40 bar, starting from a 300-nm PMMA layer. Prior to SAM formation, the residual layer was removed during 30 s in acetone. For a withdrawal speed of 0.25 µm s–1 , a bilayer of particles was deposited on the 700-nm lines, of which the top layer of particles follows a square lattice. Right: SEM image of 55-nm carboxylate-functionalized SiO2 NPs assembled on PMMA-imprinted sub-300-nm patterns with aminoalkyl SAMs. The initial polymer height was 100 nm and the residual layer was removed in O2 plasma during 10 s. The withdrawal speed was 1 µm s–1 . After particle adsorption, the polymer template was removed by sonication in acetone during 1 min. The effect of confinement on the particle assembly is apparent by using lines with linewidths of 60 nm.

substrate in solutions of the different components. Patterning LBL assemblies allows the creation of 3D nanostructures. A review by Hammond describes the different ways of patterning LBL assemblies [101, 102]. Therefore, the different techniques to pattern LBL assemblies will be only briefly mentioned here and the emphasis will be on the confinement issues in patterned LBL assemblies. In contrast to the section on block copolymers and NPs, the crucial issue here is not ordering of the building blocks but control over the selective attachment of one layer on top of the other in order to have an accurate control over the height of the assembly. In the template-assisted method, a polymer template is made usually by photolithography, EBL, or NIL (Figure 15.8a). In some experiments, a chemical functionalization is also made on the nonprotected areas of the substrate. LBL assembly is formed on the template usually by immersion. After the desired number of layers, the polymer is removed, revealing the LBL pattern. Liftoff is a crucial step, as the solvent should not induce disorder in the LBL assembly that is usually held together by electrostatic, hydrogen bonding, or hydrophobic interactions. Ultrasonication, which is normally used during liftoff can also affect the stability of the structures built in 3D. The advantage of this approach is that finding a chemical pattern to fit the requirements to attach the LBL assembly is not necessary because the LBL assembly is

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

394

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

(d )

(a) 500 nm

1 µm

2 µm (b)

(e)

500 nm

2 µm (c)

500 nm (f )

500 nm

2 µm

1 µm

Figure 15.7. SEM images of a pattern of 5-µm lines of CD layers (a) after one bilayer consisting of adamantyl-terminated PPI (poly(propylene imine)) dendrimers and 60-nm CD-functionalized SiO2 NPs, (b) after two bilayers, (c ) after three bilayers, the polymer was removed afterward. The insets show closeups of the particle areas. SEM images of 700-nm patterned CD layer after two bilayers formation before (d ) and after (e) polymer removal and (f ) of 1-µm patterned CD dots after formation of three bilayers and polymer removal.

started directly on the substrate. The polymer pattern prevents nonspecific adsorption and helps to obtain sharp edges after liftoff. The usual criterion to check that the patterned LBL assembly was not damaged by the liftoff process is to measure the height of the LBL assembly after liftoff and to compare it to the corresponding height of the nonpatterned LBL assemblies. Nevertheless, it has to be taken into account that different techniques are used to measure layer thickness on patterned substrates, typically AFM or cross-section scanning electron microscopy (SEM), and on nonpatterned samples, typically ellipsometry.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.3 COMBINING TOP-DOWN AND BOTTOM-UP TECHNIQUES

395

LBL assembly

Liftoff

(b)

(a) Figure 15.8. Utilization of polymer topographical (a) and chemical templates (b) to pattern LBL assembly.

Chemical patterning of LBL assemblies is based on a chemical contrast with a functionalization that attracts the first layer of the LBL assembly, usually a charged SAM such as with NH3 + or COO− groups, and an antiadherent functionalization for the LBL (alkyl or poly(ethylene glycol) (PEG), for instance, Figure 15.8b) [29, 103, 104]. A large number of layers can usually be attached while maintaining a good match with the chemical pattern that determines the lateral resolution. Patterning LBL assemblies on the micrometer scale has been widely studied but few attempts have been made on the sub-micrometer scale [105, 106]. Mirkin was able to construct patterned LBL assemblies up to 19 nm height on 200-nm COOHterminated thiol lines, and 16.3 nm height on 80-nm COOH-terminated thiol dots, using six bilayers. Jonas made a chemical pattern smaller than the length of the polyelectrolyte polymer, in which case the LBL assembly changed its configuration to fit the chemical pattern, resulting in a higher LBL height than for nonconfined LBL assemblies [107]. Generally, high selectivity of the LBL assembly makes patterning an easy process but structures with high aspect ratios have not been obtained yet. In another experiment, a molecular ruler was created by LBL assembly on gold structures prepared by using EBL and metal liftoff (Figure 15.9) [108]. The LBL assembly was anisotropically grown on Au features. The structure length was controlled by the number of deposited multilayers, one bilayer corresponding to 2 nm. In this way, the interfeature distance was decreased and the patterned LBL assembly was used as a mask for Au deposition and liftoff, creating nanostructures of sizes down to 40 nm. To avoid the rinsing step involved in liftoff, nTP can be used to transfer LBL assemblies. An LBL assembly was formed first on a patterned PDMS stamp [109].

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

396

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

Metal evaporation

Liftoff

Figure 15.9. Use of the molecular rulers, created by LBL assembly on polymer structures, as a way to pattern high resolution metallic structures.

The PDMS stamp was applied on a substrate and only the LBL areas that were in contact with the substrate were transferred [110]. We have recently used nTP of complete supramolecular assemblies onto molecular printboards [111]. PDMS stamps were functionalized with adamantyl-modified poly(propylene imine) dendrimers and CD-functionalized Au NPs of about 3 nm in size using a LBL assembly process [112]. These were transferred faithfully in micrometer patterns onto CD molecular printboards. The assembly and transfer were shown to be largely governed by specific host–guest interactions, although some nonspecific interactions, in particular involving the rather hydrophobic dendrimers, were observed as well [111]. Using NIL, much smaller lateral sizes of such supramolecular assemblies could be achieved [113, 114]. NIL was used to create sub-micrometer down to 50-nm feature sizes into PMMA. The substrate, after residual layer removal, was functionalized with CD monolayers, and LBL assembly of the dendrimers and Au NPs was performed yielding truly 3D supramolecular assemblies (Figure 15.10). A similar scheme was more recently applied to 60-nm CD-functionalized silica NPs [97].

(a)

(b)

(c)

Figure 15.10. Patterned assemblies of adamantyl-functionalized dendrimers and CDfunctionalized Au NPs prepared by LBL assembly on NIL-patterned CD SAMs followed by PMMA removal. Contact mode AFM images show (a) micrometer structures of four bilayers (40 × 40 µm2 , linewidth 5 µm, period 10 µm, height 10.2 nm), (b) submicrometer structures of 15 bilayers (3 × 3 µm2 , linewidth 200 nm, period 500 nm, height 19 nm), and (c ) sub-100-nm structures of 15 bilayers (1.3 × 1.3 µm2 , dot diameter 80 nm, period 500 nm, height 18 nm).

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.4 APPLICATION OF COMBINED TOP-DOWN AND BOTTOM-UP NANOFABRICATION

397

15.4 APPLICATION OF COMBINED TOP-DOWN AND BOTTOM-UP NANOFABRICATION FOR CREATING FUNCTIONAL DEVICES 15.4.1 Photonic Crystal Devices Photonic crystals (PhCs) are periodic dielectric nanostructures designed to affect the propagation of light in the same way as the periodic potential in a semiconductor crystal affects the electron transport by defining allowed and forbidden electronic energy bands. The periodicity of the PhC structure needs to be of the same order of magnitude than the wavelength of the light (in the 100-nm range). The photonic bandgap (PBG) of a PhC is the range of wavelengths for which light does not propagate through the structure. An important parameter for PhCs is the difference in refractive index of the materials used (n) between the patterned material and the surroundings, generally air. A high n is desirable because it improves the light confinement and thus the bandgap properties. Most of the 1D and 2D PhCs are made using conventional lithography, such as EBL and deep etching. 3D PhCs exhibit a PBG in all directions. The most popular method to create 3D PhCs is to make use of polymer or silica NP self-assembly because of the high order and low costs associated with this natural process. There exists a considerable number of reviews describing the performance of this type of PhCs [58, 115–119]. Due to the low refractive index of the NPs, an interesting method to increase n is to create an inorganic matrix (generally made up of metal oxide such as TiO2 ) around the colloidal assembly and to burn away the NPs afterward. In this way, a so-called inverted PhC is created [120]. In order to use these 3D PhCs as devices, they need to be structured, for example, by creating intentional defects or defining waveguides, and integrated with other devices. Examples of devices that can be fabricated with PhCs are low loss waveguides, filters, and resonators. The focus of this section is on the fabrication of 3D PhC devices by means of combinations of top-down and bottom-up techniques.

15.4.1.1 Patterning 3D Photonic Crystals. Colloidal epitaxy allows the controlled creation of 3D PhCs. The major drawback in colloidal self-assembly is the apparition of defects, usually cracks, which cause the NP domains to shrink. Cracks appear when the solvent used in the colloidal suspension evaporates. An effective method to overcome this problem is to add a slight interfeature distance between the colloids. This interfeature distance is obtained by creating a pattern underneath the colloids that directs the lattice or superlattice parameters. This method is called colloidal epitaxy (see Section 15.2.2). In this way, patterned and defect-free areas of 800 × 800 µm2 were made with 10 layers of colloids [121, 122]. An improvement of the quality of the PBG of the 3D PhCs fabricated in that way was observed. In order to produce optical devices, the most ambitious being optical chips or light-driven computers, light needs to be processed analogous to the way electrons are in current electronic circuits [87]. For that purpose, 3D PhCs need to be integrated and patterned in circuits made with different 3D PhC channels.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

398

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

A common method to pattern 3D PhCs is to grow or deposit NPs on patterned Si substrates using sedimentation, microfluidics [123], the meniscus method [124, 125], or spin-coating [55]. The PhCs can be grown on various shapes, usually rectangular channels [124, 126] but also on V grooves or squares [127, 128]. Multilayer NPs filling in closed patterns such as holes or rectangles is more difficult because the capillary forces are limited in those cases [123]. Sotomayor noticed that in the feature size range of 20 µm, cracks appeared along the sidewalls of the Si pattern [129]. Metallic PhCs can be patterned as well using gold NPs confined in photoresist channels [55]. In that case, the waveguide properties can be coupled to particle plasmon resonance. A method to construct integrated circuits of 3D PhCs is to transfer the patterned NPs onto a substrate and construct the photonic circuit by assembling channels of PhCs. Free-standing slabs of NPs can be made using various ways and later used as building blocks for PhC circuits. Sotomayor made use of cracks appearing along the sidewalls of the Si template in order to extract the patterned NPs by using adhesive tape leading to free-standing colloidal slabs [126]. Ozin used the same method after growing NPs on rectangular or V grooves. In additional experiments, NPs were grown in the pattern into an inverted lattice and etching of silica was performed to remove the template [128]. Matsuo produced free-standing pyramids [130] and Parikh used a sandwich structure formed by glass and Si substrates patterned chemically with hydrophobic SAMs with NPs grown in between to assemble free-standing NP slabs [131]. By integration of top-down and bottom-up techniques, Chua created a woodpile structure made of NP slabs [132]. Photolithography was used in order to create a polymer pattern that was used to confine a multilayer assembly of NPs. On top of the first layer, a second lithography step was performed, which was followed by filling with NPs. After the desired number of layers, the polymer was removed, leading to a woodpile of colloidal slabs.

15.4.1.2 Inserting Defects in 3D Photonic Crystals. A review about the integration of defects in PhCs appeared recently [116]. Therefore, in this section, only the integration of defects in PhCs by using combinations of top-down and bottom-up lithographies is covered exclusively. A method to propagate light along a given path inside a PhC is via defects along the path, which is known as a waveguide. The challenge of this approach is the local integration of defects of the desired shape in an otherwise perfect NP assembly. When using PMMA NPs, defects can be created on surfaces, because this material is widely used as resist for EBL. By using NPs made of PMMA, grown into a PhC on a substrate, localized defects can be made by irradiating selectively some PMMA NPs and removing them (or parts of them) by developing in the appropriate solvent [133, 134]. Using this method, single NPs can be accurately removed and patterns of voids can be fabricated on top of the colloidal crystal [135]. Trenches of the desired depth, up to the thickness of the assembly, were made inside the colloidal assembly by irradiating intensively the substrate. All irradiated NPs were dissolved during development, leading to trenches in the colloidal assembly [126, 133] In a similar

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.4 APPLICATION OF COMBINED TOP-DOWN AND BOTTOM-UP NANOFABRICATION

399

experiment, Lopez infiltrated a SiO2 matrix around the PMMA NP assembly. Then irradiation was performed, and the irradiated PMMA NPs were dissolved. The dissolved NPs led to a patterned inverted lattice [134]. Another method to create defects is forming linear voids embedded inside the NP assembly. By using planar confinement [136], a NP assembly was grown and then subjected to chemical vapor deposition (CVD) infiltration. Subsequently, a layer of silicon oxide was formed and the growth of the NP assembly was continued on top of it followed by CVD infiltration. At the end of the process, the silica was etched away and an inverted opal with a missing plane inside was obtained. Zhao and co-workers used the same principle but instead of a silica layer, they employed a single layer of silica particles sandwiched between two layers of PS NPs [137]. At the end of the process, the substrate was infiltrated with silica and the PS NPs were removed by calcination to produce inverted opals with a planar defect in the center. Chua used photoresist spin-coated on bulk silica NPs that was later patterned using photolithography [138]. NPs were assembled again around the polymer pattern. Then the polymer was removed in acetone leading to an empty rectangular line that was used as a waveguide. In a similar experiment, after removing the polymer pattern embedded inside the SiO2 opals, the process was continued with CVD of Si or Ge, and wet etching was used to remove the silica opals and to fabricate an inverted lattice [139]. An alternative method consists of embedding defects by using NPs of different sizes instead of creating voids. In that case, a bulk NP assembly was grown, on top of which photoresist was spin-coated and patterned by photolithography [140]. This template was used to confine NPs of different sizes. The polymer was removed and the colloids of the first size were grown again around the patterned NPs of the second diameter. In this way, rows of the second type of NPs were embedded in bulk NPs. In a similar experiment but with a different patterning technique, NIL was performed on top of a silica opal assembly, and the polymer pattern was used to create a single layer of silica particles with a different diameter [92]. Then the previous silica opals were grown again. At the end of the process, the polymer pattern was removed using acetone. The aim of their experiment was to create point defects on the colloidal lattice. Braun used multiphoton polymerization in order to write patterns of different shapes inside the bulk opals [63]. For that purpose, an opal assembly was grown to which a monomer was added as well as a photoinitiator. By using laser irradiation by a confocal microscope, the monomer was locally polymerized with the desired patterns, usually with a linear shape. Polymerization was induced inside the bulk using the property of a confocal microscope to focus the laser on different planes. Arbitrary features were written in this way. After polymerization, wet etching was performed in order to create inverse opals and to remove the polymer pattern. At the end of the process, empty lines integrated in the bulk assembly of inverted opals were obtained. The empty line created in the bulk of the colloidal assembly was used as a waveguide. Norris used a slight modification of this process to create an inverted lattice [141, 142]. The colloidal assembly was infiltrated with a hybrid photoresist that stands

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

400

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

high temperatures. After creating a line of defects using photopolymerization with a confocal microscope, the nonirradiated resist was removed and replaced by silicon. Colloids and irradiated resist were removed by wet etching. Embedded lines of 1 µm were created in that way. Xia used the swelling effect of PDMS when solvents such as silicon fluid, hexane, or octane are applied on it [143, 144]. NPs were assembled on a glass substrate integrated in a PDMS matrix, which was called “photonic paper.” The solvent used as an ink was applied locally by means of ink-jet printing or µCP. The PDMS regions that were inked shrank locally causing the lattice parameter and thus the color of the colloidal assembly to change. After evaporation of the solvent, the photonic paper went back to its initial state with uniform color. 15.4.2 Protein Assays Nanobiotechnology constitutes the interface between nanotechnology and biology. An extraordinary interest exists as a wide range of applications can be envisaged, ranging from biosensors, genomics, and proteomics. Several reviews deal with NPs as biosensors [145], protein chips [146], transducers [147], nanobiotechnology [148] in general, and bioengineering [71]. A widely used method to pattern biological compounds such as proteins or DNA is ink-jet printing, where a drop of solution of the compound is delivered on a substrate. This technique allows the creation of arrays of different proteins on the millimeter scale [149]. In this chapter, the focus is on the directed attachment of proteins on the micrometer and sub-micrometer scale using patterns made by combination of top-down lithography, as a way to create the lateral pattern, and bottom-up assembly to anchor selectively proteins through interactions such as physisorption, covalent attachment, and supramolecular interactions. Some requirements are needed in order to construct a suitable protein assay: (i) accurate spatial localization, (ii) specific adsorption of the selected protein, (iii) negligible nonspecific adsorption of other components, (iv) controlled orientation of the proteins, and (v) maintaining the functionality of the protein. Common patterning methods are based on the adsorption of proteins, by using topographical templates with chemical functionalization, or chemical patterns. µCP and DPN are widely used to create patterns for protein patterning. These will not be reviewed here. Molecular assembly patterning by liftoff (MAPL) is used often to fabricate chemical patterns for protein adsorption on SiO2 [150]. Nanosphere lithography is also popular as it allows easy access to sub-micrometer features over large areas [151]. The proteins are usually labeled with a fluorescent dye that allows characterization using fluorescence microscopy.

15.4.2.1 Patterned Protein Assembly Using Physisorption. The process of patterning proteins is simplified by the fact that proteins adsorb nonspecifically on substrates. µCP has been used as a way to pattern a lipid membrane followed by filling the uncovered areas with proteins, or to pattern proteins followed by filling with a lipid membrane. In the first method, protein diffusion is limited by the lipid

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.4 APPLICATION OF COMBINED TOP-DOWN AND BOTTOM-UP NANOFABRICATION

401

layer [152]. This technique provides a way to complex functionalized patterns that can be used both for protein and membrane studies. µCP was also used to pattern PEG layers that have protein-resistant properties for nonspecific protein patterning directly on the SiO2 areas [153]. Additionally, nanosphere lithography was used to create patterns of proteins [154]. The originality of this approach relies on the fact that both proteins and particles were present in the same suspension. When the suspension was allowed to dry on a mica surface, particles assembled with high order and confined the proteins at their interface. Due to the fact that particles were previously coated with a hydrophobic coating, they were removed by simple rinsing from the surface, revealing sub-micrometer protein patterns. High resolution patterning of proteins down to 60 nm was performed by using a FIB to pattern GaAs on SiO2 [155]. Human serum albumin was found to attach with high specificity on the patterned GaAs. The full integration of proteins and DNA inside a microfabricated Si device has been achieved [156]. After preparation of the microstructures, proteins or DNA was ink-jet printed on specific areas. The structures were coated with gold to preserve their functionalities while further microfabrication processes were performed. At the end of the process, after gold removal, DNA molecules recovered their full functionality while proteins recovered only to about 50%. This example offers good prospects for the future integration of proteins into devices. When using proteins attached through hydrophobic interactions, a hydrophobic/hydrophilic chemical contrast is often created to direct their deposition. A way to create the chemical contrast is by patterning a polymer. Patterns have been obtained by printing (using µCP) a protein-resistant polymer such as poly(lactic acid)–PEG (PLA–PEG) on a hydrophobic PS surface, or by patterning novolak resist using photolithography, which becomes hydrophobic after irradiation [157]. Hydrophobic patterns can be created by irradiation of LBL assemblies [158]. On the sub-micrometer scale, nanosphere lithography was used indirectly to create Au patterns on SiO2 [151]. A hydrophobic thiol was attached to Au while PEG was formed on the SiO2 areas with features down to 100 nm. Moreover, nanosphere lithography has been used directly to create a mask to form the chemical pattern [159]. PMMA-b-PS block copolymer was used in order to obtain high resolution patterning with sizes down to 45 nm easily [160]. Different proteins were adsorbed, and it was noticed that the protein height measured on the sub-50-nm patterned areas was smaller than the height of the protein when nonpatterned. NIL was used to create chemical patterns down to 20 nm for protein adsorption. In these experiments, proteins attached with high specificity on the hydrophobic pattern and fluorescence microscopy was used to check the integrity of the protein. Proteins can be attached through hydrophilic interactions. For instance, a Si surface was patterned with a carbohydrate-terminated monolayer using µCP, which specifically attached carbohydrate-binding proteins [161]. Electrostatic attachment is easy to use as proteins are usually charged. However, no specific orientation can be induced. Nanografting, which uses an AFM tip, was performed on a COOH-functionalized SAM to create a pattern. A hydrophobic SAM was formed on the uncovered areas to direct protein attachment through hydrophobic

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

402

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

interactions. The aim of the process was to create a hydrophilic/hydrophobic pattern, but it was noticed that the proteins attached selectively on the negatively charged areas formed by the carboxylate groups. Proteins patterned down to 100-nm features were created [162]. Finally, nanosphere lithography was used to pattern proteins. The NPs were used as a mask for adsorption of a PEG SAM. After NP removal, the uncovered areas were filled with a negatively charged SAM. After protein deposition, pattern sizes down to 100 nm were obtained [163].

15.4.2.2 Patterned Protein Assembly through Covalent Attachment. Covalent attachment is a good candidate when the protein needs to be firmly attached to the surface through specific functional groups. Covalent attachment can be achieved using a linker that binds to a conventional SAM and allows proteins to attach covalently via specific functional groups. A method for covalent attachment is based on photolithography, for which, after patterning a water-based resist, a SAM was formed on the unprotected areas [164]. Several functionalizations were performed allowing proteins or peptides to attach via their cysteine sites. Liftoff was performed with water in order to avoid using organic solvents that could denature the proteins. Another example is based on Au/Si patterns created using photolithography and Au liftoff [165, 166]. A PEG SAM was attached specifically on SiO2 , while a carboxylic acid thiol was attached on the Au areas. A linker was attached to the thiol with the purpose to bind the proteins through their amino groups. The selective attachment of proteins was shown by fluorescence and also by infrared spectroscopy confirming the right composition. By resolving the protein shape with high resolution AFM imaging, it was confirmed that all proteins inside the pattern take a similar orientation, confirming the attachment at a specific site. Using the same type of covalent binding, photolithography was used to create a mask to chemically pattern amino and fluoro layers on a single-crystal diamond surface [55]. DNA was attached covalently with high specificity on the chemical pattern. In an experiment based on NP assembly, carboxylate-functionalized NPs were confined with hexagonal packing inside hole patterns created by photolithography [167]. Proteins were attached on the NPs through conjugation with carboxylic acid groups. The novelty consists of attaching proteins on curved surfaces in order to increase the surface area for protein attachment. This was inspired by the positive effect of surface curvature to retain the native structure and functionality of proteins compared to planar surfaces, but the quality of the proteins patterned in the former way was not investigated [168, 169]. Nanosphere lithography was used to pattern a poly(acrylic acid) (PAA) layer with a COOH-terminated surface [170]. To fabricate this, a layer of PAA (130 nm thick) was deposited using plasma-enhanced CVD, followed by attachment of PS beads on top using spin-coating. The dispersed PS beads were used as an etch mask for O2 plasma to protect the PAA from etching. Finally, the PS beads were removed in water and ultrasonication. The resulting patterned PAA had the shape of a dome with a plateau in the center surrounded by an edge. BSA (bovine serum albumin) protein adsorption was

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.4 APPLICATION OF COMBINED TOP-DOWN AND BOTTOM-UP NANOFABRICATION

403

performed. It was noticed that proteins attached preferentially in the center of the plateau and on the edges of the PAA domes.

15.4.2.3 Patterned Protein Assembly through Supramolecular Interactions. The advantage of using supramolecular interactions between the protein and the anchoring layer is that proteins can attach selectively via a specific binding site. The orientation of the protein can thus be directed. Another advantage is that the interaction can be reversed, allowing substrate regeneration. This effect has been studied mainly on nonpatterned substrates. Antibodies are Y-shaped proteins that attach specifically to antigens. Antibodies have been patterned by printing via physisorption [171, 172], or by covalent attachment on a silane layer [170]. Morhard found a higher quality by directed printing of the antibody compared to successive covalent binding, starting from printed alkyl thiols. Antibody patterns can also be formed starting from chemical patterns made by irradiation. A simple way to produce these patterns is to include the antibodies in a polymer matrix and subsequently imprint the polymer, as reviewed by Haupt [173]. Physical confinement has been compared to chemical confinement for the guiding of microtubules attached on kinesin [174]. Topographical templates were created by a photoresist film patterned by photolithography, which was subsequently functionalized with kinesin. The chemical template was formed by kinesin and a PEG SAM. The best directing quality was provided by the topographical pattern. It was found that microtubules remained fixed on the chemical edges when using chemical confinement. An important issue for the development of protein chips is to pattern different types of proteins on the same area. To achieve this, sequential patterning was performed using photolithography [175]. A pattern was prepared by photolithography and the uncovered areas were filled with one type of protein using immersion. Then the same substrate was patterned for a second time using photolithography. Alignment marks allowed patterning in different areas than during the first lithographic step. This second pattern was filled with a different type of protein. Up to three different proteins were patterned in this way. The quality was checked by fluorescence microscopy and inhomogeneous attachment was observed. The quality of the protein attachment was improved by antibody functionalization after UV patterning on the uncovered areas providing more homogeneous fluorescence images. Bruckbauer proposed patterning on the sub-micrometer scale of the same type of proteins functionalized with different fluorescent labels, in order to demonstrate a system in which biomolecules can be addressed [176]. The process consisted of fabricating an array of holes in which proteins with different labels were addressed (Figure 15.11). To achieve this, holes of 300 nm diameter were etched using FIB on a silicon substrate covered with gold. Au was functionalized with hexa(ethylene glycol) thiol to prevent nonspecific adsorption, while the opening holes in silicon were functionalized with 3-mercaptopropionic acid to facilitate the immobilization of IgG antibodies trough electrostatic interactions. The IgG antibodies were deposited from solution. A nanopipet was used to address selectively an individual hole with a solution of anti-IgG protein. The size of the tip of the nanopipet corresponded to the

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

404

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

FIB etching Apply

Au Glass

MPA

Nanopen delivery

PEG thiol

Treat with IgG

Treat with anti-IgG

MPA thiol

Figure 15.11. Scheme showing protein with a different label addressed using an AFM tip in an array of antibody-functionalized holes [176]. Mercaptopropionic acid (MPA) is used to facilitate the immobilization of IgG antibodies.

diameter of the holes. IgG and anti-IgG were labeled with a different fluorescent label. To check their attachment, confocal microscopy was used, showing spots containing two fluorescent colors implying that both IgG and anti-IgG were present. The biotin–streptavidin interaction is widely used in protein attachment applications. Streptavidin is a protein that contains four binding sites that interact specifically with biotin through supramolecular interactions. The biotin–streptavidin complex is very well known and popular for its specificity and stability. MAPL was used to fabricate chemical patterns. First, a hydrophobic SAM was formed (fluoroalkyl or alkyl), and after removal of the patterned polymer an aminoalkyl SAM was formed. Biotin was attached covalently on the aminoalkyl SAM to specifically direct streptavidin adsorption. The polymer patterns were made using photolithography [150, 177], EBL [55, 178], NIL [179–181], and nanosphere lithography [182] with features of 100, 75, and 50 nm, respectively. Biotin-functionalized protein attached selectively to the streptavidin pattern [177–179]. Hoff demonstrated protein attachment on the micrometer scale by AFM and fluorescence, and in the sub-micrometer scale by AFM [180, 181]. Zhang noticed a decrease of fluorescence intensity for dot diameters decreasing from 500 to 100 nm [55]. Another approach consists of using PMMA-b-PS block copolymers to provide a PS pattern template down to 15 nm [183]. The PMMA pattern was removed by acetone. PS provided the hydrophobic areas and an aminoalkyl SAM was formed in the uncovered areas. In this approach, the streptavidin functionalization was achieved by adsorption of streptavidin-functionalized Au NPs (10 nm) that attached selectively on the aminoalkyl SAM. This template was used for adsorption of biotinfunctionalized molecules, followed by attachment of streptavidin. The protein attachment was demonstrated by cyclic voltammetry. In this example, regeneration of the

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

15.4 APPLICATION OF COMBINED TOP-DOWN AND BOTTOM-UP NANOFABRICATION

405

substrate was shown using surface plasmon resonance. In another study, NP deposition was directed by patterning densely packed (sometimes hexagonal) streptavidinfunctionalized 2-µm particles on chemical patterns [182]. Chemical patterns made of amino/hydrophobic or PEG have also been made using irradiation through UV exposure [177], EBL [178], µCP [59], and DPN [184]. Multilayers of proteins by alternating attachment of biotin and streptavidin have also been demonstrated [178, 183]. Doh has made use of aqueous photoresist to pattern successively two types of proteins [185]. After an initial biotin patterning followed by protein attachment, the areas protected by the resist were uncovered to be filled with biotin and followed by another protein adsorption step. In another study, a transparent substrate was used that was covered on one side by a biotin and UV irradiation was performed through the other side using a mask [186]. The biotin attachment was activated by the UV exposure. Subsequently, the biotin pattern was used to direct streptavidin attachment. Several attachment steps can be performed using UV exposure through a mask and mask alignment to prevent overlaying, as the substrate does not need processing. N-nitrilotriacetic acid (NTA) allows specific and reversible binding of His-tagged proteins through its Ni2+ complex. This technique is widely used to purify proteins. NTA, bound to a support, is complexed with a metal ion (Ni2+ or Co2+ ) that allows binding of the proteins functionalized with a histidine group. Dissociation of the proteins attached by means of NiNTA–His-tag interactions can be induced by releasing the chelate complex (using EDTA (ethylenediamine tetraacetic acid)) or by using a competitive agent such as histidine or imidazole [187]. Tampe et al. have further developed this approach on gold surfaces using µCP. The chemical pattern consisted of a hydrophobic SAM (alkyl thiol) and NTA thiols. A binding assay was created by ink-jet printing different types of proteins on the same array [187–193]. We have recently used supramolecular interactions at CD molecular printboards to immobilize proteins [194, 195]. Orthogonal linkers were designed, containing one or two adamantyl groups for interaction with the CD substrate and biotin for anchoring streptavidin [194]. This resulted in assemblies the stability of which could be tuned by the valency of the linker molecules. µCP of the linker was employed to show the specificity of the orthogonal interaction scheme as well as to achieve heterofunctionalization of the immobilized streptavidin. More recently, it was shown that nonspecific interactions could be completely suppressed using a monovalent adamantyl-oligo(ethylene glycol) molecule [195]. This created a temporary PEG surface inhibiting nonspecific interaction, but at the same time still allowing specific interactions of linker protein constructs. This was shown to work as well for His-tagged proteins using an adamantyl-NiNTA linker molecule. His-tag–NiNTA interactions (solely) were employed in the creation of patterned protein substrates using NIL [196]. Nonspecific interactions on the non-NTA areas were avoided using a fluorinated monolayer. This allowed the specific, yet reversible, attachment of different His-tagged, naturally fluorescent proteins in micrometer patterns (Figure 15.12).

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

406

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

(a)

(b)

(c)

Figure 15.12. Consecutive fluorescence microscopy images of His-tagged proteins successively attached on a chemical pattern made of NiNTA and fluoroalkyl SAMs after rinsing the substrate to regenerate the NTA arrays. (a) Substrate after DsRed-FT attachment, (b) sample after rinsing with imidazole, (c ) sample after EGFP (enhanced green fluorescent protein) attachment.

Top-down techniques have been successfully used as a way to pattern bottom-up assemblies such as SAMs, micelles, block copolymers, NPs, and LBL assemblies. Some of the assemblies (block copolymers, NPs), when their dimensions are commensurate to the patterns, are confined and thus take on different configurations. These configurations are determined by the dimensions of the pattern and can be predicted. New elements are obtained that could not be done by using only top-down or bottom-up techniques. The integration allows high resolution patterning as well, and is more versatile and cheaper. Functional devices, such as 3D PhCs, were fabricated by embedding a structure inside a colloidal assembly. Protein assays have been demonstrated by using topdown techniques to pattern with lateral resolution and bottom-up assemblies as a way to direct and anchor the protein adsorption, nonspecifically or specifically.

REFERENCES 1. Geissler, M. and Xia, Y. N. (2004) Patterning: principles and some new developments. Adv. Mater. 16, 1249. 2. Alkaisi, M. M., Blaikie, R. J., and McNab, S. J. (2001) Nanolithography in the evanescent near field. Adv. Mater. 13, 877. 3. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1996) Imprint lithography with 25nanometer resolution. Science 272, 85. 4. Guo, L. J. (2004) Recent progress in nanoimprint technology and its applications. J. Phys. D: Appl. Phys. 37, R123. 5. Ludwig, A., Cao, J., Brugger, J., and Takeuchi, I. (2005) MEMS tools for combinatorial materials processing and high-throughput characterization. Meas. Sci. Technol. 16, 111. 6. Tseng, A. A. (2004) Recent developments in micromilling using focused ion beam technology. J. Micromech. Microeng. 14, R15. 7. Tseng, A. A. (2005) Recent developments in nanofabrication using focused ion beams. Small 1, 924.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

407

8. Krishnamoorthy, S., Pugin, R., Brugger, J., Heinzelmann, H., and Hinderling, C. (2006) Tuning the dimensions and periodicities of nanostructures starting from the same polystyrene-block-poly(2-vinylpyridine) diblock copolymer. Adv. Funct. Mater. 16, 1469. 9. Bockstaller, M. R., Mickiewicz, R. A., and Thomas, E. L. (2005) Block copolymer nanocomposites: perspectives for tailored functional materials. Adv. Mater. 17, 1331. 10. Park, M., Harrison, C., Chaikin, P. M., Register, R. A., and Adamson, D. H. (1997) Block copolymer lithography: periodic arrays of similar to 10(11) holes in 1 square centimeter. Science 276, 1401. 11. Hamley, I. W. (2003) Nanostructure fabrication using block copolymers. Nanotechnology 14, R39. 12. Yang, S. M., Jang, S. G., Choi, D. G., Kim, S., and Yu, H. K. (2006) Nanomachining by colloidal lithography. Small 2, 458. 13. Decher, G. (1997) Fuzzy nanoassemblies: toward layered polymeric multicomposites. Science 277, 1232. 14. Jackman, R. J., Wilbur, J. L., and Whitesides, G. M. (1995) Fabrication of submicrometer features on curved substrates by microcontact printing. Science 269, 664. 15. Michel, B., Bernard, A., Bietsch, A., Delamarche, E., Geissler, M., Juncker, D., Kind, H., Renault, J. P., et al. (2002) Printing meets lithography: soft approaches to highresolution patterning. Chimia 56, 527. 16. Xia, Y. N. and Whitesides, G. M. (1998) Soft lithography. Annu. Rev. Mater. Sci. 28, 153. 17. Loo, Y. L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Interfacial chemistries for nanoscale transfer printing. J. Am. Chem. Soc. 124, 7654. 18. Ginger, D. S., Zhang, H., and Mirkin, C. A. (2004) The evolution of dip-pen nanolithography. Angew. Chem. Int. Ed. 43, 30. 19. Hong, S. H. and Mirkin, C. A. (2000) A Nanoplotter with both parallel and serial writing capabilities. Science 288, 1808. 20. Onclin, S., Ravoo, B. J., and Reinhoudt, D. N. (2005) Engineering silicon oxide surfaces using self-assembled monolayers. Angew. Chem. Int. Ed. 44, 6282. 21. Pan, F., Wang, P., Lee, K., Wu, A., Turro, N. J., and Koberstein, J. T. (2005) Photochemical modification and patterning of polymer surfaces by surface adsorption of photoactive block copolymers. Langmuir 21, 3605. 22. Yang, S. Y. and Rubner, M. F. (2002) Micropatterning of polymer thin films with pHsensitive and cross-linkable hydrogen-bonded polyelectrolyte multilayers. J. Am. Chem. Soc. 124, 2100. 23. Moon, J. H., Jang, S. G., Lim, J. M., and Yang, S. M. (2005) Multiscale nanopatterns templated from two-dimensional assemblies of photoresist particles. Adv. Mater. 17, 2559. 24. Yu, C. H., Parikh, A. N., and Groves, J. T. (2005) Direct patterning of membranederivatized colloids using in-situ UV-ozone photolithography. Adv. Mater. 17, 1477. 25. Park, J. W. and Thomas, E. L. (2002) A surface-reactive rod-coil diblock copolymer: nano- and micropatterned polymer brushes. J. Am. Chem. Soc. 124, 514. 26. Harada, Y., Girolami, G. S., and Nuzzo, R. G. (2003) Catalytic amplification of patterning via surface-confined ring-opening metathesis polymerization on mixed primer layers formed by contact printing. Langmuir 19, 5104.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

408

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

27. Husemann, M., Mecerreyes, D., Hawker, C. J., Hedrick, J. L., Shah, R., and Abbott, N. L. (1999) Surface-initiated polymerization for amplification of self-assembled monolayers patterned by microcontact printing. Angew. Chem. Int. Ed. 38, 647. 28. Hyun, J. and Chilkoti, A. (2001) Surface-initiated free radical polymerization of polystyrene micropatterns on a self-assembled monolayer on gold. Macromolecules 34, 5644. 29. Zhou, D. J., Bruckbauer, A., Batchelor, M., Kang, D. J., Abell, C., and Klenerman, D. (2004) Influence of the foundation layer on the layer-by-layer assembly of poly-l-lysine and poly(styrenesulfonate) and its usage in the fabrication of 3D microscale features. Langmuir 20, 9089. 30. Schmelmer, U., Jordan, R., Geyer, W., Eck, W., Golzhauser, A., Grunze, M., and Ulman, A. (2003) Surface-initiated polymerization on self-assembled monolayers: amplification of patterns on the micrometer and nanometer scale. Angew. Chem. Int. Ed. 42, 559. 31. Ahn, S. J., Kaholek, M., Lee, W. K., LaMattina, B., LaBean, T. H., and Zauscher, S. (2004) Surface-initiated polymerization on nanopatterns fabricated by electron-beam lithography. Adv. Mater. 16, 2141. 32. Kaholek, M., Lee, W. K., LaMattina, B., Caster, K. C., and Zauscher, S. (2004) Fabrication of stimulus-responsive nanopatterned polymer brushes by scanning-probe lithography. Nano Lett. 4, 373. 33. Maynor, B. W., Filocamo, S. F., Grinstaff, M. W., and Liu, J. (2002) Direct-writing of polymer nanostructures: poly(thiophene) nanowires on semiconducting and insulating surfaces. J. Am. Chem. Soc. 124, 522. 34. Hecht, S. (2003) Welding, organizing, and planting organic molecules on substrate surfaces—promising approaches towards nanoarchitectonics from the bottom up. Angew. Chem. Int. Ed. 42, 24. 35. Okawa, Y. and Aono, M. (2001) Materials science—nanoscale control of chain polymerization. Nature 409, 683. 36. von Werne, T. A., Germack, D. S., Hagberg, E. C., Sheares, V. V., Hawker, C. J., and Carter, K. R. (2003) A versatile method for tuning the chemistry and size of nanoscopic features by living free radical polymerization. J. Am. Chem. Soc. 125, 3831. 37. Boyen, H. G., Kastle, G., Zurn, K., Herzog, T., Weigl, F., Ziemann, P., Mayer, O., Jerome, C., et al. (2003) A micellar route to ordered arrays of magnetic nanoparticles: from size-selected pure cobalt dots to cobalt-cobalt oxide core-shell systems. Adv. Funct. Mater. 13, 359. 38. Spatz, J. P., Chan, V. Z. H., Mossmer, S., Kamm, F. M., Plettl, A., Ziemann, P., and Moller, M. (2002) A combined top-down/bottom-up approach to the microscopic localization of metallic nanodots. Adv. Mater. 14, 1827. 39. Sundrani, D., Darling, S. B., and Sibener, S. J. (2004) Guiding polymers to perfection: macroscopic alignment of nanoscale domains. Nano Lett. 4, 273. 40. Cheng, J. Y., Zhang, F., Chuang, V. P., Mayes, A. M., and Ross, C. A. (2006) Selfassembled one-dimensional nanostructure arrays. Nano Lett. 6, 2099. 41. Cheng, J. Y., Zhang, F., Smith, H. I., Vancso, G. J., and Ross, C. A. (2006) Pattern registration between spherical block-copolymer domains and topographical templates. Adv. Mater. 18, 597. 42. Li, H. W. and Huck, W. T. S. (2004) Ordered block-copolymer assembly using nanoimprint lithography. Nano Lett. 4, 1633.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

409

43. Rockford, L., Liu, Y., Mansky, P., Russell, T. P., Yoon, M., and Mochrie, S. G. J. (1999) Polymers on nanoperiodic, heterogeneous surfaces. Phys. Rev. Lett. 82, 2602. 44. Yang, X. M., Peters, R. D., Nealey, P. F., Solak, H. H., and Cerrina, F. (2000) Guided selfassembly of symmetric diblock copolymer films on chemically nanopatterned substrates. Macromolecules 33, 9575. 45. Stoykovich, M. P., Cao, H. B., Yoshimoto, K., Ocola, L. E., and Nealey, P. F. (2003) Deformation of nanoscopic polymer structures in response to well-defined capillary forces. Adv. Mater. 15, 1180. 46. Stoykovich, M. P., Muller, M., Kim, S. O., Solak, H. H., Edwards, E. W., de Pablo, J. J., and Nealey, P. F. (2005) Directed assembly of block copolymer blends into nonregular device-oriented structures. Science 308, 1442. 47. Kim, S. O., Kim, B. H., Kim, K., Koo, C. M., Stoykovich, M. P., Nealey, P. F., and Solak, H. H. (2006) Defect structure in thin films of a lamellar block copolymer self-assembled on neutral homogeneous and chemically nanopatterned surfaces. Macromolecules 39, 5466. 48. Edwards, E. W., Montague, M. F., Solak, H. H., Hawker, C. J., and Nealey, P. F. (2004) Precise control over molecular dimensions of block-copolymer domains using the interfacial energy of chemically nanopatterned substrates. Adv. Mater. 16, 1315. 49. Xia, Y. N., Yin, Y. D., Lu, Y., and McLellan, J. (2003) Template-assisted self-assembly of spherical colloids into complex and controllable structures. Adv. Funct. Mater. 13, 907. 50. Zhou, D., Bruckbauer, A., Ying, L. M., Abell, C., and Klenerman, D. (2003) Building three-dimensional surface biological assemblies on the nanometer scale. Nano Lett. 3, 1517. 51. Allard, M., Sargent, E. H., Lewis, P. C., and Kumacheva, E. (2004) Colloidal crystals grown on patterned surfaces. Adv. Mater. 16, 1360. 52. Kumacheva, E., Golding, R. K., Allard, M., and Sargent, E. H. (2002) Colloid crystal growth on mesoscopically patterned surfaces: effect of confinement. Adv. Mater. 14, 221. 53. Fudouzi, H., Kobayashi, M., and Shinya, N. (2002) Site-controlled deposition of microsized particles using an electrostatic assembly. Adv. Mater. 14, 1649. 54. Cui, T. H., Hua, F., and Lvov, Y. (2004) Lithographic approach to pattern multiple nanoparticle thin films prepared by layer-by-layer self-assembly for microsystems. Sensors Actuators A 114, 501. 55. Zhang, X. P., Sun, B. Q., Friend, R. H., Guo, H. C., Nau, D., and Giessen, H. (2006) Metallic photonic crystals based on solution-processible gold nanoparticles. Nano Lett. 6, 651. 56. Ye, Y. H., Badilescu, S., Truong, V. V., Rochon, P., and Natansohn, A. (2001) Selfassembly of colloidal spheres on patterned substrates. Appl. Phys. Lett. 79, 872. 57. Schaak, R. E., Cable, R. E., Leonard, B. M., and Norris, B. C. (2004) Colloidal crystal microarrays and two-dimensional superstructures: a versatile approach for patterned surface assembly. Langmuir 20, 7293. 58. Ozin, G. A. and Yang, S. M. (2001) The race for the photonic chip: colloidal crystal assembly in silicon wafers. Adv. Funct. Mater. 11, 95. 59. Manandhar, P., Huang, L., Grubich, J. R., Hutchinson, J. W., Chase, P. B., and Hong, S. H. (2005) Highly selective directed assembly of functional actomyosin on Au surfaces. Langmuir 21, 3213.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

410

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

60. Velikov, K. P., Christova, C. G., Dullens, R. P. A., and Van Blaaderen, A. (2002) Layerby-layer growth of binary colloidal crystals. Science 296, 106. 61. Dziomkina, N. V., Hempenius, M. A., and Vancso, G. J. (2005) Symmetry control of polymer colloidal monolayers and crystals by electrophoretic deposition onto patterned surfaces. Adv. Mater. 17, 237. 62. Dziomkina, N. V. and Vancso, G. J. (2005) Colloidal crystal assembly on topologically patterned templates. Soft Matter 1, 265. 63. Lee, W. M., Pruzinsky, S. A., and Braun, P. V. (2002) Multi-photon polymerization of waveguide structures within three-dimensional photonic crystals. Adv. Mater. 14, 271. 64. Vossen, D. L. J., Fific, D., Penninkhof, J., van Dillen, T., Polman, A., and Van Blaaderen, A. (2005) Combined optical tweezers/ion beam technique to tune colloidal masks for nanolithography. Nano Lett. 5, 1175. 65. Van Blaaderen, A. (2003) Colloidal molecules and beyond. Science 301, 470. 66. Hoogenboom, J. P., Vossen, D. L. J., Faivre-Moskalenko, C., Dogterom, M., and Van Blaaderen, A. (2002) Patterning surfaces with colloidal particles using optical tweezers. Appl. Phys. Lett. 80, 4828. 67. Garcia-Santamaria, F., Miyazaki, H. T., Urquia, A., Ibisate, M., Belmonte, M., Shinya, N., Meseguer, F., and Lopez, C. (2002) Nanorobotic manipulation of microspheres for on-chip diamond architectures. Adv. Mater. 14, 1144. 68. Lyles, B. F., Terrot, M. S., Hammond, P. T., and Gast, A. P. (2004) Directed patterned adsorption of magnetic beads on polyelectrolyte multilayers on glass. Langmuir 20, 3028. 69. Gu, Z. Z., Fujishima, A., and Sato, O. (2002) Patterning of a colloidal crystal film on a modified hydrophilic and hydrophobic surface. Angew. Chem. Int. Ed. 41, 2068. 70. Jonas, U., del Campo, A., Kruger, C., Glasser, G., and Boos, D. (2002) Colloidal assemblies on patterned silane layers. Proc. Natl. Acad. Sci. USA 99, 5034. 71. Aizenberg, J. (2004) Crystallization in patterns: a bio-inspired approach. Adv. Mater. 16, 1295. 72. Kaltenpoth, G., Himmelhaus, M., Slansky, L., Caruso, F., and Grunze, M. (2003) Conductive core-shell particles: an approach to self-assembled mesoscopic wires. Adv. Mater. 15, 1113. 73. Masuda, Y., Itoh, T., and Koumoto, K. (2005) Self-assembly patterning of silica colloidal crystals. Langmuir 21, 4478. 74. Fustin, C. A., Glasser, G., Spiess, H. W., and Jonas, U. (2003) Site-selective growth of colloidal crystals with photonic properties on chemically patterned surfaces. Adv. Mater. 15, 1025. 75. Fustin, C. A., Glasser, G., Spiess, H. W., and Jonas, U. (2004) Parameters influencing the templated growth of colloidal crystals on chemically patterned surfaces. Langmuir 20, 9114. 76. Zheng, H. P., Rubner, M. F., and Hammond, P. T. (2002) Particle assembly on patterned “plus/minus” polyelectrolyte surfaces via polymer-on-polymer stamping. Langmuir 18, 4505. 77. Jacobs, H. O., Campbell, S. A., and Steward, M. G. (2002) Approaching nanoxerography: the use of electrostatic forces to position nanoparticles with 100 nm scale resolution. Adv. Mater. 14, 1553.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

411

78. Suzuki, M., Yasukawa, T., Mase, Y., Oyamatsu, D., Shiku, H., and Matsue, T. (2004) Dielectrophoretic micropatterning with microparticle monolayers covalently linked to glass surfaces. Langmuir 20, 11005. 79. Feng, J., Wang, B., Gao, C. Y., and Shen, J. C. (2004) Selective adsorption of microcapsules on patterned polyelectrolyte multilayers. Adv. Mater. 16, 1940. 80. Ng, H. T., Fang, A. P., Huang, L. Q., and Li, S. F. Y. (2002) Protein microarrays on ITO surfaces by a direct covalent attachment scheme. Langmuir 18, 6324. 81. Lee, I., Zheng, H. P., Rubner, M. F., and Hammond, P. T. (2002) Controlled cluster size in patterned particle arrays via directed adsorption on confined surfaces. Adv. Mater. 14, 572. 82. Cherniavskaya, O., Adzic, A., Knutson, C., Gross, B. J., Zang, L., Liu, R., and Adams, D. M. (2002) Edge transfer lithography of molecular and nanoparticle materials. Langmuir 18, 7029. 83. Liu, S. T., Maoz, R., and Sagiv, J. (2004) Planned nanostructures of colloidal gold via self-assembly on hierarchically assembled organic bilayer template patterns with in-situ generated terminal amino functionality. Nano Lett. 4, 845. 84. Liu, S. T., Maoz, R., Schmid, G., and Sagiv, J. (2002) Template guided self-assembly of [Au5(5)] clusters on nanolithographically defined monolayer patterns. Nano Lett. 2, 1055. 85. Tzeng, S. D., Lin, K. J., Hu, J. C., Chen, L. J., and Gwo, S. (2006) Templated selfassembly of colloidal nanoparticles controlled by electrostatic nanopatterning on a Si3 N4 /SiO2 /Si electret. Adv. Mater. 18, 1147. 86. Kraus, T., Malaquin, L., Delamarche, E., Schmid, H., Spencer, N. D., and Wolf, H. (2005) Closing the gap between self-assembly and microsystems using self-assembly, transfer, and integration of particles. Adv. Mater. 17, 2438. 87. Arsenault, A., Fournier-Bidoz, S., Hatton, B., Miguez, H., Tetreault, N., Vekris, E., Wong, S., Yang, S. M., et al. (2004) Towards the synthetic all-optical computer: science fiction or reality? J. Mater. Chem. 14, 1778. 88. Yao, J. M., Yan, X., Lu, G., Zhang, K., Chen, X., Jiang, L., and Yang, B. (2004) Patterning colloidal crystals by lift-up soft lithography. Adv. Mater. 16, 81. 89. Yan, X., Yao, J. M., Lu, G., Li, X., Zhang, J. H., Han, K., and Yang, B. (2005) Fabrication of non-close-packed arrays of colloidal spheres by soft lithography. J. Am. Chem. Soc. 127, 7688. 90. Choi, D. G., Jang, S. G., Yu, H. K., and Yang, S. M. (2004) Two-dimensional polymer nanopattern by using particle-assisted soft lithography. Chem. Mater. 16, 3410. 91. Choi, J., Zhao, Y. H., Zhang, D. Y., Chien, S., and Lo, Y. H. (2003) Patterned fluorescent particles as nanoprobes for the investigation of molecular interactions. Nano Lett. 3, 995. 92. Yan, Q. F., Chen, A., Chua, S. J., and Zhao, X. S. (2005) Incorporation of point defects into self-assembled three-dimensional colloidal crystals. Adv. Mater. 17, 2849. 93. Cayre, O. J. and Paunov, V. N. (2004) Fabrication of microlens arrays by gel trapping of self-assembled particle monolayers at the decane-water interface. J. Mater. Chem. 14, 3300. 94. Maury, P., Mahalingam, V., Reinhoudt, D. N., and Huskens, J. (2005) Patterned selfassembled monolayers on silicon oxide prepared by nanoimprint lithography and their applications in nanofabrication. Adv. Funct. Mater. 15, 451.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

412

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

95. Maury, P., Escalante, M., Reinhoudt, D. N., and Huskens, J. (2005) Directed assembly of nanoparticles onto polymer-imprinted or chemically patterned templates fabricated by nanoimprint lithography. Adv. Mater. 17, 2718. 96. Mahalingam, V., Onclin, S., Peter, M., Ravoo, B. J., Huskens, J., and Reinhoudt, D. N. (2004) Directed self-assembly of functionalized silica nanoparticles on molecular printboards through multivalent supramolecular interactions. Langmuir 20, 11756. 97. Maury, P., Peter, M., Crespo-Biel, O., Ling, X. Y., Reinhoudt, D. N., and Huskens, J. (2007) Patterning the molecular printboard: patterning cyclodextrin monolayers on silicon oxide using nanoimprint lithography and its application in 3D multilayer nanostructuring. Nanotechnology 18, 044007. 98. Ludden, M. J. W., Reinhoudt, D. N., and Huskens, J. (2006) Molecular printboards: versatile platforms for the creation and positioning of supramolecular assemblies and materials. Chem. Soc. Rev. 35, 1122. 99. Huskens, J. (2006) Multivalent interactions at interfaces. Curr. Opin. Biotechnol. 10, 537. 100. Crespo-Biel, O., Ravoo, B. J., Huskens, J., and Reinhoudt, D. N. (2006) Writing with molecules on molecular printboards. Dalton Trans., 2737–2741. 101. Hammond, P. T. (2004) Form and function in multilayer assembly: new applications at the nanoscale. Adv. Mater. 16, 1271. 102. Clark, S. L. and Hammond, P. T. (1998) Engineering the microfabrication of layer-bylayer thin films. Adv. Mater. 10, 1515. 103. Jeon, N. L., Choi, I. S., Whitesides, G. M., Kim, N. Y., Laibinis, P. E., Harada, Y., Finnie, K. R., Girolami, G. S., et al. (1999) Patterned polymer growth on silicon surfaces using microcontact printing and surface-initiated polymerization. Appl. Phys. Lett. 75, 4201. 104. Liang, Z. Q. and Wang, Q. (2004) Multilayer assembly and patterning of poly(pphenylenevinylene)s via covalent coupling reactions. Langmuir 20, 9600. 105. Lee, S. W., Sanedrin, R. G., Oh, B. K., and Mirkin, C. A. (2005) Nanostructured polyelectrolyte multilayer organic thin films generated via parallel dip-pen nanolithography. Adv. Mater. 17, 2749. 106. Nyamjav, D. and Ivanisevic, A. (2004) Properties of polyelectrolyte templates generated by dip-pen nanolithography and microcontact printing. Chem. Mater. 16, 5216. 107. Pallandre, A., Moussa, A., Nysten, B., and Jonas, A. M. (2006) Nanoconfined polyelectrolyte multilayers. Adv. Mater. 18, 481. 108. Hatzor, A. and Weiss, P. S. (2001) Molecular rulers for scaling down nanostructures. Science 291, 1019. 109. Park, J. and Hammond, P. T. (2004) Multilayer transfer printing for polyelectrolyte multilayer patterning: direct transfer of layer-by-layer assembled micropatterned thin films. Adv. Mater. 16, 520. 110. Mallwitz, F. and Laschewsky, A. (2005) Direct access to stable, freestanding polymer membranes by layer-by-layer assembly of polyelectrolytes. Adv. Mater. 17, 1296. 111. Crespo-Biel, O., Dordi, B., Maury, P., Peter, M., Reinhoudt, D. N., and Huskens, J. (2006) Patterned, hybrid, multilayer nanostructures based on multivalent supramolecular interactions. Chem. Mater. 18, 2545. 112. Crespo-Biel, O., Dordi, B., Reinhoudt, D. N., and Huskens, J. (2005) Supramolecular layer-by-layer assembly: alternating adsorptions of guest- and host-functionalized

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

113.

114.

115. 116. 117. 118. 119. 120.

121.

122.

123. 124. 125.

126.

127.

128.

129.

413

molecules and particles using multivalent supramolecular interactions. J. Am. Chem. Soc. 127, 7594. Maury, P., Crespo-Biel, O., Peter, M., Reinhoudt, D. N., and Huskens, J. (2006) Integration of top-down and bottom-up nanofabrication schemes. Mater. Res. Soc. Symp. Proc. 901E, 09010-RB12-01.1. Huskens, J., Maury, P., Crespo-Biel, O., Peter, M., and Reinhoudt, D. N. (2006) Fabrication of 3D hybrid nanostructures by an integrated process comprising nanoimprint lithography and layer-by-layer assembly. Proc. Inst. Mech. Eng. Part N: J. Nanoeng. Nanosys. 220, 157. Birner, A., Wehrspohn, R. B., Gosele, U. M., and Busch, K. (2001) Silicon-based photonic crystals. Adv. Mater. 13, 377. Lopez, C. (2003) Materials aspects of photonic crystals. Adv. Mater. 15, 1679. Norris, D. J., Arlinghaus, E. G., Meng, L. L., Heiny, R., and Scriven, L. E. (2004) Opaline photonic crystals: how does self-assembly work? Adv. Mater. 16, 1393. Thylen, L., Qiu, M., and Anand, S. (2004) Photonic crystals—a step towards integrated circuits for photonics. ChemPhysChem 5, 1268. Yacaman, M. J., Ascencio, J. A., Liu, H. B., and Gardea-Torresdey, J. (2001) Structure shape and stability of nanometric sized particles. J. Vac. Sci. Technol. B 19, 1091. Tetreault, N., Mihi, A., Miguez, H., Rodriguez, I., Ozin, G. A., Meseguer, F., and Kitaev, V. (2004) Dielectric planar defects in colloidal photonic crystal films. Adv. Mater. 16, 346. Jin, C. J., Li, Z. Y., McLachlan, M. A., McComb, D. W., De La Rue, R. M., and Johnson, N. P. (2006) Optical properties of tetragonal photonic crystal synthesized via templateassisted self-assembly. J. Appl. Phys. 99, 116109. Jin, C. J., McLachlan, M. A., McComb, D. W., De La Rue, R. M., and Johnson, N. P. (2005) Template-assisted growth of nominally cubic (100)-oriented three-dimensional crack-free photonic crystals. Nano Lett. 5, 2646. Miguez, H., Yang, S. M., and Ozin, G. A. (2003) Optical properties of colloidal photonic crystals confined in rectangular microchannels. Langmuir 19, 3479. Yang, S. M., Miguez, H., and Ozin, G. A. (2002) Opal circuits of light—planarized microphotonic crystal chips. Adv. Funct. Mater. 12, 425. Ye, J. H., Zentel, R., Arpiainen, S., Ahopelto, J., Jonsson, F., Romanov, S. G., and Torres, C. M. S. (2006) Integration of self-assembled three-dimensional photonic crystals onto structured silicon wafers. Langmuir 22, 7378. Ferrand, P., Egen, M., Zentel, R., Seekamp, J., Romanov, S. G., and Torres, C. M. S. (2003) Structuring of self-assembled three-dimensional photonic crystals by direct electron-beam lithography. Appl. Phys. Lett. 83, 5289. Ferrand, P., Minty, M. J., Egen, M., Ahopelto, J., Zentel, R., Romanov, S. G., and Torres, C. M. S. (2003) Micromoulding of three-dimensional photonic crystals on silicon substrates. Nanotechnology 14, 323. Miguez, H., Yang, S. M., Tetreault, N., and Ozin, G. A. (2002) Oriented free-standing three-dimensional silicon inverted colloidal photonic crystal microribers. Adv. Mater. 14, 1805. Ferrand, P., Egen, M., Griesebock, B., Ahopelto, J., Muller, M., Zentel, R., Romanov, S. G., and Torres, C. M. S. (2002) Self-assembly of three-dimensional photonic crystals on structured silicon wafers. Appl. Phys. Lett. 81, 2689.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

414

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

130. Matsuo, S., Fujine, T., Fukuda, K., Juodkazis, S., and Misawa, H. (2003) Formation of free-standing micropyramidal colloidal crystals grown on silicon substrate. Appl. Phys. Lett. 82, 4283. 131. Brozell, A. M., Muha, M. A., and Parikh, A. N. (2005) Formation of spatially patterned colloidal photonic crystals through the control of capillary forces and template recognition. Langmuir 21, 11588. 132. Yan, Q. F., Zhao, X. S., Teng, J. H., and Chua, S. J. (2006) Colloidal woodpile structure: three-dimensional photonic crystal with a dual periodicity. Langmuir 22, 7001. 133. Ferrand, P., Seekamp, J., Egen, M., Zentel, R., Romanov, S. G., and Torres, C. M. S. (2004) Direct electron-beam lithography on opal films for deterministic defect fabrication in three-dimensional photonic crystals. Microelectron. Eng. 73–74, 362. 134. Juarez, B. H., Golmayo, D., Postigo, P. A., and Lopez, C. (2004) Selective formation of inverted opals by electron-beam lithography. Adv. Mater. 16, 1732. 135. Jonsson, F., Torres, C. M. S., Seekamp, J., Schniedergers, M., Tiedemann, A., Ye, J. H., and Zentel, R. (2005) Artificially inscribed defects in opal photonic crystals. Microelectron. Eng. 78–79, 429. 136. Palacios-Lidon, E., Galisteo-Lopez, J. F., Juarez, B. H., and Lopez, C. (2004) Engineered planar defects embedded in opals. Adv. Mater. 16, 341. 137. Wang, L. K., Yan, Q. F., and Zhao, X. S. (2006) From planar defect in opal to planar defect in inverse opal. Langmuir 22, 3481. 138. Yan, Q. F., Zhou, Z. C., Zhao, X. S., and Chua, S. J. (2005) Line defects embedded in three-dimensional photonic crystals. Adv. Mater. 17, 1917. 139. Vekris, E., Kitaev, V., von Freymann, G., Perovic, D. D., Aitchison, J. S., and Ozin, G. A. (2005) Buried linear extrinsic defects in colloidal photonic crystals. Adv. Mater. 17, 1269. 140. Yan, Q. F., Zhou, Z. C., and Zhao, X. S. (2005) Introduction of three-dimensional extrinsic defects into colloidal photonic crystals. Chem. Mater. 17, 3069. 141. Jun, Y. H., Leatherdale, C. A., and Norris, D. J. (2005) Tailoring air defects in selfassembled photonic bandgap crystals. Adv. Mater. 17, 1908. 142. Taton, T. A. and Norris, D. J. (2002) Device physics: defective promise in photonics. Nature 416, 685. 143. Fudouzi, H. and Xia, Y. N. (2003) Colloidal crystals with tunable colors and their use as photonic papers. Langmuir 19, 9653. 144. Fudouzi, H. and Xia, Y. N. (2003) Photonic papers and inks: color writing with colorless materials. Adv. Mater. 15, 892. 145. Shipway, A. N., Katz, E., and Willner, I. (2000) Nanoparticle arrays on surfaces for electronic, optical, and sensor applications. ChemPhysChem 1, 18. 146. Zhu, H. and Snyder, M. (2003) Protein chip technology. Curr. Opin. Chem. Biol. 7, 55. 147. Maes, H. E., Claeys, C., Mertens, R., Campitelli, A., Van Hoof, C., and De Boeck, J. (2001) Trends in microelectronics, optical detectors, and biosensors. Adv. Eng. Mater. 3, 781. 148. Klefenz, H. (2004) Nanobiotechnology: from molecules to systems. Eng. Life Sci. 4, 211. 149. de Gans, B. J., Duineveld, P. C., and Schubert, U. S. (2004) Inkjet printing of polymers: state of the art and future developments. Adv. Mater. 16, 203.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

415

150. Falconnet, D., Koenig, A., Assi, T., and Textor, M. (2004) A combined photolithographic and molecular-assembly approach to produce functional micropatterns for applications in the biosciences. Adv. Funct. Mater. 14, 749. 151. Agheli, H., Malmstrom, J., Larsson, E. M., Textor, M., and Sutherland, D. S. (2006) Large area protein nanopatterning for biological applications. Nano Lett. 6, 1165. 152. Kung, L. A., Kam, L., Hovis, J. S., and Boxer, S. G. (2000) Patterning hybrid surfaces of proteins and supported lipid bilayers. Langmuir 16, 6773. 153. Khademhosseini, A., Jon, S., Suh, K. Y., Tran, T. N. T., Eng, G., Yeh, J., Seong, J., and Langer, R. (2003) Direct patterning of protein- and cell-resistant polymeric monolayers and microstructures. Adv. Mater. 15, 1995. 154. Garno, J. C., Amro, N. A., Wadu-Mesthrige, K., and Liu, G. Y. (2002) Production of periodic arrays of protein nanostructures using particle lithography. Langmuir 18, 8186. 155. Bergman, A. A., Buijs, J., Herbig, J., Mathes, D. T., Demarest, J. J., Wilson, C. D., Reimann, C. T., Baragiola, R. A., et al. (1998) Nanometer-scale arrangement of human serum albumin by adsorption on defect arrays created with a finely focused ion beam. Langmuir 14, 6785. 156. Trau, D., Jiang, J., and Sucher, N. J. (2006) Preservation of the biofunctionality of DNA and protein during microfabrication. Langmuir 22, 877. 157. Nicolau, D. V., Taguchi, T., Taniguchi, H., and Yoshikawa, S. (1998) Micron-sized protein patterning on diazonaphthoquinone/novolak thin polymeric films. Langmuir 14, 1927. 158. Cao, T. B., Wei, F., Jiao, X. M., Chen, J. Y., Liao, W., Zhao, X., and Cao, W. X. (2003) Micropatterns of protein and conducting polymer molecules fabricated by layer-by-layer self-assembly and photolithography techniques. Langmuir 19, 8127. 159. Valsesia, A., Colpo, P., Meziani, T., Lisboa, P., Lejeune, M., and Rossi, F. (2006) Immobilization of antibodies on biosensing devices by nanoarrayed self-assembled monolayers. Langmuir 22, 1763. 160. Kumar, N. and Hahm, J. I. (2005) Nanoscale protein patterning using self-assembled diblock copolymers. Langmuir 21, 6652. 161. Shirahata, N., Yonezawa, T., Miura, Y., Kobayashi, K., and Koumoto, K. (2003) Patterned adsorption of protein onto a carbohydrate monolayer immobilized on Si. Langmuir 19, 9107. 162. Wadu-Mesthrige, K., Xu, S., Amro, N. A., and Liu, G. Y. (1999) Fabrication and imaging of nanometer-sized protein patterns. Langmuir 15, 8580. 163. Cai, Y. G. and Ocko, B. M. (2005) Large-scale fabrication of protein nanoarrays based on nanosphere lithography. Langmuir 21, 9274. 164. Sorribas, H., Padeste, C., and Tiefenauer, L. (2002) Photolithographic generation of protein micropatterns for neuron culture applications. Biomaterials 23, 893. 165. Veiseh, M., Wickes, B. T., Castner, D. G., and Zhang, M. Q. (2004) Guided cell patterning on gold-silicon dioxide substrates by surface molecular engineering. Biomaterials 25, 3315. 166. Veiseh, M., Zareie, M. H., and Zhang, M. Q. (2002) Highly selective protein patterning on gold-silicon substrates for biosensor applications. Langmuir 18, 6671. 167. Yap, F. L. and Zhang, Y. (2005) Protein micropatterning using surfaces modified by self-assembled polystyrene microspheres. Langmuir 21, 5233.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

416

14:9

Printer Name: Yet to Come

COMBINATIONS OF TOP-DOWN AND BOTTOM-UP NANOFABRICATION TECHNIQUES

168. Lundqvist, M., Sethson, I., and Jonsson, B. H. (2004) Protein adsorption onto silica nanoparticles: conformational changes depend on the particles’ curvature and the protein stability. Langmuir 20, 10639. 169. Vertegel, A. A., Siegel, R. W., and Dordick, J. S. (2004) Silica nanoparticle size influences the structure and enzymatic activity of adsorbed lysozyme. Langmuir 20, 6800. 170. Valsesia, A., Colpo, P., Meziani, T., Bretagnol, F., Lejeune, M., Rossi, F., Bouma, A., and Garcia-Parajo, M. (2006) Selective immobilization of protein clusters on polymeric nanocraters. Adv. Funct. Mater. 16, 1242. 171. Howell, S. W., Inerowicz, H. D., Regnier, F. E., and Reifenberger, R. (2003) Patterned protein microarrays for bacterial detection. Langmuir 19, 436. 172. Morhard, F., Pipper, J., Dahint, R., and Grunze, M. (2000) Immobilization of antibodies in micropatterns for cell detection by optical diffraction. Sensors Actuators 70, 232. 173. Haupt, K. and Mosbach, K. (1998) Plastic antibodies: developments and applications. Trends Biotechnol. 16, 468. 174. Clemmens, J., Hess, H., Lipscomb, R., Hanein, Y., Bohringer, K. F., Matzke, C. M., Bachand, G. D., Bunker, B. C., et al. (2003) Mechanisms of microtubule guiding on microfabricated kinesin-coated surfaces: chemical and topographic surface patterns. Langmuir 19, 10967. 175. Ivanova, E. P., Wright, J. P., Pham, D., Filipponi, L., Viezzoli, A., and Nicolau, D. V. (2002) Polymer microstructures fabricated via laser ablation used for multianalyte protein microassay. Langmuir 18, 9539. 176. Bruckbauer, A., Zhou, D. J., Kang, D. J., Korchev, Y. E., Abell, C., and Klenerman, D. (2004) An addressable antibody nanoarray produced on a nanostructured surface. J. Am. Chem. Soc. 126, 6508. 177. Orth, R. N., Clark, T. G., and Craighead, H. G. (2003) Avidin-biotin micropatterning methods for biosensor applications. Biomed. Microdevices 5, 29. 178. Biebricher, A., Paul, A., Tinnefeld, P., Golzhauser, A., and Sauer, M. (2004) Controlled three-dimensional immobilization of biomolecules on chemically patterned surfaces. J. Biotechnol. 112, 97. 179. Falconnet, D., Pasqui, D., Park, S., Eckert, R., Schift, H., Gobrecht, J., Barbucci, R., and Textor, M. (2004) A novel approach to produce protein nanopatterns by combining nanoimprint lithography and molecular self-assembly. Nano Lett. 4, 1909. 180. Hoff, J. D., Cheng, L. J., Meyhofer, E., Guo, L. J., and Hunt, A. J. (2004) Nanoscale protein patterning by imprint lithography. Nano Lett. 4, 853. 181. Hoff, J. D. and Hunt, A. (2003) Microscale protein patterning via nanoimprint lithography. Biophys. J. 84, 292A. 182. Michel, R., Reviakine, I., Sutherland, D., Fokas, C., Csucs, G., Danuser, G., Spencer, N. D., and Textor, M. (2002) A novel approach to produce biologically relevant chemical patterns at the nanometer scale: selective molecular assembly patterning combined with colloidal lithography. Langmuir 18, 8580. 183. Jung, J. M., Kwon, K. Y., Ha, T. H., Chung, B. H., and Jung, H. T. (2006) Goldconjugated protein nanoarrays through block-copolymer lithography: from fabrication to biosensor design. Small 2, 1010. 184. Hyun, J., Ahn, S. J., Lee, W. K., Chilkoti, A., and Zauscher, S. (2002) Molecular recognition-mediated fabrication of protein nanostructures by dip-pen lithography. Nano Lett. 2, 1203.

P1: OTA/XYZ P2: ABC c15 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

417

185. Doh, J. and Irvine, D. J. (2004) Photogenerated polyelectrolyte bilayers from an aqueous-processible photoresist for multicomponent protein patterning. J. Am. Chem. Soc. 126, 9170. 186. Holden, M. A. and Cremer, P. S. (2003) Light activated patterning of dye-labeled molecules on surfaces. J. Am. Chem. Soc. 125, 8074. 187. Valiokas, R., Klenkar, G., Tinazli, A., Tampe, R., Liedberg, B., and Piehler, J. (2006) Differential protein assembly on micropatterned surfaces with tailored molecular and surface multivalency. ChemBioChem 7, 1325. 188. Gamsjaeger, R., Wimmer, B., Kahr, H., Tinazli, A., Picuric, S., Lata, S., Tampe, R., Maulet, Y., et al. (2004) Oriented binding of the His(6)-tagged carboxyl-tail of the Ltype Ca2+ channel α(1)-subunit to a new NTA-functionalized self-assembled monolayer. Langmuir 20, 5885. 189. Kato, K., Sato, H., and Iwata, H. (2005) Immobilization of histidine-tagged recombinant proteins onto micropatterned surfaces for cell-based functional assays. Langmuir 21, 7071. 190. Yu, T. Y., Wang, Q., Johnson, D. S., Wang, M. D., and Ober, C. K. (2005) Functional hydrogel surfaces: binding kinesin-based molecular motor proteins to selected patterned sites. Adv. Funct. Mater. 15, 1303. 191. Montemagno, C. and Bachand, G. (1999) Constructing nanomechanical devices powered by biomolecular motors. Nanotechnology 10, 225. 192. Bachand, G. D., Soong, R. K., Neves, H. P., Olkhovets, A., Craighead, H. G., and Montemagno, C. D. (2001) Precision attachment of individual F-1-ATPase biomolecular motors on nanofabricated substrates. Nano Lett. 1, 42. 193. Soong, R. K., Bachand, G. D., Neves, H. P., Olkhovets, A. G., Craighead, H. G., and Montemagno, C. D. (2000) Powering an inorganic nanodevice with a biomolecular motor. Science 290, 1555. 194. Ludden, M. J. W., Peter, M., Reinhoudt, D. N., and Huskens, J. (2006) Attachment of streptavidin to β-cyclodextrin molecular printboards via orthogonal host-guest proteinligand interactions. Small 2, 1192. 195. Ludden, M. J. W., Mulder, A., Tamp´e, R., Reinhoudt, D. N., and Huskens, J. (2007) Molecular printboards as a general platform for protein immobilization: a supramolecular solution to nonspecific adsorption. Angew. Chem. Int. Ed. 46, 4104. 196. Maury, P., Escalante, M., Peter, M., Reinhoudt, D. N., Subramaniam, V., and Huskens, J. (2007) Creating nanopatterns of His-tagged proteins on surfaces by nanoimprint lithography using specific NiNTA–histidine interactions. Small 3, 1584.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16 ORGANIC ELECTRONIC DEVICES

16.1 INTRODUCTION Organic optoelectronics represents an area to which unconventional patterning techniques can uniquely contribute. Often, traditional photolithography cannot be used because, for example, the developer, which is a solvent that is typically used to dissolve photoresists in photolithography, can easily damage or degrade the organic layers in the device. For this reason, and many other related ones, it is simply undesirable to use photolithography for patterning organic layers. The materials used for organic devices are either small molecule organics or polymers. The former and latter materials are typically deposited by vacuum evaporation and solution processing, respectively. With vapor phase deposition, the organic layers are often patterned with the use of a shadow mask [1]. In fact, organic lightemitting diodes in commercial production are fabricated in this manner. This method, however, has limited resolution and it is difficult to apply it over large areas. For solution-based materials, ink-jet printing is often used as a route to printing of allpolymer transistors [2] and other types of devices. Ink-jet printing, however, has limited resolution, relatively poor pattern fidelity and nonuniform deposits, particularly for large areas. Tailored light-emitting prepolymers that can be cross-linked by ultraviolet (UV) light have been used for patterning pixels with the aid of a shadow mask [3]. Certain of the problems associated with the conventional use of the shadow mask apply here too. The problems associated with pattern fidelity and uniformity, resolution and large area applicability can be avoided, to a large extent, with the unconventional lithographies based on molds, stamps and conformable photomasks, as demonstrated recently and discussed at the end of the chapter.

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


419

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

420

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

In this chapter, applications of the unconventional patterning techniques described in the previous chapters to organic electronics and optoelectronics are presented. The devices include organic light-emitting diode (OLED) and organic thin film transistor (OTFT). Each type has its own unique characteristics to consider. Therefore, the two types of the organic devices are treated separately, first OLED and then OTFT.

16.2 ORGANIC LIGHT-EMITTING DIODES An OLED can be implemented in simple device structures. The simplest involves a thin film of an organic light-emitting material sandwiched between two film electrodes corresponding to the cathode and anode. Perhaps the first example of the application of an unconventional patterning technique was for this simplest OLED [4] in which a thickness contrast of the emitting layer (EML) gained by solution-assisted micromolding [5] was utilized to create spatial contrast in the brightness. To reduce the energy barrier against hole transport at the anode–EML interface, a hole injection layer (HIL) and hole transport layer (HTL) are inserted between the anode and EML, thereby enhancing the device efficiency. Similarly, an electron transport layer (ETL) and electron injection layer (EIL) can be inserted between the cathode and EML such that the multilayer structure can take on the structure of anode/HIL/HTL/EML/ETL/EIL/cathode. The anode is usually indium tin oxide (ITO) deposited on a glass substrate. For flexible applications, transparent plastic sheets coated with ITO can be used. Photolithography is often applicable for patterning the anode. The film thickness of any layer in the multilayer structure including the electrodes is about 100 nm or less. Although such multilayer designs can be used for OLEDs based on small molecule organics, the solvent compatibility can limit the number of layers that can be inserted, into a polymer-based device. In fact, most devices include just a single layer, because the solvent used for coating a polymer film on another polymer layer can damage the underlying layer. Partly as a result, the device efficiency of the polymer-based OLED is much poorer than that of the small molecule OLED. The patterning needs for OLEDs depend on the type of scheme by which the OLEDs are driven for display applications. When the OLEDs are driven by a passive matrix scheme [6], only the cathode and anode need to be patterned. If the OLEDs are driven by an active matrix scheme in which transistors drive the OLEDs, the patterning needs are more involved. In addition to the anode and cathode patterning, pixels of red (R), green (G), and blue (B) OLEDs for full color display have to be patterned and isolated from one another. Since the anode patterning can often be carried out with photolithography, the cathode patterning is considered first. To pattern the cathode, a separator [7, 8] can be utilized, in particular for passive or dot matrix display. Separators are rows of lines with a reverse tapered trapezoidal cross-section. They are fabricated on a patterned anode surface such that the rows of the separators are perpendicular to the anode lines. Sequential blanket depositions of organic materials and cathode metal complete the passive matrix formation for

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.2 ORGANIC LIGHT-EMITTING DIODES

421

Rigifiex or hard mold SAM Aluminum Organic Layers Glass substrate

ITO

Pressing and releasing

Figure 16.1. Schematic diagram of cathode transfer patterning. A prepatterned and pretreated mold onto which aluminum is deposited is placed on the surface of organic layers deposited onto an ITO substrate and then pressed. The difference in adhesion force between mold–aluminum and organic layer–aluminum allows the aluminum on the mold to be transferred to the surface of the organic layer upon pressing and then releasing the pressure. (Reprinted with permission from [11]. Copyright 2002 American Institute of Physics.)

the display. Because of the reverse tapered shape, the deposited films are discontinuous at the edges of the separators, thereby isolating one row of OLEDs from another. Although photolithography can form the separators [7, 8], the method of micromolding in capillaries [9], as described in Chapter 3, offers powerful capabilities for the fabrication [10]. General purpose patterning of the cathode can be accomplished by the transfer patterning discussed in Chapter 4. Such an approach as applied to cathode patterning methods is illustrated in Figure 16.1. Although a glass mold was used in the original work [11], a rigiflex mold [12] can be implemented to pattern sub100-nm features, if needed, and for wide area applicability. Typically, a mold having the desired cathode pattern is coated with a self-assembled monolayer (SAM) material such as (tridecafluoro-1,1,2,2-tetrahydrooctyl)trichlorosilane to lower the work of adhesion between the mold and the metal layer to be transferred. The cathode metal is then deposited on the SAM-coated mold surface. The mold with the SAM and metal layers is brought into contact with the organic surface and pressure is applied to the backside of the mold. Releasing the pressure and removing the mold completes the transfer. The transfer relies on the difference in the adhesion strength at the two interfaces: the metal–SAM interface and the metal–organic interface. The weaker the adhesion strength at the metal–SAM interface compared with that at the metal–organic interface, the more effective the technique becomes.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

422

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

Therefore, the role of SAM is in lowering the adhesion strength at the mold–SAM interface. Comparisons for a blue-emitting OLED showed that there are essentially no differences in the device performance [11] whether the cathode is patterned by the transfer technique or by thermal evaporation through a shadow mask. However, an anomaly appears in the current density–voltage characteristics around the turn-on voltage. This feature might be due to less than perfect contact between the metal and the underlying organic surface. A distinct advantage of transfer patterning the cathode over other methods is new opportunities for engineering the surface properties of the cathode. It is well known that treating the anode surface with oxygen plasma [13] or SF6 [14] leads to a significant improvement in the device performance. Since the metal surface on the mold can be treated before the transfer, similar strategies can be applied to the cathode surface, as a possible route to improved device performance. The subtractive transfer patterning method discussed in Chapter 4 has also been applied to patterning the cathode [15]. In one case, a silicon substrate patterned with lines and spaces served as a mold. This mold was brought into contact with an OLED surface coated with Al–Au at a pressure of 15 MPa. Upon releasing the pressure and lifting the mold, the metal in contact with the raised parts of the mold adheres to the mold, leaving behind patterned metal stripes or cathodes on the organic surface. Although the anode is usually patterned by photolithography and subsequent wet etching of ITO, imprint lithography can also be used [16–18]. This procedure is illustrated in Figure 16.2a. A poly(methylmethacrylate) (PMMA) thin film (M w = 15,000 g mol−1 ) is first spin cast onto a ITO/poly(ethylene terephthalate)

(a)

Molding

(b)

Si mold PMMA ITO PET RIE

10 mm ITO etching ( b)

PMMA removal

Figure 16.2. OLED fabrication using imprint lithography. (a) Schematic illustration of ITO anode patterning. (b) Optical image of OLEDs (emitting area = 65 × 70 mm2 ) with structure of ITO/CuPc/NPB/Alq3/LiF/Al on a flexible PET substrate. (c ) Image of the device in a rolled configuration. (Reprinted with permission from [16]. Copyright 2005 IEEE.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.2 ORGANIC LIGHT-EMITTING DIODES

423

(PET) substrate. A photolithographically patterned silicon mold and the PMMA are then pressed together and heated at 140◦ C for 2 h above the glass transition temperature of the polymer, with pressure of 0.29 MPa. The melt polymer flows into the mold patterns and conformally wets the surface. The pattern on the mold becomes physically imprinted into the PMMA after separation at ambient temperature, such that the PMMA residue in the compressed region can be removed by oxygen reactive ion etching (RIE) to open the underlying ITO surface. The exposed ITO is then removed by wet etching with oxalic acid. Stripping the PMMA resist completes the process. Sequential deposition of an HIL of copper phthalocyanine (CuPc), an HTL of N,N  -di(naphthalen-1-yl)-N,N  -diphenylbenzidine (NPB), an electron transport and emission layer of tris-(8-hydroxyquinoline)aluminum (Alq3 ), and a cathode layer of lithium fluoride (LiF)/Al onto the PET substrate with the patterned ITO anodes generates arrayed pixels of OLEDs. Figure 16.2b shows a photograph for two seven-segment light-emitting patterns on the flexible plastic for a numerical display (emitting area = 65 × 70 mm2 ). The luminous efficiency reaches 1.13 lm W−1 (3.04 cd A−1 ) at a luminance of 3.8 cd m−2 , and luminance increases to a maximum of 244 cd m−2 at a drive voltage of 30 V. This display can operate to a bending radius of curvature of 1 cm, as shown in Figure 16.2c [16]. The isolation achieved by the patterned anode alternatively can be realized by patterning an insulator film on the ITO [19]. Unconventional methods used for the insulator patterning include microcontact printing [20] and room-temperature imprint lithography [21]. In the case of microcontact printing, the process was carried out at 80◦ C to facilitate the reaction between a SAM and ITO [22]. After the printing, blanket depositions of all the layers for OLED were performed. Because the SAM acts as an insulator, light emission occurs only through the area that is not covered by SAM. A similar approach is possible with room-temperature imprinting [23]. In one example, polystyrene (PS) features consisting of lines (80 nm) and spaces (450 nm) were imprinted with a silicon mold. The 80-nm-wide OLEDs showed electrical and optical characteristics that are close to those of wide-area OLEDs. However, the periodic arrayed device layout caused a slight shift in electroluminescence spectrum and anisotropic far-field radiation due to scattering. Control over the area through which light is emitted can also be gained by patterning organic layers in the multilayer structure of OLED such as the HIL or HTL [24]. As discussed in Chapter 4, subtractive transfer patterning involving a small molecule organic layer is straightforward and quite simple. A poly(dimethylsiloxane) (PDMS) mold can be placed on the organic layer to be patterned without applying any pressure. Annealing at 90◦ C and then peeling away the mold completes the patterning. The parts of the organic layer that are in contact with the raised parts of the mold adhere to the mold, resulting in subtractive transfer patterning. The effectiveness of this patterning approach is shown in Figure 16.3. The HTL of 4,4 -bis[N-1-naphthyl-Nphenyl-amino]biphenyl (NPB or α-NPD) in a multilayer structure for green-emitting OLED based on Alq3 was patterned using three different PDMS mold patterns. After the NPB patterning, blanket depositions of all the layers for the G OLED were carried out. The dark, nonemissive area is where NPB was removed by subtractive transfer patterning.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

424

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

(a)

50 µm

(b)

50 µm

(c)

30 µm

Figure 16.3. Optical microscopy images of emission from OLED devices fabricated by subtractive transfer patterning. Green light is observed from the regions where NPB (50 nm thick) is not removed. Nonemissive areas are defined with various PDMS mold patterns: (a) 100-µm lines and circles, (b) 10-µm lines and 40-µm spaces, and (c ) 60 × 20-µm2 boxes. (Reprinted with permission from [24]. Copyright 2005 Wiley-VCH Verlag.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.2 ORGANIC LIGHT-EMITTING DIODES

425

One efficient way of fabricating flexible OLEDs is to utilize soft contact lamination [25–27] in which two PDMS sheets are brought together to form the devices. On one sheet, an ITO layer is deposited followed by spin- coating of an EML. On the other sheet, which is treated with oxygen plasma, the cathode is defined by either thermal evaporation of metal through a shadow mask or soft lithography. When the two sheets are brought together, van der Waals forces pull the electrodes into intimate contact with the EML layer at room temperature, without application of external pressure. Typically, this contact initiates on one side of the structure, a wetting front then progresses naturally across the sample until the entire surface is in contact. As an example of cathode patterning using soft lithography, Figure 16.4 demonstrates three different approaches: (i) a PDMS sheet with surface relief can generate cathode patterns on the areas contacted with the EML layer (Figure 16.4a), (ii) a blanketdeposited cathode on a flat or structured PDMS surface can be chemically modified by microcontact printing patterns of insulating thiols (C18) as barriers to charge injection (Figure 16.4b), and (iii) phase-shift lithography with conformable masks and etching procedures can be used to pattern directly the cathode on the PDMS to feature sizes down to 150 nm in width (Figure 16.4c). Each of these approaches achieves high contrast patterned emission after lamination with a polyfluorene derivative as the electroluminescent layer [25]. The performance of the device using lamination can be better than the one fabricated by thermal evaporation of metal on EML, presumably because the lamination avoids the detrimental effects caused by metal evaporation that give rise to nonradiative decay channels for excitons. A unique opportunity that the transfer patterning methods discussed in Chapter 4 present is the ability to form a solution-based layer on an existing organic layer. In particular, the method can be used where the solvent would otherwise dissolve or damage the underlying organic layer. An example is the formation of an HIL of poly(3,4-ethylenedioxythiophene)–poly(4-styrenesulfonate) (PEDOT–PSS) on NPB layer for a top emitting OLED. Because of damage that an aqueous solution of PEDOT-PSS can cause, typically an insoluble layer such as pentacene is deposited on the underlying organic layer prior to coating with PEDOT-PSS [28]. With transfer patterning, PEDOT-PSS can be delivered directly on NPB [29]. On a mold such as the one in Figure 16.1, an anode of Au is deposited first followed by spin-coating of PEDOT-PSS. After baking, a bilayer of PEDOT on Au is transferred to the underlying layer of NPB. In this transfer process, the anode is automatically patterned since the mold pattern can be made to correspond to that of the anode. Unconventional patterning techniques have also been applied to the fabrication of full color OLEDs. In a process based on dry dye printing [30], the RGB pixels are defined by introducing R and G dyes into a polymer that contains B dye. As shown in Figure 16.5a, a dye coated on glass substrate is patterned to serve as a printing plate. To prepare the plate, a layer of photoresist on a backing layer is photolithographically patterned, and then laminated onto a glass substrate with the dye layer coated on its surface. After peeling off the backing layer, the dye layer is patterned by an oxygen plasma with the laminated photoresist as an etch mask. This etch mask is removed by peeling off to complete the preparation of the printing plate. As shown in Figure 16.5b, the printing plate is used as an “inked” mold for microcontact printing

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

426

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

(a)

Stamp

(b)

Printed c18

(c) Phase shift hν PDMS Photoresist Au PDMS

EL emission

PDMS

Contrast

Developing

Au etching 0 PDMS

5

10

15

20

Distance (µm)

Figure 16.4. Patterned OLEDs fabricated by the combined use of soft contact lamination and soft lithography. (a) Molding features of relief into the PDMS, followed by blanket deposition of the electrode before lamination, generates emission in the pattern of the raised features. (b) Microcontact printing an insulating SAM (octadecanethiol: C18) before lamination yields emission only in the bare Au regions (scale bar = 100 µm). (c ) Phase-shift lithography with a conformable mask followed by etching directly pattern lines (width = 150 nm) of Au on the PDMS. The average linewidth (∼600 nm) of the pattern of emission (graph below) is comparable to the resolution of the optical imaging system (scale bar = 5 µm). (Reprinted with permission from [25]. Copyright 2004 The National Academy of Science of the USA.) The left inset in (c ) illustrates the procedure of the phase-shift lithography. (Reprinted with permission from [27]. Copyright 2005 Wiley-VCHVerlag.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.2 ORGANIC LIGHT-EMITTING DIODES

(a)

(b)

Backing layer RISTON phtoresist (i) Capping layer

427

Exposure

Device polymer blue dye Develop

(i)

ITO Display substrate

(ii)

Printing plate

Dye source Glass substrate Laminate

(ii)

ITO Display substrate

(iii)

Dye source Glass substrate

Printing plate

Peel off backing layer

(iv)

Dye

(iii)

Glass substrate Etch dye source using O2 plasma Al foli

(v)

(iv)

ITO Display substrate

ITO Display substrate

Glass substrate Remove RISTON photoresist

Figure 16.5. (a) Fabrication of printing plate: (i) pattern RISTON dry photoresist with backing layer remaining as support, (ii) laminate patterned RISTON dry photoresist onto the surface of the dye-doped polymer layer (together with backing layer), (iii) peel off the backing layer to expose the pattern on the dry photoresist, (iv) etch dye-doped polymer using oxygen plasma, and (v) remove the dry photoresist by laminating and peeling off an aluminum foil. (b) Schematic process flow of dry printing from prepatterned dye sources and subsequent solvent-enhanced dye diffusion for RGB patterns: (i) uniform spin coat of device polymer with blue dye, (ii) transfer of green dye from prepatterned dye source (at 70◦ C for 1 h), (iii) repeat the transfer step for red dye, and (iv) solvent-enhanced diffusion of dye throughout the polymer film. Vertical dimensions are exaggerated versus lateral dimensions. (Reprinted with permission from [30]. Copyright 2006 IEEE.)

the dyes on the polymer layer containing the B dye. To facilitate the transfer of the dye from the mold to the polymer surface, the sample is annealed at 70◦ C for 1 h. After the printing of R and G dyes, a drive-in process is executed in an acetone vapor ambient to diffuse the dyes throughout the bulk of the polymer film. As described at the beginning of this chapter, an OLED consists of multiple layers of organics. The transfer patterning processes discussed in Chapter 4 can be used not

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

428

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

Support film Rigiflex mold FEP AI cathode ITO

Organic multilayer

Glass substrate

Red-emitting organic multilayer

Blue-emitting organic multilayer

Green-emitting organic multilayer

Figure 16.6. Schematic illustration of procedure for full color pixelation by whole device transfer. The device transfer is repeated three times with alignment. Three separate PUA molds for red, green, and blue devices are used for the transfer. (Reprinted with permission from [31]. Copyright 2006 IOP Publishing Ltd.)

only for the transfer of one layer but also for multiple layers. Therefore, the method can be used to realize entire device transfer for R, G, and B pixels [31]. In one example the multilayer structure of the OLED including the cathode is transferred onto a patterned ITO anode on a glass substrate. As shown in Figure 16.6, fluorinated ethylene propylene is first deposited on a rigiflex mold to reduce the work of adhesion at the mold interface, followed by sequential depositions of the aluminum cathode and organic multilayer stacks. Device transfer is accomplished by pressing the coated mold onto the substrate at a pressure ranging from 1 to 5 MPa at 50◦ C for 5 min. After cooling to room temperature, the mold is simply removed, thereby transferring the device structure on the protruding parts of the mold to the substrate. After the R pixel is fabricated, the same procedure can be followed for aligning and transferring G and B OLEDs for RGB pixelation. The performance of the devices thus fabricated is equivalent to those formed in the conventional way [31]. The emission properties of and light extraction from OLEDs can be improved by introducing a photonic crystal [32] or a periodic structure [33] on the back of

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.3 ORGANIC THIN FILM TRANSISTORS

429

the glass substrate that is typically used for OLEDs. While unconventional techniques have not yet been utilized for fabricating such structures, this area is one for which these techniques could be used for easy and cost-effective realization of various structures including microlens arrays.

16.3 ORGANIC THIN FILM TRANSISTORS An OTFT usually consists of four layers including the gate electrode, gate insulator, semiconductor, and source/drain electrodes. A bottom gate OTFT is one in which the gate is patterned first on a substrate such as glass or plastic sheet. A top contact device results when an insulator and a semiconductor are deposited sequentially on the gate, followed by patterning of a conductor to form the source and drain electrodes. A bottom contact OTFT is one in which the source and drain are defined on an insulator coated on the gate, on which the semiconductor is then coated. The device performance of the top contact OTFT is typically better than that of the bottom contact OTFT [34]. The OTFT is referred to as top gate device if the source and drain are defined first on a substrate on which a semiconductor and an insulator layer are deposited sequentially, followed by patterning of a conductor film on the insulator for the gate electrode. The bottom gate OTFT is often preferred because of detrimental effects that a solvent can have on a semiconductor when a polymer insulator is coated on top of it in the course of fabricating the top gate device. When an array of OTFTs is fabricated, the insulator layer often covers the whole array area. The gate or the source/drain layer, depending on whether the bottom or top gate structure is used, is typically prepatterned either by photolithography or by an unconventional technique. Patterning of the source and drain electrodes, which defines the channel length, is more demanding than that of the gate electrode because the channel length is often much smaller than the gate size. Therefore, patterning of the source and drain electrodes and of the semiconductor layer are of main interest in the fabrication of OTFTs. A number of the unconventional patterning techniques described in this book can be used for OTFTs. These include microcontact printing (µCP) [20], soft contact lamination [35], imprint lithography [36], and transfer printing/transfer patterning. Of these, the µCP is the most widely explored. The µCP method can be used in several different ways. The pattern generated by the printing can serve as an etch mask or it can become a patterned active area or it can define the electrodes. It can also be used to create a medium by which selective deposition is possible. When the pattern generated by printing is used as an etch mask, the process is typically carried out on a metal film coated on a substrate for the purpose of defining the source and drain electrodes. This patterning and subsequent etching [37] are illustrated in Figure 16.7. A PDMS mold or stamp with the desired pattern for the source and drain and any necessary interconnections is inked with a molecular species capable of forming a SAM on the metal such as hexadecanethiol (HDT). This inked stamp is placed into contact with the surface of a gold film on a substrate

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

430

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

Ink stamp

Inked stamp •

Au (200A)

Patterned monolayer of HDT

Microcontact print

Etch unprotected Au

Remove monolayer of HDT

Figure 16.7. Schematic illustration of steps for using microcontact printing to define source and drain electrodes for organic transistors that use semiconductor material deposited on top of the source and drain. Contact of an inked (≈1 mM hexadecanethiol in ethanol) elastomeric stamp ˚ produces a patterned SAM. Etching the (PDMS) with the surface of a thin film of gold (≈200 A) gold not protected by the SAM produces a pattern of gold with the geometry of the surface relief on the stamp. Application of heat or exposure to UV light or an oxygen plasma removes the SAM. (Reprinted with permission from [37]. Copyright 1999 Wiley-VCH Verlag.)

and then removed, leaving on the gold surface a patterned SAM of HDT. With the patterned SAM as etch mask, the gold film is etched with an etching solution. The SAM is then removed by application of heat or exposure to UV light or an oxygen plasma, completing the patterning for the source and drain electrodes. In many cases, the substrate is a silicon wafer, which serves as a gate, with a top oxide layer

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.3 ORGANIC THIN FILM TRANSISTORS

431

such that the deposition of a semiconductor directly onto the patterned electrodes leads to a bottom contact, bottom gate OTFT. Also, the same bottom contact, bottom gate geometry can be implemented on plastic substrates such as PET and polyimide for flexible electronic devices. In this case, the gate is defined first on the plastic, which is then coated with an insulator before patterning of the source and drain by the procedure in Figure 16.7. Simple deposition of a semiconductor yields the bottom contact, bottom gate OTFT. As an example, Figure 16.8a shows a plastic active matrix backplane circuit (12 × 12 cm) with an array of 256 pentacene transistors fabricated in this manner [38, 39]. Here the substrate, gate, insulator, and source/drain were PET, ITO, organosilsesquioxane spin on glass, and Au/Ti, respectively. The

(a)

25 µm

(b)

Figure 16.8. Photographs of (a) a flexible active matrix backplane (12 × 12 cm) of pentacene OTFTs with bottom contacts patterned by microcontact printing and etching, and (b) electronicpaper-like display that uses electrophoretic ink and the active matrix backplane. (Reprinted with permission from [38]. Copyright 2001 The National Academy of Science of the USA.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

432

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

pentacene semiconductor was isolated by evaporation through a shadow mask. This plastic backplane was used to electrophoretic ink for a first demonstration of a flexible electronic-paper-like display as shown in Figure 16.8b. Just as a SAM can be printed on a substrate, a semiconductor or a conductor in a solvent can also be printed to form patterned active areas or source and drain electrodes. On a substrate that supports gate, insulator, and source/drain (S/D) electrodes, a PDMS stamp inked with a polymer solution of poly(3-hexylthiophene) (P3HT) can be aligned and placed into contact with the S/D electrodes to produce bottom contact OTFTs [40]. A PDMS stamp can also be inked with a conducting polymer solution such as PEDOT to define S/D electrodes [41] on an oxide surface, onto which pentacene can be deposited for a bottom contact OTFT. Energetically heterogeneous surfaces either generated by surface treatment or inherently present have been exploited in silicon technology for selective deposition. Perhaps the first application of this concept to OTFTs is one in which the channel or S/D electrodes are defined by the selective deposition in fabricating all polymer transistors [2]. In this case, hydrophobic strips of polyimide (PI) are photolithographically defined on a hydrophilic glass substrate. Ink jet printing an acqueous solution of PEDOT-PSS onto such a pattern leads to dewetting from the PI. This process defines PEDOT-PSS S/D electrodes separated by distances defined by the width of the PI strips. A simple and convenient way of preparing such an energetically heterogeneous surface is µCP. A SAM material is typically used in this application. To define the gate on a hydrophilic plastic substrate, the surface can be stamped with a fluorinated silane by µCP, which renders the stamped region hydrophobic. Upon carrying out electroless plating of nickel, the growth occurs only on the hydrophilic region for the gate formation [42]. In another approach, when a hydrophilic surface is patterned to have hydrophilic and hydrophobic regions through stamping with a SAM, the hydrophilic conductor polyaniline (PANI) remains only on the hydrophilic region upon spin-coating the surface with PANI, thereby defining the S/D electrodes [43]. The same technique can be used to coat selectively the hydrophobic regions by dipcoating with the polymer semiconductor of P3HT [44], thereby defining the active areas. In yet another application of µCP for selective deposition [45], a SiO2 surface is patterned with a thick octadecyltriethoxysilane (OTS). When organic vapor source is passed over the patterned SiO2 surface, organic crystals nucleate and grow only on the OTS-stamped region. This selectivity in the deposition was attributed to the roughness created by the thick OTS film. Simultaneous patterning of both the S/D electrodes and the active area can be accomplished for polymer OTFTs in a process called soft molding in reverse, as shown in Figure 16.9 [46]. A known amount of the polymer semiconductor solution is first dispensed into a trapezoidal void in a PDMS mold. The substrate that supports gate and insulator levels is placed on the coated mold so that the center of the trapezoid is aligned with the gate. Flipping over the substrate–mold gives the arrangement shown in the second step of Figure 16.9. The solvent in the polymer solution starts evaporating in the void, and permeates out of the mold. Because of the evaporation, the polymer solution slowly thickens and solidifies. This molding process leads to the structure shown in the third step of the figure, thereby defining the patterned active

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.3 ORGANIC THIN FILM TRANSISTORS

(a)

433

P3HT in toulene/drop PDMS mold

(b) Polymer gate dielectric AI gate Galss substrate

(c)

(d )

P3HT active layer

Au source/drain

Figure 16.9. Schematic illustration of soft molding in reverse. (a) P3HT solution in toluene is dispensed on a PDMS mold to fill trapezoidal void. (b) Glass substrate that supports gate and dielectric is then placed on the PDMS mold and the substrate–mold pair is flipped over. (c ) After several minutes, the toluene solvent evaporates completely, and only P3HT layer with tapered edges remains, thereby defining the active area. (d ) After removing the mold, a blanket deposition of metal defines patterned source/drain electrodes. (Reprinted with permission from [46]. Copyright 2006 Wiley-VCHVerlag.)

area. A blanket deposition of metal over this molded structure leads to S/D electrodes separated by the distance between the two protruding edges. Imprint lithography or embossing can also be used for the patterning. One distinct advantage of imprinting over µCP is the resolution that rigid molds used for imprinting provide, especially for defining the channel length. In one example, a P3HT-TFT with channel length as small as 70 nm was demonstrated using gold source and drain on a silicon wafer (gate) with 5-nm-thick SiO2 (insulator) [47]. Scanning electron microscopic images in Figure 16.10 illustrate high quality electrodes and channels with uniform dimensions of 70 nm and sharp edges. In a conceptually related method, a substrate with a metal film can be embossed with a V-shaped microgroove by pressing the groove against a thin metal film, thereby creating a vertical channel. This technique has been used to fabricate self-aligned, vertical polymer OTFTs [48]. Soft contact lamination [35], as discussed previously for OLEDs, is a useful technique for fabricating plastic circuits. Usually one PDMS sheet supports the S/D level

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

434

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

2 µm

L = 70 nm

200 nm Figure 16.10. Scanning electron microscopic images of a finger-type OTFT, with channel length of 70 nm and channel width of 4 µm, fabricated using imprint lithography. (Reprinted with permission from [47]. Copyright 2002 American Institute of Physics.)

and the other sheet supports the levels of gate, dielectric insulator, and organic semiconductor. These two sheets are laminated after alignment to produce plastic circuits, as shown in Figure 16.11. Spin-coating a thin (25–50-µm-thick) layer of PDMS against a fluorinated flat glass plate produces a thin PDMS sheet. A PET sheet coated with ITO, is contacted with the PDMS sheet on the glass after the PDMS and ITO surfaces are treated with an oxygen plasma, which produces an irreversible bond between the PDMS and the ITO. Peeling back the PET releases the PDMS from the glass plate. This transfer cast film of PDMS on PET is again exposed to an oxygen plasma and then coated with a bilayer film of Ti (∼1 nm) and Au (∼15 nm). The Ti enhances the adhesion to the plasma-treated surface of the PDMS. Patterning with µCP yields S/D electrodes of Ti/Au and completes the fabrication of the upper sheet or top substrate. The bottom substrate is a PET sheet that supports the other parts of circuit such as gate (patterned ITO), dielectric (spin on glass), and organic semiconductor. Aligning the top substrate with the bottom one and then bringing them together completes the circuit. Initiating contact at an edge by slightly bending one

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.3 ORGANIC THIN FILM TRANSISTORS

435

(a)

(b)

(c)

(d)

Figure 16.11. Schematic illustration of steps for building embedded plastic circuits by lamination. (a) Fabrication begins by transfer casting a thin film of the elastomer PDMS (25–50 µm thick) onto ITO/Mylar (≈175 µm thick), oxidizing the exposed PDMS surface, and depositing uniform layers of Ti (≈1 nm, adhesion promoter) and Au (15–20 nm). (b) Microcontact printing on the goldcoated PDMS film defines the transistor source/drain electrodes and related interconnections (insets provide magnified views of a pair of source/drain electrodes and a side profile of an electrode), and plasma oxidation produces hydroxyl groups on the exposed surface of the PDMS. (c ) Aligning and laminating this sheet to a bottom plastic substrate (PET; ∼175 µm thick) that supports the semiconductor, dielectric, and gate levels produces a complete circuit with top contact transistors (contacts at the edges of the bottom sheet enable electrical connection to the embedded circuit). (d ) Insets provide schematic views of a typical transistor and a side profile of a laminated electrode. (Reprinted with permission from [35]. Copyright 2002 The National Academy of Science of the USA.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

436

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

of the substrates and then allowing contact to proceed gradually across the circuit provides a convenient way to laminate over large areas without creating trapped air pockets. The thin layer of PDMS “wets” the bottom substrate to enable this intimate contact without the use of external pressure to force the two parts together. The lamination produces the circuit and at the same time imbeds it between two PET sheets. The electrical properties of the laminated transistors are similar to those observed in top contact devices that use shadow mask electrodes deposited directly on the semiconductor. While the top substrate in Figure 16.11 supports only patterned S/D electrodes, which leads to top contact OTFTs, it can also include the semiconductor film [49], which results in bottom contact OTFTs. The top substrate can be a single crystal semiconductor in place of a PDMS supported on PET. This type of lamination has been used to study charge transport in organic single crystals [50–55]. Although organic single crystals can offer intrinsic electronic properties, device fabrication requires unconventional processes to avoid damage caused by physical vacuum deposition of metals or insulators and other effects (i.e., swelling, doping) caused by chemicals such as solvents or wet etchants. Soft lithographic type lamination procedures, as illustrated in Figure 16.12, can overcome this challenge. In this approach, thin layers of metal are evaporated onto a PDMS stamp to produce electrically isolated S/D electrodes on the raised regions and a gate electrode on the recessed area. By laminating an organic single crystal onto the S/D electrodes, a transistor with an air or “free-space” dielectric is formed in a simple process that is completely noninvasive to the organic. The free-space dielectric eliminates effects caused by defects,

PDMS stamp

(a)

Master

(b)

Source

Single crystal

Ti/Au coating

Drain

Gate

5 mm Figure 16.12. Schematic illustration of the steps to fabricate organic single-crystal transistors with free-space gate dielectrics using PDMS molding and lamination. (Reprinted with permission from [51]. Copyright 2004 Wiley-VCH Verlag.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

16.3 ORGANIC THIN FILM TRANSISTORS

437

traps, and other phenomena that can occur at the semiconductor–dielectric interface. Mobilities in the range of 10–20 cm2 (Vs)−1 are possible, depending on crystallographic orientations, with rubrene organic single crystals [50, 51]. Lamination processes that exploit both donor and receiver sheets have also been used for thermal imaging [56]. The donor sheet is coated with a conducting polymer. When it is subjected to a laser beam that provides heat, it decomposes the surrounding organics creating a bubble. The expansion caused by the bubble propels the conducting layer onto the receiver. After the thermal imaging, the donor sheet is separated from the receiver. An approach that has some similarities to this imaging technique involves transfer printing S/D electrodes onto a PDMS substrate [57] as illustrated in Figure 16.13. Layers of Au (20 nm) then Ti (1 nm) are then evaporated onto a stamp. Both the

(a)

Stamp 0.05–100 µm

Deposit Au/Ti on stamp

(b) 0.2–10 µm

0.1–10 cm Plasma oxidize surface of stamp, substrate; print Substrate

(c)

Si Ti

TiOx chemically bonds to substrate

(d) Removal substrate; Au/Ti printing complete (e)

Figure 16.13. Schematic of procedure for nanotransfer printing of electrode patterns that can be used in organic devices. (a) A stamp with relief features defines the geometry of the pattern. The overall stamp dimensions are between 0.1 and 10 cm, the relief is in the range of 0.2–10 µm, and the lateral dimensions of the features are between 0.05 and 100 µm. (b) Twenty nanometer Au and 5 nm Ti are uniformly evaporated on the stamp. (c ) Both the stamp and the substrate are plasma oxidized, resulting in (−OH) groups on the surfaces; they are then brought into contact. (d ) In the regions of physical contact, an interfacial reaction produces strong chemical bonds that bind the metal on the stamp to the substrate. (e) Separating the stamp and substrate results in complete transfer of the Au/Ti pattern from the raised regions of the stamp to the substrate. (Reprinted with permission from [57]. Copyright 2002 American Institute of Physics.)

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

438

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

stamp and PDMS substrate surfaces are exposed to an oxygen plasma to create hydroxyl termination. Placing the metal-coated stamp on the substrate leads to intimate, conformal contact between the raised regions of the stamp and the substrate, without the application of any external pressure. A condensation reaction between the hydroxyl groups results in Ti–O–Si bonds at the interface. Peeling the substrate and stamp apart transfers the Au/Ti bilayer from the raised regions of the stamp to the substrate. This transfer-printed sheet can be laminated with another sheet supporting the other levels of OTFT to yield devices. As discussed in Chapter 4, printing or patterning by transferring a layer on a stamp or mold onto a substrate is always possible provided two conditions are met. The first is that the surface of the layer to be transferred has to be in intimate physical contact with that of the substrate. The other is that the work of adhesion at the mold–layer interface has to be smaller than that at the layer–substrate interface. The first condition is always satisfied if either the mold or the substrate is uncoated PDMS because of its unique physical properties. For all other cases, external pressure must be applied for intimate contact. The pressure needed for intimate contact depends on the flexibility of either the mold or the substrate, the pressure being lower for more flexible mold or substrate. The second condition on the work of adhesion can be satisfied in most cases by depositing a material of low surface energy such as fluorinated compounds on the mold before coating it with the layer to be transferred. As described in Chapter 4, kinetic effects can also be exploited when elastomeric materials are used for the stamps, molds or substrates, as illustrated in Figure 16.13. Using related methods, even organic semiconductors can be transferred to substrates, for example, that support gate electrodes and polymer dielectrics [58]. After the transfer, forming S/D electrodes on the transferred pentacene leads to an OTFT. As described in Chapter 2, use of a rigiflex mold [12] enables sub-100-nm features to be transferred over large areas with a moderate level of applied pressure. Such molds have been used successfully to pattern layers of pentacene for OTFTs [59], with sub-100 nm resolution. The process is similar to that shown in Figure 16.1 except that the material coated on the rigiflex mold is pentacene instead of aluminum. The raised part of the mold can be designed such that the transfer defines the active semiconductor area for OTFT. As indicated earlier, transfer techniques such as these offer the unique ablity to pattern a solution-based organic layer on an existing organic layer. For example, on a substrate that supports gate and polymer dielectric, a polymer semiconductor such as P3HT can be transferred from a mold [60] to form a polymer OTFT. It is always possible to spin-coat a polymer semiconductor on a polymer dielectric if the solvent for the semiconductor does not dissolve the dielectric. A case in point is the dielectric poly(2-hydroxyethylmethacrylate) (PHEMA) [61]. The toluene solvent for P3HT does not dissolve PHEMA, for which the solvent is methanol. Therefore, P3HT can be spin-coated on PHEMA. However, the mobility of the polymer OTFT thus fabricated is about one order of magnitude smaller than that fabricated by transfer patterning. Although the solvent for P3HT does not dissolve PHEMA such that the solution processing is compatible, the solvent can still change the

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

439

microscopic morphology of the underlying polymer gate dielectric. The surface roughness thus generated leads to poor device performance [61]. Another advantage is that the dielectric can be annealed prior to the transfer to improve the device performance [60]. The transfer technique can also be used to pattern all levels of OTFT by sequentially transferring each level of the device one at a time [62]. Usually, a top gate pentacene transistor fabricated with a solution-based dielectric exhibits poor device performance because of the deleterious effects of the solvent on the underlying pentacene. An evaporated layer of pentacene on a spin-cast layer of polymer on a mold can be transferred to a substrate that supports S/D electrodes for the fabrication of a top gate pentacene transistor [63]. An added advantage of this bilayer transfer is that the polymer dielectric can be selected to maximize the pentacene grain size. Unconventional patterning techniques are particularly well suited for organic devices simply because conventional photolithography cannot be used for fabricating organic devices in many cases. A broad review of patterning techniques for organic electronic and optoelectronic systems is available in reference [64].

REFERENCES 1. Tian, P. F., Bulovic, V., Burrows, P. E., Gu, G., and Forrest, S. R. (1999) Precise, scalable shadow mask patterning of vacuum-deposited organic light emitting devices. J. Vac. Sci. Technol. A 17, 2975–2981. 2. Sirringhaus, H., Kawase, T., Friend, R. H., Shimoda, T., Inbasekaran, M., Wu, W., and Woo, E. P. (2000) High resolution inkjet printing of all-polymer transistor circuits. Science 290, 2123–2126. 3. Gather, C. M., Kohnen, A., Falcon, A., Becker, H., and Meerholz, K. (2007) Solutionprocessed full color polymer organic light-emitting diode displays fabricated by direct photolithography. Adv. Funct. Mater. 17, 191–200. 4. Rogers, J. A., Bao, Z., and Dhar, L. (1998) Fabrication of patterned electroluminescent polymers that emit in geometries with feature sizes into the submicron range. Appl. Phys. Lett. 73, 294–296. 5. Kim, E., Xia, Y., Zhao, X. M., and Whitesides, G. M. (1997) Solvent-assisted microcontact molding: a convenient method for fabricating three-dimensional structures on surfaces of polymers. Adv. Mater. 9, 651–654. 6. Gu, G. and Forrest, S. R. (1998) Design of flat-panel displays based on organic lightemitting devices. IEEE J. Sel. Top. Quantum Electron. 4, 83–99. 7. Tian, P. F., Burrows, P. E., and Forrest, S. R. (1997) Photolithographic patterning of vacuum-deposited organic light emitting devices. Appl. Phys. Lett. 71, 1397–1399. 8. Nagayama, K., Yahagi, T., Nakado, H., Tohma, T., Watanabe, T., Yoshida, K., and Miyaguchi, S. (1997) Micropatterning method for the cathode of the organic electroluminescent device. Japan. J. Appl. Phys. 36, L1555–L1557. 9. Kim, E., Xia, Y., and Whitesides, G. M. (1995) Polymer microstructures formed by moulding in capillaries. Nature 376, 581–584.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

440

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

10. Rhee, J., Park, J., Kwon, S. J., Yoon, H., and Lee, H. H. (2003) Fabrication of reversely tapered three-dimensional structures and their applications to organic light-emitting diodes. Adv. Mater. 15, 1075–1078. 11. Rhee, J. and Lee, H. H. (2002) Patterning organic light-emitting diodes by cathode transfer. Appl. Phys. Lett. 81, 4165–4167. 12. Suh, D., Choi, S. J., and Lee, H. H. (2005) Rigiflex lithography for nanostructure transfer. Adv. Mater. 17, 1554–1560. 13. Wu, C. C., Wu, C. I., Sturm, J. C., and Kahn, A. (1997) Surface modification of indium tin oxide by plasma treatment: an effective method to improve the efficiency, brightness, and reliability of organic light emitting devices. Appl. Phys. Lett. 70, 1348–1350. 14. Choi, B., Yoon, H., and Lee, H. H. (2000) Surface treatment of indium tin oxide by SF6 plasma for organic light emitting diodes. Appl. Phys. Lett. 76, 412–414. 15. Hoshino, K., Hasegawa, T., Matsumoto, K., and Shimoyama, I. (2006) Organic lightemitting diode micropatterned with a silicon convex stamp. Sensors Actuators A 128, 339–343. 16. Kao, P. C., Chu, S. Y., Chen, T. Y., Zhan, C. Y., Hong, F. C., Chang, C. Y., Hsu, L. C., Liao, W. C., et al. (2005) Fabrication of large-scaled organic light emitting devices on the flexible substrates using low-pressure imprinting lithography. IEEE Trans. Electron Devices 52, 1722–1726. 17. Cheng, X., Hong, Y., Kanicki, J., and Guo, L. J. (2002) High-resolution organic polymer light-emitting pixels fabricated by imprinting technique. J. Vac. Sci. Technol. B 20, 2877–2880. 18. Cheng, C. Y. and Hong. C. F. (2006) Fabrication of organic light-emitting diode arrays on flexible plastic substrates by imprint lithography. Japan. J. Appl. Phys. 45, 8915–8919. 19. Suganuma, N., Adachi, C., Koyama, T., Taniguchi, Y., and Shiraishi, H. (1999) A 200 nm × 2 mm array of organic light-emitting diodes and their anisotropic electroluminescence. Appl. Phys. Lett. 74, 1206–1208. 20. Kumar, A. and Whitesides, G. M. (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. 21. Khang, D. Y., Yoon, H., and Lee, H. H. (2001) Room temperature imprint lithography. Adv. Mater. 13, 749–752. 22. Koide, Y., Such, M. W., Basu, R., Evmenenko, G., Cui, J., Dutta, P., Hersam, M. C., and Marks, T. J. (2003) Hot microcontact printing for patterning ITO surfaces. Methodology, morphology, microstructure, and OLED charge injection barrier imaging. Langmuir 19, 86–93. 23. Suh, D. and Lee, H. H. (2004) Sub-100 nm organic light-emitting diodes patterned with room temperature imprint lithography. J. Vac. Sci. Technol. B 22, 1123–1126. 24. Choi, J., Kim, D., Yoo, P. J., and Lee, H. H. (2005) Simple detachment patterning of organic layers and its application to organic light-emitting diodes. Adv. Mater. 17, 166–171. 25. Lee, T. W., Zaumseil, J., Bao, Z., Hsu, J. W. P., and Rogers, J. A. (2004) Organic lightemitting diodes formed by soft contact lamination. Proc. Natl. Acad. Sci. 101, 429–433. 26. Liu, J., Lewis, L. N., Faircloth, T. J., and Duggal, A. R. (2006) High performance organic light-emitting diodes fabricated via a vacuum-free lamination process. Appl. Phys. Lett. 88, 223509.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

441

27. Lee, T. W., Jeon, S., Maria, J., Zaumseil, J., Hsu, J. W. P., and Rogers, J. A. (2005) Softcontact optical lithography using transparent elastomeric stamps: application to nanopatterned organic light-emitting devices. Adv. Funct. Mater. 15, 1435–1439. 28. Dobbertin, T., Werner, O., Meyer, J., Kammoun, A., Schneider, D., Riedl, T., Becker, E., Johannes, H. H., et al. (2003) Inverted hybrid organic light-emitting device with polyethylene dioxythiophene-polystyrene sulfonate as an anode buffer layer. Appl. Phys. Lett. 83, 5071–5073. 29. Seo, S., Kim, J. H., and Lee, H. H. (2006) Dry formation of polymer hole injection layer for top emitting organic light emitting diodes. Appl. Phys. Lett. 89, 253515. 30. Long, K., Pschenitzka, F., Lu, M. H., and Sturm, J. C. (2006) Full-color OLEDs integrated by dry dye printing. IEEE Trans. Electron Devices 53, 2250–2258. 31. Choi, J., Kim, K. H., Choi, S. J., and Lee, H. H. (2006) Whole device printing for full color displays with organic light-emitting diodes. Nanotechnology 17, 2246–2249. 32. Rocha, L., Dumarcher, V., Malcor, E., Fiorini, C., Denis, C., Raimond, P., Geffroy, B., and Nunzi, J. M. (2002) Photo-induced microstructured polymers for the optimisation and control of organic devices emission properties. Synth. Met. 127, 75–79. 33. Lee, Y. J., Kim, S. H., Huh, J., Kim, G. H., Lee, Y. H., Cho, S. H., Kim, Y. C., and Do, Y. R. (2003) A high-extraction-efficiency nanopatterned organic light-emitting diode. Appl. Phys. Lett. 82, 3779–3781. 34. Jackson, T. N., Lin, Y. Y., Gundlach, D. J., and Klauk, H. (1998) Organic thin-film transistors for organic light-emitting flat-panel display backplanes. IEEE J. Sel. Top. Quantum Electron. 4, 100–104. 35. Loo, Y. L., Someya, T., Baldwin, K. W., Bao, Z., Ho, P., Dodabalapur, A., Katz, H. E., and Rogers, J. A. (2002) Soft, conformable electrical contacts for organic semiconductors: high-resolution plastic circuits by lamination. Proc. Natl. Acad. Sci. 99, 10252– 10256. 36. Chou, S. Y., Kraus, P. R., and Renstrom, P. J. (1996) Imprint lithography with 25-nanometer resolution. Science 272, 85–87. 37. Rogers, J. A., Bao, Z., Makhija, A., and Braun, P. (1999) Printing process suitable for reel-to-reel production of high performance organic transistors and circuits. Adv. Mater. 11, 741–745. 38. Rogers, J. A., Bao, Z., Baldwin, K., Dodabalapur, A., Crone, B., Raju, V. R., Kack, V., Katz, H., et al. (2001) Paper-like electronic displays: large-area rubber-stamped plastic sheets of electronics and microencapsulated electrophoretic inks. Proc. Natl. Acad. Sci. 98, 4835–4840. 39. Bao, Z., Kuck, V., Rogers, J. A., and Paczkowski, M. A. (2002) Silsesquioxane resins as high-performance solution processible dielectric materials for organic transistor applications. Adv. Funct. Mater. 12, 526–531. 40. Park, S. K., Kim, Y. H., Han, J. I., Moon, D. G., and Kim, W. K. (2002) Highperformance polymer TFTs printed on a plastic substrate. IEEE Trans. Electron Devices 49, 2008–2015. 41. Li, D. and Guo, L. J. (2006) Micron-scale organic thin film transistors with conducting polymer electrodes patterned by polymer inking and stamping. Appl. Phys. Lett. 88, 063513. 42. Zschieschang, U., Klauk, H., Halik, M., Schmid, G., and Dehm, C. (2003) Flexible organic circuits with printed gate electrodes. Adv. Mater. 15, 1147–1151.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

442

14:9

Printer Name: Yet to Come

ORGANIC ELECTRONIC DEVICES

43. Lee, K. S., Blanchet, G. B., Gao, F., and Loo, Y. L. (2005) Direct patterning of conductive water-soluble polyaniline for thin-film organic electronics. Appl. Phys. Lett. 86, 074102. 44. Briseno, A. L., Roberts, M., Ling, M. M., Moon, H., Nemanick, E. J., and Bao, Z. (2006) Patterning organic semiconductors using “dry” poly(dimethylsiloxane) elastomeric stamps for thin film transistors. J. Am. Chem. Soc. 128, 3880–3881. 45. Briseno, A. L., Mannsfeld, S. C. B., Ling, M. M., Liu, S., Tseng, R. J., Reese, C., Roberts, M. E., Yang, Y., et al. (2006) Patterning organic single-crystal transistor arrays. Nature 444, 913–917. 46. Kang, H., Park, J., and Lee, H. H. (2006) Active area and channel formation for polymer thin film transistors. Adv. Mater. 18, 1603–1606. 47. Austin, M. D. and Chou, S. Y. (2002) Fabrication of 70 nm channel length polymer organic thin-film transistors using nanoimprint lithography. Appl. Phys. Lett. 81, 4431–4433. 48. Stutzmann, N., Friend, R. H., and Sirringhaus, H. (2003) Self-aligned, vertical-channel, polymer field-effect transistors. Science 299, 1881–1884. 49. Chabinyc, M. L., Salleo, A., Wu, Y., Liu, P., Ong, B. S., Heeney, M., and McCulloch, I. (2004) Lamination method for the study of interfaces in polymeric thin film transistors. J. Am. Chem. Soc. 126, 13928–13929. 50. Sundar, V. C., Zaumseil, J., Podzorov, V., Mennard, E., Willett, R. L., Someya, T., Gershenson, M. E., and Rogers, J. A. (2004) Elastomeric transistor stamps: reversible probing of charge transport in organic crystals. Science 303, 1644–1646. 51. Menard, E., Podzorov, V., Hur, S. H., Gauer, A., Gershenson, M. E., and Rogers, J. A. (2004) High-performance N- and P-type single-crystal organic transistors with free-space gate dielectrics. Adv. Mater. 16, 2097–2101. 52. Zeis, R., Besnard, C., Siegrist, T., Schlockermann, C., Chi, X., and Kloc, C. (2006) Field effect studies on rubrene and impurities of rubrene. Chem. Mater. 18, 244–248. 53. Molinari, A., Gutierrez, I., Hulea, I. N., Russo, S., and Morpurgo, A. F. (2007) Biasdependent contact resistance in rubrene single-crystal field-effect transistors. Appl. Phys. Lett. 90, 212103. 54. Takeya, J., Yamagishi, M., Tominari, Y., Hirahara, R., Nakazawa, Y., Nishikawa, T., Kawase, T., and Shimoda, T. (2007) Very high-mobility organic single-crystal transistors with in-crystal conduction channels. Appl. Phys. Lett. 90, 102120. 55. Hulea, I. N., Fratini, S., Xie, H., Mulder, C. L., Iossad, N. N., Rastelli, G., Ciuchi, S., and Morpurgo, A. F. (2006) Tunable Fr¨ohlich polarons in organic single-crystal transistors. Nat. Mater. 5, 982–986. 56. Blanchet, G. B., Loo, Y. L., Rogers, J. A., Gao, F., and Fincher, C. R. (2003) Large area, high resolution, dry printing of conducting polymers for organic electronics. Appl. Phys. Lett. 82, 463–465. 57. Loo, Y. L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Additive, nanoscale patterning of metal films with a stamp and a surface chemistry mediated transfer process: applications in plastic electronics. Appl. Phys. Lett. 81, 562–564. 58. Ofuji, M., Lovinger, A. J., Kloc, C., Siegrist, T., Maliakal, A. J., and Katz, H. E. (2005) Organic semiconductor designed for lamination transfer between polymer films. Chem. Mater. 17, 5748–5753. 59. Park, S. Y., Kwon, T., and Lee, H. H. (2006) Transfer patterning of pentacene for organic thin-film transistors. Adv. Mater. 18, 1861–1864.

P1: OTA/XYZ P2: ABC c16 JWBK252/Lee September 18, 2008

14:9

Printer Name: Yet to Come

REFERENCES

443

60. Park, J., Shim, S. O., and Lee, H. H. (2005) Polymer thin-film transistors fabricated by dry transfer of polymer semiconductor. Appl. Phys. Lett. 86, 073505. 61. Park, J., Park, S. Y., Shim, S. O., Kang, H., and Lee, H. H. (2004) A polymer gate dielectric for high-mobility polymer thin-film transistors and solvent effects. Appl. Phys. Lett. 85, 3283–3285. 62. Hines, D. R., Mezhenny, S., Breban, M., Williams, E. D., Ballarotto, V. W., Esen, G., Southard, A., and Fuhrer, M. S. (2005) Nanotransfer printing of organic and carbon nanotube thin-film transistors on plastic substrates. Appl. Phys. Lett. 86, 163101. 63. Kwon, T., Baek, C., and Lee, H. H. (2007) Top gate pentacene thin film transistor with spin-coated dielectric. Org. Electron, 8, 615–620. 64. Menard, E., Meitl, M. A., Sun, Y., Park, J. U., Shir, D. J. L., Nam, Y. S., Jeon, S., and Rogers, J. A. (2007) Micro- and nanopatterning techniques for organic electronic and optoelectronic systems. Chem. Rev. 107, 1117–1160.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17 INORGANIC ELECTRONIC DEVICES

17.1 INTRODUCTION Large-area electronics, commonly referred to as macroelectronics, is of wide and growing interest, particularly when implemented in mechanically flexible or stretchable formats, for applications ranging from electronic paper to conformal, structural health monitors and sensitive skins for aircraft and robotic systems [1–3]. This class of circuitry is important because of its ability to offer characteristics that are not readily available with wafer-based electronics or thin film electronics on conventional substrates such as glass, including mechanical flexibility, lightweight, rugged construction, and low cost per unit area. Recent research seeks to develop unconventional materials and printing techniques for manufacturing for these systems [4–13]. Organic semiconductors represent the most widely explored materials strategy [14]. Although these materials are attractive as pixel-switching transistors in certain classes of displays, their modest performance restricts their range of applications. This drawback has created interest in the possibility for using inorganic semiconductors in these types of systems. Various approaches, as summarized in a recent review article [3], exist, ranging from the relatively conventional (e.g., low temperature amorphous Si deposited by 75◦ C) to the exploratory (e.g., films of solutiondeposited semiconductor nanocrystals). The use of high quality, single-crystalline inorganic semiconductors in the form of micro/nanoscale ribbons, wires, and sheets [15–18] represents an alternative strategy. The electrical performance of devices built by printing these materials on plastic substrates can be comparable to those fabricated on rigid wafers. Excellent bending properties can be achieved with thin substrates, such that the strains can be maintained below the fracture point of the active materials

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright 

445

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

446

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(∼1%) even for small bending radii. There are two general techniques for generating the macro/nanoscale elements: “bottom-up” approaches [9, 19, 20] that rely on vapor or liquid phase chemical synthetic techniques and “top-down” methods [13, 21–23] that use lithography and etching of conventional semiconductor wafers. The latter method has the advantage that, when combined with a soft lithographic transfer printing process described in Chapter 4 [24, 25], the lithographically defined order in the elements on the source wafer can be maintained on the target plastic substrate. In addition, high temperature processing (e.g., contact doping) can be performed on the wafer before transfer. Soft lithography is important not only for this process, but it can provide a low cost, large-area patterning capabilities for other components of macroelectronic systems, independent of the nature of the semiconductor. The features of soft lithography (i.e., high speed, parallel operation, high resolution, etc.) match better, in our view, to demands of these classes of electronic systems than they do to conventional devices on semiconductor wafers. This chapter focuses, therefore, on aspects of soft lithographic patterning of inorganic devices for large area, flexible electronics. Due to their considerable promise, we emphasize systems that use single-crystalline inorganic semiconductor micro/nanomaterials. The content begins, in Section 17.2, with descriptions of methods for fabricating such materials using “bottom-up” as well as “top-down” approaches. Section 17.3 describes strategies for device and circuit integration. Section 17.4 outlines the electronic and mechanical characteristics that can be achieved. A final section provides a summary and some perspectives on future work.

17.2 INORGANIC SEMICONDUCTOR MATERIALS FOR FLEXIBLE ELECTRONICS Inorganic semiconductor micro/nanoscale elements in the form of wires, ribbons, sheets, and bars have captured interest recently for their use as active materials in high performance large-area electronics. These semiconducting materials can provide high quality, monocrystalline transport pathways in devices; certain thin structural embodiments enable mechanical bendability and even stretchability. When on flexible substrates, such as metal foils and polymeric plastics, these elements experience strain when bent. The tensile strain at the point of fracture for most inorganic semiconductors is in the range of 0.5–1.0%. Figure 17.1 illustrates the bending mechanics of wires/ribbons with thickness t when bent to a radius of curvature r. The top and bottom surfaces are under tensile and compressive strains, respectively. For thin elements, the maximum strain can be described as: ε=

t × 100% 2r

(17.1)

Equation 17.1 also approximately explains the relationship between the top surface strain and the allowable bending radius in a system that combines these elements

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.2 INORGANIC SEMICONDUCTOR MATERIALS FOR FLEXIBLE ELECTRONICS

447

t

r

ε=

t 2r

×100%

Figure 17.1. Schematic illustration of a bendable semiconductor object.

on a thin substrate. Here, t corresponds to the summed thickness of the substrate and the material layers on top [26]. For a given bend radius, strain at the device level can be reduced by using thinner substrates, thereby enabling improved bendability. Other strategies, such as laminated structures that place the device layer near the neutral mechanical plane, are also possible [26–28]. The following sections describe “bottom-up”, or synthetic, and “top-down”, or lithographic, approaches to creating these classes of devices. 17.2.1 “Bottom-Up” Approaches Single-crystalline semiconductor elements (one or two dimensional) can be fabricated via various chemical synthetic processes that have been reviewed recently [3, 20]. Generally, the simplest and most successful strategy, known as the vapor liquid solid (VLS) method, is based on a process originally introduced by Wagner and Ellis [29]. VLS relies on controlling the reaction conditions between a reactant (i.e., semiconductor precursor gas) and catalytic metal nanoparticles such that nanowire growth is favored. Figure 17.2a depicts the growth mechanism for generating semiconducting materials (i.e., silicon nanowires) via the VLS process. The growth begins by placing mono-disperse metal nanoparticles onto a substrate, followed by the introduction of gaseous semiconductor precursors. Increasing the temperature above the eutectic point of the system, as determined from the binary phase diagram (Figure 17.2b), causes atoms of the gaseous precursor to condense onto the surfaces of the metal nanoparticles, a process that is favored because it results in the reduction of the total surface energy of the Au–Si system. When the concentration of the semiconducting atoms present on the nanoparticle surface achieves supersaturation, the metal nanoparticles melt, thereby forming liquid alloyed droplets composed of a mixture of the semiconductor and the metal. Nanowire growth results from the precipitation of the solid semiconductor phase from the alloyed droplet. Choices of the metal semiconductor precursors, metal catalysts, and reaction temperatures are determined by examining the metal/semiconductor binary phase diagram (Figure 17.2b) of the materials of interest [9, 20]. In most cases, the size of the nanoparticle determines the diameter of the as-produced nanowire to within ∼2 nm [30].

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

448

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(a)

Si Si

Au

Cluster formation Nucleation and growth Nanowire

Temperature (°C)

(b)

1414 AuSi (l) 1064

AuSi (l) + Si (s) 363 Au (s) + Si (s) Au

Atomic percentage

Si

Figure 17.2. (a) Growth of inorganic semicoductor nanowires using VLS. (b) Binary phase diagram for Au and Si illustrating the thermodynamics of VLS growth.

Well-aligned nanowire arrays can be fabricated in orientations normal to the substrate surface by epitaxial growth with patterned catalysts particles [31–33]. For example, silicon nanowires with diameters >20 nm tend to grow predominantly along the Si 1 1 1 direction [30]. Similarly, zinc oxide nanowires form preferentially along the 0 0 1 orientation or vertically on the a-plane (1 1 0) of a sapphire substrate. Control over the densities and orientations of nanowire arrays can be achieved by combing selective growth mechanisms with substrates that are prepatterned with metal nanoparticles in layouts useful for electrical device integration. For example, films of well-aligned nanowire arrays synthesized by controlled growth strategies can be integrated into three-dimensional layouts for high performance electronics [34]. The VLS process is a versatile approach that allows for the formation of a wide range of complex heterostructured semiconductor nanowires in a controlled fashion. These structures are realized by controlling the growth conditions such that the incoming gases deposit preferentially on the catalyst surface (i.e., axial growth) or the nanowire surface (i.e., radial growth) [35]. For instance, Ge/Si coreshell nanowires can be synthesized by first growing Ge nanowires via nanocluster-directed axial growth followed by decomposition of silicon atoms introduced via chemical vapor deposition (CVD) onto the Ge nanowire surface [36]. Continuous exposure of the synthesized coreshell structures to vapor phase reactants can lead to the formation of core multishelled nanowires using these same synthetic strategies. Ge/Si nanowires

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.2 INORGANIC SEMICONDUCTOR MATERIALS FOR FLEXIBLE ELECTRONICS

449

and related heterostructured nanowires have been used as active thin films for achieving high performance electrical and optical devices [37]. In a related strategy, branched nanowire structures can be synthesized via bottomup approaches. These dendritic structures can be synthesized by first growing nanowires via a VLS process followed by sequential “seeding” of the catalyst nanoparticles onto the surfaces of the as synthesized nanowires. The process can be repeated such that growth of a second set of nanowires occurs on the surfaces of the newly deposited catalyst particles [35, 38, 39]. A related process, known as solution-liquid-solid (SLS), can also create functional micro/nanoscale elements. In this case, the synthesis is carried out in a liquid phase with low melting point metal nanoparticles as catalysts combined with protic solvents such as methanol, benzoic acid, ethyl amine, and others [39, 40]. Interestingly, the SLS process can generate nanowires composed of III–V materials at lower reaction temperatures compared to VLS approaches [39, 40]. For example, GaAs nanowires with diameters ranging from 10 to 150 nm can be synthesized from the alkane elimination reaction of precursors composed of III–V materials. The use of tri-tert-butylindane and gallane and catalytic quantities of protic solvents can produce III–V nanowires with yields between 50 and 100% [39]. By implementing indium nanoparticles as catalysts, the SLS process can yield nanowires composed of GaAs with narrow diameter distributions (i.e., 14–16% variations), and dimensions down to ∼6 nm [40]. An advantage of the SLS process for silicon nanowire growth is that narrower size distributions are possible than with the conventional VLS process [39– 41]. For example, SLS has been used to synthesize Si nanowires with diameters in a narrow range between 4 and 5 nm and with lengths up to several micrometers via the use of alkanethiol-coated nanocrystals and diphenylsilane at 500◦ C [41, 42]. At this temperature, diphenylsilane decomposes to produce silicon atoms, which diffuse into the nanoparticles and are subsequently expelled to form high quality silicon nanowires [41, 42]. A challenge with both the SLS and VLS processes is that the generation of organized arrays of horizontally aligned wires, in the type of configuration that might be most useful for thin-film-type electronics, can be extremely difficult. The “top-down” approaches described in the following section, particularly when combined with soft lithographic transfer printing techniques, avoid this problem. 17.2.2 “Top-Down” Approaches Semiconducting nanowires fabricated via the bottom-up approach have certain advantages: unusual heterostructures are possible, bulk quantities can be generated, and nanowires with dimensions as small as a few nanometers can be formed. However, the maximum possible lengths of these nanowires are often shorter than ∼100 µm [9, 20, 41–44] and the characteristic sizes, especially for geometries that resemble ribbonlike structures, tend to have broad distributions [45–48]. The ability to tailor or control the nanowire surface properties, compositional purity, doping uniformity, and concentration are also much less well developed in comparison to commodity wafers used in the integrated circuit industry. Means to exploit this well-developed

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

450

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

wafer technology for applications in macroelectronics is, as a result, of considerable interest. Simple lithographic (soft lithography or conventional photolithography, as examples) approaches and chemical etching techniques can be used to fabricate ribbons wires, platelets, etc. from standard bulk or layered semiconductor wafers. We refer to these elements as microstructured semiconductors, or µs-Sc, and have explored examples of silicon (µs-Si), gallium arsenide (µs-GaAs), and several others. The conceptually simplest approaches use layered wafers such as silicon on insulator (SOI) or superlattices of GaAs/AlAs/SiGaAs or AlGaN/GaN/Si formed by molecular beam epitaxy as source substrates. Patterning a resist layer on such wafers and then etching them to remove a sacrificial layer or underlying support (e.g., SiO2 for SOI; AlAs for GaAs/AlAs/SIGaAs; Si for AlGaN/GaN/Si) produces ribbons, wires, platelets, or bars [3, 17, 59–61] with geometries defined by the lithographic step. Although this patterning process commonly uses photolithography [17, 25], techniques such as soft lithography [17, 62] and nanoimprint lithography [63, 64] can also be employed. For example, Figure 17.3 depicts images of µs-Sc’s formed via top-down approaches. This simple fabrication approach has been used to generate µs-GaAs (Figure 17.3a), µs-InP (Figure 17.3b), and µs-GaN (Figure 17.3c) [13, 66]. High performance electrical devices that use µs-Sc’s as the active thin film have been achieved [17, 18, 21–23, 67–72], including device layouts that are both stretchable and bendable [3, 60, 73]. The lateral dimensions possible using top-down fabrication approaches range from very small (i.e., wires with thicknesses and lateral dimensions down to ∼10 nm range [15, 17, 74] and ∼17 nm [25, 64, 75], respectively) to very large (i.e., membranes with dimensions of several centimeters). Very small wires have been used as thin-film chemical sensors on plastic substrates [64]. Advantages for using wafers with embedded liftoff layers include the simplicity of the technique and the versatility in designing the spatial features of the fabricated micro/nanoscale elements. A disadvantage is that the cost associated with fabricating large quantities of these elements can be high. As a result, fabrication approaches that use less expensive bulk wafers are of special interest. We recently developed etching techniques that can generate single crystal silicon ribbons, platelets, and bars from bulk silicon (1 1 1) wafers [22, 23, 70] as well as wire structures from wafers of certain III–V materials [18]. Figure 17.4a depicts the fabrication sequence for the case of silicon. The process begins with the patterning of lines of resist perpendicular to the Si (1 1 0) planes using conventional lithography [23] or soft lithographic methods [22, 70]. Next, dry etching creates trenches into the exposed silicon (top frame of Figure 17.4a). The top surface of the wafer and the trenches are coated with passivation layers (i.e., etching mask for anisotropic etching step) of materials such as SiO2 or Si3 N4 . Depositing metal protecting layers via electron-beam evaporation at an oblique angle to the wafer surface and subsequently removing the unprotected passivation material exposes the silicon planes (i.e., Si (1 1 0)), as illustrated in the middle frame in Figure 17.4a. For anisotropic wet etchants for silicon such as KOH or tetramethyl ammonium hydroxide, these groups of planes etch must faster (i.e., >100×) than the other silicon planes. Compared to KOH, etchants such as tetramethyl ammonium hydroxide are complementary

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.2 INORGANIC SEMICONDUCTOR MATERIALS FOR FLEXIBLE ELECTRONICS

451

Figure 17.3. SEM images of microstructured semiconductor objects. (a) GaAs, (b) InP, (c ) GaN. (Reprinted with permission from [21] and [65]. Copyright 2005 Wiley-VCH Verlag.)

metal-oxide semiconductor (CMOS) compatible and avoid the introduction by KOH of mobile ions that are detrimental to integrated circuits [70, 76]. As the etching progresses, the Si (110) etch fronts proceed in a horizontal fashion until they meet, thereby creating freestanding single crystal silicon ribbons with rectangular crosssections (third frame in Figure 17.4a). Figure 17.4b shows a random array of 500nm-thick, 6-cm–long, and 7-µm-wide silicon ribbons produced in this manner. By carefully controlling the processing parameters, ribbons, platelets, and bars with

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

452

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

Dry etch trenches

(a)

(111) Si

(110) Passivate

Anisotropic wet etch

(b)

20 µm Figure 17.4. Schematic illustration of processes for fabricating silicon wire/ribbons from bulk wafers.

dimensions between 100 nm and the size of the wafers can be formed in flexible and stretchable configurations [70]. Interestingly, by modulating the dry etching step (top frame of Figure 17.4a) using an inductively coupled plasma reactive ion etching system, well-controlled ripples can be introduced on the sidewalls of the silicon trenches. These ripples can be used to generate, in a single processing sequence, multilayered stacks of silicon elements. These elements can be released from the wafer or they can be removed in a layer-by-layer fashion for integration onto a device substrate [23]. Micro/nanostructures fabricated by these types of approaches can exhibit good surface uniformity, morphologies, and materials quality, with geometries that can be easily controlled. The main drawbacks include compositions that are limited to materials that are readily available and finite roughness on certain surfaces, generated by the etching processes.

17.3 SOFT LITHOGRAPHY TECHNIQUES FOR GENERATING INORGANIC ELECTRONIC SYSTEMS Soft lithographic techniques, including various types of printing, molding, and optical approaches described in previous chapters, are valuable for the classes of inorganic electronic systems described here both for their use in generating semiconductor elements and in patterning electrodes, dielectrics, and other components of the

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.3 SOFT LITHOGRAPHY TECHNIQUES FOR GENERATING

453

devices. Their scalability to large areas and high throughputs, their high resolution, and their potential for low cost operation represent key features [77–79]. The following sections describe representative applications of soft lithographic techniques including micromolding in capillaries, imprint lithography, and dry transfer printing. 17.3.1 Micromolding in Capillaries Micromolding in capillaries (MIMIC) uses an elastomeric object to create micro channels on a substrate as described in detail in Chapter 3. Elastomers such as poly(dimethylsiloxane) (PDMS) are an excellent choice for the mold material because the low surface energy and modulus of PDMS (ca. 21.6 dyn cm−1 ) can lead to a soft, nondestructive, reversible and a liquid tight seal against a flat surface. Also, PDMS is transparent to ultraviolet (UV) light suitable for typical photocurable polymers [77]. Figure 17.5a depicts a typical MIMIC process. The process begins by placing a mold with appropriate relief structure onto a target substrate. Next, liquid solutions (e.g., polymers, conductive nanoparticle inks, etc.) are placed at one end of the mold. Capillary action or the application of vacuum fills the channels created between the PDMS relief and the substrate. The solvent then evaporates or the material is otherwise cured into a solid form. Removing the mold completes the fabrication process (third frame Figure 17.5a). This type of MIMIC process can be used to generate features of an electronic circuit (see Figure 17.10 and Section 17.4.1) or patterns of etch resist for forming µs-Sc.

(a)

PDMS mold

(b) Photoresist plastic

Place a drop of prepolymer Fill channels

Cure, remove mold

Pattern transfer

RIE, metal evaporation Lift off

Figure 17.5. Two lithographic techniques valuable for patterning microstructured semiconductor elements and other material elements needed to construct devices with them. (a) MIMIC. (b) Imprint lithography.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

454

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

In a related approach, the microfluidic channels associated with the MIMIC technique can be filled with solution suspensions of nanowires. Shear forces associated with flow during deposition of the wires onto the underlying surface can create aligned arrays [80]. Multilayers of aligned or crossed arrays of wires can be achieved through the repeated, sequential application of this process. 17.3.2 Imprint Lithography Imprint lithography, particularly when performed with a soft mold, is another useful method for creating functional electronic systems. The process involves embossing relief features into films on a target substrate, as illustrated in Figure 17.5b and described in detail in Chapters 3 and 6. Typically, these films consist of soft or liquid layers that can be manipulated by a comparatively hard mold. Next, an etching step removes the thin regions to yield isolated features that can be used as resists to guide the deposition or removal of other materials. The resulting structures can serve, for example, as metal contacts (i.e., source drain or gate electrodes) integrated with semiconducting materials for devices, or they can be used in the etching procedures to create these materials according to procedures described previously [79]. Nanoscale features that function as source and drain contacts with dimensions as small as 200 nm can be achieved with this method over millimeter lengths scales on plastic substrates [81]. 17.3.3 Dry Transfer Printing In its earliest form, soft lithography involved the use of soft stamps to print molecular inks that covalently bond to target substrates (e.g., alkanethiols on gold). This method, known as microcontact printing, can be used to form various components of electronic systems. In fact, the use of microcontact printing for the fabrication of organic electrical and optical electronic devices appears in Chapter 16 and is also reviewed elsewhere [79]. The same method can also be inserted into the fabrication processes for semiconductor nanomaterials described previously. Relatively new printing approaches use stamps, similar to those originally developed for microcontact printing, that enable printing of thin, solid layers as described in Chapter 4. In addition to the stamp, these methods also require some mechanism for the transfer of the ink from the stamp to the substrate. Some combination of surface chemistries, conformable adhesive layers, or other means can provide preferential adhesion to ensure high yield transfer. Vacuum, solution, or solid transfer methods can be used to prepare the printable material on the stamp (“inking the stamp”). The first type of transfer printing technique, known as nanotransfer printing, used thin metal layers deposited onto the stamp to provide resolution that exceeds that possible with microcontact printing, due to the avoidance of vapor phase transport and other ink spreading mechanisms that are often present with molecular inks. (See references in Chapter 4.) Figure 17.6 depicts a schematic illustration of the process flow for the transfer printing approach of interest here. As with microcontact printing, this type of technique can be used in various ways to produce electronic

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.3 SOFT LITHOGRAPHY TECHNIQUES FOR GENERATING

455

Figure 17.6. Schematic illustration of the generic process flow for transfer printing solid objects.

devices. Here, we focus on the use of transfer printing for the manipulation of µs-Sc. In this case, a donor substrate provides the inking material, which consists of µs-Sc’s undercut from the mother wafer but tethered to it at certain lithographically defined “anchor” points [12, 13, 17, 21–23, 61, 66, 70, 65]. The process begins by placing a stamp (e.g., PDMS) with suitable relief features into conformal contact with the top surfaces of the µs-Sc [24]. Peeling back the stamp can lead to fracture at the anchor

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

456

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

points and liftoff of the µs-Sc from the mother wafer to the stamp surface. Careful design of these anchors can lead to higher retrieval rates via stress focusing centers, which can help facilitate the fracture process from the anchor [82]. The adhesion to the stamp can be controlled in a variety of ways, ranging from kinetic control that exploits the viscoelastic nature of the stamp to surface chemistries that partially bind the elements to the stamp as described in detail in Chapter 4. Figure 17.7 illustrates, as an example, the rate dependence of the adhesion energy between PDMS and glass, where v and G represent the peel speed and energy release rate, respectively. At low speeds, the energy required to separate PDMS from a solid body is relatively low, due to the low surface energy (∼20 mJ m−2 ) of PDMS [83]. At higher speeds, however, the energy required to separate the two is much greater. As a result, fast removal of a stamp from a mother substrate can lead to an increase in the retrieving yield of the µs-Sc’s. Figure 17.7a depicts this process where low peel speeds lead to poor or virtually no retrieval of silicon bars from the mother wafer to the stamp, and in comparison fast peel rates accomplish complete retreival. The stamp, “inked” in this manner with µs-Sc, can be used to print these elements onto a target substrate (i.e., glass, plastic, etc.). In the simplest approach, the

Energy release rate, G (J m–2)

(a)

5 Retrieving regime 4 Printing regime 3

v

2

mgh 1

0 0

θ

1

3 2 Separation speed, v (cm s–1)

4

5

(b)

Figure 17.7. The dependence of stamp adhesion on peel rate, v . (a) Calculated energy release rate, G, from the loss of gravitational potential energy. (b) Low peel speeds lead to poor retrieval (left) but faster peel rates cleanly break the structures from their anchors (right). (Reprinted with permission from [25]. Copyright 2006 Nature Publishing.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

457

17.3 SOFT LITHOGRAPHY TECHNIQUES FOR GENERATING

kinetic control can be used to print µs-Sc’s onto smooth surfaces without the use of an adhesive layer [25]. In many cases, however, a separate adhesive layer is used to ensure high yield transfer. Adhesives based on polyimide [67–69], polyurethane [13, 24, 65], benzocyclobutene-containing siloxane polymers [84], PDMS [70], and epoxies [17, 22, 72] can work well in this process. Alternatively, surface chemistries, cold welding, kinetic control, and other mechanisms have also been demonstrated. Figure 17.8 illustrates on example of a surface chemistry that has been used to bind µs-Sc’s to PDMS. PDMS is composed of three-dimensional cross-linked structures with repeating units of –(CH3 )2 SiO2 –. The untreated PDMS surface is hydrophobic due to the high density of methyl moieties (–CH3 ) [85]. Oxidizing the methyl groups on the PDMS surface by exposure to highly active oxygen species, such as O2 + , O2 − , O generated from oxygen plasma and ozone generated by UV irradiation, leads to a PDMS surface that will react with a variety of different materials that contain exposed silanol groups, such as oxides. For instance, placing the oxidized stamp in physical contact with a SiO2 layer can lead to a very strong bond via condensation reactions as illustrated in Figure 17.8. This adhesion is strong enough such that it enables structural form factors for applications in stretchable electronics [3]. Although most examples of transfer printing use soft, elastomeric stamps, in certain cases rigid stamps are possible. Such rigid stamps often enable resolution and feature definition at precisions that are not possible with elastomers. In one example, a rigid GaAs stamp in the geometry of a photonic crystal was used to print thin layers of Au in the geometry of photonic crystals, with edge resolution better than 20 nm. (See references in Chapter 4.) In a different approach, known as superlattice nanowire pattern transfer (SNAP) [16], the stamp consists of a superlattice of GaAs/AlGaAs layers etched on one edge to forming a comblike structure. A thin layer of metal is evaporated on the top surface of the superlattice layers, which is PDMS CH3

Weak generalized adhesion

O

CH3

Si O Si CH3

Oxidation

O,Si(OH)H

m

CH3

OH

O

O

Si

OH OH Si OH

OH H

H

O

O O

Si

O

O

Si

PDMS SiO2/Si

O O

O

Si OH

O

Si

O

Si

O

Condensation Si reaction

O

Si Si Si

O

Strong adhesion

O n

O n

SiO2 layer

Figure 17.8. Schematic illustration of the surface chemistry of PDMS and semiconductor ribbons covered with thin SiO2 layers.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

458

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

placed into contact with an adhesive-coated substrate and the metal layer is lifted off from the superlattice via wet etching of the template layer (i.e., GaAs layer). The transferred metal film on the SOI substrate serves as a mask for subsequent etching steps to define the nanowire spatial dimensions. Recently, wires produced via the SNAP approach have been used as thin-film chemical sensors on plastic substrates [64]. Figure 17.9 illustrates some representative examples where semiconducting objects were transfer printed on various rigid and flexible substrates with and without (a)

(b)

(c)

(d)

(e)

(f)

Figure 17.9. Semiconductor objects transfer printed on various rigid and flexible substrates with and without the use of adhesives. (a) GaN ribbons printed onto a silicon wafer. (b) Printed silicon nanobeams bridging printed silicon micro beams. (c ) Silicon beams on a glass lens. (d ) Printed array of silicon nanoribbon p–n diodes formed on a double-convex polycarbonate magnifying glass. (Reprinted with permission from [25]. Copyright 2006 Nature Publishing.) (e) Large area (15 cm × 15 cm) transfer of silicon ribbons onto a PET sheet coated with PU. (f ) Crossbar array of single walled carbon nanotubes printed onto a plastic substrate. (Reprinted with permission from [24] Copyright 2005 Wiley-VCH Verlag.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

459

adhesive coatings. Figure 17.9a shows GaN ribbons printed onto a silicon wafer, silicon nano beams printed onto silicon micro beams (Figure 17.9b), silicon ribbons printed onto a glass lens (Figure 17.9c), and silicon ribbons printed onto a polycarbonate magnifying glass (Figure 17.9d). Structures shown in frames Fig. 17.9a–17.9d were printed without adhesive layers. When shape complimentarily cannot be achieved or if surface roughness is large enough to affect the conformal contact, then an adhesive layer is often required. The printing yields can be remarkably improved in this way because the adhesive layers ensure intimate contact. For example, Figure 17.9e depicts silicon ribbons printed onto a polyethylene terephthalate (PET) sheet coated with polyurethane (PU). Figure 17.9f displays images of single-walled carbon nanotubes (SWNT) transferred to a PET substrate coated with thin epoxy layer, with repetitive transfer for crossed arrays. Dry transfer printing provides a well controlled, highly deterministic approach to the assembly of functional electrical systems from micro/nanomaterials such as, but not restricted to, µs-Sc. Fully automated, high yield printing systems exist, and various classes of high performance devices and circuits have been demonstrated, as described in subsequent sections. 17.4 FABRICATION OF ELECTRONIC DEVICES 17.4.1 Transistors on Rigid Substrates via MIMIC Processing MIMIC as described in Section 17.3.1 and more extensively in Chapter 3 has been used to form a variety of functional devices such as Schottky diodes [86, 87], half-wave rectifier circuits [88], GaAs high electron mobility transistors (HEMTs) [89, 90], and Si p channel metal oxide field effect transistors (pMOSFETs) [91]. As an example, Figure 17.10 shows the cross-sectional view of a Si pMOSFET where the source, drain, and gate metal electrodes were defined by MIMIC. Figures 17.10b and c show images of an array of transistors with channel lengths (Lc ) of 20 µm and channel widths (W c ) of 200 µm, together with current–voltage characteristics of a representative MOSFET fabricated using this procedure. The device properties are similar to those of MOSFETs fabricated via conventional photolithographic techniques. 17.4.2 Flexible Inorganic Transistors Micro/nanoscale semiconductor materials generated via bottom-up and top-down approaches described in Section 17.2 can be integrated into conventional electronic devices by using the classes of soft lithography, nanoimprinting, and dry transfer printing methods highlighted in previous sections and chapters. Figures 17.11a and b show an optical image and current–voltage characteristics of Si nanowire FETs on plastic substrates fabricated by combining nanoimprint lithography and microfluidic assembly [81]. Here, nanoimprint defined the gate electrode lines. Besides patterning metal electrodes for device integration, transfer printing can deliver nanowires on target device substrates. Figure 17.11c represents a precisely ordered set of Si nanowires formed by SNAP and transferred to a PET sheet coated with a thin film

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

460

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(b)

(a) Gate

AI

Drain

Source

SiO2 n-Si 200 µm (c)

0.5

VGS = 7 V

−IDS (mV)

0.4 0.3

−6V

0.2

−5 V

0.1

−4 V

0 0

4

8 –VDS (V)

12

16

Figure 17.10. (a) Schematic illustration of a Si pMOSFET. (b) Optical micrograph of array of devices (Lc = 20 µm and W c = 200 µm) fabricated by MIMIC. (c ) Current–voltage characteristic of a typical device. (Reprinted with permission from [91]. Copyright 1998 Wiley-VCH Verlag.)

of an epoxy (i.e., SU-8) [64]. FETs fabricated from these transferred nanowires using conventional device processes yield high carrier mobilites and transconductances due to the single crystalline nature of these wires (Figure 17.11d). Also, these high performance nanowire transistors can be used as chemical sensors, in rigid or mechanically flexible formats. Such nanowire building blocks with diverse functional properties can be vertically stacked up to 10 or more layers, which could enable novel circuit concepts such as three-dimensional (3D) integrated electronics (Figures 17.11e and f ) [34]. Similar processing steps can be implemented with other semiconductor nanomaterials. Figures 17.12 a–c show optical microscope images and DC electrical characterizations of ribbon-and wire-based µs-Si MOSFETs [69], µs-GaN HEMTs [61], µs-GaAs MESFETs [18], respectively. In these cases, high temperature processes (>900◦ C) required for the contacts were completed on the mother wafer prior to the transfer to plastic. Figure 17.12a shows an optical image of a polyimide sheet (with a thickness of 25 µm) covered with an array of Si MOSFETs fabricated by printing Si ribbons (290 nm thick) with predefined contact-doped regions. These substrates can be bent to a small radius of curvature (∼3 mm) without delaminating or fracturing the ribbons [67–69]. The adhesive layer in this case was a liquid precursor to polyimide (i.e., polyamic acid) that is converted into polyimide by thermal treatment after printing the semiconductor layer. The gate dielectric material consisted of SiO2 (100 nm

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

461

17.4 FABRICATION OF ELECTRONIC DEVICES

(b) 103 Conductance (nS)

(a)

102

101

100 −6 (c)

−2

−4

2 0 VGS (V)

6

4

(d) 101 4

10−1

−IDS (µV)

−IDS (µV)

100

10−2

0 0

10−3 10−4 −5.0

−2.5

0.75 1.5 −VDS (V)

0

2.5

VGS (V)

(e)

(f)

5 1.5 VVGS steps

−IDS (mV)

4 3 2

1

0

1

)

(V

4

DS

3

2

−V

10

5 Layer

1

Figure 17.11. (a) SEM image of a 20-nm Si NW crossing an imprint-patterned metal gate electrode. (b) Current–voltage characteristic of a typical device. (Reprinted with permission from [81]. Copyright 2003 American Chemical Society.) (c ) SEM image of SNAP nanowires on plastic. (d ) Electrical characterization of nanowire transistors on plastic. (Reprinted with permission from [64]. Copyright 2005 Nature Publishing.) (e) Optical image of 10 layers of Ge/Si nanowire transistors. (f ) Current–voltage characteristics for nanowire transistors from layers 1, 5, and 10. (Reprinted with permission from [34]. Copyright 2007 American Chemical Society.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

462

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

thick) grown by PECVD using SiH4 and N2 O at 250◦ C. A Cr (5 nm)/Au (100 nm) layer deposited by electron-beam evaporation formed the source, drain, and gate electrodes. Electrical measurements on these devices show excellent performance. For example, effective mobilities in µs-Si nMOSFETs are ∼600 cm2 (V s)–1 in the linear regime and ∼530 cm2 (V s)–1 in the saturation regime, as computed with standard models (Figure 17.12a) [92]. These values approach those of similarly scaled single-crystalline silicon nMOSFETs on wafers and they outperform those observed 3

(a)

VGS = 3 V IDS (mA)

2 2V

1

1V 0 0

IDS (mA)

(b)

1

3

2 VDS(V)

6

VGS = 1 V

4

0V −1 V

2

−2 V 0 0

1

2

3 VDS(V)

4

5

(c) VGS = 0.5 V

2 IDS (mA)

0V 1

−1 V −2 V

0 0

1

2 3 VDS(V)

4

5

Figure 17.12. Images, cross-section, and electrical characteristics of flexible devices that use semiconducting wires/ribbons on plastic substrates. (a) µs-Si MOSFETs. (Reprinted with permission from [69]. Copyright 2006 IEEE.) (b) µs-GaN HEMTs. (Reprinted with permission from [61]. Copyright 2006 American Institute of Physics.) (c ) µs-GaAs MESFETs. (Reprinted with permission from [18]. Copyright 2006 American Institute of Physics.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

463

in the most sophisticated forms of laser annealed polycrystalline silicon transistors on glass [93]. For the µs-GaN HEMTs, the ohmic contacts consisted of Ti (15 nm)/Al (60 nm)/Mo (35 nm)/Au (50 nm) deposited on a layer of GaN on a Si wafer, thermally annealed at 850◦ C for 30 s in a N2 environment. For the µs-GaAs MESFETs, ohmic contacts used AuGe (120 nm)/Ni (20 nm)/Au (120 nm) annealed at 450◦ C for 60 s. The transconductance of a typical µs-GaN HEMT is 1.6 mS for a channel length, channel width, and gate length, of 20, 170, and 5 µm, respectively (Figure 17.12b). This transconductance is lower (by about 40%) than those observed in similar waferbased devices, primarily because of degradation of the ohmic contacts caused by exposure to etchants used during fabrication of the µs-GaN. Process improvements can likely eliminate these effects. µs-GaAs MESFETs on plastic with gate lengths of 2 µm and channel lengths of 50 µm exhibit pinch-off voltages at V DS = 0.1 V (i.e., linear region), ON/OFF current ratios, and maximum transconductances of −2.7 V, ∼106 , and ∼800 µS, respectively (Figure 17.12c). Figure 17.13 shows radio frequency performance of these transistors [18, 69]. For µS-Si MOSFETs and µS-GaAs MESFETs, the unity current gain frequency, f T , in common-source configuration was measured by extracting S-parameters using a vector network analyzer with a standard short-open-load-thru calibration. The f T of typical µs-Si nMOSFETs with Lc of 2 µm, channel overlap distance (Lo : defined by the distance that the gate electrode extends over the doped source/drain regions) of 1.5 µm, and W c of 200 µm reaches ∼500 MHz, at gate and drain biases of 2 V. The frequency is, as expected, highly dependent on Lc and Lo . Figure 17.13b indicates reasonable agreement between measured (filled) and calculated (open) f T values for devices with different Lc and Lo (the measurements do not involve de-embedding). GaAs MESFETs with Lc of 50 µm, W c of 150 µm, and gate lengths of 2 µm fabricated with GaAs wires (width of ∼2 µm) integrated with ohmic stripes exhibit f T in gigahertz regime, i.e., 1.55 GHz (Figure 17.13c). This level of high frequency operation indicates a potentially promising pathway to the fabrication and assembly of large area, flexible active antennas operating in the UHF regime [1]. 17.4.3 Flexible Integrated Circuits Macroelectronic systems require the integration of transistors with other passive components including capacitors, diodes, inductors, and resistors. Microstructured ribbon and wires composed of µs-Si MOSFETs and µs-GaAs MESFETs, respectively, can be integrated into analog and digital circuits [68, 71]. Figure 17.14a shows, as an example, optical images, schematic illustrations, and electrical measurements of differential amplifiers built with µs-Si nMOSFETs on flexible PI substrates [68]. This differential amplifier consists of four different components including a current source (three TFTs (thin film transistors) with Lc = 30 µm and W c = 80 µm), current mirror (two TFTs with Lc = 40 µm, W = 120 µm and Lc = 20 µm, W c = 120 µm), differential pair (two TFTs with Lc = 30 µm and W c = 180 µm), and load (two TFTs with Lc = 40 µm and W c = 80 µm). The Lo values of these transistors are 15 µm. This amplifier exhibits an output voltage gain of ∼1.3 for an input signal

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

464

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(a)

0.6

fT(GHz)

0.4

0.2

0.0

0

1

3

2

4

IDS (mA) (b)

0.8

L0 = 1.5 µm

fT(GHz)

0.6

L0 = 2.5 µm 0.4 0.2

L0 = 5.5 µm

0.0 0

2

4

6

8

10

12

14

L0 (µm) (c)

H21(dB)

40

SDS

20

fT = 1.55 GHz (experiment) fT = 1.68 GHz (simulation)

SGS

Gate length = 2 µm 0

0.1

0.5 ( Freq GHZ)

1

3

Figure 17.13. High frequency responses of silicon ribbon and GaAs wire transistors. (a) Unity current gain frequency (f T ) of µs-Si MOSFETs as a function of I DS . (Reprinted with permission from [68]. Copyright 2007 American Institute of Physics.) (b) Dependence of f T on channel length of µs-Si MOSFETs. The different symbols represent measurements on different devices; the dashed line corresponds to calculation. (Reprinted with permission from [69]. Copyright 2006 IEEE.) (c ) Experimental (black) and simulated (gray) results of rf responses of GaAs-wire MESFETs. (Reprinted with permission from [18]. Copyright 2006 American Institute of Physics.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

Signal Amp (V)

(a)

465

0.5

0.0

−0.5 0.0 6 4

V (V)

(b)

0.2 0.4 Time (mS) (0,0)1 (0,1)1

(1,0)1 (1,1)0

2 0 −2

VA VB

−4 −6

VO Time

Figure 17.14. Images and electrical measurements of integrated circuits that use semiconductor wires and ribbons on plastic. (a) Silicon ribbon differential amplifier. (Reprinted with permission from [68]. Copyright 2007 American Institute of Physics.) (b) GaAs wire logic gates. (Reprinted with permission from [71]. Copyright 2006 Wiley-VCH Verlag.)

with amplitude 0.5 V peak-to-peak (VPP), consistent with the P-SPICE simulations used to design the circuit (Figure 17.14a, right). These bendable circuits are useful building blocks for structural health monitor systems in which the circuit is wrapped around a curved rigid structure where it remains for monitoring purposes such as the real time, in-flight monitoring of the deformation of aircraft panels. As another example, µs-GaAs MESFETs can be integrated into different digital logic gates such as NOR gates, NAND gates, and inverters [71]. Due to the nature of depletion-mode operation in GaAs MESFET, these logic gates are built using both µs-GaAs MESFETs and µs-GaAs Schottky diodes. Figure 17.14b shows a collection of transistors and simple circuits on a PET substrate, the electrical characteristics and circuit diagram of the NAND gate. This NAND gate is constructed with three transistors; one serves as the load and the others serve as the switching transistors, which have channel lengths of 100 and 50 µm, respectively. All of these devices have channel widths of 150 µm and gate lengths of 5 µm. In the configuration of a NAND gate, when both switching transistors (V A and V B ) are turned on by applying a high voltage, a large current flows through the drain (V DD ) of the load transistor to ground (GND), resulting in a low output voltage (V O ) (logic 0). On the other hand, a high output results if one or both the inputs to the gate are low (logic 1). By electrically interconnecting nMOSFETs with pMOSFETs from µs-Si ribbons, low power and high performance CMOS logic structures on flexible plastic substrates can be achieved [94]. Figure 17.15a shows a CMOS type three-stage ring oscillator on a 25-µm-thick sheet of PI. The Lc and Lo are 12 and 10 µm, respectively; the channel widths of the pMOSFET and nMOSFET are 300 and 100 µm, respectively. The circuits exhibit oscillation frequencies of 2.6 MHz, corresponding to a

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

466

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(a)

0.5 mm

Freq (MHZ)

(b)

Bending radius (mm) 3.25 4

∞

6.5

6.5

3.25

0.0 0.2 Strain (%)

0.4

3

2

1 −0.4

−0.2

Figure 17.15. (a) Images of an array of three-stage ring oscillators on a PI substrate. (b) Frequency responses collected at different bending configurations. The insets show the bending system and probing apparatus.

stage delay of 64 ns. Additional improvements in device design, such as reduction of contact overlap and channel length, should lead to significantly higher oscillation frequencies [95]. Good mechanical bendability is a critical feature of many of the envisioned applications described in the introduction. This aspect can be systematically evaluated by bending the plastic substrates with specially designed mechanical stages to generate concave (compressive strains on top surface of the device) and convex surfaces (tensile strain; inset of Figure 17.15b). Figure 17.15b presents the results of bending tests on ring oscillators at different strain values between 0.051% and 0.29%, corresponding to bending radii of 25.5 and 4.5 mm, respectively, with a 25-µm-thick substrate. These results show slight but nonsystematic variations of the electrical properties.

17.4.4 Heterogeneous Electronics The dry transfer printing approach provides a promising route to heterogeneously integrated, 3D electronic systems for interesting applications that cannot be addressed

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

SWNTs

Source wafer

GaN

Print GaAs

467

Stamp

Si Process devices; repeat printing

Device substrate

Nanotubes, wires, and ribbons 3D-HGI

Figure 17.16. Schematic illustration of a printed semiconductor nanomaterials based approach to 3D heterogeneous electronics. (Reprinted with permission from [67]. Copyright 2006 American Association for the Advancement of Science.)

with conventional technologies. Examples include flexible displays, large area solar cells, conformable x-ray imagers, distributed structural and personal health monitors, curved surface imagers as electronic eyes, and others. As an example of this capability, broad classes of materials ranging from SWNT, silicon, GaAs, and GaN were combined into functional circuits and device arrays by transfer printing [67]. Figure 17.16 presents representative processing steps. The process involves the repeated application of the printing and device fabrication steps described previously, but where different semiconductor nanomaterials for different devices come from separate substrates. The soft, elastomeric stamps are critical for these applications because they enable nondestructive contacts with underlying device layers and applicability to surfaces that have some topography. After the first layer of devices is printed, the substrate is coated with a thin layer of polymer, which planarizes the first layer, and forms an insulating adhesive layer for the next layers of devices. Because the polymer is thin (<1 µm), small via holes can easily be etched to allow connections between selected devices in different layers. This process is then repeated to create 3D functional devices. Figure 17.17 shows an optical microscope image, a schematic cross-sectional view, and electrical characteristics of heterogeneous triple layers on PI with a bottom layer of µs-GaN HEMTs, followed by layers of SWNT TFTs and µs-Si MOSFETs. The µs-GaN HEMTs and µs-Si MOSFETs use the same design as those shown in Figure 17.12, respectively. The SWNT TFTs use electrodes of Cr/Au (2/20 nm) and the gate dielectric of SiO2 /Epoxy (300/150 nm) with channel lengths and widths of 50 and 200 µm, respectively. These devices have stable characteristics comparable to those in simple single layer configurations: the µs-GaN HEMTs have threshold voltages (V th ) of −2.4 ± 0.2 V, on/off ratios >106 , and transconductances of 0.6 ± 0.5 mS; the SWNT TFTs have V th = −5.3 ± 1.5 V, on/off ratios >105 and linear mobilities of 5.9 ± 2.0 cm2 (V s)−1 ; and the µs-Si MOSFETs have V th = 0.2 ± 0.3 V, on/off ratios >104 and linear mobilities of 500 ± 30 cm2 (V s)−1 .

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

468

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(c)

(a)

100 3rd Si 50

−6 −3 0 3

3rd Si 2nd SWNT

1st GaN

IDS (µA)

−0.2

S

G

600

3V

100

300

2V

2 10−

0

(b)

2

10

2nd SWNT

−0.1

10

−1

10−3 10

−5

−10 −5 Epoxy SiO2 Cured PI PU PI

1

0V 3

2

7V

−1.2

−0.6 5 V 0.0 0 V

0.0

D

0 0

6

0

−6

5

1000 1st GaN

−2

10

0

10

0 1

−1 V

500

−2

0 0

0 0V

102 1000

500

−7 −5 −3 −1 VGS (V)

−4

−3 V 2 4 VDS (V)

6

Figure 17.17. Three-dimensional heterogeneous integrated devices on plastic including µsGaN HEMTs, µs-Si MOSFET, and SWNT TFTs, in a three layer stack. (a) Large-area image of the 3D circuit. (b) Schematic cross-sectional view. (c ) Full transfer and current–voltage characteristics of three layers. (Reprinted with permission from [67]. Copyright 2006 American Association for the Advancement of Science.)

Electrical interconnects between these multistacked devices on different levels can create interesting examples of circuits. Figure 17.18 presents some examples. The first, shown in Figure 17.18a, consists of GaAs metal-semiconductor-metal (MSM) infrared (IR) detectors integrated with µs-Si nMOSFETs on PI substrates. Back-to-back Schottky diodes with a channel of 10 µm are formed by Ti/Au (5/70 nm) electrodes deposited on the ends of 270-nm-thick GaAs nanoribbons. The resulting GaAs detector displays current enhancement as a function of IR illumination power. A measured response of about 0.30 A W−1 at a wavelength of 850 nm is observed in the range from 1 to 5 V. This bendable IR detector system could allow advanced systems such as curved focal plane arrays for wide-angle IR night vision imagers. Another example is a hybrid inverter (Figure 17.18b) with a complementary design (CMOS) by use of integrated n-channel µs-Si MOSFETs with Lc of 75 µm and W c of 50 µm and p-channel SWNT TFTs with Lc of 30 µm and W c of 200 µm. This combination of dimensions was chosen to equalize the output currents from the nMOS and pMOS elements. With a supply voltage of 7 V, these CMOS inverters exhibit good transfer characteristics with gains of ∼7, which are qualitatively consistent with numerical circuit simulations. Besides these examples, the same methods can enable integration of optical, sensing, and micromechanical devices with these electronics to yield complete, multifunctional systems.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

(a) 12

IDS (mA)

9

469

Increasing illumination Si GaAs

6 VG

3

400 µm GND

0 −6

−4

−2

0

2

VDD 4

6

VGS (V)

(b) 6

SWNT

Si

6

VDD

VIN

4

2

VOUT

200 µm 0

Gain

Vout (mA)

4

8

2 0

−2

0

2 4 Vin (V)

6

8

Figure 17.18. (a) The transfer characteristic of a GaAs MSM-Si TFT IR detector built on a polyimide substrate; insets show current–voltage response at different levels of illumination with an IR light source at 850 nm, optical image, and a circuit schematic. (b) The transfer characteristic of a hybrid SWNT-Si CMOS inverter built on a silicon wafer substrate; insets show optical micrograph and a circuit schematic. (Reprinted with permission from [67]. Copyright 2006 American Association for the Advancement of Science.)

17.4.5 Stretchable Electronics Simple mechanical bendability, as allowed by the use of thin plastic substrates, is useful for many applications. These types of systems can allow, however, conformal wrapping only of cylinders and cones; spheres or more complex curvilinear surfaces (e.g., aircraft wings, human body parts, etc.) are not possible. For these cases, mechanical stretchability, which is a much more challenging characteristic than flexibility, is needed. Stretchability can also provide extremely high levels of bendability. One strategy to achieve such stretchability in inorganic electronics is to use semiconductor nanomaterials configured into “wavy” shapes. These layouts can accommodate large applied strains without fracturing the materials, with a deformation physics that is very similar to that of an accordion bellows. These wavy layouts can be fabricated by exploiting nonlinear buckling phenomena, well known in systems comprised of rigid thin layers on compliant supports as described in the context of patterning in Chapter 3. This section provides a brief summary of stretchable µs-Sc elements and their implementation in stretchable electronic devices.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

470

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

When a thin layer of stiff material bonded to a soft substrate such as rubber is subjected to compressive stress, the layer relieves the surface strain by mechanical buckling. The phenomenon can provide a route to stretchability in electronics [73, 96] and devices [60, 73], or in passive interconnects [97, 98]. The process for fabricating stretchable devices begins with µs-Sc in the form of ribbons, platelets or wires, followed by bonding to prestrained elastomeric (PDMS) substrates using the chemistries described previously [99]. Relaxing the prestrain forms buckled and “wavy” device elements that can accommodate externally applied strains by changing their shapes, such as their wavelengths and amplitudes, without inducing significant strains in the materials themselves. Figure 17.19a shows a schematic illustration of the fabrication. Here µs-Sc ribbons are first fabricated from a source wafer. A substrate of PDMS is then prestretched, either by thermal or mechanical means, and then bonded to the ribbons. Peeling back the PDMS and then releasing the prestrain leads to buckling of the ribbons to produce very regular, sinusoidal “wavy” shapes that relieve the compressive strain acting on them by the PDMS (Figure 17.19b). In this implementation, the mechanics define the wavelengths, and the ribbons are intimately bonded to the (a) Stretched PDMS

Single crystal ribbon

L+

∆L

(c)

(b)

10 µm

50 µm

100 µm

Figure 17.19. (a) Schematic illustration of procedures for fabricating “wavy” and “buckled” semiconductor nanoribbons on elastomeric PDMS substrates. (b, c ) Scanning electron micrographs of wavy Si (b) (Reprinted with permission from [96]. Copyright 2006 Wiley-VCHVerlag.) and buckled GaAs ribbons (c ) on elastomeric substrates. (Reprinted with permission from [60]. Copyright 2006 Nature Publishing group.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

471

PDMS along their entire lengths. To achieve high stretchability, the PDMS can be selectively activated by UV/ozone treatment so that only certain regions of the ribbons bond strongly to PDMS according to chemistries described previously in this chapter and in Chapter 2. Upon releasing the prestrain in PDMS, the weakly bonded areas of ribbons delaminate from the PDMS and form bridgelike structures that are capable of accommodating strains of up to 100% (Figure 17.19c). In practical embodiments, such structures are encapsulated on top with additional PDMS to eliminate the air gaps and to provide fully reversible stretching behaviors. These “wavy” structures of inorganic semiconductors on PDMS can be reversibly stretched or compressed; changes in amplitude and wavelengths accommodate the externally applied strain [73]. Figure 17.20 shows the response of Si ribbons to strain. When the initially wavy Si ribbons (middle) are compressed, the amplitudes increase and wavelengths decrease (top). The opposite is true for stretching (bottom). For functional, stretchable electronic devices on PDMS, all the device processing steps, especially those such as doping and contact metallization that can require high temperatures, are performed on the source wafer. Subsequently, ribbons with integrated device layers are configured into wavy geometries using the processes mentioned above. As an example, Figure 17.21 shows the stretchable Si nMOSFET: (a) shows the schematic cross-sectional geometry (top) and optical images (bottom) and (b) shows the output characteristics of a transistor. Upon applying compressive or tensile strains of ∼10%, these devices show good electrical performance without significant changes.

Compressed 1 0 −1

Unperturbed Height (µm)

1 0

−1

Stretched

1 0 −1

0 0 50 10 ) m Distance (µ

0

50 100 Distance (µm)

Figure 17.20. AFM images of wavy Si ribbons formed on a PDMS substrate. (Reprinted with permission from [73]. Copyright 2006 American Association for the Advancement of Science.)

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

472

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(a)

G

S

D

SiO2 Si

Compressed Stretched

(b)

2.0

−9.9% 0% 9.9%

IDS (µA)

1.5 1.0

0.0

0

1

2

3

VDS(V) Figure 17.21. (a) Schematic illustration (top) and optical images (bottom) of a stretchable Si transistor. (b) Electrical characteristics of a stretchable Si ribbon MOSFET on a PDMS substrate stretched at different levels. (Reprinted with permission from [73]. Copyright 2006 American Association for the Advancement of Science.)

Two-dimensionally buckled, or “wavy”, membranes or platelets provide a route to high performance electronics with full, two-dimensional stretchability [100]. Figure 17.22 shows some representative optical images of Si nanomembranes with herringbone wavy layouts, with a thickness of 100 nm (lateral dimension of about 4 × 4 mm2 ) collected at different stages of compression. These images indicate wavy formation in two stages—first, one-dimensional waves are predominantly formed over large areas (Figure 17.22b–c) and subsequently, herringbone layouts emerge from the bending of these wave structures upon reaching room temperature (Figures 17.22d–f ) corresponding to a total thermally induced compressive strain of 3.8%. Figure 17.23 presents AFM (atomic force microscope) and SEM (scanning electron microscope) images of structures similar to those illustrated in the fully cooled state of Figure 17.22. These data clearly show that the herringbone patterns are characterized by zigzag structures that define two characteristic directions, even though the compressive strain is completely isotropic. The herringbone structures represent a minimum elastic energy configuration that reduces the overall in-plane stress in the system and relieves biaxial compression in both directions. These geometries allow the sheets of silicon to stretch in two distinct dimensions, which can provide an effective route for building stretchable two-dimensional (2D) electronic systems. For

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

17.4 FABRICATION OF ELECTRONIC DEVICES

(a)

(b)

(c)

(d)

(e)

(f)

473

Figure 17.22. Optical images of 2D wavy structures in silicon nanomembranes at various stages of biaxial compression, ranging from 0% to 3.8%. (Reprinted with permission from [100]. Copyright 2007 American Chemical Society.)

example, a stretchable ring oscillator made of Si MOSFETs and metal interconnections with a similar design to those depicted in Figure 17.15 can be achieved using these approaches, where the device fabrication is performed on other substrates prior to transferring onto a PDMS substrate [101]. The resulting device shows wavy morphologies as shown in Figure 17.24a. Circuits fabricated in this fashion show different wave geometries, which depend on the thicknesses of the components of the circuit. Metal interconnects that were relatively thin displayed smaller wave geometries, while the thicker Si regions showed larger or an absence of waves, which suggests that the applied strain is delivered to the metal interconnects while the Si experiences a small amount of strain. Figure 17.24b shows optical images of a CMOS inverter stretched to a tensile strain of 5%. As shown in the images, the circuit can be stretched up to ∼5% without mechanical failure.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

474

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

(ii) 1 (i)

0 −1 1

20

40

50 µm

60

80

µm

0 −1 0

10 µm

50

100

2 µm

Figure 17.23. AFM and SEM images of 2D wavy structures in silicon nanomembranes on PDMS. (Reprinted with permission from [100]. Copyright 2007 American Chemical Society.) (a)

(b)

Figure 17.24. (a) Optical images of 2D wavy Si circuits under a biaxial compressive strain of 4%. (b) Images of the relaxed (left) and the stretched (vertical direction) state (right) of a pair of CMOS inverters.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

REFERENCES

475

In this chapter, we reviewed some recent work in the areas of micro/nanomaterials and fabrication methods for high performance devices and circuits on flexible and even stretchable substrates. The fabrication methods use soft lithography techniques including MIMIC, nanoimprint lithography, and dry transfer printing, in steps that define electrodes, semiconductor layers, and other components of the circuits. These types of patterning methods offer experimentally simple, high throughput, and potentially low cost operation, all of which are important for the unusual classes of electronic systems that represent the focus of this chapter. The wide-ranging application possibilities of such technologies motivate the growing interest in this field. Although the engineering challenges of developing full manufacturing capacities for electronics of this type are considerable, many basic aspects of technically feasible approaches are now emerging, including but not restricted to the methods outlined here. For these reasons, the future of electronics beyond the semiconductor wafer appears bright.

REFERENCES 1. Reuss, R. H., Chalamala, B. R., Moussessian, A., Kane, M. G., Kumar, A., Zhang, D. C., Rogers, J. A., Hatalis, M., et al. (2005) Macroelectronics: perspectives on technology and applications. Proc. IEEE 97, 1239–1256. 2. Service, R. F. (2006) Materials science-inorganic electronics begin to flex their muscle. Science 312, 1593–1594. 3. Sun, Y. and Rogers, J. A. (2007) Inorganic semiconductors for flexible electronics. Adv. Mater. 19, 1897–1916. 4. Mitzi, D. B., Copel, M., and Chey, S. J. (2005) Low-voltage transistor employing a high-mobility spin-coated chalcogenide semiconductor. Adv. Mater. 17, 1285– 1289. 5. Mitzi, D. B., Kosbar, L. L., Murray, C. E., Copel, M., and Afzali, A. (2004) Highmobility ultrathin semiconducting films prepared by spin coating. Nature 428, 299–303. 6. Kane, M. G., Goodman, L., Firester, A. H., Van Der Wilt, P. C., Limanov, A. B., and Im, J. S. (2007) 100-MHz CMOS circuits directly fabricated on plastic using sequential laterally solidified silicon. J. Soc. Inform. Display 15, 471–478. 7. Dai, Z. R., Pan, Z. W., and Wang, Z. L. (2003) Novel nanostructures of functional oxides synthesized by thermal evaporation. Adv. Funct. Mater. 13, 9–24. 8. Law, M., Goldberger, J., and Yang, P. (2004) Semiconductor nanowires and nanotubes. Annu. Rev. Mater. Res. 34, 83–122. 9. Lieber, C. M. (1998) One-dimensional nanostructures: chemistry, physics & applications. Solid State Comm 107, 607–616. 10. Briseno, A. L., Mannsfeld, S. C. B., Ling, M. M., Liu, S., Tseng, R. J., Reese, C., Roberts, M. E., Yang, Y., et al. (2006) Patterning organic single-crystal transistor arrays. Nature 444, 913–917. 11. Crone, B., Dodabalapur, A., Lin, Y.-Y., Filas, R. W., Bao, Z., LaDuca, A., Sarpeshkar, R., Katz, H. E., et al. (2000) Large-scale complementary integrated circuits based on organic transistors. Nature 403, 521–523.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

476

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

12. Menard, E., Lee, K. J., Khang, D. Y., Nuzzo, R. G., and Rogers, J. A. (2004) A printable form of silicon for high performance thin film transistors on plastic substrates. Appl. Phys. Lett. 84, 5398–5400. 13. Sun, Y. and Rogers, J. A. (2004) Fabricating semiconductor nano/microwires and transfer printing ordered arrays of them onto plastic substrates. Nano Lett. 4, 1953–1959. 14. Reese, C. and Bao, Z. (2007) Organic single-crystal field-effect transistors. Mater. Today 10, 20–27. 15. McAlpine, M. C., Friedman, R. S., and Lieber, C. M. (2005) High-performance nanowire electronics and photonics and nanoscale patterning on flexible plastic substrates. Proc. IEEE 93, 1357–1363. 16. Melosh, N. A., Boukai, A., Diana, F., Gerardot, B., Badolato, A., Petroff, P. M., and Heath, J. R. (2003) Ultrahigh-density nanowire lattices and circuits. Science 300, 112–115. 17. Menard, E., Nuzzo, R. G., and Rogers, J. A. (2005) Bendable single crystal silicon thin film transistors formed by printing on plastic substrates. Appl. Phys. Lett. 86, 093507. 18. Sun, Y., Menard, E., Rogers, J. A., Kim, H.-S., Kim, S., Chen, G., Adesida, I., Dettmer, R., et al. (2006) Gigahertz operation in flexible transistors on plastic substrates. Appl. Phys. Lett. 88, 183509. 19. Wang, Z. L. (2003) Nanowires and Nanobelts: Materials, Properties and Devices, Kluwer, Boston, Vols I and II. 20. Xia, Y. N., Yang, P., Sun, Y., Wu, Y., Mayers, B., Gates, B., Yin, Y., Kim, F., et al. (2003) One-dimensional nanostructures: synthesis, characterization, and applications. Adv. Mater. 15, 353–389. 21. Lee, K. J., Lee, J., Hwang, H., Reitmeier, Z. J., Davis, R. F., Rogers, J. A., and Nuzzo, R. G. (2005) A printable form of single-crystalline gallium nitride for flexible optoelectronic systems. Small 1, 1164–1168. 22. Mack, S., Meitl, M. A., Baca A. J., Zhu, Z. T., and Rogers, J. A. (2006) Mechanically flexible thin-film transistors that use ultrathin ribbons of silicon derived from bulk wafers. Appl. Phys. Lett. 88, 213101. 23. Ko, H. C., Baca, A. J., and Rogers, J. A. (2006) Bulk quantities of single-crystal silicon micro-/nanoribbons generated from bulk wafers. Nano Lett. 6, 2318–2324. 24. Lee, K. J., Motala, M. J., Meitl, M. A., Childs, W. R., Menard, E., Shim, A. K., Rogers, J. A., and Nuzzo, R. G. (2005) Large-area, selective transfer of microstructured silicon: a printing-based approach to high-performance thin-film transistors supported on flexible substrates. Adv. Mater. 17, 2332–2336. 25. Meitl, M. A., Zhu, Z.-T., Kumar, V., Lee, K. J., Feng, X., Huang, Y. Y., Adesida, I., Nuzzo, R. G., et al.(2006) Transfer printing by kinetic control of adhesion to an elastomeric stamp. Nat. Mater. 5, 33–38. 26. Suo, Z., Ma, E. Y., Gleskova, H., and Wagner, S. (1999) Mechanics of rollable and foldable film-on-foil electronics. Appl. Phys. Lett. 74, 1177–1179. 27. Loo, Y. L., Someya, T., Baldwin, K. W., Bao, Z., Ho, P., Dodabalapur, A., Katz, H. E., and Rogers, J. A. (2002) Soft, conformable electrical contacts for organic semiconductors: high-resolution plastic circuits by lamination. Proc. Natl. Acad. Sci. USA 99, 10252–10256.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

REFERENCES

477

28. Sekitani, T., Iba, S., Kato, Y., Noguchi, Y., Someya, T., and Sakurai, T. (2005) Ultraflexible organic field-effect transistors embedded at a neutral strain position. Appl. Phys. Lett. 87, 173502. 29. Wagner, R. S. and Ellis, W. C. (1964) Vapor-liquid-solid mechanism of single crystal growth. Appl. Phys. Lett. 4, 89–90. 30. Wu, Y., Cui, Y., Huynh, L., Barrelet, C. J., Bell, D. C., and Lieber, C. M. (2004) Controlled growth and structures of molecular-scale silicon nanowires. Nano Lett. 4, 433–436. 31. Huang, M. H., Mao, S., Feick, H., Yan, H., Wu, Y., Kind, H., Weber, E., Russo, R., et al. (2001) Room-temperature ultraviolet nanowire nanolasers. Science 292, 1897– 1899. 32. Wu, Y., Yan, H., Huang, M., Messer, B., Song, J. H., and Yang, P. (2002) Inorganic semiconductor nanowires: rational growth, assembly, and novel properties. Chem. Eur. J. 8, 1261–1268. 33. Wu, Z. H., Mei, X. Y., Kim, D., Blumin, M., and Ruda, H. E. (2002) Growth of Aucatalyzed ordered GaAs nanowire arrays by molecular-beam epitaxy. Appl. Phys. Lett. 81, 5177–5179. 34. Javey, A., Nam, S., Friedman, R. S., Yan, H., and Lieber, C. M. (2007) Layer-by-layer assembly of nanowires for three-dimensional, multifunctional electronics. Nano Lett. 7, 773–777. 35. Lu, W. and Lieber, C. M. (2006) Semiconductor nanowires. J. Phys. D: Appl. Phys. 39, R387–R406. 36. Lauhon, L. J., Gudiksen, M. S., Wang, D., and Lieber, C. M. (2002) Epitaxial core-shell and core-multishell nanowire heterostructures. Nature 420, 57–61. 37. Lieber, C. M. and Wang, Z. L. (2007) Functional nanowires. MRS Bull. 32, 99–108. 38. Dick, K. A., Deppert, K., Karlsson, L. S., Larsson, M. W., Seifert, W., Wallenberg, L. R., and Samuelson, L. (2007) Directed growth of branched nanowire structures. MRS Bull. 32, 127–133. 39. Trentler, T. J., Hickman, K. M., Goel, S. C., Viano, A. M., Gibbons, P. C., and Buhro, W. E. (1995) Solution-liquid-solid growth of crystalline of III–V semiconductors-an analogy to vapor-liquid-solid growth. Science 270, 1791–1794. 40. Yu, H. and Buhro, W. E. (2003) Solution-liquid-solid growth of soluble GaAs nanowires. Adv. Mater. 15, 416–419. 41. Holmes, J. D., Johnston, K. P., Doty, R. C., and Korgel, B. A. (2000) Control of thickness and orientation of solution-grown silicon nanowires. Science 287, 1471–1473. 42. Lu, X., Hanrath, T., Johnston, K. P., and Korgel, B. A. (2003) Growth of single crystal nanowires in supercritical silicon solution from tethered gold particles on a silicon substrate. Nano Lett. 3, 93–99. 43. Morales, A. M. and Lieber, C. M. (1998) A laser ablation method for the synthesis of crystalline semiconductor nanowires. Science 279, 208. 44. Westwater, J., Gosain, D. P., Tomiya, S., Usui, S., and Ruda, H. (1997) Growth of silicon nanowires via gold/silane vapor-liquid-solid reaction. J. Vac. Sci. Technol. B 15, 554–557. 45. Shi, W., Peng, H., Wang, N., Li, C. P., Xu, L., Lee, C. S., Kalish, R., and Lee, S.-T. (2001) Free-standing single crystal silicon nanoribbons. J. Am. Chem. Soc. 123, 11095–11096.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

478

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

46. Zhang, R.-Q., Lifshitz, Y., and Lee, S.-T. (2003) Oxide-assisted growth of semiconducting nanowires. Adv. Mater. 15, 635–640. 47. Shan, Y., Kalkan, A. K., Peng, C.-Y., and Fonash, S. J. (2004) From Si source gas directly to positioned, electrically contacted Si nanowires: the self-assembling “grow-in-place” approach. Nano Lett. 4, 2085–2089. 48. Pan, Z. W., Dai, Z. R., and Wang, Z. L. (2001) Nanobelts of semiconducting oxides. Science 291, 1947–1949. 49. Li, Y. B., Bando, Y., Sato, T., and Kurashima, K. (2002) ZnO nanobelts grown on Si substrate. Appl. Phys. Lett. 81, 144–146. 50. Arnold, M. S., Avouris, P., Pan, Z. W., and Wang, Z. L. (2003) Field-effect transistors based on single semiconducting oxide nanobelts. J. Phys. Chem. B 107, 659–663. 51. Dai, Z. R., Pan, Z. W., and Wang, Z. L. (2002) Gallium oxide nanoribbons and nanosheets. J. Phys. Chem. B 106, 902–904. 52. Wen, X., Wang, S., Ding, Y., Wang, Z. L., and Yang, S. (2005) Controlled growth of large-area, uniform, vertically aligned arrays of α-Fe2 O3 nanobelts and nanowires. J. Phys. Chem. B 109, 215–220. 53. Kong, X. Y. and Wang, Z. L. (2003) Structures of indium oxide nanobelts. Solid State Commun. 128, 1–4. 54. Bae, S. Y., Seo, H. W., Park, J., Yang, H., Park, J. C., and Lee, S. Y. (2002) Singlecrystalline gallium nitride nanobelts. Appl. Phys. Lett. 81, 126–128. 55. Wang, Z. Q., Gong, J. F., Duan, J. H., Huang, H. B., Yang, S. G., Zhao, X. N., Zhang, R., and Du, Y. W. (2006) Direct synthesis and characterization of CdS nanobelts. Appl. Phys. Lett. 89, 033102. 56. Kar, S. and Chaudhuri, S. (2005) Controlled synthesis and photoluminescence properties of ZnS nanowires and nanoribbons. J. Phys. Chem. B 109, 3298–3302. 57. Ma, C., Ding, Y., Moore, D., Wang, X., and Wang, Z. L. (2004) Single-crystal CdSe nanosaws. J. Am. Chem. Soc. 126, 708–709. 58. Ding, Y., Ma, C., and Wang, Z. L. (2004) Self-catalysis and phase transformation in the formation of CdSe nanosaws. Adv. Mater. 16, 1740–1743. 59. Melosh, N. A., Boukai, A., Diana, F., Geradot, B., Badolato, A., Petroff, P. M., and Heath, J. R. (2003) Ultrahigh-density nanowire lattices and circuits. Science 300, 112–115. 60. Sun, Y., Choi, W.-M., Jiang, H., Huang, Y., and Rogers, J. A. (2006) Controlled buckling of semiconductor nanoribbons for stretchable electronics. Nat. Nanotechnol. 1, 201– 207. 61. Lee, K. J., Meitl, M. A., Ahn, J.-H., Rogers, J. A., Nuzzo, R. G. , Kumar, V., and Adesida, I. (2006) Bendable GaN high electron mobility transistors on plastic substrates. J. Appl. Phys. 100, 124507. 62. Yin, Y., Gates, B., and Xia, Y. N. (2000) A soft lithography approach to the fabrication of nanostructures of single crystalline silicon with well-defined dimensions and shapes. Adv. Mater. 12, 1426–1430. 63. Austin, M. D., Zhang, W., Ge, H. X., Wasserman, D., Lyon, S. A., and Chou, S. Y. (2005) 6 nm half-pitch lines and 0.04 mu m(2) static random access memory patterns by nanoimprint lithography. Nanotechnology 16, 1058–1061.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

REFERENCES

479

64. Mcapline, M. C., Ahmad, H., Wang, D., and Heath, J. R. (2007) Highly ordered nanowire arrays on plastic substrates for ultrasensitive flexible chemical sensors. Nat. Mater. 6, 379–384. 65. Sun, Y., Khang, D.-Y., Hua, F., Hurley, K., Nuzzo, R. G., and Rogers, J. A. (2005) Photolithographic route to the fabrication of micro/nanowires of III–V semiconductors. Adv. Funct. Mater. 15, 30–40. 66. Sun, Y., Graff, R. A., Strano, M. S., and Rogers, J. A. (2005) Top-down fabrication of semiconductor nanowires with alternating structures along their longitudinal and transverse axes. Small 1, 1052–1057. 67. Ahn, J.-H., Kim, H.-S., Lee, K. J., Jeon, S., Kang, S. J., Sun, Y. G., Nuzzo, R. G., and Rogers, J. A. (2006) Heterogeneous three-dimensional electronics by use of printed semiconductor nanomaterials. Science 314, 1754–1757. 68. Ahn, J.-H., Kim, H.-S., Menard, E., Lee, K. J., Zhu, Z., Kim, D.-H., Nuzzo, R. G., Rogers, J. A., et al. (2007) Bendable integrated circuits on plastic substrates by use of printed ribbons of single-crystalline silicon. Appl. Phys. Lett. 90, 213501. 69. Ahn, J.-H., Kim, H.-S., Lee, K. J., Menard, E., Nuzzo, R. G., and Rogers, J. A. (2006) High-speed mechanically flexible single-crystal silicon thin-film transistors on plastic substrates. IEEE Electron. Device Lett. 27, 460–462. 70. Baca, A. J., Meitl, M. A., Ko, H. C., Mack, S., Kim, H.-S., Dong, J., Ferreira, P. M. and Rogers, J. A. (2007) Printable single-crystal silicon micro/nanoscale ribbons, platelets and bars generated from bulk wafers. Adv. Func. Mater. 17, 3051–3062. 71. Sun, Y., Kim, H.-S., Menard, E., Kim, S., Adesida, I., and Rogers, J. A., (2006) Printed arrays of aligned GaAs wires for flexible transistors, diodes, and circuits on plastic substrates. Small 2, 1330–1334. 72. Zhu, Z. T., Menard, E., Hurley, K., Nuzzo, R. G., and Rogers, J. A. (2005) Spin on dopants for high-performance single-crystal silicon transistors on flexible plastic substrates. Appl. Phys. Lett. 86, 133507. 73. Khang, D.-Y., Jiang, H., Huang, Y., and Rogers, J. A. (2006) A stretchable form of single-crystal silicon for high-performance electronics on rubber substrates. Science 311, 208–212. 74. Lagally, M. G. (2007) Silicon nanomembranes. MRS Bull. 32, 57–63. 75. Wang, D., Sheriff, B. A., and Heath, J. R. (2006) Silicon p-FETs from ultrahigh density nanowire arrays. Nano Lett. 6, 1096–1110. 76. Madou, M. (1997) Fundamentals of Microfabrication, CRC Press, Boca Raton, FL. 77. Xia, Y. N. and Whitesides, G. M. (1998) Soft lithography. Angew. Chem. Int. Ed. 37, 551–575. 78. Xia, Y. N., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1999) Unconventional methods for fabricating and patterning nanostructures. Chem. Rev. 99, 1823–1848. 79. Menard, E., Meitl, M. A., Sun, Y., Park, J.-U., Shir, D., Nam, Y.-S., Jeon, S., and Rogers, J. A. (2007) Micro- and nanopatterning techniques for organic electronic and optoelectronic systems. Chem. Rev. 107, 1117–1160. 80. Huang, Y., Duan, X., Wei, Q.Q., and Lieber, C.M. (2001) Directed assembly of onedimensional nanostructures into functional networks. Science 291, 630–633. 81. Mcalpine, M. C., Friedman, R. S., and Lieber, C.M. (2003) Nanoimprint lithography for hybrid plastic electronics. Nano Lett. 3, 443–445.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

480

14:10

Printer Name: Yet to Come

INORGANIC ELECTRONIC DEVICES

82. Meitl, M. A., Feng, X., Dong, J. Y., Menard, E., Ferreira, P. M., Huang, Y. G., and Rogers, J. A. (2007) Stress focusing for controlled fracture in microelectromechanical systems. Appl. Phys. Lett. 90, 083110. 83. Hur, S. H., Khang, D.-Y., Kocabas, C, and Rogers, J. A. (2004) Nanotransfer printing by use of noncovalent surface forces: applications to thin-film transistors that use singlewalled carbon nanotube networks and semiconducting polymers. Appl. Phys. Lett. 85, 5730–5732. 84. Vitale, S. A., Chae, H., and Sawin, H. H. (2000) Etching chemistry of benzocyclobutene (BCB) low-k dielectric films in F2 +O2 and Cl2 +O2 high density plasmas. J. Vac. Sci. Technol. A 18, 2770–2778. 85. Kim, J., Chaudhury, M. K., and Owen, M. J. (1999) Hydrophobicity loss and recovery of silicone HV insulation. IEEE Trans. Dielectr. Electr. Insul. 6, 695–702. 86. Hu, J., Beck, R. G., Westervelt, R. M., and Whitesides, G. M. (1998) The use of soft lithography to fabricate arrays of Schottky diode. Adv. Mater. 10, 574–577. 87. Hu, J., Deng, T., Beck, R.G., Westervelt, R.M., and Whitesides, G. M. (1999) Fabrication of arrays of schottky diodes using microtransfer molding. Sensors Actuators A. 75, 65–69. 88. Deng, T., Goetting, L. B., Hu, J., and Whitesides, G. M. (1999) Microfabrication of half-wave rectifier circuits using soft lithography. Sensor Actuat. A. 75, 60–64. 89. Hu, J., Beck, R. G., Deng, T., Westervelt, R. M., Maranowski, K. D., Gossard, A. C., and Whitesides, G. M. (1997) Using soft lithography to fabricate GaAs/AlGaAs heterostructure field effect transistors. Appl. Phys. Lett. 71, 2020–2022. 90. Hu, J., Beck, R. G., Deng, T., Westervelt, R. M., Maranowski, K. D., Gossard, A. C., and Whitesides, G. M. (2000) Fabrication of GaAs/AlGaAs high electron mobility transistors with 250 nm gates using conformal phase shift lithography. Sensor. Actuat. A. 86, 122–126. 91. Jeon, N. L., Hu, J., Whitesides, G. M., Erhardt M. K., and Nuzzo R. G. (1998) Fabrication of silicon MOSFETs using soft lithography. Adv. Mater. 10, 1466–1469. 92. Sze, S. M. (1981) Physics of Semiconductor Devices, Wiley, New York. 93. Angelis, C. T., Dimitriadis, C. A., Miyasaka, M., Farmakis, F. V., Kamarinos, G., Brini, J., and Stoemenos, J. (1999) Effect of excimer laser annealing on the structural and electrical properties of polycrystalline silicon thin-film transistors. J. Appl. Phys. 86, 4600–4606. 94. Kim, D.-H., Ahn, J.-H., Kim, H.-S., Lee, K. J., Kim, T.-H., Yu, C.-J., Nuzzo, R. G., Rogers, J. A. (2008) Complementary logic gates and ring oscillators on plastic substrates by use of printed ribbons of single-crystalline silicon. IEEE Electron Device Lett. 29, 73–76. 95. Yuan, H.-C. and Ma, Z. (2006) Microwave thin-film transistors using si nanomembranes on flexible polymer substrate. Appl. Phys. Lett. 89, 212105. 96. Sun, Y., Kumar, V., Adesida, I., and Rogers J. A. (2006) Buckled and wavy ribbons of GaAs for high-performance electronics on elastomeric substrates. Adv. Mater. 18, 2857–2862. 97. Lacour, S. P., Jones, J., Wagner, S., Li, T., and Suo, Z. (2005) Stretchable interconnects for elastic electronic surfaces. Pro. IEEE 93, 1459–1466. 98. Lacour, S. P., Wagner, S., Huang, Z., and Suo, Z. (2003) Stretchable gold conductors on elastomeric substrates. Appl. Phys. Lett. 82, 2404–2406.

P1: OTA/XYZ P2: ABC c17 JWBK252/Lee September 18, 2008

14:10

Printer Name: Yet to Come

REFERENCES

481

99. Sun, Y. and Rogers, J. A. (2007) Structural forms of single crystal semiconductor nanoribbons for high-performance stretchable electronics. J. Mater. Chem. 17, 832–840. 100. Choi, W.-M., Song, J., Khang, D.-Y., Jiang, H., Huang, Y., and Rogers, J. A. (2007) Biaxially stretchable “wavy” silicon nanomembranes. Nano Lett. 7, 1655–1663. 101. Kim, D. Y., Ahn, J.-H., Choi, W.-M., and Rogers, J. A., unpublished data.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18 MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES Hanqing Jiang, Jizhou Song, Yonggang Huang, and John A. Rogers

18.1 INTRODUCTION The buckling of stiff thin films on compliant substrates can be controlled in microand nanoscale systems to generate interesting structures with well-defined geometries and dimensions in the 100 nm to 100 µm range [1]. This has generated numerous theoretical and experimental studies of such systems [2–21] because their important applications in stretchable electronics [11–14, 22], micro- and nanoelectromechanical systems [23], tunable phase optics [3, 24], force spectroscopy in cells [25], biocompatible topographic matrices for cell alignment [26, 27], high precision micro- and nanometrology methods [8, 9, 28], and pattern formation for micro/nanofabrication [1, 29–34]. The controlled buckling is realized in thin films deposited onto prestrained elastomeric substrates. The release of prestrain in substrates leads to buckling of thin films. Figure 18.1 illustrates the fabrication of buckled stiff thin ribbons on compliant substrates [11], where the thin ribbons of single crystal silicon are chemically bonded to flat, prestrained elastomeric substrates of poly(dimethylsiloxane) (PDMS). The release of prestrain leads to the compressive strain in ribbons that generate the wavy layouts. Such a structure is of interest for applications in stretchable electronics.

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


483

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

484

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

Bond elements to prestrained elastomeric substrate L + dL Si

PDMS

Release L

Stretchable Si devices

Figure 18.1. Schematic illustration of the process for fabricating buckled, or “wavy”, single crystal Si ribbons on a PDMS substrate. (Reprinted with permission from [21]. Copyright 2008 Elsevier Ltd.)

Field-effect transistors, p–n diodes, and other devices for electronic circuits can be directly integrated into the wavy Si to yield fully stretchable components. Integrated electronics that use such components could be important for devices such as flexible displays [35], eyelike digital cameras [36], conformable skin sensors [37], intelligent surgical gloves [38], and structural health monitoring devices [39]. This chapter provides a review on the buckling analyses of this class of systems illustrated in Figure 18.1. The theory and experimental results for small prestrains and applied strains are presented in Section 18.2, while their counter parts for large strains are given in Section 18.3. The edge effect around the end of the ribbon and effect of finite ribbon width and spacing are discussed in Sections 18.4 and 18.5, respectively. The buckling of thin membranes is studied in Section 18.6, where the unique two-dimensional buckling pattern named the herringbone structure is observed and studied. Section 18.7 presents a method to precisely control the buckling of such systems to achieve large stretchability. 18.2 BUCKLING ANALYSIS OF STIFF THIN RIBBONS ON COMPLIANT SUBSTRATES We consider a stiff thin film (ribbon) of thickness hf and elastic modulus Ef on a prestretched (prestrain, εpre ), compliant substrate of modulus Es , E f  E s

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.2 BUCKLING ANALYSIS OF STIFF THIN RIBBONS ON COMPLIANT SUBSTRATES

485

[4–5, 10]. One example is the silicon ribbon (Ef = 130 GPa) on a PDMS substrate (Es = 1.8 MPa) with the elastic moduli different by five orders of magnitude [11]. The thin ribbon is modeled as a beam since the wavelength of buckled ribbon (∼15 µm) is much larger than the ribbon thickness (0.1 µm). The beam, however, undergoes large rotation once the ribbon buckles. The membrane strain ε11 is related to the in-plane displacement u1 and out-of-plane displacement w by ε11 =

du 1 1 + dx1 2




dw dx1

2 (18.1)

where x1 is the coordinate along the beam direction. The ribbon is linearly elastic, and its deformation is assumed to be plane strain. The membrane force N 11 is then given by N11 = E f h f ε11

(18.2)

where E f is the plane-strain modulus of the ribbon. The shear and normal tractions at the ribbon–substrate interface is obtained from the force equilibrium as T1 =

dN11 dx1

(18.3)

and E f h 3f d4 w d T3 = − + 12 dx14 dx1




dw N11 dx1

 (18.4)

The strain energy density in the ribbon consists of the membrane energy density W m and the bending energy density W b , which are given by Wm =

E fhf 2 1 N11 ε11 = ε , 2 2 11

(18.5)

and E f h 3f Wb = 24




d2 w dx12

2 (18.6)

The out-of-plane displacement of the buckled thin ribbon can be represented by


2π x1 w = A cos (kx1 ) = A cos λ

 (18.7)

where the amplitude A and wavelength λ (or wave number k) are to be determined. The bending energy U b is the integration of W b in equation 18.6 over the buckle

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

486

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

wavelength, and is given in terms of the buckle amplitude A and wavelength λ by Ub =

π 4 E f h 3f A2 3λ3

(18.8)

The effect of interface shear is negligibly small on the buckling of stiff thin ribbon–compliant substrate system [10]. The vanishing shear T 1 = 0 in equation 18.3 gives the in-plane displacement 
π A2 4π x1 − εpre x1 u1 = sin 4λ λ

(18.9)

where the last term represents the uniform displacement field in the ribbon if the ribbon does not buckle after the prestretched PDMS is relaxed, and the first term on the right hand side is the axial displacement associated with the buckling. The membrane strain is ε11 =

π 2 A2 − εpre λ2

(18.10)

The membrane energy U m is the integration of W m in equation 18.5 over the buckle wavelength, and is given in terms of A and λ by 1 Um = E f h f 2




π 2 A2 − εpre λ2

2 λ

(18.11)

The substrate is linear elastic and is modeled as a semi-infinite solid since it is six orders of magnitude thicker than the thin ribbon. For the semi-infinite solid subjected to the normal displacement in equation 18.7 and vanishing shear on its boundary, the strain energy over the buckle wavelength in the substrate is [10, 11] Us =

π E s A2 4

(18.12)

where E s is the plane-strain modulus of the substrate. The amplitude A and wavelength λ are then determined by minimizing the total energy ∂ ∂ (Um + Ub + Us ) = (Um + Ub + Us ) = 0 ∂A ∂λ

(18.13)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.2 BUCKLING ANALYSIS OF STIFF THIN RIBBONS ON COMPLIANT SUBSTRATES

487

This gives a constant wavelength independent of the prestrain  λ0 = 2π h f

Ef

1/3

3E s

(18.14)

and the amplitude  A0 = h f

εpre −1 εc

(18.15)

 2/3 s where εc = 14 3E is the critical buckling strain (the minimal prestrain to induce Ef buckling), which is 0.034% for the Si ribbon (Ef = 130 GPa, ν f = 0.27) [40] on PDMS substrate (Es = 1.8 MPa, ν s = 0.48) [28]. For the prestrain εpre < εc (= 0.034% for the Si–PDMS system), relaxing the prestrain does not lead to buckling. Instead, the ribbon supports a small compressive strain −εpre , which we refer to as membrane strain. When εpre > εc , the ribbon buckles to relieve some of the strain. Equation 18.1, together with the wavelength and amplitude in equations 18.14 and 18.15, gives a constant membrane strain εmembrane = −εc

(18.16)

where the membrane strain is evaluated at the plane that lies at the midpoint of the ribbon thickness. 2 The bending strain in the buckled ribbon is obtained from the maximum of h2f ddxw2 1 as εbending =

2π 2 Ah f λ2

(18.17)

The maximum strain in the ribbon, also called the peak strain εpeak , is the sum of membrane strain and bending strain. In most cases of practical interest, the bending strain is much larger than the membrane strain; thus, this peak strain can be written as √ εpeak ≈ 2 εpre εc

(18.18)

It is typically much smaller than the prestrain that the ribbon accommodates by buckling. For example, in the case of εpre = 29.2%, εpeak is only 1.8% for the system of Figure 18.1. This mechanical advantage provides an effective level of stretchability/compressibility in materials that are intrinsically brittle. For the buckled system subjected to the applied strain εapplied (i.e., postbuckling behavior), the above results still hold except the prestrain εpre is replaced by εpre − εapplied . The wavelength is still independentof the strain. The membrane strain

remains at −εc and the bending strain becomes 2 εpre − εapplied εc .

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

488

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

18.3 FINITE-DEFORMATION BUCKLING ANALYSIS OF STIFF THIN RIBBONS ON COMPLIANT SUBSTRATES The recent experiments at large strains [11, 15] showed a clear and systematic decrease in wavelength with increasing prestrain as shown in Figure 18.2. This straindependent wavelength has also been reported for layers of polystyrene on PDMS

Wavelength (µm)

15

10 Experiment Finite deformation FEA Previous model

5

0 0

0

20

30

40

Prestrain (%) (a)

Amplitude (µm)

3

2 Experiment Finite deformation FEA Previous model

1

0 0

0

20

30

40

Prestrain (%) (b) Figure 18.2. (a) Wavelength and (b) amplitude of buckled structures of Si (100 nm thickness) on PDMS as a function of the prestrain. The finite-deformation buckling theory yields wavelengths and amplitudes that both agree well with experiments and finite element analysis. Results from previous mechanics models (i.e., small deformation limit) are also shown. (Reprinted with permission from [21]. Copyright 2008 Elsevier Ltd.)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.3 FINITE-DEFORMATION BUCKLING ANALYSIS OF STIFF THIN RIBBONS

489

substrates for the prestrain up to 10% [3] and in platinum ribbons on rubber substrates for prestrains up to 400% [41]. Jiang et al. [15] and Song et al. [21] attributed the strain-dependent wavelength to the finite deformation (i.e., large strain) in the compliant substrate and established a buckling theory that accounts for finite deformations and geometrical nonlinearities to yield a quantitatively accurate description of the system. It is different from all previous buckling analyses in the following three aspects. (i) Finite geometry change. The initial strain-free (or stress-free) states are different for the PDMS substrate and Si thin ribbon. As illustrated in Figure 18.1, the Si thin ribbon is strain free in the top configuration, but becomes compressed in the bottom configuration. On the contrary, the PDMS substrate is stretched in the top configuration and becomes relaxed in the bottom one. This will be further illustrated in Figure 18.3. (ii) Finite strain. The strain–displacement relation in the PDMS substrate becomes nonlinear since the maximum prestrain in the experiments is 28% [15]. (iii) Constitutive model. The stress–strain relation in the PDMS substrate becomes nonlinear at the large prestrain. The top figure in Figure 18.3 shows the initial, strain-free state of the PDMS at the original length L0 before stretching. The middle figure shows the stretched PDMS attached to a strain-free Si thin ribbon. The length of the PDMS becomes

1 + εpre L 0 , which is also the original length of the Si ribbon, where εpre is the prestrain in the stretched PDMS. Releasing the prestrain buckles the Si ribbon, as  illustrated in the bottom figure. The coordinate x1 in the middle figure is related to  x1 in the top figure by x1 = 1 + εpre x1 . The thin ribbon is still modeled as a beam. Equations 18.1 to 18.6 hold except that the coordinate x1 is replaced by x1 . For example, the membrane strain in equation 18.1 now becomes ε11

du 1 1 =  + dx1 2




dw dx1

2 (18.19)

The out-of-plane displacement of the buckled thin ribbon can be represented by


2π x1 w = A cos λ





2π x1

= A cos 1 + εpre λ

(18.20)

in the strain-free configuration (middle figure, Figure 18.3) as well as in the relaxed configuration (bottom figure, Figure 18.3). The bending energy U b is the

integration of bending energy density W b over the length of strain-free thin ribbon 1 + εpre L 0

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

490

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

L0 x1

Strain-free PDMS

x3 Stretch (1 + εpre)L0 x'1 Strain-free Si

Stretched PDMS

x'3

Release L0 x1

Relaxed PDMS

Buckled Si film x3

Figure 18.3. Three sequential configurations for the thin film–substrate buckling process. The top figure shows the undeformed substrate with the original length L0 , which represents the zero strain energy state. The middle figure shows the substrate deformed by the prestrain and the integrated film, which represents its zero strain energy state. The bottom figure shows the deformed (buckled) configuration. (Reprinted with permission from [21]. Copyright 2008 Elsevier Ltd.)

(middle figure, Figure 18.3) as Ub =



E f h 3f A2 π4



4 1 + εpre L 0 3 1 + εpre λ

(18.21)

where L0 is the length of strain-free PDMS substrate (top figure, Figure 18.3). The neglect of interface shear gives the in-plane displacement  εpre 4π x1 π A2

sin

x u1 = − 1 + εpre 1 1 + εpre λ 4 1 + εpre λ

(18.22)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.3 FINITE-DEFORMATION BUCKLING ANALYSIS OF STIFF THIN RIBBONS

491

where the last term represents the uniform displacement field in the ribbon if the ribbon does not buckle after the prestretched PDMS is relaxed, and the first term on the right hand side is the axial displacement associated with the buckling. The membrane strain becomes εpre π 2 A2 − ε11 =

2 2 1 + εpre 1 + εpre λ

(18.23)

The membrane energy U m is the integration

of membrane energy density W m over the length of strain-free thin ribbon 1 + εpre L 0 (middle figure, Figure 18.3),  2

εpre π 2 A2 1 Um = E f h f 1 + εpre L 0 −

2 2 2 1 + εpre 1 + εpre λ

(18.24)

The displacements in the substrate are denoted by u 1 (x1 , x3 ) and u 3 (x1 , x3 ), where x1 and x3 are the coordinates for the strain-free configuration of PDMS substrate (top figure, Figure 18.3). For large stretch, the Green strains EIJ in the substrate are related to the displacements as E IJ =

1 2




∂u I ∂u J ∂u K ∂u K + + ∂xJ ∂ xI ∂ xI ∂ xJ

 (18.25)

where the subscripts I and J are 1 or 3. The neo-Hookean constitutive law [42], which is the simplest nonlinear elastic constitutive relation, is used to represent the substrate TIJ =

∂ Ws ∂ E IJ

(18.26)

where TIJ is the 2nd Piola–Kirchhoff stress, and the strain energy density W s takes Es Es the form Ws = 6(1−2ν ( I¯1 − 3). Here J is the volume change at (J − 1)2 + 4(1+ν s) s) a point and is the determinant of deformation gradient FiJ , I¯1 is the trace of the left Cauchy–Green strain tensor Bij = FiK FjK times J −2/3 , Es and ν s are Young’s modulus and Poisson’s ratio of the substrate, respectively, and Es = 1.8 MPa and ν s = 0.48 for PDMS [28]. The force equilibrium equation for finite deformation is (FiK TJK ),J = 0

(18.27)

and the traction on the surface is Ti = Fi K TJ K n J , where nJ is the unit normal vector of the surface. Song et al. [21] obtained the analytical solution in the substrate and

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

492

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

gave the strain energy in the substrate as 
5 π 2 A2 π E s A2 L0 1+ Us = 3 λ 32 λ2

(18.28)

where L0 is the original length of the substrate. Minimization of the total energy (sum of the membrane and bending energy in the thin ribbon and the strain energy in the substrate) gives the wavelength and amplitude1 λ0 λ=

1 + εpre (1 + ξ )1/3

(18.29)

1 A0 A≈  1 + εpre (1 + ξ )1/3

(18.30)

where λ0 and A0 respectively are the wavelength and amplitude in equations 18.14

 and 18.15 based on the small-deformation theory and ξ = 5εpre 1 + εpre 32. Contrary to the small-deformation theory, the wavelength depends on the prestrain. As shown in Figure 18.2, both the wavelength and amplitude agree well with the experimental data. The finite element method is used to study the buckling of stiff thin ribbon– compliant substrate system [21]. The modeled system consists of a 3.02-mmthick PDMS substrate and a 100-nm-thick Si thin ribbon. The length of the twodimensional system is 1 mm. The Si thin ribbon is modeled by the beam elements. The substrate is modeled by the 4-node plane-strain element with the smallest element size of 0.4 µm × 0.4 µm. Once the substrate is subjected to the prestrain εpre , its top free surface is attached to the thin ribbon by sharing the ribbon and substrate with the same nodes at the interface. The numerical results are also shown in Figure 18.2 and they agree well with the experimental data and the analytical solution. The membrane strain becomes εmembrane

ξ 3 ε =− 1/3 c (1 + ξ ) 1+

(18.31)

which remains essentially a constant, −εc , for the prestrain up to 100%. The bending strain in the buckled ribbon is  

1+ξ 3 εpre 1/3 √ εbending = 2 (1 + ξ ) εc − εc (18.32) 1/3 1 + εpre (1 + ξ ) 

εpre 1 + ξ/3 − 1 + εpre εc (1 + ξ )1/3 . For εpre  εc , the  1 + εpre (1 + ξ )1/3

exact solution of the amplitude is A = h f  εpre numerator is approximately − 1, which gives the amplitude in equation 18.30. εc

1 The

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.3 FINITE-DEFORMATION BUCKLING ANALYSIS OF STIFF THIN RIBBONS

Strain (%)

2

493

Finite deformation εpre = 29.2% FEA

Peak strain

1

Memrane strain 0 10

0

20

30

40

Prestrain (%) Figure 18.4. Membrane and peak strains in the Si as a function of prestrain for a system of buckled Si ribbons (100 nm thickness) on a PDMS substrate. The membrane strain is a small and constant throughout this range. (Reprinted with permission from [21]. Copyright 2008 Elsevier Ltd.)

The peak strain is given by (1 + ξ )1/3 √ εpeak = 2 εpre εc  1 + εpre

(18.33)

Figure 18.4 shows εpeak and εmembrane as a function of εpre . Both the membrane and peak strains agree well with finite element analysis. The membrane strain is negligible compared to the peak strain. Likewise, the peak strain is much smaller than the prestrain, such that the system can accommodate large strains. As a result, εpeak determines the point at which fracture occurs in the ribbon. For Si, the fracture strain is around εfracture  = 1.8%. The maximum allowable prestrain is, therefore, approxi2 2 εfracture 43 εfracture mately 4εc 1 + 48 4εc , which, for the system examined here, is approximately 29% or almost 20 times larger than εfracture . For the system subjected to the applied strain after buckling (i.e., postbuckling behavior), the minimization of total energy gives the wavelength and amplitude2  2

 εpre − εapplied εc − 1 A ≈ h f 

1 3 1 + εpre 1 + εapplied + ζ / (18.34)



λ0 1 + εapplied  λ =



1 3 , 1 + εpre 1 + εapplied + ζ / 

εpre − εapplied 1 + εapplied + ζ /3 −

1/3 1 + εpre εc 1 + εapplied + ζ 2 The exact solution of the amplitude is A = h . For f

1/3  1 + εpre 1 + εapplied + ζ  εpre − εapplied − 1, which gives the amplitude in εpre − εapplied  εc , the numerator is approximately εc equation 18.34.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

494

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

Wavelength (µm)

15

10

Experiment Finite deformation FEA Previous model

5

0 −10

−5

0

5

10

15

10

15

Applied strain (%) (a)

Amplitude (µm)

3

2

Experiment Finite deformation FEA Previous model

5

0 −10

−5

0

5

Applied strain (%) (b) Figure 18.5. (a) Wavelength and (b) amplitude of buckled structures of Si (100 nm thickness) on PDMS formed with a prestrain of 16.2%, as a function of the applied strain. The measured wavelength increases for tensile strain and the measured amplitude decreases, reaching zero once the tensile strain reaches the prestrain. The finite-deformation buckling theory yields wavelengths and amplitudes that both agree well with experiments and finite element analysis. Results from previous mechanics models (i.e., small deformation limit) are also shown. (Reprinted with permission from [21]. Copyright 2008 Elsevier Ltd.)



 where ζ = 5 εpre − εapplied 1 + εpre 32. Figure 18.5 shows the experimentally measured and theoretically predicted wavelength and amplitude versus applied strain εapplied for a buckled Si thin ribbon–PDMS substrate system formed at the prestrain of 16.2%. The constant wavelength and the amplitude predicted by the previous mechanics models, with εpre replaced by εpre − εapplied , are also shown as well as the finite element results. The measured wavelength increases with tension and the amplitude decreases and becomes zero once the tensile strain reaches the prestrain. The

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.4 EDGE EFFECTS

495

finite-deformation buckling theory agrees well with experiments and finite element analysis for both amplitude and wavelength. The previous mechanics models also capture the amplitude trend but deviate from the experimental results for large tensile strain (>10%). The amplitude A vanishes when the applied strain reaches the prestrain plus the critical strain εc . This gives the stretchability (maximum applied tensile strain) as εpre + εfracture + εc , which varies linearly with the prestrain. The peak strain in the ribbon is εpeak

1 3 

1 + εapplied + ζ /  ≈ 2 εpre − εapplied εE 1 + εpre

(18.35)

The compressibility, which is the maximum applied compressive strain when the peak Si strain reaches εfracture , decreases almost linearly with  increasing prestrain and 2 2 εfracture 43 εfracture is reached. vanishes when the maximum applicable prestrain 4εc 1 + 48 4εc   2 2 εfracture 43 εfracture Therefore, the compressibility is well approximated by 4εc 1 + 48 4εc − εpre . As the prestrain increases, the stretchability improves but the compressibility worsens.

18.4 EDGE EFFECTS Although Sections 18.2 and 18.3 capture the effects in systems that are far from boundaries, they do not apply to regions of the ribbons that lie near the edges. In these regions, the amplitudes of the waves decrease gradually to zero at the edge, i.e., the edge effect, which can be seen clearly in atomic force microscope (AFM) images of Figure 18.6 for 100-nm-thick ribbons of Si with widths of 20 µm and separations of 20 µm, bonded to a 3.5-mm-thick PDMS substrate [19]. The frames on the left (images and linecuts) correspond to periodic structures that exist at regions away from the free edge; those on the right show regions near the free edge. The bottom images are obtained from finite element analysis, as described in the following. The modeled two-dimensional system consisted of a 1.3-mm-long and 3.5-mmthick PDMS substrate subjected to a prestrain εpre imposed at its two edges. A shorter, 1-mm-long, and much thinner, 100-nm-thick, Si thin ribbon resides on and is intimately coupled mechanically to the top surface of the stretched PDMS substrate. The thin ribbon covers the center part of the substrate surface, leaving bare regions of the substrate (uncovered by the ribbon) near the two edges. The wavelength λ and amplitude A near the center of ribbon (i.e., away from the edges) obtained from the finite element analysis agree well with the AFM measurement and with the analytical solution of λ and A according to equations 18.14 and 18.15. This provides validation of the numerical method. The edge-effect length Ledge , defined as the distance from the edge to the midpoint between the first peak and valley nearest to the free edge (Figure 18.7, top inset), is shown in Figure 18.7 (top main frame) versus εpre . The experimentally measured

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

496

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

(a)

Top Mid. Bot.

Height (µm)

(b)

Height (µm)

1.0

1.0

0.5 0.0 −1.5 −1.0 0.00

0.03

0.06

0.09

Top Mid. Bot.

0.5 0.0 −1.5 −1.0 0.00

Axial distance (mm)

0.03

0.06

0.09

Axial distance (mm)

(c)

0.1822 mm

0.1464 mm

Figure 18.6. (a) Images and (b) linecuts of atomic force micrographs, and (c ) finite element results of buckled single crystal Si ribbons on PDMS substrate. The images in the left and right columns are for the center part and edge of the Si thin film, respectively. (Reprinted with permission from [19]. Copyright 2007 American Institute of Physics.)

Ledge from Figure 18.6 agrees well with the numerical results in Figure 18.7 without any parameter fitting. The wavelength near the center of the ribbon is about λ = 14 µm. The bottom left panel in Figure 18.7 shows the edge-effect length Ledge versus the prestrain εpre for two different values of the PDMS modulus E s = 2 and 3 MPa. The value of Ledge for E s = 2MPa is higher than that for 3 MPa, which suggests that the edge effect increases as the substrate modulus decreases. In fact, the dimensional analysis shows the edge-effect length to be proportional to the ribbon thickness hf ,

L edge = h f f εpre , εc

(18.36)

where f is a nondimensional function of the prestrain εpre and critical strain for buckling εc , and it decreases as εpre or εc increases.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

497

18.4 EDGE EFFECTS

Prestrain (%) 0

5

10

15

20

80

Ledge (µm)

60

Displacement W(µm)

Experiments Finite element Power-law fit.

40

20

−2.4

First peak First valley Edge

−2.8 −3.2

Ledge εPre = 2.92% 1.00 0.92 0.96 Axial distance (mm)

0

Ledge (µm)

60

Edge

40

Edge

εPre = 2.92%

EPDMS = 2 MPa

4

20 0

2

6 EPDMS = 3 MPa 0

5

15 10 Prestrain (%)

20 0.0

0.2

0.4

0.6

0.8

Axial forxce (mN)

0

80

8 1.0

Axial distance (mm)

Figure 18.7. The edge-effect length Ledge versus prestrain for the Young’s modulus 3 MPa of the PDMS substrate (top panel). Inset gives the definition of Ledge . The edge-effect length is shown in the bottom left panel for the Young’s modulus 2 and 3 MPa of the PDMS substrate. The distribution of axial force in the Si thin film is shown in the bottom right panel. (Reprinted with permission from [19]. Copyright 2007 American Institute of Physics.)

The distribution of axial force in the buckled thin ribbon, as shown in Figure 18.7 (bottom right), explains the edge effect. Except near the edges, the axial force is a constant in the thin ribbon, and this compressive force causes the thin ribbon to buckle. However, this compressive force decreases to zero at the free edges such that the ribbon in these regions does not buckle. Therefore, the flat region of the thin ribbon results from the traction-free boundary condition at edges. Besides the linear elasticity model described above, we implemented a hyperelasticity model for the PDMS. For the same Young’s modulus and Poisson’s ratio, the two constitutive models give essentially the same edge-effect length. This outcome is reasonable because, for up to 20% strain, the difference between the two constitutive models is small. In summary, the flat region around the edges of buckled thin ribbons on compliant substrates results from the traction-free edges. The edge-effect length Ledge is proportional to the thin-ribbon thickness and decreases with the increasing prestrain

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

498

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

and substrate modulus. Such results might provide useful design guidelines for implementation of these edge regions in certain classes of applications in stretchable electronics and other areas. For example, well-placed edges can lead to flat regions in a larger scale buckled system where, for example, planarity is required for efficient photodetection or other functions. These and other possibilities appear to be interesting topics for further study.

18.5 EFFECT OF RIBBON WIDTH AND SPACING The analyses in the above sections involve one critical assumption that the thin ribbon width (the dimension perpendicular to the prestrain direction) is much larger than the wavelength such that the deformation is plane strain. Such an assumption, however, does not hold in many applications. For instance, the thin ribbon in stretchable metal interconnects [6] is a one-dimensional-like stripe, for which the plane-strain assumption does not hold. Our own experiments to be discussed in the following also show the strong effect of ribbon width. Figure 18.8 shows some experimental results of the ribbon width effect on the buckling profile. An optical microscope image of buckled, 5-µm-wide Si ribbons (25 µm spacing) is shown in Figure 18.8a, and a three-dimensional AFM perspective view of a buckled 100-µm-wide ribbon in Figure 18.8b. In Figure 18.8c, plane-view AFM images of Si ribbons with different widths (2, 5, 20, 50, and 100 µm, from top to bottom) are stacked together. The peaks of waves in each ribbon are aligned at the left side and marked with long, vertical red line, and 4th wave peaks are marked with short red lines on each ribbon. From this series of images, the variation of wavelength can clearly be seen; the wavelength increases with the ribbon width and then seems to saturate at a finite value. For the quantitative comparison of wave profile for each ribbon width, the linecut profiles from the AFM measurements are replotted in Figure 18.8d for the 2- and 20-µm-wide ribbons. The data are shifted to make the peaks at the same location, thereby making it easy to observe that the buckling amplitude and wavelength increase with the ribbon width, i.e., strong ribbon width effect. Jiang et al. [17] obtained the analytical solution for the buckling of a finite-width stiff ribbon on a compliant substrate, shown schematically in Figure 18.9. The ribbon width is denoted by W. The ribbon is modeled as a beam such that equations 18.1 to 18.11 still hold except that the bending energy in equation 18.8 and membrane energy in equation 18.11 need to be multiplied by the ribbon width W. The substrate is modeled as a three-dimensional, semi-infinite solid with traction-free surface except for the portion underneath the ribbon. The strain energy in the substrate can be analytically as

Us =

1 π Es






1 2 2 1 2 2 k h f + A k − εpre E f kh f A 12 4

2 ρ (W k)

(18.37)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

499

18.5 EFFECT OF RIBBON WIDTH AND SPACING

(b)

(a)

800 W = 2 µm W = 20 µm

Height (nm)

400 0 −400

−800

(c)

0

30 60 Distance (µm)

90

(d)

Figure 18.8. Some representative experimental results on the width-dependent buckling profile. (a) Optical microscopy image of buckled 5-µm-wide Si ribbons (spaced 25 µm apart) on PDMS (scale bar = 25 µm). (b) Three-dimensional atomic force microscope (AFM) perspective view of buckled 100-µm-wide Si ribbons. (c) Stacked plane-view AFM images of buckled Si ribbons, having different width of 2, 5, 20, 50, and 100 µm (from top to bottom). A wave peak in each image is aligned at the left and marked with long line, and their 4th peaks are marked with short lines indicating the variation of the wavelength with respect to the ribbon width. It can be clearly seen that the wavelength increases and then seems to saturate at/above a certain value, as the Si ribbon width increases. (d) AFM line-cut profiles along the buckled wavy Si ribbons for 2-µm-wide and 20-µm-wide ribbons. For comparison purpose, those profiles are shifted to have peaks at the same origin; again, the wavelength and amplitude increase as the ribbon width increases. (Reprinted with permission from [17]. Copyright 2008 Elsevier Ltd.)

where ρ (x) = −1 + xY1 (x) + x 2 Y0 (x) +

π 2 x [H1 (x) Y0 (x) + H0 (x) Y1 (x)] 2

(18.38)

is a nondimensional function, Yn (n = 0, 1, 2,. . .) is the modified Bessel function of the second kind, and Hn (n = 0, 1, 2,. . .) denotes the Struve function [43]. The energy minimization gives the following governing equation for the wave number k W 2 k 2 [ρ (W k)]2 EsW 3 2 = 3 3π ρ (W k) + 1 − W kY1 (W k) Ef hf

(18.39)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

500

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

x2 W x1 x3 PDMS

Figure 18.9. Schematic illustration of the geometry and coordinate system for a buckled single thin film on PDMS substrate. W is the width of the thin film. (Reprinted with permission from [17]. Copyright 2008 Elsevier Ltd.)

The wave number k has the following dependence on the ribbon and substrate elastic moduli and film thickness hf and width W,

1 hf

k=



3E s Ef

1/3



Es f Ef

1/3

 W hf

(18.40)

 1/3 W where f is a nondimensional function of its variable EEfs to be determined hf numerically from equation 18.38. Figure 18.10 shows the nondimensional func 1/3 W . It is a universal relation for all ribbon and tion f versus its variable EEfs hf substrate elastic properties, as well as ribbon width and thickness. As shown in Figure 18.10,this universal relation is very well approximated by the simple relation  16 14 f (x) ≈ coth 15 x , where coth is the hyperbolic cotangent function (cosh/sinh). Therefore, the wavelength λ = 2π /k can be obtained as
λ = 2π h f

Ef 3E s

1/3

    1/3 1/4    16 W   Es tanh   Ef hf  15 

(18.41)

which suggests that the wavelength depends on the film width W through the nondi 1/3 W mensional combination EEfs . hf The amplitude is obtained from energy minimization as  2 ε − F εpre ≥ F A = k pre εpre < F 0

(18.42)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.5 EFFECT OF RIBBON WIDTH AND SPACING

501

10

f x ≈ coth

f

Es Ef

1/3

16 1/4 x 15

W hf

1 1E-4

1E-4

0.01

0.1

Es Ef

1/3

1

10

100

W hf

   1/3 Es W versus a combination parameter Figure 18.10. Plot of dimensionless function f Ef hf   1/3  Es W 16 1/4 is given in dotted line. (Reprinted E h . An approximated expression f (x) ≈ coth 15 x f

f

with permission from [17]. Copyright 2008 Elsevier Ltd.)

where F=

1 π W Es + h2k2 4h E f ρ (W k) 12 f

(18.43)

Figures 18.11a and b respectively show the buckling wavelength and amplitude versus the ribbon width given by equations 18.41 and 18.42 (solid line) as well as the experimental results (filled circles). The thickness of Si ribbon is 100 nm and the prestrain is 1.3%. It is clear that the buckling profile depends strongly on film width since the buckling wavelength varies from 15.5 µm for 100-µm-wide ribbon to 12.5 µm for 2-µm-wide ribbon. The analytical model in this section agrees very well with experiments. The effect of ribbon spacing is studied via the model of two ribbons with the same thickness hf and width W shown schematically in Figure 18.12. The ribbon spacing is denoted by s. For large ribbon spacing, the ribbons buckle independently. For small spacing, the two ribbons have strong interactions and therefore buckle together with the same wavelength and same phase. Jiang et al. [17] obtained the analytical solution for this problem. Figure 18.13 shows the wavelength versus the ribbon spacing s for two moderately wide ribbons (width W = 20 µm) and two narrow ribbons (width W = 2 µm). For the two limits of ribbon spacing s approaching infinity and zero, the wavelength λ becomes that for a single ribbon of width 2W and W, respectively. For moderately wide ribbons (W = 20 µm), the effect of ribbon spacing is almost negligible since the wavelength varies from 15.4 µm (s → ∞) to 15.0 µm (s → 0). For narrow ribbons (W = 2 µm), the effect of ribbon spacing is significant, 12.5 µm

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

502

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

16

Wavelength (µm)

15

Analytical modeling Experimental data

14 13 12 11 10

0

40

80 120 160 Ribbon width (µm)

200

(a) 0.60

Amplitude (µm)

0.55 0.50

Analytical modeling

0.45

Experimental data

0.40 0.35 0.30

0

20

40 60 80 Ribbon width (µm)

100

(b) Figure 18.11. The buckling profile, wavelength for (a) and amplitude for (b), as a function of the width of silicon thin films. The theoretical analysis is shown in solid line and the experimental data are shown in filled circles. (Reprinted with permission from [17]. Copyright 2008 Elsevier Ltd.)

(s → ∞) to 11.2 µm (s → 0). Only when the ribbon spacing reaches about three times the width (i.e., 6 µm) the effect of ribbon spacing disappears.

18.6 BUCKLING ANALYSIS OF STIFF THIN MEMBRANES ON COMPLIANT SUBSTRATES Choi et al. [14] and Song et al. [20] produced biaxially stretchable wavy silicon nanomembranes on elastomeric PDMS substrate to provide full two-dimensional (2D) stretchability. As illustrated in Figure 18.14, the approach involved first

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.6 BUCKLING ANALYSIS OF STIFF THIN MEMBRANES ON COMPLIANT SUBSTRATES

503

W W

x2

s

x3

x1

PDMS

Figure 18.12. Schematic illustration of the geometry and coordinate system for two buckled thin films on PDMS substrate, with identical thickness and width W . s is the spacing between two thin films. (Reprinted with permission from [17]. Copyright 2008 Elsevier Ltd.)

the delineation of nanomembranes of Si (thickness between 55–320 nm) from silicon-on-insulator wafers (top silicon is (1 0 0)) by photolithographic processing and etching of the top silicon. Next, the buried SiO2 layer is removed by hydrofluoric acid to yield membranes that rest on, but are not bonded to, the underlying wafer. The lateral dimensions of these membranes are typically a few millimeters by a few millimeters. Casting and curing prepolymers of PDMS against polished silicon wafers generated flat, elastomeric substrates (about 4 mm thick). Heating the substrates in a convection oven induced a controlled degree of isotropic thermal expansion. Contacting the prestrained PDMS to the Si nanomembranes formed strong chemical bonds between these materials. Peeling back the PDMS and flipping it over, yielded Si–PDMS structures (Figure 18.14a). Cooling to room temperature released the thermally induced prestrain; thereby, causing the PDMS to relax back to its unstrained state (Figure 18.14b). This relaxation led to the spontaneous formation of 2D wavy patterns on the surface. These patterns exhibited different behaviors near the edges, where one-dimensional (1D) periodic waves (shown schematically in Figure 18.15a) predominated, at inner regions, where 2D herringbone layouts (shown schematically in Figures 18.14c and d and Figure 18.15c) were typically observed, and near the centers, where disordered herringbone structures often occurred. The images in Figure 18.14 and Figure 18.15 clearly show that the herringbone patterns are characterized by zigzag structures that define two characteristic directions, even though the compressive strain is completely isotropic. The herringbone region is characterized by the perpendicular distance between adjacent sinusoidal contours, which we refer as the short wavelength λ, the amplitude A  of wave out of the plane of the membrane, and a longer distance λ2 = 2π k2 associated with the separation between adjacent jogs in the herringbone structure, which we refer to as the long wavelength. Other characteristic lengths are the jogs wavelength λ1 = 2π k1 , the amplitude B of the jogs in the plane of the membrane, and the jog angle θ . Except for the amplitude A, all parameters are illustrated in Figures 18.14c and d, where λ2 and λ1 are along the x2 and x1 directions, respectively.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

504

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

Equivalent to a single film with 40 µm width

Wavelength λ (µm)

15.4 15.3

W = 20 µm 15.2 15.1

Equivalent to a single film with 20 µm width

15.0 0

10

20

30

40

50

Film spacing, s (µm) (a)

12.6 Equivalent to a single film with 4 µm width

Wavelength λ (µm)

12.4 12.2 12.0

W = 2 µm

11.8

Equivalent to a single film with 2 µm width

11.6 11.4 11.2 0

2

4

6

8

10

12

Film spacing, s (µm) (b) Figure 18.13. The buckling wavelength as a function of spacing, s, between two thin films with identical width of (a) 20 µm and (b) 2 µm. (Reprinted with permission from [17]. Copyright 2008 Elsevier Ltd.)

18.6.1 One-Dimensional Buckling Mode The out-of-plane displacement of the 1D buckling mode, as shown in Figure 18.15a, is the same as equation 18.7 for the ribbon. The wavelength is also given by equation 18.14 and is independent of the prestrains. For the plane-strain deformation pre (ε22 = 0), the amplitude is still given by equation 18.15. For the equi-biaxial

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.6 BUCKLING ANALYSIS OF STIFF THIN MEMBRANES ON COMPLIANT SUBSTRATES

505

L+∆L (a)

L+∆L

Si membrane Stretched PDMS slab Release the strain of PDMS L

(b)

L

(d)

(c)

λ1 λ1

2B

λ2 θ

λ2

θ

λ

λ

Figure 18.14. Schematic illustration of the process for fabricating two-dimensional (2D) wavy Si nanomembranes on a PDMS substrate. (a) Si membrane is bonded on the stretched PDMS. (b) Formation of 2D wavy patterns when PDMS is relaxed. (c ) Herringbone mode. (d ) Top-down view of the herringbone mode. The parameters are illustrated in (c ) and (d ), including the short wavelength λ, long wavelength λ2 , jogs wavelength λ1 , and amplitude B of the jogs in the plane of the film, and the jog angle θ. (Reprinted with permission from [20]. Copyright 2008 American Institute of Physics.) pre

pre

prestrains ε11 = ε22 = εpre , the amplitude becomes  A = hf 

εpre c −1 ε1D

(18.44)

3E s E f ) / c where ε1D = ( 4(1+ν is the critical strain for the 1D buckling mode subjected to f) equi-biaxial prestrains. 23

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

506

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

(c )

(b)

(a)

Figure 18.15. Schematic illustrations of different buckling modes. (a) 1D mode, (b) checkerboard mode from equation 18.45, and (c ) herringbone mode from equation 18.48 with A = 1.0µm, k1 = 0.309µm−1 , B = 10µm, k2 = 0.139µm−1 . (Reprinted with permission from [20]. Copyright 2008 American Institute of Physics.)

18.6.2 Checkerboard Buckling Mode The out-of-plane displacement of generalized checkerboard mode, as shown in Figure 18.15b, is given by w = A cos (k1 x1 ) cos (k2 x2 )

(18.45)

The checkerboard mode is the mode for k1 = k2 . For the equi-biaxial prestrains pre pre ε11 = ε22 = εpre , the minimization of total energy gives the wave numbers and amplitude as 1 1 k1 = k2 = √ 2 hf  A = hf 

8 (3 − νf ) (1 + νf )



3E s

1/3 (18.46)

Ef


εpre c εcheckerboard

 −1

(18.47)

3E s E f ) / c = ( 4(1+ν is the critical strain for the checkerboard buckling where εcheckerboard f) mode and is the same as that for the 1D buckling mode. 23

18.6.3 Herrington Buckling Mode The out-of-plane displacement w of the herringbone mode, as shown in Figure 18.15c, is periodic in both long wavelength direction x2 and its perpendicular direction x1 (also see Figures 18.14c and d). The contour line of constant w is

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.7 PRECISELY CONTROLLED BUCKLING OF STIFF THIN RIBBONS

507

sinusoidal in the x1 –x2 plane (Figures 18.14c and d) and therefore can be represented by x1 + B cos (k2 x2 ) = constant, where B is the amplitude of the sinusoidal line and k2 = 2π λ2 is the corresponding wave number. The displacement of the herringbone mode is also sinusoidal in the x1 direction (λ direction in Figure 18.14) and therefore can be represented by w = A cos {k1 [x1 + B cos (k2 x2 )]}

(18.48)

The minimization of total energy with respect to A, B, k1 , and k2 gives their governing equations, which are solved by the quasi-Newton and finite difference gradient method. These three different buckling modes, 1D, checkerboard, and herringbone, are studied for the Si film–PDMS substrate. The critical buckling strain is 0.0267% for the 1D and checkerboard modes and the numerical results give the same (0.0267%) for the herringbone mode. The prestrain is introduced by thermal expansion, which pre pre gives equi-biaxial compression (ε11 = ε22 = εpre ) in Si. Figure 18.16a shows the ratio of the total energy U total to U 0 versus the prestrain for three different buckling modes: 1D, checkerboard, and herringbone, where 2 is the energy for the unbuckled state. The herringbone mode U0 = E f h f (1 + νf ) εpre gives the lowest energy and is therefore energetically favorable mode in 2D buckling. Figures 18.16b, c, and d provide an explanation by giving the ratios of substrate strain energy U s , thin film bending energy U b , and membrane energy U m to U 0 . The film membrane energy of the herringbone mode is much lower than other two modes (see Figure 18.16d) although its substrate strain energy and film bending energy are slightly higher than their counterparts (Figures 18.16b and c). The herringbone mode significantly reduces the thin film membrane energy at the expense of slight increase of the thin film bending energy and substrate strain energy.

18.7 PRECISELY CONTROLLED BUCKLING OF STIFF THIN RIBBONS ON COMPLIANT SUBSTRATES The approach discussed in prior sections has the following limitations (i) The “wavy” Si ribbons are formed from spontaneous buckling with amplitudes and wavelengths determined by materials properties (e.g., moduli and thickness), without any direct control over the geometries; (ii) Although the range of acceptable strain is improved significantly (to ∼20%) compared to that of silicon itself (∼1%) the stretchability is still too small for certain applications. In order to control the buckle geometries and improve the stretchability, Sun et al. [12] used a mechanical strategy to create precisely controlled buckle geometries for nanoribbons of GaAs and Si, which combines lithographically patterned surface

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

508

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

1.0

0.4 1.0

0.8

0.8

Checkerboard

0.6

Herringbone

0.4 0.02

0.04

0.06

0.08

0.3

0.10

1D mode 0.4

Checkerboard

Herringbone

0.2

Checkerboard 1D mode

0.1

0.2 0.0 0.0

U U

U

U

0.6

1D mode

Herringbone

0.5

1.0

1.5

2.0

2.5

0.0 0.0

0.5

1.0

Prestrain (%)

2.0

2.5

(b)

(a) 0.20

1.0 0.8

0.15 Herringbone

0.6

0.10

U U

U U

1.5

Prestrain (%)

Checkerboard

1D mode

Checkerboard

0.4

1D mode

0.05

0.2 Herringbone

0.00 0.0

0.5

1.0

1.5

Prestrain (%)

(c)

2.0

2.5

0.0 0.0

0.5

1.0

1.5

2.0

2.5

Prestrain (%)

(d)

Figure 18.16. Ratios of energy in the buckled state to that in the unbuckled state U 0 versus the prestrain εpre for 1D, checkerboard, and herringbone modes. (a) Total energy U total in the Si film–PDMS substrate system, (b) strain energy U s in the PDMS substrate, (c ) bending energy U b in the Si film, and (d ) membrane energy U m in the Si film. (Reprinted with permission from [20]. Copyright 2008 American Institute of Physics.)

bonding chemistry and a buckling process similar to that reported in Khang et al. [11]. As illustrated in Figure 18.17, the photolithograph process to define the bonding chemistry is conducted on a stretched PDMS substrate subject to prestrain εpre = L L along the ribbon direction to form periodic interfacial patterns with activated sites where the chemical bonding occurs between ribbon (GaAs or Si) and PDMS substrate, as well as inactivated sites where there is only weak van der Waals interactions at the interface. The widths of activated and inactivated sites are denoted as W act and W in , respectively (Figure 18.17a). Thin ribbons oriented parallel to the prestrain direction are attached to the prestrained and patterned PDMS substrate (Figure 18.17b). The relaxation of the prestrain εpre in PDMS leads buckling of these ribbons due to the physical separation of the ribbons from the inactivated sites on the Win and PDMS (Figure 18.17c). The wavelength of the buckled structures is 2L 1 = 1+ε pre its amplitude A depends on the geometries of the interfacial patterns (W act and W in ) and the prestrain. Figure 18.17d shows a tilted-view scanning electron microscope

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.7 PRECISELY CONTROLLED BUCKLING OF STIFF THIN RIBBONS

509

(SEM) image of buckled GaAs ribbons on PDMS, in which εpre = 60%, W act = 10 µm, and W in = 400 µm. Details of the experimental procedures and observations were presented elsewhere [12]. Jiang et al. [16] developed a model to study the buckling behavior of such systems and to predict the maximum strain in the ribbons as a function of interfacial pattern. The buckling profile can be expressed as

w=

  

w1 =


π x1 1 , −L 1 < x1 < L 1 A 1 + cos 2 L1 L 1 < |x1 | < L 2 w2 = 0,

where A is the buckling amplitude, 2L 1 =

Win 1+εpre

(18.49)

is the buckling wavelength, and

2L 2 = + Wact is the sum of activated and inactivated regions after relaxation (Figure 18.17c). The minimization of total energy with respect to the buckling amplitude A gives Win 1+εpre

A= where εc =

h 2f π 2 12L 21

4 π





L 1 L 2 εpre − εc ,

for εpre > εc

(18.50)

is the critical strain for buckling, which is identical to the Euler h3 E

f f . buckling strain for a doubly clamped beam with length 2L1 and bending rigidity 12 The ribbon does not buckle when εpre < εc . Once εpre exceeds εc , the membrane strain in the ribbon is ε11 = −εc , i.e., the ribbon buckles and adjusts its buckling amplitude A such that the (compressive) membrane strain remains at εc . This analytical solution agrees very well with experiments as shown in Figure 18.18. The red lines are the profiles of the buckled GaAs ribbons given by equations 18.49 and 18.50 for different prestrain levels with the same layout of interfacial patterns W act = 10 µm and W in = 190 µm. The experimental images are also shown for comparison. Good agreement between analytical solutions and experiments are observed for both amplitude and wavelength, except for low prestrain (e.g., εpre = 11.3%). This discrepancy is due to the assumption that the ribbon buckles over the entire inactivated region, which may not hold at low prestrain. h 2f π 2 −6 for the buckling wavelength The critical strain εc = 12L 2 is on the order of 10 1 h ≈ 0.1 µm. The buckling amplitude A in 2L1 ≈ 200 µm and ribbon thickness f 

  equation 18.50 becomes A ≈ π4 L 1 L 2 εpre = π2 Win (Win + Wact ) εpre 1 + εpre , which is independent of the ribbon properties (e.g., thickness, Young’s modulus) and is completely determined by the layout of interfacial patterns (W in and W act ) and the prestrain. This generic conclusion suggests a broader application of this approach: ribbons made of any materials will form into almost the same buckled geometries (shown in Figure 18.17d) for the same interfacial patterns. Sun et al. [12] used this approach to obtain very similar buckled geometries for Si and GaAs. One of the most important issues in stretchable electronics is to reduce the strain level in electronics. Since the membrane strain ε11 is negligible (∼10−6 ), the

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

510

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

Wact

Win

Inactivated region Activated region PDMS L+∆L

(a)

Thin film

PDMS L+∆L (b)

A x1

2L1 2L2 L

(c)

10 µm

100 µm (d )

Figure 18.17. Processing steps for precisely controlled thin film buckling on elastomeric substrate. (a) Prestrained PDMS with periodic activated and inactivated patterns. L is the original length of PDMS and L is the extension. The widths of activated and inactivated sites are denoted as W act and W in , respectively. (b) A thin film parallel to the prestrain direction is attached to the prestrained and patterned PDMS substrate. (c ) The relaxation of the prestrain εpre in PDMS leads to buckles of thin film. The wavelength of the buckled film is 2L1 , and its amplitude is A. 2L2 is the sum of activated and inactivated regions after relaxation. (d ) SEM image of buckled GaAs thin films formed using the previous procedures. The inset shows the GaAs–PDMS substrate interface. (Reprinted with permission from [16]. Copyright 2007 American Institute of Physics.)

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

18.7 PRECISELY CONTROLLED BUCKLING OF STIFF THIN RIBBONS

511

Wact = 10 mm Win = 190 mm Prestrain εpre = 11.3%

25.5%

33.7%

56.0% 100 µm Figure 18.18. Buckled GaAs thin films on patterned PDMS substrate with W act = 10 µm and W in = 190 µm for different prestrain levels, 11.3%, 25.5%, 33.7%, and 56.0% (from top to bottom). The dotted lines are the profiles of the buckled GaAs thin film predicted by the analytical solution. (Reprinted with permission from [16]. Copyright 2007 American Institute of Physics.)

maximum strain in the ribbon is the bending strain that results from the ribbon cur vature d2 w dx12 . This gives εmax =


2  hfπ  dw hf = 2 L 1 L 2 εpre max 2 2 dx1 L1

(18.51)

For the activated region W act much smaller than the inactivated region W in , equa√ tion 18.51 can be approximated by εmax ≈ hLf π1 εpre . It is much smaller than the prestrain for ribbon thickness h (∼0.1 µm) much less than the wavelength 2L1 (∼200 µm). For a 0.3-µm-thin GaAs ribbon buckled on a patterned PDMS substrate with W act = 10 µm, W in = 400 µm, and εpre = 60%, the maximum strain is only 0.6%, two orders of magnitude smaller than the 60% prestrain. Therefore, the precisely controlled buckling can significantly reduce the maximum strain in thin ribbon and improve the system stretchability. The buckled ribbons (such as GaAs nanoribbons) need protection in practical applications. This is achieved by embedding the buckled ribbons in PDMS (i.e., ribbons sandwiched between PDMS) via the casting and curing of a prepolymer. The prepolymer is fluid that can flow and fill the air gaps between buckled thin film and PDMS substrate [12]. Such a procedure does not change the wavelength (2L1 ) nor the amplitude (A) because the liquid prepolymer does not impose any deformation to the ribbon. Jiang et al. [18] obtained the analytical solution for the postbuckling of such systems.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

512

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

18.8 CONCLUDING REMARKS We have reviewed the theories, numerical and experimental studies of the buckling of stiff thin ribbons or membranes on compliant substrates. Both the analytical solutions and numerical results show the strain-dependent buckling wavelength, which agree well with the experiments without any parameter fitting. The strains are accommodated through changes in the amplitudes and wavelengths of buckled geometries. Once the ribbon buckles, its membrane strain remains a constant (the critical strain for buckling), and peak strain (due to bending) increases very slowly with the applied strain. The effect of free edge near the ribbon end is studied numerically, while the analytical solution is obtained for the ribbon of finite width and spacing. For thin membranes, buckling takes the form of herringbone pattern. Finally, the precisely controlled buckling of thin ribbons provides a simple method to significantly increase the system stretchability/compressibility. These conclusions and the detailed analyses are important for the many envisioned applications for buckled thin film–substrate systems. ACKNOWLEDGMENTS We acknowledge the support from the National Science Foundation under grant DMI-0328162, the U.S. Department of Energy, Division of Materials Sciences under Award No. DEFG02-91ER45439, through the Frederick Seitz MRL and Center for Microanalysis of Materials at the University of Illinois at Urbana-Champaign. HJ acknowledges the support from NSF CMMI-0700440.

REFERENCES 1. Bowden, N., Brittain, S., Evans, A. G., Hutchinson, J. W., and Whitesides, G. M. (1998) Spontaneous formation ordered structures in thin films of metals supported on an elastomeric polymer. Nature 393, 146–149. 2. Huang, R. and Suo, Z. (2002) Instability of a compressed elastic film on a viscous layer. Int. J. Solids Struct. 39, 1791–1802. 3. Harrison, C., Stafford, C. M., Zhang, W. H., and Karim, A. (2004) Sinusoidal phase grating created by tunably buckled surface. Appl. Phys. Lett. 85, 4016–4018. 4. Huang, R. (2005) Kinetic wrinkling of an elastic film on a viscoelastic substrate. J. Mech. Phys. Solids 53, 63–89. 5. Chen, X. and Hutchinson, J. W. (2004) Herringbone buckling patterns of compressed thin films on compliant substrates. J. Appl. Mech. 71, 597–603. 6. Lacour, S. P., Jones, J., Wagner, S., Li, T., and Suo, Z. (2005) Stretchable interconnects for elastic electronic surfaces. Proc. IEEE 93, 1459–1467. 7. Lacour, S. P., Wagner, S., Narayan, R. J., Li, T., and Suo, Z. (2006) Stiff subcircuit islands of diamondlike carbon for stretchable electronics. J. Appl. Phys. 100, 6. 8. Stafford, C. M., Harrison, C., Beers, K. L., Karim, A., Amis, E. J., Vanlandingham, M. R., Kim, H. C., Volksen, W., et al. (2004) A buckling-based metrology for measure the elastic moduli of polymeric thin films. Nat. Mater. 3, 545–550.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

15:32

Printer Name: Yet to Come

REFERENCES

513

9. Stafford, C. M., Vogt, B. D., Harrison, C., Julthongpiput, D., and Huang, R. (2006) Elastic moduli of ultrathin amorphous polymer films. Macromolecules 39, 5095–5099. 10. Huang, Z. Y., Hong, W. and Suo, Z. (2005) Nonlinear analyses of wrinkles in a film bonded to a compliant substrate. J. Mech. Phys. Solids 53, 2101–2118. 11. Khang, D. Y., Jiang, H. Q., Huang, Y., and Rogers, J. A. (2006) A stretchable form of single-crystal silicon for high-performance electronics on rubber substrate. Science 311, 208–212. 12. Sun, Y., Choi, W. M., Jiang, H., Huang, Y., and Rogers, J. A. (2006) Controlled buckling of semiconductor nano ribbons for stretchable electronics. Nat. Nanotechnol. 1, 201–207. 13. Sun, Y., Kumar, V., Adesida, I., and Rogers, J. A. (2006) Buckled and wavy ribbons of GaAs for high-performance electronics on elastomeric substrate. Adv. Mater. 18, 2857–2862. 14. Choi, W. M., Song, J., Khang, D. Y., Jiang, H., Huang, Y., and Rogers, J. A. (2007) Biaxially stretchable “wavy” silicon nanomembranes. Nano Lett. 7, 1655–1663. 15. Jiang, H., Khang, D. Y., Song, J., Sun, Y., Huang, Y., and Rogers, J. A. (2007) Finite deformation mechanics in buckled thin films on compliant supports. Proc. Natl. Acad. Sci. USA 104, 15607–15612. 16. Jiang, H., Sun, Y., Rogers, J. A., and Huang, Y. (2007) Mechanics of precised controlled thin film buckling on elastomeric substrate. Appl. Phys. Lett. 90, 133119. 17. Jiang, H., Khang, D. Y., Fei, H., Kim, H., Huang, Y., Xiao, J., and Rogers, J.A. (2008) Finite width effect of thin films buckling on compliant substrate: experimental and theoretical studies. J. Mech. Phys. Solids. 56, 2585–2598. 18. Jiang, H., Sun, Y., Rogers, J. A., and Huang, Y. (2008) Post-buckling analysis for the precisely controlled buckling of thin film encapsulated by elastomeric substrates. Int. J. Solids Struct. 45, 2014–2023. 19. Koh, C. T., Liu, Z. J., Khang, D. Y., Song, J., Lu, C., Huang, Y., Rogers, J. A., and Koh, C. G. (2007) Edge effects in buckled thin films on elastomeric substrate. Appl. Phys. Lett. 91, 133113. 20. Song, J., Jiang, H., Choi, W. M., Khang, D. Y., Huang, Y., and Rogers, J.A. (2008) An analytical study of two-dimensional buckling of thin films on compliant substrates. J. Appl. Phys. 103, 014303. 21. Song, J., Jiang, H., Liu, Z. J., Khang, D. Y., Huang, Y., Rogers, J. A., Lu, C., and Koh, C. G. (2008) Buckling of stiff thin film on a compliant substrate in large deformation. Int. J. Solids Struct. 45, 3107–3121. 22. Wagner, S., Lacour, S. P., Jones, J., Hsu, P. H. I., Sturm, J. C., Li, T., and Suo, Z. (2004) Electronic skin: architecture and components. Physica E 25, 326–334. 23. Fu, Y. Q., Sanjabi, S., Barber, Z. H., Clyne, T. W., Huang, W. M., Cai, M., Luo, J. K., Flewitt, A. J., et al. (2006) Evolution of surface morphology in TiNiCu shape memory thin films. Appl. Phys. Lett. 89, 3. 24. Efimenko, K., Rackaitis, M., Manias, E., Vaziri, A., Mahadevan, L., and Genzer, J. (2005) Nested self-similar wrinkling patterns in skins. Nat. Mater. 4, 293–297. 25. Harris, A. K., Wild, P., and Stopak, D. (1980) Silicon-rubber substrata-new wrinkle in the study of cell locomotion. Science 208, 177–179. 26. Jiang, X. Y., Takayama, S., Qian, X. P., Ostuni, E., Wu, H. K., Bowden, N., LeDuc, P., Ingber, D. E., et al. (2002) Controlling mammalian cell spreading and cytoskeletal arrangement with conveniently fabricated continuous wary features on poly(dimethylsiloxane). Langmuir 18, 3273–3280.

P1: OTA/XYZ P2: ABC c18 JWBK252/Lee September 20, 2008

514

15:32

Printer Name: Yet to Come

MECHANICS OF STRETCHABLE SILICON FILMS ON ELASTOMERIC SUBSTRATES

27. Teixeira, A. I., Abrams, G. A., Bertics, P. J., Murphy, C. J., and Nealey, P. F. (2003) Epithelial contact guidance on well-defined micro- and nanostructured substrate. J. Cell Sci. 116, 1881–1892. 28. Wilder, E. A., Guo, S., Lin-Gibson, S., Fasolka, M. J., and Stafford, C. M. (2006) Measuring the modulus of soft polymer networks via a buckling-based metrology. Macromolecules 39, 4138–4143. 29. Bowden, N., Huck, W. T. S., Paul, K. E., and Whitesides, G. M. (1999) The controlled formation of ordered, sinusoidal structures by plasma oxidation of an elastomeric polymer. Appl. Phys. Lett. 75, 2557–2559. 30. Huck, W. T. S., Bowden, N., Onck, P., Pardoen, T., Hutchinson, J. W., and Whitesides, G. M. (2000) Ordering of spontaneously formed buckles on planar surfaces. Langmuir 16, 3497–3501. 31. Sharp, J. S. and Jones, R. A. L. (2002) Micro-buckling as a route towards surface patterning. Adv. Mater. 14, 799–802. 32. Yoo, P. J., Suh, K. Y., Park, S. Y., and Lee, H. H. (2002) Physical self-assembly of microstructures by anisotropic buckling. Adv. Mater. 14, 1383–1387. 33. Schmid, H., Wolf, H., Allenspach, R., Riel, H., Karg, S., Michel, B., and Delamarche, E. (2003) Preparation of metallic films on elastomeric stamps and their application for contact processing and contact printing. Adv. Funct. Mater. 13, 145–153. 34. Moon, M.-W., Lee, S. H., Sun, J.-Y., Oh, K. H., Vaziri, A., and Hutchinson, J. W. (2007) Wrinkled hard skins on polymers created by focused ion beam. Proc. Natl. Acad. Sci. USA 104, 1130–1133. 35. Crawford, G. P. (2005) Flexible Flat Panel Display Technology, John Wiley & Sons, New York. 36. Jin, H. C., Abelson, J. R., Erhardt, M. K., and Nuzzo, R. G. (2004) Soft lithographic fabrication of an image sensor array on a curved substrate. J. Vac. Sci. Technol. B 22, 2548–2551. 37. Lumelsky, V. J., Shur, M. S., and Wagner, S. (2001) Sensitive skin. IEEE Sens. J. 1, 41–51. 38. Someya, T., Sekitani, T., Iba, S., Kato, Y., Kawaguchi, H., and Sakurai, T. (2004) A largearea, flexible pressure sensor matrix with organic field-effect transistors for artificial skin applications. Proc. Natl. Acad. Sci. USA 101, 9966–9970. 39. Nathan, A., Park, B., Sazonov, A., Tao, S., Chan, I., Servati, P., Karim, K., Charania, T., et al. (2000) Amorphous silicon detector and thin film transistor technology for large-area imaging of x-rays. Microelectron. J. 31, 883–891. 40. INSPEC (1988) Properties of Silicon, Institution of Electrical Engineers, New York. 41. Volynskii, A. L., Bazhenov, S., Lebedeva, O. V., and Bakeev, N. F. (2000) Mechanical buckling instability of thin coatings deposited on soft polymer substrates. J. Mater. Sci. 35, 547–554. 42. Symon, K. (1971) Mechanics, Addison-Wesley, Reading, MA. 43. Abramowitz, M. and Stegun, I. A. (1972) Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, Dover, New York.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19 MULTISCALE FABRICATION OF PLASMONIC STRUCTURES Joel Henzie, Min H. Lee, and Teri W. Odom

19.1 INTRODUCTION Plasmonics is a field that encompasses the science and applications of noble metal structures that can guide and manipulate light at the nanometer scale. This area was so-named because it is an analog of photonics; instead of controlling light with dielectric materials, light is controlled by metal structures via plasmons, collective free electron oscillations [1]. Beyond these semantics, one of the early driving forces for research in plasmonics was the discovery that highly confined electromagnetic fields, in the form of propagating surface plasmons, could provide an alternative approach for miniaturizing optical devices while carrying electrical signals at optical frequencies [2]. Such properties are possible because plasmons, unlike photons, are not subjected to the classical diffraction limit (∼λ/2) [3]. Not surprisingly, the design of plasmonic devices was inspired by the fabrication techniques and device architectures of photonic ones, and plasmonic devices could also perform similar functions, but at physical dimensions approaching the length scales of the modern transistor [4, 5]. Prototype devices such as waveguides [6, 7], interferometers [8], demultiplexers [9], and plasmonic bandgap crystals [10] have been demonstrated (Figures 19.1a–c) [15].

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright 

515

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

516

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

(a)

(b)

(c)

(d)

(e)

(f)

Figure 19.1. Plasmonic nanostructures. (a) Gold stripe and nanoparticle waveguides. (Reprinted with permission from [7]. Copyright 2001 APS and [1] Copyright 2001 Wiley.) (b) Ring resonator. (Reprinted with permission from [11]. Copyright 2006 Nature Publishing.) (c ) Subwavelength bull’s-eye structure. (Reprinted with permission from [12]. Copyright 2002 Science.) (d ) Silver cubes. (Reprinted with permission from [13]. Copyright 2004 Wiley.) (e) Silver nanowires. (Reprinted with permission from [14]. Copyright 2001 RSC.) (f ) Gold stars (courtesy of J. Hafner).

Rapid progress in plasmonics can be attributed to a combination of breakthroughs over the past decade, including: 1. Imaging tools. Near-field scanning optical microscopy enabled direct imaging of surface plasmon waves [16], and optical microscopes outfitted with dark field scattering capabilities allowed localized surface plasmon resonances to be correlated with individual metal nanoparticles [17]. 2. Theoretical tools. Optimized electrodynamics calculation methods as well as improved computational resources have made detailed theoretical descriptions of the observed optical properties possible. 3. Nanomaterials methods. Advances in synthetic and fabricated approaches to produce nanostructures have allowed control over the shape and size of noble metal (Au, Ag, Cu) structures (Figures 19.1d–f ). In particular, the development of nanofabrication tools capable of sub-100-nm resolution has played a key role in enabling groundbreaking discoveries in plasmonics: the demonstration of enhanced optical transmission through subwavelength hole arrays [18], the collimation of light through a subwavelength bull’s-eye structure [12],

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.1 INTRODUCTION

517

negative permeability and refraction at visible wavelengths [19, 20], and second harmonic generation from magnetic metamaterials [21]. This chapter will focus on how new types of nanofabrication tools, based on soft lithography, can generate a wide range of plasmonic structures with exceptional optical properties. There are two important features of these tools: (1) they provide a scalable and inexpensive approach to create arrays of complex metal structures (nanoholes and nanoparticles), and (2) they expand the types of plasmonic metamaterials that are possible because the metallic building blocks can now be organized over multiple length scales and over macroscale areas.

19.1.1 Brief Primer on Surface Plasmons The optical properties of metals are determined in part by the resonant interaction between light and their surface free electrons at a metal–dielectric interface. These collective electron oscillations or charge density waves—surface plasmon polaritons (SPPs)—exist as propagating waves on planar metal films (Figure 19.2a). The amplitude of the SPPs extends farther in the dielectric region compared to the metal region. In order to excite SPPs on a surface using free space light, additional momentum must be provided either by patterning a grating structure on the film or by evanescent coupling of light into the metal [22]. The propagation distance of SPPs depends primarily on the absorption of the metal and the thickness and surface roughness of the film; for SPPs on Au films excited using 633-nm light, this decay length is around 10 µm [23, 24]. Metal particles (<250 nm in diameter) also interact strongly with light through plasmons that are confined within about 10 nm of the particle surface (Figure 19.2b); these resonant optical fields are called localized surface plasmons (LSPs). LSP resonances are highly sensitive to the size [25], shape [26], and dielectric environment of the metal particles and can be tuned from ultraviolet (UV) to near-infrared (NIR) wavelengths [27]. Spherical particles with sizes less than 50 nm support single LSP resonances that are dipolar in character, and their optical properties can be described reasonably well by the lowest order term in Mie theory [28]. In addition, slightly larger anisotropic particles (100-nm-diameter Au pyramids) exhibit only a dipole

(a) z

(b)

Dielectric

E

E

εd

Hy

x εm

z

Metal

Figure 19.2. Different types of surface plasmons. (a) Surface plasmon polaritons at a metal–dielectric interface. (b) Localized surface plasmons on a metal nanoparticle excited by free space light.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

518

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

resonance in contrast to pyramids with sizes of 250 nm or larger [29]. Rod-shaped Au nanoparticles can support both transverse and longitudinal modes depending on polarization [30, 31]. Larger metallic particles (overall sizes >100 nm) with anisotropic shapes can exhibit multiple LSP resonances [32, 33] that correspond to higher order plasmon modes [34]. These multipolar excitations depend on the direction of the wave vector as well as the polarization vector [35]; thus, certain excitation angles can allow selected resonances to be more pronounced [36].

19.1.2 Conventional Methods to Plasmonic Structures Chemical synthesis has been the primary means to grow a wide variety of metal nanoparticle shapes (Figure 19.1d–f ), including prisms, stars, rods, boxes, and cages, because reaction conditions such as temperature, surfactants, and precursors can be independently controlled [37]. Although solution-based methods are scalable, the large distribution in nanoparticle shape and size within a single reaction vessel is a challenge. Metal particles can also be assembled into two-dimensional (2D) and three-dimensional (3D) lattices using a variety of techniques; both the interparticle distances and the geometry of the lattices can be used to tune the optical properties of the assemblies [37–40]. Nanofabrication methods offer an alternative strategy to organize plasmonic structures into arrays. Direct-write techniques such as electron-beam (e-beam) lithography can fabricate linear and 2D arrays of nanoparticles with different spacings [38, 39], which typically need to be on the order of the size of the particle or less (<200 nm) for efficient dipolar coupling [40]. Focused ion beam (FIB) milling has been the most common method to fabricate hole and slot structures in optically thick metal films [18]. This method can drill all the way or partially through the film and can control the diameter and spacing of the structures with reasonable precision (ca. 100 nm). Free-standing suspended films have also been fabricated by FIB and reactive ion etching, but the generation of multilayered metal films is difficult and laborious [41]. Although serial lithographic methods have been instrumental for fabricating prototype structures whose unique properties opened the field of plasmonics, the patterned areas were small (hundreds of square micrometers) and access to multiscale structures is limited. Soft lithography can meet these challenges.

19.2 SOFT LITHOGRAPHY AND METAL NANOSTRUCTURES Soft lithography is a catchall term for fabrication techniques that use wafer scale (several square inches) poly(dimethylsiloxane) (PDMS) stamps or masks to transfer patterns from one material or surface to another [42]. To extend these capabilities to produce sub-100-nm features, the mechanical properties of the PDMS were improved [43], and composite PDMS stamps composed of a thin, stiff layer of hard PDMS (h-PDMS) supported by a thick slab of 184 PDMS were developed [44–49].

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.2 SOFT LITHOGRAPHY AND METAL NANOSTRUCTURES

519

Soft lithographic methods offer three important advantages over the serial ones described in Section 19.1.2 to produce plasmonic structures. 1. Parallelism. PDMS molded against a master forms a transparent, elastomeric mask that can be used to create arrays of structures simultaneously. 2. Simplicity. Only widely available microfabrication techniques are required, including photolithography, wet chemical etching, e-beam deposition, reactive ion etching, and microtoming [50]. 3. Flexibility. Patterns can be generated with various symmetries and spacings, out of multiple materials, and with exquisite control over the thickness of each layer. The most widely used soft lithographic approach to fabricate metal nanostructures starting from metal films involves microcontact printing followed by chemical etching [51, 52]. Here, a PDMS stamp is “inked” with an alkanethiol solution and then brought into conformal contact with a noble metal surface; the molecules are transferred from the stamp to the metal and self-assemble into a monolayer (SAM) with the same pattern as the stamp. These patterned SAMs can function as etch masks for the underlying metal film, and thus after subjected to a chemical etch, the protected metal patterns have the same characteristics as the PDMS stamp. A range of metal nanostructures have been generated, from arrays of 90-nm lines that acted as linear polarizers to slots in metal films that could be used as frequency selective surfaces [53]. One variation on this printing and etching method is decal transfer lithography, where the bas relief patterns on a PDMS stamp are first chemically bonded to the metal surface [54]. When the PDMS stamp is removed from the surface, remnants of PDMS remain on the substrate (and, as a drawback, the pattern on the stamp is destroyed). These PDMS surface patterns can then act like etch masks to produce metal patterns identical to those on the original PDMS stamp. Another strategy to produce metal nanostructures is to deposit the metal directly on patterned PDMS stamps or polymeric structures using sputtering, e-beam, or thermal evaporation. In nanotransfer printing, a thin layer of Au is deposited on the patterned PDMS using line-of-sight deposition and then placed on a chemically modified surface [55]. When the stamp is removed, only the gold portions in contact with the surface remain. If, however, the deposited metal is allowed to accumulate on the sidewalls of the pattern (through a combination of sample rotation and angled deposition), films that follow the complex 3D topography of the stamp can be fabricated and then released to yield free-standing structures [56]. High aspect ratio metal nanostructures can also be produced if the metal patterns on the PDMS mold are sectioned with a microtome tool; the optical properties of such structures depend strongly on the polarization of the incident light [57]. Composite structures can also be fabricated by depositing a thin (50 nm) Au layer on arrays of 500-nm-diameter wells molded in a polymeric substrate. This hole–particle system exhibits interesting properties in the NIR because the LSPs of the nanoparticles in the wells interact with the SPPs of the hole film on the top surface [58].

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

520

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

19.3 A PLATFORM FOR MULTISCALE PATTERNING The remaining sections of this chapter will focus on a new variation of soft lithography—soft interference lithography (SIL)—that can pattern plasmonic structures such as nanoscale holes and particles over multiple length scales. We have developed a suite of nanofabrication tools that use PDMS masks to generate freestanding nanoholes in films and nanoparticles with arbitrary spacings (ranging from a0 = 400 nm to 25 µm) over macroscale areas (several square inches). The most pressing reason that multiscale patterning is necessary for designing and understanding new plasmonic systems is that surface plasmons can interact over multiple length scales: tens of nanometers for LSPs and tens of micrometers for SPPs. The interplay between LSPs and SPPs—and SPPs with other SPPs—has not been studied extensively, however, because conventional patterning tools cannot easily produce the necessary metal patterns and structures. 19.3.1 Soft Interference Lithography: Patterns on a Nanoscale Pitch SIL uses nanoscale patterns generated by interference lithography (IL) as high quality masters for soft lithography. Figure 19.3a shows an IL master patterned with Si posts (diameter d = 100 nm, pitch a0 = 400 nm, height h = 400 nm). Hundreds of SIL photomasks can be replicated from one master pattern without exhibiting observable defects. To produce patterns in photoresist identical in size and shape to the IL master, the SIL photomask is first placed in conformal contact with a positive photoresist and then exposed with broadband UV light. If the photoresist is developed at this point, posts with the same lateral dimensions as the IL master (called infinite arrays) are produced. If the photoresist is exposed again through a Cr photomask patterned with microscale (1–10 µm) features and then developed, patches of post arrays (called finite arrays) are formed. The overall size of a SIL master is typically large (>2 in2 ) and can be aligned with the top edge of the Cr mask to achieve misalignments less than 5% in rotation; the use of a mask aligner reduces misalignment to less than 1%. 19.3.2 Phase-Shifting Photolithography: Patterns on a Microscale Pitch Standard phase-shifting photolithography (PSP) is an edge photolithography technique that produces narrow features in photoresist at the edges of the patterns in a PDMS mask. Typically, these PSP masks are fabricated by replicating photoresist masters (h = 400 nm) with microscale (a0 = 1–25 µm, d = 1–25 µm) patterns. Edge features as narrow as 50 nm can be produced in positive-tone photoresist by exposing UV light through a composite PDMS mask. When two exposures are made through a mask patterned with microscale lines (the mask is rotated after the first exposure and then exposed again before development), photoresist posts are formed at the intersection of the line patterns [60]. If the second exposure is through the same mask

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.3 A PLATFORM FOR MULTISCALE PATTERNING

521

Figure 19.3. Masters with macroscale dimensions and patterned with nanoscale features. (a) Silicon master made by interference lithography. The silicon posts are 400 nm tall, 100 nm in diameter, and on a square array of a0 = 400 nm. (Reprinted with permission from [59]. Copyright 2007 Nature Publishing.) (b) Photoresist master made by phase-shifting photolithography. The photoresist posts are on a microscale pitch and can be (c ) circular, 400 nm tall, and 250 nm in diameter or (d ) anisotropic, 400 nm tall, and 90 nm × 900 nm in size.

and rotated by 90◦ , square arrays of circular dots are formed; when rotated by 15◦ , more complex arrays of anisotropic dots are formed (the smallest distance between dots is determined by the periodicity of the line mask) (Figures 19.3b–d) [61]. 19.3.3 PEEL: Transferring Photoresist Patterns to Plasmonic Materials PEEL is a nanopatterning procedure that transfers patterned features in photoresist into free-standing, functional materials (Figure 19.4). This method combines Phaseshifting photolithography, Etching, Electron-beam deposition, and Liftoff [23, 62]. First, PSP is used to define the photoresist pattern on a Si (1 0 0) surface and then deposition and liftoff transfers the resist pattern into holes in a Cr film. The film of Cr holes acts as an etch mask to form pyramidal pits in the Si (1 0 0) surface beneath the holes and then as an e-beam deposition mask to produce perforated metal films over pyramidal particles. The metal nanohole film can be released from the surface by etching the Cr layer and placing the film on a glass substrate. The nanopyramids formed within the Si pits can be embedded in other materials such as PDMS [29], or released from the Si template, suspended in solution, and dispersed on a substrate [48].

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

522

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

PSP; electron-beam Cr

Liftoff resist

Etch Si

Electron-beam Au

Etch Cr

Figure 19.4. PEEL procedure to transfer patterns in photoresist to functional materials such as gold or silicon. Arrays of nanoscale holes and arrays of pyramidal nanoparticles are fabricated simultaneously.

19.4 SUBWAVELENGTH ARRAYS OF NANOHOLES: PLASMONIC MATERIALS In the early 20th century, the intensity of transmission through subwavelength apertures in opaque metal films was expected to be small, proportional to (r/λ)4 based on geometric optics [63]. The discovery of enhanced optical transmission through subwavelength hole arrays exceeded these predictions by orders of magnitude [18], and moreover, revealed that the wavelengths of highest transmission could be tuned by changing the material of the film, the hole size, and the array spacing and geometry [64]. Surface plasmons play a major role in this enhanced transmission. Light incident on the hole array excites surface plasmons on one side of the film, which then tunnel through the holes and/or increase the efficiency of light transmitted through the holes. Then, the surface plasmons that emerge at the opposite side of the film scatter from the hole array structure and are converted into free-space light [65].

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.4 SUBWAVELENGTH ARRAYS OF NANOHOLES: PLASMONIC MATERIALS

523

We define subwavelength nanohole arrays as metal films perforated by holes whose diameter and spacing are smaller than the wavelength of incident and transmitted light. Arrays of such metal nanostructures can also be considered as plasmonic metamaterials because their effective materials parameters (ε, permittivity and µ, permeability) are determined by geometry and not by the bulk properties of the materials; that is, the subwavelength nanohole arrays act as an effective medium at optical frequencies [66, 67].

19.4.1 Infinite Arrays of Nanoholes To fabricate subwavelength nanohole arrays using SIL and PEEL, we first exposed a thin (ca. 200 nm) layer of photoresist through a SIL photomask (Section 19.3.1) and then developed it to produce an array of photoresist posts with the same lateral sizes and array structure as the master. Next, we transferred the nanopatterns into Au using PEEL to generate macroscale (>2 in2 ) areas of films perforated with arrays of 100-nm-diameter circular holes separated center to center by 400 nm (infinite arrays). Finally, the films were placed on a glass substrate (refractive index n = 1.523) (Figure 19.5a). We characterized the optical properties of the nanohole arrays by illuminating the films under normal incidence with collimated white light and collecting the transmitted light through a microscope objective coupled to a spectrometer. When the incident light is normal to a square array of nanoholes of periodicity a0

Figure 19.5. Macroscale areas of gold films perforated with 100-nm holes. (a) Optical micrograph of a gold nanohole array supported on glass. (b) Zero-order transmission spectra of an infinite Au nanohole array. The gold film was 50 nm thick. The surface plasmon Bloch modes are labeled as (i , j ) at the relevant metal–dielectric interface. (c ) Zero-order transmission spectra of a patch Au nanohole array. The patches are separated by 4.5 µm. (Reproduced with permission from [59]. Copyright 2007 Nature Publishing.)

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

524

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

in a gold film, an approximate equation for the SPP Bloch waves (λSPP ) is given by [68] a0



λSPP =
i2 + j2

εAu εd εAu + εd

(19.1)

where i and j are integers that define the particular order of the Bloch modes, εAu is the dielectric constant of the Au, and εd is the dielectric constant of the surrounding dielectric material (air or glass). Figure 19.5b indicates that infinite Au nanohole array exhibited peaks character(1,1) istic of SPPs at the Au–glass interface of λ(1,0) SPP = 690 nm and λSPP = 588 nm, where the subscripts (1,0) and (1,1) correspond to the (i, j) pairs. The minimum in the spectra λ = 659 nm can be associated with a Wood’s anomaly (light diffracted parallel to the surface) [69], and the peak at 500 nm is the bulk plasmon resonance of an Au film. Compared to the calculated transmission of hole arrays based only on geometry, the infinite Au nanohole arrays exhibited enhanced optical transmission factors as large as 11, providing evidence that SIL can produce hole arrays of optical quality similar to those fabricated by FIB milling. Figure 19.6 shows how the properties of the Au infinite subwavelength nanohole arrays can be tuned by changing the refractive index of the top surface from air (n = 1) to different refractive index oils (n = 1.5–1.7). As expected, the SPP peaks of the infinite Au nanohole arrays (a)

(b)

Figure 19.6. Refractive index sensing using infinite Au nanohole arrays. (a) SEM image of a portion of the infinite Au nanohole array. (b) Zero-order transmission of infinite Au hole arrays on glass in the presence of higher index immersion oils (n = 1.50–1.70). (Reproduced with permission from [59]. Copyright 2007 Nature Publishing.)

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.4 SUBWAVELENGTH ARRAYS OF NANOHOLES: PLASMONIC MATERIALS

525

redshifted with increased n, but the bulk plasmon remained unchanged. Numerical calculations have revealed that at n = 1.7, the peak at λ = 563 nm is a localized resonance, the peak at λ = 620 nm is the (1,1) Au–oil resonance, the sharp peak at λ = 681 nm arises, in part, from first-order diffraction, and the peak around 800 nm can mostly be attributed to the (1,0) Au–oil resonance [70]. 19.4.2 Finite Arrays (Patches) of Nanoholes Light transmitted through subwavelength apertures should diffract in all directions [63]. The relationship between the enhanced optical transmission of light and its spatial output, however, suggests that light traveling through subwavelength holes can be manipulated. One striking demonstration of this effect is the case of a subwavelength bull’s-eye structure, where a series of circular gratings around a single 250-nm hole confined the light to the center of the structure and collimated it out to the far field (Figure 19.1c) [12]. We designed microscale patches (overall sizes ranging from d = 2.5–10 µm) of 100-nm holes to test how light emerged from these structures. Interestingly, we discovered that light emerged most strongly from the center of the patches and appeared to focus to a point; the wavelength-dependent intensity of these spots also correlated with the plasmon resonances of the finite arrays of nanoholes [59]. This result has important implications because it suggests that plasmons—an inherently near-field phenomena—can strongly affect the far-field transmission patterns [12]. Figure 19.7a shows that finite-sized nanohole arrays patterned into circular regions (d = 2.5 µm, a0 = 4.5 µm) of patches contained approximately 30 nanoholes. Because the size and pitch of the circles on the Cr mask were incommensurate with the photoresist post lattice (400 nm), neighboring patches had slightly different configurations of holes. The spectra acquired for single patches indicated, however, that these small structural variations did not noticeably affect the optical properties. Surprisingly, the nanohole patch spectra exhibited a decrease in the relative intensity of the (1,0) resonance (Figure 19.5c), indicating that coupling between the nearest-neighbor holes was reduced, and the width of this peak was significantly narrower than that of the infinite array (full width at half maximum, FWHM = 19.3 vs. 34.1 nm). A new, narrow peak also emerged at λ = 660 nm with a FWHM of 16.8 nm. To test whether this narrow peak could be attributed to either patch–patch coupling or the finite area of the nanohole array, we fabricated arrays of patches with different sizes (d = 2.5, 5, and 10 µm diameters) and separated by 15 µm (edge to edge). This distance is larger than the decay length of the Au plasmon at optical wavelengths (ca. 10 µm), and hence the neighboring patches should not couple to each other. Normal incidence transmission spectra for all isolated patches did not reveal any new resonances unlike the closely spaced patches, although the (1,0) resonances were redshifted slightly for the 2.5- and 5-µm patches compared to the infinite array. We found that the spectra of the patches began to resemble the infinite nanohole array as the number of holes increased, and that 10-µm patches with approximately 400 holes exhibited features nearly identical to those of the infinite array [59].

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

526

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

(a)

(b)

Figure 19.7. Refractive index sensing using patches of Au nanohole arrays. (a) SEM image of a portion of an array of 2.5-µm Au patches of nanoholes separated by 4.5 µm. (b) Zeroorder transmission of Au patches on glass in the presence of higher index immersion oils (n = 1.50–1.70). (Reproduced with permission from [59]. Copyright 2007 Nature Publishing.)

Also, similar to the infinite arrays, changing the refractive index of the top surface from air (n = 1) to different index oils caused all the SPP peaks to shift to longer wavelengths. Noticeably, several peaks remained very narrow over the range of n tested (Figure 19.7b); for example, the peak that shifted from λ = 690 nm (n = 1.5) to λ = 756 nm (n = 1.7) had an average FWHM of 14.8 ± 0.6 nm, which is one of the narrowest spectral widths reported for a surface plasmon resonance.

19.5 MICROSCALE ARRAYS OF NANOSCALE HOLES Although most work to date has focused on subwavelength hole arrays, where both the size of the holes and the spacing of the array are subwavelength in scale, nanoscale holes patterned with microscale spacings (microscale arrays) are becoming important for understanding in detail the optical properties of nanostructured metallic surfaces [23, 62]. These structures—whose hole–hole spacing is greater than the wavelength of incident light—show unique spectral characteristics compared to subwavelength hole arrays because of high order Bragg coupling. Furthermore, anisotropic nanoholes patterned into different geometries display striking, tunable polarization-dependent colors.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.5 MICROSCALE ARRAYS OF NANOSCALE HOLES

(a)

(c)

(b)

(d)

527

λ

Figure 19.8. Microscale arrays of nanoscale holes. Nanohole arrays in 170-nm-thick Au films perforated with (a) circular and (b) slitlike holes. Zero-order transmittance of arrays of (c ) 250-nm circular holes and (d ) slitlike holes under illumination of light with different polarizations (•, θ = 0◦ ; , θ = 45◦ ; and , θ = 90◦ ). (Reprinted with permission from [61]. Copyright 2007 Wiley.)

To fabricate microscale arrays of nanoholes, we subjected a thin (400 nm) layer of photoresist to two exposures through a PDMS mask of bas relief lines 2 µm wide spaced by 2 µm; the second exposure was carried out after the line mask was rotated an angle φ = 90◦ or 15◦ . The resist was then developed to produce an array of photoresist posts (Section 19.3.2), which were then transferred using PEEL into holes in free-standing, 170-nm-thick Au films. Two hundred fifty nanometer circular holes were constructed in a square array (φ = 90◦ ) and 90-nm × 950-nm slitlike holes were fabricated in a sharp-diamond array (φ = 15◦ ) (Figures 19.8a–b); these hole shapes with sharp corners and features are difficult to construct using FIB. Figure 19.8c indicates that circular holes in microscale arrays exhibited markedly different optical transmission characteristics in the visible wavelength range compared to subwavelength hole arrays (Figure 19.5b) [71, 72]. The microscale spacing of the array produced ultranarrow features in the transmittance (some had FWHM <15 nm) because of high order Bragg modes, where the square lattice has a spacing of a0 = 4 µm (the unit cell size, since the two sets of spacings are 1.8 and 2.2 µm). Thus, the SPP Bragg modes are constrained by 17 ≤ (i 2 + j 2 ) ≤ 58 at the Au–air interface and 40 ≤ (i 2 + j 2 ) ≤ 200 at the Au–glass interface in the range of 550–950 nm. High order surface plasmon Bragg modes is thus expected with microscale arrays, in contrast with the low order modes (e.g., (1,0) and (1,1)) observed

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

528

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

Figure 19.9. Bright field optical micrographs of colors transmitted by anisotropic nanohole arrays under different polarizations of incident light. (Reprinted with permission from [61]. Copyright 2007 Wiley.)

in subwavelength nanohole arrays. Similar to circular hole arrays, anisotropic hole arrays (with a larger and more complicated unit cell) exhibited complex spectral features (Figure 19.8d). Also, slitlike hole arrays showed a wavelength of maximum transmittance at λmax = 730 nm, which was 27 ± 3 times larger for θ = 90◦ () compared to θ = 0◦ (•), where θ is the angle between the long axis of the hole and the polarization direction of the incident light. Circular nanohole arrays illuminated with polarized white light transmitted color that appeared yellowish white and no noticeable difference was observed under polarization because of the symmetry of the holes. In contrast to circular holes, slitlike hole arrays exhibited dramatic color changes tunable from green to red as the polarization direction of the incident light rotated from θ = 0◦ to 90◦ . For slitlike hole arrays, at θ = 0◦ , the transmitted light was green; at θ = 45◦ , the holes transmitted green-orange; and at θ = 90◦ , the holes appeared red-orange. (Figure 19.9).

19.6 PLASMONIC PARTICLE ARRAYS 19.6.1 Metal and Dielectric Nanoparticles We can take advantage of the free-standing nature of the Au subwavelength nanohole films and use them as deposition masks to fabricate macroscale areas of metal and dielectric particles (Figure 19.10a). If the mask is approximately 2 in2 , over 1010 nanoparticles can be fabricated at the same time. Figure 19.10b depicts a portion of an array of Au nanoparticles (d = 100 nm, h = 50 nm) supported on a glass substrate (no adhesion layer is necessary) with the same pitch (a0 = 400 nm) as the SIL nanohole mask. Nanoparticles of other single component materials (Si, Ag, Cu) were fabricated and exhibited characteristic dark field (DF) scattering peaks from UV to visible wavelengths (Figure 19.10c). In addition, Au films perforated with patches of nanoholes could be used as masks for multiscale patterning of different nanoparticles on the same substrate. Figure 19.10d is an SEM (scanning electron microscope) image of an Au nanoparticle array generated from a single 10-µm square hole patch

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.6 PLASMONIC PARTICLE ARRAYS

(a)

(d)

(b)

529

(c)

(e)

(f)

Figure 19.10. Infinite and multiscale nanoparticle arrays. (a) Scheme depicting the fabrication of nanoparticle arrays using SIL Au nanohole films as deposition masks. (b) Au nanoparticle array (h = 50 nm) fabricated using an infinite nanohole film; (inset) image zoom. (c ) DF scattering spectra for metal (Ag, Cu, Au) and dielectric (Si) nanoparticles (h = 50 nm). (d ) 10 × 10-µm square of Au nanoparticles (h = 65 nm) fabricated using a nanohole patch film with a 25-µm pitch. (e) DF scattering image of Cu (left) and Ag (right) nanoparticles (h = 15 nm) fabricated on a different portion of the mask used in (d ). (f ) DF scattering image of Au (left) and Ag (right) nanoparticles (h = 65 nm) fabricated using a 10-µm circular nanohole patch film. (Reproduced with permission from [59]. Copyright 2007 Nature Publishing.)

within a patch array (a0 = 25 µm). A different portion of this patch array mask was also used to fabricate 100 µm2 areas of thin (h = 15 nm) Cu and Ag nanoparticles side by side; DF scattering images indicated that some patches at the interface contained both materials (Figure 19.10e). Finally, Figure 19.10f shows that thicker (h = 65 nm) Au and Ag particles can be patterned into arrays of 10-µm circular regions. Multilayered nanoparticles can also be fabricated by sequential deposition through SIL Au nanohole films (Figure 19.11a). We selected Au and alumina (Al2 O3 ) as our building materials to investigate how metal nanoparticles coupled through a dielectric spacer. Figure 19.11b shows scattering spectra of three types of structures, all with Au layers 25 nm thick. The nanoparticles with two layers,

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

530

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

(c)

Scattering intensity (a.u.)

(a)

Scattering intensity (a.u.)

(b)

Wavelength (nm)

Wavelength (nm)

Figure 19.11. Multilayered metal and dielectric nanoparticles and refractive index sensing. (a) Scheme depicting the fabrication of multilayered nanoparticle arrays using SIL Au nanohole films as deposition masks. (b) DF scattering spectra for multilayered metal/dielectric nanoparticle arrays in air (refractive index, n = 1); Au/Al2 O3 (h = 25/30 nm), Al2 O3 /Au (h = 55/25 nm), and Au/Al2 O3 /Au (h = 25/30/25 nm). (c ) DF scattering spectra of the multilayered nanoparticles in (b) surrounded by different refractive index liquids. (Reproduced with permission from [59]. Copyright 2007 Nature Publishing.)

Au/Al2 O3 (h = 25/30 nm) and Al2 O3 /Au (h = 55/25 nm), exhibited single, dipolar resonance peaks of comparable widths, with the former occurring at longer wavelengths as expected. The properties of nanoparticles in a sandwich structure (Au/Al2 O3 /Au, h = 25/30/25 nm), with upper and lower layers identical to the twolayered nanoparticles above, supported three peaks; two of the peaks were similar to those from the two-layered nanoparticles, while the third peak at shorter wavelengths (λ = 600 nm) might be attributed to coupling between the upper and lower Au disks or a hybridized plasmon resonance. Furthermore, we surrounded the nanoparticle arrays with different refractive index oils to determine how their spectra would evolve. As expected, the dipolar LSP resonances shifted to longer wavelengths as n increased, and the apparent blueshift of the short wavelength peak from the sandwich structure (Figure 19.11c, bottom graph) could be due to the appearance of a multipolar resonance.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

19.6 PLASMONIC PARTICLE ARRAYS

(b) Scattering intensity (a.u.)

(a)

531

Wavelength (nm) Figure 19.12. Anisotropic nanoparticles from anisotropic Au nanohole masks. (a) Ag particles after deposition of 50 nm of Ag through a Au nanohole film in Figure 19.8b. (b) DF scattering spectra under different polarizations (•, θ = 0◦ ; , θ = 45◦ ; and , θ = 90◦ ).

19.6.2 Anisotropic Nanoparticles Isolated submicrometer spheres and close-packed sphere arrays can act as templates or masks for metal deposition to obtain particles with crescentlike [73, 74] and triangular shapes [75], respectively. Anodized aluminum oxide membranes are another widely used template for producing nanorods as long as 10 µm [76]. We have used free-standing Au films of slitlike holes as deposition masks to create arrays of planar, highly anisotropic particles using the same process described above in Section 19.6.1. Because of their shape, the optical scattering spectra of these particles depended strongly on the polarization of incident light; light polarized parallel to the long axis of anisotropic Ag particles excited longitudinal plasmons at red wavelengths, and light polarized perpendicular to the long axis excited transverse plasmons at shorter wavelengths (Figure 19.12).

19.6.3 Pyramidal Nanostructures Serial lithographic techniques such as e-beam lithography can fabricate structures with arbitrary sizes and shapes in 2D [38, 77], although low throughput and small write areas (hundreds of square micrometers) are current challenges. The deposition of materials through Au nanohole masks as described above solves the scalability issues—and extends materials combinations—but does not solve the limitation of nanostructures with planar shapes. The PEEL process described in Section 19.3.3 enables an innovative approach to fabricate 3D nanostructures. At the same time that arrays of nanoholes are generated, pyramidal particles are formed directly beneath the holes and in the pyramidal pits of the Si substrate (Figure 19.4). This procedure thus provides a unique, template-based approach to generate monodisperse, anisotropic particles with smooth facets and sharp tips (r < 2 nm) with high densities [78].

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

532

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

(d)

(b)

Intensity (a.u.)

Intensity (a.u.)

(a)

14:12

(e) Scat. σ (µm2)

Scat. σ (µm2)

(c)

Wavelength (nm)

Wavelength (nm)

Figure 19.13. Arrays of pyramidal Au nanopyramids. (a) Au pyramids supported on Si pedestals. (b) Measured and (c ) calculated scattering properties of arrays of 100-nm-diameter Au pyramids (h = 40 nm) surrounded by PDMS. (d ) Measured and (e) calculated scattering properties of arrays of 250-nm-diameter Au pyramids (h = 50 nm) in PDMS. (Reprinted with permission from [29]. Copyright 2007 ACS.)

Figure 19.13a displays arrays of Au pyramids supported by Si pedestals; such architectures are achieved by anisotropic etching of the Si template containing the pyramids [79]. These pyramidal arrays were then embedded in PDMS for two reasons: (1) to create a uniform dielectric environment around the particles, and (2) to control the orientation of the pyramids in order to correlate specific plasmon modes with a well-defined orientation. To fix the pyramids at defined angles relative to a substrate, we sliced the PDMS film with nanopyramids at different angles relative to the plane of the particle array [80]. We first characterized a planar array of 100-nm Au pyramids (h = 50 nm) in PDMS whose tips were oriented perpendicular to the surface and whose base planes were oriented perpendicular to the optical axis of the microscope. These pyramids supported only a single resonance around 720 nm (Figure 19.13b). Numerical calculations of a single pyramid indicated that the lone scattering peak originated from a dipole resonance (Figure 19.13c), which suggested that the 100-nm pyramidal shells were not large enough to support multipolar resonances. This result is consistent with other reports that particles less than 100 nm do not usually support multipolar plasmon resonances [30]. In contrast, planar arrays of 250-nm Au pyramids (h = 50 nm) exhibited a strong peak at red wavelengths (∼650 nm) and another that appeared to extend into the NIR region (Figure 19.13d). Figure 19.13e shows the calculated spectrum of a single Au pyramidal shell, which exhibited a weak peak at 680 nm superimposed on a background that increased with increasing wavelength. An analysis of the induced polarizations indicated that the peak at 680 nm was from a quadrupole resonance—localized in the base plane of the pyramid—and that the background increased at longer wavelengths because of the tail of a dipolar excitation that peaks in the NIR. Although these calculations determined the Rayleigh scattering integral cross-section, which is not the same as the angle-resolved cross-section measured

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

REFERENCES

533

in the scattering spectra, the same resonance structure should be present. Figures 19.13d and e indicate that experiment and theory are in qualitative agreement, especially regarding the position of the quadrupole resonance, with the main difference being the intensity of the resonance. This difference could be attributed to several factors, including the difference between the calculated integral and angle-resolved cross-sections noted above or the sensitivity of the calculations to subtle structural features of the pyramids. Fundamental plasmonics research is laying the foundation for future technologies in diverse areas ranging from optics to electronics to imaging. To meet these challenges and to advance the science forward, however, the methods by which plasmonic nanostructures are constructed need to be simplified and made scalable. Multiscale patterning—and SIL in particular—provides a first step toward realizing new types of plasmonic architectures because (1) microscale ordering can be imposed on the nanoscale patterns, and thus plasmonic coupling strengths can be tuned and (2) complexity of plasmonic metamaterials can be built up simply by starting with a different master. The prospect for multiscale plasmonic structures is considerably bright. In the short term, because they support intense and concentrated electromagnetic fields, plasmonic materials can be used for highly sensitive refractive index sensing using films with arrays of nanoholes as well as multiplexed, label-free detecting and screening of analytes using arrays of particles. In the long term, active or hybrid plasmonic metamaterials will become important for understanding how plasmons can enhance or manipulate the properties of nonlinear media such as ferroelectric and optoelectronic materials. Furthermore, such metamaterials combined with super-resolution imaging techniques could yield an unprecedented platform for studying biological processes of a cell.

ACKNOWLEDGMENTS This work was supported by the National Science Foundation and the David and Lucile Packard Foundation. We thank George C. Schatz for numerous useful discussions. T.W.O. is a DuPont Young Professor, a Cottrell Scholar of Research Corporation, and an Alfred P. Sloan Research Fellow.

REFERENCES 1. Maier, S. A., Brongersma, M. L., Kik, P. G., Meltzer, S., Requicha, A. A. G., and Atwater, H. A. (2001) Plasmonics—a route to nanoscale optical devices. Adv. Mater. 13, 1501–1505. 2. Ozbay, E. (2006) Plasmonics: merging photonics and electronics at nanoscale dimensions. Science 311, 189–193. 3. Barnes, W. L., Dereux, A., and Ebbesen, T. W. (2003) Surface plasmon subwavelength optics. Nature 424, 824–830.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

534

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

4. Brongersma, M. L., Hartman, J. W., and Atwater, H. A. (2000) Electromagnetic energy transfer and switching in nanoparticle chain arrays below the diffraction limit. Phys. Rev. B 62, 16356–16359. 5. Quinten, M., Leitner, A., Krenn, J. R., and Aussenegg, F. R. (1998) Electromagnetic energy transport via linear chains of silver nanoparticles. Opt. Lett. 23, 1331–1333. 6. Maier, S. A., Barclay, P. E., Johnson, T. J., Friedman, M. D., and Painter, O. (2005) Experimental demonstration of fiber-accessible metal nanoparticle plasmon waveguides for planar energy guiding and sensing. Appl. Phys. Lett. 86, 071103. 7. Bozhevolnyi, S. I., Erland, J., Leosson, K., Skovgaard, P. M. W., and Hvam, J. M. (2001) Waveguiding in surface plasmon polariton band gap structures. Phy. Rev. Lett. 86, 3008–3011. 8. Ditlbacher, H., Krenn, J. R., Schider, G., Leitner, A., and Aussenegg, F. R. (2002) Twodimensional optics with surface plasmon polaritons. Appl. Phys. Lett. 81, 1762–1764. 9. Drezet, A., Koller, D., Hohenau, A., Leitner, A., Aussenegg, F. R., and Krenn, J. R. (2007) Plasmonic crystal demultiplexer and multiports. Nano Lett. 7, 1697–1700. 10. Kitson, S. C., Barnes, W. L., and Sambles, J. R. (1996) A full photonic band gap for surface modes in the visible. Phys. Rev. Lett. 77, 2670–2673. 11. Bozhevolnyi, S. I., Volkov, V. S., Devaux, E., Laluet, J.-Y., and Ebbesen, T. W. (2006) Channel plasmon subwavelength waveguide components including interferometers and ring resonators. Nature 440, 508–511. 12. Lezec, H. J., Degiron, A., Devaux, E., Linke, R. A., Martin-Moreno, L., Garcia-Vidal, F. J., and Ebbesen, T. W. (2002) Beaming light from a subwavelength aperture. Science 297, 820–822. 13. Wiley, B., Sun, Y., Mayers, B., and Xia, Y. (2004) Shape-controlled synthesis of metal nanostructures: the case of silver. Chem.—Eur. J. 11, 454–463. 14. Jana, N. R., Gearheart, L., and Murphy, C. J. (2001) Wet chemical synthesis of silver nanorods and nanowires of controllable aspect ratio. Chem. Commun. 7, 617–618. 15. Saleh, B. E. A. and Teich, M. C. (1991) Fundamentals of Photonics, Wiley-Interscience, New York. 16. Kawata, S. (2001) Near-Field Optics and Surface Plasmon Polaritons, Springer, Osaka. 17. Mock, J. J., Barbic, M., Smith, D. R., Schultz, D. A., and Schultz, S. (2002) Shape effects in plasmon resonance of individual colloidal silver nanoparticles. J. Chem. Phys. 116, 6755–6759. 18. Ebbesen, T. W., Lezec, H. J., Ghaemi, H. F., Thio, T., and Wolff, P. A. (1998) Extraordinary optical transmission through sub-wavelength hole arrays. Nature 391, 667–669. 19. Grigorenko, A. N., Geim, A. K., Gleeson, H. F., Zhang, Y., Firsov, A. A., Khrushchev, I. Y., and Petrovic, J. (2005) Nanofabricated media with negative permeability at visible frequencies. Nature 438, 335–338. 20. Dolling, G., Wegener, M., Soukoulis, C. M., and Linden, S. (2006) Negative-index metamaterial at 780 nm wavelength. Opt. Lett. 32, 53–55. 21. Klein, M. W., Enkrich, C., Wegener, M., and Linden, S. (2006) Second-harmonic generation from magnetic metamaterials. Science 313, 502–504. 22. Raether, H. (1988) Surface Plasmons, Springer, Berlin. 23. Gao, H., Henzie, J., and Odom, T. W. (2006) Direct evidence for surface plasmonmediated enhanced light transmission through metallic nanohole arrays. Nano Lett. 6, 2104–2108.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

REFERENCES

535

24. Lamprecht, B., Krenn, J. R., Schider, G., Ditlbacher, H., Salerno, M., Felidj, N. L. A., and Aussenegg, F. R. (2001) Surface plasmon propagation in microscale metal stripes. Appl. Phys. Lett. 79, 51. 25. Daniel, M.-C. and Astruc, D. (2004) Gold nanoparticles: assembly, supramolecular chemistry, quantum-size-related properties, and applications toward biology, catalysis, and nanotechnology. Chem. Rev. 104, 293–346. 26. Xia, Y. and Halas, N. J. (2005) Shape-controlled synthesis and surface plasmonic properties of metallic nanostructures. MRS Bull. 30, 338–348. 27. Haes, A. J., Haynes, C. L., McFarland, A. D., Schatz, G. C., Van Duyne, R. P., and Zou, S. (2005) Plasmonic materials for surface-enhanced sensing and spectroscopy. MRS Bull. 30, 368–375. 28. Bohren, C. F. and Huffman, D. R. (1983) Absorption and Scattering of Light by Small Particles, Wiley, New York, p. 82. 29. Henzie, J., Shuford, K. L., Kwak, E.-S., Schatz, G. C., and Odom, T. W. (2006) Manipulating the optical properties of pyramidal nanoparticle arrays. J. Phys. Chem. B 110, 14028. 30. Jain, P. K., Lee, K. S., El-Sayed, I. H., and El-Sayed, M. A. (2006) Calculated absorption and scattering properties of gold nanoparticles of different size, shape, and composition: applications in biological imaging and biomedicine. J. Phys. Chem. B 110, 7238–7248. 31. Murphy, C. J., Sau, T. K., Gole, A. M., Orendorff, C. J., Gao, J., Gou, L., Hunyadi, S. E., and Li, T. (2005) Anisotropic metal nanoparticles: synthesis, assembly, and optical applications. J. Phys. Chem. B 2005, 13857. 32. Kumbhar, A. S., Kinnan, M. K., and Chumanov, G. (2005) Multipole plasmon resonances of submicron silver particles. J. Am. Chem. Soc. 127, 12444–12445. 33. Millstone, J. E., Park, S., Shuford, K. L., Qin, L., Schatz, G. C., and Mirkin, C. A. (2005) Observation of a quadrupole plasmon mode for a colloidal solution of gold nanoprisms. J. Am. Chem. Soc. 127, 5312–5313. 34. Shuford, K. L., Ratner, M. A., and Schatz, G. C. (2005) Multipolar excitation in triangular nanoprisms. J. Chem. Phys. 123, 114713. 35. Kelly, K. L., Coronado, E., Zhao, L. L., and Schatz, G. C. (2003) The optical properties of metal nanoparticles: the influence of size, shape, and dielectric environment. J. Phys. Chem. B 107, 668. 36. Krenn, J. R., Schider, G., Rechberger, W., Lamprecht, B., Leitner, A., Aussenegg, F. R., and Weeber, J. C. (2000) Design of multipolar plasmon excitations in silver nanoparticles. Appl. Phys. Lett. 77, 3379–3381. 37. Wiley, B., Sun, Y., Chen, J., Cang, H., Li, Z., Li, X., and Xia, Y. (2005) Shape-controlled synthesis of silver and gold nanostructures. MRS Bull. 30, 356–361. 38. Hicks, E. M., Zou, S., Schatz, G. C., Spears, K. G., Van Duyne, R. P., Gunnarson, L., Rindzevicius, T., Kasemo, B., et al. (2005) Controlling plasmon line shapes through diffractive coupling in linear arrays of cylindrical nanoparticles fabricated by electron beam lithography. Nano Lett. 5, 1065–1070. 39. Haynes, C. L., McFarland, A. D., Zhao, L., Van Duyne, R. P., Schatz, G. C., Gunnarsson, L., Prikulis, J., Kasemo, B., et al. (2003) Nanoparticle optics: the importance of radiative dipole coupling in two-dimensional nanoparticle arrays. J. Phys. Chem. B 107, 7337–7342. 40. Gunnarsson, L., Rindzevicius, T., Prikulis, J., Kasemo, B., Kall, M., Zou, S., and Schatz, G. C. (2005) Confined plasmons in nanofabricated single silver particle pairs:

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

536

41.

42. 43. 44.

45.

46.

47. 48. 49. 50. 51.

52. 53.

54. 55. 56.

57.

58.

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

experimental observations of strong interparticle interactions. J. Phys. Chem. B 109, 1079–1087. Grupp, D. E., Lezec, H. J., Thio, T., and Ebbesen, T. W. (1999) Beyond the Bethe limit: tunable enhanced light transmission through a single sub-wavelength aperture. Adv. Mater. 11, 860–862. Xia, Y. and Whitesides, G. M. (1998) Soft lithography. Annu. Rev. Mater. Sci. 28, 153–184. Schimd, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. Odom, T. W., Love, J. C., Paul, K. E., Wolfe, D. B., and Whitesides, G. M. (2002) Improved pattern transfer in soft lithography using composite stamps. Langmuir 18, 5314. Kwak, E.-S., Henzie, J., Chang, S.-H., Gray, S. K., Schatz, G. C., and Odom, T. W. (2005) Surface plasmon standing waves in large-area subwavelength hole arrays. Nano Lett. 5, 1963–1967. Odom, T. W., Thalladi, V. R., Love, J. C., and Whitesides, G. M. (2002) Generation of 30–50 nm structure using easily fabricated, composite PDMS masks. J. Am. Chem. Soc. 124, 12112–12113. Stender, C. L., Greyson, E. C., and Odom, T. W. (2005) Patterned MoS2 nanostructures over centimeter-square areas. Adv. Mater. 17, 2837–2841. Henzie, J., Kwak, E.-S., and Odom, T. W. (2005) Mesoscale metallic pyramids with nanoscale tips. Nano Lett. 5, 1199–1202. Greyson, E. C., Babayan, Y., and Odom, T. W. (2004) Directed growth of ordered arrays of small-diameter ZnO nanowires. Adv. Mater. 16, 1348–1352. Gates, B. D., Xu, Q., Love, C. J., Wolfe, D. B., and Whitesides, G. M. (2004) Unconventional nanofabrication. Ann. Rev. Mater. Res. 34, 339–372. Love, C. J., Wolfe, D. B., Chabinyc, M. L., Paul, K. E., and Whitesides, G. M. (2002) Self-assembled monolayers of alkanethiolates on palladium are good etch resists. J. Am. Chem. Soc. 124, 1576–1577. Xia, Y., Zhao, X.-M., and Whitesides, G. M. (1996) Pattern transfer: self-assembled monolayers as ultrathin resists. Microelectron. Eng. 32, 255–268. Odom, T. W., Love, C. J., Wolfe, D. B., Paul, K. E., and Whitesides, G. M. (2002) Improved pattern transfer in soft lithography using composite stamps. Langmuir 18, 5314–5320. Childs, W. R. and Nuzzo, R. G. (2005) Large-area patterning of coinage-metal thin films using decal transfer lithography. Langmuir 21, 195–202. Loo, Y.-L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Interfacial chemistries for nanoscale transfer printing. J. Am. Chem. Soc. 124, 7654–7655. Zaumseil, J., Meitl, M. A., Hsu, J. W. P., Acharya, B. R., Baldwin, K. W., Loo, Y.-L., and Rogers, J. A. (2003) Three-dimensional and multilayer nanostructures formed by nanotransfer printing. Nano Lett. 3, 1223–1227. Xu, Q., Bao, J., Rioux, R. M., Perez-Castillejos, R., Capasso, F., and Whitesides, G. M. (2007) Fabrication of large-area patterned nanostructures for optical applications by nanoskiving. Nano Lett. 7, 2800–2805. Stewart, M. E., Mack, N. H., Malyarchuk, V., Soares, J. A. N. T., Lee, T.-W., Gray, S. K., Nuzzo, R. G., and Rogers, J. A. (2006) Quantitative multispectral biosensing

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

REFERENCES

59. 60. 61. 62.

63. 64. 65.

66. 67. 68.

69. 70.

71.

72.

73. 74.

75. 76.

77.

537

and 1D imaging using quasi-3D plasmonic crystals. Proc. Natl. Acad. Sci. USA 103, 17143–17148. Henzie, J., Lee, M. H., and Odom, T. W. (2007) Multiscale patterning of plasmonic metamaterials. Nat. Nanotechnol. 2, 549–554. Greyson, E. C., Babayan, Y., and Odom, T. W. (2004) Directed growth of ordered arrays of small-diameter ZnO nanowires. Adv. Mater. 16, 1348–1352. Lee, M. H., Gao, H., Henzie, J., and Odom, T. W. (2007) Microscale arrays of nanoscale holes. Small 3 (12), 2029–2033. Kwak, E. S., Henzie, J., Chang, S. H., Gray, S. K., Schatz, G. C., and Odom, T. W. (2005) Surface plasmon standing waves in large-area subwavelength hole arrays. Nano Lett. 5, 1963–1967. Bethe, H. A. (1944) Theory of diffraction by small holes. Phys. Rev. 66, 163–182. Genet, C. and Ebbesen, T. W. (2007) Light in tiny holes. Nature 445, 39–46. Barnes, W. L., Murray, W. A., Dintinger, J., Devaux, E., and Ebbesen, T. W. (2004) Surface plasmon polaritons and their role in the enhanced transmission of light through periodic arrays of subwavelength holes in a metal film. Phys. Rev. Lett. 92, 107401. Pendry, J. B., Mart´ın-Moreno, L., and Garcia-Vidal, F. J. (2005) Mimicking surface plasmons with structured surfaces. Science 305, 847–848. Garcia-Vidal, F. J., Mart´ın-Moreno, L., and Pendry, J. B. (2005) Surfaces with holes in them: new plasmonic metamaterials. J. Opt. A 7, S97–S101. Chang, S.-H., Gray, S. K., and Schatz, G. C. (2005) Surface plasmon generation and light transmission by isolated nanoholes and arrays of nanoholes in thin metal films. Opt. Express 13, 3150–3165. Wood, R. W. (1935) Anomalous diffraction gratings. Phys. Rev. 48, 928–936. McMahon, J. M., Henzie, J., Odom, T. W., Schatz, G. C., and Gray, S. K. (2007) Tailoring the sensing capabilities of nanohole arrays in gold films with Rayleigh anomaly-surface plasmon polaritons. Opt. Express 15, 18119–18129. Ghaemi, H. F., Thio, T., Grupp, D. E., Ebbesen, T. W., and Lezec, H. J. (1998) Surface plasmons enhance optical transmission through subwavelength holes. Phys. Rev. B 58, 6779–6782. Dintinger, J., Klein, S., Bustos, F., Barnes, W. L., and Ebbesen, T. W. (2005) Strong coupling between surface plasmon-polaritons and organic molecules in subwavelength hole arrays. Phys. Rev. B 71, 035424. Shumaker-Parry, J. S., Rochholz, H., and Kreiter, M. (2005) Fabrication of crescentshaped optical antennas. Adv. Mater. 17, 2131–2134. Liu, G. L., Lu, Y., Kim, J., Doll, J. C., and Lee, L. P. (2005) Magnetic nanocrescents as controllable surface-enhanced Raman scattering nanoprobe for biomolecular imaging. Adv. Mater. 17, 2683–2688. Willets, K. A. and Van Duyne, R. P. (2006) Localized surface plasmon resonance spectroscpy and sensing. Annu. Rev. Phys. Chem. 58, 267–297. Nicewarner-Pena, S. R., Freeman, R. G., Reiss, B. D., He, L., Pena, D. J., Walton, I. D., Cromer, R., Keating, C. D., et al. (2001) Submicrometer metallic barcodes. Science 294, 137–141. Canfield, B. K., Kujala, S., Kauranen, M., Jefimovs, K., Vallius, T., and Turunen, J. (2005) Remarkable polarization sensitivity of gold nanoparticle arrays. Appl. Phys. Lett. 86, 183109.

P1: OTA/XYZ P2: ABC c19 JWBK252/Lee September 18, 2008

538

14:12

Printer Name: Yet to Come

MULTISCALE FABRICATION OF PLASMONIC STRUCTURES

78. Henzie, J., Kwak, E.-S., and Odom, T. W. (2005) Mesoscale metallic pyramids with nanoscale tips. Nano Lett. 5, 1199–1202. 79. Seidel, H., Csepregi, L., Heuberger, A., and Baumg¨artel, H. (1990) Anisotropic etching of crystalline silicon in alkaline solutions. J. Electrochem. Soc. 137, 3612. 80. Henzie, J., Barton, J. E., Stender, C. L., and Odom, T. W. (2006) Large-area nanoscale patterning: chemistry meets fabrication. Acc. Chem. Res. 39, 249-257.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20 A RIGIFLEX MOLD AND ITS APPLICATIONS Se-Jin Choi, Tae-Wan Kim, and Seung-Jun Baek

20.1 INTRODUCTION Over the past decade, top-down lithographic approaches with a mold or stamp have led to various routes to unconventional patterning such as soft lithography [1–3], nanoimprint lithography (NIL) [4–6], and others. In these methods, various patterns are simultaneously formed by the physical contact of a hard or soft mold with the targeted substrate. The characteristics inherent in these contact-dependent methods give rise to certain limitations and disadvantages. The imprint lithography based on a hard, stiff master mold can provide a resolution down to several nanometers scale but it requires a very high pressure and an extremely flat surface [4–6]. The soft lithography based on a soft mold, typically made of poly(dimethylsiloxane) (PDMS), does not necessarily require a pressure or a flat surface but its resolution capability is quite limited [7–10]. Although PDMS is the most popular choice because of its low modulus for conformal contact, low surface tension for easy release, and high elongation at break for patterning capability on curved surface, there still remain some limitations to overcome. Its low modulus causes applications to be restricted in many cases only to microscale regime and its poor solvent resistance [11] leads to reproducibility problems due to degradation (i.e., swelling) in the course of patterning repeatedly. Besides, the dimensional change (i.e., thermal shrinkage) due to thermal expansion after thermal curing makes it difficult to apply it for large-area mass production, especially multilevel pattern registration over a large area.

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


539

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

540

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

On the other hand, rigid mold made of quartz glass or silicon, which have been utilized typically in imprint lithography, are superior to PDMS in modulus and swelling resistance, making it possible to accomplish sub-100-nm patterning and to use the mold repeatedly. However, an inherent drawback is the direct use of the hard mold that is costly to fabricate, typically made by electron-beam lithography. It also requires additional surface treatment step to reduce the surface energy of the mold surface for easy release. To overcome these problems, several approaches have been taken. These include step and flash method [12], low pressure NIL [13], and utilization of a modified PDMS [14–17] or new materials for soft lithography [18–20]. Nevertheless, some basic problems still remain, such as the use of an expensive quartz mold and a relatively long processing time for the preparation of mold including fluorination step [12] in the step and flash imprinting, or construction of a more complex elastomeric mold [14–17] and pattern collapse when the features are fine and dense with a high aspect ratio [7–10] in soft lithography. As these problems are inherently related to the mold material, there has been a need for a new mold material that can provide a rigidity high enough for fine and dense features with a high aspect ratio and yet a degree of flexibility for conformal contact over a nonflat, large area. The improvement in mold properties for mechanical and/or chemical requirements appears to be a necessity for industrial applications. In this chapter, various approaches and strategies for mold fabrication are introduced, which would open opportunities from a practical point of view. In particular, the adaptability of a versatile mold material to various contrasting unconventional lithographic techniques is described, from soft lithography to NIL.

20.2 MODULUS-TUNABLE RIGIFLEX MOLD Photocuring for mold fabrication is an avenue that can be used to avoid thermal curing process, which can lead to thermally induced distortion and shrinkage of the mold. The thermal curing also requires a long curing time, up to several hours. A common underlying concept for photocuring mold is to incorporate a reactive group for radical chain reaction such as acryloxy into a backbone with low surface energy such as fluoropolymer [21–23] or siloxanepolymer [24], which facilitates easy release of mold after patterning. Subsequent preparation of the mold material involves blending with a photoinitiator, which creates radicals as an initiator for photopolymerization of acryloxy moiety upon exposure to ultraviolet (UV) in the wavelength range of 300–400 nm. The physical properties such as toughness and modulus of mold material system can be adjusted by cross-linking density as well as chemical structure in the backbone. The photocuring PDMS elastomer [24] with a modified PDMS chemistry showed many physical properties that meet many of the requirements for soft lithography, especially for nanopatterning applications. Photocurable PDMS provides a tensile modulus higher than that of 184-PDMS and an elongation at break that is much higher than that of hard PDMS (h-PDMS) [14–17, 25], which makes it easier to handle than h-PDMS and to be less susceptible to mechanical deformation. A pattern

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.2 MODULUS-TUNABLE RIGIFLEX MOLD

541

consisting of 600-nm-thick and 300-nm-wide lines with 300-nm spacing between lines has been replicated, which cannot be accomplished with 184-PDMS. Despite numerous attractive advantages of PDMS-based chemistry, a serious drawback is that this material swells significantly when brought into contact with even a small amount of curable organic monomer or most organic solvents, which severely limits its capabilities in soft lithography [11]. One of the approaches to this problem has been to replace PDMS with photocurable perfluoropolyethers (PFPEs), which exhibits the attractive properties of PDMS along with a resistance to swelling to most organic liquid compounds [21, 22]. The reaction involves the methacrylate functionalization of a commercially available hydroxy-terminated PFPE with isocyanatoethyl methacrylate to yield a methacryloxy-functionalized polymer, which is a photocurable liquid. PFPEs are a unique class of fluoropolymers that exhibit lower surface energy and excellent chemical resistance compared to PDMS. They presented a novel solvent-resistant microfluidic devices fabricated from PFPEbased elastomer, showing a remarkable resistance to organic solvents, i.e., negligible swelling of less than 3% after immersion in dichloromethane for 94 h. A mold material should possess a number of desirable properties such as physical rigidity for high resolution, flexibility for conformal contact, small curing-induced shrinkage for dimension accuracy, and chemical inertness for durability. The mold material system introduced in this section provides a flexibility for adjusting the mold properties to match the demands of a particular patterning task. Furthermore, photocuring is used to prepare the mold to avoid thermally induced shrinkage and to decrease the preparation time from several hours to a matter of minutes [26]. The modulus-tunable mold material shown in Figure 20.1 consists of a functionalized

Functionalized prepolymer with acrylated group O Cycloaliphatic unit

HN

C

O R1

O

O

C

NH

Cycloaliphatic unit

NH O

NH

C

O

O

O

C O

R2

R2

O

O

C

O

C m

UV (250–400 nm)

Cross-linked, cured mold

CH3

O C

C

C

O

CH3

OH

Photoinitiator

O R3 CH3 H3C

3

Si

O

Si

O

CH3

Si p

CH3

O

Si

CH3

3

q

Releasing agent: organo-modified silicone acrylate

Figure 20.1. Formula of the UV-curable mold in liquid precursor. (Reprinted with permission from [26] and The American Chemical Society.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

542

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

prepolymer with acrylate group as a main binder, a photoinitiator, and a radiationcurable releasing agent for surface activity. Additionally, a monomeric cross-linking modulator can be included in the mold precursor for further tuning the modulus of the fabricated mold. Even at 0.4 wt% loading of the releasing agent in the mold composition, in particular, the surface energy of the cure mold decreases to around 23 dyn cm–1 , comparable to that of PDMS (∼21.6 dyn cm–1 ), which is sufficiently low enough for clean release [27]. A small amount of the additive, which participates in the cross-linking, serves the purpose of lowering the surface energy while not affecting other properties. Therefore, the mold can be easily prepared as hard or soft one, depending on the choice of a suitable prepolymer (i.e., chemical structure and/or the number of reactive site for cross-linking). Basically, the excellent properties of the mold are a result of the fact that the preploymer contains both cycloaliphatic and linear long chains [28]. The former provides the mechanical rigidity while the latter provides the flexibility. Therefore, the mold is hard and yet flexible enough for molding. Shown in Figure 20.2 is the procedure for replicating the composite mold. The liquid mold mixture is drop dispensed on a master pattern. A flexible and transparent support made of polyurethane elastomer or soft epoxy resin is brought into contact

(a)

(c)

(b) Master

Master Coating UV-curable material

Attaching supporting substrate

First replica

UV radiation curing and peeling off Second replica Edge cutting

First replica

Second replica

Figure 20.2. Schematic illustration for the fabrication of UV-curable molds. (a) The first replica is the negative replica of the master pattern. (b) Process of self-replication. The second replica is the positive replica. (c ) SEM images, from the top down, of the master, the first replica, and the second replica. (Bar scale is 500 nm) (Reprinted with permission from [26] and The American Chemical Society.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.2 MODULUS-TUNABLE RIGIFLEX MOLD

(a)

(b)

(c)

(d)

543

Figure 20.3. SEM and optical microscopic images of replication results. (a) Master pattern pf 75-nm line/space polarizer. (b) Replicated pattern of (a). (c ) Replicated pattern of a 100-nm line/space circuit pattern. Inset is the cross-sectional SEM image. (Bar scale in the inset is 500 nm.) (d ) Example of large-area replication of hologram gratings. (Reprinted with permission from [26] and The American Chemical Society.)

with the coated mold material. Subsequently, it is exposed to UV (250–400 nm) for a few tens of seconds through the transparent backside. After the UV curing, the mold is peeled from the master. One notable feature of the mold material is that it is stiff enough for replicating dense, very fine sub-100-nm features, which is difficult to achieve with other molding materials because of the pattern-collapsing problem arising from their low mechanical strength. The SEM (scanning electron microscopy) micrographs in Figure 20.2 for the master and for the replicated pattern with the mold demonstrate the replication capability, respectively. To show the replication capability of the mold for sub-100-nm feature, a 75-nm-wide polarizer pattern was used. The master and replicated pattern are shown in Figures 20.3a and b, respectively. Another example of the replication is shown in Figure 20.3c for a 100-nm equal-spaced line and space pattern with a step height of 330 nm. This result demonstrates the capability to replicate patterns with a high aspect ratio. The mold is also flexible such that large-area replication can be accomplished, as shown in Figure 20.3d. The fabricated mold is inert to some representative polar and nonpolar solvents and is durable in that 500 replication cycles for repeated UV curing did not result in notable degradation of the mold.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

544

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

20.3 APPLICATIONS OF RIGIFLEX MOLD 20.3.1 From Nanoimprint to Microcontact Printing NIL is a technique to generate nanostructures by mechanically pressing a hard mold with surface-relief features into a thin thermoplastic polymer film that is then heated above the glass transition temperature (T g ) of the polymer. Microcontact printing as one of the typical soft lithography is a technique to define a pattern by transferring a molecular ink from a patterned elastomeric mold onto substrate by conformal contact. Here, we describe the use of the mold material for two contrasting lithographic techniques [29]: physical patterning of imprinting that is usually carried out with a hard mold and chemical patterning of microcontact printing that is usually carried out with an elastomeric soft mold. For the two contrasting methods, the mechanical properties can be tuned to each application by choosing a cross-linking modulator that can modulate the chemical structure of the UV-curable mold material, as shown in Figure 20.4. A radiation-curable modulator plays a key role in preparing the mold material. Mixing the monomeric modulator with the prepolymer, which possess trifunctionality in the form of cycloaliphatic and high molecular structure for a balance between rigidity and flexibility, results in a reduction of viscosity that allows the mixture to fill very fine structures of sub-100-nm channels or holes [29]. More importantly, the modulator allows one to change the chain length or cross-linking density, which further leads to a modification in the mechanical properties. For a relatively hard mold that is needed for imprint lithography, which is a filmtype mold shown in Figure 20.5b, trimethylolpropane (ethoxylated)3 triacrylate was chosen as the modulator since its relatively short chain length would lead to a dense cross-linking. For a soft mold to be used for microcontact printing, which uses a block-type mold as shown in Figure 20.5c, trimethylolpropane (ethoxylated)15 triacrylate was chosen because of its high content of ether linkage. The long ether linkage can have a high degree of rotational movement that allows stretching. As shown in Figure 20.5b, therefore, the 2-mm-thick soft block mold can be bent or deformed freely and yet it still maintains the pattern shape and enables a conformal contact with other substrates. Moreover, the modulus can also be tuned by changing the amount of the modulator in the mold material and/or by mixing the two modulators. Shown in Figure 20.6 are the patterning results obtained by imprint lithography with the hard rigiflex mold [29]. While maintaining the temperature above T g of polystyrene (M n = 404,200 g mol–1 ), typically 120◦ C, a pressure of 3–4 bar was applied for 20-min imprinting. Both a high aspect ratio (>4) pattern in Figure 20.6a and a sub-100-nm-dense pattern in Figure 20.6b were successfully patterned by imprinting. Compared to the original pattern height of master, the lower relative pattern height in Figure 20.6b than that in Figure 20.6a (filling ratio to the channel: 60% vs. 40%) may have to do with the difference in the mass transfer into the channel due to the pattern density. The more sparsely spaced pattern in Figure 20.6a (line CD (critical dimension)/space CD = 1.0:2.5) would have more mass to draw from the space region than the more densely spaced pattern in Figure 20.6b (line CD/space CD = 1:1).

Prepolymer

C

3

CH2CH2

C O

O

CH2

O

O

C O

O

O C NH

O

O CH2

C

CH3

CH2

C

CH2

O

CH2 O C

O

C

O

CH3

O

O

O

CH2 O C

3

O

NH CH2

CH2CH2

O 2

CH3

O

CH2CH2

CH2 CH2

CH2 C

CH2

O

2

Trimethylolpropane (ethoxylated) 6 triacrylate

C

O

C

O

CH2

O

O

CH2CH2

O

O

3

2

O

NH

O C NH

O C

CH2CH2

Modulator for soft mold

NH C O

O

CH2CH2 O C NH

O

CH2 O C NH CH2

CH2

CH2 C

CH2

O C O

C

C

O

CH2

Figure 20.4. Chemical structure of the UV-curable mold material for modulus tuning. (Reprinted with permission from [29] and The American Chemical Society.)

C

O

O

Trimethylolpropane triacrylate

CH2 NH C O

O

O C O

NH

CH2

NH

O

14:12

Modulator for hard mold

NH

CH2

NH

Aliphatic urethane triacrylate oligomer

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008 Printer Name: Yet to Come

545

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

546

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

Figure 20.5. (a) Comparison of tensile stress–strain relationship for three mold materials from the UTM (universal testing machine) analysis. (b) Example of hard mold prepared by a filmsupported type. (c ) Example of the soft mold prepared by a single block type. (Reprinted with permission from [29] and The American Chemical Society.)

It is notable that the hard rigiflex mold allows for a low pressure imprinting just as a flexible Teflon mold [13, 30] does. The applied pressure of 3–4 bar contrasts the high pressure in excess of 100 bar needed in the conventional NIL [4–6]. The mold prepared in a film form, which has a degree of flexibility inherent in the mold material, provides the necessary flexibility for the low pressure imprinting. This low pressure condition makes it possible for the mold to maintain its mechanical integrity during the imprinting process. Figure 20.7 shows the results of the gold surface patterned by microcontact printing with the soft rigiflex mold [29]. In microcontact printing, the swelling characteristics of the mold are important because the inking material is transferred to the mold as a solution and the solvent can induce unexpected deformation of the mold. To check the effect of solvent swelling, a soft block-type mold was immersed in pure ethanol for 3 h and then weighed for the mass difference by swelling. Only a 0.4 wt% increase in the weight was observed, which would have little effect on the pattern deformation by swelling. This slight solvent retention was rather helpful for preserving the hexadecanethiol (HDT) on the mold surface and continuous release of HDT during the contact. Although there are some defects and deterioration along the edges as revealed in Figure 20.7, the pattern transfer by the printing was successfully demonstrated even for the densely spaced sub-micrometer pattern (250-nm line and space pattern in Figure 20.7a). There have been a number of results for nanoscale patterning by microcontact printing method, but they were mostly for sparsely separated and isolated patterns [31–33]. Only recently has there been a successful patterning by the microcontact printing for a 300-nm line and space pattern with a UV-curable PDMS mold [24]. However, the conventional PDMS and other materials [14–19] fail to maintain the mechanical integrity during the contact process when the pattern is densely spaced, which leads to a pattern collapse or nonselective diffusion of self-assembled monolayer material to the gold surface [34].

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

(a)

Master

(b)

547

Master

Replica

Replica

Polymer

Polymer

(c)

– Figure 20.6. Imprinting results with hard rigiflex mold. (a) SEM images for a high aspect ratio pattern (master: 100-nm line × 250-nm space × 400-nm depth). Bar scale is 1 µm. Inset in the master image (bar scale is 500 nm) shows a magnification for a clear view of the pattern shape. (b) SEM images for a sub-100-nm pattern (master: 75-nm line × 75-nm space × 100-nm depth). Bar scale is 500 nm. (c ) Three-dimensional (top) and cross-sectional (bottom) atomic force microscopy images for the patterned polymer in (b). Z scale in the top image is 300 nm. (Reprinted with permission from [29] and The American Chemical Society.)

20.3.2 Rapid Flash Patterning for Residue-Free Patterning Presented in this section is a rapid patterning technique [35], which allows for the exposure of the substrate surface. Although many techniques are available for large area, general purpose patterning, there are a number of aspects yet to be resolved such as the speed, which the patterning can be accomplished, and exposure of the

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

548

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

(a)

(b)

(c)

Figure 20.7. SEM (top) and atomic force microscopy (AFM) (bottom) micrographs for the patterning results by microcontact printing method. Z scale in all AFM images is 500 nm. (a) Gold line pattern with 500-nm period (stamping mold: 250-nm line × 250-nm space × 400-nm depth). Bar scale in the SEM image is 2 µm. (b) Gold line pattern with 1.2-µm period (stamping mold: 700-nm line × 500-nm space × 1000-nm depth). Bar scale in the SEM image is 5 µm. (c ) Gold dot pattern with 1.2-µm period (stamping mold: 750-nm dot × 450-nm space × 400-nm depth). Bar scale in the SEM image is 5 µm. (Reprinted with permission from [29] and The American Chemical Society.)

substrate surface in the course of the patterning. Sub-100-nm structure can be fabricated in tens of seconds with an aspect ratio much lager than unity by the general purpose-patterning method introduced here. Unlike other unconventional methods, the substrate surface can be made exposed and the resulting pattern height is sufficiently high for subsequent etching of the substrate. A rigiflex mold and a rapid flash heating by an infrared lamp are the main ingredients of the technique. The rapid flash patterning (RFP) method is shown schematically in Figure 20.8. A thin rigiflex mold is placed on the polymer layer that has been spin-coated onto a substrate and dried. To assure a uniform pressure distribution when pressed, a PDMS block is placed on the mold followed by a glass plate, onto which pressure is applied, typically 3–4 bars. In the pressed state, a halogen lamp in infrared (IR) range is flashed for a short period of time, typically 60 s with a 500-W lamp, but 10 s with 10-kW lamp array. The temperature reached is typically around 260◦ C. After a few minutes of cooling, the pressure is relieved, followed by removal of the rigiflex mold, leaving behind a patterned polymer layer on the substrate. Shown in Figure 20.9 is a sub-100-nm pattern formed by RFP on poly(vinylalcohol) (PVA) surface with a 500-W IR lamp. Figure 20.9a is the SEM image of the patterned PVA. The cross-sectional SEM image of the line and space pattern in Figure 20.9a is shown in Figure 20.9b. For the 80-nm-wide lines, the patterned line height is 440 nm, which gives an aspect ratio in excess of 5. One-step reactive ion etching was carried out with the polymer pattern formed in Figure 20.9a

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

549

Infrared ray Pressing (3–4 bar)

PDMS block Rigiflex mold Polymer Substrate

Figure 20.8. Schematic illustration for the rapid flash patterning method. (Reprinted with permission of from [35] and American Institute of Physics.)

to etch into the underlying silicon layer. The plasma-etchant gas was CF4 at a rate of 15 sccm and a total pressure of 5.4 mTorr with a power of 300 W. The measured etch rate was 5 and 10 nm s–1 for the silicon and the PVA polymer, respectively. The result is shown in Figure 20.9c. Even though the etch rate of the polymer was twice the rate of silicon, the silicon was etched to a 200 nm depth because of the high aspect ratio of the patterned polymer. The essence of RFP lies in the use of a flexible, rubbery mold. When a hard mold is used as in the ultrafast technique [36], the substrate surface does not get exposure because the dewetting cannot take place due to the rigidity of the hard mold. Furthermore, any possible local contact of the hard mold with the underlying hard substrate, even if it is possible, would lead to breakage of one of the hard plates. The rigiflex mold used for RFP has a Young’s modulus of 0.3 GPa, compared to 107 GPa for silicon, that is high enough for fine patterning and yet low enough for flexibility. The RFP is an efficient and versatile patterning technique that has been developed to meet some of the patterning requirements for large-area applications. 20.3.3 Continuous Rigiflex Imprinting A continuous rigiflex imprinting (CRIM) [37] is another method by which the substrate surface can be made exposed in the course of the imprinting with rigiflex mold. This technique is suitable for mass production of nanometer scale patterns in that a “continuous” process could provide high throughput. There was an earlier study on making the imprinting a continuous process, called roller nanoimprint lithography (RNIL) [38]. The main issue in any form of RNIL, however, has to do with the damage to the hard master mold that can occur under high pressure. Any hard master mold with nanometer scale feature over a large area is quite expensive to fabricate and therefore such a mold is not suitable for a mass production process because of the damage problem. In contrast, if damaged in the imprinting process, the mold can economically be replaced because the rigiflex mold can be easily prepared from a master mold.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

550

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

(a)

(b)

(c)

Figure 20.9. PVA pattern formed by RFP in which the substrate surface is exposed (80-nm-wide lines and 270-nm space between the lines). The patterned 80-nm lines have a height of 440 nm. (a) SEM image of patterned polymer. (b) Cross-sectional SEM micrograph of the lines in (a). (c ) Silicon pattern etched with polymer resist of (a). (Reprinted with permission from [35] and American Institute of Physics.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

(a)

551

PDMS-coated roller Pressure

PUA mold with PET film

Cooling roller Polymer-coated substrate

Heating roller

(b)

(c)

P(t)

PDMS PET

V

PUA mold

W S

h0 Polymer

H

h

L Substrate

Figure 20.10. (a) Schematic illustrations of continuous rigiflex imprinting process. (b) SEM image of a flexible PUA (polyurethane acrylate) mold that was replicated from a hard mold. (c ) Schematic illustration of imprinting with rigiflex PUA mold. The PUA mold moves at a speed of V and the pressure is applied to the PDMS buffer layer. (Reprinted with permission from [37] and Elsevier Ltd.)

The CRIM process is shown schematically in Figure 20.10a. The roller system in the figure consists of two parts: two rollers for heating and another two rollers for cooling. In the heating part, only the bottom roll is heated by the heating wires that come into contact with the roller externally as it is rotated. While air cooling was sufficient for the cooling rollers, additional cooling could be used. The upper rollers were coated with PDMS. This PDMS layer provides a conformal contact with the rigiflex mold and a uniform pressure distribution, the flexible PDMS acting as a

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

552

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

buffer when pressed. The role of the cooling rollers is to maintain contact between the rigiflex mold and the substrate long enough for the temperature to go down below the glass transition temperature (T g ) of the polymer film being patterned so that no lateral collapse of the patterned features occurs. As pointed out earlier, the rigiflex mold in its film form is quite flexible as shown in Figure 20.10b but it is also rigid enough for imprinting. The imprinting that takes place as the mold comes into contact with the underlying substrate with a polymer film is illustrated in Figure 20.10c. It was stated earlier that the substrate surface gets exposed when imprinted by CRIM process. The condition for the exposure is that the pattern depth is large enough to accommodate all polymers on a substrate since otherwise the amount of polymer in excess of the void volume of the mold has no place to move into and remains on the substrate surface. To demonstrate this surface exposure, a line and space pattern, for which L = S = W = 500 nm, was used with an initial polymer film thickness of 100 nm. The patterning results are shown in Figure 20.11a. As shown, the substrate surface was exposed with the patterned line height of 200 nm.

(a)

(b)

Figure 20.11. SEM image showing the exposed substrate surface. (a) The polystyrene lines are 500 nm wide with a spacing of 500 nm. (b) Tapered cylinders (300 nm in diameter spaced by 400 nm) formed by CRIM without cooling rollers. (Reprinted with permission from [37] and Elsevier Ltd.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

553

Also shown in Figure 20.11b is a corrugated structure formed by CRIM without the cooling rollers. Although the polymer pattern collapsed partially after imprinting as intended for the corrugation, the substrate surface was also exposed. The process leading to the exposure of the substrate surface could be considered a two-step process. In the initial phase of process, the polymer melt is squeezed out of the space between the protruding parts of the mold and the substrate because of the pressure applied. When the remaining polymer film becomes sufficiently thin by the squeezing out, the second step of dewetting takes over, leaving behind dry spots or area. An idealized model in which the polymer melt is purely viscous and incompressible gives the time it takes, tf , for a film thickness initially at h0 to reduce to the final thickness hf as follows [39]: η(T )S 2 tf = 2p




1 1 − 2 2 hf h0

 (20.1)

where S is the mold width, P is the applied pressure, and η(T ) is the viscosity of the polymer melt at the temperature, T. According to the theory, it would take infinite time for the film thickness to be reduced to zero, i.e., hf = 0 when the surface is exposed. Therefore, the squeeze out cannot be responsible for the surface exposure. When a free liquid film on a solid substrate is sufficiently thin, dewetting takes place and the dewetting creates dry spots or area on the substrate [40]. Likewise, when the protruding part of the rubbery mold, such as rigiflex mold, approaches the substrate by the applied pressure such that the film thickness gets sufficiently small and the radius of this thin layer exceeds a critical value through a squeeze out, then a dry spot forms instantaneously and this dry spot expands outward at a speed of 1–10 µm s–1 [41]. Therefore, the film thickness decreases according to equation 20.1 up to a critical thickness and then the spontaneous dewetting takes over, making the surface exposed. This two-step process of thinning of the film followed by the spontaneous dewetting could explain why the substrate surface is exposed in a short time. 20.3.4 Soft Molding Application The soft molding process [42, 43] is a soft lithography technique by solvent-induced capillarity using soft elastomeric mold. Soft molding was developed to substantially reduce the high pressure needed for molding by using a soft mold that is capable of absorbing the solvent from sufficiently solvent-contained polymer film by the time the polymer is molded. However, the PDMS molds typically used in previous work on soft molding has generally difficulty in producing a densely arrayed nanopattern (e.g., alternating 100-nm line/space) because the PDMS mold collapses laterally as pointed out before. Presented here is soft molding method with a highly accurate rigiflex mold for multistep nanopattern fabrication with a high feature density [44]. More interestingly, the rigiflex mold softens dramatically upon reaching a temperature of 50◦ C whereas it is harder than the usual PDMS mold. Figure 20.12 illustrates schematically the soft molding process. First, a polyelectrolyte multilayer (polydiallydimethyl ammonium chloride/sulfonated polystyrene

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

554

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

Polyurethane mold Polyurethane mold

Resin solution (SPS/H2O) Adhesion promotion layer

(a) Multilayer (PDAC/SPS)

Soft molding

(b)

Peel off

Figure 20.12. An illustration of the procedure for fabricating a replica on a substrate from rigiflex polyurethane mold. SEM images of (a) rigiflex polyurethane mold and (b) a replicated polymer nanopattern on silicon substrate. (Reprinted with permission from [44] and IOP Publishing Ltd.)

(SPS) layer) as an adhesion promotion layer was formed onto a substrate via an alternating adsorption process [45, 46]. The warm rigiflex mold at 50–60◦ C is placed onto an SPS film immediately after the film is spin-coated onto an adhesion promotion layer from a 20 wt% SPS aqueous solution as a transfer layer. At the softening temperature, the mold becomes conformal with the underlying polymer film that is to be molded. At this point, the spin-coated film is still wet with residual water (25–35 wt% SPS wet films). The rigiflex mold is then pressed lightly onto the substrate of the SPS film at a pressure of less then 1 N cm–2 , and the substrate is allowed to remain in contact undisturbed for a period of time (∼20 min) for solidification on a hot plate at 50–60◦ C, during which time the solvent in the molded structure is absorbed into the mold. The rigiflex mold is then peeled off the substrate while hot.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

555

In this soft molding process, the solvent absorption of the rigiflex mold is an important factor. According to this experiment, the rigiflex mold made of polyurethane (PU) chemistry can absorb 1% of its weight in water at 50◦ C. About 90 wt% of the absorbed water diffuses out into air in 30 min at 50◦ C. The selection of 20 wt% aqueous solution of SPS as transfer layer was advantageous for the use of the rigiflex mold because the PU acrylate-based rigiflex mold is permeable to water vapor, providing for absorption and evaporation of the residual solvent under the mold conditions. A comparison of the rigiflex mold (Figure 20.12a) with the transferred nanopattern (Figure 20.12b) reveals that the pattern of the mold is well replicated in the SPS layer that is transferred to the substrate. Figure 20.13 shows the nanostructures transferred using this technique. Dense nanolines with a high aspect ratio are shown in Figures 20.13a and b. As shown in Figure 20.13b, the top linewidth is 80 nm (line spacing ∼220 nm) and the height is 400 nm. Figures 20.13c and d show multilevel and complex nanopatterns on a glass substrate. In the transfer of these dense, multilevel, and complex nanostructures with high aspect ratio, other techniques have some challenges to overcome, but

(a)

(b)

(c)

(d)

Figure 20.13. SEM images of replicated various polymer nanostructures. (a) Replicated polymer nanostructures on a Si wafer. (b) Magnified view of replicated nanostructure in (a) (the upper top width of the periodic nanoline is 80 nm and the height of the nanolines is 400 nm). (c , d ) Multilevel and complex nanostructures on a glass substrate. (Reprinted with permission from [44] and IOP Publishing Ltd.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

556

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

one can easily transfer a densely nanostructured polymer film to the substrate with good fidelity. These results show the effectiveness of this technique in dense, multilevel, and complex nanopattern fabrication such as periodic 80-nm lines with 400-nm height using a highly accurate rigiflex mold and a polyelectrolyte multilayer as an adhesion promotion layer. 20.3.5 Capillary Force Lithography Applications Recently, by combining essential features of nanoimprint and capillarity, the capillary force lithography (CFL) was developed for fabricating polymeric micro/ nanostructures over large areas [47–49]. In this method, one directly places a patterned elastomeric mold onto a spin-coated thermoplastic resin on a substrate and then produces a negative replica of the mold by raising the temperature above the resin’s glass transition temperature after solvent evaporation (temperature-induced capillarity) [47], which contrasts the direct soft molding prior to solvent evaporation (solvent-induced capillarity) described in the previous section [42, 43]. There has always been a prerequisite to the mold that is used for CFL, which is the permeability of the mold with respect to gas. If the mold is impermeable, the air trapped in the voids between the mold and the substrate surface gets compressed as the voids get filled by capillary action, which prevents the capillary rise. In this section, we present CFL in which impermeable or less permeable mold is used [50]. The driving force for the capillary rise in a capillary of radius r is the Laplace pressure PL that is given by PL =

2γ cos θ r

(20.2)

where γ is the surface tension and θ is the contact angle that the material filling up the pore makes with the capillary. If the capillary rise takes place at atmospheric pressure and the capillary rise is to a height of z in a capillary tube of length Z, the pressure of the trapped air in the tube Pa is given by Z /(Z − z) or 1/(1 − z m ) where zm is z/Z. The capillary rise will cease when Pa equals the Laplace pressure. Therefore, the normalized height zm to which the material of interest rises at atmospheric pressure (1 bar) is given by zm = 1 −

1 PL

(20.3)

where PL is in bar unit. The geometric factor r/2 in equation 20.2 is the ratio of the exposed surface area to the circumference that is in contact with the wall. When a channel infinitely long is involved, the geometric factor becomes L/2 where L is the channel width. Therefore, equation 20.2 applies to the channel if r is replaced by L. As the relationship shows, there is no capillary rise taking place if the Laplace pressure is less than 1 bar. On the other hand, equation 20.3 tells us that the Laplace pressure increases inversely with the capillary radius or equivalently with the feature

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

557

size of a pattern and that as the radius decreases, the pressure should exceed 1 bar. In fact, the Laplace pressure can far exceed atmospheric pressure if the radius or feature size is sufficiently small, ensuring that CFL should be possible with impermeable mold. When an impermeable hard mold is pressed against a hard substrate as in NIL, however, the pressure needed for the intimate contact all over the pressed areas is so high that it far exceeds the Laplace pressure because of the inherent roughness present on the mold as well as on the substrate. Therefore, the capillary force is still inconsequential. Accordingly, we used the rigiflex mold for resolving an intimate contact problem, which is rigid enough for fine patterning and yet flexible when it is made in a film form. In fact, this impermeable rigiflex mold has been shown to be rigid enough to be used for NIL. For the system under consideration where the CFL was carried out at 150◦ C with polystyrene (PS), the capillary pressure can be calculated from equation 20.2 with γPS−air = 40 mN m–1 and contact angle (θ = 56.6◦ ) of PS with the rigiflex mold as a function of the pattern size, resulting in exceeding the atmospheric pressure for the feature size smaller than about 440 nm. Therefore, the capillary rise should ensue at atmospheric pressure for the feature size smaller than 440 nm. The patterning result by CFL with the rigiflex mold is compared according to feature size in Figures 20.14a through d. As can be seen in Figures 20.14a and b,

(a)

(b)

(c)

(d )

Figure 20.14. SEM images of the masters and patterning results. (a) The master pattern is an array of slightly tapered cylinders (The bottom diameter is 6 µm and the height is 10 µm). (b) The patterning result by CFL with impermeable rigiflex mold replicated from the master in (a). It has a diameter of 5.78 µm and a depth at the center of 865 nm. (c ) The master pattern is a 80-nm line/60-nm space pattern made of aluminum. (d ) The patterning result by CFL with impermeable rigiflex mold replicated from the master in (c ). (Height is approximately 150 nm). (Reprinted with permission from [50] and American Institute of Physics.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

558

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

where the pattern of large feature size was used (bottom diameter of 6 µm and height of 10 µm), little capillary rise should have occurred because the diameter is not small enough for sufficient capillary pressure. In contrast, it can be seen from Figure 20.14d that the PS pattern formed has a height of approximately 150 nm where the master shown in Figure 20.14c was used, which is a 80-nm line/60-nm space pattern made of aluminum. For the 80-nm-wide channel in the rigiflex mold, the corresponding capillary pressure is about 5.5 bar. According to equation 20.3, capillary rise should be 139 nm since the channel depth is 170 nm. This calculated capillary rise compares with the experimental result of about 150 nm in Figure 20.14d. A simple capillary kinetic model has also been presented to describe the capillary rise of polymer melt into a less permeable mold [51]. The kinetic model predicts that the capillary rise is linearly proportional to the time for less permeable molds, which has been verified using rigiflex molds with poly(ethylene terephthalate) (PET) film as a supporting layer. It has been shown that for a given mold geometry, the height of nanostructure is tunable by changing the annealing time. Also, the tip shape can be rounded or dimpled depending on the temperature as a result of reflows of polymer melt. Figure 20.15 shows SEM images for the evolution of 150-nm nanopillars with time obtained at 120◦ C with poly(methylmethacrylate) (PMMA). It can be seen from the figure that the height or aspect ratio was proportional to annealing time and the PMMA film reached the ceiling of the mold after 60 min. At initial stage, the rate of capillary rise at the mold wall was faster than that at the center due to the presence of a meniscus as seen from Figures 20.15a–c. After reaching the ceiling of the mold, the PMMA melt reflowed to fill the entire space of the void in Figure 20.15d. The maximum height was measured to be ∼460 nm, in good agreement with the height of the original rigiflex mold (∼480 nm). For nanotribological applications, the effects of tip shape of nanostructure have been tested. It was found that both adhesion and friction forces were significantly reduced on the nanostructured surface due to the reduced contact area and lubrication effects at the tip. It is envisioned that this capillarity-assisted approach using less permeable rigiflex mold provides a convenient route to shape-engineering surface nanotopography for better nanotribological properties. 20.3.6 Transfer Fabrication Technique Fabrication of micro- and nanostructures by transferring a layer on a mold to a substrate has received much attention recently [52–54]. Transfer patterning method for the purpose of fabricating the inverted and embedded structures that is not readily applicable with conventional photolithography or unconventional lithographies are addressed here [55]. Shown in Figure 20.16 is the schematic representation of the technique. As shown in Figure 20.16a, the fabrication of an inverted structure starts with patterning the usual structure on a flat mold, which is then brought in contact with a substrate. With a slight pressing for intimate physical contact, the temperature is raised slightly above the glass transition temperature (T g ) of the patterned polymer. When the

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

20.3 APPLICATIONS OF RIGIFLEX MOLD

(a) t = 5 min

(c)

t = 30 min

(b)

559

t = 15 min

(d ) t = 60 min

Figure 20.15. Tilted SEM images for the evolution of 150-nm PMMA pillars with time at 120◦ C. A meniscus was seen from (a) to (c ) while the polymer completely filled the cavity after 60 min in (d ). (Reprinted with permission from [51] and American Institute of Physics.)

temperature is cooled down and the flat mold is removed, the pattern on the mold is transferred to the substrate due to larger work of adhesion with the substrate than with the flat mold, resulting in the inverted structure formed on the substrate. To accomplish the patterning on the flat mold, micromolding in capillaries (MIMIC) [56] or soft molding [42, 43] can be used. The method for fabricating embedded structure is the same as in the inverted structure except for the first patterning step. In this case, the flat mold surface should not be exposed and thus any of a number of techniques can be used for the patterning. The essence of the methods, as in any transfer patterning, lies in a difference in the work of adhesion between the two interfaces of the layer being transferred. Let W 12 be the work of adhesion [57] at the interface between the flat mold and the transfer layer, A12 its contact area, W 23 the work of adhesion at the interface between the transfer layer and the substrate, and A23 its contact area. Then, the following condition has to be satisfied for the transfer to be successful: W12 A12 < W23 A23

(20.4)

Shown in Figure 20.17 are the cross-sectional SEM micrographs of the embedded and inverted microstructures formed by the transfer fabrication. To obtain the

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

Flat mold

14:12

Printer Name: Yet to Come

Flat mold

Patterning

Patterning

Flat mold

Flat mold

Contacting with substrate

Contacting with substrate

Slight pressure Flat mold

Flat mold

Heating (T >Tg)

Heating (T >Tg)

Cooling and removing the flat mold

Cooling and removing the flat mold

(a)

(b)

Figure 20.16. Schematic illustration of transfer fabrication technique for (a) inverted structure and (b) embedded structure. (Reprinted with permission from [55] and American Institute of Physics.)

(a)

(b)

(c)

(d)

Figure 20.17. Embedded and inverted structures obtained by transfer fabrication. (a) Crosssectional SEM micrograph of an array of void boxes fabricated on silicon substrate. (b) Crosssectional SEM micrograph of a multilayer void channels fabricated on a thin film on silicon substrate. (c ) SEM micrograph of a trapezoidal pattern made by soft molding on flat rigiflex mold. (d ) SEM micrograph of the inverted structure that was transferred to PEDOT (poly(3,4ethylenedioxythiophene)) coated ITO (indium-tin oxide). The inset shows the edge of the structure. (Reprinted with permission from [55] and American Institute of Physics.)

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

REFERENCES

561

embedded microstructure in Figure 20.17a, an array of recessed boxes, each 3 µm × 3 µm in area with a depth of about 500 nm and with a separation of 2 µm between the boxes, was patterned by soft molding [26] of a photoresist (M w < 10,000, T g ∼ 80◦ C) on a flat PDMS film. This patterned box array was then transferred onto a bare silicon wafer surface with a native oxide by the transfer fabrication under an applied slight pressure at 90◦ C. It is quite satisfying that the transfer fabrication allows fabrication of three-dimensional embedded structures through multiple stacking. An example is shown in Figure 20.17b. Periodically spaced straight channels were formed following the same procedure as in Figure 20.17a. For transfer of a defect-free embedded nanostructure, in particular, a tough rigiflex mold rather than hPDMS as a flat mold could be more effective. An inversely tapered structure also was fabricated by the transfer fabrication. Shown in Figure 20.17c is the usual tapered structure patterned on a flat rigiflex mold that is supported by PET film. Soft molding [42, 43] was used to fabricate the structure on the flat mold with a PR (photoresist). This pattern was then transferred to the substrate, the result of which is shown in Figure 20.17d. The pattern formed by soft molding may have a very thin (∼10 nm) residual layer. However, the polymer can easily fracture along the contact line when slightly pressed, provided its molecular weight is less than about 10,000 [58]. Spin transfer printing [59] and rigiflex lithography [60] using a rigiflex mold have also been introduced for the transfer of complex multilayer nanostructures.

REFERENCES 1. Xia, Y. and Whitesides, G. M. (1998) Soft lithography. Angew. Chem. Int. Ed. 37, 550–575. 2. Xia, Y., Rogers, J. A., Paul, K. E., and Whitesides, G. M. (1999) Unconventional methods for fabricating and patterning nanostructures. Chem. Rev. 99, 1823–1848. 3. Kumar, A. and Whitesides, G. M. (1994) Patterned condensation figures as optical diffraction gratings. Science 263, 60–62. 4. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1995) Imprint of sub-25 nm vias and trenches in polymers. Appl. Phys. Lett. 67, 3114–3116. 5. Chou, S. Y., Krauss, P. R., and Renstrom, P. J. (1996) Imprint lithography with 25nanometer resolution. Science 272, 85–87. 6. Khang, D. Y., Yoon, H., and Lee, H. H. (2001) Room-temperature imprint lithography. Adv. Mater. 13, 749–752. 7. Kumar, A., Biebuyck, H. A., Abbott, N. L., and Whitesides, G. M. (1992) The use of self-assemble monolayers and a selective etch to generate patterned gold features. J. Am. Chem. Soc. 114, 9188–9189. 8. Love, J. C., Wolfe, D. B., Chabinyc, M. L., Paul, K. E., and Whitesides, G. M. (2002) Self-assembled monolayers of alkanethiolates on palladium are good etch resists. J. Am. Chem. Soc. 124, 1576–1577. 9. Hui, C. Y., Jagota, A., Lin, Y. Y., and Kramer, E. J. (2002) Constraints on microcontact printing imposed by stamp deformation. Langmuir 18, 1394–1407. 10. Delamarche, E., Schmid, H., Michel, B., and Biebuyck, H. (1997) Stability of molded polydimethylsiloxane microstructures. Adv. Mater. 9, 741–746.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

562

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

11. Lee, J. N., Park, C., and Whitesides, G. M. (2003) Solvent compatibility of PDMS-based microfluidic devices. Anal. Chem. 75, 6544–6554. 12. Bailey, T., Choi, B. J., Colburn, M., Meissi, M., Shaya, S., Ekerdt, J. G., and Wilson, C. G. (2000) Step and flash imprint lithography: template surface treatment and defect analysis. J. Vac. Sci. Technol. B 18, 3572–3577. 13. Khang, D. Y., Kang, H., Kim, T. I., and Lee, H. H. (2004) Low-pressure nanoimprint lithography. Nano Lett. 4, 633–637. 14. Schmid, H. and Michel, B. (2000) Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules 33, 3042–3049. 15. Odom, T. W., Love, J. C., Wolfe, D. B., Paul, K. E., and Whitesides, G. M. (2002) Improved pattern transfer in soft lithography using composite stamps. Langmuir 18, 5314–5320. 16. Gates, B. D. and Whitesides, G. M. (2003) Replication of vertical features smaller than 2 nm by soft lithography. J. Am. Chem. Soc. 125, 14986–14987. 17. Odom, T. W., Thalladi, V. R., Love, J. C., and Whitesides, G. M. (2002) Generation of 30–50 nm structures using easily fabricated, composite PDMS masks. J. Am. Chem. Soc. 124, 12112–12113. 18. Cscus, G., Kunzler, T., Feldman, K., Robin, F., and Spencer, N. D. (2003) Microcontact printing of macromolecules with submicrometer resolution by means of polyolefin stamps. Langmuir 19, 6104–6109. 19. Trimbach, D., Feldman, K., Spencer, N. D., Broer, D. J., and Bastiaansen, C. W. (2003) Block copolymer thermoplastic elastomers for microcontact printing. Langmuir 19, 10957–10961. 20. McClelland, G. M., Hart, M. W., Rettner, C. T., Best, M. E., Carter, K. R., and Terris, B. D. (2002) Nanoscale patterning of magnetic islands by imprint lithography using a flexible mold. Appl. Phys. Lett. 81, 1483–1485. 21. Rolland, J. P., Hagberg, E. C., Denison, G. M., Carter, K. R., and DeSimone, J. M. (2004) High-resolution soft lithography enabling materials for nanotechnologies. Angew. Chem. 116, 5920–5923. 22. Rolland, J. P., Van Dam, R. M., Schorzman, D. A, Quake, S. R., and DeSimone, J. M. (2004) Solvent-resistant photocurable liquid Teflon for microfludic device fabrication. J. Am. Chem. Soc. 126, 2322–2323. 23. Truong, T. T., Lin, R., Jeon, S., Lee, H. H., Maria, J., Gaur, A., Hua, F., Meinel, I., et al. (2007) Soft lithography using acryloxy perfluoropolyether composite stamps. Langmuir 23, 2898–2905. 24. Choi, K. M. and Rogers, J. A. (2003) A photocurable PDMS chemistry designed for soft lithographic molding and printing in the nanometer regime. J. Am. Chem. Soc. 125, 4060–4061. 25. Kang, H., Lee, J., Park, J., and Lee, H. H. (2006) An improved method of preparing composite PDMS molds. Nanotechnology 17, 197–200. 26. Choi, S. J., Yoo, P. J., Baek, S. J., Kim, T. W., and Lee, H. H. (2004) An ultra-violet curable mold for sub-100 nm lithography. J. Am. Chem. Soc. 126, 7744–7745. 27. Struck, S. and Heilen, W. (2002) Tego Journal, 2nd edn., Tego Chemie Service GmbH, Essen, pp. 68–70. 28. Decker, C. (2002) Kinetic study and new applications of UV radiation curing. Macromol. Rapid Commun. 23, 1067–1093.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

14:12

Printer Name: Yet to Come

REFERENCES

563

29. Yoo, P. J., Choi, S. J., Kim, J. H., Suh, D., Baek, S. J., Kim, T. W., and Lee, H. H. (2004) Unconventional patterning with a modulus tunable mold: from imprinting to microcontact printing. Chem. Mater. 16, 5000–5005. 30. Khang, D. Y. and Lee, H. H. (2004) Sub-100 nm patterning with an amorphous fluoropolymer mold. Langmuir 20, 2445–2448. 31. Xia, Y. and Whitesides, G. M. (1995) 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. 32. Suh, K. Y., Langer, R., and Lahann, J. (2003) Fabrication of elastomeric stamps with polymer-reinforced sidewalls via chemically selective vapor deposition polymerization of poly(p-xylylene). Appl. Phys. Lett. 83, 4250–4252. 33. Geissler, M., Wolf, H., Stutz, R., Delamarche, E., Grummt, U. W., Michel, B., and Bietsch, A. (2003) Nanocontact printing: a route to sub-50-nm-scale chemical and biological patterning. Langmuir 19, 1963–1965. 34. Delamarche, E., Schmid, H., Bietsch, A., Larsen, N. B., Rothuizen, H., Michel, B., and Biebuyck, H. (1998) Transport mechanisms of alkanethiols during microcontact printing on gold. J. Phys. Chem. B 102, 3324–3334. 35. Yoon, H., Lee, K. M., Khang, D. Y., and Lee, H. H. (2004) Rapid flash patterning of nanostructures. Appl. Phys. Lett. 85, 1793–1795. 36. Xia, Q., Keimel, C., Ge, H., Yu, Z., Wu, W., and Chou, S. Y. (2003) Ultrafast patterning of nanostructures in polymers using laser assisted nanoimprint lithography. Appl. Phys. Lett. 83, 4417–4419. 37. Seo, S. M., Kim, T. I., and Lee, H. H. (2007) Simple fabrication of nanostructures by continuous rigiflex imprinting. Microelectron. Eng. 84, 567–572. 38. Tan, H., Gilbertson, A., and Chou, S. Y. (1998) Roller nanoimprint lithography. J. Vac. Sci. Technol. B 16, 3926–3928. 39. Heyderman, L. J., Schift, H., David, C., Gobrecht, J., and Schweizer, T. (2000) Flow behavior of thin polymer films used for hot embossing lithography. Microelectron. Eng. 54, 229–245. 40. Xie, R., Karim, A., Douglas, J. F., Han, C. C., and Weiss, R. A. (1998) Spinodal dewetting of thin polymer films. Phys. Rev. Lett. 81, 1251–1254. 41. Martin, P. and Brochard-Wyart, F. (1998) Dewetting at soft interfaces. Phys. Rev. Lett. 80, 3296–3299. 42. Kim, Y. S., Suh, K. Y., and Lee, H. H. (2001) Fabrication of three-dimensional microstructures by soft molding. Appl. Phys. Lett. 79, 2285–2287. 43. Kim, Y. S., Park, J., and Lee, H. H. (2002) Three-dimensional pattern transfer and nanolithography: modified soft molding. Appl. Phys. Lett. 81, 1011–1013. 44. Kim, Y. S., Lee, H. H., and Hammond, P. T. (2003) High density nanostructure transfer in soft molding using polyurethane acrylate molds and polyelectrolyte multilayers. Nanotechnology 14, 1140–1144. 45. Lee, I., Zheng, H., Rubner, M. F., and Hammond, P. T. (2002) Controlled cluster size in patterned particle arrays via directed adsorption on confined surfaces. Adv. Mater. 14, 572–577. 46. Hammond, P. T. (2000) Recent explorations in electrostatic multilayer thin film assembly. Curr. Opin. Collid Interface Sci. 4, 430–442.

P1: OTA/XYZ P2: ABC c20 JWBK252/Lee September 18, 2008

564

14:12

Printer Name: Yet to Come

A RIGIFLEX MOLD AND ITS APPLICATIONS

47. Suh, K. Y., Kim, Y. S., and Lee, H. H. (2001) Capillary force lithography. Adv. Mater. 13, 1386–1390. 48. Suh, K. Y. and Lee, H. H. (2002) Capillary force lithography: large-area patterning, selforganization, and anisotropic dewetting. Adv. Funct. Mater. 12, 405–413. 49. Jeong, H. E. and Suh, K. Y. (2005) On the thickness uniformity of micropatterns of hyaluronic acid in a soft lithographic molding method. J. Appl. Phys. 97, 114701. 50. Yoon, H., Kim, T. I., Choi, S. J., Suh, K. Y., Kim, M. J., and Lee, H. H. (2006) Capillary force lithography with impermeable molds. Appl. Phys. Lett. 88, 254104. 51. Suh, K. Y., Jeong, H. E., Kim, D. H., Singh, R. A., and Yoon, E. S., (2006) Capillarityassisted fabrication of nanostructures using a less permeable mold for nanotribological applications. J. Appl. Phys. 100, 034303. 52. Rhee, J. and Lee, H. H. (2002) Patterning organic light-emitting diodes by cathode transfer. Appl. Phys. Lett. 81, 4165–4167. 53. Loo, Y. L., Willett, R. L., Baldwin, K. W., and Rogers, J. A. (2002) Additive, nanoscale patterning of metal films with a stamp and a surface chemistry mediated transfer process: applications in plastic electronics. Appl. Phys. Lett. 81, 562–564. 54. Hur, S. H., Khang, D. Y., Kocabas, C., and Rogers, J. A. (2004) Nanotransfer printing by use of noncovalent surface forces: applications to thin-film transistors that use singlewalled carbon nanotube networks and semiconducting polymers. Appl. Phys. Lett. 85, 5730–5732. 55. Seo, S. M., Kim, J. H., Kim, T. I., and Lee, H. H. (2006) Transfer fabrication technique for embedded and inverted micro/nanostructure. Appl. Phys. Lett. 88, 023118. 56. Kim, E., Xia, Y., and Whitesides, G. M. (1995) Polymer microstructures formed by moulding in capillaries. Nature 376, 581–583. 57. Wu, S. (1982) Poly Interface and Adhesion, Marcel Dekker, New York. 58. Seo, S. M., Park, J. Y., and Lee, H. H. (2005) Micropatterning of metal substrate by adhesive force lithography. Appl. Phys. Lett. 86, 133114. 59. Kim, Y. S., Baek, S. J., and Hammond, P. T. (2004) Physical and chemical nanostructure transfer in polymer spin-transfer printing. Adv. Mater. 15, 581–584. 60. Seo, D. C., Choi, S. J., and Lee, H. H. (2005) Rigiflex lithography for nanostructure transfer. Adv. Mater. 17, 1554–1560.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21 NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY Jong M. Kim, Hwan Y. Choi, Moon-G. Lee, Seungho Nam, Jin H. Kim, Seongmo Whang, Soo M. Lee, Byoung H. Cheong, Hyuk Kim, Ji M. Lee, and In T. Han

21.1 INTRODUCTION Nanoimprint (NI) technology, originally proposed by Chou in 1995, creates the pattern by stamping with inscribed specific image on a mold, generating fine micro/nano images and linewidth to construct various devices such as displays, micro-optics, high density patterned media, polarizing plates, and light guide plates (LGPs) with submicron wavelength (Figure 21.1). NI technology mechanically generates the pattern by pressing, curing, and demolding using an irregular pattern stamp, and therefore requires less energy consumption, considered to be more environmental friendly than the traditional lithographic processes. Also, it can reduce the cost by simplifying the manufacturing process and realize the facile implementation of three-dimensional shapes. NI technology can be classified into either a thermal type or an ultraviolet (UV) type, suggested by Haisma in 1996, depending on curing methods. The disadvantage of the thermal type is difficulty in the formation of a multilayer arrangement due to thermal deformation and high pressure (10–30 bar) requirement, whereas the UV type

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


565

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

566

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Resin

Mold

Pattern transfer completed

Figure 21.1. Nanoimprint.

has the advantage of processing at room temperature under lower pressure using an optically curable resin with a low viscosity (Figure 21.2). The nanoimprint technology has various applications in displays, semiconductors, and biotechnologies (Figure 21.3). The nanoimprint technology has gained more attention in the field of display application as the size of the original glass substrate is getting larger. In order to commercialize the nanoimprint technology, several issues must be addressed first, such as mold manufacturing method, durability, repeatable demolding property, the development of related materials, transferring patterns without defects, accurate and reproducible alignment, and applicability to a larger substrate, and many research centers are actively participating in the development of technology to overcome these issues (Figure 21.4). In Korea, LG Electronics is developing a polarizing film for the projection TV and LED elements using the nanoimprint technology. LG Philips LCD (LPL) in cooperation with LG Chemistry is conducting researches on the application of the nanoimprint technology for the LCD production. Samsung Electronics is in the process of developing components for their display products using nanoimprint technology for transferring pattern. LCD has been extensively accepted by the market for several advantages and continuous performance improvement. It is safe to say that LCD has secured a firm position for mobile device application with its thin and small size and low energy consumption. The LCD panel requires a back light unit (BLU) for the panel lighting. The BLU is a core part of LCD, of which requirements are to be mechanically thin, bright to increase luminance, and small for less power consumption. The BLU consists of a light source and an optical system controlling the light properties (refraction, reflection, diffraction, polarization, etc.). The conventional BLU consists of the following components: a light source such as cold cathode fluorescence lamp (CCFL), a LGP with a pattern for the complete reflection of the light and a partial dispersion of the light to the front, a diffuser sheet to hide the pattern and equalizes the light from the LGP, a prism sheet for concentrating the light to the front, a reflective polarizer (such as DBEF (dual brightness enhancement film) from 3M Company) for transmitting a linearly polarized light and reflecting the vertical linearly polarized light, and a reflector at the bottom of the LGP (Figure 21.5).

Cooling and demolding

Thermosetting type

Substrate

Mold

Pressing

Heating and embossing

Figure 21.2. Nanoimprint curing type.

Polymer deposition and alignment

Substrate

Mold

Substrate

Mold

UV

Substrate

Mold

Substrate

Mold

UV curing type

Demolding

UV exposure

Soft contact aligning and UV exeposure

Polymer dispensing

Relesaing layer

14:13

Substrate

Pressure

Substrate

Mold

Polymer Substrate

Mold

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008 Printer Name: Yet to Come

567

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

568

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Figure 21.3. Application of Nanoimprint (Reprinted with permission from [1], Copyright 2005 Institute of Physics Publishing; from [2], Copyright 2003 American Institute of Physics; from [3, 4], Copyright 1997 American Institute of Physics; and from [5], Copyright 2004 Nikkei Electronics.)

Figure 21.4. Nanoimprint consortium.

Rear polarizer S

P

S P

S

Reflective polarizer Prism sheet Diffuser sheet P+S

Lamp

P+S

P+S

LGP

Reflector Figure 21.5. Conventional BLU.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.2 HOLOGRAPHIC LGP

569

The light guide and the polarizing plates in an optical system typically have a regular pattern arrangement in sub-micrometer range, and the light properties can be adjusted by controlling the pattern and the frequency. Also, the optical system with the current NI technology makes the implementation possible since most of the optical systems do not include the overlay function and do not require a precise alignment, and maximizes the contribution of NI technology to the simple implementing process for a large area display. The LGP, among many other applications, is considered to have the best potential for the application of NI technology due to the increase in light efficiency and the requirement for a thinner component. It is well known that the light efficiency of the LCD is 5–7%, which is quite low. The main reason for the low efficiency is the limited use of the polarized light in one direction by the light crystal switching element. In other words, the existing LCD uses the absorbing polarizing plate where the absorbing polarizer transmits a linearly polarized element in one direction and the other polarizing element is absorbed by the iodine aligned in the direction of the polarizing plate. In this case, the transmitting efficiency is only about 43%. The reflective polarizer is introduced to improve the efficiency up to 150% or more by transmitting the polarized light in one direction, reflecting the other polarized light to the backlight, and rereflecting the polarized light from the backlight from the back of the LCD panel. This method improves the efficiency of light use, in addition to the reduction of power consumption in the mobile displays such as in the laptop computers, PMP (portable multimedia player) and PDA’s (personal digital assistants). This section will illustrate several examples of NI technology application in the holographic LGP by realizing the hologram pattern to improve the light use efficiency, the polarized LGP to separate the polarized light, and the reflective polarizer as well as in the reflective plate for the transflective LCD.

21.2 HOLOGRAPHIC LGP The LGP, a part of the optical system in LCD BLU, disperses the incident light from the light source. Recently, the addition of a prism sheet or the polarizing function to the LGP has been studied. It is concluded from these studies that the power consumption can be reduced using a hologram element [6–12] and the dispersed light can be vertically collected [13]. Figure 21.6 shows the comparison between a conventional light guide light plate and a holographic LGP. This hologram pattern is implemented via the formation of a submicrometer sized structure on the surface of the LGP, and the incident light on the hologram is diffracted and outcoupled to the LGP, depending on the diffraction property of the surface-deformed hologram. The hologram pattern formed on the LGP should be optimized to improve the outcoupling efficiency. The technology applied to the holographic LGP largely consists of an optical design technology for a light guide, a master pattern manufacturing technology, a nanoimprint and copy technology, and an evaluation technology. Fabrication of a high quality LGP is possible only when these various technologies are interconnected systematically. This section describes the geometrical structure of the

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

570

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

(a) LCD Panel Diffuser 2 Prism V Prism H Diffuser 1 LGP Reflector

Conventional light guide plate

(b)

Holographic light guide plate Figure 21.6. (a) Conventional light guide plate and (b) holographic light guide plate.

holographic LGP, implementing the structure using NI technology, and the property of the fabricated LGP. 21.2.1 Design and Properties of Holographic LGP The light exiting from a holographic LGP is influenced by the wavelength, the incident angle, the polarization of the incident light, and the period and the shape of the diffraction grating of the hologram pattern. Figure 21.7 shows a LGP structure for a mobile display consisting of a holographic LGP and LED. The diffraction grating of this holographic LGP consists of a sinusoidal shape with 0.45 µm period. Statistically, the strongest incident light from the LED source is at incident angle of about 55◦ . This incident light from LED is optically coupled to the diffraction grating and emitted through the LGP vertically. Figure 21.8 shows the results of the angular luminance distribution. Near LED

Far LED

LED LGP

Hologram Figure 21.7. Light guide plate structure for the mobile display.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.2 HOLOGRAPHIC LGP

(a)

571

(b)

Figure 21.8. Angular luminance distribution of Holographic light guide plate (a) area near to the LED (b) area far from the LED.

The brightness shows a ununiformity depending on the location of the LGP due to energy being concentrated on the center of the angular luminance distribution for the points far from the LED. One of the major objectives of the LGP design is elimination of the ununiformity in the brightness of a LGP. The holographic LGP is based on the diffraction property of the surface-deformed diffraction grating. Figure 21.9 shows the diffraction efficiency plotted against the pattern depth of a fine diffraction grating with the sinusoidal shape with a period of 0.45 µm and the wavelength of the incident light. (This result is obtained at the 55◦ incident angle, TE refers to transverse electric wave and TM refers to transverse magnetic wave.) Since the period of the diffraction grating is small and similar to the wavelength of the light, the coupling efficiency should be calculated based on the vector diffraction theory [14–16]. It can be seen that the calculated result and the measured result are in good agreement. From the result, it can be concluded that 0.2–0.3 µm is the appropriate

(a) Mesured result

(b) Calculated result

Figure 21.9. First diffraction efficiency of sinusoidal shape hologram. (a) Measured result (b) calculated result.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

572

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Master HOE recording

He-Cd laser Master HOE

PR Glass sub

PR coating

Glass sub

Glass sub

Glass sub

Prebaking

Exposing

Developing

Figure 21.10. Hologram master fabrication process.

depth of diffraction grating, producing the most even uniformity and maximum efficiency. 21.2.2 NI Technology for the Holographic LGP The fabrication process for the holographic LGP is divided into two processes: fabrication of the master hologram pattern and application of the fabricated master to NI technology. The master fabrication process consists of applying the photoresist on the glass substrate followed by setting it thermally. After dividing a laser light with good interfering property into two beams, the beams are allowed to interfere on the photoresist. The two laser lights interfere to create local light intensity modulation by generating an area with alternative strong and weak light intensity. The degree of photoresist cured varies depending on the light intensity and the position allowing the hologram master with the sinusoidal shape to be fabricated on the surface of the photoresist (Figure 21.10). The fabricated hologram master is coated with Ni by sputtering and plating processes for fabricating the imprinting mold. The photopolymer is applied to the fabricated nickel mold (or nickel shim) and imprint the pattern on the glass substrate. With the glass substrate mounted, the hologram pattern of the nickel mold is imprinted and copied to the polymer on the glass substrate by applying a proper pressure and UV light. For large-scale manufacturing process in companies, the massive production of the nickel mold should be feasible and an injection molding should be used to improve productivity (Figure 21.11). UV embossing

Injection machine

UV-radiation

Glass substrate UV-curable photopolymer Ni shim Figure 21.11. Process for imprinting the holographic light guide plate.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.3 POLARIZED LGP

573

X = 0.4360 µm Y = 0.1120 µm D = 0.4500 µm

1.00 µm

Figure 21.12. Hologram pattern formed with the nanoimprint (SEM view).

Figure 21.12 shows the electron microscope image of the hologram pattern formed by NI on the surface of the holographic LGP. Utilization of the existing geometric optics has limits of controlling visible light rays in the 400–700-nm wavelength range. For this reason, the diffraction phenomenon and the local field enhancement phenomenon, which are beyond the diffraction phenomenon, are important application areas of optical components. In order to realize these phenomena, it is essential to design and manufacture structures with the size of visible light wavelengths or smaller, and to develop the optical material with high transmissivity and excellent demolding property for the nanoimprint. Although challenges to achieve reliability and mass productivity for commercialization are still remained, there would be no questions about NI technology considered as possible alternatives for patterning and copying the structures smaller than the wavelength.

21.3 POLARIZED LGP In the conventional BLU, more than 50% of the unpolarized beam is absorbed into the rear polarizer. This absorption is one of the major drawbacks for the low light efficiency of about 5% in the LCDs. Many studies have been demonstrated to provide different polarized light in the BLU to increase the light use efficiency and to achieve higher brightness and lower power consumption. Commercially available and widely adopted approach is the usage of reflective polarizers transmitting one linear or circular polarized light and reflecting its orthogonally polarized light [17, 18]. However, such reflective polarizer is expensive and results in poor recycling efficiency of light, thus limiting the overall efficiency of the BLU. Another approach is to integrate both the polarizing function and the recycling function in the LGP to separate one linear polarized light within the LGP and reuses the orthogonally polarized light. In the previous studies, they have either formed the

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

574

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

microstructure in the inclined surface of the LGP or used the polarizing beam splitter with the multilayered thin film to separate the polarized light satisfying the Brewster angle [19, 20]. However, the disadvantages of this method are intricate fabrication process and heavy dependency on the incident angle and the wavelength. Another type of polarized LGP, based on the selective total internal reflection (TIR) at a microstructured interface between isotropic and anisotropic layers, was also proposed [21–23]. However, only the feasibility was demonstrated in a small-scale sample and the microstructures were not optimized. Furthermore, the manufacturing process requiring diamond-cutting anisotropic material [21, 22] is not easily scalable for large LGPs, and the lamination using the liquid crystalline polymer [23] is not cost effective and may cause delamination resulting in poor reliability. In order to overcome the points at issue, a highly efficient BLU prototype with a polarization-separating anisotropic layer (AL) was developed. The microstructure on the uniaxially drawn polymer film by the hot-embossing process was replicated. The measured optical performances show that the BLU with the asymmetric sawtooth microstructures produces the polarized light having improved luminance along the normal direction by reducing the stray intensity at high polar angles. The BLU has over 30% higher luminance and integrated intensity than the BLU adopting a reflective polarizer. The LGP design, the LGP fabrication process, and the optical performance results are described in the following subsections. By applying the micro/nanoimprinting technology to the fabrication of the polarized LGP, the problems of existing processes such as difficulties in scaleup for larger LGPs and reliability issue were cleared. Furthermore, the manufacturing process was proved to be cost effective because the overall steps of uniaxial stretching, hot embossing, and lamination of the AL have the potential of an in-line roll-to-roll process. 21.3.1 Design and Properties of Polarized LGP Figure 21.13 shows the schematic drawing of the polarized LGP. This polarized LGP consists of a CCFL as a light source, a back reflector, and a flat LGP on which a microstructure AL is laminated using a photocurable isotropic adhesive for coupling the isotropic LGP and the AL (Figure 21.13). The extraordinary refractive index ne of the coupled AL, is aligned to the s-polarization direction in the LGP. The goal is to replace the optical films such as a diffuser sheet, one or two prism sheet(s) and a reflective polarizer with the polarized Rear polarizer

Reflective polarizer Prism sheet Diffuser sheet

S

S P

P+S P+S Lamp

P P+S

S

Rear polarizer no = 1.53 Anisotropic layer no ne = 1.66 LGP

Reflector

Figure 21.13. Schematic of the polarized BLU.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.3 POLARIZED LGP

575

LGP. The unpolarized light from the light source is selectively separated at the interface between the LGP and the microstructured AL and supplied to the liquid crystal panel. The Monte Carlo ray tracing method for the polarized light in the anisotropic material is utilized for numerical simulations of the optical performances such as the polarization separation, the total amount of extracted light and the angular distribution of luminance. As stated above, only the extraordinary rays experience the TIR at the microstructured interface between the isotropic layer and the AL and are extracted from the AL. The uniaxially drawn polyethylene terephthalate (PET) is used as anisotropic material due to the cost competitiveness and the adequate birefringence of the material. The extraordinary refractive index and the ordinary refractive indices of the uniaxially drawn PET, close to the uniaxial crystal structure, are ne = 1.67, no1 = 1.56, and no2 = 1.54, respectively. The photocurable adhesive layer has the refractive index (nadh = 1.50) similar to the poly(methylmethacrylate) (PMMA) LGP (nLGP = 1.49). Because the ordinary light has almost no refractive index difference at the interface, it is continuously guided by the TIR within LGP. All related refractive indices are measured using the prism coupler refractometer (2010M, Metricon) at the 532 nm wavelength. The earlier works [21–23] using the AL with the single prism array show the increased luminance along the normal direction, but suffer from serious drawback such as the stray light flux produced at high polar angles. The stray light is caused by the lights slightly deviated from the TIR condition at the microstructure interface. These lights are passed through the prism structure and out coupled at the high polar angle of the LGP. In the case of the out coupled light with its luminance peak at a high polar angle, the BLU should adopt additional optical films to redirect that flux to the normal direction. The additional optical films reduce the light use efficiency, in addition to increase in the total cost of the BLU. A new type of the AL with an asymmetric sawtooth dual structure was introduced to reduce the stray rays and to increase luminance along the normal direction [24]. In this dual structure, the light passing through the first prism can be redirected to the normal direction by the adjacent higher prism. With the definition of the period and the heights of the dual microstructure as
, h1 , and h2 , respectively, numerical simulations were performed to calculate the distribution of the luminous intensity as a function of the polar angle at the central wavelength of 550 nm. The values for the geometrical parameters from the simulations were
= 60–30 µm, h1 = 12 µm, and h2 = 20 µm, respectively. All apex angles of the prisms were optimally designed to be 50◦ . It is found that at high polar angles above θ = 40◦ , the luminous intensity for the asymmetric dual structure array becomes significantly smaller than that for the conventional single prism array. 21.3.2 Fabrication of the Polarized LGP The newly designed LGP in 7-in. diagonal is fabricated based on the simulation results as described above. The uniaxially drawn PET film with about 100 µm thickness is used as the polarization-separating AL. The asymmetric microstructure is formed on the PET via the hot-embossing process using the stamp with the pattern

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Temperature (degree)

250

350 300

200

250 150

200

100

150 100

Temperature

50

Pressure (kg /cm–2)

576

14:13

50

Pressure

0

0 0

3

6

9

12

15

Time (min) (a)

20 cm (b)

Figure 21.14. (a) Schematic profile of hot-embossing process. (b) Cross-sectional view of PET film with the asymmetric dual sawtooth microstructure formed.

of the designed microstructures [25–27]. The line-shaped features of the stamp were positioned parallel to the orientation direction of the PET film. The basic process of the hot embossing is that a polymer substrate is heated above its glass transition temperature (T g , about 80◦ C for the PET), followed by pressing the stamp against the substrate for the complete transfer of the patterns onto the substrate. After holding for a specific time, the system is cooled below T g , and the stamp is separated from the substrate. It is important that the film should not be contracted or the order of the oriented molecules should not be relaxed by heat during the process. The schematic profile of the hot embossing process is shown in Figure 21.14a. The hot embossing was performed for about 5 min under about 290 kg cm−2 at 200◦ C. The PET anisotropic film with uniformly embossed microstructures over the entire area could be obtained by the optimized process. The optical microscope cross-section image of the hot-embossed PET film with the asymmetric dual microstructure is shown in Figure 21.14b. Measured values for the embossed microstructures are period
= 30 µm, and heights of h1 = 12 µm and h2 = 20 µm. These values indicate that the microstructured formed on the stamp are completely transferred to the film. The hotembossed PET film is then laminated onto the PMMA LGP using an adhesive resin. The adhesive layer was subsequently cured by the exposure to UV light. 21.3.3 Optical Performance of the Polarized LGP The optical performances of the polarized BLU with the polarization-separating AL in the CCFL edge-lit configuration were measured using the conoscope

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.3 POLARIZED LGP

(a)

577

(b)

Figure 21.15. Angular luminance distribution of the fabricated polarized BLU (a) s-polarization (b) p-polarization. (Reprinted with permission from [24]. Copyright 2007 Society for Information Display.)

(EZContrast 160R, Eldim). No additional optical film except the white paper reflector at the bottom of the BLU was used. The absorbing polarizer was placed between the detector assembled to the conoscope and the BLU. The transmission axis of the polarizer is either in parallel or perpendicular to the optic axis of the AL. The angular luminance distribution L(ϕ, θ ) for the azimuth angle 0◦ –360◦ and the polar angle 0◦ –80◦ was measured. The angular luminance distributions of the polarized BLU for the s-polarization and the p-polarization are shown in Figures 21.15a and b, respectively. It is clear that the fabricated BLU produces significantly more s-polarized light (when the transmission axis of the polarizer is parallel to the optic axis of the AL) than the orthogonal p-polarized light along the normal direction. The measured luminances for the s-polarized light and the p-polarized light along the normal direction are 3060 and 347 cd m–2 , respectively. Thus, the ratio of s-polarized luminance to p-polarized luminance is approximately 9:1. Note that in the BLU configuration discussed here, the light absorption through the rear polarizer can be minimized by using the s-polarized light for the LCD illumination. In order to determine the effectiveness of the AL with asymmetric sawtooth dual structures, the measured normalized luminous intensity for the s-polarized light is compared with that for the light through a conventional single prism as a function of the polar angle. The total luminous intensity at high polar angles is over θ = 40◦ for the conventional single prism is clearly reduced by more than 50% when the approach discussed here is used. This is due to the redirection of the stray rays propagating through the first prism to the normal direction of the AL by the adjacent second prism. And these results are in good agreement with the simulation results. The reduction of the stray light increases the luminance along the normal direction, contributing directly to the improvement of the contrast in LCD using the polarizing BLU with the asymmetric dual microstructure formed. The luminance for both along the normal direction and the integrated intensity over all angles in the 7-in. polarized BLU prototype is at least 30% higher than those of a conventional BLU consisting of an LGP, a diffuser sheet, a prism sheet, and a reflective polarizer such

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

578

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

as DBEF (3M Company). It means that the polarized BLU with the asymmetric dual microstructure increases the light use efficiency by eliminating the optical films and directly producing the polarized light through an LGP along the direction normal to the AL. The highly efficient 7-in. BLU prototype based on the AL with polarizationseparating microstructures was designed and fabricated. The BLU with the AL of the newly designed asymmetric dual microstructure produces the polarized beam with the improved luminance along the normal direction by reducing the stray intensity at high polar angles. The measured optical performances show that the BLU has over 30% higher luminance and integrated intensity than a conventional BLU adopting a reflective polarizer. The manufacturing process is quite simple and cost effective. Also, because the overall steps of uniaxially drawing, hot embossing, and lamination of the AL have the potential of the in-line roll-to-roll process, it presents the facile applicabilities for larger LGPs used in the LCD monitors and the like. In addition to the application shown in this chapter for the fabrication of the microstructures, the formation of the micro/nanostructure is one of the core processes in fabrication of the conventional or polarized LGP. Thus, there are still large field of applications open for nanoimprint technology. For example, it is possible to form the alignment layer of the liquid crystal as shown in Figure 21.16 via the imprint method

(a) Imprinter I

Imprinter II

Press and UV exposure UV-curable polymer Plastic substrate

Press and UV exposure UV-curable polymer Plastic substrate

Remove imprinter I

Bottom substrate

Remove imprinter II

Top substrate

(b) Plastic substrate 1 µm

Microgrooves Liquid crystal Spacer 100 µm

Plastic substrate Figure 21.16. Forming the liquid crystal alignment layer by using the imprinting technology. (a) Imprint process using the bilevel imprinter. (b) Schematic drawing of fabricated liquid crystal cell and magnified image of microstructure. (Reprinted with permission from [28]. Copyright 2006 American Institute of Physics.)

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.4 REFLECTIVE POLARIZER: WIRE GRID POLARIZER

579

no no ne

Figure 21.17. Polarized LGP using the imprinting technology and the photocurable liquid crystal polymer.

using the stamp with the micro/nano structure formed [28]. This imprint method, by forming stamp using the elastic material such as poly(dimethylsiloxane) (PDMS), can be effectively applicable to the plastic substrate as well as the thin LGP. The thin-type polarized LGP as briefly shown in Figure 21.17 can be more easily fabricated by using the liquid crystal alignment method with this imprint technology. First, by using the PDMS or others, the stamp with both the hundreds of nanometers to one-micrometer microstructure for aligning the liquid crystal (e.g., grating structure with the right angle or the sinusoidal function) and the structure for separating the polarization (e.g., triangular prism shape) is fabricated. The master mold for the stamp is fabricated just once, and the stamp can be used repeatedly. And then, coat the UV-curable resin with the refractive index similar to the LGP on the LGP, locate the stamp on the top side, expose the UV, and then separate the stamp. Upon coating, i.e., spin-coating, the photocurable liquid crystalline reactive mesogen (RM) on the UV-cured microstructure is formed with a proper thickness; the RM is aligned in accordance with the aligning direction of the microstructure. After that, when the UV light with a specific wavelength is applied, the RM is polymerized and cured. With this method, it is possible to form the AL with the polarization separating function on the conventional LGP with several micrometer thicknesses. As stated above, this imprint technology can be easily applied to the ultrathin LGP and the flexible LGP whose demand is recently being increased.

21.4 REFLECTIVE POLARIZER: WIRE GRID POLARIZER The wire grid polarizer (WGP), a reflective polarizer, has the potential to replace the existing absorbing polarizer as a core optical element in the micro display-based projectors, LCD, military applications, x-ray imaging systems, ellipsometry, and spectrometer. If it can replace the existing absorbing polarizer for the LCD, it can be used as the alternative method in improvement of the low light use efficiency, which is an inherent drawback of a LCD. Furthermore, if it is implemented as an in-cell polarizer taking advantage of its superiority of being thinner than the existing absorbing

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

580

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

polarizer, it is expected to have a far-reaching effect. However, even with the advantages and expectations, its use has been limited to the infrared area due to the limitations of existing process technology, which has been evolving for over 40 years since it was first introduced in 1970s. However, the development of lithography technology and NI technology in the mid 1990s made it possible to implement a microwidth wire to be used in the visible light spectrum. Specifically, patents on the WGP submitted by leading companies such as Sony in 1995 and Moxtek in 1998 has initiated the accelerating development of the WGP at the visible light spectrum for the LCD. 21.4.1 Design and Properies of WGP Perhaps, WGP is the first form birefringence element using a metal grating smaller than wavelength. WGP consists of an infinitely long and uniform metal wire grating with a period less than the incident wavelength. If the metal grating is much smaller than the incident wavelength, it exhibits a polarization separating property instead of a general diffraction phenomenon. When the electromagnetic wave is applied to this structure, the vibrating electric field induces the electric dipole at the surface of WGP. The free electrons freely vibrate under the influence of incident light and reradiate, in other words, reflected. However, the light is transmitted at the dielectric space with no free electrons. As shown at Figure 21.18, the electrons at WGP freely vibrate along the metal wire while the light vertical to the grating is transmitted for invisible structure due to the grating width smaller than the wavelength. The polarized light parallel to the wire (in case of the TE polarization) is reflected because it senses the structure as a conductor while the vertically polarized light (in case of the TM polarization) is transmitted because it sensed the structure as a dielectric resulting in the polarizing separation [29]. WGP that is possible to drive in the entire area of 380–750 nm visible light spectrum, which is meaningful as a display element, can have a property of 2000:1 polarizing extinction ratio or higher and 85% transmittance or higher only when it has a metal wire grating structure with 100 nm wire width or

Incident light

S-pol.

Reflected light

P-pol.

S-pol.

P-pol. Transmitted light Figure 21.18. Structure of wire grid polarizer.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.4 REFLECTIVE POLARIZER: WIRE GRID POLARIZER

581

1.E+07

Extinction ratio

1.E+06 1.E+05 1.E+04 650 nm

1.E+03 100 nm

1.E+02 1.E+01 40

550 nm 450 nm

80

120 160 200 Grating pitch (nm)

240

Figure 21.19. Polarizing extinction ratio according to the period by Al metal wire rating wavelength. (Reprinted with permission from [1]. Copyright 2005 Institute of Physics Publishing.)

thinner and 3:1 aspect ratio or higher. These properties imply a level equal to or higher than the existing absorbing polarizer. As shown in Figure 21.19, the period of the pattern should be 100 nm or less at the minimum in order to have a high polarizing extinction ratio for the entire wavelength of R, G, B visible light spectrum. Figure 21.19 shows the value derived by G-solver, a commercial diffusion grating simulation tool, based on the rigorous coupled wave analysis [30]. In this simulation, the height of the Al metal wire is 140 nm and the duty cycle is fixed to 0.5 [1]. However, if the grating period becomes relatively larger, the Rayleigh resonance, a diffusion property, occurs at the condition shown in equation 21.1, which may prevent it from being used as a polarizer. In the equation 21.1,
is the period of the grating, λ is the incident wavelength, η is the refractive index of the substrate, and θ is the incident angle of the incident wavelength. Figure 21.20 shows the simulation results for identifying the required minimum period to avoid Rayleigh resonance using G-solver.
=

λ (n + sin θ )

(21.1)

21.4.2 Fabrication and Applications To achieve commercialization, three technologies are necessary to fabricate a high performance WGP—a mold fabricating process, a pattern transferring process, and a metal wire grating forming process. In order to form a stable WGP, the metal wire grating pattern with a period of 100 nm or less is required. First of all, the electron-beam (e-beam) lithography for fabricating the nanowire grating, the interference lithography, and EUV (extreme ultraviolet radiation) lithography are used to fabricate the master pattern, followed by the transfer of the pattern using the fabricated mater pattern in NI technology [31, 32]. The step and flash process is used to

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

582

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Limits of Rayleigh resonance 500

650 550 450 350

450

Period (nm)

400 350 300 250 200 150 100 50 0 0

20

40 60 Incident angle

80

Figure 21.20. Rayleigh resonance condition by period according to incident wavelength.

make a large-area process simple for lfabricating WGP. The step and flash Lithography (S-FIL) process, a method in forming a large-area micro pattern with the imprint process, is a method for making a large area by performing NI process repeatedly [33]. This method is based on the UV-curable material with the low viscosity. Also, the S-FIL process coats an UV-curable material for the step and repeat patterning by field-to-field drop dispensing method (Figures 21.21 and 21.22). The paper presented by N. Khusnatdinov et al. shows WGP fabricating example by S-FIL method, using a dry etching process to fabricate a metal pattern with 100-nm period and aspect ratio

(b) Lower template into resist

(a) Dispense droplets of resist

Step and repeat

(d ) Separate template and wafer

(c) UV illuminates to cure resist

Figure 21.21. S-FIL process for making the 10 mm × 10 mm area by the step and repeat process. (Reprinted with permission from http://www.molecularimprints.com. Copyright 2005 Molecularimprints Co.]

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.4 REFLECTIVE POLARIZER: WIRE GRID POLARIZER

583

Figure 21.22. Real sample fabricated by the S-FIL process. (Reprinted with permission from [33]. Copyright 2006 SPIE.)

of 4:1 (Figure 21.23). Finally, the formation of metal wire grating mainly depends on an oblique-angle deposition on the side or on the top of the patterned NI pattern or a dry etching of the previously formed metal deposition layer by using the imprint pattern as a mask pattern. The structure and the physical properties of the metal grating are important in influencing the optical properties of WGP. Major WGP manufacturing companies are Moxtek, LG Electronics, NanoOpto, and Asahi Kasei. Among them, Proflux polarizer from Moxtek is the only WGP

Figure 21.23. Cross-sectional view of the WGP with the Al layer of the substrate etched with 4:1 aspect ratio by using the nanoimprint pattern fabricated by the S-FIL process.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

584

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Figure 21.24. Cross-sectional view of Moxtek WGP. (Reprinted with permission from [35]. Copyright 2005 Moxtek Co.)

commercialized product for the display application in the visible light spectrum [34] and the aluminum nanowire is fabricated by the e-beam lithography and the dry etching process. It has 85–90% transmissivity and 400:1–1000:1 extinction ratio for the TM polarization. The commercialized size ranges from 10 mm × 10 mm to 140 mm × 140 mm and the antireflection coating is applied as required [35]. The Figure 21.24 shows the cross-sectional view of the fabricated WGP. This section describes the development trend of WGP by using NI technology. The structure and the physical property of this metal grating formed with various methods are important to influence the optical property of WGP. In case of an oblique-angle deposition used mostly in the early period, an absorption and dispersion within the metal due to an incomplete shape caused degradation of optical property as a polarizer. The issues for design and processes still remain to be seen and further optimization is required. In addition, utilization of products as a polarizing element or as an optical component in a display requires easy implementation to a larger area pattern from several centimeters at a minimum to a several tens of centimeters. Since the pattern formed from NI is directly reflected at the metal deposition or etching process, it is critical to establish uniformity. As stated in the previous section, this S-FIL process is proposed as an alternative for a larger area, but the problems such as a misalignment between patterns and a specific gap caused by the alignment limits still remains to be solved. Recently, Asahi Kasei Corp. presented a 50-cm-wide WGP production plan via roll-to-roll process at the 2007 FPD International. It also exhibited a film-type trial sample using actual production equipment. The wire grating pattern was formed by NI process using an actual fabricating method with little difference from the one presented in 2005 and then the metal layer was selectively formed on the wire grating pattern by an oblique-angle deposition. The pattern had about 100–150-nm period and dual structure with 150-nm polymer and 150-nm metal (Figure 21.25). Although

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.5 TRANSFLECTIVE DISPLAY

585

Figure 21.25. Cross-sectional view of Asahi Kasei WGP.

the details of the performance have not been released, this is reported to be the largest NI process. In order to use NI process as a large area display element, a consistent effort is required, in addition to the equipment development [36]. Another challenge in functional implementation of WGP using NI process is a requirement for a specific pressure or a hard mold to transfer the complete pattern. In this case, the life span of a mold is reduced considerably by repeated processes using high pressure, which may also increase the process cost. Also, it is fundamentally difficult to say that NI process is stable for forming a high aspect ratio pattern 50-nm half-pitch pattern. Especially, the pattern distortion such as a slack, may have great influence on the optical performance and the uniformity. In order to alleviate these problems, it is necessary to have a concerted effort at various levels including equipment development, curing resin development, and improvements in the coating material and the mold material for solving the adhesion problem between the substrate and the mold, etc. If these problems are solved, it will become the de facto standard for lithography in the future by allowing implementation of a micro pattern as well as providing a lower cost process than any other technology. In addition, it will help to accelerate the commercialization of WGP.

21.5 TRANSFLECTIVE DISPLAY The mobile phone display should be highly visible regardless of any lighting condition since the mobile phones are used under various lighting conditions including bright outdoor conditions such in the beach and on the ski slopes as well dark indoor conditions such as in the movie theaters. Also, the visibility of the LCD has become an important issue as the application is rapidly moving away from consumer applications to commercial applications such as outdoor advertisements or public displays. As shown in Figure 21.26, the display luminance should be 5–50% of the ambient light [37]. If the luminance is lower, the information is not visible, and if the luminance of the display is too high, the information will be visible but the consumers will end up with fatigued eyes.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

586

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

1,00,000 10,000 1000

Display is too bright—readable, but annoying

Luminance (cd m)

100 10 1

Optimum visual comfort

Display is too dark — not readable

Indoor

Daylight

0.1 Car

0.01 0.001 0.1

1

10

100

1000

10,000

1,00,000

1M

Ambient Illumination — Lux Figure 21.26. Correlation between the external brightness and the display brightness. (Reprinted with permission from [37]. Copyright 2004 Society for Information Display.)

In general, since an ambient light condition without a direct sunlight has a brightness of about 10,000–20,000 lx (during cloudy day or in a shade), the display luminance should be at least 500–1000 cd m–2 . However, the readability of current mobile phone display is quite low, with BLU luminance of about 200–300 cd m–2 . To overcome the low readability issue, a transflective display is used instead of increasing the BLU luminance. The transflective display uses a part of the LCD panel’s reflective area to reflect the incoming ambient light, resulting in increase of the luminance without requiring additional power consumption (Figure 21.27). Daylight

Reflective mode

Transmittive mode

Figure 21.27. Cross-sectional view of transflective LCD.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.5 TRANSFLECTIVE DISPLAY

(a) Diffusive type

(b) Asymmetrically concave type

587

(c) Diffractive type

Figure 21.28. Reflecting patterns. (a) Diffusive type, (b) asymmetrically concave type, and (c ) diffractive type.

As shown in Figure 21.27, the transmissive area has a structure that allows the light from BLU at the bottom to be transmitted directly, while the reflecting area at the left has a structure that allows the external light to be reflected off the reflective metal in the center of the liquid crystal layer. In order to maintain a separate light path for the reflecting and transmitting light to avoid polarization, the reflecting area is generally located at the center of the liquid crystal. The image quality of the transflective LCD is closely related to the surface structure of the reflecting plate. The reflection efficiency of a structure with an irregular scattering pattern is superior to a flat surface. However, since the size and the type of a pattern have great influences on the distribution of reflection angles and the brightness of reflections, various types of reflecting plate have been developed. 21.5.1 Design and Optical Properties of Reflecting Pattern The reflecting pattern can be largely classified into (a) a diffusive type, (b) an asymmetrically concave type, and (c) a diffracting type as shown in Figure 21.28. The majority of reflecting pattern used in the transflective LCD employ the diffusive type developed by Sharp. The diffusive pattern shown in Figure 21.28a is fabricated by using a photolithography process to form a convex pattern, and then shaped into a round form by reflowing the pattern. It is widely used for the simple processibility and facile control of pattern arrangement. However, it is difficult to change the shape of each pattern due to the characteristics of a reflow process and to control reflective properties due to multiple reflections from adjacent patterns. Figure 21.29a shows the surface image of a diffusive-type reflecting pattern and Figure 21.29b shows a diffusive angle distribution for 30◦ parallel incident beam [38]. The graph shows that the reflecting light is mostly concentrated on the specular reflecting angle (−30◦ ), but the reflecting ratio is rapidly decreased as the angle is changed. Because it is difficult to modify the shape of a diffusive-type reflecting

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Scattering light intensity (a.u.)

588

14:13

MRS MgO

−60 −40 −30 −15

0

15

30

45

60

Scattering angle (degree)

Figure 21.29. (a) Surface image of dispersing-type reflecting pattern. (b) Reflection distribution of dispersing-type reflecting pattern. (Reprinted with permission from [38]. Copyright 1997 Sharp Co.)

pattern, it is a challenge to improve the scattering angle distribution. The imprint method allowing fabrication of random shapes for the reflecting pattern has been recently developed to address this issue. The imprint method has a reflecting property superior to the conventional diffusive-type reflection for the facile control of the accurate patterning. Currently utilizing methods are an asymmetrically concavetype reflecting pattern (b) and a diffracting-type reflecting pattern (c) as shown in Figure 21.29. The asymmetrically concave type uses a micro tip to form concave patterns directly on the reflecting surface, and the distribution of reflecting angles is controlled by making the shape of the micro tip asymmetrical. As shown in Figure 21.30, the symmetric concave pattern has a low light efficiency because the reflection distribution is wide while the reflection ratio of the asymmetric reflection distribution increases because the reflection distribution is concentrated between 0◦ –30◦ as shown in the graph [39]. However, the diffractive-type reflecting pattern in Figure 21.29c allows adjustment of the direction of the reflection angle distribution by periodically making the diffraction pattern smaller than the wavelength of the visible light spectrum to diffract the external light. Because the reflecting light in the diffracting method is concentrated at the diffraction angle, the reflection efficiency of the light at the desired direction and angle can be increased by proper use of a diffraction pattern [40, 41].

21.5.2 Fabrication of the Reflecting Pattern The microhologram method using a laser writer is used for the fabrication of a master pattern for the imprint process. As shown in Figure 21.31, the pattern is generated by coating the PR (photoresist) on the glass substrate, and then, using a spin coater, AZ P4210 PR is coated to 1 mm in thickness, and finally it is soft baked in the oven for 30 min at 90◦ C. After soft-baking step, the temperature is lowered to a room

Coventional reflectors(µm)

0

10

reflection

20

Conventional

Acceptance angle

−20 −10

1

2

3

30

40

Newly developed reflector

Figure 21.30. Asymmetric concave reflecting pattern and the reflection distribution.

Newly developed reflector(µm)

AFM (atomic force microcopy) images of the reflector panel

Reflector

4

50

60

70

Reflective propeities of the newly developed rellector panel (angle of incidence 30°)

3. Improved contrast of the reflactive display

2. Improved color reprodaction of the reflactive display

1. Improved lightnees of the reflactive display

The eharacteristies of the LCD featuring a newly developed reflector panel

Direction of direct reflection

Newly developed reflector Effective visual angle

14:13

Dispersion range

Optical loss range

Direction of direct reflection

Conventional reflectors

Effective visual angle

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008 Printer Name: Yet to Come

589

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

590

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

Glass substrate

HMDS : 3500 RPM (30 s) PR [AZ P4210] : 4000 RPM (60 s) :~1 µm thickness

Spin-coating PR

PR master

Soft baking

LASER exposure

Development

@oven 90°C (30 min)

Micro-grating pattern writing intensity: s = 5 mW, R = 5 mW Expose time = 50–200 ms Developer [AZ 400 K(1:3.5)]: 1 min

n =159, thickness = 0.5 mm

UV replica

Hard baking

UV curing

Replica

@oven 120°C (30 min)

Polycarbonate LGP + UV resin

n =159, thickness = 0.5 mm

Figure 21.31. Process for fabricating the master pattern with the laser.

temperature and the micrograting pattern is written using a laser. At this time, both the signal beam and the reference beam are set to 5 mW of light intensity and 50–200 ms of exposure time. Then, PR is developed using AZ 400 K developer diluted with DI water in 1:3.5 ratio, and finally PR is cured by hard-baking process at 120◦ C for 30 min in the oven to complete the process. Writing the grating pattern with a laser is performed using test equipment setup shown in Figure 21.32. The 458 nm wavelength Ar+ laser is used as a light source and the light source is spatially filtered and collimated by installing an object glass and a pinhole. The same light intensity of two beams divided by a beam-splitter is maintained by a variable attenuator and the distribution of diffracting angles is maximized by adjusting the grating interval, the depth and the direction of the micrograting pattern on the attached PR surface using the x–y stage installed vertically on the floor. Figure 21.33 shows an example of a fabricated pattern. The sample size is 20 mm × 70 mm and a beam spot with about 30 µm in diameter is arranged in a hexagonal structure. It can be seen from the AFM (atomic force microscopic) and SEM (scanning electron microscopy) images that an interference pattern of about 450 nm in size is formed on each spot surface. Laser intensity was fixed at 10 mW cm–2 because the pattern depth changes based on the exposure time of the laser beam, and the exposure time was varied from 100, 125, 150 to 200 ms to get the optimum diffraction efficiency by measuring each shape formed by changing the exposure times. The imprint process is a useful technology in the improvement of reflectivity by fabricating a highly precise reflection pattern on a reflecting area of the transflective

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

21.5 TRANSFLECTIVE DISPLAY

M OL P

S

OL M

M

x–y-φ stage

BS PR BS

Shutter BS M

Figure 21.32. How to fabricate the microhologram.

Microscope

Microscope image

Diameter: 300 µm 500 µm

Period: 450 nm

SEM

AFM

Figure 21.33. Surface image of sample fabricated by microhologram method.

591

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

592

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

LCD. However, in order to apply the process to a large-area display, it is necessary to develop an imprint process for a larger area, including replication technology as well as master fabrication technology. At present, transflective LCD is used in the mobile phones and integrated into Ultra Mobile PC (UMPC) and laptop PC displays. Therefore, it is anticipated that the need for the development of a larger area process will only be in high demand.

REFERENCES 1. Ahn, S.-W., Lee, K.-D., Kim, J.-S., Kim, S. H., Park, J.-D., Lee, S.-H., and Yoon P.-W. (2005) Fabrication of a 50 nm half-pitch wire grid polarizer using nanoimprint lithography. Nanotechnology, 16, 1874–1877. 2. Vratzov, B., Fuchs, A., Lemme, M., Henschel, W., and Kurz, H. (2003) Large scale ultraviolet-based nanoimprint lithography. J. Vac. Sci. Tech. B 21, 2760–2764. 3. Guo, L. Krauss, P. R., and Chou, S. Y. (1997) Nanoscale silicon field effect transistors fabricated using imprint lithography. Appl. Phys. Lett., 71, 1881–1883. 4. Krauss, P. R. and Chou, S. Y. (1997) Nano-compact disks with 400 Gbit/in2 storage density fabricated using nanoimprint lithography and read with proximal probe. Appl. Phys. Lett. 71, 3174–3176. 5. Nanoimprint Technology to be Commercially Available in 2004, Nikkei Electronics, April 26th 2004, p. 67. 6. Trout, T. J., Gambogi, W. J., Steijn, K. W., and Mackara, S. R. (2000) Volume holographic components for display applications. SID Digest, 31, 202–205. 7. Chen, A. G., Jelley, K. W., Valliath, G. T., Molteni, W. J., Ralli, P. J., and Wenyon, M. M.,(1995) Holographic reflective liquid-crystal display. J. SID 3, 159–163. 8. Sterling, R. D. and Bleha, W. P. (2000) D-ILA technology for electronic cinema. SID Digest, 31, 310–313. 9. Cornelissen, H. J., Greiner, H., and Dona, M. J. (1999) Frontlights for reflective liquid crystal display based on lightguides with micro-grooves. SID Digest, 30, 912–915. 10. Oki, Y. (1998) Novel backlight with high luminance and low power consumption by prism-on-light-pipe technology. SID Digest, 29, 157–160. 11. Pang, Z. and Li, L. (1999) Novel high efficiency polarizing backlight system with a polar beam splitter. SID Digest, 30, 916–919. 12. Tedesco, J. M., Bardy, L. A. K., and Colburn, W. S. (1993) Holographic diffusers for LCD back lights and projection screens. SID Digest, 24, 29–32. 13. Choi, H. Y., Lee, M. G., Min, J. H., and Choi, J. S. (2001) Hologram based light-guide plate for LCD-backlights. IDW ‘01, 20, 521–524. 14. Moharam, M. G., Grann, E. B., Pommet, D. A., and Gaylord, T. K. (1995) Formulation for stable and efficient implementation of the rigorous coupled-wave analysis of binary gratings. J. Opt. Soc. Am. A 12, 1068–1076. 15. Lalanne, P. and Morris, G. M. (1996) Highly improved convergence of the coupled-wave method for TM polarization. J. Opt. Soc. Am. A 13, 779–784. 16. Peng, S. and Morris, G. M. (1995) Efficient implementation of rigorous coupled-wave analysis for surface-relief gratings. J. Opt. Soc. Am. A 12, 1087–1096.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

14:13

Printer Name: Yet to Come

REFERENCES

593

17. Weber, M. F., Stover, C. A., Gilbert, L. R., Nevitt T. J., and Ouderkirk, A. J. (2000) Giant birefringent optics in multilayer polymer mirrors. Science 287, 2451–2456. 18. Broer, D. J., Lub, J., and Mol, G. N. (1995) Wide-band reflective polarizers from cholesteric polymer networks with a pitch gradient. Nature 378, 467–469. 19. Tanase, H., Mamiya, J., Suzuki, M., Nakatani, H., Hatazawa T., and Watanabe, T. (1997) A new backlighting system with a polarizer light pipe for enhanced light output from LCDs. SID Intl. Symp. Digest Tech. Papers 28, 365–368. 20. Pang Z. and Li, L. (1999) Novel high-efficiency polarizing backlighting systems with a polarizing beam splitter. SID Intl. Symp. Digest Tech. Papers 30, 916–919. 21. Jagt, H. J. B., Cornelissen, H. J., Broer, D. J., and Bastiaansen, C. W. M. (2002) Microstructured polymeric linearly polarized, light emitting light guide for LCD illumination. SID Intl. Symp. Digest Tech. Papers 33, 1236–1239. 22. Chien, K.-W., Shieh H.-P. D., and Cornelissen, H. (2004) Polarized backlight based on selective total internal reflection at microgrooves. Appl. Opt. 43, 4672–4676. 23. Cornelissen, H. J., Huck, H. P. M., Broer, D. J., Picken, S. J., Bastiaansen, C. W. M., Erdhuisen E., and Maaskant, N. (2004) Polarized light LCD backlight based on liquid crystalline polymer film. SID Intl. Symp. Digest Tech. Papers 35, 1178– 1181. 24. Hwang, S., Kim, Y.-T., Kim, Y., Nam, S., and Lee, S.-D. (2007) Highly efficient backlight unit with a polarization-separating anisotropic layer. SID Intl Symp. Digest Tech. Papers 38, 476–479. 25. Shan, X. C., Maeda, R., and Murakoshi, Y. (2003) Micro hot embossing for replication of microstructures. Japan J. Appl. Phys. 42, 3859–3862. 26. Juang, Y.-J., Lee, L. J., and Koelling, K. W. (2002) Hot embossing in microfabrication. Part I: experimental. Polym. Eng. Sci. 42, 539–550. 27. Stutzmann, N., Tervoort, T. A., Bastiaansen, C. W. M., Feldman, K., and Smith, P. (2000) Solid-state replication of relief structures in semicrystalline polymers. Adv. Mater. 12, 557–562. 28. Kim, Y.-T., Hwang, S., Hong, J.-H., and Lee, S.-D. (2006) Alignment layerless flexible liquid crystal display fabricated by an imprinting technique at ambient temperature. Appl. Phys. Lett. 89, 173506. 29. Zhang, L., Zhang, F., Li, C., Liu, W., and Wang, D. (2006) High performance polarizing beam splitter with embedded metal-wire nanograting. Proc. IEEE Intl. Conf. Nano/Micro Eng. Mol. Syst. 947–950. 30. http://www.gsolver.com. 31. Wang, J., Shablitsky, S., Yu, Z. N., Wu, W., and Chou, S. Y. (1999) Fabrication of a new broadband TM-pass waveguide polarizer with a double-layer 190 nm metal grating using nanoimprint lithography. J. Vac. Sci. Technol B 17, 2957–2960. 32. Yu, Z., Deshpande, P., Wu, W., Wang, J., and Chou, S. Y. (2000) Reflective polarizer based on a stacked double-layer subwavelength metal grating structure fabricated using nanoimprint lithography. Appl. Phys. Lett. 77, 927–929. 33. Khusnatdinov, N., Doyle, G., Miller, M., Stacey, N., Watts, M., and LaBrake, D. L. (2006) Fabrication of nano and micro optical elements by step and flash imprint lithography. Proc. SPIE 6110, 61100K-1. 34. Perkins, R. T., Hansen, D. P., Gardner, E. W., Thorne, J. M., and Robbins, A. A. (2000) Broadband wire grid polarizer for the visible spectrum, U. S. Patent 6,122,103, Sep. 19.

P1: OTA/XYZ P2: ABC c21 JWBK252/Lee September 18, 2008

594

14:13

Printer Name: Yet to Come

NANOIMPRINT TECHNOLOGY FOR FUTURE LIQUID CRYSTAL DISPLAY

35. http://www.moxtek.com/visible light.html 36. Technology Trend Analysis Report: Nano Imprint Lithography. (2003) KISTI 37. Laaperi, A. (2004) The present and future of handheld display appliances. SID Biz Conf. B-04. 38. Nakamura, K., Nakamura, H., and Kimura, N. (1997) Development of HR-TFT. Sharp Tech. J. 33. 39. Yoshii, K. and Chie, C. (2005) Reflector and liquid crystal display device, US Patent 6,966,662. 40. Huang, Y.-P., Chen, J.-J., Ko, F.-J., and David Shieh, H.-P. (2002) Multidirectional asymmetrical microlens array light control film for improved image in reflective color liquid crystal displays. Japan J. Appl. Phys., 41, 646–651. 41. Tokumaru, T., Iwauchi, K.-I., and Higashigaki, Y. (1999) Holographic diffusive reflectors for reflective color LCDs. SPIE 3637, 196–203.

P1: OTA/XYZ P2: ABC ind JWBK252/Lee September 18, 2008

13:51

Printer Name: Yet to Come

INDEX

2-D soft lithography, 100 3-D soft lithography, 110 Acryloxy PFPE (a-PFPE), 16 Anisotropic buckling, 59 Anisotropic nanoparticles, 531 BCP microdomains, 234 Bendable semiconductor, 447 Bilayer transfer, 75 Biological macromolecules, 332 Biomolecular nanoarray, 338 Bio-nanoelectromechanical systems (BioNEMS), 325 Black light unit (BLU), 566, 569, 573, 576, 586 Bloch wave surface plasmon polaritons, 315 Block copolymer (BCP), 233, 383, 387 Block copolymer assembly, 387 Bottom-up approach for nanowires, 447 Bottom-up process/approach for BCP, 233, 257 Bottom-up technique for nanofabrication, 383 Buckling, 10, 58, 483 Buckling analysis, 484 Capillary completely closed, 40 closed permeable, 31 open-ended, 29 Capillary force, 28 Capillary force lithography (CFL), 37, 329, 556

Capillary kinetics, 45 Cathode patterning, 421 Cell adhesion control, 369 Chemical synthesis using microchemical chips, 368 Cleavable junction, 272 Colloidal lithography, 383 Combined nanoimprint and photolithography, 155 Composite PDMS stamp, 10, 518 Conformity criteria, 10, 11 Continuous rigiflex imprinting (CRIM), 549 Controlled buckling, 507 Controlled undercutting, 175 Copolymer thermoplastic resist, 141 Copper sulfide stamp, 201 Counter-propagating RD fronts, 222, 226 Covalent attachment, 402 Critical peel velocity, 73, 87 Deflection, 75 Device transfer, 428 Dimensional stability, 11 Dip-pen lithography (DPN), 385 DNA nanoarray, 338 DNA sequencing, 332 DNA stretching, 332 Drug delivery, 336 Edge effect, 495 Edge lithography, 167 Edge-spreading lithography (ESL), 176 Edge-transfer lithography (ETL), 178 Elastomeric photomask, 96, 520

Unconventional Nanopatterning Techniques and Applications by John A. Rogers and Hong H. Lee C 2009 John Wiley & Sons, Inc. Copyright


595

P1: OTA/XYZ P2: ABC ind JWBK252/Lee September 18, 2008

596

13:51

Printer Name: Yet to Come

INDEX

Elastomeric substrates, 483 Electrical microcontact printing (e-µCP), 185 Electrochemical imprinting, 203 Electrochemistry, 300 Electrodeposition, 175, 180 Electron beam lithography (EBL), 381, 387, 388, 393, 518 Electron injection layer (EIL), 420 Electron transport layer (ETL), 420 Embedded plastic circuit, 435 Embossing, 433 Emitting layer (EML), 420 Epitaxy, 245 ETFE mold, 138, 158 External field, 239 Extracellular matrices (ECM), 344, 347 Extreme ultraviolet (EUV) radiation, 381, 388 Filament stability, 48 Finite deformation, 488 Flexibility of mold, 75 Flexible electronics, 446 Flexible fluoropolymer mold, 137 Flexible inorganic electronics, 446, 459 Flexible integrated circuits, 463 Focused ion beam (FIB) lithography, 382, 518 Fractional RD, 222 Fracturing, 182 Fuel cell, 306 Glass transition temperature, 32 Gradient generation, 297 Graphoepitaxy, 256 Hard mold, 19 Hard PDMS, 10, 133 Heterogeneous electronics, 466 Heterogeneous surface, 432 Hierarchical structure, 270 Highly oriented pyrolytic graphite (HOPG), 180 Hole filling, 51 Hole injection layer (HIL), 420 Hole transport layer (HTL), 420 Holographic light guiding plate (LGP), 569 Homopolymer, 237

Imprint lithography, 19, 129, 130, 429, 433 Improved method of preparing PDMS mold, 14 Ink-jet, 419, 432 Intrinsic wavelength of wrinkles, 58 Inverter circuit, 117 Ion transport, 197 Kinetically controlled adhesion, 71 Laminar flow, 298 Lamination, 429, 434 Laplace pressure, 30 Laser interference lithography (LIL), 381 Lateral collapse (merging), 10 Lattice-gas model, 221, 227 Layer-by-layer (LBL) assembly, 383, 392 Light-stamping lithography (LSL), 332 Line calibration, 204 Lipid array, 345 Liquid crystalline polymer, 242 Liquid filament, 53 Liquid handling, 362, 370 Liquid microspace, 361 Localized surface plasmon resonances, 315 London force, 48 Long period fiber grating, 311 Long-range ordering, 254 Low pressure imprint lithography, 137 Macroelectronics, 445, 450 Mechanical stress, 56 MESFETs, 460 Metal patterning, 56 Micells, 387 Microchannel, 365 Microchemical chips, 368 Microchip, 361, 365 Microcontact printing (µCP), 185, 296, 384, 544 Microdominos, 186 Microfluidics, 118, 306 microfluidic chip, 361 microfluidic device, 297 microfluidic fiber, 311 Microlense, 218 Micromolding in capillaries (MIMIC), 29, 328, 453, 459

P1: OTA/XYZ P2: ABC ind JWBK252/Lee September 18, 2008

13:51

Printer Name: Yet to Come

INDEX

Microstructured semiconductor (µs-Sc), 450 Microstructured silicon (µs-Si), 450 Microtome, 184 Micro total analysis systems µ-TAS), 359 Micro unit operation (MUOs), 363 Mixed technique for nanofabrication, 384 Modified soft molding, 36 Modulator, 545 Mold materials, 8 Mold preparation, 8 MOSFETs, 459 Multilayer transfer, 79 Nanochannel, 326 Nanochannel delivery systems (nDSs), 336 Nanofabrication, 379 Nanofluidics, 326 Nanohole array, 523 Nanoimprint lithography (NIL), 129, 327, 340, 381, 392, 396, 454, 544 molds, 133 mold surface preparation, 137 resist, 138 process, 149 roll-to-roll NIL, 156 Nanomolding, 328 Nanoparticle array, 528 Nanoparticle assembly, 389 Nanoporous structure, 269 Nanoribbons, 446, 470 Nanostencil lithography, 381 Nanostructured materials, 247 Nanotophography, 347 Nanotransfer printing (nTP), 75, 296, 384, 396 Nanowire pattern transfer (SNAP), 457, 459 Nanowires, 446 Optical soft lithography, 95 Optical waveguide, 308 Organic light emitting diodes (OLED), 420 Organic thin film transistors (OTFT), 429 Orientation of BCP microdomains, 235 Orthogonal field, 260 Oxygen plasma, 23 Oxygen reduction reaction, 302, 304 Ozonolysis, 264

597

PEEL method, 521 Perfluoropolyether (PFPE) stamp, 15, 329, 540 Permeability of PDMS, 302, 306 Phase-selective cross-linking chemistry, 270 Phase-shift photolithography, 170, 520 pH gradient, 300, 304 Photolithography, 295, 380 Photonic bandgap (PBG), 397 Photonic bandgap material, 120 Photonic crystals (PhCs), 397 Physisorption, 400 Plasmonic structure, 515 Polarized light guiding plate (LGP), 573 Poly(dimethylsiloxane) (PDMS), 7, 296, 331, 423, 429 Polymer template, 380 Polymer TFT, 433 Potassium hexacyanoferrate, 218, 222 Protein assembly, 400 Protein nanoarray, 340 Protein separation, 334 Proximity field nanopatterning, 111 PUA mold, 19 Pyramical nanostructures, 531 Quasicrystal, 115 Rapid flash patterning (RFP), 44, 547 Reaction-diffusion (RD), 216 Reaction-diffusion fabrication (RDF), 218 Reflective polarizer, 579 Reverse nanoimprint, 152 Ribbon width and spacing, 498 Ribbon/wire- based FETs, 460 Rigiflex lithography, 75, 133, 544 Rigiflex mold, 19, 75, 137, 540 Rod-coil BCP, 245 Roof collapse, 10 Room temperature imprinting, 18, 141 Selective deposition, 432 Selective dewetting, 49 Selective extraction, 271 Self-assembled monolayers (SAMs), 172, 300, 383 Self-directing, 247 Shape engineering, 60 Silicon films, 483

P1: OTA/XYZ P2: ABC ind JWBK252/Lee September 18, 2008

598

13:51

Printer Name: Yet to Come

INDEX

Silver nitrate, 218 Silver sulfide stamp, 201 Soft chemical etch, 272 Soft contact lamination, 425 Soft interference lithography (SIL), 520 Soft lithography, 296, 331, 342 Soft mold, 8 Soft molding, 13, 32, 553 Soft molding in reverse (SMIR), 433 Soft PDMS (s-PDMS), 8 Soft-UV nanoimprint lithography, 296, 314 Solid electrolytes, 199 Solid-state superionic stamping (S4), 197 S4 stamp, 199 Solubility parameter, 9, 10 Solution-liquid-solid (SLS) method, 449 Solvent annealing, 254 Solvent-assisted micromolding (SAMIM), 31, 329 Spreading coefficient, 53 Step-and-flash imprint lithography (S-FIL), 130, 327, 582 Step-edge decoration, 175, 180 Stretchable electronics, 469, 483 Subtractive transfer patterning, 79 Supramolecular interaction, 403 Supramolecules, 252 Surface plasmon, 315, 517 Surface plasmon polaritons, 314 Surface treatment and modification, 21 Swelling, PDMS, 9, 10 SWNT TFTs, 468

Talbot effect (self-imaging effect), 111 Teflon mold (solvent casting, compression molding), 20 Tensile stress-strain relationship, 546 Thermal-curable resist, 142 Thermal degradation, 264 Thermoplastic resist, 139 Top-down approach for nanowires, 449 Top-down process/approach for BCP, 233, 257 Top-down technique for nanofabrication, 380 Topographic reorientation, 186 Topographically-directed etching (TODE), 175 Topography-directed pattern transfer, 169 Transfer patterning/printing, 74, 82, 437, 454, 558 Transflective display, 585 Tunable-oxidized PDMS nanochannel, 332 Two-photon, 99, 100, 112, 113, 115 Ultraviolet (UV) degradation, 267 UV-curable mold, 540 UV-curable resist, 146 UV ozone, 308 Vapor liquid solid (VLS) method, 447 Washburn kinetics, 30 Wet stamping (WETS), 218 Williams-Landel-Frerry equation, 58 Wire grid polarizer, 579 Wood’s anomalies, 315 Work of adhesion, 68