Electrochemical Microsystem Technologies (New Trends in Electro Chemical Technology, 2)

Electrochemical Microsystem Technologies New trends in electrochemical technology Edited by Tetsuya Osaka Waseda Univ...

Author: J. Walter Schultze | Tetsuya Osaka | Madhav Datta

45 downloads 1703 Views 18MB Size Report

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

Report copyright / DMCA form

DOWNLOAD PDF

Electrochemical Microsystem Technologies

New trends in electrochemical technology Edited by Tetsuya Osaka Waseda University, Tokyo, Japan

Madhav Datta Portland Technology Development, Intel Corporation, Hillsboro, USA

Volume 1 Energy Storage Systems for Electronics Edited by Tetsuya Osaka and Madhav Datta Volume 2 Electrochemical Microsystem Technologies Edited by J. Walter Schultze, Tetsuya Osaka and Madhav Datta

Electrochemical Microsystem Technologies

Edited by J. Walter Schultze, Tetsuya Osaka and Madhav Datta

London and New York

First published 2002 by Taylor & Francis 11 New Fetter Lane, London EC4P 4EE Simultaneously published in the USA and Canada by Taylor & Francis Inc, 29 West 35th Street, New York, NY 10001 Taylor & Francis is an imprint of the Taylor & Francis Group This edition published in the Taylor & Francis e-Library, 2005.

“To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” © 2002 Taylor & Francis All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Every effort has been made to ensure that the advice and information in this book is true and accurate at the time of going to press. However, neither the publisher nor the authors can accept any legal responsibility or liability for any errors or omissions that may be made. In the case of drug administration, any medical procedure or the use of technical equipment mentioned within this book, you are strongly advised to consult the manufacturer’s guidelines. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data A catalog record for this book has been requested ISBN 0-203-21921-X Master e-book ISBN I

ISBN 0-203-23435-0 (Adobe eReader Format)

(Print Edition) ISBN 0-415-27355-2 I

Contents

Editorial board List of contributors Preface to the series Preface

xi xii xv xvii

PART I

Fundamentals

1

1

3

Electrochemical microsystem technologies: principles J. WALTER SCHULTZE

1.1 1.2 1.3 1.4 1.5 2

Introduction: subjects and definitions 3 Electrochemistry in microscopic dimensions 8 Reactions and materials 17 Special aspects of EMST 23 Conclusions 27

Application for homogeneous electrochemistry: measurements of fast reaction kinetics

32

J. HEINZE AND K. BORGWARTH

2.1 2.2 2.3 2.4 2.5 2.6 2.7 3

Introduction 32 Basic principles 32 Kinetics under steady-state conditions 50 Kinetics in fast-scan voltammetry 52 Kinetics under transient conditions 58 Kinetics between two electrodes 59 Conclusions 61

Fundamentals of scanning electrochemical microscopy MICHAEL V. MIRKIN AND BENJAMIN R. HORROCKS

3.1 Introduction 66 3.2 Instrumentation 73 3.3 Theory 75

66

vi

Contents 3.4 Selected applications 81 3.5 Advances in SECM imaging 97

4

Electrochemical microcells and surface analysis

104

M. M. LOHRENGEL AND A. MOEHRING

4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5

Definitions 104 Classification 105 Thin film cells 107 Mobile and immobile mask techniques 108 Droplet cells 108 Scanning electrochemical microscope 115 Other classes of microcells 115 Borderline cases of microcells 117 Conclusions 117

Application of optical micro-methods and lasers in electrochemistry

122

A. MICHAELIS AND S. KUDELKA

5.1 5.2 5.3 5.4 5.5

Introduction 122 Fundamentals of optical/electrical methods 126 Experimental details 130 Optical/optical and optical/electrical measurements 132 Photoresist microelectrochemistry (nano – liter droplet method) 141 5.6 Summary and conclusions – Ti/TiO2 model 152 6

Nucleation and growth in microsystem technology

156

GEORGI STAIKOV

6.1 6.2 6.3 6.4 6.5

Introduction 156 Fundamentals of phase formation 157 Deposition processes in electrochemical systems 165 Electrochemical microstructuring 176 Conclusions 181

PART II

Micropatterning

185

7

187

High resolution lithography SHINJI MATSUI

7.1 Introduction 187 7.2 Nanofabrication using electron beam 189 7.3 Nanofabrication using focused ion beam 206

Contents vii 7.4 Nanofabrication using a de Broglie wave (material wave technology) 213 7.5 Summary 220 8

Advanced plating technology for electronics packaging

224

HIDEO HONMA AND HIDETO WATANABE

8.1 Introduction 224 8.2 Fabrication of gold bumps using the disulfiteaurate complex 224 8.3 Nickel bump formation on aluminum substrate using electroless plating 231 8.4 Via-filling using electroplating for build-up printed circuit boards 236 8.5 Preparation of anisotropic conductive particles by electroless nickel plating 240 9

Micro-electroforming of miniaturized devices for chemical applications

245

HOLGER LÖWE, WOLFGANG EHRFELD AND JÖRG SCHIEWE

9.1 9.2 9.3 9.4 9.5

Microreactors – a novel concept in chemical engineering 245 Fabrication by LIGA-technology 246 Alternative fabrication technologies 254 Potential and applications of microreactors 254 Conclusions 265

PART III

Integration of systems

269

10 Capacitors and micropower systems

271

AKIHIKO YOSHIDA

10.1 10.2 10.3 10.4 10.5 10.6 10.7

Introduction 271 The principle of electric double-layer capacitors 271 History and present status of the capacitors 273 Types of capacitors 273 Effective factors to capacitor characteristics 278 Application of the capacitors 282 Future prospects of the capacitors 284

11 Batteries for micropower applications BOONE B. OWENS, CURTIS L. HOLMES, JOHN BATES, WILLIAM H. SMYRL, N. J. DUDNEY, B. J. NEUDECKER, AND STEFANO PASSERINI

11.1 Introduction 286 11.2 Lithium cells for cardiac pacemakers (C. L. Holmes) 288

286

viii

Contents 11.3 Micropower lithium cells for solid-state hazardous gas monitor (B. B. Owens) 293 11.4 Lithium-ion cells for rechargeable hearing aids (B. B. Owens and S. Passerini) 295 11.5 Recent developments in thin-film lithium and lithium-ion batteries (J. Bates, B. J. Neudecker, and N. J. Dudney) 303 11.6 High energy amorphous metal oxides (W. H. Smyrl and B. B. Owens) 313 11.7 Micropower fuel cells (W. H. Smyrl) 316 11.8 Conclusions 317

12 Micro flow systems for chemical and biochemical applications

321

SHUICHI SHOJI

12.1 12.2 12.3 12.4

Introduction 321 MEMS type micro flow systems 321 Chip CE type micro flow systems 325 Conclusions 327

13 Corrosion of microsystems

329

GUENTER SCHMITT

13.1 13.2 13.3 13.4 13.5

Introduction 329 Corrosion systems 330 Corrosion of electronic and photonic devices 338 Electrochemical sensors 350 Conclusions 365

PART IV

Microanalysis and microsensors

369

14 Electrochemical microanalysis

371

HENDRIK EMONS

14.1 14.2 14.3 14.4

Analytical chemistry and the microworld 371 Electroanalysis and microscales 373 Electrochemical devices for microanalysis 377 Perspectives 380

15 Novel approaches to design silicon-based field-effect sensors M. J. SCHÖNING

15.1 Introduction 384 15.2 Theory of silicon-based field-effect sensors 386 15.3 Sensor fabrication and measurement 390

384

Contents ix 15.4 Sensor properties of the planar and porous EIS structures 394 15.5 Sensor performance of the insect antenna-based BioFET 400 15.6 Conclusions 404 16 Miniaturization of biosensors

409

WOLFGANG SCHUHMANN AND KATJA HABERMÜLLER

16.1 16.2 16.3 16.4

Introduction 409 Fundamentals 411 Selected examples of micro-biosensors 417 Summary and outlook 423

17 Scanning probe microscopy as an analysis tool

429

LARRY A. NAGAHARA

17.1 17.2 17.3 17.4 17.5 17.6 17.7

Introduction 429 Fundamentals of scanning probe microscopy 430 Electrochemical analysis on metal surfaces 442 SPM analysis of adsorbates on surfaces 446 SPM directed modification 449 Future directions 454 Summary 458

18 Microelectrode techniques for characterization of advanced materials for battery and sensor applications

465

MATSUHIKO NISHIZAWA AND ISAMU UCHIDA

18.1 18.2 18.3 18.4

Introduction 465 Single particle measurement 466 Microarray electrode for in situ conductance measurement 473 Summary 480

PART V

Biological systems

483

19 Microsystems for biosensing nucleic acids and immuno proteins

485

TADASHI MATSUNAGA AND TAE-KYU LIM

19.1 Introduction 485 19.2 Microsystem for biosensing nucleic acid 486 19.3 Electrochemical microsystems for biosensing immuno proteins 496 19.4 Conclusions 507

x

Contents

20 New microelectrode arrays for biosensing and membrane electroporation

512

EBERHARD NEUMANN AND KATJA TÖNSING

20.1 Introduction 512 20.2 Microsensing cholinergic effector substances 512 20.3 Capillary electrode array for clinical electroporation 518 21 Multi-barrelled ion-selective microelectrodes: measurements of cell volume, membrane potential, and intracellular ion concentrations in invertebrate nerve cells

526

PAUL WILHELM DIERKES, SUSANNE NEUMANN, ANJA MÜLLER, DOROTHEE GÜNZEL, AND WOLF-RÜDIGER SCHLUE

21.1 Introduction 526 21.2 Ion-selective microelectrodes 527 21.3 Application of ion-selective microelectrodes in biological tissues 531 21.4 Conclusions 539 22 Nerve cells and lipid vesicles on silicon chips – considerations on ionoelectronic sensors

541

PETER FROMHERZ

22.1 22.2 22.3 22.4 22.5 22.6 22.7 Index

Introduction 541 Planar core-coat conductor 542 Distance of membrane and chip 545 Conductance of junction 547 Distribution of channels 552 Action potentials 554 Outlook 558 561

Editorial board

Alkire, R. C., University of Illinois, USA Cairns, E. J., Lawrence Berkeley Laboratory, USA Datta, M., Intel Corporation, USA Fukunaka, Y., Kyoto University, Japan Ito, S., Nippon Steel Corporation, Japan Ito, Y., Kyoto University, Japan Landolt, D., Ecole Polytechnique, Federal de Lausanne, Switzerland Masuko, N., University of Tokyo, Japan Muller, R. H., University of California, Berkeley, USA Nakahara, S., Bell Laboratory, AT&T, USA Nihei, K., Oki Electric Industry Co. Ltd., Japan Osaka, T., Waseda University, Japan Plieth, W. W., Dresden University of Technology, Germany Romankiw, L. T., IBM Corporation, USA Sadoway, D. R., MIT, USA

Contributors

John Bates, Thin Film Battery Inc., 2130 Northwest Parkway, Suite F, Marietta, GA 30067, USA K. Borgwarth, Institut für Physikalische Chemie and Freiburger Materialforschungszentrum, Albertstr. 21, D-79104 Freiburg, Germany Paul Wilhelm Dierkes, Institut für Neurobiologie Heinrich-Heine-Universität, Düsseldorf Universitätsstr. 1, 40225 Düsseldorf, Germany N. J. Dudney, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA Wolfgang Ehrfeld, Ehrfeld Mikrotechnik GmbH, Mikroforum Ring 1, D 55234 Wendelsheim, Germany Hendrik Emons, Forschungszentrum Jülich, Institut Phytosphäre, D 52425 Jülich, Germany Peter Fromherz, Department of Membrane and Neurophysics, Max-PlanckInstitute for Biochemistry, D-82152 Martinsried/München, Germany Dorothee Günzel, Institut für Neurobiologie Heinrich-Heine-Universität Düsseldorf Universitätsstr. 1, 40225 Düsseldorf, Germany Katja Habermüller, Analytische Chemie-Elektroanalytik and Sensorik Ruhr-Universität Bochum, D-44780 Bochum, Germany Jürgen Heinze, Institut für Physikalische Chemie and Freiburger Materialforschungszentrum, Albertstr. 21, D-79104 Freiburg, Germany Curtis L. Holmes, Greatbatch-Hittman, 9190 Red Branch Rd., Columbia MD 21045, USA Hideo Honma, Faculty of Engineering and Graduate School, Kanto Gakuin University 4834 Mustuura-cho, Kanazawa-ku, Yokohama, Kanagawa 236–8501, Japan Benjamin R. Horrocks, Department of Chemistry, University of Newcastle upon Tyne, Newcastle upon Tyne, NE1 7RU, UK

Contributors xiii Stephan Kudelka, Infineon Technology, MPTIINN, Postfach 10 09 40, D 01076 Dresden, Germany Tae-Kyu Lim, Department of Biotechnology, Tokyo University of Agriculture and Technology, Koganei, Tokyo 184-8588, Japan Manuel M. Lohrengel, Institute for Physical Chemistry and Electrochemistry, Heinrich-Heine-University Duesseldorf, 40225 Duesseldorf, Germany Holger Löwe, Institut für Mikrotechnik Mainz GmbH Carl-Zeiss-Str. 18-20 55129 Mainz, Germany Shinji Matsui, Laboratory of Advanced Science and Technology for Industry, Himeji Institute of Technology 1479-6 Kanaji, Kamigori, Ako, Hyogo 6781201, Japan Tadashi Matsunaga, Department of Biotechnology, Tokyo University of Agriculture and Technology, Koganei, Tokyo 184-8588, Japan Alexander Michaelis, H.C. Starck, Im Schleeke 78-91, D 38642 Goslar, Germany Michael V. Mirkin, Department of Chemistry and Biochemistry, Queens College-CUNY, Flushimg, NY 11367, USA Andreas Moehring, Institute for Physical Chemistry and Electrochemistry, Heinrich-Heine-University Duesseldorf, 40225 Duesseldorf, Germany Anja Müller, Institut für Neurobiologie Heinrich-Heine-Universität, Düsseldorf Universitätsstr. 1, 40225 Düsseldorf, Germany Larry A. Nagahara, Physical Science Research Laboratories, Motorola Labs Tempe, Arizona, 85284, USA B. J. Neudecker, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA Eberhard Neumann, Department of Physical and Biophysical Chemistry, Faculty of Chemistry, University of Bielefeld, PO Box 100 131, D-33501 Bielefeld, Germany Susanne Neumann, Institut für Neurobiologie Heinrich-Heine-Universität, Düsseldorf Universitätsstr. 1, 40225 Düsseldorf, Germany Matsuhiko Nishizawa, Department of Applied Chemistry, Graduate School of Engineering, Tohoku University, Aramaki-Aoba, Sendai 980-8579, Japan Boone B. Owens, Corrosion Research Center, University of Minnesota, 221 Church St. SE, Minneapolis, MN 55455, USA Stefano Passerini, ENEA, ERG-TEA-ECHI, CR Casaccia, Via Anguillarese 301, 00060 Rome, Italy

xiv Contributors Jörg Schiewe, Institut für Mikrotechnik Mainz GmbH Carl-Zeiss-Str. 18-20 55129 Mainz, Germany Wolf-Rüdiger Schlue, Institut für Neurobiologie Heinrich-Heine-Universität Düsseldorf Universitätsstr. 1, 40225 Düsseldorf, Germany Guenter Schmitt, Laboratory for Corrosion Protection Iserlohn, University of Applied Sciences, Frauenstuhlweg 31, D-58644 Iserlohn, Germany Michael J. Schöning, Institute of Thin Films and Interfaces, Research Centre Jülich and University of Applied Sciences Aachen, Division Jülich, Ginsterweg 1, D-52428 Jülich Wolfgang Schuhmann, Analytische Chemie-Elektroanalytik and Sensorik RuhrUniversität Bochum, D-44780 Bochum, Germany J. Walter Schultze, AGEF eV-Institute an der Heinrich-Heine-Universität, Düsseldorf D-40225 Düsseldorf, Germany Shuichi Shoji, Department of Electronics, Information and Communication Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan William H. Smyrl, Corrosion Research Center, University of Minnesota, 221 Church St. SE, Minneapolis, MN 55455, USA Georgi Staikov, Institut für Physikalische Chemie und Elektrochemie, HeinrichHeine-University Düesseldorf, 40225 Düesseldorf, Germany Katja Tönsing, Department of Physical and Biophysical Chemistry, Faculty of Chemistry, University of Bielefeld, PO Box 100 131, D-33501 Bielefeld, Germany Isamu Uchida, Department of Applied Chemistry, Graduate School of Engineering, Tohoku University, Aramaki-Aoba, Sendai 980-8579, Japan Hideto Watanabe, Faculty of Engineering and Graduate School, Kanto Gakuin University, 4834 Mustuura-cho, Kanazawa-ku, Yokohama, Kanagawa 236-8501, Japan Akihiko Yoshida, Matsushita Electric Industrial Co., Ltd. 3-1-1 Yagumonakamachi, Moriguchi-shi, Osaka 570-0005, Japan

Preface to the series

While electrochemical technology has been playing an important role in the manufacturing of chemicals, metals, batteries, fuel cells and other electric utility industries, its contribution in the evolution of electronics and communication industries has been phenomenal. During the past several decades the electronics industry has been through a very rapid evolution from thick to thin films and to ever increasing miniaturization. Electrochemical technology played a decisive role in the direction of this evolution. With an increasing understanding of transport processes, current distribution problems, process monitoring and control issues, and the ability to develop environmentally friendly processes, electrochemists and electrochemical engineers have been able to face the challenges presented by the electronics industry. Several new and unique electrochemical processes and technologies have been introduced. They include advanced plating and etching technologies, cleaning and planarization technologies, and silicon processing technologies. These technologies have made a significant impact in storage, packaging, interconnects, and several other aspects of the microelectronics industry. The importance of smart energy storage systems in the sophistication of advanced portable electronic devices is becoming increasingly recognized. Microelectromechanical systems, sensors, and nanotechnology are some of the other evolving areas where electrochemical processing technologies are playing an increasingly important role. In the aerospace and automobile industries, machining and finishing of complicated shaped components of different materials including high strength alloys have been possible due to development and implementation of advanced electrochemical machining, polishing, anodization and plating technologies. The series is intended to report recent advances made in the development of electrochemical technologies as applied to microelectronics, information and other high-tech industries as well as to heavy industries. The emphasis will be on the application of newly developed concepts and materials. Internationally recognized experts from scientific and industrial communities will provide the current status and future trends of a technology area. The first volume is devoted to energy storage systems for electronics. Future volumes in the series will include topics such as electrochemical processing technologies, advanced

xvi Preface to the series manufacturing, microelectronic packaging, silicon processing, micromachines, and electrochemically fabricated advanced materials. Tetsuaya Osaka, Waseda University Madhav Datta, Intel Corporation

Preface

During the past decade, electrochemical technology has played a decisive role in the phenomenal growth of storage, chip interconnects, microelectronic packaging, microelectromechanical systems, sensors, actuators, and many other microelectronic and micromechanic components. These advances have been possible due to continuing efforts by academia and industry to understand the underlying physical principles of micro and nanoscale processing, to implement innovative ideas, and to improve stability of microsystems including last results of corrosion research in microscopic dimensions. Several years ago, we started a Symposium on “Electrochemical Microsystem Technologies”. The first meeting was held in 1996 in Grevenbroich, the second one in 1998 in Tokyo, and the third one in 2000 in Garmisch-Partenkirchen. These symposia provided a forum for presentation and discussion of novel approaches in the use of electrochemical techniques for micro fabrication and integrated microsystems. Advances in topics such as microelectronics, sensors, material science, corrosion, and micro- and nanostructures were discussed. Another key aspect of these symposia was to initiate interactions with researchers in microbiological sciences to learn from their experiences thus providing deeper insight into molecular and nanoscopic processes. Experts from neurobiology, and medicine were invited to present their opinions on topics such as the understanding of electron transfer reaction on DNA, electrophoresis, and new microanalytical techniques in microbiological sciences. Such interactions have given a new dimension to the field of Electrochemical Microsystems Technologies. These advances in micro and nanotechnologies are now making an enormous impact on the novel approaches to miniaturization with increasing ability to fabricate structures with high lateral and vertical resolution. The present book on “Electrochemical Microsystems Technologies” is based on the foundations of the above symposia. The chapters in the book are clustered together under five different headings: I Fundamentals, II Micropatterning, III Integration of systems, IV Microanalysis and microsensors, and V Biological systems. The chapters, therefore, cover a wide spectrum of issues in electrochemical technologies ranging from fundamental understanding of electrochemical processes to their applications in micro- and nanofabrication, microanalyses, microsensing and interaction of electrochemistry and biological systems.

xviii Preface We express our thanks and gratitude to all authors, referees and advisors for their help and support in the making of this book. J. Walter Schultze Tetsuya Osaka Madhav Datta

Part I

Fundamentals

1

Electrochemical microsystem technologies Principles J. Walter Schultze

1.1

Introduction: subjects and definitions

1.1.1

Subjects of electrochemical microsystem technologies

Traditional macroscopic electrochemistry started 200 years ago with a strong technological connection to energy technology and electrolysis. Though electrochemical nano technology has been favoured in the last 10 years it is, in fact a field for fundamental science. As a link between these different fields, micro electrochemistry is a fast growing part of electrochemistry forming interdisciplinary bridges to science and medicine.1–9 Moreover, it finds growing application in various fields of technology. The new subject of Electrochemical microsystem Technologies (EMST), includes and extends the fundamental knowledge of electrochemistry, but in difference to fundamental science, it emphasises the technological application.6–9 This is an urgent activity since otherwise electrochemists do not realise the wide application of their own field, and, on the other hand, electrochemical methods are used without sufficient background and efficiency. Special developments in corrosion research, microelectronics and biology have characterised this insulation of highly specialised fields of engineers and scientists in recent years. During the first International Symposium on Electrochemical microsystem Technologies in Düsseldorf-Grevenbroich,7 the interdisciplinary character of EMST became evident. It is explained in Figure 1.1 and will form the concept of this book. The encouraging aspect is the reversibility of information: electrochemistry as well as the special field take profit from new solutions of problems and new technologies. It is a challenge to integrate all these experiences from various fields, and to use it for a general improvement of EMST. Moreover, the technologies and the systems should be miniaturised to close the recent gap between micro- and nanotechnology. In the present stage this cannot be done as an encyclopedy, except with typical examples of highly developed processes or systems. 1.1.1.1

Fundamentals

First, the relevant principles of physical chemistry, electrochemistry, kinetics and experiments are described. This is necessary due to the decrease of electrode size

4

J. Walter Schultze

Fundamentals Nucleation, Growth materials Surface analysis

Methods Electrodes Kinetics

Microgalvanics

Electrochem. Machining Packaging MEMS, Microelectronics

Electroanalysis Electrochemical microsystem technologies (EMST)

Microsystems

Energy storage microreactors

Bacteria Medical therapy Pharmacy

Biosensors Neurons on chips

Biology Medicine

Figure 1.1 EMST with connections to five special topics of fundamentals, galvanics, systems, analysis and biology. Many special subjects are located in between, because they belong to various topics.

which causes the change from linear to spherical symmetry and the decrease of system dimensions to that of the diffusion layer of the electrolyte and the grain size of materials. Micro electrochemistry started with kinetics of homogeneous reactions.1,10,11 This is a field which profits from micro electrochemistry, but simultaneously the clarification of complex mechanisms is also very helpful, especially for the Scanning Electro Chemical Microscope (SECM),12 and for EMST in general. Electrochemical Materials Science (EMS) is a new field of materials science13 which yields fundamental information on mechanical, electronic and optical properties. But it needs microscopic information on the surface, and, therefore, it profits from the progress of EMST14 and optical measurements.15 On the other hand, fundamentals of nucleation and growth determine the deposition rate as well as the quality of various microstructures.16 1.1.1.2

Microgalvanics

The technologies of microgalvanics are now well established.6–9,17–22 EMST started in electronics with the invention of the printed circuit board in the mm-range, but the dimensions were continuously decreased due to the requirements of electronics.6 Lithography is now a dominating technology.23 The number of interconnections increased, and they had to be miniaturised. While galvanic processes are applied in the nm-range for micro- and nano electronics,24 packaging in m dimensions is still a limiting factor.25 Simultaneously, the development of the LIGA-process19,26 opened up the electrochemical mass production of micro

Principles of EMST 5 mechanic and opto-electronic tools, which now enhance the development of micro reactors for chemical synthesis,27 mixers and other tools. 1.1.1.3

Systems

Chemical reactors and mixers already represent microsystems with growing importance.27 Micro flow systems are required for analytical systems.28 Other, active micro devices need energy delivered by batteries or condensors within the microsystem.29,30 Finally, the application of such systems presumes a high lifetime and corrosion stability which is a challenge for construction of microsystems which allow corrosion protection in the m scale only.31 Other systems will be described briefly in Section 1.4.5. For definitions, characterisation and examples see Section 1.1.2 and Figures 1.2 and 1.3. 1.1.1.4

Electroanalysis

Another highly developed field of EMST can be found in electroanalysis.32–39 Requirements of mass production can be solved by miniaturisation.40–44 Terms like ‘Micro Total Analysis Systems’ (TAS) and ‘lab on the chip’ represent the successful miniaturisation of electrodes and sensors which is of special interest for biosensors.44,45 New concepts of combination of natural systems like a beetle antenna and semiconductor technology41 form a bridge to the application in biology and medicine which is continued in the paper by Fromherz on neurons on a chip.46 1.1.1.5

Biology/Medicine

Figure 1.4 will present some similarities between inorganic and biological systems. The Nobel prize given to Sackmann and Neher3 demonstrated that EMST has been highly developed in biology and medicine 15 years ago.47 The patch clamp technique and the detection of ionic currents through channels in membranes mark the first achievements. Some aspects will be discussed in this book. The control of cell properties like changes of its volume show the progress.48 Micro electroporation will be applied in tumour therapy.49 Other examples refer to bacteria manipulation and deterioration50 which could not be discussed here in spite of great future potential. But it should be emphasised that in biological systems the relevant field strength is often much smaller than in common electrochemical systems (see Section 1.2.4 and Figure 1.9).

1.1.2 1.1.2.1

Definitions for EMST Definition of EMST

After this description of the scope of the subjects we have to define the term EMST to distinguish it from electrochemistry and the microsystem technologies. Electrochemistry covers the whole field from the molecular reaction to the

6

J. Walter Schultze

process, the product and the final system. For example, the electrochemical detection of NO molecule includes mechanistic models and transport reactions as well as material aspects of the sensor. EMST, on the other hand, concentrates on the miniaturisation of single parts and the whole system. Photo electrochemistry treats the general aspects of semiconductor electrochemistry and photons, while EMST uses the principles for microscopic surface analysis and modification. Following the description by Schultze and Tsakova6 we define EMST in the following way. Electrochemistry means all processes of ions or molecules at charged interfaces, that is, faradaic as well as electroless processes. They take place usually at solid/liquid interfaces, but liquid/liquid interfaces may be included, too.51 Micro limits the systems and processes to the range below 1 mm in all three dimensions x, y, z. This yields a clear difference to the well known field of thin film technology which limits only the vertical dimension (z). Therefore, lateral effects in x, y-direction become dominant in EMST. For examples see Figure 1.3. The lower limit of 1m is not sharp. For example, the miniaturisation to ultra micro electrodes of nm dimensions is in progress with common technologies. Moreover, molecular roughness and nucleation have a large impact on EMST in spite of their dimensions in the nm range. System means a combination of two or more parts or devices. A single microelectrode is essential for an electrochemical microsystem, but it represents only a part of it. Due to the large applications, EMST includes electrochemical processes for systems and systems as well. Technology finally limits the field of processes and procedures with general relevance and good reproducibility. As will be discussed in Section 1.4, we interpret EMST in two ways: electrochemical processes for microsystem technologies as well as microsystem technologies for electrochemistry.

1.1.2.2

Definition and characterisation of systems

The discussion of any system needs a characterisation at first.52 Between fundamental micro electrochemistry and EMST there is a large difference of relevant systems. As is shown in Figure 1.2(a) on the right side, fundamental micro electrochemistry prefers homogeneous, symmetric single systems which are flat in most cases. In EMST, on the other hand, technical requirements dictate the topography and other parameter like degree of heterogeneity and symmetry (Figure 1.2(a), left side). According to multiplicity, we have to distinguish single and multiple systems. While single systems depend on the radius r only, r and the distance d have to be given for multiple systems. The difference has a large impact on current density or signal intensity as shown in Figure 1.2(b). If the signal intensity L0 is constant, the resulting signal L of the microelectrode with a radius r is almost constant for periodic systems, and the common equipment can be used. For single systems, on

Principles of EMST 7 (a)

(b)

Order

Sym. (topview) Aspect ratio z A= x

Research

Statistic Irregular Periodic system system system

0-dim.

1-dim.

Single system

2-dim.

+10

0

Periodic systems L~r0 Masking

–10 Positive

Negative

L ~ r2

Flat

1 µm

1 nm Gaussian

Shell

Rectangular

Profile

Lateral Material

L ~r –2

Focussing

log L/L 0

Production

–10

–8

–6 log r(m)

1 mm –4

–2

0

Vertical M

M

I

M

M

Figure 1.2 (a) Classification of micro electrochemical systems by order, symmetry, aspect ratio and material combination; (b) Signal intensity L/L0 in dependence on the electrode radius r. L0standard intensity. For masking and focussing signals see Figure 1.17.

the other hand, the signal L decreases with miniaturisation (L r2 ), for example, for masking techniques. In case of focussing techniques, the signal can be maintained, but the signal intensity L increases with decreasing r, L 1/r 2. This has an important consequence, for example, for laser focussing, when the enhanced signal density causes an increase of temperature.53 Periodic systems (e.g. arrays54,55), irregular systems (printed circuit boards56), and statistic systems (e.g. phosphate layers6) differ by their order. In electroanalysis, disc electrodes are preferred, while in technical systems lower symmetries occur. Another important character is given by the aspect ratio Adepth z/width x which often increases up to 100 for technical systems. Finally, the material composition in vertical or lateral direction is important. We will distinguish insulators I, metals M and semiconductors S according to their electronic conductivity (see Section 1.3.2 and Figure 1.12 for further examples). Applying these characteristics, the Pt disc microelectrode used for amperometric measurements is described as a single, flat (A0), metallic microsystem with 2D-(circular) symmetry suitable for high current densities and exact simulations. In contrast to that, the through hole plating of printed circuit boards represents a multiple, irregular insulator/metal system with a high aspect ratio (A5) suitable for chemical deposition of polymers and galvanic processes at low current densities only.56 Electrochemical machining57–59 yields negative microstructures in a homogeneous material with aspect ratios near 1.

8

J. Walter Schultze

1.2

Electrochemistry in microscopic dimensions

1.2.1

Fundamentals

In microscopic dimensions, there are similarities and differences to the macroscopic electrochemistry. Thermodynamics predict an increase of the chemical potential for a particle radius r 0.1 m, but micro materials have the same as the bulk. Further, charge transfer reactions and nucleation phenomena at the solid/liquid interface are the same. Principle transport and reaction mechanisms in solution are also the same, but the dimensions of the system become comparable Micro- and nano-systems Homogeneous systems

Hetero systems

x

Substrate

Substrate

A>1

A<1

–9

1 µm

1 nm

As

µ-Technologies

–6

Inhibitors Oxides, Membranes

1 mm

log x (m)

–3

Lateral 1 m systems

Double layer

0

Phosphates Protecting films Nernst layer Car technologies

Thin films

pe ct ra z/ x = tio A 1

z

Circuit boards

Whirls LIGA Porous n-Si

Por. p-Si Por. Al2O3 Nano materials

Vertical systems

Atom 1 nm

1 µm

–9

–6

1 mm –3 log z(m)

1m 0

Figure 1.3 Survey of electrochemical phenomena in a double logarithmic plot: HHlayer, oxide films, Nernst diffusion layer, whirls, car technologies and some vertical systems.

Principles of EMST 9 with that of the diffusion layer. Due to the transition from linear to spherical symmetry of single electrodes, the diffusion is enhanced, the limiting current density increases, id 1/r, and the ohmic drop decreases with decreasing r, R r. 6,10 This difference between macroscopic and microscopic systems is maintained for electrode arrays with distances d >> r, but it vanishes for periodic systems with dr or at least similar to r. The influence of the cell diameters on the diffusion layer becomes also apparent in hydrodynamics (see Section 1.2.7). Another important difference arises, if random phenomena such as nucleation, pit formation or accumulation of impurities become dominant. For example, at a typical nucleation rate of 106/cm2 ˙ s, only a single nucleus is formed on 1 m2 in 100 s! In contrast to the thin film technologies, the lateral dimensions have to be considered, if the components of x and y become smaller than that of z. The aspect ratio Az/x (more general, A2z/(xy) for systems with lower symmetry) is important for transport reactions in the electrolyte. Figure 1.3 shows a summary of relevant sizes in a double logarithmic plot. At a birds eye view, a common treatment of scientific, engineering and biological problems seems to be impossible. However, this is not true, as will be Electrochemical processes Inorganic (a)

Metal

Biology/Medicine

H2O

ETR

(b) Electrophoresis

Metal

H2O

e–

+

(c) + Field/ capacity

–

– –

–

Oxide +

e– DNA

Bacteria

– +/–

–

– – –

+

+/–

– – – –

+ –+ –+ – +

–+ –+ –+

+ + +

(d) Metal

Oxide

H2O

Pits/ channels

Figure 1.4 Similarity of fundamental processes of inorganic and biological systems: (a) electron transfer reactions (ETR); (b) electrophoresis and localisation of bacteria by AC polarisation; (c) polarisation of thin films (oxide or cells); (d) formation of pits in passive films or channels in membranes.

10

J. Walter Schultze

explained in Figure 1.4 for some processes of inorganic and biological systems and later in Figure 1.9. All of them are governed by the electric field and mobility of electrons, ions or larger particles. Electron transfer reactions (ETR) (a) take place in the double layer. Transport of larger molecules is well known from electrophoresis (b), and similar phenomena occur during localisation of bacteria in the electric field.50 Polarisation (c) is important for insulating films and biological cells . Channel formation in thin films (d) is known from pitting in corrosion and lipid membranes. 1.2.2

Microelectrodes and microcells

During the nineties, large progress was made by miniaturisation of electrodes.6,10 Problems result from mechanics of electrode production, sensitivity and insulation or leakage currents. Glass technologies compete with etching of metals and silicon technology. Figure 1.5 shows some examples of microelectrodes and electrode arrays. Microelectrodes have been successfully applied in electroanalysis40 and studies of nucleation and growth.16 In biological research, pH-sensitive electrodes in glass technology have been applied. Even four-barrelled, multifunctional electrodes are available.48 In Scanning Tunnelling Microscope (STM) experiments, nano sized metal electrodes are applied. The resolution measured by the tip is in atomic dimensions, while the exposed electrode surface remains in the m range. (a)

(b)

(c) Ag wire

Pt-microdisk electrode

Wax seal Pt-microwire Tip 20 µm

Pt-microwire Lead

AgCl coating Buffer

Tip (magnified) 10 µm

pH sensor

20 µm

Glass capillary

1µm

Tip

(d)

(e)

(f) A

D Ag/AgCl-reference electrode

1 5

3 2

4

Al-bondpads

Pt-counter electrode UMA Alternative Ag/AgCl-reference electrode

Figure 1.5 Microelectrodes and arrays: (a) microdisc electrode for amperometric measurements;39 (b) STM-Tip; (c) pH-sensitive electrode; (d) microelectrode array for research;60 (e) multibarrelled, io selective electrode;48 (f) electrode array for glucose measurements.54,55

Principles of EMST 11 (a)

(b) RE

CE

(c) RE

(d) RE

CE

CE

CE

RE

Gold H2O

WE

WE

Oil

WE

WE (e)

(f)

(g)

(h) CE

RE

CE

RE WE

CE

H2O Silicon rubber

WE

neuron

WE

ox.

red. WE

Figure 1.6 Cell constructions for micro electrochemical experiments (a) water droplet on a photoresist electrode;15 (b) water droplet in oil;14 (c) movable mask;6 (d) scanning droplet or capillary cell;14 (e) optical micro cell;62 (f) biological cell;47 (g) vapour condensation cell with two electrodes, the so-called nanocell; 61b (h) SECM.12

For measurements in microsystems with high aspect ratio, needle shaped, modified or metal disc electrodes are applicable.38,39 For research and development, an irregular electrode array was developed in silicon technology,60 while periodic microelectrode arrays are used in medical research.54,55 Various cells for surface modification or investigation were developed which are summarised in Figure 1.6. The main characteristics are limitation of cell volume, distance of the electrodes, electrolyte flow, and optical transparency.14 The SECM12 is realised in a macroscopic cell, but the local resolution is achieved by localized diffusion. Microelectrodes surrounded by a photoresist can be measured wetted by a nl droplet with introduced counter and reference electrodes. Various modifications of the capillary cell and the optical microcell are described by Lohrengel and Moehring.14 They are similar to the concept of a movable mask. The smallest electrochemical cell with two electrodes is realised by vapour condensation between a substrate and a SPM-tip.61 It is useful for preparation of S/I structures in the nm range.61b Biological cells represent microcells with a large electrolyte volume outside of the cell. They are described in Chapter 21, this volume. 1.2.3

Current and charge densities

In EMST, current densities in the range between A/cm2 and A/cm2 are applied in potentiostatic or galvanostatic experiments. The measured current, I r2

12

J. Walter Schultze

I =1MA/cm2

0

1A/cm2

–2 EC Machining

log I (A)

–4

Reactions

Common laboratory experiments

PCB Microelectrochemistry

–6

–8

Fast kinetics

Analysis LIGA Modification

STM –10 Faradaic Biol. nano channel structure currents

–12

–14

–12

–10

1µA/cm2

Amperom. Sensors Potentiometric

Surface analysis

–8 –6 log S (cm2)

–4

–2

0

Figure 1.7 Current I in dependence on the surface area S for various processes in microelectrochemistry (108 cm2 S 102cm2). Electron tunnelling in STM and faradaic processes in the range of 1010 cm2. Typical ranges for EMST processes according to Ref. 6.

decreases with miniaturisation. Figure 1.7 shows some typical examples. The lower limit for electrochemical experiments is in the range of about 10 fA63 for dc and 100 pA for ac measurements. Otherwise, such measurements have to be carried out on multiple microstructures. Technical processes can be carried out with common techniques, since they are usually taking place at periodic systems. Faradaic, capacitive or electronic tunnel currents have to be distinguished. In STM experiments, currents are measured at single surface atoms, that is, for 1015 cm2, but they are caused by very large electronic tunnel currents of I 10 pA.6 The minimum quantity which can be measured is also limited by the minimum charge which has to flow: this is about a few fC as long as no multiplication techniques are available. Thus, fast faradaic processes may be detected for S 1010 cm2, but capacitive currents for 107 cm2 only. Surface reactions could be measured in case of oxide formation down to 100 nm,61b while A. Bard could detect single ions due to repeated transfer between two electrodes.64 A verification of such considerations on the experimental limitations was carried out by Buß65 with various microelectrode arrays for research and development. On gold electrodes, the charge of oxide formation was determined as well as the double layer capacity. Results are shown in Figure 1.8 in dependence on the

Principles of EMST 13 (b) –4.5

(a) 4

–5.0 (a)

log q

2 1 0 –1

–5.5

(b)

–6.0

Theory m = 2 (a)

–6.5

(b) log Q

–2

–7.0 log (C/F )

log Q (µc); log q (µC cm–2)

3

–7.5 –8.0 –8.5 –9.0 –9.5 C (exp.)

–10.0 –10.5

–3

–11.0 –4 –4.0 –3.5 –3.0 –2.5 –2.0 –1.5 –1.0 –0.5 log a (cm)

–11.5 –4.0

C (theo.) –3.5

–3.0

–2.5 –2.0 log a (cm)

–1.5

–1.0

Figure 1.8 Test of electrochemical data of quadratic electrodes in dependence on electrode length a 艐2r. (a) Charges Q and charge densities q of oxide formation on gold microelectrodes measured on various electrodes (b) Double layer capacity. Deviations at a 10 m are caused by parasitic capacities.65

electrode length a (a 艐2r) of quadratic microelectrodes. The faradaic charge of oxide formation is proportional to r2 for r2 m. Edge effects (e.g. delamination) cause errors for the oxide charge at small r in dependence on the passivation quality.31b,65 But for double layer capacities, the surface area of the electrode is not the only source. Parasitic capacities of leads and the Si substrate have also to be considered. Therefore, deviations occur for r10 m already. Thus, the minimisation of the RC-value of microelectrodes requires special care. 1.2.4

Field strength in microstructures

Another important parameter is given by the potential U and the field strength F (V/m), respectively. The potentials are in the range of some V for most electrochemical experiments. For conducting electrolytes of sufficient concentration, electrochemistry is concentrated at the interfaces in the nm range. In systems with insulating substances, however, some kV are necessary. The correspondent field strength decreases with increasing z as long as no space charges are present. Figure 1.9 gives some examples of inorganic and biological systems. For the growth of insulating oxides, a typical field strength of 108–109 V/m is necessary. 1.2.5

Spectro electrochemical and other in situ techniques

Full understanding of electrochemical phenomena requires energetic, topographic and other information available from spectroscopy, mechanics and

14

J. Walter Schultze HH layer Membrane –10

log F (V×m–1)

8

Bacteria Neurons

ETR Ionic migration Photo desinfection Electroporation

6

Membranes

4

Field desinfection

Capillary electroBacteria phoresis localization Field potentials

2

1 kV Electric fish 1V Conductance

Nano Micro Electrochemistry

0 1nm –9

1 µm –6 log z (m)

1 mV 1 mm –3

Figure 1.9 Double logarithmic plot of field strength F vs distance z for biological systems and insulating oxides. Typical cell voltages (kV) and applied electrode potential differences (mV) as parameter. The field strength of photodesinfection (e.g. on TiO266) and field desinfection is similar, but the voltages differ. For electroporation see Ref. 49. Localisation of bacteria,50 field potentials in brain.67

other fields.13,15 Here we will mention the in situ methods which are available for EMST. The interaction with photons yields structural, energetic and kinetic information. A classification can be derived from the input and output signal, differentiating between optical (photons) and electrical (i, U, C) signals. Table 1.1 gives a survey on typical methods, their information and resolution. For the optical methods a description will be given by Michaels et al.15 Mechanical properties like hardness, friction and other interactions in the m and nm range can be measured by Atomic Force Microscopy (AFM), nano indenters and similar techniques. This is described by Nagahara43 and in Refs. 68, 69. 1.2.6

Resolution of time and space

The resolution of processes in time and space is a complex problem. Figure 1.10 gives a double logarithmic plot of general phenomena. At first we will discuss the resolution in time, since EMST takes place in three different time scales. ●

Preparation of the microstructure. That includes nucleation (within s), growth or etching (up to some ks) and post treatment (usually ks to Ms). The

Principles of EMST 15 Table 1.1 Electrical and optical in situ methods for EMST Process

Signal in

Signal out

Result

Information

Resolution

Faradaic process Electrical Electrical

I(U)

Kinetics

Double layer charging e-Tunnelling

Electrical Electrical

Light emission Photocurrent

Electrical Optical Optical Electrical

AME, Ellipsommetry -reflectometry

Optical

Optical

Optical

Optical

-Raman

Optical

Optical

U(q) or q(U) Capacity C, thickness I(U, z) Topographic, electronic h Electronic I(U, h) Electronic, band structure Delta, psi, Structural alfa Intensity () Electronic, optical V Vibration, structure

1 m (phoresist) nm (STM) 10 m

Electrical Electrical

nm (STM) 10 m 50 m 10 m m

0 Application medicine

Information –2

log z (m)

El. conduction

Microtechnology

–4

–6

Pits

Devices

Neurons Bacteria Lipid films

–8 Nucleation –10

Clusters Channels

Atoms Vibration

–10

–5

0 log t (s)

5

10

Figure 1.10 Resolution of electrochemical effects in time t and depth z in a double logarithmic plot. Effects of inorganic systems: atom vibrations, nucleation, pits etc. Effects of electronic systems: devices, electronic conduction, information. Effects in biological systems: channel formation in membranes, measurements on bacteria and neurons, application in medicine.

16

J. Walter Schultze adjustment of nucleation, diffusion, migration and ion transfer reactions requires sometimes special potential time programs as in pulse plating16 or electrochemical machining, Period of standby of the structure (up to 10 Ms), and Application (between s and Ms) which depends very much on the type of application.

● ●

An example proving the extreme conditions is given by the airbag sensor which will be electrochemically produced in some 1 ks, stands by for (hopefully) many Ms encapsulated in an inert atmosphere, but has (mechanically) to react within ms. The electrochemistry of the microstructure changes in dependence of the surrounding electrolyte and gas atmosphere, respectively. Due to the small dimensions of the microstructure, pits, cracks and other corrosion phenomena can cause a complete failure within some seconds. Therefore, the efficient and economic preparation of corrosion stable micro structures requires a lot of special know-how.31 1.2.7

Hydrodynamics

Micro fluidics are of great interest for special problems, but they have scarcely been investigated. Fields of interest exist for micro reactors, sensors, electrochemical machining (ECM)57–59 and fast galvanics.70–72 The thickness of the Nernstian diffusion layer and the Prandtl layer are in the same order of magnitude as the structure. Therefore, a laminar flow is assumed for most systems. Moreover, many systems are operated in quiescent solution. Experiments were carried out in the optical micro cell at flow rates up to 1 m/s. Calculated Reynolds numbers (a) 0.2

(b) 2.0 U = –0.7 V

0.0 1.0

–0.2 I (µA)

–0.6

6 µl s–1

–0.8

9 µl s–1

–I diff.(µA)

0 µl s–1

–0.4

–1.0 –1.2 –1.4 –1.6 –0.8 –0.6 –0.4 –0.2 U (V)

27 µl s–1

1.2 0.8 Fig (a)

0.4 0.0

0.0

0.2

0

5

10

15 20 v (µl/s–1)

25

30

Figure 1.11 [Fe(CN6)]3 reduction in a microcell (described in Ref. 62); (a) Experimental current density in dependence on the flow rate. A flow rate of 1 l/s corresponds roughly to a velocity of 2.2 cm/s; (b) Limiting current in dependence on the flow rate under potentiostatic (U0.7 V) and potentiodynamic conditions from (a).68

Principles of EMST 17 Table 1.2 Typical reactions for microstructuring: ITR, ETR and photoreactions Process

Type of reaction

Formula

i: Metal deposition i: etching, ECM Anodic oxidation Anodic oxidationh

ITR

Mzze → M

ITR ITR induced by laser light Cathodic ITR, anodic ETR Precipitation induced by ETR ITR, induced by laser light ITR/ETR, induced by laser light

MzH2O → MOz2zH2ze MzH2O → MOz2zH2ze

Chemical metal deposition Electrodeposition of paint Photodissolution, photocorrosion Photopolymerisation

Mzze → M CH2OH2O → CO24H4e Rx–COO–H → RxCOOH He → s1 H2 Si6HF4ph → SiF62 6H R–H2h → Rn2H2e

R100 Rcrit 2300 indicate a laminar flow.70 In jet experiments by Unwin71 and ECM,73 velocities up to 30 m/s are realized. Then microscopic whirls have to be taken into account which can cause rough surfaces (Figure 1.11).

1.3 1.3.1

Reactions and materials Reactions of micro electrochemical processes

In principle, all electrochemical reactions known in a macroscopic system can occur in micro structuring processes. For the formation of positive or negative structures, reactions involving Ion Transfer Reactions (ITR) from the electrolyte to the solid phase or vice versa can be used. Besides that, ITR can be induced by ETR, for example, in chemical reactions. Electrophoretic reactions are also used. Typical reactions are realised in metal deposition or dissolution, oxide formation and polymerisation. Gas evolving reactions like hydrogen evolution, or pure ETR take place and have to be taken into account in electroless processes. Pure ETR are important for the SECM, but not for micro structuring. Table 1.2 gives a summary of most important reactions. 1.3.2

Materials properties and combinations in EMST

As can be seen from Table 1.2 and Figure 1.13, the choice of materials for combination of substrate and product materials allows a large variety of ITR, ETR or photo electrochemical reactions. Systems can be built up by a single element, compounds of the same element or various element/compound combinations. Even for the same element, the preparation of a microstructure can produce combinations of metals or semiconductors with insulators. The formation of conducting, semiconducting or insulating oxides on a metal or semiconductor yields very different combinations.16

18

J. Walter Schultze

(a)

(b) I2

I2

M

(c) I

Comp.

I

M2 M1

I1 (e)

(d) M3 M2 M1

I

(f) M

M⬘ Pol. I

M

I

M X I1

I2

I2

(g) O

M1 M2

O

Pol.

O

Pol.

S

Figure 1.12 Typical combinations of materials for microstructures: M metal, I insulator, S semiconductor, O oxide, Pol. conducting polymer, Comp. composite film; (a) lateral metal structure M insulated laterally by I2 on top of insulator I1 (b) composite film on metal M2 for sensors; (c) microstructure with three metals, M2 as sacrifice layer; (d) metal on insulator with an interfacial layer X for adhesion or nucleation; (e) vertical structure metal/conducting polymer/insulator, for example, in holes of Printed circuit boards with M, M Cu; (f) vertical MOM structure, for example, Cu/TiO2/Ti;84 (g) lateral structure conducting polymer Pol/O/Pol on S, for example PBT/SiO2/PBT on Si.52

Applying the physical nomenclature, possible substrate/product combinations can be characterised by the electronic conductivity, using the symbols M (metal), S (semiconductor), I (insulator), O (Oxide), Pol (polymer), X (others), E (electrolyte), numbers indicating different material (Figure 1.12). For systems like batteries or sensors, the ionic conductivity has also to be defined. For preparations, we have to distinguish between lateral and vertical structures according to Figure 1.2. For the rate of vertical structure formation (deposition or removal) the local field distribution at the interface and within the structure is important. While at the surface of conducting phases, the potential drop is concentrated in the Helmholtz layer, in semiconducting or insulating phases a large potential drop may occur in space charge layers or across the whole insulator. Therefore, the formation of thicker deposits is easy for metallic phases and at small overpotentials for semiconductors. But for non-conducting phases it is limited as can be seen from Figure 1.13.13 For insulating films on metals, the high field law causes a constant growth of a typical thickness proportional to U, for example, d2 nm/V. For n-Si and other n-type semiconductors, the potential drop within the semiconductor has to be considered. It can be eliminated, however, by illumination. The field distribution has a large impact on micro- and nano-structuring. By analogy to the so-called Wagner number, the EMT number EMT R(pol)/R(el) was defined by the ratio of the polarisation resistance of the microstructure and the electrolyte resistance.9c For a localisation of the electrode reaction in an electrolyte, the condition EMT << 1 must be fulfilled. For example, vertical nano-structures of insulating SiO2 can be prepared on p-Si (and on n-Si by illumination) by a field dependent process only in almost monomolecular films of condensed water, but not in an extended electrolyte which dissipates the field necessary for film growth.61 On the other hand, self catalysing reactions like pitting enhance the localisation.

Principles of EMST 19

Coating

Semiconductors

2

Spark oxidation (‘Anof’)

EDP (paint) ‘Eloxal’

Oxide growth

Metals (Al)

High field

1

0

Semiconductors (n-Si)

Nuclei

log z (nm)

3

Galv. dep.

Metals

4

0

1 2 log (U +U0) (V)

3

Figure 1.13 Double logarithmic plot of the thickness z vs overpotential UU0 for vertical microstructures. Galvanic deposition of conducting metals and semiconductors at low overpotentials up to large z-values. Electrodeposition of paint.74,75 Formation of insulating phases limited by the field strength.13 Influence of the potential drop within the substrate.

Electrodeposition of paint74,75 is used to insulate metal tips or for fixation of functional substances, for example, on ion selective electrodes.76 1.3.3

Substrate properties and preparation

Typical substrates of EMST are Si, Ti, Fe or insulating polymers. In general a clean, homogeneous flat surface is preferred, but adhesion requires a rough surface or a seed for nucleation of the microstructure. Therefore, electrochemical surface treatment of the substrate is often necessary. For negative microstructures, on the other hand, ECM represents a successful technology for high aspect ratios.57 The great advantage of EMST is given here by the attack of the surface atoms without mechanical stress within the substrate. Figure 1.14 gives a survey of smoothing and etching technologies. Chemical properties of the substrate are important for the interaction with the first layer of product and therefore, for the adhesion. Dislocations and surface inhomogeneities determine the nucleation. EDX and Scanning Auger Microscopy will be used to check chemical impurities and surface layers. Details of the electronic structure are important for rate of processes, especially at semiconductor surfaces. Porous silicon is a typical example. Physical properties include the surface roughness in microscopic or atomic dimensions. In the atomic scale, the crystallographic orientation or atom density of

20

J. Walter Schultze Smooth surfaces M

H2O

A = z/x =1

en

El. chem. mach. Etching porous Si

gh R

ou

log x (m)

–3

in g

Electropolishing

Brigthener

0

–6

Vertical structures

Eloxal

Pores

M

–9 Membr.

Atom –9

–6

–3 log z (m)

0

Figure 1.14 Electrochemical preparation of structured surfaces by ECM, roughening, etching, electropolishing, etc. EloxalElectrolytically oxidised aluminium.

Micro

Phosphates Metallic layers paint Pores

Structured pores Si, Al2O3 –9

–6 –3 log z (m)

Oxide films

log x (m) –6

A =1

Nano

Screw disloc.

Micro –3

Passivation layers

Macro Micro

Grains

A =1

Crevices

Atom Nano –9 metals –9

Macro 0

Inhibitors

–6

Steps, double layer

log x (m)

Micro –3

(b)

Foils

Weld

0

Roughness

Mole- Ultra- Thin films cular thin

Nano

(a)

Atom –9

–6 –3 log z (m)

Figure 1.15 Log/log plot for materials properties: (a) metals (grains, screw dislocations, crevices, welds), (b) coatings. A aspect ratio.

Principles of EMST 21 Grain b~(1010)

(b)

i (mA cm–2)

(a) Grain a~(0001)

500 µm

4 3.5 3 2.5 2 1.5 1 0.5 0

Beginning O2-evolution 0–8 V

Only oxide formation

0–10 V a

0–6 V

0–2 V 0–4 V b 0

2

4

6

8

10 U (V)

Figure 1.16 Inhomogeneities of polycrystalline substrates and their role in electrochemical processes: (a) microscopic picture of Ti oxidised to 15 V. Interference colours of single grains can be correlated with the crystallographic orientation of the grain; (b) cyclovoltamograms show oxide formation and oxygen evolution, respectively.15,78

the surface plane determines the rate of processes. Since single crystals and homogeneous surfaces are expensive, the surface structure of polycrystalline materials becomes important. This can be studied by Anisotropy Micro Ellipsometry (AME) in case of anisotropic materials15 or by Electron Back Scattering Diffraction (EBSD). The latter can determine all three Euler angles with an accuracy of 5º with a high local resolution of 1 m. Based on such a principal investigation, other effects which are strongly connected with the grain orientation, can be used, too. As an example, the EBSD-analysis of a Ti-surface and the interference colours of anodic films on it can be used for determination of the surface orientation which has a great impact on corrosion and reactivity (Figures 1.15 and 1.16).77 1.3.4

Localisation of processes

For the preparation of microstructures, the localisation of electrochemical reactions has to be realised. Depending on the intended profiles shown in Figure 1.2, methods with sharp blocking or a focussed signal can be chosen. Figure 1.17 gives a summary of principles. Geometric blocking, for example, by a photo resist,23 gives the sharpest structures. Therefore, lithography is widely applied with all variations. However, complications arise for thicker structures, if the reaction mainly takes place at the boundary which is observed for chemical deposition, or formation of conducting polymers. Chemical modification can be used for local inhibition or catalysis. Local induction of nucleation and growth by foreign clusters is used for electroless deposition, for example, by Pd deposition on insulators, etc. Localised deposition of a solid state reactand causes a good localisation of the product.52 Local insulation by adsorbed poisons, inhibitors, or by surface modification, for example, ion implantation, gives sharp boundaries for the first atomic layers only.52 Focussed signals, for example, electric field, laser light, etc., usually yields Gaussian profiles. Localisation by surface tension is well known from preparation

22

J. Walter Schultze (a)

(b)

(c)

Inhibition Photoresist

Catalysis

X

Laser focus

X

Adsorbed poison

(d)

Jet

Nuclei

Movable mask

HA

X

Wetting droplet

Implanted ions

X

SECM CE

Red

Educt

X

Electric field +

Ox Red Educt

Ox –

–

X

–

Figure 1.17 Principles of localisation of electrochemical reactions by (a) geometric blocking,52 (b) chemical modification, (c) localised signals52 or (d) transport limitation by a jet or the SECM.12,69

Lens

Light scattering

Gas x(prim.)

x(sec.) + Secondary

+

Redeposition of products Concentration shifts

∆H Semiconductor

Primary

Figure 1.18 Schematic representation of primary and secondary effects of microstructuring: primary effects by focussing of the signal, secondary effects by dissipation of heat, migration of holes and light scattering.

of insulated STM-tips, and it is used for the scanning droplet cell.14 Transport of educts can be focussed by the SECM or in a jet.71 In case of focussed signals, primary and secondary effects have to be distinguished. Primary effects are due to focussed signal. Usually, secondary effects broaden the microstructure. They can be caused by transport of educts, products or heat as is schematically shown in Figure 1.18. Typical examples are given by laser induced reactions.79 Diffusion of holes or electrons in the substrate dissipate the focussed signal. Formation of gas bubbles with a scattering effect gives

Principles of EMST 23 another example. In the anodic polymerisation, on the other hand, diffusion of oligomers usually causes broad deposits. The heat transport is important for laser induced metal deposition.80 1.3.5

Corrosion of microsystems

After all these discussions of reactions (Table 1.2), time scale (Figure 1.10), and materials properties (Figure 1.15), it becomes clear that the stability of microstructures in electrolytes or wet atmospheres requests special care. In principle, corrosion in the gas phase can be avoided by encapsulation, for example, for electronic devices, but simpler passivation films are preferred for economic reasons.31 In electrolyte, penetration of water and aggressive ions is fast. Thus, a stability of sensors for some days is a good result.

1.4 1.4.1

Special aspects of EMST Scaling down and multiplication

The scaling up of a process for industrial realisation is well known as a standard problem of chemical engineering. In EMST, the problem is opposite: the first step consists of scaling down or miniaturisation of products (Figure 1.19). After clarification of the microsystem, the technical realisation does not require the scaling up in a single process, but the multiplication (numbering up) of the micro technology by a more or less periodic system. The phosphating process is a good example.6,13 The first step required scaling down from a laboratory single crystal of phosphate to crystals on a steel. In the next step, nucleation density was adjusted to the technical system. Then, layer properties were improved by microscopic local elements. Finally, the multiplication in the technical car production with about 1012 crystals per car body has to be achieved. Another example is given by the fuel cell development: while the catalyst has dimensions of few nm, the Membrane Electrode Assembly (MEA) requires micro technological procedures, but the production of the fuel cell presumes a lateral scaling up to a cell in the cm-range. Finally, these cells have to be multiplied in the stack to the final aggregate. Figure 1.19 gives the relation between particles, sizes and area. This demonstrates the long way from fundamental science via materials science to microsystem technologies and the industrial production at the end. 1.4.2

Flow diagrams

In EMST, various process step in series are necessary for realisation. Typical examples are substrate preparation, prestructuring of the substrate, modification by activation which all take place before the main process of materials deposition or removal. Then, one or more process steps of deposition follow. After that, a post treatment (annealing, illumination, fixation) is often applied.6,52

24

J. Walter Schultze

(g) 18

(f)

Ballard MK-5 fuel cell stack

6

Phosphating Layer (c)

S

(b)

Materials science

Nanomaterials

(e)

Micromaterials

log Ncryst

T m ech ul n tip ol lic og at y: io n

12

(a) single crystal

(d)

Course grain 0

Surface science

Scaling down 1 nm

1 µm2

–12

–6

2

log (S/mm2)

1 mm2

1 m2

0

6

Figure 1.19 From Materials Science to technical production: schematic representation of scaling down in research, and multiplication in technology. Nnumber of crystals, Ssurface area. Examples are given for phosphating (micromaterial, (a–c, g) and fuel cells (nanomaterials, (d–g)); (a) single crystal of phosphate grown in solution; (b) crystals grown on a steel surface with defined orientation,6 (c) technical layer; (d) TEM-picture of a nanocatalyst for fuel cells; (e) membrane electrode assembly with a microtechnology to combine the catalyst, the coal, and nafion membrane, cell stack and (g) car with about 1018 crystals of catalyst and 1012 phosphate crystals.

Processes take place at different chemical and electrochemical conditions. Hence, an extended description of regulation of temperature, potential (or current) pH etc. is necessary.6 Flow diagrams and a description in the Pourbaix diagram is appropriate. Figure 1.20 gives an example for the production of an irregular, negative, lateral microstructure I/Pol/M with an aspect ratio up to 5. Typical production rates of a factory are 1000 m2 of PCBs per day.

Principles of EMST 25

1 MANOX

PEDT

MnOx

3

PEDT

Cu EPOXY

pH

(b)

3 CUPROSTAR

5

U (SHE)/V

EPOXY

7

5

1.1

1

1

1.0 0.9

0.8 1 µm

6 min

Rinsing

d (MnOx) 0 d (Polymer) 0 0 d (Copper)

EPOXY

3 min

0 –1.5 200 nm 20 µm 15 min Process step

2

U (SHE)(V)

Cu MnOx

2 DMS-E

Rinsing

(a)

MnO–4

2 1 1

MnO2

Mn2+ 2+ Cu 3 Cu

0 0

2

Mn2O3 CuO 4

6 pH

8

10

12

Figure 1.20 Flow diagram and Pourbaix diagram of a micro structuring process: through hole plating of Printed circuit boards according to the Blasberg process (‘Manox, DMS and Cuprostar’). 1: oxidation of epoxide substrate by (KMnO4), 2: deposition of Poly-Ethylene Dioxy Thiophene PEDT, 3: metalisation by copper deposition.52,56 (a) flow diagram;6 (b) representation of the electrochemical processes in the Pourbaix-diagram.

1.4.3

Micro technologies for EMST

In multistep processes of microsystem productions, electrochemical technologies are involved in various steps of pretreatment, structuring, posttreatment and application. Therefore, EMST has three large fields: ● ● ●

microsystem Technologies (MST) for electrochemistry Electrochemical processes for MST (see Section 1.4.4) Electrochemical microsystems (see Section 1.4.5).

MST for electrochemistry covers the necessary background knowledge of electrochemists: it includes all non-electrochemical techniques for deposition, removal or bonding of materials like PVD, silicon technologies, glass etching (Foturan), spark erosion, milling, drilling, laser pulling of metal wires, glueing, passivation etc. The relevance of MST is due to their impact on the following electrochemical processes and applications. A summary is given in Ref. 6. 1.4.4

Electrochemical processes for MST

Electrochemical Micro Technologies for non-electrochemical systems include all galvanic and electroless processes, material preparations and anodic bonding. Typical examples can be found in electronics. A survey is given in Figure 1.21. For example, a large group of electrochemical work has been done

J. Walter Schultze

log x (m) log dy (m)

Substrate

x

Packaging Mechanics Laser Foturan LIGA –6 1 µm Si Glass, wire Vertical techniques

–9 1 nm 1 nm

z

A =1

–9

1 µm

1 mm

–6 –3 log z (m)

Lateral techniques

A =1

–3 1 mm

–3 1 mm

A >1 A >1

(b)

Lateral techniques

–6 1 µm

SECM

A = z/x

Polishing

(a)

log x (m)

26

ECM LIGA Galvanising Nanopores

–9 1 nm

Vertical techniques 1 nm

–9

1 µm

1 mm

–6 –3 log z (m)

Figure 1.21 Survey of some microtechnologies: (a) microsystem technologies for electrochemistry;6 (b) electrochemical technologies for MST: galvanising in electronics,24 ECM,57 LIGA.19,24

by preparation of nano- and micro materials. Magnetic materials are especially important.13,81 The LIGA process is used for clock manufacturing. Magnetic materials are necessary for heads.81 Further, the production of airbag sensors, jets and the application of ECM for razor heads, masks and jets have to be mentioned. An example for large scale microprocesses is given by the phosphating process. For electronics, the production of PCBs and packaging is very important.6,25 1.4.5

Electrochemical microsystems

Electrochemical microsystems and products of EMST are listed out below. Many of them are just under development. ●

●

●

●

Electrochemical sensors and other devices for analysis, for example electrochemical TAS, electrochemical ‘lab on chip’ devices,40 the multifunctional electrodes of biologists.48 In addition to that physical detectors like angle sensors based on conductivity etc. should be mentioned here. Finally, capillary electrophoretic systems belong to that group.82 Systems for chemical microproduction include microcells for local galvanic processes or modifications of surfaces.14 The jets for electrolyte allow high current densities for galvanic processes.72 Systems for chemical and electrochemical reactions are closed cells for special application. They include electrochemical microreactors27,83 and fuel cells. Systems for charge storage include condensors for small charges and high voltages and batteries and supercaps for large charges and low voltages. Figure 1.22 shows the opposite dependence of these systems on the thickness of the functional layer.13 Microcoulometers can be discussed here, too.

Principles of EMST 27

Active material

Macro system (1×1 cm2)

0

Battery C =nzF/Vm – –

m = +1

–2

Charged

–4

Discharged

+ +

d

Double layer

log C (F)

+–+–+– –+–+–+ +–+–+– –+–+–+

–6

zsc (U, N) m = –1

–8 –10

Micro system (10 × 10 µm)2 Insulator + – – + + DD – 0 Condensor + – – + C =DD0 /z – +

d zsc (U, N)

–12

–10

–5 log z (m)

0

Figure 1.22 Capacitive systems for MST in a double logarithmic plot of C vs z. Active materials for batteries with C z (m 1), and condensors with C 1/z (m 1). Full line: micro capacitor with 10 10 m2, dotted line for a macrosystem with 1 1 cm2. For semiconductors the space charge layer dsc f(U, N) must be considered. ●

1.5

Systems for fluid systems are strongly connected to electrochemical systems, even if the working principle is not electrochemical. Mixers, emulsifiers,27 ink jets, dosimeters and flow detectors should be mentioned in this group.

Conclusions

The young field of EMST is fast developing. It enriches many fields of science and technology, since it offers alternative approaches to mechanical or physical techniques. Advantages are due to the ● ● ● ● ● ●

localisation of the electric field at the solid/liquid interface absence of mechanical stress and tenses below the surface regulation of process rates by potential or current high sensitivity large variability of materials with differing electronic or ionic conductivity flexibility of chemical processes and properties.

On the other hand, EMST is not an unique, isolated technology. It needs other technologies and integrates them, and often it has to compete. The dimensions and experimental limits of EMST become smaller and smaller. Higher sensitivity of electrochemical equipment, progress in mechanical technologies by application of piezo techniques, and improvement of other MST

28

J. Walter Schultze

allows a continuous progress of miniaturisation. Starting from the mm-range about 20 years ago, many techniques are now available in the m-range. In contrast to the – at first – insulated success of STM in the nm-range, EMST maintained the connection to neighbouring dimensions. This gives EMST the chance to form a bridge between the traditional macroscopic electrochemistry and the electrochemical nano science. That was the reason to change the title of the International Symposium on Electrochemical microsystem Technologies into ‘International Symposium on Electrochemical Micro and Nano Technologies’ which will be held in Düsseldorf 2002. We are sure that this book will catalyse further progress in that field.

Acknowledgement The support of this work by the Ministerium für Schule, Wissenschaft und Forschung of Nordrhein-Westfalen by the joint projects ‘Integrierte Mikrosystemtechnik für Fest/Flüssig-Systeme IMST’ und ‘Elektrochemische Mikro- und Nano-Systeme Elminos’ is gratefully acknowledged. We obliged to Elsevier for permission to reproduce some figures taken from Electrochimica Acta.

References 1. S. Pons, M. Fleischmann, Anal. Chem. 59 (1987) 1391A 2. The liquid/solid interface at high resolution. Faraday Discuss. 94 (1992) 3. E. Neher, Angew. Chemie 104 (1992) 837; B. Sakmann, Angew. Chemie 104 (1992) 844 4. J. W. Schultze, M. Schweinsberg, Electrochim. Acta 43 (1998) 2761 5. W. J. Lorenz (ed.), ‘Scales in Electrochemical Systems – From Angstroms to meters’, Electrochim. Acta 43 (1998), Special issues 19 and 20 6. J. W. Schultze, V. Tsakova, Electrochim. Acta 44 (1999) 3605 7. J. W. Schultze (ed.), ‘Electrochemical Microsystem Technologies’, Electrochim. Acta 42 (1997), Special issues 20–22, 2981–34061 8. T. Osaka (ed.), ‘Electrochemical Applications of Microtechnology’, Electrochim. Acta 44 (1999), Special issues 21 and 22, 3603–3952 9. (a) J. W. Schultze, G. Staikov (eds), ‘Electrochemical Micro- and Nano-System Technologies’, Electrochim. Acta 47 (2001), Special issues 1 and 2, 1–385; (b) J. W. Schultze, G. Staikov (eds), ‘Scaling Down in Electrochemistry: Electrochemical Micro- and Nano-System Technology’, Elsevier Amsterdam, 2001; (c) J. W. Schultze, A. Bressel, Electrochim. Acta 47 (2001) 3 10. J. Heinze, Angew. Chemie, Int. ed. Engl. 32 (1993) 1268 11. J. Heinze, K. Borgwarth, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, p. 32–65 (this volume) 12. M. V. Mirkin, B. R. Horrocks, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 66–103 (this volume) 13. J. W. Schultze, Electrochim. Acta 45 (2000), 3193 in: J. W. Schultze, M. Musiani, T. Osaka (eds), ‘Electrochemical Material Science’, Electrochim. Acta 45 (2000), Special issue 20, 3193–3438

Principles of EMST 29 14. M. M. Lohrengel, A. Moehring, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 104–121 (this volume); M. M. Lohrengel, Electrochim. Acta 42 (1997) 3265; A. W. Hassel and M. M. Lohrengel, Electrochim. Acta 42 (1997) 3327 15. A. Michaelis, S. Kudelka, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 122–155 (this volume); S. Kudelka, J. W. Schultze, Electrochim. Acta 42 (1997) 2817 16. G. Staikov, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 156–184 (this volume); E. Budevski, G. Staikov, W. J. Lorenz, Electrochemical Phase Formation and Growth, VCH, Weinheim, 1996 17. L. Romankiw, Electrochim. Acta 42 (1997) 2985; L. T. Romankiw, D. R. Turner (eds), Electrodeposition Technology. Theory and Practice, Electrochemical Society, Pennington, NJ, 1987 18. N. Masuko, T. Osaka, Y. Ito (eds), Electrochemical Technology: Innovation and New Developments, Kodansha and Gordon and Breach, Tokyo and Amsterdam, 1996 19. W. Bacher, in: N. Masuko, T. Osaka, Y. Ito (eds), Electrochemical Technology: Innovation and New Developments, Kodansha and Gordon and Breach, Tokyo and Amsterdam, 1996, p. 159; Prospectus ‘Microfabrication by the LIGA process’, Kernforschungszentrum Karlsruhe, Institut für Mikrostrukturtechnik 20. T. Osaka, in: N. Masuko, T. Osaka, Y. Fukunaka (eds), New Trends and Approaches in Electrochemical Technology, Kodansha and VCH, Tokyo and Weinheim, 1993, p. 13 21. K. Ouchi, in: N. Masuko, T. Osaka, Y. Ito (eds), Electrochemical Technology: Innovation and New Developments, Kodansha and Gordon and Breach, Tokyo and Amsterdam, 1996, p. 197 22. O. Shinoura, in: N. Masuko, T. Osaka, Y. Ito (eds), Electrochemical Technology: Innovation and New Developments, Kodansha and Gordon and Breach, Tokyo and Amsterdam, 1996, p. 219 23. S. Matsui, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 187–223 (this volume) 24. P. Andricacos, in: The Electrochemical Society Interface, 1998, p. 2; P. Andricacos, in: IBM J. Res. Devel. Electrochem. Microfab. (Special Issue) September 1998 25. H. Honma, H. Watanabe, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 224–244 (this volume) 26. W. Ehrfeld, in: H. Reichl (ed.), Microsystem Technologies, Springer Verlag, Heidelberg, 1990, p. 521; P. Bley, ibid., p. 302 W. Menz, P. Bley, Mikrosystemtechnik für Ingenieure, VCH, Weinheim, 1993, p. 189 27. H. Löwe, W. Ehrfeld, J. Schiewe, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 245–268 (this volume) 28. S. Shoji, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 321–328 (this volume) 29. A. Yoshida, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 271–285 (this volume) 30. B. B. Owens, C. L. Holmes, J. Bates, W. H. Smyrl, N. J. Dudney, B. J. Neudecker, S. Passerini, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 286–320 (this volume)

30

J. Walter Schultze

31. (a) G. Schmitt, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 329–368 (this volume); (b) G. Schmitt, J. W. Schultze, F. Faßbender, G. Buß, H. Lüth, M. J. Schöning, Electrochim. Acta 44 (1999) 3865 32. R. M. Wightman, in: A. J. Bard (ed.), Electroanalytical Chemistry, Vol. 15, Marcel Dekker, New York, Basel, 1989 33. R. C. Thomas, Ion-Sensitive Intracellular Microelectrodes. How to make and use them, Academic Press, London, 1978 34. A. J. Bard, G. Denault, Ch. Lee, D. Mandler, D. O. Wipf, Acc. Chem. Res. 23 (1990) 357 35. W.-R. Schlue, W. Kilb, D. Günzel, Electrochim. Acta 42 (1997) 3197 36. J. O. Park, M. Verhoff, R. Alkire, Electrochim. Acta 42 (1997) 3281 37. R. Rapp, W. Hoffmann, W. Süß, H. J. Ache, H. Gölz, Electrochim. Acta 42 (1997) 3391 38. E. Klusmann, J. W. Schultze, Electrochim. Acta 42 (1997) 3123 39. M. Küpper, J. W. Schultze, Electrochim. Acta 42 (1997) 3023, 3085 40. H. Emons, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 369–383 (this volume) 41. (a) M. J. Schöning, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 384–408 (this volume); (b) M. Schöning et al., Electrochim. Acta 47 (2001), 243, 251, 259, 293 42. (a) W. Schuhmann, K. Habermüller, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 409–428 (this volume); (b) W. Schuhmann et al., Electrochim. Acta 47 (2001) 265 43. L. A. Nagahara, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 429–464 (this volume) 44. M. Nishizawa, I. Uchida, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 465–482 (this volume) 45. T. Matsunaga, T.-K. Lim, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 485–511 (this volume) 46. P. Fromherz, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 541–559 (this volume); P. Fromherz, Ber. Bunsenges. Phys. Chem. 100 (1996) 1093 47. J. W. Deitmer, W.-R. Schlue, Pfluegers. Arch. Europ. J. Physiol. 397 (1983) 195 48. P. W. Dierkes, S. Neumann, A. Müller, D. Günzel, W.-R. Schlue, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 526–540 (this volume) 49. E. Neumann, K. Tönsing, in: J. W. Schultze, T. Osaka, M. Datta (eds), Electrochemical Microsystem Technologies, Taylor & Francis, London and New York, 2002, pp. 512–525 (this volume) 50. T. Matsue, N. Matsumoto, I. Uchida, Electrochim. Acta 42 (1997) 3251 51. F. Silva, M. J. Sousa, S. M. Pereira, Electrochim. Acta 42 (1997) 3095 52. J. W. Schultze, T. Morgenstern, D. Schattka, S. Winkels, Electrochim. Acta 44 (1999) 1847 53. J. W. Schultze, K. Bade, A. Michaelis, Ber. Bunsenges. Phys. Chem. 95 (1991) 1349 54. R. Kakerow, Y. Manoli, W. Mokwa, M. Rospert, H. Meyer, H. Drewer, J. Krause, K. Camman, Sensors Actuators A 43 (1994) 296

Principles of EMST 31 55. H. Meyer, B. Naendorf, M. Wittkampf, B. Gründig, K. Cammann, R. Kakerow, Y. Manoli, W. Mokwa, M. Rospert, in: A. van den Berg, P. Bergveld (eds), Micro Total Analysis Systems, Kluwer, 1995, p. 245 56. D. Schattka, S. Winkels, J. W. Schultze, Metalloberfläche, 1997, p. 51 57. M. Datta, in: M. Datta, K. Sheppard, D. Snyder (eds), Electrochemical Microfabrication, The Electrochemical Society Proceedings, Pennington, New York, 1992, p. 61; M. Datta, J. Electrochem. Soc. 142 (1995) 3801; M. Datta, in: N. Masuko, T. Osaka, Y. Ito (eds), Electrochemical Technology: Innovation and New Developments, Kodansha and Gordon and Breach, Tokyo and Amsterdam, 1996, p. 137 58. M. Datta, D. Harris, Electrochim. Acta 42 (1997) 3007 59. D. Landolt, Electrochim. Acta 32 (1987) 1 60. G. Buß, M. J. Schöning, H. Lüth, J. W. Schultze, Electrochim. Acta 44 (1999) 3899 61. (a) H. Sugimura, T. Uchida, N. Kitamura, H. Masuhara, Appl. Phys. Lett. 63 (1993) 1288; H. Sugimura, K. Okiguchi, N. Nakagiri, M. Miyashita, J. Vac. Sci. Technol. B 14 (1996) 4140; (b) H. Bloeß, J. W. Schultze, G. Staikov, Electrochim. Acta 47 (2001) 335 62. A. Vogel, J. W. Schultze, Electrochim. Acta 44 (1999) 3751 63. M. M. Lohrengel, A. Moehring, M. Pilaski, Electrochim. Acta 47 (2001) 137 64. A. Bard, Lecture given at the 1st EMT in Grevenbroich (1996) 65. G. Buß, Thesis, Heinrich-Heine-Universität Düsseldorf, 2000 66. A. Fujishima, K. Hashimoto, T. Watanabe, TiO2 Photocatalysis, BKC, Inc., Tokyo 67. K. Holthoff, O. Witte, Electrochim. Acta 42 (1997) 3241 68. M. Seo, M. Chiba, Electrochim. Acta 47 (2001) 319 69. C. Kobusch, J. W. Schultze, in: W. J. Lorenz, W. Plieth (eds), Iupac-Monography, Series ‘Chemistry for 21st century’, ‘Electrochemical Nanotechnology – In-situ Local Probe Techniques at Electrochemical Interfaces’ (1998) 225 70. A. Vogel, Thesis, Heinrich-Heine-Universität Düsseldorf, Shaker Verlag, Aachen 2000 71. J. V. McPherson, N. Simjee, P. R. Unwin, Electrochim. Acta 47 (2001) 29 72. De Vogelaere, V. Sommer, H. Springborn, U. Michelsen-Mohammadein, Electrochim. Acta 47 (2001) 109 73. T. Haisch, E. J. Mittemeijer, J. W. Schultze, Electrochim. Acta 47 (2001) 235 74. F. Beck, H. Guder, J. Appl. Electrochem. 15 (1985) 825 75. K. Arlt, Electrochim. Acta 39 (1994) 1189 76. M. Küpper, J. W. Schultze, J. Electroanal. Chem. 427 (1997) 129 77. (a) U. König, B. Davepon, Electrochim. Acta 47 (2001) 149; (b) J. W. Schultze, M. M. Lohrengel, M. Pilaski, U. König, Faraday Discussions 121, in print 78. O. Karstens, J. W. Schultze, W. Bacher, R. Ruprecht, Metalloberfläche 48 (1994) 148 79. K. Bade, O. Karstens, A. Michaelis, J. W. Schultze, Faraday Discuss. 94 (1992) 45 80. R. J. von Gutfeld, E. E. Tynan, R. L. Melcher, S. E. Blum, Appl. Phys. Lett. 35 (1979) 651; M. Datta, L. T. Romankiw, D. R. Vigliotti, R. J. von Gutfeld, J. Electrochem. Soc. 136 (1989) 2251 81. T. Osaka, Electrochim. Acta 45 (2000) 3311 82. A. Manz, E. Verpoorte, C. S. Effenhauser, N. Burggraf, D. E. Rymond, H. M. Widmer, Fresenius J. Anal. Chem. 348 (1994) 567 83. A. Vogel, J. W. Schultze, M. Küpper, in: Proc. 2nd Int. Conf. on Microreaction Technology, 1998, p. 323 84. O. Voigt, B. Davepon, G. Staikov, J. W. Schultze, Electrochim. Acta 44 (1999) 3731

2

Application for homogeneous electrochemistry Measurements of fast reaction kinetics J. Heinze and K. Borgwarth

2.1

Introduction

Research in the field of microelectrodes intensified in the beginning of the 1980s. At that time, it became clear that extremely small electrodes with diameters below 25 m possess quite a large number of unconventional properties which allowed new applications in chemistry, life sciences and materials research. Electrochemists are interested in ultramicroelectrodes mainly because their application in accordance with the general measuring principles used in voltammetry and chronoamperometry opens totally new possibilities of studying electrode reactions. The results of dynamic measurements in solvents with low electrolyte concentrations or in poorly conducting media,1 or in the solid state,2 or even in the gas phase3 have been particularly spectacular. On top of this line, using microelectrodes expands the timescale of measurements by several orders of magnitude, which makes it much easier to study rapid homogeneous or heterogeneous reactions than with, for example, rotating electrodes. Over and above this, the combination of exceptionally high current densities and low measuring currents promises new application in the fields of analysis, sensors and scanning electrochemical microscopy (SECM).4–8 At the same time, the smaller dimensions of the electrodes guarantee that the experiment will not alter or destroy the object measured, a very important aspect in, for instance, biophysical examinations. This chapter reviews in detail the principles and applications of ultramicroelectrodes and their application to studies of homogeneous chemical reactions induced by electron transfer. The first section summarizes the basic principles. It is followed by sections each dedicated to experimental techniques: steady-state voltammetry, fast-scan voltammetry (FSV), chronoamperometry and generation/ collection experiments. The final section deals with the experiments defining the state-of-the-art in this field and the outlook for some future activities.

2.2 2.2.1

Basic principles Mass transport

In principle, experiments using ultramicroelectrodes are similar to those using conventional electrodes. A stationary electrode immersed in an unstirred electrolyte

Homogeneous reaction kinetics 33 solution is given either a constant potential or one that changes linearly over time. Provided there is an electroactive substance, that is one that can undergo oxidation or reduction, in the electrolyte solution, a heterogeneous charge transfer will occur at the metal/liquid interface, during which, in the case of reduction, for example, electrons will be transferred from the electrode to the electroactive substance.9 At the same time, the concentrations at the electrode surface start to change, which sets off diffusive mass transport to and from the electrode. Usually, contributions of migration and convection can be neglected when using ultramicroelectrodes. The phenomena at ultramicroelectrodes are caused by time-dependent changes in mass transport, in which one-dimensional diffusion fields transform into spatial fields of convergent mass transport. In the case of planar electrodes, besides the usual axial diffusion, an additional, radial diffusive component becomes effective parallel to the electrode surface causing a convergent diffusion. In the case of curved electrode surfaces, spherical or cylindrical diffusion fields form over time rates that depend on the electrode size. In mathematical terms, the diffusion is described by Fick’s second law, which varies for spatial fields (Eqns (1)–(5)) according to the electrode geometry:10 ⭸2c ⭸c D 2 ⭸t ⭸x

冤冥 ⭸c ⭸c ⭸c D 冤 2r 冥 ⭸t ⭸r ⭸r ⭸c ⭸ c ⭸c ⭸c D 冤 1r 冥 ⭸t ⭸r ⭸z ⭸r ⭸c ⭸c ⭸c D冤冥 ⭸t ⭸x ⭸ z ⭸c ⭸2c 1 ⭸c D ⭸t ⭸r2 r ⭸r

(planar diffusion)

(1)

(cylindrical electrode)

(2)

(spherical electrode)

(3)

(disk)

(4)

(band)

(5)

2

2

2

2

2

2

2

2

2

2

The introduction of further boundary conditions determines whether the diffusion will be finite (thin-layer cell) or semi-infinite and whether the electrode reaction will proceed galvanostatically (constant current), potentiostatically (constant potential) or potentiodynamically (variable potential). When ultramicroelectrodes are used, measurements are generally potentiostatic (chronoamperometry) or potentiodynamic (cyclic voltammetry). Applying a potential step at t 0 to a spherical electrode of radius r0 and assuming the following boundary conditions for t 0 and r r0: c c*, for t 0 and r : c c*; and for r r0: c 0) the solution of Eqn (3) is

冢

rr 冣冤兹4Dt 冥

r C(r, t)c* 1r erfc 0

0

(6)

34

J. Heinze and K. Borgwarth

where t is the time, r is the distance from the center of the sphere, D and c* are the diffusion coefficient and the initial bulk concentration of the electroactive species. When the dimension of the electrode is large, or at short times, the expression simplifies to C(r, t)c*erfc

rr 冤兹4Dt 冥 0

(7)

Complementary, at ultramicroelectrodes or electrodes after a long time the concentration is given by:

冢冣

r C(r, t)c* 1r 0

(8)

The planar disk is the most popular electrode in experiments (Figure 2.1). It is constructed relatively easy by encasing a metal wire or a carbon fiber in glass or embedding it in plastic; the upper surface of the insulated wire serves as the active electrode surface.11–15 To some extent, use is being made of band16–19 and ring electrodes,20,21 whose surface can be enlarged by altering the length of the band or the circumference of the ring without losing the specific properties of ultramicroelectrodes. The same objective can be achieved with electrode arrays22–31 that is, several electrodes arranged, for example, in the form of an interdigitated structure. Recently, nanoelectrode ensembles (NEE) became popular which enable to monitor very fast processes, but often lack of a defined geometry.32–34 However, with arrays the diffusion conditions change depending on the arrangement of the electrodes. This means that, in the case of very long experiments, the diffusion field eventually returns to a planar form.35–37 From a theoretical point of view, spherical (Hg drop)37 and cylindrical (wire) electrodes18,38,39 are particularly interesting as they allow one-dimensional solutions to the diffusion problem.10

Disk

Ring

Sphere

Array

Band

Wire

Interdigitated structure

Figure 2.1 Typical configurations of ultramicroelectrodes.

Homogeneous reaction kinetics 35 In the simplest case of a chronoamperometric (potential step) experiment, a working electrode is given a potential sufficient to completely reduce or oxidize at the electrode surface an electroactive substance of concentration c* in solution. Experiment and theory show that under these conditions a diffusion-controlled, faradaic current i flows over the electrode. Its value is directly proportional to the concentration gradient at the electrode surface (Eqn (9)):

冤冥

⭸c i j(0, t) D ⭸x nFA

(mol s1 m2)

(9)

x0

where n is the number of electrons transferred per species, F is the faradaic constant, D the diffusion coefficient, and A the area of the electrode. This gradient causes the formation of a diffusion layer in front of the electrode, which gradually spreads into the solution (Figure 2.2). In the case of purely planar diffusion, (a)

1.00

0.001 s

0.01 s

0.1 s

c /c*

0.75 0.50

1s

0.25 0.00 0

(b)

1.00

8

16 x (µm)

0.001 s

0.01 s

24

32

0.1 s 1s

c /c*

0.75 0.50 0.25 0.00 0

8

16 x (µm)

24

32

Figure 2.2 Concentration profiles of diffusion layers (D 1 109 m2 s1) in a chronoampero-metric experiment for different times t after application of a potential step, concentration of the electroactive species at electrode surface 0; (a) semi-infinite planar diffusion, (b) spherical diffusion for a spherical microelectrode with r0 0.5 m.

36

J. Heinze and K. Borgwarth

the time-dependent current obeys the Cottrell equation: it1/2 1 nFAD1/2c* ␲1/2

(10)

This predicts that the current in the chronoamperometric experiment diminishes inversely proportional to t1/2. As the electrode radius decreases, the semi-infinite planar diffusion gradually transforms into a semi-infinite hemispherical diffusion (Figure 2.3). An increase in time produces the same phenomenon. Due to the time-dependent change in the

r0 = 3 ×10–3 m

r0 = 3 ×10–4 m 100

x (µm)

x (µm)

100

50

0.5

50

1.0

0.5

r0 = 3 ×10–5 m

100

x (µm)

1.0 r /r0

r /r0

50

0.5

1.0

1.5

2.0 r /r0

2.5

3.0

3.5

Figure 2.3 Normalized concentration contours c/c* for electrodes with different radii (r0 3 103, 3 104, 3 105 m, D 3 109 m2 s1) 1 s after start of a chronoamperometric experiment; concentration of the electroactive species at the electrode is zero because of a high potential pulse. Curves of the same concentration separated by 0.1 of a unit.

Homogeneous reaction kinetics 37 diffusion shape, the solid angle occupied by the diffusion layer in front of the electrode expands, and, hence, grows considerably relative to the electrode surface. As a consequence, many more electroactive particles reach the electrode than in the case of a pure planar diffusion. On the other hand, the growing volume within the prescribed solid angle means that the particle flux in and out of the volume eventually becomes stationary (Figure 2.2(b)) and the diffusion layer stops growing. The flux rate in the stationary case is given by the mass transport coefficient m: m

␲ 1 (m s1) Dr 2 0

(11)

In some cases the mass transport is accelerated by using a microjet directed towards the UME.40 Its value in a (hemi)spherical field is indirectly proportional to the electrode radius r0 (Eqn (11)). Consequently, in the case of extremely small electrodes the diffusion rate can be extraordinarily high. Hence, the theoretical calculation of the current–time curve produces a modified Cottrell equation (Eqn (12)). For the disk electrode this reads:41

冢冢冣冣

it1/2 Dt 1 1b 2 r0 nFAD1/2c* ␲1/2

1/2

(12)

where ␲1/2 b 4␲⫺1/2. The value of the pre-factor b increases in the transition from the planar to the hemispherical diffusion field. Approximate formula for this intermediate case may be found in the literature.37,42,43 By analogy, the equation for the spherical electrode is:

冢

冢冣冣

it1/2 Dt 1 1␲1/2 2 r0 nFAD1/2c* ␲1/2

1/2

(13)

As long as planar diffusion dominates and the measuring time is short, the second term on the right-hand side of Eqns (12) and (13) can be neglected, which results in the limiting case of Eqn (10). By contrast, if the measuring time is very long, the first term on the right-hand side becomes negligible, and the result is a stationary current. The smaller the electrode surface is, the shorter the transition time to the stationary state becomes. Oldham’s comprehensive theoretical studies44–46 have shown that one can set up an identity for the stationary limiting currents for the spherical, hemispherical and disk electrodes by introducing a surface diameter d (Eqn (14)). iLt 2nFc* Dd

(14)

for the sphere, d 2␲r0, for the hemisphere d ␲r0, and for the disk electrode d 2r0. The attainable stationary limiting currents for the different electrode forms are listed in Table 2.1. The time lapse before a stationary state is reached

38

J. Heinze and K. Borgwarth

Table 2.1 Limiting currents for ultramicroelectrodes Sphere iGr 4r0nFDc* Hemisphere iGr 2r0nFDc* Disk iGr 4r0nFDc* is identical with the sphere when (/2)r0 (hemisphere) Cylinder (approximate) 1 i 2nFADc* ln[4(Dt)/r20] where A cylinder surface area (2r0l) or hemicylinder surface area (r0l), l length of the cylinder Band electrode (approximate) 1 i 2nFDc*l ln[4(Dt)/r20] l length of the band Ring ␲2(ab) limt→ i(t ) ln [16(ba)/(ba)] nFDc* where ab << a and b inner radius of the ring, a outer radius of the ring

depends both on the size of the electrode and on the diffusion coefficient of the electroactive system. Calculations of Oldham et al.46 show that for an average diffusion coefficient of 109 m2 s1, a 10 m disk electrode takes approx. 1.3 s and a 1 m electrode 0.01 s to reach a stationary state, with a deviation of 5%. As it is obvious that several parameters determine the specific properties of such electrodes it is very difficult to define the term ‘ultramicroelectrode’ in terms of a precise limit for its characteristic dimension. In the literature it has become common practice to use the term ‘UME’ for disk electrodes with a diameter of 20 m or less. The term ‘nanode’ is used for electrodes with a diameter of less than 1 m.47,48 As one might expect, the influence of electrode shape is greatest in the case of spherical diffusion fields. The comparison with planar diffusion Figure 2.3 shows that already after 1 s, despite a small absolute current, the current density at a spherical microelectrode with a diameter of 1 m is nearly 100 times greater than that at a normal electrode (Eqn (15)):

冢冣

ispha¨r Dt 1 ␲ 2 iplan r0

1/2

(15)

The calculations for cylindrical, band and ring electrodes, prove that these attain only a quasi-stationary state, as the equations for their currents contain

Homogeneous reaction kinetics 39 time-dependent terms. On the other hand, the current response can be considerably increased via band length or ring circumference without changing the temporal response diffusion field. This opens interesting perspectives for analytical applications. The most important technique to study fast homogeneous reactions is voltammetry. In voltammetry, the potential of the working electrode changes linearly over time. Starting from an initial potential Ei, a linear potential sweep (potential ramp) is applied to the electrode. After reaching a switching potential E␭, the sweep is reversed and the potential returns linearly to its initial value,9 which is why one speaks of cyclic voltammetry. The experimental timescale is determined by the potential scan rate (also known as the sweep rate) v |E/t|. Ultramicroelectrodes in cyclic voltammetry produce, in principle, the same phenomena as in chronoamperometry. The transition to the stationary state is achieved by reducing the scan rate – and the smaller the characteristic radius of the electrode is, the faster the transition. Initially, the cyclic voltammograms at ‘high’ v values are peak-shaped; as v declines they become sigmoid polarograms (Figure 2.4). As such, they are comparable with the steady-state voltammograms at rotating disk electrodes. This resemblance is due to the formation of stationary (so-called Nernst) diffusion layers, which, in the case of rotating electrodes, are caused by rapid convectional mass transport, and in the case of microelectrodes are a result of the particular shape of the diffusion field and the high diffusion mass transport rate.

v =10 mV s–1

v =100 mV s–1

2 nA

2 nA

–0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 E (vs Ag/AgCl)/(V)

v =1 V s–1

–0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 E (vs Ag/AgCl)/(V)

v = 10 V s–1

10 nA 2 nA

–0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 E (vs Ag/AgCl)/(V)

–0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 E (vs Ag/AgCl)/(V)

Figure 2.4 Experimental cyclic voltammograms of the oxidation of ferrocene (c 3.2 mM) in CH2Cl2/0.1 M TBAPF6 at different scan rates. Measurements taken with a Pt microdisk electrode (r0 6 m).

40

J. Heinze and K. Borgwarth

In the stationary state, the currents are time-independent and the scan no longer affects the shape and size of the voltammograms. The conditions for the formation of a steady-state situation differ according to the geometry of the electrode used. Whereas the diffusion-limited current or the current density corresponds to the values obtained for the stationary state in chronoamperometry (Table 2.1), theoretical modeling of the entire current–voltage curve has to be done numerically.49–54 The only exception is the case of the reversible redox process during spherical diffusion to a spherical electrode. For this, Reinmuth55 has developed a correction formula that can be used to calculate the dimensionless current 1/2 ␹s(at) (which corresponds to a normalized current – S stands for spherical) from the known current ␲1/2 ␹p(at) (p stands for planar) for planar diffusion (Eqn (16a) where ␴ is the sphericity (16b) and the spherical correction factor (16c)). Where a is the normalized scan rate in cyclic voltammetry defined as a vnF/RT and ␩0 E 0nF/RT is the normalized potential. 兹␲␹s(at) 兹␲␹ p(at)␴(at)

冪arD

(16b)

1exp(at) 1 exp(␩␩0 )

(16c)

␴

(16a)

2

The sphericity ␴ characterizes the extend of spherical diffusion. For ␴ 0, the limiting case of normal planar diffusion (r → ␣) characterized by the typical transient voltammogram is observed, whereas from ␴ 1 the stationarity ensues. The transition to the stationary state takes place at somewhat higher ␴ values for other electrode geometries. Larger ␴ values, usually a consequence of smaller electrode radii, imply higher mass transfer coefficients m and therefore higher current densities. A comparison with rotating-disc electrodes (RDE) reveals that the transport coefficient of a spherical electrode with a diameter of 10 m corresponds to a speed of 10 000 rpm. Whereas this is the maximum speed for rotating electrodes in a laminar flow, the dimension of ultramicroelectrodes can be considerably smaller. 2.2.2

Determination of reaction kinetics

One property that distinguishes ultramicroelectrodes from stationary electrodes of conventional size is the increase in the transport rate of the electroactive particles as the electrode radius decreases. The high transport rates mean that, with very small electrodes this process can take place as fast or even faster than heterogeneous charge transfer, that is, instead of being diffusion controlled, the process becomes kinetically controlled. Unlike the diffusion-controlled case, which is in thermodynamic (Nernstian) equilibrium at the electrode, the corresponding stationary current–voltage curves shift and change their shapes (Figure 2.5). Thus,

Homogeneous reaction kinetics 41 12.0

I

9.0

6.0

3.0

0.0 0.2

0.1

0.0

–0.1

–0.2

–0.3

–0.4

–0.5

E – E 0 (V)

Figure 2.5 Steady-state cyclic voltammograms with an ultramicrodisk electrode (␴ 11.1, D 1 1010 m2 s1, r 0.5 m, (hemi)spherical diffusion field): k0s : (——) 1, (-------) 0.011, 0.0035, 0.0011, 0,0035, 0.00011, 0.000035, 0.000011 m s1. For k0s 1 m s1 and a mass transport coefficient of m 0.002 m s1, the electrode process is diffusion controlled (Nernst case). For k0s 0.002 m s1, the reaction is kinetically controlled, and the steady-state voltammograms shift with simultaneous minor changes in shape along the potential axis. With the help of theory,56–58 k0s can be determined from the curves’ potential shift.

ultramicroelectrodes are a very elegant instrument for determining kinetic parameters of rapid electron transfer reactions. As theoretical considerations show,56–58 rapid transfer processes with heterogeneous rate constants of the order of 102 m s1 can be analyzed by disk electrodes with diameters between 5 and 0.05 m. For slower processes, electrodes with significantly larger diameters are adequate. By applying a suitable set of ultramicroelectrodes of different sizes to steady-state voltammograms, one can determine the kinetic parameters of the heterogeneous charge transfer (␣, k0s ) as well as the thermodynamic standard redox potential.56–60 The ability to change the time window – expressed by the ␴-value (Eqn (16b)) or the mass transport coefficient m (Eqn (11)) – of the experiment by using ultramicroelectrodes of different sizes, is exploited for different purposes. On the one hand, very small electrodes enable the accurate determination of standard redox potentials. As a result of the high transport rates, the homogeneous chemical processes associated with the heterogeneous charge transfer in the experimental timescale take place to a minor extent (Figure 2.6). Because the diffusion rates rise with decreasing electrode radius, while the rates of homogeneous chemical

J. Heinze and K. Borgwarth Dimer, = 5.0, k f × c* = 20.0 Concentration

(a) 1.0

Anion, = 5.0, k f × c * = 20.0 Concentration

42

0.5 0.0

1.0 0.5 0.0

0.0

l

ia Ax

2.0

2.0 4.0

2.0 Radial

4.0

Concentration

1.0 0.5 0.0 12.5 Axial 25.0

0.0 12.5 Radial

2.0 Radial

Anion, = 0.1, k f × c * = 20.0 Concentration

Dimer, = 0.1, k f × c* = 20.0

(b)

0.0

Axial

1.0 0.5 0.0 12.5 Axial 25.0

0.0 12.5 Radial

Figure 2.6 Two-dimensional concentration profiles for an electron transfer (R → R), to which an irreversible dimerization reaction is coupled (R → R22, EC2irr, kf c/a 20.0). (a) ␴ 5.0 (microdisk electrode); (b) ␴ 0.1 (macrodisk electrode). The axial and radial coordinates are normalized to the radius of the disk electrode. The formation of the dimer close to the macroelectrode is very clear; in the case of the microelectrode, the concentration profile of the first anion formed predominates.

processes do not change, their effect on the shape and position of the voltammetric current–voltage curves decreases continuously. Provided that the heterogeneous charge transfer is very fast, appropriately small electrodes enable one to directly determine the formal redox potential of an electrode reaction by measuring the half-wave potential E1/2.59,60 On the other hand, the kinetic parameters of these homogeneous reactions (kf, kb) can be determined when using electrodes of different sizes. As has been shown in Figure 2.6, the stationary current under diffusion control is influenced by chemical follow-up reactions and further electron transfer steps. Regarding a three-step reaction of the electrochemical, chemical, electrochemical (ECE) type, which means an electron transfer step followed by a chemical reaction and a second electron transfer, large electrodes with a diameter less than 100 m reveal a diffusion-limited current indicating a transfer of n 2 electrons. Using smaller electrodes, the rate of the chemical reaction becomes small as compared

Homogeneous reaction kinetics 43 to the rate of mass transport. Consequently, part of the species generated in the first step do not undergo the following reactions and the number of electrons transferred per species n as indicated by the current tends to unity. This dependence of the current on the electrode radius can be exploited to determine the kinetic parameter of the chemical step in the mentioned ECE type reactions59–62 as well as in case of the catalytic EC mechanism61–70 and CE mechanism symbolizing a chemical reaction proceeding the electron transfer. Contrary, in case of an EC mechanism with a chemical follow-up reaction the diffusion limited current does not depend on the rate of the chemical reaction. Information about kinetics is only accessible from the shift of steady-state voltammograms depending on the electrode size.62 The approach of using steady-state voltammetry for the analysis of homogeneous kinetics was strongly pushed forward by the pioneering work of Fleischmann. As an example the study of an EC mechanism will be given here considering a typical reaction O ne 7 R

(17a)

k OP RZ →

(17b)

It is assumed that the bulk concentration of Z is much larger than that of species O, so that the concentration of species Z at the electrode surface is almost constant. Under steady state conditions, the generated R diffuses over a distance of (Dt1/2)1/2, where t1/2 is the reaction half time before reacting with Z. Only if this distance is small compared to the electrode dimension, its current will be influenced by the chemical reaction. Assuming a spherical electrode with radius rs and a pseudo-first order reaction, the following current is measured: 1(nFDc*/rs)[1r0兹k /D]

(18)

k is the pseudo-first order rate. More compact, the current can be expressed in terms of the apparent number of electrons transferred, napp. nappn[1r0兹k /D]

(19)

Similar equations were obtained for ECE and DISP1 mechanisms.61 Careful discussions of theoretical concepts have been presented by Zhuang et al.71–73 The application of steady-state voltammetry to the analysis of the simple ECirr mechanism is more demanding because the diffusion-controlled current does not change as a function of the electrode radius. However, the kinetic parameters can be obtained from the slope and the intercept of the steady-state current voltage curves according to (Eqn (20)).71 EE0

I I ln 冢冪Dk 冣 RT nF 冢 I 冣

RT ln 1rs nF

d

(20)

44

J. Heinze and K. Borgwarth

where rs is the radius of a spherical microelectrode, k the first-order rate constant and I the steady-state current. For second-order reactions more complex expressions have been derived.73 An alternative approach to determine reaction kinetics is to apply very high scan rates by bringing the timescale of the experiment in the range of the reaction rate to be studied. Under these conditions of the FSV the diffusion field around ultramicroelectrodes becomes planar again and the voltammograms resemble those well known from large electrodes. Practically, up to 1 MV s1 can be applied and first-order rate constants of 6 106 s1 respectively second-order rate constants of 6 109 M1s1 can be measured. At higher scan rates the propagation of the diffusion layer and of the double layer become comparable and cannot be separated, as theory requires.74 Moreover, experimental obstacles such as high capacitive currents and iR drops become increasingly important. In principle, the fast-scan technique provides the same kind of information as the conventional voltammetry with one important distinction: the timescale for experiments extends below 1 s (e.g. at 1 MV s1 ␶ 25 ns where ␶ RT/vF). Thus, it has become possible to analyze complex mechanisms including chemical steps.74–76 Quantitative evaluation on diffusion-controlled steps of second-order reactions are now also reality. Nor is there any difficulty in evaluating simple EC reactions.77,78 Recently, a new potentiostat was developed which allowed the recording of undistorted ohmic drop-free voltammograms up to 2.25 MV s1.79 Under these conditions, first-order follow-up reactions with rate constants larger than 108 s1 may be accessible. Similarly, chronoamperometry can be performed by applying potential jumps on the timescale of several micro- or milliseconds. Recently, generation/collection experiments using two working electrodes in combination with a bipotentiostat became popular. Facing two oppositely polarized UMEs and bringing them close together increases the mass transfer coefficient even more, and it becomes inversely proportional to the distance d from the sample.4,80 mD d

for d << r0

(21)

In practice, it is much easier to bring a flat-ended electrode within very small distances than to make electrodes with a defined geometry and a similar value as tip radius. Such setups are realized by the SECM,5–8 dual-disk electrodes,81,82 and ring-disk electrodes.83 Thus, the SECM strongly facilitates the determination of high rate constants. Bringing the tip down to the present technical limit of about 10 nm distance and assuming a typical diffusion coefficient D 105 cm2 s1, mass transfer coefficients of 10 cm s1 can be achieved. As the separation between both electrodes is changed – either by switching between different sets of electrodes or, much easier, by altering the distance between two electrodes by a piezo driver or stepper motor in a SECM configuration – so does the probability for the tip-generated species to undergo chemical reactions while diffusing through the gap between the electrodes. To determine rate constants of homogeneous reactions two modes of operation can be applied. On the one hand, using the generation/collection mode, the first electrode converts the given species, for

Homogeneous reaction kinetics 45 example, by a reduction into such, which may undergo chemical follow-up reactions, while the second electrode monitors the fraction of unreacted species by their reoxidation current. The rate of chemical steps in EC, ECE and EC mechanisms can be extracted by measuring the collection efficiency, meaning the absolute ratio between collection and generation current, as a function of electrode separation. On the other hand, in the feedback mode, species shuttle between both electrodes transporting charge across the gap.84 If either the reduced or the oxidized form undergoes chemical reactions the concentration of the mediator changes in the gap and the charge transported per time becomes reduced. From current–distance curves the order and rate of the chemical reaction step can be extracted. Using one ultramicroelectrode as a tip and a large sample electrode, the latter does not need to be biased to enable the opposite reaction as occurring at the tip. It may take up charge on a place far from the tip, conduct it into the region of interest and provide it for electron transfer reactions. Alternatively, two ultramicroelectrodes oppositely biased by a bipotentiostat can be used. Note, that due to the redox cycling at small separations highly amplified currents can be detected at the ultramicroelectrode. Corresponding theoretical curves have been calculated numerically by Mirkin and Bard for different EC mechanisms of first85–87 and second order.88–90 Unwin and Bard published simulations for an ECE/DISP mechanism.91 2.2.3

Capacitive current and the ohmic drop

As with all dynamic electrochemical experiments, chronoamperometry and cyclic voltammetry have to take account of the faradaic current as well as the capacitive current iC that flows along the electrode when the electrical double layer is charged. The simplest of simulating the capacitive current is by applying an electrical equivalent circuit, with resistance that corresponds to the resistance of the solution RU and a capacity in series that corresponds to the double layer capacity Cd. For a single potential step E, iC is:

冢

t iC E exp RU RUCd

冣

(22)

As Eqn (22) shows, the capacitive charging current falls exponentially over time. The smaller the corresponding time constant for charging the double layer is – contained in the term RC – the faster the process. As RUCd ~ r0

(23)

the drop-off time for the charging current for ultramicroelectrodes becomes extremely short, they hold great potential for measurements in very small time domains. Because of continual potential sweep (Ei initial potential), the capacitive current in cyclic voltammetry has two components, one transient, the other steady state (Eqn (24)): iC

冤冢RE vC 冣 exp 冢R tC 冣冥vC i

d

U

d

U

d

(24)

46

J. Heinze and K. Borgwarth

The latter predominates at medium and low scan rates, so that, as a rule: iC vCd

(25)

The capacitive current component iC disturbs studies on electrochemical induced electron transfer reactions, the so-called faradaic processes, because voltammetric measurements always give the sum of both currents. The capacitive current increases linearly with the scan rate v, whereas the faradaic current if changes at v1/2. Moreover, in the case of low concentrations of the active redox system (105 M), if and iC cannot be adequately separated. Although there is some capacitive effect when ultramicroelectrodes are used in the stationary state, the current ratio if /iC improves dramatically, as the faradaic current density increases at r01, whereas the capacitive current density is independent of this. The following relationship holds either exactly or approximately for most geometric shapes: if 1 ⬇ iC r0Cdv

(26)

The smaller the characteristic dimension of the electrode and the lower the scan rate selected, the better the current ratio if /iC in the stationary state of a voltammetric measurement. As a consequence of the ohmic resistance in an electrolyte, in every electrochemical experiment involving current flow, the effective potential will not be the applied (time-dependent) potential because of an ohmic potential drop (Eqn (27)). The lower the conductivity of the electrolyte system and the larger the total current, the larger this effect will be. Eeff (t) Ea(t) iRU

(27)

This influence is particularly conspicuous when, in addition to the faradaic current, there is also a strong capacitive component, as observed, for instance, at high scan rates in cyclic voltammetry (Figure 2.7). As, moreover, the combination of R and Cd – its properties are given by the time constant RCd – acts as a low pass filter on the current response, sudden changes in current are more or less strongly distorted.59,92 Because of these various effects, current and potential are mutually dependent in voltammetric analysis, and instead of the set potential scan rate v, one is working with an effective rate v (Eqn (28), Figure 2.8). Eeff (t) Estart v t RU(Cdv if)

(28)

Hence, even in electrolyte solutions with high conductivity, the maximum scan rate for ordinary electrodes with a radius greater than 1 mm is approx. 400 V s1. 9,93 In solutions with low conductivity, this value falls dramatically. For example, in benzene the current–voltage curves measured at scan rates of less than 0.1 V s1 are so distorted that they cannot be interpreted, even though the time constant is negligible and the only effect is the ohmic potential drop.

Homogeneous reaction kinetics 47 1.5

1.0

I

0.5

0.0

–0.5

–1.0 –1.5 0.3 0.2 0.1 0.0 –0.1 –0.2 –0.3 –0.4 E – E 0(V)

Figure 2.7 Simulated cyclic voltammogram for a reversible charge transfer with planar diffusion (——). Addition of a normalized capacity 1 (-------),92 a normalized uncompensated solution resistance ␳ 10.0 (…….) and both effects together (—.—).

–0.3

(b)

–0.3 –0.2

–0.1

–0.1

0.0

E (V)

–0.2

E (V)

(a)

0.0

0.1

0.1

0.2

0.2

0.3 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 t (s)

0.3 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 t (s)

Figure 2.8 Potential–time curves for the cyclic voltammograms in Figure 2.10; measured potential (——), applied potential (-------). (a) only resistance ␳ 10; (b) resistance ␳ 10 and capacity ␥ 1.92

In the past, one attempted to reduce the iR drop by improving equipment, usually a potentiostat in combination with a three-electrode.9,94 Via the potentiostat, a reference electrode with high impedance provided a correction signal large enough to compensate for most of the loss of potential at the working electrode. The remaining uncompensated solution resistance was further reduced by an

Reaction type

EC1ps

EC2

EC

EC

ECDim

Entry no.

1

2

3

4

5

H

H

C

+

C

C

H

+ N C CH3

H

RCH3.

ArNH2.

.

TPB

2TPA

TPA

2e

k

e

TPB2

TPB 2H

TPA.

k ArNH. H ArNH2. Triphenylamine oxidation

ArNH2

k RCH2. H RCH3. 1-Naphthylamine oxidation

RCH3e

Hexamethylbenzene oxidation

e (C6H6)Cr(CO)3 (C6H6)Cr(CO)3 Nu (C6H6)Cr(CO)3Nu k

k

CH3CN

–e–

Antracene oxidation in ACN

Reaction scheme

Pt disc r 0.3–20 m

Pt disc r 0.5–12.5 m

Pt disc r 0.3–25 m

Pt disc r 10; 25 m

Pt disc r 0.3–25 m

Electrode

Table 2.2 Homogeneous rate constants obtained under steady state conditions

0.1 M TEAClO4/ACN

0.1 M TBAClO4/ DMSO

0.1 M TBAClO4/ACN

0.1 M Bu4NClO4/ CH2Cl2

0.1 M TEAClO4, 4% TFA anhydride/ACN

Medium electrolyte/ solvent

20

25 0.2

20

20

T (ºC)

2103 M1 s1

59–61

160

61

720 s1

4.1103 s1

159

61

190 s1

30 s1

Ref.

Rate constant (k)

EC

EC

EC

EC

CE

6

7

8

9

10

k

e

Ag2

H e

CH3COOH

6

k2

k1

0.5H2

CH3COO H

6

k A BX A B• X Ferrocyanide oxidation in the presence of aminopyrine e Fe(CN)4 Fe(CN)3 6 6 k 3 Fe(CN) AP Fe(CN)4 AP.

Electron transfer from radical anions to sterically hindered alkyl halides or 4,4-dibromodiphenyl e A A.

Ag Mn3 Ag2 Mn2 Ascorbic acid (AA) oxidation at a Prussian Blue (PB) film e PB PB k PB AA PB AA

Ag

Pt disc r 0.3–25 m

Pt band pair gap 2–12 m thickness 0.01 – 6 m

r 2.5–432 m

Pt disc r 0.3–25 m

Pt disc r 0.25 m r 6–9 m

PB film on Pt disc a 2.5; 25 m

Pt disc a 3; 13.7 m

1 M CH3COONa/H2O

1 M KOH/H2O

0.1 M TBABF4/DMF

200 mM KCl (pH 3.7)/H2O

10 mM H2SO4/H2O

20

20

22 25 0.2

20

25

k2 4.11010 M1 l s1

800 M1 s1

3.0103 M1 s1 1.8103 M1 s1

9104–1.7 104 M1 s1 3.88 M1 s1

1.3105 M1 s1

106 s1

63

17,63,116

70,161,162

115

65

50

J. Heinze and K. Borgwarth

electronic feedback,95,96 so that it was possible to obtain scan rates of over 1000 V s1 in high conductivity electrolyte solutions using conventional electrodes. Using ultramicroelectrodes, the iR drop is reduced so strongly that it is possible to interpret the results of measurements taken even in extremely diminished electrolyte solutions. One must distinguish between several limiting cases, depending on the experimental conditions. Theoretical97,98 and experimental99 studies show that in the stationary state with diffusion-controlled limiting current the ohmic potential drop is no longer dependent on the size and geometry of the microelectrode, but solely on the properties of the electrolyte: iR nFc*D␬1

(29)

(␬ specific conductivity of the electrolyte system). The iR drop in the case of a low conductivity electrolyte with a conductivity of ␬ 102 m2 s1 (D 109 m2 s1, c* 0.1 mol m3) is, according to Eqn (29), then 1 mV. The situation becomes more complicated when the level of the conducting electrolyte in the solution is reduced, for then, in addition to the diffusive mass transport, migration effects can also play a role.100,101 As Oldham’s,11,102,103 calculations and experiments show, the quantitative ratio of conducting electrolyte to electroactive substance can fall to 1 : 1 without any significant distortion of the current–voltage curve in the stationary state. One reason for this is that the charge transfer at the solution/metal interface raises the ionic density in front of the electrode, which improves the local conductivity ␹. The following relationship holds for purely planar diffusion, as exists in the case of high scan rates: iR ~ r0

(30)

Accordingly, as electrode size decreases, so does the ohmic potential drop. As, according to Eqn (23), the time constant of the cell decreases parallel to this, the smaller the radius of the ultramicroelectrode, the more suitable it is for measurements at high scan rates.

2.3

Kinetics under steady-state conditions

As has been shown above, the diffusion-limited current depends not only on the simple charge transfer, but on whether chemical reactions take place, possibly coupled with further electron transfer steps. This dependence of the current on electrode radius can be exploited to determine the kinetic parameter of the chemical step in the CE, ECE and catalytic EC mechanisms. Moreover, kinetic information about an EC mechanism can be gained from the shifts of the voltammograms using differently sized electrodes. In the following examples illustrating the reaction schemes in studies of organic and inorganic species are presented (Table 2.2). To determine the competition parameter in a reaction scheme at high rates of mass transport Daasbjerg applied both ultramicroelectrodes and rotating disk

Homogeneous reaction kinetics 51 electrodes.104 Aromatic compounds, after reduction to radical anions, could either reduce alkyl halides or undergo coupling reactions with alkyl radicals formed thereby. The reduction leads to the initial compounds and, thus, represents an EC mechanism, the possible subsequent coupling corresponds to an EC reaction. Co(II) (salen) can be reduced to Co(I) (salen) which catalytically reduces alkyl halides. The rate constant of the second-order reaction between Co(I) salen and the alkyl halide can be determined from the positive shift in the potential of the steady-state reduction wave.73 Rate constants between 104 and 105 M1 s1 are obtained.105 The method of steady-state voltammetry has been extended by Compton et al.106 introducing a sonic horn. The high density of energy enables the formation of hydroxyl radicals from water. The presence of these species lead to an anodic shift of the reduction wave of ferrocyanide with increasing power in correspondence with an EC mechanism. The rate constant of the latter was found to increase linearly with the applied ultrasonic power. A typical example of an ECE mechanism is the dimerization of triphenylamine (TPA). In the past, it was studied by several groups.107–109 TPA, after oxidation, couples forming the tetraphenylbenzidine (TPB), which immediately becomes oxidized to its dication, since its redox potential lies below that of TPA. Figure 2.9 shows the change in the number of electrons napp at an 8 m disk electrode as a function of the TPA concentration. Because the dimerization is a second-order reaction, the rate of the coupling step and thus the number of electrons transferred increases with the concentration of TPA. Fitting led to a rate constant of 2.2 103 M1 s1 59–61 in good agreement with values given by Marcoux et al. and Debrodt et al.107,108 At larger concentrations apparently larger coupling constants have been detected;61 it turned out that this effect is due to a polymer deposition on the electrode surface.59,60

1.20

napp

k = 2.2 ×10–3 M–1s–1

0.95 0.0

10.0

5.0

15.0

10 3c (M)

Figure 2.9 Numbers of electrons transferred (napp) for the oxidation of triphenylamine for different concentrations in CH2Cl2/0.1 M TBAF6. Experimental points (o) and simulated curve (——) Diffusion-controlled limiting currents measured at a Pt microdisk electrode with diameter 8 m.59,60

52

J. Heinze and K. Borgwarth

The mechanisms of a whole series of processes used in industrial electrosynthesis have been analyzed.110 The chief advantage of ultramicroelectrodes is absence of iR distortions in the current–voltage curves, even in extremely high concentrations of the reactants, which, of course, enables kinetic and thermodynamic data to be accurately evaluated. For instance, steady-state measurements revealed the experimental conditions under which hydrodimerization of acrylonitrile to adiponitrile (Monsanto process) dominates over the simple reduction of acrylonitrile to propionitrile.111 Other examples include the catalytic mechanism methoxylize furane by indirect oxidation of bromide and methanol112 and the reductive hydrodimerization of formaldehyde to ethylenglycol.113 Lehmann and Evans114 analyzed the two wave reduction of 7,7,8,8-tetracyanoquinodimethane (TCNQ) in acetonitrile while varying the nature and concentration of the supporting electrolyte. The results were in good agreement with Debye–Hückel theory. Except in case of lithium salts significant shifts of the potential were observed, which were attributed to the known one-to-one ion pairing between lithium ions and the TCNQ dianions. Again, the measurements rely on the remarkable ability of microelectrodes to reduce the error due to solution resistance to a negligible extent.114 Finally, reports on microelectrodes with modified surfaces include studies of electrocatalytic effects of simple redox systems115,116 as well as charging processes at conducting polymers or their formation mechanisms.117,118 The reduction of protons originating from acids provides information about the dissociation kinetics involved. Using strong acids, the preceding chemical reaction does not become visible; instead the protons can be reduced to hydrogen under diffusion control. The theoretical simulations of these reactions have reached a mature state and may consider migration effects occurring in the absence of supporting salts.119 Daniele et al.120 worked on the oxidation of hydroxide ions to form oxygen. The steady-state currents observed depended on the actual concentration of the electrolyte, as the diffusion coefficient of the electroactive species varies with both the ionic strength and viscosity of the medium. Equations for the consideration of the ionic strength have been derived. The reduction of 2,4,6-trinitrotoluene (TNT) to amine compounds could be performed in the gaseous phase to maintain a detection level of 7 ppb.121 (Methylcyclopentadienyl) Mn(CO)NO undergoes a one-electron reduction yielding a 19-electron neutral radical that rapidly dissociates CO and dimerizes. In the presence of P-donor nucleophiles (L) the reduction initiates an electron transfer chain (ETC) CO substitution to give a quantitative yield of (MeCp)Mn (CO)(L)NO.122 (Cyclopentadienyl) Fe(CO)3 and (Indenyl) Fe(CO)3 reveal similar effects.123 Steady-state and FSV delivered details of the ETC reaction.

2.4

Kinetics in fast-scan voltammetry

At present, the fast-scan rather than the steady-state technique is preferred for studying reaction mechanisms and measuring kinetic parameters of homogeneous

Homogeneous reaction kinetics 53 and heterogeneous electrochemical processes. Its crucial advantage is that all methods of evaluation used in classic voltammetry, including simulation, can be applied. Performing voltammetry at high scan rates provides a source of mechanistic information on the very short timescale, still following the theory for the established technique. In some cases, the fast-scan technique, has made it possible to obtain direct evidence for the existence of reactive intermediates as well as new, unequivocal findings of reaction mechanisms. For example, in the very rapid homolytic cleavage of halide ions from radical ions of aryl halides, it was not always clear, whether the bond-breaking step occurred concurrently with the charge transfer as an inner sphere reaction124 or whether the cleavage took place after the formation of the radical anion, that is, in a two-step reaction. Measurements of Savéant et al.125,126 of the reduction of 9-bromoanthracene confirm that at scan rates above 100 000 V s1 the formation of the radical anion is reversible. This proves the intermediacy of a highly reactive anion; the rate constant for the cleavage of the bromide ion is kf 5.9 106 s1. For a long time it was disputed whether the cation of the dimerization of radical cations of aromatic amines and heterocyclic compounds first coupled with its neutral species (RS dimerization) or immediately underwent radical cation coupling (RR coupling). Several independent studies support the assumption of RR coupling in most cases.127–129 The electropolymerization of heterocycles such as pyrrole also involve dimerization steps. Again, Savéant et al.130 have demonstrated the existence of highly reactive radical cations of pyrrole and some derivatives within the voltammetric timescale. The dimerization of pyrrole at a concentration of 4 mM becomes partly reversible at 18 000 V s1, which corresponds to an average cation lifetime of 30 s (Figure 2.10). The oxidation of 3,3 -dimethoxy-2,2 -bithiophene131 demonstrates the complexity of the reaction path involving coupling steps in a growing polymer chain. Fast-scan experiments show that the cations formed at the electrode with a rate constant of more than 105 initially couple to form a ␴-“dimer”, which, in turn, after the loss of two protons, also transforms into a dication. At scan rates of 5000 V s1 and a concentration of 1 mM the reaction at the electrode is chemically reversible. Further measurements at high concentrations and slow-scan rates confirm that the dications comproportionate in solution with unchanged starting monomer, thereby autocatalytically initiating new coupling step. Very recent experiments at low temperatures give evidence that the ␴-dimer intermediate is relatively stable, and that a slow first-order proton elimination takes place (Figure 2.11).132 Amatore et al.133 studied the reactivity of different p-halideanilines with respect to reactions induced by their oxidation. Following the authors, this covers a fast deprotonation first, the coupling of radical and the parent radical cation by the formation of new CN-bonds, and, finally, a dehalogenation in the para position. To detect the kinetics of these fast reactions scan rates up to 100 000 V s1 were applied and the coupling reaction was identified to be the limiting step.

(a)

(b)

N H

(d)

(c)

CH3

H3C (CH2)5OH N H H

N H

1.0 1.5 E (vs Ag/AgCl)(V)

1.0 1.5 E (vs Ag/AgCl)(V)

Figure 2.10 Fast-scan voltammograms for the oxidation of pyrrole and substituted derivatives, scan rate: (a) v 18 kV s1, r0 5 m; (b) v 1600 V s1, r0 5 m; (c) v 2800 V s1, r0 8.5 m; (d) v 3600 V s1, r0 8.5 m. All measurements were performed in ACN/Et4NClO4 at room temperature.130

20 V s–1

100 V s–1

5 000 V s–1

–0.1

0.5 1.1 E (vs Ag/AgCl)(V)

Figure 2.11 Experimental cyclic voltammograms for the oxidation of 3,3 -dimethoxy2,2 -bithiophene in acetonitrile/0.1 M TBAPF 6, c 2 103 M, disk electrode r0 100 m or 2.5 m (5000 V s1).132

Homogeneous reaction kinetics 55 Cyclic nitroxide radicals (analogs of tetramethylpiperidinyl-1-oxy free radical TEMPO) formed by oxidation may undergo a second-electron transfer in combination with a protonation to yield a hydroxylamine compound. The sensitivity of electrodes in FSV has been increased by a factor of 10 due to a nafion coating.134 The electroreduction of p-benzoquinone and anthracene in supercritical CO2 for analytical purposes represents quite a challenging task since this medium reveals poor conductivity and solubilty of ions. Nevertheless, using modified microelectrodes this could be realized, and, additionally, the electrophilic addition of CO2 was monitored.135 The reaction of C60 with dimethylamine (DMA) has been studied using FSV. The results demonstrate dynamic equilibria between C60 and DMA through production of “complexes” in the uncharged or monoanionic states.136 Isomerization reactions of ionic species can easily be resolved with the help of FSV. It has been used to study conformational changes and equilibria of 1,1 dimethylbianthrone and 1,2-diiodocyclohexane during reduction.137,138 The reduction of CpFe(CO)2CH3 occurs at very low potentials and causes a set of parallel reactions including emission and uptake of CO. Amatore et al.139 were able to identify the reaction steps and to quantify the kinetics by varying the scan rate. Charge transfer to macrocylic tetrathiaether of copper (I)/(II) leads to rapid conformational changes. Using scan rates up to 60 kV s1 a rate of 5 104 s1 was determined.140 The fourfold negatively charged ferrocyanide tends to form ion pairs with alkali cations whereas ferricyanide does not. This leads to a chemical reaction step of formation and destruction of ion pairs. Pletcher et al.141 have determined the kinetics of the homogeneous and heterogeneous reactions as a function of the electrolyte composition (Table 2.3). The oxidation of dibenzo-(c,e)-1,2-diselenide in acetonitrile may follow an ECE mechanism due to reactions with residual water. Müller et al.142 investigated this phenomenon and were able to draw conclusions on the stability of radical cations depending on structural variations. The same group extensively studied the electrochemical oxidation of 2-phenyl-1,2-benzisothiazol-3(2H)-ones and various related compounds using steady-state and FSV at microelectrodes and rotating-disk electrodes.143,144 Again, residual water causes an ECE type reaction. The oxidation of thioselenanthrene followed a DISP2 mechanism forming the corresponding selenoxide in the presence of water; the life time of the radical cation was found to be 78 10 ms using conventional and FSV.145 Species, which undergo oxidation and reduction reactions in different waves and largely separated potentials (2 V) often show a comproportionation reaction, which is accompanied by the emission of light. This well-known phenomenon of electrochemiluminescence (ECL) became the basis of numerous analytical applications, especially in the determination of ultralow-level concentrations. Here, we will focus on the observation of emitted light as an additional source of information on the kinetics of homogeneous electron transfer reactions. To obtain the ECL of the dissolved species in most cases, the electrode potential has to be varied fast between the largely separated oxidation and reduction waves. Principally, the working electrode potential can be altered by continuous FSV or,

Reaction type

EC

CEC

EC

ECDim

Entry no.

1

2

3

4

2,6-Diphenyl-pyrylium cation reduction DPP e DPP• k 2 DPP. dimer

trans-1,2-Dimethyl-1,2dinitrocyclohexane reduction RL e RL. k RL . R. L

trans-1,2-Diiodocyclohexane reduction and conformational change

9-Hal-anthracene reduction ArHaI e ArHaI. k . ArHaI Ar . HaI

Reaction scheme

Pt disc r 5 m

Hg film on Pt disc r 12.5; 50 m

Pt disc r 25; 50 m

r 52 m

Au disc r 6 m

Electrode

Table 2.3 Homogeneous rate constants obtained by fast-scan voltammetry v (V s1)

7.5104 – 20 2.5105

0

102–104

TEAPF6/DMF

0.1 M TBABF4/ACN

40 to 0 20–103

20

T (ºC)

0.3 M TEAPF6/DMF

0.6M TEA/ACN 105 0.5 M TBAClO4/DME

Medium electrolyte/solvent

76

129

2.5109 M1s1

138

78,163

Ref.

2.8103 s1

0.36103 – 4.8103 s1 (kf) 0.11103 – 1.7103 s1(kb)

6105 s1 (Br) 60 s1 (Cl)

Rate constant (k)

(C)EC

ECEC

EC2

EC

5

6

7

8

• Fe(CO)5 e Fe(CO)5 fast • Fe(CO) 4 CO • Fe(CO) Fe(CO)2 4e 4 k 2 Fe(CO) 4 Fe(CO)5 Fe2(CO)2 8 CO 10-Methylacridan oxidation and deprotonation by substituted pyridines AHe AH. k AH. B A• BH. Methylbenzene oxidation and deprotonation by substituted pyridines Ar-CH3 Fe(III) Ar-CH3 Fe(II) k Ar-CH.3 B ArCH.2 BH

Dixanthelene reduction and isomerization

Pt disc r 5 m

Au disc r 2.5; 8.5 m

Au disc r 100 m

Hg hemisphere r 12.5 m

0.1 M TBAClO4/ACN or THF

TEABF4/ACN

0.3 M TBABF4/THF

0.3 M TBAPF6/DMF

20

20

22

2103 – 4 104

5103 – 5104

50 to 20

2 102 – 5103

102–103

166

165

164

3102 – 109 M1 s1 167,168

2104 – 2.5107 M1 s1

1.3 l03 – 1.4 104 s1(kf) 1.1 l02 – 1.4 l03 s1 (kbneutral) 6 106 M1 s1

58

J. Heinze and K. Borgwarth

alternatively, by potential jumps. The use of microelectrodes strongly facilitates both approaches, since it provides small RC cell constants and iR drops to enable the precise control of the experimental parameters, which is essential for kinetic studies. Obviously, the homogeneous reaction between the generated strong oxidant and strong reductant takes place in the diffusion field around the electrode. The size of the reaction zone is determined by the potential sequence applied as well as the kinetics involved. The group of Wightman followed a very direct approach and determined the distance of the light emitting zone from a microelectrode’s surface while oxidizing and reducing a mixture of benzonitrile and 9,10-diphenylanthracene (DPA) with a microscope.146 The sixth reduction wave of fullerene C60 in a mixture of toluene and acetonitrile occurs as a catalytic reaction. Its rate constant has been determined applying steady-state and fast-scan cyclovoltammetry as 85.8 s1.147 An extended review on mechanistic studies in organometallic chemistry with special attention to FSV has been written by F. Battaglini et al.148 In this field, ECE represents the most common mechanism involved in electrochemically initiated ET chain catalysis. The oxidation and reduction of W(CO)2(dppe)2 causes an isomerization since the cis-form is favored in the neutral complex whereas the trans-form predominates in the oxidized species. The isomerization could be observed at 1000 V s1.149

2.5

Kinetics under transient conditions

The group of McCreery150–152 performed some pioneering work to prove through chronoamperometry and spectroscopy that within 150 ns concentration profiles of electroactive particles build up in the diffusion layer region of ultramicroelectrodes. They observed the electron transfer between the radical cation of chlorpromazine (CZP) and dopamine, present in equimolar quantities, and found a rate constant of 6.2 107 M1 s1 at pH 6.8. The electrochemical oxidation of pyrrole derivatives undergoing radical– radical coupling and deprotonation has been analyzed using double stepchronoamperometry. The lifetimes in 10 mM solution were determined ranging from 7.7 to 400 s depending on the alkyl- (slowest) phenylsulfonyl- (fastest), and alkylsilyl-substituents.153 A theoretical study of the Fleischmann group154 was devoted to the deposition and dissolution of catalytic centers at constant potentials. Mathematical solutions for the nucleation and three-dimensional growth of a new phase’s single growth center are given and cover the range of kinetically controlled to diffusioncontrolled regime. The quantitative electrochemical detection of catecholamine and related compounds from individual biological cells is strongly influenced by the following disproportionation yielding in a DISP1 scheme. Therefore, the kinetics have been investigated using chronoamperometry at platinum and glassy carbon microelectrodes. It turned out that shortly after applying an adequate potential only one electron per species has been transferred. This figure continuously increased due

Homogeneous reaction kinetics 59 to the insetting disproportionation up to two electrons after 10 s duration of the experiment due to the sluggish heterogeneous rate at carbon electrodes no DISP reaction could be observed.155

2.6

Kinetics between two electrodes

Two electrodes facing each other were first used by Engström et al.85 to investigate reaction mechanisms. In an SECM configuration they investigated the four electron reduction of epinephrine to adrenochrome with adrenolinequinone as an intermediate. While a large sample electrode served as the generator, the tip ultramicroelectrode acted as the collector (SG/TC). The current measured as a function of electrode separation followed the expectations for the ECE mechanism qualitatively (Figure 2.12). But at that time, a precise interpretation of the experiments lacked of reliable simulation. The oxidation of N,N-dimethyl-pphenylenediamine (DMPPD), which occurs as a two-electron transfer followed by a deamination (ErC1i),86,156 was studied by Bard et al. in dependence on the pH value. The steady-state feedback and chronoamperometric experiments above a conducting sample are shown in Figure 2.13.86 The authors were able to determine first-order rate constants of kc 19 s1 (pH 11.2), kc 5.8 s1 (pH 10.8), kc 1.4 s1 (pH 10.2) in good agreement with theoretical tip current transients. Furthermore, the SG/TC mode of the SECM has been applied in investigations of the sodium borohydride oxidation to distinguish whether the intermediates formed in a two-electron step87 react directly on the electrode surface or inside the solution by a chemical reaction. In the latter case, intermediates leaving the generator should be detected by a reduction current at the tip. This was

0.5

0

0

10 Distance (µm)

20

Figure 2.12 Dependence of adrenalinequinone concentration on the distance from the conductive substrate in a SG/TC experiment. ( ), experimental; (- - - - - -), first-order simulation; (———), second-order simulation.85

……

60

J. Heinze and K. Borgwarth 3

2 IT

Conductor

1

Insulator 0

0

10 d (µm)

20

Figure 2.13 Steady state IT –d curves for the oxidation of DMPPD (0.5 mM) at a 25 m Pt tip over an unbiased substrate at pH 10.20 (嘷嘷嘷嘷), pH 10.78 (▲▲▲), and pH 11.24 ( ): The solid lines represent the theoretical behavior for insulating and conductive materials.86

indeed the case; the rate constant was estimated to be in the order of 500 s1 in agreement with results obtain in FSV. Concerning second-order reactions, theoretical considerations treat dimerization reactions occurring after the electron transfer.88 Experiments have been conducted to study the reductive coupling of dimethyl fumarate (DF) respectively fumaronitrile (FN). Both systems had been topic of previous studies.157,158 In accordance with these results, rate constants of 180 M1 s1 (DF) respectively (2.0 0.4) 105 M1 s1 (FN) have been found. Especially, the determination of fast reaction kinetics was strongly facilitated by the fast mass transport that has been enabled by the SECM approach using a 10 m tip in a distance of about 1 m. Later, the same group worked on determining the very fast kinetics of the electrohydrodimerization of acrylonitrile (AN).89 This compound was found to undergo a dimerization after a one-electron reduction and, subsequently, to become protonated twice yielding an uncharged product. The time necessary for the diffusion through the gap between generator and collector electrode could be reduced significantly enabling the amperometric detection of the intermediate radical anion. Effectively, beside other relevant parameters like the diffusion coefficient of the intermediate the rate constant for the coupling has been determined to be as high as (6 3) 107 M1 s1. An even faster reaction has been observed when oxidizing 4-nitrophenolate in acetonitrile; the phenoxyl radical cations irreversibly dimerize at a rate of (1.2 0.3) 108 M1 s1.90 Steady state current–distance curves were used to differentiate between possible ECE and DISP1 mechanisms of the reduction of anthracene in DMF in the presence of phenol. Bard and Unwin found that the reaction follows a DISP1 pathway. The rate constant for the protonation of the anthracene radical cation by phenol has been determined to be (4.4 0.4) 103 M1 s1.91

Homogeneous reaction kinetics 61

2.7

Conclusions

Microelectrodes are powerful probes for the elucidation of homogeneous reaction mechanisms. The results obtained up to now indicate that the fast-scan technique may be superior to steady-state methods. Its crucial advantage is that all methods of evaluation used in classical cyclic voltammetry, including simulation, can be applied. Thus, all the usual mechanisms can be analyzed, and scan rates up to 106 Vs1 are accessible. The new SECM method which involves the application of nanodes will enlarge this scope, because both steady-state and fast-scan techniques are accessible within the same experimental setup.

References 1. A. M. Bond and T. F. Man, Electrochim. Acta 32, 863 (1987). 2. A. M. Bond, M. Fleischmann and J. Robinson, J. Electroanal. Chem. 180, 257 (1984). 3. J. Ghorogchian, F. Sarfarazi, T. Dibble, J. Cassidy, J. J. Smith, A. Russell, G. Dunmore, M. Fleischmann and S. Pons, Anal. Chem. 58, 2278 (1986). 4. A. J. Bard, F.-R. F. Fan, J. Kwak and O. Lev, Anal. Chem. 61, 132 (1989). 5. A. J. Bard, F.-R. F. Fan and M. Mirkin, Physical Electrochemistry, Marcel Dekker, New York, 209 (1995). 6. A. J. Bard, F.-R. F. Fan and M. Mirkin, Electroanalytical Chemistry, Marcel Dekker, New York, 243 (1994). 7. H.-Y. Liu, F.-R. F. Fan, C. W. Lin and A. J. Bard, J. Am. Chem. Soc. 108, 3828 (1986). 8. R. C. Engstrom, M. Weber, D. J. Wunder, R. Burgess and S. Winquist, Anal. Chem. 58, 844 (1986). 9. A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and Applications, Wiley, New York (1980). 10. J. Crank, The Mathematics of Diffusion, Clarendon Press, Oxford (1976). 11. A. M. Bond, M. Fleischmann and J. Robinson, J. Electroanal. Chem. 168, 299 (1984). 12. M. A. Dayton, J. C. Brown, K. J. Stutts and R. M. Wightman, Anal. Chem. 56, 946 (1980). 13. J. O. Howell and R. M. Wightman, Anal. Chem. 56, 524 (1984). 14. P. Tschuncky and J. Heinze, Anal. Chem. 67, 4020 (1995). 15. D. T. Miles, A. Knedlik and D. O. Wipf, Anal. Chem. 69, 1240 (1997). 16. K. R. Wehmeyer, M. R. Deakin and R. M. Wightman, Anal. Chem. 57, 1913 (1985). 17. T. V. Shea and A. J. Bard, Anal. Chem. 59, 2101 (1987). 18. P. M. Kovach, W. L. Caudill, D. G. Peters and R. M. Wightman, J. Electroanal. Chem. 185, 285 (1985). 19. J. S. Symanski and S. Bruckenstein, J. Electrochem. Soc. 131, 110 (1984). 20. D. R. MacFarlane and D. K. Y. Wong, J. Electroanal. Chem. 185, 197 (1985). 21. M. Fleischmann, S. Bandyopadhyay and S. Pons, J. Phys. Chem. 89, 5537 (1985). 22. T. Gueshi, K. Tokuda and H. Matsuda, J. Electroanal. Chem. 89, 247 (1978). 23. T. Gueshi, K. Tokuda and H. Matsuda, J. Electroanal. Chem. 101, 29 (1979). 24. T. Gueshi, K. Tokuda and H. Matsuda, J. Electroanal. Chem. 102, 41 (1979). 25. F. Belal and J. L. Anderson, Analyst 110, 1493 (1985). 26. M. Ciszkowska and Z. Stojek, J. Electroanal. Chem. 191, 101 (1985). 27. H. Meyer, H. Drewer, B. Gründig, K. Cammann, R. Kakerow, Y. Manoli, W. Mokwa and M. Rospert, Anal. Chem. 67, 1164 (1995). 28. K. L. Drake, K. D. Wise, J. Farraye, D. J. Anderson and S. L. Bement, IEEE Transact. Biomed. Eng. 35, 713 (1988).

62

J. Heinze and K. Borgwarth

29. N. A. Blum, B. G. Carkhuff, J. H. C. Charles, R. L. Edwards and R. A. Meyer, IEEE Transact. Biomed. Eng. 38, 68 (1991). 30. D. J. Anderson, K. Najafi, S. J. Tanghe, D. A. Evans, K. L. Levy, J. F. Hetke, X. Xue, J. J. Zappia and K. D. White, IEEE Transact. Biomed. Eng. 36, 693 (1989). 31. K. Najafi and K. D. Wise, IEEE Transact. Electron. Dev. 32, 1206 (1985). 32. V. P. Menon and C. R. Martin, Anal. Chem. 67, 1920 (1995). 33. J. C. Hulteen, V. P. Menon and C. R. Martin, J. Chem. Soc., Faraday Trans. 92, 4029 (1996). 34. P. Ugo, L. M. Moretto, S. Bellomi, V. P. Menon and C. R. Martin, Anal. Chem. 68, 4160 (1996). 35. H. Reller, E. Kirowa-Eisner and E. Gileadi, J. Electroanal. Chem. 138, 65 (1982). 36. H. Reller, E. Kirowa-Eisner and E. Gileadi, J. Electroanal. Chem. 161, 247 (1984). 37. D. Shoup and A. Szabo, J. Electroanal. Chem. 140, 237 (1982). 38. K. R. Wehmeyer and R. M. Wightman, Anal. Chem. 57, 1989 (1985). 39. P. M. Kovach, M. R. Deakin and R. M. Wightman, J. Phys. Chem. 90, 4612 (1986). 40. J. V. Macpherson, S. Marcar and P. R. Unwin, Anal. Chem. 66, 2175 (1994). 41. J. Heinze, J. Electroanal. Chem. 124, 73 (1981). 42. K. Aoki and J. Osteryoung, J. Electroanal. Chem. 122, 19 (1981). 43. K. Aoki and J. Osteryoung, J. Electroanal. Chem. 160, 335 (1984). 44. K. B. Oldham, J. Electroanal. Chem. 122, 1 (1981). 45. K. B. Oldham and C. G. Zoski, J. Electroanal. Chem. 256, 11 (1988). 46. C. G. Zoski, A. M. Bond, E. T. Allinson and K. B. Oldham, Anal. Chem. 62, 37 (1990). 47. R. M. Penner, M. H. Hebden, T. L. Longin and N. S. Lewis, Science 250, 1118 (1990). 48. Y. Shao, M. V. Mirkin, G. Fish, S. Kokotov, D. Palanker and A. Lewis, Anal. Chem. 69, 1627 (1997). 49. J. Heinze and M. Störzbach, Ber. Bunsenges. Phys. Chem. 90, 1043 (1986). 50. M. Störzbach and J. Heinze, J. Electroanal. Chem. 346, 1 (1993). 51. C. Amatore, M. R. Deakin and R. M. Wightman, J. Electroanal. Chem. 206, 23 (1986). 52. M. R. Deakin, R. M. Wightman and C. A. Amatore, J. Electroanal. Chem. 215, 49 (1986). 53. C. Amatore, B. Fosset, M. R. Deakin and R. M. Wightman, J. Electroanal. Chem. 225, 33 (1987). 54. C. Michael and C. A. Amatore, J. Electroanal. Chem. 267, 33 (1989). 55. W. H. Reinmuth, J. Am. Chem. Soc. 79, 6358 (1957). 56. K. B. Oldham, C. G. Zoski, A. M. Bond and D. Sweigart, J. Electroanal. Chem. 248, 467 (1988). 57. K. B. Oldham, J. C. Mayland, C. G. Zoski and A. M. Bond, J. Electroanal. Chem. 270, 79 (1989). 58. Z. Galus, J. Golas and J. Osteryoung, J. Phys. Chem. 92, 1103 (1988). 59. J. Heinze, in Microelectrodes: Theory and Applications, Kluwer Academic Publishers, Dordrecht, 283 (1991). 60. M. Störzbach, Dissertation, Freiburg (1991). 61. M. Fleischmann, F. Lasserre and J. Robinson, J. Electroanal. Chem. 177, 115 (1984). 62. K. B. Oldham, J. Electroanal. Chem. 313, 3 (1991). 63. M. Fleischmann, F. Lasserre and J. Robinson, D. Swan, J. Electroanal. Chem. 177, 97 (1984). 64. G. Denuault, M. Fleischmann, D. Pletcher and O. R. Tutty, J. Electroanal. Chem. 280, 243 (1990). 65. G. Denuault, M. Fleischmann and D. Pletcher, J. Electroanal. Chem. 280, 255 (1990).

Homogeneous reaction kinetics 63 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80.

81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105.

G. Denuault and D. Pletcher, J. Electroanal. Chem. 305, 131 (1991). C. G. Phillips, J. Electroanal. Chem. 296, 255 (1990). M. A. Dayton, A. S. Ewing and R. M. Wightman, Anal. Chem. 52, 2392 (1980). M. Dong and G. Che, J. Electroanal. Chem. 309, 103 (1991). C. L. Miaw, J. F. Rusling and A. Owlia, Anal. Chem. 62, 268 (1990). Q. Zhuang and C. Hongyan, J. Electroanal. Chem. 346, 29 (1993). Q. Zhuang and H. Chen, J. Electroanal. Chem. 346, 471 (1993). Q. Zhuang and D. Sun, J. Electroanal. Chem. 440, 103 (1997). C. Amatore and C. Lefrou, J. Electroanal. Chem. 296, 335 (1990). C. P. Andrieux, P. Hapiot and J.-M. Saveant, Chem. Rev. 90, 723 (1990). W. J. Bowyer and D. H. Evans, J. Org. Chem. 53, 5234 (1988). D. O. Wipf and R. M. Wightman, Anal. Chem. 60, 2460 (1988). D. O. Wipf and R. M. Wightman, J. Phys. Chem. 93, 4286 (1989). C. Amatore, E. Maisonhaute and G. Simonneau, Electrochem. Comm. 2, 81 (2000). K. Borgwarth, Oberflächenanalyse und Strukturierung mit Hilfe des Rasterelektrochemischen Mikroskops und des Rastertunnelmikroskops, Shaker Verlag, Aachen, Germany (1996). C. Zhang and X. Zhou, J. Electroanal. Chem. 415, 65 (1996). F.-M. Matysik, Electrochim. Acta 42, 3113 (1997). G. Zhao, D. M. Giolando and J. R. Kirchhoff, Anal. Chem. 67, 1491 (1995). J. Kwak and A. J. Bard, Anal. Chem. 61, 1221 (1989). R. C. Engstrom, T. Meany, R. Tople and R. M. Wightman, Anal. Chem. 59, 2005 (1987). P. R. Unwin and A. J. Bard, J. Phys. Chem. 95, 7814 (1991). M. V. Mirkin, H. Yang and A. J. Bard, J. Electrochem. Soc. 119, 829 (1972). F. Zhou, P. R. Unwin and A. J. Bard, J. Phys. Chem. 96, 4917 (1992). F. Zhou and A. J. Bard, J. Am. Chem. Soc. 116, 393 (1994). D. A. Treichel, M. V. Mirkin and A. J. Bard, J. Phys. Chem. 98, 5751 (1994). C. Demaille, P. R. Unwin and A. J. Bard, J. Phys. Chem. 100, 14137 (1996). J. C. Imbeaux and J. M. Savéant, J. Electroanal. Chem. 28, 325 (1970). P. Delahay, New Instrumental Methods in Electrochemistry, Interscience, New York (1954). J. A. Frauenhofer and C. H. Bank, Potentiostats and its Applications, Butterworth, London (1972). D. Britz, J. Electroanal. Chem. 88, 309 (1978). D. Garreau and J. M. Saveant, J. Electroanal. Chem. 50, 1 (1974). K. B. Oldham, J. Electroanal. Chem. 237, 303 (1987). S. Bruckenstein, Anal. Chem. 59, 2098 (1987). A. M. Bond, D. Luscombe, K. B. Oldham and C. G. Zoski, J. Electroanal. Chem. 249, 1 (1988). K. B. Oldham and C. G. Zoski, Comprehensive Kinetics, Elsevier,, Amsterdam, 79 (1986). C. Amatore, M. R. Deakin and R. M. Wightman, J. Electroanal. Chem. 220, 49 (1987). K. B. Oldham, Microelectrodes: Theory and Applications, Kluwer Academic Publishers, Dordrecht, 83 (1991). K. B. Oldham, J. Electroanal. Chem. 250, 1 (1988). K. Daasbjerg, Acta Chem. Scand. 47, 398 (1993). D. Pletcher and H. Thompson, J. Chem. Soc., Faraday Trans. 99, 3669 (1997).

64

J. Heinze and K. Borgwarth

106. 107. 108. 109.

R. G. Compton, J. C. Eklund and S. D. Page, J. Phys. Chem. 99, 4211 (1995). L. S. Marcoux, R. N. Adams and S. W. Feldberg, J. Phys. Chem. 73, 2611 (1969). K. Debrodt and K. Heussler, J. Phys. Chem. 125, 35 (1981). E. T. Seo, R. F. Nelson, J. M. Fritsch, L. S. Marcoux, D. W. Leedy and R. N. Adams, J. Am. Chem. Soc. 88, 3498 (1966). I. Montenegro, Microelectrodes: Theory and Applications, Kluwer Academic Publishers, Dordrecht, 429 (1991). I. Montenegro and D. Pletcher, J. Electroanal. Chem. 248, 229 (1988). M. J. Medeiros, I. Montenegro and D. Pletcher, J. Electroanal. Chem. 290, 155 (1990). I. Montenegro, D. Pletc, E. A. Liolios, D. J. Mazur and C. Zawodzinski, J. Appl. Electrochem. 20, 54 (1990). M. W. Lehmann and D. H. Evans, J. Phys. Chem. B 102, 9928 (1998). S. Dong and G. Che, J. Electroanal. Chem. 315, 191 (1991). S. Dong and G. Che, Electrochim. Acta 37, 2695 (1992). M. Kalaji, L. Nyholm and L. M. Peter, J. Electroanal. Chem. 325, 269 (1992). R. John and G. G. Wallace, J. Electroanal. Chem. 306, 157 (1991). A. Jaworski, M. Donten, Z. Stojek and J. G. Osteryoung, Anal. Chem. 71, 167 (1999). S. Daniele, M. A. Baldo, C. Bragato, G. Denuault and M. E. Abdelsalam, Anal. Chem. 71, 811 (1999). M. Krausa and K. Schorb, J. Electroanal. Chem. 461, 10 (1999). Y. Huang, C. C. Neto, K. A. Pevear, M. M. B. Holl, D. A. Sweigart and Y. K. Chung, Inorg. Chim. Acta 226, 53 (1994). K. A. Pevear, M. M. B. Holl, G. B. Carpenter, A. L. Rieger, P. H. Rieger and D. A. Sweigart, Organometallics 14, 512 (1995). H. Taube and H. Myers, J. Am. Chem. Soc. 76, 2103 (1954). C. P. Andrieux, P. Hapiot and J.-M. Saveant, J. Phys. Chem. 92, 5987 (1988). C. P. Andrieux, P. Hapiot and J.-M. Saveant, J. Phys. Chem. 92, 5992 (1988). D. Larumbe, I. Gallando and C. P. Andrieux, J. Electroanal. Chem. 304, 241 (1991). H. Yang and A. J. Bard, J. Electroanal. Chem. 331, 913 (1992). C. Amatore, A. Jutand and F. Pfluger, J. Electroanal. Chem. 218, 361 (1987). C. P. Andrieux, P. Audebert, P. Hapiot and J.-M. Savéant, J. Am. Chem. Soc. 112, 2439 (1990). P. Tschuncky and J. Heinze, Synth. Met. 55, 1603 (1993). J. Heinze, H. John, M. Dietrich and P. Tschucky, Synth. Met., 119, 49 (2001). C. Amatore, G. Farsang, E. Maisonhaute and P. Simon, J. Electroanal. Chem. 462, 55 (1999). J. E. Baur, S. Wang and M. C. Brandt, Anal. Chem. 68, 3815 (1996). E. F. Sullenberger, S. F. Dressman and A. C. Michael, J. Phys. Chem. 98, 5347 (1994). K. Meerholz, P. Tschuncky and J. Heinze, J. Electroanal. Chem. 347, 425 (1993). A. Fitch and D. H. Evans, J. Electroanal. Chem. 202, 83 (1986). W. J. Bowyer and D. H. Evans, J. Electroanal. Chem. 240, 227 (1988). C. Amatore, M. Bayachou, J.-N. Verpeaux, L. Pospíˇsil and J. Fiedler, J. Electroanal. Chem. 387, 101 (1995). P. V. Robant, R. R. Schroeder and D. R. Rorabacher, Inorg. Chem. 32, 3957 (1993). C. Beriet and D. Pletcher, J. Electroanal. Chem. 361, 93 (1993). R. Müller, L. Lamberts and M. Evers, J. Electroanal. Chem. 401, 183 (1996). R. Müller, B. Dakova, L. Lamberts and M. Evers, Electrochim. Acta 39, 1961 (1994).

110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143.

Homogeneous reaction kinetics 65 144. R. Müller, L. Lamberts and M. Evers, J. Electroanal. Chem. 407, 175 (1996). 145. R. Müller, L. Lamberts and M. Evers, J. Electroanal. Chem. 417, 35 (1996). 146. R. M. Wightman, C. L. Curtis, P. A. Flowers, R. G. Maus and E. M. McDonald, J. Phys. Chem. B 102, 9991 (1998). 147. G. Diao and Z. Zhang, J. Electroanal. Chem. 429, 67 (1997). 148. F. Battaglini, E. J. Calvo and F. Doctorovich, J. Organometallic. Chem. 547, 1 (1997). 149. A. Vallat, M. Person, L. Roullier and E. Lavrion, Inorg. Chem. 26, 332 (1987). 150. R. S. Robinson and R. L. McCreery, Anal. Chem. 53, 997 (1981). 151. R. S. Robinson, C. W. McCurdy and R. L. McCreery, Anal. Chem. 54, 2356 (1982). 152. R. S. Robinson and R. L. McCreery, J. Electroanal. Chem. 182, 61 (1985). 153. P. M. Narula and R. E. Noftle, J. Electroanal. Chem. 464, 123 (1999). 154. M. Fleischmann, S. Pons, J. Sousa and J. Ghoroghchian, J. Electroanal. Chem. 366, 171 (1994). 155. E. L. Ciolkowski, K. M. Maness, P. S. Cahill, R. M. Wightman, D. E. Evans, B. Fosset and C. Amatore, Anal. Chem. 66, 3611 (1994). 156. L. K. J. Tong, J. Phys. Chem. 58, 1090 (1954). 157. J. V. Puglisi and A. J. Bard, J. Electrochem. Soc. 119, 829 (1972). 158. I. B. Goldberg, D. Boyd, R. Hirasama and A. J. Bard, J. Phys. Chem. 78, 295 (1974). 159. C. G. Zoski, D. A. Sweigart, N. J. Stone, P. H. Rieger, E. Mocellin, T. F. Mann, D. R. Mann, D. K. Gosser, M. M. Doeff and A. M. Bond, J. Electrochem. Soc. 110, 2109 (1988). 160. S. Daniele, P. Ugo and G. A. Mazzocchin, J. Electroanal. Chem. 267, 129 (1989). 161. S. U. Pedersen and K. Daasbjerg, Acta Chem. Scand. 43, 301 (1989). 162. K. Daasbjerg, S. U. Pedersen and H. Lund, Acta Chem. Scand. 43, 876 (1989). 163. D. O. Wipf and R. M. Wightman, Anal. Chem. 62, 98 (1990). 164. W. J. Bowyer, E. E. Engelman and D. H. Evans, J. Electroanal. Chem. 262, 67 (1989). 165. C. Amatore, J. N. Verpeaux and P. J. Krusic, Organometallics 7, 2426 (1988). 166. P. Hapiot, J. Moiroux and J. M. Savéant, J. Am. Chem. Soc. 112, 1337 (1990). 167. C. J. Schlesener, C. A. Amatore and J. K. Kochi, J. Phys. Chem. 90, 3747 (1986). 168. C. Amatore and C. Lefrou, J. Electroanal. Chem. 325, 239 (1992).

3

Fundamentals of scanning electrochemical microscopy Michael V. Mirkin and Benjamin R. Horrocks

3.1

Introduction

The production and characterisation of electrochemical microsystems requires the ability to do chemistry in micrometer and submicrometer domains. Many conventional electrochemical techniques are not useful in this area because they lack spatial resolution due to the large electrode size employed (millimetres). More recently, electrochemists have begun to appreciate the advantages in electroanalytical applications of ultramicroelectrodes (UMEs), that is, those with at least one dimension in the micrometer range.1 Their attractive features include the virtual absence of errors due to ohmic resistance and double layer charging current and the availability of a well-defined steady state relatively insensitive to convection. An attempt to combine the advantages of the UMEs with the precise positioning capabilities of the scanning probe microscopes (such as scanning tunnelling microscope (STM)) resulted in the development of scanning electrochemical microscopy (SECM; this abbreviation is also used to signify the device, that is, the microscope).2,3 In SECM one scans an amperometric or potentiometric UME (the ‘tip’) across the surface of the sample (the ‘substrate’) and monitors the current flowing at the tip or its potential (Figure 3.1). The dependence of the recorded electrochemical signal on the tip/substrate distance is used to image the surface topography and map chemical reactivity. In addition, the faradaic process at the tip may be used to induce reactions at the substrate for the purpose of studying interfacial kinetics or for microfabrication. While almost any kind of electrochemical measurement, like cyclic voltammetry, ac voltammetry or potentiometry, may be carried out in the SECM set-up, the addition of spatial resolution greatly increases the utility of this technique for studying various technologically important systems such as electrocatalysts and biosensors. The precise positioning capability of the SECM enables a broad range of applications in the area of microsystems. The SECM can be used to produce microstructures, visualise their topography and to probe local chemical reactivity on a micrometer or submicrometer scale. Unlike many scanning probe microscopes, quantitative theoretical models are available for many of the possible SECM experiments. In the simplest case the variation of tip current with tip/substrate distance provides an accurate calibration

Fundamentals of SECM 67 Bipotentiostat Tip ET, IT Reference 3-axis stage/ micropositioner

Auxiliary Substrate ES, IS ET, ES, IT, IS

To/from PC

Control signals from PC/ controller

Figure 3.1 Schematic diagram of the SECM apparatus.

of the distance scale, d, in units of tip radius, a. Quantitative models of SECM have been used to characterise various interfaces, evaluate the kinetics of electron transfer (ET) at metal and semiconductor electrodes, measure the rates of intermolecular ET at redox and conductive polymers, and immobilised enzymes.3 More complex heterogeneous processes, such as corrosion and crystal dissolution reactions, have also been studied.4,5 Finally, it is worth describing some of the limitations of the SECM technique, although some of these are technological rather than fundamental and it is likely that many will be overcome as the technology develops. First, the smallest tip size which is routinely employed is limited by the diameter of available microwire to c.1 m. Strategies for fabricating much smaller tips have been reported, however, the tip geometry is either less well characterised or the process requires specialised equipment. Recent work on laser-pulled micropipette tips has some promise for the production of submicrometer tips of well-defined geometry.6 Second, the time required to capture an image is much longer than other scanning probe microscopes (SPMs), for example, STM or atomic force microscopy (AFM) and these in turn are slower than optical methods. The basic reason is that the communication between tip and substrate relies on diffusion and therefore the velocity of the tip, v, must be small compared to the mass transport coefficient, D/d where d is the diffusion coefficient. Depending on the tip size and area scanned, the typical image acquisition times range from a few minutes to an hour. Again, smaller tips would improve the capture rate, though the technique will likely remain slow compared to other SPMs. Third, it may often be difficult to deconvolute topographic information and chemical reactivity. For example, determining tip height information over a mixed conductive/insulating sample. This remains an area of significant activity and approaches based on tip height modulation or the development of current-independent distance calibration methods have been reported (Section 3.5).

68

M. V. Mirkin and B. R. Horrocks

The theory and applications of SECM have been reviewed recently.3 The focus of this introduction is on the basic principles of the technique and more recent applications relevant to electrochemical microsystems, such as studies of heterogeneous reactions, surface imaging and microfabrication. 3.1.1

Modes of operation

The SECM can be used to study heterogeneous and homogeneous reaction kinetics, as a microscope, and as a tool for microfabrication. These applications make use of different modes of the SECM operation which are named according to the manner in which the tip interacts with the substrate. 3.1.1.1

Feedback mode

In the feedback mode the tip serves as the working electrode in a three- or four-electrode cell. The substrate may also be biased and serve as a second working electrode. In this case a four-electrode potentiostat (or bipotentiostat) is required to control the tip and substrate potentials (Figure 3.1). The standard type of tip is the cross-section of a micron-sized wire sealed in an insulator such as glass and the glass sheath is bevelled to allow close approach of the tip to the substrate. This tip geometry is favoured because the tip surface is parallel to the substrate and maximises the feedback effect discussed below. In some applications, for example, where nanometer-sized electrodes are desired, conically or spherically shaped electrodes can be used, although the magnitude of the feedback is smaller. Tip and substrate are immersed in a solution containing a redoxactive molecule, which we take to be oxidisable in the example below. When the tip potential is sufficiently positive, the oxidation of R occurs via the reaction Rne →O

(1)

at a rate governed by the diffusion of R to the tip. If the tip/substrate separation is much greater than the tip diameter (Figure 3.2) the steady-state current, iT, , is given by: iT, 4nFDca

(2)

F is the Faraday constant, n is the number of electrons transferred in the tip reaction, D is the diffusion coefficient of R, c is the bulk concentration of R, and a is the tip radius. Equation (2) applies to a disk-shaped tip, but similar equations hold for other geometries. When the tip is brought within a few tip radii of a conductive substrate, the O species formed in the reaction (1) diffuses to the substrate where it may be reduced back to R O ne →R

(3)

Fundamentals of SECM 69 2a Tip

ne–

iT

R

O

R

O

Positive feedback Negative feedback

d

L = d/a

ne–

Positive feedback

Substrate

Figure 3.2 Schematic diagram of the SECM feedback mode. R and O are respectively the reduced and oxidized forms of a redox mediator.

This reaction produces an additional flux of R to the tip and hence an increase in tip current (iT iT, ). This phenomenon is termed ‘positive feedback’ (Figure 3.2) and reaction (3) may be either an electrochemical or chemical process. As the tip-substrate distance (d ) decreases, iT increases without limit. If the substrate is an inert electrical insulator or the oxidation of R is irreversible, R is not regenerated at the substrate. At small d, iT iT, because the substrate simply hinders diffusion of species R to the tip from the bulk solution (Figure 3.2). Indeed, the closer the tip to the substrate the smaller will be iT (‘negative feedback’). Variations of iT can be related to changes in d and, hence, the substrate topography can be imaged by scanning the tip over the substrate surface. Since the mass transport rate in positive feedback experiments at small d is of order D/d, convective effects due to the motion of the tip during imaging at typical speeds of 10 m s1 are not usually important. The rate of mediator regeneration at the substrate determines the magnitude of the positive (or negative) feedback, therefore the measured feedback-distance behaviour (‘approach curve’) provides information on the kinetics of the process at the substrate.7 If reaction (3) is an electrochemical process, its rate can be controlled by applying a suitable potential to the substrate. Alternatively, if the solution contains only the reduced form of the redox species, most of the substrate, which is usually much larger than the tip, is in contact with a solution of R. The substrate in this case is poised at a potential much more negative than the standard potential of the mediator (E 0), and reaction (3) is usually diffusion-controlled. 3.1.1.2

Generation/collection mode and potentiometric SECM

Generation/collection (G/C) mode refers to experiments where species generated at the substrate are collected at the tip or vice-versa (Figure 3.3(a)). The ratio of

70

M. V. Mirkin and B. R. Horrocks (a)

R

Tip – Generator

Tip – Collector

ne–

O R

O

ne– Substrate – Collector

Substrate – Generator

(b)

IT

k f large

–

ne

k f small O

R kf

kb

L = d /a Phase 1 Phase 2

R Substrate

Figure 3.3 (a) Generation-collection mode of SECM operation. The UME tip collects species R generated at the substrate (SG/TC mode) or the substrate may collect species R generated at the tip (TG/SC mode). (b) Equilibrium perturbation mode. Species R is in equilibrium between two phases (drawn as bulk phases, but phase 2 could be an adsorbed layer). The tip current is due to electrolysis of R and the magnitude depends on the rate constant, kf, for the interfacial transfer/desorption reaction.

fluxes at the tip and the substrate defines the collection efficiency. The two different G/C modes are referred to as substrate generation/tip collection (SG/TC) and tip generation/substrate collection (TG/SC). Owing to the relative sizes of tip and substrate, the collection efficiency approaches 100% for stable species in TG/SC mode. Deviations of the collection efficiency from 100% can be used to investigate homogeneous chemical kinetics in the tip/substrate gap.8 Unlike the feedback (and TG/SC) mode, in SG/TC experiments the tip travels within a thick diffusion layer generated by the substrate. In fact, the first

Fundamentals of SECM 71 SECM-type measurements were aimed at probing such a layer.9 If the tip is an amperometric sensor, the reaction occurring at its surface may significantly disturb the substrate diffusion layer. This and other problems (e.g. stirring of the substrate diffusion layer by the moving tip and low collection efficiency, iT/iS) limit the applicability of the SG/TC mode. A compensating feature of the SG/TC mode is that the tip response is zero when there is no substrate reaction, whereas in feedback experiments the tip current is measured on top of a background due to diffusion of species from the bulk solution. The SG/TC mode is often simplest when employing a potentiometric tip, which does not consume analyte and disturbance to the concentration profile of electroactive species is minimized. An analytical treatment of the steady-state SG/TC experiment is available when a small substrate (a microdisk or a spherical cap) generates stable species which are sensed by a smaller tip, for example, an ion-selective microelectrode.10 For large substrates the diffusion field of the substrate does not attain a steadystate, however, for reversible ET at tip and substrate, the transient response to a potential step has been treated numerically and the tip response is found to reach a pseudo-steady state.11a This experiment was found to be able to distinguish the diffusion coefficients of the oxidised and reduced forms of a redox couple and cases where the usual assumption in electroanalysis, that these are equal, breaks down were reported. A treatment of the feedback mode for this problem has also been reported.11b While feedback mode is certainly advisable for heterogeneous kinetic measurements, SG/TC can be used for monitoring many heterogeneous processes such as corrosion and enzymatic reactions. Slevin and Unwin12 used a generation-collection approach to study chemical reactions at the liquid/liquid interface. They employed a Ag/AgCl potentiometric tip to determine Cl in the diffusion layer of a previously developed droppingelectrolyte electrode13 and fitted the resulting concentration profile to determine the Cl flux. This technique was used to study the kinetics of hydrolysis of triphenylmethyl chloride at the dichloroethane/water interface. 3.1.1.3

Penetration experiments

In this mode, a small SECM tip is used to penetrate the substrate, for example, a redox-active polymer film, and extract information about the variation of concentrations, kinetic and mass-transport parameters with depth.14 When the tip is relatively far from the underlying conductive or insulating support, conventional voltammetric measurements can be carried out. As the tip approaches the support, feedback effects occur and the current–distance curves are similar to those obtained in solution, if the film is homogeneous and reasonably conductive. Eventually, tunnelling effects may be important for very short distances; d 1 nm. 3.1.1.4

Ion-transfer feedback mode

The difference between this recently developed technique15 and other modes of the SECM is that the tip current is produced by ion-transfer processes. An ITIES

72

M. V. Mirkin and B. R. Horrocks

(interface between two immiscible liquids) supported at a micropipette is used as an amperometric tip. Either a liquid/liquid or a solid/liquid interface may serve as a substrate. The theoretical treatment is formally the same as for the feedback mode, although the ratio of the tip orifice diameter to total tip diameter is c.1.1, whereas the metal/glass ratio (RG) is c.10 for conventional UMEs. This has comparatively little effect for positive feedback experiments, but strongly influences negative feedback experiments where diffusion from the back of the tip becomes important.15,16 3.1.1.5

Equilibrium perturbation mode

This mode has been pioneered by Unwin and coworkers3f and is illustrated in Figure 3.3(b). It is a technique in which the electrochemical process at the tip is used to perturb an equilibrium at a second interface, by depletion of a component of the solution. The tip current transient is sensitive to the flux of that component as the position of the equilibrium adjusts. The steady-state tip current will lie between the value for negative feedback and positive feedback depending on the kinetics of the reaction. If the kinetics are rapid, pure positive feedback, numerically the same as for a simple redox couple at a metallic substrate, is observed. The use of smaller tips and decreased tip/substrate distances results in a higher mass transport rate and increases the range of measurable rate constants. This mode is a flexible method for obtaining heterogeneous kinetic information; systems studied include adsorption/desorption, solubility and partition equilibria. A related technique is double potential step chronoamperometric SECM in which the first potential step generates a reactive species which diffuses to the interface of interest and then undergoes a reaction or is adsorbed or transferred to a second phase. The second potential step collects the remaining reactive species. In combination with a kinetic model for the interfacial reaction the amount of reactive species ‘lost’ as a function of the time delay gives a measure of the interfacial kinetics. 3.1.2

SECM imaging

As should be apparent from the discussion of the various modes of SECM, the instrument may be used in different ways; as an imaging tool in which the feedback current or tip potential reflects both topography and chemical reactivity or with a stationary tip as a means to initiate and monitor chemical reactions with high spatial resolution (m–nm) and high mass transport rates of order D/d 艐1 cm s1. In the following sections we outline the instrumentation required (Section 3.2), briefly survey the methods which have been applied to quantitative modelling of various SECM modes (Section 3.3), discuss selected applications of relevance to electrochemical microsystems including SECM microfabrication (Section 3.4) and finally we describe some advances in SECM imaging technology (Section 3.5).

Fundamentals of SECM 73

3.2 3.2.1

Instrumentation Apparatus

The basic SECM apparatus consists of a tip position controller, electrochemical cell (including tip, substrate, counter electrode, and reference electrode), bipotentiostat, and data acquisition system (Figure 3.1). The tip is usually mounted on the z-axis of an x–y–z stage controlled by piezoelectric actuators, although DC servo motors or stepper motors have also been employed. The substrate is attached to the bottom of the electrochemical cell which is mounted on a stable platform. A vibration-free table may be required for high resolution work. An airtight SECM cell which leaves the tip mobile is hard to design, therefore the SECM has to be housed inside a glove box or bag if anaerobic conditions are required. A programmable controller is used to control the movement of all three axes. A bipotentiostat controls the potentials of the tip and substrate vs the reference electrode. Unless transient measurements are intended, the bipotentiostat does not have to be very fast. The capability to make low current measurements (nA–pA) is more important. SECM instruments suitable for 3D imaging require a PC equipped with an interface board to synchronise acquisition of the electrochemical data with the movement of the tip. Building a manually controlled SECM unit from commercially available components is relatively easy, but fairly sophisticated software and some electronic work are required to construct a computercontrolled apparatus. Details on the construction of SECM instruments can be found elsewhere.3,21,22 The first prototype of a commercial SECM instrument has been recently built by CH Instruments, Inc. (Cordova, TN). The instrument is based around Inchworm™ translators, a 3-axis stage and a bipotentiostat controlled by an external PC under a 32-bit Windows environment. Various standard electrochemical techniques are incorporated along with SECM imaging, approach curves and many of the modes described in Section 3.1.1.

3.2.2

Tips

The information obtainable from SECM measurements depends essentially on the tip type and size. Most SECM experiments employ m-sized metal disks. Such tips are fabricated by heat sealing a fine Pt or Au wire or a carbon fibre of the desired radius in a glass capillary to which a vacuum is applied using a simple oil pump. The glass is polished to expose the cross-section of the wire and the glass wall surrounding the conductive disk is bevelled to allow the tip to approach close to the substrate. The ratio of the radius of metal glass to metal alone is termed RG and is typically ~10. Detailed information on preparation of such tips can be found elsewhere.2a,3d Nanometer scale tips have been fabricated by electrochemically etching 125-mdiameter Pt–Ir rods in a solution of saturated CaCl2 and HCl at about 20 V rms ac.23,24 The tip was insulated with molten Apiezon wax and the very end of the tip

74

M. V. Mirkin and B. R. Horrocks

was exposed by placing it in an STM. These nm-sized conical tips are useful for penetration of soft interfacial films and for imaging.14a,b Small disk-shaped tips were prepared recently by sealing metal wires in glass using a laser-based micropipette puller.6 The tip was exposed either by etching away a small portion of glass insulator or by micropolishing. Several of these tips exhibited well-behaved voltammetry with effective radii from 2 to 500 nm. However, only relatively large tips (a 100 nm) were found to be suitable for quantitative electrochemical measurements on the basis of voltammetric, SEM, and SECM feedback characterisation. A novel method for preparation of micrometer and submicrometer spherical tips based on self-assembly from Au sols was reported by Demaille et al.25 The self-assembly of c.8 nm diameter Au particles occurred at the tip of a 0.1–1 m glass pipette containing a solution of 1,9-nonanedithiol. A spherical Au tip was formed in 2 h. Steady-state voltammograms and SECM approach curves obtained with these tips were in good agreement with the theory. Other tips which extend the range of analytes to species not accessible to metal tips have been reported. Horrocks et al.26 prepared biosensor tips for hydrogen peroxide by coating carbon fibres with an electrically wired enzyme (horseradish peroxidase). The liquid-membrane (micropipette-type) ion-selective electrodes (ISEs) can also serve as SECM tips. Micropipette ISEs for K, Zn2 and NH4 have been used to map local concentrations of ions with micrometer resolution.10c A disadvantage of a potentiometric SECM probe is the difficulty in determination of the tip-substrate distance which is often required for quantitative data analysis. Antimony tips can be used as both amperometric electrodes and potentiometric pH sensors.10b Another approach is to make a dual-channel tip, in which one channel operates as an ISE and the other as a distance sensor.10c,d A hybrid, optical/electrochemical microprobe can be used for SECM studies of photoelectrochemical reactions on semiconductors.27 This consists of an optical fibre coated with gold and insulated with a polymer film.27a The fibre creates a micrometer-scale focussed light spot on the substrate surface and the concentric gold ring serves as a tip electrode and monitors the products of the photoelectrochemical reaction at a substrate. Both the tip current and the substrate photocurrent can be used to image the photoelectrochemical response of the substrate. A micropipette filled with a solvent immiscible with the outer solution can also serve as an amperometric UME. In Ref. 28, interfacial electron transfer (ET) occurred between the aqueous ferrocyanide inside a micropipette and tetracyanoquinodimethane (TCNQ) in the outer dichloroethane solution. Similarly, to a conventional metal tip, the tip current increases due to the positive feedback as a micropipette approaches a conductive substrate. This type of tip may have some advantages since sub-micron pipettes of regular geometry are fabricated more readily than metal tips of comparable size. Ion-transfer at the liquid/liquid interface, can also be used as a source of the SECM signal. For example, SECM experiments with the tip current produced by ion-transfer of K from water to 1,2-dichloroethane facilitated by dibenzo-18-crown-6 (DB18C6) were carried out.15a The resulting approach curves at a second liquid/liquid interface, which

Fundamentals of SECM 75

was a source of K , showed normal positive feedback. This modification allows the SECM to probe interfacial ion-transfer reactions in a way similar to conventional feedback measurements of ET.

3.3

Theory

Recent reviews have discussed various aspects of the SECM theory.3 Here we briefly survey the different approaches to quantitative modelling of SECM experiments. The theoretical expressions are written in terms of a normalised tip/ substrate distance, L d/a, and a current, IT, normalised by the diffusion limited current to a disk-shaped UME, IT iT/4nFDac. The applications of nondisk-shaped tips (e.g. shaped as a cone or spherical cap) are discussed in Refs. 3d, 6 and 24. 3.3.1

Numerical simulation of SECM experiments

A particular feature of the SECM compared to other scanning probe microscopes is that the interaction of the tip and sample can often be accurately modelled. Such modelling typically involves numerical solution of a reaction–diffusion equation with the appropriate boundary conditions to describe the interfacial kinetics. This enables quantitative information to be obtained on many chemical and electrochemical parameters, for example, interfacial rate constants at the tip and substrate or homogeneous kinetics in the tip/substrate gap. Compared to the numerical simulation of standard electrochemical experiments, however, simulation of SECM experiments is computationally much more demanding. There are two basic reasons for this: (1) diffusion in at least two dimensions must be considered and (2) the concentration profiles are very steep at the boundary between the tip metal and insulating sheath and therefore a fine mesh is required in this region for accurate results. These factors significantly increase the size of the matrix problem to be solved and may severely restrict the size of the time step for which the algorithm is numerically stable. 3.3.1.1

Finite difference methods

The most common approach to simulation of the SECM has been via the adaptation of finite difference techniques to the cylindrical geometry of the SECM with a microdisk-shaped tip. Typically, a non-uniform grid is used in order to produce a fine mesh near the edge of the microdisk and a comparatively coarse mesh further away in both the radial or perpendicular direction. Depending on whether steady-state or time-dependent phenomena are of interest several approaches have been employed to solve the difference equations. Simple iterative methods such as successive over-relaxation can be used to obtain the steady-state concentration profile.29,30a This technique is very simple to implement either in code or on a spreadsheet. However, it is not particularly efficient nor does it provide directly the transient response, although it may be used to do so in combination

76

M. V. Mirkin and B. R. Horrocks

with a fully implicit time step. The majority of the work carried out by the finite difference method has employed the alternating direction implicit method which is at least as efficient in calculating the steady-state concentration profiles and in addition provides the complete transient response.8 In essence, for alternate half-time steps diffusion in the radial or the perpendicular direction is treated implicity. The difference equations which result are of the form which can be solved efficiently by the Thomas algorithm. Since the technique generates the whole current–time response the steady-state approach curves were obtained from the long-time limit of the current transient. Unwin and coworkers have applied this technique to the simulation of many different interfacial kinetic experiments with the SECM including finite heterogeneous ET kinetics, following chemical reactions, crystal dissolution and several types of equilibrium perturbation experiments.31 Some simulations are possible with explicit time stepping methods,33 these are however restricted to short time steps if a fine mesh is used, or conversely suffer from poor accuracy if the mesh is coarsened. The FQEFD method was employed by Matsue and coworkers in their treatment of enzyme-mediated feedback.32b This method updates the concentration profile using concentration values at the previous two time points. Although more sophisticated and efficient finite difference methods are available, for example, multigrid, or Krylov integrators37 for most axisymmetric SECM problems the techniques above are sufficient for solution on a modern PC in a reasonable time. 3.3.1.2

Integral equations

An analytical solution for the typical SECM problem of a disc-shaped tip over a flat substrate, with quasi-reversible ET at both tip and substrate, can be expressed in terms of multidimensional integral equations through the application of integral transforms to the diffusion equation.34,35 This solution applies to both transient and steady-state feedback experiments. The integral equations, however must be solved numerically. This was achieved using a finite difference technique and the resulting simulations were observed to be in good agreement with standard finite difference solution of the diffusion equation. Although this technique has not been widely applied, it does offer some advantages at the cost of some mathematical complexity. Since one spatial variable may be integrated out exactly the dimensionality of the problem is therefore reduced and the difference equations can be solved rapidly. As for convolution voltammetry at planar macroscopic electrodes, the basic equations are written in a form independent of the shape of the applied voltage perturbation. 3.3.1.3

Finite element methods

Although the first numerical treatment of the SECM feedback experiment was carried out by the finite element method (FEM),2b it has seldom been employed subsequently. The major reason for this seems to be that the finite difference techniques above are easy to implement and very familiar to most electrochemists.

(a) 1.8

1.6

1.4

IT 1.2

1.0 2

4 6

8 Normalised 10 distance

0.8 12 15

14 20

16

5 10 Normalised distance

25

(b) 1.8

1.6

1.4

IT 1.2

1.0 2

4 6

8 Normalised 10 distance

0.8 12 14

15 20

16

5 10 Normalised distance

25

Figure 3.4 SECM feedback mode images simulated by the boundary element method. (a) image of two inlaid microdiscs and (b) an image of a recessed (left) and raised microhemisphere (right). Each image consists of 100 steady-state current values across the substrate surface. (Reprinted with permission from Ref. 36. Copyright 1999 American Chemical Society.)

78

M. V. Mirkin and B. R. Horrocks

The FEM is however well suited to handle complex geometries and adaptive algorithms are able to refine the mesh where the concentration profiles are steep at boundaries between conductive and insulating regions. In two papers the boundary element method (BEM) has been employed to model SECM.36 In this technique the weighting function is chosen so that the weighted residual formulation of the problem can be written in terms of quantities evaluated only on the boundary of the simulation domain. The advantage of the method is the reduction of the dimensionality of the problem which enabled the simulation of SECM linescans and images. Since the mesh is generated only on the boundaries, that is, the surface of the electrode and insulator, the number of elements required is reduced. The SECM image of an inlaid disc electrode and raised or recessed hemispherical electrodes in an insulating plane were computed assuming a steady-state condition during imaging, that is, tip speed less than the diffusion mass transport coefficient (Figure 3.4). This type of simulation is not as easily carried out by the finite difference method since the symmetry of the problem is reduced when the tip moves away from the centre of the substrate disc. This would require a complex 3D grid and therefore greatly increase the computational burden. Although most workers have written their own code to solve SECM diffusion problems, finite element programs are widely available and some have proven useful for the simulation of microelectrode and cylindrically symmetrical SECM problems (2D).15,30b Such programs easily generate concentration profiles, however not all of these programs may be suited to electrochemical simulation where the concentration gradient is required and routines to integrate this over the electrode surface may need to be written.

3.3.2

Steady-state SECM experiments: analytical approximations

Since transient SECM measurements are more complex and charging current and uncompensated resistance effects more significant, quantitative studies are generally carried out under steady-state conditions whenever possible. The steadystate theory is therefore simpler and often approximate analytical expressions suitable for least-squares fitting of data can be found.

3.3.2.1

Diffusion-controlled reactions

The dimensionless current–distance curves for steady-state processes were obtained numerically for both insulating and conductive substrates at several values of RG assuming a diffusion-controlled mediator turnover, equal diffusion coefficients, and an infinitely large substrate2b (the current–distance curves for microscopic substrates can be found in Ref. 35). An analytical expression for an infinite conductive substrate is24 I Tc (L) iT/iT, 0.78377/L 0.3315 exp(1.0672/L) 0.68

(4)

Fundamentals of SECM 79 which fits the IT – L curve over an L interval from 0.05 to 20 to within 0.7%. For an insulating substrate, Eqn (5) is accurate to within 0.5% over the same L interval. ITins (L) 1/(0.15 1.5385/L 0.58 exp(1.14/L) 0.0908 exp[(L 6.3)/(1.017L)])

(5)

By fitting an experimental current–distance curve to theory, one can determine the zero tip-substrate separation point (L 0), which in turn allows the determination of L values essential for any quantitative SECM measurement. Both Eqns (4) and (5) were derived using the data tabulated in Ref. 2b for RG 10. Although the effect of RG is often negligible, when RG 10 the effect may be significant, especially in experiments with insulating substrates. Very small RG values (~1.1) are typical for micropipette-based tips. The appropriate values of the normalised steady-state SECM current for various values of RG have been calculated using a finite element technique.15b Approximate theory for tips shaped as a cone or a spherical cap was discussed in Ref. 24 and more recently using the boundary element method.36 If a small (submicrometer) tip is to be used in quantitative measurements, it is important to show that the metal surface is not recessed into the insulator. This is possible via a study of the approach curve, whereas voltammetry in bulk solution is less sensitive to tip geometry. On the other hand, a recessed tip may be produced intentionally, for example, for single molecule detection.38 Fan et al.38b simulated a special case of the recessed electrode with the radius of the circular hole in the insulator equal to the disk radius. An analytical approximation for recessed tips with the hole in the insulator sheath different from the disk radius is also available.6 3.3.2.2

Finite kinetics

For finite kinetics at the tip and a diffusion-controlled mediator regeneration at the substrate one can obtain an approximate equation for the tip current at any potential assuming no variation of current density across the tip surface:39 IT(E, L) [0.68 0.78377/L 0.3315 exp(1.0672/L)]/( 1/)

(6)

where 1 exp[nf (E E 0 )]DO/DR, k 0exp[␣nf (E E 0 )/mO, k0 is the standard rate constant, E is the tip potential, E 0 is the formal potential, ␣ is the transfer coefficient, n is the number of electrons transferred per redox event, and f F/RT (where F is the Faraday, R is the gas constant, and T is temperature), and the effective mass-transfer coefficient for SECM is mO 4DO[0.68 0.78377/L 0.3315 exp(1.0672/L)]/(a) iT(L)/(a2nFc)

(7)

One can see from Eqn (7) that at L >> 1, mO ~ D/a (as for a microdisk electrode alone), but at L << 1, mO ~ D/d, a thin-layer cell (TLC) type behaviour. At constant

80

M. V. Mirkin and B. R. Horrocks

L, Eqn (6) describes a quasireversible steady-state tip voltammogram. Such a curve can be obtained by scanning the potential of the tip while the substrate potential is held constant. By decreasing the tip-substrate distance, the mass-transport rate can be increased sufficiently for quantitative characterisation of the electrontransfer kinetics, preserving the advantages of steady-state methods, that is, the absence of problems associated with ohmic drop, adsorption, and charging current. When the kinetics of the tip reaction are fast, the local rate of an irreversible heterogeneous reaction occurring at a substrate can be extracted by fitting an experimental current–distance curve to Eqn (8):40 IT(L) IS(1 ITins/I Tc ) I ins T

(8a)

IS 0.78377/L(1 1/) [0.68 0.3315 exp(1.0672/L)]/[1 F(L, )] (8b) where I cT and I ins T are given by Eqns (4) and (5), respectively, and IS is the kinetically controlled substrate current; k f d/D, kf is the apparent heterogeneous rate constant (cm/s), and F(L, ) (11/ 7.3)/(110 40L). The numerical results fit Eqn (8) over an L interval from 0.1 to 1.5 and 2 log ␬ 3 within ~1–2%. Equation (8) is equally applicable to finite electrochemical and chemical kinetics at the substrate. The only difference is that an electrochemical rate conkf O) kf stant is a function of electrode potential, while for a chemical process (R → is potential-independent. Equation (8a) splits IT into two components; the fraction of the substrate current, IS, which is collected by the tip plus the contribution due to diffusion into the tip/substrate gap, I ins T. This is exact for a simple positive feedback experiment over a conductive substrate when all of the tip current is collected at the substrate; I Tc IS. However, it can be applied approximately as long as the product of the substrate reaction is stable and forms the basis for a semi-analytical treatment of chemical reactions following ET at the tip.8d 3.3.2.3

Generation-collection mode

In SG/TC mode the flux of the reaction at the substrate can be evaluated from the concentration profile of generated species. A typical experimental situation is a collection mode measurement of the products of an enzymatic reaction where the enzyme is immobilised in a small disc-shaped spot on the substrate. This problem has been treated analytically and the full concentration profile expressed as an integral which can be evaluated numerically.10b,e When the enzyme kinetics are rate limiting and the flux is uniform across the substrate, the concentration profile over the centre of the disk is:10b,e c(Z) (rf/D){(Z 2 1)1/2 Z}

(9)

where r is the substrate radius, Z d/r, D is the diffusion coefficient of the species sensed by the tip and f is the reaction flux/mol cm2 s1.

Fundamentals of SECM 81

3.4

Selected applications

SECM has found considerable use as a tool for the investigation of various technologically important materials and interfaces, for example, metallic corrosion,44–49 semiconductor photocatalysts,27,42,50 conducting polymers,14,17,41,43 liquid/liquid and liquid/gas interfaces.15,31 In these applications SECM has been employed to image the substrate or with the tip at a fixed location to study the kinetics of the interfacial reactions of interest.

3.4.1 3.4.1.1

Solid/liquid interfaces Corrosion

The SECM is one of a number of scanning probe techniques which have been employed to study the corrosion of metals, especially steel, in aqueous solution. The scanning reference electrode technique has provided direct evidence for the existence of local anodes and cathodes on a corroding iron surface and potentiometric microelectrodes were used to monitor pH and Cl in corrosion pits.51 More recently, the SECM has also been employed to image localised corrosion processes at higher resolution.44–49 Several workers have investigated the identity of precursor sites using the SECM to locate active areas on the substrate surface. Using redox couples such as Fe(CN)64/3 and Br/Br2 strong enhancement of the SECM feedback was observed at a small number of microscopic sites on oxide-covered titanium foils (Figure 3.5). The diameters of these active sites were in the range c.10–50 m at a density on the surface of 20 5 sites cm2 and in a later report 3–30 m at 180 sites cm2.52 Application of a positive potential to initiate pitting and subsequent examination by a range of techniques (energy dispersive X-ray analysis, SEM, AFM and confocal laser microscopy) showed that the active sites were usually associated with inclusions of Al and Si.47 A novel SECM probe consisting of an optical fibre with a concentric gold microring electrode was used to obtain simultaneous SECM and photoelectrochemical activity images of the oxide film. The use of the optical fibre to provide a scanning light spot provides similar information to the scanning laser spot techniques for mapping photoelectrochemical activity across a surface. The photocurrent was found to be lower over the inclusions, consistent with a metallic particle.27b Pitting precursors were also identified on nitrogen-bearing austenitic steel.54 The SECM was used to monitor pH and Cl at the pit precursors in combination with Raman spectroscopy. An increase in pitting resistance was observed as the nitrogen content increased. The authors concluded from the SECM and Raman data that nitrate ions were formed and incorporated into the passive film. Work at very high (submicron) resolution identified MnS inclusions as pitting precursors on stainless steel.46 Current densities of 1 A cm2 were observed for some MnS inclusions. It was suggested that the flux of Cl required to support this current density leads to an enhanced Cl concentration near the inclusion and in

82

M. V. Mirkin and B. R. Horrocks i L (nA)

(a)

18 16 14 12 10 8 6 4 2 0

300

µm

SECM tip

(b)

Bulk solution Br–

–

Br –

Br TiO2 Film

Br2

d

Ti

Figure 3.5 (a) SECM image of an electroactive site on a Ti/TiO2 substrate in a solution containing 50 mM KBr and 10 mM H2SO4. The substrate potential was 1.5 V and the tip potential was 0.0 V to reduce Br2 at the diffusion limited rate. The tip/substrate separation, d, was c.5 m. (b) Schematic of the imaging mechanism, the diagram is drawn approximately to scale, but the TiO2 film thickness (6.5 nm) is exaggerated. (Reprinted with permission from Ref. 52b. Copyright 1998 American Chemical Society.)

combination with a drop in the local pH this causes the steel to depassivate and a pit to form. The SECM has also been used to initiate pitting events at predetermined locations rather than to search for natural pitting sites.44 The basis of the technique is the generation of a pulse of Cl via reduction of trichloroacetic acid at the tip. This technique enabled the study of the effect of the time and potential of passivation as well as spatial location on the ease of pit initiation. Large tip current fluctuations were also observed (presumably from soluble iron(II) species) and assigned to the precursors to breakdown of the passive film.

Fundamentals of SECM 83 3.4.1.2

Semiconductors

Several studies of ET at semiconductor substrates via SECM techniques have been reported. Studies of the dark electrochemical behaviour of WSe2, TiO2 and Ta2O5 have shown that the ET rate is spatially heterogeneous when the semiconductor is in depletion.42,52,53 In the case of iodide, bromide and ferrocyanide on n-TiO2 the SECM images showed active sites for oxidations which occurred positive of the flatband potential. The reduction of ferricyanide which occurs at potentials where the n-TiO2 is in accumulation was observed to be spatially uniform. These results are consistent with metal-like properties for a small number of active sites. The current at these sites is therefore swamped unless the semiconductor is in depletion and the mean current across the semiconductor/liquid interface is small. In the case of ET at the interface p-WSe2/Ru(NH3)63(aq), the feedback current as a function of substrate potential was determined by the Boltzmann factor for the concentration of majority carriers at the interface at potentials negative of the flatband potential. This corresponds experimentally to an apparent transfer coefficient of 1. Such ideal behaviour was observed only on undamaged portions of the van-der Waal’s surface; at steps or pits the currents were much greater and increased less steeply with potential (Figure 3.6).42 It was suggested that the general observation in semiconductor electrochemistry of apparent transfer coefficents less than 1 could be explained by the existence of microscopic active sites which may dominate the measured dark current density under depletion conditions. The photoelectrochemical behaviour of titanium dioxide has also been studied; the scanning microring/optical fibre tip of Smyrl and coworkers was used to map the photoactivity of titanium dioxide films.27b More recently, SECM was used to study the mechanisms of oxidation and reduction at unbiased titanium dioxide photocatalysts.50 A model for metallised TiO2 photocatalysts was prepared by spray pyrolysis of titanyl acetylacetonate onto indium tin oxide glass (ITO). The half of the ITO uncoated with TiO2 served to mimic the action of metallic particles deposited on TiO2 particles. With the tip far from the semiconductor surface and oxidising ferrocyanide at the diffusion limited rate, bandgap illumination of the semiconductor caused a depletion of ferrocyanide and a reduction in the tip current. However, when the tip was within feedback range of the substrate, the tip current increased upon illumination indicating feedback of ferrocyanide via reduction of ferricyanide at the TiO2. This result demonstrated that both oxidation and reduction reactions occur on the illuminated semiconductor and that any Schottky barrier at the photocatalyst surface must therefore be relatively small. The extent of the increase in feedback current was greater when the bare ITO portion was insulated with epoxy. Saturation of the solution with oxygen inhibited the positive feedback and in fact, the tip current now decreased on illumination. The role of oxygen and the ITO is similar to that of an external bias, removing photogenerated electrons and favouring the oxidation reaction at the TiO2 surface. These studies provided support for the mechanism of enhancement of activity of TiO2 photocatalysts by deposition of small noble metal particles.

84

M. V. Mirkin and B. R. Horrocks

+0.3 V +0.2 V

0.2 nA

+0.1 V 0.0 V –0.1 V

0

30

60 90 120 150 Relative distance (µm)

180

210

Figure 3.6 SECM linescans across a stepped portion of a p-WSe2 surface in 0.1 M KCl, with the tip poised at a potential to reduce Ru(NH3)63 (4.39 mM) at the diffusion limited rate. The tip was 2 m diameter Pt and iT, was c.1.5 nA. The potential of the semiconductor substrate (vs SCE) is indicated on the right and the linescans are offset by c. 50 pA for clarity. The valence band edge is at 0.86 V. (Reprinted with permission from Ref. 42. Copyright 1994 American Chemical Society.)

3.4.1.3

Conductive polymers

Ionically conducting polymers (e.g. perfluoronated sulfonate polymers and dialysis membranes) and electronically conductive polymers (e.g. polyvinylferrocene and polypyrrole) are often important components of electrochemical sensors. Characterisation of their behaviour with regard to charge and mass transport is an ongoing area of research. The SECM provides some unique capabilities to image these processes as well as to study their kinetics. Typically, the various steps involved in charge and mass transport processes in these systems may involve interfacial ET or ion transfer (IT), diffusion, migration, solvation changes and ejection or incorporation of counterions in the polymeric structure. A complete understanding of such a complex process requires not only kinetic measurements, but also a determination of the location of the rate limiting processes, that is, in the bulk polymer, at the electrode/polymer interface or the polymer/solution interface. In addition, SECM measurements of ionic fluxes from polymeric films on electrode surfaces show that these polymer films may also show heterogeneity across their surface.17 The IT that occurs on electrochemical switching of redox-active polymers has been studied by tip/substrate cyclic voltammetry (T/S CV) in which the tip current is monitored against substrate potential and time whilst the tip

Fundamentals of SECM 85 potential is poised at a constant value to detect particular ions near the polymer from NaFion coated electrodes showed that the film. Ejection of Os(bpy)2/3 3 ions ejected from the film were mostly Os(III).55 T/S CV was also used to detect proton fluxes during the redox cycling of polyaniline and to differentiate Cl and H fluxes.57 Potentiometric tips were also used to detect Cl fluxes and provided direct evidence in support of the accepted mechanism of polyaniline oxidation.10 Electrochemical reduction of polypyrrole (PPY) similarly requires a compensating ion flux to maintain electroneutrality in the bulk polymer. Using a platinum tip poised at a potential sufficient to oxidise Br ions, it was demonstrated that reduction of PPYBr initially involves an influx of cations and that Br is ejected only at more extreme negative potentials (0.3 V vs SCE).41 The SECM can obtain spatially resolved information in the direction parallel to the film surface in a straightforward manner. However, it is also possible to probe the location of reactions inside the film in two ways; (1) through penetration experiments with nm-scale tips and (2) through approach curves. The penetration experiments are limited to soft films which neither damage the tip, nor are themselves strongly disturbed by its presence. The oxidation of Os(bpy)2 3 inside a 218-nm-thick NaFion film was studied directly at a 30 nm radius Pt tip.14a The method enables determination of the diffusion coefficient and kinetic parameters of the redox couple inside the film directly without the complication of processes at the polymer/liquid interface. In addition, the film thickness is measured by approaching to within tunneling distance of the underlying metallic substrate. The same method was applied in a study of the redox switching of polyvinylferrocene and the depth variation of the kinetic properties in a 300 nm film.14b On a coarser scale, SECM has been used to study the transport of ruthenium hexaamine and methylene blue in polyacrylamide gels using a 25 m Pt tip.58 A chronoamperometric experiment inside the gel yields the diffusion coefficient directly, independent of concentration and was used to observe the interaction of the redox species with DNA in the gel by measurement of the decrease in effective diffusion coefficient. The second method for studying spatial location of reactions in polymer films via approach curves is based on the idea that the magnitude of the feedback depends on the distance between the tip and the site at which the regeneration of the mediator takes place. For example, the effective tip/substrate distance appearing in the equation for feedback may correspond to the distance between the tip and the polymer/solution interface or the tip and the buried polymer/metal interface. A set of diagnostic criteria has been produced which enables the experimenter to distinguish between regeneration of the mediator at the polymer/ solution interface or in the bulk polymer.41 Analysis of approach curves was used to show that reduction of ferrocenium and Os(bpy)3 3 at oxidised PPY occurs within a submicrometer layer of polymer at the polymer/solution interface.41b This methodology works well when an independent measurement of the film thickness and the tip-film/solution interface separation is available to calibrate the distance scale. The latter can sometimes be obtained by directly touching the film and then retracting the tip.

86

M. V. Mirkin and B. R. Horrocks

3.4.1.4

Immobilised enzymes

The SECM has been employed to study the catalytic activity of immobilised redox enzymes. When the enzyme kinetics are too slow for feedback measurements, generation/collection mode can be used. In this section we discuss kinetic investigations and leave patterning of activity to Section 3.4.3 on microfabrication. The first example studied was the catalytic oxidation of -D-glucose to D-glucono--lactone inside a micrometer-thick porous layer of immobilised glucose oxidase (GOx) by Pierce et al. using feedback.56a The oxidised form of the mediator produced by the diffusion controlled tip reaction was reduced at the substrate surface by GOx as long as there is a sufficient concentration of glucose present. Zero-order enzyme-mediator kinetics were established and similar apparent heterogeneous rate constants evaluated for several mediators indicating saturation with respect to substrate and mediator under the conditions of the experiment. In general, the flux of the enzymatic process must be able to compete with the diffusion of mediator from bulk solution. A quantitative detection relation was reported which indicates that low mediator concentrations and large tip radius decrease the limit of detection. kcatenz 103 (DRed CRed)/a

(10)

where kcatenz is the maximum flux of the enzyme reaction and the right-hand side is proportional to the flux of mediator from the bulk solution to the tip. This flux is the background signal and should be less than kcatenz if the enzyme is to be detected. A comparable feedback detection scheme was used to observe the localized reaction of GOx in the pores of track-etched polycarbonate membranes and membrane-bound NADH-cytochrome c reductase in rat liver mitochondria.56b The SECM can also be used to monitor antibody–antigen complexation and DNA–protein interactions, through the use of labelling techniques to generate redox active species which can be detected. For example, Wittstock et al.85 imaged an enzymatic label, alkaline phosphatase, which was covalently attached to the antigen, digoxin, and was observed to bind to the corresponding antibody immobilised on a glass slide. 4-Aminophenol, an electroactive product of the enzymatic hydrolysis of 4-aminophenyl phosphate, was detected by an amperometric SECM tip. Shiku et al. demonstrated a dual immunoassay for the polypeptide hormones human placental lactogen (HPL) and human chorionic gonadotrophin (HCG). A glass substrate was spotted with anti-HPL and anti-HCG. After incubating the substrate in a solution containing the antigens, horseradish peroxidase (HRP)-labelled antibodies were used to form sandwich complexes as in standard enzyme-linked immunosorbent assay (ELISA). The presence of HRP was detected using a Pt tip to observe ferrocenium produced by HRP in a solution containing hydrogen peroxide. The tip current was used to quantify the amount of antigen and the two species were distinguished by their spatial position on the substrate as indicated in the SECM image. The study of non-redox enzymes by SECM is possible using potentiometric collection mode.86 An ammonium selective SECM tip was used to monitor

Fundamentals of SECM 87 the flux of ammonium ions produced via enzymatic hydrolysis of urea by urease immobilised on a gold microdisc substrate. Reversible inactivation of the enzyme, but no faradaic current, was observed when the potential of the gold substrate was scanned across the range 0.0 to 0.4 V vs Ag/AgCl. These phenomena were attributed to changes in the enzyme quaternary structure due to the changing electric field and ionic composition at the interface. 3.4.1.5

Dissolution of ionic crystals

The kinetics of dissolution of ionic solids has traditionally been studied via measurement of bulk concentration changes in, for example, stirred suspensions. Using the equilibrium perturbation mode of the SECM, the experimenter can control directly the interfacial undersaturation and make measurements at single crystal surfaces or with micrometer spatial resolution.5g Investigation of the dissolution of ionic solids by SECM is based on the use of the UME tip to oxidise or reduce a component of the crystal in order to generate a controllable undersaturation at the crystal surface.5 This induces crystal dissolution and the rate of dissolution determines the extent of feedback.5a Two limiting cases of the Burton–Cabrera–Frank model corresponding to detachment (n 1) and surface diffusion (n 2) as the rate limiting steps were simulated. j kn␴ n

(11)

where j is the flux of dissolving species, kn is the rate constant, n is the reaction order, and ␴ is the undersaturation produced by the tip reaction. The reaction orders were determined for dissolution of several different salts, for example, n 1 for the (100) face of copper sulfate pentahydrate5a and n 2 for the (010) face of potassium ferrocyanide.5c The spatial distribution of dissolution activity can also be imaged by scanning the SECM tip over the surface. Images of the dissolution rate around a single pit on the surface of potassium ferrocyanide trihydrate showed an increase in tip current near the edge of the pit indicating that the local dissolution rate there was more rapid than for the planar surface.5c Higher spatial resolution, was achieved using a combined AFM–SECM probe where the dissolution was initiated on the SECM scale by the faradaic tip current, but with simultaneous AFM imaging of the KBr crystal surface.5f These studies were extended to the case of salt dissolution in a solution containing no supporting electrolyte.5e Although the theory describing masstransport via diffusion/migration in such systems is more complicated, it was possible to fit the experimental current–distance curves for AgCl dissolution and to determine the rate constant and reaction order, n 2. 3.4.2

Liquid/liquid and liquid/gas interfaces

The reproducibility and smoothness of liquid/liquid interfaces makes the interface between two immiscible electrolyte solutions (ITIES) an interesting object

88

M. V. Mirkin and B. R. Horrocks

(a)

(b)

–ve

Liquid phase 1 ne– R1

Metal tip: in liquid phase 1

O1 ET ne–

IT

IT M+

O2

Pipette tip: in liquid phase 2

IT

R2

M+

Substrate: liquid phase 2

Substrate: liquid phase 3

Figure 3.7 Schematic diagrams of SECM experiments to probe (a) SECM electron transfer (ET) at liquid/liquid interface with coupled ion transfer (IT). (b) SECM ion transfer (IT) at liquid/liquid interface using pipette-type tip.

for fundamental studies of charge transfer (Figure 3.7). In addition, adsorption of surface active molecules can make a useful model of biological membranes. In this section we briefly discuss selected SECM studies of ET, IT, chemical and enzymatic reactions and the transfer of neutral molecules occurring at the liquid/air interface.

3.4.2.1

Electron-transfer at the ITIES

The SECM is capable of probing charge-transfer reactions at the ITIES 40 because the ITIES was found to be mechanically stable and sharp on a submicrometer scale. SECM theory at the liquid/liquid interface differs from conventional feedback mode theory at a metallic substrate because several charge transfer processes occur simultaneously. When SECM is used to probe ET at the ITIES, a metal tip is placed in the upper liquid phase containing, for example, the reduced form, R1, of a redox species (Figure 3.7). At positive tip potentials, the oxidised form of the species, O1, is produced. As the tip approaches the ITIES, R1 can be regenerated at the interface via the bimolecular ET reaction between O1 in the organic phase (o) and R2 in the aqueous phase (w) O1(o) R2(w)

12

R1(o) O2(w)

(12)

and positive feedback is observed. In SECM measurements, a nonpolarizable ITIES is poised by the concentrations of the potential-determining ions which provides a constant driving force for the ET process.59,60 Four steps may influence

Fundamentals of SECM 89 the tip current: organic mediator diffusion between the tip and the ITIES, interfacial reaction (Eqn (12)), diffusion of R2 in water, and charge-compensation by IT. The electrical current across the ITIES (iS) due to this multistage serial process can be expressed as40 1/iS 1/iTc 1/iET 1/id 1/iIT

(13)

where iTc , iET, id, and iIT are the characteristic limiting currents for the above four stages, respectively. The concentration of R2 is usually made sufficiently high to exclude the possibility of diffusion limitations in the bottom phase. If the common ion (e.g. ClO 4 which is also potential determining) concentration in both phases is sufficiently high then IT is fast. Under these conditions, the effective heterogeneous ET rate constant can be obtained by fitting experimental current– distance curves to Eqn (8).40,59 Recently an alternative strategy to study ET at ITIES using thin organic layers adhering to graphite electrodes has been demonstrated.61 In a study of ET from aqueous redox couples to zinc porphyrin in benzene, phospholipids were adsorbed at the interface and found to reduce the ET rate. The ET rate decreased as the number of methylene groups in the lipids increased, consistent with ET through the monolayer rather than at pinholes. The driving force dependence of the ET rate followed the prediction of Marcus theory with a transfer coefficient c.0.5 at low driving force and a decrease in ET rate at very high driving force where Marcus inverted behaviour was obtained.59 3.4.2.2

Ion-transfer at the ITIES

The IT mode of the SECM should extend the applicability of SECM techniques to many processes which cannot be studied by conventional metal electrodes.15 A micropipette-based electrode was used as the tip, and the tip current was produced by facilitated transfer of K from the aqueous solution inside the pipette into 1,2-dichloroethane (DCE) assisted by dibenzo-18-crown-6 (DB18C6)15a K(w) DB18C6 (DCE) [KDB18C6](DCE)

(at the pipette tip) (14)

With the concentration of K inside a pipette at least fifty times higher than the concentration of DB18C6 in DCE, the tip current was limited by diffusion of DB18C6 to the pipette orifice. When the tip approached the aqueous layer, the regeneration of DB18C6 occurred via an interfacial dissociation mechanism:60b [KDB18C6](DCE) K(w) DB18C6(DCE)

(at the ITIES)

(15)

Thus a positive feedback current was observed when the tip approached the water/DCE interface. Negative feedback was observed when the tip approached a glass insulator.

90

M. V. Mirkin and B. R. Horrocks

The SECM response depends on rates of both heterogeneous reactions (14) and (15). Thus the kinetics of facilitated IT and reverse IT reactions can be measured using methodologies developed previously for studies of ET processes.40,59 While facilitated transfer of potassium is rapid and diffusion-controlled feedback was observed, a kinetically controlled regeneration of the mediator was observed for the slower reaction of proton transfer assisted by 1,10-phenanthroline.15a The mass-transfer rate for IT measurements by SECM is similar to that for heterogeneous ET measurements, and the standard rate constants of the order of 1 cm/s should be measurable. The developed technique could also be used for studies of IT processes at the liquid/membrane interfaces and for imaging of biological and artificial membranes. Similar SECM experiments can be performed using a simple (unassisted) IT process.15b In this case both the top and the bottom phases contain the same ion at equilibrium. The micropipette tip is used to deplete concentration of this common ion in the top solvent near the ITIES in a type of equilibrium perturbation experiment. This depletion causes IT across the ITIES, which is observed as positive feedback (Figure 3.7).

3.4.2.3

Transfer of neutral molecules

The SECM tip can be used to locally deplete the concentration of an electroactive species near the ITIES and induce the transfer of this species from the second liquid phase. If the interfacial transfer is known to be rapid, this provides a methodology (called SECM-induced transfer by the original authors) for analysis of species in the second liquid phase without the tip penetrating that phase.87 The theoretical modelling of this mode was carried out using a finite difference technique, however for the particular case of slow interfacial transfer the theory is essentially the same as for unassisted IT experiments.15b In a general case in which diffusion in the second phase and interfacial transfer may limit the rate, the separation of these effects may be experimentally difficult. If the partition coefficient or diffusion coefficient ratio is known, the interfacial kinetics can be determined. The method was applied to oxygen transfer between water and non-aqueous solvents87 or water and laryngeal cartilage62 as well as charged species, for example the transfer of copper(II) ions between water and a 1,2dichloroethane solution of an oxime ligand.87 Transfer of oxygen across the air/water interface in the presence of a monolayer of 1-octadecanol was studied by steady-state SECM-induced transfer. The depletion of oxygen in the interfacial region by reduction at an inverted UME tip in the aqueous phase was used to drive the transfer of oxygen from air into water. The dependence of the tip current on the monolayer compression showed that the monolayer strongly reduced the transfer rate and the rate was in fact essentially determined by the fraction of accessible interface.88 Liquid/gas interfaces have also been studied by double potential step chronoamperometry; the transfer of tip generated bromine across the water/air interface was however

Fundamentals of SECM 91 observed to be diffusion controlled with heterogeneous rate constant 0.5 cm s1.63

3.4.3

Microfabrication

Scanning probe microscopes (SPMs) have often been applied to microfabrication. The motivation for the use of SPMs in microfabrication has often been the potential of these techniques to achieve higher spatial resolution than optical methods, for example, photolithography, which are limited by diffraction effects. The spatial resolution of SPM is limited only by the tip size, however, the typical tip size in SECM is c.1 m and therefore a significant increase in resolution over standard methods is not routinely obtained. Instead, SECM fabrication is often motivated by the possibility to use the technique to control chemical processes at the substrate. The use of the SECM for fabricating microscopic structures and patterning surfaces is based on its capacity for changing local chemical properties in small (micrometer or submicrometer-sized) domains. Several approaches to surface modification by SECM are outlined in this section. These may be divided roughly into methods in which the tip/substrate are the working electrode and auxiliary electrodes of a cell and methods in which the chemistry of the system is chosen so that the SECM feedback process leads to deposition or etching of the substrate (Figure 3.8).

(a)

(b)

ne– R

Tip

ne–

O

R

Tip

O

Mn+

n+

M

Substrate

ne– Substrate

ne–

Figure 3.8 Schematic diagrams of SECM microfabrication experiments. R and O are respectively the reduced and oxidized forms of a redox mediator. (a) Microfabrication with direct tip/substrate current flow. (e.g. metal deposition). (b) Microfabrication via tip/substrate feedback (e.g. metal etching).

92

M. V. Mirkin and B. R. Horrocks

3.4.3.1

Methods based on direct current flow between tip and substrate

In the direct mode of the SECM operation the current is passed between the conductive substrate and the tip which acts as a working electrode or as the counter electrode.64 In this way lines were etched in illuminated n-GaAs with the tip biased negative. The lack of tip/substrate distance control was addressed by coating the substrate with an ionically conductive polymer and maintaining the constant current in an electronic feedback loop to adjust the penetration of the STM-like tip into the film. Etched lines and high resolution deposition of Cu, Ag and Au were obtained by choosing the sign of the potential between tip and substrate. The tip scan rate however was limited to c.0.1 m s1.65 More recently, this method has been applied to deposit silver particles on mica substrates in a humid atmosphere where a thin water film on the substrate plays the role of the NaFion in the previous example (see Section 3.5.3, Figure 3.10).66 Microfabrication of conductive polymers has also been carried out by SECM and this application may be of greater technological significance since PPY, polyaniline and related compounds cannot be deposited by evaporation as metals. Schuhmann and coworkers have used a galvanostatic pulse sequence to ensure that the monomer is not depleted in the tip/substrate gap and used this approach to connect two gold electrodes with a line of PPY drawn on the insulator between.67 This technique was combined with a current-independent distance measurement and a feedback loop to create PPY towers of several hundred microns height and about 80 m diameter.67 An important area of SECM fabrication where conventional methods suffer from serious problems is the patterning of biological molecules for biosensor development, for example, SECM immunoassay. In general, photolithographic methods can denature or disrupt the function of many enzymes and proteins. An example of the patterning of glucose oxidase was demonstrated by usng the tip to electrochemically desorb alkanethiols from a self-assembled monolayer on a gold substrate. Well-defined areas of desorbed thiol were obtained in alkaline media (0.2 M KOH) on which cystamine could subsequently be adsorbed. Periodate-treated glucose oxidase was then attached to the terminal amino function via Schiff-base chemistry and the enzymatic spots were imaged by detection of hydrogen peroxide in generation/collection mode.83a Another approach to enzyme micropatterning was developed by Shiku et al.32a who used a scanned micropipette to create an array of ~20-msized droplets of deposition solution on a glass slide. Two different immobilisation mechanisms (i.e. direct immobilisation from solutions containing albumin and glutaraldehyde, and an antigen-antibody based method) resulted in a formation of well-defined micrometer-sized spots of active horseradish peroxidase which was then imaged with the SECM.

3.4.3.2

Methods based on feedback or localised generation of reagents

SECM fabrication using the feedback mode is straightforward in the case of etching (Figure 3.8); strong oxidants generated at the tip can be used to etch metals68

Fundamentals of SECM 93 69

or semiconductors. The approach curve (IT vs L) can be used to study the etching kinetics and in the case of semiconductors is sensitive to the position of the band edges and the majority carrier type.69b Injection of holes into n-type semiconductors by oxidants produced at the tip leads to the local build-up of positive charge in the semiconductor and the etching reaction proceeds with regeneration of the reduced form of the mediator. This results in positive feedback at the tip and the etching rate can in principle be obtained by analysing the tip current, negative feedback indicating that no etching occurs. On the other hand, injection of holes into p-type material does not lead to etching although a positive feedback is observed, because the charge rapidly diffuses away. It is also possible for this process to happen at n-type material if the oxidant is sufficient to produce an inversion layer at the surface. A partial list of semiconductors which have been etched by SECM feedback methods includes GaAs, Si, CdTe, Hg1 xCdxTe and GaP. In more recent work, Mandler and coworkers succeeded in etching silicon by oxidation of bromine at the tip in acidic fluoride solutions which ensure the surface is not protected by an oxide layer.71 One-electron oxidants were not able to etch silicon, presumably due to inversion of the surface layer and rapid dissipation of the injected charge. The authors suggested that bromine etches silicon owing to the trapping of holes at surface Si–Br bonds. Etching of metals by one-electron oxidants is also possible and Cu surfaces were etched by generation of Fe(phen)33 and Os(bipy)33. Again, there is the possibility for the injected charge to diffuse laterally. Using two microdisc electrodes in a twin electrode thin-layer cell configuration, this may be avoided and it was shown that in fact the etching of Cu by Ru(bipy)33 and bromine was diffusion controlled at the available mass transport rates for SECM.71 Metal deposition in the feedback mode is not entirely straightforward because generation of reductants in solutions of metal ions can lead to deposition on the tip as well as the substrate. However, gold structures were electrodeposited in this way in polymer films containing AuCl 4 by generating a reducing agent, Ru(NH3)62, at the tip.68a An alternative is to produce the metal ions locally by dissolution of a gold tip in bromide-containing solution.72 A positive feedback current is observed because the bromide ions are regenerated when the AuBr 4 electrodeposits on the substrate. The resolution in this mode is of the order of the tip diameter. Another approach is to use the tip to alter the local chemical composition of the solution in the tip/substrate gap.73 The deposition of Ag on Au was carried out in this way using the oxidation of nitrite at the tip to lower the pH and induce dissociation of Ag(NH3) 2 ; the free Ag ions are then reduced at the substrate. This system is an example of a ‘chemical lens’; the concentration profile of Ag is focussed to a zone of radius smaller than the tip because diffusion of ammonia from the bulk solution continuously removes free Ag. The technique was also applied to writing lines of Ag in AgCl films by generation of hydroquinone at the tip with free Ag as the ‘focussing’ agent.74 The chemical lens is similar to the confined etchant layer technique (CELT), though in the latter the focussing agent is consumed at the tip and the resolution of the technique is determined by the tip diameter.77,75 Inorganic materials can be deposited through the

94

M. V. Mirkin and B. R. Horrocks

use of the tip to generate a reagent at the substrate; deposition of Ni(OH)2 on Pt from a solution of Ni2 was obtained by reduction of methyviologen (MV2) at the tip which was in turn reoxidised at the unbiased Pt substrate with consumption of H. This produced a higher pH at the substrate than the tip or the bulk and Ni(OH)2 deposited chemically on the Pt.76 Two feedback-mode strategies have been reported for the deposition of conductive polymers. Heinze chose a monomer which is soluble in organic solvents, but not in aqueous solutions (2,5-bis(1-methyl-pyrrol-2yl)-thiophene).73 The substrate was coated with the monomer by thermal evaporation and polymerised by generation of oxidants in aqueous solution by feedback SECM. The substrate was finally washed in organic solvent to remove the excess monomer. The polymerised patterns were insoluble and remained on the substrate. Wipf and Zhou deposited polyaniline onto noble metals by applying a positive potential to the substrate and consuming protons at the tip in order to locally shift the oxidation potential of the monomer to a value negative of the substrate potential.78 Several methods have recently developed for the patterning of biomolecules or locally modulating their activity on surfaces. Matsue and coworkers showed that local generation of bromine or chlorine can be used to inactivate immobilised enzymes. The system studied was diaphorase which catalyses the oxidation of NADH and its activity was therefore imaged by the SECM using ferrocenemethanol as the mediator.32b The inactivated lines and spots showed up in the feedback mode images as low currents. The technique of deactivating the enzyme locally is ingenious, though rather wasteful, and these workers’ technique have more recently reported a different strategy in which self-assembled monolayers (SAMs) of the silane type on glass surfaces were locally modified (Figure 3.9).79 This was achieved using Fenton chemistry to generate hydroxyl radicals at the tip by reduction of Fe3 in peroxide containing solutions. Either positive or negative patterns could be made. Amino terminated silanes were used to covalently immobilise the enzyme and the SECM-generated hydroxyl radical could make a negative by destruction of the amino-terminated SAM. Alternatively, a thiol terminated monolayer could be locally etched away to produce a hydoxylic surface leaving the surrounding monolayer terminated by unreactive sulfonate groups. Subsequent adsorption of (3-aminopropyl)trimethoxysilane in the etch spots enabled the formation of positive patterns of enzymatic activity. These patterns of enzymatic activity were also imaged using the SECM, nicely illustrating the capability of both fabrication and characterisation by the technique through choosing the solution chemistry. Local activation of an enzyme, alcohol dehydrogenase immobilised on agarose beads, has also been demonstrated. The pH in the tip/substrate gap was raised from the bulk value of 6 towards the optimum for the enzyme, that is, 9, by reduction of oxygen and water, though in this case the enzymatic activity was observed by monitoring the reduced cofactor, NADH, via fluorescence microscopy.84 Another strategy for the immobilisation of enzymes has been described in which the pulse method of deposition of conductive polymers above has been used to pattern functionalised pyrroles for subsequent attachment of glucose

Fundamentals of SECM 95

HO (a)

E

CH3CH3CH3CH3

CH3 CH3 OH OH

HO

E

CH3

CH3 OH OH

E

(b) NH2 NH2 NH2 NH2

NH2 NH2 OH OH

E OH OH

HO SO3– SO3– OH OH

(c) SH SH SH SH

E SO3–

SO3–

NH2 NH2 SO3–

E SO3–

E ; Diaphorase

Figure 3.9 Preparation of micropatterns with diaphorase at SAM-immobilised glass surfaces by electrogenerated hydroxyl; (a) diaphorase is physicallyadsorbed onto the hydrophobic area; (b) diaphorase is chemically-linked to the hydroxyl-radical-nonattacked area to give a negative pattern; (c) diaphorase is chemically-linked to the hydroxyl-radical-attacked area to give a positive pattern. (Reprinted with permission from Ref. 79. Copyright 1997 American Chemical Society.)

oxidase.80 The authors were also able to image the enzyme activity by SECM and discussed the possibilities for fabrication of miniaturised biosensors by this method. Recently,81 a strategy for immobilisation of biomolecules which promises to be of considerable generality has been demonstrated; biotin hydrazine oxidised at the tip was observed to attach to carbon substrates and retained its

96

M. V. Mirkin and B. R. Horrocks

ability to bind avidin. The biotin–avidin combination is widely used for immobilisation of biomolecules by conjugation of the biomolecule to avidin. It should be noted though that the substrate potential had to be carefully controlled to prevent extensive deposition of biotin and the tip/substrate distance was kept small because of the instability of the oxidised biotin. The authors also showed that tipgenerated hydroxide was capable of removing the biotin/avidin complex. Despite the considerable technical difficulties, the area of biosensor miniaturisation using SECM techniques seems to offer promise for the future since the conventional lithographic methods often involve harsh chemical treatments and/or high temperatures whereas most of the strategies outlined above function in aqueous solution under ambient conditions. In all modes of surface modification mentioned above the spatial resolution is governed by the size of the SECM microprobe. With the commonly used micrometer-sized tips fabricating submicrometer structures is difficult. This may be possible to do using smaller metal tips6 and recently introduced micropipette tips which could be made as small as a few nm radius.15b 3.4.4

Single molecule electrochemistry

As the size of SECM tips is reduced, the number of molecules in the tip/substrate gap decreases strongly, although the currents remain measurable because each molecule makes many more trips between the tip and substrate per second. In fact, the single molecule limit can be reached using nm-sized tips fabricated by electrochemical etching of Pt–Ir wires.38 The etched Pt/Ir tip was coated with apiezon wax and the apex of the tip exposed by placing the insulated tip in the feedback loop of an STM. Some control over the exposed area was possible through adjustment of the tunnelling current. These tips exhibited normal voltammetry in bulk solution, however for several tips the approach curve showed a decrease near a conductive substrate (ITO-glass) rather than the expected positive feedback. This was attributed to compression of the wax insulator resulting in a reduction of the volume of solution under the tip. At fixed tip position, the current fluctuated between discrete current levels of c.1.6 pA which were interpreted as the presence of zero, one or two molecules in the tip/substrate gap turning over at a rate of 107 s1. Other approaches to single molecule experiments have been discussed; the mean number of molecules in the diffusion field around a nm-scale UME may be of the order of unity and this suggests that individual electrochemical events could be distinguished if the current is observed with sufficient time resolution. A variation of this idea which has been implemented is to observe electrogenerated chemiluminescence at a UME with a sample rate of c.125 kHz.89 Potentiometric experiments may also be possible with a suitable high-impedance voltmeter as long as the double layer capacitance scales with electrode area for the nm-scale tips.82 The Coulomb staircase which arises from the finite voltage change produced by single electron charging of the tip double layer capacitance has in fact been observed at nm UMEs in dilute (M) solutions of redox couples

Fundamentals of SECM 97 at room temperature.91 In this experiment the linear region of the current–voltage curve around the rest potential shows discrete steps of roughly equal height as the interfacial potential responds to single electron charging events.

3.5

Advances in SECM imaging

SECM images are obtained by rastering the tip above the substrate and recording the variations in the tip current (constant height mode) or z-coordinate (constant current mode). While constant height mode is simpler to implement and is the usual method, the constant current mode may offer a higher resolution and is preferable for imaging substrates where the topography and conductivity are both variable. 3.5.1

Constant height imaging

Constant height images are obtained by rastering the tip in a fixed x–y plane above the substrate surface. Variations in the current or potential at the tip during the scanning are converted to grayscale images or 3D plots. An image recorded in feedback mode can be processed to yield the substrate topography (height) using Eqn (4) for a conducting substrate or Eqn (5) for an insulator. This obviously fails if the reactivity or conductivity of the substrate is non-uniform. Constant height images of metals,2,28 ionic crystals,5,90 polymer films,17,20,92,93 and biological specimens32,56,94 have all been obtained. The image resolution depends upon the tip radius, and the distance between tip and sample.20 The use of nm-sized tips has been used to increase the lateral resolution from the usual micrometer range to c.30–50 nm.95 Further dramatic improvements for the usual amperometric feedback mode seem unlikely because at conductive substrates, smaller tips would require tip/substrate separations within tunnelling range and the tunnelling current will eventually dominate. On the other hand, the resolution at insulators is usually not as good in negative feedback imaging because the diffusion field spreads laterally within the tip/ substrate gap to a greater extent. However, some of the newer modes based on IT or experiments in thin water films (see Section 3.5.3) may circumvent these limitations. A method to improve resolution, the ‘chemical lens’, through the use of a following chemical reaction was described in Section 3.4.3 on microfabrication.20,77 Since the apparent size of an object in SECM is somewhat exaggerated because of lateral diffusion, when the mediator is chemically unstable the extent of lateral diffusion is reduced and the image ought to be sharpened. This can be achieved using a bulk concentration of a reagent which reacts with the form of the mediator generated at the tip. The price to be paid may be that the feedback effect decreases if the mediator is too unstable. A somewhat similar approach has been previously discussed by Engstrom et al.19b and also in generation-collection experiments where pH images were sharpened by the careful control of the buffer capacity of the solution.42

98

M. V. Mirkin and B. R. Horrocks

SECM imaging of surface reactivity rather than topography can be done either in a feedback or collection mode. The former provides the spatial distribution of the rate of a redox reaction responsible for mediator regeneration at the substrate. Using the collection mode of the SECM one can image fluxes of species produced or consumed at the substrate, for example, localised iontophoretic fluxes of electroactive species through porous polycarbonate membrane or hairless mouse skin.18 The rate of ion transport and radius of membrane pores can be determined from the analysis of SECM images (Section 3.3.2.3). In generation-collection mode, the diffusional broadening of the image is dependent on the flux at the substrate and is therefore virtually independent of the tip size. If the features on the substrate are too large, a steady-state condition is not reached. For this reason, feedback mode is preferred for imaging experiments. However, in certain situations where the substrate kinetics are slow, generation-collection mode may be essential. Immobilised enzymes are a typical case, in particular when the enzymes are adsorbed or bonded to a metallic surface on which redox mediator regeneration is fast and dominates the feedback. An example is the collection mode imaging of H2O2 fluxes produced by GOx adsorbed on Au.83a The image resolution was enhanced by addition of small amounts of catalase to the medium. 3.5.2

Constant current imaging

In constant current imaging, the piezoelectric actuator is part of an electronic feedback loop which maintains a fixed setpoint tip current, as in conventional STM. The difficulty arises for mixed insulator/conductor substrates for which the feedback loop cannot be set simply to withdraw the tip according to increases in current (conductors) or decreases in current (insulators). Constant current imaging, which is routine in STM, is therefore much harder to implement for SECM unless the conductivity of the sample is uniform. To control the piezoelectric movement over mixed conductivity substrates one needs a method for recognising the local nature of the substrate. Wipf et al. have described a constant current imaging device which uses an automatically switched servosystem in combination with tip position modulation (TPM).22 In TPM SECM the the tip-substrate distance is modulated by application of a small amplitude ac signal to a piezoelectric pusher. The modulated tip current which results is detected by a lock-in amplifier and the phase angle of this signal discriminates between conductors (in-phase) and insulators (anti-phase). A TPM image of a Kel–F/Au composite surface was able to accurately show the topography of both conductive (Au) and insulating (Kel–F) regions. However, some problems are encountered at the boundaries of the insulating/conducting domains. More recent work has instead concentrated on the development of tip current– independent distance measurements.43 The basic idea is to use a piezotube to make an UME tip oscillate at its resonant frequency. When the tip approaches the sample, hydrodynamic forces alter the frequency. Using a laser beam focussed on the tip and a photodiode, the change in tip oscillations can be detected with a lock-in amplifier and used to regulate the tip/substrate distance. Another

Fundamentals of SECM 99 possibility would be to utilise the frequency change of a quartz tuning fork to control tip/substrate distance.96 3.5.3

High resolution SECM imaging

The highest resolution images to date have employed a type of penetration mode in which a conical tip partially penetrates an Å-thick layer of water which is present on a solid surface in humid air. In fact, images of insulating surfaces, for example, mica, with nm-scale resolution were obtained in a constant current mode. There is however still discussion about the imaging mechanism in these experiments. Originally Guckenberger et al. obtained DNA images on mica and attributed the tip current to tunnelling processes although the distance between the tip and the metallic contact on the sample was of the order of mm.97 Fan and Bard 93 proposed instead that the images occurred by an electrochemical mechanism and suggested that a faradaic current begins to flow when a nm-sized conical tip, positioned 1–2 mm laterally from the Au contact, touches the liquid layer on the atomically smooth surface of mica.93 This arrangement was used to image molecular steps on the mica surface as well as DNA cast on mica. Other workers have also observed that a water bridge between the tip and the sample surface is necessary for successful imaging in this way.98 More recently, Ag clusters were deposited on mica as direct evidence for a faradaic component to the imaging mechanism (Figure 3.10).66 An alternative is to combine electrogenerated chemiluminescence (ECL) and SECM to produce a new type of optical microscopy.99 A small metal tip was used to generate ECL in close proximity to the substrate and the transmitted or reflected light was detected with a photomultiplier. The ECL and tip current provide topographic and chemical information. Although this technique may solve the problem of fabricating a small light source, the resolution currently remains at the micrometre level.

Tip

Au

O

R

e–

e–

R O Ag+

Ag0

Mica

Figure 3.10 Schematic diagram showing the mechanism proposed for deposition of Ag at very high resolution (nm) in humid air on insulating mica surfaces. The tip travels in a thin water layer and reduces species O with a corresponding faradaic process at the gold contact. (Adapted with permission from Ref. 66. Copyright 1997 American Chemical Society.)

100

M. V. Mirkin and B. R. Horrocks

Note added in proof Rapid progress in SECM has been made and new areas of application in cell biology, membranes and liquid/liquid interfaces have opened. In our opinion, the most significant advances from the point of view of researchers in microsystem technology have been the technical achievements that have resulted in combined AFM-SECM probes and the development of fabrication strategies for nm-scale SECM tips and new current-independent distance control methods which should allow imaging of rough surfaces with spatial resolution in the 10 nm range.100–103

Acknowledgements The support of our research in SECM by grants from NATO (95-0009), Action Research (BRH) the Petroleum Research Fund administrated by the American Chemical Society (MVM) and PSC-CUNY Research Award Program (MVM) is gratefully acknowledged.

References 1. (a) Wightman, R. M., Wipf, D. O. in Electroanalytical Chemistry, Vol. 15 (A. J. Bard, ed.), Marcel Dekker, New York, 1988, p. 267. (b) Bard, A. J., Faulkner, L. R. Electrochemical Methods, Wiley, New York, 1980. 2. (a) Bard, A. J., Fan, F.-R. F., Kwak, J., Lev, O. Anal. Chem. 1989, 61, 132. (b) Kwak, J., Bard, A. J. Anal. Chem. 1989, 61, 1221. 3. (a) Mirkin, M. V. Anal. Chem. 1996, 68, 177A. (b) Bard, A. J., Fan, F.-R. F., Mirkin, M. V. in Physical Electrochemistry: Principles, Methods and Applications (I. Rubinstein, ed.), Marcel Dekker, New York, 1995, p. 209. (c) Arca, M., Bard, A. J., Horrocks, B. R., Richards, T. C., Treichel, D. A. Analyst 1994, 119, 719. (d) Bard, A. J., Fan, F.-R. F., Mirkin, M. V. in Electroanalytical Chemistry, Vol. 18 (A. J. Bard, ed.), Marcel Dekker, New York, 1993, p. 243. (e) Bard, A. J., Fan, F.-R. F., Pierce, D. T., Unwin, P. R., Wipf, D. O., Zhou, F. Science 1991, 254, 68. (f ) Barker, A. L., Gonsalves, M., Macpherson, J. V., Slevin, C. J., Unwin, P. R. Anal. Chim. Acta 1999, 385, 223. 4. Casillas, N., Charlebois, S., Smyrl, W. H., White, H. S. J. Electrochem. Soc. 1994, 141, 636. 5. (a) Macpherson, J. V., Unwin, P. R. Phys. Chem. 1994, 98, 1704. (b) ibid., p. 11764. (c) ibid., 1995, 99, 3338. (d) ibid., p. 14824. (e) ibid., 1996, 100, 19475. (f ) Macpherson, J. V., Unwin, P. R., Hillier, A. C., Bard, A. J. J. Am. Chem. Soc. 1996, 118, 6445. (g) Unwin, P. R., Macpherson, J. V. Chem. Soc. Rev. 1995, 24, 109. 6. Shao, Y., Mirkin, M. V., Fish, G., Kokotov, S., Palanker, D., Lewis, A. Anal. Chem. 1997, 69, 1627. 7. Wipf, D. O., Bard, A. J. J. Electrochem. Soc. 1991, 138, 469. 8. (a) Unwin, P. R., Bard, A. J. J. Phys. Chem. 1991, 95, 7814. (b) Zhou, F., Unwin, P. R., Bard, A. J. J. Phys. Chem. 1992, 96, 4917. (c) Zhou, F., Bard, A. J. J. Am. Chem. Soc. 1994, 116, 393. (d) Treichel, D. A., Mirkin, M. V., Bard, A. J. J. Phys. Chem. 1994, 98, 5751. (e) Unwin, P. R., Bard, A. J., Demaille, C. J. Phys. Chem. 1996, 14137. 9. (a) Engstrom, R. C., Weber, M., Wunder, D. J., Burgess, R., Winquist, S. Anal. Chem. 1986, 58, 844. (b) Engstrom, R. C., Meaney, T., Tople, R., Wightman, R. M. Anal. Chem. 1987, 59, 2005. (c) Engstrom, R. C., Wightman, R. M., Kristensen, E. W. Anal. Chem. 1988, 60, 652.

Fundamentals of SECM 101 10. (a) Denuault, G., Troise-Frank, M. H., Peter, L. M. Faraday Discuss. Chem. Soc. 1992, 94, 23. (b) Horrocks, B. R., Mirkin, M. V., Pierce, D. T., Bard, A. J., Nagy, G., Toth, K., Anal. Chem. 1993, 65, 1213. (c) Wei, C., Bard, A. J., Nagy, G., Toth, K. Anal. Chem. 1995, 67, 1346. (d) Wei, C., Bard, A. J., Kapui, I., Nagy, G., Toth, K. Anal. Chem. 1996, 68, 2651. (e) Klusmann, E., Schultze, J. W. Electrochim. Acta 1997, 42, 3123. 11. (a) Martin, R. D., Unwin, P. R. Anal. Chem. 1998, 70, 276. (b) Martin, R. D., Unwin, P. R. J. Electroanal. Chem. 1997, 439, 123. 12. Slevin, C. J., Unwin, P. R. Langmuir 1997, 13, 4799. 13. Koryta, J., Vanysek, P., Brezina, M. J. Electroanal. Chem. 1976, 67, 263. 14. (a) Mirkin, M. V., Fan, F.-R. F., Bard, A. J. Science 1992, 257, 364. (b) Fan, F.-R. F., Mirkin, M. V., Bard, A. J. J. Phys. Chem. 1994, 98, 1475. 15. (a) Shao, Y., Mirkin, M. V. J. Electroanal. Chem. 1997, 439, 137. (b) Shao, Y., Mirkin, M. V. J. Phys. Chem. 1998, 102, 9915. 16. Amphlett, J. L., Denuault, G. J. Phys. Chem. B 1998, 102, 9946. 17. Jeon, I. C., Anson, F. C. Anal. Chem. 1994, 64, 2021. 18. Scott, E. R., White, H. S. Anal. Chem. 1993, 65, 1537. 19. (a) Wipf, D. O., Bard, A. J. J. Electrochem. Soc. 1991, 138, L4. (b) Engstrom, R. C., Small, B., Kattan, L. Anal. Chem. 1992, 64, 241. 20. Borgwarth, K., Ricken, C., Ebling, D. G., Heinze, J. Fresenius J. Anal. Chem. 1996, 356, 288. 21. Kwak, J., Bard, A. J. Anal. Chem. 1989, 61, 1794. 22. (a) Wipf, D. O., Bard, A. J. Anal. Chem. 1992, 64, 1362. (b) Wipf, D. O., Bard, A. J., Tallman, D. E. Anal. Chem. 1993, 65, 1373. 23. Nagahara, L. A., Thundat, T., Lindsay, S. M. Rev. Sci. Instrum. 1989, 60, 3128. 24. Mirkin, M. V., Fan, F.-R. F., Bard, A. J. J Electroanal. Chem. 1992, 328, 47. 25. Demaille, C., Brust, M., Tsionsky, M., Bard, A. J. Anal. Chem. 1997, 69, 2323. 26. Horrocks, B. R., Schmidtke, D., Heller, A., Bard, A. J. Anal. Chem. 1993, 65, 3605. 27. (a) Casillas, N., James, P., Smyrl, W. H. J. Electrochem. Soc. 1995, 142, L16. (b) James, P., Casillas, N., Smyrl, W. H. J. Electrochem. Soc. 1996, 143, 3853. (c) Pennarun, G. I., Boxall, C., O’Hare, D. Analyst 1996, 121, 1779. (d) Shi, G. D., GarfiasMesias, L. F., Smyrl, W. H. J. Electrochem. Soc. 1998, 145, 2011. 28. Solomon, T., Bard, A. J. Anal. Chem. 1995, 67, 2787. 29. Jeffrey, D., Horrocks, B. R., Armstrong, R. D., Walsh, A. B. Electrochim. Acta 1997, 42, 3637. 30. (a) http://www.staff.ncl.ac.uk/b.r.horrocks/ (b) http://www.fortranlib.com/ 31. Unwin, P. R. J. Chem. Soc. Faraday Trans. 1998, 94, 3183. 32. (a) Shiku, H., Matsue, T., Uchida, I. Anal. Chem. 1996, 68, 1276. (b) Shiku, H., Takeda, T., Yamada, H., Matsue, T., Uchida, I. Anal. Chem. 1995, 67, 312. 33. Nugues, S, Denuault, G. J. Electroanal. Chem. 1997, 408, 125. 34. Mirkin, M. V., Bard, A. J. J. Electroanal. Chem. 1992, 323, 29. 35. Bard, A. J., Mirkin, M. V., Unwin, P. R., Wipf, D. O. J. Phys. Chem. 1992, 96, 1861. 36. Fulian, Q., Fisher, A. C., Denuault, G. J. Phys. Chem. B, 1999, 103, 4387, ibid., p. 4393. 37. Bard, A. J., Denuault, G., Friesner, R. A., Dornblaser, B. C., Tuckerman, L. S. Anal. Chem. 1991, 63, 1282. 38. (a) Fan, F.- R. F., Bard, A. J. Science 1995, 267, 871. (b) Fan, F.-R. F., Kwak, J., Bard, A. J. J. Am. Chem. Soc. 1996, 118, 9669. 39. Mirkin, M. V., Bard, A. J. Anal. Chem. 1992, 64, 2293. 40. Wei, C., Bard, A. J., Mirkin, M. V. J. Phys. Chem. 1995, 99, 16033. 41. (a) Mirkin, M. V., Arca, M., Bard, A. J. J. Phys. Chem. 1993, 97, 10790, (b) Arca, M., Mirkin, M. V., Bard, A. J. J. Phys. Chem. 1995, 99, 5040.

102

M. V. Mirkin and B. R. Horrocks

42. Horrocks, B. R., Mirkin, M. V., Bard, A. J. J. Phys. Chem. 1994, 98, 9106. 43. (a) Ludwig, M., Kranz, C., Schuhmann, W., Gaub, H. E. Rev. Sci. Instr. 1995, 66, 2857. (b) Kranz, C., Gaub, H. E., Schuhmann, W. Advanced Materials 1996, 8, 634. (c) James, P. I., GarfiasMesias, L. F., Moyer, P. J., Smyrl, W. H. J. Electrochem. Soc. 1998, 145, L64. 44. Still, J. W., Wipf, D. O. J. Electrochem. Soc. 1997, 144, 2657. 45. Fushimi, K., Azumi, K., Seo, M. ISIJ International 1999, 39, 346. 46. Williams, D. E., Mohiuddin, T. F., Zhu, Y. Y. J. Electrochem. Soc. 1998, 145, 2664. 47. GarfiasMesias, L. F., Alodan, M., James, P. I., Smyrl, W. H. J. Electrochem. Soc. 1998, 145, 2005. 48. Tanabe, H., Togashi, K., Misawa, T., Mudali, U. K. J. Mat. Sci. Lett. 1998, 17, 551. 49. Razzini, G., Maffi, S., Mussati, G., Bicelli, L. P., Mitsi, G. Corrosion Science 1997, 39, 613. 50. Maeda, H., Ikeda, K., Hashimoto, K., Ajito, K., Morita, M., Fujishima, A. J. Phys. Chem. B 1999, 103, 3213. 51. (a) Luo, J. L., Lu, Y. C., Ives, M. B. J. Electroanal. Chem. 1992, 326, 51. (b) Luo, J. L., Lu, Y. C., Ives, M. B. Materials Performance 1992, 31, 44. 52. (a) Basame, S. B., White, H. S. J. Phys. Chem. 1995, 99, 16430. (b) Basame, S. B., White, H. S. J. Phys. Chem. 1998, 102, 9812. 53. Basame, S. B., White, H. S. Langmuir 1999, 15, 819. 54. Misawa, T., Tanabe, H. ISIJ International 1996, 36, 787. 55. Lee, C., Anson, F. C. Anal. Chem. 1992, 64, 528. 56. (a) Pierce, D. T., Unwin, P. R., Bard, A. J. Anal. Chem. 1992, 64, 1795. (b) Pierce, D. T., Bard, A. J. Anal. Chem. 1993, 65, 3598. 57. (a) Troise Frank, M. H., Denuault, G. J. Electroanal. Chem. 1993, 354, 331. (b) Troise Frank, M. H., Denuault, G. J. Electroanal. Chem. 1994, 379, 405. 58. Pyo, M., Bard, A. J. Electrochim. Acta 1997, 42, 3077. 59. (a) Tsionsky, M., Bard, A. J, Mirkin, M. V. J. Phys. Chem. 1996, 100, 17881. (b) Tsionsky, M., Bard, A. J, Mirkin, M. V. J. Am. Chem. Soc. 1997, 119, 10785. 60. For reviews of the electrochemistry of ITIES, see (a) Girault, H. H., Schiffrin, D. J. in Electroanalytical Chemistry, Vol. 15 (A. J. Bard, ed.) Marcel Dekker. New York, 1989, p. 1. (b) Girault, H. H. in Modern Aspects of Electrochemistry, Vol. 25 (J. O’M. Bockris, B. E. Conway, R. E. White, eds), Plenum Press, New York, 1993, p. (c) Liquid-Liquid Interfaces. Theory and Methods, Volkov, A. G., Deamer, D. W., eds, CRC Press, Boca Raton, 1996. 61. Shi, C. N., Anson, F. C. J. Phys. Chem. B 1998, 102, 9850. 62. Macpherson, J. V., O’Hare, D., Unwin, P. R., Winlove, C. P. Biophys. J. 1997, 73, 2771. 63. Slevin, C. J., Macpherson, J. V., Unwin, P. R. J. Phys. Chem. B 1997, 101, 10851. 64. Craston, D. H., Lin, C. W., Bard, A. J. J. Electrochem. Soc. 1988, 135, 785. 65. Husser, O. E., Craston, D. H., Bard, A. J. J. Vac. Sci. Tech. B 1988, 6, 1873. 66. Forouzan, F., Bard, A. J. J. Phys. Chem. B 1997, 101, 10876. 67. Kranz, C., Ludwig, M., Gaub, H. E., Schuhmann, W. Advanced Materials, 1995, 7, 38, ibid., p. 568. 68. (a) Mandler, D., Bard, A. J. J. Electrochem. Soc. 1990, 137, 1079. (b) Mandler, D., Bard, A. J. J. Electrochem. Soc. 1989, 136, 3143. 69. (a) Mandler, D., Bard, A. J. J. Electrochem. Soc. 1990, 137, 2468. (b) Mandler, D., Bard, A. J. Langmuir 1990, 6, 1489. 70. Meltzer, S., Mandler, D., J. Chem. Soc. Faraday Trans. 1995, 91, 1019. 71. Macpherson, J. V., Slevin, C. J., Unwin, P. R. J. Chem. Soc. Faraday Trans. 1996, 92, 3799.

Fundamentals of SECM 103 72. Meltzer, S., Mandler, D., J. Electrochem. Soc. 1995, 142, L82. 73. Borgwarth, K., Ricken, C., Ebling, D. G., Heinze, J. Ber. Bunsenges. Phys. Chem. 1995, 99, 1421. 74. Hess, C., Borgwarth, K., Ricken, C., Ebling, D. G., Heinze, J. Electrochim. Acta 1997, 42, 3065. 75. Zu, Y. B., Xie, L., Mao, B. W., Tian, Z. W. Electrochim. Acta 1998, 43, 1683. 76. Shohat, I., Mandler, D. J. Electrochem. Soc. 1994, 141, 995. 77. Tian, Z. W., Faraday Discuss. 1992, 94, 37. 78. Zhou, J., Wipf, D. O. J. Electrochem. Soc. 1997, 144, 1202. 79. Shiku, H., Uchida, I., Matsue, T. Langmuir 1997, 13, 7239. 80. (a) Kranz, C., Wittstock, G., Wohlschlager, H., Schuhmann, W. Electrochim. Acta. 1997, 42, 3105. (b) Schuhmann, W., Kranz, C., Wohlschlager, H., Strohmeier, J. Biosensors & Bioelectronics 1997, 12, 1157. 81. Nowall, W. B., Wipf, D. O., Kuhr, W. G. Anal. Chem. 1998, 70, 2601. 82. Bard, A. J., Fan, F.-R. F. Accts Chem. Res. 1996, 29, 572. 83. (a) Wittstock, G., Schuhmann, W. Anal. Chem. 1997, 69, 5059. (b) Wittstock, G., Hesse, R., Schuhmann, W. Electroanalysis 1997, 9, 746. 84. O’Brien, J. C., Shumaker-Parry, J., Engstrom, R. C. Anal. Chem. 1998, 70, 1307. 85. Wittstock, G., Yu, K., Halsall, H. B., Ridgway, T. H., Heineman, W. R. Anal. Chem. 1995, 67, 3578. 86. Horrocks, B. R., Mirkin, M. V. J. Chem. Soc. Faraday Trans. 1998, 94, 1115. 87. Barker, A. L., Macpherson, J. V., Slevin, C. J., Unwin, P. R. J. Phys. Chem. B 1998, 102, 1586. 88. Slevin, C. J., Ryley, S., Walton, D. J., Unwin, P. R. Langmuir 1998, 14, 5331. 89. Collinson, M. M., Wightman, R. M. Science 1995, 268, 1883. 90. Unwin, P. R., Bard, A. J. J. Phys. Chem. 1992, 96, 5035. 91. Fan, F.-R. F., Bard, A. J. Science 1997, 277, 1791. 92. Lee, C., Bard, A. J. Anal. Chem. 1990, 62, 1906. 93. Fan, F.-R. F., Bard, A. J. Science 1995, 270, 1849. 94. Lee, C., Kwak, J., Bard, A. J. Proc. Natl. Acad. Sci. USA 1990, 87, 1740. 95. Bard, A. J., Fan, F.-R. F. Faraday Discuss. Chem. Soc. 1992, 94, 1. 96. (a) Karrai, K., Grober, R. D. Appl. Phys. Lett. 1995, 66,1842. (b) Edwards, H., Taylor, L., Duncan, W., Melmed, A. J. J. Appl. Phys. 1997, 82, 980. 97. Guckenberger, R., Heim, M., Cevc, G., Knapp, H. F., Wiegrabe, W., Hillebrand, A. Science 1994, 266, 1538. 98. Patel, N., Davies, M. C., Lomas, M., Roberts, C. J., Tendler, S. J. B., Williams, P. M. J. Phys. Chem. B. 1997, 101, 5138. 99. Fan, F.-R. F., Cliffel, D., Bard, A. J. Anal. Chem. 1998, 70, 2941. 100. Kranz, C., Friedbacher, G., Mizaikoff, B. Anal. Chem. 2001, 73, 2491. 101. Hengstenberg, A., Kranz, C., Schuhmann, W. Chem-Eur. J. 2000, 6(9), 1547. 102. Macpherson, J. V., Unwin, P. R. Anal. Chem. 2001, 73(3), 550–557. 103. Macpherson, J. V., Unwin, P. R. Anal. Chem. 2000, 72(2), 276–285.

4

Electrochemical microcells and surface analysis M. M. Lohrengel and A. Moehring

4.1

Definitions

A scheme for geometric classification of electrochemical microcells is given in Figure 4.1. A rectangular shape is chosen to simplify comparisons, in special applications, however, any design is possible. In Figure 4.1(a) the size of the electrolyte volume is determined by x , y , and z , the size of the working electrode by x and y. The distance from the working electrode to the counter electrode is given by z. In many cases, however, the corresponding dimension are equal to each other, that is, xx , yy , and zz (Figure 4.1(b)). This means only three free parameters, which again simplifies comparisons. The fact that some cells have no cubic but cylindrical, spherical, or hemispherical shapes will not be considered, as the difference is small (less than a factor of 2). All dimensions not specified are assumed to be equal to 1 cm.

(b)

(a)

CE z

B z⬘

CE WE

y x

WE y⬘ A

x⬘

Figure 4.1 Typical geometric parameter of electrochemical microcells (a) and simplified geometry (b) with different possible positions of the reference electrode (A and B).

Microcells 105 Table 4.1 The complete range of typical cell parameter of microcells from large thin film cells to a fictive ultimate microcell Type

Size

Thin film cell

x 1 cm y 1 cm z 100 m

Ultimate microcell

x 1 nm y 1 nm z 1 nm

A (cm2) 1

1014

V (l)

R ()

105

0.1

1024

108

r (cm2) 0.1

106

imin (A cm2) 1013

10

Note: The dimensions x, y, and z according to Figure 4.1(b), the working electrode area A, the cell volume V, the absolute cell resistance R (between working and counter electrode) for a specific conductivity of 0.1 1 cm1, this resistance normalized to the surface area A, and the detectable minimum of current density at the working electrode (assuming a resolution of 1013 A) are given.

Best results are obtained in a 3-electrode arrangement. Therefore, this work will focus on this arrangement, but in many cases it can be easily developed from 2-electrode systems, and so these systems will be also reviewed briefly. There are two common positions for the (Luggin capillary of the) reference electrode, either between working and counter electrode (B in Figure 4.1) or behind an opening in the working electrode (A). A microcell will be defined as an electrochemical cell, if at least one of the parameter x, y, or z is smaller than 1 mm. This determines the range of typical parameters of microcells. An overview is given in Table 4.1 according to the scheme in Figure 4.1(b). The cell volume V varies from 104 l (the largest microcell) to perhaps 1024 l (the ultimate microcell, the electrolyte would just contain one anion and cation), the cell resistance R from 1 to 108 , if a specific conductivity of 0.1 1 cm1 is assumed which is typical for 1 M solutions. The normalized resistance rR/A is helpful for the investigation of kinetics because Uri is the potential drop in the cell for a given current density i at the working electrode. In potentiostatic experiments U is the potential error due to current flow that must be compensated if it is not negligible. The absolute current I at the working electrode decreases with decreasing electrode area for a constant current density i. Assuming a current resolution of 1013 Acm2 which is typical for modern electrochemical equipment we can calculate the minimum current density imin which is detectable. These values are also given.

4.2

Classification

Miniaturization of cell components is an important trend in electrochemistry.1–3 A small cell volume has several attractive aspects: 1

The wetted area of the sample is small and the microcell enables spatially resolved investigations.

106 2 3

4

5

M. M. Lohrengel and A. Moehring The cell resistance and, thus, the potential drop and the evolution of heat are small (Microreactors, Section 4.7.2). The ratio of vessel surface to electrode volume is large, that is, a good heat emission and an intensive relaxation of intermediates, for example, radicals. The accumulation of critical amounts of products or educts can be avoided (Microreactors, Section 4.7.2). A small cell volume is important for the processing of rare and expensive materials (e.g. enzymes) or for analytical purposes, if only small electrolyte samples are available. Processes with common current densities cause detectable changes of the product/educt concentration in the electrolyte (Thin film cells, Section 4.3).

Miniaturization of electrochemical cells can start with single component, for example, the working electrode. In this case we get microelectrodes of different shapes: Microelectrodes,2,4 usually of circular shape,5–7 sometimes shielded to reduce noise or parasitic capacities,8 but also of different geometry: bands,9–11 cylinders,10 tubes,10,12 rings,6,13 recessed disks,7 hemispheres,6 dual disks,14 arrays,15 or interdigitated arrays.16 These microelectrodes are usually not used in microcells, which mean small, well-defined volumes. An actually very important field is the spatially resolved surface analysis.17 A sufficient resolution is usually realized by micro components, usually scanning probes.18 Microelectrodes are also used in scanning devices which enable a local analysis. Scanning devices include various types of probes: ● ● ● ● ● ● ●

scanning reference electrodes;19 scanning microelectrodes;20–22 especially the Scanning Tunnel Microscope (STM;18, 23, 26); scanning Kelvin probes;24 ion sensitive capillary electrodes;25 the Atomic Force Microscope (AFM26); focused radiation (laser beams,27 microellipsommetry28).

All these probes are commonly used together with macrocells and, consequently, should not be considered further in this work. The volume of a real microcell can be defined by different methods: 1

2

By phase boundaries, for example, walls of solid material (usually insulating materials like glass or plastic, but also metal; this can be true for solid cells like movable masks (Section 4.3) or capillary cells (Section 4.5.1). By the interface electrolyte/gas. In some droplet cells (Section 4.5) the volume is defined by the surface tension of the electrolyte and the contact angles.

Microcells 107 3

By the interface aqueous electrolyte/oil. This is true for ion sensitive membrane electrodes or to avoid evaporation of the solvent (Section 4.5.3).

But in some cases microcells are formed within macrocells by weak boundaries: 1 2

By hindrance of diffusion, for example, in capillary based droplet cells or in the electrochemical microscope (Section 4.6). By the electric field, for example, with a small counter electrode in a weakly conductive electrolyte.

In some cases microcells are formed as subsets of macrocells, that is, the electrolyte volume corresponds to a macrocell, but due to some special structures within the system small compartments are formed which have properties of microcells. In the following these cells are called quasi-microcells. One example is the surface of porous electrodes. If the potential within one pore is measured by microelectrodes we get a similar geometry as in the case of capillary based cells (Section 4.5.1). The separation of the microcell electrolyte from the bulk electrolyte is usually done by a hindrance of diffusion (e.g. Electrochemical microscope, Section 4.6). From this definition flow-through microcells are quasimicrocells if the inlet and outlet are continuously open and real microcells if inlet and outlet are closed by valves. Microcells are mainly used in four applications: 1

2 3

4 5

4.3

Electrochemical microsensors measure the concentration of special substances, typically in macroscopic volumes. The microcell is separated from this larger volume by a selective barrier, often a membrane. Microbatteries are simply small devices for energy storage. Analytical microcells are used to determine concentrations within the cell, usually sometimes by electrochemical29 or spectroscopic methods. Flowthrough microcells are used for pre-concentration of substances (stripping techniques) or to get a fast response if the concentration changes with time. Non-flow microcells allow an analysis of small electrolyte samples. Microreactors for electrochemical production. Microcells for surface analysis which are reviewed, in detail, in Section 4.5.

Thin film cells

Thin film cells are used since 1960 and represent the first micro technique which became popular in electrochemistry. They were derived from macrocells by reduction of the distance working to counter electrode (z in Figure 4.1) to some m. Thereby, the cell volume was reduced to values where Faradaic processes at the working electrode (e.g. film formation in the monolayer region) caused notable concentration changes which could be detected.30,31 Thin film cells are usually not used for surface analysis.

108

M. M. Lohrengel and A. Moehring

In spectroscopic investigations of surface layers the absorption of light in the electrolyte becomes a problem, especially in the IR-range. In thin film cells absorption is minimized as the ratio electrolyte/layer is small. Some of the probe techniques presented in Section 4.2 form thin film cells if the tip–sample distance is smaller than the tip radius (typically some 10 nm or larger, for example, STM32).

4.4

Mobile and immobile mask techniques

The insulation of parts of an electrode with paints, resins, or passivation films has a long tradition (e.g. Lajain33). The insulation of cutting edges or backsides or the embedding of wires to get well-defined surfaces, especially microelectrodes, are examples for immobile masks fixed to the sample surface. An attractive method for micro structuring of such masks is the modification of photo resins with a focused laser beam.34 Disadvantageous is the developing process of the resin which can modify the surface of interest. These principles are also used in silicon chip technology in electronics. Several recent electrochemical developments are based on silicone technology (Cellon-a-Chip), for example, microcells,35 microelectrode arrays,36 or combinations of flow-through concepts with arrays.37 Fundamentally all these mask technique produce microelectrodes or microelectrode arrays rather than microcells. Microcells are formed if the mask remains on the sample surface and if its thickness is equal or larger than the diameter of the openings. Then the transport of ions can be limited by diffusion and we get quasi-microcells (Section 4.2). Another method of addressing small parts of an electrode is the use of mobile masks. The typical problem is seepage, the mobile masks show typically weak adhesion only and thus prevent a precisely limited access to the surface. Mobile masks with repeated patterns (often honeycomb style) can be easily produced by LIGA-technology in metal or plastics38 and allow the parallel processing of hundreds or thousands of spots. For some applications this can be more important than perfect boundaries.39 Similar cell arrays are produced during passivation in corrosive media of some metals (e.g. Al) where a self-organizing pore formation is observed. An electrochemical deposition of metals or polymers in these pores shows a different behavior of different pores,40 indicating that the pores behave more or less like independent cells. Special cases of mobile masks are capillaries with silicone rubber gaskets (Section 4.5.1).

4.5

Droplet cells

Pipetting tiny droplets on non-wetting surfaces is a convenient and simple method to realize microcells (e.g. Schreiber et al.41). A similar technique is used to perform a local analysis or preparation of structured surfaces: an electrolyte droplet is positioned at the site of interest and forms the cell.11,42

Microcells 109 Advantages of this arrangement are given by the simple setup, the free choice of the position, and the low mechanical stress of the base material. This is important for preparation and experiments of ultrathin film sandwiches (nm scale), for example, tunnel junctions.43 Furthermore this setup enables an illumination and/or optical inspection of the surface. Special class of these droplet cells are capillary-based microcells which are described in the next section. 4.5.1

Capillary-based microcells

In the last decade special variants of droplet cells were introduced: capillarybased cells. They enable a spatially resolved analysis of conducting surfaces and the investigation of small inhomogeneities. Common aspect of these techniques is a capillary filled with electrolyte. It is positioned (a) closely above (usually some 10 m) the surface, the droplet is simply held by its surface tension (free droplets), and the spot diameter is almost equal to the outer diameter of the capillary (Figure 4.2) or (b) if fitted with a gasket ring at the mouth, in contact with the investigated surface. The spot diameter is given by the inner diameter of the gasket. The wetted area defines the working electrode. Reference and counter electrodes are combined with the capillary. As these cells are based on the conventional 3-electrode arrangement, all the fundamental electrochemical methods (galvanostatic, potentiostatic, or potentiodynamic techniques, pulse transients, impedance spectroscopy) become possible. These capillary-based techniques enable straightforward experiments because no special time-consuming pretreatment of the sample surface is required. Several variants were developed. The free droplet technique44 is usually based on conducting capillaries (metal covered glass or plastic capillaries, metal capillaries45) or insulating capillaries, for example, glass.46 The capillary acts as xyz-stage

Video microscope

Reference electrode Counter electrode Capillary

Droplet

Cold light lamp

Working electrode

Figure 4.2 Fundamental setup of a scanning microcell (shown with a free droplet): capillary, xyz-stage, video microscope, and illumination.

110

M. M. Lohrengel and A. Moehring

counter electrode and as Luggin capillary of the reference electrode. In a typical arrangement the gold capillary has an outer diameter of 150 m and a channel diameter of 50 m. The size of the droplet can be varied by the electrolyte pressure, best results are obtained if the diameter of the droplet equals the outer capillary diameter at a distance capillary/specimen of 10–50 m. This arrangement is applicable to various technical surfaces provided that they are conducting and that the contact angle is in the range of 90º, which is true for many electropolished metals. In fact, the contact angle is less important since the droplet size is mainly determined by the capillary diameter. Very rough or hydrophilic surfaces, however, cannot be investigated by this technique. The stability of free droplets depends on the evaporation of the solvent (usually water). Electrolyte droplets of hygroscopic solutions like sulfuric or acetic acid remain stable for some minutes, other droplet can vanish within some 10 s. This is prevented by a slow stream of solvent saturated nitrogen which is blown across the droplet. This method guarantees stable droplets for several 1000 s. Independently, a Swiss group47 introduced a setup with a silicone rubber gasket at the mouth of glass capillaries. The gasket is formed by dipping the capillary into liquid silicone. The capillary is pressed onto the investigated surface and the gasket is slightly compressed. A well reproducible wetted area results without evaporation of solvent. The hydrophobic silicone ring forms a perfect border even on rough or hydrophilic surfaces (e.g. etched samples, catalysts etc.). The local pressure of the silicone ring, however, is notable. A modification of thin film devices, for example, the partial oxidation of 20 nm Al films on glass resulted in flaking-off due to the mechanical stress.48 Such fragile layers should be investigated or modified with the free droplet. If larger numbers of investigations are intended it is necessary to realize a fast exchange of the capillaries. Glass capillaries tend to break and the silicone ring’s lifetime is limited on hard and rough surfaces to some 100 experiments. Therefore, special carriers made of plastics were developed. They hold the capillary by one or two O-rings which are compressed by a female screw. Furthermore, the carrier contains a micro reference electrode, counter electrode, electric connectors to the potentiostat, and fittings for the (stepper motor operated) syringes for electrolyte supply and droplet control. The capillary carrier is combined with an xyz positioning system and an optical control. Two variants can be used: mobile carrier or mobile sample. A mobile carrier is suitable if large technical parts are investigated. The carrier with the capillary is fixed to the mobile stage of an xyz positioning system used for computer aided machining. These devices enable large working regions of more than 20 cm in every direction. The stage is moved by stepper motors with a typical resolution of some m. The stage is strong enough to carry some kg, therefore it is possible to mount also a video microscope and cold light illumination. With small samples it is more convenient to make the sample mobile. Small xyz systems enable resolutions up to some nm. The carrier and the video setup remain at fixed positions.

Microcells 111 The video microscope is used to control the droplet size and to identify structures on the investigated surface. An electrochemical surface image can be obtained in different manners. Some techniques are limited to free droplets. The droplet can be shifted across the surface (scanning droplet, this is only possible with free droplets) during a continuous measurement, for example, of the electrode capacity. This method is fast, but only applicable to detect large variations of surface impedance. The moving droplet tends to oscillate slightly, thus yielding fluctuations of the wetted area. Better results are obtained if a stop-and-go technique is used with a fixed position during measurements. This procedure, however, enables an accumulation of corrosion products and contamination in the tiny droplet. Suitable for all capillary techniques is the hopping droplet. The droplet is positioned, measurements are performed, then the electrolyte droplet is sucked back or simply left at this position, the capillary may be rinsed in a special reservoir, and a new position is chosen. For most measurements it is necessary to determine the spot size. With a free droplet the area may differ up to 30% according to local variations of the wettability of the surface. With silicone gaskets this reduces to about 5%. A sufficient value can be calculated from the inner diameter of the gasket ring or, in the case of free droplets, from the droplet diameter, measured, for example, with a video microscope. On some technical surfaces, however, the wettability varies locally and, hence, the wetted area depends in form and size on the position. This requires an optical labeling: after the local analysis the droplet size is marked by strong anodization or cathodization which means optical changes of the surface, for example, local roughening or colored surface film formation (Figure 4.3).

Grain 1

6

7

Grain 2

8

9

10

Figure 4.3 Experiments with the scanning droplet cell; microscopic view of a sequence of experiments on two different grains on a polycrystalline Zr sheet. The spots (mean diameter 150 m) are marked by anodic formation of about 15 nm oxide which is visible due to interference colors.

112

M. M. Lohrengel and A. Moehring

Table 4.2 Typical cell parameter of capillary based droplet cells R () r (cm2)

imin (A cm2)

Design

Size

A (cm2) V (l)

Capillary with silicone rubber gasket

x y 10 m z1 cm

106

109

107

10

107

Free droplet, insulated capillary

xy 100 m z 1 cm

104

107

105

10

109

Free droplet, metal capillary

x y 100 m z 10 m

104

1010

102

109

100

Note: The working electrode area A, the cell volume V, the absolute cell resistance R (between working and counter electrode) for a specific conductivity of 0.1 1 cm1, this resistance r normalized to the surface area A, and the detectable minimum of current density at the working electrode (assuming a resolution of 1013 A) are given.

The area is determined by computer image processing of digitized photos or video images. The droplet cell is a micro technique, but one of the typical advantages of microelectrodes, a hemispherical electrolyte geometry close to the electrode and, therefore, a very small electrolyte resistance, is not valid for the droplet (Table 4.2). Furthermore, the surface area of the counter electrode is typically similar or equal to that of the working electrode. Nevertheless, the droplet cell is suitable for fast transient experiments at large current densities or high frequencies: The absolute electrolyte resistance between Luggin-capillary and working electrode is normally in the range of some 100 which corresponds to a value of, for example, 0.05 cm2 referred to the working electrode area. This value is significantly better than that of thin wire working electrodes in macrocells (typically 1 cm2). Typical applications of the scanning droplet cell are presented in Figure 4.4. An electrochemical surface analysis of almost randomly shaped objects like sheets, rods, tubes, die cast housings becomes possible. Bulk material is analyzed at the bottom of blind holes. Tiny structures on technical surfaces like solder pads on printed circuit boards, single electrodes in micro electrode arrays, or inhomogeneities in technical surfaces (corrosion pits,47 passivation kinetics of single grains in polycrystalline material49). Single crystals, for example, silicon wafers can be investigated without embedding of the sample. The same is true for coupons from coated sheets, for example, catalyst surfaces. No insulation of cutting edges or the backside is necessary and hundreds of experiments become possible with one coupon. Another interesting aspect is an electrochemical investigation of single microparticles. A small number of particles is suspended, brought to a metal sheet by sedimentation or by a centrifuge, fixed and connected by galvanizing few m of the sheet metal.50 Then an addressing of single particles becomes possible with the droplet cell.51 The alternative method is contacting with a metal filament or carbon fiber.21,52

Microcells 113

Sheets, rods

Metal die cast objects

Welding seam

Blind holes in metal parts

Solder and contact pads (PCB)

Single crystals

Single grains

Micro electrodes Micro electrode arrays

Tunnel contacts Sandwich structures

Modification and structuring

Figure 4.4 Typical applications of the scanning droplet cell.

The droplet cells were not only used for a surface analysis but also for structuring of surfaces. With the droplet local modifications like passivation, metal or polymer deposition, etching, etc. become possible. In some special cases the investigation of well-defined surface films (e.g. oxides on Al or Ta) enables a significantly higher resolution in lateral and normal direction. This was presented in detail in a recent paper.53 4.5.2

Optical flow cell

Flow-through concepts of microcells are common in sensor or analytical applications.54 A different class of microcells was developed for optical purposes, for example, Raman spectroscopy55 or electroluminescence.56 An effort to develop an all-purpose microcell to combine a flow-through concept with optical and electrochemical investigations was done by Vogel et al.57 A relative complex design was obtained, which combined a gasket technique similar to that in Section 4.5.1 with a flow cell and an optical window for microscopic control and/or illumination of the investigated surface (Figure 4.5). The cell was realized as a sandwich

114

M. M. Lohrengel and A. Moehring

Electrolyte inlet Acryl glass cube Electric connector

Optical window Electrolyte outlet (fittings removed)

Silicone gasket (bottom)

Foturan panes

1 cm

Figure 4.5 View of the optical flow microcell. The silicone rubber gasket and the electrolyte window which connects the cell to the sample surface are positioned at the bottom.

Position market Openings for CE and RE

Cell chamber ( 300 µm) Electrolyte inlet Electrolyte channel Outlet

1 cm Window to sample surface

Figure 4.6 Stack of structured Foturan glass panes which form the optical flow microcell.

Microcells 115 of a Plexiglass block and structured glass panes. These glass sheets (thickness 100 m) were made of Foturan (Schott, Mainz), a special glass which can be microstructured by UV-light and subsequent etching (Figure 4.6). 4.5.3

Electrolyte droplets in oil

The problem of evaporation of solvent of tiny electrolyte droplets can also be solved by embedding the droplet in insulating oil, for example, paraffin oil.58 Usually the electrolyte droplet is positioned by a capillary at a chlorinated silver wire in an oil bath. Then the capillary is removed. The wire acts as a reference electrode or counter electrode as well. Further microelectrodes are inserted. This type of setup is mainly used in biology to test multibarreled capillary electrodes for the investigation of membrane potentials in living cells.

4.6

Scanning electrochemical microscope

The scanning electrochemical microscope (SECM) consists of a microelectrode positioned by an xyz-stage above the sample. The distance is typically some m. The diameter of the insulating material surrounding the microelectrode is larger than this distance (typically 5–10 times). Therefore, a flat cylindrical quasimicrocell is formed because a mass exchange by diffusion between the microelectrode and the bulk of electrolyte is hindered due to this special geometry. The SECM is used in 2-electrode arrangements or in potentiostatic 3- or 4-electrode configurations. Three classes of experiments were carried out: in the scanning mode an electrochemical image or local modification is produced, in the distance mode the tip response is recorded as a function of the tip–sample distance, and in the redox mode the microelectrode is positioned as close as possible and the electrochemical interaction between tip and substrate is monitored. With nm-sized tips the detection of single molecules becomes possible. If only one or two molecules move to the sample and back to the tip, discrete current level of about 1.5 pA are detected which means an oscillation of the molecule with a rate of 107 s1.59 This technique was introduced by Bard et al.60 and is presented in detail in reviews and many special papers.61

4.7

Other classes of microcells

In the following sections some microcells are briefly described which are not used in surface analysis. Furthermore these devises usually use a 2-electrode arrangement. 4.7.1

Microsensors

Usually the space where sensors should be positioned is limited. This is true, for example, for sensors which shall be implanted in biological material, for example,

116

M. M. Lohrengel and A. Moehring

blood sensors in the human body. Such applications of microsensors62 require in addition the miniaturization of the measurement equipment and the power sources. This is also true for technical sensors which measure environmental parameter. They should be small to be easily handled and exchanged. Furthermore the sensitivity against shock and other mechanical stress decreases with decreasing size. Finally, small sensors are suitable for mass production and, hence, cheaper. In macrosystems the material costs are an important part of the price, in microsystems usually only the production costs must be regarded. The use of very expensive materials becomes possible because the amounts in one microsensor are negligible. Biosensors63 are usually based on enzymes and ion sensitive electrodes, recent development couple silicon-based amplifiers to insect antennas or even to living insects.64 4.7.2

Microreactors

Microreactors65 are usually based on 2-electrode arrangements. They are typically used for three reasons: ● ● ●

the amount of reacting species is very small (expensive or rare educts); intermediates of the process are extremely dangerous (poisonous or explosive); dissipation of the reaction heat is a problem.

For more details see Lowe et al.66 4.7.3

Microbatteries

Batteries neither use a 3-electrode arrangement nor allow surface analysis, therefore, they are not respected here. Nevertheless, a short overview of typical parameters shall be given. Table 4.3 compares different sizes of microbatteries. For comparison a charge density of 100 W h kg1 and a density of 2 g cm3 was assumed. The corresponding lifetimes were calculated for typical standby values of an electronic device, for example, 0.1 A at 2 V. The source resistance R was based on an electrolyte conductivity of 1 1 cm1 and a geometry as in Section 4.1. Table 4.3 Typical parameter of microbatteries Type

Size

V (l)

Lifetime

Thin film battery

x y 1 cm z 100 m

105

104 h

Microbattery

x y z 10 m

1012

Microbattery

x y z 1 m

15

10

R () 0.01

3.6 s

103

3.6 ms

104

Note: Volume V, lifetime for a typical standby current and source resistance R. For more parameters see text.

Microcells 117

4.8

Borderline cases of microcells

In this section some very special electrochemical microsystems are presented. The (presumably) smallest microcell in use is formed under the tip of an STM. Hydrogen-terminated silicon can be anodized in normal atmosphere.67 Due to the presence of adsorbed water on the silicon surface an electrochemical cell between tip and sample is formed.68 SiO2-traces with a width of 20 nm can be written. The width decreases with decreasing humidity and, thus, proves an electrochemical process. Assuming a mean distance between tip and sample of some nm and a diameter of the electrolyte spot of 50 nm we get a cell volume of about 1020 l. This is a cell in a 2-electrode configuration. In biology the membrane potentials of isolated living cells (often neurons) are measured. These cells have typical diameters of some 10 m and volumes in the range of nl. The cell membrane is a self-assembling bilayer of lipids which forms an almost perfect barrier against electron exchange (negligible tunnel probability at a distance of 6–12 nm) or ion transfer (hydrophobic inner part). Different concentrations of special ions (e.g. Na and K) cause potential differences of 40–100 mV across the membrane. In a typical setup the potential of the intercellular electrolyte is held at a constant potential by a reference electrode connected to ground potential. An access to the potential inside the cell is possible by glass capillaries with tip diameters down to less than 1 m.69 Concentrations of the cell interior are measured with ion sensitive membrane electrodes (see below). Several capillaries are usually combined in one multibarrel capillary.70 Such multielectrode assemblies allow potential control in 3-electrode arrangements (two capillaries connect counter and reference electrodes, the cell membrane is the working electrode) or even pH-control by migration of H or OH.44 Ion sensitive membrane electrodes25,71 are capillaries with a tiny droplet of a special oil-like nonaqueous electrolyte in the mouth. Within this viscous ionophor (the “membrane”) only one special ion is transported. The potential difference between the inner part of the capillary (which is filled with a solution of known concentration of the ion) and the capillary’s environment is a measure of the concentration in the environment. The membrane with a typical diameter of some m resembles a 2-electrode microcell with an electrolyte volume of about 1013 l and the two interfaces membrane/electrolyte (“electrodes”). These ion sensitive electrodes are common in biology and are now used to investigate technical microsystems, for example, the phosphating kinetics of steel surfaces.72 Special microcells for an investigation of the liquid/liquid interface of immiscible electrolytes were designed.73

4.9

Conclusions

Microcells cover a wide range of cell volumes (from 105 to 1019 l). They are used wherever miniaturized electrochemistry is demanded: batteries, reactors, sensors, bulk, and surface analysis. A very important field is the spatially resolved surface analysis, usually in a 3-electrode configuration. This is the

118

M. M. Lohrengel and A. Moehring

special domain of droplet cells. This technique has a smaller resolution (10 m) than some of the probe techniques (e.g. STM) but enables the complete range of potentiostatic and galvanostatic techniques including impedance spectroscopy. Future developments of the droplet cells will include: ●

●

● ●

enhanced lateral resolution (this is limited by the detection of absolute currents, which is in the moment about 10 fA for dc signals in electrochemical environments); computer controlled automatic measurements (this becomes possible with a force sensor indicating the mechanical contact between capillary and sample); coupling of light to the capillary to enable photoelectrochemical measurements; realization of a flow-through concept to avoid enrichment or depletion of electrolyte components or contaminations.

Acknowledgment The financial support of the Ministerium für Schule, Wissenschaft und Forschung des Landes Nordrhein-Westfalen is gratefully acknowledged.

References 1. B. R. Scharifker, in J. O’M. Bockris, B. E. Conway, and R. E. White (eds) Modern Aspects of Electrochemistry No. 22, Plenum Press, New York 1992, S467–519. 2. J. Heinze, Angew. Chem. 105 (1993) 1327. 3. J. W. Schultze and V. Tsakova, Electrochim. Acta 44 (1999) 3605. 4. J. Heinze, Angew. Chem. 103 (1991) 175; R. J. Forster, Chem. Soc. Rev. 23 (1994) 289; K. Aoki, Electroanalysis 5 (1993) 627. 5. J. Heinze, J. Electroanal. Chem. 124 (1981) 73. 6. C. G. Zoski, J. Electroanal. Chem. 296 (1990) 317. 7. A. M. Bond, D. Luscombe, K. B. Oldham, and C. G. Zoski, J. Electroanal. Chem. 249 (1988) 1. 8. D. O. Wipf, A. C. Michael, and R. M. Wightman, J. Electroanal. Chem. 269 (1989) 15; S. Nomura, K. Nozaki, and S. Okazaki, Anal. Chem. 63 (1991) 2465; P. Tschuncky and J. Heinze, Anal. Chem. 67 (1995) 4020. 9. H. Gerischer, I. Mattes, and R. Braun, J. Electroanal. Chem. 10 (1965) 553. 10. P. M. Kovach, W. L. Caudill, D. G. Peters, and R. M. Wightman, J. Electroanal. Chem. 185 (1985) 285. 11. W. J. Bowyer, M. E. Clark, and J. L. Ingram, Anal. Chem. 64 (1992) 459; D. M. Odell and W. J. Bowyer, Anal. Chem. 62 (1990) 1619; T. Matsue, A. Aoki, E. Ando, and I. Uchida, Anal. Chem. 62 (1990) 409; A. M. Bond, T. L. E. Henderson, and W. Thormann, J. Phys. Chem. 90 (1986) 2911; K. R. Wehmeyer, M. R. Deakin, and R. M. Wightman, Anal. Chem. 57 (1985) 1913. 12. W. J. Blaedel, C. L. Olsen, and L. R. Sharma, Anal. Chem. 35 (1963) 2100. 13. D. R. MacFarlane and D. K. Y. Wong, J. Electroanal. Chem. 185 (1985) 197. 14. F.-M. Matysik, Electrochim. Acta 42 (1997) 3113. 15. M. Wittkampf, K. Cammann, M. Amrein, and R. Reichelt, Sensors and Actuators B40 (1997) 79.

Microcells 119 16. O. Niwa, M. Morita, and H. Tabei, Anal. Chem. 62 (1990) 447; M. Morita, O. Niwa, and T. Horiuchi, Electrochim. Acta 42 (1997) 3177; C. E. Chidsey, B. J. Feldman, C. Lundgren, and R. W. Murray, Anal. Chem. 58 (1986) 601. 17. J. W. Schultze, Mater. Sci. Forum 185–188 (1995) 377. 18. A. A. Gewirth and H. Siegenthaler (eds) Nanoscale Probes of the Solid/Liquid Interface, NATO ASI Series E288, Kluwer, Dordrecht 1995. 19. H. S. Isaacs and B. Vyas in F. Mansfeld, and U. Bertocci (eds) Electrochemical Corrosion Testing ASTM STP 727 (1981) 3; H. S. Isaacs and G. Kissel, J. Electrochem. Soc. 119 (1972) 1628. 20. M. Fleischmann, S. Pons, D. Rolison, and P. Schmidt, Ultramicroelectrodes, Datatech Systems Inc., Morganton 1987; R. M. Wightman, Anal. Chem. 53 (1981) 1125. 21. M. Nishizawa and I. Uchida, Electrochim. Acta 44 (1999) 3629. 22. M. Küpper and J. W. Schultze, Electrochim. Acta 42 (1997) 3023, 3085; E. Klusmann and J. W. Schultze, Electrochim. Acta 42 (1997) 3123; A. W. Hassel, K. Fushimi, and M. Seo, Electrochem. Comm. 1 (1999) 180. 23. G. Binnig and H. Rohrer, Helv. Phys. Acta 55 (1982) 726; D. A. Bonell (ed.), Scanning Tunnel Microscopy and Spectroscopy, VCH Verlag, Weinheim 1993; C. Kobusch and J. W. Schultze, Electrochim. Acta 40 (1995) 1395. 24. M. Stratmann and H. Streckel, Corros. Sci. 30 (1990) 679; M. Stratmann, K. T. Kim, and H. Streckel, Z. Metallkunde 81 (1990) 715; K. Saurbier, J. W. Schultze, and J. Geke, Electrochim. Acta 39 (1994) 1171; A. Thies and J. W. Schultze, Mater. Corros. 47 (1996) 146; O. A. Semenikhin, L. Jiang, T. Iyoda, K. Hashimoto, and A. Fujishima, Electrochim. Acta 42 (1997) 3321. 25. M. Küpper and J. W. Schultze, J. Electroanal. Chem. 427 (1997) 129. 26. R. Wiesendanger and H.-J. Güntherodt (eds), Scanning Tunneling Microscopy, Springer Verlag, Berlin 1992 (3 Volumes). 27. J. W. Schultze, K. Bade, and A. Michaelis, Ber. Bunsenges. Phys. Chem. 95 (1991) 1349; K. Bade, O. Karstens, A. Michaelis, and J. W. Schultze, Faraday Discuss. 94 (1992) 45; T. Morgenstern, A. Thies, and J. W. Schultze, Sensors and Actuators A51 (1995) 103. 28. A. Michaelis and J. W. Schultze, Ber. Bunsenges. Phys. Chem. 97 (1993) 431; A. Michaelis and J. W. Schultze, Thin Solid Films 274 (1996) 82; S. Kudelka, A. Michaelis, and J. W. Schultze, Electrochim. Acta 41 (1996) 863; M. Schweinsberg, A. Michaelis, and J. W. Schultze, Electrochim. Acta 42 (1997) 3303. 29. Ya. I. Tur’yan, Talanta 44 (1997) 1. 30. F. E. Woodard and C. N. Reilley in E. Yeager, J. O’M. Bockris, B. E. Conway, and S. Sarangapani (eds) Comprehensive Treatise of Electrochemistry Vol. 9, Plenum Press, New York 1984, S353–392. 31. E. Schmidt and H. R. Gygax, Chimia 16 (1962) 165; E. Schmidt and H. R. Gygax, Helv. Chim. Acta 48 (1965) 1178; C. R. Christensen and F. C. Anson, Anal. Chem. 35 (1963) 205; A. T. Hubbard and F. C. Anson, Anal. Chem. 36 (1964) 723; A. T. Hubbard and F. C. Anson, J. Electroanal. Chem. 9 (1965) 163; E. Schmidt and H. R. Gygax, J. Electroanal. Chem. 12 (1966) 300; T. P. DeAngelis and W. R. Heineman, Anal. Chem. 48 (1976) 2262; T. P. DeAngelis, R. E. Bond, E. E. Brooks, and W. R. Heineman, Anal. Chem. 49 (1977) 1792. 32. D. M. Kolb, Bunsen-Magazin 2 (2000) 4. 33. H. Lajain, Werkst. Korros. 23 (1972) 537. 34. S. Kudelka, A. Michaelis, and J. W. Schultze, Ber. Bunsenges. Phys. Chem. 99 (1995) 1020.

120

M. M. Lohrengel and A. Moehring

35. M. Koudelka and C. G. Francis, Anal. Chim. Acta 219 (1989) 45; M. Morita, M.L. Longmire, and R.W. Murray, Anal. Chem. 60 (1988) 2770. 36. G. Schmitt, J. W. Schultze, F. Faßbender, G. Buß, H. Lüth, and M. J. Schöning, Electrochim. Acta 44 (1999) 3865; G. Buß, M. J. Schöning, H. Lüth, and J. W. Schultze, Electrochim. Acta 44 (1999) 3899. 37. R. Herber, S. Ernst, G. Buss, M. J. Schöning, and H. Baltruschat, The Electrochem. Soc., Proc. 99–5 (1999) 168. 38. O. Karstens, J. W. Schultze, W. Bacher, and R. Ruprecht, Metalloberfläche 48 (1994) 148. 39. L. Dunsch, P. Rapta, A. Neudeck, A. Bartl, R.-M. Peters, D. Reinecke, and I. Apfelstedt, Synth. Met. 85 (1997) 1401; P. Rapta, A. Neudeck, A. Bartl, and L. Dunsch, Electrochim. Acta 44 (1999) 3483; T. Morgenstern and J. W. Schultze, Electrochim. Acta 42 (1997) 3057. 40. D. Ebling and J. W. Schultze, Metalloberfläche 44 (1990) 491. 41. R. Schreiber and W. D. Cooke, Anal. Chem. 27 (1955) 1475; R. T. Iwamoto, R. N. Adams, and H. Lott, Anal. Chim. Acta 20 (1959) 84. 42. K. Bitterling and F. Willig, J. Electroanal. Chem. 204 (1986) 211. 43. S. Schatteburg, D. Diesing, and A. Otto, Appl. Phys. B70 (2000), 573. 44. M. M. Lohrengel, Electrochim. Acta 42 (1997) 3265. 45. H. Forker, Diplomarbeit, Heinrich-Heine-Universität Düsseldorf 1995. 46. A. W. Hassel and M. Seo, Electrochim. Acta 44 (1999) 3769. 47. T. Suter and H. Böhni, Electrochim. Acta 42 (1997) 3275. 48. M. Pilaski, D. Diesing, G. Kritzler, and M. M. Lohrengel, A. Moehring in Full Papers on CD-ROM of EUROCORR 1999, Aachen. 49. A. Moehring and M. M. Lohrengel, GDCh Monogr. 14 (1998) 424; M. M. Lohrengel, A. Moehring, and M. Pilaski, Fresenius J. Anal. Chem. (2000), 334. 50. M. Keddam, S. Senyarich, and H. Takenouti, J. Appl. Electrochem. 24 (1994) 1037. 51. T. Hamelmann and M. M. Lohrengel, Electrochim. Acta 47 (2001) 117. 52. M. Perdicakis, N. Grosselin, and J. Bessière, Electrochim. Acta 42 (1997) 3351. 53. A. W. Hassel and M. M. Lohrengel, Electrochim. Acta 42 (1997) 3327. 54. E. Bychkov, M. Bruns, H. Klewe-Nebenius, G. Pfennig, K. Raptis, W. Hoffmann, and H. J. Ache, Sensors and Actuators B26–27 (1995) 384; E. Bulska, M. Walcerz, W. Jedral, and A. Hulanicki, Anal. Chim. Acta 357 (1997) 133. 55. H.-Y. Chen and Y.-T. Long, Anal. Chim. Acta 382 (1999) 171. 56. Y.-T. Hsueh, S. D. Collins, and R. L. Smith, Sensors and Actuators B49 (1998) 1. 57. A. Vogel and J. W. Schultze, Electrochim. Acta 44 (1999) 3751. 58. R. C. Thomas, J. Physiol. 255 (1976) 715. 59. F.-R. F. Fan, J. Kwak, and A. J. Bard, J. Am. Chem. Soc. 118 (1996) 9669. 60. M. V. Mirkin and B. R. Horrocks in J. Walter Shultze, Tetsuya Osaka, Madhav Datta (eds) Electrochemical Microsystem Technologies, Taylor & Francis, London and New York 2002; A. J. Bard, F. F. Fan, and M. V. Mirkin in A. J. Bard (ed.) Electroanalytical Chemistry 18, p. 243, Marcel Dekker, New York 1994; G. Denuault, M. H. Troise Frank, and S. Nugues in A. A. Gewirth, and H. Siegenthaler (eds) “Nanoscale Probes of the Solid/Liquid Interface”, NATO ASI Series E288, Kluwer, Dordrecht 1995. 61. D. Mandler and A. J. Bard, J. Electrochem. Soc. 136 (1989) 3143; A. J. Bard, F.-R. F. Fan, J. Kwak, and O. Lev, Anal. Chem. 61 (1989) 132; J. Kwak and A. J. Bard, Anal. Chem. 61 (1989) 1221; A. J. Bard, G. Denuault, C. Lee, D. Mandler, and D. O. Wipf, Acc. Chem. Res. 23 (1990) 357; D. Mandler and A. J. Bard, J. Electrochem. Soc. 137 (1990)

Microcells 121

62. 63.

64.

65. 66.

67. 68.

69. 70.

71.

72. 73.

2468; M. Pyo and A. J. Bard, Electrochim. Acta 42 (1997) 3077; K. Borgwarth, C. Ricken, D. G. Ebling, and J. Heinze, Ber. Bunsenges. Phys. Chem. 99 (1995) 1421; C. Heß, K. Borgwarth, C. Ricken, D. G. Ebling, and J. Heinze, Electrochim. Acta 42 (1997) 3065; C. Kranz, G. Wittstock, H. Wohlschläger and W. Schuhmann, Electrochim. Acta 42 (1997) 3105. W. Göpel, J. Hesse, and J. N. Zemel (eds) “Sensors: A Comprehensive Survey”, Vol. 8, Micro- and Nanosensor, VCH Verlag, Weinheim 1995. W. Schuhmann and K. Habemüller, in J. Walter Shultze, Tetsuya Osaka, Madhav Datta (eds) Electrochemical Microsystem Technologies, Taylor & Francis, London and New York 2002; P. Fromherz, in J. Walter Schultze, Tetsuya Osaka, Madhav Datta (eds) Electrochemical Microsystem Technologies, Taylor & Francis, London and New York 2002. M. J. Schöning, in J. Walter Schultze, Tetsuya Osaka, Madhav Datta (eds) Electrochemical Microsystem Technologies, Taylor & Francis, London and New York 2002. H. Löwe and W. Ehrfeld, Electrochim. Acta 44 (1999) 3679; W. Ehrfeld, V. Hessel, H. Möbius, and Th. Richter, Dechema Monogr. 132 (1996). H. Löwe, W. Ehrfeld, and J. Schiewe, in J. Walter Shultze, Tetsuya Osaka, Madhav Datta (eds) Electrochemical Microsystem Technologies, Taylor & Francis, London and New York 2002. J. A. Dagata, J. Schneir, H. H. Harary, C. J. Evans, M. T. Postek, and J. Bennett, J. Appl. Lett. 56 (1990) 2001. H. Sugimura and N. Nakagiri, Jpn. J. Appl. Phys. 34 (1995) 3406; H. Sugimura, N. Kitamura, and N. Nakagiri, Jpn. J. Appl. Phys. 33 (1994) 143; R. Garcia, M. Calleja, and F. Pérez-Murano, Appl. Phys. Lett. 72 (1998) 2295; P. Avouris, T. Hertel, and R. Martel, Appl. Phys. Lett. 71 (1997) 285; A. E. Gordon, R. T. Fayfield, D. D. Litfin, and T. K. Higman, J. Vac. Sci. Technol. B13 (1995) 2805. G. Ling and R. W. Gerard, J. Cell. Comp. Physiol. 34 (1949) 383. P. W. Dierkes, S. Neumann, A. Müller, D. Günzel, and W.-R. Schlue, in J. Walter Shultze, Tetsuya Osaka, Madhav Datta (eds) Electrochemical Microsystem Technologies, Taylor & Francis, London and New York 2002; D. Günzel, A. Müller, S. Durry, and W.-R. Schlue, Electrochim. Acta 44 (1999) 3785, W.-R. Schlue, W. Kilb, and D. Günzel, Electrochim. Acta 42 (1997) 3197; H. Matsuoka, M. Saito, T. Watabe, A. Ohi, H. Sotoyama, and K.-B. Oh, Electrochim. Acta 44 (1999) 3801. D. Ammann, Ionselective Microelectrodes: Principles, Design and Application, Springer Verlag, Berlin 1986; R. C. Thomas, Ion-sensitive Intracellular Microelectrodes, Academic Press, London 1978. J. W. Schultze, N. Müller, Metalloberfläche 53 (1999) 17. Z. Samec, A. Trojánek, and E. Samcová, J. Electroanal. Chem. 386 (1995) 225.

5

Application of optical micro-methods and lasers in electrochemistry A. Michaelis and S. Kudelka

5.1

Introduction

5.1.1

Optical methods in electrochemistry

Conventional electrochemical methods show an excellent vertical resolution for the investigation of thin films on solid surfaces but their application to microsystems is limited because of the poor lateral resolution with typical electrode areas in the mm2-range. In order to gain deeper insight into local electrode reactions, microscopic in situ methods must be developed and employed which feature both high lateral and vertical resolution. We describe an important approach to address and solve this issue, that is, the combination of conventional electrochemistry with optical micro-methods. This enables characterization or/and modification of solid surfaces at high lateral resolution which is only limited by refraction of light to about 1 m. Some fundamental aspects of optical methods employing focused light and their application to electrochemistry are discussed with emphasis on laser light sources. Depending on the input/output signals, the electrochemical micro-methods presented here can be divided into three different classes: 1

2

Optical/optical methods (exciting and measured signal light beam): examples are the application of well established optical methods such as ellipsometry, reflection spectroscopy, beam deflection, interferometry, Raman spectroscopy, etc. along with electrochemistry.1–6 The incident light beam is focused on a microscopic portion (e.g. a single substrate grain) of a macroscopic electrode and the optical information in the reflected or deflected light beam is analyzed and correlated to the simultaneously recorded electrochemical signals. For instance, micro-ellipsometry can be performed during cyclovoltammetric formation of a passive layer to record film thickness growth on a particular grain of a polycrystalline macroscopic electrode. Examples for this important application will be discussed later. In order to allow for an accurate correlation of the signals to the deliberately chosen electrode areas, microscopic control is essential. Figure 5.1 shows a typical experimental setup for this purpose. Optical/electrical or optical/chemical methods (measured signals such as electric current (photocurrent iph), voltage, reaction rates, etc.). Characteristic

Application of optical methods and lasers 123

CCD Monitor

Ar+–laser

Frequency doubler

Semitransparent mirror UV 99%

Telescope +pinhole

UV-objectvie 257 nm CE

RE Electrochemical cell Prism mounting platform

ME

z x, y Stage

Figure 5.1 Typical experimental setup used for the combined application of micro-optical and electrochemical methods along with in situ microscopic sample monitoring. The electrochemical cell is mounted on an x-y-z stage and can be scanned with sub-micron resolution.

3

of this class is that the light beam is used to locally induce a reaction.7–12 The type of reaction depends on thermal and electronic properties of the solid, the neighboring phase, and energy or power density of the light beam. In metals, energy dissipation is very fast. Consequently, most of the energy is transformed into heat resulting in melting or even evaporation of the substrate. In dielectric materials bond breaking with following reactions dominates. In semiconductors dissociation and ionization reactions are often observed if high power densities are used. However, at low power densities electron/hole pair formation with an associated change of the potential distribution at the interface dominates. Both electron transfer (ETR) as well as ion transfer (ITR) reactions can be induced. For instance, hole accumulation can cause redox reactions as well as film growth or removal (corrosion, etching) as will be discussed later. In this paper we will emphasize laser-induced reactions under electrochemical conditions at low power densities, that is, at quasi-isothermal conditions. Electrical/electrical microelectrode methods under optical (microscopic) control. This class covers the conventional electrochemical methods, that is, application of an electrical signal (e.g. potential) to the electrode to induce a reaction. The role of optics here is twofold. 3.1 Samples are prepared by means of photolithography, that is, the local resolution of the electrical measurement is determined by focused laser radiation. In this context it has to be kept in mind that the high lateral

124

A. Michaelis and S. Kudelka resolution achieved in modern device and micro-system technology and research is mainly due to optical processing, that is, photolithography. Detailed examples for application of photoresist microelectrochemistry namely the new photoresist nl-droplet method are given in this overview. 3.2 In order to measure on defined areas the electrode must be microscopically controlled (see Figure 5.1).

Examples for a combined application of these methods to characterize and modify local areas on technically relevant polycrystalline materials such as the valve metal/valve metal oxides-systems are presented. All experiments are performed under microscopic control on well-defined and deliberately chosen single substrate grains. As an optical/optical method micro-ellipsometry will be emphasized since it allows for a complete determination of both the optical parameters of the system and the film thickness d. Moreover, the new method called anisotropy micro-ellipsometry (AME) will be introduced.13–16 AME uses a generalized ellipsometry approach and allows for determination of crystallographic properties such as the orientation of substrate grains which then can be correlated to the film properties. This allows quantifying the effect of the substrate/film interface on film properties. For instance, recrystallization of semiconducting films or epitaxial film growth can be studied. Examples for combination of AME with photoresist microelectrochemistry are presented for the Ti/TiO2 system. The electronic film parameters are correlated to the substrate texture yielding insight into the growth mechanisms of this system. It is important to note that photoresist microelectrochemistry is not limited to this application but can be used to perform all kinds of electrochemical measurements at high local resolution in a droplet of nl-volume, for example, on single substrate grains. Figure 5.2 illustrates the

(a)

(b) Photoresist-microelectrodes

Micro-ellipsometry x,y,z

x,y,z

RE and pipette

CE Laser beam Lens

Resist

Electrolyte Grain a WE

Grain a Laserspot d = 50 µm

Figure 5.2 Electrochemical and optical micro-methods under microscopic control. (a) Schematic depiction of the photoresist nl-droplet method; (b) optical micromethods can be applied on the same portion of the heterogeneous sample. Measurements on a single grain of technically relevant polycrystalline samples can be performed.

Application of optical methods and lasers 125 combined application of ellipsometry and the nl-photoresist electrochemistry on technically relevant polycrystalline electrode materials. As examples for application of optical/electrical micro-methods, photocurrent spectroscopy and UV-laser scanning measurements are chosen. Fundamental aspects, which are important for application of photoelectrochemistry to thin films, are discussed. A new model involving both an exact optical and electrical treatment of the sample system is presented. Consequently, both electronic and optical film properties are available allowing for mutual confirmation and consistence checks of the pure optical and electrical methods. 5.1.2

Application of lasers in electrochemistry

Since laser radiation plays an important role for all presented applications some fundamental aspects of lasers are briefly discussed here. The fact that lasers are more and more replacing conventional light sources in all kinds of optical applications can be understood by considering their fundamental advantages. Perhaps the most striking feature of laser radiation is its directionality due to a very low angle of divergence typically in the order of 1 rad. This has basically two important implications: 1 2

Laser radiation can be focused down to dimensions of the wavelength of light. A large amount of energy can be concentrated in small spots yielding high power densities.

As a matter of fact the focusability of laser radiation is the most applied feature. Optimum focusing of laser radiation demands high mode quality of the laser beam. Best suited is the TEM00-mode, which shows a Gaussian spatial intensity distribution. This fundamental mode can always be obtained by using intra-cavity apertures, which are small enough to suppress the higher modes. For lasers operated in the TEM00-mode the minimum obtainable spot diameter s (diffraction limit) is given by Airy’s formula: s␭

f D␲

(1)

with f, focal length of the used focusing system; ␭, wavelength of laser; D, maximum effective beam diameter (aperture) of focusing system (beam expanders are of advantage). However, it is not only the spot size s that governs the final spatial resolution of the electrode reaction but illumination of the sample is just the first step in a whole series of induced processes resulting in the final state of the system. For instance, in semiconductor interfaces the lateral hole drift extends the reaction zone. In case of thermally activated processes, the heat conductivity of the sample determines the reaction zone. Therefore, both optical and physical effects determine the final spatial resolution of optical/electrical laser methods.

126

A. Michaelis and S. Kudelka

Other important properties of laser radiation are its narrow spectral bandwidth, coherence and polarization. These features are used for numerous spectroscopic and interferometric applications such as Raman, polarization, interference spectroscopy, spectroscopic ellipsometry, holography, etc. Moreover, the possibility of creating very short light pulses by Q-switch or mode-locking need to be mentioned. Pulses below fs duration are possible and allow elucidating fast transient phenomena such as short spontaneous lifetimes or fast relaxation processes. Laser illumination results in thermal heating, chemical reactions or/and potential changes. This report emphasizes laser-induced reactions under electrochemical conditions at low power densities, that is, at quasi-isothermal conditions. Thermal processes are briefly discussed below. 5.1.2.1

Thermal effects

With the assumption of a Gaussian temperature distribution (TEM00-mode) on the laser affected zone the temperature increase T can be estimated by the following equation.17 T

8␬t 冪␳cr

A · P · r · arctan 2␬兹␲/2

2

(2)

with A, absorptance; ␳, density; c, specific heat; ␬, heat conductivity; r, spot radius s/2; P, power density. For times t r2␳␹/(8␬(t102s)) the temperature becomes time independent (continuous wave (cw)-condition). Then the local temperature is given by T A Pr 2␬

冪␲2

(3)

The temperature effect often disturbs precise measurements if isothermal conditions cannot be maintained and if it leads to a damage of the sample. Equation (2) shows that at constant power density the temperature effect decreases with decreasing pulse times. Therefore, the application of short pulses may be of advantage to avoid damages. If, however, the modification of the surface needs a large total energy amount, it should be delivered with low power density. On the other hand, there are numerous applications of thermal heating: it can be used to evaporate or to dissociate the substrate (LAMMA),18 to enhance reaction rates at the surface or the convection of the electrolyte.19–21 Finally, it can be employed to study electrode reaction rate constants as well as the dynamics of the double layer.22

5.2

Fundamentals of optical/electrical methods

The theoretical formalisms for the interpretation of the pure optical (optical/ optical) and the pure electrochemical methods are well established and are to be found in numerous publications. It is far beyond the scope of this chapter to review

Application of optical methods and lasers 127 the corresponding models. However, the quantitative interpretation of optical/ electrical measurements (photoelectrochemistry) is still incomplete and shall be treated here. For instance, multiple internal reflections within the thin films must be considered, causing iph to depend on the layer thickness.23–25 For this, an exact optical model has to be applied26–28 and combined with the Butler–Gärtner model29,30 which itself holds only for crystalline bulk materials. The extended model describes iph as a function of the layer thickness df, the optical constants of substrate and layer, the extension of the space charge layer dscl, and such electronic properties as the defect state concentration ND and the permittivity number ␧(0). However this model is still not accurate for an exact description of the iph behavior of the amorphous TiO2 layers, but another, E-field depending term, taking the recombination efficiency r(E) into account, must be coupled to the model. This yields a further dependence of iph on the film thickness and on the applied anodic potential. This complete photocurrent model consisting of three different parts namely the Gärtner-, optics- and r(E)-part allows for a complete description of the experimental results. Consequently, both electronic and optical film properties are available allowing for a mutual confirmation of the pure optical and electrical methods. We will discuss in which potential and film thickness ranges the different parts of the model dominate and whether a quantitative evaluation of optical and electronic properties is possible. 5.2.1

Photocurrent model for ultrathin, amorphous films

Illumination of anodically polarized n-type semiconducting TiO2 films with photon energies above the band gap energy of Eg ⬇3.5 eV results in electron/hole pair formation. These charges are separated within the space charge layer (scl) causing a photocurrent iph. According to the Butler–Gärtner model, the main contribution to iph is due to electron/hole pairs, which are directly generated within the space charge layer namely iscl. Another contribution is due to charges diffusing into the scl, that is, iph iscl idiff. The iph generation is in competition with recombination processes throughout the entire layer. These recombination processes are not explicitly considered in the Gärtner approach. The electronic layer parameters enter into the Gärtner model just by their effect on the extension of the scl (dscl): dscl

冪

2␧(0)␧0(UUFB) eND

(4)

U is the applied potential and UFB the flat band potential, e1.6 1019 C, ␧0 8.859 pF m1. However, in the discussion so far the optical film properties describing the absorption of light, that is, the number of photons g generating the electron/hole pairs are not considered yet. This generation number g is plainly given by the change of the light intensity I per unit length z, that is, g(z)

d I dz

(5)

128

A. Michaelis and S. Kudelka

Ambient 0

scl Layer 1

d

Substrate 2

Figure 5.3 Multiple internal reflections within ultrathin semiconducting layers (penetration depth of light ␦ >> film thickness d. The extension of the space charge layer scl depends on the applied potential U as well as on the electronic layer properties (defect density ND, dielectric constant ␧(0)). It is assumed that only electron/hole pairs generated within the scl contribute to the photocurrent.

In the classical Butler–Gärtner model, light absorption is simply described in terms of Lambert–Beer’s law, containing just one optical constant, the extinction coefficient k. But this approach holds for bulk semiconductors only. In the case of ultra thin semiconducting layers dealt with here, multiple internal reflections must be considered26–28 if the penetration depth ␦ of the light is larger than the film thickness df. This is illustrated in Figure 5.3. The light energy flux I into the → volume has now to be described by the pointing vector I . A detailed description of the derivation of this quantity can be found in Ref. 28. In this case, n and k of both layer and substrate enter into the expression for iph and, even more important, the layer thickness df is now quantitatively considered. In principle, this optical model can be coupled to any electronic model. Here, the Butler–Gärtner approach shall be kept, yielding

冕冢冣

iphe

dscl

0

l

dI(z) dz dz

(6)

if only the contribution of electron/hole pairs generated within the scl is considered. It can be shown, that the neglection of the diffusion contribution is small.23,28 More severe and decisive is the complete neglection of recombination processes in Eqn (6). Recombination processes are of particular interest in case of amorphous semiconductors because of the large number of recombination centers. The recombination, that is, the efficiency of the electron/hole pair separation must clearly depend on the E-field gradient, which itself is a function of the applied potential and the layer thickness. Therefore, a recombination factor r (E) must be coupled to the above equation. For r (E) the following approach obeying the boundary conditions of r (E)0 for vanishing E-fields and r (E)1 for high

Application of optical methods and lasers 129 fields can be chosen:

冢

r(E ) 1

1 const · E1

冣

(7)

yielding

冢

iphe 1

dI(z) 冕冢冣 dz 冣 dz

1 · const [UUFB/dscl]1

dscl

l

(8)

0

with a linear E-field approximation. The unknown constant in this equation can be determined by fitting of the experimental iph(U) curves. The used function for r(E) is a good approximation of the expression for the recombination efficiency derived by Pai and Enck31 for amorphous materials which had already been discussed in this context by Quarto et al.23,30 The comparison of both equations shows that the const has to be a function of the thermalization length which itself depends on the mobility of the charge carriers and the photon energy of the exciting light. In Table 5.1 the contributions of the three different parts of the model to the observed signals are summarized for different cases. The three parts are ●

●

●

The classical Gärtner part describing the effect of the extension of the space charge layer dscl which itself depends on the electronic properties ␧(0) and ND. The exact optical model describing multiple internal reflection which are important for thin films where the penetration depth of the light ␦ is larger than the layer thickness. Due to this part iph depends on the layer thickness and the optical constants of both substrate and layer. This optical part mainly explains the observed iph(df) variation found in the photocurrent spectra. The third part describes the electron/hole pair recombination efficiency r(E) which causes a further dependence of iph on the layer thickness for very thin films, high potentials respectively (dscl df) and superimposes to the thickness dependence due to the optical part.

The “arrow up”, “arrow down” symbols in the table denote an increase or decrease of the applied potential U and df or iph respectively. For example, the iph thickness dependence in case 1 and 4 (dscl df) is given solely by the effect of multiple internal reflections or Lambert–Beer law depending on whether the penetration depth ␦ of the light is larger or smaller than df. Therefore, these cases can be used to judge the optical model and to determine the optical parameters. In the case of the ultrathin films df << ␦ only applies. The experiments presented below can be understood by referring to particular cases of this table.

130

A. Michaelis and S. Kudelka

Table 5.1 Contributions of the different parts (dscl change, recombination efficiency r(E), optics) of the presented photocurrent model to the potential and thickness dependence of the photocurrent (iph(U) and iph(d)). The “arrow up”, “arrow down” symbols refer to an increase, decrease of the corresponding parameters Case

Varied parameter

Effect on iph divided into the different parts of the model

dlay

U

dscl

r(E)

Multiple Lambert– reflections Beer

1 2 3

dscl dlay and dscl

↑ const ↑

const ↑ ↑

const ↑ ↑

const ↑ ↑

↑↓ const ↑↓

not valid (n.v.) n.v. n.v.

4 5 6

dscl dlay and dlay ␦

↑ const ↑

const ↑ ↑

const ↑ ↑

const ↑ ↑

n.v. n.v. n.v.

↓ const ↓

7 8 9

dscl dlay and dlay␦

↑ const ↑

const ↑ ↑

↑ const ↑

↓ ↑

↑↓ const ↑↓

n.v. n.v n.v.

10 11 12

dscl dlay and dlay ␦

↑ const ↑

const ↑ ↑

↑ const ↑

↓ ↑

n.v. n.v. n.v.

↓ const ↓

Figure 5.4 shows iph(df) simulations according to the model for two different wavelengths (280 and 375 nm) under potentiostatic conditions (applied potential 2 V). The optical constants used for these calculations were determined by ellipsometry (isotropic approximation) and are given in the insets of the figures. Due to the strong light absorption at 280 nm (k1.4) no interference is observed at this wavelength. The decrease of the photocurrent at high film thicknesses is due to the dominating E-field effect r(E) corresponding to cases 7 and 10 of Table 5.1. For this, an insulating TiO2 film behavior was assumed, that is, the E-field gradient reaches through the entire film (dscl df). At 375 nm (3.3 eV) case 7 clearly dominates (␦<< df), that is, the interference effect is obvious. Note that in this case the photon energy lies below the direct bandgap energy of TiO2. In this spectral range the photocurrent is caused by indirect or intraband excitations. Since this intraband mechanism differs from the direct interband one, the electronic part of the model is not fully applicable and slight deviations between measurement and simulation have to be expected (see experiments below).

5.3

Experimental details

As an example for the application of the optical micro-methods the Ti/TiO2 system was chosen. Due to its pronounced heterogeneity this system is particularly suited to demonstrate the advantages of micro-methods. The heterogeneity of the Ti/TiO2 system can immediately be observed by looking at an electropolished (open grain boundaries) Ti/TiO2 surface through a microscope. The TiO2 passive

Application of optical methods and lasers 131 0.07 Ti: n =1.13 k =1.33 TiO2: n =1.95 k =1.4 = 280 nm

Quantum yield (a.u.)

0.06 0.05 0.04 0.03 0.02 0.01 0 0

20

40

60 80 100 Film thickness (nm)

120

140

160

Quantum yield (a.u.)

0.006 Ti: n =1.47 k =1.33 TiO2: n = 2 k = 0.05 = 375 nm

0.005 0.004 0.003 0.002 0.001 0 0

20

40

60 80 100 120 Film thickness (nm)

140

160

Figure 5.4 Simulations according to the photocurrent model (Eqn (8)) for two different wavelengths. The used optical constants are given in the insets. It was assumed that the E-field gradient reaches through the entire film (dielectric behavior), applied potential 2 V.

film properties significantly depend on the substrate texture resulting in a variation of the TiO2 interference colors varying from grain to grain. The objective of the presented experiments is the quantification of this phenomenon. For all experiments Ti high purity (99.98%) samples with a coarse grain texture were used (grain diameter 600 m). This allows performing several independent experiments on each single grain. The crystallographic orientation of the substrate grains was determined in situ by AME as described below. In order to prepare samples with open grain boundaries the surfaces were electropolished. As the

132

A. Michaelis and S. Kudelka

electrolyte 0.5 M H2SO4 was chosen. Potentials are given with respect to the standard hydrogen electrode (SHE). As the optical/optical methods micro ellipsometry and micro-reflection spectroscopy were used. For ellipsometry a rotating analyser configuration with a HeNe laser (632.8 nm) as the light source was employed. AME requires rotation of the sample around the surface normal (denoted by an angle ␣). In order to allow this sample rotation, a particular in situ cell was constructed. For this, the cell windows were mounted on the ellipsometer arms using an alignable connector and were simply dipped into the electrolyte. Therefore, rotation of the entire cell containing the sample in three-electrode configuration is possible without affecting the window alignment. More details on AME can be found in Refs. 13 and 14. For optical/electrical (photoelectrochemical) experiments at high lateral resolution, an in situ UV-laser scanning apparatus was developed32,33 (see also Figure 5.1). The light of a frequency doubled Ar-laser (␭257 nm, i.e. 4.8 eV) is focused onto the sample surface by means of the same microscope which is used for control of the optical/optical and electrical/electrical methods, that is, the measurements can simultaneously be carried out. A minimum spot radius of about 1 m was achieved yielding power densities up to some kW cm2. The sample/cell configuration can be scanned normal to the light beam using a computerized x, y-stage (step size1 m). For the detection of photocurrent spectra, the laser is substituted by a standard 1 kW Xe high pressure arc lamp and a monochromator. In this case the lateral resolution is about 50 m. The whole setup is based on a modular concept, that is, all measurements can be performed simultaneously under microscopic control and further tools such as reflection-spectroscopy or video-microscopy can be added. The new nl-droplet photoresist method (electrical/electrical) is described in Section 5.4 in full detail. (The photoresist microelectrodes were prepared by standard mask-photolithography. An electrolyte droplet is dispensed onto the microelectrodes by means of a microcapillary, which is connected to a nanopipette and a miniaturized reference electrode. A Pt-wire (50 m diameter) is used as the counter electrode which is bent to a loop and dipped into the droplet. Measurements and preparation are performed under microscopic control).

5.4 5.4.1

Optical/optical and optical/electrical measurements Determination of Ti substrate grain orientation by AME

In order to determine the crystallographic orientation of single Ti grains, the new AME method was applied. The fundamentals of this method and its application for the determination of crystal orientations of optically anisotropic systems (Ti crystallizes in the hcp-lattice) are described in recent publications.13,14 AME yields the angle ␸ between the surface normal and the optical axis (c-axis). For this, the ellipsometric measurables and are measured as a function of the angle ␣ which describes rotation of the sample around the surface normal. The

110

34

109 (0001)

33.5

108

33 32.5 (xxx0)

106

32 105

Ψ (°)

∆ (°)

107

31.5

104

31

103 102

(0001)

101 0

60

120

240

180 α (°)

300

30.5 30 360

(0001)

c

(×××0)

a3 a1

(0111)

a2

113

34 (0111)

112

33.5

111

33 32.5

109 32 108 31.5

107

31

106

30.5

105 104

Ψ (°)

∆ (°)

110

0

60

120

180 α (°)

240

300

30 360

Figure 5.5 Anisotropy Micro-ellipsometry (AME) on three different Ti grains with (0001), (0111) and (xxx0) orientation. The crystallographic orientations are illustrated in the displayed Ti-lattice.

134

A. Michaelis and S. Kudelka

corresponding experimental curves are shown in Figure 5.5 for the example of two different Ti-grains. The grains are covered by anodically formed 4 V TiO2 layers which themselves are amorphous and therefore do not contribute to the (␣) and (␣) variation. The quantitative analysis of these curves allows determination of the crystallographic orientation. The planes determined from the curves of Figure 5.5 are illustrated in the shown Ti-crystal structure. The (0001)-surface (␸ 0º) is the closest packed surface with 1.15a2 (a 0.2951 nm, c0.4679 nm), (xxx0) refers to any plane perpendicular to the c-axis (␸90º). 5.4.2

Photocurrent spectra and iph(U) measurements on single Ti/TiO2 grains

Micro-photocurrent spectra (quantum yield Q) taken on different oriented single Ti grains in 0.5 M H2SO4 are shown in Figure 5.6. The TiO2 films were formed potentiodynamically in the same solution at the indicated formation potentials (2–20 V) before the spectra were taken at a constant anodic potential of 1 V. Consequently, the data show the dependence of the photocurrent spectra on both the TiO2 film thickness and the grain orientation. The formation factor was measured for each of the different orientations by ellipsometry (value averaged over all orientations: 2 nm/V). This allows for a quantitative correlation of the spectra with film thickness. Evaluation of the Q2- and Q0.5-curves allows determination of the direct and indirect bandgap energies. Edir varies between 3.7 and 3.4 eV for the different orientations and shows a slight decrease with increasing film thickness. The same holds for Eindir which varies between 3.3 and 3 eV. For energies below the band gap an Urbach tail was found with a slope of 120 meV for the disorder energy. This indicates the amorphousity of the TiO2 passive films. Consequently, there is no sharp band gap but a mobility gap. The mobility gap energy is close to the band gap energies of the TiO2 modification anatase, therefore indicating a relation between this modification and the amorphous TiO2 layers found here (near order). The variation of iph with the Ti grain orientation confirms that in spite of the amorphicity of the TiO2 films, their properties depend sensitively on the substrate texture. It has to be emphasized that the dependence of iph on df caused by internal reflections and the recombination processes mentioned in the above photocurrent model is evident across the entire spectral range: up to 10 V iph increases, thereafter iph decreases. This interference pattern is in excellent agreement with the simulation curves shown in Figure 5.4, therefore confirming the model. In order to allow for a more quantitative check of the model, the photocurrent was measured as a function of the formation potential, film thickness respectively. Four experimental curves covering the spectral range from 280 to 375 nm are shown in Figure 5.7 for an 0001 oriented grain. During illumination an anodic potential of 2 V was applied. For the wavelength of ␭ 280 nm clearly case 10 of Table 5.1 applies, that is, the r(E) term dominates the entire thickness range. Since the penetration depth of the light increases with increasing wavelength, the interference effect becomes more and more prominent at larger wavelengths confirming the contribution of internal reflections to iph. The simulations shown in Figure 5.4 agree well with the experimental curves. The

Application of optical methods and lasers 135 0.03 (0001) 8V 0.02

4V

Q

10 V

20 V

0.01

6V 2V

0

3

3.5

4 Energy (eV)

5

4.5

0.03 (xxx0)

0.025

10 V

Q

0.02 0.015

6V

8V

0.01 0.005 0

2V 3

3.5

4 Energy (eV)

20 V 5

4.5

0.03

Q

(0111) 0.02

8V 10 V 6V

0.01 2V 0

3

3.5

4 Energy (eV)

20 V

4.5

5

Figure 5.6 Photocurrent spectra (quantum yield Q) of TiO2 passive films for different film thicknesses (formation potential, respectively). The measurements were carried out on three differently oriented Ti single grains. Electrolyte 0.5 M H2SO4, anodic potential 2 V.

minor deviations between model and experiment can be attributed to the different mechanisms for direct and subband excitations. 5.4.3

Microscopic modification of the TiO2 films by means of laser scanning

The migration of holes to the oxide/electrolyte interface leads to an accumulation of positive charge at the surface, which causes a change of the potential distribution within the oxide, namely, the photopotential. Under potentiostatic conditions, the

136

A. Michaelis and S. Kudelka

0.07

0.05

280 nm/4.439 eV 0.06

315 nm/3.946 eV

0.04

0.05 0.03 Q

Q

0.04 0.03

0.02

0.02 0.01

0.01 0.00

0.00 0 10 20 30 40 50 60 70 80 U P (SHE)/V

0.05

0 10 20 30 40 50 60 70 80 U P (SHE)/V 0.005

335 nm/3.71 eV

0.03

0.003

375 nm/3.315 eV

Q

0.004

Q

0.04

0.02

0.002

0.01

0.001

0.00

0.000 0 10 20 30 40 50 60 70 80 U P (SHE)/V

0 10 20 30 40 50 60 70 80 U P (SHE)/V

Figure 5.7 Photocurrent (quantum yield Q) as a function of the TiO2 formation potential, film thickness respectively. During illumination a potential of 2 V was applied to the electrode.

potential drop in the Helmholtz layer will be increased accordingly enhancing the charge transfer of oxygen through the oxide/electrolyte interface. This affects the film formation which itself is a simple anodic ion transfer reaction without direct contribution of holes as: Ti 2H2O ⇒ TiO2 4H 4e

(9)

On the other hand, the holes are directly involved in two reactions, oxygen formation and photo-corrosion, as: 4h 2H2O ⇒ O2 4H

(10)

4h TiO2 ⇒ Ti4 O2

(11)

and

Application of optical methods and lasers 137 As will be shown below, at high anodic potentials reaction (9) dominates allowing for a light-induced modification of the surfaces, that is, laser-induced film growth. This is demonstrated in Figure 5.8 where the results of so called locodynamic experiments across grain boundaries between single Ti grains of known orientation are shown. For this, the illuminated samples were scanned in m steps normal to the laser beam while the electrode potential was increased as in a conventional cyclovoltammogram. The advantage of this procedure is that the ray always hits an almost fresh surface area allowing to investigate the potential dependence of the photo-reactions without previous modification of the initial semiconducting films. Figure 5.8(a) shows micrographs of the resulting modified Ti surfaces on four substrate grains. The surface orientations and the grain boundaries between the single grains are indicated. The locodynamic experiments were performed on anodic TiO2 films formed potentiodynamically at 20 V. The UV-laser power was 50 kW/cm2, the applied potential Uloco during illumination was cycled between 0 V and 3 V. The lateral dimensions of the laser modified line (parallel to the grain boundaries) exceeds significantly over the

(a)

(0001)

0v

(xxx1) oko 0v

3v 3v 0v 0v (xxx0) (b)

(0111)

0.4 0.35 0.6 0.5

0.1

1

(0111)

2

x (µm) 0

100

200

300

3 400

500

Ulo

0.15 (xxx0)

co (V)

0

0.4 0

0.3 (0001) 0.2 0.1

(xxx0)

1

2 x (µm) 3 0 50 100 150 200 250 300 350 400 450 500

co (V)

0.2

Ulo

0.25

iph (µA)

iph (µA)

0.3

Figure 5.8 (a) Micrographs of the (0001)/(xxx0) and (0111)/(0001) grains (anodic 20 V TiO2 films). The applied electrode potential Uloco was cycled between 0 and 3 V, laser power50 KW cm2. The visible laser traces indicate formation of a gradient in the TiO2 film thickness. (b) Laser scanning analysis across the laser traces and the grain boundaries (imaging of the film properties).

138

A. Michaelis and S. Kudelka

illuminated area (spot radius 2 m) indicating lateral transport mechanisms such as hole diffusion beyond the laser focus. The color of the laser traces varies with the applied anodic potential Uloco indicating formation of a gradient in the film thickness.

5.4.4 5.4.4.1

Characterization of the modified TiO2 films UV-laser scanning analysis

For a quantitative analysis of the laser-modified films, laser-scanning experiments normal to the laser traces and across the grain boundaries were carried out using reduced laser power below the threshold of modification (50 W). The results (iph(x, y)) are shown in Figure 8(b). The scale of the x-axis holds for both micrograph and the corresponding iph(x)-curve. These curves give a sectional profile of the laser-modified site corresponding to the potential Uloco applied in the locodynamic experiment before. From the x-scan the lateral dimensions of the laser-induced modification can be evaluated. Scanning of the surface iph f (x, y) in both dimensions allows an imaging of the entire potential dependence iph f(x, yUloco) of the laser-induced modification. The photocurrent fluctuations on the unmodified crystal surfaces are below 1%. Crossing from one crystal surface to the other (scan across the grain boundary) yields an immediate iph change of about a factor of 2 due to the dependence of iph on the texture as discussed previously. From the decay of this sudden change, the lateral resolution of the method can be estimated to be about 2 m which agrees well with the laser focus. Hence, it can be concluded that laser scanning allows imaging of passive layer properties at heterogenieties like grain boundaries. For instance, in case of Figure 5.8(a) the fringed structure of the grain boundary between the (0001)/(xxx0) surfaces, visible in the micrograph can be imaged accurately. 5.4.4.2

Micro-reflection spectroscopy

Figure 5.9 shows the film thickness measured on top of the modified region by means of micro-reflection spectroscopy. For the evaluation of the film thicknesses averaged optical constants were used which were determined by ellipsometry. At locodynamic potentials exceeding 0.25 V an increase of the film thickness is observed proving reaction (9) to be dominant in this potential regime. However, below 0.25 V a thinning of the TiO2 film took place which can be attributed to photocorrosion, that is, reaction (11). This leads to the important conclusion that UV-laser illumination can be used for both a thickening (writing) or thinning (erasing) of semiconducting films. The applied potential (Uloco) determines which reaction dominates and therefore can be used to control the layer thickness. One important application of this method is the generation of semiconducting films with thickness gradients. Such gradients can be realized by means of locodynamic illumination of the sample surface.

Application of optical methods and lasers 139 90 Micro-reflection spectra

80

(0001)

d (nm)

70 60

(xxx0)

50

(0111)

40 30

0

0.5

1

1.5 U loco (V)

2

2.5

3

Figure 5.9 Micro-reflection spectroscopically determined film thickness of the UV-laser generated TiO2 gradients as a function of the applied electrode potential Uloco.

55 50 WHM (µm)

45 (0111)

40 35 30

(xxx0)

25 20 15

(0001) 0

0.5

1

1.5 U loco (V)

2

2.5

3

Figure 5.10 Lateral extension (full width at half maximum) of the TiO2 gradients as a function of the applied electrode potential Uloco. The focus spot diameter during illumination was 2 m.

Additionally, df as well as the lateral extension of the laser-modified films depends on the substrate orientation, which therefore must be known. For this, AME can be applied in situ. For all substrate orientations it was found that the lateral extension of the modified regions exceeds the illuminated site significantly. In the case of the (0001) and (xxx0) orientation, the lateral extension of the modified site is almost independent of Uloco. In contrast, the (0111) surface shows a significant dependence of the laser line extension on Uloco. This is quantified in Figure 5.10 showing the full width at half maximum of the laser scans shown in Figure 5.8(b) as a function of Uloco. These curves correlate well with the reflection spectra results shown in Figure 5.9, that is, the laser induced oxides are vertically thicker but less extended for the (0001) and (xxx0) than for the (0111) orientation. This suggests that the amount of the laser-induced oxide is determined

140

A. Michaelis and S. Kudelka

by the number of generated electron/hole pairs and is almost constant for the different orientations. The difference in the lateral and vertical extension of the laser-modified region for the different grains is due to the different electronic properties of the original semiconducting films determining the lateral hole diffusion. 5.4.4.3

AME analysis of the modified films

In order to analyze the structure of the UV laser-modified anodic TiO2 films (formation potential: 20 V), AME was carried out (Uloco 2 V). Figure 5.11 shows the

40

82 Unmodified area 78 ∆ (°)

(d)

38 37

74

b

(b)

(0111) 0

50

100

35 150 200 α (°)

250

300

34 350

43 42.5 42 41.5 41 40.5 40 39.5

61 Near the laser induced oxide

60

∆ (°)

59 58 57 56 (0111)

Anatase

55 54

(c)

0

50

100

292

150 200 α (°)

250

300

39 38.5 350

45

On top of laser oxide

290

–c

Ψ (°)

70

50 µm

36

66

Region a

39

Ψ (°)

(a)

44.5 44

286 284

43.5

282

43

280 278

Ψ (°)

∆ (°)

288

(0111) 0

50

100

150 200 α (°)

250

300

42.5 350

Figure 5.11 AME curves recorded on potentiodynamically formed 20 V TiO2 films on a (0111) oriented Ti substrate grain. (a) in large distance from a laser modified spot, (b) close to the modifed region and (c) on top of the laser modified arc.

Application of optical methods and lasers 141 resulting AME-curves (␣) and (␣) for the (0111) grain as an example. Three AME experiments at different sites of the sample were performed, that is: 1 2 3

at large distance from the laser modified region (characterization of unmodified passive film) close to the illuminated area (effect of lateral hole diffusion on passive film) on top of the illuminated area.

At large distance from the laser illumination, the typical AME curve for an amorphous TiO2 with 35 nm thickness was measured. The film properties completely correspond to the films measured previously in Figure 5.5. More interesting were the AME results of experiment 2. The AME curves show a significant change in amplitude, phase and average value in comparison to the AME curve of experiment 1. Now, the optical parameters of crystalline anatase were derived suggesting a recrystallization of the TiO2 films. However, this phenomenon did not occur on the other two grain orientations. Lateral hole diffusion is most prominent on the (0111) orientation. This agrees with the laser scanning measurements of Figure 5.8 where the largest extension of the laser traces appears on the (0111) surface. Experiment 3 again shows a clear change in amplitude, phase and average value of the (␣) and (␣)-curves. The increase of the (␣) and (␣) average values can be attributed to an increased film thickness (reaction (9), i.e. film growth). The alteration of both phase and amplitude again indicates a structure change of the TiO2 films. Quantitative AME evaluation reveals that now a twolayer model has to be applied. It was found that the initial films (metal/oxide interface) recrystallized to anatase analogues to experiment 2 (optical parameters nao 2.48, no 2.55 (ordinary and extraordinary refractive index)). On top of this crystallized film an amorphous layer was found exhibiting a low density of about 3 g cm3 (estimated by analysis of the determined optical constants41). Therefore, these oxides can clearly be distinguished from each other and from the original anodic films. These experiments nicely illustrate how even on homogeneous electrode areas (in this case, one single (0111)-oriented grain optical micro-methods can be applied to characterize and elucidate local electrode reactions.

5.5 5.5.1

Photoresist microelectrochemistry (nano-liter droplet method) Introduction

As described in the preceding chapters, the application of focused laser light is a powerful instrument for the detection of local thin film properties such as morphology, thickness and photocurrent. In order to complement this information in terms of electrochemical reactivity, such as electron and ion transfer rates, capacitance and potential, a flexible photolithographic technique in combination with electrochemical measurements in small droplets under optical control has been applied to scale down classical electrochemical experiments.34 In comparison to

142

A. Michaelis and S. Kudelka

the well established scanning electrochemical microscope (SECM),35 the technique employed here, uses a patterned fixed ultramicroelectrode with a well defined surface area and a three-electrode configuration. Hence, the local resolution is limited by the lithographic resolution in the sub micrometer range and by the electronic detection limit of the current amplifier. 5.5.2

Experimental details

5.5.2.1 Preparation and measurement technique In order to allow the preparation of the photoresistmicroelectrode at the point of interest, preparation as well as the measurement were carried out under microscopic control as described in the flow chart in Figure 5.12. Following the resistcoating (positive tone, dresist 1 m) and prebake, the microelectrode is exposed under proximity conditions at the desired spot of the sample. A preconditioning of a highly reflective metal surface with HMDS (Hexamethyldisilane) and an antireflective coating was avoided due to the possible impact on the electrochemical reactivity of the surface, though limiting the performance of the lithography. Pinhole masks are positioned by a x, y, z stage and exposed by UV light, that is fed through the microscope. The distance between two adjacent microelectrodes can be as low as 50 m. A postbake and the development completes the preparation. A schematic drawing and the microscopic view of the experimental setup are shown in Figures 5.12 and 5.13. A glass capillary is filled with electrolyte and is connected to the reference electrode (e.g. Hg/Hg2SO4, U0 0.68 V) and an electrolyte reservoir with a nanopipette, so that nanoliter droplets can be dispensed on the photoresistmicroelectrode. The diameter of the droplet is usually set twice the diameter of the microelectrode. Micromanipulators are used to position the electrodes. A Pt-wire is dipped into the droplet as a counterelectrode, completing the experimental setup. Microelectrodes with a diameter of 5 m can be realized using proximity exposure with reproducible surface areas. Figure 5.13 shows an example of a microelectrode preparation on single grains of a polycrystalline Ti sample as viewed with the microscope. Individual grains are identified on the electropolished surface when the resist is applied. The localization of the electrode reaction is demonstrated by the different colors of the anodic oxide layers. Each electrode can be addressed separately. A more detailed description including the effects of the photoresistpreparation on the electrochemical measurements is given in Ref. 36. 5.5.2.2 General aspects It is beyond the scope of this treatment to discuss the current and potential distribution for photoresist microelectrodes in detail, as this will be covered by other contributors to this series. Therefore, only a few remarks are made on the influence of the geometry and the resist barrier on electrochemical measurements.

Application of optical methods and lasers 143

Resistapply

Prebake x,y,z

Mask Mask alignment

Exposure Postbake Development

x,y,z

GE

x,y,z

Electrochemical measurement

RE Electrolyte

WE=Substrate Resist strip

Figure 5.12 Process steps for the preparation and electrochemical measurement of photoresistmicroelectrodes. All steps are carried out under microscopic control with micromanipulators to handle the mask and the microelectrodes (GE, Counter electrode; RE, Reference electrode; WE, Working electrode).

Photoresistmicroelectrodes can be treated as a recessed microdisk electrode. The diffusion field and the current distribution are not hemispherical, as the aspect ratio increases. Model calculations36,37 have shown that in case of a microelectrode with an aspect ratio of 0.1 (diameter of the electrode d 10 m, dresist 1 m) the resistivity increases by 15% in comparison to a planar microdisk electrode. For an aspect ratio of 1 the value increases to 50%. These limitations are negligible unless very high current densities in the order of A cm2 occur. For potential step measurements, high currents occur due to the

144

A. Michaelis and S. Kudelka

(a)

(b)

200 µm

200 µm

(c)

(d)

200 µm

200 µm

Figure 5.13 Microscopic view of sample preparation and measurement as expressed in Figure 5.12. The same area of a polycrystalline Ti sample is shown after different treatments: (a) with 20 V anodic oxide layer formed in 0.5 M H2SO4 potentiodynamically with 60 mV s1, (b) after electropolishing, (c) during the electrochemical measurement on a 100 m photoresistmicroelectrode and (d) after growing anodic oxide layers and stripping of the photoresist.

capacitate charging of the double layer, microelectrodes with d 50 m have been used in this study. Here, the assumptions of planar microelectrodes are still applicable. The time constant for charging a capacitance of CH 20 F cm2 in 0.5 M H2SO4 (3.7 cm) as a 50 m microelectrode is less than 1 s, thus allowing kinetic studies in the ms range as will be demonstrated in Section 5.5.3.3. In addition, the distance between the microelectrode and the reference and counterelectrode was adjusted to some m, which helps to control the potential of the electrode. Due to the finite electrolyte volume of the nanoliter droplet, the diffusion boundary layer reaches the dimension of the droplet within seconds, assuming a diffusion coefficient of, for example, D1105 cm s1. A constant diffusion

Application of optical methods and lasers 145 gradient can not be established because the bulk concentration is reduced. Thus, measurements of diffusion limited currents are not stable over time. The electron transfer reaction discussed in Section 5.5.3.4 were carried out on anodic oxide layers with maximum current densities 3–4 orders of magnitude lower than the diffusion limited current densities, thus ensuring a stable support of consumables over the time of measurement. In order to prevent evaporation of electrolyte, the ambient was saturated with water vapor. The use of a small electrolyte covered resist area around the microelectrode is essential when capacitance measurements are performed, since the resist capacitance is parallel to the electrode capacitance. With a specific resistance of ␴1012 cm and a dielectric permittivity of ␧1.5 for the resist and assuming a typical electrode capacitance of 10 F cm2 with a 50 m electrode, an error of 5% is obtained, if the electrolyte covered surface is 103 cm2 ( f1000 Hz).36 Thus, for capacitance measurements, the use of nanoliter droplets is essential.

5.5.3

Applications of photoresistmicroelectrodes

In the following chapters the electrochemical reactivity of single grains of polycrystalline Ti is explored by using the nl-droplet method. The results from electrochemical measurements and the optical laser techniques from Chapter 4 are combined to yield a band structure model for anodically grown anodic oxide layers. Other applications of this method to study laser-induced corrosion, texture-dependent photocurrent and corrosion of anodic oxide films are described in Refs. 38–40.

5.5.3.1

Cyclovoltammetry

Anodic passive layer formation on single Ti grains The following measurements were performed on single grains of polycrystalline Ti. In some cases the experimental results are discussed for two deliberately selected grains (a) and (b), because the aspects of the different behavior are represented sufficiently. Prior to the electrochemical measurements, the surface orientation of these grains and the thickness of the anodic oxide layers was determined by microellipsometry as described in Chapter 4. In Figure 5.14 several cyclovoltammograms are shown, that were obtained on a fresh electropolished Ti sample. Obviously, the texture has a significant influence on the anodic current density. The total anodic current density i is composed of four contributions: iiOx iO2 icorr iC with iOx oxide formation, iO oxygen evolution, icorr corrosion current and iC due to capacitive 2 charging. The latter two contributions can be neglected since iC does not exceed 2 A cm2 (dU/dt50 mV s1, CH 20 F cm2) and icorr was found to be less than 8% in the potentiodynamic preparation mode.41 In the potential region from 0 to 3 V SHE the oxygen evolution can be excluded and the anodic current density is purely the oxide formation iOx. Thus, the anodic charge of the oxide layer formation depends on the texture of the

146

A. Michaelis and S. Kudelka (a)

Ti/0.5 M H2SO4

4 i ≈ iO2 + iOx

Grain b

i (mA cm–2)

3

2 i ≈ iOx

1

Grain a

0 0 (b)

2

4 6 U (SHE)(V)

8

10

–1

log q (C cm–2)

q ≈ qO2 + qOx q ≈ qOx

Grain a –2 Grain b

–3 0

2

4 6 U (SHE)(V)

8

10

Figure 5.14 Current density (a) and charge (b) during the anodic oxidation on different grains of a polycrystalline Ti sample (diameter of the microelectrode 20 m, dU/dt50 mV s1 in 0.5 M H2SO4).

underlying Ti. For grain (a), with an orientation angle ␸85º, the current density is independent of the potential, corresponding to a proportional growth of the oxide thickness with the potential U. For grain (b), with an orientation angle ␸37º, i decreases with U, thus the oxide formation rate is potential dependent. The relationship between surface orientation, expressed by the orientation angle ␸, and anodic oxide charge qOx is systematic, as plotted in Figure 5.15. At orientation angles 30º, (near the (0001) orientation) the anodic charge is 30% less than at orientation angles 50%. A sharp transition is observed at 40º. The values for the (0001) and (xxx0) orientation are taken from measurements at macroscopic single crystal electrodes. These experiments prove that standard electrochemical measurements can be extended into the m range.

Application of optical methods and lasers 147 (0001)

(xxx0)

15

qOx, pot.-stat.

10

log N d (cm–2)

qOx(mC cm–2)

20

qOx, pot.-dyn. ND, pot.-dyn. 5

19

ND, pot.-stat. 0 0

10

20

30

40 50 (°)

60

70

80

90

Figure 5.15 Anodic charge qOx and donor density ND as a function of the orientation angle ␸ for potentiodynamically (slow dU/dt50 mV s1) and potentiostatically (fast) grown oxide layers. For the potentiodynamically grown oxides, the oxide charge was integrated from 0.5 to 3 V (SHE), for the potentiostatic grown oxides from 0 to 4 V. Values for (xxx0) and (0001) orientation were taken form measurements on macroscopic single crystal electrodes.

In the potential region above 3 V, additional oxygen evolution may contribute, that also depends on the surface orientation. On grain (a) (close to (xxx0)) only a slight increase in the current density can be observed at U 8 V, whereas on grain (b) (close to (0001)) the current rises up to 10 times the oxide formation current. A correlation with the oxide formation current can be found again. The oxygen evolution increases with a decrease of the anodic charge between 0.5 and 3 V. As demonstrated below, this is due to an increased conductivity of the oxide layers. 5.5.3.2

Capacitance measurements

As outlined in Section 5.5.2, the resist capacitance can be neglected, if the electrolyte covered resist surface is reduced, for example, 3104 cm2 for a 50 m electrode. Typical capacitance values of a 50 m diameter electrode at anodic potentials of U5 V are 100 pF. Figure 5.16 shows the reciprocal capacitance as a function of the formation potential. Experimental values are represented by a straight line in case of grain (a) (xxx0) orientation. The dielectric permittivity of ␧34 can be calculated by a simple capacitor model. In contrast, for

148

A. Michaelis and S. Kudelka 0.7 Grain a ≈ 34

0.6

C –1(µF–1 cm2)

0.5 0.4

Grain b ≈ 40

0.3 0.2 0.1 0.0 0

2

4

6 8 U (SHE)(V)

10

12

Figure 5.16 Reciprocal capacitance C plotted against the formation potential U for grain a and b with two different orientations. The oxide layers were grown in 0.5 M H2SO4 with 20 mV s1. The capacitance was measured at the upper formation potential with f1030 Hz. The assumption of a simple capacitor model yields in combination with ellipsometric determined thicknesses a dielectric permittivity of 34 for the oxide film on grain a and a mean value of 40 for grain b.

example, for grain (b) (0001) a nonlinear behavior reflects the nonlinear growth. The average dielectric constant of ␧40 can be estimated. At lower potentials than the formation potential, for films formed at 2, 4, 6 and 10 V the capacitance are plotted according to the Schottky–Mott equation in Figure 5.17. The slope of the curves is a function of the dielectric permittivity ␧ and the donor density ND :1/C2 1/CHL 2/e␧␧0ND (UUFBkT/e) with UFB the flatband potential, e the elementary charge and k the Boltzmann constant. Below 0.5 almost straight lines are observed. Here, only the slopes of the curves were evaluated. In the case of grain (a) (xxx0), the slope increases with the formation potential. In contrast, grain (b) (0001) shows an almost constant slope up to 4 V formation potential. According to the Schottky–Mott equation, donor densities are calculated and plotted as a function of the formation potential in Figure 5.18. In the potential region between 3 and 5 V (SHE) donor densities differ by almost one order of magnitude, underlining the texture-dependent electronic properties of the passive layers. Again, the change in the donor density is a systematic function of the crystal orientation, as indicated in Figure 5.15 for potentiodynamically formed oxides. 5.5.3.3

Transient measurements

One of the main features of microelectrodes is the suitability for the study of fast electrode reactions. Due to a low potential drop in the electrolyte and low absolute

Application of optical methods and lasers 149 0.02

Grain b 10 V

C –2 (µF–2 cm4)

Grain a 10 V 6V

6V

4V

0.01

2V 4V 2V 0.00 0.0

0.2 0.4 U⬘(SHE)(V)

0.0

0.2 0.4 U⬘(SHE)(V)

Figure 5.17 Evaluation of the capacitance according to the Schottky–Mott equation for grain a and b for oxide layers grown potentiodynamically at different formation potentials U (100 m electrodes, f1030 Hz, dU/dt20 mVs1).

20

log ND (cm–3)

Grain b

19 Grain a

18 0

2

4

6 8 U (SHE)(V)

10

12

Figure 5.18 Donor densities of the passive films on grain a (xxx0) and b (0001) as a function of formation potential derived form the Schottky–Mott plots in Figure 5.17, with dielectric permitivities from Figure 5.16.

currents, the potential of the microelectrode can be controlled very accurately. Here, the formation of anodic oxide layers was investigated as a current response to a potential step experiment. The potential of the oxide free electrode (natural oxide) is jumped to the formation potential. Figure 5.19 shows time and current in a double logarithmic plot for an anodic pulse form U0–4 V (SHE). The

A. Michaelis and S. Kudelka

3 2

log i (A cm–2)

14

4

d c

1

16

U (V)

a e

0

0

12

t (s)

0

10

10

8

–1

6

–2

c d e a

–3 –4

q (mC cm–2)

150

4 2 0

–6

–5

–4

–3 –2 log t (s)

–1

0

1

Figure 5.19 Double logarithmic plot of current density and time t for a potentiostatic pulse from 0 to 4 V (SHE) on four different grains with different orientations. Microelectrode diameter was 100 m.

potential control of the electrode was achieved within a few microseconds. High current densities of more than 100 A cm2 could be realized, with the absolute values of only some mAcm2. Time as well as current could be followed over almost 7 orders of magnitude by using a fast transient recorder in combination with automatic amplifier sense switching. In contrast to the slowly grown potentiodynamically formed oxides, for all grain orientations, charge and current is almost identical. Again, prior to the measurement, the orientation of each grain was determined by microellipsometry. Thus, for fast grown oxides the influence of the texture is decreased drastically. The slope d log i/d log t1 is in agreement with an oxide growth according to a high field mechanism. Additional evidence for the grain independent properties of the fast grown oxides can be found from capacitance and electron transfer reaction measurements, as will be shown later. The donor densities of the fast grown oxide layers are lower than for the potentiodynamically formed oxides, indicating a more isolating character (see Figure 5.15).

5.5.3.4

Electron transfer measurements

ETR measurements were carried out in a 0.5 M Fe2/Fe3 redoxsystem in 0.5 M H2SO4 after the oxide layers were formed in a slow growth mode also in 0.5 M H2SO4 with a formation potential of 4 V. Capacitive charging and ionic currents are subtracted, so that the Tafel plots in Figure 5.20 represent only the contribution of ETR. Again each curve represents a grain with a different crystal orientation. Since it is not the subject of this treatment to elaborate the different mechanisms of ETR, only a few remarks are made here. A more detailed description is

–2 30°

–3

(a) 37° log i(A cm–2)

–4 40° (b) 45°

–5

(c) 55° (e) 85°

–6 –7 –8 –9 0.0

0.5

1.0 1.5 U (SHE)(V)

2.0

2.5

Figure 5.20 Tafel plots of electron transfer reaction measurements with 0.5 M Fe2/Fe3 in 0.5 M H2SO4 on different grains of a polycrystalline sample. The orientation of the grains is given as the angle p. The current without the redox system is subtracted. The current detection limit is indicated with the horizontal line (d200 m, dU/dt5 mVs1).

–1 Anodic = +1.3 V ∆Et = 4 meV

–2

log i (A cm–2)

–3 –4 –5 –6 –7 –8

Cathodic = –0.2 V ∆Et = 20 meV

–9 0.00 0.04 0.08 0.12 0.16 0.20 0.24 C –1(µF–1 cm2)

Figure 5.21 Cathodic and anodic ETR current densities at a constant overvoltage of ␩ 0.2 V and ␩1.3 V taken from Figure 5.20, plotted vs the corresponding reciprocal capacitance C1 at the same ␩, measured on different grains.

152

A. Michaelis and S. Kudelka Ti

EF

TiO2 CB 1 – 2 – – –

H2O + Ox/Red Fe2+

Eg = 3.2 eV VB d = 6–10 nm

Figure 5.22 Schematic representation of electron transfer reaction to the conduction band for direct (1) and resonance-tunneling (2).

available in Ref. 42. A direct electron exchange with the metal can be excluded due to the oxide thickness of 6.5–10 nm. On the cathodic as well as on the anodic site, log i decreases with increasing orientation angle ␸, that means from (0001) to (xxx0) orientation. In the same direction, the donor density decreases which results in a increasing extension of the space charge layer. Since the extension of the space charge layer is proportional to 1/C at a given potential, the plot of log i vs C1 in Figure 5.21 suggests that a tunneling mechanism is the rate-determining step. Oxide layers with the highest donor concentration show the highest anodic current densities because of the reduced tunneling distance to the conduction band (reduced Debye length). In case of the (xxx0) orientation (grain e), the anodic reaction is totally blocked. On the cathodic branch at high overvoltages (␩500 mV), the ETR is independent of texture, which is typical for a direct elastic tunneling process (see process 1 in Figure 5.22). At lower overvoltages a texture-dependent shoulder in the current density is observed, that can be explained by a resonance tunneling process (see process 2 in Figure 5.22). Oxide layers formed by a potentiostatic pulse experiment (see Section 5.5.3.3) show the behavior of grain (e), independent of the crystallographic orientation, underlining the isolating character.

5.6

Summary and conclusions – Ti/TiO2 model

We have shown that the employment of micro-optical methods along with conventional electrochemistry represents a powerful approach for the study of local electrode reactions on technically relevant heterogeneous solid surfaces. Due to its directionality, laser radiation is especially well suited for focusing to minimum dimensions. A local resolution in the order of the wavelength of light can be obtained. For instance, focused laser radiation can be applied to prepare photoresist microelectrodes with m dimensions. The resulting photoresist nl-droplet method allows for the direct electrochemical study of electrode reactions on single grains of polycrystalline material as is demonstrated with the Ti/TiO2 system

Application of optical methods and lasers 153 in this paper. Moreover, focused laser radiation allows obtaining high power densities enabling an efficient induction of local electrode reactions, which can be used to characterize and modify the electrode surfaces. It is essential to perform all measurements under microscopic control. This allows precise measurement on deliberately chosen electrode areas. For instance, all measurements presented in this study were carried out on the same single grain of only one electropolished Ti-sample. The properties of the semiconducting TiO2 films sensitively depend on the substrate texture. The new AME-method was used for in situ determination of the substrate grain orientation. The electronic and optical TiO2 film parameters were determined by means of photoresist micro-electrochemistry and micro-ellipsometry and then correlated with the Ti substrate grain orientation. The results using photoresist microelectrodes show that conventional electrochemical measurement techniques provide valuable information on the electronic properties even in m areas. The passive layer formation on Ti depends on the grain structure in terms of anodic charge, growth factor, oxygen evolution rate, electron transfer rate and donor density. The most important results can be summarized in a geometric film model shown schematically in Figure 5.23. Because of the semiconducting

Ti

TiO2

Ti ∆E (eV)

TiO2

H2O

CB e∆Ox

2 eV EF Eg = 3.2 eV

Grain b Na ≈(0001)

VB

OH2 O2, H+

2 4 6 X (nm) ∆E (eV)

CB e∆Ox = 2 eV

Grain a Ne (xxx0)

EF Eg = 3 eV

VB 2 4 6 8 X (nm)

Figure 5.23 Calculated band structure of anodically grown oxides on (xxx0) and (0001) grains at a potential of U2 V (SHE). The oxides are formed at the same upper potential of UOx 4 V with dOx 10 and 6.5 nm. Dielectric permittivity, donor density and flat band potential were taken from measurements using photoresist microelectrodes.

154

A. Michaelis and S. Kudelka

character of the passive film, the potential drop over the metal electrolyte interface is mainly located in the space charge layer. The extension of this space charge layer and the corresponding distribution and availability of energetic states determine the rate of electron transfer. The electron exchange takes place over the conduction band. For the low density packed Ti surface (xxx0), the oxide growth is a linear function of the potential. The donor density is comparably low, 1019 cm3. Anodic reactions such as the oxygen evolution reaction are blocked. On the close packed (0001) surface, the films grow more slowly above 2 V, are highly doped, 1020 cm3, and allow anodic reactions due to a thinner space charge layer and reduced tunnel barrier. High growth rates reduce the effect of texture and show a more isolating character. This explains the systematic variation of the film properties such as layer thickness and defect state concentration with the substrate orientation. The micro electrochemically determined electronic and optical film parameters were confirmed by photocurrent measurements. The anodic TiO2 films were found to be amorphous with mobility gap energies close to those of the TiO2 modification anatase. The TiO2 films were modified at high local resolution by means of focused UV-laser illumination. Both a thickening (writing) or thinning (erasing) of the films could be realized depending on the applied potential which therefore could be used to control the layer thickness. This allows preparation of unique structures such as semiconducting films with thickness gradients. Pitting could be induced as well. We conclude that the presented experimental approach can significantly contribute to the fast evolving micro-system technology in terms of both monitoring and preparation of microstructures.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

M. West (ed.): Lasers in Chemistry, Elsevier, Amsterdam (1977) K. A. Friedrich and G. L. Richmond: Ber. Bunsenges., 97 (1993) 389 P. Dhamelincourt and P. Bisson: Microsc. Acta, 79 (1977) 267 O. Acher, E. Bigan and B. Drevillon: Rev. Sci. Instrum., 60 (1989) 65 R. Kötz: Ber. Bunsenges., 97 (1993) 427 White: Anal. Chem., 62(11) (1990) 1133 A. M. Butler: J. Electrochem. Soc., 130 (1983) 2358 A. M. Butler: J. Electrochem. Soc., 131 (1984) 2185 J. W. Schultze, K. Bade and A. Michaelis: Ber. Bunsenges. Phys. Chem., 95 (1991) 1349 K. Leitner and J. W. Schultze: Ber. Bunsenges. Phys. Chem., 92 (1988) 181 J. W. Schultze and J. Thietke: Electrochim. Acta, 34 (1989) 1769 K. Bade, O. Karstens, A. Michaelis and J. W. Schultze: Faraday Discussion, 94/6 (1992) A. Michaelis and M. Schweinsberg: Thin Solid Films, 313 (1998) 756 A. Michaelis and J. W. Schultze: Thin Solid Films, 274 (1996) 82 R. M. A. Azzam and N. M. Bashara: Ellipsometry and Polarized Light, North Holland Publ. Company (1977)

Application of optical methods and lasers 155 16. M. Erman and J. B. Theeten: J. Appl. Phys., 60(3) (1986) 859 17. H. S. Carlslav and J. C. Jaeger: Conduction of Heat in Solids, New York, Oxford United Press (1959) 18. F. Hillenkamp: In “Microbeam Analysis” (1989) 277 19. H. Yoneyama, K. Kawai and S. Kuwabata: J. Electrochem. Soc., 128 (1988) 1699 20. F. Friedrich and C. J. Raub: Metalloberfläche, 38 (1984) 237 21. H. R. Khan, M. Kittel and C. J. Raub: Surf. Coat. Tech., 35 (1988) 215 22. V. A. Benderski and G. I. Velichko: J. Electroanal. Chem., 140 (1982) 1 23. F. Di. Quarto, S. Piazza and C Sunseri: Ber. Bunsenges. Phys. Chem., 91 (1987) 437 24. F. Di. Quarto, S. Piazza R. D Agostino and C. Sunseri: Electrochim Acta, 34 (1989) 321 25. F. Di. Quarto, S. Piazza and C. Sunseri: J. Chem. Soc. Faraday Trans. I, 85 (1989), 3309 26. J. P. H. Sukamto, W. H. Smyrl, C. S. McMillan and M.R. Kozlowski: J. Electrochem. Soc., 130 (1992) 265 27. J. P. H. Sukamto, C. S. McMillan and W. H. Smyrl: Electrochim. Acta, 38 (1993) 15 28. A. Michaelis and J. W. Schultze: Applied Surface Science, 106 (1996) 483 29. W. W. Gärtner: Phys. Rev., 116 (1959) 84 30. F. Di. Quarto, S. Piazza and C. Sunseri: Electrochim. Acta, 38 (1993) 29 31. D. M. Pai, R. C. Enck: Phys. Rev. B, 11 5136 32. St. Kudelka, A. Michaelis and J. W. Schultze: Electrochim. Acta, 40 (1995) 33. J. W. Schultze, K. Bade and A. Michaelis: Ber. Bunsenges. Phys. Chem., 95 (1991) 1349 34. S. Kudelka, A. Michaelis and J. W. Schultze: Ber. Bunsenges. Phys. Chem., 99 (1995) 1020 35. A. J. Bard, F. R. Fan and M. V. Mirkin: Electroanalytical Chemistry, (1994) 244–273 36. S. Kudelka, Dissertation 1996, Verlag Shaker, ISBN# 3-8265-2324-5 37. P. O. Brunn, V. I. Fabrikant and T. S. Sankar: Quart. J. Mech. Appl. Math., 37 (1984) 311 38. J. W. Schultze, S. Kudelka, B. Davepon and R. Krumm: In “Passivity and its Breakdown”, Electrochemical Society Proceedings, Vol. 97–26, (1997), pp. 725–739 39. S. Kudelka and J. W. Schultze: Electrochim. Acta, 42 (1997) 2817–2825 40. S. Kudelka, A. Michaelis, E. Smailos and J. W. Schultze: Metalloberflaeche, 50 (1996) 5, 369–378 41. A. Michaelis, J. L. Delplancke and J. W. Schultze: Materials Science Forum, Vol. 185–188 (1995) 471–480 42. S. Kudelka, A. Michaelis and J. W. Schultze: Electrochimica Acta, 41 (1996) 863–870

6

Nucleation and growth in microsystem technology Georgi Staikov

6.1

Introduction

Nucleation and growth of new phases play an important role in many electrochemical processes used in microsystem technology.1–4 Typical examples are electrodeposition and electroless deposition of metals and alloys which have a wide application in microelectronics and in the so-called “LIGA-technique”.5–7 An exciting new application of electrodeposition which has recently become of significant importance for microelectronics is the fabrication of copper interconnections on integrated circuit (IC) chips based on an original electroplating technology developed by IBM.8,9 Beside metal deposition, anodic deposition of oxides and conducting polymers10,11 is also of interest due to the various possibilities for application of these materials in different microelectronic devices and microelectro-mechanical systems. Localized electrochemical deposition is successfully applied in microsystem technology for both creation of positive microstructures and filling of negative microstructures. The localization of deposition processes is realized using traditional photolithographic techniques, microdroplet techniques, local probe techniques, localized laser signals, etc.12,13 A variety of novel materials and structures with unconventional mechanical, chemical, electrical and magnetic properties can be deposited applying different polarization routines. Many deposition reactions in electrochemical systems are used for an industrial production of functional coatings with well defined microstructure (grain size, grain distribution and orientation) and can also be considered as microstructuring processes. Deposition processes taking place in electrochemical systems include a number of phase formation phenomena. The understanding of nucleation and growth processes, which control the overall deposition kinetics and the evolution of structure and morphology of deposits, is the key to successful application of electrochemical deposition. The purpose of this chapter is to discuss the role of nucleation and growth in different deposition techniques used in electrochemical microsystem technology. The fundamentals of phase formation processes are reviewed in Section 6.2. Section 6.3 presents typical mechanisms of nucleation and growth of metals, oxides and conducting polymers in electrochemical systems. In Section 6.4 some specific features of deposition in microstructures are discussed on the basis of experimental results obtained in selected systems.

Nucleation and growth 157

6.2

Fundamentals of phase formation

6.2.1

Thermodynamic aspects

Formation of condensed phases in supersaturated systems is a first order phase transition involving nucleation phenomena. The initial stages of this process are characterized by the appearance of small clusters of monomer particles (atoms, ions or molecules) of the new phase.14 The Gibbs energy of formation of a cluster consisting of N monomer particles can be generally described by the equation G(N) (N)

N ␮ ˜ NA

(1)

where NA is the Avogadro constant and (N) represents the energy contribution connected with the creation of new interface boundaries and with the deviation of the structure and/or interparticle interactions of the small cluster from the condensed bulk phase. The driving force of the phase transition, the so-called supersaturation is denoted as ␮ ˜ . In electrochemical systems, charged monomer particles are involved in the process and the supersaturation is given by ␮ ˜ ␮ ˜i(El)␮˜i(CP) where ␮ ˜ i(El) and ␮ ˜i(CP) are the electrochemical potentials of the charged species i in the parent electrolyte phase (El) and in the condensed bulk phase (CP), respectively. If uncharged monomer particles take part in the process, as in the case of phase formation from a gas phase, the supersaturation is expressed as a difference of the chemical potentials ␮. In a supersaturated system (␮ ˜ 0) the first term in Eqn (1), (N/NA)/␮ ˜, decreases linearly with N, whereas the second term, (N), is in the simple case proportional to the cluster surface and increases with N. For small N the second term (N) dominates and the cluster formation and growth can occur only as a result of thermal fluctuations in the sense of statistical thermodynamics. For sufficiently large N the first term in Eqn (1) is dominating and the clusters can grow spontaneously. Thus, for a given supersaturation G(N) displays a maximum for some critical value of N Ncrit. The corresponding critical cluster with the size Ncrit is in metastable equilibrium with the ambient phase, that is, the probability for the further growth of the critical cluster is equal to the probability for its decay. The concept of critical cluster called nucleus of the new phase is an important concept in the nucleation theory. The energy barrier for nucleation Gcrit and the nucleus size Ncrit are determined by the maximum condition, [dG(N)/dN]N Ncrit 0, and can be calculated from Eqn (1) applying two different theoretical approaches.14–16 In the classical approach (N ) is derived using macroscopic thermodynamic quantities such as specific surface energy, specific adhesion energy, specific edge energy, etc. The clusters are assumed to satisfy the condition for the equilibrium form, that is, minimal total surface energy at constant volume. The equilibrium form of a crystallite on a foreign substrate can be described by the distances hi of the corresponding crystal faces i from the so-called Wulff point (Figure 6.1) using the Gibbs–Wulff–Kaischew theorem:14 hj* hi const ␴i (␴j ␤)

(2)

158

G. Staikov

(a)

(b) hj hj d = hj +hj *

Wulff point hj *

Substrate

d = hj –hj*

hj*

Wulff point

Substrate

Figure 6.1 Schematic representation of the equilibrium form of a crystallite on a foreign substrate showing the influence of the adhesion energy ␤. (a) 0 ␤ ␴j ; (b) ␴j ␤ 2␴j.

In this equation ␴i are the specific surface energies of crystal faces, ␴j is the specific surface energy of the crystal face parallel to the substrate surface, hj* is the distance to the contact crystal/substrate interface, and ␤ is the specific adhesion energy defined by ␤ ␴j ␴sub ␴j*

(3)

where ␴sub and ␴j* represent the specific surface energy of the substrate and the specific energy of the crystal/substrate interface, respectively. Then the total surface energy of a three-dimensional (3D) crystallite staying in contact with a foreign substrate can be described by 3D

兺 ␴ A A * (␴ ␤)

i ⴝ j*

i

i

j

j

(4)

where Ai and Aj* denote the surface areas of the crystal faces i and the crystal/ substrate interface j*. Introducing an average specific surface energy ␴ 3D /兺iAi one obtains from Eqns (1) and (4) for the energy of formation Gcrit and the size Ncrit of a 3D nucleus 4BVm2␴ 3 ˜ )2 27(␮

(5)

8BNAVm2␴ 3 ˜ )3 27(␮

(6)

Gcrit Ncrit

where Vm is the molar volume of the condensed new phase and B is a factor depending on the geometrical form of the nucleus (B 62␲ for a sphere and B 63 for a cube). Equation (6) corresponds to the so-called Gibbs–Thomson

Nucleation and growth 159 equation, which gives the stability ranges of the condensed, infinitely large, 3D bulk phase and the small (finite size) 3D clusters. The distances hi in Eqn (2) are given by the moduli of radius vectors normal to the corresponding faces. Depending on the specific adhesion energy ␤, that is on the crystal–substrate interaction, the distance to the crystal/substrate interface hj* can obtain either positive or negative values. The thickness of crystallite d decreases with increasing adhesion energy ␤ (Figure 6.1) and can be described by d hj hj*

hj(2␴j ␤) ␴j

(7)

At ␤ → 2␴j, however, d reaches atomic dimensions and the 3D treatment based on surface energies is no longer valid. Moreover, in systems with adhesion energy larger than the cohesion energy (␤2␴j), a formation of condensed twodimensional (2D) phases (monolayers) involving 2D nucleation can occur even at undersaturations (␮˜ 0) with respect to the condensed 3D phase.13,14 A special case of 2D nucleation can be obtained in the growth of singular crystal faces. In this case the equilibrium form of a 2D cluster on native substrate is defined in full analogy to the 3D case and the total surface energy is given by the periphery energy. Equations (5) and (6), however, can not be applied for nuclei consisting of small number of monomer particles (atoms) because the use of macroscopic quantities such as specific surface energy and specific edge energy loses its physical meaning. In this case the so-called atomistic approach14–16 can be used and (N) in Eqn (1) can be calculated from (N) N␾kink

N

兺␾

i1

(8)

i

where ␾kink is the binding energy in the so-called kink-site position and 兺i1 ␾i is the dissociation energy of the cluster which includes the interaction energies between the monomer particles and the interaction energy with the substrate. In fact, (N) represents the difference between the interaction energies of N monomer particles in the bulk phase and in the cluster. The calculations show that in contrast to the classical nucleation model the nucleus size Ncrit determined by the maximum of G(N ) remains constant in relatively large supersaturation intervals where Gcrit is given by N

Gcrit

Ncrit Ncrit ␮ ˜ Ncrit␾kink 兺 ␾i NA i1

(9)

An important advantage of the atomistic approach is the possibility to include in the calculations structures and interparticle (interatomic) interactions which differ for the clusters and for the bulk crystal.

160

G. Staikov

6.2.2

Kinetic aspects

In the nucleation theory the formation of clusters with different sizes is described by a set of bimolecular reactions of the type ␻

att, N1 N N1 ⇔ ␻ det, N

␻att, N ⇔ ␻det, N1

N1

(10)

The rates of transition of a N-sized cluster to a higher (N1) or a lower (N1) size class are determined by the corresponding probabilities (frequencies) of attachment ␻att, N or detachment ␻det, N of one monomer particle to or from the cluster. The time dependence of the concentration of differently sized clusters is described by a set of differential equations. The rate of nucleation is defined by the net flux of clusters through the critical size (the nucleus size). The solution of the problem gives generally for the steady state nucleation rate14–16

冢

J ␻att, NcritZo ⌫ exp

Gcrit kT

冣

(11)

where ␻att, Ncrit is the frequency of attachment of monomer particles to the nucleus (critically sized cluster) and Zo can be considered in the case of heterogeneous nucleation as a number of nucleation sites of the substrate. The so-called nonequilibrium or Zeldovich factor ⌫ is a weak function of the supersaturation and has usually an order of magnitude of 102. For sufficiently large nuclei of the new phase, Gcrit in Eqn (11) can be taken from Eqn (5). For small critical clusters, however, the atomistic approach should be used and Eqns (12) and (14) predict a linear dependence of ln J on ␮˜ in supersaturation intervals where Ncrit remains constant. The number of monomer particles in the nucleus Ncrit can be determined with a sufficient accuracy using the relation14,17 d ln J Ncrit RT d ␮˜

(12)

which follows from Eqns (5), (6), (9) and (11). Equation (12) neglects the supersaturation dependence of the pre-exponential factors ␻att,Ncrit and ⌫ in Eqn (11) but is general and can be applied without any restrictions for the size and form of the nucleus. Because of its fluctuation and stochastic character nucleation is a random phenomenon. At constant supersaturation the process can be considered as a sequence of independent random nucleation events characterized by a mean ␶nuc which is directly related to the nucleation rate given by Eqn (11). nucleation time In the case of nucleation on a substrate the mean nucleation time is given by ␶nuc1/JA, where J (cm2 s1) is the nucleation rate and A (cm2) is the substrate area. A characteristic feature of all nucleation processes is the existence of a critical supersaturation ␮ ˜ crit. The nucleation rate J is practically zero at

Nucleation and growth 161 ˜ ␮ ˜ crit) and supersaturations ␮˜ lower than the critical supersaturation (␮ ˜ crit can rises exponentially at (␮˜ ␮ ˜ crit). It should be noted, however, that ␮ not be considered as a characteristic constant for a given system because its value depends on the time scale of observation and on the experimental technique used to detect the nucleation. Nucleation on real substrates usually occurs on preferred nucleation sites. The process of nucleation on a substrate with a fixed number Zo of randomly distributed and equally active nucleation sites can be described by a first order low:18,19 Znuc(t) Zo[1 exp(fnuc t)]

(13)

where Znuc(t) denotes the number of nuclei, that is, the number of nucleation sites converted to nuclei at time t and fnuc is the frequency of nucleation per nucleation site. If fnuc is very large, all nucleation sites Zo are converted to nuclei virtually instantaneously. This case is known as instantaneous nucleation and the corresponding nucleation rate can be expressed by dZnuc/dt Zo␦(t) with ␦(t) being the Dirac delta function (␦(t) 1 for t 0 and ␦(t) 0 for t 0). For relatively low values of fnuc the nucleation sites are converted to nuclei progressively with time and the process is called progressive nucleation characterized by a nucleation rate dZnuc/dt Zo fnuc J. The rate of growth of a crystal depends on its size, degree of perfection and surface structure. The incorporation of the monomer particles in the crystal lattice occurs at the so-called “kink positions” (half crystal positions) on monatomic steps.14 Two mechanisms of incorporation are generally possible: (i) a mechanism of direct incorporation (direct transfer from the ambient phase to a kink) and (ii) a surface diffusion mechanism, that is, formation of adatoms (adsorbed monomer particles) followed by surface diffusion to the steps. In an advanced stage of growth the crystal surface consists usually of singular crystal faces, so that the growth kinetics is determined mainly by the mechanism of growth of these faces. The normal growth rate vn of a singular crystal face is proportional to the frequency of generation of monolayers fmon and in the case of – electrocrystallization is related to the average current density i by vn hmon fmon

Vm 兩兩 zF i

(14)

where hmon is the thickness of a monolayer. On a quasi-perfect (screw dislocation free) singular crystal face the generation of new monolayers takes place by 2D nucleation. The growth kinetics is determined by the mean nucleation time ␶nuc (JA)1 , the surface area A of the crystal face, and the propagation time defined by ␶p A1/2v1 p where vp is the prop␶nuc ␶p the growth occurs according to the agation rate of the monolayers.14 At mononuclear layer-by-layer growth mechanism (Figure 6.2(a)) and the frequency of generation of monolayers in Eqn (14) is given by the nucleation frequency

162

G. Staikov (a)

(b)

(c)

Figure 6.2 Growth mechanisms of singular crystal faces. (a) mononuclear layer-by-layer growth; (b) multinuclear multilayer growth; (c) spiral growth.

␶nuc ␶p operates the multinuclear multilayer growth mechanism fmon JA. At (Figure 6.2(b)) characterized by fmon J1/3v2/3 p . If a singular crystal face is intersected by screw dislocations the growth occurs according to the so-called spiral growth mechanism (Figure 6.2(c)). The frequency of generation of new monolayers fmon is given by the frequency of rotation of the spiral.14 At relatively high supersaturations the formation of a new phase on foreign substrates occurs by multiple nucleation and growth. In the case of electrochemical phase formation the current density ifree resulting from the nucleation and growth of isolated (independent) islands without taking into account overlapping effects and nucleation on top of already formed islands is given by ifree(t) I1(t) * (dZnuc/dt)

(15)

where I1(t) is the growth current of an isolated island, (*) represents the convolution product and (dZnuc/dt) is the nucleation rate defined by Eqn (13). The growth current I1(t) is related to the normal vn and lateral vl growth rates, which determine the preferred growth direction and the growth mode (1D, 2D, 3D) of the islands. The ifree t relationship depends on the rate-determining step of deposition (charge transfer, diffusion, migration) and on the nucleation type (instantaneous or progressive nucleation). Typical ifree(t) expressions derived on the basis of Eqn (15) are summarized in Table 6.1.18–21 Conditions corresponding to Eqn (15), however, can be realized only in the initial stages of deposition, that is, for relatively short times. For longer times, effects connected with the interference and overlapping of growing islands have also to be taken into account. Assuming a random spatial distribution of the nuclei the actual fractional surface coverage S(t) can be described by the well-known Avrami equation S(t) 1 exp[Sex(t)]

(16)

where Sex(t) represents the so-called “extended” fractional surface coverage corresponding to an island growth without overlapping. This approach was used by various authors to account for the interference of growing islands and/or their fields occurring at longer times.18–24 Considering the growing centers as right circular cones, the deposition kinetics in the cases of progressive and instantaneous

Nucleation and growth 163 Table 6.1 Initial current density, |ifree|, as a function of time, t, for different growth modes, rate-determining steps and nucleation types.18–21 The symbols used are explained at the bottom of the table* Growth mode

Growth shape

Rate-determining step

Nucleation type

|ifree(t)|

1D

Needles

Charge transfer

Instantaneous

␲

Progressive 2D

Discs

Charge transfer

Instantaneous Progressive

Cylindrical diffusion

Instantaneous Progressive

Cones

Charge transfer

Instantaneous Progressive

3D

Charge transfer

Instantaneous Progressive

Hemispheres

Diffusion

Instantaneous Progressive

Migration

Instantaneous Progressive

zF 2 r Zv Vm n o n zF ␲ rn2 J vnt Vm

zF h Z v2 t Vm o l zF ␲ h J vl2 t 2 Vm 2␲

zF h 21 D Zo Vm zF ␲ h 21 D J t Vm ␲

zF Z v v2 t 2 Vm o n l ␲ zF 2 3 v t 3 Vm Jvn l

␲

zF Z v3 t2 Vm o 2 zF 3 3 ␲ Jv t 3 Vm

2␲

zF 3 3/2 D Zo t1/2 Vm 2 2 zF 3 3/2 3/2 ␲ D Jt 3 Vm 2

␲

zF 3 3/2 ␬ Zo t1/2 Vm 3 2 zF 3 3/2 3/2 ␲ ␬ Jt 3 Vm 3

␲

* Vm is the molar volume, Zo is the number of nucleation sites; J is the nucleation rate; vn and v1 denote the normal and lateral growth rate of crystallites; v is the growth rate by hemispherical growth (vvnv1); rn is the radius of the growing needles; h is the height of growing discs; 1, 2 and 3 are terms defined in Refs. 19, 21; D is the diffusion coefficient; ␬ is the electrolyte conductivity.

3D nucleation and charge transfer controlled crystal growth can be described by

冤

冣冥

(17)

zF 兩i兩 vn[1 exp(␲Zo v2l t2)] Vm

(18)

冢

␲J v2l 3 zF t 兩i兩 vn 1 exp Vm 3

164

G. Staikov

where vn and vl represent the normal and lateral growth rates, that is, the rates in the directions perpendicular and parallel to the substrate surface.22 The cases of diffusion or migration controlled crystal growth are more complicated due to formation of “exclusion” zones around the growing centers.23,24 6.2.3

Structural aspects

The microstructure of a polycrystalline deposit is determined mainly by the rates of nucleation and normal and lateral growth of the crystallites. Two types of 3D nucleation can be considered by the deposition of a new crystalline phase on foreign substrates: (i) primary nucleation taking place on the foreign substrate in the initial stages (Figure 6.3(a)) and (ii) secondary nucleation occurring on top of already formed crystallites (Figure 6.3(b)). At relatively low supersaturations the 3D nucleation stops after the first layer of crystallites is deposited on the substrate. The further deposition proceeds only by the growth of the crystallites initially formed by primary nucleation and a near-columnar microstructure develops (Figure 6.4(a)). The mean diameter of the columns is determined by the rates of nucleation and the normal and lateral growth of the crystallites. If the primary nucleation occurs instantaneously on a large number of uniformly distributed nucleation sites and the subsequent growth is enhanced in the normal to the substrate direction the deposit obtains a fiber-like structure. The growth of individual columns and/or fibers in the structure in Figure 6.4(a) occurs by formation of layers generated by a 2D nucleation and/or screw-dislocation mechanism (cf. Figure 6.2). In many cases the growth involves a “bunching” of monatomic steps leading to formation of macrosteps. At high supersaturations and particularly in the presence of strong inhibitors the deposition process involves also a secondary 3D nucleation on top of already deposited crystallites (cf. Figure 6.3(b)). Generally, an equiaxed microstructure consisting of randomly oriented crystallites with an average grain size determined by the nucleation rate is formed in this case (Figure 6.4(b)). If the crystal growth (a)

Primary nucleation Vn VI Substrate

(b)

Secondary nucleation

Substrate

Figure 6.3 (a) Primary and (b) secondary 3D nucleation.

Nucleation and growth 165 (a)

(b)

Substrate

Substrate

Figure 6.4 Typical deposit microstructures. (a) columnar microstructure; (b) equiaxed microstructure.

is diffusion limited isolated needle-like deposit structures (dendrites, whiskers) can be obtained. At even higher supersaturations and strong inhibition a formation of 2D nuclei in a twinning position becomes also possible and can lead to the formation of different multi-twinned structures. The structure, orientation and adhesion of polycrystalline deposits depend strongly on the chemical, crystallographic and morphological properties of the substrate surface. In the case of a weak deposit–substrate interaction (weak adhesion) the structure and orientation of the deposit only weakly depend on the substrate surface structure. In systems with a strong deposit–substrate interaction, however, the epitaxy of the initially formed crystallites as well as the structure, energy and orientation of the compact deposit strongly depend on the substrate surface structure and the deposit–substrate lattice misfit. In absence of deposit– substrate misfit the crystallites formed by primary nucleation grow epitaxially in registry with the substrate. The final orientation and the microstructure of the compact deposit, however, are determined also by the conditions of subsequent deposition. The presence of significant deposit–substrate lattice misfit usually induces considerable strain in the initially formed crystallites and leads to preferred azimuthal orientations. After reaching a certain deposit thickness the lattice misfit is removed by a formation of so-called misfit or interfacial dislocations.25

6.3

Deposition processes in electrochemical systems

6.3.1

Electrodeposition of metals

The overall reaction describing the electrodeposition or dissolution of a bulk metal Me can be expressed by z Mez solv(El) ⇔ Me (Me)

(19)

where Mez solv(El) are the solvated metal ions in the electrolyte phase (El) and Mez(Me) represents the metal ions in the 3D metal phase (Me) which are coupled to the electrons e in the Me-crystal lattice are Mez ze Me. At the so-called Nernst potential EMe/Mez the electrochemical reaction (19) is in

166

G. Staikov

˜ (EI) ˜ (Me) z ␮ z Me Me equilibrium which is defined by ␮ . The deviation from equilibrium z ␮ z is related to the actual electrode potential E by ␮ ˜ ␮ ˜Me(EI) ˜ (Me) Me ␮ ˜ zF(E EMe/Mez) zF␩

(20)

where ␩ is the overpotential. If the kinetics of metal deposition and dissolution are not controlled by charge transfer, ion diffusion and migration or additional chemical reaction steps the overpotential represents the so-called crystallization overpotential and Eqn (20) defines the supersaturation (␮˜ 0 for ␩ 0) or the undersaturation (␮˜ 0 for ␩ 0) in the system.14 The first step of electrodeposition of metals on foreign substrates is the formation of single adatoms. Thus, it is very suitable to express the supersaturation as a difference between the chemical potentials of adatoms at the actual electrode potential and the equilibrium potential of the 3D Me bulk phase ␮Meads ␮0,Meads zF(E EMe/Mez).14 As already discussed a critical supersaturation, that is, a critical adatom concentration, is needed to initiate a nucleation process occurring with a measurable rate. In the initial stages of electrodeposition of metals on foreign substrates the overall deposition mechanism depends strongly on the deposit–substrate interaction.13,14 In systems with weak deposit–substrate interaction (weak adhesion) the metal deposition starts at supersaturations (␮˜ 0) in the overpotential deposition (OPD) range E EMe/Mez+ following a 3D island growth mode (Volmer–Weber growth mode). In the case of a strong deposit– substrate interaction (strong adhesion), however, the metal deposition can start even at undersaturations (␮˜ 0) in the so-called underpotential deposition (UPD) range E EMe/Mez+ with a formation of metal overlayers (2D metal phases). Depending on the deposit–substrate lattice misfit the deposition at low overpotentials usually follows a Frank-van der Merwe growth mode (layerby-layer growth mode) or a Stranski–Krastanov growth mode (3D Me island formation and growth on top of predeposited 2D Me overlayers). These deposition mechanisms and the involved phase formation phenomena have been extensively studied in various systems using single crystals as foreign substrates.14 New important information on the role of substrate surface inhomogeneities in the deposition process as well as on the structure and morphology of metal phases was obtained at an atomic level combining electrochemical methods with modern in situ SPM (scanning probe microscopy) techniques.14 Experimental results show, that in systems with a strong deposit–substrate interaction substrate surface inhomogeneities can induce a stepwise formation of metal phases with different dimensionality.13 As a typical example Figure 6.5 shows an in situ AFM line scan image of the initial stages of Pb electrodeposition on stepped Ag(111) single crystal surfaces.26 In this case the deposition starts in the UPD range at E E1D Pb with a decoration of monatomic steps of the substrate representing an 1D Pb phase formation. This process is followed at E E2D Pb by a formation of a condensed Pb monolayer (condensed 2D Pb phase) which acts as a precursor for the nucleation and growth of 3D Me bulk phase occurring in the OPD range E E3D Pb ≡ EMe/Mez+.

Nucleation and growth 167 E2D Pb

(a)

E1D Pb

50 40 30

D2

i (µA cm–2)

20 10 D3

D1

0 A3

A1

–10 –20 A2

–30 –40

100 200 E–EPb/Pb2+ (mV)

(b)

300

Time

Terraces

2D Pb

1D Pb 1s E2D Pb E1D Pb Potential

0 nm

50 nm

Figure 6.5 Formation of low-dimensional metal phases on a stepped Ag(111) substrate in the system Ag(111)/5mM Pb(ClO4)2 0.5M NaClO4 5mM HClO4. (a) cyclic voltammogram in the UPD range (scan rate |dE/dt| 1mV/s); (b) in situ AFM line scan image showing the step decoration by a 1D Pb phase and the formation of a condensed 2D Pb monolayer on substrate terraces (the applied polarization routine is indicated).

Electronic properties of the foreign substrate can also influence significantly the kinetics and mechanism of metal electrodeposition. In the case of deposition on foreign metal substrates the substrate surface is a continuum of electronic states and thus, sufficient supersaturations for nucleation of the Me bulk phase are usually reached by application of relatively low cathodic overpotentials. The situation, however, becomes more complicated in the case of electrodeposition of

168

G. Staikov (a)

0.15

i (mA cm–2)

0.1 0.05 0

–0.05 ECd/Cd2+ << Efb

–0.01 –0.15 –60 –40 –20

(b)

0.5

–0.6

0

–0.4 Efb

20 40 (mV) /V –0.2

60

0

80

0.2

100

0.4

ECu/Cu2+

i (mA cm–2)

0.0

–0.5 –1.0

–1.5 –0.5

ECu/Cu2+ >>Efb –0.3

–0.1

0.1 0.3 E/V (SHE)

0.5

0.7

Figure 6.6 Electrodeposition of Cd and Cu on H-terminated n-Si(111). (a) cyclic voltammogram in the system n-Si(111)/5mM CdSO4 0.5M Na2SO4 5mM H2SO4 (scan rate |dE/dt| 1mV/s); (b) cyclic voltammogram in the system n-Si(111)/10mM CuSO4 0.5M H2SO4 (scan rate |dE/dt| 5mV/s).

metals on semiconductor substrates. The influence of the space charge layer in the semiconductor has to be taken into account in this case.27–29 For a not very high doped n-type semiconductor, the surface concentration of electrons and the band bending at the actual electrode potential play an important role in the electrodeposition process. Neglecting the influence of surface states, the mechanism of metal deposition in this case depends strongly on the relative position of the equilibrium potential EMe/Mez+ with respect to the flat band potential Efb of the semiconductor substrate. If EMe/Mez+ << Efb accumulation layer is formed at E EMe/Mez+ and the deposition of the Me bulk phase usually occurs at relatively small cathodic overpotentials, such as on metal substrates. If the equilibrium potential EMe/Mez+, however, is located in the band gap and is more positive than the flat band potential Efb (EMe/Mez+ >> Efb) depletion layer is formed in the n-type semiconductor at E EMe/Mez+. Thus, electrode potentials more negative

Nucleation and growth 169 than the flat band potential (E Ef b) are needed in this case to reach a sufficient surface electron density (accumulation conditions) and electrodeposition of metals in such systems occurs at high cathodic overpotentials. This specific behavior of metal electrodeposition on semiconducting substrates is illustrated in Figure 6.6 showing the deposition of Cd and Cu on H-terminated n-Si(111) substrates.29 If a Schottky barrier is formed at the metal–semiconductor interface, the metal deposit can not be stripped anodically. The system n-Si(111)/Cu2 shown in Figure 6.6b is an example of this behavior. Generally, the electronic properties of metal/semiconductor interface can influence not only the process of dissolution but also the growth kinetics and structure of metal deposits. The mechanisms of crystal growth shown schematically in Figure 6.2 have been demonstrated and verified quantitatively in a very convincing way in the case of electrocrystallization of silver on “quasi-perfect” silver single crystal faces (free of screw dislocations or with extremely low dislocation density) prepared according to the so-called capillary technique.14 Experimental results indicated that in this case the surface diffusion plays a subordinate role and the incorporation of Ag in the crystal lattice occurs predominantly by a direct ion transfer at kink sites on monatomic steps. The processes of nucleation in electrodeposition of metals are studied by different experimental pulse techniques based on the specifics of nucleation kinetics, in particular on the existence of a critical cathodic overpotential (critical supersaturation).14 In the so-called double pulse technique the nucleation is initiated by a short cathodic nucleation pulse with a sufficiently high amplitude exceeding the critical overpotential. The formed nuclei then continue to grow during a second growth pulse with an amplitude much lower than the critical overpotential. The number of nuclei initiated by the nucleation pulse corresponds to the number of growing crystallites during the second pulse and can be detected microscopically or by analysis of the current transient corresponding to the growth process. The nucleation rate is determined by the density of nuclei and the duration of the nucleation pulse. Another technique for estimation of nucleation rate is based on a direct analysis of initial current transients in the case of progressive nucleation. The size of the critical cluster (nucleus) Ncrit can be determined from the nucleation rate-overpotential dependence using Eqn (12). Table 6.2 summarizes experimental results for Ncrit obtained in various systems with metal and semiconductor substrates.29–39 The relatively small number of atoms Ncrit generally found in the case of metal deposition on foreign substrates indicates that in this case the nucleation process has to be described by the so-called small cluster or atomistic model.14–16 Nucleation and crystal growth are the main processes determining the structure and properties of compact polycrystalline metal deposits obtained by elecrtroplating. A phenomenological classification of the general texture types of electrodeposits has been proposed by Fischer.40 The field-oriented texture (FT) type and the field-oriented isolation (FI) type structures in this classification are formed by an electric-field enhanced normal growth. The so-called basis-oriented reproduction (BR) type and unoriented dispersion (UD) type correspond to the microstructures

170

G. Staikov

Table 6.2 Ncrit-values for electrochemical nucleation of Cd, Cu, Pb and Tl on various substrates in different overpotential (supersaturation) ranges Metal

Substrate

Overpotential range (mV)

Supersaturation range (kJ/mol)

Cd

n-Si(111) Pt

15 to 21 22 to 30 30 to 38

2.89 to 4.05 4.25 to 5.79 5.79 to 7.33

Cu

n-Si(111) n-GaAs(100)

W

474 to 520 510 to 570 631 to 881 71 to 82 40 to 54 54 to 82 82 to 140 22 to 32 50 to 100 40 to 55

91.47 to 100.34 98.41 to 109.99 121.76 to 170.01 13.70 to 15.82 7.72 to 10.42 10.42 to 15.82 15.82 to 27.02 4.25 to 6.17 9.65 to 19.30 7.72 to 10.61

1 to 0 1 to 0 0 2 4 1 0 11 0 4

29 31 32 33 34 34 34 35 36 35

Pb

n-Si(111) Ag(111) Ag(100) C HOPG

6 to 10 15 to 19 13 to 18 50 to 300 4 to 7

1.16 to 1.93 2.89 to 3.67 2.51 to 3.47 9.65 to 57.89 0.77 to 1.35

11 11 13 0 11

37 38 38 39 37

Tl

n-Si(111)

3 to 7

0.29 to 0.68

18

29

Au(111) Pd Pt

Ncrit (atoms) 6 2 1

Refs. 29 30 30

shown in Figures 6.4(a) and (b), respectively. Under real conditions, nucleation and growth can be influenced significantly by adsorption of inhibitors, hydrogen evolution, ion complexation, mass transport and electric field. An analysis of different contributions in the total overvoltage, however, is very difficult, so that the plating conditions to obtain a given deposit structure are usually optimized using as parameters the inhibition intensity, the applied current density, the diffusion current density and the exchange current density. A very useful correlation between these parameters and the Fischer’s main structure types of metal electrodeposits has been introduced by Winand considering the large number of experimental results obtained in different systems.41 6.3.2

Electroless metal deposition

The processes of electroless metal deposition take place in the absence of an external source of electric current under mixed potential control.42–46 Figure 6.7 shows schematically some mechanisms of electroless metal deposition involving different partial anodic reactions. In the displacement deposition mechanism (Figure 6.7(a)) the partial anodic oxidation reaction is a dissolution of the substrate leading to an apparent replacement of substrate surface atoms (S) with deposited metal atoms (Me). Mechanisms presented in Figures 6.7(b) and (c)

Nucleation and growth 171 (a)

Mez+

Deposited metal Sn+

e–

Substrate (b)

Mez+

Red

e–

Substrate (c)

Ox

Mez+

Red

Ox

e

Substrate

Figure 6.7 Mechanisms of electroless metal deposition. (a) displacement deposition mechanism; (b) deposition induced by a substrate catalyzed redox reaction; (c) autocatalytic deposition.

involve a catalytic oxidation reaction of a reducing agent (Red) in solution. In the first case (Figure 6.7(b)) the redox reaction is catalyzed by the substrate whereas in the second case (Figure 6.7(c)) the deposited metal plays a catalyzing role itself. In the mechanisms shown in Figures 6.7(a) and (b) the substrate (electronic conductor or semiconductor) is involved in the occurring charge transfer processes. Thus, in these cases the process of electroless metal deposition stops after formation of a thin compact metal film covering completely the substrate surface. In contrast, the autocatalytic electroless plating (Figure 6.7(c)) takes place without involvement of the substrate in the charge transfer process and can be successfully applied for deposition of thick uniform metal layers on both electronic conductors and isolators. The latter is an important advantage of this technique over electrodeposition of metals, which can be applied only on electronically conducting substrates. In the following, the term electroless deposition will be used for the autocatalytic plating process.

172

G. Staikov

Generally, the process of electroless metal deposition is described on the basis of the so-called mixed potential concept.42–44 In the simplest case the mixed potential (Emix) is determined by the two main partial electrochemical reactions: the anodic oxidation of the reducing agent and the cathodic reduction of metal ions. In real electroless systems, however, the metal deposition process is more complicated and involves a number of additional reactions and reaction steps. A uniform mechanism has been proposed by van den Meerakker:47 Anodic reactions: Me

RH l R H

(21)

R OH l ROH e

(22)

H H l H2

(23)

H OH l H2O e ¬ (alkaline media)

(24a)

H l H e

(24b)

(acidic media)

Cathodic reactions: Mez ze l Me

(25)

2H2O 2e l H2 2OH ¬ (alkaline media)

(26a)

2H 2e l H2

(26b)

(acidic media)

Along with the main anodic oxidation and cathodic reduction reactions (22) and (25), this mechanism includes a preceding dehydrogenation of the reducing agent RH (21), recombination (23) and/or anodic oxidation (24) of the hydrogen produced by (21) as well as cathodic hydrogen evolution (26). A cathodic co-deposition of phosphorus or boron has also to be taken into account in electroless systems with reducing agents containing these elements. Nucleation and growth phenomena are related to two important characteristics of electroless plating systems: (i) the stability of solutions and (ii) the microstructure of metal deposits. The stability of an electroless deposition solution is directly connected to the possibility and rate of formation of small metal clusters in the solution bulk (homogeneous nucleation) as well as to their further behavior (growth or dissolution). A similar problem has been treated in the case of latent image development in photographic processes by Konstantinov et al.,48 who showed that thermodynamic aspects of stability of small silver clusters in solutions containing silver ions and a Red/Ox system can be discussed on the basis of the Gibbs–Thomson relation (6). This approach has been later applied by Vashkyalis49 to the stability of electroless metal plating solutions. According to Eqns (6) and (20) the size Ncrit of the critical metal cluster, which can grow autocatalytically in an electroless

Nucleation and growth 173 solution, can be expressed by Ncrit

8BNAVm2␴ 3 27z3F3(EMe/Mez ERed/Ox)3

(27)

where EMe/Mez and ERed/Ox represent the equilibrium potentials of the infinitely large Me phase and the Red/Ox system, respectively. As already discussed, the application of Gibbs–Thomson relations (6) and (27) is strictly justified only for large metal clusters. Statistical calculations, however, showed that the Gibbs–Thomson relation can be applied for qualitative estimations even for small clusters consisting of only few atoms.50 An electroless deposition solution is unstable and can decompose spontaneously if metal clusters larger than the critical cluster defined by Eqn (27) form in the solution bulk. The solution stability can be increased by decreasing the difference (EMe/Mez ERed/Ox) and/or inhibiting the autocatalytic growth of metal clusters. Different additives called stabilizers are usually used to prevent the spontaneous decomposition of electroless plating baths.43–46 The microstructure and morphology of electroless metal deposits are determined by nucleation and growth phenomena which occur during the deposition process. Thus, the electroless deposition of metal films with good quality usually requires a continuous control of the plating rate and the bath activity. The properties of the substrate surface are crucial for the initial (primary) nucleation and the adhesion of metal deposit. To initiate the process of electroless metal deposition on dielectric substrates, the substrate surface is modified with small catalytic metal clusters (usually Pd clusters). Therefore, the number of catalytic metal clusters per unit area and their size play an important role in the initial nucleation and growth processes. Information on deposition mechanisms and on the structure and properties of metal deposits in various electroless metal deposition systems can be found in reviews43–46 and in the references given therein. 6.3.3

Anodic oxide formation

Anodic oxides and/or hydroxides are formed by electrochemical surface oxidation (passivation) of a substrate (S) or by electrodeposition according to: mS nH2O ⇔ SmOx(OH)n x (n x)H (n x)e

(28)

mMez nH2O ⇔ MemOx(OH)n x (n x)H (n x mz)e

(29)

The mechanism and kinetics of these processes are complex and depend strongly on the substrate nature, electrolyte composition and electrode potential, as well as on the electronic and ionic conductivity of the oxide phase.51–54 The first step in both processes is usually the adsorption of hydroxyl ions OH on the substrate. In the case of surface oxidation (28) oxide molecules are formed by place

174

G. Staikov

exchange reactions with the substrate, whereas in the case of oxide electrodeposition (29), the formation of oxide molecules takes place after charge transfer of Mez from the electrolyte. In the simple case, the current density iox of oxide formation can be directly measured. In some systems, however, the electrochemical oxide formation is accompanied by other electrochemical processes such as corrosion and oxygen evolution, which have to be taken into account by the analysis of experimental results. Additional important information about the nature and thickness of oxide layers can be obtained by capacitance measurements. Kinetic aspects of anodic oxide layer formation and involved nucleation and growth phenomena have been extensively studied by Schultze et al.54–57 in various electrochemical systems. Two main groups of systems have been distinguished on the basis of experimental results obtained:54 (i) autocatalyzing systems characterized by (dlgiox/dlgt) 0 and (ii) self-inhibiting systems characterized by (dlgiox/dlgt) 0. Typical examples illustrating this electrochemical behavior are shown in Figure 6.8. The potential drop within the growing oxide phase is an important factor determining the difference between lateral and normal growth rate. In the case of formation of conducting oxide phases the lateral and normal growth rate are usually comparable and the initial kinetics can be described by nucleation and growth models presented in Table 6.1. Typical examples for such systems are Au/NiOOH54 and Pt/PbO2.58 In systems with insulating anodic oxides, however, the normal growth is strongly inhibited, whereas the lateral growth and nucleation of the oxide phase on the oxide free substrate surface are very fast. Thus, the substrate is covered almost instantaneously by a thin continuos oxide film and a detection of the initial nucleation and growth processes is practically impossible. The further normal growth of the oxide film is usually described by the so-called high field model including an electric field induced ionic transport within the oxide.51–54 According to this model the anodic current density of oxide formation

–2 Au/Au2O3 log i ox (A cm–2)

Al/Al2O3 –3

Pt/PbO2

Au/FeOOH

–4

Au/Au2O3/Au(OH)3

–5

Au/NiOOH

–6 –3

–2

–1

0 log t (s)

1

2

3

Figure 6.8 Kinetics of anodic oxide formation in the systems Au/NiOOH,54 Pt/PbO2,58 Au/FeOOH,57 Au/Au2O3/Au(OH)356 and Al/Al2O3.59

Nucleation and growth 175 iox depends exponentially on the applied electric field strength E and can be expressed by iox io exp( ␤E)

(30)

where io and ␤ are material specific constants. Although the high field model was applied to describe the oxide film growth in many systems, experimental results often showed deviations from Eqn (30). Thus, various new model approaches have been proposed to explain the observed deviations. The established theoretical models and the experimental results obtained in different oxide systems have been reviewed recently by Lohrengel.60 6.3.4

Electrodeposition of electronically conducting polymers

The discovery more than 20 years ago of the unique electronic properties of molecularly doped polyacetylene established the field of electronically conducting polymers and opened the way to research and development of new systems based on these materials.61–63 Electrochemical research groups have been attracted to this field both because of the possibility for preparation of conductive polymers by anodic electropolymerization, and because of their potential application in advanced batteries and microsystem technologies. Extensive experimental studies performed in recent years showed that electropolymerization is an irreversible phase formation process, which includes a number of electrochemical and chemical reaction steps with complex interactions between involved species. The influences of electrode potential, monomer concentration, solvent nature and electrolyte composition on the kinetics and mechanism of electrodeposition of conductive polymers such as polypyrrole, polyaniline and polythiophene have been studied in different systems using cyclic voltammetry and potentiostatic transient techniques.64–74 Experimental results obtained in these studies are usually discussed on the basis of nucleation and growth models for electrochemical phase formation on foreign substrates (Table 6.1) and show that in the rule the deposition process occurs by a similar mechanism. A schematic representation of this mechanism is given in Figure 6.9.70,71 The electropolymerization starts with the oxidation of the monomer species at the electrode surface and a formation of linear (1D) polymer chains, which is thermodynamically favored. After this initial stage the deposition process occurs by an appearance of stable growth centers (Figure 6.9(a)), which grow preferentially in the lateral direction (Figure 6.9(b)). This process can be described by the theoretical models for nucleation and 2D growth (Table 6.1). Generally, at relatively low electrode potentials and/or low monomer concentrations the formation of the first compact polymer film involves progressive nucleation, whereas an instantaneous nucleation takes place at high electrode potentials and/or high monomer concentrations. As pointed out by Tsakova and Schultze,71 however, due to irreversibility of the phase transition, the rate of “nucleation” in these models should be considered as a rate of appearance of growth centers. In

176

G. Staikov (a) Substrate (b) vn vl

(c)

Figure 6.9 Mechanism of electrodeposition of conducting polymers according to Ref. 70 (a) nucleation; (b) preferential growth in lateral direction (vl >> vn); (c) normal growth.

a more advanced stage the polymer growth in the normal direction is accelerated and polymer deposits with a 3D structure are formed (Figure 6.9(c)). The acceleration of the polymer growth rate in this stage is related to a progressive branching of the growing polymer chains, which leads to an increase of the growth sites.70,71 The experimental results discussed above show that the process of electrodeposition of conducting polymers is characterized by some specific features. A better understanding of the correlation between the mechanism of formation and the structure and properties of electrodeposited polymer films requires further developments of theoretical models and experiments on a molecular level.

6.4 6.4.1

Electrochemical microstructuring Deposition in negative microstructures

The most important application of electrochemical deposition techniques in microsystem technologies is the filling of various negative microstructures.1–12 Important parameters characterizing the microstructures are the aspect ratio (depth to width ratio) and electronic properties of involved materials (dielectrics, electronic conductors or semiconductors). Some typical examples of negative microstructures and the specific features of their filling are presented schematically in Figure 6.10. Generally, the filling behavior of a deposition process depends on the localization and rate of involved nucleation and growth phenomena.10–12 The filling of the microstructure shown in Figure 6.10(a) can be achieved

Nucleation and growth 177 (a) Microstructure

Concentration distribution

Convection influence

Dielectric

Dielectric

c c0

Deposit Electronic conductor (semiconductor)

(b)

(c)

Microstructure

Imperfect filling

Perfect filling (“superfilling”)

Microstructure

Activation

Electroless deposition

Start layer

Figure 6.10 Specific features of metal deposition in negative microstructures.

successfully by electrodeposition of metals, alloys or conducting polymers. In this case the initial nucleation takes place only at the bottom of the microstructure and it is filled gradually from bottom to top by further deposition. The arising problems are connected mainly to the non-uniform concentration distribution in the microstructure and have been recently studied for the case of cathodic alloy deposition using spatially resolved concentration measurements.75 As indicated in Figure 6.10(a) the formation of vertical and lateral concentration gradients can lead to a preferential deposition at certain sites. The distribution of the deposit inside the microstructure can be also significantly influenced by electrolyte convection. In electrodeposition of alloys, concentration gradients result in composition gradients within the deposit. The problems discussed above arise in diffusion controlled electrodeposition. In galvanic deposition a change from diffusion to activation control can be achieved by reducing the cathodic current density. Such an approach, however, can lead to a drastic change of deposit microstructure and properties due to the simultaneous change of the ratio between the rates of nucleation and growth. In addition, in the case of a negative microstructure with a relatively high aspect ratio, the current distribution and the filling can be significantly influenced by the ohmic drop in the electrolyte inside the structure.

178

G. Staikov

A completely conducting microstructure as shown in Figure 6.10(b) is very suitable for electrodeposition because in the very initial stages itself the process starts uniformly at all points of the structure by primary nucleation. However, the appearance of concentration gradients of depositing species during the further deposition usually results in a preferential growth at the top of the microstructure and its imperfect filling. A perfect filling requires an increasing deposition rate from top to bottom of the microstructure. This problem has been solved in a very elegant way in the so-called Damascene copper electroplating,8,9 which was developed for IBM’s Cu chip interconnection technology. The key point in this technique is the use of plating additives which inhibit Cu electrodeposition and are simultaneously consumed at the deposit/electrolyte interface under diffusion control. The appearing gradient of the additive concentration along the microstructure leads to an increasing inhibition of Cu electrodeposition from the bottom to the top of the structure and to a perfect filling called “superfilling”. A theoretical model describing the superfilling phenomenon has been developed on the basis of theoretical approaches, previously applied to levelling in conventional electroplating.8,76 Anodic oxide formation can be used for modification of negative metal or semiconductor microstructures by an insulating oxide film. Homogeneous thin oxide films of equal thickness at all points of the microstructure can be obtained in the case of a high field controlled oxide growth.11,54,60 Completely insulating negative microstructure shown in Figure 6.10(c) can be filled successfully with metals or alloys by electroless deposition, which does not depend on the substrate conductivity. Thus, no electric contact to the microstructure is needed in this case. To start the process, the surface is sensitized by a start layer consisting of small catalytic metal clusters (usually Pd clusters). The best conditions for an optimal filling are obtained by a selective sensitization only at the bottom of the microstructure (cf. Figure 6.10(c)).45 Electroless deposition can also be applied for filling of electronically conducting microstructures. A detailed comparative study of filling behavior of electrodeposition and electroless deposition of nickel has been recently published by Thies et al.77 6.4.2

Microstructuring by selective electrochemical deposition

In comparison with other deposition techniques, electrodeposition offers unique possibilities for an exact and easy control of both, the deposition rate by the current density and the supersaturation by the overpotential. A characteristic feature of electrodeposition on foreign substrates is the existence of a critical overpotential (supersaturation) to initiate nucleation of the new phase. The critical overpotential |␩crit| depends on the nature of the deposit as well as on the size and properties of the foreign substrate. Microstructured surfaces exhibit different microareas, which can be considered as microsubstrates with respect to the process of electrodeposition. Based on the critical overpotential concept, various techniques for modification of microstructures by selective electrodeposition can be developed and applied in electrochemical microsystem technologies. Recently, this possibility has been demonstrated in the case of copper electrodeposition on

Nucleation and growth 179 (a) TiO2

TiN Ti

(b) TiO2

Cu TiN Ti

(c) TiO2

Cu TiN Ti

Figure 6.11 Selective electrochemical deposition of Cu on a microstructured Ti substrate.78 (a) microstructure; (b) low cathodic overpotentials (|␩| 50mV); (c) high cathodic overpotentials (|␩| 220mV).

microstructured Ti(0001) substrates.78 The microstructure studied (Figure 6.11(a)) consists of a TiN-microarea with metallic conductivity behavior surrounded by a homogeneous n-type semiconducting TiO2-layer with a thickness of 3nm. The TiN-microarea is produced by local N2 -implantation in UHV whereas the TiO2film is formed electrochemically by a high field controlled growth. Experimental results obtained show clearly that at low cathodic overpotentials (|␩| 50mV) Cu electrodeposition occurs selectively only on the TiN-microarea (Figure 6.11(b)). Deposition of Cu on n-TiO2-modified surface can be achieved only at significantly higher cathodic overpotentials |␩| 220 mV (Figure 6.11(c)). The much higher critical cathodic overpotential |␩crit| for Cu electrodeposition on n-TiO2surface results from semiconducting properties of this material. These results demonstrate that selective electrochemical deposition and dissolution processes can be successfully applied for preparation of different heterostructures involving metals, semiconductors, conducting polymers and insulators. 6.4.3

Electrochemical structuring by local probe techniques

The development of various local probe techniques opened a new window not only for characterization of surfaces at an atomic level but also for local surface structuring and modification in submicrometer and nanometer ranges. In situ application of these techniques in electrochemical systems offers a great advantage, since the potentials of both substrate and probe can be controlled independently of each other using reference and counter electrodes.79 Different approaches have been applied during the past 10 years for electrochemical structuring of

180

G. Staikov

solid surfaces by local probe techniques.13,14,80–82 Penner et al.80,83–85 have used a scanning tunneling microscope (STM), operating in two-electrode mode (tip and substrate), to structure HOPG by local electrodeposition of Ag and Cu. The structuring is based on a successive pit formation and preferential nucleation at the pit, both induced by the applied pulse polarization routine. Kolb et al.81,86,87 developed a technique for local generation of metal clusters using an electrochemical STM, operating in the four-electrode mode (tip, substrate, reference and counter electrodes). In this case an initial electrodeposition on the tip is followed by a controlled tip approach to the surface leading to a “jump-to-contact” and formation of a connective neck between tip and substrate. The generation of metal clusters on the substrate occurs during withdrawing of the tip. This technique can be successfully applied for structuring in systems with a high metal/substrate adhesion energy. Another approach using in situ STM involves an initial cathodic metal deposition on a STM tip followed by an anodic pulse dissolution of the deposit.82,88 At constant substrate potential this polarization routine leads to a significant local increase of the supersaturation resulting in an enhanced local nucleation and growth. This structuring technique is more universal and can be used also in systems with weak metal/substrate adhesion. Many techniques employing the scanning electrochemical microscope (SECM) for local microstructuring are based on a similar approach.89–92 The atomic force microscope (AFM) has also been used as a powerful tool for electrochemical structuring on a submicrometer scale.13,14,80,93–100 One of the most extensively investigated electrochemical processes is the AFM-tip-induced anodic oxidation.94–100 In this case, AFM operates in an ambient humidity and involves an electronically conductive tip. Condensation of water between tip and substrate leads to the formation of a meniscus representing a two-electrode “nanoelectrochemical” cell, which localizes the process (Figure 6.12). This technique is used successfully for local oxidation of various metals and silicon. As an example Figure 6.13 shows oxide lines written with an AFM-tip on n-Si and p-Si substrates.98 As expected the illumination does not influence the anodic oxidation process on p-Si but leads to a significant change of the width and height of the oxide line on the n-Si substrate. The AFM-tip-induced anodic oxidation has been applied for fabrication

Conductive AFM-tip

H2O-condensation

– Oxide +

Substrate

Substrate

Figure 6.12 Tip-substrate “nanoelectrochemical” cell for AFM tip-induced anodic oxidation.

Nucleation and growth 181 (a)

(b)

Dar

10 nm

Illu

min

k

Dar

10 nm

Illu

min

ate

d

k

ate

d

p-Si

n-Si 2.5 µm

2.5 µm

Figure 6.13 Silicon oxide lines obtained on n-Si (a) and p-Si (b) by AFM tip-induced anodic oxidation in a nitrogen atmosphere with a relative humidity of 55% (tip scan rate 1m/s, applied substrate-tip voltage 10V).98

of different submicrometer- and nanometer-scale electronic devices, such as metal oxide semiconductor field-transistor (MOSFET)99 and single electron transistor (SET) operating at room temperature.100

6.5

Conclusions

A large amount of knowledge has been accumulated in recent years on the mechanisms of formation of metals, oxides and polymers in different electrochemical systems. Although the fundamentals of involved nucleation and growth phenomena seem to be well understood, further developments of new theoretical models and sophisticated experiments are necessary to understand the mechanisms of various phase formation processes occurring under the specific conditions of electrochemical microstructuring.

Acknowledgments The author is indebted to J. W. Schultze, M. M. Lohrengel and V. Tsakova for helpful discussions and to K. Márquez, R. Krumm, H. Bloeß and A. Kurowski for their scientific cooperation and technical assistance. Financial support given by the “Deutsche Forschungsgemeinschaft” (DFG), the “Arbeitsgemeinschaft industrieller Forschungs-vereinigungen” (AiF-FV-Nr. 11955 N), the “Bundesministerium für Wirtschaft” (BMWi) and the “Ministerium für Schule und Weiterbildung, Wissenschaft und Forschung des Landes Nordrhein-Westfalen” (MSWWF NRW) is also gratefully acknowledged.

References 1. H. Reichl (ed.), microsystem Technologies (Springer-Verlag, Heidelberg, 1990). 2. N. Masuko, T. Osaka and Y. Ito (eds), Electrochemical Technology: Innovation and New Developments (Kodansha Ltd., Tokyo and Gordon & Breach Publ., Amsterdam, 1996).

182

G. Staikov

3. J. W. Schultze (ed.), Electrochemical Microsystem Technologies Special Issue of Electrochim. Acta 42(20–22) (1997). 4. T. Osaka (ed.), Electrochemical Applications of Microtechnology Special Issue of Electrochim. Acta 44(21–22) (1999). 5. L. T. Romankiw, Electrochim. Acta 42, 2985 (1997). 6. E. W. Becker, W. Ehrfeld, P. Hagmann, A. Maner and D. Münchmeyer, Microelectron. Eng. 4, 35 (1986). 7. W. Menz and P. Bley, Mikrosystemtechnik für Ingenieure (VCH, Weinheim, 1993). 8. P. C. Andricacos, C. Uzoh, J. O. Ducovic, J. Horkans and H. Deligianni, IBM J. Res. Develop. 42, 567 (1998). 9. P. C. Andricacos, Interface 8(1), 32 (1999). 10. J. W. Schultze and K. G. Yung, Electrochim. Acta 40, 1369 (1995). 11. J. W. Schultze, T. Morgenstern, D. Schattka and S. Winkels, Electrochim. Acta 44, 1847 (1999). 12. J. W. Schultze and V. Tsakova, Electrochim. Acta 44, 3605 (1999). 13. G. Staikov, W. J. Lorenz and E. Budevski, in Imaging of Surfaces and Interfaces (Frontiers of Electrochemistry, Vol. 5), J. Lipkowski and P. N. Ross (eds) (Wiley-VCH, Inc., New York, 1999). 14. E. Budevski, G. Staikov and W. J. Lorenz, Electrochemical Phase Formation and Growth – An Introduction to the Initial Stages of Metal Deposition (VCH, Weinheim, 1996). 15. S. Stoyanov, in Current Topics in Material Science, Vol. 3, E. Kaldis (ed.) (NorthHolland, Amsterdam, 1979). 16. A. Milchev, Contemp. Phys. 32, 321 (1991). 17. D. Kashchiev, J. Chem. Phys. 76, 5098 (1982). 18. M. Fleischmann and H. R. Thirsk, in Advances in Electrochemistry and Electrochemical Engineering, Vol. 3, P. Delahay (ed.) ( Wiley, New York, 1963). 19. J. A. Harrison and H. R. Thirsk, in Electroanalytical Chemistry, Vol. 5, A. J. Bard (ed.) (Marcel Dekker, New York, 1971). 20. G. J. Hills, D. J. Schiffrin and J. Thompson, Electrochim. Acta 19, 657 (1974). 21. D. Kashchiev and A. Milchev, Thin Solid Films 28, 189, 201 (1975). 22. M. Y. Abyaneh and M. Fleischmann, Electrochim. Acta 27, 1513 (1982). 23. G. A. Gunavardena, G. Hills, I. Montenegro and B. Scharifker, J. Electroanal. Chem. 138, 225 (1982). 24. B. Scharifker and G. Hills, Electrochim. Acta 28, 879 (1983). 25. J. H. van der Merwe, in Single Crystal Films, M. H. Frankombe and H. Sato (eds) (Pergamonn Press, Oxford, 1964). 26. G. Staikov, J. Sackmann and W. J. Lorenz, in preparation. 27. P. Bindra, H. Gerischer and D. M. Kolb, J. Electrochem. Soc. 124, 1012 (1977). 28. P. Allongue, in Modern Aspects of Electrochemistry Vol. 23, B. E. Conway, T. H. White and J. O’M. Bockris (eds) (Plenum Press, New York, 1992). 29. R. Krumm, B. Guel, C. Schmitz and G. Staikov, Electrochim. Acta 45, 3255 (2000). 30. S. Toshev and I. Markov, Electrochim. Acta 12, 281 (1967). 31. G. Sherb and D. M. Kolb, J. Electroanal. Chem. 396, 151 (1995). 32. P. M. Vereecken, K. Strubbe and W. P. Gomes, J. Electroanal. Chem. 433, 19 (1997). 33. M. H. Hölzle, V. Zwing and D. M. Kolb, Electrochim. Acta 40, 1237 (1995). 34. P. M. Rigano, C. Mayer and T. Chierchie, J. Electroanal. Chem. 248, 219 (1988). 35. E. Michailova, I. Vitanova, D. Stoychev and A. Milchev, Electrochim. Acta 38, 2455 (1993).

Nucleation and growth 183 36. A. I. Danilov, E. B. Molodkina and Ju. M. Polukarov, Russ. J. Electrochem. 30, 674 (1994). 37. B. Rashkova, B. Guel, R. T. Pötzschke, G. Staikov and W. J. Lorenz, Electrochim. Acta 43, 3021 (1998). 38. H. Bort, K. Jüttner, W. J. Lorenz, G. Staikov and E. Budevski, Electrochim. Acta 28, 985 (1983). 39. J. Mostany, J. Mozota and B. R. Scharifker, J. Electroanal. Chem. 177, 25 (1984). 40. H. Fischer, Elektrolytische Abscheidung und Elektrokristallisation von Metallen (Springer, Berlin, 1954). 41. R. Winand, Electrochim. Acta 39, 1091 (1994). 42. M. Paunovic, Plating 55, 1161 (1968). 43. M. Paunovic, in Electrochemistry in Transition, O. J. Murphy, S. Srinivasan and B. E. Conway (eds) (Plenum Press, New York, 1992). 44. Y. Okinaka and T. Osaka, in Advances in Electrochemical Science and Engineering, H. Gerischer and Ch. Tobias (eds) (VCH, Weinheim, 1994). 45. Y. Shacham-Diamand, V. Dubin and M. Angyal, Thin Solid Films 262, 93 (1995). 46. C. J. Sambucetti, in Electrochemical Technology: Innovation and New Developments, N. Masuko,T. Osaka and Y. Ito (eds) (Kodansha Ltd., Tokyo and Gordon & Breach Publ., Amsterdam, 1996). 47. J. E. A. M. van der Meerakker, J. Appl. Electrochem. 11, 395 (1981). 48. I. Konstantinov, A. Panov and J. Malinowski, J. Phot. Sci. 21, 250 (1973). 49. A. Vashkyalis, Elektrokhimiya 14, 1770 (1978). 50. A. Bonissent and B. Mutaftschiev, J. Chem. Phys. 58, 3727 (1973). 51. L. Young, Anodic Oxide Films (Academic Press, New York, 1961). 52. K. J. Veter, Elektrochemische Kinetik (Springer, Berlin, 1961). 53. D. A. Vermilyea, in Advances in Electrochemistry and Electrochemical Engineering, Vol. 3, P. Delahay (ed.) (Wiley, New York, 1963). 54. J. W. Schultze, M. M. Lohrengel and D. Ross, Electrochim. Acta 28, 973 (1983). 55. J. W. Schultze and M. M. Lohrengel, Ber. Bunsenges. Phys. Chem. 80, 552 (1976). 56. M. M. Lohrengel and J. W. Schultze, Electrochim. Acta 21, 957 (1976). 57. M. M. Lohrengel, P. K. Richter and J. W. Schultze Ber. Bunsenges. Phys. Chem. 83, 490 (1979). 58. J. A. Harrison, S. K. Rangarajan and H. R. Thirsk, J. Electrochem. Soc. 113, 1120 (1966). 59. M. M. Lohrengel, Electrochim. Acta 39, 1265 (1994). 60. M. M. Lohrengel, Materials Science and Engineering R11, 243 (1993). 61. T. A. Skotheim (ed.), Handbook of Conducting Polymers (Marcel Dekker, New York, 1986). 62. M. G. MacDiarmid and M. R. Maxfield, in Electrochemical Science and Technology of Polymers, Vol. 1, R. G. Linford (ed.) (Elsevier, London, 1987). 63. J. Roncali, Chem. Rev. 92, 711 (1992). 64. S. Azavapiriyanont, G. K. Chandler, G. A. Gunawardena and D. Pletcher, J. Electroanal. Chem. 177, 229, 245 (1984). 65. A. J. Downard and D. Pletcher, J. Electroanal. Chem. 206, 139 (1986). 66. A. R. Hillman and E. F. Mallen, J. Electroanal. Chem. 220, 351 (1987). 67. A. Hamnett and A. R. Hillman, J. Electrochem. Soc. 135, 2517 (1988). 68. B. R. Scharifker, E. Garcia-Pastoriza and W. Marino, J. Electroanal. Chem. 300, 85 (1991). 69. J. W. Schultze and A. Thyssen, Synth. Met. 41–43, 2825 (1991).

184 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100.

G. Staikov K. Bade, V. Tsakova and J. W. Schultze, Electrochim. Acta 37, 2255 (1992). V. Tsakova and J. W. Schultze, Bulg. Chem. Comm. 27, 138 (1994). F. Li and W. J. Albery, Electrochim. Acta 37, 393 (1992). R. Córdova, M. A. del Valle, A. Arratia, H. Gómez and R. Schrebler, J. Electroanal. Chem. 377, 75 (1994). R. Schrebler, P. Grez, P. Cury, C. Veas, M. Merino, H. Gómez, R. Córdova, M. A. del Valle, J. Electroanal. Chem. 430, 77 (1997). M. Küpper and J. W. Schultze, Electrochim. Acta 42, 3023 (1997). J. O. Dukovic and C. W. Tobias, J. Electrochem. Soc. 137, 3748 (1990). A. Thies, G. Schanz, E. Walch and J. Konys, Electrochim. Acta 42, 3033 (1997). O. Voigt, B. Davepon, G. Staikov and J. W. Schultze, Electrochim. Acta 44, 3731 (1999). H. Siegenthaler, in Scanning Tunneling Microscopy II, Springer Ser. Surf. Sci., Vol. 28, R. Wiesendanger and H.-J. Güntherodt (eds) (Springer, Berlin, 1992). R. M. Nyffeneger and R. M. Penner, Chem. Rev. 97, 1195 (1997). D. M. Kolb, R. Ullmann and T. Will, Science 275, 1097 (1997). R. T. Pötzschke, G. Staikov, W. J. Lorenz and W. Wiesbeck, J. Electrochem. Soc. 146, 141 (1999). W. Li, J. A. Virtanen and R. M. Penner, Appl. Phys. Lett. 60, 1181 (1992). W. Li, J. A. Virtanen and R. M. Penner, J. Phys. Chem. 96, 6529 (1992). R. M. Penner, Scanning Microsci. 7, 805 (1993). R. Ullmann, T. Will and D. M. Kolb, Chem. Phys. Lett. 209, 238 (1993). D. M. Kolb, R. Ullmann and J. C. Ziegler, Electrochim. Acta 43, 2751 (1998). D. Hoffmann, W. Schindler and J. Kirschner, Appl. Phys. Lett. 73, 3279 (1998). S. Meltzer and D. Mandler, J. Electrochem. Soc. 142, L82 (1995). A. J. Bard, M. V. Mirkin, P. R. Unwin and D. O. Wipf, J. Phys. Chem. B 96, 1861 (1992). K. Borgwarth, C. Ricken, D. Ebling and J. Heinze, Ber. Bunsen-Ges. Phys. Chem. 99, 1421 (1995). C. Heß, K. Borgwarth, C. Ricken, D. Ebling and J. Heinze, Electrochim. Acta 42, 3065 (1997). J. W. LaGraff and A. A. Gewirth, J. Phys. Chem. 98, 11246 (1994). J. A. Dagata, J. Schneir, H. H. Harray, C. J. Evans, M. T. Postek and J. Bennet, Appl. Phys. Lett. 56, 2001 (1990). H. Sugimura, T. Uchida, N. Kitamura and H. Masuhara, Appl. Phys. Lett. 63, 1288 (1993). L. Tsau, D. Wang and K. L. Wang, Appl. Phys. Lett. 64, 2133 (1994). T. Hattori, Y. Ejiri, K. Saito and M. Yasutake, J. Vac. Sci. Technol. A 12, 2586 (1994). H. Bloeß, G. Staikov and J. W. Schultze, Electrochim. Acta 47, 335 (2001). S. C. Minne, H. T. Soh, Ph. Flueckiger and C. F. Quate, Appl. Phys. Lett. 66, 703 (1995). K. Matsumoto, M. Ishii, K. Segawa, Y. Oka, B. J. Vartanian and J. S. Harris, Appl. Phys. Lett. 68, 34 (1996).

Part II

Micropatterning

7

High resolution lithography Shinji Matsui

7.1

Introduction

Recent years have witnessed a number of investigations concerning nanostructure technology. The objective of research on nanostructure technology is to explore the basic physics, technology, and applications of ultrasmall structures and devices with dimensions in the sub-100-nm regime. Today, the minimum size of Si and GaAs production devices is down to 0.25 m or less. Nanostructure devices are now being fabricated in many laboratories to explore various effects, such as those created by downscaling existing devices, quantum effects in mesoscopic devices, or tunneling effects in superconductors, etc. In addition, new phenomena are being explored in an attempt to build switching devices with dimensions down to the molecular level. Figure 7.1 summarizes the resolution capabilities of several lithography processes that use electrons, ions, and photons. It includes the narrowest line

Submicron technology EB

Electron

Atom (angstrom) technology

Nano technology SEM

STEM

STM

SET

Atom-device

FIB

Ion

g, i-line Excimer, SR

Photon

Si ULSI

1 Gb

Sub-0.1µm CMOS Mesoscopic devices Quantum dot

256 Mb Bacterium

64 Mb

16 Mb

Hemoglobin Ribosome

Insulin

Benzene c9c bond

4 Mb 1000

100

10 Size (nm)

Figure 7.1 Microfabrication using electrons, ions, and photons.

1

0.1

188

S. Matsui

width of feature size obtained with each process. Microfabrication can be classified into three regimes: submicron (1000–100 nm), nano (100–1 nm), and atom (or angstrom, less than 1 nm). A 256-Mb dynamic random access memory (DRAM) Si ULSI of 0.25-m dimensions can be fabricated by using an i-line stepper with a phase shift mask, or an excimer laser stepper. An excimer laser or synchrotron radiation (SR) lithography can be applied to a 1 Gb DRAM with a 0.15-m feature size. Electron beam (EB) lithography is the most widely used and versatile lithography tool used in fabricating nanostructure devices. Because of the availability of high-quality electron sources and optics, EB can be focused to diameters of less than 10 nm. The minimum beam diameters of scanning electron microscopes (SEM) and scanning transmission microscopes (STEM) are 1.5 and 0.5 nm, respectively. While focused-ion beam (FIB) can be focused close to 8 nm. EB and FIB can be used to make nanoscale features in the 100–1-nm regime. Scanning tunneling microscope (STM) is used for atomic technology in the region of 1–0.1 nm. Figure 7.2 shows the resolution of various resists, which were confirmed by experiment for electrons and ions. Minimum sizes of 8 nm for PMMA,1,2 10 nm for ZEP (Nippon Zeopn Co.) positive resists,3 20 nm for SAL601 (Shipley Co.),4 and 10 nm for Calixarene negative resists5 have been demonstrated using EB lithography. Nanoscale patterns have also been written in inorganic resists such as AlF3, NaCl, and SiO2 using STEM6,7 and SEM.8 Furthermore, carbon contamination patterns of 8 nm have been fabricated with SEM,9 and 8-nm PMMA patterns have been demonstrated by using Ga FIB.10 In this chapter, recent progress in nanofabrication using EB and its application to devices are described. 25

Resolution (nm)

20

Organic Inorganic Deposition

SAL601 5 kV SEM

15 ZEP 25 kV SEM

Calixarene 30 kV SEM

SiO2 300 kV STEM

10

5

PMMA 50 kV FIB

PMMA 50 kV SEM

Carbon deposition 45 kV SEM (Li, Al)F 30 kV SEM

NaCl AlF3

100 kV STEM

0 10–3 10–2 10–1 100 101 102 103 104 Sensitivity (nC cm–1)

Figure 7.2 Resolution of various resists for electrons and ions.

High resolution lithography 189

7.2 7.2.1

Nanofabrication using electron beam Nanometer electron beam direct writing system

There are some reports on EB nanolithography system capable of exposure with a sub-10-nm beam in nanofabrication.3,11–15 Figure 7.3 shows a photograph and a design target of a 50-kV EB lithography system developed by modifying the JEOL-5FE 25 kV EB system.16 The major modified points of this EB system are acceleration voltage and a gas feed system. The acceleration voltage has been changed from 25 to 50 kV. The column design with respect to the electron gun chamber is modified to withstand the high voltage in order to avoid the field emission from the electrode surface. Because a Zr/O/W thermal field emission (TFE) electron gun is used in this system, the vacuum in the electron gun chamber must be maintained under 109 Torr. The TFE gun has a small virtual source size and a high angular current intensity. Various gas species can be introduced into this specimen chamber to investigate the EB-induced surface reactions. The gas feed system is shown in Figure 7.4. The 50-cc-capacity gas cylinder contains the reaction gas. The gas is introduced through a variable leak valve and a fine

Performance Acceleration = 50 kV Beam diameter (at field center) X = 4.67 nm, Y = 4.68 nm Beam current :100 pA Direct writing accuracy (field : 80 µm) (a) stitching : 2σ =0.021 µm (b) overlay : 2σ =0.016 µm Beam current : 800 pA

Figure 7.3 50-kV nano-EB direct writing system.

190

S. Matsui 200 µm Nozzle 50 kV Electron beam d = 5 nm

Base pressure 4.5 × 10–7 Torr Gas injection Styrene gas pressure 5 × 10–6 Torr 3 mm

Sample

Figure 7.4 Photograph and schematic diagram of gas introduction system. Gas injection system for EB-induced surface reaction.

nozzle with an inner diameter of 200 m. The gas line nozzle is placed at a distance of 3 mm from the wafer. A carbon deposition pattern with 14 nm in size was made by using STYRENE (C6H5CH CH2) gas. The EB diameter was measured by using a knife-edge method. The result of the measurement for both the x and y directions was less than 5 nm. Overlay and stitching accuracy were evaluated by exposing 16 chips (4 4 in array) with a size of 80 80 m in

High resolution lithography 191 PMMA resist at 50 kV. The stitching accuracy and overlay accuracy were 0.021 and 0.016 m at 2␴, respectively. 7.2.2

Ten-nanometer lithography using organic resist

A PMMA positive resist was exposed out to evaluate the fine pattern exposure characteristics using the above EB system. PMMA is known as the positive resist with high resolution. Thirty-nm-thick PMMA was spin-coated on a bare thick Si wafer. After PMMA was prebaked at 170C for 20 min, EB exposure was carried out. The line dose was 0.8 nC cm1. The PMMA was developed in a mixture of MIBK : IPA 1 : 3 for 1 min and was then rinsed in IPA for 1 min. The point EB was line-scanned with a period of 100 nm. Ten-nm width line patterns in PMMA resist were obtained as shown in Figure 7.5. Fine metal patterns are useful as conductive wires or gate metals for the investigation of mesoscopic devices and other nanostructure physics. Au–Pd metal was delineated by a liftoff method. For the exposure of dot patterns, the resist was exposed with a shot time of 75 s for each dot. After development, Au–Pd metal was deposited on the resist, at a thickness of 3 nm for dot patterns. The liftoff was performed in acetone. Figure 7.6 shows the SEM photographs of Au–Pd metal dot patterns on a Si wafer. Twentynm-diameter dot patterns with a period of 100 nm were successfully fabricated. Nanodevice fabrication requires not only high resolution but also high overlay accuracy. High-speed exposure very effectively meets the requirements because overlay accuracy is improved due to less beam drift on the nanometer scale. Moreover, it enables the use of a highly sensitive resist such as ZEP520,17 which 10 nm

300 nm

Figure 7.5 10-nm line width PMMA patterns.

192

S. Matsui 20 nm

100 nm

Figure 7.6 20-nm diameter Au–Pd patterns.

has sufficient resolution and high dry etching durability for nanolithography. A 10-nm scale resist pattern was obtained using ZEP520 positive resist. The ZEP520 resist was spin-coated onto an Si wafer to a thickness of 50 nm, and prebaked at 200ºC. After EB exposure, the ZEP520 was developed with hexyl acetate for 2 min and rinsed with 2-propanol. Figure 7.7 shows a ZEP520 resist pattern, in which the lines are 10 nm wide and have a pitch of 50 nm.3 Calixarene has a cyclic structure, as shown in Figure 7.8, and works as an ultrahigh resolution negative EB resist. Such characteristics seem to be convenient for a nanodevice fabrication process. It is roughly a ring-shaped molecule with about a 1-nm diameter. The basic component of calixarene is a phenol derivative that seems to have high durability and stability, originating from the strong chemical coupling of the benzene ring. The threshold of sensitivity was about 800 C cm2, which is almost 20 times higher than that of PMMA. Calixarene negative resist exposure was carried out. A 30-nm thick resist was coated on a bare Si wafer. After prebaking at 170ºC for 30 min, EB exposure was carried out and then the resist was developed in xylene for 20 s and was rinsed in IPA for 1 min. The etching durability of calixarene was tested using a DEM-451 (ANELVA Corp.) plasma dry-etching system with CF4 gas. The etching rate of calixarene is almost comparable to that of Si, and the durability is about four times higher than that of PMMA. This durability seems to be sufficient to make a semiconductor or a metal nanostructure. Nanodot arrays are useful not only for quantum devices but also for studying exposure properties. In this experiment, the EB current was fixed to 100 pA at 50 kV accelerating voltage, for which the spot size is estimated to be about 5 nm.

High resolution lithography 193

0.1 µ

Figure 7.7 10-nm line width ZEP patterns.

CH3

CH2

6

OCOCH3

Figure 7.8 Structure of calixarene.

All the dot arrays were fabricated on Si substrates. And the typical exposure dose (spot dose) was about 1 105 electrons/dot. Figure 7.9 shows typical dot array patterns having 15-nm diameter with 35-nm pitch. Germanium pattern transfer is shown in Figure 7.10. The 20-nm thick Ge layer requires at least a 5-nm thick calixarene layer to be etched down, and the resist thickness was 30 nm. Figure 7.10(a) shows the line patterns of the resist on Ge film exposed at a line dose of 20 nC cm1. Delineation was done using the S-5000 (Hitachi corp.) SEM with a beam current of 100 pA at a 30-kV acceleration voltage. A 10-nm line width and a smooth line edge were clearly observed. This smoothness is the key point in fabricating quantum nanowires by etching processes. Figure 7.10(b) shows the transferred pattern treated by 1 min of over etching, followed by oxygen–plasma treatment to remove the resist residues.

Figure 7.9 Calixarene dot array patterns with 15-nm diameter and 35-nm pitch.

(a)

10 nm

(b)

7 nm

Figure 7.10 Pattern transfer to Ge. (a) 10 nm line width calixarene pattern and (b) transferred 7-nm line width Ge pattern.

High resolution lithography 195 A Ge line of 7-nm width was clearly observed without short cutting. Narrowing by over etching is a standard technique to obtain a fine line, however, sidewall roughness limits the line width.18 The smoothness of the calixarene sidewall enables the line width to be narrowed below the 10-nm region by over etching. Calixarene is a single molecule and thus is monodispersed with a molecular weight of 972. In contrast, other phenol-based resists have dispersive weights from 1000 to 100 000, which set a resolution limit. The molecular uniformity of calixarene and its small molecular size is the origin of such surface smoothness and the resulting ultrahigh resolution. 7.2.3

Sub-10-nm lithography using inorganic resist

An inorganic resist seems to be the most promising material to achieve sub10-nm lithography. Many previous works concerning the inorganic resist were carried out using STEM,5,6,19,20 and many have attained nanometer-scale delineation. However, usage of the membrane-film substrate, which is commonly used in STEM studies, causes a crucial difficulty in device fabrication due to its delicate handling. In contrast to this STEM lithography, the conventional scanning SEM should give many advantages for nanosize device fabrication if one could achieve an equally fine pattern delineation on the standard Si substrate. In general, inorganic resists have a finer resolution than organic resists. An encouraging result, using AlF3-doped LiF resist, was reported.21,22 It suggested the resist grain size was reduced to below 10 nm on a nitrogen cooled substrate and successfully demonstrated sub-10 nm lithography using a STEM system. Furthermore, electron stimulated desorption should occur under relatively lowenergy (20–50 keV) electron irradiation in a standard SEM lithography system. The basis for the self-developing properties on LiF(AlF3) was studied, and sub-10 nm lithography was demonstrated using a standard SEM beam writing system.8 The AlF3 partially doped LiF inorganic-resist films were fabricated by using conventional multi-target (LiF, (Li0.9Al0.1)Fx, and (Li0.7Al0.3)Fx) ion beam sputtering. The chemical composition of the films was adjusted by controlling the flux ratio from each target. The sputtered particles have energy of several eV. As a result, the ion beam sputtering effectively reduced the grain size below 10 nm even at room temperature deposition, and even with an extremely slow growth rate of about 1 nm min1. The required dose for the (Li, Al)F resist in terms of AlF3 concentration is summarized in Figure 7.11, where sensitivity of the LiF and AlF3 are cited from the STEM work, and the hatched area was obtained by extrapolating these data. Sensitivity of 0.1 C cm2 on the 10-nm thick film was just below critical dose, but the other films do not show the perfect developing. The curve obtained from the 10-nm thick films shown with solid circles are very closed to those of the STEM data. This suggests that the desorption mechanism was not influenced by the electron energy in principle, suggesting the possibility of lithography of the (Li, Al)F resist under low-energy irradiation using SEM.

196

S. Matsui

Required dose (C cm–2)

102

Grain size effect

Deterioration by Al 60 nm thick w =1 µm

100

10–2 0

10 nm thick w =1 µm

STEM lithography

50 Aluminium concentration (%)

100

Figure 7.11 Required dose vs aluminum concentration.

60 nm

5 nm

Figure 7.12 5-nm line width patterns with 60-nm period fabricated at 100-nC cm1 line dose by 30 kV and 1.5-nm beam diameter SEM.

By optimizing the film quality, sub-10-nm lithography was demonstrated by using a Hitachi S-4200 (Hitachi Corp.) SEM with a fine EB produced by a field emission gun. The accelerating voltage of 30 kV and the beam diameter of 1.5 nm are the specification of the SEM. The line dose was about 100 nC cm1. Figure 7.12 shows the best result of the line delineation, where a fine line of 5-nm line width was clearly observed. This result demonstrates that sub-10-nm lithography can be achieved by SEM using an inorganic resist.

High resolution lithography 197 7.2.4

Nanometer fabrication using EB-induced deposition

In situ processes using beam-induced chemistry are promising technologies to fabricate ultimate fine patterns and to reduce process steps. In situ observation of W deposition by EB irradiation using a WF6 gas source was carried out by TEM to study the growth mechanism. The experimental arrangement is illustrated in Figure 7.13. The employed electron microscope was an EM-002A (Akashi Beam Technology Corp.) equipped with a real time TV monitor system specially improved for in situ observation. The microscope resolution was 0.23 nm at 120 kV, which allowed imaging individual rows of atom columns in W crystals. To reduce problems in regard to specimen contamination, the instrument was operated under ultrahigh vacuum of 3 108 Torr, attained by a dry vacuumpumping system with turbo pumps and ion pumps. The gas injection tube had a 3-mm inner diameter and was 5 mm from the sample surface. Fine and spherical Si particles were used as TEM specimens. The small particles were made by a gas-evaporation method in argon under a reduced atmosphere, where Si vapor was condensed into small particles. They were less than 100 nm in diameter and were usually covered with 1–3-nm-thick SiO2. The gas molecules were adsorbed on the fine Si particles, which were on the TEM specimen grid. The gas molecules were excited by the TEM EB and dissociated into W and F2 gas. Tungsten metal was deposited on the Si surface and growth was started. The growth process was observed in situ at single atom resolution level by an electron microscope equipped with a TV monitor system. First the EB was irradiated on a WF6 adlayer, formed on the fine Si particle surface, in order to clarify an initial growth process of EB-induced deposition. Second, the focused EB was irradiated on the Si fine particle surface, while WF6 was flowing on the surface, to study the resolution of EB-induced deposition.23,24 Figure 7.14 shows a typical series of electron micrographs of in situ TEM observation during EB irradiation of the WF6 adlayer. These micrographs were selected from a VTR tape which ran for 30 min. TEM e–

IP Specimen TMP

Source gas

Fluorescent screen Prism

SIT camera

TMP

Figure 7.13 In situ observation TEM system of EB deposition.

198

S. Matsui

(a)

(b)

0.31 nm

(c)

(d) 1

2

Figure 7.14 EB exposure time dependence of W cluster growth on an Si particle. Exposure times: (a) 0, (b) 3, (c) 15 and (d) 30 min.

W

15 nm

Si

Figure 7.15 W rod with 15-nm diameter.

High resolution lithography 199 (a)

(b)

(c)

0.5 µm

Figure 7.16 SEM micrograph of a branched EB-deposition tip.

Electron beam irradiation times for Figures 7.14(a)–(d) were 0, 3, 15, 30 min, respectively. These results indicated that W atoms, dissociated by EB irradiation from the WF6 adlayer, coalesced and grew under EB irradiation. According to the real time observation on a TV screen, moving clusters often collided with each other, causing coalescence. Figure 7.15 shows an electron micrograph of a W rod on a fine Si particle. The W rod was made using a focused EB (STEM mode: 3 nm in diameter) scanning manually at 1 nm s1 speed on the Si surface under 1 106 Torr source gas pressure. The W rod radius was 15 nm. This result indicated that a three-dimensional nanostructure could be fabricated using this technique. Three-dimensional STM tips as shown in Figure 7.16 25 and electron field emitters26 have successfully been made by EB-induced deposition with a computer-controlled writing system. Furthermore, single-electron transistor (SET) fabrication with 10-nm dimension was reported by using EB deposition with WF6 gas.27 7.2.5

Forty-nanometer-gate-length MOSFETs

Minute CMOS devices with a gate length of 100 nm or less are under extensive examination.28–30 This is because it is expected that small feature size devices do not only realize very high-density integrated circuits, but also high switching speed with low power consumption. Forty-nm-gate MOSFETs have been fabricated using excimer-laser lithography and resist-thinning technology.30 However, to accelerate investigation of these devices, it is important to develop sub-100-nm direct EB lithography process with good line width control. A 40-nm-gate NMOS device has been demonstrated by using direct EB lithography.31 Figure 7.17 shows the fabrication process of the sub-100-nm MOS transistors by EB direct writing. A 3.5-nm-thick gate oxide was used to obtain high current drivability. The polysilicon thickness was 150 nm. The single-layer resist

200

S. Matsui Negative resist (200 nm) Polysilicon (150 nm) Gate oxide (3.5 nm)

Resist coating (SAL601) p-Si substrate

EB exposure (50 kV EB) Development p-Si substrate

Pattern transfer (Cl2 plasma etching) p-Si substrate

SiO2 sidewall Source

n+-polysilicon gate 3.5 nm gate oxide Drain

n+ p-Si substrate

Sidewall deposition Ion implantation

Figure 7.17 Gate fabrication process of a MOSFET.

was adopted as a mask to make the fabrication process simple. A SAL601 (Shipley Ltd.) chemically-amplified negative resist with 200-nm thickness was coated on a 6-inch Si wafer. The thickness of resist was decided on the basis of the etching selectivity of five for polysilicon over the resist to obtain high-aspect ratio patterns. The gate-resist patterns were exposed by a 50-kV EB lithography system as shown in Figure 7.3. After developing the resist, gate pattern was transferred into polysilicon by plasma etching in which Cl2SF6O2 etching gas was used. The etching rate of polysilicon was 130 nm min1 with a uniformity of less than

5%. The etching selectivity of polysilicon to SiO2 is over 40. The other lithography processes were performed using optical steppers. Figures 7.18(a) and (b) show a 40-nm-gate resist pattern on polysilicon/SiO2/ Si and on an NMOS transistor. The resist thickness was about 200 nm. A 40-nm line was exposed at 300 C cm2 by a single line scan. NMOS FETs with various gate lengths of above 40 nm on 6-inch wafers were fabricated. Source-to-drain resistance at VD 0.1 V vs designed gate-pattern width is plotted in Figure 7.19. The good linearity indicated that the gate length was successfully controlled even for a less than 100-nm gate by using proper proximity effect correction and a high-energy nanometer EB. ID–VD characteristics are shown in Figure 7.20. Well-behaved short-channel characteristics were obtained down to 60-nm gate length. Operation of a 40-nm gate FET was also

High resolution lithography 201 (a)

40 nm

(b)

Figure 7.18 (a) 40-nm-gate resist pattern with a height of 200 nm on a polysilicon layer and (b) on an NMOS transistor.

confirmed, though weak punch-through occurred. Maximum transconductance (gm) at VDS was 580 mS mm1 for the 40-nm NMOSFET. 7.2.6

Fourteen-nanometer gate-length MOSFETs

An electrically variable shallow junction MOSFET (EJ-MOS) with an ultrashallow source/drain junction has been fabricated to investigate transistor characteristics and physical phenomena in ultrafine gate MOSFETs.32 Figure 7.21 shows a schematic cross section of the EJ-MOSFET. The lower gate, which corresponds to the “gate” in conventional MOSFETs, controls the drain current. A positive upper-gate bias induces source/drain regions at the silicon surface. Because the source/drain regions are electrically induced, they are

202

S. Matsui

Channel resistance (Ω µm–1)

250 Vg-Vth = 2.0 V Vg-Vth = 0.8 V

200 150 100 50 0 0.0

0.2 0.4 0.6 0.8 Designed gate length

1.0

Figure 7.19 Channel resistance vs designed gate length for different gate voltage.

(a)

0.8 ID (mA µm–1)

(b) 1.0

1.0

0.6 0.4

VG(V) 1.5 1.25 1.0 0.75 0.5 0.25 0

Lg = 60 nm

0.8 0.6 0.4

VG(V) 1.5 1.25 1.0 0.75 0.5 0.25 0 –0.25

Lg = 40 nm

0.2

0.2

0.0 0.0 0.2 0.4 0.6 0.8 1.0 VD (V)

0.0 0.0 0.2 0.4 0.6 0.8 1.0 VD (V)

Figure 7.20 ID –VD characteristics of NMOSFETs using direct EB exposure and a single-layer resist.

extremely shallow, typically 5 nm deep. The EJ-MOSFET was fabricated in a similar way as conventional Si-MOSFETs. To suppress short-channel effects (SCEs) caused by the lateral expansion of the depletion layers, a relatively high boron concentration of 2 1018 cm3 was used within the substrate. The boron concentration was controlled by means of the boron ion implantation and the thermal drive-in. The n regions were formed by arsenic ion implantation. A gate oxide (tox 5 nm) was formed by thermal oxidation and a 40-nm thick poly-Si layer was grown by chemical vapor deposition (CVD). Phosphorus was doped into the poly-Si film in a POCl3 atmosphere. The ultrahigh resolution EB resist

High resolution lithography 203 Intergate oxide

Lower gate

Gate oxide

Upper gate

n+ p-sub

n+ Channel Ultra-shallow source/drain

Electron

VLG = 0 VLG > 0

Conduction band

Figure 7.21 Schematic cross section of an EJ-MOSFET.

14 nm

Figure 7.22 TEM cross-sectional view of a 14-nm-gate-length poly-Si lower gate.

calixarene5 was spin-coated onto the poly-Si film and EB direct writing with a 5-nm diameter and 50-kV acceleration energy was performed. After the developing procedure, the resist pattern was transferred to the poly-Si film by reactive ion etching (RIE) with CF4 gas. Figure 7.22 shows a TEM cross-sectional view of a 14-nm long poly-Si lower gate. The lower gate was well defined. The 20-nm thick integrate oxide layer was grown by CVD, which was followed by N2 annealing and H2 annealing. Finally, the upper gate and source/drain electrodes were formed by Au/Al evaporation.

204

S. Matsui 10

Drain current (A)

8

LLG = 14 nm T = 300 K VUG = 7V

VLG = 1 V 0

6

4 –1 2 –2 0

1 Drain voltage (V)

Figure 7.23 I–V characteristics of a 14-nm-gate-length EJ-MOSFET at 300 K.

The electrical characteristics of the fabricated devices with gate length (LLG) ranged from 14 to 98 nm at 300 K. To form source/drain regions, an upper-gate voltage (VUG) of 7 V was applied. Figure 7.23 shows the I–V characteristics of a 14-nm gate-length devices. Although the device does not exhibit clear saturation characteristics at a high-drain bias, a transistor operation in the 14-nm gate length was confirmed by this experiment. 7.2.7

Room temperature operation of an Si-SET

An Si-SET was fabricated by converting a one-dimensional Si wire on a SIMOX (separation by implanted oxygen) substrate into a small island with a tunneling barrier at the etch end by means of pattern-dependence oxidation.33–35 A SIMOX wafer with a very smooth superficial Si/buried SiO2 interface, which was obtained by 40-h annealing at 1350ºC, was used as the base material. A 30-nm thick superficial Si layer with a 30-nm thick top SiO2 layer was prepared. The thickness of the buried SiO2 layer was 400 nm. In order to fabricate the one-dimensional Si wire, EB nanolithography combined with an image reversal process using ECR plasma oxidation was used. The SET fabrication process as shown in the schematic diagram of Figure 7.24 is as follows. Just after the deposition of a 40-nm thick amorphous Si layer (a-Si) on the top SiO2 layer, and a 50-nm thick ZEP positive resist was coated on the a-Si layer, and then the resist pattern shown in Figure 7.25 was formed by EB lithography. Here, the resist on the one-dimensional wire and two-dimensional source and drain regions was removed and the surface of the a-Si on the area appeared since these areas were exposed on the EB. Then, the a-Si surface was oxided by an ECR-O2 plasma

High resolution lithography 205 (b)

1-dimensional Si wire

1-dimensional Si wire

Drain

20 nm

Gate electrode Source

400 nm

(a)

Drain

Source Buried SiO2

Back gate

Si substrate SIMOX wafer

Superficial Si layer

Ld 2-dimensional Si layer

Figure 7.24 (a) Schematic diagram of an Si-SET and (b) an EB pattern.

Figure 7.25 SEM photograph of SET structure corresponding to Figure 24(b).

stream, where the lateral shrinkage of the resist pattern due to O2-plasma etching was very small because of the highly directional nature of the ECR plasma stream. Next, the resist was removed with acetone to avoid oxidizing the a-Si surface under the resist. The a-Si layer was etched by ECR Cl2-based plasma using very thin plasma oxidized SiO2 as a mask. Then, the 30-nm thick SiO2 layer and the superficial Si layer were etched by RIE and ECR Cl2-based plasma using a-Si and SiO2 as a mask, respectively. A one-dimensional Si wire, with twodimensional Si layers for the source and drain regions, was formed, on which the

206

S. Matsui

T = 300 K 241 K

6

203 K

Conductance (µs)

175 K 4

146 K 118 K 89 K

2 61 K 42 K 0

2

4 6 Gate voltage (V)

8

Figure 7.26 Conductance oscillations of SET with LD50 nm at various temperatures. The curves are offset vertically.

SiO2 layer remained. However, the a-Si layer was completely removed by the final ECR etching. The width of the wires was 20 nm while the length LD varied in the range of 50–200 nm. The width of the two-dimensional regions, which were connected to each end of the wire, was 400 nm. After the pattern was formed, the wafer was oxidized in dry oxygen ambient at 1000ºC, which not only reduced the width and height of the wire, but also constricted the wire at its ends, by pattern dependent oxidation.33 Finally, the SET fabrication was completed after the phosphorous-doped poly-Si gate electrode was formed over the wire, followed by the phosphorous ion implantation. The source–drain conductance was derived from I–V characteristics measurement. The conductance characteristics at various temperatures for SETs with a wire length LD of 50 nm are shown in Figure 7.26. Clear conductance oscillations are observed even at high temperatures of 30 K and above. These oscillations are due to the Coulomb blockade by tunneling barriers with a very small capacitance fabricated by the procedure described above. The conductance oscillations remain even at 300 K for the SET, though they almost disappear at around 200 K for the SET in Figure 7.26. From these data, the total capacitance of these SETs is estimated to be about 1.5 aF.

7.3

Nanofabrication using focused ion beam

A focused ion beam (FIB) is a very attractive tool in lithography, etching, deposition, and doping.36–38 The liquid metal ion (LMI) source was developed in 1975,39 and within four years Seliger et al.40 had been able to produce a Ga FIB with a diameter of only 0.1 m and a current density of 1.5 A cm1. Their work

High resolution lithography 207 was very impressive to device engineers and researchers because it indicated that FIB technology would be applicable in device processes. The Ga LMI source41 and the Au–Si–Be alloy LMI source42 are popular for use in FIB processes because of their long lifetime (over 500 h) and high stability variation (less than

1%). Column design has also progressed. Kubena et al.,10 for example, developed a two-lens column system and have produced line and dot PMMA patterns with 8 nm dimensions by using a 50 kV, 100 pA Ga FIB. FIB technology is now used in various device processes, and FIB etching and deposition have become indispensable in such fields as failure analysis, mask repair, and transmission electron microscope (TEM). 7.3.1

FIB lithography

The sensitivity of a resist used for FIB lithography is two or more orders of magnitude higher than that of a resist used for EB lithography. And FIB lithography has another advantage that the ion-scattering in the resist is negligible and there is very little backscattering from the substrate. A novolak-based negative resist (Shipley Co. SAL601-ER7) was used to demonstrate high resolution capability of FIB lithography.43 After prebaking for 30 min at 80ºC in a N2 atmosphere, the resist on the wafer was exposed to 260-keV Be and Si FIB. Post-exposure baking was carried out for 7 min at 105ºC in a N2 atmosphere. The wafers were then developed for 3 min by using SAL-MF622 developer at 25ºC, and were rinsed with deionized water. The SAL601-ER7 resist sensitivity for 260-keV FIB exposure was measured in Figure 7.27 as compared with that for EB lithography. These resist sensitivity

SAL601

Normalized thickness

260 keV FIB with PEB

50 keV EB

without PEB

with PEB

1 Si++ 0.5

0 1011

Be++

e–

Si++

1012

Be++

1013 1014 Dose (ions, electrons cm–2)

Figure 7.27 Resist sensitivities of SAL601-ER7.

1015

1016

208

S. Matsui

curves with and without post exposure baking show that the resist sensitivity for FIB is about 100 times greater than that for EB and that the sensitivity obtained with postexposure baking is about 100 times greater than that obtained without baking. Figure 7.28 shows a SAL-601 negative resist pattern fabricated at a dose of 1.2 1012 ions cm2 by using a 260 keV Be FIB. Line width and resist thickness are 0.1 and 0.6 m, respectively. Exceptionally excellent patterns are produced, as demonstrated by the vertical sidewall profile. To fabricate semiconductor devices we not only need to make high resolution patterns but also to control the resist pattern profiles. When we make a GaAs microwave FET with a gate length less than 0.5 m, for example, to keep the gate resistance low enough we need to fashion a gate electrode with a mushroomshaped cross section. A multilayer resist structure and a side-etching method might be used to make such a gate, but it is not easy to obtain an accurate gate length when using these methods. An advantage of FIB lithography is that it enables both the lateral and longitudinal dimensions of patterns to be controlled precisely. Figure 7.29 shows the detailed process flow for FIB fabrication of

0.5 µm

Figure 7.28 0.1 m line width, 0.6 m thick SAL601-ER7 resist patterns fabricated by 260 keV Be FIB.

High resolution lithography 209 FIB mixed exposure

Epitaxial layer Source Drain

Resist

1. Source–drain formation 2. Resist coating

Damage FIB 0.25 µm

3. Exposure 1 (Be++) Ni/Al

5. Wafer immersing

FIB

0.7 µm

4. Exposure 2 (Si++)

Source Gate

6. Recess etching 7. Gate metal deposition

Drain

8. Liftoff

Figure 7.29 Process flow of mushroom-shaped gate fabrication using FIB lithography.

(a)

(b)

0.1U

0.1U Resist pattern

Gate pattern

Figure 7.30 SEM micrograph of a mushroom gate: (a) resist pattern and (b) gate pattern.

a mushroom-shaped gate of a high electron-mobility transistor (HEMT).44 A 1.05-m thick PMMA layer was exposed successively with 192 keV Be and 260 keV Si ion beams (each to a dose of 2.01013 ions cm2). The resist was developed by immersing the wafer in a 2 : 3 mixture of methylisobuthyl keton (MIBK) and isopropyl alcohol (IPA) (resist profiles were almost independent of developing conditions). The gate region was then recessed to the optimum depth by wet chemical etching, and Ni–Al alloy metal gate about 0.6 m thick was deposited. A liftoff procedure finished the gate fabrication process. An SEM photograph of the mushroom-shaped gate fabricated this way is shown in Figure 7.30. A 0.25-m gate with a mushroom shape with a cap width

210

S. Matsui

of 0.7 m was obtained. The size of the fabricated chip was 350 380 m and the gate width was 200 m. For passivation, a plasma CVD SiN about 0.1 m thick covered the gate metal and the epitaxial wafer surface. Excellent microwave performance, minimum noise figures of 0.68 dB at 12 GHz and 0.83 dB at 18 GHz were obtained. 7.3.2

FIB etching

FIB direct etching was applied as diagnostics tools for multilevel metallization VLSI.45 The main FIB techniques utilized when an FIB is used as a diagnostics

(a)

(b)

(c)

Figure 7.31 Illustration of microscopic cross sectioning by FIB etching. (a) Before etching; (b) after rough etching; and (c) after fine etching.

High resolution lithography 211 tool are microscopic selective etching, microscopic partial deposition, and scanning ion microscopy (SIM). These three techniques are used to form probing pads and to monitor the pad formation for failure analysis. They have also been used in circuit repair. FIB techniques have become essential tools for VLSI failure analysis, circuit repair, and process monitoring. There are several reports on failure analysis and circuit repair, and the etching function and the SIM function are used for microscopic cross sectioning and in situ observation. Figure 7.31 illustrates a procedure of microscopic cross sectioning and in situ observation. A roughly etched rectangular hole is first formed at the area of interest on the

(a)

A

Upper level metal Insulator Lower level metal Insulator

5 µm (b) Upper level metal Insulator Void

Voids

Lower level metal Insulator

1 µm

Figure 7.32 Electromigration-induced voids at lower level metallization. (a) SIM image after microscopic cross sectioning by FIB, tilted angle is 60; (b) high magnification of the A area in (a).

212

S. Matsui

VLSI chip, then the sidewall of the hole is etched finely and finally the chip is then tilted (0–60) and the sidewall (cross section) is observed using the SIM. A VLSI memory with multilevel metallization was subjected to a high-temperature bias test and failed electrically. The failed area in the device was then located by means of electrical analysis and was suspected to have failed because of gaps in its lower-level aluminum stripes. These suspect areas on the devices were then examined using a metallurgical microscope. The suspected area was cross sectioned using an FIB, and many voids were found in the lower metal (Figure 7.32). These voids were found, from this result and other related information, to be caused by electromigration. This cross-sectioning technique has also been used to fabricate cross-sectional TEM specimens by Kirk et al.46–48 Figure 7.33 shows the fabrication procedure. The T-shaped bar is first cut to a length of 3 mm using a holder, and this structure is mounted on the x – y table of the FIB machine. The area of interest is then thinned by FIB etching, after which the structure is placed in a specimen holder and observed by TEM. Figure 7.34 shows a TEM specimen of a 16 MDRAM with a wall only 0.1 m thick.

(a)

(b)

FIB

(c)

E-Beam

Figure 7.33 A schematic diagram of preparation of a cross-sectional TEM specimen by FIB etching: (a) mechanical slice, (b) FIB slice and (c) TEM observation.

High resolution lithography 213

Figure 7.34 A scanning ion image of the thin wall etching by FIB for TEM observation.

7.3.3

FIB deposition

Chemical assisted deposition uses chemical reactions between the substrate surface and molecules adsorbed on the substrate. FIB chemical assisted deposition has been used to repair clear defects in photo and X-ray masks. Blauner et al.49 demonstrated the high spatial resolution of FIB deposition process using a 100 keV, 13 pA Ga FIB of about 0.1 m in diameter. A regular array of thirtysix gold pillars, each corresponding to an individual beam spot, is shown in Figure 7.35. The pillars are about 0.15 m in diameter and 10 m high, deposition yield was approximately 75 atoms per incident ion, and the diameter of the gold pillars is 1.5–2 times the beam diameter. This result indicated that chemical assisted deposition had very high yield and can be used in mask repair, circuit repair, and micromechanics.50

7.4

Nanofabrication using a de Broglie wave (material wave technology)

Material-wave nanotechnology using electron- and atomic-beam holography has been developed. This technique allows nano- and atomic scale structures to be produced simultaneously.

214

S. Matsui

(a)

(b)

Figure 7.35 (a) SEM micrograph showing high resolution features written with 100 keV Ga FIB-induced gold deposition; (b) High magnification of the partial area in (a).

7.4.1

Electron beam holography

Holographic lithography has an advantage that it can produce a number of periodic patterns simultaneously. Electron holographic lithography was applied to nanofabrication. Electron interference fringes were recorded on a PMMA resist by using of a W(100) TFE gun and an electron biprism, and the fabricated patterns were observed by a conventional TEM and AFM.51,52 The electron optics of TEM with a W(100) TFE gun for electron holographic lithography is schematically illustrated in Figure 7.36. An electron beam of 40 kV is focused above an electron biprism with two condenser lenses. The Möllenstedttype electron biprism is constructed of two grounded plane electrodes and a finewire electrode, called a filament, between them. When a positive voltage, VB, is supplied to the filament, electron waves traveling on both sides of the filament are deflected and superimposed to form interference fringes on an observation plane. A Pt wire of 0.6 m in diameter was used as the filament. As is well known, two coherence waves overlapping at an angle of ␪ produce interference fringes with spacing, s, represented by s (␭/2)/sin(/2)

(1)

where ␭ denotes the de Broglie wavelength of 6.0 103 nm in this case. Figure 7.37 shows four-wave interference fringes through an X-biprism. Setting an X-biprism below two condenser lenses instead of the Möllenstedt-type

High resolution lithography 215 Source

Condenser lens 1

Condenser lens 2 Electron biprism

Interference fringes PMMA SiN

Figure 7.36 Scheme of electron optics of TEM for electron holographic lithography.

X-biprism 4

2 4 plane waves

3

1

Interference fringes

Figure 7.37 Four-wave interference fringes through an X biprism.

biprism, which has two filaments placed normal to each other and both are supplied VB, four coherent waves produce fringes like a checkerboard below the intersection of filaments, with the same spacing, s, as given by Eqn (1). Thus, electron holographic lithography would be, in principle, possible to generate line and dot patterns whose minimum spacing is ␭/2, which is comparable to the crystal lattice spacing. A 30-nm-thick PMMA, spin-coated on a 50-nm-thick self-supporting SiNx membrane and prebaked at 170ºC for 20 min, was set on the observation plane 70 mm below the biprism. The self-supporting nitride (SiN) membrane was about 60 m square and used to place the PMMA below interference fringes appropriately. Electron exposure to produce line patterns was carried out for 18 s with a dose of 25 C cm2, which was measured at the fringe part. Then, the PMMA

216

S. Matsui

was developed in MIBK : IPA 1 : 3 for 1.0 min and rinsed in IPA for 30 s. Similarly, PMMA dot patterns were exposed, at half the dose as that for the line patterns, in order to maintain whole dots. The electron exposure to produce dot patterns was carried out for 9.0 s with a dose of 13 C cm2. The PMMA was developed in MIBK : IPA for 3.0 min and rinsed in IPA for 1.0 min. Figure 7.38(a) shows interference fringes of the Möllenstedt-type electron biprism, which was magnified 530 times by the lenses below the observation plane and recorded on a photoplate with 1.0 s exposure. Figure 7.38(b) shows the AFM image of the same interference fringes as those in (a), which was recorded on PMMA. The thickness of PMMA is represented by a photo-contrast in (b), and the thicker PMMA corresponds to the brighter part of the image. The supplied voltage to the filament of electron biprism, VB, was 5.3 V and the spacing of fringes, s, was 108 nm in Figures 7.38(a) and (b). Figure 7.39(a) shows interference fringes of the X-biprism magnified and recorded on a photoplate, and (b) shows the AFM image of interference fringes recorded on PMMA. The supplied

(a)

1 µm

(b)

0.1µm 1µm

Figure 7.38 (a) Two-wave interference fringes magnified and recorded on a photoplate. (b) Interference fringes corresponding to (a) recorded on PMMA. VB: 5.3 V and s: 108 nm.

High resolution lithography 217 voltage to the filament, VB, was 5.0 V and the spacing of fringes, s, was 125 nm in Figures 7.39(a) and (b). In Figure 7.39(a), dot patterns are found at the intersection where four-wave interference occurred and line patterns around the dot patterns where two-wave interference occurred. While, in Figure 7.39(b), about 10 10 dots are recognized, but lines are not observed, owing to the reduction of dose. Consequently, Figures 7.38(b) and 7.39(b) show that line and dot patterns were fabricated successfully, and the dose needed for lines is about twice as that for dots. More precise fabrication would be possible by optimizing the dose. In order to produce finer patterns than 100-nm in period, the larger overlapping angle ␪, that is, the larger supplied voltage to the filament VB, should be selected.

(a)

1 µm

(b)

0.1µm 1µm

Figure 7.39 (a) Two- and four-wave interference fringes magnified and recorded on a photoplate. (b) Interference fringes corresponding to (a) recorded on PMMA. VB: 5.0 V and s: 125 nm.

218

S. Matsui

A simple assessment suggests that the spacing, s, becomes 1 nm when VB is 2.4 kV with the same electron optics. Carbon contamination line patterns with a period of 20 nm was fabricated by a 30-kV SEM.53 7.4.2

Atomic beam holography

Atomic manipulation based on a holographic principle has been demonstrated by using a laser trap technique and a computer generated hologram (CGH) made by EB lithography.54 One approximation of a CGH is the binary hologram, in which the hologram takes a binary value, either 100% transparent or 100% opaque. This hologram can be directly translated to a hologram for atomic de Broglie waves, by cutting out the pattern on a film that is equal to the pattern of the 100% transmission area of the binary hologram. A monochromatic atomic wave reconstructs an atomic pattern by passing the hologram. The hologram used in this experiment was a Fourier hologram, which produced the Fourier-transformed wavefront of the object. When the hologram is illustrated with a plane wave, the far-field pattern of the diffracted wave produces an image of the object. The object used in this experiment was a transparent F-shaped pattern, in which the transparent portion had a constant amplitude and random phase distribution. The object was represented by the complex transmission amplitude at points on a 128 128 matrix covering the F-shape pattern. The two-dimensional array of numbers was Fourier-transformed using a fast Fourier transform (FFT) algorithm, and the resulting 128 128 complex areas (cells) of the Fourier hologram. The transmission function of each cell of the hologram was expressed by a matrix of 4 4 subcells. A 100-nm-thick SiN menbrane was used for the hologram. The binary pattern was transferred to a ZEP resist on the SiN membrane by an EB writing system. Subsequent CF4 plasma etching created through-holes in the membrane. A scanning electron micrograph of the hologram is shown in Figure 7.40. The size of the subcell was 0.3 0.3 m square, so the size of the entire hologram was 153.6 153.6 m. To increase the intensity of the deflected beam, the same pattern was repeated 10 times along the x and y directions, making the overall size of the hologram 1.5 1.5 mm. A schematic diagram of this experiment is shown in Figure 7.41. The ultra-cold Ne atomic beam was generated by the reported method.55 The cloud of Ne atoms in the trap was ~0.3 mm in diameter, and the one-directional average velocity of the atoms was 20 cm s1. The hologram was placed 40 cm below the trap and was mounted on the top of a 0.2-mm-diameter diaphragm. The size of the diaphragm limited the resolution of the image of the Fraunhofer hologram. The position of the hologram was not adjusted because any small portion of the hologram could produce the same image. The average atomic velocity at the hologram was 2.8 m s1, corresponding to a de Broglie wavelength ␭ of 7.1 nm. The acceleration due to gravity reduced the relative velocity spread to ~0.28%. To detect the Fraunhofer diffracted pattern from the hologram, multi-channel plate (MCP) detector was placed 45 cm below the hologram. Figure 7.42(a) shows the reconstructed F-pattern. The data was accumulated for 10 h, and the total atom number

High resolution lithography 219

6 µm

Figure 7.40 Binary CGH hologram on SiN membrane made by EB lithography. Transfer laser

Zeeman slower

Magneto-optic trap

Deflector 1s5 Ne*

–

Cooling laser

+

DC-discharge

1s3 Ne* Hologram

MCP

Figure 7.41 Experimental apparatus of atomic beam holography.

of spots on the figure was 6 104. Figure 7.42(b) shows another example of reconstructed patterns, which represents characters of “atom, Ne, and ”. In this experiment, a focusing lens for imaging was not used, but it is possible to combine the function of a focusing lens into the hologram.56 In such a hologram, the resolution is determined by the same rule as applies to an optical lens.

220

S. Matsui (a)

(b)

Figure 7.42 Reconstructed image: (a) “F” pattern and (b) “atom, Ne, and ” pattern.

The binary hologram does not control the phase and amplitude of the wave inside a hole. When the hologram is the sole component for atomic-beam manipulation, therefore, the practical limit is approximately the minimum size of through-holes, which is in the range 10–100 nm.

7.5

Summary

Nanofabrication and its application to nanodevices using EB and FIB have been discussed. Comparison of the sizes between artificial and biological structures is shown in Figure 7.43. EB lithography with a commercial available machine using an organic resist has already achieved 10 nm features, which is the same size as a virus. Features 10–1 nm in size can also be fabricated by using inorganic resists. This indicates that even a minimum pitch of 3.4 nm of DNA can be delineated. Furthermore, a new atomic structure on a Si surface can be created by atomic

High resolution lithography 221

100

100 nm

10 nm Calixarene pattern

Rotavirus particles

Size (nm)

60 nm 10

5 nm 5 nm (Li, Al)F pattern

1 DNA of bacteriophage

New structure formation on Si(111) by STM 0.1 Biological

Microfabricated Structure type

Figure 7.43 Comparison of the sizes between artificial and biological structures.

manipulation with STM.57 Electron- and ion-beam-induced surface reaction is a very interesting and promising three-dimensional process, and much progress is expected in the nanoscale regime. As applications to nanodevices of EB lithography, a 40-nm-gate NMOS, a 14-nm-gate-length EJ-MOSFET, and a hightemperature operation SET have been demonstrated by EB nanolithography using an organic resist. FIB technology has been important as diagnostic and micromechanics tool. As a new approach, material-wave nanotechnology using electron-and atomic-beam holography has been developed. This technique allows nano- and atomic-scale structures to be produced simultaneously. The low cost and high throughput are very important for commercialization of electronic, optoelectronic, and magnetic nanodevices. Nanoimprinting lithography, which is expected to achieve both cost and throughput efficiencies, has been developed to fabricate practical nanodevices.58

222

S. Matsui

References 1. F. Emoto, K. Gamo, S. Namba, N. Samoto, and R. Shimizu, Jpn. J. Appl. Phys. 24, L809 (1985). 2. W. Chen and H. Ahmed, Appl. Phys. Lett. 63, 1116 (1993). 3. K. Kurihara, K. Iwadate, H. Namatsu, M. Nagase, H. Takenaka, and K. Murase, Jpn. J. Appl. Phys. 34, 6940 (1995). 4. T. Yoshimura, Y. Nakayama, and S. Okazaki, J. Vac. Sci. Technol. B10, 2615 (1992). 5. J. Fujita, Y. Ohnishi, Y. Ochiai, and S. Matsui, Appl. Phys. Lett. 68, 1297 (1996). 6. M. Isaacson and A. Murray, J. Vac. Sci. Technol. 19, 1117 (1981). 7. D. R. Allee and A. N. Broers, Appl. Phys. Lett. 57, 2271 (1990). 8. J. Fujita, H. Watanabe, Y. Ochiai, S. Manako, J. S. Tsai, and S. Matsui, Appl. Phys. Lett. 66, 3065 (1995). 9. A. N. Broers and W. W. Molzen, J. J. Cuomo, and N. D. Wittles, Appl. Phys. Lett. 29, 596 (1976). 10. R. L. Kubena, J. W. Ward, F. P. Stratton, R. J. Joyce, and G. M. Atkinson, J. Vac. Sci. Technol. B9, 3079 (1991). 11. H. Nakazawa, H. Takemura, M. Isobe, Y. Nakagawa, M. Hassel Shearer, and W. Thompson, J. Vac. Sci. Technol. B6, 2019 (1988). 12. Z. W. Chen, G. A. C. Jones, and H. Ahmed, J. Vac. Sci. Technol. B6, 2009 (1988). 13. M. Gesley, J. Vac. Sci. Technol. B10, 2451 (1992). 14. B. H. Koek, T. Chisholm, J. P. Davey, J. Romijin, and A. J. V. Run, Jpn. J. Appl. Phys. 32, 5982 (1993). 15. H. Hiroshima, S. Okayama, M. Ogura, and M. Komuro, J. Vac. Sci. Technol. B13, 2514 (1995). 16. Y. Ochiai, M. Baba, H. Watanabe, and S. Matsui, Jpn. J. Appl. Phys. 30, 3266 (1991). 17. T. Nishida, M. Notomi, R. Iga, and T. Tamamura, Jpn. J. Appl. Phys. 31, 4508 (1992). 18. T. Yoshimura, H. Shiraishi, J. Yamamoto, and S. Okazaki, Appl. Phys. Lett. 63, 764, (1993). 19. A. Murry, M. Isaacson, and I. Adesida, Appl. Phys. Lett. 45, 589 (1984). 20. E. Kratschmer and M. Isaacson, J. Vac. Sci. Technol. B5, 369 (1987). 21. W. Langheinrich, A. Vescan, B. Spangenberg, and H. Beneking, Microelectron. Eng. 7, 287 (1992). 22. W. Langheinrich and H. Beneking, Jpn. J. Appl. Phys. 32, 6218 (1993). 23. S. Matsui and T. Ichihashi, Appl. Phys. Lett. 53, 842 (1988). 24. S. Matsui, T. Ichihashi, and M. Mito, J. Vac. Sci. Technol. 7, 1182 (1989). 25. A. Sato, Y. Tsukamoto, M. Baba, and S. Matsui, JJAP Series 5, Proc. 1991 Intern. MicroProcess Conference pp. 409–412 (1991). 26. H. W. P. Koops, J. Kretz, M. Rudolph, M. Weber, G. Dahm, and K. L. Lee, Jpn. J. Appl. Phys. 33, 7099 (1994). 27. M. Komuro and H. Hiroshima, Microelectron. Eng. 35, 273 (1997). 28. Y. Taur, S. Wind, Y. J. Mii, Y. Lii, D. Moy, K. A. Jenkins, C. L. Chen, P. J. Coane, D. Klaus, J. Bucchignano, M. Rosenfield, M. G. R. Thomson, and M. Polcari, IEDM93 Technical Digest, p. 127 (1993). 29. K. F. Lee, R. H. Yan, D. Y. Jeon, G. M. Chin, Y. O. Kim, D. M. Tennant, B. Razavi, H. D. Lin, Y. G. Wey, E. H. Westerwick, M. D. Morris, R. W. Johnson, T. M. Liu, M. Tarsia, M. Cerullo, R. G. Swartz, and A. Ourmarzd, IEDM93, Technical Digest, p. 131 (1993). 30. M. Ono, M. Saito, T. Yoshitomi, C. Fiegna, T. Ohguro, and H. Iwai, IEDM93, Technical Digest, p. 119 (1993).

High resolution lithography 223 31. Y. Ochiai, S. Manako, S. Samukawa, K. Takeuchi, and T. Yamamoto, Microelectron. Eng. 30, 415 (1996). 32. H. Kawaura, T. Sakamoto, Y. Ochiai, J. Fujita, and T. Baba, Extended Abstract of the 1997 International Conference on Solid State Devices and Materials, Hamamatsu, Japan, pp. 572–573 (1997). 33. Y. Takahashi, M. Nagase, H. Namatu, K. Kurihara, K. Iwadake, Y. Nakajima, S. Horiguchi, K. Murase, and M. Tabe, Electron. Lett. 131, 136 (1995). 34. A. Fujiwara, Y. Takahashi, K. Murase, and M. Tabe, Appl. Phys. Lett. 67, 2957 (1995). 35. Y. Takahashi, H. Namatsu, K. Kurihara, K. Iwadate, M. Nagase, and K. Murase, IEEE Trans. Electron. Device 43, 1213 (1996). 36. J. Melngailis, J. Vac. Sci. Technol. B5, 469 (1987). 37. L. R. Harriott, Nucl. Instrum. Methods B55, 802 (1991). 38. S. Matsui, Nanotechnology 7, 247 (1996). 39. V. E. Krohn and G. R. Ringo, Appl. Phys. Lett. 27, 479 (1975). 40. R. L. Seliger, J. W. Ward, V. Wang, and R. L. Kubena, Appl. Phys. Lett. 34, 310 (1979). 41. L. W. Swanson, G. A. Schwind, and A. E. Bell, J. Appl. Phys. 51, 3453 (1980). 42. E. Miyauchi, H. Arimoto, H. Hashimoto, T. Furuya, and T. Utsumi, Jpn. J. Appl. Phys. 22, L287 (1983). 43. S. Matsui, Y. Kojima, Y. Ochiai, and T. Honda, J. Vac. Sci. Technol. B9, 2622 (1991). 44. H. Morimoto, H. Onoda, T. Kato, Y. Sasaki, K. Saitoh, and T. Kato, J. Vac. Sci. Technol. B4, 205 (1986). 45. K. Nikawa, K. Nasu, M. Murase, T. Kaito, T. Adachi, and S. Inoue, Proc. 27th Int. Reliability Physics Symp. IEEE, New York, p. 43 (1989). 46. E. C. G. Kirk, R. A. McMahan, J. R. A. Cleaver, and H. Ahmed, J. Vac. Sci. Technol. B6, 1940 (1988). 47. R. Pantel, G. Auvert, G. Mascarin, and J. P. Gonchond, 1994 Interdisciplinary developments and tools Proc. Electron Microscopy, ICEM 13, Paris, 17–22 July, 1994, 1, p. 1007. 48. H. Saka, K. Koroda, M. H. Hong, T. Kamino, T. Yaguchi, H. Tsuboi, T. Ishitani, H. Koike, A. Shibuya, and Y. Adachi, 1994 Interdisciplinary development and tools Proc. Electron Microscopy, ICEM 13, Paris, 17–22 July 1994, 1, p. 1009. 49. P. G. Blauner, Proc. 1991 Int. MicroProcess Conf. p. 309 (1991). A. Wagner, J. P. Levin, J. I. Mauer, P. G. Blauner, S. J. Kirch, and P. Longo, J. Vac. Sci. Technol. B8, 1557 (1991). 50. S. Matsui, T. Kaito, J. Fujita, M. Komuro, K. Kanda, and Y. Haruyama, J. Vac. Sci. Technol. B18, 3181 (2000). J. Fujita, M. Ishida, T. Sakamoto, Y. Ochiai, K. Kaito, and S. Matsui, J. Vac. Sci. Technol. B19, 2834 (2001). 51. K. Ogai, S. Matsui, Y. Kimura, and R. Shimizu, Jpn. J. Appl. Phys. 32, 5988 (1993). 52. K. Ogai, Y. Kimura, R. Shimizu, J. Fujita, and S. Matsui, Appl. Phys. Lett. 66, 1560 (1995). 53. S. Fujita, S. Maruno, H. Watanabe, Y. Kusumi, and M. Ichikawa, Microelectron. Eng. 30, 435 (1996). 54. J. Fujita, M. Morinaga, T. Kishimoto, M. Yasuda, S. Matsui, and F. Shimizu, Nature 380, 691 (1996). 55. F. Shimizu, K. Shimizu, and H. Takuma, Opt. Lett. 16, 339 (1991). 56. M. Morinaga, M. Yasuda, T. Kishimoto, F. Shimizu, J. Fijita, and S. Matsui, Phys. Rev. Lett. 77, 802 (1966). 57. M. Baba and S. Matsui, Appl. Phys. Lett. 65, 1927 (1994). 58. S. Y. Chou, P. R. Krauss, and P. J. Renstrom, Science 72, 85 (1996).

8

Advanced plating technology for electronics packaging Hideo Honma and Hideto Watanabe

8.1

Introduction

Recently, miniaturization of integrated circuit (IC) and large-scale integrated (LSI) circuit has become essential with the downsizing of electronics devices. Wire bonding has been applied to connect the external circuit and the IC chips. However, the miniaturization of the chip substrate distance is demanded with the downsizing of circuit packaging. Accordingly, the connection between the circuit and IC by using microbumps or anisotropic conductive film becomes an indispensable technology. On the other hand, PWBs are shrunk in size and increase in circuit densities with the progress of electronics devices. Therefore, the development of build-up PWBs as the alternative technology is replaced instead of the conventional multi-layer PWBs with through-holes. In this new technology, the plating of micro-vias is applied to the connection between each conductive layers. However, the planarization of dielectric layer on the micro-vias filling is very difficult in the case of the conventional plating process. Therefore, the demand of micro-via filling by electroplating increases. In this chapter, microconnection of circuit packaging, especially, microbump formation, preparation of conductive particles, and via-filling using wet process will be described.

8.2

Fabrication of gold bumps using the disulfiteaurate complex1,2

Generally, soft gold plating is used in the electronics industry, especially for semiconductor devices. Baths for the soft gold deposition are generally based on KAu(CN)2, KH2PO4, potassium citrate, and trace amounts of additives.3 However, cyanide in the electrolytes tends to penetrate the organic photoresist, which is applied to fabricate metal bumps or conductive circuit patterns, which causes insulating problems. In comparison, cyanide-free gold plating solution is not sensitive to the photoresist; therefore, the straight walled gold bumps can be formed. Recently, the straight walled gold bump formation using cyanide-free gold plating was studied.1,2,4,5 However, the bath stability of the cyanide-free gold plating is lower than the baths containing cyanide. In this section, the improvement of bath stability and the techniques of microbump formation are described.

Advanced plating technology for packaging 225 8.2.1

Stability of cyanide-free gold plating bath

Metallic and black colored colloids are gradually formed as the plating proceeds and finally plating bath is decomposed. The plating bath was separated into cathode and anode chambers by an ion-exchange membrane and determine the decomposition reaction. The decomposition reaction by colloidal precipitate was detected only in the cathode chamber. Therefore, only cathodic reaction of gold (I) ions seems to be involved in the decomposition reaction. A color change from transparence to purple is also found before decomposition of the solution. As shown in Figure 8.1, a new absorption peak appears at 313 nm after plating, and the peak gradually increased with the increase of the plating time, and finally a black colloidal precipitate is formed. The absorption peak originated from a change in the gold complex ions. The blackish purple precipitates were analyzed by energy dispersive spectrometer, and only gold was detected. From these results, it is confirmed that the metallic gold and gold (III) ions accumulate in the bath with passing of the electrolysis. Shirai and co-workers6 reported that decomposition process of gold sulfite complex ions is composed of the three following steps, and they verified that the charge transfer is a rate-determining step. Na3Au(SO3)2 l 3Na Au(SO3)3 2

(1)

2 Au(SO3)3 2 l AuSO3 SO3

(2)

AuSO 3

(3)

e l Au

313 nm

4.0

SO2 3

Absorption

4

5

1.750 3 2 6 1

–1.5 200

300 400 Wave length (nm)

500

Figure 8.1 Absorption spectrum curves for decomposing gold sulfite complex solution. 1: Standard, 2: 50 C, 3: 100 C, 4: 400 C, 5: HAuCl4 (as Au 80 ppm), 6: 2,2 -dipyridine (100 ppm), 900 C.

226

H. Honma and H. Watanabe Table 8.1 Addition effects of various stabilizers on improvement of bath stability Stabilizer

Quantity of electricity (C)

Standard bath 1,5-Naphthalenedisulfonic acid (1 g/L) Benzensulfonic acid (1 g/L) m-Aminobenzensulfonic (1 g/L) Pyridine (100 ppm) 2,2 -Dipyridine (1 ppm) 2,2 -Dipyridine (10 ppm) 2,2 -Dipyridine (100 ppm) o-Phenanthroline (200 ppm) Neocuproin (200 ppm)

30 30 33 55 40 60 400 No decomposition 300 260

Free gold (I) ions accumulate at the reaction interface and cause a disproportionation reaction proceeds according to Eqn (4). 3Au l 2Au Au3

(4)

The bath stability can be improved by the retardation of the disproportionation reaction. Various kinds of organic compound that can form complexes with gold (I) ions were added to the plating solution in order to improve the bath stability. As shown in Table 8.1, the plating bath without additives decomposed after passing 30 C of charge. On the other hand, a remarkable improvement in stability was achieved by the addition of 2,2 -dipyridine to the bath. The adsorption peak based on gold (III) ions at 313 nm did not appear and no blackish purple colloidal precipitate was formed, even after passing 900 C in the presence of 2,2 -dipyridine addition. As shown in Figure 8.1, 2,2 -dipyridine and its derivatives cause a retardation effect on the decomposition reaction. It is believed that the stability improvement is due to the presence of unshared electron pair associated with the 2,2 -dipyridine and derivatives that allows the formation of complex ion with gold (I) at the discharge interface. 8.2.2

Selection of supporting electrolyte for bump formation

Generally, a supporting electrolyte is added in the plating solution to improve the electric conductivity and characteristics of the deposited films. Selection of supporting electrolyte is an important factor to obtain the smooth and soft gold films. Appearances of photo resist and gold bumps after gold electroplating are shown in Figure 8.2. The smooth gold deposition is obtained by using the phosphorus acid of 0.05–0.15 mol/L. However, swelling of the photo resist is observed when phosphorus acid concentration reached to over 0.30 mol/L. Therefore, the straight walled gold bumps which followed the patterns are not obtained in this case. When phosphoric acid is used as the supporting electrolyte, the swelling phenomenon did not occur, however, gold deposits exhibited dendritic feature with

Advanced plating technology for packaging 227

Phosphorus acid Phosphoric acid

Sulfuric acid

Boric acid

(a)

(b)

(c)

10 µm

(d) 80 µm

Figure 8.2 Influences of conductive salt concentration in gold bump formation. (a) 0.05 mol/dm3, (b) 0.15 mol/dm3, (c) 0.30 mol/dm3, (d) appearances of photoresist after gold electroplating (0.30 mol/dm3).

an increase in the phosphoric acid concentration. On the other hand, smooth gold bumps without swelling of the photoresist can be formed using sulfuric acid or boric acid. From these results, selection of the supporting electrolyte is important to form more uniform and smooth straight walled gold bumps, which followed the thick patterns. 8.2.3

Influence of crystal adjustment agent in bump formation

It is reported that the addition of heavy metal ions such as thallium and arsenic is effective to adjust the surface morphology of deposited gold films.7,8 However,

228

H. Honma and H. Watanabe

the properties of deposited gold films are affected by addition of the crystal adjustment agent. Accordingly, the effects of the crystal adjustment agents for the gold bump formation are evaluated. Figure 8.3 shows the SEM images of deposited gold films. Uniform gold films are obtained by adding 1–10 ppm of lead, thallium, cerium or tungsten ions into the plating bath. However, these crystal adjustment agents are codeposited in gold films, which may cause an increase in the hardness of films. Soft gold films are necessary for connection between the wiring and driver IC on the display devices. As shown in Figure 8.4, hardness of deposited gold increased by the addition of lead or thallium ions in the bath. On the other hand, hardness of the gold films decreases by the addition of bismuth, cerium or tungsten ions. From these results, the selection of crystal adjustment 1 ppm

10 ppm

100 ppm

Pb

Tl

Bi

Bath decomposition

Ce

Bath decomposition

W

5 µm Additive free

Figure 8.3 Addition effects of crystal adjustment agents on gold bump formation.

Advanced plating technology for packaging 229

Additive free

87.8

Pb 1 ppm

100.6

Pb 10 ppm

96.6

Pb 100 ppm

97.7

Tl 1 ppm

86.8

Tl 10 ppm

106.8

Tl 100 ppm Bi 1 ppm

96.9 59.7

Ce 1 ppm

75.2

Ce 10 ppm

77.7

W 1 ppm W 10 ppm

92.4 74.5

W 100 ppm

85.1

50

100

150

Hardness (Hk)

Figure 8.4 Influences of crystal adjustment agents on gold film hardness.

agents and additive concentrations are important to fabricate the smooth and soft gold bumps. 8.2.4

Microbump formation by pulse plating

Generally, the straight walled gold bump formation must be carried out under the low current density since the photoresist is deformed under the high current density. Accordingly, the long plating duration is required to form high walled gold bumps. It is reported that the finer grain, higher purity and smoother gold films are rapidly produced by pulse plating.9,10 Figure 8.5 shows the effects of pulse current density and duty ratio on microbump formation. The deposition rate increased with an increase in the pulse current density and the smoothness of deposited gold films decreased. On the other hand, the smoothness of gold deposits was improved by extending the off time of the pulse cycle. This is because the diffusion layer of cathode interface recovers sufficiently during the off time. Gold bumps can be formed at higher speed with increasing the current density and duty ratio. However, striped patterns associated with many hydrogen gases are recognized on the sidewall of deposited gold bumps. The gold ions of the plating reaction interface decreased with an increase in the pulse current density and hydrogen gas evolved vigorously. Gold ions at the vicinity of the electrode decrease if the off time is shortened, therefore evolution of hydrogen gas is intensified and pH increases at the plating reaction interface. As a result the photoresist is attacked; therefore the gold bumps were deformed. On the other

230

H. Honma and H. Watanabe = 0.2

= 0.1

= 0.05

= 0.02

= 0.01

(a)

(b)

10 µm

Figure 8.5 Effect of pulse current density and duty ratio on microbumps formation. (a) 5 A/dm2, (b) 10 A/dm2.

(a)

(b)

(c)

(d)

10 µm

Figure 8.6 Appearances of microbump by pulse plating. (a) i0.5 A/dm2, 60 min, (b) ip10 A/dm2, ␪0.1, 8 min → ip2 A/dm2, ␪0.25, 55 min (c) ip10 A/dm2, ␪0.1, 15 min→ ip2 A/dm2, ␪0.25, 32 min (d) ip10 A/dm2, ␪0.1, 22 min→ ip2 A/dm2, ␪0.25, 10 min.

hand, excellent appearances of gold bumps were formed with pulse current density at 5 A/dm2, and duty ratio of 0.02. The formation of straight gold bumps with a smooth wall is difficult to obtain under the single pulse condition. Accordingly, high pulse current density condition (ip 10 A/dm2, ␪0.2) and low pulse current density condition (ip 2 A/dm2, ␪0.25) were concurrently applied and the fabrication of smooth gold bump at high speed was attempted. As shown in Figure 8.6, the gold bumps following the patterns can be formed without damaging the organic photoresist by concurrently applying the pulse plating condition. Furthermore, plating time is reduced to half in comparison with the direct current plating by initially applying high current followed by pulse plating with low current.

Advanced plating technology for packaging 231

(a)

Photoresist

(b)

Dissolved oxygen

(c)

Electroplating

(d)

No deposition

Deposited gold

40 µm

Figure 8.7 Influences of dissolved air on gold bump formation.

8.2.5

Influence of micro bubbles in the plating bath for bump formation

Skip deposits and gas pits were often observed even at the optimum plating conditions. As shown in Figure 8.7, micro bubbles produced by air in the gold plating bath tend to adhere to the organic photoresist patterns and gas pits, and pattern deformation of resist were observed. The elimination of micro bubbles from the plating bath is evaluated by ultrasonic vibration (100 Hz, 15 min) low frequency vibration (80 Hz), and heating the plating bath after construction of the bath. No significant effect was observed and the deformation of organic photoresist and skip plating of gold were confirmed by the ultrasonic vibration treatment and the low frequency vibration. On the other hand, micro bubbles stuck on the substrate were completely eliminated by raising the temperature to 80C. In this way, dissolved air in the plating bath can be removed and straight walled bumps following the patterns are obtained.

8.3

Nickel bump formation on aluminum substrate using electroless plating11,12

Generally, aluminum is used as the electrode or interconnections material, and sputter-deposited Cr or Ti under layer is applied as a first barrier layer in order to

232

H. Honma and H. Watanabe

improve the adhesion strength between the aluminum substrate and the deposited metal. Subsequently, W, Cu or Ni is formed as the second barrier layer for the prevention of metal diffusion and is followed by the bump formation with Au, Cu or solder plating. Recently, the bump formation without these sputtering processes has been demanded. Usually, zincate process has been applied to initiate electroless nickel plating on aluminum substrate. However, an organic photoresist cannot be applicable to the bump formation, because the zincate bath is a strong alkali. Accordingly, inorganic resist such as phosphosilicate glass (resist film thickness: 0.9 m) has been used. However, only the mushroom bumps can be formed by this process. Therefore, electoroless nickel plating on the aluminum substrate without zincate process is required to fabricate the straight walled nickel bumps. 8.3.1

Bump formation by nickel displacement process

Electroless nickel plating reaction on the aluminum is not initiated because aluminum does not have a catalytic action on the oxidation reaction of hypophosphite. Accordingly a catalytic nickel site is formed only in a required area by displacement between nickel ions and the aluminum substrate. As shown in Table 8.2(a) dissolved oxygen concentration (DO) is reduced with nitrogen purging in each rinsing step in order to prevent the oxidation of the aluminum substrate. Figure 8.8 shows SEM images of the initial displaced nickel and the following electroless nickel films. Fine nickel particles were formed in the early stages of the reaction. These particles however, were agglomerated as the reaction proceeds. Therefore, the roughness of the deposited nickel films increased with the reaction time since electroless nickel plating reaction progresses on the displaced nickel particles. Table 8.2 Electroless nickel plating process (a) Nickel displacement process

(b) Direct nickel plating process

Conditioning ↓ Rinsing (Deaerated by N2, R.T., 15 s) ↓ Acid etching (5%-HF, R.T., 30 s) ↓ Rinsing (Deaerated by N2, R.T., 15 s) ↓ Nickel displacement plating

Conditioning ↓ Rinsing (Deaerated by N2, R.T., 15 s) ↓ Acid etching (5%-HF, R.T., 30 s) ↓ Rinsing (Deaerated by N2, R.T., 15 s) ↓ Activating (0.05 mol/dm3-DMAB, Deaerated by N2, 40 s) ↓ Electroless nickel plating

↓ Electroless nickel plating

Advanced plating technology for packaging 233

5 min

10 min

20 min

(a)

10 µm

(b)

10 µm

(c) 0.5 µm

Figure 8.8 Appearances of initial deposited nickel and electroless nickel film on various displacement plating time. (a) After nickel displacement plating, (b) after electroless nickel plating, (c) roughness of electroless nickel film.

Figure 8.9 shows SEM images of the initial displaced nickel particles and the electroless nickel films on the pH value of nickel displacement plating. The nickel displacement plating rate increased by increasing the pH values. Therefore, displaced nickel was agglomerated like island at pH 10 and the exfoliation of the deposited nickel film was recognized during the electroless nickel plating and resulted in poor adhesion. Furthermore, penetration of the plating solution into the resist and the distortion of the resist also occurred at pH 10. On the other hand, uniform and fine nickel displaced particles were formed at pH 8 and smooth electroless nickel films were formed since the electroless plating reaction occurred on the finely displaced nickel particles. The penetration of the plating solution and the distortion of the resist were not observed under these conditions. Based on the results nickel bumps were formed on the aluminum substrate without the zincate process as shown in Figure 8.10. The straight walled nickel bumps with smooth surfaces were formed by controlling the nickel displacement rate. In this process control of the nickel displacement conditions is important to improve the adhesion between aluminum and nickel and roughness of electroless nickel films.

234

H. Honma and H. Watanabe

pH = 8

pH = 9

pH = 10

(a)

10 µm

(b)

10 µm

(c) 0.5 µm

Figure 8.9 Appearances of initial deposited nickel and electroless nickel film on various pH values of displacement plating bath. (a) After nickel displacement plating, (b) after electroless nickel plating, (c) roughness of electroless nickel film. (a)

(b)

20 µm

Figure 8.10 Effects of displaced nickel conditions on nickel bump formation. (a) Basic displacement bath composition, (b) controlled displacement bath composition.

8.3.2

Bump formation by direct nickel plating

If the active surface of aluminum substrate is maintained electroless nickel films can be deposited on the aluminum without nickel displacement process. In this

Advanced plating technology for packaging 235 section, electroless nickel plating on aluminum substrate without nickel displacement deposition is described. Direct electroless nickel plating process is shown in Table 8.2(b). In this process, the activation treatment with 0.05 mol/L-dimethyl amine borane (DMAB) was performed pretreatment in order to obtain the active aluminum surface and was followed by the electroless nickel plating. Figure 8.11 shows the appearances of nickel deposits in the early stages. Nickel particles were coalesced with increasing nickel displacement plating time (Figure 8.11(a)). On the other hand, nickel particles from direct nickel plating were finer than that of nickel displacement plating and were not coalesced with the progress of the electroless nickel plating. Therefore, as shown in Figure 8.12, uniform nickel bumps can be fabricated on the aluminum electrode. However, nodules were often observed around corner areas of the patterns under the optimum conditions as shown in Figure 8.13(a). It is reported that the linear diffusion reaction proceeds on a plane surface and nonlinear diffusion reaction occurred on the corner areas.

(a) Ni displacement plating

(b) Direct Ni plating

3 min

5 min

10 min

20 µm

Figure 8.11 Appearances of nickel particles in the early stages. (a) Nickel displacement plating process, (b) direct nickel plating process.

236

H. Honma and H. Watanabe

(a)

(b)

20 µm

Figure 8.12 Appearances of initial deposited nickel and nickel bumps after peeling photoresist. (a) Nickel displacement plating process, (b) direct nickel plating process.

(a)

(b)

40 µm

Figure 8.13 Effects of additives on prevention of extraneous nickel deposition. (a) Additive free, (b) Bi2 0.1 ppm.

Therefore, the nonlinear diffusion reaction causes the nodular nickel deposition, because the electroless plating reaction is enhanced at the corner by promoting the reaction (Figure 8.14). Accordingly, lead bismuth, or arsenic ions as the catalytic poisoner were added into the electroless nickel plating bath for the prevention of the nodular deposition. As shown in Figure 8.13(b), nickel deposition without nodule can be obtained by adding catalytic poisoner since the catalytic poisoner adsorbed at the corner and the deposition reaction is inhibited.

8.4

Via-filling using electroplating for build-up printed circuit boards13,14

Recently, PWBs have been downsized with miniaturization of electronic devices. Conventional multi-layer PWBs have a limitation to the higher packaging densities. Therefore, build-up process has been adopted as a new multi-layered PWBs manufacturing process.15,16 The process is complicated because planarization can be accomplished by polishing after an insulation layer formation. Accordingly,

Advanced plating technology for packaging 237 Catalytic poisoner

Figure 8.14 Effects of nonlinear diffusion on microbump formation.

(a)

(b)

50 µm

Figure 8.15 Cross-sectional views of plated copper with applied dc plating followed by PR plating. (a) 2 h, (b) 3 h.

the layer-to-layer connection with via-holes is developed.17,18 As the conductor patterns become finer a filling of via-holes with copper plating has become an effective process. Therefore, the via-filling by electroless plating,19 periodical reverse (PR) plating20 and direct current (dc) plating were evaluated. In this section, especially via-filling using dc plating is described. 8.4.1

Via-filling by low copper concentration bath

The via-filling can be achieved by PR plating.20 However, supplemental equipment is necessary besides the existing lines. Also, an irregularity or roughness at

238 (a)

H. Honma and H. Watanabe (b)

50 µm

Figure 8.16 Cross-sectional views of plated via-hole by dc plating. (a) 1 A/dm2, (b) 5 A/dm2.

deposited copper surface increased since anodic current must be periodically applied to dissolve deposited copper at an open aperture of via-holes as shown in Figure 8.15. Therefore, dc plating is indispensable technology for the via-filling. Via-filling with low copper concentration bath and higher copper concentration bath will be described. Figure 8.16 shows a relationship between current density and filling condition after 100 C of electricity by dc plating using conventional high throwing power bath (low copper concentration bath). Filling ratio is about 16% at 1 A/dm2 of current density and is decreased with increasing current density; therefore, it reached about 3% at 5 A/dm2. In those conditions no appropriate filling ratio can be obtained, because the current lines are more concentrated at the aperture area of via-holes. As mentioned above, it is difficult to fill the viahole by dc plating. However, via-filling ratio can be improved by the addition of leveling agent. Figure 8.17 shows the addition effects of leveling agents into low copper concentration bath on the via-filling process. A thickness at the center of via-holes decreases under the condition of high current density. On the other hand, the thickness increases with a decrease in the current density and filling ratio increases to 37% under the current density at 0.1 A/dm2. However, over 30 h are necessary to fill out the via-holes. Figure 8.18 shows the applicability of imposing the pulse plating to the low copper concentration bath with additives. The filling ratio is improved to 44% since deposits at aperture areas of via-holes are accelerated by the concentration of the current lines. From the above results, decreasing the current density, adding the leveling agent, imposing the pulse plating and optimizing the pulse conditions are necessary for improving the filling condition in case of filling of via-hole using low copper concentration bath. 8.4.2

Via-filling by high copper concentration bath

Figure 8.19 shows cross-sectional views of plated via-hole using high copper concentration bath with additives. Filling ratio is improved with decreasing the

Advanced plating technology for packaging 239

(a)

(b)

(c)

(d)

50 µm

Figure 8.17 Effects of current density on filling ratio for dc plating. (a) 0.1 A/dm2, (b) 1 A/dm2, (c) 3 A/dm2, (d) 5 A/dm2. (a)

(b)

(c)

50 µm

Figure 8.18 Applicability of imposing pulse plating to via-filling. (a) 1 A/dm2 (60 s) → 3 A/dm2 (0.1 s), (b) 1 A/dm2 (60 s) → 1.5 A/dm2 (0.1 s), (c) 1 A/dm2 (60 s) → 1.25 A/dm2 (0.1 s).

current density even for low copper concentration bath. Via-holes are buried completely and a flat surface is formed at current density of 1 A/dm2, and the filling ratio indicated 94%. On the other hand, the filling ratio decreases with imposing pulse plating. It is supposed that the filling ratio is improved with increasing

240

H. Honma and H. Watanabe

50 µm

Figure 8.19 Cross-sectional views of plated via-hole using high concentration bath.

copper concentration of plating bath because reacting species are easy to supply into via-holes at high copper concentration.

8.5

Preparation of anisotropic conductive particles by electroless nickel plating21,22

Solder is widely applied as a mounting technology for electronics components, because connection and repair are easily accomplished. However, not only fine pitch connections become difficult with the progress of high density mounting, but also the environmental problems based on regulation of lead and volatile organic compound have occurred. A leadless connection such as conductive pastes, anisotropic conductive films and other mechanical connection method has been examined. Especially, mechanical solderless bonding with small gold bumps and metal balls has increased in the electronics devices.23 The conductive particles (5–7 m diameter) are prepared by electroless nickel plating and displacement gold plating. Generally, batch type electroless plating is applied to provide conductivity on the nonconductor. However, the plating bath becomes unstable particles, because the surface areas of particles are much larger than the bulk substrate. Accordingly, the continuous dropping method as the plating technology is applied for the preparation of the conductive particles. In this section, the plating techniques for the formation of nickel particles are introduced. 8.5.1

Catalyzing treatment of fine particles

Recently, resin particles, metal particles and carbon particles are being used for mechanical bonding.24 Especially, the metalized resin particles have been applied for the anisotropic conductive films. The resin particles are coated with electroless nickel plating, subsequently, displacement gold plating is done because these particles are nonconductors. Optimization of catalyzing process is necessary to form the conductive particles that have uniform surface. Therefore, effects of

Advanced plating technology for packaging 241 Table 8.3 Relationship between SnCl2 and PdCl2 concentration and uniformity of deposited nickel films. (a) Excellent, (b) Good, (c) Fair

SnCl2 0.5 g/dm3 SnCl2 1.0 g/dm3 SnCl2 2.0 g/dm3

PdCl2 0.025 g/dm3

PdCl2 0.05 g/dm3

PdCl2 0.1 g/dm3

B

B

B

C

A

C

B

C

C

catalyzing solution on uniformity of deposited nickel films were evaluated. When fine particles are catalyzed with palladium ion solution, palladium dissolved into the plating solution. The palladium that dissolved into the bath become nuclei and oxidation of hypophosphite progresses and leads to bath decomposition. Also, a colloidal catalyst and an alkaline catalyst process are applied for the catalyzing treatment; however, nodules are observed extensively on nickel films. On the contrary, uniformity is excellent by using tin palladium two-step process in comparison with other catalyst processes. In the case of tin palladium two-step process, optimization of tin chloride and palladium chloride concentration is important in order to obtain uniform nickel films on the particles. Table 8.3 shows effect of tin chloride and palladium chloride concentration on the uniformity of deposited nickel films. Extraneous deposits are obtained on the nickel films along with an increase of each concentration. Also, uniformity of nickel films decrease with a decrease in tin chloride concentration. On the other hand, uniform nickel films can be obtained by using 0.1 g/L of tin chloride and 0.05 g/L of palladium chloride. Furthermore, uniformity of nickel films on the particles can be improved by adding cathode surfactant into tin chloride solution. 8.5.2

Metalization method of fine particles

The deposition rate is enhanced and self-decomposition is frequently observed because the specific surface area of particles is large. Accordingly, the continuous dropping method is applied for the formation of conductive particles as shown in Figure 8.20. In this process, the reaction solution is fed only in necessary amounts. The electroless plating reaction progresses properly since each chemical is stored separately and mixed into the reaction solution. The plating solution is divided into three kinds of chemicals, 0.85 mol/dm3 of nickel sulfate with 1.0 mol/dm3 of complexing agent as a metal salt solution, 2.0 mol/dm3 sodium hypophosphite as a reducing agent solution, and 1.0 mol/dm3 of sodium hydroxide as a pH adjustment solution are used. 8.5.3

Surface morphology of deposited nickel

Surface morphology was evaluated by changing pH value, a complexing agent and bath temperature. Figure 8.21 shows effects of pH values and complexing

242

H. Honma and H. Watanabe

Dropping solution NaH2PO2 . H2O 2.0 mol/dm3 NaOH 1.0 mol/dm3

3 ml/min Particle Micro tube pump Dispersant solution NaH2PO2 . H2O 0.2 mol/dm3

NiSO4 . 6H2O 0.85 mol/dm3 Complexing agent 1.0 mol/dm3

pH 4–6, Temperature 25–70°C

Figure 8.20 Experimental apparatus for conductive particle formation by electroless plating. Glycine

Sodium tartrate

Ammonium acetate

pH 6

pH 5

pH 4

5 µm

Figure 8.21 Effects of pH values on uniformity of nickel plated particles.

agent on the surface morphology of deposited nickel films. Only extraneous deposition can be formed at pH 6.0 because the nickel deposition rate is fast. At pH 5.0, the nickel deposition rate decreases and the extraneous deposition is suppressed. However, no deposition by coalescence of particles is observed using glycine as a complexing agent. On the other hand, uniform nickel films can be formed without depending on complexing agents at pH 4.0. Figure 8.22 shows effects of plating bath temperature on the surface morphology of deposited nickel films. The extraneous nickel deposits are suppressed with lowering the bath temperature in the case of the bath using sodium tartrate as a complexing agent. When glycine is used as a complexing agent, the

Advanced plating technology for packaging 243

Glycine

Sodium tartrate

Ammonium acetate

70°C

50°C

25°C

5 µm

Figure 8.22 Effects of bath temperature on uniformity of nickel plated particles.

Deformation of particle (µm)

6.0

3.0 Dimple deposition Extraneous deposition Uniform deposition Brank (resin particle)

0.0 0.0

0.5

1.0 Long (gf)

1.5

2.0

Figure 8.23 Relationship between surface morphology of deposited nickel film and flexibility of conductive particle. O: Crack point.

uniformity of deposited nickel films is improved by decreasing bath temperature from 70- to 50C. However, the uniformity of deposited nickel films decreases in case of decreasing bath temperature under 25C. On the other hand, uniform nickel films can be formed at 50–70C using ammonium acetate. However, the exfoliation of the deposited nickel films occurs at 25C. From these results, the

244

H. Honma and H. Watanabe

optimization of pH and bath temperature, and the selection of complexing agent are important in order to form uniform nickel fine particles. 8.5.4

Flexibility of the nickel fine particles

Generally, anisotropic conductive films which contain fine metalized particles are used in the connection between LCD (liquid crystal display) and TCP (tape carrier package), PWBs and TCP, IC and LCD. Therefore, the flexibility of deposited nickel films is required because an anisotropic conductive fine particle must compressively adhere as an electrode connection material. The flexibility of nickel films was evaluated using the micro compression test device. As shown in Figure 8.23, cracks of deposited nickel films occur under the loads of 0.45 gf on the extraneous deposition. Also, dimple deposits developed over 0.55 gf on the uniform metalized particles.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

H. Honma and Y. Kagaya, J. Electrochem. Soc., 140, L135 (1993). H. Honma and K. Hagiya, J. Electrochem. Soc., 142, 81 (1995). P. Wilkinson, Gold Bull, 19, 3 (1986). A. Gemmler, W. Keller, H. Richer and K. Ruess, Plating and Surface Finishing, 81(8), 52 (1994). T. Osaka, K. Kodera, T. Misato, T. Honma and Y. Okinaka, J. Electrochem. Soc., 144, 3462 (1997). N. Shirai, S. Yoshimura and E. Sato, J. Sur. Finish. Soc., Jpn., 40, 543 (1989). J. D. E. Mclntyre and W. F. Peak, Jr., J. Electrochem. Soc., 123, 1800 (1976). T. E. Dinan and H. Y. Cheh, J. Electrochem. Soc., 139, 410 (1992). O. Izumi, J. Met. Finish. Soc. Jpn., 139, 3 (1988). K. Hosokawa, J. Met. Finish. Soc. Jpn., 139, 10 (1988). H. Honma, H. Watanabe and T. Kobayashi, J. Electrochem. Soc., 141, 1791 (1994). H. Watanabe and H. Honma, J. Electrochem. Soc., 144, 471 (1997). T. Kobayashi, J. Kawasaki, J. Ishibashi, K. Tanaka and H. Honma, J. Sur. Finish. Soc. Jpn., 49, 1327 (1998). T. Kobayashi, J. Kawasaki, J. Ishibashi, K. Tanaka and H. Honma, Proc. Pan Pactific Microelectronics Symp., p. 345 (1999). Y. Tsukada, J. Jap. Inst. Electro. Pack., 11, 308 (1996). Y. Tsukada, J. Jap. Inst. Electro. Pack., 11, 472 (1996). S. Nakatani, S. Nakura and A. Wada, The 9th National Convention Record JIPC, p. 57 (1995). H. Odaira, E. Imamura and H. Wada, The 9th National Convention Record JIPC, p. 55 (1995). S. Abe, M. Ohkubo, T. Fujinami and H. Honma, Trans IMF, 76(1), 12 (1998). T. Fujinami, T. Kobayashi, A. Maniwa and H. Honma, J. Sur. Finish. Soc. Jpn., 48, 660 (1997). K. Hagiwara, J. Watanabe and H. Honma, Plating and Surface Finishing, 84(4), 74 (1997). I. Mochizuki, K. Izawa, J. Watanabe and H. Honma, J. Sur. Finish. Soc. Jpn., 48, 429 (1997). H. Nishida and K. Sakamoto, J. Jpn. Inst. Electro. Peck., 10, 377 (1995). J. Takeshita, J. Sur. Finish. Soc. Jpn., 47, 896 (1996).

9

Micro-electroforming of miniaturized devices for chemical applications Holger Löwe, Wolfgang Ehrfeld and, Jörg Schiewe

9.1

Microreactors – a novel concept in chemical engineering

The use of miniaturized reactors comprising fluidic components, with characteristic dimensions below one millimeter, for various purposes in chemical engineering recently attracted great attention. In the last five years, microreaction technology has been showing a fast development, first from concepts to devices that are now commercially available.1,2 Important milestones during this period were the development of metal microreactor components such as micro heat exchangers and micromixing devices and the integration of these components to build up integrated microreaction systems.3,4 In the field of electrochemistry the micro gap cell has to be regarded in relation to the development of micro-electroorganic reactors.5 Microreactors offer many advantages, in particular for the performance of heat and mass transfer limited reactions. Reduction of the characteristic dimensions in microreactors leads to large gradients, for example, in concentration and temperature, being the driving forces for increased mass and heat flow rates, respectively. This is especially advantageous in the case of transport-limited multiphase reactions as well as for highly exothermal processes. In addition, the fact that very high surface-to-volume-ratios can be achieved in microchannels leads to a highly efficient exchange of matter and heat. Based on these technological advantages, the performance of new process regimes becomes technically feasible. For instance, the fluorination of toluene using fluorine as reactant was carried out successfully in microreactors comprising reaction channels and heat exchanger structures in close proximity. Due to a well-defined residence time along the reaction co-ordinate and a fast and efficient heat transfer, high selectivities with respect to the monofluorinated products could be achieved even at high conversion rates.6 Besides the application of microreactors as a tool for research and development, very recent examples show the advantages of microreactors for a small lot production of chemicals.7,8 In order to qualify microreactors for the specific needs of production processes, the volume or mass throughput in the reactor has to be increased significantly. This can be achieved by “numbering up”, that is, parallel operation of a certain number of microreaction units, thus sparing the time-consuming conventional scale-up process. In principle, identical operating conditions can be assumed for any of these units (microchannels). Therefore

246

Löwe et al.

process development using laboratory scale microreactors can be transferred directly into the production scale by “numbering up”. Thus, microreactors can be profitably utilized in the chemical industry saving development costs and time. Another important motivation for utilizing microreactors for technical processes arises from safety considerations. Small hold-ups of hazardous reactants and the possibility of distributed production in microreactors significantly decrease the expenditure for safety installations.9–11 Even working with oxygen in the explosive regime in a standard laboratory environment becomes possible since microchannels smaller than the quenching distance act as flame arresters.12,13 On the other hand, microreaction technology offers advantages due to the large number of reactions to be performed parallelly in defined units with respect to space and time as a result of mere volume reduction. Consequently, high throughput synthesis of substance libraries and subsequent screening regarding certain material properties is profitably performed in microreactors. These devices providing chemical information are designed in order to contain as small volumes of each substance as is technically feasible. This allows for a high degree of parallelization since information is not related to molar quantity. Laboratory high throughput screening systems, first developed and utilized for drug discovery in pharmaceutical industries, have nowadays found widespread application in material science for identification of new catalysts, ceramics or polymers with designed material properties.14 Microreactors of different design concepts can be profitably utilized for these applications. The development of microreactors was highly dependent on the development of suitable fabrication methods. Powerful fabrication processes originally developed for semiconductor industries, such as silicon micromachining find their applications more and more in other fields including microreaction technology.15 However, the low chemical stability, especially with respect to highly concentrated acids and bases, prevents the widespread use of this material for fabrication of microreactors. In the last decade, a number of new microstructuring techniques essentially broadened the range of materials available.16,17 For use in microreactors, chemically inert and pressure resistant materials, like metals, alloys or ceramics would be preferable. Among others, LIGA technology, ultra precision micromachining and micro electro discharge machining are the most important fabrication methods to form metal microstructures.18,19

9.2 9.2.1

Fabrication by LIGA-technology Introduction

Microstructures in various metals can be obtained using the LIGA method, based on a combination of lithography, electroforming and molding (German acronym: Lithographie, Galvanoformung, Abformung; Figure 9.1). LIGA yields highly precise metal structures in a nearly unrestricted choice of contour.20 The “classical” LIGA process starts with the irradiation of a radiation-sensitive polymer resist layer through a mask.21–23 This can be achieved by different types

Micro-electroforming 247

(a)

(b)

Deep X-ray or UV-lithography

Microstructure

Resist LASER Ablation Substrate Ultra-precision micromachining Mould insert

Figure 9.1 Scheme of LIGA process steps: (a) lithography and (b) electroforming.

of radiation, like synchrotron radiation in the X-ray range or UV-light. Subsequently, the irradiated (or in case of image reversal resists the non-irradiated) parts of the resist can be dissolved leading to a three-dimensional microstructure. A resist material commonly used in deep X-ray lithography (DXRL) is poly(methyl methacrylate) (PMMA) since it can be casted in thick films showing a high resolution at a comparatively low sensitivity. In a second step, a complementary metal structure is generated by means of electroforming. This metal structure can be used as the final product or as a tool for replication processes like injection molding or embossing. When extreme precision and structural depth is not required, a cost-saving alternative to DXRL is photolithography with UV-light.24,25 For rapid prototyping, laser micromachining,26 micro electro discharge machining as well as silicon micromachining,27 for example, reactive ion etching, can be utilized for direct structurization of various materials. A combination of these prototyping techniques with subsequent electroforming results in metal microstructures or mold inserts in analogy to the “classical” LIGA process. 9.2.2

Electroforming of metal microstructures

Electrodeposition offers the possibility of manufacturing miniaturized components using a number of other metals and special alloys.28,29 For fabrication of microreaction devices, electrodeposition of gold, copper, silver and nickel by standard electroplating processes is often used and will be discussed in detail. Defect-free fabrication of metal microstructures requires clean electrolytes adapted to the specific demands of micro-electroforming.28,30 Complete wetting of the resist structures is indispensable and wetting agents are added to the electrolyte. The growth of the metal layers within the microstructure is mainly controlled by diffusion of the metal ions into cavities and may not be influenced by changes in hydrodynamic conditions. Therefore, electrolytes should have a high throwing power.31 Contamination of the electrolyte, for example, by dust from the environment, precipitation of electrolyte components, decomposition of

248

Löwe et al. Wafer Process cell

Rinsing cell

Microstructures De-ionized water Dosing unit Processing unit

Microfilter

Supply unit Regeneration

Filter

Exhaust air

Reservoir

To neutralization Filling Outflow

Figure 9.2 Apparatus for micro-electroforming (flow chart).

Figure 9.3 View of a micro-electroforming apparatus for use in a clean room.

additives or fouling needs to be avoided. Therefore, the online monitoring of the electrolyte is necessary. Moreover, clean-room conditions are indispensable for all fabrication steps. For this purpose, a new electroforming apparatus was developed for mass

Micro-electroforming 249 production of LIGA microstructures. The modular concept of this device is shown schematically in Figure 9.2. This apparatus allows flexible production and automated process control requiring a minimum of floor space in the clean room (Figure 9.3). Closed electrolyte circulation and integrated recycling of rinsing water reduce the risk of contamination and, furthermore, prevent environmental pollution.32 9.2.2.1

Micro-electroforming of gold

Gold electroforming is an important process in LIGA technology. Both, masks for deep X-ray lithography, containing gold absorber structures, and threedimensional microstructures, such as precisely defined pore filters are examples of gold plating. In contrast to the well established techniques in jewelry and semiconductor industry, thick layers in the range of 30–1000 m, depending on the specific application, have to be deposited. Due to the instability of X-ray and photoresists against electrolytes containing cyanides or citric acid only gold sulfite electrolytes can be used.33,34 The electrodeposition of gold from sulfitic electrolytes has been described in detail by Gemmler et al.35 Basic processes are the dissociation of the disulfitoaurate(I)complex (Eqn (1)) and the cathodic reduction (Eqn (2)) of gold(I): [Au(SO3)2]3 i Au 2 SO2 3

(1)

Au e i Au

(2)

Sulfitic gold electrolytes are commercially available (Table 9.1), enabling a current efficiency of 100% and show very good results regarding homogeneity, ductility and minimal internal stress. Nevertheless, these electrolytes are very sensitive to changes in process parameters. Thus, these have to be kept within close margins in order to guarantee the high quality of the gold deposits. Since the growth rate of gold is low and cannot be enhanced by changes in process conditions, electroplating of microstructures usually takes a long time. For instance, the electrodeposition of structures with heights larger than 100 m Table 9.1 Composition of the sulfitic gold electrolyte (IMABRITE, Schlötter) Parameter

Range of values

Gold content Sulfite as potassium sulfite Leveler pH-value Temperature Current density Anode material Growth rate

8–12 g/l 40 g/l 5 mg/l 9.2–9.6 50C 0.1–0.4 A/dm2 Platinated titanium mesh 3.8–15.2 m/h

250

Löwe et al.

Figure 9.4 Gold micro-sieve fabricated utilizing a LIGA process (hole diameters: 8 m).

Figure 9.5 Mask for deep X-ray lithography fabricated by gold micro-electroforming.

requires approximately 24 h. In order to maintain optimum process conditions these long periods of time, the concentration of complexing agents (sulfite), gold, the leveler as well as the pH-value and the temperature need to be monitored permanently. Therefore, in situ detection by electrochemical methods such as cyclic voltammetry is convenient.36 Besides the application of gold as a coating material for microstructures in order to yield chemically inert surfaces, functional gold microstructures with

Micro-electroforming 251 Table 9.2 Composition of the nickel sulfamate electrolyte Parameter

Range of values

Nickel content (as sulfamate) Nickelbromide Boric acid Wetting agent: Perfluorinated alkylsufate (FT 248 Bayer) Saccharin pH-value Temperature Cathodic current density Anodic current density Electrolyte flow Filter Anode material Geometry of plating bath Growth rate

105–110 g/l 5 g/l 40 g/l 10 ml/l in 2% solution 2–20 mg/l 3.8 50C 1.0–7.5 A/dm2 0.1–1 A/dm2 500 l/min 0.2 m S-Ni rounds in a Ti-basket Plate-to-plate arrangement 10 m/h at 1 A/dm2

structural depths up to 1000 m can be fabricated by electroforming. For instance, a self-supporting micro-sieve with a hole diameter of 8 m for applications in chemistry and biotechnology was obtained using the process described above (Figure 9.4). Since gold is an ideal X-ray absorbing material, mask absorber structures with a thickness of up to 100 m can be realized by micro-electroforming as well (Figure 9.5). The electrodeposition process in this case requires an even higher quality since the tiniest defects in the X-ray absorber will be copied to the final structure of the LIGA process. Therefore, electrodeposited absorber structure must be homogeneous and free of defects. 9.2.2.2

Micro-electroforming of nickel

Nickel, being inert against most chemicals, is a metal commonly applied in microtechnology. It can be electrodeposited easily by a standard process using a nickel sulfamate bath (Table 9.2). Fluorinated compounds are used as wetting agents due to their capability of reducing the internal stress of the deposited microstructures.32,37,38 In contrast to conventionally applied wetting agents, the internal stress of the deposited microstructures decreases with the electrolyte load (Figure 9.6). As the cathodic current density for nickel deposition is limited to approximately 1 A/dm2, the growth rate of the metal layer of 15 m/h is rather slow. In principle, the fabrication of monolithic nickel microstructures with a thickness of several millimeters is possible, but in practice the fabrication procedure is very time-consuming. Additional problems may arise during the growth of nickel layers on large non-conducting areas. The lateral growth is not sufficient for connection of single microstructures in order to build a uniform metal layer (Figure 9.7). These difficulties may be avoided by applying combined fabrication processes using different layers and materials. Preferably, the microstructures are

(b)

40 20 0

5

Thickness (µm) 10 15 20 25 30

35

4

3 2 1

–20 –40 –60

Internal stress (N/mm2) Pressure Tensile

(a)

Löwe et al.

Internal stress (N/mm2) Pressure Tensile

252

–80

40 20 0

5

10

Thickness (µm) 15 20 25 30

35 2 3 4 1

–20 –40 –60 –80

Figure 9.6 Influence of electrolyte load and wetting agent on internal stress of electroplated nickel coatings: (a) fluorinated wetting agent; (b) standard wetting agent (current density: 1 A/dm2; curve 1: without wetting agent; 2: with wetting agent; 3: with wetting agent after 12.23 Ah/l; 4: with wetting agent after 24.36 Ah/l).

Reinforcement with copper Metalization Nickel microstructure Substrate

Resist

Substrate

Figure 9.7 Electroforming of nickel structures with copper reinforcement (left: nickel microstructure; right: reinforcement with copper).

electrolytically filled with nickel. A copper coating, only a few micrometers thick, prevents passivation of the nickel surface. Subsequently, the complete surface is covered with copper by physical coating techniques in order to render the insulating resist areas conductive. This layer may easily be reinforced by electrodeposition of copper at high growth rates by using standard acid sulfate baths, thus, resulting in layers several millimeters thick (Section 9.2.2.3). 9.2.2.3

Micro-electroforming of copper

Copper as a highly conductive metal in certain cases is a desirable material for microstructures themselves, for example strip conductors, coils, electrical

Micro-electroforming 253 Table 9.3 Composition of the standard acid sulfate electrolyte Parameter

Range of values

Copper sulfate Sulfuric acid Chloride Bath temperature Current density Anodic material Growth rate

30–60 g/l 200–250 g/l 10–60 mg/l 20–25ºC 1–4 A/dm2 Electrolytic copper 10–40 m/h

Table 9.4 Composition of the adapted acid sulfate electrolyte Parameter

Range of values

Copper content Sulfuric acid Sodium chloride Leveler Cuprostar LP1 (Blasberg Oberflächentechnik GmbH) Bath temperature Current density Anodic material Growth rate

15–25 g/l 200–250 g/l 60–100 mg/l 5 ml/l 20–25ºC 1–4 A/dm2 Phosphorus depolarized copper 12.5–50 m/h

contacts and circuit boards. It is preferred for the production of microstructured electrodes by micro electro discharge machining techniques developed recently. Electroplating of copper microstructures can be performed using different electrolyte types like copper tetrafluoroborate baths, methane sulfonate baths, cyanide baths, alkaline pyrophosphate baths or acid sulfate baths. Organic additives like polyalcohols or mercaptanes have to be avoided in order to prevent damage of commonly used UV- and X-ray resists, respectively. High copper contents in the electrolyte are used for reinforcement of nickel microstructures as well as for applications with low aspect ratios (Table 9.3). A modified acid sulfate electrolyte with low copper content and comparatively high acid concentration is used for electroforming of microstructures with high aspect ratios (Table 9.4). This modified electrolyte has a very high throwing power and allows the fabrication of fine-grained and silken copper deposits.39 9.2.2.4

Micro-electroforming of silver

Silver as a noble metal is, in a number of cases, a material advantageous to microchemical devices since it is inert to organic solvents as well as many inorganic acids. Furthermore, silver plays an important role as a catalyst material for various gas phase reactions with industrial relevance.

254

Löwe et al. Table 9.5 Composition of the cyanidic silver electrolyte Argophan (Blasberg Oberflächentechnik GmbH, Solingen) Parameter

Range of values

Silver content Free potassium cyanide Organic additives pH-value Bath temperature Current density Anodic material Ratio anode/cathode

30 g/l 120 g/l 3% 12–12.5 18–30C 1–2.5 A/dm2 High purity silver 1/1

For the fabrication of silver microreactor components a commercial cyanidic electrolyte is used (Table 9.5). In analogy to gold and copper electrolytes it has a high throwing power.40 The applicable growth rate of 35 m/h is comparatively high. Contrary to other cyanide electrolytes damage of the resist structures is avoided by a high content of organic additives adsorbing on the resist surface.

9.3

Alternative fabrication technologies

Apart from lithographic techniques, a number of alternatives for fabrication of metal microstructures have been reported in the literature. Mechanical techniques such as drilling, milling, turning, dicing and slicing used in precision engineering for machining of hard materials have been modified to be suitable for micro technology.41 This enables the use of materials like stainless steel and special alloys for applications requiring a very high mechanical or chemical resistance which is important for the design and fabrication of microchemical devices. Electro discharge machining (EDM) is a well introduced technology for the processing of conductive materials. It is based on an erosion process due to a high energy discharge between a die-sinking electrode and a workpiece and is also applicable for microfabrication (-EDM).19 Using the LIGA process for the fabrication of the electrodes, the so-called LIGA–EDM method, results in much smaller tolerances than conventional EDM techniques, and has been successfully applied for the manufacture of microreactor components (Figure 9.8).42 EDM-grinding using fast rotating tungsten disc electrodes, for instance, is a convenient method for generating fluid microchannels with aspect ratios of approximately 15.43 On the other hand, utilizing a high speed rotating cylindrical electrode (EDM-milling) is favorable to realize microchannels in TiB2 ceramic, a material suitable for high temperature microreactor applications.

9.4

Potential and applications of microreactors

In the following sections, the operational principle, the construction as well as applications of microreactor components and integrated microreactors fabricated

Micro-electroforming 255 Dielectric liquid Workpiece

Flash over generates plasma channel

Electrode

Spark erosion Using various electrodes of complex shape

Plasma energy removes small amounts of material

Figure 9.8 Working principle of EDM for the fabrication of microstructures (scheme).

by LIGA technology are discussed in detail. Important components of reactors in chemical engineering are mixing devices and heat exchangers. Miniaturization in both cases results in a significant improvement of performance in terms of mass and heat transport, respectively. Furthermore, a microreactor for a heterogeneously catalyzed gas phase reaction is described. This reactor comprises electroformed microstructures for gas mixing as well as a catalytically active reaction zone illustrating advantages of a modular design concept.11 9.4.1

Micromixing devices

A micromixing device utilizing a static operational principle is shown in Figure 9.9. It consists of a microstructured mixing element fabricated by deep X-ray lithography and electroforming. On its upper side, a wound microstructure with corrugated walls divides a hollow into two fluid compartments forming 2 15 interdigital microchannels in between (Figure 9.10). Typically, these microstructures with channel widths in the range of 10–50 m are formed in nickel on a copper support in order to yield a mechanically robust mixing element. Silver is also a suitable material which can be electroplated uniformly with thicknesses of several millimeters. The mixing element tightly fits into a stainless steel housing comprising the top and bottom parts. The fluids to be mixed are fed through inlets in the top part of the steel housing extending into the fluid compartments on the mixing element and subsequently flow in a counter-current configuration into the multichannel system. The fluids are then withdrawn perpendicular to the direction of the feed flows through a discharge slit in the upper housing connected to the outlet. Thereby, inside the slit zone, a system of alternating lamellae of the two fluids is formed. The mixing process itself is based on diffusion from one lamella into the other. Since the width of these lamellae is in the order of the channel width

256

Löwe et al.

(a)

(b) Top part

Outlet

Bottom part Outlet

Inlet 2

Inlet 1

LIGA device

Nickle on copper

Housing

Silver

Figure 9.9 Micromixer fabricated by LIGA-technology. (a) parts of the single mixer unit. (b) LIGA mixing devices and assembled housing. (a)

(b) 500 µm

100 µm

Width of microchannels: 40 µm Material: Silver

Figure 9.10 Mixing element containing 2 15 microchannels.

(depending on the ratio of the flow rates), that is, the diffusional length amounts to only 10–50 m. Therefore, complete mixing is achieved within milliseconds. Mixing time as well as the prevention of inhomogeneities are major parameters influencing the mixing quality. The experimental evaluation is based on the investigation of two parallel reactions with different rate constants according to Villermaux.44 The reaction system used is the ultrafast ionic reaction of protons with acetate to form acetic acid and the comproportionation of iodide and iodate to form iodine in case of proton excess. The amount of iodine generated can be measured easily by UV/VIS spectrometry. If fast and complete mixing is achieved, in theory no iodine formation should occur and a colorless solution should be obtained. Based on this experimental method, the mixing quality of the micromixer described above has been compared to various laboratory mixing devices. Although, the micromixer can be operated only in the laminar flow regime, it gives, by far, the best results, even compared to turbulent systems like the laboratory-scale magnetic stirrer and a mixing tee (Figure 9.11).45

Micro-electroforming 257 Mixing-tee laminar

Mixing quality UV-absorption

Unstirred vessel

Heavily stirred vessle Mixing-tee turbulent Micromixer laminar

Figure 9.11 Comparison of different mixing devices in terms of mixing quality using the method by Villermaux.

For industrial scale applications different micromixing devices have been realized with different volume throughput, which were utilized for small lot production as well as lab-scale production. According to the concept of numbering up, an increase in the number of mixing units (microchannels or mixing elements) working in parallel yields a volume throughput of 1–100 l/h. In Figure 9.12, a micromixer array with ten single mixing elements each comprising 2 15 microchannels is shown. It has been formed also by the LIGA process using nickel on copper electroplating. The device enables a volume throughput up to 10 l/h at a pressure difference of less than 2 bar. Typical flow velocities in the microchannels, feeding a larger mixing zone, amount to 5–50 ml (referring to volume flows of 10–1000 ml/h for aqueous systems), whereas in the mixing zones lower velocities between 0.005 and 5 ml can be reached. The numbering up concept circumvents usual scale-up problems, since the physico-chemical boundary conditions in each of the microchannels are equal regardless of their number. Thus, a major advantage of numbering up instead of scaling up is the ability to transfer results of R&D work on lab scale into production scale very fast. In chemical engineering, fast and efficient mixing is an important prerequisite for process optimization of many reactions. Due to the advantages discussed above, micromixers have found first applications for production purposes on a larger scale. In case of very fast consecutive and parallel reaction sequences, bad mixing can cause a non-desired decrease in yield and selectivity. For instance, radical polymerization reactions performed on an industrial scale can be optimized in terms of the products’ molecular weight distribution by use of micromixers.7 Thus, the more efficient mixing process leads to a higher product quality and prevention of reactor fouling. Another similar example is the utilization of micromixers for fast organometallic reactions leading to increased yields of about 20% compared to the standard production process.8

258

Löwe et al.

(a)

(b) Outlet channel Outlet Upper housing

Inlet 2 Inlet 1

Lower housing

Micromixer array

Figure 9.12 Micromixer array comprising ten mixing elements with 2 15 microchannels working in parallel. (a) Micromixer array and single mixer unit. (b) micromixer array with upper and lower housing. Width of outlet channel: 350 m.

(b)

50 µm

Number density distribution

(a)

0.25 0.20 0.15 0.10 0.05 0.0

0

5

10 15 20 Droplet size (µm)

25

Figure 9.13 Silicon oil in water emulsion generated using the micromixer: (a) microscope image at 500-fold magnification; (b) number density distribution.

Besides homogeneous mixing processes, the generation of gas/liquid and liquid/liquid dispersions, respectively, is of major importance in chemical technology. Recently, first investigations have shown that micromixers can be advantageously utilized to form gas/liquid dispersions, emulsions and creams as well.46–48 Results obtained for a model system forming a creamed silicon oil in water emulsion have shown that small droplets having a remarkably narrow number density distribution (Figure 9.13) can be formed utilizing the micromixer shown in Figure 9.10. Furthermore, the influence of the total flow rate and the ratio of water to oil flow rates as well as the microchannel width of the mixing element have been investigated leading to the following conclusions.

Micro-electroforming 259 The interdigital flow configuration in the micromixer leads to an alternating arrangement of small lamellae of different phases which subsequently are split into small droplets due to shear forces created by periodical velocity gradients. Higher total volume flows therefore increase the shear forces, resulting in smaller droplet sizes. Since the thickness of the lamellae is determined by the channel dimensions, droplet sizes are proportional to the channel width. Furthermore, under conditions of large ratios of the flow rates, the thickness of the lamellae of the phase with lower flow rate becomes smaller than the channel width due to the effect of hydrodynamic focussing.49 As a result, emulsion systems with droplet sizes ranging from 4 to 50 m having narrow size distributions with typical standard deviations of 4–8 m can be formed in a continuous process. The mean droplet size can be adjusted easily by the choice of the operating parameters and the channel width of the mixing element. In addition to the formation of emulsions, micromixers can be advantageously used for the generation of gas/liquid dispersions.47 These dispersions with small bubble sizes and, therefore, large contact areas between gas and liquid phase are particularly interesting for gas phase reactions with high demands with regard to mass and heat transfer. In analogy to liquid two-phase systems, bubble sizes depend on the flow rate, the channel width but also on the fluid properties of the solution, like viscosity and surface tension. Table 9.6 illustrates the dependence of the bubble sizes formed on liquid flow rate and channel width for an argon/ water system containing glycerol and surfactants. The bubble size distribution was found to be narrow and symmetrical (Figure 9.14). Furthermore, the same gas/liquid dispersion system with low to medium content of glycerol, formed at a low to medium ratio of gas and liquid flow rates shows a continuous flow pattern of high geometric order, the so-called hexagon flow (Figure 9.14). This highly ordered flow pattern allows a well defined residence time, being a crucial process parameter for fast and exothermic chemical reactions. Moreover, microreactors can be used for the generation of creams and semi-solid formulations.48 These oil in water as well as water in oil dispersions are used as cream bases for various pharmaceutical and cosmetic applications, respectively. Creams were formed by contacting liquid oil and water phases at higher temperatures (60–80C) using micromixers. After passing the mixer, the emulsion formed cools down to yield a thixotropic cream at room temperature.

Table 9.6 Bubble sizes as a function of liquid flow rate and channel width of the micromixer Bubble size (m)

Flow rate (ml/h)

Channel width

120–350 170–470 400–650 400–800

2–23.4 2–34 2–16 3–78

25 40 25 40

260

Löwe et al.

1 mm

220 ± 50 µm

Figure 9.14 Bubble sizes and flow pattern of gas/liquid dispersions using a micromixer.

Experimental results show qualitatively the same dependence of the droplet size on operating conditions as found for the model system described above. Mean droplet sizes of 0.8–2.5 m with standard deviations ranging from 0.4 to 1.3 m were measured (Figure 9.15). In addition to these investigations, stability studies were carried out at various emulsifier concentrations. The comparison of two sets of samples showing different mean droplet diameters revealed an increase in stability with decreasing droplet size. This is in accordance with the theory for non-electrostatically stabilized systems.50 Thus, temporary stable creams (weeks to months) with low or zero content of emulsifiers and preservatives, as relevant for clinical applications, for example, for patients suffering from allergic reactions or neurodermitis, can be generated in a continuous flow set-up on-site and on-demand. Using a single micromixer as shown in Figure 9.12 a total mass throughput of 250 g/h up to 1.5 kg/h can be achieved which is comparable to typical batch sizes of small scale production. 9.4.2

Micro heat exchanger

Fast and efficient heat transfer is a major prerequisite for many applications requiring defined thermal control. In chemical engineering, efficient thermal

Micro-electroforming 261 (a)

0.30 0.25 0.20 0.15 0.10 0.05 0.00 0

Cumulative frequency

40 µm

Frequency distribution

(b)

1

2 3 4 Diameter (µm)

5

6

1.0 0.8 0.6 90% value

0.4 0.2 0.0 0

1

2 3 4 Diameter (µm)

5

6

Figure 9.15 Cream base for pharmaceutical applications: (a) microscope images recorded using the differential interference contrast method and crossed polarizers, respectively; (b) number density distribution and cumulative frequency.

control of highly exothermic reactions in most cases is a crucial parameter influencing yield and selectivity. Utilizing miniaturized heat exchangers, therefore, offers certain advantages. Since high surface to volume ratio combined with high temperature gradients are achievable, very high heat transfer efficiencies can be realized. In Figure 9.16, a micro heat exchanger is shown comprising a stack of microstructured platelets. These platelets containing microchannels were formed by the LIGA method using UV- or deep X-ray lithography depending on the accuracy requirements. Subsequent electroplating of nickel, followed by copper reinforcement, yields a heat exchanger platelet with typical channel widths and depths of 100–300 m. Two sets of microstructured platelets of mirror image symmetry are assembled alternately to form a counter-current micro heat exchanger (see Figure 9.17). For such a micro heat exchanger with four elements, each containing 450 m thick fluid compartments, a heat transfer coefficient of 2.4 kW/m2K was measured when operated in a counter-current mode (water flow rate: 0.9 l/h; temperature difference: 30 K). Degreasing the fluid layer thickness to 100 m leads to an increased heat transfer coefficient of 11.6 kW/m2K due to the higher temperature gradient between adjacent layers of the heat exchanger. These values are in good agreement with theoretical calculations based on laminar flow conditions. The results show that the heat transfer efficiency of a micro heat exchanger with small

262

Löwe et al.

Figure 9.16 Micro heat exchanger.

Layer A Inlet fluid B Layer B Inlet fluid A

Opening for fixing pin

Outlet fluid A

Outlet fluid B

Figure 9.17 Scheme of working principle of a counter-current micro heat exchanger.

lateral dimensions is much higher than for conventional plate heat exchangers with unshaped (0.4–1.2 kW/m2K) or shaped (1.0–4.0 kW/m2K) metal sheets.3 A decrease in fluid layer thickness on the other hand results in a higher pressure loss across the micro heat exchanger. Therefore, the design and the characteristic dimensions have to be optimized in terms of heat transfer efficiency and total volume throughput for each specific application. 9.4.3

Integrated microreaction system

A microreaction system for the heterogeneous gas phase synthesis of ethylene oxide from ethylene and oxygen was realized in order to investigate new process regimes. The ethylene oxide synthesis in the gas phase is an industrially relevant process with an annual turnover of 11.2 million tons, and therefore, the object

Micro-electroforming 263 of many scientific investigations. The microreactor designed, as shown in Figure 9.18, comprises a gas mixing unit, a diffusion zone and a catalytic reaction zone.18 Both, the mixing unit and the catalyst were realized as stacks of fourteen microstructured foils fabricated by LIGA technology. The principle of mixing is based on the multilamination of the reactant gas streams, which is similar to that introduced in Section 9.4.1. Since the reactor operates in the laminar regime, mixing is accomplished by diffusion. The length of the diffusion zone was chosen in order to achieve at least 99% homogeneous gas mixing before entering the catalyst zone for all operating conditions. Microstructures on each foil of the mixing unit have the shape of nine quadrants with varying radii (see Figure 9.19(a)). In order to achieve equipartition of the volume flows in each channel having a different length, the channel width need to be increased with increasing radius from 150 to 470 m. Each foil was fabricated by a Laser-LIGA process using a combination of Eximer laser ablation and milling to form a polymer master for electroplating. Subsequent nickel electroforming yields the metal structure which was additionally gilded electroless with a 5 m gold layer to avoid catalytic interference of nickel. The fourteen identical foils were stacked in an alternate manner with every second foil being turned 90 degrees. Thus, two inlet zones on opposite ends of the stack were formed, while the outlet channels extended towards the diffusion zone. The microstructured catalyst foils were fabricated by Laser-LIGA and silver electroplating. Each element comprises nine microchannels 500 m wide and

Inner cover

Cover Seal C2H2

O2

Gold plated mixer structures

Catalyst structures (silver) Outlet

Figure 9.18 View of the microreactor components for ethylene oxide synthesis.

264

Löwe et al.

(a)

(b)

Nickel

Silver

Figure 9.19 Microstructured parts of the ethylene oxide microreactor: (a) Mixer structures; (b) Catalyst structures.

50 m deep (see Figure 9.19(b)). Simulations of the amount of heat released due to the heterogeneous reaction process and the efficient heat conduction within the silver catalyst resulted in a nearly uniform temperature distribution with a minimal hot spot of 0.2 K. Therefore, the integration of heat exchanger microstructures in the catalyst compartment was not necessary. Both, mixer and catalyst foils were inserted into a corrosion resistant stainless steel housing fabricated by precision engineering and -EDM. Attention was drawn to a tight fit of all microstructured foils in the housing recess. Furthermore, the housing consists of an inner cover plate and a top plate. Both are screwed onto the reactor and sealed with graphite washers. The microreactor was designed for operation at pressures of 1–25 bar, temperatures ranging from 200 to 300ºC and a total volume flow of up to 5 l/h. Due to the flame arrester effect of the microchannels, experiments within the explosive regime of the reaction was possible without any problems. Thus, reactant concentrations of 0.75–15% ethylene and 5–85% oxygen could be realized. The results of a comprehensive experimental characterization show that using the microreactor with a polycrystalline silver catalyst ethylene oxide yields up to 5% at a selectivity of 50% were achieved.9 The yield was found to be a function of temperature, and therefore, the highest yield and selectivity were measured at 300C and 33% oxygen. On the basis of the space–time yield, the microreactor performance can be compared to conventional reactors used in the industry, which typically range between 0.13 and 0.26 t/hm3. For the microreactor space– time yields of 0.25–0.67 t/hm3 related to the internal reactor volume were measured operating with a gas mixture of 15% ethylene and 85% oxygen at 5 bar and a temperature range from 240 to 300ºC.9 Thus, the performance is comparable to state-of-the-art reactors. Especially, when operated in process regimes (e.g. high oxygen partial pressure) not accessible with conventional technology, the performance of the microreactor is superior.

Micro-electroforming 265

9.5

Conclusions

Metal microstructures are crucial and important components of microreactors for chemical processes. Devices for unit operations like mixing and heat exchange with structural dimensions of less than one millimeter offer advantages when performing transport-limited or highly exothermal reactions. Therefore, chemically inert as well as temperature and pressure resistant materials are necessary. Today, a range of fabrication techniques is available for microstructuring. Among others, LIGA technology as well as micro electro discharge machining are the most prominent techniques to form microstructures in a variety of metals, alloys and special materials. Lithography and subsequent electroforming of structures with micrometer accuracy in gold, nickel, copper and silver as well as various alloys are standard processes. The structural quality of electrodeposits depends highly on the operating parameters such as electrolyte composition, pH value, etc. These parameters have to be kept within close margins. Future activities concern the extension of microstructuring techniques to broaden the variety of materials available. For instance, microelectro discharge machining of conducting ceramic materials, like titanium diboride, leads to microstructures suitable for high temperature applications. The utilization of microstructures for chemical applications has begun to play an important role in chemical processing leading to the development of novel reaction regimes. Thus, a high impact of this new field of microreaction technology on chemical engineering is proposed in the near future. The commercialization of microreactors and their consequent application in R&D work as well as in small scale chemical production saves development time, costs, energy and material resources. Therefore, the utilization of microreactors will be a future trend for sustainable development in the field of chemical engineering, pharmacy, cosmetic industry and life sciences, respectively.

References 1. Ehrfeld, W. “Die Chemieanlage aus der Retorte / Chemical plants from test tubes”, ACHEMA Magazin 94 (1994) 8–11. 2. Ehrfeld, W., Golbig, K., Hessel, V., Löwe, H. and Richter, T. “Microfluidic reaction systems: LIGA-a favourable process for microreactors”, mst news 17(5) (1996) 5. 3. Ehrfeld, W., Löwe, H., Hessel, V. and Richter, T. “Anwendungspotentiale für chemische und biologische Mikroreaktoren”, Chemie IngenieurTechnik 69(7) (1997) 931–934. 4. Löwe, H. and Ehrfeld, W. “State of the art in microreaction technology: concepts, manufacturing and applications”, Electrochim. Acta 44 (1999) 3679–3689. 5. Beck, F. and Guthke, H. “Entwicklung neuer Zellen für elektro-organische Synthesen”, Chemie Ing. Techn. 41(17) (1969) 943–990. 6. Hessel, V., Ehrfeld, W., Golbig, K., Haverkamp, V., Löwe, H., Storz, M., Wille, C., Guber, A., Jähnisch, K. and Baerns, M. “Gas/liquid microreactors for direct fluorination of aromatic compounds using elemental fluorine”, in Ehrfeld, W. (ed.) Microreaction Technology: 3rd International Conference on Microreaction Technology, Proceedings of IMRET 3, pp. 526–540, Springer-Verlag, Berlin (2000).

266

Löwe et al.

7. Bayer, T., Pysall, D. and Wachsen, O. “Micro mixing effects in continuous radical polymerization”, in Ehrfeld, W. (ed.) Microreaction Technology: 3rd International Conference on Microreaction Technology, Proceedings of IMRET 3, pp. 165–170, Springer-Verlag, Berlin (2000). 8. Krummradt, H., Kopp, U. and Stoldt, J. “Experiences with the use of microreactors in organic synthesis”, in Ehrfeld, W. (ed.) Microreaction Technology: 3rd International Conference on Microreaction Technology, Proceedings of IMRET 3, pp. 181–186, Springer-Verlag, Berlin (2000). 9. Kestenbaum, H., Lange de Olivera, A., Schmidt, W., Schüth, H., Ehrfeld, W., Gebauer, K., Löwe, H. and Richter, T. “Synthesis of ethylene oxide in a catalytic microreactor system”, in Proceedings of the 12th Int. Conf. on Catalysis 2000, Granada, Spain (2000). 10. Kestenbaum, H., Lange de Oliveira, A., Schmidt, W., Schüth, F., Gebauer, K., Löwe, H. and Richter, T. “Synthesis of ethylene oxide in a catalytic microreacton system”, in Ehrfeld, W. (ed.) Microreaction Technology: 3rd International Conference on Microreaction Technology, Proceedings of IMRET 3, pp. 207–212, Springer-Verlag, Berlin (2000). 11. Richter, T., Ehrfeld, W., Gebauer, K., Golbig, K., Hessel, V., Löwe, H. and Wolf, A. “Metallic microreactors: components and integrated systems”, in Ehrfeld, W., Rinard, I. H., Wegeng, R. S. (eds) Process Miniaturization: 2nd International Conference on Microreaction Technology, IMRET 2; Topical Conference Preprints, pp. 146–151, AIChE, New Orleans, USA (1998). 12. Schubert, K., Bier, W., Brandner, J., Fichtner, M., Franz, C. and Linder, G. “Realization and testing of microstructure reactors, micro heat exchangers and micromixers for industrial applications in chemical engineering”, in Ehrfeld, W., Rinard, I. H., Wegeng, R. S. (eds) Process Miniaturization: 2nd International Conference on Microreaction Technology, IMRET 2; Topical Conference Preprints, pp. 88–95, AIChE, New Orleans, USA (1998). 13. Hagendorf, U., Janicke, M., Schüth, F., Schubert, K. and Fichtner, M. “A Pt/Al2O3 coated microstructured reactor/heat exchanger for the controlled H2/O2-reaction in the explosion regime”, in Ehrfeld, W., Rinard, I. H., Wegeng, R. S. (eds) Process Miniaturization: 2nd International Conference on Microreaction Technology; Topical Conference Preprints, pp. 81–87, AIChE, New Orleans, USA (1998). 14. Jandeleit, B., Schaefer, D. J., Powers, T. S., Turner, H. W. and Weinberg, W. H. “Kombinatorische Materialforschung und Katalyse”, Angew. Chem. 111(17) (1999) 2649. 15. Asmead, J. W., Blaisdell, C. T., Johnson, M. H., Nyquist, J. K., Perotto, J. A. and Ryley, J. F. “Integrated chemical processing apparatus and processes for the preparation thereof ”, EP 0 688 242 B1, (19.03.1993); E.I. Du Pont de Nemours and Company; Wilmington (USA). 16. Ehrfeld, W. and Lehr, H. “Deep X-ray lithography for the production of three dimensional microstructures from metals, polymers and ceramics”, Radiat. Phys. Chem. 45(3) (1995) 349–365. 17. Bauer, H.-D., Weber, L. and Ehrfeld, W. “Formeinsätze für die Massenfertigung von hochpräzisen Kunststoffteilen: Die LIGA-Technik”, Werkzeug und Formenbau 4 (1994) 22–26. 18. Löwe, H., Ehrfeld, W., Gebauer, K., Golbig, K., Hausner, O., Haverkamp, V., Hessel, V. and Richter, T. “Microreactor concepts for heterogeneous gas phase reactions”, in Ehrfeld, W., Rinard, I. H., Wegeng, R. S. (eds) Process Miniaturization: 2nd

Micro-electroforming 267

19.

20.

21.

22.

23. 24. 25. 26.

27.

28.

29. 30. 31.

32.

33. 34.

International Conference on Microreaction Technology, IMRET 2; Topical Conference Preprints Process Miniaturization: 2nd International Conference on Microreaction Technology; Topical Conference Preprints, pp. 63–74, AIChE, New Orleans, USA (1998). Ehrfeld, W., Lehr, H., Michel, F. and Wolf, A. “Micro electro discharge machining as a technology in micromachining”, in Proceedings of the “SPIE Symposium on Micromachining and Microfabrication”, 14–15 Oct., 1996; pp. 332–337; Austin TX, USA. Ehrfeld, W. and Becker, E.-W. “Das LIGA-Verfahren zur Herstellung von Mikrostrukturkörpern mit großen Aspektverhältnis und großer Strukturhöhe”, KfK – Nachrichten 19(4) (1987) 167. Ehrfeld, W., Bley, P., Götz, F., Hagmann, P., Maner, A., Mohr, J., Moser, H. O., Münchmeyer, D., Schelb, W., Schmidt, D. and Becker, E. W. “Fabrication of microstructures using the LIGA process”, in Proceedings of the “IEEE Micro Robots and Teleoperator Workshop”, 9–11 Nov., 1987; p. 1; Hyannis, Cape Cod MA, USA. Ehrfeld, W. and Münchmeyer, D. “LIGA method – Three-dimensional microfabrication using synchrotron radiation”, Nuclear Instruments and Methods in Physics Research A 303 (1991) 523–531. Ehrfeld, W. and Lehr, H. “Synchrotron radiation and LIGA-technique”, Synchrotron Radiat. News 7(5) (1994) 9–13. Bischofberger, R., Zimmermann, H. and Staufert, G. “Low-cost HARMST-Process”, Sensors and Actuators A 61 (1997) 392–399. Puippe, J.-C. “UV-LIGA-Mikrogalvanoformung”, Metalloberfläche 52(11) (1998) 871. Arnold, J., Dasbach, U., Ehrfeld, W., Hesch, K. and Löwe, H. “Combination of excimer laser micromachining and replication processes suited for large scale production (Laser-LIGA)”, Appl. Surf. Sci. 86 (1995) 251. Sander, D., Hoffmann, R. and Relling, V. M., J. “Fabrication of metallic microstructures by electroplating using deep-etched silicon moulds”, J. Microelectromechanical Systems 4(2) (1995) 81. Löwe, H., Ehrfeld, W. and Diebel, J. “Ultraprecision microelectroforming of metals and metal alloys”, in Proceedings of the “Micromachining and Microfabrication Process Technology 3”, 29–30 Sept., 1997; pp. 168–175; Austin TX, USA. Bratoeva, M. and Atanassov, N. “Nickel-tungsten alloy electrodeposition from sulfamate electrolyte”, Metal Finishing 96(6) (1998) 92. Stark, W., Saumer, M. and Matthis, B. “Nickelsulfamat-Elektrolyte für die Mikrogalvanoformung”, Galvanotechnik 86(12) (1995) 3931. Leyendecker, K., Bacher, W. and Bade, K. “Untersuchungen zum Stofftransport bei der Galvanoformung von LIGA-Mikrostrukturen”, Galvanotechnik 89(2) (1998) 382–391. Löwe, H., Mensinger, H. and Ehrfeld, W. “Galvanoformung in der LIGA-Technik”, in Zielonka, A. (ed.) Jahrbuch Oberflächentechnik, Vol. 50, pp. 77–95, Metall Verlag, Heidelberg (1994). Chiu, S.-L. and Acosta, R. E. “Electrodeposition of low stress gold for X-ray masks”, J. Vac. Sci. Technol. B 8(6) (1990) 1589. Bacher, W., Bley, P., Hein, H., Klein, U., Mohr, J., Schomburg, W. K., Schwarz, R. and Stark, W. “Herstellung von Röntgenmasken für das LIGA – Verfahren”, KfK – Nachrichten 23(2–3) (1991) 76.

268

Löwe et al.

35. Gemmler, A., Keller, W. and Ruess, K. “Mikrostrukturen”, Metalloberfläche 47(9) (1993) 461. 36. Küpper, M., Baltrunas, G. and Löwe, H. “Elektrolytische Abscheidung von Gold in der Mikrotechnik”, Galvanotechnik 88(9) (1997) 2906. 37. Ehrfeld, W., Hessel, V., Löwe, H., Schulz, C. and Weber, L. “Materials of LIGA technology”, Microsystem Technologies (1997) 112. 38. Abel, S., Ehrfeld, W., Lehr, H., Möbius, H. and Schmitz, F. “Charakterisierung von Materialien zur Fertigung elektromagnetischer Mikroaktuatoren in LIGA-Technik”, in Proceedings of the “International Conference on Micro Materials”, 1996; p. 413; Dresden, Germany. 39. Aroyo, M. and Zonev, N. “Über das Mikrostreuvermögen bei der elektrochemischen Abscheidung von glänzenden Kupferschichten aus sauren Sulfatlösungen mit Pulsstrom”, Galvanotechnik 83(3) (1992) 855. 40. Leisner, P., Tang, P. and Bech-Nielsen, G. “The optimization of throwing power in pulse reversal plating from silver cyanide solution”, in Proceedings of the “81st AESF Annual Technical Conference, SUR/FIN ‘94”, 1994; pp. 79–84; Indianapolis, USA. 41. Linder, G., Bier, W., Schaller, T., Schubert, K. and Seidel, D. “Mikrowärmeübertrager und Mikroreaktoren”, in Proceedings of the “ACHEMA 94, Internationales Treffen für chemische Technik und Biotechnologie; Tagungsband des Symposiums ‘Mikrotechnik’”, 6.–10. Mai, 1994; Frankfurt/M., Germany. 42. Wolf, A., Ehrfeld, W., Lehr, H., Michel, F., Richter, T., Gruber, H. and Wörz, O. “Mikroreaktorfertigung mittels Funkenerosion”, F & M, Feinwerktechnik, Mikrotechnik, Meßtechnik 6 (1997) 436–439. 43. Wolf, A., Ehrfeld, W., Michel, F., Koch, O., Preuß, S., Soultan, H. and Gruber, H. P. “Application of new actuator and vision control systems for micro electro discharge machining”, in Proceedings of the “Intelligent Systems and Advanced Manufacturing”, 1–6 Nov., 1998; Boston, USA. 44. Villermaux, J., Falk, L., Fournier, M.-C. and Detrez, C. “Use of parallel competing reactions to characterize micromixing efficiency”, AIChE Sym. Ser. 88(286) (1991) 6. 45. Ehrfeld, W., Golbig, K., Hessel, V., Löwe, H. and Richter, T. “Characterization of mixing in micromixers by a test reaction: single mixing units and mixer arrays”, Ind. Eng. Chem. Res. 38(3) (1999) 1075–1082. 46. Haverkamp, V., Ehrfeld, W., Gebauer, K., Hessel, V., Löwe, H., Richter, T. and Wille, C. “The potential of micromixers for contacting of disperse liquid phases”, Fresenius J. Anal. Chem. 364 (1999) 617–624. 47. Hessel, V., Ehrfeld, W., Golbig, K., Haverkamp, V., Löwe, H. and Richter, T. “Gas/ liquid dispersion processes in micromixers: the hexagon flow”, in Ehrfeld, W., Rinard, I. H., Wegeng, R. S. (eds) Process Miniaturization: 2nd International Conference on Microreaction Technology, IMRET 2; Topical Conference Preprints, pp. 259–266, AIChE, New Orleans, USA (1998). 48. Hessel, V., Ehrfeld, W., Haverkamp, V., Löwe, H. and Schiewe, J. “Generation of dispersions using multilamination of fluid layers in micromixers”, in Müller, R. H., Böhm, B. (eds) Dispersion Techniques for Laboratory and Industrial Production, Wissenschaftliche Verlagsgesellschaft, Stuttgart, vol. 42, pp. 45–59 (2001). 49. Knight, J. B., Vishwanath, A., Brody, J. P. and Austin, R. H. “Hydrodynamic focussing on a silicon chip: mixing nanoliters in microseconds”, Phys. Rev. Lett. 80(17) (1998) 3863. 50. Lagaly, G., Schulz, O. and Ziemehl, R. Dispersionen und Emulsionen, Steinkopff Verlag, Darmstadt (1997).

Part III

Integration of systems

10 Capacitors and micropower systems Akihiko Yoshida

10.1

Introduction

An electric double-layer capacitor (EDLC) is an energy storage device, in which an electric charge is stored in the electric double layer formed at the interface between carbon materials and electrolytic solutions when dc voltage is applied. The capacitor has a pair of polarizable electrodes with collector electrodes, a separator, and an electrolytic solution. The capacitor is charged and the electrical energy stored in the capacitor is discharged at loads. Since we started to produce EDLCs in 1978, they have been widely used as memory back-up devices in many electrical appliances, for example, VCRs, cameras, etc. Many studies on the capacitors have been carried out in correlation with physical and chemical properties of carbon electrode materials. Recently, the development of capacitors with ultra-high capacitances for high current load over 1 A was actively carried out around the world. They are for the short period back-up at very high current loads, electric power storage uses, EV systems, etc. In this chapter, the principle of the capacitor, its history and present status, structure and characteristics, application, and future prospects are outlined. Both the terms “electric double-layer capacitor” and “electrochemical capacitor” are used to denote the capacitor in this chapter.

10.2

The principle of electric double-layer capacitors

At the interface between different phases, for example, solid and liquid, an electric double layer of positive and negative charges is formed when dc voltage is applied. Several types of models are suggested for the electric double layer structure. The parallel-plate condenser model of Helmholtz,1 the diffuse-charge model of Gouy–Chapmann,2 and the Stern model in which some ions are stuck to the electrode while others are scattered in thermal disarray,3 are the representative models. Electric charge is stored at the electric double layer of the Helmholtz model at a voltage application with ␺1 as shown in Figure 10.1. Equation (1) shows an electric capacitance C in farad, where ␧ is a dielectric constant of an electrolytic solution, C

冕 ␧ · (4 ␲)

1

dS

(1)

272

A. Yoshida

Load Polarizable electrode Electrolyte

Electrolyte Charge

Discharge

Electric double layer

Electric double layer 0 + 1

0 0 – 1

Figure 10.1 Principle of an electric double-layer capacitor. (a) electric potential with no voltage application; (b) electric potential with a voltage application.

where ␦ is the thickness of the electric double layer, and S the surface area of the polaraizable electrode. As the electric double layer capacitance obtained at an aqueous solution and mercury system ranges between 20 and 40 F cm2, polarizable electrodes of activated carbon material with a high specific surface area, a high electric conductivity, and an electrochemical stability will give capacitors with ultra high capacitances, an “electrochemical capacitor”.4 Several features of the electrochemical capacitors are described as follows compared with rechargeable batteries. Abrupt charges and discharges at high current loads can be carried out in the capacitor. Stable charge–discharge performances are obtained in a wide temperature range for the capacitor. A short circuit of the terminal of capacitors does not destroy the cell fatally. A replacement of the capacitor is not required for long period uses. No hazardous material, for example, lead or cadmium is used in the capacitor. These features of the electrochemical capacitors are attributed to a charge–discharge mechanism with a physical adsorption–desorption of electric charges to polarizable electrodes of carbon materials in the capacitor. The capacitors are classified into two types derived from electrolytic solutions used, that is, an “organic-type” and an “aqueous-type”. The organic-type capacitor with a propylene carbonate based electrolytic solution shows higher energy density compared with the aqueous-type capacitor with an aqueous sulfuric acid solution, because the organic-type capacitor cell can have higher rated voltage.

Capacitors and micropower systems 273 Equation (2) shows an electric energy E stored in a capacitor cell, where V is a rated cell voltage, C a capacitance: E 2 CV 2 1

10.3

(2)

History and present status of the capacitors

In 1954, Becker of GE made an application of patent on an electrolytic capacitor with porous carbon electrodes and an aqueous sulfuric acid solution.5 However, in the early 1950s, capacitors with low rated voltages did not play any effective role in electric circuits with vacuum valves. In the late 1970s, with increasing application of ICs and LSIs, micro power sources were required to back up for microprocessors with lower voltage and current. Thus, “electrochemical capacitors” started to be used as micro power sources in the memory back-up application from 1978. In the 1980s, the capacitors were used for the energy source to drive wristwatches with solar cells. In the early 1990s, capacitors were used as actuator back-up sources for toys, electric appliances, home equipment, etc. Recently, capacitors with higher capacitances and lower resistances have been under development for higher electric power sources in electric vehicle systems, electric power storage system, etc. Figure 10.2 summarizes the structures and characteristics of electrochemical capacitors available in the market and under development.

10.4

Types of capacitors

In this section, three types of representative capacitors, coin-type, tubular-type, and power-type capacitors are discussed. 10.4.1

Coin-type capacitors

Figure 10.3 shows the manufacturing process of the coin-type capacitor. Phenolic resin-based fiber cloth is carbonized and activated. The obtained activated carbon fiber cloth (ACF cloth) is coated with an aluminum layer by a plasma spraying method. The ACF cloth with the aluminum layer is punched into disks and assembled on to the capacitor with metal cases, electrolytic solution, and gasket ring. The mixture of tetraethylammonium–tetrafluoroborate and propylene carbonate is used for the electrolytic solution. Figure 10.4 shows the SEM appearance of the electrode.6,7 10.4.2

Tubular-type capacitors

Figure 10.5 shows the manufacturing process of the tubular-type capacitor. The capacitor consists of aluminum foils as collector electrodes, activated carbon layers with binding material on the aluminum foils as polarizable electrodes,

Commercially available

Stack square type 2.3 V/100 F Jelly-roll Tubular type ~2.5 V/6000 F Stack square type 12 V/200 F

Organic

Organic

Aqueous

Organic

AC/binder film

AC/carbon composite

AC/binder composite

Rubber

Stack square type ~2.5 V/3000 F

Al plate

Separator

AC electrodes

Al lead

Terminal electrode

Lead

Case

Lead

Anode Separator Cathode

Safety valve

AC polarizable electrodes Separator Collector electrode

Case

Sealing plate

Gasket Bolt Terminal

Cu terminal

AC with H2SO4

Separator

Packing Al case Al foil AC polarizable electrodes

Separator Rubber gasket

AC block

Stack tubular type ~5.5 V/10 F

Aqueous

Conductive rubber

ACF polarizable electrodes

Gasket ring

AC powder pellet

Jelly-roll tubular type ~2.5 V/100 F

Coin type ~5.5 V/1 F

Metal case Separator Al electrodes

Construction of cells

Organic

Organic

Electrolyte

AC/binder film

AC/binder pellet

fiber cloth

Activated carbon (AC)

Polarizable electrodes

Figure 10.2 Summary of electric double-layer capacitors.

Under development

Matsushita Asahi glass

NEC

Maxwell

Matsushita Elna Nippon-chemicon

Fujidenkikagaku

Tokin

Matsushita Elna

Matsushita Elna Hitachi Maxell

Development

Capacitors and micropower systems 275 Phenolic resin-based cloth Carbonization activation Polarizable electrode (Activated carbon fiber cloth) Aluminum plasma spraying Separator Electrolytic solution Metal case Sealing

Figure 10.3 Manufacturing process of the coin-type capacitor.

Figure 10.4 SEM appearance of the ACF electrode.

and separators. A pair of the aluminum foils with activated carbon layers is wound with separators and the whole construction is assembled with an organic electrolytic solution in an aluminum tubular can with a rubber-sealing cap. Figure 10.6 shows discharge characteristics of the 2.5 V–30 F capacitor at constant currents of 100 mA, 500 mA, and 1 A. Figure 10.7 shows temperature characteristics of the capacitor. Both the capacitance change and resistance change is very small in a wide temperature range.8,9

276

A. Yoshida

Binding material

Activated carbon

Solvent

Mixing Aluminum foil Coating Drying Activated carbon electrodes Separator Winding

Electrolytic solution

Immersing Casing New capacitor

Figure 10.5 Manufacturing process of the tubular-type capacitor.

Cell voltage (V)

3

New capacitor Conventional capacitor

2 100 mA

1

100 mA

500 mA 1A

0 0

200

400 600 Time (s)

800

1000

Figure 10.6 Discharge characteristics of the tubular-type capacitors.

10.4.3

Power-type capacitors

Power-type capacitor has almost the same structure as the tubular-type capacitor, except for the terminal and case construction to be used at a higher current flow over several amperes. Figure 10.8 shows charge–discharge characteristics of the 2.3 V–470 F capacitor. The cell was quickly charged at a high current over several amperes and discharged with a linear cell voltage decrease. The inner resistance of the cell calculated from the initial voltage drop in the discharge curve is less than 5 m. With the cell size of 50 mm in diameter and 120 mm in height,

Capacitors and micropower systems 277 600

10

400

0

300 200

–10

DC resistance (mΩ)

Capacitor change (%)

500

100 –20 –40

–20

0 20 40 Temperature (°C)

60

0 80

Figure 10.7 Temperature characteristics of the tubular-type capacitor.

2.5

30 A

100 A 70 A 50 A

Cell voltage (V)

2.0

10 A

1.5 1.0 0.5 0 0

10

20

30

40

60 50 Time (s)

70

80

90

100

Figure 10.8 Charge–discharge characteristics of the power-type capacitor.

the energy density and power density of the capacitor are 1.5 Wh/L and 500 W/L, respectively. This type of capacitor can be applied to high power back-up in electric plants. The development of power-type capacitors with energy densities above 5 Wh/L is strongly expected.

278

A. Yoshida

Figure 10.9 Appearance of the capacitors.

Construction of three types of the cell is shown in Figure 10.2 and the appearances of the cells are shown in Figure 10.9.10

10.5

Effective factors to capacitor characteristics

Figure 10.10(a) and (b) show a structure model and an equivalent circuit model for the electrochemical capacitor, respectively. The capacitor is supposed to be assembled micro-capacitors connected in series and parallel. Physical, chemical, and electrical properties of activated carbon materials, electrolytic solutions, collector electrode materials, separator materials, etc. are very important to determine the capacitor characteristics. In this section, relationships of material properties and capacitor characteristics are discussed. 10.5.1

Activated carbon electrodes

Figure 10.11 shows a relationship between a specific capacitance per weight of activated carbons and a specific surface area of activated carbon fiber materials. The specific capacitance increases with increasing specific surface areas.11,12 Figure 10.12(a) shows pore size distributions of ACF (A) and (B). The ACF (A) has a pore size distribution with larger diameters than ACF (B). The result of temperature dependence of capacitance of the capacitor with ACF (A) and ACF (B) is summarized in Figure 10.12(b). The ratio of capacitance at 25ºC and at 25ºC becomes lower as the accumulated pore volume of pores with diameters larger than 2 nm decreases. The ACF (A) with a pore size distribution of larger pores

Capacitors and micropower systems 279 Collector electrode

(a)

Separator

Electrolyte

A1

Activated carbons

Collector electrode

A2 An

Rsn

Resistance attributed to collect charges in pores

⊕ ⊕ ⊕ ⊕ ⊕

Cn ⊕ RLn ⊕ ⊕

Capacitor element

Resistance attributed to ion migration

RL (b)

R1

C1 R2

C1

Rn

Cn

Figure 10.10 Equivalent circuit of the electric double-layer capacitor.

Capacitance (F · g–1)

40

: Phenolic resin based : Rayon based : PAN based

30

20

10

0

0

500 1000 1500 Specific surface area (m2 · g–1)

2000

Figure 10.11 Relationship between capacitor characteristics and a specific surface area of activated carbons.

280

A. Yoshida (b) (A)

(C at –25°C)/(C at –25°C)

1.0

(B) 0.5

Accumulated pore volume (cc · g–1)

(a)

(A)

1.0

(B)

0.5

0 0 1.0 0 0.5 Pore volume (d > 2 nm))/(Total pore volume)

1

5 10 Pore diameter (nm)

20

Figure 10.12 Relationship between capacitor characteristics and a pore size distribution of activated carbons. (a) The ratio of capacitance of 25ºC to capacitance at 25ºC vs the ratio of pore volume of the pores with diameters larger than 2 nm to the total pore volume; (b) pore size distribution of two kinds of ACF (A) and (B).

shows good capacitor characteristics at low temperatures. At low temperatures the electrical resistivity and viscosity of the organic electrolytic solution become higher, suppressing the formation of an electric double layer in micropores with small diameters.11 A study on the relationship between a capacitance density and a differential specific surface area of activated carbons was reported to obtain the optimum pore size distribution for the activated carbons of EDLC.13 A concentration of surface acidic functional groups such as carboxyl or acidic hydroxyl groups of ACF is also one of the significant factors in forming an electric double layer. Figure 10.13 shows the relationship of a leakage current of the capacitor vs a concentration of the surface acidic functional group of ACF. Lower concentration of the acidic functional groups on the surface of ACF resulted in the lower leakage current of the capacitor. The heat treatment of the ACF decreased the concentration of the surface acidic functional groups of ACF, which resulted in the capacitor with a low leakage current.14–16 The micro-structure of activated carbons significantly affects a formation of electric double layers. Randin et al.17 reported that higher capacitance per area was obtained in the edge structure of carbon than in the plane structure of carbon.17 Figure 10.14 shows a relationship between specific capacitance per volume of electrodes and a filling ratio of activated carbons in the electrodes. Higher filling ratios of activated carbons are strongly required to obtain a capacitor with a high capacitance density per volume.18 10.5.2

Electrolytic solutions

An electrochemical decomposition of electrolytic solutions during charge– discharge cycles of the capacitors determines a rated cell voltage. A wider voltage

(a)

(b)

Apparent leakage current (µA)

70 60 50 40 30 20 10 0 0 0.1 0.2 0.3 Concentration of SAFG (mmol · g–1 of ACF)

Concentration of SAFG (mmol · g–1 of ACF)

Capacitors and micropower systems 281

0.3

7 min

0.2

10 min

0.1

12 min 0 as prepared 800 1000 1200 Heat treatment temperature (°C)

Capacitance density (F · cc–1)

Figure 10.13 Relationship between capacitor characteristics and a concentration of surface acidic functional groups of activated carbons. (a) Concentration of SAFG vs leakage current; (b) heat treatment temperature vs concentration of SAFG.

20

15

10

5

0 55

60 65 70 Filling ratio of activated carbons (%)

75

Figure 10.14 Relationship between a capacitance density and a filling ratio of activated carbons in polarizable electrodes.

window in a CV curve is needed for higher rated voltage capacitors. Figure 10.15 shows a representative CV curve of electrolytic solutions. As the electrochemical stability at anodic and cathodic regions is different in solvent species, an appropriate combination of electrolyte, solvent, and activated carbon species have to be chosen. The inclusion of water in the solution will cause significant damage to the cell voltage.19

282

A. Yoshida

3 mA

(a) Propylene carbonate/Et4NBF4

Current

(b) Propylene carbonate/Et4NBF4 with 1%wt H2O added

WE: activated carbon fiber CE: Pt Scan rate: 5 mV · sec–1

(c)

Sulfolane/Et4NBF4

– 4.0

–3.0

1.0 2.0 –2.0 –1.0 0 Voltage vs Ag/Ag+ electrode (V)

3.0

Figure 10.15 CV curve of electrolytic solutions.

10.6

Application of the capacitors

Although the energy density of electrochemical capacitors is considerably low compared with that of rechargeable batteries, the capacitor shows many other features described in the previous section, that is, a higher power density, a stable charge–discharge performance in a wide temperature range, an environmental advantage, etc. The electrochemical capacitors have been applied to many kinds of electrical equipment and systems as energy back-up devices. 10.6.1

Low rate uses

Small coin-type capacitors are used as memory back-up devices for microcomputers in many electronic equipments. The capacitors discharge at less than mA-rate-load in VCRs, handy telephones, pagers, etc. to back-up memories. 10.6.2

High rate uses

Tubular-type capacitors have such a low inner resistance that they are used as back-up devices for actuators in many electric appliances. The capacitors

Capacitors and micropower systems 283

Figure 10.16 Appearance of toys with the capacitors. Table 10.1 Summary of application of the electric double-layer capacitors

Memory back-up Actuator back-up High power back-up Hybrid power source Electric power storage

System construction

Application

Capacitor/AC power source /Battery Capacitor/AC power source /Battery /Solar cell Capacitor/AC power source /ICE /Battery Capacitor/ICE /Battery Capacitor/AC power source /Battery /Solar cell /Windmill power generator

VCR, Camera, Clock Handy phone, Pager Home appliance Toy Toy, Watch Elevator Automobile EHC Hybrid engine EV, Power assist system Load leveling Output leveling

discharge at less than A-rate-load in toys, measuring equipments, traffic signals, electric thermo pots, etc. to drive motors or LEDs.9,18 Figure 10.16 shows an appearance of toys with the capacitors to drive a motor. The power-type capacitors can be charged and discharged at higher than several hundred A loads. Many applications of the power-type capacitor have been studied in the systems of electric heated catalysts, EVs, elevator systems, AC-cordless devices, etc.

284

A. Yoshida

In all these capacitor-systems, the capacitors work as energy-storage devices for a temporary period in which the capacitors are charged by solar cells, batteries, AC sources, inner combustion engines, fuel cells, etc. Many advantages are obtained in the capacitor hybrid system; one of the examples of the advantage is that the discharge period for the battery voltage to get to the cut-off value at a highly fluctuated load can be lengthened in the capacitor–battery hybrid system.20 Table 10.1 summarizes the application of the electrochemical capacitors; both the examples that are shown are actually used now and can be potentially applied in the future.

10.7

Future prospects of the capacitors

Many studies have been carried out for the electrochemical capacitors to be applied in the field of high current load. Novel technology to attain high energy density, high power density, high reliability, low cost of the capacitors, and electric circuits for efficient charges–discharges of capacitors, are required. The development of new material for electrodes and electrolytic solutions, and power electronics study with inverter–converter components are significant.21–23 In the 21st century, the electrochemical capacitor system will play an important role in the field of energy storage or energy management. A load leveling of electric power plants, an output leveling of solar cells or windmill power generators are potential applications with the capacitors. Application of the capacitor in EV system is another important expectation for the capacitor system.

Acknowledgement The author would like to express his sincere appreciation to Mr. K. Yoshioka, Mr. S. Nomoto, Mr. H. Handa of Matsushita Electric Ind. and Mr. M. Okamoto of Matsushita Electronic Components for their helpful discussion.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

H. L. F. von Helmholtz, Ann. Phys., 3, 337 (1879). G. Gouy, J. Phys., 4, 457 (1910); D. L. Chapman, Phil. Mag., 25, 475 (1913). O. Stern, Z. Elektrochem., 30, 508 (1924). A. Yoshida, Denki Kagaku, 66, 884 (1998). H. I. Becker, US Patent 2,800,616 (1957). A. Yoshida, I. Tanahashi, and A. Nishino, IEEE Trans., CHMT-11, 318 (1988). A. Nishino, Tanso, 132, 57 (1988). A. Yoshida, I. Imoto, A.Nishino, and I. Yoneda, IEEE Trans., CHMT-15, 133 (1992). A. Yoshida, I.Aoki, S. Nonaka, and A. Nishino, J. Power Sources, 60, 213 (1996). M. Okamoto, Proc. Denkijidoushakennkyuukai (1996). A. Yoshida, I. Tanahashi, and A. Nishino, IEEE Trans., CHMT-10, 100 (1987). I. Tanahashi, A. Yoshida, and A. Nishino, Denki Kagaku, 56, 892 (1988). S. Nomoto, H. Handa, K. Yoshioka, and A. Yoshida, Denkikagaku Fall Meeting, 1E13 (1998).

Capacitors and micropower systems 285 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

A. Yoshida, I. Tanahashi, and A. Nishino, Carbon, 28, 611 (1990). S. Nonaka, I. Aoki, and A. Yoshida, Denkikagaku Fall Meeting, 2D25 (1993). T. Momma, T. Osaka, and X. Liu, Denki Kagaku, 64, 143 (1996). J. Randin and E. Yeager, J. Electroanal. Chem., 36, 257 (1972). A. Yoshida, S. Nonaka, and A. Nishino, ECS Proc., 95(29), 210 (1995). H. Handa, S. Nomoto, K. Yoshioka, and A. Yoshida, Denkikagaku Fall Meeting, 1E11 (1998). S. Nomoto, S. Nonaka, K. Nishida, M. Ikeda, and A. Yoshida, ECS Proc., 96(25), 268 (1996). K. Sakai, New Capacitor, 2, 48 (1995). A. Yoshida, J. IEE Japan, 117(3), 171 (1997). F. Matsui, K. Watanabe, T. Nitta, N. Jinbo, K. Shibano, and G. Yamada, J. Society of Automotive Eng. Japan, 51, 24 (1997).

11 Batteries for micropower applications Boone B. Owens, Curtis L. Holmes, John Bates, William H. Smyrl, N. J. Dudney, B. J. Neudecker, and Stefano Passerini

11.1

Introduction

The development of advanced batteries for micropower applications is intimately linked to the availability of new materials and the development of miniature battery designs. The need for such new power sources results directly from the continuing evolution of smaller electronic circuitry with very high component density. The range of power requirements is illustrated by the variety of relatively new miniature portable electronic devices such as cardiac pacemakers, hearing aids, smart cards, personal gas monitors, and remote sensors with RF capability. This chapter will not purport to present a complete review of the topic, but will summarize the broad areas of battery requirements and give some detailed summaries of selected areas, based on the authors’ specific interests. The individual authors of this chapter are recognized experts in one or more of the following areas, and will be identified in the respective sections of the review: Primary batteries ● Prismatic, long life batteries for biomedical implantable applications ● Thick film primary, pouch cells Secondary batteries ● Miniature button cells ● Thin film, rechargeable microbatteries ● High capacity amorphous oxide cathode materials Fuel cells ● Microfuel cells A micropower battery may be simply defined as a battery used to provide power in the range of microwatts. This power requirement may be for intermittent or continuous operation over short or long periods, and in some cases there may be high power pulses superimposed on the low microwatt background power level. The batteries may be single cycle, low rate primary cells or multicycle secondary cells capable of being recharged for many cycles. Because of the variety of combinations of possible use conditions and the wide range of environmental

Sensor card; smart card electronic key Rechargeable hearing aid Sensor RF ID tags implantable medical devices IC power source Research cells

Li Pouch, primary cell

Sol gel processed thick film Li or Li-ion cells with amorphous cathode films Methanol/air fuel cells, Si wafers, MEMS technology

Research cells

Cardiac pacemakers

Li prismatic, primary cell

Li-ion button, secondary cell Li, thin-film microbattery, secondary cell

Application

Technology

Table 11.1 Micropower batteries

MeOH/O2 NAFION membrane

Li/V2O5

Li-ion button cell Li/glass electrolyte/ metal oxide

Li/MnO2 liquid electrolyte

Li/I2, in situ formed “solid” electrolyte

Battery

Thin or thick film, prismatic cells; composite electrode button or coated electrode cells Cell on a chip

4.5 mm ht., 8 mm diameter 1 m films, large area cells

4–8 mm thick; 2–10 cm2 area; irregular custom shapes to conform to product case 1–5 mm thick, large area cells

Shape factor 1–5 Wh

1 Wh

40 mWh rechargeable 100 Wh, rechargeable

20–200 W

150 W monitor; 30 mW alarm 500 W; pulses to 40 mW 1 W–20 mW

1 mW

24 Wh/kg 180 Wh/l

10 W–10 mWcm2 High energy cathode materials

Energy

Power

In research stage of development

In research stage of development

Up to 5 years; 1200–2000 cycles 1000s of cycles; 5 y 20–150C

Single cycle primary, 1 year life

Single cycle primary, 3–10 year life, 37ºC isothermal operation

Life, environment

288

Owens et al.

conditions that may be required, this chapter will best be viewed as an illustration of types of batteries appropriate to selected applications. Designers of devices and of batteries should work closely together, early in the development cycle to obtain the optimal final configuration of product. Miniature fuel cells have recently emerged as a possible power source for applications that range from cell phones and small digital devices to autonomous sensors to embedded monitors or to microelectromechanical system (MEMS) devices. Their use in these applications will require small volumes and small mass, but the performance should approach or exceed that of large systems when scaled for size or they may not be attractive. In many of these applications for micropower, there is direct competition with miniature batteries. In miniature fuel cells, there is interest in adapting silicon microtechnology to the formation of viable single cells as well as arrays of cells (i.e. stacks) to deliver the desired power and energy. The topics to be addressed in the present review are summarized in Table 11.1. Also shown are the nominal specifications for the power, in each example. The power levels range from a few microwatts, up to about 1 mW, and may include intermittent pulse requirements of 10–100 mW. Lithium-based batteries have demonstrated very high values of energy density. However, the values are generally based on the performance of larger cells with capacities of up to several ampere-hours (Ah). For small microbatteries the achievable power and energy densities are diminished because the packaging and internal hardware will determine the size and mass of the completed package. As a consequence of this, the more adaptable the cell system is to miniaturization, the more useful it will be in the evolving area of new micropower devices. For this reason, the use of solid electrolytes (and thus the absence of liquid electrolytes) within the electrochemical cell system may be expected to dominate this area in the future.

11.2

Lithium cells for cardiac pacemakers C. L. Holmes

The cardiac pacemaker is a mature product that has been used for the treatment of a heart arrhythmia known as bradycardia since the early 1960s. The first such devices were typically powered with zinc/mercuric oxide batteries, which showed insufficient longevity and a sudden approach to the end of battery life. The development of low power electronic components in the 1970s permitted the use of high impedance batteries, and the first Li-battery powered cardiac pacemaker was implanted in Italy in 1972.1 The battery was the lithium/iodine-polyvinylpyridine (PVP) system. In the earlier days of lithium-battery powered pacemakers, several chemical systems were used. These included lithium/iodine-PVP, lithium/silver chromate, lithium/cupric sulfide, and lithium/thionyl chloride. By the late 1970s virtually all cardiac pacemakers were powered by Li batteries, and greater energy density

Micropower batteries 289 and reliability were observed. It could be argued that implantable cardiac pacemaker devices represented the first commercially successful application of Li batteries. Gradually, the chemical systems other than lithium/iodine-PVP fell out of use, and by the mid-1980s virtually all pacemakers were powered by this system. From 1972 until today over five million pacemakers using the lithium/iodine-PVP batteries have been implanted, and this system remains the power source of choice for cardiac pacemakers today. The cardiac pacemaker is truly a micropower system today, operating at a level of 20–200 W. Advances in electronic components, pacemaker lead technology, and patient management have made possible devices requiring less than 20 A current drain. Typical battery requirements for a pacemaker are shown in Table 11.2. The lithium/iodine-PVP battery was invented by Schneider and Moser 2,3 based on earlier work of Gutmann et al.4 In the 28 years following the first implant, this system has been the subject of much development, improvement, and characterization, and today there exists a rather impressive body of knowledge concerning the behavior of this system.5,6 Lithium/iodine-PVP batteries in use today typically consist of a centrally placed Li anode corrugated to increase surface area and coated with a thin layer of PVP. This central anode is surrounded by the cathode material, which consists of one part PVP mixed with 30–50 parts by weight elemental iodine and thermally treated to form the cathode. The iodine-PVP material is poured into the battery in the molten state and freezes to form a tar-like material that consists mostly of elemental iodine. The anode current collector is brought out of the battery through a glass-to-metal seal, and the cell is welded shut to insure a hermetic seal. The iodine-PVP material shows a unique property of electronic conductivity, which allows the proper operation of the battery. This electronic conductivity is a function of the ratio of iodine to PVP in the cathode material. It shows a maximum conductivity at a weight ratio of about eight parts iodine to one part PVP, and decreases rapidly as the iodine is depleted past this point during cell discharge.7 Table 11.2 Pacemaker battery requirements Parameter

Value

Voltage Power Current Capacity Temperature Internal impedance Operating life time Shelf life

2.5–3 V 20–200 W 10–100 A 300–2000 mAh 37ºC 10–1000 3–10 years 2 years

290

Owens et al.

The chemical and physical properties of this cathode material have been extensively studied. Brennen and Untereker8 have reported that one iodine atom replaces a hydrogen on the “backbone” of the PVP, and a molecule of iodine is bonded to the nitrogen in the pyridine ring of the polymer. Thus, three atoms of iodine per monomer unit of PVP are unavailable for cell reaction, and the remainder of the iodine, at unit thermodynamic activity, is available to react with the lithium. The cell reaction is simply the formation of lithium iodide from the elements, that is, Li 12 I2 LiI

(1)

The Gibbs free energy change for this reaction is 64.45 kcal/mole, yielding the observed open circuit voltage of 2.8 V. The electrolyte material is the solid lithium iodide that is formed in situ as the battery is discharged. The anode coating procedure mentioned above results in profound changes in the morphology of the lithium iodide and therefore the performance of the battery.9,10 Indeed, it has been shown that this coating procedure leads to the formation of a highly conductive liquid phase in the electrolyte,11 and this phase coupled with the macroscopic effects observed in the electrolyte, lead to a lower Li-ion conductivity than would be expected from that of pure lithium iodide, which in turn leads to a lower impedance and better current-delivery capabilities of the battery. The ever-increasing thickness of lithium iodide coupled with the impedance characteristics of the cathode material lead to a discharge curve in which the voltage begins near the open circuit voltage of 2.8 V and decreases somewhat linearly throughout most of the discharge. Near the end of service, the increasing cathode resistance causes a more dramatic but still gradual decrease to the point at which the pacemaker requires elective replacement. A typical pacemaker battery that has seen considerable use for about 15 years is known as the Model 8402. Its characteristics are shown in Table 11.3, and in Figure 11.1. The figure illustrates the degradation of performance at the higher current drains. The relation between delivered capacity (and energy) and current Table 11.3 WGL Model 8402 pacemaker battery Property

Value

Weight Volume Operating temperature Storage temperature range Gravimetric energy density Volumetric energy density Width Height Thickness

14.4 g 3.72 cc 37ºC 40 to 52ºC 230 Wh/kg 900 Wh/l 45 mm 23 mm 5 mm

Micropower batteries 291 3000

2500

5.49 k Ω 12.4 k Ω 21.0 k Ω 47.5 k Ω 100 k Ω

Voltage (mV)

2000

1500

1000

50 0

0 0

200

400

600 800 Capacity (mAh)

1000

1200

1400

Figure 11.1 Discharge curve of the Model 8402 battery under selected constant resistive loads.

drain can be summarized by a Selim–Bro12 curve. This curve plots delivered capacity of the battery as a function of the current drain (at beginning of life). Figure 11.2 shows the Selim–Bro curve for the battery discussed above. The capacity is decreased at higher current drains because the internal resistance of the battery causes the voltage to decline more rapidly. The slight decrease in capacity seen at the very low current drains is caused by self-discharge. In the first part of the cell’s discharge, the lithium/iodine layer is thin, and iodine molecules can diffuse through to react directly with the Li anode. This process is very slow and decreases to negligible levels as the electrolyte layer becomes thicker. However, at very low current drains, this self-discharge causes a reduction in delivered capacity. Today’s pacemakers operate at very low current drains. The actual current drain is determined by many parameters, including the frequency at which the patient needs to receive cardiac stimulation, the programmed settings of energy, pulse width, and rate, the effects of variable-rate pacing, and the impedance of the lead implanted in the heart. A typical use profile for a pacemaker can be approximated by a 15 A constant current discharge. For the Model 8402 battery discussed above, the capacity delivered under this load is between 1200 and 1300 mAh to a 1.8-V cutoff. The time required for this discharge regime is approximately 8 years. A typical discharge profile is depicted in Figure 11.3. Lithium/iodine-PVP batteries have demonstrated excellent reliability in the cardiac pacemaker application. Their unique ability to provide microampere

292

Owens et al. 1400

Delivered capacity (mAh)

1200 1000 800 600 400 200 0 1

1000

10 100 Beginning of life current (µA)

Figure 11.2 Selim–Bro curve for the Model 8402 battery.

3000

Voltage (mV)

2500 2000 1500 1000 500 0

0

200

400

600 800 1000 Capacity (mAh)

1200

1400

Figure 11.3 Typical discharge curve of the Model 8402 battery at 15 A constant current.

currents for many years has confirmed they were a good choice for this application. The system will not provide high current drains, does not operate well at either low temperatures or temperatures above about 60ºC, and thus has very limited application. However, held at 37ºC and asked to deliver 5–50 A, the system will provide many years of reliable performance. The low current delivery capability of the lithium/iodine-PVP system does cause some inconvenience in special pacemaker applications such as programming and telemetry. For these purposes, the battery can be called on to provide temporary currents of 100 A or higher. As the cell impedance rises, the voltage

Micropower batteries 293 of the lithium/iodine-PVP system may fall below desired levels during these higher current demands. This requires special voltage-management techniques which may result in longer than desired telemetry times. For this reason, it is possible that at least some pacemakers of the future will be powered by lower impedance systems such as lithium/manganese dioxide or lithium/carbon monofluoride. Such systems are available and may see use in pacemakers of the future.13 Pacemakers represent a unique requirement for micropower. This application requires a power source which can provide low current drains for many years. This in turn requires a thorough understanding of the chemistry, materials compatibility, long-term performance, and stability of the system. It also requires the ability to model long-term performance based on short-term accelerated tests. The lithium/iodine-PVP system meets all of these requirements. The system is well understood, many years of real time and field data are available, and excellent modeling and predictive capabilities have been developed. It is likely that this system will see many more years of use in this interesting and important application.

11.3

Micropower lithium cells for solid-state hazardous gas monitor B. B. Owens

Smart cards, similar in form-factor to a credit card, have been of interest for a variety of applications such as gas monitoring and entry keys, in addition to smart credit cards. Some such gas monitors require the development of an extremely compact, low cost, and long lived chemical detection system in a credit card format that can be either attached to the clothing of field personnel or affixed to a convenient mobile or stationary surface. Micropower circuitry including sensor, alarm, and display elements can be assembled to provide a viable device that requires a power level of 150 W for monitoring and 20–30 mW for alarm. Such a device named the Hazard Card was developed to incorporate an electrochemical sensor for carbon monoxide, an LCD display and a piezoelectric alarm.14 This instrument is shown in Figure 11.4. The design goal was to provide the product in the package envelope of a credit card. The time average power requirement of the Hazard Card was 150 W (or 50 A at 3 V), and this could be provided by a 1–2 mAh thin-film (10 m thick) lithium solid state microbattery, provided the cell could be recharged daily. Alternatively, if the Hazard Card was powered by a lithium primary cell, with a capacity of 450 mAh, then it would provide continuous, maintenance-free operation for about 1 year. Primary cells with this capacity are available in the configuration of pouch cells or large coin cells. A thin-film rechargeable microbattery was initially evaluated for this Hazard Card. The thin film, solid state microbattery technology is described by Bates in Section 11.5. The trade-off between using the rechargeable microbattery and using a primary cell is that the primary cell results in a thicker card device

294

Owens et al.

(8–10 mm thick) that is ‘maintenance free’ whereas the solid-state microbattery could reduce the thickness by 2 mm. The total device thickness would be about 6–8 mm, and the power cell would have to be recharged on a daily basis. Flat prismatic batteries from several suppliers are commercially available, for example, Pouch Cell X758C (MSA) thickness of 2.1 mm Powerdex 3C120M (Gould) thickness of 1.2 mm U3VF-G-T (Ultralife) thickness of 1.6 mm These primary pouch cells are available in a range of sizes (capacities), and provide stable micropower outputs for extended time periods. Output voltages, determined by the LiMnO2 chemistry range between 3 and 2.7 V, and the pouch cells readily sustain current pulses in the milliampere range. Short pulses in the milliwatt range were required to drive the alarm function of the Hazard Card. The cells are sealed in aluminized polymer foil flat-packs. Such flat cells were readily designed into a “thick” Hazard Card with final dimensions of 6 9 1 cm3, as illustrated in Figure 11.4.

Figure 11.4 Hazardous gas monitoring badge.

Micropower batteries 295 Rechargeable pouch cells are also available, based on Li-ion gel electrolyte systems. These cells are designed for notebook and cellular phone applications and are not intended to be a micropower battery. However, the Li-ion chemistry may be designed into a micropower cell, as will be described in the following section.

11.4

Lithium-ion cells for rechargeable hearing aids B. B. Owens and S. Passerini

The hearing aid is an example of a low power device that incorporates small batteries. Typically, hearing aids are powered by small zinc air cells with capacities ranging from 50 mAh up to several hundred milliampere-hour. The more energy efficient hearing aids draw currents of a few hundred microampere, equivalent to power levels of less than 500 W. The incidence of hearing disorders is widespread, affecting about 10% of the population. It has been estimated that over 25 million individuals in the United States suffer hearing loss and approximately 7 million use hearing aids. The zinc air hearing aid batteries are not rechargeable and must be replaced after a few days or weeks. As a result, the annual consumption of these batteries is over 200 million in the United States alone. There are some disadvantages to the consumer that result from the use of these throwaway batteries in hearing aids. These include: 1

2 3

4

Cost of replacement cells. The average user will purchase about 150 replacement cells during the 4 year lifetime of the hearing aid (nominal 4 years). Battery disposal. 250 million cells are disposed of each year. Consumer safety. The small button cells can be accidentally swallowed or inserted into the nose or ears. This presents a medical problem, especially with small children and the elderly. Cognitive/dexterity impairment. Individuals with poor eyesight and/or limited dexterity have difficulty in replacing the small cells in the hearing aid. Battery replacement in a hearing aid is an increasing problem as the trend in product design has been to make the devices smaller and less obtrusive in the ear, the so-called ITC (in the canal) or CIC (completely in the canal) aids. A smaller hearing aid requires a smaller battery, and this requires more skill to handle and replace the cell. Again, these problems are more often found with the elderly patient.

A rechargeable hearing aid would resolve the above problems, and some such aids have been available for many years, in a relatively large size of device. These products were powered by NiCd batteries that did not permit the design of a small hearing aid. Also, the lifetime of the device was too short for trouble-free use over a 4- or 5-year period. Consequently, rechargeable hearing aids are not widely used today, and the technology limiting component has been the battery.

296

Owens et al.

In order to overcome the above indicated battery-related problems, a small, 3.6 V Li-ion button cell was developed.15–18 The cell has a typical Li-ion chemistry, with a carbon anode, and lithiated cobalt oxide cathode, and may be schematically depicted as: Cu/Graphite//1 M LiAsF6 in EC-DMC// LiCoO2/Al

(2)

where copper (foil) and aluminum (foil) serve as the current collectors and a typical electrolyte is 1 M lithium hexafluoroarsenate (LiAsF6) dissolved in an ethylene carbonate (EC)-dimethyl carbonate (DMC) organic solvent. Both of the active components of the battery are Li-ion insertion compounds that permit a highly reversible set of Faradaic electrode reactions. A conventional design for the external envelope of this cell was selected and the materials used for the internal cell design were similar to those of standard commercial Li-ion batteries. Consequently, the electrochemistry of the cell is very similar to that of the larger 500–1000 mAh Li-ion batteries used in consumer electronics. However, in scaling down the size of the cell to a 10 mAh capacity, it was necessary to maintain a low internal resistance (5–10 ) so that the cell output voltage would be stable when pulse loads were required. Normally button cells have disk or slug electrodes, and with organic electrolytes one might anticipate an internal resistance of at least 100 . The internal cell structure was selected to maximize the inter-electrode area. A schematic of the cell crosssection design is shown in Figure 11.5. The early prototype cells (Model 312) had a diameter of 8 mm and a height of 3.6 mm. The final design for cell Model 845 maintained the 8 mm diameter but the height was increased to 4.5 mm. The size perspective of the Model 845 battery is illustrated in Figure 11.6. The battery was designed as a 10 mAh, 3.6 V cell. The nominal mass and volume are 0.5 g and 0.2 cm3, respectively. The actual capacity depends on a number of factors, including the voltage limits for charge and discharge, the rates for each of these, the cycle number and the temperature. Typical discharge and charge profiles are shown in Figure 11.7 for a Model 312 cell operating at

Gasket

Lid (negative)

Insulating tape Electrode assembly

4.5 mm

Case (Positive) 8.0 mm

Figure 11.5 Schematic of cell section. Model 845 Li-ion cell.

Micropower batteries 297

Figure 11.6 External view of 10 mAh Li-ion cell.

Cell voltage (V)

5

(c) 4

3

(a)

(b)

2 0

2

6 4 Discharge capacity (mAh)

8

10

Figure 11.7 Discharge and charge curves for the Model 312 cell; (a) C/12 rate discharge; (b) C rate discharge; (c) C/12 rate charge. (By permission Elsevier Science.)

ambient temperature. The two discharge curves are for rates of C/12 (0.75 mA) and C (9 mA). The single charge curve is for constant current (CC) charging at 0.75 mA, up to the voltage limit of 4.1 V. The capacity is a function of the charge protocol, and also is dependent upon the discharge rate and the lower voltage limit for discharge. The cell of

298

Owens et al.

Figure 11.7 was charged by CC to an upper limit of 4.1 V at the C/12 rate. For the conditions of Figure 11.7, the discharge capacity values were 8.5 mAh at the C rate and 9.5 mAh at the C/12 rate, to the cutoff voltage of 3.0 V. The preferred charge protocol for Li-ion cells is CC charging to an upper voltage limit, followed by constant voltage charging to some lower limit of the charge current (or to a fixed time). This two-step charge protocol, referred to as CCCV, permits a more rapid initial CC charging of the cell, at the C or C/3 rate, followed by a constant voltage step that “tops off ” the capacity as the charge current falls off to a low value such as C/100 (100 A). In this case, the cell of Figure 11.7 would have a capacity of over 10 mAh, at the rate of C/12. The pilot-line Model 845 cells exhibit a good reproducibility of performance, as shown in Figure 11.8, for a group of seven cells, discharging at the C/3 rate, after CC charge at C/3, to 4.1 V. One of the important characteristics of a battery is the impedance, and also the frequency dependence of the impedance. The impedance was determined for one of the prototype button cells (Model 312), and was compared to that of similar miniature cells based on NiMH, Zinc Air, and Li/V2O5 chemistries. The results are shown in Figure 11.9. The Li-ion button cell had an impedance of about 10 at the lower frequencies and 6 at the high frequencies near 105 Hz. With the internal impedance of under 10 , the cell not only functions well at the hearing aid rate of 300 A, but may also be discharged at the C rate, and even pulse discharged at the 10C or 20C rates. The cell may be discharged at a low background rate such as 10 or 100 A, with intermittent pulses of 10 or 100 mA, for example. A pulse discharge test is illustrated in Figure 11.10. The cell was charged by CC, 3 mA, 4.1 V, and then discharged down to the lower voltage limit of 2.75 V. The dashed curve of Figure 11.10 shows the cell discharging at a constant current of 2 mA and delivering a capacity of 7.3 mAh. Then the cell was recharged and discharged under a continuously repeating pulse mode that 4.2

Cell voltage (V)

4 3.8 3.6 3.4 3.2 3 2.8

0

1

2

3

4 5 6 Capacity (mAh)

7

8

9

10

Figure 11.8 Discharge curves for a set of eight Model 845 cells.

Micropower batteries 299 1000

Impedance (Ω)

100

10

312 A Li-ion Ni-MH Zn-Air Vl 1620

1

0.1

10

0

100 1000 Frequency (Hz)

10000

Figure 11.9 Impedance curves for hearing aid size cells. (By permission Elsevier Science.)

Cell voltage (V)

4

3

2

0

1

2 3 4 5 6 Discharge capacity (mAh)

7

8

Figure 11.10 10 mA pulses superimposed on a 2 mA constant current discharge. (By permission Elsevier Science.)

consisted of 2 mA for 8 s, followed by 10 mA for 2 s. The capacity under this higher rate was 7.2 mAh, in good agreement with the value at the lower rate of discharge. The cell would easily function at a lower background rate of, for example, 10 A, with intermittent pulses at 10 mA, and be expected to deliver substantially all of the capacity. Other pulse tests were run on the cell at nominal currents of 50 and 100 mA. These were constant resistance loads of 70 and 30 that were placed across the

300

Owens et al.

cell for 10-s periods, with a recovery time of 15 s between the loads. The results are shown in Figure 11.11. This micropower battery may also provide power at levels of several hundred milliwatts, and thus it may also be useful for a wide range of applications. Self-discharge of Li-ion batteries has been reported to be in the range of a few percent a month up to 10% per month. Estimation of the rate of self-discharge in the Model 312 cell was determined by measuring the limiting current required to maintain the cell at various voltage levels (or states of charge). The self-discharge current was about 0.1 A which would amount to a loss of about 10% of the cell capacity in a period of 1 year. Direct tests of capacity loss over real time should be performed to obtain direct confirmation of the self-discharge losses at temperatures specific to any application. Cycle life is an important characteristic of the Li-ion battery. Commercial cells of this chemistry have cycle lives reported to be from 500 up to about 2000

4.0

Cell voltage (V)

(a)

3.8

3.6

3.4

(b)

120

160 Time (s)

200

Cell voltage (V)

4.0

3.6

3.2

2.8

220

260 Time (s)

300

Figure 11.11 50 and 100 mA pulse discharge tests on Model 312 cell. Constant Resistance Pulse Tests. 312 A cell voltage behavior during nominal 50 and 100 mA pulse discharge (10 s) on 70 constant load (panel (a)) and 30 constant load (panel (b)). The 10-s pulses were alternated with open circuit rest periods of 15 s. (By permission Elsevier Science.)

Micropower batteries 301 cycles. Life tests have been reported for this hearing aid cell, under various conditions, and they are in agreement with those of the larger commercial cells. Five hundred cycles is readily obtained, and has been determined by an independent test lab as well as by the developer.14,17 Results for cells cycling at 60% depth of discharge, as obtained at the independent test laboratory are shown in Figure 11.12.17 Other life tests were performed on cells cycling at the 50% depth of discharge, with a periodic full charge and discharge to identify any loss in the full capacity. Figure 11.13 shows these results; up to 2000 cycles, 80% of the initial cycle capacity was still delivered during a full discharge.15 These cycles were continued at the 50% DOD, with intermittent complete cycles to assess capacity degradation.

10 9 Capacity (mAh)

8 7 6 5 4 3 2 1 0

0

100

200

300 400 Cycle number

500

600

Cell voltage (V)

Figure 11.12 Independent test lab results for 500 cycles.

4

3 1600 2700 2000 0

2

4 6 Capacity (mAh)

1

8

Figure 11.13 Cycle life tests; discharge curves for cycles 1, 1600, 2000, and 2700.

302

Owens et al.

After 3000 cycles, the full discharge capacity was still 70% of the initial capacity of this cell, as shown in Figure 11.14. Figure 11.15 shows the diagnostic cycle 3000, and the 50% DOD cycle 3001. The cell still delivers the 5 mAh as it has done throughout the 3000 cycles of testing. However, whereas initially the discharge curve was much flatter with the load voltage remaining above 3.6 V, it has degraded to a lower limit of 3 V as shown for cycle 3001. The plot of capacity delivered per cycle has been constant, at 5 mAh as shown in Figure 11.16 for cycles 2900–3000. Points A and C are the full discharges that were obtained every 100 cycles for diagnostic purposes. Preliminary safety test results were reported for the button cell.16 Because of the relatively large thermal mass of the cell as compared to its mass of active materials, it appears to be safe, even though it may be pulsed at rates of as high as 20C (200 mA). The small button cell with Li-ion chemistry has shown a high

Cell voltage (V)

4.5

4

3.5

CN 3000

3

2.5 0

2

4 Capacity (mAh)

6

8

Figure 11.14 Cycle life tests; discharge curve for cycle number 3000.

Cell voltage (V)

4.5

CN 3000

CN 3001

3.5

2.5

0

10 Time (h)

Figure 11.15 Cycle 3000 and 3001, for Model 312 cell.

20

Micropower batteries 303

Capacity (mAh)

A

C

5 B

0 2900

3000 Cycle number

Figure 11.16 Capacity vs cycle number (CN) for cycles 2900–3000. Cell Model 312: Cycles A and C were for 100% discharge following a full charge to 4.1 V; cycles B were for 5 mAh discharge, following charge to 4.0 V.

level of performance that is consistent with that of larger cells of this chemistry. The specific energy and the energy density of this cell are 70 Wh/kg and 200 Wh/l, respectively comparing favorably with the values of 110 Wh/kg and 250 Wh/l that are typical for larger cells.17

11.5

Recent developments in thin-film lithium and lithium-ion batteries J. Bates, B. J. Neudecker, and N. J. Dudney

Research on thin-film solid state Li batteries begun a decade ago at the Oak Ridge National Laboratory19–22 has resulted in the development of practical solid-state thin-film rechargeable lithium and Li-ion batteries. The batteries, which are less than 15 m thick, have important applications in a variety of consumer and medical products, and several companies in the United States are now moving toward production of these devices. The long cycle and shelf life of the batteries result from the properties of the glassy lithium phosphorus oxynitride (“Lipon”) electrolyte23,24 which is stable in contact with metallic lithium at potentials from 0 to nearly 5.5 V and has an acceptable conductance in thin-film form. Responding to the need for thin-film batteries that can tolerate heating to 250–260ºC so they can be integrated into circuits using the solder reflow process, we have synthesized several inorganic anode materials25,26 which yield thin-film Li-ion cells that are stable at these temperatures. Recently, we have investigated batteries with in situ plated lithium anodes27 which, prior to the initial plating, also are not affected by the solder reflow conditions.

304

Owens et al.

In a previous volume of this series22 we discussed our research on thin-film Li batteries that was current to about March of 1998. The purpose of this contribution is to summarize the progress made from this date through about August 1999. Emphasis will be placed on the performance and applications, as we have recently reviewed the use of thin-film batteries in the study of lithium intercalation compounds.28

11.5.1

Battery fabrication

Details of the fabrication steps have been discussed earlier.19–22 With the exception of the Li anode and the parylene protective coating, all of the layers are deposited by magnetron sputtering. The slowest steps in the fabrication are deposition of the cathode and electrolyte films at rates of up to 75 and 180 Å/min, respectively. Deposition of the cathode at higher rates onto uncooled substrates causes the substrate temperature to rise above 100ºC, leading to excessive grain growth and a low-density microstructure at the film surface. This results in a significantly higher cell resistance. With the possibility of rapid post deposition thermal annealing of the LiCoO2 or LiMn2O4 cathodes, it should be possible to fabricate a complete battery in 10 h in a production environment.

11.5.2

Lithium and lithium-ion cells with textured polycrystalline LiCoO2 cathodes

The highest performance of thin-film batteries has been achieved with polycrystalline LiCoO2 cathodes with a high degree of preferred orientation (i.e. textured films) and metallic lithium anodes. In a recent study,29 we found that the charge–discharge rates are limited by the electrolyte resistance for cells with cathodes of up to 3 m thick. For more than 50 Li–LiCoO2 cells, the average cell resistance was about 110 cm2. Discharge curves and graphs of capacity and energy vs discharge current in mA and C-rate for discharge between 4.2 and 3 V for one of the best cells we have fabricated having a resistance of 70 cm2 are shown in Figures 11.17 and 11.18, respectively. At a discharge current of 6 mA/cm2 (30 C), for example, the cell delivered 50% of its maximum capacity and energy. The cycle life of Li/LiCoO2 thin-film batteries is exceptionally high. As illustrated in Figure 11.19, lithium cells with capacities of up to 0.15 mAh/cm2 showed no measurable capacity loss within experimental error after 4000 cycles at discharge rates of 0.1 mA/cm2 and 100% depth of discharge (DOD). At higher discharge rates, up to 1.5 mA/cm2, some capacity fade was observed, but the loss per cycle diminished to zero on returning to more moderate discharge rates. Earlier 21 we showed that with very thin cathodes, in excess of 40 000 cycles could be obtained with less than 5% total capacity loss when thin-film lithium cells were cycled at about 40 C rate and 100% DOD. Some integrated circuits are fabricated using solder reflow (surface mount) assembly in which the IC is heated to 250–260ºC for a short time causing all of

Micropower batteries 305 4.2

Potential (V)

i (mA): 0.1, 0.4, 1, 2, 4 3.8

3.4

3.0 0

40

80 120 Capacity (µAh)

160

Figure 11.17 Discharge at different currents of a lithium cell with a 1 cm2 2.5 m thick crystalline LiCoO2 cathode.

0

Current (mA) 4 6

2

8

10

100 Capacity Energy

Maximum (%)

80 60 40 20 0 0

10

20

30 C-rate

40

50

60

Figure 11.18 Capacity and energy as a percentage of the maximum vs discharge rate from the data in Figure 11.17. The maximum values of 169 Ah and 2.4 J for the capacity and energy respectively were obtained at a discharge current of 10 A.

the components to be soldered at once. Because of the low melting point of lithium, 180ºC, lithium batteries fail when heated to these temperatures, and so we sought to replace the pure metallic anode with a high melting and less reactive inorganic compound. We have investigated several oxynitride25 and nitride26 compounds as possible anode materials. Low current discharge curves of LiCoO2 Li-ion cells with silicon-tin oxynitride (SiTON), tin nitride, and zinc nitride anodes are compared in Figure 11.20 with a discharge curve for a cell with a lithium metal

306

Owens et al. 160 0.1 mA

0.2 mA 120

1 mA 1 mA

80

Capacity (µAh)

1.5 mA 40

60

0.1 mA

0.1 mA 0.6 mA

40

1 mA

20

0.1 mA

0

2000 Cycle

4000

Figure 11.19 Discharge capacity vs cycle number for lithium, lithium-free, and Li-ion cells with LiCoO2 cathodes of different thickness cycled at different currents. (●) Lithium cells and ( ) lithium-free cells cycled between 4.2 and 3 V; Li-ion cells with () Sn3N4 and (▲) Zn3N2 anodes cycled between 4.2 and 2.7 V. The cycle performance of cells with silicon–tin oxynitride anodes is similar to that of the Sn3N4 cells. The dotted horizontal lines drawn through the data for the lithium cells emphasize the negligible capacity loss.

anode. Silicon-tin oxynitride25 with a typical composition SiSn0.87O1.20N1.72 is deposited by sputtering SnSiO3 in N2, while tin and zinc nitride are deposited by sputtering of the base metals in N2. On the initial charge, 40–50% of the lithium is irreversibly trapped in the anodes by the formation Li2O and/or Li3N. Because of the lower discharge potentials and smaller capacities per unit area and thickness of the cathode, the energy delivered by Li-ion cells is lower than that of the lithium batteries. For example, the energy of the Sn3N4 cell calculated from the curves in Figure 11.18 is about 67% of that of the lithium anode cell. However, in optimized Sn3N4 cells in which lithium is plated on the anode during the charge step, the relative energy is increased to 83%.26 More importantly, the Li-ion cells are not affected by heat treatment at 250ºC for times much longer than required

Micropower batteries 307

4.2 Li

Potential (V)

3.9 Zn3N2

3.6 3.3 SiTON 3.0

Sn3N4

2.7 0

20 40 60 Capacity (µAh/cm2 µm )

80

Figure 11.20 Comparison of low current discharge curves of lithium-ion and lithium or lithium-free cells with crystaline LiCoO2 cathodes. Capacity in Ah per unit of area in cm2 and thickness in m of the cathode.

for solder reflow, and they can deliver more than 50% of their maximum capacity and energy at discharge rates up to 5 mA/cm2. The secondary performance of Li-ion cells with Zn3N2 and Sn3N4 anodes26 discharged at 0.1 and 1 mA/cm2, respectively, is illustrated in Figure 11.19. At 0.1 mA, the capacity of the Zn3N2 cell decreased by 27% over 1000 cycles while that of the Sn3N4 cell discharged at 1 mA decreased by 43% over 1000 cycles. Recently, we have fabricated27 cells that exhibit the performance of thin-film lithium batteries, yet can tolerate the solder reflow conditions. In these cells, a metal current collector such as Cu is deposited over the electrolyte in place of the anode. On the initial charge, lithium from the LiCoO2 cathode is plated between the electrolyte and the Cu current collector as we have observed by viewing a fracture cross section in a scanning electron microscope during charge and discharge. Cycling then involves repeated stripping and plating of lithium under the Cu. For cathodes of comparable thickness, the discharge curves of these “lithiumfree” lithium cells are identical to those for lithium anode cells such as those shown in Figure 11.17. Before the initial charge, the “lithium-free” cells can be heated to solder reflow temperatures without degrading subsequent performance. An example of cycle data for a lithium-free cell is shown in Figure 11.19. At a discharge current of 1 mA/cm2, the capacity decreased by 20% over 1000 cycles. The problem remaining to be solved before Li-ion and lithium-free cells can be considered for applications is to reduce the percentage of cells that develop a leak on the initial charge, about 30–50% for both types. By comparison, the yield for cells with lithium anodes is over 90%. 11.5.3

Cells with crystalline LiMn2O4 cathodes

The Li–Mn–O system is significantly more complex than Li–Co–O system as a consequence of the large number of phases and the wide variations in

308

Owens et al.

stoichiometry possible within the spinel structure.30 We have deposited thin films by e-beam evaporation31,32 of LiMn2O4 and by RF magnetron sputtering33,34 of LiMn2O4 in Ar N2 and Ar O2. Post deposition annealing of the nanocrystalline films in O2 at temperatures from 400–900ºC yields well-crystallized films with the spinel structure. The films are manganese deficient-lithium rich with compositions described by Li1xMn2yO4, where the values of x and y depend on the deposition conditions. For example, for films sputter deposited in Ar O2, y ~ 0.3 and 0.2 x 1.2. These cathodes exhibit a significant capacity at 5 V at the expense of the capacity at 4 V. Recently35 we have found that doping the film with Mn effectively increases the 4-V capacity and improves the discharge rate capability. Examples of 4.5 Battery potential (V)

i (mA): 0.04, 0.2, 1, 2, 3 4.0

3.5

3.0

0

5

10 15 Capacity (µAh)

20

Figure 11.21 Discharge at different currents of a Li–LiMn2O4 cell with a 1 cm2 0.5 m thick cathode.

60 2 µm

50 Capacity (µAh)

1 µm 40

0.5 µm

30 20 10 0

0

1

2 3 Current (mA)

4

5

Figure 11.22 Capacity vs current of Li–LiMn2O4 cells with 1 cm2 cathodes of different thickness.

Micropower batteries 309 discharge curves of a Li–LiMn2O4 cell at different rates and graphs of capacity vs discharge current are shown in Figures 11.21 and 11.22, respectively. While the energy delivered by the Li–LiMn2O4 thin-film batteries are not as high as those delivered by Li–LiCoO2 cells, their performance is satisfactory for many applications. 11.5.4

Metal foil substrates

The high specific energy and power densities inherent to thin-film solid-state batteries can be realized only if they can be directly fabricated on a device package therefore requiring no separate substrate. For stand alone batteries, the mass and volume of the battery is essentially determined by the substrate which typically is a 0.1–0.3 mm thick ceramic. Batteries fabricated on metal foils 25 m thick (1 mil) on the other hand can deliver significantly higher energy and power per unit of mass and volume. In Figure 11.23 we compare the specific energy vs specific power and energy density vs power density for a single thin-film Li– LiCoO2 battery with no substrate and double-sided batteries fabricated on 1 mil Ti and 5 mil alumina substrates. Titanium was chosen because it is a light metal that can withstand the 700ºC anneal of the cathode. The calculated values were based on the data in Figure 11.17 using the masses and volume for each layer of a parylene coated 1 cm2 cell and the masses and volumes of 1 cm2 substrates. We have fabricated Li/LiCoO2 batteries on 1 mil Ti and stainless foils. The batteries were flexible and could be rolled into 1 cm diameter cylinders without affecting performance. The yield of batteries on metal foils was much less than those on ceramic substrates. About 70% of the cells were shorted as deposited or shorted on the initial charge. We believe the poor yield was caused by flaws in the metal foils that either penetrated the films or resulted in pin holes after deposition of the electrolyte. 11.5.5

Applications and scale-up

The performance of thin-film solid-state lithium and Li-ion batteries makes them attractive for application in many consumer and medical products. Manufacturing scale-up is underway at several US companies, and at presently estimated production costs, the products targeted first for commercial application include implantable medical devices, CMOS-based integrated circuits, and RF identification tags. The batteries will be incorporated as separate components, or they will be integrated into devices by direct fabrication on IC packages or chips. We show in Figure 11.24 photographs of a stand-alone prototype “D”-shaped battery, a battery fabricated on a multichip module package, and a cell with a 1 cm2 active area we use for our experiments. For high-power applications, it will be necessary to connect many thin-film cells in parallel. Recently, in collaboration with Dr Leo Kwak formerly of Teledyne Electronic Technologies, scaling of battery performance with increasing active area was demonstrated by connecting ten 7.5 cm2 “D”-shaped Li–LiCoO2 cells in parallel giving a battery with a total active area of

Owens et al.

Specific energy (Wh/kg)

310

100

10 cell only 2 cells on 1 mil Ti 2 cells on 5 mil alumina

1 100

101

102 103 104 Specific power (W/kg)

105

Energy density (Wh/l)

103

100 cell only 2 cells on 1 mil Ti 2 cells on 5 mil alumina 10 100

101

102 103 104 Power density (W/I)

105

Figure 11.23 Ragone plots of the Li–LiCoO2 cell Figure 11.17. The mass and volume of the cell calculated for 1 cm2 areas of the current collector, cathode, electrolyte, anode, and protective coating. The substrate areas were taken as 1 cm2.

75 cm2. The battery was required to supply on a single charge a minimum of five pulses of 2.5 mA/cm2 amplitude, 8.5 s duration, followed by a 2 s rest with a minimum of 60 mJ/cm2 of energy per pulse. For 75 cm2, these requirements translate to a pulse amplitude of 187.5 mA and an energy of 4.5 J per pulse. As shown in Figure 11.25, the battery supplied twelve pulses of the required amplitude and energy on a single charge. Single laboratory cells with 1 cm2 active areas typically could supply 10–15 pulses of the required current and energy densities. There are many other possible applications of thin-film lithium and Li-ion batteries in consumer products such as cellular telephones and notebook computers. The types of applications will be limited only by the cost of production. Based on the deposition rates we presently use in our small research chambers, it could be possible to produce 40 000 cm2 of battery per 24 h period with a single 1 m2 in-line deposition system. With a 3-m thick crystalline LiCoO2 cathode, this is equivalent to a production of about 34 Wh per day.

Micropower batteries 311

(a)

(b)

(c)

(d)

Figure 11.24 Photograph of Li–LiCoO2 cells. (a) “D” cell for an implantable defibrillator; (b) battery fabricated on the backside of a multichip module shown in (c); (d) laboratory cell with a 1 cm2 active area.

200

10 × 7.5 cm2 LiCoO2

6.0

4.0 100

3.0 2.0

50

Energy (J)

Current (mA)

5.0 150

1.0 0 0

50

100

150

0.0

Time (s)

Figure 11.25 Current pulses and energy per pulse for a battery consisting of 10 D-cells (Figure 11.24) connected in parallel.

For many applications, the question often arises about the time required to recharge thin-film batteries. We show in Figure 11.26 the current vs time for charging of a 1 cm2 Li–LiCoO2 cell at initial rates of 1 and 0.2 mA and the percentage charge vs time for the two rates. This cell, which had a resistance of 280 , had been discharged at 1 mA from 4.2 V to a cutoff potential of 3 V on

312

Owens et al.

the previous half-cycle delivering a capacity of 107 Ah in about 6.5 min. On charging Li–LiCoO2 cells, it is important for the potential not to exceed 4.2 V in order to prevent degradation of the cathode and a consequent decrease in cycle life. Because of cell resistance, the 4.2 V cutoff can be reached rapidly on charging at high currents. For example, with an initial charging current of 1 mA, the cutoff potential of the cell in Figure 11.26 was reached within about 70 ms where the current decreased to about 0.55 mA and continued to decrease with further charging. After 30 min 94% of the capacity was recovered while about 75 min of charging was needed to recover 100%. At a charging current of 0.2 mA, about 82% of the discharge capacity was recovered after 30 min when the 4.2 V cutoff was reached. For cells with lower resistances, the initial charge current can exceed 1 mA such that 95% of the discharge capacity can be recovered in 10–20 min. The time to recover most or all of the discharge capacity should be considered in applications requiring a rapid charge. In order to minimize charge time, the battery area should be as large as possible.

Charge current (mA)

(a) 0.6

1 mA charge 0.2 mA charge

0.4 0.2 0 0

(b)

10

20 30 40 50 60 Time (min)

70 80

Charge (%)

100 80

1 mA charge

60

0.2 mA charge

40

Q = 107 µAh

20 0

0

10

20 30 40

50 60

70 80

Time (min)

Figure 11.26 (a) Charge current vs time after discharge of a cell at 1 mA to 3 V. The arrow indicates that the initial current of 1 mA decreased to 0.55 mA after about 70 ms when the potential reached 4.2 V. This potential was reached at about 30 min on charging at 0.2 mA. (b) Percent charge recovered on charging at the two rates. The discharge capacity was 107 Ah.

Micropower batteries 313

11.6

High energy amorphous metal oxides W. H. Smyrl and B. B. Owens

When a battery is scaled down to a miniature size, the energy density of the battery may be drastically reduced because of the requirement to still include a reliable package design in the final product. With larger batteries the non-active elements of the package do not dominate the mass and the volume of the product. Consequently the energy density may approach 25–50% of the theoretical value based on the electroactive materials. However, if one reduces the scale of the cell design to the micron level, and the current collectors and external package still are on the millimeter level, then the energy density is relatively independent of the intrinsic energy density of the cell couple, but is determined by other design factors. The investigations of high capacity cathode materials at the University of Minnesota are beginning to show promise for practical batteries based on lithium and Li-ion systems. Sol–gel processing of intercalation oxide materials like V2O5 and LixMnO2 is easily carried out under ambient conditions. In this way, one forms a solid network of the oxide that is bicontinuous with an interpenetrating liquid (usually H2O) that fills the pores. The connected oxide network forms spontaneously in the process and the network becomes the pathway for electrons and intercalated guest cations when used as an electrode material. The sol–gel process affords real flexibility in forming electrode materials and structures with controlled oxidation state, structure, morphology, and porosity. One may obtain the intercalation electrode material in the form of powder, films, and freestanding membranes. We have utilized sol–gel procedures to synthesize and process a family of V2O5 and MnO2 materials that have high intercalation capacity and that are exceptionally reversible. V2O5 is very amenable to sol–gel processing by several different routes.36 The normal alkoxide route, however, yields a product with a low oxidation state because the oxide reacts with product alcohol as well. V2O5 gels and films are routinely made industrially by pouring the molten oxide into water. The resulting material is used as an anti-static coating for photographic film and for magnetic tapes. The third process that we use at the University of Minnesota is an ionexchange process that yields a precursor HVO3 solution that self-assembles into the V2O5 hydrogel which is further processed into powders, films, and free-standing membranes. Most of the work, however, has used the gel-based precursor to make materials of different porosity. Xerogel V2O5 is formed by directly drying the hydrogel under vacuum, and the product is compact with low porosity. Spincoated thin films of V2O5 xerogel have been found to reversibly intercalate four Li ions per V2O5 as shown in Figure 11.27. These film electrodes have been cycled for several 1000 insertion/release cycles of Li ions 37 and have been found to be completely reversible in charge, mass, stress, and local atomic structure. The spin-coated films (50–500 nm thick) could be used directly for microbatteries as described in an earlier section, but our attention has been directed to systems of a larger scale.

314

Owens et al.

Highly porous forms of V2O5 are formed from the hydrogel by first replacing the water which fills the pores with acetone. Acetone is then replaced with liquid CO2 and the CO2 is removed under supercritical conditions to form an aerogel with a surface area that approaches 500 m2/g. Aerogel V2O5 has reversible intercalation capacity for Li ions that exceeds that of V2O5 xerogel (i.e. four Li per mole of V2O5).38 In fact, as shown in Figure 11.27, the aerogel material may intercalate up to six Li ions, and at a higher potential than that of the xerogel material. A second type of highly porous material is formed from the acetone-gel of V2O5 by replacing the acetone with hexane and drying at ~80ºC under vacuum. Since the pore liquid (hexane) is removed under low surface tension conditions, the highly porous solid network is retained on drying, and the surface area of the solid is about 200 m2/g. The product V2O5 (aerogel-like) also has reversible intercalation capacity of four Li ions per mole of V2O5.39 The excellent intercalation capacity and reversibility of the material have led us to make composites of the oxides with conductive carbon and a binder in order to test their scale-up and rate capabilities in coin cell configurations. The composites have been fabricated into pellets, cast sheets, and spray-coated films from slurries. The pellets are typically 200– 400 m thick, the cast sheets are 150– 400 m, and the spray-coated films are 20–200 m. The latter films are formed by spray coating a film onto a suitable metal foil such as aluminum. Coin cells were constructed with 2016 hardware using composite electrodes from above, with lithium foil as the anode, and 1 M LiClO4/PC/EC as the electrolyte. Three electrode cells were assembled with Li foils as both counter and reference electrodes. Specific capacity and specific energy are 500 Ah/kg (active material) and 1200 Wh/kg (active material), respectively, for the xerogel. For aerogel and aerogel-like V2O5, the specific capacity and specific energies are 500 Ah/kg (active material) and 1600 Wh/kg, respectively, for both materials

Equilibrium potential (V vs Li)

4.0 Electrochemical insertion Chemical insertion 3.5 Aerogel

3.0 2.5

Spin-coated Xerogel

2.0 Amorphous 1.5 0.0

1.0

2.0 3.0 x (Lix V2 O5)

4.0

5.0

6.0

Figure 11.27 Equilibrium potential vs Li content in V2O5 aerogel, xerogel, and amorphous V2O5.

Micropower batteries 315 (Figure 11.28). The reasons for the larger specific energy for the highly porous materials are beyond the scope of the present discussion. The rate of insertion and the associated intercalation capacity have been determined as well. The specific capacity and specific energy quoted above are for near-equilibrium or very slow insertion/release. As the rate is increased, the capacity decreases to 380 Ah/kg (at C/20), 320 Ah/kg (at C/5), and 200 Ah/kg (at C/1), respectively.40 The stoichiometric rate is calculated for one Li per V2O5. The difference in capacity at different rates is significant, illustrating that kinetic and diffusion limitations are important factors that control the performance of the xerogel material. The porous aerogel-like material has a better rate response and shows very high capacity,39 approaching 400 Ah/kg for a C/10 rate. The capacity decreased as the rate of insertion increased, that is, 390 Ah/kg (at C/5), 350 Ah/kg (at C/2.5), and 300 Ah/kg (at C/1). However, the full capacity of the porous materials is not utilized in the composite electrode configurations. In other work in which aerogel nanofibers were grown on a porous Ni current collector, the rate response was found to increase substantially. The configuration shows that the V2O5 behaves like a “redox capacitor” in which the measured capacitance approaches 20 000 F/g.41 This value is two orders of magnitude larger than any other system reported previously. The specific energy approaches 800 Wh/kg (active material) and the specific power is as large as 6 MW/kg (active material). In addition, V2O5 may be doped and chemically modified by prior reaction in the precursor hydrogel stage. Reaction of hydrogel V2O5 with metallic Cu, Ag, Al, Ni, Mg, and Zn has been found to enhance the electronic conductivity (i.e. doping) 2000

Specific energy (Wh/kg)

500 Ah/kg

V2O5 (ARG)

V2O5 (XRG)

1000

V6O13 LiMn2O4 LiCoO2 LiNiO2 TiS2

0 Specific capacity (width of bar)

Figure 11.28 Specific energy and capacity of selected cathode materials, calculated on the basis of the mass of the cathode active material.

316

Owens et al.

by up to three orders of magnitude. Subsequent processing to the aerogel state gives material, which supports facile intercalation that is highly reversible. For example, Cu0.1V2O5 aerogel-like samples were found to have capacities of 170 Ah/kg (active material) at a rate of C/1 for more than 400 cycles.42 Cu0.1V2O5 xerogel samples had similar capacity–cycle life behavior. In addition, the latter materials showed very good performance in pulsed insertion tests. When pulses of about 1 mJ in 1 s were imposed, more than 40 000 pulses could be sustained in one discharge (insertion) cycle. The manganese oxide-based aerogels42 also perform well as lithium insertion hosts in terms of insertion capacity (400 Wh/kg), but not as well in terms of rate performance. On the other hand, manganese-based materials offer advantages in terms of cost, safety, and low environmental impact. The methods to improve the kinetics of gel-based vanadium oxide positive electrodes that have been found to be effective will now be applied to manganese oxide-based materials to improve their rate performance, and the results will be reported later. Because of their electrochemical properties, these amorphous metal oxide materials may function as high capacity (500–600 mAh/g), high-energy positive electrodes for lithium or Li-ion batteries. The synthesis procedures are simple, inexpensive, and reproducible. The techniques are general and can be applied to other intercalation compounds synthesized via sol–gel routes.

11.7

Micropower fuel cells W. H. Smyrl

In many applications for micropower, there is direct competition with miniature batteries. We have recently reported results for a miniature methanol/air fuel cell based on Si technology.44 Although the concept is not new, and other research groups are interested in related devices, there are no previous studies of miniature fuel cells in the current literature. The cells for our studies were fabricated from silicon wafers using processing developed for microelectronics and MEMS technology. Fuel flow-fields, electrode structures, and assembly techniques were used to fabricate individual electrodes on a chip. Two such chips were sandwiched on either side of a NAFION membrane to form single cells. The cross-section area was 0.25 cm2, but the design could be used for further miniaturization by three orders of magnitude. When tested, the cells achieved performance that was within a factor of two of a state-of-the-art large fuel cell. In particular, at 23, 50 and 70ºC, the specific energies are 23.6, 132, and 189 Wh/kg, respectively, when based on the mass of the device and the mass of methanol consumed in 1 h. The analogous energy densities (Wh/l) are 61.2, 361, 535 at 23, 50 and 70ºC, respectively. These performances compare favorably with the NEMI roadmap goals for miniature fuel cells of 180 Wh/l and 300 Wh/kg for a 10 W/70 Wh system. The present cells deliver about 1 mW at 23ºC and 6 mW at 70ºC. The present design successfully miniaturizes a methanol/air fuel cell. A second example is provided in the chapter by Bostaph et al.45 It is a thin-film cell based

Micropower batteries 317 on earlier work by one of the authors. Alternative designs are being developed to fabricate the cells on an interdigitated electrode array, for example, with a polymer electrolyte covering the array. There is considerable activity in this area and new designs are being developed for proprietary positions.

11.8

Conclusions

Many types of power sources are available to provide micropower. These power sources are not limited to microbatteries, as there are many applications in which the level of power required is in the microwatts region, but the device also requires unattended operation over an extended period of time, even many years of maintenance-free performance. In this case, a primary battery with a capacity of 0.5–2 Ah will be the power source of choice, as was described for the cardiac pacemaker and the hazardous gas personal monitor. Utility gas meters incorporating electronic circuit design and remote interrogation capability are also powered by lithium primary cells with operating life times in the 5–10-year time frame. Lithium primary cells perform very well for this type of application. However, these batteries must not only provide a high-energy density, but the selfdischarge rate must be low. The latter is implicit in the specified value of energy density, provided the energy density is for the 5–10-year rate of discharge. With the continuing reduction in scale of electronic devices, there has been a concurrent reduction in the size of the device. In this case, the use of a relatively large primary battery is not practical as the battery size would dominate the package. Miniature rechargeable batteries and microbatteries have been developed for such applications. A rechargeable Li-ion hearing aid button cell has been tested up to 3000 cycles and could provide operation over a 5-year period. Thin-film microbatteries with cell profiles measured in microns are moving from engineering to the pilot-production stage. This will provide a true microbattery for new applications. Microfuel cells that offer an interesting alternative to the batteries are also described.

Acknowledgments The authors gratefully acknowledge financial support from (a) the US Army Research Office (DAA-H04-93-R-BAA10); (b) the Department of Energy for (i) Contract No. DE-FG02-93ER14384, to the University of Minnesota, (ii) Contract No. DE-AC05-96OR22464, to Oak Ridge National Laboratory; (c) Wilson Greatbatch Ltd., (d) Grant No. 5 R44 AG12711-03 from the National Institute on Aging, to Research International, Inc., Woodinville WA, USA. Figures 11.7, 11.9, 11.10, and 11.11 are reprinted from the Journal of Power Sources, 89(1), S. Passerini, B. B. Owens, and F. Coustier, “Lithium-Ion Batteries for Hearing Aid Applications: I. Design & Performance”, 29–39 (2000) and 90(2) S. Passerini, F. Coustier and B. B. Owens, “Lithium-Ion Batteries for Hearing Aid Applications: II. Pulse Discharge and Safety Tests”, 142–152 (2000), with permission from Elsevier Science.

318

Owens et al.

The contents of this review article are solely the responsibility of the authors and do not necessarily represent the official views of the funding agencies.

References 1. G. Antonioli, F. Baggioni, F. Consiglio, G. Grassi, R. LeBrun, and F. Zanardi, Stimulatore cardiaco impiantabile con nuova battaria a stato solido al litio. Minerva Med. 64, 2298 (1973). 2. J. R. Moser, Solid state lithium iodine primary battery, US Patent 3,660,163 (1972). 3. Schneider and J. R. Moser, Primary cells and iodine-containing cathodes therefor, US Patent 3,674,562 (1972). 4. F. Gutmann, A. Herman, and A. Rembaum, Solid-state electrochemical cells based on charge transfer complexes, J. Electrochem. Soc. 114, 323 (1967). 5. W. Greatbatch and C. F. Holmes, The Lithium/Iodine battery – an historical perspective, Pacing and Clinical Electrophysiology, 15, 2034 (1992). 6. Holmes, R. A. Leising, D. M. Spillmar, and E. S. Takeuchi, “Batteries for Biomedical Implantable Devices”, 12th IBA Grenoble-Annecy Battery Materials Symposium”, 20 September 1998. 7. M. Phillips and D. F. Untereker, Phase diagram for the poly-2-vinylpyridine and iodine system, in: B. B. Owens and N. Margalit (eds), Proceedings of the Symposium on Power Sources for Biomedical Implantable Applications and Ambient Temperature Lithium Batteries, The Electrochemical Society, Princeton, NJ, pp. 195–206 (1980). 8. K. R. Brennen and D. F. Untereker, Iodine utilization in Li/I2-poly-2-vinylpyridine batteries, in: B. B. Owens and N. Margalit (eds), Proceedings of the Symposium on Power Sources for Biomedical Implantable Applications and Ambient Temperature Lithium Batteries, The Electrochemical Society, Princeton, NJ, pp. 161–173 (1980). 9. W. Greatbatch, R. T. Mead, and F. Rudolph, Lithium iodine battery having coated anode, US Patent No. 3,957,533 (1976). 10. W. R. Brown, W. R. Fairchild, H. A. Hornung, and C. F. Holmes, The effects of anode precoating on the structural and electrical properties of the solid electrolyte formed in lithium/iodine-polyvinylpyridine batteries, in: Extended Abstracts of the Fall 1984 Electrochemical Society Meeting, The Electrochemical Society, Princeton, NY, p. 257 (1977). 11. B. Phipps, T. G. Hays, P. M. Skarstad, and D. F. Untereker, Lithium/Iodine batteries with poly-2-vinylpyridine coated anodes: a microstructural investigation, Ibid., p. 158. 12. R. Selim and P. Bro, Performance domain analysis of primary batteries, J. Electrochem. Soc., 118, 829 (1971). 13. W. Greatbatch, C. F. Holmes, E. S. Takeuchi, and S. J. Ebel, Lithium/carbon monofluoride: A new pacemaker battery, Pacing and Clinical Electrophysiology 19, 1836 (1996). 14. Research International, “Micropower Solid State Hazard Card”, Final Technical Report, DAARPA Contract No. DAAH01-95-C-R168, 27 March (1998). 15. S. Passerini, B. B. Owens, and F. Coustier Journal of Power Sources, 89(1) 29–39 (2000). 16. S. Passerini, F. Coustier, and B. B. Owens, Journal of Power Sources, 90(2) 142–152 (2000). 17. “Lithium-Ion Button Cells for Low Rate and Pulse Applications”, B. B. Owens, S. Passerini and E. Saaski, Proceedings of the 39th Power Sources Symposium, Cherry Hill, NJ, p. 240 (2000).

Micropower batteries 319 18. E. Saaski, B. B. Owens, and S. Passerini, Rechargeable Battery, US Patent No. 6,265,100 (2001). 19. J. B. Bates, N. J. Dudney, D. C. Lubben, G. R. Gruzalski, B. S. Kwak, X. Yu, and R. A. Zuhr, J. Power Sources 54, 58 (1995). 20. N. J. Dudney, J. B. Bates, and D. Lubben, in Role of Ceramics in Advanced Electrochemical Systems, American Ceramic Society, Westerville, Ohio, p. 113 (1996). 21. B. Wang, J. B. Bates, F. X. Hart, B. C. Sales, R. A. Zuhr, and J. D. Robertson, J. Electrochem. Soc. 143, 3203 (1996). 22. J. B. Bates, N. J. Dudney, B. J. Neudecker, and B. Wang, “Thin-Film Lithium Batteries,” in New Trends in Electrochemical Technology: Energy Storage Systems in Electronics, T. Osaka and M. Datta (eds), Gordon and Breach, Amsterdam (2000). 23. J. B. Bates, N. J. Dudney, G. R. Gruzalski, R. A. Zuhr, A. Choudhury, and C. F. Luck, J. Power Sources 43–44, 103 (1993). 24. X. Yu, J. B. Bates, G. E. Jellison, and B. C. Sales, J. Electrochem. Soc. 144, 524 (1997). 25. B. J. Neudecker, R. A. Zuhr, and J. B. Bates, Journal of Power Sources 81–82, 27 (1999). 26. B. J. Neudecker, R. A. Zuhr, and J. B. Bates, Meeting abstracts for 195th Meeting of the Electrochemical Society, Seattle, WA, May 2–6, 1999. Meeting abstracts for the 196th Meeting of the Electrochemical Society, Honolulu, Hawaii, 17–22 October (1999). 27. B. J. Neudecker, N. J. Dudney, and J. B. Bates, Journal of the Electrochemical Society 147, 517 (2000). 28. J. B. Bates, N. J. Dudney, B. Neudecker, A. Ueda, and C. D. Evans, “Thin Film Lithium and Lithium-Ion Batteries”, Proceedings of the 12th International Conference on Solid State Ionics, Thessaloniki, Greece, 6–12 June (1999). 29. J. B. Bates, N. J. Dudney, B. J. Neudecker, F. X. Hart, H. P. Jun, and S. A. Hackney, “Preferred Orientation of Polycrystalline LiCoO2 Films,” Journal of the Electrochemical Society 147, 59 (2000). 30. M. M. Thackeray, A. de Kock, M. H. Rossouw, D. Liles, R. Bittihn, and D. Hoge, J. Electrochem. Soc. 139, 363 (1993). 31. J. B. Bates, D. Lubben, N. J. Dudney, and F. X. Hart, J. Electrochem. Soc. 142, L149 (1995). 32. J. B. Bates, D. Lubben, N. J. Dudney, R. A. Zuhr, and F. X. Hart, in: Thin Film Solid Ionic Devices and Materials, J. B. Bates (ed.), The Electrochemical Society, Pennington, New Jersey, p. 215 (1996). 33. A. Ueda, J. B. Bates, and R. A. Zuhr, in Lithium Batteries Electrochemical Society Proceeding, Boston, MA, P.V. 98–16, p. 279 (1999). 34. N. J. Dudney, J. B. Bates, R. A. Zuhr, S. Young, J. D. Robertson, H. P. Jun, and S. A. Hackney, J. Electrochem. Soc. 146, 2455 (1999). 35. N. J. Dudney, private communication (1999). 36. J. Livage, Chem. Mater 3, 578 (1991). 37. H.-K. Park, Ph.D. Thesis, University of Minnesota, 1993; H.-K. Park and W. H. Smyrl, J. Electrochem. Soc. 141, L25 (1994). 38. D. B. Le, S. Passerini, J. Guo, J. Ressler, B. B. Owens, and W. H. Smyrl, J. Electro chem. Soc. 143, 2099 (1996). 39. F. Coustier, S. Passerini, and W. H. Smyrl, J. Electrochem. Soc. 145, L73 (1998). 40. L. Manhart, J. Xu, F. Coustier, S. Passerini, B. B. Owens, and W. H. Smyrl, Proceedings Solid State Ionics Symposium, MRS, Fall 1998.

320

Owens et al.

41. M. J. Parent, S. Passerini, B. B. Owens, and W. H. Smyrl, J. Electrochemical Soc. 146, 1346 (1999); S. Passerini, J. J. Ressler, D. B. Le, B. B. Owens, and W. H. Smyrl, Electrochim. Acta 44, 2209 (1999). 42. F. Coustier, J. Hill, B. B. Owens, S. Passerini, and W. H. Smyrl, J. Electrochem. Soc. 146, 1355 (1999). 43. S. Passerini, F. Coustier, M. Giorgetti, and W. H. Smyrl, Electrochem. Sol. State Lett. 2, 483 (1999); J. Xu, A. J. Kinser, B. B. Owens, and W. H. Smyrl, Electrochem. Sol. State Lett. 1, 1 (1998). 44. S. Kelley, G. Deluga, and W. H. Smyrl, Electrochem. Sol. State Lett. 3(9), (2000). 45. J. Bostaph, C. K. Dyer, S. P. Rogers, D. Gervasio, A. M. Fisher, S. Tasic, and R. G. Lawrence, Proceedings 39th Power Sources Conference, Cherry Hill, NJ, p. 152 (2000).

12 Micro flow systems for chemical and biochemical applications Shuichi Shoji

12.1

Introduction

Micromachining, based on photolithography commonly used in the integrated circuit, has been applied to miniaturize chemical and biochemical total analysis systems (TAS). Micro/miniaturized total analysis systems (TAS) realize very small sample volume, fast response and reduction of reagents which is very useful in biochemical analysis. Reduction of waste is also an advantage when toxic reagents are required. The majority of TAS developed so far are flow type analysis systems in which the analyses are completed in continuous flow, for example, a flow injection analysis (FIA) and a gas chromatography (GC). The flow type TAS are classified into two groups. One is a MEMS (Micro Electro Mechanical System) type system including mechanical flow control devices of microvalves and micropumps.1,2 The other uses electrically driven liquid handling without mechanical elements. Typically, electroosmotic flow has been used so far. Planer capillary electrophoresis (chip CE) type systems are the common application of this method.3 Micromachining technologies are diversified to fabricate chemical and biochemical micro reactors as well as TAS.4 Examples of micro flow systems and their components for chemical and biochemical applications are described.

12.2 12.2.1

MEMS type micro flow systems System considerations

An example of MEMS type micro flow systems is a conventional FIA system that consists of mechanical micro flow control devices of micropumps, an injection valve, a mixer/reactor, a separator and a detector (a chemical sensor) (Figure 12.1). In case of MEMS type TAS, a hybrid system in which each element is fabricated on the different chips and integrated on a board having microchannels connecting them is realistic.1 Micro O-rings are used for the connecting ports. In sophisticated systems, the electronic circuits for controlling the fluidic devices and detecting signals from the sensors are also located on the board. Chemical and biochemical micro reactors using MEMS elements have also been developed. Since the Reynolds number of liquid flow in the micro channel

Carrier reservoir

Carrier pump

Sample Injector Mixer/Reactor

Sample inlet Sample outlet

Filter Sample Pump

Separator

Waste Detector Optical Electrochemical

Figure 12.1 Schematic of the MEMS type micro total analysis system.

Cleaning solution (Ringer solution) reservior

Calibration solution reservior

Injection pumps

Suction pump

3-way valve Highest sample level Micro sensors 3-way valve

Waste

Partly disposable detector flow cell

Sample inlet

Figure 12.2 Schematic of partly disposable micro total analysis system.8

Micro flow systems 323 is usually below 200, the flow is laminar. Special design concepts are necessary for the fluidic elements of mixers and reactors to reduce the reaction time.5,6 Disposable TAS will be ideal in practical use. However, high fabrication cost of sophisticated TAS including micropumps and microvalves is an actual problem. One of a simple setup of TAS considering this problem is illustrated in Figure 12.2.2 A detector cell consists of micro sensors and a 3-way microvalve is placed at the sample inlet. The sample flow and cleaning solution flow are controlled by a suction pump and an injection pump connected to the detector cell. The calibration solution flow is also controlled by an individual pump and a 3-way valve. The sample after measurement and the calibration solution and cleaning solution flows into the sensor part are wasted through the 3-way microvalve on the cell. In this system, the sample flow comes up only to the detector cell. The upper parts of the system are free from contamination so that it can be reused many times, while the detector cell has to be disposed. A continuous flow micro reactor was fabricated to achieve the polymerase chain reaction (PCR) which amplifies DNA in abundance.7 In this system, DNA samples are introduced into a winding channel having three thermostated temperature zones (60ºC, 77ºC, 95ºC) fabricated on a glass substrate. A biochemical micro flow cell realizing fast mixing and temperature controlled reaction was also fabricated.4 12.2.2

Micro flow devices

Various types of micro flow devices of microvalves and micropumps have been developed in MEMS. Since the structure of the microvalve has to be designed considering the application, many types of microvalves have been developed using various types of micro actuators: electrostatic, magnetic, piezoelectric, thermopneumatic, etc.2 Micropumps developed so far, are classified into two groups: mechanical and non-mechanical. Mechanical micropumps are the main stream of MEMS type flow systems. Electrohidrodynamic pumps, electroosmotic pumps, ultrasonic pumps and thermocapillary pumps are non-mechanical micropumps. The electroosmotic pump is widely used in chip CE systems, described in the next section. Typical examples of microvalves and micropumps considering practical application are described as follows. 12.2.2.1

Partly disposable microvalves

A separable channel type microvalve whose channel part is disposable while actuator part is reusable was developed.8 Mechanically fixed stack structures including disposable parts will be practical in chemical and biochemical TAS. The structure of the 3-way microvalve is illustrated in Figure 12.3. The pneumatic actuator was chosen because of its large displacement and relatively high available pressure. The flow channel and the pneumatic chambers are made between the polymer membrane and the silicon substrate using the sacrificial process. One inlet and four outlet injection and switching microvalves were also fabricated with similar microstructures.9

8.5 mm

(a)

4.2 mm

Flow channel (b) Port 1

Port 2

Port 3

Flow channel (Depth: 30 µm) Channel part (Si) Polymer membrane (Silicone rubber) Actuator part (Si) Pneumatic chamber

(c)

Pneumatic chamber

Figure 12.3 Structure of the pneumatic 3-way micro valve.8 (a) Bottom view of the channel part; (b) cross-sectional view; (c) top view of the actuator part.

Piezoelectric disk

Pressure chamber

Check valves

Inlet

Outlet

Figure 12.4 Structure of the self-priming and bubble-tolerant micropump.11

Micro flow systems 325 Table 12.1 Structures and characteristics of the micropumps using a piezoelectric disk actuator.10–12 Actuator

Check valve

Maximum flow range

Maximum output pressure

2.3 ml (min) (1 kHz)

7.6 mH2O

1 ml (min) (120 V, 220 Hz)

9.0 mH2O

400 µl (min) (70 Hz) 3.5 ml (min) (Air)

21 mH2O

Diffuser/Nozzle

Piezoelectric disk

Cantilever (Bulk Si)

Membrane (Polycarbonate)

12.2.2.2

500 hPa (Air)

Diaphragm type micropumps

The majority of the micropumps developed so far have been diaphragm micropumps. The diaphragm micropump consists of a pressure chamber driven by an actuator and inlet, and outlet check valves. In many TAS, high output pressure and wide controllable flow range are required for micropumps. High performance micropumps using a piezoelectric disk as an actuator and nozzle/diffusers instead of the check valves were developed.10 Two types of self-priming and bubbletolerant micropumps were recently developed by minimizing the volume of the pressure chamber.11,12 Schematic of these micropumps is shown in Figure 12.4. These micropumps can be used for gas flow control as well as for liquid flow control. The characteristics of the micropumps are listed in Table 12.1.

12.3

Chip CE type micro flow systems

12.3.1 System consideration The chip CE system has a very simple structure and has no mechanical flow control devices. The electrophoresis system has only reservoirs for carrier, sample and regent, and connecting microchannels typically fabricated on the glass substrate as shown in Figure 12.5. Since this system uses electrokinetic flow of electroosmotic flow, the flow in the microchannels is controlled by just applying electric voltage between the inlet and outlet of the channels. In these systems, external sample preparation chambers, a mechanical sample induced system is necessary. Optical detection of fluorescence or absorption is employed in most of this type of systems.3 The DNA analysis is one of the main target of this system. To achieve high-throughput DNA analysis, multiplexed channel systems of maximum 96 samples and 96 channels have been fabricated on glass substrates.13 In the DNA analysis systems, micromachining has been applied to integrate

326

S. Shoji

Photomultiplier Confocal pinhole Focusing lens Band pass filter 488 nm laser

Dichronic beam splitter

Sample injector Samples

Carrier supply

Voltage supply

Microfabricated chip

Figure 12.5 Schematic of the chip capillary electrophoresis system.

micro chambers which realize PCR or efficient biochemical reaction.14 Much sophisticated systems are demonstrated to realize the sample preparation required in the immunoassay.15 12.3.2

Fabrication of the capillary

Since fabrication of microchannels are important in any micro flow systems, various micromachining technologies have been applied. The basic method is to bond two glass wafers which have etched through-holes for inlet and outlet, and etched micro grooves for microchannels. Isotropic wet etching is used for fabricating the structures. The bonding is performed at a very high temperature of 1100ºC when the quartz (fused silica) substrates are used as the substrate. A new method of low temperature (80ºC) bonding using diluted HF which can apply SiO2–SiO2 bonding was developed.16 The bond temperature can be reduced (500 ~ 600ºC) when the Pyrex glass is used. To be competitive on market products, it is cheaper the fabrication cost the better. Thermoplastic replication methods of injection molding and hot embossing are applied to fabricate microchannels and micro flow control devices. Micromachining, for example, thick photoresist patterning (IBM SU-8) and deep reactive ion etching (DRIE) methods are used to fabricate the fine mold inserts.17 Chip CE systems consist of thermoplastic base plate of polymethylmethacrylate (PMMA) and elastomeric polymethylsiloxane (PDMS) cover are fabricated.18 Higher aspect ratio microchannels can be fabricated

Micro flow systems 327 using metal replica formed by X-ray lithography and electroplating, the so called LIGA process.19 12.3.3

Detection methods

Optical detection of fluorescence or absorption has been commonly used in DNA analysis using chip CE systems. In this case, a one-to-one imaging detector having large optical elements of excitation laser source, a dicromic beam spritter, focusing and objective lenses, etc. are indispensable. To realize the parallel detection, a confocal laser scanning detection is utilized. On the other hand, new fluorescence detection methods for multiplexed CE array are developed. Excitation light beam matrix is formed using micro lens array which is focused on the appropriate points of the channel array. The fluorescence from each point is detected by an arrayed photo-detector or CCD camera.20,21 Recently, mass spectrometry has been used for the analysis of lager molecules of proteome, because of its high sensitivity and high resolution.22,23 Micromachining is currently applied to fabricate the electrospray used in the sample inlet.24

12.4

Conclusions

Micro flow systems described here have many advantages compared to the conventional systems. TAS has promising fields on chemical and biochemical applications. However, high resolution analysis is sometimes difficult because of the small sample volume. High sensitive detectors/sensors are required for high performance TAS. Precise flow control of the order of l/min is also required in MEMS type micro flow systems. Micropumps and microvalves are currently developed for this purpose. High output pressure micropumps are key components in micro flow systems. Chip CE systems are very simple and easy to fabricate. Many research works have been continued on chip CE systems. Useful TAS will be realistic by combining a MEMS type sample preparation system and a chip CE analysis system.

Acknowledgement I would like to thank Mr. H. Nakanishi and Mr. T. Nishimoto in Shimadzu Co. and Mr. E. Shinohara in Olympus Optical Co. for their technical support. I also thank Mr. T. Ohori, Mr. K. Miura, Mr. A. Yotsumoto and T. Saitoh for their contributions.

References 1. A. van den Berg and T. S. J. Lammerink, Topics in Current Chemistry, 194 (Springer, Berlin, 1998), 22–49 2. S. Shoji, Topics in Current Chemistry, 194 (Springer, Berlin, 1998), 163–188 3. C. S. Effenhauser, Topics in Current Chemistry, 194 (Springer, Berlin, 1998), 51–82

328

S. Shoji

4. A. Yotsumoto, R. Nakamura, S. Shoji and T. Wada, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 185–8 5. J. Branebjerg, B. Fabius and P. Gravesen, Micro Total Analysis Systems’94 (Kluwer Academic Pub., Dordrechit, 1994), pp. 141–151 6. R. Miyake, T. S. Klammerink, M. Elwenspooek and J. H. J. Fluitman, Proc. Micro Electro Mechanical Systems; MEMS-93, (Trabemunde, 1993), pp. 248–253 7. M. U. Kopp, M. B. Luechinger and A. Manz, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 7–10 8. T. Ohori, S. Shoji, K. Miura and A. Yotsumoto, Sensors & Actuators, A64 (1998), 57–62 9. K. Miura and S. Shoji, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 85–88 10. A. Olsson, O. Larsson, J. Holm, L. Lundbladh, O. Ohman and G. Stemme, Proc. Micro Electro Mechanical Systems; MEMS-97, (Nagoya, 1997), pp. 305–310 11. R. Linnemann, P. Woias, C-D. Senfft and J. A. Ditterich, Proc. Micro Electro Mechanical Systems; MEMS-97, (Heidelberg, 1998), pp. 532–537 12. A. K.-P. Kamper, J. Dopper, W. Ehrfeld and S. Oberbeck, Proc. Micro Electro Mechanical Systems; MEMS-97 (Heidelberg, 1998), pp. 432–437 13. R. A. Mathies, P. C. Simpson and A. T. Woolley, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 1–6 14. W. E. Lee, D. E. Bader, D. Jed, Harrison, et al., Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 423–426 15. S. Attiya, X. C. Qiu, G. Ocvirk, N. Chiem, W. Lee and D. J. Harrion, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 231–234 16. S. Ike, S. Shoji, H. Kudo, R. Nakamura, T. Saitoh, H. Nakanishi and T. Nishimoto, Proc. Microstructure & Microfabricated Systems IV, Electrochem. Soc., 98(14) (Boston, 1998), 70–76 17. H. Becker and U. Heim, Tech. Dig. Micro Electro Mechanical Systems (Orlando, 1999), pp. 228–231 18. T. D. Boone and H. H. Hopper, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 257–260 19. Y. Baba and O. Tabata, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 331–334 20. A. E. Bruno, E. Baer, R. Volkel and C. S. Effenhauser, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 281–285 21. J.-C. Roulet, K. Fluri, E. Verpoorte, R. Volkel, H.-P. Herzig, N. F. de Rooij and R. Dandliker, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 287–290 22. N. H. Bing, C. D. Skinner, C. W. Wang, C. L. Colyer, D. J. Harrison, J. Li and P. Thibault, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 141–144 23. D. Figeys, Micro Total Analysis Systems’98 (Kluwer Academic Pub., Dordrechit, 1998), pp. 457–462 24. A. Desai, Y.-C. Tai, M. T. Davis and T. D. Lee, Int. Conf. Solid-State Sensors & Actuators;Transducers’97 (Chicago, 1997), pp. 927–930

13 Corrosion of microsystems Guenter Schmitt

13.1

Introduction

The world of today relies on the integrity of microelectronic, microphotonic, micromagnetic, and micromechanic devices: in the information technology (radio, telephone, television, internet, prints), finance systems (banking, stock market), systems in air-, water- and groundborne transport, in chemical engineering, in military installations, medical technology, energy production, and power generation technologies, to name only the most important fields where microsystems are applied. Malfunction of such devices, which by itself constitute in many cases only a marginal value, can cause in the worst cases worldwide disasters and millions of fatalities. Continuing miniaturization of microsystems – integrated circuits are now approaching the 0.1 mm size and gate oxide thicknesses are demonstrated at 2 nm – further increases the susceptibility to system failures. There are no sound statistical data available to quantify the percentage of microsystems failures due to corrosion. However, there is no doubt that corrosion is one of the most important and decisive reasons for malfunctioning of microsystems. Despite this fact, it seems that the corrosion issue generally experiences only a low priority in design, materials selection, and corrosion mitigation in processing and application of microsystems. And this cannot be attributed to insufficient knowledge on the basic corrosion processes impairing the materials properties and functioning of such systems. The knowledge is there, because the mechanisms of environmental interactions of materials used in microsystems are basically the same as in engineering systems. Only two factors have to be considered additionally: the small, frequently microscopic dimensions of the device features and the presence of applied voltage or intense light beams. The small dimensions make the devices intolerant to any corrosion attack, while electric fields and photonic interaction may enhance the materials deterioration. This chapter is written as a contribution to increase the awareness of corrosion problems in microsystems, to enhance the mechanistic understanding of environmental materials degradation, and to outline protection strategies. In addition to electronic and photonic devices, problems in the application of electrochemical microsensors in corrosive environments will be covered.

330

G. Schmitt

13.2

Corrosion systems

Corrosion is the reaction of a material with its environment which occurs at the interface and results in measurable changes of the properties of the material and/or the environment.1 These changes can lead to an impairment of the material and/or the environment and can cause a corrosion damage. Not all corrosion reactions lead to a corrosion damage. The ensemble of material and environment constitutes the corrosion system (Figure 13.1) which is further determined by additional system parameters like temperature, flow velocity, applied potentials, etc. The understanding of corrosion mechanisms requires the definition of the corrosion systems, that is, the definition of the material (including its composition, microstructure, internal and external stresses, surface state, etc.), the environment (including its composition, type of reactive compounds, temperature, flow regime, etc.) and the situation at the interface (e.g. applied potentials, presence of adsorbates, films, tarnishes, reaction products from the interaction between material and environment, etc.).

Medium (environment)

Material

Phase boundary

Figure 13.1 The corrosion system.

Materials selection

Elimination of corrosive species

Design, construction

Material

Medium (environment) Addition of inhibitors

Phase boundary

Electrochem. protection

Coatings, linings

Figure 13.2 Strategies of corrosion prevention.

Corrosion of microsystems 331 The definition of a corrosion system includes also strategies for corrosion mitigation. They can concentrate on the material, the environment and/or the interface (Figure 13.2). The risk assessment of a corrosion damage in a given microsystem needs the definition of the corrosion systems in which the microsystem is likely to be used. Thus, electronic and photonic devices are mostly exposed to ambient environments and, therefore, will most frequently suffer from atmospheric corrosion. Electrochemical sensors for use in liquid environments, most frequently in aqueous electrolyte solutions, have to be resistant to the specific liquid medium, that is, its corrosive ingredients. The exposure to aqueous electrolyte solutions opens a new dimension to the corrosion likelihood of microsystems.

13.2.1

Atmospheric corrosion

If for technical reasons a material or a set of materials is given in the device of a specific microsystem, the risk of corrosion depends on the type and concentration of corrosive components in the environment (Figure 13.1). 13.2.1.1

Environmental contamination

To protect from direct impact of rain, snow, and wind, electronic and photonic devices are for the most part housed in enclosures. However, this does not mean that at the same time the humidity and the access of environmental pollutant are controlled in all cases. Therefore, the humidity and environmental components influence the risk of corrosion, which is further increased in the presence of electrical bias. In atmospheric corrosion differentiation is necessary between outdoor and indoor environments. Outdoor environments may contain primarily gases derived from fossil fuel combustion (sulfur dioxide (SO2) and nitrogen oxides (NOx)), particulate matter and microbes,2 while indoor environments may contain vapors from cooking, cleaning solutions, waxes and wax strippers, paints, carpets, plastic furnishings, appliances, office equipment, and manufacturing activities.3,4 Both outdoor and indoor environments are influenced by the temperature, relative humidity (including condensed moisture), presence of particles, electric field, and air velocity. In some cases vibration, mechanical stress, thermal shock, and solar irradiation contribute to the corrosion severity of the environment.

13.2.1.1.1

PARTICULATE MATTER

A widely overlooked and underestimated environmental component which decisively contributes to the degradation of microsystems is particulate matter.5,6 Most of the mass of particles in the atmosphere exist in the size range of 0.1–15 mm. Within this range, the mass exhibits a bimodal distribution. Particles of 2.5–15 mm are called coarse particles and are derived largely from natural

332

G. Schmitt

materials. They are largely mineralogical in composition and form predominantly as a result of human activity (mechanical processes) or by the action of wind and other abrasion processes on soils. Particles from 0.1–2.5 mm are called fine particles and are primarily of anthropogenic origin, though volcanic activity can sometimes be a significant source. Indoor and outdoor mass concentrations of fine and coarse particles can vary significantly from place to place. Table 13.1 summarizes data collected in a US telecommunications center and an electronics assembly plant.2,7 The particles carry varying amounts of ionic species. Sulfate and nitrate stem from fossil combustion derived sulfur dioxide and nitrous oxides which are further oxidized to sulfuric acid (H2SO4) and nitric acid (HNO3). These acids are neutralized by ambient ammonia from fertilizers, feedlot activity, and humans. The corrosivity of the particles increases with increasing content of water-soluble ionic species. It is seen from Table 13.1 that fine particles are more corrosive than coarse particles. The corrosivity depends also on the relative humidity. Table 13.2 gives critical relative humidities (RH) above which saturated salt solutions are produced on the particles at 24ºC due to the hygroscopic properties of the salts.1,8,9 Thus, if particles containing ammonium hydrogensulfate deposit on electronic equipment, it needs only relative humidities above 40% to produce a saturated salt

Table 13.1 Airborne concentrations (ng/m3) of fine and coarse particlesa Subject

Fine particles Chloride Sulfate Sodium Ammonium Potassium Magnesium Calcium Coarse particles Chloride Sulfate Sodium Ammonium Potassium Magnesium Calcium

Telecommunications center

Manufacturing plant

Indoor

Outdoor

Indoor

Outdoor

2880 4 721 14 168 19 2 10 605 4 25 3 0 2 1 20

18 090 66 5214 103 1631 53 14 25 12 340 502 725 446 1 30 90 297

12 200 15 2840 39 846 36 4 17 3150 43 104 35 13 24 5 101

13 600 37 4339 69 1573 52 8 31 8000 300 376 269 26 28 29 245

Note a The ionic components were determined by ion chromatography of water extracts of particles collected on PTFE filters. The nitrate content of the particles has not been reported.

Corrosion of microsystems 333 Table 13.2 Critical relative humidity for formation of saturated salt solutions (24C) Salt

Relative humidity (%)

Ammonium hydrogensulfate (NH4HSO4) Ammonium nitrate (NH4NO3) Sodium chloride Ammonium sulfate ((NH4)2SO4)

40 65 75 81

Table 13.3 Deposition velocities of airborne particles (cm/s) Location

Air velocity (cm/s)

Fine (0.1–2.5 mm)

Coarse (1–15 mm)

Outdoor Office building Cleanroom

250–1000 2–4 50

0.6 0.006 0.009–0.09

0.6 0.6 0.6

solution at the interface between particle and the surface of the electronic device. If this happens between mating pairs of contacts on connectors or relays the electrolyte solution can cause interrupting electrical contact and thus malfunction of the device. The deposition of particles depends strongly on the local air velocity which determines the arrival rate of the particles at the surface (ratio of airborne particle concentration and deposition velocity). This fact is important for standardized laboratory lifetime testing.2,10 Table 13.3 summarizes deposition velocities of airborne particles at different air velocities in different systems. The values given in Table 13.3 are suggested as a guideline.2,10 It is interesting to note that the arrival rate of particles on surfaces within forced-air cooled electronic equipment cabinets is 20–200 times that for normal office building environment having the same airborne particle concentration. These deposition velocities apply only in the absence of thermophoretic and electric forces. While these forces can usually be ignored for coarse particles, they play a significant part for fine particles. If the surface of a device is in a field at 10ºC cooler or warmer than the ambience, or if the surface is 100 V above or below ground potential, these effects need to be considered. Surfaces biased at a few hundred volts will collect fine particles at 5 times the rate of grounded surfaces. Not all arriving particles will stick to the surface. Therefore, the sticking probability is another important factor together with the deposition velocity. On clean surfaces both fine and coarse particles have sticking probabilities close to one, recognizing that coarse particles collect only on top-side horizontal surfaces.

334

G. Schmitt

However, on uncleaned surfaces accumulation of particles will be continuous and linearly related to the airborne concentration. Thus, it can be summarized: the higher the deposition rate of fine particles (generally carrying high content of ionic species), the sticking probability (unclean surfaces), and the relative humidity, the higher the corrosivity based on particulate matter. 13.2.1.1.2

GASEOUS POLLUTANTS

The relative humidity also determines the system’s corrosivity if corrosive gases are present. These include SO2 and NOx, as mentioned, as well as organic acids [formic acid (HCOOH), acetic acid (CH3COOH)], hydrochloric acid (HCl), carbonyl sulfide (COS), alkyl mercaptanes (R-SH), and organic sulfides [e.g. dimethylsulfide ((CH3)2S) or dimethyl disulfide ((CH3)2S2)]. While the concentration of SO2 and NOx is generally 2–10 times lower in indoor than in outdoor environments, this must not necessarily be the case for the other pollutants named above. The indoor concentrations can be even higher than the outdoor concentrations. Typical indoor and outdoor concentrations of gaseous pollutants that are estimated to be tolerable for electronic and photronic equipment without significant corrosion attack are collected in Table 13.4.11 As for salts carried on fine particles there exist also for gaseous pollutants critical relative humidities above which critical reaction rates occur with metal surfaces.12 For example, silver is very reactive to H2S, with a strong dependence on the relative humidity.13 However, no RH dependence was found for the reaction of silver with elemental sulfur vapors. The critical relative humidity for the reaction of steel with SO2 is in the order of 75%.

Table 13.4 Design benchmarks for atmosperic pollutants Species

SO2 H2S NH3 NOx HNO3 O3 Reactive chlorineb Volatile organicsc

Concentration (ppbv)a Outdoor

Indoor

50 2 50 60 4 50 4 1200

25 2 0.5 40 2 30 2 1500

Notes a Parts per billion of air by volume. b Mostly HCL, but includes Cl2 and HOCl. c Methane is excluded. Concentrations are expressed in ppbv of carbon.

Corrosion of microsystems 335 Although organic gases (except vapors of formic acid and acetic acid) are usually not directly corrosive, they can play an important role in modifying the surface, for example, changing the wettability by coadsorption. Organic gases can also affect the performance of electrical contacts, particularly contacts found in electrochemical relays. To prevent failures of electronic and photonic devices by corrosive gases long years of engineering experience in designing and packaging devices is available to mitigate these effects. Therefore, materials deterioration by corrosive gases should no longer be an issue. However, cost considerations, particularly with respect to consumer electronics, sometimes cause designers to make compromises that overlook the hard lessons of the past.

13.2.1.1.3

MICROBES

The experience of the past 25 years shows that the likelihood of materials degradation due to microbial activity is generally relatively low in electronic equipment and devices. A few instances have been reported in indoor environments. Designers should be aware of microbial-influenced corrosion and consider if the activity of microbes or other life forms (e.g. insects or even rodents) could be an issue during application of the devices.

13.2.1.2

Corrosion mechanisms

The corrosion mechanisms encountered in electronic and photonic devices are generally similar to those responsible for materials degradation in macrosystems (chemical plants, power generation systems, buildings etc.). However, special situations can occur due to the presence of electric fields. In metallic corrosion the oxidation of metal atoms needs the presence of an electron acceptor. In electrolytic corrosion of metals additionally water and ionic species are necessary to fulfil the condition of local electroneutrality. The corrosion mechanism is then electrochemical in nature consisting of an anodic metal dissolution (1) and a cathodic consumption of electrons (2) by an oxidant (Ox) which is thus transformed into the reduced species (Red). Me l Men ne Ox ne l Red(ne)

(1) (2)

The most prevalent oxidant to be considered here is oxygen from the ambient air. The cathodic reduction reaction of oxygen is given in Eqn (3). However, water also can act as an electron acceptor according to Eqn (4). O2 2 H2O 4e l 4 OH 2H2O 2e l H2 2OH

(3) (4)

336

G. Schmitt

Additionally, acidic air pollutants (e.g. HCl, H2SO3, H2SO4, HNO3, H2S) can carry the cathodic reaction, for example, 2 H 2e l H2 2H2S 2e l H2 2SH

(5) (6)

The higher the concentration of electron acceptors and the higher the conductivity of the water (due to dissolved salts, acids or bases), the higher the metal dissolution rate. Formation of insoluble reaction products (hydroxides, oxides, sulfides, carbonates etc.) will influence the reaction rate. At the surface sites where the metal is in contact with water (containing dissolved oxidants and salts) the corrosion attack would be termed uniform in case of macrosystems. Where the device features have, however, microscopic dimensions, even down to the nanoscale, the corrosion attack will always be more or less localized and, therefore, detrimental in short times. Additionally, it must be considered that not all surface sites experience the same corrosion environment at the interface. Thus, the deposition of particulate matter (carrying hygroscopic ionic species) is local and, therefore, humidity dependent saturated salt solutions will be formed locally at the deposition site. Localized corrosion attack is therefore the preferred mode of corrosion. Pitting is the predominant appearance of localized attack if local breakdown of passive layers occurs. This is generally initiated by residual halide ions. Examples for pitting on aluminum will be given later in this chapter. Enhancement of the corrosion reaction occurs under conditions of bimetallic (galvanic) corrosion at contact sites of metals with different corrosion potentials in the corrosive medium. This is the case, for example, at connectors, joints, solders, metallization layers of different metals with pores or pinholes in the top layer, intermetallic phases in pure metals, etc. The manifestation of the bimetallic corrosion reaction is generally shallow or – in some cases – deep pitting with preferential dissolution of the material with the more negative corrosion potential. Migration is a failure mechanism typical of electronic and photonic devices. This is frequently found at silver conductors. The principle is sketched in Figure 13.3. In the presence of surface moisture and an electric field silver ions can form at the positive (anodic) conductor and migrate from the anode to the cathode where they plate out, forming a dentritic structure of silver deposits that eventually grows back towards the anode, bridging the gap and causing an electric short circuit and an arc. Arcs often carbonize organic circuit board materials, producing a permanent leakage path even if the dendrite is vaporized by the arc. Metal migration shorts can also occur with gold, tin, lead, palladium, and copper. In these cases usually chlorides have to be present in the moist environment. Selective corrosion is favored if one phase of a multiphase material has a less noble corrosion potential in the given corrosion system than the rest of the material. This less noble phase will experience preferential corrosion. Selective corrosion is basically driven by the same mechanism as the bimetallic (galvanic) corrosion and can be taken as a special case in bimetallic corrosion.

Corrosion of microsystems 337 SilverAnode

Cathode Silver ion Migration

Positive potential

Negative potential

Figure 13.3 Migration at adjacent silver conductors with dendrite formation.

In some rare cases also stress corrosion cracking (SCC) is possible. SCC needs the combination of a material and a material-specific corrodent which can induce SCC at this specific material. Additionally, certain environmental boundary conditions have to be fulfilled like: transgression of a critical stress level, a critical concentration of the SCC-specific corrodent, a critical corrosion potential, and a critical temperature. SCC in electronic devices has been reported at soldered connections (e.g. at pins of ICs of surface mounted devices (SMD),14,15 or at iron–nickel–cobalt alloys with low coefficient of thermal expansion, used for metal glass seals.16,17 13.2.2

Corrosion in liquid media

In recent years the microsystem technology has gained increasing importance due to existing and extraordinary developments in the field of sensors, actuators, and micromechanical tools. Miniplants are designed and prepared aiming at miniaturizing production plants into the liter, milliliter and even microliter scale.18 A powerful microsystem technology needs, however, a great variety of different miniaturized sensors to cope with the requirements of integrated analytical, monitoring, and process control systems. Miniaturization of sensors down to the micrometer scale is easily achieved using the methods well established in silicon planar technology.19 Microsensors exhibit a wide application potential20 ranging from medical diagnosis, biotechnology, and biological research, over food control to environmental monitoring.21 However, the application of these silicon based sensors is severely hampered by the fact that the sensors generally fail too fast (within minutes, or only a few hours) when used in liquid media, specifically in electrolyte solutions. This is due to the poor barrier efficiency of the passivation layers (thickness 0.25–1 mm) generally used to insulate conducting tracks and integrated functional electronic devices on the chip. Passivation layers also provide scratch and particle protection to the circuit during chip mounting, bonding, and packaging. The poor performance of the passivation layers relates to the fact that they were originally developed and

338

G. Schmitt

optimized to serve as dielectrics in clean, noncorrosive environments. Exposure to liquid media was not intended and/or anticipated. On the contrary, microelectronic chips are very often hermetically encapsulated to protect from environmental degradation, for example, by moisture and/or electrolytes. Therefore, the multitude of potential applications of silicon based microsensors can only be transformed into reality, if the barrier performance of passivation layers is increased considerably. This has been successfully achieved in recent years.22–26 Details are described later (Section 13.4).

13.3

Corrosion of electronic and photonic devices

13.3.1

Design and materials selection

The design and materials selection for electronic and photonic devices is generally driven by their function and the electrical, resistive, magnetical and/or optical properties of their functional elements needed to achieve the desired functional performance. Additionally, the cost effective producibility plays an important role. In the past, considerations at the design stage on the corrosion susceptibility or resistance of the ensemble have been more the exception than the usual procedure. However, costly failures caused by corrosion-related malfunction of such devices increased the awareness that corrosion prevention and mitigation starts at the design and materials selection stage itself (Figure 13.2). Today, appropriate accelerated-life testing has become an important issue in fit-for-purpose design of electronic and photonic devices. The following chapters shall indicate the fields of corrosion concern and outline corrosion protection strategies based on failure analysis. 13.3.1.1

Metal migration

Since the earliest installations of electrochemical switches metal migration induced electrical short circuits have been a hazard to communications equipment.27–29 Silver-plated finishes, fired silver pastes, and silver filled epoxies are all susceptible to this problem. The boundary conditions for metal migration have already been outlined in Section 13.2.1.2. Migrate formation is possible on surfaces of insulators or even within insulators, particularly when the insulator is a composite. It was found that fiber-filled insulators can fail under high humidity conditions due to metal migration along the interface between the fiber and the polymer. Silver-filled epoxy, sometimes used for IC chip mounting, can contribute to silver migration when excessive water is present in a hermetic package. For the case of silver migration laboratory investigations revealed that the migration process can be accelerated by elevated voltage and humidity stressing. It was found that a bias of 130 V applied at 90% relative humidity for 30 days produces silver migration equivalent to 4 years in a typical office environment. In extreme cases, migration has also been observed on gold-metallized integrated circuits (ICs) and on hybrid ICs with soldered conductors. Migration at

Corrosion of microsystems 339 conductors of gold, palladium, copper, tin, or lead are favored in the presence of high concentrations of chloride or bromide. 13.3.1.2

Electrical contacts and metallic joints

Reliable connector contacts used on printed circuit boards are produced by electroplating gold over copper or copper alloy substrate which – prior to the gold plating – has been coated with a thin diffusion barrier layer (1.2–2.5 mm) of electroplated nickel or electroless nickel–phosphorous alloy. The intervening barrier layer is important because even the best quality gold electroplates of economically appropriate thicknesses (0.25–1.25 mm) are inevitably porous with pore densities in the range of 10 pores/cm2. Without the barrier layer, corrosive environmental pollutants (hydrogen sulfide, organic sulfur containing gases, elemental sulfur gas, hydrogen chloride) could reach the surface of the copper or copper alloy substrate through the pores in the corrosion resistant gold plate to form copper sulfide, copper oxide or copper chloride. This corrosion reaction is bimetallically enhanced as the corrosion potential of the copper substrate is considerably more negative than the corrosion potential of the contacting gold plate. A further problem arises from the fact that copper sulfide and copper oxide formed near a gold surface tend to creep over the surface of gold in periods as short as a few days to weeks, forming an insulating film which impairs the connector performance. Nickel-based barrier layers have the advantage of forming stable passive oxide films on exposure to air and oxygen containing moisture, and protect the substrate metal from reacting with corrosive gases even in exceptional environments as encountered in oil fields, refineries, pulp mills, and packaging in sulfur-rich cardboard due to formation of stable sulfides. Electroless nickel–phosphorous barrier layers are more corrosion resistant than those of electroplated nickel. Severe reliability problems have been experienced when instead of nickel coatings electroplated silver was used as diffusion barrier layer in such connector contacts. Even in ordinary environments creep failure occurred in a very short time because silver readily reacts with gases containing sulfur in the oxidation state 2 and zero even when present at low concentration. The silver sulfide migrates very rapidly across the gold surface and produces an insulating surface. Failure mitigation strategies include: (i) low porosity gold plating, (ii) nickel or (even better) nickel–phosphorous alloy as diffusion barrier layer, (iii) avoid electroplated silver as diffusion barrier layer, (iv) use sulfur-free packaging. High contact resistance problems due to formation of silver sulfide have also been encountered in mechanical relays with gold–10% silver contacts. Such relays are used to switch redundant standby circuits of transmission systems for telecommunication into primary use, if one circuit fails. They are sometimes open for long periods of time, perhaps months or years. During this time the silver in the gold–silver alloy slowly diffuses to the surface where it reacts with sulfur containing environmental pollutants forming an insulating silver sulfide film. In frequent relay operation this sulfide layer would be destructed with each opening and closing of the contact surface. However, a sulfide film slowly grown over

340

G. Schmitt

a long period of time is hard and strongly adherent on the gold–silver alloy contact surface. Therefore, the closing action will not dislodge the sulfide surface film, and a high resistance connection results. Failure prevention includes: (i) regularly exercising the relays, (ii) avoiding exposure to sulfur-rich environments as found, for example, near oil refineries or pulp mills, (iii) use of sealed relays. Materials selections dictated by cost reduction considerations have in many cases provoked reliability problems due to corrosion. This was, for example, the case when in the 1970s tin finishing or solder alloys were used instead of gold plating for separable connectors. Microscopic motion of one contact with respect to the other (as induced by ordinary vibrations) caused fretting corrosion with formation of tin oxide. The result was contact failure due to high resistance. It appeared that tin/gold contact pairs are actually more susceptible to fretting failures than tin/tin pairs because of the difference in hardness of the two metals. During use some tin transfers to the gold, and then the tin/tin interface is readily abraded. Other typical fretting corrosion situations were encountered in the use of fuses with tin-plated end-caps, and the use of low cost tin-plated connectors between subassemblies and circuit boards. The use of contact lubricant can postpone the failure. But replacement of the part is inevitably required. The design of electronic and photonic devices must consider materials selection of the complete ensemble which includes not only the metallic, but also all nonmetallic parts. An example of improper design was found in a coaxial cable used in a subset detection system. The design is sketched in Figure 13.4. The copper cable was insulated with a rayolin dielectric which was surrounded by a silverplated copper braid (electric shield and ground return) followed by a rayolin jacket.

(a) Coaxial cable6 5 (cross-section) 4 3 2 1

(b)

1. 20 gauge copper 2. Rayolin dielectric 3. Silver-plated copper braid 4. Rayolin jacket 5. Silicone oil 6. Butyl rubber

Silver

(c)

Copper No free sulfur present

Sulfide tarnish Crack

Free sulfur present

Figure 13.4rubber Feature of coaxial cable failure by sulfur-containing rubber in outer jacket in outercaused rubber jacket jacket. (a) Cross-section of coaxial cable housing; (b) No free sulfur present in outer rubber jacket; and (c) Schematic of silver-plated copper wire used for the shielding braid. Sulfide tarnished wire cracks during flexing. Free sulfur present in outer rubber jackets.

Corrosion of microsystems 341 Silicon oil was introduced between the outer jacket and the braid to increase the flexibility in handling. The jacket material was made from chlorobutyl rubber which unfortunately was sulfur-vulcanized. Sulfur compounds and – if present – also free elemental sulfur dissolve in the silicone oil and diffuse to the braid where they react with the silver-plated copper braid forming sulfide tarnish. Since sulfides of silver and copper are very brittle, cracking of the shield occurred resulting in open circuits. This kind of failure can easily be prevented by using sulfur-free rubber jacketing. Designers should also know that even gold metallizations may corrode and may cause failures even in the absence of water-soluble ionic contaminants.30 In the presence of only moisture and applied voltage, positively biased gold conductors may corrode with formation of voluminous, solid gold hydroxide [Au(OH)3]. With progressing reaction the corrosion product becomes thicker and eventually spalls. Finally, the conductor resistance is increased due to loss of metallic gold. Also electrical leakage between conductors is possible due to corrosion product spread over the device surface. Corrosion prevention includes coating the device surface with adhering dielectrics or polymers. 13.3.2

Processing

Processing-related corrosion failures at electronic and photonic devices are caused by pollutants residual from the different production and fabrication processes in the component and assembly manufacturing lines. Thus, during formation of the conducting tracks on the circuit board etching solutions containing chemicals like FeCl3, CuCl2, or Cu(NH3)4Cl2 are used and may pollute the surface in case of improper cleaning. The surface activation prior to the deposition of bond surfaces includes the use of ammonium or sodium persulfate solutions, chromic acid, sulfuric acid, nitric acid, or fluoroboric acid. Metallizations on the circuit board consist of tin–lead or nickel–gold layers including electro-plating or electroless plating processes which use fluoroborate or sulfate solutions together with brighteners. Improper handling or insufficient cleaning leads to residues from these processes on the circuit board which in combination with humidity start corrosion processes under service conditions. Further contamination on the circuit board relate to the storage conditions and the handling prior to soldering processes, for example, air pollutants, dusts, hand sweat. The major part of the circuit board pollutants, however, stem from the soldering process and the no-clean processing commonly applied today which means that the assemblies are generally not cleaned after the soldering process. This leaves flux residues on the surface which concentrate at the bonding areas of the assemblies. However, even if cleaning processes are used after the soldering, flux residues can still remain on the board if the cleaning parameters are not optimized. Flux activators are inevitable to remove oxides from the metallization surface and assure wetting by the liquid solder. Flux activators today contain collophonium additives of chlorine-free organic acids like adipic acid, propionic acid, citric acid or apple acid. In former times flux activators used to contain chloride salts too.

342

G. Schmitt

A prominent part of the corrosion failures with electronic and photonic devices is caused by halide ions (mainly chlorides, but also fluorides) stemming from i halogenated solvents used in surface cleaning procedures (e.g. 1,1,1trichloroethane, fluor containing solvents like Freon TF, Freon TA, Freon TE, Freon TMS). These solvents decompose slowly, for example, by hydrolysis (moisture is enough) under formation of chloride and fluoride ions, a process which is catalyzed in the presence of metallic surfaces of iron, copper, copper alloys, aluminum, etc.31 Therefore, these solvents generally contain stabilizers. Halogenated solvents are also sometimes used as diluent for protective coatings; ii metal reactive ion etching (e.g. plasma enhanced) using fluoro compound like trifluoromethane; iii solder flux-activators if they contain chlorides (NaCl, NH4Cl). Minute amounts of halides are enough to activate metallic surfaces by breakdown of passive layers. Aluminum is very susceptible to halide-induced corrosion causing pitting and selective corrosion attack already under common processing conditions. Corrosion failures at aluminum caused by chlorinated solvents during the processing of circuit boards (components containing aluminum in non-hermetic packages) and ICs (aluminum and aluminum–copper connections) have been frequently found.32 Residues of free chloride from decomposition of chlorinated hydrocarbons resulted in pitting specifically on the interconnection patterns. The corrosion products at pitted areas include chlorides, oxychlorides and oxides of aluminum and copper. In order to inhibit electromigration in ICs a small amount of copper (0.5–2%) is added to aluminum metallizations. This, however, increases the pitting susceptibility during rinsing after metal reactive-ion etching, specifically at edgdes of the metallizations.33 The reason is that copper – due to low solubility in aluminum – precipitates as intermetallic theta-phase Al2Cu along grain boundaries. Because the Al2Cu-phase has a higher corrosion potential than the surrounding pure aluminum, the intermetallic phase acts as local cathode for the cathodic water reduction or oxygen reduction, and the surrounding pure aluminum is anodically dissolved (Figure 13.5). The process is favored by residual fluoride from the rinse operation which destructs the passive oxide film on the aluminum. It was found that saturation of the rinsing water with CO2 inhibits the cathodic reaction and prevents pitting.34 Halogenated solvents may also be responsible for aluminum electrolytic capacitors commonly used on circuit boards. If the capacitor comes in contact with halogenated solvent, the solvent may diffuse through or along the edges of rubber sealing gaskets (Figure 13.6) causing breakdown of the protective oxide layer on the positive aluminum leads.35 This starts a rapid corrosion between the halogenated solvent and the underlying aluminum metal resulting in an open circuit.

Corrosion of microsystems 343

Al3+

Al3+

H 2O OH–

H2

Al/Cu-Particle

e–

e–

Aluminum

Figure 13.5 Localized corrosion at aluminum–copper intermetallic phase.

Seal rubber gasket Lead wire

Aluminum case

Sleeve

Aluminum lead

Figure 13.6 Schematic cross section of a typical aluminum electrolyic capacitor.

Further acceleration of the corrosion process can relate to a venting of the capacitor due to a corrosion-related increase of the internal pressure. Failure prevention can be achieved by improving the seal between the gasket and the electrode leads to reduce the diffusion of solvent into the capacitor. Cadmium plating of the positive aluminum leads prevents halide-induced aluminum activation. Preferential corrosion of aluminum can also become a problem in metallization layer combinations on silicon integrated circuits with submicron feature. It has been observed that the aluminum layer in a Ti/TiN/Al/TiN metallization stack experiences enhanced corrosion with undercutting of the TiN top layer up to 20 mm (Figure 13.7).36 The reaction is initiated by residual chloride from reactive-ion etching and is stimulated by galvanic coupling to titanium nitride which has the more noble corrosion potential in this system and acts as the cathode for the electrochemical water (or oxygen) reduction. Further increase of the corrosion rate is due to the unfavorable anode/cathode surface area ratio, that is, very small anode (aluminum), relatively large cathode (titanium nitride) area. In order to avoid

344

G. Schmitt

H2O + e–

OH– +1/2 H2

TiN (cathode)

Al (anode)

Al

Al+3 + 3e–

TiN Ti SiO2

Figure 13.7 Bimetallic (galvanic) corrosion of Ti|TiN|Al|TiN| stack integrated circuit on Si chips. Corrosion occurs during water rinsing that follows patterning in a reactive plasma.

the bimetallic corrosion enhancement, the processing sequence has to be changed in such a way that the top TiN layer is not fully exposed to water until all chloride residue has been removed completely from the side wall of the stack. Stress corrosion cracking is a failure mode which has been observed at pins of ICs.14 It appeared that the intermetallic phases (type Cu3Sn and Cu6Sn5) formed during soldering of the base material CuFe2P with Sn62Pb36Ag2 were susceptible to SCC, for example, in ammonia containing environments. The same was true for connections at CuZn37 produced with the same solder. It is assumed that mechanical stress induces microcrack formation in the intermetallic zone. After crack initiation cracks propagate along the intermetallic zone (IZ) perpendicular to the main stress direction through the solder zone. This initiates crevice corrosion with reduction of pH and increase of copper ion concentration at the interface IZ/copper base material. The crack propagation becomes intermittent. Prevention measures include optimizing the soldering conditions (reduction of temperature and time to reduce the formation of intermetallic phases) and changing the lead construction (e.g. J-string, TAB lead). One example that should remind that corrosion of metals is not only caused by halide-induced passive breakdown or surface reactions with acids but also with bases if the metal can be dissolved as hydroxo complexes (aluminum, tin, lead, zinc, tantalum etc.) features a failure analysis of thin-film tantalum nitride resistors. These resistors are used in hybrid IC and silicon thin-film IC applications. After sputter deposition of tantalum nitride and subsequent patterning the resistors are stabilized by heating in air to form protective tantalum pentoxide layers. In the present environmental and manufacturing contamination, atmospheric moisture, and applied voltage, the surface pH may become basic enough to dissolve the tantalum pentoxide to non-passivating tantalate salt (Figure 13.8). Adsorbed moisture and oxygen oxidize the underlying tantalum nitride to tantalum pentoxide which is again transformed into the tantalate salt. Thus, the degradation cycle continues increasing the resistance of the thin film. The basic pH is produced by electrochemical water reduction at the negative (cathodic) end

Corrosion of microsystems 345 Basic solution or contaminant Unstable Tantalum pentoxide

OH–

OH–

Tantalate

OH –

Tantalum pentoxide

Tantalum nitride

Moving front of Ta2O5 prior to conversion to tantalate

Figure 13.8 Schematic illustration of a basic corrosion of tantalum nitride thin-film resistor.

of the resistor producing hydroxide ions. Corrosion prevention includes cleanliness and humidity control during processing and appropriate packaging. Hazards are detected by life testing, using bias, humidity, and temperature stressing. Organic coatings (mainly polyurethane- or acrylate-based) are frequently applied to prevent functional failures due to deposition of water films below the dew point. The application occurs by dipping or spraying. Generally the assemblies are not cleaned prior to the coating process. This means that pollutant residues from the manufacturing process remain in place and are covered by the coating. In the presence of water or moisture all organic coatings absorb more or less amounts of water causing swelling and diffusion of water to the interface base material/coating. Even without any pores, cracks or fissures in the coating water reaches the hygroscopic contaminants on the surface and concentrates there with formation of saturated salt solutions. This finally leads to osmotic-driven formation of delaminations, blisters, and cracks in the coating giving rise to enhanced corrosion attack. This process is supported by bias as well as by temperature changes and temperature gradients at the interface base material/coating. Therefore, organic coating is only useful on clean surfaces. 13.3.3

Shipping, storage, and application

In times of globalization it is very common that the devices (from computer chips to equipment frames) needed for end-assembly in the manufacturing plant are produced at different places, sometimes thousands of kilometers away. This need shipping and storing, and frequently extended periods of time. Although corrosion problems connected with shipping and storage are well known there are generally no standards specifying the composition of the packaging materials and shipping containers, and the storage conditions. It is well known that corrosive substances can emanate from packaging materials. Thus, sulfur containing compounds from paper and card board can tarnish copper and silver. Contacts,

346

G. Schmitt

connectors, equipment frames, and other parts can be corroded by gases emanating from adhesives, plastisizers, molding compounds, improperly cured plastics etc. High humidity and elevated temperatures favor the corrosion process. Thus, the shipping and storage situation can be responsible for the fact that the devices (and also the final manufacturing products) reach their final destination in an already corroded condition. Mitigation strategies include (i) selecting packaging materials that are compatible with metallic parts, (ii) making certain that no corrosive gases emanate from the surrounding material (like plastics plastisizer, molding compounds), and (iii) avoiding shipping and storage at elevated temperature and humidity. While the conditions during processing, manufacturing, shipping, and storage can basically be controlled by the manufacturer, the service conditions, however, are determined by the end user. Today, electronic and photonic devices and assemblies are used in environments (e.g. in the transport, communication, information, or building technology or even in toys) which some time ago would have been considered as unacceptable with respect to humidity, unavoidable gaseous and particulate contaminants, temperature levels and gradients. The belief of the end user in the reliability of electronic equipment has risen so high that the risks of failures and malfunction due to detrimental environmental effects is more and more underestimated. Therefore, the awareness of the risk parameters should be increased by the manufacturers and the end user. And the risk of failure increases with the development of the microtechnology into the nano scale where the tolerance to undesired reactions with the environment is close to zero. 13.3.4

Corrosion testing

The introductory remarks must start with the statement that there is no single “shorttime” test to evaluate all possible environmental interactions which could shorten the life of electronic and photonic devices. It needs always a series of tests to elucidate possible degradation processes. The selection of tests should be based on the existing knowledge and experience in corrosion sciences, chemistry, and physics. The first step in deciding on a test protocol is to conduct a literature search on the atmospheric corrosion of each material involved. Consultation with experts in the corrosion of each material is also advised. Finally, internationally standardized tests which include the state-of-the-art in a specific field of environmental interaction of materials should help to get unique, reliable, and predictive information on appropriate testing procedures and protocols. There are already a number of such testing standards available. In some cases the debate is still going on with respect to the application limits and life-time predictiveness of the test. Unique aspects of testing electronic and photonic devices are outlined in the literature.2 13.3.4.1

Accelerated life time testing – general remarks

Accelerated life time testing requires extrapolating results from short-time tests in a specific test environment (exposure time: days to a few weeks or months) to the behavior of the tested material, device or ensemble over long times (e.g. a life

Corrosion of microsystems 347 time of several years) in an environment which is generally only to a certain extent comparable with the test medium and may, additionally, change with time. This is indeed difficult, however, it has nevertheless been achieved in some cases where the performance in long-term practical application could be correlated with laboratory test results. Thus, test procedures are available which accelerate the effects of temperature, humidity, and bias.37 Protocols exist for corrosive gas testing but there is still some disagreement about what combinations of corrosive gases in what concentration are appropriate for each type of environment. A few of these tests are found in Refs.38–45 Testing for the effects of particles is much less advanced than corrosive gas testing, but further advanced than organic vapors or microbial testing. For testing the effect of particles some approaches were made with respect to coarse particles alone or mixed effects of fine and coarse particles.46,47 In view of the discussion above, testing for the effects of fine particles should be a priority. However, the development of test methods is only in the initial stages. 13.3.4.2

Testing for thermal, humidity, and bias effects

Thermal, humidity, and bias testing can be done individually or in any combination. Simple thermal testing without humidity or bias is usually performed in combination with mechanical stress or as a temperature cycling experiment. Humidity testing is nearly always carried out at temperatures ranging from 50ºC to 100ºC with relative humidity from nearly zero to 100%. For tests above 100ºC steam bombs can be used. The non-condensing humidity is then generally 85%. Such tests are applied to evaluate coatings, papers, adhesives, adhesion between mechanically or chemically bonded materials, and in electronics, for evaluating the effectiveness of devices encapsulation. Bias is frequently added for testing of electronic devices, printed wiring boards, and assemblies of electronic equipment. The 85ºC, 85% RH, bias test has been the predominant one in electronics for many years.37 It is generally able to find weak points in new products and is especially useful for quality control of products with known long-term reliability. 13.3.4.3

Corrosive gas testing

Comprehensive studies comparing laboratory data with field data in the years 1975–90 yielded test protocols which allow quite accurately to simulate a 20years field performance in 5–30 days laboratory exposure tests.38–42 Based on metal coupon exposures and simultaneous air pollutant measurements from over a hundred locations around the world, a four-level environmental classification scheme for environmental severity was developed which is now in common use around the world: Class I Class II

No significant corrosion observed. Mild pore corrosion of Au/Ni/Cu with chloride, oxide, and smaller amounts of sulfide in corrosion product; corrosion product on unprotected copper contains oxide and chloride.

348

G. Schmitt Table 13.5 Corrosive gas test conditions2 Location

Atmosphere

Test duration (days)

Indoors (environmentally controlled) Outdoors (uncontrolled)

25C; 75% RH; 10 ppb Cl2; 10 ppb H2S; 200 ppb NO2; 200 ppb SO2; balance air 25C; 75% RH; 10 ppb Cl2; 10 ppb H2S; 200 ppb NO2; 200 ppb SO2; balance air

10

20

Class III Pore and creep corrosion of Au/Ni/Cu with major amounts of sulfide, chloride, and oxide in corrosion product; corrosion product on unprotected copper is rich in sulfide and oxide. Class IV Severe creep corrosion of Au/Ni/Cu with major amounts of sulfide, oxide, and chloride in the corrosion product; corrosion product on copper is primarily a sulfide film with some oxide. Suggested conditions for corrosive gas testing that cover both indoor and outdoor environments are given in Table 13.5. It can reasonably be argued both for or against using the same humidity for indoor and outdoor testing. Since most chambers have been set up to operate stably at 70 or 75% RH, one possible approach is to do all corrosive gas exposure testing at this RH, but for outdoor simulations, to follow the corrosive gas exposure with a cyclic RH exposure that covers the full range of expected outdoor conditions. Helpful guidance for conducting corrosive gas tests is provided in ASTM B 810,43 ASTM B 827,44 and ASTM B 845.45 According to the literature2 most failures in electronic assemblies attributable to the environment are due to ionic particle contamination in conjunction with atmospheric moisture. Therefore, valid accelerated testing of electronic components, circuit boards, and assemblies should include ionic contamination, mainly stemming from fine particles. 13.3.4.4 13.3.4.4.1

Particle testing COARSE PARTICLE TESTING

Occasionally, some materials have been subjected to coarse particle dust tests in which the test dust was frequently a “real world” standard material such as “Arizona Road Dust”, talc, silica, or other mineralogically-derived substances. Relevant background information is given in the literature.46,47 Typically, the test dust is dispersed by mechanical means into a rapidly moving air stream that is directed towards the test specimen. Most existing or emerging tests use, however, particle sizes much larger than would ever be encountered in indoor environments, except perhaps for some manufacturing processes. Therefore, internationally

Corrosion of microsystems 349 recognized standards for indoor coarse particle testing are not likely to be available for many years. In the absence of such tests, the test engineer will have to use best judgment in deciding how to carry out coarse particle testing. Fortunately, the effects of coarse particles indoors are usually small. 13.3.4.4.2

FINE PARTICLE TESTING

It has already been outlined in Section 13.2.1.1.1 that fine particles are more corrosive than coarse particles because fine particles carry more ionic species and have a higher deposition velocity. Therefore, fine particle testing is more important in lifetime testing. However, due to the small size of fine particles (0.1–2.5 mm) their standardized preparation is not trivial. Over the last ten years, a series of methods for depositing particles on surfaces that are representative of submicron atmospheric particles has been developed. The first uses a synthetic dust produced by hand-grinding of a mixture of ammonium sulfate (29 mass-%), ammonium bisulfate (3 mass-%), sodium chloride (1 mass-%), potassium bicarbonate (1 mass-%), and talc (66 mass-%) in dry nitrogen atmosphere. The dust with a particle size of typically 1–5 mm is then mechanically dispersed on the test substrate. This test is useful for solving field problems because it is quick. The limitations in using this test are (i) the particle size is too large, so that gravitational settling dominates the deposition process, and (ii) the deposition on partial shielded surfaces, especially vertical and top-side horizontal surfaces, is inadequate. To overcome this problem, the generation of submicron particles is achieved by atomizing an aqueous solution of the appropriate salt or salt mixture, followed by thorough drying of the resulting aerosol-laden air stream. A special test chamber has been developed for this kind of testing. Details are described in the literature.2 It has already been demonstrated that this method accurately simulates the deposition and electrical effects of fine particle dust as are found in a forced-air-cooled equipment frame. The resulting electrical effects on circuit boards and equipment assemblies are comparable to those found in field environments. At the very least, use of this test demonstrates a state-of-the-art goodwill effort to show that the equipment will operate reliably in the presence of fine, hygroscopic dust. The acceleration factor for fine particle deposition is about 200, depending on the surface area and configuration of the equipment tested. 13.3.5

Future developments

With the continuing shrinkage in the feature size, the tolerance of environmental interactions of electronic and photonic devices is further decreased. The integrity of metal conductors in the submicron scale demands zero corrosion. New challenges will be associated with the replacement of aluminum by copper as the conductor material and the anticipated change to a polymeric dielectric material in place of the currently used silicon dioxide. The biggest challenge for photonic devices will be in the packaging which currently need hermetic enclosures because

350

G. Schmitt

of the extreme sensitivity of lasers and diodes to moisture and contamination. A breakthrough in the passivation technology for compound semiconductor materials would reduce the cost of the packaging considerably because it would allow to tolerate non-hermetic enclosures. In all processing steps from the device to the assembly, more cleanliness is mandatory to remove as much processing-related residual contaminants as possible. New processes need a thorough systems analysis to avoid corrosion problems already in the design stage.

13.4

Electrochemical sensors

The use of electrochemical sensors – based on silicon planar technology – in liquid media, especially in electrolyte solutions, is another challenge which is closely related to the barrier properties of the passivation layers used to protect the microstructure from interaction with the extremely hostile environment. While in the technology of electronic and photonic devices it is aimed to eliminate or prevent the interaction of the structures with ionic species, it is the direct intention in electrochemical sensoring to use the microelectronic sensors in ionic media. Therefore, specific protection strategies have to be developed which focus on the improvement of the environmental resistance of the passivation layers. 13.4.1

Passivation layers

In order to understand corrosion problems with passivation layers on sensor devices and to develop appropriate mitigation methods it is necessary to know the chemical nature of passivation layers and their physical properties. As literature on this subject with respect to corrosion aspects is scarce, an overview is given here before discussing mechanistic questions on passivation layer deterioration and recent progress in improving their barrier properties. 13.4.1.1

Chemical nature

Passivation layers can be organic and inorganic in nature. Organic passivation layers can consist of polyimides as well as photoresist lacquers. While polyimides are formed by polycondensation reactions of (generally aromatic) polycarboxylic acids (e.g. pyromellitic anhydride) and (generally aromatic) polyamines (e.g. p-phenylene diamine) (Figure 13.9), the photoresist films are quite often novolack-based and formed by polycondensation of phenolic compounds (e.g. phenol, bisphenol A) with formaldehyde and subsequent reaction with unsaturated epoxy esters of the acrylic type (Figure 13.9). The organic passivation layers are generally applied by spin coating processes. Inorganic passivation layers include silicon dioxide (SiO2), silicon nitride (Si3N4), silicon oxinitride (SixOyNz), and silicon carbide (SiC). They are applied as mono layers, duplex layers (e.g. SiO2/Si3N4) or triplex layers (e.g. oxide/nitride/oxide (ONO) layers in the sequence SiO2/Si3N4/SiO2).

Corrosion of microsystems 351 CH2

CH2

CH2

CR

CR

CR

C

O

C

O

H

CH2

CH2

C

C

OH H

O

N

N

O

O

C

O

CH2 OH H

C

OH

CH2

O

O

CH2

CH2

n

O

O

CH2

CH2

O

O

O

CH2R

n

Polyimides

Novolack-based photoresist

Figure 13.9 Chemical structures of organic passivation films.

H

H H

Si

H

H

N

Si Si

H

H

+ NH

N

+ SiH2

Si

H

N

N

Si

Si

Si

H H Si H N

Si

1

2 H

cross-

H

linking

N Si

H

H N

Si Si

H Si + H2 N

Si 3

Figure 13.10 Growth reaction (1 → 2) and cross-linking reaction (2 → 3) during deposition of plasma Si–N.48

These layers are generally prepared by plasma-enhanced chemical vapor deposition (PECVD) via decomposition of appropriate gas mixtures. Due to the specific deposition method PECVD films consist of non-stoichiometric compounds containing varying amounts of other elements (hydrogen, chlorine, or fluorine) depending on the specific deposition conditions.48 Thus, PECVD silicon nitride layers can contain Si–H and N–H bonds as schematically shown in Figure 13.1048 and could be best described as polysilazane (SixHyNz), that is, as an inorganic polymer.50 Plasma SiO2 may contain 5–10 at-% hydrogen, all bonded as OH.50 PECVD silicon carbide layers can consist of low cross-linked polysilicon or nearly completely cross linked polycarbon depending on the silicon or carbon content of the deposition gas mixture.51 The carbon content can amount to 40 at-%52 and the hydrogen content to 49 at-%.53–55

352

G. Schmitt

Post-treatment of PECVD layers with reactive gases (CF4, HCF3, oxygen plasma, etc.) can change the chemical nature and composition in near-surface zones up to depths of several tenths of nanometers. 13.4.1.2

Physical properties

The non-stoichiometry of the PECVD layer composition influences the density, the mechanical properties (Young’s modulus, ultimate tensile stress, microhardness, intrinsic growth stress, fracture stress, fracture strain), the electrical properties49 (electrical resistance, permittivity), and the growth structure56 of the passivation layers. Specifically, the mechanical properties and growth structures of inorganic passivation layers directly influence their performance under corrosive environmental conditions. Thus, films in tension tend to develop cracks, whereas those with high compressive stress tend to delaminate and peel off. The stress state of the dielectric can also affect the metal layers it isolates. It is known that intrinsic stresses in plasma SiN passivation layers can have tremendous influence on the stress induced voiding as well as hillock growth on aluminum runners in double level metal integrated circuits.57 Lowering the residual stress of the film is also desirable to prevent the underlying aluminum film from notching.57 Depending on the deposition conditions the mechanical properties of plasma passivation layers can vary within broad limits. This has been widely investigated during recent years, as reviewed in the literature.22 It can be summarized that PECVD silicon nitride may have intrinsic stress levels between 600 and 1200 Pa on silicon substrates.58 PECVD silicon oxide films generally contain compressive stresses (100 Pa).59 In PECVD silicon carbide layer stresses generally range between 80 and 100 MPa53,60 but can also reach 1 GPa.61 13.4.1.3

Corrosion properties

The protective properties of inorganic and organic films differ significantly. Organic films are very sensitive to diffusion of water. For a 25-m film it may take only a few seconds to reach 90% saturation at the substrate surface.62 Corrosion can be prevented only if there is perfect adhesion of the polymeric film to the metallic underground. Degradation of organic films occurs by splitting of cross-linking bonds between molecule chains and between the polymer and the substrate. In contrast to this a 1-m thick inorganic film will need several hundred years62 to reach saturation with water. Therefore, diffusion is negligible in a crack and pinhole-free film. Chemical dissolution of PECVD films can be achieved by etching. The etching rates were mostly determined in solutions containing hydrofluoric acid, due to the use of hydrofluoric acid as a selective etchant in microtechniques. There are less investigations on other corrosive solutions. For PECVD silicon nitride a corrosion rate of 41 nm/h was found for 30% KOH at 70ºC.62 LPCVD silicon nitride which is known to be less sensitive to corrosion than PECVD silicon nitride showed corrosion rates of 0.051 nm/h in

Corrosion of microsystems 353 deionized (DI) water at 90ºC, up to 0.7 nm/h in diluted KOH at pH 11 and no detectable loss in diluted HNO3 at pH 2.63 Silicon oxide layers are usually stable at low pH values and easily etched at high pH values. This was shown for thermally grown SiO2 layers on silicon wafers.63 The etching rates are below 1 nm/h in DI water (90ºC) and diluted HNO3 (90ºC, pH2) and increase to 100 nm/h in diluted KOH (pH 11, 140ºC). Obviously, the best chemical resistance is found for PECVD SiC layers. In DI water at 90ºC and diluted HNO3 at 140ºC and pH2 no etching rate was detectable.63 In 40% HF or HCl/HNO3 mixtures no corrosive effect was observed after 70 min at room temperature.53 In alkaline medium (diluted KOH, pH 11, 140ºC) the etching rate was below 0.26 nm/h.63 In 45% KOH at 80ºC the etching rate of SiC layers was found to increase above 600 nm/h. The corrosion rate of SiC layers in KOH correlates with the carbon and hydrogen content, and the number of CH3-Groups in the layer.53 Only recently, the corrosion resistance and barrier properties of passivation layers on Si-based microelectrode arrays were improved considerably.22–26,75–78 Passivation layers tested included organic and inorganic films. The organic passivation layers were applied by spin coating and included polyimide films (Selectilux HTR 3/ Merck; thickness: 800 and 1000 nm, respectively), a novolackbased photoresist (AZ 5214/Hoechst; thickness:1400 nm), an epoxy resin (Probimer, Ciba-Geigy, thickness 1400 nm). The inorganic layers were deposited by the PECVD process and included mono layers of SiO2 (800 nm), Si3N4 (800 nm), and SiC (1000 nm), duplex layers (400 nm SiO2 capped with 400 nm Si3N4), and oxide–nitride–oxide (ONO) triplex layers (130 nm SiO2 540 nm Si3N4 130 nm SiO2).64 The barrier properties of the passivation layers were tested by exposing test chips at ambient temperature in aerated 1 M NaCl solution at pH values of 2, 7, and 10.22–26 Organic passivation layers like polyimide and photoresist quickly undergo delamination at edges and steps in 1 M NaCl at pH 7 and protect only for a few hours (Figure 13.11). Only in the case of epoxy-based layers the time-tofailure ranged in the order of 500 h. Optimization of polyimide films was achieved by inert gas plasma treatment. The time to failure was improved to 600 h. From the inorganic passivation layers the mono layers of SiO2 and Si3N4 and the duplex layer exhibited only poor protection in the as-deposited state (Figure 13.11). The lowest time to failure showed the SiO2 film (a few hours), while the plasma Si–N layer protected only for less than a day. Even the duplex layer failed after one day. Annealing of plasma Si–N layers for 10 min at 400ºC caused spalling due to high intrinsic stresses. Annealing at higher temperatures yielded an even higher spalling intensity. SiO2 layers could be annealed up to 800ºC without spalling. The time to failure of such films was much higher (31 h) than in the as-deposited state (3 h). The highest effect was observed when duplex layers were annealed for 10 min at 400ºC. In this case the time to failure amounted to 340 h. Very successful were attempts to improve the barrier properties of the duplex layers by modifying the PECVD deposition conditions. By decreasing the

354

G. Schmitt

deposition rate a better growth structure with less intrinsic stress was obtained. This increased the time to failure of the as-deposited layers to 500 h. Silicon carbide layers protected 700 h in the as-deposited state without further processing. The best results were obtained with the ONO-triplex layer. Using the new modified conditions for the deposition of the intermediate plasma Si–N film, the protection of this passivation layer lasted nearly 1200 h (Figure 13.11). The better performance of both duplex and ONO layers compared to monolayers of the same thickness is obviously due to the balancing of the counteracting intrinsic stresses in the SiO2 and Si3N4 layers. SEM investigations on the failure modes of the tested chips revealed that surface steps, for example, at non-buried conducting tracks (Figure 13.12(c)) with a height of 200 nm, induced irregularities in the growth structure of the films (nodular growth) and structural intrinsic stress with high susceptibility to cracking, specifically after prolonged exposure. Therefore, to avoid intrinsic stress at step structures, chips were fabricated which contained buried conducting tracks (Figure 13.12(d)). This was achieved by etching grooves into the thermal SiO2 layer and subsequently filling the grooves with the deposit of the conducting track. In the ideal case this chip production process can completely avoid the formation of steps and edges. Exposure tests with a duplex layer (plasma Si–N film produced with the improved PECVD process) on buried conducting tracks increased the times to failure from 500 h up to 2000 h, that is, close to 3 months, in the as-deposited state. Figure 13.13 compares in SEM photographs of deliberately broken chips with duplex passivation the appearance of non-buried (Figure 13.13(a)) or buried (Figure 13.13(b)) conduction tracks. The duplex layer on non-buried conduction tracks shows globulitic growth structures at the edges

Novolack based Polyimide based Polyimide (plasma treated) Epoxy based SiO2 SiO2 (annealed) Si3N4 SiC SiC (buried conducting tracks) Duplex Duplex (annealed) Duplex (opt. Si3N4) Duplex (opt. Si3N4) + Novolack Duplex (buried conducting tracks; opt. Si3N4) ONO (opt. Si3N4) 0

500

1000 1500 Time to failure

2000 (h)

Figure 13.11 Corrosion resistance of passivation layers in 1 M NaCl at 25ºC.

Corrosion of microsystems 355 (a)

(c)

10 mm

Passivation layer

10 mm

Electrode Conducting track

Cr Au Cr Therm. SiO2

Contact pad Mask-alignment marker

(b)

(d) Passivation layer Microelectrode

Intermediate layer

Cr Au Cr Therm. SiO2

Therm. SiO2

Wafer (n-Si<100>)

Figure 13.12 Design structures of microelectrode array. (a) Top view; (b) cross section; (c) non-buried conducting track; (d) Buried conducting track.

(a)

(b) 2 µm

Duplex passivation layer surface

480 h

2 µm

c Fracture Au Substrate (c)

(d) 2600 h

1100 h

200 nm

200 nm

Figure 13.13 (a,b) Deliberate fracture through duplex-passivated chips after exposure to 1 M NaCl. Chips with (a,c) non-buried conducting tracks exposed 1100 h; (b) buried conducting tracks exposed 480 h; (d) buried conducting tracks exposed 2600 h.

356

G. Schmitt

of the not buried conducting tracks (Figure 13.13(a)). Opposed to that, the film morphology is much smoother at the edges of buried conducting tracks (Figure 13.13(b)). At the edges of the metallic structure a small groove is visible in the passivation layer (Figure 13.13(b)). This is due to the isotropic etching of the trenches into the thermally grown silicon dioxide layer with dilute HF. Therefore, the trenches are not completely filled with the deposited metals and the PECVD passivation layer copies the groove as sketched in Figure 13.12. For the silicon carbide layer burying the conducting tracks extended the time-to-failure up to 1000 h. ONO layers on buried conduction tracks promise even better results (Figure 13.11). The performance of the optimized duplex layers was lower at test chips with opened, non-buried electrode surfaces under these conditions due to delamination. However, using buried electrode surfaces delamination was reduced considerably. 13.4.1.4

Failure modes and mechanisms

The failure mechanisms of organic and inorganic passivation layers can be divided into three groups (Table 13.6): Group 1: Group 2: Group 3:

Failures due to mechanical stress and film defects. Failures due to chemical, physicochemical and electrochemical reactions. Failures due to the combined action of mechanical stress and chemical interactions.

Within Group 1 cracking of passivation layers due to stresses, either extrinsic (externally applied) or intrinsic due to film growth, is a very important and common failure mode of inorganic and organic passivating layers. Stress produces strain in the passivation layers which can lead to delamination, the intensity of which depends on the kind of stress and the adhesion of the film to the substrate. For a given layer there exists always a thickness-dependent critical strain, above which the layer cracks and may spall depending on the adhesion to the substrate and the ductility of the substrate. A general failure mode diagram for critical strain vs layer thickness is given in the literature.65 In case of tensile stress only cracking of the film with no delamination or spalling will occur, if the film adhesion is good and the substrate exhibits a higher ductility than the film (Table 13.6, Group 1.1(a)). Delamination and spalling will, however, occur if the film adhesion is poor, independent of the ductility ratio between film and substrate (Table 13.6, Group 1.1(b)). In case of compressive stress the film will spall if its adhesion to the substrate is good (Table 13.6, Group 1.2(a)), or will buckle if its adhesion is poor (Table 13.6, Group 1.2(b)). A failure due to internal stress in an inorganic passivation layer is illustrated in Figure 13.14. In this case a chip with plasma Si–N passivation had been annealed for 10 min at 400ºC. The spalling of the film occurred preferentially over the

Table 13.6 Failure modes of passivation layers Group

Failure mode

Schematic sketch

1 1.1 1.1.1 1.1.2

Failures due to mechanical stress and film defects Cracks due to tensile stress Intrinsic stress Extrinsic stress

1.2 1.2.1 1.2.2

Cracks due to compressive stress Intrinsic stress Extrinsic stress

(a) without delamination (b) with delamination

(a) spalling

(b) buckling

1.3 1.3.1 1.3.2

Cracks due to intrinsic stress Induced by surface topography Induced by swelling

1.4

Pinholes

1.5

Particle inclusion

2 2.1

Failures due to chemical, physicochemical and electrochemical reactions + – Adsorption/Desorption processes H , OH H2O

2.2 2.2.1 2.2.2

External corrosion Hydrolysis Oxidation

2.3

Absorption, swelling

2.4

Diffusion processes

2.5 2.5.1 2.5.2

Internal corrosion Hydrolysis Oxidation

O2, H2O, products

2.6 2.6.1 2.6.2

Sublayer corrosion Without delamination With delamination

O2, H2O, C+, A–

2.7

Sublayer corrosion due to easier swelling and hydrolysis of plasma Si–O

3

Failures due to combined action of mechanical stress and chemical interaction A (H2O, NH3) (A=adsorbens) Stress corrosion cracking (adsorption induced stress corrosion cracking)

H2O

(a) without delamination (b) with delamination

3.1

H2O, O2

SiO2. aq, NH3

H2O

+

O2, H2O, C , A

–

O2, H2O, C+, A–

H2O Si-N Si-O

358

G. Schmitt

100 µm

Figure 13.14 Spalling of plasma Si–N layer due to intrinsic stress after annealing at 400ºC (10 min) (no electrolyte exposure).

metallized surface areas indicating that the adhesion to the chromium interlayer was not appropriate. Figure 13.14 shows in part a triangular electrode on the test electrode array. This type of plasma Si–N could not be improved in its fracture mechanical properties by annealing. Stresses (generally intrinsic stresses) in passivating layers can be induced by the substrate topography, for example, at edges of conducting tracks or other step structures, where the growth morphology of the film is more irregular and nodular growth is more likely to occur than on plain surfaces (Table 13.6, Group 1.3). Optimized chips will have buried conducting tracks and electrodes. Under these conditions the risk of delamination of the passivating layer will be considerably lower, even in case of opened electrodes. On chips with classical non-buried conducting tracks and with opened electrodes delamination of the passivation film was observed after longer exposure times (1350 h) in neutral 1 M NaCl at ambient temperature (Figure 13.15(a–c)). The delamination started at the overlay on the electrode and progressed to the conducting tracks. The close-up in Figure 13.15(c)) clearly shows that the rest of the duplex layer which remained on the electrode surface, buckled and cracked in the corner. This indicates that the layer contained intrinsic compressive stresses which probably increased due to swelling caused by absorption of water. The cracks occurred in the corner, that is, in the area of highest stress intensity. Uptake of water along the interface metallization/ plasma deposited silica stimulates the progress of delamination via hydrolysis reactions. Figure 13.15(d) shows the as-deposited state before the exposure. No cracks and buckling are visible. Figure 13.16 gives further examples of passivation layer performance at opened electrodes for different types of passivation layers. While delamination occurs on non-buried conduction tracks (Figure 13.16(b)) no cracking or spalling

Corrosion of microsystems 359 (a)

(b) 2 µm

30 µm (c)

(d)

2 µm

4 µm

Figure 13.15 Delamination of duplex passivation layer on non-buried conducting track and opened Au electrode after 1350 h in 1 M NaCl at 25ºC (a–c). For comparison: opened electrode before exposure (d).

(a)

(b) Cracks

2 µm

800 µm

(c)

(d)

2 µm

2 µm

Figure 13.16 Chips with opened electrodes (SEM images): (a) duplex passivated chip with non-buried conducting tracks (120 h); (b) duplex passivated chip with non-buried conducting tracks (135 h); (c) SiC passivated chip with nonburied conducting tracks (380 h); (d) triplex passivated chip with buried conducting tracks (2400 h); electrolyte: 1 M NaCl, pH 7, 25ºC.

360

G. Schmitt

or significant corrosion attack is visible at SiC on non-buried conducting tracks (Figure 13.16(c)). The surface appearance of the triplex layer after 2040 h electrolyte exposure (Figure 13.16(d)) is still comparable with the as-deposited state, which demonstrates clearly the superior performance of such type of passivation layer. While extrinsic tensile or compressive stresses can be avoided in passivating layers by appropriate design of the system in which a chip is integrated, it is much more complicated to avoid intrinsic stresses. It has been shown that in PECVD processes intrinsic stresses can be influenced by the deposition parameters.48,57 It is possible to obtain stress-free silicon nitride films by a proper choice of process parameters under given boundary conditions. The formation of pinholes (Table 13.6, Group 1.4) is influenced by the parameters of the PECVD deposition process, for example, the deposition temperature. Thus, for plasma Si–N the pinhole density is less than 1 pinhole per cm2 for deposition temperatures above 300ºC, while it is unacceptably high for deposition temperatures below 300ºC.66 Failures of passivation films due to particle inclusion (Table 13.6, Group 1.5) is generally a matter of clean processing. This includes not only cleanliness in the rooms where the wafers are handled, but also in the PECVD chamber where particles of previous deposition processes could become attached to the chamber walls and fall onto a new wafer during the next passivation run. In this case the particle will be incorporated in the growing passivation film. As an example Figure 13.17 shows particle inclusions in a duplex passivation layer. Group 2 failures include chemical, physicochemical, and electrochemical reaction processes in their mechanisms (Table 13.6). Adsorption of water via hydrogen bridging at hydrogen bonds at the surface of the film, for example, Si–H,

400 µm Blank

150 µm 1100 h

4 µm

Figure 13.17 Particle inclusions in duplex passivation layer.

6 µm

Corrosion of microsystems 361 Si–O–H, Si–N–H, or at oxygen or nitrogen bonds, for example, Si–O–Si or Si–N–Si, is an important first step for further chemical degradation of SiO2 or Si–N films, for example, by hydrolysis or by etching in hydrofluoric acid. External corrosion of PECVD films, that is, degrading the passivation layer from the surface, can be achieved by hydrolysis at different pH values. Acid hydrolysis is the degradation mechanism during etching. Oxidation of Si–N films in most environments (moist air, water vapor in the PECVD chamber) can occur by absorption of water with subsequent hydrolysis of Si–H bonds, thus introducing oxygen into the film (as Si–O–H groups).67 Absorption of small molecules (e.g. water molecules) in the film will cause swelling. This phenomenon can be observed both at organic and inorganic films. Swelling with subsequent formation of blisters and delaminations is the typical failure mode of organic passivations layers. Figure 13.18 gives an example of polyimide films which failed after 12 days in neutral 1 M NaCl solution at room temperature. It was found that blistering, buckling, and delamination of the polyimide film started always on surface areas where the film was directly contacted with the thermal silica layer. Only at a later stage did the polyimide films show blistering over the gold conducting tracks. Detailed investigations proved that the delamination on the metallized surface areas started from the edges of the metallization. It is assumed that blistering, delamination, and buckling of polyimide films over thermal silica is due to osmosis. Diffusion of water through the polyimide film to hydrolysable centers on the silica surface and formation of hydrated or

Swelling and buckling of PI layer PI on SiO2 (Thermal)

PI on Au

100 µm

Borderline of delamination over Au-electrode

Figure 13.18 Swelling and buckling of polyimide (PI) film, predominantly on thermal SiO2 layer.

362

G. Schmitt

dissolved silicic acid at the interface produces a water concentration gradient which is the driving force for enhanced water diffusion. This explains why the polyimide film over the metallized areas is delaminated preferentially from the edges. Also inorganic films can absorb and swell. Thus, plasma PECVD SiO2 films can absorb water which may hydrolyze Si–O–Si bonds to form silanol groups. These silanol groups, due to their high polarity, can bind water molecules and cause swelling of the film increasing the intrinsic compressive stress. An important precondition for internal and sub-layer corrosion is the diffusion of water, oxygen, and/or ions through the film. It is well known that all organic coatings, specifically those with low thicknesses (e.g. less than 1 mm), exhibit only a limited barrier against diffusion of water. The diffusion constant of water in polymers (acceptable as passivation layers) is in the range of 107–105 cm2 s1, which means that it takes from 1 s to 2 h to reach 90% saturation at the substrate surface through a 25-mm thick film.62,68 Polymers are also diffusible for oxygen and smaller ions, for example, sodium or chloride ions. For inorganic thin films such as SiO2 and SixHyNz the diffusion constant is several orders of magnitude less, reaching down to 1018 cm2 s1.62,68 However, as mentioned above, for example, plasma SiO2 films can become more diffusible with increasing concentration of silanol groups, as these groups can bind water molecules and form domains with higher local water concentrations which allow diffusion of more water and ions. Thus, SiO2 passivation films have only a limited barrier function for diffusion of sodium ions. In contrast to plasma SiO2 films plasma Si–N layers provide some resistance to sodium ions and moisture permeation.69,70 Absorption of water can lead to internal corrosion by hydrolysis and oxidation. Water reacts with strained Si–O or Si–N bonds and forms silanol groups. Further hydrolysis reactions eliminate Si and N atoms from the network, for example, in the form of Si(OH)4 and NH3 which are dissolved in the aqueous medium (Figure 13.19). Internal and external corrosion can proceed by the same mechanisms. On exposure to neutral 1 M NaCl at room temperature plasma Si–N films experience materials dissolution which starts with the formation of a high density of small etching pits which grow together to yield general corrosion. Figure 13.20 shows the state of etching pits in a duplex layer which had been exposed for 1100 h in the neutral electrolyte solution. Deliberate breaking of the chip in liquid nitrogen revealed the structure and thickness of the different layers on the chip. The magnification at site 1 shows the sequence of the different layers at the edge of a triangular electrode which had not been opened and is, therefore, still covered with the duplex layer. It was not possible to distinguish between the two types of plasma layers in the crack surface of the duplex layer. At the edge of the metallization the typical nodular growth of the plasma-deposited duplex layer can be observed. A close-up of such nodular growth at step structures here after 1100 h corrosion gives the low left picture in Figure 13.20, which was made from site 2. The low right picture demonstrates that the etching pits have an opening of

Corrosion of microsystems 363

H2O HO HO

OH H Si

N

N

N

Si

SiO2 · aq + NH3

H

Si N N

N

HO OH HO Si

H2O

Si N

H

H

N

Si

N

Si

N

Si

H N

N

N Si

Figure 13.19 Hydrolytic dissalution of plasma Si–N.

600 µm

2 µm

1

Surface

Duplex Si3N4 SiO2

Fracture

Au Surface Therm. Fracture SiO2

1

Surface

2

n-Si

Fracture

600 nm

2

2

200 nm

Figure 13.20 Duplex passivation after 1100 h in 1 M NaCl at 25ºC. Deliberate fracture through chip for fracture surface studies.

100–200 nm and a depth of 100–150 nm. This corresponds to a local corrosion rate of 0.1–0.2 nm/day. During an exposure time of 2600 h these pits grew together and formed a rough surface. Scans carried out with an atomic force microscope (Figure 13.21) illustrate the difference between the smooth surface of a plasma Si–N layer in the as-deposited state and the rough surface of the same type of layer after an exposure time of 7600 h (1 M NaCl, pH 7, 25ºC). The surface appearance of a duplex layer on a buried conducting track after 2600 h in neutral 1 M NaCl at ambient temperature is demonstrated in Figure 13.22. The close-up from the corner of an electrode – in this case buried in the thermal silica layer – shows the slightly rough surface of the generally corroded surface. The groove at the edge of the electrode and the conducting track was caused by the fact that the etch groove in the thermal silica layer had been a little wider than the diameter of the conducting track due to isotropic etching with diluted HF.

364

G. Schmitt (a)

as-deposited nm 30

nm 50 0

0

–50

–30 0.5

200 1.0

400 nm

µm (b)

7600 h nm 50

nm 30

0

0

–50

–30 100

0.5

200

1.0

300

µm

nm

Figure 13.21 AFM scans of duplex passivation (a) as-deposited, (b) after 7600 h in neutral 1 M NaCl at 25ºC.

2600 h

4 µm

300 nm

Figure 13.22 Duplex passivation layers on buried conducting tracks after 2600 h in 1 M NaCl at 25ºC.

Nevertheless, the long lasting barrier effect (more than 1250 h, that is, nearly two months) of the duplex layer on buried conducting tracks and electrodes clearly demonstrates that minimizing or avoiding topographically induced intrinsic stresses at step structures is an important necessity to produce microelectrode sensor arrays with high service life under corrosive environmental conditions. If layer imperfections like pinholes or particle inclusions (Figure 13.17) are avoided, even longer corrosion protection can be achieved with optimized chip design and improved plasma deposition process of the passivation layer.

Corrosion of microsystems 365 If water, oxygen, and ions can diffuse through the passivation layer and reach metal sublayers (e.g. conducting tracks) corrosion of the metallic sublayer can occur. Depending on the adhesion strengths between the layer systems this sublayer corrosion proceeds with or without delamination of the passivation layer. Group 3 corrosion failures are related to the combined action of mechanical stress and chemical interactions and follow the mechanism of stress corrosion cracking. This type of corrosion mechanism has been intensively studied in recent years at silica glass and amorphous silica.71,72 Cracking occurs due to water attack at strained Si–O–Si bonds existing either at the plain surface or at a crack tip. A good understanding of water-induced stress corrosion cracking of amorphous silica was obtained by a ring size distribution analysis in amorphous silica via molecular orbital calculations.73 It was also found that adsorbed ammonia favors cleavage of strained Si–O–Si bonds. Physisorbing molecules such as pyridine also adsorb on active sites but will not directly result in bond cleavage, instead it will block dissociatively adsorbing species from reaction.71 This seems to open up the possibility for inhibition of stress corrosion cracking at passivating amorphous films. The stress corrosion mechanism proposed for amorphous SiO2 should be analogously transferable to amorphous Si–N material. The mechanism underlines the importance of intrinsic stress in passivation layers and explains the fact that the thermal history and the processing greatly affects the cracking susceptibility of amorphous passivation layers.

13.5

Conclusions

The barrier performance of passivation layers to prevent corrosion of sensor chips can be optimized by the chip design, the appropriate selection of the type of passivation layer, and by optimizing the deposition conditions. Organic passivation layers like photoresist or polyimide suffer from water absorption with subsequent swelling, delamination, and formation of blisters. The time to failure was in the range of a few hours. However, in case of polyimide films it can be increased to 400 h by plasma surface treatment. Inorganic passivation layers exhibit a much higher optimization potential, for example, by optimizing the deposition process and the post-deposition treatment conditions, for example, annealing conditions. The best results were obtained with SiO2/Si3N4 duplex and SiO2/Si3N4/SiO2 (ONO) triplex layers, which gave protection for more than 1200 and 1000 h, respectively. However, problems for the film integrity arise from intrinsic stresses (specifically, topographically induced stresses at step structures), water uptake, cracking, and delamination. They can be minimized by burying the conducting tracks in the insulating layer of thermal SiO2. Impairments of the passivation film can be observed by SEM or by decoration with cathodic metal deposition.74 Experiments proved that the stability of amorphous, non-stoichiometric plasma Si–O and Si–N films in aqueous solutions is significantly lower than the corresponding stoichiometric crystalline compounds.

366

G. Schmitt

Reactions at the surface and diffusion of water and electrolyte into the film play an important role.

References 1. ISO 8044 Corrosion of Metals and Alloys – Vocabulary (International Standard Organization, Geneva, Switzerland, 1999). 2. J. D. Sinclair, Indoor Atmospheres (Corrosion Test Standards Manual, ASTM Manual 20, American Society of Testing Materials, Philadelphia, 1995), pp. 295–306. 3. H. C. Shields and C. J. Weschler, J. Air and Waste Management Assoc. 42, 796 (1992). 4. A. J. Muller, L. A. Psota-Kelty, H. W. Krautter, and J. D. Sinclair, Solid State Technology, September, 61 (1994). 5. J. D. Sinclair, J. Electrochem. Soc. 135, 89C (1988). 6. R. P. Frankenthal, D. J. Siconolfi, and J. D. Sinclair, J. Electrochem. Soc. 140, 3129 (1993). 7. J. D. Sinclair, L. A. Psota-Kelty, G. A. Peins, and A. O. Ibidunni, Atmospheric Environment 26A, 871 (1992). 8. D. R. Lide, Handbook of Chemistry and Physics (CRC Press, Inc., Cleveland, Ohio, 1994–1995), 71st edn. 9. G. O. Nelson, Controlled Test Atmospheres (Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, 1972). 10. R. B. Comizzoli, R. P. Frankenthal, R. E. Lobnig, and J. D. Sinclair, The Electrochemical Society Interface 2(3), 26 (1993). 11. B. Comizzoli and J. D. Sinclair, Encyclopedia of Applied Physics 6, 21 (1993). 12. W. H. J. Vernon, Transactions Faraday Soc. 31, 1668 (1935). 13. B. T. Reagor and J. D. Sinclair, J. Electrochem. Soc. 128, 703 (1981). 14. K. G. Schmitt-Thomas, Werkstoffe und Korrosion 42, 381 (1991). 15. S. Wege and M. Hilfer, AVT Report (1991) No. 5. 16. R. G. Baker and A. Mendizza, Electro-Technology 72, 11 (1963). 17. M. J. Elkind and H. E. Hughes, Physics of Failure in Electronics 5, 477 (1969). 18. W. Ehrfeld, V. Hessel, H. Moebius, Th. Richter, and K. Russow, Workshop on Microsystems Technology, Dechema Monographs, Vol. 132, Mainz/Germany, 20–21 February 1995. 19. A. Uhlig, M. Paeschke, K. Schakenberg, R. Hintsche, D.-J. Diederich, and F. Scholz, Sensors Actuators B24–25, 899 (1995). 20. W. Goepel, J. Hesse, and J. N. Zemel, Sensors: A Compresensive Story (VCH-Verlag, Weinheim, 1996), Vols. 2–3, p. 1ff. 21. U. Wollenberger, R. Hitsche, and F. Scheller, Microsystems Technology 1, 75 (1995). 22. G. Schmitt, J. W. Schultze, F. Fassbender, G. Buss, H. Lueth, and M. J. Schoening, Electrochimica Acta 44, 3865 (1999). 23. F. Faßbender, G. Schmitt, M. J. Schöning, H. Lüth, G. Buss, and J.-W. Schultze, EUROSENSORS XIII, Proc. 13th European Conf. On Solid-State Tranceducers, 12–15, September 1999, The Hague, The Netherlands, Paper 25P40 (ISBN 90-76699-02-X). 24. G. Schmitt, F. Faßbender, H. Lüth, M. J. Schöning, J.-W. Schultze, and G. Buß, EUROCORR’99, August 30–September 2, 1999. 25. G. Schmitt, F. Faßbender, H. Lüth, M. J. Schöning, J.-W. Schultze, and G. Buß, Materials and Corrosion 51, 1 (2000). 26. F. Faßbender, G. Schmitt, M. J. Schöning, H. Lüth, G. Buss, and J.-W. Schultze, Sensors and Actuators B68, 128–133 (2000).

Corrosion of microsystems 367 27. G. T. Kohman, H. W. Hermance, and G. H. Downes, Bell System Tech. Journal 34, 11159 (1955). 28. A. Shumka and R. Piety, 13th Ann. Int. Reliab. Physics Symp (1975), p. 93. 29. R. C. Benson et al., IEEE Trans. Compon., Hybrids, Manuf. Technol., CHMT-11, 363 (1988). 30. R. P. Frankentahl and W. H. Becker, J. Electrochem. Soc., 126, 1718 (1979). 31. E. Heitz and E.-M. Horn, in: Korrosion verstehen – Korrosionsschäden vermeiden (Verlag Irene Kuron, Bonn, 1994), Vol. 1, Chapter 7.2, pp. 267; see also: Werkstoffe und Korrosion 31, 157 (1980). 32. H. H. Ulig and R. W. Revie, Corrosion and Corrosion Control (J. Wiley & Sons, New York, 1985), 3rd edn., pp. 347. 33. J. R. Scully, R. P. Frankenthal, K. J. Hanson, D. J. Siconolfi, and J. D. Sinclair, J. Electrochem. Soc. 137, 1365 (1990). 34. J. R. Scully, R. P. Frankenthal, K. J. Hanson, D. J. Siconolfi, and J. D. Sinclair, J. Electrochem. Soc. 137, 1373 (1990). 35. P. A. Totta, J. Vac. Sci. Technol. 13, 26 (1976). 36. D. A. Jones, Principles and Prevention of Corrosion (Macmillan Publ., New York, 1992), Chapter 6. 37. R. R. Tummula and E. J. Rymaszewski, Microelectronis Packaging Handbook (Van Nostrand Reinhold Publ., New York, 1989), pp. 629. 38. D. W. Rice, B. P. B. Philipps, and R. Tremouroux, J. Electrochem. Soc. 126, 1459 (1979). 39. D. W. Rice, R. J. Cappel, W. Kinsolving, and W. Laskowski, J. Electrochem. Soc. 127, 892 (1980). 40. D. W. Rice, B. P. B. Philipps, and R. Tremouroux, J. Electrochem. Soc. 127, 563 (1980). 41. R. R. Gore, R. W. Witska, J. R. Kirby, and J. L. Chao, IEEE Transactions on Compounds, Hybrids, and Manufacturing Technology 13, 27 (1990). 42. W. H. Abbott, IEEE Transactions on Compounds, Hybrids, and Manufacturing Technology 11, 22 (1988). 43. ASTM B 810, American Society of Testing Materials, Philadelphia. 44. ASTM B 827, American Society of Testing Materials, Philadelphia. 45. ASTM B 845, American Society of Testing Materials, Philadelphia. 46. EIA Standard RS-364–50, Electrical Industries Association, Washington, DC, 7 October 1983. 47. MIL-Standard-810E, 14 July 1993. 48. W. A. P. Claassen, W. G. J. N. Valkenburg, M. F. C. Willemsen, and W. M. v. d. Wijgert, J. Electrochem. Soc. 132, 893 (1985). 49. G. Kelm and G. Jungnickel, Mater. Sci. and Eng. A139, 401 (1991). 50. W. A. Lanford and M. J. Rand, J. Appl. Phys. 49, 2472 (1978). 51. W. Y. Lee, J. Appl. Phys. 51, 3365 (1980). 52. I. Solomon, M. P. Schmidt, and H. Tran-Quoc, Phys. Rev. B38, 9895 (1988). 53. A. F. Flannery, N. J. Mourlas, C. W. Storment, S. Tsai, S. H. Han, J. Heck, D. Monk, T. Kinn, B. Gogoi, and G. T. A. Kovacs, Sensors & Actuators A70, 48 (1998). 54. M. Fathalla, R. Gharbi, E. Crovini, F. Demichelis, F. Giorgis, C. F. Pirri, E. Tresso, and P. Rva, J. Non-Crystal Solids 198–200, 490 (1996). 55. J. Huran, J. Safankowa, and. A. P. Kobzev, Vacuum 50, 103 (1998). 56. A. K. Sinha, H. J. Levinstein, T. E. Smith, G. Quintana, and S. E. Haszko, J. Electrochem. Soc. 125, 601 (1978).

368

G. Schmitt

57. D. C. H. Yu, and J. A. Taylor, Mat. Res. Soc. Symp. Proc., Vol. 188, Paper 1202 (1990). 58. A. Stoffel, A. Kovács, W. Kronast, and B. Mueller, J. Micromech. Microeng. 6, 1 (1996). 59. W. Kern, G. L. Schnable, and A. W. Taylor, RCA Reviews 37, 3 (1976). 60. C. W. Lim, C. C. Toh, J. K. Fu, and S. K. Lahiri, J. Mater. Sci. Lett. 15, 2177 (1996). 61. P. M. Sarro, C. R. deBoer, E. Korkmarz, and J. M. W. Iaros, Sensors & Actuators A67, 175 (1998). 62. K. Dyrbye, T. Romedahl-Brown, G. F. Eriksen, J. Micromech. Microeng. 6, 187 (1996). 63. G. F. Eriksen and K. Dyrbye, J. Micromech. Microeng. 6, 55 (1996). 64. M. J. Schöning, G. Buß, F. Faßbender, O. Glück, H. Emons, G. Schmitt, J. W. Schultze, H. Lüth, Sensors and Actuators B65, 284 (2000). 65. R. Robertson, M. I. Manning, Mater. Sci. Tech. 6, 81 (1990). 66. K. Allaert, A. van Calster, H. Loos, and A. Lequesue, Solid-State Sci. Technol. 132, 1763 (1985). 67. W.-S. Liao, C.-H. Lin, and S.-H. Lee, Appl. Phys. Lett. 65, 2229 (1994). 68. R. K. Ulrich, W. D. Brown, S. S. Ang, S. Yi, J. Sweet, and M. D. Peterson, IEEE, 05695503, 738 (1991). 69. J. V. Dalton and J. Drobek, J. Electrochem. Soc. 115, 865 (1969). 70. M. T. Duffy and W. Kern, RCA Reviews 31, 742 (1970). 71. T. A. Michalske and S. W. Freiman, J. Amer. Ceram. Soc. 66, 284 (1983). 72. T. A. Michalske, B. C. Bunker, J. Appl. Phys. 56, 2686 (1984). 73. J. K. West, L. L. Hench, Philosophical Magazine A77, 85 (1998). 74. G. Buß, M. J. Schöning, H. Lüth, and J. W. Schultze, Electrochimica Acta 44, 3899 (1999). 75. C. Haan, Diploma Thesis (Aachen University of Technology, Aachen (1996)). 76. G. Schmitt, F. Faßbender, H. Lüth, M. J. Schöning, and G. Buß, Werkstoffwoche’98, Vol. III, Symposium 7: Materials and Corrosion (A. Kranzmann, U. Gramberg, eds), (Wiley-VCH, Weinheim, New York, 1998), pp. 345–350. 77. M. J. Schöning, G. Buß, F. Faßbender, O. Glück, G. Schmitt, J. W. Schultze, and H. Lüth, Proc. 7th Intern. Meeting on Chemical Sensors, 27–30 July, 1998; Beijing, China, Technical Digests (International Acad. Publishers, Beijing, China), pp. 100–102. 78. F. Faßbender, PhD Thesis (Aachen University of Technology, Aachen, 1999).

Part IV

Microanalysis and microsensors

14 Electrochemical microanalysis Hendrik Emons

14.1

Analytical chemistry and the microworld

At present we are in the midst of a rapid development in several areas of measuring sciences and one of them is certainly the experimental investigation and exploitation of the atomic and molecular structures of matter. Several methodic breakthroughs such as scanning probe microscopies (STM, AFM, etc.) and patch-clamp techniques have sharpened our eyes at the microscopic scale. This development has also been stimulated by demands to one of the basic disciplines of chemistry – analytical chemistry. If one realizes that “Analytical Chemistry is a scientific discipline that develops and applies methods, instruments and strategies to obtain information on the composition and nature of matter in space and time”,1 the importance of analytical developments and investigations at the microscopic scale becomes evident. This concerns now more the quantitative part of chemical analysis, because the structural chemical analysis has been always directed by definition to the qualitative characterization of samples at the molecular/atomic level. The term “microanalysis” has already been used for a long time in analytical chemistry. Originally it was devoted to the quantitative analysis in small sample amounts but the targets of microanalysis have been expanded over the years. An overview of objects which can be studied by microanalytical methods in solution is presented in Figure 14.1. Small sample volumes are targets as well as very low total amounts of the analyte of interest in a larger sample size which requires usually the accumulation of the analyte species in a small sample pretreatment device. Distribution analysis is attracting more attention with the increasing emphasis on the analytical investigation of natural samples in environmental and life sciences. Therefore, chemical functionalities at surfaces and interfaces have to be characterized in biological, geological and technical systems. Further progress in the understanding of biogeochemical cycles and the development of environmentally sustainable anthropogenic activities is also relying on the analysis of liquid/solid microcompartments such as pore water in soil or sediment and of liquid/liquid microdomains such as micelles in solution. Such microenvironments are often subject to dynamic changes and one has to probe the local concentrations at the microscopic level with respect to their temporal changes.

372

H. Emons Microanalysis Distribution analysis

Concentration analysis Integral concentrations of sampled: • Microvolumes • Microamounts

S t a t i c

Heterogeneous macroscopic samples: • Microstructures at surfaces • Microdomains in l/s or l/l systems

Dynamic Concentration profiles: • At fixed position • Spatially resolved

Figure 14.1 Targets of modern microanalysis (l – liquid; s – solid).

Therefore, concentration variations caused by physical effects (mass transport) or localized chemical reactions (e.g. immobilized enzymes) have to be analyzed. This can be performed after positioning the microprobe at a particular location, for instance above a specific biomembrane channel, and during the controlled movement of the probe within or near the sample of interest, respectively. A major stimulation for microanalytical developments arises from the demands of in vivo and single cell analysis.2 Despite the intellectual attraction of pushing measuring capabilities to their limits the driving force of analytical developments has been shifted more to problem-solving approaches in recent years. Therefore, one has to consider the problem-oriented design of the whole analytical process. Its general steps are illustrated in Figure 14.2. Several specific aspects have to be taken into account before a microanalytical technique can be applied for the analytical determination. They will be discussed in the frame of electrochemical microanalysis in the following section. But progress in microanalysis cannot be restricted to the analytical target. Exciting developments have also been reported with respect to the miniaturization of analytical tools.3,4 Analytical Chemistry has been advanced in various areas of sampling, separation as well as detection techniques both by the miniaturization of existing methods such as dialysis, high performance liquid chromatography (HPLC) or fluorescence measurements and by the introduction of new techniques which are only feasible with devices having at least one dimension in the micrometer range such as capillary electrophoresis (CE). A few years ago a new level of miniaturization in analytical instrumentation has been introduced with the concept of “micro-total analysis systems (TAS)”.5,6 The integration of successive analytical operations from sample preparation to the quantification of the analytes of interest into a miniaturized system should offer a number of advantages such as: ● ●

reduced amount of sample and reagents leading also to waste reduction shorter time of analysis because of diminished transport lines

Electrochemical microanalysis 373

Problem

Sampling Sample processing

Storage Analytical sample preparation

Identification Analytical determination

Quality assurance/control

Conservation

Data evaluation Auswertung Assessment Bewertung

Figure 14.2 Scheme of the total analytical process. ● ● ●

higher sample throughput due to parallelized analytical modules improved precision because of smaller dead volumes portability and reduced energy consumption.

Electrochemical methods can play an important role in microanalysis as explained in the following.

14.2

Electroanalysis and microscales

Most of the electrochemical methods are based on interfacial processes as shown in Figure 14.3. This “electroanalytical tree” contains the main electrochemical detection methods which are presently used for analytical purposes. There is a renaissance of techniques which are based on the measurement of electrical conductivity during recent years because of the increasing application of corresponding flow-stream detection methods7 and the development of new sensors.8 Many of the latter are actually exploiting changes of the surface conductivity at the transducer and belong also to the “interfacial branch” in Figure 14.3. Overall the dominant role of interfaces provides a distinct advantage of electrochemical methods in microanalysis: a very thin layer of sample at the electrode is sufficient

374

H. Emons Electroanalysis

Electrodics

Ionics

Dynamic (I ≠ 0)

Static (I = 0)

Conductometry

Potentiometry Controlled current

Controlled potential

Amperometry

Potentiostatic coulometry

AC techniques

AC Impedance voltammetry measurement

Voltammetry

Potential scan

Cyclic Linear sweep voltammetry voltammetry

Chronopotentiometry

Coulometry (galvanostatic)

Potential step

Differential SquareNormal pulse wave pulse voltammetry voltammetry voltammetry

Figure 14.3 “Tree of methods” of electroanalysis.

for sensitive measurements and the power of detection is not restricted by a bulk phase dimension such as for methods based on optical absorption (Beer’s law). On the other hand one has to realize high surface-to-volume ratios of the microsample for powerful electrochemical measurements and the composition as well as stability of the detecting interface is of critical importance for its analytical performance. Before the selection, optimization and application of microelectrochemical methods for the quantification of the analyte of interest (i.e. the “determination” step) one has to design and consider the preceding steps of the total analytical process (Figure 14.2). Sampling is not only the first, but very often also the most underestimated step of the whole analysis. A fundamental prerequisite for meaningful analysis consists in the representativity of the sample with respect to the problem of interest, the selected population of specimens to be studied, the total amount of material to be characterized and its variations in time and space. Therefore, microamounts of sample material which are finally analyzed with microanalytical methods are only useful if one is asking for average concentrations of very homogeneous study material or if one is interested in the spatial distribution of certain analytes in a heterogeneous object. Presently, the majority of analytical determinations is dealing with relatively heterogeneous samples (on the microscale) such as biological tissues and body fluids in clinical analysis, environmental samples such as surface water, sediments or soil in monitoring

Electrochemical microanalysis 375 programs and a broad range of specimens in food control. Therefore, the minimum amount of sample material which is representative for the whole sample lot has to be estimated. Corresponding statistical approaches have been reported in the literature.9,10 Another aspect determining the sample in-weight consists in the power of detection of the measuring principle. Electrochemical methods which are based on faradaic reactions at the electrode, namely amperometry and voltammetry, belong to the analytical detection methods with the lowest limits of detection. This makes them well-suited for measuring microsamples which contain electroactive, that is, oxidizable or reducible, analytes. One can estimate that already 600 000 reacting species (1018 mol) carry in a one-electron process enough charge across the electrochemical interface to obtain a reasonable analytical signal of 10 pA within a typical voltammetric measuring time window of 10 ms. The accumulation of such a number of species in a monolayer at the electrode would only require an electrode area of about 100 m2. Therefore, the theoretical limit of detection does usually not represent the limiting factor for the application of such electrochemical methods in microanalysis. More serious problems can originate from the increasing probability of sample contamination with decreasing sample size. This is a phenomenon which has to be obeyed for all microanalytical procedures, but interfacial detection principles as in electroanalysis are especially sensitive, in that respect. For instance, many materials used in sample preparation and handling still contain heavy metals which can easily be transferred into the microsample because of its large contact area to container walls, etc. in relation to the sample volume. Such contaminants are often electroactive and can give rise to interference signals. Another source of concentration error for liquid microsamples consists in the loss of solvent due to uncontrolled evaporation and the adsorption of analyte species at surfaces of the various devices. Operations of sample preparation belong to the most crucial steps of the analytical process (Figure 14.2) both for large and small samples. They include frequently the separation of matrix components, the transfer of the sample into a measurable state (by dissolution, extraction, filtration, derivatization, etc.) and possibly also the separation and/or accumulation of the various analytes of interest. Electrochemical procedures such as electrolysis or electric osmosis can be part of the preparation scheme, but their realization for the analysis of microsamples is still on the horizon. The development of electrochemical sensors was actually initiated for the purpose of avoiding sample preparation (see also Chapter 16), but their limited stability and lifetime under harsh conditions (such as whole blood, waste water, etc.) is still asking for efficient and reliable miniaturized operations of sample preparation. The identification of the compounds of interest is always necessary for samples with varying chemical composition and represents a major challenge for complex materials such as environmental samples. Unfortunately, the capabilities of methods for structural analysis such as nuclear magnetic resonance spectroscopy, organic mass spectroscopy or infrared spectroscopy are still too limited, in particular, with respect to their sensitivity, for the identification of a few

376

H. Emons

analyte species in complex microsamples. This holds true even if the detection of a single molecule has been realized by laser-induced fluorescence measurements,11 but only in standard solution. Therefore, the qualitative analysis in microdomains constitutes at present a serious limitation for the further development of quantitative microanalytical methods with limited selectivity such as amperometry, voltammetry and, in particular, for conductometry. Electroanalytical methods are mainly used for trace analysis which means analyte concentrations below 100 ppm (1 mmol L1). Because of the complexity of most of the analytical samples a separation step has often to be performed. In microanalysis this can either be performed by the integration of receptor and transducer components into a chemical microsensor or by the on-line coupling of separation and detection units resulting in flow-through systems. The first approach has been extensively discussed elsewhere12 and also the possibilities and limitations of electrochemical detectors for normal-sized HPLC have not only been investigated in research laboratories, but also during their application in commercial instruments for almost 20 years.13,14 Their miniaturization is under development and will be discussed in the following section. At this point it should be kept in mind that a major advantage of electroanalysis consists in the fact that these interfacial methods are species-selective and not only element-selective as other analytical methods with much larger excitation energies (atomic absorption spectrometry, ICP-MS, etc.). Several separation methods are not only applied for the discrimination between different sample components, but also for providing an accumulation of the analyte at the same time. This is of particular importance for microanalysis where the number of species available for quantification is small even at larger concentrations (see Section 14.1). Electrochemical approaches for analyte accumulation have been realized within the broad range of stripping procedures.15,16 This accumulation of an increasing variety of chemical species has pushed the analytical quantification limits into the microdimensions with respect to concentration (for instance 20 ppq or 1013 mol L1 for rhodium17) and absolute amounts.18 Anodic, cathodic and adsorptive stripping voltammetry or chronopotentiometry are also well suited for microanalysis because of the in situ attachment of the analyte to the measuring interface, the working electrode. Therefore, electrochemical stripping analysis at microelectrodes is attracting more attention during recent years.19,20 The necessary calibration represents another challenge for microelectrochemical methods. Most of the “real world” analysis has to be performed on complex samples containing matrix or trace components which influence the electrochemical process at the measuring electrode. Therefore, “macroanalytical calibration approaches” are using either appropriate matrix-matched calibrants or the method of standard additions. For microvolume analysis the dilution of the original sample by adding pure standards has a much more pronounced effect on the solution properties and one should actually mimic the sample composition also for the standard addition technique. A general problem results from the necessity of very precise pipetting of solution volumes in the nanoliter range and below. Despite

Electrochemical microanalysis 377 the considerable progress in this area the pipetting error can still contribute significantly to the total uncertainty budget U of the analytical procedure. This can be expressed as 2 2 2 2 2 2 2 Uanalysis Umaterial Usampling Ustability Utreatment Umeasurement Ucalibration

and has to be estimated according to21 as one of the quality parameters of the analytical procedure. An interesting challenge for the statistical evaluation of microanalytical data is hidden in one of the fundamental prerequisites of Gaussian statistics: very small numbers of analyte species cannot be properly described by a normal distribution using parameters such as arithmetic mean and standard deviation. Therefore, approaches of discrete counting statistics have to be applied and new methods for outlier testing and uncertainty calculations have to be developed. Unfortunately, the analytical performance of many of the microelectroanalytical methods proposed in the literature has not been rigorously validated according to state-of-the-art requirements.22,23 This is a considerable hindrance for their wider acceptance and application in analytical laboratories outside the developing group. Quality parameters such as limit of detection/quantification, sensitivity, selectivity, dynamic range, precision and robustness have to be determined not only for the electrochemical measuring step, but for the whole analytical procedure. Its accuracy has to be checked by analyzing certified reference materials (CRM) with well-known analyte content and intermethodic comparisons with analytical approaches based on different physicochemical principles.24 But the necessity of CRMs for microanalytical techniques has only been recognized recently25,26 and appropriate materials are not available at present. Moreover, different microanalytical methods for dissolved analytes with the detection power of electroanalysis are rarely available. Special sample introduction devices allow the analysis of several heavy metals in microliter samples by atomic absorption spectrometry with electrothermal vaporization27 and possibly also by mass spectrometry after analyte ionization in the inductively coupled plasma (ICP-MS) in the near future.

14.3

Electrochemical devices for microanalysis

During the last years, various microelectroanalytical components and a few prototype systems have been developed. Their intended application was mainly the trace determination of inorganic ions and several small organic molecules such as phenols, catecholamines or saccharides in solution for the purpose of clinical and environmental analysis. Actually most of the developments have focused on the preparation of specific working electrodes or electrode arrangements by using existing microtechnologies for manufacturing. The design of complete measuring setups on the miniaturized scale and the further integration of preceding and subsequent steps of the total analytical process (Figure 14.1) into compact microanalytical systems (TAS) is just in the starting phase.

378

H. Emons

A compilation of existing microanalytical components which are exploiting electrochemical principles is shown in Figure 14.4. One may be surprised that separation techniques such as CE and capillary electrochromatography (CEC) are listed among electrochemical devices. But both are based on the application of an external electric field and, in particular, on the separation effect of the electroosmotic flow for charged analyte species. The most interesting developments are presently directed to the modification of the interface capillary/solution which include the knowledge transfer not only from HPLC (in case of CEC)28 but also from interfacial electrochemistry. Both separation approaches are true microanalytical techniques with sample volumes in the nanoliter range and below. Therefore, miniaturized detectors are necessary for an efficient quantification of analytes in CE, and electrochemical detection at microelectrodes has attracted considerable attention for CE.29 For instance, the amperometric detection of daunorubicin (an antibiotic agent) in human urine has been reported by using a carbon disk electrode as CE detector in the wall-jet configuration achieving a limit of detection of 8 . 107 mol L1.30 The on-line coupling of successive microanalytical operations can even be extended to microdialysis sampling. This has been demonstrated by the in vivo sampling of nicotin and its determination in a CE system with amperometric detection at a carbon fiber electrode.31 Even the injection mode may be varied, as shown in a comparison of capillary batch injection analysis and capillary flow injection analysis with amperometric detection at platinum disk electrodes and potentiometric detection of chloride at an Ag/AgCl microdisk electrode, respectively.32

Batch cells Capillary electrophoresis

• Single microelectrode • Microelectrode array Flow cells

Capillary electrochromatography

Sensors/ test strips

Detectors

• Thin-layer electrode • Wall-jet electrode • Tubular electrode

Scanning probes • Single microelectrode • Multielectrode tip

Figure 14.4 Electrochemical devices for microanalysis.

Conductometric–potentiometric– amperometric/voltammetric

Separators

Surface-modified thin-film electrode

Surface-modified 2D microelectrode

Surface-modified microelectrode array

Electrochemical microanalysis 379 Microelectrochemical flow-stream detection offers also the possibility to combine miniaturized analytical operations directly on a chip. First examples have been reported recently, such as the determination of chlorophenolic compounds in river water with a CE chip including amperometric detection at a screen-printed thick-film carbon electrode coated with gold,33 and the analysis of sucrose and glucose on a microchip-based CE system with amperometric detection at a copper microdisk electrode.34 Obviously, the geometric features, the power of quantification and the reduced convection dependence35 of single microelectrodes, in particular, microdisk and microfiber designs, are advantageous for incorporating them into capillary-based analytical systems. But microelectrode arrays offer also favorable mass transfer characteristics (for fundamentals, see the Chapter 2) in flow-through systems of more conventional size such as HPLC. Moreover, they can be individually addressed with respect to the applied detection potential and different detector cells of the thin-layer and wall-jet configuration have been designed. For instance, Lenigk et al.36 have constructed a microflow-through system with a dead volume below 1 L and four platinum microelectrode assemblies each consisting of 192 microhole electrodes (3.6 m in diameter). Microelectrodes are also increasingly used as detecting devices in batch cells, that is, for measurements in quiescent solution. From the point of microanalysis they are mainly applied for trace and ultratrace determinations with the help of stripping voltammetry and stripping chronopotentiometry (also called “potentiometric stripping analysis”).19,20 But the analytical signal is often very small (in the low pA range) at single microelectrodes which prevents their use in analytical laboratories outside the specialized R&D environment. Therefore, microelectrode arrays have been developed for various stripping procedures. For example, Nolan and Kounaves37 have determined ultratraces of Cu2 and Hg2 by anodic stripping voltammetry at a microlithographically fabricated array of 564 iridium microdisk electrodes (each 12 m in diameter). Silva et al.38 have described the electrochemical and analytical performance of a microdisk array consisting of 900 iridium electrodes (5 m in diameter) and plated mercury on them for the stripping voltammetry of heavy metals. A systematic comparison of different excitation signals for the voltammetric detection step has been performed by Bond et al.39 also under the aspect that oxygen removal and stirring of solution can be avoided in field-based voltammetric stripping analyzers with microelectrode arrays as working electrode. Specially designed nanometer-thin working electrodes allow also the detection of several heavy metals by electrochemical stripping analysis with measurements of the surface resistance instead of the faradaic current at the working electrode.40 This alternative detection principle is called “voltohmmetry”.41 Moreover, first attempts have been reported to apply iridium-based microelectrode arrays with 64 individually addressable lines for the investigation of concentration profiles at liquid–liquid and liquid–solid interfaces (as models for natural aquatic systems) with spatial resolution by simultaneous recording of 64 complete voltammograms.42 This represents an alternative approach for

380

H. Emons

achieving local resolution with electrochemical detection in a sampling environment which is not suitable for scanning probe techniques such as scanning electrochemical microscopy (see also Chapter 3). The analysis of microenvironments by the positioning of single microelectrodes represents certainly one of the most exciting applications of electrochemical microdetection. Amperometric and voltammetric measurements at a carbon-fiber electrode which was placed in contact with a chromaffin cell or near a synaptic gap can now be realized for monitoring the neurotransmitter release in vivo.43 Individual exocytosis events of a single cell have been identified by voltammetric measurements of only 1019 mol (about 60 000 analyte species) at a carbon disk electrode of 5 m in diameter.44 In environmental analysis the local positioning of an amperometric microsensor (tip diameter 25 m) for the determination of H2S in pore water of marine sediments has been reported.45 On the other hand, electrodes with small overall dimensions allow the quantitative analysis in microvolumes of sample, such as in single raindrops46 or in microvials with sizes down to 1 pL.47 The possibility of chemical measurements in ultrasmall sample containers offers the opportunity of characterizing samples which are only available in small sizes and it has also an important consequence for ultratrace analysis: the rapid dilution of the sample by diffusion is physically restricted and ultraamounts are becoming detectable at the microprobe surface. Electrochemical measuring principles have not only been developed at bare metal or carbon microelectrodes, but also at electrode surfaces modified with a chemical or biological receptor component. The resulting chemical sensors or biosensors represent basically a close integration of separator and detector functions (Figure 14.4) into a single device. If they are not designed as reversible systems but rather disposable for single use, such products are often also called “test strip”. Several examples for electrochemical microsensors and the application of various microtechnologies for their production are presented in Chapters 15 and 16.

14.4

Perspectives

All in all electrochemical microanalysis is still a very young and emerging field. From the beginning it is in heavy competition with other microanalytical approaches such as laser induced fluorescence, etc. Nevertheless, the inherent advantages of combining electrochemistry and analytical chemistry at the microscale can open new opportunities for satisfying demands in several areas as shown in Figure 14.5. For the next decade one can expect two main routes of developments: on the one hand, advances in the field of chemical microsensors with a tendency towards the molecular design of electrode surfaces, that is, the realization of the bottom-up principle in sensor fabrication. On the other hand, electrochemical components for TAS (“lab-on-chip”), in particular, miniaturized flow detectors, will progress. One may have doubts about currently projected numbers of items for specific microanalytical systems and their real need because of reasons discussed in Sections 14.1 and 14.2. But miniaturized standard

Electrochemical microanalysis 381

In vivo analyzer Homecare monitors

Clinical bedside analyzer

Life Environment Industry

Food analyzer

Product analyzer

Water monitors

Process monitors

Figure 14.5 Vision of future applications for electrochemical microanalysis.

components for chemical analyzers will facilitate the problem-specific design of the whole systems in an efficient and economic manner and will stimulate the more widespread use of such devices. Another promising direction is just starting with the development of sensor arrays of “macroscopic dimensions” which are composed of a broad range of different single microsensors using different electrochemical detection principles (potentiometry, voltammetry, conductometry) such as the “hybrid electronic tongue”.48 In general, microarrays of sensors will allow the simultaneous measurement of different analytes in microvolumes, the increase of analytical selectivity and the identification of components in complex samples by “fingerprint analysis” as well as improved analytical reliability because of redundant signals.49 Miniaturized analyzers and monitors for in-field use will be applied in food control as well as agricultural and industrial production control in the future. Electroanalytical methods should contribute for instance to the on-line process control of corrosion and catalytic effects. Microelectroanalysis for environmental monitoring will focus on the further development of portable analyzers for largescale on-site screening and the installation of permanent monitoring systems for the control of drinking water, ground water and waste water. From the point of view of analytical targets one can envisage that the most promising contributions of microelectroanalysis will take place in the areas of ultratrace determinations of small molecules and ions, real time monitoring and the quantification of analytes with spatial resolution both for synthetic materials and natural samples. Electrochemical detection principles should especially play their role for the trace determination of electroactive compounds and the evolving part of speciation analysis in waters and body fluids, which discriminates

382

H. Emons

between different oxidation states of metals and metalloids. This will be important for human health control because of the large variations in toxicity of the different redox species of the same element. Miniaturized electroanalytical devices will certainly be pushed by increasing demands from health care, ranging from acute surgery control in the hospital to permanent biomonitoring devices for urine control at home and workplace monitors for occupational health control. The research in genomics and proteomics and the development of new pharmaceuticals require a large variety of high-throughput screening methods which can only be realized with miniaturized analyzers50 and electrochemical components will be at least partially integrated. Overall electrochemical microtechnologies possess promising potentials with respect to future investigations of the microworld as analytical target as well as for their utilization to advance miniaturized analytical devices and instrumentation and more of them will be revealed in the coming years.

References 1. Division of Analytical Chemistry, Federation of European Chemical Societies, in R. Kellner et al. (eds), Analytical Chemistry (Wiley-VCH, Weinheim, 1998). 2. R. M. Wightman, P. Runnels and K. Troyer, Anal. Chim. Acta 400, 5 (1999). 3. A. Manz and H. Becker (eds), Microsystem Technology in Chemistry and Life Sciences (Springer-Verlag, Berlin/Heidelberg, 1999). 4. G. Henze, M. Köhler and J. P. Lay (eds), Umweltdiagnostik mit Mikrosystemen (Wiley-VCH, Weinheim, 1999). 5. A. Manz, N. Graber and H. M. Widmer, Sens. Actuat. B 1, 244 (1990). 6. A. Manz, D. J. Harrison, E. M. J. Verpoorte, J. C. Fettinger, A. Paulus, H. Ludi and H. M. Widmer, J. Chromatogr. 593, 253 (1992). 7. J. Weiß, Ion Chromatography, 2nd edn. (Wiley-VCH, Weinheim, 1994). 8. K. Cammann, B. Ahlers, D. Henn, C. Dumschat and A. Shulga, Sens. Actuat. B 35–36, 26 (1996). 9. P. Gy, Sampling for Analytical Purposes (J. Wiley & Sons, Chichester, 1998). 10. J. Pauwels and C. Vandecasteele, Fresenius’ J. Anal. Chem. 345, 121 (1993). 11. C. Zander, Fresenius’ J. Anal. Chem. 366, 745 (2000). 12. A. J. Bard, Integrated Chemical Systems (Wiley-Interscience, New York, 1994). 13. K. Stulik and V. Pacakova, Electroanalytical Measurements in Flowing Liquids (Ellis Horwood, Chichester, 1987). 14. S. M. Lunte, C. E. Lunte and P. T. Kissinger, in P. T. Kissinger and W. R. Heineman (eds) Laboratory Techniques in Electroanalytical Chemistry, 2nd edn. (Marcel Dekker, New York, 1996), pp. 813–854. 15. J. Wang, Stripping Analysis (VCH Publ., Weinheim, 1985). 16. K. Brainina and E. Neymann, Electroanalytical Stripping Methods (J. Wiley & Sons, New York, 1993). 17. C. Leon, H. Emons, P. Ostapczuk and K. Hoppstock, Anal. Chim. Acta 356, 99 (1997). 18. H. Emons, in H. Günzler et al. (eds), Elementaranalytik (Springer-Verlag, Berlin, 1996), pp. 191–219. 19. J. Wang, B. Tian, J. Wang, J. Lu, C. Olsen, C. Yarnitzky, K. Olsen, D. Hammerstrom and W. Bennett, Anal. Chim. Acta 385, 429 (1999).

Electrochemical microanalysis 383 20. C. M. A. Brett, Electroanalysis 11, 1013 (1999). 21. S. L. R. Ellison, M. Rosslein and A. Williams (eds), EURACHEM/CITAC Guide: Quantifying Uncertainty in Analytical Measurement (Laboratory of the Government Chemist, London, 2000). 22. R. J. Mesley, W. D. Pocklington and R. F. Walker, Analyst 116, 975 (1991). 23. EURACHEM/CITAC Working Group, Quality Assurance for Research and Development and Non-routine Analysis (LGC, Teddington, 1998). 24. H. Emons, P. Ostapczuk, M. Rossbach and J. D. Schladot, Fresenius’ J. Anal. Chem. 360, 398 (1998). 25. M. Rossbach and K.-H. Grobecker, Accred. Qual. Assur. 4, 498 (1999). 26. M. Rossbach, P. Ostapczuk and H. Emons, Fresenius’ J. Anal. Chem. 360, 380 (1998). 27. A. Baade, PhD Thesis, University of Essen (1999). 28. H. Sawada, K. Jinno, Electrophoresis 20, 24 (1999). 29. F.-M. Matysik, Electroanalysis 12, 1349 (2000). 30. Q. Hu, L. Zhang, T. Zhou and Y. Fang, Anal. Chim. Acta 416, 15 (2000). 31. J. Zhou, D. M. Heckert, H. Zuo, C. E. Lunte and S. M. Lunte, Anal. Chim. Acta 379, 307 (1999). 32. U. Backofen, F.-M. Matysik, W. Hoffmann and H.-J. Ache, Fresenius’ J. Anal. Chem. 362, 189 (1998). 33. J. Wang, M. P. Chatrathi and B. Tian, Anal. Chim. Acta 416, 9 (2000). 34. C. G. Fu and Z.-L. Fang, Anal. Chim. Acta 422, 71 (2000). 35. F.-M. Matysik and H. Emons, Electroanalysis 4, 501 (1992). 36. R. Lenigk, H. Zhu, T.-C. Lo and R. Renneberg, Fresenius’ J. Anal. Chem. 364, 66 (1999). 37. M. A. Nolan and S. P. Kounaves, Anal. Chem. 71, 3567 (1999). 38. P. R. M. Silva, M. A. El Khakani, M. Chaker, G. Y. Champagne, J. Chevalet, L. Gastonguay, R. Lacasse and M. Ladouceur, Anal. Chim. Acta 385, 249 (1999). 39. A. M. Bond, W. A. Czerwinski and M. Llorente, Analyst 123, 1333 (1998). 40. O. Glück, M. J. Schöning, H. Lüth, A. Otto and H. Emons, Electrochim. Acta 44, 3761 (1999). 41. H. Emons, B. Hüllenkremer and M. J. Schöning, Fresenius’ J. Anal. Chem. 369, 42 (2001). 42. J. Pei, M.-L. Tercier-Waeber, J. Buffle, G. C. Fiaccabrino and M. Koudelka-Hep, in H. Emons and P. Ostapczuk (eds) Electroanalysis (Forschungszentrum Juelich GmbH, Juelich, 2000), p. B04. 43. R. M. Wightman, P. Runnels and K. Troyer, Anal. Chim. Acta 400, 5 (1999). 44. K. D. Kozminski, D. A. Gutman, V. Davila, D. Sulzer and A. G. Ewing, Anal. Chem. 70, 3123 (1998). 45. P. Jeroschewski, C. Steuckart and M. Kühl, Anal. Chem. 68, 4351 (1996). 46. H. Emons, A. Baade and M. J. Schöning, Electroanalysis 12, 1171 (2000). 47. R. A. Clark, P. B. Hietpas and A. G. Ewing, Anal. Chem. 69, 259 (1997). 48. F. Winquist, S. Holmin, C. Krantz-Rülcker, P. Wide and I. Lundström, Anal. Chim. Acta 406, 147 (2000). 49. O. Shulga and K. Cammann, in G. Henze, M. Köhler and J. P. Lay (eds.), Umweltdiagnostik mit Mikrosystemen (Wiley-VCH, Weinheim, 1999), p. 18. 50. S. C. Jakeway, A. J. de Mello and E. L. Russell, Fresenius’ J. Anal. Chem. 366, 525 (2000).

15 Novel approaches to design silicon-based field-effect sensors M. J. Schöning

15.1

Introduction

Miniaturizable chemical sensors and biosensors have attracted more and more attention during the last few years as analytical devices.1,2 The main reason is that two successful technologies were combined in the late seventies, the technique of solid-state integrated circuits, for example, the field-effect transistor (FET), and the manufacturing of conventional ion-selective electrodes.3 This means that it is now possible to fabricate chemically sensitive materials by means of methods of microelectronic technology. As a result, such silicon-based sensors can be fabricated as small, rugged and reliable chip devices with a broad field of applications in medicine, biotechnology, food analysis and environmental monitoring.4 The integration of an ion-selective membrane with a FET leads to chemically sensitive solid-state devices, which can be distinguished into two main groups: Chemically sensitive field-effect transistors and chemically sensitive capacitors.5 The chemically sensitive field-effect transistor (ChemFET) reacts sensitively towards ions or biomolecules. In this case, it is titled as ion-sensitive field-effect transistor (ISFET), or biologically sensitive field-effect transistor (BioFET) and enzyme field-effect transistor (EnFET), respectively.6–9 If ChemFETs are sensitive to gases, they are called GasFETs (gas-sensitive field-effect transistors).10,11 In contrast, chemically sensitive capacitors, the so-called EIS (electrolyte-insulator-semiconductor) structures are much simpler to fabricate, since usually no photolithographic process steps are required.12–14 Their build-up corresponds to the gate region of the ChemFET and they also utilize the capacitive field-effect as the measuring principle (see also Section 15.2). Both sensor types represent potentiometrically operating devices, and in spite of the long period of research and development of more than twenty years, they still suffer from a number of problems and limitations. For example, most of these sensors are exposed to a chemically aggressive environment, and consequently, a high, long-term stable encapsulation of the electronics from the analyte is required. In some cases, the unstable attachment and insufficient mechanical adhesion of the sensitive layer on top of the electronic devices leads to instabilities that produce high sensor drifts and decrease the sensor performance.15–17 Moreover, with regard to the determination of lowest concentrations in trace

Novel sensors 385 analysis the sensitivity and the lower detection limit have to be improved.18 Therefore, the Institute of Thin Film and Ion Technology (Research Centre Jülich) is working on novel silicon-based sensor concepts of chemically sensitive solidstate devices that deal with the improvement of the characteristics of potentiometric sensors, like stability in the long term, the stabilization of biomolecules onto silicon surfaces and the employment of intact chemoreceptors for highly sensitive bioelectronic approaches. In order to realize a highly long-term stable, capacitive pH sensor, the pulsed laser deposition process (PLD) is suggested as a novel thin-film preparation method for chemically sensitive layers. Besides its high flexibility, the main advantages of this process are the well-defined and stoichiometric deposition of even complicated target compositions as well as the short deposition times due to the high growth rates.19–20 Moreover, the thin-film preparation can be performed in various atmospheres (O2, N2, etc.) at elevated temperatures up to 1200ºC, and depending on the wavelength and the pulse energy, a large variety of materials have yet been deposited.21 The presented pH sensor consists of a capacitive EIS structure of Al/p-Si/SiO2/Al2O3 or Al/p-Si/SiO2/Ta2O5. The SiO2 layer was prepared by means of thermal oxidation, and the pH-sensitive layers were deposited by the PLD process. In a second approach also, based on capacitive EIS sensors, porous silicon has been investigated for the purpose of application as a substrate material for potentiometric biosensors. Porous Si and its preparation by an anodic etching process in concentrated hydrofluoric acid (HF) solutions has been known for many years.22 The formation of porous Si in terms of its practical application as a transducer material clearly depends on the preparation conditions such as the doping concentration of the Si, the HF acid concentration, the anodization current, etc.23,24 Consequently, the microscopic structure of porous Si can be classified into microporous (d 2 nm), mesoporous (2 d 50 nm) and macroporous (d 50 nm), where d describes the mean pore diameter.25 Porous Si has become a promising material for electrooptical investigations, for gas-sensing devices with regard to humidity determination and even for Si-based capacitors.26–29 In the presented application, porous Si exhibits the advantages of the protected embedment of biomolecules as sensing material inside the pores and the enlargement of the active sensor area due to the porous structure.30 The microscopic ‘sponge-like’ structure with pores of easily adjustable size allows the tailoring of a matrix that serves as a support for a large variety of biologically and chemically sensitive membrane types. The increase of the sensor surface provides an increase in the measuring value – the capacitance – which is essential when the miniaturization of the porous EIS sensor to a porous EIS microsensor is considered. Porous pH sensors based on n-Si/SiO2/Si3N4 have been developed. The macroporous n-Si layer has been realized by means of an anodic etching process, the double-insulating layer of SiO2/Si3N4 has been deposited by chemical vapour deposition. Adsorptively immobilized penicillinase yielded a porous biosensor, and by means of photolithographical patterning macroporous microsensors with ‘spot’ sizes down to 10 m 10 m have been successfully fabricated.

386

M. J. Schöning

To simultaneously detect different species by utilizing the same measuring set-up, a capacitive EIS sensor array has been proposed. This array was exemplarily demonstrated for the combination of a pH- and a penicillin sensor.31 In this case, the enzyme penicillinase was immobilized in a stable manner by use of a heterobifunctional cross-linker onto a planar p-Si/SiO2/Si3N4 EIS structure.32 Similar as for ISFETs, this EIS arrangement that is based on the C/V measuring technique offers the possibility of eliminating disturbing effects such as thermal and light sensitivity.3,7,33 All these capacitive chemical sensors and biosensors use specifically designed sensor layers as transducer materials in order to quantitatively determine analyte compositions. Beneath them, there exists also a strong interest in highly sensitive and selective potentiometric biosensors. One possibility is to utilize insects’ extraordinary sensory abilities. Insect antennae are known as specific and highly sensitive detectors tuned to perceive, for example, sex pheromones and host plant odours in gaseous analytes. To study this phenomenon, electroantennographic measurements of the electric activity of insect antennae have already been performed.34–35 Considering that the detection limit of the analytes is typically in the ppb range or even lower, such sensors could serve as analytical devices with unrivalled data acquisition time. However, those measurements suffer from the fact that a highly sophisticated set-up for connecting the antenna to an electrical amplifier was required. One promising strategy is to develop a biosensor that includes the chemoreceptive organ of the insect on the one hand, and the miniaturized preamplification of the antenna signal on the other hand. Therefore, a direct FET – insect antenna junction that represents the first BioFET on the basis of an intact insect antenna that has been designed.36 Two types of preparation of the biocomponent were examined yielding a ‘whole-beetle’ BioFET and an ‘isolated-antenna’ BioFET. This hybrid biosensor also belongs to the first type of a bioelectronic nose. The realization of a hybrid biosensor requires a specifically adapted signal interfacing, where the working point of the FET represents one of the most important parameters. As a model system, the insect antennae of the potato beetle and the steelblue jewel beetle were connected by means of an electrolyte to the gate of a n-channel FET with a double-layer insulator of SiO2/Si3N4. The number of applications of such a biosensor reaches from agricultural problems in integrated pest management, like the detection of insect pheromones or plant damages up to the food quality assessment, the medical research and the fire detection.37–38

15.2 15.2.1

Theory of silicon-based field-effect sensors Capacitive (electrolyte-insulator-semiconductor (EIS)) sensor

The theory of capacitive EIS sensors is very well developed and can be derived from a fundamental metal-insulator-semiconductor (MIS) structure, a most useful tool in order to study semiconductor surfaces.39 This device consists of a parallelplate capacitor, where both electrodes (the metal and the semiconductor) are

Novel sensors 387 separated by an insulating layer (dielectric layer). The MIS structure is in equilibrium when the Fermi level in the metal and the Fermi level in the semiconductor are equal. Therefore, when analysing the MIS structure one has to consider the calculation of the charge distribution, the electric field distribution and the potential distribution within this capacitor depending on the externally applied voltage difference between the semiconductor and the metal. In case of the capacitive EIS sensor the metallic gate electrode of the MIS structure is replaced by an electrochemical system, consisting of the sensitive layer, the analyte (electrolyte) and the reference electrode.3 The schematic buildup of an EIS sensor and the measuring principle is given in Figure 15.1(a). The sensor consists of a p- or n-type semiconductor (silicon) covered by a thermally grown SiO2 insulating layer (100 nm) and the sensitive layer that is directly immersed into the analyte. Usually, the sensor is contacted via a conventional reference electrode. The properties of this sensor can be explained by the charge carrier distribution at the insulator/semiconductor interface, which is controlled by both the external dc voltage (VB) and the electrochemical interaction between the analyte and the sensitive layer (V). For a p-Si substrate, a negative VB (VB 0) on the reference electrode accumulates mobile charge carriers (i.e. positive holes) at the Si/SiO2 interface (accumulation). When VB becomes positive (VB 0), the

(a)

Reference electrode (e.g. Ag/AgCl)

Electrolyte (e.g. test solution)

VB

V Sensitive layer

∆V Insulating layer

p-Si Contact (e.g. Al)

C/V Voltage (V )

Capacitance (C )

(b)

Conc.

VB

Concentration

Figure 15.1 Schematic diagram of (a) a capacitive EIS (electrolyte-insulator-semiconductor) sensor; (b) measuring principle of the EIS sensor in the C/V (capacitance/voltage) mode and resulting calibration curve.

388

M. J. Schöning

holes are displaced from the interface, forming a space charge region (depletion) at the semiconductor interface. If the potential gets more positive (VB >> 0), an inversion layer of accumulated electrons at the interface is generated (inversion). The electrical behaviour is given by the small-signal capacitance of the EIS structure. Depending on the applied voltage VB and a superimposed ac voltage (e.g. 1 kHz, 20 mV), a characteristic C/V (capacitance/voltage) curve results (Figure 15.1(b), left). The integral capacitance C that corresponds to VB is: 1 1 1 1 C CL CI CS

(1)

where CL, CI, and CS are the capacitance values of the sensitive layer, the insulator, and the space charge region, respectively, with: C

␧0␧R A d

(2)

where A is the area, d the thickness, ␧R the dielectric permittivity and ␧0 the dielectric constant. Due to the electrochemical interaction (V), a horizontal shift of the C/V curve is provided, depending on the change of the ion concentration in the electrolyte. As resulting measuring signal (calibration curve), this shift can be evaluated at a fixed capacitance value within the linear region of the C/V curves (e.g. 60% of the maximum capacitance, Figure 15.1(b), right). Thus, a relationship between the analyte concentration and the output voltage VB exists. Using a feedback circuit, the measured capacitance can be adjusted at a fixed value in the Concap (constant capacitance) mode and thus, concentration changes can be dynamically recorded.40

15.2.2

Biologically sensitive field-effect transistor

Biologically sensitive field-effect transistors can react sensitive to biomolecules in aqueous media or gases. They incorporate the sensitive layer directly on the gate area of a FET. A schematic diagram of a BioFET with a SiO2 gate insulator (about 100 nm thickness) and the biologically sensitive layer is given in Figure 15.2(a). The sensor is immersed into the test solution and contacted via a reference electrode. In this configuration a potential shift (V) occurs at the interface between the biologically sensitive layer and the electrolyte depending on the concentration of the analyte, which results in a change of the drain current ID flowing between source and drain. In principle, the analyte also consists of an aqueous solution or a gaseous medium. After calibration of the BioFET with a standard solution of known ion activity, the variation of ID can be used to determine the ion concentration in the test solution (Figure 15.2(b)). In the presented approach as the biocomponent, an intact insect antenna of, for example, a potato beetle is directly connected via an electrolyte to the gate of the FET. The realization

Novel sensors 389

(a)

Reference electrode (e.g. Ag/AgCl)

VG Biologically sensitive layer

Electrolyte Insulating layer ∆V p-Si

Source (n+–Si) (b)

ID

Drain (n+–Si)

VDS

ID conc.1

conc.2 ∆V

VG

Figure 15.2 (a) BioFET (biologically sensitive field-effect transistor) configuration, and (b) input characteristic. The metallic electrode of a MOSFET (metal-oxide-semiconductor FET) is replaced by the configuration biologically sensitive layer/electrolyte/reference electrode (VG: gate voltage, VDS: drain source voltage, ID: drain current).

of such a hybrid biosensor requires a specifically adapted signal interfacing, which will be discussed in detail in Section 15.5.41 The operation principle of BioFETs can be derived from the essential electronic behavior of MOSFET (metal-oxide-semiconductor FET) devices,5 where the drain current ID is expressed by the equation:

冢

ID Kd (VG VT)VDS

2 VDS 2

冣

(3)

for the nonsaturated region (VDS VG VT), and ID Kd ·

(VG VT)2 2

(4)

for the saturated region (VDS VG VT) with Kd ␮

␧i b b ␮Ci L di L

(5)

390

M. J. Schöning

The proportional coefficient Kd includes the geometric factors that influence the signal behaviour (input and output characteristics) of the MOSFET; ␮ is the mobility of the electrons in the channel between source and drain, b is the width, and L is the length of the channel. Ci represents the gate capacitance of the insulator per unit area. The gate capacitance, and consequently the gate region of the BioFET exactly corresponds to the build-up of capacitive MIS structures in Section 15.2.1. In the Eqs (4) and (5) the threshold voltage VT is the needed gate voltage in order to create a conductive channel between source and drain (i.e. when an inversion layer at the surface of the semiconductor is formed). For example, a positive gate voltage VG is applied in Figure 15.2(a) that causes an n-inversion layer between the two n-regions (highly n-doped silicon) of source and drain. The additional positive drain source voltage VDS controls the measured current in a kind that the FET is operated in the saturated region. Thus, a small change in VG results in a significant change in ID.

15.3 15.3.1

Sensor fabrication and measurement Sample preparation of the EIS structures

Two different types of pH-sensitive EIS structures, planar samples consisting of p-Si/SiO2/Al2O3 and p-Si/SiO2/Ta2O5, respectively, and porous samples built-up of n-Si/SiO2/Si3N4 have been realized. For both approaches novel techniques for the sensor fabrication have been applied: PLD process and porous silicon technology. The planar EIS sensors were fabricated from p-type Si (18–24 cm) with (100)-orientation. The SiO2 layer was thermally grown, the pH-sensitive layer of Al2O3 was deposited by means of the PLD process.42 A schematic diagram of the sensor and the deposition process is shown in Figure 15.3(a) and (b). In the PLD process, the layer deposition takes place with a KrF excimer laser in oxygen atmosphere. Detailed process parameters are listed in Figure 15.3(c). N-type silicon (10–15 cm, (100)-orientation) was used as substrate material for the porous EIS sensors. A schematic diagram of this sensor type is given in Figure 15.4(a). The porous silicon layer was formed by an anodic etching process (Figure 15.4(b), (c)) in an ethanol/HF solution. Layers of SiO2 and Si3N4 were deposited by means of plasma-enhanced chemical vapour deposition (PECVD). The metallic rear side contact was deposited by vacuum evaporation.43 For both the planar and the porous samples penicillin biosensors have been realized by immobilizing the enzyme penicillinase (EC 3.5.2.6) from bacillus cereus (Sigma) onto the pH-sensitive structures.30,32 15.3.2

Sample preparation of the BioFET

The BioFETs were fabricated by means of standard semiconductor technology. Figure 15.5(a)–(d) present selected process steps of the BioFET preparation. Each chip contains ISFETs and MOSFETs with identical gate dimensions and different channel layouts such as linear gates, U-shaped gates and meander

Novel sensors 391 (a)

Electrolyte pH-sensitive layer SiO2 p-Si Al

Vacuum chamber

(b) Lens

Pulsed laser

Target

SiC heater

Substrate

(c) Energy density : Wavelength : Pulsed period : Pulse frequency : Substrate : Target : Deposition rate : Oxygen pressure : Substrate temperature :

2–4 J/cm2 248 nm (KrF-Excimer) 40 ns 10 Hz Si/SiO2 Al2O3, Ta2O5 1–5 nm/s 2–4 * 10–3 mbar 20–1200°C

Figure 15.3 Schematic diagram of (a) a planar pH-sensitive EIS structure; (b) principle of the pulsed laser deposition (PLD) process; (c) process parameters.

gates.41,44,45 As the biocomponent, the insect antenna of the Colorado potato beetle (Leptinotarsa decemlineata Say) or the steelblue jewel beetle (Phaenops cyanea), respectively, was directly connected to the FET device via an electrolyte solution (Figure 15.5(d)). In the experiments two different configurations have been investigated: the wholebeetle BioFET (Figure 15.6(a)) and the isolated-antenna BioFET (Figure 15.6(b)). In Figure 15.6(a) a 100-m platin wire is placed between the neck and the head of the beetle as a reference electrode in order to adjust a defined gate voltage. In Figure 15.6 (b) the antenna is removed from the beetle and mounted in a home-made

392

M. J. Schöning (a) Si3N4 SiO2 n-Si

Contact

(b)

Pt electrode

Current source

PTFE cell

Cu contact O-ring

Porous Si Si sample Electrolyte (c)

* Anodic etching in a 1:1 mixture of 50% HF and pure ethanol, current density 6.4 mA/cm2 * Deposition of the SiO2 layer and Si3N4 layer by means of PECVD (plasmaenhanced chemical vapour deposition)

Figure 15.4 Schematic diagram of (a) a porous pH-sensitive EIS structure; (b) principle of the anodic etching process; (c) process parameters.

antenna holder. Both ends of the antenna are dipped into the electrolyte that is connected to the FET gate and to an Ag/AgCl reference electrode, respectively. 15.3.3

Measuring conditions

In the case of the capacitive EIS structures, typical sensor properties such as pH sensitivity, stability and selectivity, have been performed by C/V measurements with an impedance analyser (Zahner Elektrik) at a frequency of 120 Hz.46 In order to investigate the sensor characteristics of the BioFET, it is operated at the constant voltage mode.41

(a)

(b) Si3N4 SiO2

SiO2 n+ Drain

+

+

n Source Substrate (p-silicon)

n Drain

n+ Source Substrate (p-silicon)

(d) Antenna Electrolyte

(c) Passivation

Al Ti Si3N4

Al Ti Si3N4 n+ Drain

n+ Drain

n+ Source Substrate (p-silicon)

n+ Source Substrate (p-silicon) Al

Al

Figure 15.5 Selected process steps of the BioFET preparation: (a) defining the source and drain area and n-doping; (b) defining of the gate and SiO2/Si3N4 deposition; (c) defining of the drain and source contacts, metallisation and passivation; (d) FET-antenna junction by means of a haemolymph Ringer’s solution.

(a)

(b)

VG

Reference electrode

Reference electrode

‘Odour’

VG Antenna

Antenna ‘Odour’

+ VDS

+ VDS

Potato beetle Electrolyte D

S n-Si

n-Si p-Si

FET

ID Electrolyte S

ID

D n-Si

n-Si p-Si

FET

Figure 15.6 Schematic diagram of the BioFET: (a) for the whole-beetle BioFET, the antenna is connected via the electrolyte to the FET and the reference electrode is contacted to the beetle; (b) for the isolated-antenna BioFET, the antenna is mounted in a measuring cell together with the reference electrode.

394

M. J. Schöning

For the pH measurements, buffer solutions (Titrisol, Merck) in the concentration range from pH 2 to pH 12 were used. For the penicillin measurements, 0.01–100 mM penicillin G solution (benzylpenicillin, sodium salt, Sigma) was dissolved in 10 mM TRIS buffer, pH 7, containing 100 mM KCl as an ionic strength adjuster. In case of the BioFET arrangement, different concentrations from 1 ppt to 100 ppm of Z-3-hexen-1-ol, 1-octen and guaiacol are added to an air current that has been flushed over the antenna.47

15.4

Sensor properties of the planar and porous EIS structures

The physical structure and the stoichiometric composition of the PLD-deposited layers have been studied by RBS (Rutherford backscattering spectrometry) measurements and ion channeling, X-ray diffractometry and TEM (transmission electron microscopy) cross-sectional analysis. Figure 15.7 exemplarily shows the RBS and the cross-sectional TEM of the sensor after the Al2O3 deposition. The sensor was built-up of 70 nm of SiO2 and 50 nm of Al2O3. The micrograph demonstrates the very sharp interfaces between the single layers (Si/SiO2 and SiO2/Al2O3) as well as the flat surface of the deposited Al2O3 layer. In the RBS diagram, the full line corresponds to the measured spectrum that is compared to a simulation (open circles). For the simulation, a density of 2.2 g cm3 for SiO2 and 3.98 g cm3 for Al2O3 have been considered. Due to the perfect fitting for

Energy (MeV) 60

0.4

0.6

0.5

0.7 Silicon

O

Normalised yield

50

0.8 70 nm 50 nm

Substrate SiO2

Al2O3

40

30 Al

Al2O3 20

Si

SiO2

10

Si 50 nm

0 250

300

350

400 Channel

450

500

550

Figure 15.7 RBS (Rutherford backscattering spectrometry) spectrum of a planar pH sensor prepared by the PLD process (circles: theoretically calculated values; full line: measurement), and TEM (transmission electron microscopy) cross section (inlet).

Novel sensors 395 both curves, the stoichiometry of the pH-sensitive Al2O3 layer can be assumed to be identical with the original target composition. Ion channeling experiments as well as X-ray diffraction investigations could validate the amorphous structure of the pH-sensitive Al2O3 layer. Comparable results have been obtained for Ta2O5 as pH-sensitive material. For the electrochemical sensor characterization, C/V- and Concap measurements have been performed. Figure 15.8 shows a typical set of C/V curves of the developed capacitive pH sensor. Depending on the pH value, the position of the C/V curves shifts along the voltage axis. An increase of the pH leads to a shift towards positive voltage values. This shift is due to the variation of the ionization states of the Al2O3 surface groups described in the site-binding theory.48 To obtain the resulting calibration curve (Figure 15.8, inlet), the sensor output signal was calculated at 60% of the maximum capacitance value. The evaluated calibration curve between pH 2 and pH 12 results in a linear, nearly-Nernstian pH response of about 56 mV/pH. The pH sensitivity of Ta2O5 was slightly higher with about 56–58 mV/pH. Moreover, the pH sensitivity is highly stable in the long term. During a measuring period of more than 1200 days, an average pH sensitivity for Al2O3 of about 55 1 mV/pH without any significant loss in the sensitivity was found.49 Within this period, only an extremely small drift of the calibration curve of about 0.35 mV per day was observed. With regard to practical applications, the developed planar pH sensor was tested in a continuous flow-through arrangement consisting of the pH sensor, a Ag/AgCl reference electrode, a six-part valve and a pump. The dynamic sensor response was in the order of about 1 min with a small hysteresis of less than 3 mV between pH 4 and pH 9. However, the response time is clearly influenced by the tube diameter and the tube length as well as the flow velocity of the solution. When comparing the developed pH sensor to a conventional pH electrode the resulting detection accuracy is better than 3% in deviation.42 30

pH

20 1000

15

10

Voltage (mV)

Capacitance (nF)

25

800 600 400 200

2

4

6

8

10

12

pH

5 –1500

–1000

–500

0

500

1000

1500

2000

Voltage (V)

Figure 15.8 Typical set of C/V curves and resulting calibration curve (inlet) of a planar pH sensor in the concentration range from pH 2 to pH 12.

396

M. J. Schöning

In order to realize miniaturized capacitive sensors, the formation of porous Si as a novel Si-based transducer material for ions and biomolecules has been investigated. The formation takes place by an anodic etching process in hydrofluoric solution as described in the experimental section. A scanning electron micrograph (SEM) and a transmission electron micrograph (TEM) of the prepared macroporous silicon layer are presented in Figure 15.9(a,b). The dark areas in the top (a) pore

Si

10 µm

(b)

SiO2

Si3N4

50 nm

Figure 15.9 (a) SEM (scanning electron micrograph) of a porous silicon layer (top view); and (b) TEM of a single pore (cross section) covered with 50 nm SiO2 and 30 nm Si3N4.

Novel sensors 397 view (a) show the homogeneous distribution of the circular pores which possess a mean pore diameter of about 1 m. The depth of the pores was about 1 m to 2 m. The TEM of a single pore after the deposition of a layer sequence of 50 nm SiO2 and 30 nm Si3N4 by means of the PECVD process is shown in Figure 15.9(b). As can be seen, both the pore walls as well as the pore bottom are completely covered with the insulating layers. Due to the macroporous formation of the silicon layer a deposition of sensitive materials inside the pores is possible. The passivation of the pore walls with the two gate-insulating materials SiO2 and Si3N4 has been demonstrated by X-ray photoelectron spectroscopy (XPS) sputter depth profiles and cyclic voltammograms in earlier investigations.43 To study the sensor characteristics, C/V measurements were performed in the concentration range from pH 4 to pH 8 and vice versa. Figure 15.10(a) represents a set of C/V curves of the porous EIS sensor built up of Si/SiO2/Si3N4. Here, the parallel shift of the C/V curves that is also caused by the flat-band voltage shift towards positive values with increasing the pH can be explained by the presence of both Si–OH and Si–NH2 groups at the sensor surface.48 The average sensitivity of 54 mV/pH is in good agreement with results obtained for non-porous pH sensors.8 Besides the linear pH response (Figure 15.10, inlet), the maximum capacitance value is increased to about 900 nF with respect to approximately 30 nF (a) 1000 pH 8 900

pH 7 pH 6

Capacitance (nF)

800

pH 5 200

pH 4

700

(b) 150

600

100 50

500

0

400

~54 mV/pH

–50 –100

300

3

–0.4

–0.2

0.0

Cporous –900 nF Cplanar–30 nF

0.2 Voltage (V)

0.4

4

5

6

0.6

7

8

9

0.8

“Embedment”: Porous biosensor

“Scaling down”:

Porous microsensor

Figure 15.10 (a) Set of C/V curves of a porous EIS sensor with 50 nm SiO2 and 30 nm Si3N4; and (b) resulting calibration curve.

398

M. J. Schöning

for non-porous samples with comparable layer sequence. This increase of the capacitance by a factor of about 30 can be explained by the surface enlargement of the active sensor area. Thus, besides the ‘embedment’ of biomolecules inside the porous structure, a miniaturization of the porous EIS sensor in its size, that is, a ‘scaling down’ without any loss of the sensor performance should be possible. Therefore, the silicon wafer was structured by means of mask-alignment techniques.43,50 Porous EIS microsensors with different geometries and shapes such as rectangles, triangles, crosses and circles have been fabricated.51 It was possible to realize porous microsensors with ‘etching spot’ sizes down to 10 10 m2.52 The experiment did also indicate that the different shapes could be formed independent of the crystallographic orientation of the silicon substrate unlike the anisotropic etching technique with KOH.43 For pH calibration measurements, a porous EIS microsensor of 100 100 m2 spot size was mounted in a specifically designed microcell of Foturan53 yielding an average pH sensitivity of about 56 mV/pH in the concentration range from pH 4 to pH 8. Besides the possibility of miniaturization, porous EIS sensors should offer the advantage of fixing biomolecules inside the porous surface just by means of a physical adsorption process. Therefore, a porous biosensor based on the enzyme penicillinase that has been immobilized into the porous pH-sensitive EIS structure of Si/SiO2/Si3N4 has been developed. A typical Concap measurement in the concentration range from 10 M to 1 mM is given in Figure 15.11. The porous biosensor principle can be explained in the way that the enzyme penicillinase hydrolyzes penicillin in the electrolyte in order to liberate hydrogen ions. The more penicillin is present in the solution to be investigated, the more molecules are converted into penicilloic acid, resulting in a decrease of the pH value near the

0.01 mM

–1940

0.05 mM

0.1 mM

0.25 mM

–20 Sensor signal (mV)

Voltage (mV)

0

–1980

–40 –60

0.5 mM

–80

–100

–2020

0.75 mM 0.0

0

0.2 0.4 0.6 0.8 Penicillin concentration (mM)

10

20 30 Time (min)

1.0

1 mM 40

50

Figure 15.11 Typical penicillin Concap measurement of a porous EIS biosensor in the concentration range from 0.01 to 1 mM, and corresponding calibration curve (inlet).

Novel sensors 399 sensor surface, that is, the pH-sensitive Si3N4 layer. Thus, the decrease of the measured voltage as the sensor signal depends on the penicillin concentration. The corresponding calibration curve (Figure 15.11, inlet) possesses a wide linear range from 0.01 M to 1 M with an average penicillin sensitivity of about 90 mV/mM (regression coefficient r2 0.98499). In spite of the mild immobilization procedure, a long-term stability of more than 50 days during continuous electrolyte exposure was achieved, which can be assumed due to the more stable anchoring of the biomolecules inside the porous structure.54–55 The different developed modules of a pH- and a penicillin microsensor can also be combined to arrays in order to simultaneously detect both the variation of the pH value and the penicillin content. In order to check the principal availability, a capacitive EIS sensor array has been suggested.56 The sensor array contains two EIS structures, namely a penicillin-sensitive one and a pH-sensitive one with identical layer structure (p-Si/SiO2/Si3N4), but without the enzyme penicillinase.

(a)

70

Capacitance (nF)

60 50 40

pH 5

pH 8

pH sensor Penicillin sensor pH 5

30

pH 8

20 –3

–2

–1 0 Voltage (V)

1

2

70

(b)

Capacitance (nF)

60 50 40

pH sensor Penicillin sensor

30 10 mM

0 mM

20 –3

–2

–1 0 Voltage (V)

1

2

Figure 15.12 Set of C/V curves of a capacitive EIS sensor array for the simultaneous pH and penicillin determination; (a) change of the pH; and (b) the penicillin concentration.

400

M. J. Schöning

Both sensors are subjected to an additional dc voltage source with opposite polarities and are connected in parallel.31 The experiment was performed with two planar EIS sensors. The resulting ‘step-like’ C/V curves of the EIS sensor array are demonstrated in Figure 15.12(a,b). Since the two EIS structures are placed in a parallel connection, the final C/V curve is equivalent to the serial addition of the independent capacitance values for each sensor. Thus, the integral step-like C/V curve of the EIS sensor array has two linear regions, where each part of the curve can be allocated to either the pH sensor (upper part in both diagrams) or the penicillin biosensor (lower part in both diagrams). A pH change in the electrolyte affects the sensor signal of the pH as well as the penicillin sensor, whereas the variation of the penicillin content only influences the signal of the penicillin sensor. Consequently, the pH sensor within the array arrangement is able to serve as a reference sensor in order to distinguish between pH disturbances of the analyte and variations of the penicillin content.

15.5

Sensor performance of the insect antenna-based BioFET

In order to realize an intact ‘beetle/chip’ biosensor, first of all, the bioelectronic interface of such a novel sensor type has to be studied extensively. Figure 15.13 presents a typical drain current (ID)/drain source voltage (VDS) output characteristic (a) and the corresponding drain current (ID)/gate voltage (VG) input characteristic (b) of the whole-beetle BioFET as well as representative transconductance curves (c) of the isolated-antenna BioFET with and without the insect antenna. To study the intrinsic transistor characteristics, additional odour samples were first suppressed. The curves, that is, the changes of ID vs VDS can be clearly divided into the saturated and the non-saturated region. The respective input characteristic (Figure 15.13(b)) corresponds very well to the input characteristic of identical MOSFETs. In addition, the input characteristic can be used to calculate the transconductance of the FET by taking the first derivative of each curve.45 As can be seen in Figure 15.13(c), the evaluated transconductance of the isolatedantenna set-up with and without the insect antenna shows no significant difference. That means that the electronic properties of the FET are not affected by the coupled insect antenna and vice versa, yielding a nearly perfect signal transfer inside the bioelectronic interface. One of the main goals in constructing a ‘beetle/chip’ sensor is to achieve an effective coupling between the insect antenna and the gate of the FET. Therefore, the influence of the working point (WP) of the FET on the biosensor characteristic has been studied in the isolated-antenna set-up. To adjust the proper WP in terms of the gate voltage and the drain source voltage, the biosensor was stimulated at defined times (see Figure 15.14, marked by arrows) by various concentrations of the ‘green-leaf odour’ Z-3-hexen-1-ol from 1 ppb to 100 ppm that have been applied to the antenna. Additionally, blank stimuli of air served as the reference signal. For these investigations, different working points between 1.5 and 3 V were chosen. For each working point, the gate voltage was chosen equal to the drain source voltage. The working point variations show that the height of

Novel sensors 401 (a)

(b)

3 VG = 2 V

2 1 0

VG = 1V VG = 0 V

0

4 1 2 3 Drain source voltage (V)

4 VDS = 3 V

VG = 3 V

4

Drain curent (mA)

Drain curent (mA)

5

5

3 VDS = 2 V

2 VDS = 1V

1 0

0

0.5

1 1.5 2 Gate voltage (V)

2.5

3

2.0 Transconductance (mS)

(c)

Without antenna With antenna

VG = 3 V

1.5 1.0 0.5 0.0 –2

VG = 0 V

–1

0 1 2 3 Gate voltage (V)

4

5

Figure 15.13 (a) ID-VDS output characteristic; and (b) ID-VG input characteristic of the whole-beetle BioFET; and (c) transconductance curves of the isolatedantenna BioFET with and without insect antenna.

recorded signals, that is, the drain current changes (ID) of the BioFET, increases by raising the WP from 1.5 to 3 V. This can be explained due to the increase of the transconductance of the FET. On the other hand, for all WPs the signal-tonoize ratio remains approximately constant. However, as a negative effect the peak height of the drain current no longer depends on the odour concentration when increasing the WP voltages significantly above VG VDS 2 V. Especially for the WP of VG VDS 3 V, every odour stimulus results in a drain current signal of nearly the same height, independent of the chosen odour concentration. A possible explanation for this behavior might be the voltage stress to the antenna due to the effective external gate voltage. In case of the potato beetle, an optimal signal transfer was found for a WP of VG VDS 2 V. Here, a clear dependence of the biosensor signal on the stimulus intensity exists (Figure 15.15). Even test odour concentrations down to 1 ppb are detectable with a response time of less than one second. The calibration curve also demonstrates that the voltage steps induced by the antenna result in a maximum biosensor signal of about 2.5 A. This corresponds to an antennagenerated voltage of approximately 2 mV, which is in good accordance with EAG measurements.35,57 The lifetime of the biocomponent varied between 5 and 18 h,

402

M. J. Schöning

2 µA

WP 4: VDS = 3 V, VG = 3 V

2 µA

WP 3: VDS = 2.5 V, VG = 2.5 V

2 µA

WP 2: VDS = 2V, VG = 2 V

2 µA

WP 1: VDS = 1.5 V, VG = 1.5 V

Air

1ppb 10 ppb 0.1 ppm1 ppm 10 ppm 100 ppm

200 s

Figure 15.14 Influence of the different working points (WP) with VG and VDS between 1.5 V and 3 V, respectively, on the dose-response characteristic of the isolated-antenna BioFET in the concentration range from 1 ppb to 100 ppm Z-3-hexen-1-ol.

taking into account that the antenna preparation has to be further improved. For the steelblue jewel beetle, even detection limits down to 1 ppt are achievable. One interesting application of the developed biosensor is the detection of damaged potato plants, which have been investigated in greenhouse experiments using a BioFET-based biosensor system (Figure 15.16). The compact head of the biosensor system houses the BioFET. A constant flow of air across the antenna is generated by a suction pump. In the calibration mode the incoming air runs through a charcoal filter which removes all stimulating components in the air.58 The calibration signals were generated using glass syringes containing defined dilutions of the odour Z-3-hexen-1-ol.59 When the piston of a syringe was moved by the step motor drive, a reproducible amount of the green-leaf odour was added

Novel sensors 403

Drain current changes (µA)

3.0 2.5 2.0 1.5 1.0 0.5 1E-3

0.01 0.1 1 10 Z-3-hexen-1-ol concentration (ppm)

100

Figure 15.15 Typical calibration curve of the isolated-antenna BioFET in the concentration range from 1 ppb to 100 ppm Z-3-hexen-1-ol. (a)

(b) 500

Charcoal

Odour 1

0

r 2

Down

∆ID (nA)

Up

Odour

–500 1

2

–1000

3 4

–1500 (c) 0

e

Antenna holder FET

conc. 2

Antenna

∆ID (nA)

conc. 1

–500

1⬘ 2⬘

–1000

3⬘

BioFET

–1500

20

40 Time (s)

4⬘ 60

Figure 15.16 Schematic diagram of the BioFET-based biosensor system consisting of the sampling and calibration unit (with filter and reference concentrations); (a) the antenna-holding device and the biosensor head, and measurements of plant damages in greenhouses. Four filtered reference concentrations were applied to the isolated-antenna BioFET; (b) (1st cycle); (c) in the 2nd cycle the same concentrations were added to the ambient air of damaged plants.

to the main air stream as a 0.5 s puff generating the biosensor response peak. In the second step, the odour concentrations in ambient air were measured by removing the charcoal filter, thus allowing the airborne odour of the damaged plants to reach the antenna. Here, the calibration pulses also produce a biosensor

404

M. J. Schöning

response, as described above. In contrast, a certain amount of the green-leaf odour in the surrounding will change the biosensor signal. The ambient odour concentration can be determined using the dose-response function and the amplitudes of the responses to the superimposed calibration stimuli.60 A typical measuring cycle for four reference concentrations (0.5, 5, 50, 500 ppb) of Z-3-hexen-1-ol once detected in clean (prepared) air and once in ambient air (charcoal filter removed) by using the BioFET-based biosensor system is presented in Figure 15.16(b,c). In this greenhouse experiment, one damaged potato plant was placed within a field of several hundreds of undamaged plants. It can be seen from the diagram in Figure 15.16(c) that the biosensor signal for the lowest odour concentration (first peak), that is, 0.5 ppb, is significantly reduced due to the released odour signal of the damaged plant. The extension of this application to microbiologically diseased plants and spoilt food can be demonstrated now.

15.6

Conclusions

The applicability of different capacitive EIS sensors for various biochemical sensor approaches has been demonstrated. High, long-term stable pH sensors could be prepared by using the pulsed laser deposition technique. These EIS sensors show an average sensitivity of 56–58 mV/pH. Practical experiments could validate the high detection accuracy. All these results favour the employment of the PLD process as an alternative deposition method in order to fabricate pHsensitive gate-insulator materials. Due to the identical layer sequence, in future experiments the resulting technique will be transferred to pH-sensitive ISFETs. On the other hand, the proposed approach can be extended to the thin-film preparation of a wide range of different inorganic compounds as materials for miniaturized chemical sensors. Since the PLD process represents a high-energy, non-equilibrium process, the deposition of complex multicomponent systems such as chalcogenide glass-based miniaturized ISEs (ion-selective electrodes) is feasible. Chalcogenide glass-based electrodes usually consist of up to five different materials in a well-defined stoichiometric composition, like Cd-Ag-As-I-S, Tl-Ag-As-I-S, etc. in order to guarantee their sensor performance. In general, they are sensitive towards their respective primary ions such as Cd, Tl, etc. The proper combination of various types of such thin film ISEs as useful instruments for the heavy metal determination in aqueous solutions can aim in the design of an ‘electronic tongue’ device. Therefore, in a first step single thin film sensors have to be realized, in order to subsequently combine them in multisensor arrangements. Recently performed experiments could validate the successful preparation of single sensors, which are based on these materials.61 Porous EIS structures were realized utilizing inexpensive process steps of Si planar technology. The porous Si substrate was formed by an anodic etching process. The sensor, made of porous Si/SiO2/Si3N4 shows an average pH sensitivity of 54 mV/pH. By adsorptive immobilization of the enzyme penicillinase into the macroporous structure, a stable biosensor that is sensitive to penicillin has

Novel sensors 405 been achieved. The increase of the active sensor area due to the porous layer structure raises the capacitance by a factor of up to 30. Thus, a proportional miniaturization in the sensor size down to a 1/30 is possible without any loss in the sensor behaviour, for example, by means of photolithographical structuring. Characteristic C/V curves could be obtained down to etching spots of 10 10 m2, a linear pH dependence was obtained down to 100 100 m2 spots. The spongelike structure of the porous silicon can further be used as a stable matrix for various chemically and biologically sensitive materials. Moreover, by tailoring the porous layer structure, the size of the macropores can be adjusted to the molecular size of, for example, enzymes and ionophores. Since the formation of porous Si allows to develop three-dimensional sensor structures, two additional perspectives can be of strong interest: If one deals with chemical microsensors, the lack of a suitable microreference electrode often hinders a commercial application. Hence, preliminary experiments could prove the fabrication of a chip-integrated reference, based on macroporous Si and covered by a Nafion membrane.52 On the other hand, due to its biocompatibility, porous Si as well as anisotropically etched Si62 can be favoured as transducer layer for an interaction of, for example, cells with Si, yielding a bioelectronic interface down to the molecular level. To distinguish between pH fluctuations, which result from the enzymatic reaction and those which are measured due to a pH variation, a capacitive EIS sensor array has been designed that is based on a modified C/V measurement technique. Therewith, simultaneous pH- and penicillin detection is possible. This modified principle of C/V measurements is fully compatible with any combination of at least two various EIS sensors. With regard to an attractive sensor approach, the interaction of different capacitive sensors onto a single sensor chip will be pursued. In this way, the main goal could be based on a single ‘lab on chip’ sensor that combines various chemically and biologically sensitive materials for the simultaneous detection of various quantities. As all these sensors are based on the same transducer principle, namely the field effect, a higher reliability and detection accuracy might be possible. A novel biosensor type consisting of the bioelectronic interface of a wholebeetle and an isolated-antenna BioFET, respectively, has been developed. For this purpose, the intact chemoreceptor was directly connected to the gate of the FET using an electrolyte solution. The resulting bioelectronic interface was studied with regard to the intrinsic BioFET characteristics and the BioFET performance. Both sensor arrangements are highly sensitive and selective towards Z-3-hexen1-ol, 1-octen and guaiacol down to the sub ppb range. Due to its compatibility with techniques of semiconductor processing, the sensor has been miniaturized and integrated into a portable BioFET-based biosensor system. An example for an application of this kind of BioFET is the detection of plant damages in greenhouse experiments under real-life conditions. The BioFET showed that the sensitivity and the response time are appropriate for a fast screening of potato plants, which are damaged mechanically or by insect infestation. Such a warning system could be a valuable tool in plant and in stored food protection. As a second kind

406

M. J. Schöning

of application, the insect antenna of the steelblue jewel beetle, which is able to detect volatiles released by smouldering fires, can be utilized for an early fire warning system. Further investigations are necessary to optimize the stability of this hybrid biosensor. Increased ruggedness will be achieved by a specific design of the measuring cell including an easy-to-handle preparation system for the insect antenna. Possible future developments such as the oriented growing of single neuron bundles can combine both the advantages of EAG measurements and silicon–neuron approaches. Moreover, the combination of several BioFETs can be used as a bioelectronic nose that can be fabricated for the detection and the identification of complex odour mixtures.63 This kind of electric nose includes the advantage of markedly higher operation times of the sensor heads. But they also have the disadvantage of needing calibration and training of the neuronal networks for the proper identification of special odour bouquets, which are suggested, for example, for bacteria and spoilage moulds. Future progress will be in the direction of designing a synthetic insect antenna.64 Therefore, however, it is necessary to understand the genetic, physiological and structural build-up of the odour-binding proteins (ODP) inside the insect antenna that are responsible for the recognition process. A recombinant production of the ODPs combined with a proper stabilization in an artificial membrane provides the perspective of rebuilding the natural signal cascade as a ‘nanodimensional’ biosensor.

Acknowledgements The author greatfully thanks H. Ecken, P. Kordos, H. Lüth, A. Poghossian, P. Schroth, J. Schubert, A. Steffen, S. Schütz, M. Thust and W. Zander for valuable discussions and technical support.

References 1. W. Göpel, J. Hesse and J. N. Zemel, Sensors A Comprehensive Survey, Chemical and Biochemical Sensors Part I and II, VCH Verlag, Weinheim, New York, Basel, Cambridge (1991). 2. The Measurement, Instrumentation and Sensors Handbook, CRC Press, Boca Raton, FL (1999). 3. J. Janata and R. J. Huber, Ion Selective Electrodes in Analytical Chemistry, New York (1980). 4. A. P. F. Turner, Advances in Biosensors, Vol. II, JAI Press Ltd., London (1992). 5. M. J. Madou and S. R. Morrison, Chemical Sensing with Solid State Devices, Academic Press, San Diego, CA (1989). 6. P. Bergveld, IEEE Trans. Biomed. Eng., 17 (1970) 70–71. 7. B. H. van der Schoot and P. Bergveld, Biosensors, 3 (1987/88) 161–186. 8. T. Matsuo and M. Esashi, Sensors and Actuators, 1 (1981) 77–96. 9. A. Sibbald, J. Mol. Electron., 2 (1986) 51–83. 10. T. Seiyama, Chemical Sensor Technology, Vol. 1, Elsevier, Amsterdam (1988). 11. I. Lundström, M. S. Shiravaraman, C. S. Svenson and L. Lundkvist, Appl. Phys. Letters, 26 (1975) 55–57. 12. W. Moritz, I. Meierhöfer and L. Müller, Sensors and Actuators, 15 (1988) 211–219.

Novel sensors 407 13. P. Fabry and L. Laurent-Yvonnou, J. Electroanal. Chem., 286 (1990) 23–40. 14. M. Beyer, C. Menzel, R. Quack, T. Scheper, K. Schügerl, W. Treichel, H. Voigt, M. Ullrich and R. Ferretti, Biosensors & Bioelectronics, 9 (1994) 17–21. 15. Y. Hanazato, G. Nakako and S. Shiono, IEEE Trans. Electron. Dev., 33 (1986) 47–51. 16. J. Anzai, J. Hashimoto, T. Osa and T. Matsuo, Analytical Sciences, 4 (1988) 247–250. 17. Y. Miyahara, T. Moriizumi and K. Ichimura, Sensors and Actuators, 7 (1985) 1–10. 18. R. Renneberg, Spektrum der Wissenschaft, 4 (1996) 87–92. 19. J. Cheung and J. Horwitz, MRS Bulletin (1992) 30–36. 20. K. L. Saenger, Proc. Advanced Materials, 3 (1993) 1–24. 21. G. K. Hubler, MRS Bull. (1992) 26–27. 22. Y. Arita, J. Cryst. Growth, 45 (1978) 383–392. 23. R. L. Smith, Sensors and Actuators A, 21–23 (1990) 830–834. 24. R. C. Anderson, R. S. Muller and C. W. Tobias, J. Electrochem. Soc., 138(11) (1991) 3406–3411. 25. R. L. Smith and S. D. Collins, J. Appl. Physics, 71 (1992) R1–R22. 26. R. L. Smith and D. C. Scott, IEEE Trans. Biomed. Eng., 33(2) (1986) 83–90. 27. S. Barret, F. Gaspard, R. Herino, M. Ligeon, F. Muller and I. Roga, Sensors and Actuators, 33 (1992) 19–24. 28. Z. M. Rittersma, PhD-Thesis, Shaker Publishing, Maastricht, Netherlands (1999). 29. V. Lehmann, W. Hönlein, H. Reisinger, A. Spitzer, H. Wendt and J. Willer, Thin Solid Films, 276 (1996) 138–145. 30. M. Thust, M. J. Schöning, S. Frohnhoff, R. Arens-Fischer, P. Kordos and H. Lüth, Meas. Sci. Technol., 7 (1996) 26–29. 31. M. J. Schöning, M. Thust, M. Müller-Veggian, P. Kordos and H. Lüth, Sensors and Actuators B, 47 (1998) 225–230. 32. M. Thust, M. J. Schöning, J. Vetter, P. Kordos and H. Lüth, Anal. Chim. Acta, 323 (1996) 115–121. 33. S. Caras and J. Janata, Anal. Chem., 52 (1980) 1935–1937. 34. A. I. Spielmann and J. G. Brand, Experimental Cell Biology of Taste and Olfaction, CRC Press, Boca Raton (1995). 35. A. E. Sauer, G. Karg, U. T. Koch, J. J. De Kramer and R. Milli, Chem. Senses, 17 (1992) 543–553. 36. S. Schütz, B. Weißbecker, H. E. Hummel, M. J. Schöning, A. Riemer, P. Kordos and H. Lüth, Naturwissenschaften, 84 (1997) 86–88. 37. S. Schütz, H. E. Hummel, M. J. Schöning, P. Schroth, P. Kordos, H. Lüth, S. Zimmermann, A. Schwarz, D. Kohl and D. Koczan, Proc. 2, 9th Int’l Trade Fair and Conference for Sensors Transducers & Systems, Nürnberg, Germany, May 18–20 (1999) 99–104. 38. T. C. Baker and K. F. Haynes, Physiol. Entomol., 14 (1989) 1–12. 39. S. M. Sze, Physics of Semiconductor Devices, John Wiley and Sons, New York (1981). 40. M. Klein, Sensors and Actuators B, 1 (1990) 354–356. 41. M. J. Schöning, S. Schütz, P. Schroth, B. Weißbecker, A. Steffen, P. Kordos, H. E. Hummel and H. Lüth, Sensors and Actuators B, 47 (1998) 235–238. 42. M. J. Schöning, D. Tsarouchas, L. Beckers, J. Schubert, W. Zander, P. Kordos and H. Lüth, Sensors and Actuators B, 47 (1996) 228–233. 43. M. J. Schöning, F. Ronkel, M. Crott, M. Thust, J. W. Schultze, P. Kordos and H. Lüth, Electrochimica Acta, 42 (20–22) (1997) 3185–3193. 44. P. Schroth, M. J. Schöning, S. Schütz, Ü. Malkoc, A. Steffen, M. Marso, H. E. Hummel, P. Kordos and H. Lüth, Electrochimica Acta, 44 (1999) 3821–3826.

408

M. J. Schöning

45. P. Schroth, M. J. Schöning, P. Kordos, H. Lüth, S. Schütz, B. Weißbecker and H. E. Hummel, Biosensors & Bioelectronics, 14 (1999) 303–308. 46. M. J. Schöning, A. Steffen, M. Sauke, P. Kordos, H. Lüth, A. Zundel and M. MüllerVeggian, SEE Proc. Sensors for The Environment, Grenoble, France, March 30–31 (1995) 55–59. 47. S. Schütz, B. Weißbecker and H. E. Hummel, Biosensors & Bioelectronics, 11 (1996) 427–433. 48. D. L. Harame, L. J. Bousse, J. D. Shott and J. D. Meindl, IEEE Trans. Electron. Dev., 24 (1987) 1700–1707. 49. M. J. Schöning, J. Schubert, W. Zander, M. Müller-Veggian, A. Legin, Yu. G. Vlasov, P. Kordos and H. Lüth, Proc. Chemical Microsensors and Applications II, Sept. 19–20, Boston, USA (1999) 124–133. 50. G. Buß, M. J. Schöning, H. Lüth and J. W. Schultze, Electrochimica Acta, 44 (1999) 3899–3910. 51. M. J. Schöning, O. Glück, P. Schroth, S. Schütz, A. Steffen, M. Thust, U. Windirsch, P. Kordos and H. Lüth, Biocybernetics and Biomedical Engineering, 19(1) (1999) 105–129. 52. M. J. Schöning, A. Kurowski, M. Thust, P. Kordos, J. W. Schultze and H. Lüth, Sensors and Actuators B, 64 (2000) 59–64. 53. A. Vogel and J. W. Schultze, Electrochimica Acta, 44 (1999) 3751–3759. 54. M. Thust, M. J. Schöning, P. Schroth, Ü. Malkoc, C. I. Dicker, A. Steffen, P. Kordos and H. Lüth, J. Molecular Catalysis B: Enzymatic, 7 (1999) 77–83. 55. H. Lüth, M. Thust, A. Steffen, P. Kordos and M. J. Schöning, Materials Science & Engineering B, 69–70 (2000) 104–108. 56. M. J. Schöning, M. Thust, P. Kordos and H. Lüth, Advances in Science and Technology, 26 (1999) 55–62. 57. P. Färbert, U. T. Koch, A. Färbert, R. T. Staten and R. T. Carde, Environ. Entomol., 26 (1997) 1105–1106. 58. S. Schütz, M. J. Schöning, P. Schroth, B. Weißbecker, P. Kordos, H. Lüth and H. E. Hummel, Sensors and Actuators B, 65 (2000) 291–295. 59. S. Schütz, B. Weißbecker, U. T. Koch and H. E. Hummel, Mitt. Dtsch. Ges. Allg. Ang. Ent., 10 (1995) 231–236. 60. S. Schütz, U. T. Koch and M. J. Schöning, Patent DE-19629338. 61. J. Schubert, M. J. Schöning, C. Schmidt, M. Siegert, S. Mesters, W. Zander, P. Kordos, H. Lüth, A. Legin, Yu. G. Mourzina and Yu. G. Vlasov, Applied Physics A, 69 (1999) 803–805. 62. M. J. Schöning, Ü. Malkoc, M. Thust, A. Steffen, P. Kordos and H. Lüth, Sensors and Actuators B, 65 (2000) 288–290. 63. C. Ziegler, W. Göpel, H. Hämmerle, H. Hatt, G. Jung, L. Laxhuber, H.-L. Schmidt, S. Schütz, F. Vögtle and A. Zell, Part II, Biosensors & Bioelectronics, 13 (1998) 539–571. 64. M. J. Schöning, P. Schroth and S. Schütz, Electroanalysis, 12(9) (2000) 645–652.

16 Miniaturization of biosensors Wolfgang Schuhmann and Katja Habermüller

16.1

Introduction

A biosensor is usually defined as a sensing device consisting of a biological recognition element in intimate contact with a transducer, which is able to convert a primary biological process into a measurable electronic signal. Thus, a biosensor combines the specificity and selectivity of biological components – from enzymes to whole cells – with various optimized transducer elements. This implies the possibility to develop a big variety of biosensors based on different transduction methods and adapted to specific applications in clinical1–4 and environmental analysis.5–11 Miniaturization is one of the most important trends in current biosensor research. Not only because of the small geometrical dimensions that allow the development of micro-sensor devices for many new applications, such as microelectrochemical systems for local detection, multi-sensor arrays, implantable or portable devices, but also because of the improved electrochemical properties of micro-electrodes. Increased mass transport due to a hemispherical diffusion profile, small currents and reduced capacitance of micro-electrodes lead to faster response, higher sensitivity and reduced iR-drop. Thus, micro-electrodes are optimal for measurements in small volumes, solutions with low conductivity and/or extremely small concentrations, as well as for detection of fast electron-transfer reactions using methods such as fast-scan cyclic voltammetry. Additionally, they exhibit an improved signal-to-noise ratio. Besides new applications and optimized transducer properties which are anticipated by the application of microbiosensors, it is equally important that miniaturization allows to save significant amounts of expensive and rare biological material and thus lowers development and production costs. On the other hand, due to the small surface area of miniaturized transducers the number of immobilized biological recognition elements is significantly reduced leading to a decrease of sensor-integrated biological activity and concomitantly to a loss of long-term stability. The biggest market for miniaturized biosensors is seen in clinical applications12,13 aiming mainly on improvement of patient self-control, especially for diabetic patients, the establishment of so called ‘bed-side’ diagnostics in intensive care, and the real-time control of important metabolites during operations.

410

W. Schuhmann and K. Habermüller

In addition, the possibility to determine various neurotransmitters by means of in vivo application of micro-electrodes or micro-dialysis has become an important topic in current brain research. In fields like environmental and industrial analysis14 development of micro-biosensors mainly aims on on-line analysis, automated systems and portable devices which improve comfort and may lower the costs per measurements, but the minor importance of these needs as compared to health care is reflected by a much smaller number of related publications. However, despite the variety of possible applications and the obvious advantages of micro-biosensors, most of the devices described in literature are still quite far away from commercialization. The field of micro-biosensors – shown as an overview in Figure 16.1 – offers a broad variability of possible sensor Enzyme Oligonucleotide Antibody

Cell

Tissue Coenzyme

Conducting polymer

Biological component

Gel

Entrapment

Environmental Applications

Biomedical Food analysis

Immobilization

Clinical

Micro-biosensor

Crosslinking Adsorption Microencapsulation Covalent binding

Automatization

Transducer

Microelectrodes

IDA

FET

Self-assembled Conducting polymer monolayer Arrays

Detection

Pulse voltammetric

Conductometric Potentiometric

Compatibility

Amperometric

Microtechnology

Figure 16.1 Variability of electrochemical micro-biosensors.

Compatibility

Miniaturization of biosensors 411 architectures, but not all of them fulfil the demands of industry and end-users concerning sensor characteristics, reproducibility, mass production capability and possibilities to develop multi-sensor arrays or other more sophisticated devices. The following sections will start with a short description of the fundamental principles of (micro) biosensors, pointing out the advantages and problems of miniaturized devices as well as their special demands concerning the sensor architecture, and end up with some selected examples of sophisticated microbiosensor designs, evaluated with respect to their commercialization potential.

16.2 16.2.1

Fundamentals Biological recognition

There are many different complementary recognition systems in nature such as enzyme/substrate, antigen/antibody15 and base/correlated base,16,17 which have been used for the development of related micro-biosensors. Not only the isolated recognition elements themselves but also whole cells,18–20 bacteria21–23 and tissue material24–27 have been immobilized on a variety of transducer surfaces in order to obtain specific biosensors for related substrates. However, irrespective of the chosen biological recognition element, one of the key problems in the design of a biosensor is to transduce the primary biological recognition process – which is the source of the selectivity of the sensor set-up – into a measurable signal. In this respect, enzyme-based biosensors possess significant advantages since the active recognition site is regenerated by the catalytic conversion of the primary substrate. This principally implies the reversibility of these devices concomitant with an intrinsic amplification based on the turnover frequency of the enzyme. Taking this into account, it is obvious why the most frequently used biological recognition components in biosensors are enzymes. The biological recognition between an enzyme and its substrate is mainly based on the three-dimensional structure of the protein and especially of the catalytic centre which is complementary to that of the transition state of the corresponding substrate in the enzyme-catalysed reaction. This recognition process occurs due to the simultaneous action of a number of weak intermolecular forces (Coulomb forces, van-der-Waals interactions, hydrogen bridges) between the substrate and the complementary binding pocket. Among the enzymes, which are categorized according to the reaction they are catalysing, the most frequently used are oxidoreductases. Usually, these enzymes have an integrated cofactor (e.g. FAD, PQQ, haem, a transition metal) that is called prosthetic group if it is tightly bound within the active centre. After conversion of the substrate (e.g. oxidation) the intermediately reduced prosthetic group can be recycled (i.e. reoxidized) by a co-substrate (e.g. FMN, NAD, NADP, O2) thus closing the catalytic cycle. Due to the importance of glucose determinations the most popular enzyme for micro-biosensor development is glucose oxidase (GOD), which catalyses the oxidation of ␤-D()-glucose to D-glucono-␦-lactone. Glucose oxidase is commercially available, it is comparably cheap and remarkably stable. Thus, it is

Polyphenol oxidase (PPO)

Glutamate oxidase (GMO)

Cholin oxidase (ChOx)

Acetylcholine esterase (AchE)

Peroxidase

Galactose oxidase (GalOx)

Xanthine oxidase (XO)

Cholesterol oxidase (COD)

Urease

Lactate oxidase (LOD) Penicillinase

28–67

99

dopamine O2 ¶l dopamine quinone H2O2

PPO

50,94–96 97–99

GMO

glutamate O2 H2O ¶l 2-oxoglutarate NH3 H2O2

92–94

88–91

50,85–87

82–84

choline O2 ¶l betaine H2O2

ChOx

acetylcholine H2O ¶l choline acetate

AchE

peroxide H2 ¶l alcohol H2O

PO

Dgalactose O2 ¶l Dgalacto-hexodialdose H2O2

GalOx

hypoxanthine H2O O2 ¶l urate H2O2

XO

cholesterol O2 ¶l cholest-4-en-3-one H2O2

81

53,57,76–80

COD

urea O2¶l CO2 2 NH3

urease

42,58,68–73 74,75

lactate O2 ¶l pyruvate H2O2 ␤-lactam H2O ¶l ␤-amino acid

LOD

␤-D-glucose O2 ¶l D-glucono--lactone H2O2

Glucose oxidase (GOD)

GOD(FAD)

References

Catalysed reaction

Enzyme

Table 16.1 Enzymes used for micro-biosensor development

Miniaturization of biosensors 413 a suitable enzyme both for basic research, investigation of new sensor architectures and development of sensing devices for applications in clinical chemistry or as implantable devices for diabetic patients. However, besides glucose, there are many other enzymes with substrates having medical relevance or which are interesting for environmental or food analysis, such as urea, cholesterol, lactate, acetylcholine and glutamate, that can be detected using electrochemical microbiosensors. A list of these enzymes, the reactions they catalyse, and references to their use in micro-biosensor development are given in Table 16.1. Since, as has been pointed out before, enzymes are the most frequently used biological recognition elements in electrochemical biosensors, main emphasis of the following sections will be laid on enzyme-related biosensors. 16.2.2

Communication between enzyme and transducer

The redox equivalents which are intermediately stored in the reduced prosthetic group of the oxidoreductase have to be transferred to the electrode in order to regenerate the active site of the enzyme and simultaneously to detect the biocatalytic process. Consequently, from the technical point of view, the most interesting aspect in the design of electrochemical biosensors is to link the primary biological recognition event and subsequent catalytic process to an observable current through the sensor electrode. At the first sight, the easiest way would be re-oxidation of the prosthetic group directly at the electrode surface via an electron-tunnelling mechanism leading to a current flux in the outer circuit (Figure 16.2(a)). However, in most enzymes the prosthetic group is deeply buried within the protein shell leading to an electrontransfer distance, which is – according to Marcus theory100,101 – too long for tunnelling processes with a significant electron-transfer rate. Hence, other fast electron-transfer pathways between enzyme and transducer have to be created to obtain finally a measurable signal in correlation with the primary biological recognition process, which is dependent on the substrate concentration. ‘First generation’ biosensors made use of the fact that some of the natural co-substrates, that are able to recycle prosthetic groups, can be directly reduced/ oxidized at a suitable electrode surface (noble metals, carbon). These co-substrates have been used as ‘electron-transfer shuttles’ between enzyme and electrode and (a)

(b) e

(c)

Enzyme

e

e

ea Co-substrate

Co-product

Mediator (ox.)

Enzyme

Substrate

Mediator (red.)

Enzyme

Product Substrate

Product

Substrate

Product

Figure 16.2 Possible electron-transfer pathways between enzyme and transducer.

414

W. Schuhmann and K. Habermüller

either the decrease of co-substrate (e.g. O2) or the increase of co-product (e.g. H2O2) has been detected by applying a suitable oxidation/reduction potential to the electrode leading to a current through the sensor electrode (Figure 16.2(b)). The use of free-diffusing co-substrates is not only the oldest, but until now the most-frequently applied electron-transfer principle in amperometric biosensors. Nowadays, most of the commercially available biosensors still make use of this approach. This is equally true for micro-biosensors. The majority of microsensors developed until now are glucose sensors based on glucose oxidase,35,40,42,50,55,58,60,65–67,70 but also other oxygen-dependent enzymes like lactate oxidase,42,51,69 glutamate oxidase,97–99 galactose oxidase50 or xanthine oxidase83 have been used (see Table 16.1). However, limitations inherent to this electron-transfer principle are the dependence of the sensor signal from the co-substrate concentration, which is especially important for in vivo measurements, and the high working potential necessary for the oxidation of H2O2 or reduction of O2 causing interference by other electrochemically active compounds. Therefore in ‘second generation’ biosensors the natural co-substrates were exchanged against artificial electron-transfer mediators (e.g. ferrocene derivatives, transition-metal complexes, quinones, K4(Fe(CN))6, conducting salts, redox dyes) that were added to the sample and used as free-diffusing electron-transfer mediators (Figure 16.2(c)). Another possible micro-sensor architecture, which is essentially using the same electron-transfer mechanism based on free-diffusing artificial redox mediators, makes use of adsorption of mediators on the electrode surface followed by immobilization of the enzyme in a second layer87 or adsorption of the mediator molecules at the protein shell of the enzyme.48 Even for sensors with higher mediator loading, such as modified carbon pastes67 or other composite materials,81 as well as for electrochemically deposited tetrathiafulvalen-tetracyanoquinodimethan (TTF-TCNQ),71,96 an electron-transfer mechanism comparable to that of freediffusing mediators has to be assumed caused by a slow leakage of mediator molecules into the overlying enzyme layer. Disadvantage of all sensors based on a shuttle-mechanism is the leakage of mediator causing sample contamination102 and decreased long-term stability of the biosensor.103 Consequently, in ‘third-generation’ biosensors emphasis was laid on covalent fixation of mediators and all other necessary sensor components on the transducer surface (‘reagentless biosensors’). Thus, an ‘electron-hopping’ mechanism (Figure 16.3) implying a direct communication between enzyme and transducer should be realized. As the electron transfer takes place via a sequence of several self-exchange reactions between adjacent redox relays, a high mediator loading in the immobilization matrix is an indispensable prerequisite for a fast and efficient electron transfer between enzyme and electrode. Additionally, as can be derived from the properties of sensors using freely diffusing or slowly diffusing redox mediators, it is obvious that a high flexibility of the covalently bound mediator – similar to diffusional movement – is important to attain a productive electron transfer. This can be achieved by using so called ‘redox hydrogels’, such as

Miniaturization of biosensors 415

Electrode

Matrix

Figure 16.3 Electron-hopping mechanism in a redox mediator-modified polymer film.

cross-linked poly(vinyl pyridine)89,95 or poly(vinyl imidazole)47,64 with covalently bound osmium-complexes, as immobilization matrix for enzymes. While these reagentless sensors allow a quasi-direct communication between enzyme and transducer a direct electron transfer via a tunnelling mechanism is difficult to obtain. Besides the implications imposed – according to Marcus’ theory – by the electron-transfer distance, possible denaturation of the enzyme at the bare electrode surface prevents productive electron-transfer via tunnelling processes. Nevertheless, direct electron transfer between redox proteins and various electrode materials (macroscopic) has been intensively studied during the past 20 years, and the state-of-the-art was comprehensively reviewed in 1992104 and 1997.105 In order to also make other enzymes applicable to direct electron transfer several attempts have been made to optimize the immobilization aiming on a shortening of the electron-transfer distance, for example, by oriented immobilization of enzymes on self-assembled monolayers.106–108 Nevertheless, until now only few examples have been reported describing miniaturized biosensors based on direct electron transfer.109 The reason for this becomes obvious, if one considers, that in monolayer architectures the decrease of the electrode surface has a drastic effect on the number of immobilized enzyme molecules contributing to the sensor signal. For example the surface coverage of cytochrome c (12.4 kDa) on a thiol monolayer was estimated to be about 15 1012 mol cm2.110 This would be about 7 1012 molecules on a disk electrode with 1 cm diameter (0.785 cm2 surface area) but only 7 104 protein molecules on a micro-electrode with 1 m diameter. Thus, for a micro-biosensor one has to cope with a significant decrease of the signal, which supposes sophisticated amplifiers. In addition, due to the limited number

416

W. Schuhmann and K. Habermüller

of active bio-molecules on the sensor surface denaturation of only a few enzyme molecules would have a significant impact on the over-all activity leading to a decreased long-term stability of such devices. As a matter of fact, this would be even more dramatic for bulkier enzymes like glucose oxidase (160 kDa). Consequently, optimization of the sensor design with respect to bio-compatibility, long-term stability, reproducibility, electron-transfer properties and transport limitations are even more important for micro-biosensors. Another approach to establish a direct electron transfer makes use of the conductivity of several polymers such as polypyrrole91 or its derivatives.77 Here, the electron transfer takes place via the chains of the conducting–polymer network, which can be seen as an increased ‘virtual’ electrode surface. This is especially important for the development of miniaturized sensors, as a higher number of enzyme molecules can be immobilized on such an electrode surface as compared with monolayer configurations. 16.2.3

Miniaturized transducers

Transducers available for the development of micro-biosensors can be divided into two main groups, needle type and planar electrodes. Needle-type sensors are mostly used for neuro-biological research, for example, direct electrochemical111,112 or enzymatic93,94,97,99,113 detection of neurotransmitters, but also for medical applications, for example, glucose sensors.34,35,40,48,49,51 Needle-type electrodes consist of a conducting material, mostly carbon fibres40,49,55,65,88,92–94,99 or Pt-wires,35,81,93,98 embedded within a glass capillary or an insulating coating. The remaining active electrode surfaces are usually planar discs or cylinders. Needletype electrodes are in general handcrafted, fabricated with the help of a pipette puller, or by means of a variety of different etching methods. The fabrication of planar transducers is either possible using thin-film or thickfilm technology. Thin-film structures are produced using methods of current semiconductor technology in clean-room facilities, such as the standard (complementary metal oxide semiconductor process (CMOS)) or bipolar technology. In contrast, thick-film electrodes are obtained using different printing techniques, such as screen-printing, micro-contact printing or ink-jet printing.114 Especially photolithographic techniques have been successfully used for the production of planar electrochemical transducers. By means of sophisticated micro-machining or lithographic techniques (e.g. lift-off) even three-dimensional transducer structures can be reproducibly produced. Thus, a great variety of planar transducer designs are currently available, including single free-standing electrodes,58,60,69,115 micro-electrode arrays,42,52,62,77,116,117 field-effect transistors (FETs)57,75,79,80,117 and miniaturized electrochemical cells/chambers.31,57,61,118,119 16.2.4

Immobilization of the biological recognition component on miniaturized transducer surfaces

In general, most of the immobilization methods developed and optimized for macro-biosensors can be somehow adapted to miniaturized devices. However,

Miniaturization of biosensors 417 taking into account the special prerequisites for mass production or modification of individual electrodes on an array chip, reproducible immobilization methods are required. These methods have to be compatible with the micro-technological processes applied for the fabrication of the transducer (see Section 16.2.3). Hence, non-manual immobilization methods are highly needed for the development of micro-biosensors, and it would be straightforward to directly integrate the immobilization procedure into the fabrication process of the transducer.120 Methods for the lithographic structuring of enzyme membranes by means of ‘lift-off’-techniques and/or the use of photo-polymerizable materials have been realized.57,58,79 For the fabrication of thick-film biosensors, a broad variety of modified carbon pastes have been developed, containing enzymes, mediators and other additives. In addition, screen printing of enzyme-containing polymer structures has been proposed. Another non-manual immobilization method is the electrochemical polymerization of conducting polymers, such as polypyrrole and its derivatives,97,99,121 poly(m-phenylene diamine)49 or polyaniline.52 The electrochemical-induced formation of conducting-polymer films is well suited for the modification of microbiosensors and sensor arrays. The polymerization process takes place exclusively on top of the active metal surface, and the thickness and the physical properties of the polymer film can be controlled via proper choice of various polymerization parameters.

16.3 16.3.1

Selected examples of micro-biosensors Implantable glucose and lactate sensors based on detection of molecular O2

As already mentioned, until now most biosensors, which have been evaluated as implantable devices use the natural co-substrates of oxidases, O2 or H2O2, as electron-transfer mediators. Lactate and glucose sensors designed, for example, by Gough and his group28,70 are based on the co-immobilization of lactate oxidase or glucose oxidase together with catalase in order to avoid denaturation of the oxidase by H2O2. The enzymes are cross-linked on the surface of an oxygen electrode by means of glutaraldehyde. Implantation of such glucose sensors in the vena cava of dogs30 and of the lactate sensor in dogs and in the canine right atrium70 was reported. In the latter sensor configuration70 the enzyme (lactate oxidase) is immobilized in a gel coupled to an oxygen-sensitive electrode. Lactate and oxygen from the body react within the enzyme gel and the oxygen, which is not converted in the enzyme-catalysed reaction is detected by the electrode. The ambient oxygen partial pressure is indicated at a second oxygen reference electrode. The currents at both electrodes are subtracted to give a lactate-dependent difference current. A second, additional requirement is the need to maintain a stoichiometric excess of oxygen relative to lactate within the enzyme gel to assure that the reaction/ diffusion process is limited by lactate, rather than by oxygen. To circumvent this problem a sensor design has been evaluated in which a cylindrical sensor body

418

W. Schuhmann and K. Habermüller Lactate electrode

Oxygen

Oxygen

Hydrophobic tube

Oxygen reference electrode

Lactate and oxygen Hydrophobic layer Immobilized enzyme

Silicon Epoxy rubber fill

Insulated leadwires

Epoxy

Figure 16.4 Schematic drawing of the sensor design allowing axial diffusion of substrate and oxygen and additional radial diffusion of oxygen alone (reproduced with permission of the American Chemical Society from Ref. 28).

allows the axial diffusion of O2 together with the substrate but in addition a radial diffusion of O2 alone, in order to obtain a higher O2 partial pressure close to the electrode surface (Figure 16.4). The lactate electrode and oxygen reference electrode are located in series in a single silicone rubber tube. The silicon rubber tube is impermeable for lactate but highly permeable to oxygen. Both lactate and oxygen can diffuse into the gel layer through the exposed annular end parallel to the axis of the oxygen sensor, but only oxygen can diffuse radially to the gel through the silicon rubber surface. The oxygen reference electrode is based on radial oxygen diffusion. This sensor design can eliminate an oxygen deficit without introducing an unacceptable response lag. Disadvantages of this sensor architecture are the complicated manual fabrication of the sensor (welding of short pieces of Pt- and Ag-wire to stainless-steel wires, encapsulation, dip-coating etc.), the need for a second O2-electrode to determine the O2-background concentration, and the manual immobilization of the enzymes. 16.3.2

Reagentless glucose sensors based on redox hydrogels

As has been pointed out in Section 16.3.1, problems implied by the dependence of the enzyme reaction from the O2 partial pressure can effectively be circumvented using covalently bound redox mediators. A related sensor architecture has been realized by entrapping glucose oxidase into a redox hydrogel consisting of {poly[(1-vinylimidazolyl) osmium (4,4 -dimethylbipyridine)2Cl]}(/2) (PVI13dme-Os) cross linked with poly(ethylene glycol) diglycidyl ether (PEGDE).47 The

Miniaturization of biosensors 419 0.09 mm

0.29 mm

0.25 mm

Gold

15 µm 15 µm 20 µm

Sensing layer Glucose oxidase 110 ng PVI13-dme-Os 552 ng Peg 42 ng

Mass transport limiting layer EAQ PVPA PAZ

Biocompatible layer 560 ng 26 ng 62 ng

PEO/TA

885 ng

Figure 16.5 Schematic drawing of a multi-layer biosensor architecture based on Oscomplex-modified hydrogels with entrapped glucose (reproduced with permission of the American Chemical Society from Ref. 47).

amperometric glucose sensor has a decreased working potential (200 mV vs SCE) as compared with those based on the oxidation of H2O2, which diminishes interference by co-oxidizable compounds such as ascorbate and paracetamol. For in vivo evaluation of such sensor designs multi-layer sensors containing a first layer of ‘wired’ glucose oxidase and several anti-interference and diffusion layers (Figure 16.5). The glucose mass transport restricting layer consisted of a poly(ester sulfonic acid) film (Eastman AQ29D) and a copolymer of polyaziridine and poly(vinyl pyridine) partially quaternized with methylene carboxylate. The outer biocompatible layer was formed by photocross-linking of tetraacrylated poly(ethylene oxide). At a working potential of 200 mV vs a saturated calomel reference electrode the sensor showed a sensitivity of 1–2.5 nA mM1, a linear range of up to a glucose concentration of 60 mM, a response time of less than 60 s and a decrease of sensitivity of only 4% over a test period of one week. Such sensors were implanted subcutaneously in anaesthetized rats47 and in a diabetic chimpanzee.64 16.3.3

Micro-biosensors for neurotransmitter detection

The investigation of neurotransmitter release from neuronal cells using microelectrodes is an important aspect in current brain research. Micro-biosensors for this application are based on needle-type transducers on which enzymes can be immobilized by entrapment into electrochemically deposited polymers. This non-manual immobilization method is highly reproducible and can be directly integrated into the fabrication process of the transducer. Examples for such micro-sensors are glutamate and dopamine sensitive electrodes obtained by immobilization of glutamate oxidase or polyphenol oxidase,

420

W. Schuhmann and K. Habermüller

respectively, in amphiphilic polypyrrole films.99 While the glutamate sensor is based on amperometric detection of enzymatically produced H2O2, in the dopamine sensor the chinoid reaction product of the enzyme-catalysed reaction is detected at the electrode surface. Instead of amphiphilic polypyrrole derivatives also polypyrrole, over-oxidized polypyrrole, and poly(o-phenylene diamine) have been used as immobilization matrices for glutamate oxidase on platinum micro-electrodes.97 16.3.4

Implantable sensors produced with photolithographic methods

In contrast to the hand-made transducers described in the first two examples, silicon-technology allows the easy and reproducible fabrication of sensor devices with a variety of different geometries like electrode arrays or planar threeelectrode electrochemical cells. Additionally, on-chip production of the sensor together with circuits for signal processing is possibly resulting in ‘smart’ sensor devices. Using this technology, a small three-electrode transducer (0.55 mm width) comprising a linear Pt-band electrode, an L-shaped Pt-counter electrode and an integrated Ag/AgCl reference electrode in-between the Pt-micro-electrodes was produced on a silicon substrate by thin-film and photolithographic techniques (Figure 16.6). Glucose oxidase was immobilized on the three-electrode chip by crosslinking with albumine using glutaraldehyde, electrophoretic coagulation and covered by means of a poly(urethane) membrane as bio-compatible diffusion barrier.37,39 Implanting these sensor structures subcutaneously into rats showed satisfying results over a measuring period of several days. The complicated manual

Figure 16.6 Photograph of a three-electrode electrochemical micro cell fabricated with photolithographic and lift-off techniques (courtesy of M. KoudelkaHep, IMT, Neuchâtel, CH).

Miniaturization of biosensors 421 encapsulation and immobilization method, however, remained an unsolved problem. Using the same three-electrode device as the basis transducer an improved reproducibility and an exclusive localization of the enzyme on the working electrode area could be achieved by means of electrochemically-induced entrapment of glucose oxidase within a growing polypyrrole film.39 Moreover, by combination of photolithographic transducer fabrication with photo-polymerizable materials like poly(acryl amide) hydrogels as immobilization matrix for the enzyme strategies for mass production of micro-sensors with high reproducibility could be demonstrated.57 Here, an enzyme-containing poly(acryl amide) gel is spincoated on the electrode structures on wafer level. The polymer layer is structured using conventional photolithographic and lift-off techniques leading to a parallel modification of all sensor structures on the wafer in one step. 16.3.5

Single-cell measurements using micro-machined electrochemical sensors

Application of three-dimensional transducer structures122,123 for the growth of single cells (e.g. neurons,124 tumour cells,124,125 heart cells126) in so-called ‘electronic Petri dishes’118 has been proposed aiming on the non-invasive detection of released neurotransmitters or metabolites. Figure 16.7 shows a nano-volumetric device for the detection of purine secretion from a myocyte. The heart cell is located on top of a circular platinized

Figure 16.7 Photograph of a single myocyte in a micro electrochemical chamber. The view shows the platinum working electrode situated at the bottom of the chamber being surrounded by counter and reference electrode, as well as the heart cell (reproduced with permission of the American Chemical Society from Ref. 126).

422

W. Schuhmann and K. Habermüller

Au-working electrode, surrounded by counter and reference electrode. The electrodes form the bottom of a micro-chamber of 200 m diameter and 20 m depth (volume 0.6 nl), fabricated by means of standard photolithographic techniques, such as lift-off and the use of photocurable polyimide. For the detection of purines, a solution containing three enzymes (adenosine deaminase, nucleoside phosphorylase and xanthine oxidase) was injected into the chamber. If the cell is releasing adenosine after stimulation (poisoning), H2O2, which is produced as the final product at the end of the enzymatic cascade, could be detected amperometrically at the underlying platinized working electrode. It could be shown that after a rigour contracture was induced in a single myocyte, adenosine (but not inosine) reached the extracellular space after the cell membrane had been permealized using a detergent. This result provided important information about the origin of ischaemic adenosine, which challenges the established assumption that purine release is an early retaliatory response from intact anoxic myocytes. 16.3.6

Miniaturized total chemical analysis systems (TAS)

The integration of micro-sensors into miniaturized flow cells and their combination with miniaturized actuators such as reactors, valves and pumps leads to TAS. Due to their small size, their low reagents and sample volumes, the decreased energy consumption, and often the shortened measuring time, TAS allow in principal multi-parameter measurements. Integration of different sensors into the TAS together with the possibility to develop portable devices, are attractive features of these systems for applications in medicine59 and environmental research.127,128 A modular TAS adaptable for a variety of different transducers and sensing principles had been proposed recently. The sensors (amperometric microelectrode arrays, ISFET-based potentiometric sensors and a micro-cuvette for spectrophotometric determinations) can be integrated into a base plate consisting of a micro-channel system, which has been obtained by anisotropic etching of silicon (Figure 16.8(a)). Different amperometric biosensor modules have been designed which are compatible with the dimensions of this micro-flow system ranging from a single planar macro electrodes to arrays of nine micro-electrodes (10 or 20 m diameter) which are arranged within an area of 500 500 m (Figure 16.8(b)). The electrode arrays are produced on silicon chips by means of CMOS-technology. Platinum and silver layers are sputtered on top of the wafer and structured in a lift-off process.129,130 Enzymes, such as tyrosinase for detection of phenolic compounds, have been suggested to be immobilized on these micro-electrode arrays by entrapment into electrochemically deposited polypyrrole films. An example of a TAS for medical applications is a system for continuous online monitoring of lactate and glucose.59 Here, photo-polymerizable poly(hydroxyethyl methaacrylate) (polyHEMA) was used for immobilizing glucose oxidase and lactate oxidase on an integrated biosensor array produced by lithographic

Miniaturization of biosensors 423 (a)

(b) Sample calibrant and carrier solution

Micropumps

Tubing

Fluid connector

System carrier

Optical absorption cell

Amperometric electrode

ISFET/ReFET 30

.5

mm

32

m .0 m

Silicon channel system

Backplane

Figure 16.8 (a) Schematic drawing showing the modules of a TAS with integrated potentiometric, amperometric and optical transducers, micropumps and an anisotropically-etched base plate (courtesy of S. Drost, FhG-IMS, Munich, D). (b) Microscopic photo of the planar nine-electrode array used in the TAS (courtesy of O. Köster, FhG-IMS, Duisburg, D).

methods (see Section 16.3.5). Enzymatically produced H2O2 was detected amperometrically. The whole system consists of a microdialysis probe for intravascular monitoring in an ex vivo mini-shunt arrangement, a flow-through detection system based on a three-dimensional flow circuit incorporating silicon chips with stacked micro-machined channels and an integrated biosensor array placed at the base of the stack.

16.4

Summary and outlook

Electrochemical micro-biosensors are a forthcoming trend in analytical chemistry. These special electrochemical sensors make use of the unique selectivity of biological recognition elements and combine them with the experiences obtained from modern electrochemical micro-technology. Although the number of commercially available micro-biosensors is still small, there is a big potential market for such devices especially in clinical analysis. The biggest problem that all biosensors – whether miniaturized or not – have in common, is their limited long-term stability imposed by the liability of the

424

W. Schuhmann and K. Habermüller

biological compound. However, research is directed to optimized immobilization methods, improvement of sensitivity into the nano- and picomolar measuring range, and hence, without any doubt many of the current problems concerning sensitivity and mass production will be solved within the next years. Until now, the smallest biosensors have sizes of several micrometre and have been realized mainly using glucose oxidase as biological recognition element. Further minaturization has to take into consideration the even further decrease of the immobilized enzyme activity and hence increased problems with the sensitivity and long-term stability of these sensor structures.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

V. M. Owen and A. P. F. Turner, Endeavour 11, 100–104 (1987). G. A. Rechnitz and R. M. Nakamura, J. Clin. Lab. Anal. 2, 131–133 (1988). G. G. Guilbault and R. D. Schmid, Biotechnol. Appl. Biochem. 14, 133–145 (1991). D. Pfeiffer, in Frontiers in Biosensorics II (F. W. Scheller, F. Schubert, J. Fedorowitz, eds) Birkhäuser, Basel, 1997. R. D. Schmid, GBF-Monogr. 13, 271–280 (1989). F. Scheller, K. Riedel, B. Neumann and D. Pfeifer, DECHEMA Biotechnol. Conf. 4, 1045–1048 (1990). E. B. Nikolaskaya, G. A. Evtyugin and T. N. Shekhovtsova, J. Anal. Chem. USSR 49, 408–416 (1994). K. R. Rogers and L. R. Williams, Trends Anal. Chem. 14, 289–294 (1995). I. Karube, Y. Nomura and Y. Arikawa, Trends Anal. Chem. 14, 295–299 (1995). M. P. Marco and D. Barcelo, Meas. Sci. & Technol. 7, 1547–1562 (1996). J. L. Marty, B. Leca and T. Noguer, Analusis 26, M144–M149 (1998). W. Göpel, Mikrochim. Acta 125, 179–196 (1997). J. Wang, J. Pharm. Biomed. Anal. 19, 47–53 (1999). A. S. Bassi, D. Tang, E. Lee, J. X. Zhu and M. A. Bergougnou, Food Technol. Biotechnol. 34, 9–22 (1996). M. Aizawa, A. Morioka, S. Suzuki and Y. Nagamura, Anal. Biochem. 94, 22–28 (1979). T. Livache, A. Roget, E. Dejean, C. Barthet, G. Bidan and R. Teoule, Synth. Met. 71, 2143–2146 (1995). T. Vo-Dinh, J. P. Alarie, N. Isola, D. Landis, A. L. Wintenberg and M. N. Ericson, Anal.Chem. 71, 358–363 (1999). J. Katrlik, R. Brandsteter, J. Svork, M. Rosenberg and S. Miertus, Anal. Chim. Acta 356, 217–224 (1997). L. Campanella, G. Crescentini, M. G. Donorio, G. Favero and M. Tomassetti, Annali di Chim. 86, 527–538 (1996). M. Naessens and A. Tran-Minh, Anal. Chim. Acta 364, 153–158 (1998). G. Y. Tai, M. L. Wen and C. Y. Wang, Microchem. J. 53, 152–157 (1996). L. H. Larsen, N. P. Revsbech and S. J. Binnerup, Appl. Environ. Microbiol. 62, 1248–1251 (1996). L. H. Larsen, T. Kjaer and P. Revsbech, Anal. Chem. 69, 3527–3531 (1997). M. Ozsoz, A. Erdem, E. Kilinc and l. Gokgunnec, Electroanal. 8, 147–150 (1996). B. R. Eggins, C. Hickey, S. A. Toft and D. M. Zhou, Anal. Chim. Acta 347, 281–288 (1997).

Miniaturization of biosensors 425 26. G. A. Rechnitz, D. Coon, C. Babb, A. Ogunseitan and A. Lee, Anal. Chim. Acta 337, 297–303. 27. A. W. O. Lima, E. K. Vidsiunas, V. B. Nascimento and L. Angnes, Analyst 123, 2377–2382 (1998). 28. D. A. Gough, J. Y. Lucisano and P. H. S. Tse, Anal. Chem. 57, 2351–2357 (1985). 29. J. Brueckel, H. Zier, W. Kerner and E. F. Pfeiffer, Hormone Metab. Res. 22, 382–384 (1990). 30. J. C. Armour, J. Y. Lucisano, D. B. McKean and D. A. Gough, Diabetes 39, 1519–1526 (1990). 31. E. Mann-Buxbaum, F. Pittner, T. Schalkhammer, A. Jachimowicz, G. Jobst, F. Olcaytug and G. Urban, Sens. Actuators B 1, 518–522 (1990). 32. A. Amine, J. M. Kauffmann and G. J. Patriarche, Talanta 38, 107–110 (1991). 33. T. Abe, Y. Y. Lau and A. G. Ewing, J. Am. Chem. Soc. 113, 7421–7423 (1991). 34. D. S. Bindra, Y. Zhang, G. S. Wilson, R. Sternberg, D. R. Thevenot, D. Moatti and G. Reach, Anal. Chem. 63, 1692–1696 (1991). 35. C. Cronenberg, B. Van Groen, D. De Beer and H. Van den Heuvel, Anal. Chim. Acta 242, 275–278 (1991). 36. M. Hämmerle, W. Schuhmann and H.-L. Schmidt, Sens. Actuators B 6, 106–112 (1992). 37. M. Koudelka, F. Rohner-Jeanrenaud, J. Terrettaz, E. Bobbioni-Harsch, N. F. deRooij and B. Jeanrenaud, Advances in Biosensors 2, 131–149 (1992). 38. S. A. M Van Stroe-Biezen, F. M. Everaerts, L. J. J. Janssen and R. A. Tacken, Anal. Chim. Acta 273, 553–560 (1993). 39. M. Koudelka-Hep, D. J. Strike and N. F. deRooij, Anal. Chim. Acta 281, 461–466 (1993). 40. J. W. Furbee Jr., T. Kuwana and R. S. Kelly, Anal. Chem. 66, 1575–1577 (1994). 41. L. Coche-Guerente, A. Deronzier, P. Mailley and J. C. Moutet, Anal. Chim. Acta 289, 143–153 (1994). 42. J. Wang and Q. Chen, Anal. Chem. 66, 1007–1011 (1994). 43. H. Sakslund, J. Wang, F. Lu and O. Hammerich, J. Electroanal. Chem. 397, 149–155 (1995). 44. D. J. Strike, N. F. deRooij and M. Koudelka-Hep, Biosens. Bioelectron. 10, 61–66 (1995). 45. C. Jiménez, J. Bartolí, N. F. deRooij and M. Koudelka-Hep, Sens. Actuat. B 26–27, 421–424 (1995). 46. T. Hoshi, J. Anzai and T. Osa, Anal. Chem. 67, 770–774 (1995). 47. E. Csöregi, D. W. Schmidtke and A. Heller, Anal. Chem. 67, 1240–1244 (1995). 48. M. A. McRipley and R. A. Linsenmeier, J. Electroanal. Chem. 414, 235–246 (1996). 49. L. I. Netchiporouk, N. F. Shram, N. Jaffrezic-Renault, C. Martelet and R. Cespuglio, Anal. Chem. 68, 4358–4364 (1996). 50. S. F. Peteu, D. Emerson and R. M. Worden, Biosens. Bioelectron. 11, 1059–1071 (1996). 51. H. Sakslund, J. Wang and O. Hammerich, J. Electroanal. Chem. 402, 149–160 (1996). 52. H. Sangodkar, S. Sukeerthi, R. S. Srinivasa, R. Lal and A. Q. Contractor, Anal. Chem. 68, 779–783 (1996). 53. N. Peled, Pure Appl. Chem. 68, 1837–1841 (1996). 54. P. D. Voegel, W. Zhou and R. P. Baldwin, Anal. Chem. 69, 951–957 (1997). 55. J. Wang, G. Rivas and M. Chicharro, J. Electroanal. Chem. 439, 55–62 (1997).

426

W. Schuhmann and K. Habermüller

56. R. Freaney, A. Mcshane, T. V. Keaveny, M. Mckenna, K. Rabenstein, F. W. Scheller, D. Pfeiffer, G. Urban, I. Moser, G. Jobst, A. Manz, E. Verpoorte, M. W. Widmer, D. Diamond, E. Dempsey, F. J. S. Deviteri and M. Smyth, Annals Clin. Biochem. 34, 291–302 (1997). 57. C. Jiménez, J. Bartrol, N. F. deRooij and M. Koudelka-Hep, Anal. Chim. Acta 351, 169–176 (1997). 58. D. Wilke, H. Muller and N. Kolytsheva, Fresenius J. Anal. Chem. 357, 534–538 (1997). 59. E. Dempsey, D. Diamond, M. R. Smyth, G. Urban, G. Jobst, I. Moser, E. M. J. Verpoorte, A. Manz, H. M. Widmer, K. Rabenstein and R. Freaney, Anal. Chim. Acta 346, 341–349 (1997). 60. T. Matsumoto, M. Furusawa, H. Fujiwara, Y. Matsumoto and N. Ito, Sens. Actuat. B 49, 68–72 (1998). 61. C. Y. Tian, N. Q. Jia, R. Wang, Z. R. Zhang, J. Z. Zhu and G. X. Zhang, Sens. Actuat. B 52, 119–124 (1998). 62. S. P. Yang, Y. F. Lu, P. Atanassov, E. Wilkins and X. C. Long, Talanta 47, 735–743 (1998). 63. J. Wang and F. Lu, J. Am. Chem. Soc. 120, 1048–1050 (1998). 64. J. G. Wagner, D. W. Schmidtke, C. P. Quinn, T. F. Flemming, B. Bernacky and A. Heller, Proc. Natl. Acad. Sci. USA 95, 6379–6382 (1998). 65. X. J. Zhang, J. Wang, B. Ogorevc and U. E. Spichiger, Electroanalysis 11, 945–949 (1999). 66. Q. S. Li, B. C. Ye, B. X. Liu and J. J. Zhong, Fresenius J. Anal. Chem. 363, 246–250, (1999). 67. J. Wang, X. J. Zhang and M. Prakash, Anal. Chim. Acta 395, 11–16 (1999). 68. K. Hajizadeh, H. B. Halsall and W. R. Heineman, Talanta 38, 37–47 (1991). 69. N. Ito, T. Matsumoto, H. Fujiwara, S. Kayashima, T. Arai, M. Kikuchi and I. Karube, Anal. Chim. Acta 312, 323–328 (1995). 70. D. A. Baker and D. A. Gough, Anal. Chem. 67, 1536–1540 (1995). 71. S. A. M. Marzouk, V. V. Cosofret, R. P. Buck, H. Yang, W. E. Cascio and S. S. M. Hassan, Anal. Chem. 69, 2646–2652 (1997). 72. N. F. Shram, L. I. Netchiporouk, C. Martelet, N. Jaffrezic-Renault, C. Bonnet and R. Cespuglio, Anal. Chem. 70, 2618–2622 (1998). 73. Q. L. Yang, P. Atanasov and E. Wilkins, Electroanal. 10, 752–757 (1998). 74. H. Taguchi, N. Ishihara, K. Okumura and Y. Shimabayashi, Anal. Chim. Acta 228, 159–162 (1990). 75. M. J. Schöning, F. Ronkel, M. Crott, M. Thust, J. W. Schultze, P. Kordos and H. Lüth, Electrochim. Acta 42, 3185–3193 (1997). 76. M. Trojanowicz, A. Lewenstam, T. Krawczynski vel Krawczyk, I. Lähdesmäki and W. Szczepek, Electroanalysis 8, 233–243 (1996). 77. A. Guiseppi-Elie, J. M. Tour, D. L. Allara, and N. F. Sheppard Jr., Mater. Res. Soc. Symp. Proc. 413, 439–444 (1996). 78. A. S. Jdanova, S. Poyard, A. P. Soldatkin, N. Jaffrezic-Renault and C. Martelet, Anal. Chim. Acta 321, 35–40 (1996). 79. C. Puig-Lleixa, C. Jimenez, J. Alonso and J. Bartroli, Anal. Chim. Acta 389, 179–188 (1999). 80. A. Senillou, N. Jaffrezic-Renault, C. Martelet and S. Cosnier, Talanta 50, 219–226 (1999). 81. J. Motonaka and L. R. Faulkner, Anal. Chem. 65, 3258–3261 (1993).

Miniaturization of biosensors 427 82. J. M. Kim, M. Suzuki and R. D. Schmid, Anal. Lett. 22, 2433–2443 (1989). 83. M. Suzuki, H. Suzuki, I. Karube and R. D. Schmid, Anal. Lett. 22, 2915–2927 (1989). 84. S. D. Haemmerli, A. A. Suleiman and G. G. Guilbault, Anal. Lett. 23, 577–588 (1990). 85. G. Fortier, R. Beliveau, E. Leblond and D. Belanger, Anal. Lett. 23, 1607–1619 (1990). 86. S. Cosnier and C. Innocent, Anal. Lett. 27, 1429–1442 (1994). 87. K. Miyata, M. Fujiwara, J. Motonaka, T. Moriga and I. Nakabayashi, Bull. Chem. Soc. Jpn. 68, 1921–1927 (1995). 88. E. Csöregi, L. Gorton, G. Marko-Varga, A. J. Tüdös and W. T. Kok, Anal. Chem. 66, 3640–3610 (1994). 89. H. Sakai, R. Baba, K. Hashimoto, A. Fujishima and A. Heller, J. Phys. Chem. 99, 11896–11900 (1995). 90. I. C. Popescu, E. Csöregi and L. Gorton, Electroanal. 11, 1014–1019 (1996). 91. V. Lvovich and A. Scheeline, Anal. Chem. 69, 454–462 (1997). 92. E. Tamiya, Y. Sugiura, E. N. Navera, S. Mizoshita, K. Nakajima, A. Akiyama and I. Karube, Anal. Chim. Acta 251, 129–134 (1991). 93. I. Karube, K. Yokoyama and E. Tamiya, Biosens. Bioelectron. 8, 219–228 (1993). 94. Q. Xin and R. M. Wightman, Brain Res. 776, 126–132 (1997). 95. M. G. Garguillo and A. C. Michael, J. Am. Chem. Soc. 115, 12218–12219 (1993). 96. Q. Xin and R. M. Wightman, Anal. Chim. Acta 341, 43–51 (1997). 97. J. M. Cooper, P. L. Foreman, A. Glidle, T. W. Ling and D. J. Pritchard, J. Electroanal. Chem. 388, 143–149 (1995). 98. S. Poitry, C. Poitry-Yamate, C. Innocent, S. Cosnier and M. Tsacopoulos, Electrochim. Acta 42, 3217–3233 (1997). 99. S. Cosnier, C. Innocent, L. Allien, S. Poitry and M. Tsacopoulos, Anal. Chem. 69, 968–971 (1997). 100. R. A. Marcus and N. Sutin, Biochim. Biophys. Acta 811, 265–322 (1985). 101. R. A. Marcus, Angew. Chem. Int. Ed. English 32, 1111–1121 (1993). 102. S. L. Brooks and A. P. F. Turner, Meas. Contr. 20, 37–43 (1987). 103. W. Schuhmann, H. Wohlschläger, R. Lammert, H. L. Schmidt, U. Löffler, A. D. Wiemhöfer and W. Göpel, Sens. Actuat. B 1, 537–541 (1990). 104. F. A. Armstrong, Adv. Inorg. Chem. 38, 117–163 (1992). 105. A. L. Ghindilis, P. Antanasov and E. Wilkins, Electroanal. 9, 661–675 (1997). 106. T. Lötzbeyer, W. Schuhmann and H.-L. Schmidt, Sens. Actuat. B 33, 50–54 (1996). 107. I. Willner, V. Helegshabtai, R. Blonder, E. Katz and G. L. Tao, J. Am. Chem. Soc. 118, 10321–10322 (1996). 108. H. Zimmermann, A. Lindgren, L. Gorton and W. Schuhmann, Chem. Eur. J. 6, 592–599 (2000). 109. M. M. Correira dos Santos, P. M. Paes de Sousa, M. L. Simoes Goncalves, H. Lopes, I. Moura and J. J. G. Moura, J. Electroanal. Chem. 464, 76–84 (1999). 110. S. Song, R. A. Clark, E. F. Bowden, and M. J. Tarlov, J. Phys. Chem. 97, 6564–6572 (1997). 111. D. Bruns and R. Jahn, Nature 377, 62–65 (1995). 112. P. S. Cahill, Q. D. Walker, J. M. Finnegan, G. E. Mickelson, E. R. Travis and R. M. Wightman, Anal. Chem. 68, 3180–3186 (1996). 113. P. Pantano, W. G. Kuhr, Electroanal. 7, 405–416 (1995).

428

W. Schuhmann and K. Habermüller

114. M. Lambrechts and W. Sansen, in Biosensors: Microelectrochemical Devices, Institute of Physics Publishing, Bristol, 1992. 115. H. Tap, P. Gros and A. M. Gue, Electroanal. 11, 973–977 (1999). 116. T. Hermes, M. Buehner, S. Buecher, C. Sundermeier, C. Dumschat, M. Borchardt, K. Cammann and M. Knoll, Sens. Actuat. B 21, 33–37 (1994). 117. W. H. Baumann, M. Lehmann, A. Schwinde, R. Ehret, M. Birschwein and B. Wolf, Sens. Actuat. B 55, 77–89 (1999). 118. J. M. Cooper, Trends Biotechnol. 17, 226–230 (1999). 119. B. Wolf, M. Brischwein, W. Baumann, R. Ehret and M. Kraus, Biosens. Bioelectron. 13, 501–509 (1998). 120. K. Habermüller, C. Kranz and W. Schuhmann, in Umweltdiagnostik mit Mikrosystemen (G. Henze, M. Köhler, P. Lay, eds.) Wiley-VCH, Weinheim, 1999. 121. T. Osaka, S. Komaba and A. Amano, J. Electrochem. Soc. 145, 406–408 (1998). 122. M. Brischwein, W. Baumann, R. Ehret, A. Schwinde, M. Kraus and B. Wolf, Naturwissenschaften 83, 193–200 (1996). 123. C. D. T. Bratten, P. H. Cobbold and J. M. Cooper, Anal. Chem. 69, 253–258 (1997). 124. R. S. Pickard, Anal. Chim. Acta 385, 73–77 (1999). 125. B. Wolf, M. Brischwein, W. Baumann, R. Ehret, T. Henning, M. Lehmann and A. Schwinde, Tumor Biol. 19, 374–383 (1998). 126. C. D. T. Bratten, P. H. Cobbold, and J. M. Cooper, Anal. Chem. 70, 1164–1170 (1998). 127. S. Drost, W. Wörmann, B. Ross, G. Chemnitius, M. Rospert, W. Konz, F. Frimmel, W. Schuhmann, R. Ferretti and L. Meixner, AMI, 199–202 (1996). 128. A. Schütze, in Umweltdiagnostik mit Mikrosystemen (G. Henze, M. Köhler, P. Lay, eds) Wiley-VCH, Weinheim, 1999. 129. M. Rospert, Fortschritts-Berichte VDI Reihe 9: Elektronik, 212, VDI Verlag, Düsseldorf, 1995. 130. O. Köster, Dissertation Universität – Gesamthochschule Duisburg (2000).

17 Scanning probe microscopy as an analysis tool Larry A. Nagahara

17.1

Introduction

Numerous specialized scientific and technological fields, such as microelectronics, sensors, microfluidics, and medical diagnostics, are currently focusing on ways to further miniaturize structures, components, and even entire systems to mesoscopic dimensions. Traditionally, many of these fields use one or more aspects of electrochemistry in their processes to fabricate a working device. The challenge for electrochemistry is to keep pace with this miniaturization trend by developing, improving, elucidating, and applying novel electrochemical techniques that function at the sub-micrometer to nanometer length scale. In order for electrochemistry to meet these challenges, analytical tools must be readily available to characterize, and in some cases, implement electrochemical reactions at these small dimensions. One analytical tool that is meeting this challenge is scanning probe microscopy (SPM). SPM’s role as an analysis tool can take on various meanings depending upon the application. In the area of electrochemical microsystems technology (EMT), SPM can provide information and perform various functions that are critical to EMT. For instance, the measurement and fabrication of both positive and negative structures, the monitoring and localizing of reactions even in the case of high aspect ratios, and the precise determination of nucleation sites and growth rate by potential or current control are all areas in which SPM have already played and will continue to play an important role in EMT. However, as with any analysis tool, SPM has certain advantages and disadvantages over other analytical techniques and should be regarded as a complementary tool. In this section, SPM analysis will be described in three parts. In the first part, the most common types of SPMs and considerations for operating an SPM in relation to EMT is discussed. Next, the SPM analysis will be illustrated with several experimental examples. Finally, the SPM as a manipulation tool cannot be overlooked. A few manipulation experiments and subsequent analysis of the manipulated structure will be discussed briefly. Before summarizing, the prospects for the future of SPM as an analysis tool will be given. SPM continues to be a rapidly expanding area of research. This chapter cannot possibly cover all the topics about SPM as an analytical tool in electrochemistry; rather, the main principles

430

L. A. Nagahara

and problems as well as the strategy of probe microscopy are elucidated. For further detailed information, numerous books1–10 and reviews11–15 on the subject as well as the references contained in this chapter will be a helpful starting point for the reader.

17.2

Fundamentals of scanning probe microscopy

Scanning probe microscopy comprises a family of approximately twenty proximal probe techniques that provides information about the topography and as well as surface/sub-surface properties of a material with resolution ranging from mesoscopic dimensions (i.e. hundreds of nanometer) down to atomic-scale dimensions (see Table 17.1). The widespread use of SPM methods in research and characterization labs around the world is based on their ease of implementation, extremely high resolution imaging capability, operation under a variety of environmental conditions, and relatively low costs as compared with techniques of comparable spatial resolution such as scanning electron microscopy (SEM). In addition, SPM is also being used for nanolithography and atomic/molecular manipulation.

Table 17.1 List of scanning probe microscope ac STM AFM BEEM ECAFM ECSTM CFM EFM FFM FMM LFM MFM MRFM NSOM PSTM SCM SCPM SECM SICM SKM SNAM SNOM SMSM SPSTM SSPM STM SThM UFM

Alternating current scanning tunneling microscopy Atomic force microscopy Ballistic electron emission microscopy Electrochemical atomic force microscopy Electrochemical scanning tunneling microscopy Chemical force microscopy Electrostatic force microscopy Frictional force microscopy Force modulation microscopy Lateral force microscopy Magnetic force microscopy Magnetic resonance force microscopy Near-field scanning optical microscopy Photon scanning tunneling microscopy Scanning capacitance microscopy Scanning chemical potential microscopy Scanning electrochemical microscopy Scanning ion conductance microscopy Scanning Kelvin probe microscopy Scanning near-field acoustic microscopy Scanning near-field optical microscopy Scanning Maxwell stress microscopy Spin-polarized scanning tunneling microscopy Scanning surface potential microscopy Scanning tunneling microscopy Scanning thermal microscopy Ultrasonic force microscopy

SPM as an analysis tool 431 The common features in all SPM methods are the use of a sharp probe that is rastered across a sample and a feedback mechanism to keep the probe either in contact or very nearly in contact with the surface. The probe’s interaction with the sample surface is recorded thus providing a two-dimensional map of the surface/ sub-surface properties. This interaction between the sample and probe can be in various forms. For example, as will be illustrated shortly, scanning tunneling microscopy (STM) uses conductance as the interacting media, while atomic force microscopy (AFM) uses van der Waals forces as its interaction. The resolution limit for a given SPM technique is strongly dependent on the radius of curvature of the probe end and the amount of probe area interacting with the sample as depicted schematically in Figure 17.1. Since the probe is kept in such close proximity to the surface, typically 1–10 nm separation, atomic scale resolution is routinely achieved with certain types of SPM techniques. Figure 17.1 illustrates also another key advantage of SPM, that is, the correlation of surface topography with surface properties (e.g. magnetic, chemical species). In many instances, a topographic image may be featureless while a simultaneously acquired image of a particular surface property taken over the same area shows tremendous variations. Figure 17.2 illustrates an example where the topography image shows featureless surface; however, a simultaneously acquired magnetic force image reveals the interaction of a magnetized probe with the magnetic fields generated from a buried metallic structure carrying a current (i.e. B oI/2 . a). The bright region in the image corresponds to the magnetic field of the sample repelling the field

Probe

Scan direction

0–10 nm

Sample Topography profile

Surface interaction profile

Figure 17.1 Schematic illustration of a SPM probe interacting with the surface. The SPM can simultaneously acquire topographic information as well as surface property.

432 (a)

L. A. Nagahara (b)

1µm

Figure 17.2 (a) AFM image taken over a metallic wire buried beneath an oxide layer. The surface has no particular features; (b) MFM image taken simultaneously over the same region. The bright and dark portions of the image reveal the magnetic field generated when a current is passed through a wire.

of the SPM probe, while the dark region represents attraction between probe and sample. Prior to the development of SPM, high resolution structural information about a sample surface immersed in an electrochemical environment had to be inferred from spectroscopic measurements or elucidated using ex situ methods. For an EMT related process, this required the sample to be removed from solution and transferred into another environment such as a vacuum system. As a result, questions were always raised as to whether the obtained data actually represented the in situ conditions. SPM offers the opportunity to observe the sample surface in situ and eliminates many of these concerns. One SPM technique that will not be discussed here is scanning electrochemical microscopy (SECM).16–19 The SECM probes the Faradaic currents that flow through a small electrode probe in close proximity to a sample in solution. SECM is a technique that is useful for both as a form of SPM imaging of surfaces and as an electrochemical tool. Due to its unique significance in EMT systems, a more specific description of SECM is written in detail elsewhere in this book. 17.2.1

Scanning tunneling microscopy

The “founding” member in the SPM family is the STM.20,21 Developed two decades ago by Nobel Prize winners G. Binnig and H. Rohrer in 1982, the STM and its subsequent family members have helped revolutionize mesoscopic surface characterization to the point where it has become a common analysis tool in practically every characterization lab around the world. The STM was originally used to analyze surface structures of conductive materials such as metals and semiconductors in ultra high vacuum (UHV). The power of STM was quickly

SPM as an analysis tool 433 realized when the microscope was used to show conclusively an unresolved phenomena that existed in surface science for more than twenty years, namely identifying the correct surface reconstruction model for the Si(111)-77 structure.22 Before the invention of the microscope, it was known from electron spectroscopy that the surface of Si(111) reconstructed with the unit cell containing forty-nine atoms. Various models had been proposed; but the correct model23 did not conclusively emerge until the STM took real space images of the Si(111)-7 7 surface. It was also soon discovered that the STM works equally well under a variety of other environmental conditions such as in ambient air,24 and in liquids.25 This flexibility to operate under various environmental conditions makes the STM (and other SPM techniques) ideal for investigation of surfaces in their natural environment. The first report of using a STM to study a surface immersed in a liquid environment was by Sonnenfeld and Hansma.25 They observed the carbons atoms on a highly oriented pyrolytic graphite (HOPG) surface under water. The principle of the tunneling microscope is quite simple to express in terms of quantum mechanics. When a potential is applied between a metal probe and a conducting substrate and the separation distance between the probe and sample is decreased to a few tens of Ångstroms, the probability rate for an electron to “tunnel”, hence the name “tunneling” microscope, from one side of the gap to the other can be approximately expressed as: It 艐 Ve兹␾z where z is the distance of the gap, ␾ is the effective work function of the probe and sample, V is the applied bias between the probe and sample and It is the tunneling current. In the case of an UHV–STM, the gap would correspond to the vacuum; while in air or under a solution, a contamination or water layer exists between the gap of the probe and sample. In all of these cases, the gap corresponds to a region between the probe and sample where no state exists for an electron to occupy. This is one of the reasons as to why the STM can operate in all these different environments. The number of electrons tunneling across the gap is very strongly dependent on the gap distance (i.e. a smaller gap gives rise to a higher current) and to a lesser extent to the applied potential as illustrated in Figure 17.3. Under these conditions, the tunneling current changes an order of magnitude with every change of approximately 1 Å. Recall, the diameter of a single atom is approximately 2–3 Å. This tremendous sensitivity in current to changes in height allows the STM to routinely achieve atomic resolution on well-characterized conducting surfaces. In the most common used mode of operation, a feedback loop is established so as to maintain a constant current flow between the probe and sample while raster scanning the surface. A piezoelectric scanner is used to maintain the proper gap distance as well as to scan the region of interest. It is very important to note that the acquired STM image represents a map of the electronic density of states at the surface of the sample at a particular voltage and not necessarily the spatial

434

L. A. Nagahara (a) ∆Z

e– Gap ∆Z

Tip

Sample

∆V

(b)

∆Z

e– Gap ∆Z

Tip

∆V

Sample

Figure 17.3 Schematic drawing of the electron tunneling across the gap between a STM tip and sample. (a) As the STM approaches the surface, electrons can tunnel across the gap, as the distance is decreased; (b) the number of electrons tunneling across the gap dramatically increases.

location of the atoms. Although the electronics states correspond to the position of the atoms in many cases, there are just as many cases where this is simply not the case due to delocalization of the electron or charge transfer. For example, the graphite (HOPG) unit cell, is a hexagonal close packed (hcp) structure which consists of six carbon atoms equally spaced from each other (1.42 Å). An atomic scale resolution STM image of HOPG will also show a hcp structure but with a nearest neighbor distance of 2.46 Å. In other words, only every other atom is observed with the STM. This difference arises from the influence of the underlying graphite layer. The graphite layer below is shifted relative to the top layer such that three of the carbon sites have a carbon atom directly below and the remaining three sites are located above a hollow site. Another illustration is STM images taken of the GaAs(110) surface. Due to charge transfer, the STM images do not show both the Ga and As atoms but only the Ga atoms at a positive sample bias and the As atoms at a negative sample bias.26 The absence of the other atoms demonstrates that the STM is monitoring the electronic states and not simply the location of each individual atom. In these particular cases, the physical structure and electronic properties of graphite and

SPM as an analysis tool 435 GaAs are well known; but in many cases (e.g. organic/bio-molecules) the structure and/or charge transfer is not known and one can easily misinterpret the STM image for structural morphology. 17.2.2

STM probe insulation

For STM operation in liquid environments, an additional consideration needs to be considered, namely probe insulation. Whenever a potential is applied between two electrodes in a solution, there exists a possibility for electrochemical reactions to occur at the electrode interface (e.g. probe or sample). These electrochemical reactions are dependent upon such parameters as potential and result in a current, known as a Faradaic current, which flows between the two electrodes. Since a potential is applied between the STM tip and substrate in order to establish a tunneling current, unwanted electrochemical reactions can also take place simultaneously. The Faradaic currents from a bare metal wire immersed in a solution can be in the order of milliamps; in contrast, tunneling currents are typically in the order of nanoamps to hundreds of picoamps. To decrease the unwanted contribution from the Faradaic current, the amount of exposed surface area of the STM tip to the solution needs to be reduced, while still having some metal exposed to obtain a tunneling current. As a consequence, STM tips are insulated by coating all but the very ends with an insulating material. Materials that have been used to coat tips include polymers,27 glass,28 and Apiezon™ wax.29,30 Figure 17.4 shows a drawing for a method of insulating STM probes. A solder iron is used to heat a notched metal block and Apiezon™ wax is

Figure 17.4 Schematic rendering of insulating a STM tip with Apiezon™ wax.

436

L. A. Nagahara

melted on top of the metal block. The notch provides a temperature gradient for the melted wax and thus different viscosity depending upon location. An uninsulated STM tip is brought through the wax via a goniometer. If the wax is too hot, it will flow easily away from the apex of the tip and leave too much tip area exposed. If the temperature is too cold, the wax is too viscous and will completely cover the STM tip. At the correct point (i.e. temperature) along the notch, the tip will be insulated with wax except for a small area at the apex. After coating with the insulting material, only 0.01–10 m of the tip end is exposed, resulting in a residual Faradaic current below 0.1 nA.

17.2.3

Atomic force microscopy

The second member to the SPM family, developed in 1986 by Binnig, Quate, and Gerber, is the AFM.31 Perhaps more significant in terms of applicability, the AFM has had a larger and wider impact in material analysis than the STM. This is due in part to the capability of the microscope to image non-conducting materials with comparable resolution to the STM. The principle of AFM operation is as follows: a cantilever arm with a sharp probe protruding from one end of the cantilever deflects due to an interaction (e.g. van der Waals) between the probe and sample atoms. Recording the cantilever deflection as the probe rasters over the sample, a high resolution surface topography of the sample is obtained. Depending on the regime of interaction, the force acting between the probe and sample may be attractive or repulsive. The first AFMs used an STM to monitor the deflection of the cantilever. A STM tip was positioned on the top of an AFM cantilever opposite to the probe side of the cantilever. A thin metal coating was evaporated onto the backside of the cantilever in order to have a conducting surface for the STM. A deflection in the cantilever arm would result in a change in the gap distance between the STM tip and cantilever surface and hence a change in tunneling current would be observed. Since the STM has tremendous sensitivity to vertical displacement, this design of the AFM would have, in principle, similar sensitivity. However, in practice, this scheme had numerous problems such as instability in the tunneling signal and a contamination layer between the tip and sample. Today, the most common detection scheme used in the AFM is based on an optical deflection approach. A laser beam is focused on the backside of the cantilever and the deflected laser light is monitored by a split photodiode. As the cantilever is deflected, the laser beam spot changes position on the split photodiode and is registered as a voltage difference. A feedback loop is established such that the voltage difference is kept constant (i.e. constant force is maintained) during imaging. Within the AFM itself there are several variations of the technique. The original AFM was a “contact” AFM in which the probe was in physical contact with the sample. An AFM can also be operated in a “non-contact” mode. In this mode, the cantilever is oscillated near its mechanical resonance frequency with an amplitude of about 1 nm using a piezo oscillator. The probe is placed about 1 nm

SPM as an analysis tool 437 above the surface and the force gradient is monitored. A shift in the resonance frequency occurs as the probe interacts with the sample. Another method uses intermittent contact of the probe with the surface to maintain feedback and is popularly known as “Tapping Mode”™.32 As with non-contact AFM, the AFM cantilever is oscillated near its mechanical resonance frequency. However, the free standing oscillation amplitude is an order of magnitude larger, several tens of nanometers, than for non-contact AFM. When the probe approaches a surface, the oscillating probe begins to touch or “taps” the surface which results in a significant reduction in the amplitude. For instance, as the probe passes over a protrusion on the surface, the amplitude decreases; conversely, as the probe passes over a depression on the surface the amplitude tends to increase. Maintaining a constant oscillation amplitude is the mechanism for maintaining feedback in Tapping Mode™. Since the amplitude of oscillation is much larger than for non-contact AFM, Tapping Mode™ AFM is easier to implement from an instrumentation point of view; however, the resolution of the Tapping Mode™ is about an order of magnitude less than conventional contact mode AFM. A variation to Tapping Mode™ is another ac technique that is gaining popularity known as magnetic ac mode.33–35 In this mode, a cantilever coated with a magnetic material is oscillated at high frequency using an oscillating magnetic field. One of the major differences between magnetic ac mode and Tapping Mode™ or non-contact mode is the advantage of operating in liquid environments fairly easily. Since the cantilever is driven directly by the magnetic field, this allows for better control of the oscillation parameters such as amplitudes and drive frequency. Smaller amplitudes equate to smaller vertical forces on the sample which allow for high resolution imaging of soft samples. There are several drawbacks in operating the AFM in contact-mode as seen by the numerous developments of non-contact technique. First, the probe can damage the surface of soft samples since it is in intimate contact. Second, the microscope is usually operated in the repulsive force regime, which is not the most sensitive (i.e. high-resolution) regime. Third, AFM operations under ambient conditions need to be concerned with unwanted adhesive forces, such as the capillary forces. One of the main contributors to this adhesion force is a thin water layer that exists at the surface of materials at ambient conditions. Capillary action of the water layer pulls the probe toward the sample, which is an attractive force. In order to have a net repulsive force, the force applied to the sample needs to overcome this attractive force contribution. Depending upon the humidity and probe radius, the capillary force can be as large as several hundred nN.36 Let us examine the characteristics of a force vs cantilever-displacement plot of an AFM probe approaching a surface to get a better understanding of this adhesive force. When the cantilever is far away from the surface (Region I, Figure 17.5(a)), it is freestanding and the net force acting on the cantilever is zero. As the cantilever moves closer to the surface, there is a sudden attraction of the cantilever to the surface and the cantilever “drops” into contact with the surface (Region II). This “drop” to contact is not a result of the attractive potential

438

L. A. Nagahara (a) Probe

Water layer

III II

I

IV

(b) Probe

Water layer

Figure 17.5 Schematic drawing of a force–distance plot for (a) an AFM probe interacting with a thin water layer on the surface and (b) with a AFM probe with a high aspect ratio.

between the probe and sample but rather from the probe being wetted and forming a meniscus as it makes contact with the thin water layer that exists on surfaces under ambient conditions. The meniscus pulls the cantilever toward the surface and this shows up as a sudden jump to contact in the force–distance plot. As will be shown later, the probe can be intentionally modified such that the adhesive force is significantly reduced. As the sample continues to push against the cantilever, the cantilever bends in the opposite direction and there exists a net repulsive force on the cantilever (Region III). In this regime, the cantilever’s behavior follows Hooke’s law (i.e. F kz) with the slope of the plot corresponding to the spring constant of the cantilever. As the sample or cantilever is retracted, the system enters a regime where again the water layer on the surface

SPM as an analysis tool 439 plays a role in the plot. In this case, the cantilever is prevented from leaving the surface due to the attractive force on the cantilever from the meniscus (Region IV). Eventually, the cantilever snaps back and is separated from the sample. There are different approaches to overcome the problem of capillary forces between the probe and sample. One way is to reduce the radius of curvature on the AFM probe as shown in Figure 17.5(b). Surface treating the AFM probes to make the surface more hydrophobic has also been used. Cleaning the AFM probes in an oxygen-plasma37 as well as coating the probe with a self assembled monolayer (SAM) of octadecyltricholorosilane (OTS)38 are some of the treatment reported in the literature. Immersing the sample in liquid and imaging under solution is one other way of eliminating this capillary force. Since the cantilever is immersed in the same media, Regions II and IV are no longer present. Note, one may still observe a small jump to contact in the force distance plot but this may be due to other contribution such as an electrostatic interaction. Another way to overcome adhesive forces is to reduce the amount of time the probe is in contact with the sample; hence the advent of Tapping Mode™ and magnetic ac mode. Since the tip is no longer in permanent contact with the surface during scanning, the probe is prevented from adhering to the surface and thus damaging the surface. 17.2.4 17.2.4.1

Other probe techniques Lateral force microscopy

Lateral force microscopy (LFM), also known as frictional force microscopy (FFM), differs from conventional AFM in that the lateral or torsional deflection of the cantilever is monitored as it interacts with the surface, instead of the forces acting normal to the cantilever. As with AFM, the most common detection scheme uses an optical deflection method as shown schematically in Figure 17.6. Torsional motion on the cantilever deflects the laser light as monitored by a split photodiode. Using a photodiode divided into four quadrants, both normal and lateral forces can be acquired simultaneously. This interaction is related to the mechanical (e.g. frictional, adhesive, elastic) properties of the surface and hence requires intimate contact with the surface. As fabricated structures become smaller, the mechanical properties of surfaces on the nanometer scale will play a more critical role and will need characterization. 17.2.4.2

Magnetic force microscopy

Magnetic force microscopy (MFM) is an expanding field since magnetic memory cells used in storage media continue to shrink. MFM probes the dipole interaction between a magnetized tip and a magnetic sample. Since the magnetic dipole interaction is long-range force interaction, spatial resolution of the microscope is limited to about 20 nm. The MFM probes are typically microfabricated silicon probes that are magnetically sensitized with a sputtered ferromagnetic

440

L. A. Nagahara

Figure 17.6 Schematic rendering showing the different behaviors between AFM and LFM. In the case of LFM, the cantilever undergoes torsional movement as the probe interacts with different materials, while in surface topography the cantilever will deflect normal to the surface.

material. The probe is scanned several tens or hundreds of nanometers above the sample. To enhance sensitivity, most MFM instruments use ac techniques as described previously. Gradient variations in the magnetic forces result in a shift in the resonant frequency (either amplitude or phase) of the cantilever which is recorded to produce a magnetic force image. 17.2.4.3

Force modulation microscopy

In force modulation microscopy (FMM),39 the AFM probe or sample is oscillated at a frequency significantly higher than the scan rate, while in contact with the surface. The average cantilever deflections are used as an input signal into the feedback loop similar to conventional contact mode to maintain a constant applied force. The r.m.s. amplitude deflection of the cantilever provides information about the mechanical properties of the sample with a lateral resolution of about 10 nm. For soft materials, the amplitude modulation will be less than for a hard material, thus a spatial map of the local differences in the elasticity of the sample can be obtained. In order for this technique to work properly, the spring constant for the cantilever needs to be much larger (i.e. stiffer) than the material under investigation. 17.2.4.4

Adhesion force microscopy

Adhesion force microscopy (AFM) is another probe technique to map the spatial arrangement of chemical functional groups at the surface. The ratio of the

SPM as an analysis tool 441 adhesion forces for different functional groups can be directly compared with the interfacial energy for the system. The lateral resolution for this technique is approximately 20–30 nm. In adhesion AFM, force vs cantilever-displacement curves show the cantilever deflection as the AFM probe approaches the sample, makes contact and is retraced again. The hysteresis in this curve is a measure for the adhesive probe–sample interaction and is related to the chemical nature of the functional groups at the sample and probe surface. This technique has been applied to observe differences between doped and undoped polymer films.40 17.2.4.5

SPM probe fabrication/geometry

One of the critical points in achieving high resolution imaging/sensitivity with SPM is the shape and property of the probe used to investigate the surface. Today, SPM cantilevers are mass produced using semiconductor microfabrication techniques. The cantilevers are generally made out of silicon nitride, silicon oxide, or silicon.41,42 The AFM probes made out of silicon nitride are pyramidal in shape. Etch pits are formed into a Si(100) wafer with the sides of the etch pits being the Si(111) planes. Silicon nitride is then deposited onto the top of the Si(100) layer and the Si wafer subsequently removed. Since the Si(111) planes define the probes shape, the angle of the probe is 70.6. As a comparison, STM probes are considerably sharper (typical probe angle 10) than silicon nitride AFM probes. The large probe angle is a disadvantage for imaging high aspect ratio structures43,44 and when large capillary forces are involved. The sharpened silicon oxide probes have a sharper apex than silicon nitride probes. In this process, silicon oxide is grown on the etch pitted Si(100) wafer. The oxide growth is different along the Si(111) planes than the Si(100) planes, resulting in a sharper apex at the bottom of the etch pit. A silicon nitride layer is then deposited on top of the silicon oxide surface. Further improvements in tip radii are achieved in silicon probes.42 Here, a mask layer is patterned onto the top of the Si(100) surface. Initially, etching occurs at an opening in the mask layer and subsequent isotropic etching undercuts the Si layer until an apex appears underneath the mask layer. Another method of increasing the probe sensitivity is a technique known as chemical force microscopy (CFM).45 CFM is based upon a chemically functionalized SPM probe interacting with the sample. The degree of interaction between the probe and sample generates contrast in the image. For instance, an AFM probe can be functionalized to be either hydrophobic or hydrophilic. Several researchers have used molecular self assembly techniques to form a monolayer of CH3 or COOH-terminated molecules on an AFM probe.46–48 The most common technique for attachment is using molecular self-assembly to attach a thiol molecule for a gold coated probe or a silane molecule for a silicon oxide treated probe. CFM can be applied to the various types of SPM techniques. In the case of contact-mode AFM in air, chemical modification to the SPM probe can modify the interaction of the thin water layer with the probe/sample as illustrated in Figure 17.7. The most common usage of CFM comes with using it as a LFM. Using these probes in LFM, it is possible to map the spatial arrangement and

442

L. A. Nagahara

Probe

Sample

Sam molecule with specific end-group

Figure 17.7 Schematic illustration of modifying an AFM probe to change the chemical nature of the probe interacting with a substrate.

fraction of areas with similar or different chemical functional groups. In conjunction, measurements made in the normal force mode can determine the adhesive forces between the probe and sample. Researchers have used the CFM to study the interactions between charged groups, nucleic acid, and oligomers.47

17.3

Electrochemical analysis on metal surfaces

One of the areas of science that has benefited the most by the advent of SPM is electrochemistry. The SPM provides direct structural information of electrochemical reactions at the solid–liquid interface. Moreover, in situ reactions such as corrosion, deposition, and etching can also be observed in real time. Gold is one of the most studied materials in electrochemistry due to its unique physical and chemical properties. It was only natural that the first electrochemical studies using SPM was conducted on a gold surface.29,49,50 Due to limited space and for pedagogical purposes, electrochemical analysis using SPM will be confined to various aspects of gold investigation of the Au(111) surface. For further information related to other gold faces as well as various metals and semiconductor materials, numerous reviews have been written on SPM applications to electrochemistry.51–62 17.3.1

Surface reconstruction

It is well known that certain crystallographic faces of metals and semiconductors placed in vacuum undergo a surface reconstruction in order to minimize their surface energy. Similar surface reconstructions also occur in electrochemical environments as well. For instance, Au is the only face center cubic (fcc) metal whose (111) surface has been observed to undergo a reconstruction. This reconstruction

SPM as an analysis tool 443 involves a shift from a fcc to hcp, causing ~4% compression of the Au lattice in the [11¯0] direction such that 23 atoms fit into the space of twenty-two atoms in the bulk. As a result an appearance of a 0.15 Å high herringbone pattern is seen on the surface. The Au(111) surface reconstructs into a ( p 兹3) where p 22 or 23 in UHV and p 22–30 in solution.63 Investigations on these reconstructed surfaces provide information about the distribution and density of state at the interface. Unlike UHV, electrochemistry offers the opportunity to vary the electronic state in a systematic manner with the application of an applied potential. Surfaces that undergo reconstructions in electrochemical environments have been reviewed by Kolb.64 STM observation of the Au(111) reconstruction was conducted initially in UHV65,66 and subsequently in situ with electrochemical STM.67–70 Oden et al.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

Figure 17.8 A series of STM images showing the lifting of the Au(111) reconstruction as a function of applied potential. The locations of the dislocations (“D”) and the steps (“S”) are pointed by arrows and a kink is marked by a pair of arrows. (a) VS 290 mV; (b) VS 440 mV; (c) VS 490 mV; (d) VS 540 mV; (e) VS 590 mV; (f) VS 590 mV 2 min later; (g) VS 590 mV 10 min later; (h) 4 min after VS was changed to 190 mV; (i) 3 min after VS was changed quickly to 590 mV (Tao and Lindsay, Ref. 67).

444

L. A. Nagahara

observed in situ the Au reconstruction with both AFM and STM.71 Tao and Lindsay,67,70 as well as Weaver and co-workers,68 used STM to observe the Au(111) surface undergoing a 22(23) 兹3 reconstruction to a 1 1 structure with applied potential. Relaxation of the compressed surface generates excess Au atoms. STM images revealed changes in the shape of the step edges on the surface. This indicated that the extra material required in the compress structure is taken from and returns to the step edges. Tao and Lindsay showed in a subsequent study the dynamics of the reconstruction on the time scale between tens of seconds and hours.70 The reconstruction can be cycled repeatedly between the 22(23) 兹3 reconstruction and 1 1 structure within a small potential window below ~240 mV (vs a saturated calomel electrode). Figure 17.8 shows a sequence of images showing the reconstruction process. 17.3.2

Oxidation

SPM has been used to understand fundamental processes, such as oxidation, at the nanometer scale. These studies are essential to better understand processes at mesoscopic dimensions which are important to EMT. As an example, the oxidation of gold has been extensively studied.72–75 It is well known that an oxidation–reduction cycle (OCR) causes roughening of an electrode surface. Trevor and co-workers used in situ STM to observe monolayer pit formation in a Au(111) surface after an OCR cycle.76 They observed that pit formation occurred only after the potential was swept to a potential more positive than the potential for the formation of a single monolayer of Au oxide. Itaya and coworkers showed that pits are formed during reduction rather than oxidation.77 The pits could be removed by applying a continuous potential where the Au oxide was reduced. Under potentiostatic control, Gao et al. used in situ STM to show that the gold surface initially begins to roughen as the oxide forms.78 The oxide surface became less rough as the oxide layer spread laterally across the surface but remained disordered. As the potential was swept to a potential where the oxide was reduced, the formation of pits occurred on the surface. They also observed that pit formation was enhanced underneath the STM tip. Schneeweiss et al. used both STM and AFM to study the Au oxidation process.79 The AFM was used to look at thicker (i.e. more insulting) films and to eliminate misinterpretation of the topographic data that might arise from variations in the local density of states from STM imaging. They observed the formation of small hillocks about 0.1–0.25 nm in height and 2– 4 nm in width. 17.3.3

Electrochemical deposition

One of the most intensively studied electrochemical processes is underpotential deposition (UPD) as well as adsorption of ionic species (e.g. halogens). UPD occurs when a monolayer or submonolayer of a foreign metal adatoms with a low work function is deposited on a surface with a higher work function. The canonical system is the UPD of Cu onto Au(111).

SPM as an analysis tool 445 (b) E = 180 mV

Cu/Au(111)

7 nm

0.0

0.1

0.2

0.3 E /V vs Cu/Cu2+

(c) E = 5 mV

(a) E = 300 mV

7 nm

7 nm

Figure 17.9 Cyclic voltammogram of Au(111) in 0.05 M H2SO4 and 1 mM CuSO4 and STM images taken at (a) 300 mV vs Cu/Cu2 showing the bare Au(111) surface; (b) 180 mV vs Cu/Cu2 revealing the (兹3 兹3)R30 structure, and (c) 5 mV vs Cu/Cu2 showing a complete Cu monolayer coverage on Au (Batina et al., Ref. 84).

Magnessen et al. observed the different phases of UPD Cu on Au(111) with in situ STM as shown in Figure 17.9.80 At potential more positive than 300 mV (vs Cu/Cu2), Cu deposition does not occur and the atomic structure of the Au(111) is observed. Between 200 and 100 mV, a submonolayer of Cu is deposited with a structure of (兹3 兹3)R30. Just prior to bulk deposition of Cu, the (兹3 兹3)R30 structure is converted to a (1 1) structure. Both Cu

446

L. A. Nagahara

UPD phases on Au(111) was also observed using AFM.81,82 Other groups have investigated the influence of coadsorbed anionic species as well as the kinetic effects of phase transformation from the (兹3 兹3)R30 to (1 1) structure as a function of step edges.83–85 Halogens adsorption onto noble metals have been extensively studied with a variety of analytical techniques. In situ SPM allowed direct correlation of surface charge with structure orientation. Schardt et al. were the first to observe halogen adsorption, in this case I, onto Pt(111) with STM.86 Iodine adsorption onto Au(111) has been subsequently studied by several groups.69,87–90 Yamada and co-workers observed a compression from a c( p 兹3)R30 structure, where p 3 to p 2.5, at low potentials for iodine onto a Au(111) single crystal.87,88 They also observed a moiré pattern with increasing positive potential.

17.4

SPM analysis of adsorbates on surfaces

Molecular adsorbates will play an increasingly important role in nanotechnology and EMT in such areas as catalysis, corrosion inhibitors, sensors, and for structural modification. Since molecular orientation is often critical to device performance and functionality, SPM studies on these systems will continue to contribute significantly to this area. The number of organic and inorganic adsorbates is diverse and cannot be covered in this limited space. Only a few number examples of adsorbates that self-assemble into organized structures (e.g. alkanethiol selfassembled monolayer (SAM) films, and Langmuir Blodgett films) will be presented.

17.4.1

SAM films

SAM films have perhaps attracted the most interest in regards to SPM analysis of molecular adsorbates.91,92 As the name implies, molecules self assemble onto surfaces forming a monolayer coverage. SAM films have potential applications in sensors, non-linear optical devices, lithography, molecular electronics and microelectronics. They are excellent model systems for studies of fundamental phenomena, such as molecular adsorption, wetting and tribology of organic surfaces, due to their relative ease of preparation and reproducible structure. SPM have been routinely used to characterize the surface of SAM films. Poirier and co-workers observed the initial stages of SAM formation on a Au(111) reconstructed surface in UHV.93–95 The formation of a densely packed, alkanethiol SAM film was usually accompanied by the formation of pits in the gold surface. Edinger et al.96 have used STM to investigate the corrosion of gold by alkane thiols. In conjunction with atomic absorption spectroscopy (AAS), the authors describe the formation pits in the gold surface as the result of a corrosion process in the ethanol–thiol solution. The amount of corrosion was greatly reduced in a hexane solvent which implied a polarity dependence. From a chemical point of view the oxidation of gold thiols was surprising because of the high redox potential.

SPM as an analysis tool 447 Hara et al. looked at the initial stages of alkanethiol deposition onto Au(111) in solution with in situ STM.97 At submonolayer coverages, small islands of ordered 4-mercaptopyridine were observed to initiate at the turns in the herringbone structure of the Au(111)-23 兹3 surface reconstruction. Others have looked at the same initial deposition from vapor phase and observed that the reconstruction influences the SAM domain structure.98 In addition, etch pits generated during vapor deposition of the SAM molecule were not solely the result of chemical etching. The etch pits predominantly followed the reconstruction surface pattern which precluded chemical etching as the dominant formation mechanism. Pan et al. investigated the effects of surface charge on the stability of an alkane thiol SAM film using in situ AFM.99 They observed that the film “disintegrated” at negative potential between 370 and 755 mV vs Ag/AgCl. The wide range in negative potential value was attributed to the amount of dissolved oxygen in the solution. The removal of oxygen shifted the disintegration value to more negative potentials. Ohno and co-workers have used SPM to investigate the formation of nanometer-scale wires on cleaved AlGaAs/GaAs heterostructures.100,101 An eight period of alternating undoped Al0.3Ga0.7As/GaAs heterostructure layer at various thickness was grown by metal organic chemical vapor deposition (MOCVD). The grown sample was cleaved in an ethanol solution containing 1 mM octadecylthiol (ODT) and left immersed for periods ranging from 12 to 48 h. Contact-mode AFM showed alternating bands oriented along the110 direction were clearly observed with a height difference of about 1.0 nm as shown in Figure 17.10. A comparison between the banded structure and the structure of the AlGaAs/GaAs heterostructures revealed that the ODT molecules preferentially attached to the GaAs surface, while the AlGaAs surface had no ODT molecules attached. LFM measurements showed that the AlGaAs region had a torsional force about nine times larger than that over the GaAs regions. The LFM results are consistent with the notion that the ODT adsorbed only to the GaAs regions since ODT is terminated with a hydrophobic CH end group while the AlGaAs oxide is hydrophilic. In order to further ascertain the selective growth on the GaAs region, AFM/LFM measurements on a sample immersed in pure ethanol revealed a height difference of ~0.5 nm and no frictional difference, respectively. This height difference was presumably due to the different degree of oxidation between the GaAs and the AlGaAs. Akari et al. used a CFM to distinguish between a CH3-terminated and a NH2terminated surface.46 The AFM probe was modified with a COOH modified tip. Using an unmodified Si3N4 probe, no contrast between the CH3 and NH2 regions was observed as shown in Figure 17.11(a). However, with a COOH-terminated probe, a clear contrast between the two regions was observed (Figure 17.11(b)). Force distance spectroscopy revealed that the COOH tip/NH2 surface had an attractive force (Figure 17.11(c)), while the COOH tip/CH3 surface was a repulsive interaction (Figure 17.11(d)). Several groups have used SPM to map chemical group distribution of surfaces.47,102–104 Wong et al. have used FFM to look at orientational differences of the terminal groups at the surface.104 It is known from IR studies that for

448

L. A. Nagahara 150 nm

GaAs

AlGaAs

GaAs (sub.)

(a)

(b)

A

B

[110]

[001] 1.2

Height (nm)

(c)

0.8

0.4

0.0

600

400

200

0

Distance (nm)

Figure 17.10 (a) Schematic illustration of the AlGaAs/GaAs(110) heterostructure; (b) an AFM image after immersion in 1 mM ODT ethanol solution; (c) cross sectional profile taken along the cleaved surface reveal a 1.0 nm height difference over the GaAs regions. This height difference is due to the ODT preferentially adsorbing to the GaAs region (Ohno et al. Ref. 100).

n-alkanethiol films the tilt angle for the methyl-terminated group is different depending on whether n is odd or even. In the case where n is an even number, the methyl-terminated group is tilted ~53, while for n as an odd number the tilt angle is ~26. Berger et al. used AFM to image the different adhesion forces between the two different phases of a Langmuir–Blodgett (LB) film.105 The two phases consisted of CH2 and CH3 groups. Most of the studies that have used FMM have been in the area of polymer research.106,107 However, a few studies have been on SAM108 and LB films.109 Bar

SPM as an analysis tool 449

(a)

8

(c)

COOH-modified tip on NH2-region

Force [a.u]

6

151.0 µm

151.0 µm

75.5 µm

4 2 Tip moves toward the surface 0 Tip moves away from the surface

75.5 µm

–2 0 µm 0 µm

0

(b)

100

COOH-modified tip on CH3-region

306 µm

Force [a.u]

6

612 µm

400

8

(d)

612 µm

200 300 Distance (nm)

4 Tip moves toward the surface 2 Tip moves away from the surface 0

–2

306 µm

0 0 µm 0 µm

100

200 300 Distance (nm)

400

Figure 17.11 (a) AFM image of the stamped surface containing CH3 regions and NH2 regions shows no contrast with an uncoated Si3N4 cantilever; (b) using a chemically modifying COOH cantilever, the contrast between the CH3 regions and NH2 regions is clearly resolved; (c and d) force–distance plots taken over the NH2 region and CH3 region, respectively. The adhesive force associated with the COOH interacting with the NH2 region is much stronger than the CH3 region (Akari et al., Ref. 46).

et al.108 used FMM in combination with LFM to detect differences in packing density in a patterned alkanethiol film. A patterned film consisted of two types of n-alkanethiol molecules (CH3(CH2)n1SH) with different chain length (n 7 and 18). The octadecylthiol (n 18) was stamped onto a gold surface using microcontact printing (CP) and subsequently a solution deposition of heptylthiol (n 7) molecule in an ethanol solution. Comparing the Young’s modules for the two molecules, the octadecylthiol has a higher (i.e. stiffer) value, yet the FMM measurement shows the region to be less stiff. The reason for the reduced stiffness is that the degree of packing for the CP region is less dense as compared to the solution deposited region. This was confirmed by using the same alkanethiol for CP and solution deposition and observing a similar difference in the FFM and LFM image but with no changes in the topography.

17.5

SPM directed modification

Besides being an analytical instrument for analysis, the SPM is also a powerful lithography tool.110–116 The demonstration of the STM to manipulate individual atoms on a surface can be considered the ultimate in lithography resolution whereby individual atoms can be built up (positive structures) or removed

450

L. A. Nagahara

(negative structures) to form a desired pattern.117,118 The types of modification using SPM can be grouped into two main categories: mechanical and electrical. In the mechanical modification mode, the SPM probe acts as a plow to displace material or as a tweezer to extract material. The aspects of mechanical modification will not be discussed here due to its limited role in EMT processes. In electrical modification, a current is typically passed between the SPM probe and sample. A large number of studies have been conducted on using SPM as proximal e-beam lithography tools. However, besides passing a current, electrical modification can also result or be enhanced from the intense electric fields that exist between the probe and substrate. A STM tip typically generates an electric field on the order of 107 V/cm between the tip and substrate, enough to desorb some physisorb species and modify surfaces.119–121 As in the case of SPM imaging, AFM lithography has had a wider impact in this field since first being reported in 1992.122,123 The reasons are as follows: the STM cannot be operated under tunneling mode over insulating materials (e.g. polymer resists or oxide layer); however, STM has been used in a limited role in field emission modes.124–126 There is no control on the force (i.e. distance feedback regulation) between the STM tip and resist film. As such, surface irregularites or variations in film thickness can therefore push the tip into the film thereby damaging it. Finally, inspection of the exposed area is sometimes necessary. Unfortunately, imaging in a field emission mode can inadvertently chemically modify the film. AFM lithography circumvents these problems altogether. In STM, the same feedback mechanism used to maintain a constant tip/sample distance is directly coupled with the current (dosage). However, in AFM, the principle for maintaining tip/sample distance depends only on the interacting forces between the tip and sample which is completely independent from any current or applied voltage. In other words, the AFM has complete control over the amount of exposure to a surface while simultaneously maintaining control over the amount of force applied by the AFM tip to the surface. However, STM lithography does have some advantages, especially in aqueous environments, over AFM lithography. Unlike STM tips, there is no technique available to insulate a metallized AFM probe such that only the apex is uninsulated. An uninsulated AFM probe would not confine any electrochemical reactions to a localized area beneath the probe.

17.5.1

Semiconductor/metal substrate modification

Field-induced oxidation of silicon is among the most widely used methods to fabricate nanometer-scale patterns and nanoelectronic devices. Dagata et al. were the first to report the desorption of hydrogen from a hydrogen passivated Si(111) surface and the subsequent growth of a silicon oxide layer.120 Independently, Nagahara and co-workers also observed a similar electric field phenomena on Si(100) and GaAs(100) in solution with in situ STM.127 They immersed the samples in a dilute HF solution and observed a depression in the area directly below

SPM as an analysis tool 451 (a)

(b)

1000

1000 750

750 500

500 250

250 0 0

1000

1000 750

750 500

500 250

250 0 0

Figure 17.12 Three-dimensional view of STM image before (a) and after (b) patterning in a dilute solution of HF. A 200 nm 200 nm region clearly shows the modification done by the STM (Nagahara, Ref. 121).

the STM tip. The authors proposed that an oxide layer was formed directly underneath the STM tip and the HF solution subsequently removes the oxide layer, leaving behind a depression as shown in Figure 17.12. In many surface modification experiments conducted under ambient conditions, the modified regions are really the result of a thin water layer that exists at the surface. Hence, many of these results can be applicable to EMT systems. Various semiconductors have been modified with SPM under a thin water layer or in aqueous environments.120,127–134 Wang et al. have used the AFM to anodically oxidize a thin film of Cr deposited onto Si.135 The smallest feature size obtained was about 20 nm. Sugimura and Nakagiri and co-workers have done extensive studies and tested various application on these substrates.136–155 Pérez-Murano and co-workers have used various modes of AFM to locally oxidize hydrogen passivated Si surfaces.129–131,156–159 As in contact AFM, they observed that a minimum voltage must be applied for starting the oxidation process. A short voltage pulse (5 ms) was needed to polarize the water layer adsorbed onto the sample and to form a water bridge between the tip and sample. Two mechanisms for oxide formation have emerged from these SPM studies. In one case, electrochemical oxidation appears to be the formation mechanism;160–164 while in the second case, a field-induced oxidation is the modification process.129–131,138,149,156–159,165–177 17.5.2 Adsorbate modification Zamborini and Crooks have described a method to pattern a SAM coated film with Ag using a STM tip.178 The authors first electrochemically deposited a small amount of Ag onto the STM tip, before forming a pattern in a SAM resist film of alkanethiol on Au(111) substrate. Next, the lithography step was performed with the STM tip, by increasing the bias between the tip and sample to 2.7 to 3.0 V. The SAM layer and a few layers of the Au directly below the tip were removed during this process. Finally, the tip was biased from 0.6 to 1.0 V to oxidize and remove the Ag on the tip and redeposited only on the exposed Au region as shown in Figure 17.13. Electrodeposition occurred only under humid conditions

452

L. A. Nagahara

(a)

(b) 100 nm

Eb = 400 mV

Eb = –400 mV

Figure 17.13 STM image of modification performed by the STM over a 100 nm 100 nm, 50 nm 50 nm, and 25 nm 25 nm region. (a) Initially hexadecanethiol was selectively removed from these regions by applying a sample bias 2.7 V over the region; (b) after applying a sample bias of 400 mV, the same region shows Ag deposition in the areas where the hexadecanethiol was removed (Zamborini and Crooks, Ref. 178).

and the authors attributed the reaction to a thin water condensation layer between the tip and sample. Patterns as small at 25 nm were produced in this manner. Most of the deposited Ag can be removed from the Au surface at positive biases (400 to 1200 mV). However, some of the deposited Ag in the modified region could not be removed easily, even with high positive potential bias. Moreover, the Ag appeared to spread underneath the SAM layer and the SAM molecules appeared to cover Ag in the pattern region. The authors speculated that the SAM molecules act to stabilize the Ag deposit surface which is supported by the fact that the Ag became more stable with time. Mizutani and co-workers used an STM to selectively remove nonanethiol from a Au(111) surface by applying a voltage pulse between the STM tip and sample.179 A pristine film of nonanethiol was formed on the Au surface. By applying 2.6–3.0 V to the sample, the nonanethiol molecules were removed leaving behind a 2–5 nm diameter hole. After hole formation, the void left behind was filled in with existing nonanethiol molecules in solution. Several groups have used SAM as a resist layer but this is believed to be the first report of a SAM surface “healing” after a surface modification. Presumably, surface contamination or residual from the SAM molecules may contribute to the stability of the modified region. 17.5.3

Metal deposition

Kolb’s group have used the STM in an electrochemical environment to deposit small Cu clusters (two to four atomic layers in height) onto a Au(111)

SPM as an analysis tool 453 180–183

This was achieved by initially depositing Cu onto the STM tip, surface. which acted as the source for the Cu to be placed on the surface (see Figure 17.14). Next the Cu was transferred from the STM tip to the Au(111) surface as the tip/sample distance was reduced to the point where the Cu on the STM tip made contact with the Au(111) surface. Subsequently, a small Cu cluster was deposited onto the surface and the tip/sample distance was then increased. The mechanism for the cluster formation is a phenomenon known as “jump-tocontact”. As the tip/sample distance is reduced, a connecting neck is formed between the tip and sample. During retraction, the metallic neck breaks and leaves a cluster behind on the surface. It should be noted that the Cu cluster is not deposited directly onto the Au(111) surface but rather on a monolayer of Cu. The potential at which the Au(111) surface is held is at the UPD potential for Cu. The Kolb group has automated this procedure by having a separate microprocessor unit control the voltages applied to the x-, y-, and z-piezos attached to the STM tip. With this system, they have been able to make an array (100 100) of 10 000 Cu isolated clusters in less than 20 min as shown in Figure 17.15.182 Each of the clusters had an average height of 1.1 0.2 nm and was spaced 11 nm apart from each other. The size of the cluster can be controlled in a limited manner by varying the tip/sample separation distance. The size of the cluster increased linearly over a limited range, then decreased as the tip penetrated the Au surface. In addition to isolated clusters, the group has also fabricated “conducting wires” decreasing the separation distance between clusters to the point where they contact each other. In a subsequent paper, the Kolb group investigated depositing the Cu cluster onto a Ag(111) surface.183 In contrast to the Au(111) surface, the Cu cluster could not be formed uniformly onto the Ag(111) surface. The authors attributed this

STM tip

Cu2+

Cu cluster

Au surface

Figure 17.14 Schematic illustration of the method used to transfer Cu from the Cucovered STM tip to the Au(111) substrate (Kolb et al., Ref. 181).

454

L. A. Nagahara

2 nm 0 0

286 nm

Figure 17.15 STM image of an array of 100 100 Cu dots on a Au(111) surface in a solution of 0.05 M H2SO4 and 1 mM CuSO4. The clusters have an average height of 0.7 nm and are separated by 11 nm (Engelmann et al. Ref. 182).

difference to the absence of a UPD Cu layer. In the case of Au(111), a copper monolayer was formed, however this monolayer did not form in the case of Ag(111). Instead of small clusters forming on the surface, Cu islands with a diameter as large as 45 nm were observed. The larger amount of material transferred was further complicated by the fact that the islands were not uniform in size.

17.6

Future directions

With such widespread use of SPM around the world, there are a number of directions in which this field is pursuing that can aid EMT in the future. For example, novel SPM techniques that further elucidate analysis are continuously being reported in the literature. For example, one scanning probe technique noticeably absent in the previous section is scanning near-field optical microscopy (SNOM) or near-field scanning optical microscopy (NSOM). This recent member to the SPM family has much potential as an analysis tool as will be described shortly. Other areas that will further aid SPM advancement are probe development and materials that are ideally suited for SPM application. Finally, continual improvement in instrumentation, such as automation and parallel operation, will also further help SPM’s role as an analysis tool. 17.6.1

Near-field optical microscopy

Optical characterization of materials is probably the most widely used tool in an analysis lab. A typical analysis lab would have an optical setup to do one or more

SPM as an analysis tool 455 of the following: fluorescence, photoluminescence, IR and Raman, FTIR, ellipsometry, UV–Vis spectrometry. Optical microscopes are routinely used with various contrast mechanisms such as polarization, phase contrast, absorption, reflection, and transmission. In all of these techniques, the resolution is limited by diffraction to about ␭/2, where ␭ is the wavelength of light. This limitation arises from the fact that the detection element, typically an optical lens, is many wavelengths away from the sample of interest. If, on the other hand, the illumination source or detector is positioned to less than the wavelength of light away from the sample, the resolution limit is dependent on only the probe size and the probe/sample distance. This small gap (␭/2) distance is known as the “near-field” regime and hence the name near-field optics. A typical NSOM probe consists of a tapered single mode optical fiber coated with Al or some other reflecting metal. The purpose of the metal film is to replace the cladding material and act as a waveguide to confine the light to within the fiber optics as it travels through the tapered portion of the fiber before exiting at the apex. Resolution as high as 12 nm has been reported in the literature. One complication with NSOM is that the amount of light transmitted through the probe decreases exponentially with decreasing probe diameter. As such, high resolution imaging will require considerable amount of acquisition time. Nevertheless, for mesoscopic scales, NSOM can play an important role in elucidating EMT related issues.

17.6.2

Carbon nanotubes

Carbon nanotubes (CNTs)184,185 are the latest addition to the carbon fullerene family and show remarkable structural, mechanical, and electrical properties. A simplistic way to describe a CNT is a rolled up single cylindrical sheet of graphene (single-walled) or as concentric cylindrical sheets (multi-walled). A singlewalled nanotube (SWNT) is typically about one nanometer in diameter and can stretch for several hundred micrometers in length (i.e. up to 10 000 to 1 aspect ratio). CNTs are also identified as being well suited for use as SPM probe tips.186–192 One such CNT attached AFM probe is shown in Figure 17.16. The nanotube is ideal for probing structures due to its high aspect ratio and thus providing structural information with minimal probe/sample convolution in the image. The extended length of the CNT can also be beneficial in LFM in which the degree of torsional motions is amplified by the extended tube length. The radius of curvature of the CNT is smaller than commercially available LFM probes. Moreover, the smaller radius of curvature aids in reducing adhesive forces during contact AFM. CNTs are also quite elastic and buckle easily under a small force. This is particularly suited for imaging soft materials, such as biological and organic materials, which tend to distort quite easily with an applied force in the nN range. Previously, amorphous carbon tips on the end of AFM probes induced by electron beam deposition (EBD) have been reported.193–196 The electron beam decomposes the residual gases in an SEM chamber resulting in the formation of

456

L. A. Nagahara (a)

µm

15

10

5

0 (b)

(c) µm µm

0.2 6

4 0.1 2

0

0

Figure 17.16 (a and b) SEM images of carbon nanotube (CNT) bundle attached to a silicon cantilever using an acrylic adhesive from an adhesive-coated carbon tape. The CNT are bundled but a single CNT is extended at the apex of the CNT bundle; (c) TEM image showing a single CNT (5 nm diameter) extending 250 nm (Dai et al.186).

amorphous carbon. The tip is formed by holding the electron beam stationary at the AFM probe apex and allowing the amorphous carbon to build up. In order for CNT to be really accessible for general use, a method for fabrication that will generate hundreds to thousands of nearly identically prepared probes needs to be developed. At present, extremely laborious methods of attaching CNT to the ends of AFM probes exists. One popular method is to attach a small drop of adhesive to the end of the AFM probe and attaching a probe to the drop. CNTs are generally too long to be useful as AFM probes. Typically, the CNT attached probe is shorted with voltage pulse applied between the probe and sample.

SPM as an analysis tool 457 CNT probes are already being used in many SPM applications. The Leiber group have chemically functionalized the end of multi- and single-walled CNTs for use in CFM.45,48,188,191 The end of the CNT has a carboxylic acid group attached. Amines are then coupled to the carboxyl group to make the probe hydrophobic or with benzylamine to make the probe hydrophilic. The lateral resolution obtained using the single-walled CNT is approximately 3 nm and is significantly better than those obtained using other methods.188 In addition, they have measured the binding force of a biotin–streptavidin system by covalently attaching a biotin moiety to a CNT probe and approaching the probe to a streptavidin covered mica surface.191 Well defined binding events (~200 pN) were recorded and attributed to the biotin–streptavidin interaction. Wong et al. have estimated that their SWNT probe had an estimated radius of curvature of 3.4 nm. The authors have used these probes to image other nanotubes and DNA. Dai and co-workers have attached CNT to SPM probes for nanolithography.192 They have routinely achieved 10 nm resolution with multi-walled CNT and expect to achieve 1 nm resolution using a single-walled CNT. The key advantage for using CNT as probes for SPM nanolithography is durability. CNT has the highest Young’s modulus (~1 TPa) known and no noticeable degradation in the CNT probe was observed after prolonged patterning. 17.6.3

Multi-probe development

SPM-related techniques can clearly pattern smaller linewidth features than any other existing technique as demonstrated by the numerous publications on atomic and molecular manipulation. In actual device manufacturing, linewidth and critical dimension tolerance are important parameters and become more critical with miniaturization. The semiconductor industry provides a good example of this trend that will occur to EMT systems in the future. According to the 2001 Semiconductor Industry Association (SIA) roadmap, feature size of 0.15 m linewidths will require 5.3 nm tolerance and feature size of 45 nm linewidths will require 1.5 nm control.197 Although optical lithography is not capable of fabricating 50 nm feature, electron beam lithography (EBL) has been patterning such linewidth for many years. Scanning probe lithography (SPL) is now developed to the point where it is comparable to patterning features written currently with EBL.198 Another of the main driving forces in device fabrication is the speed or throughput of manufacturing. However, the inability to pattern features over an entire wafer size at high speed will ultimately limit the role of SPM as an on-line fabrication tool. Conventional EBL provides a comparative example. Although EBL has shown superior linewidth resolution and has been in use for several decades, optical steppers are still the workhorses for lithography done in semiconductor manufacturing due to throughput concerns. This is not to say that SPM has no future in device related processing. One of the future directions for SPM-related technology is the development of faster systems.199–202 Quate and co-workers have shown that multiple AFM cantilevers can be microfabricated and operated simultaneously to form patterned features.133,201,203 Operating an array of AFM cantilevers in parallel offers a method to drastically increase fabrication

458

L. A. Nagahara

speed and several groups around the world are pursuing a similar parallel probe approach.204 Marrian et al. have estimated that an array of 1000 STM tips can write a mask having a minimum feature of 30 nm over a 4 cm2 area with a 50% fill factor with 5 h.205, 206 This speed is comparable to current microfabrication technology.

17.7

Summary

The future for SPM as an analysis tool looks promising in assisting EMT systems. Examples of the measurement and fabrication of both positive and negative structures, localization of reactions even in the case of high aspect ratios, precise determination of nucleation sites, and growth rate by potential or current control were given. Besides imaging surface features, one area in which SPM can also provide a considerable impact is the observation of events in real time. Obvious examples are corrosion and deposition. The observation of dynamical processes at the micro- to nanometer scale will further elucidate electrochemical microsystems technologies.

Acknowledgment L.A.N. would like to thank Dr Jing Shi (Motorola, Inc.) for providing the MFM image, Dr Patrick Oden (Oak Ridge National Labs) for helpful discussions, and Prof. Nongjian Tao (Florida International University) for providing the Au(111) reconstruction images. L.A.N. is grateful to his wife, Aya, for having the patience and giving the time to write this section.

References 1. R. J. Behm, N. Garcia, and H. Rohrer, Scanning Tunneling Microscopy and Related Methods (Kluwer, Dordrecht, 1990). 2. D. Sarid, Scanning Force Microscopy: With Application to Electric, Magnetic, and Atomic Force (Oxford University Press, New York, 1991). 3. S. Gauthier and C. Joachim, Scanning Probe Microscopy: Beyond the Images (Les editions de physique, Les Ulis, 1992). 4. H.-J. Güntherodt and R. Wiesendanger, Scanning Tunneling Microscopy I (Springer, Berlin, 1992). 5. R. Wiesendanger and H.-J. Güntherodt, Scanning Tunneling Miroscopy II (Springer, Berlin, 1992). 6. R. Wiesendanger and H.-J. Güntherodt, Scanning Tunneling Microscopy III (Springer, Berlin, 1993). 7. D. A. Bonnell, Scanning Tunneling Microscopy and Spectroscopy: Theory, Techniques, and Applications (VCH, New York, 1993). 8. J. Stroscio and W. Kaiser, Scanning Tunneling Microscopy (Academic, Boston, 1993). 9. C. J. Chen, Introduction to Scanning Tunneling Microscopy (Oxford University Press, New York, 1993).

SPM as an analysis tool 459 10. R. Wiesendanger, Scanning Probe Microscopy and Spectroscopy: Methods and Applications (University Press, Cambridge, 1994). 11. R. Wiesendanger, J. Vac. Sci. Technol. B 12, 515 (1994). 12. L. A. Bottomley, J. E. Coury, and P. N. First, Anal. Chem. 68, 185R (1996). 13. S. O. Vansteenkiste, M. C. Davies, C. J. Roberts, S. J. B. Tendler, and P. M. Williams, Prog. Surf. Sci. 57, 95 (1998). 14. K. D. Jandt, Materials Science and Engineering R21, 211 (1998). 15. L. A. Bottomley, Anal. Chem. 70, 425R (1998). 16. A. J. Bard, F.-R. F. Fan, J. Kwak, and O. Lev, Anal. Chem. 61, 132 (1989). 17. A. J. Bard, G. Denault, C. Lee, D. Mandler, and D. O. Wipf, Acc. Chem. Res. 23, 357 (1990). 18. A. J. Bard, F.-R. F. Fan, D. T. Pierce, P. R. Unwin, D. O. Wipf, and F. Zhou, Science 254, 68 (1991). 19. A. J. Bard, F.-R. F. Fan, and M. V. Mirkin, in Electroanalytical Chemistry (Dekker, New York, 1994), pp. 243. 20. G. Binnig and H. Rohrer, Helv. Phys. Acta. 55, 726 (1982). 21. G. Binnig, H. Rohrer, C. Gerber, and E. Weibel, Phys. Rev. Lett. 49, 57 (1982). 22. G. Binnig, H. Rohrer, C. Gerber, and E. Weibel, Phys. Rev. Lett. 50, 120 (1983). 23. K. Takayanagi, Y. Tanishiro, S. Takahashi, and M. Tahahashi, Surf. Sci. 164, 367 (1985). 24. S. I. Park and C. F. Quate, Appl. Phys. Lett. 49, 1172 (1986). 25. R. Sonnenfeld and P. K. Hansma, Science 232, 211 (1986). 26. R. M. Feenstra, J. A. Stroscio, J. Tersoff, and A. P. Fein, Phys. Rev. Lett. 58, 1192 (1987). 27. M. J. Heben, M. M. Dovek, N. S. Lewis, R. M. Penner, and C. F. Quate, J. Microsco. 152, 651 (1988). 28. J. Schneir, P. K. Hansma, V. Elings, J. Gurley, K. Wickramasinghe, and R. Sonnenfeld, Proc. SPIE 897, 16 (1988). 29. J. Wiechers, T. Twomey, D. M. Kolb, and R. J. Behm, J. Electroanal. Chem. 248, 451 (1988). 30. L. A. Nagahara, T. Thundat, and S. M. Lindsay, Rev. Sci. Instrum. 60, 3128 (1989). 31. G. Binnig, C. F. Quate, and C. Gerber, Phys. Rev. Lett. 56, 930 (1986). 32. Q. Zhong, D. Inniss, K. Kjoller, and V. B. Elings, Surf. Sci. Lett. 290, L688 (1993). 33. S. M. Lindsay, Y. L. Lyubchenko, N. J. Tao, Y. Q. Li, P. I. Oden, J. A. DeRose, and J. Pan, J. Vac. Sci. Technol. 11, 808 (1993). 34. M. Lantz, S. J. O’Shea, and M. E. Welland, Appl. Phys. Lett. 65, 409 (1994). 35. W. Han, S. M. Lindsay, and T. Jing, Appl. Phys. Lett. 69, 4111 (1996). 36. B. Drake, C. B. Prater, A. L. Weissenhorn, S. A. C. Gould, T. R. Albrecht, C. F. Quate, D. S. Cannell, H. G. Hansma, and P. K. Hansma, Science 243, 1585 (1989). 37. T. Thundat, X. Y. Zheng, G. Y. Chen, S. L. Sharp, R. J. Warmack, and L. J. Schowalter, Appl. Phys. Lett. 63, 2150 (1993). 38. R. L. Alley, K. Komvopoulos, and R. T. Howe, J. Appl. Phys. 76, 5731 (1994). 39. P. Maivald, H. J. Butt, S. A. C. Gould, C. B. Prater, B. Drake, J. A. Gurley, V. B. Elings, and P. K. Hansma, Nanotechnology 2, 103 (1991). 40. H. A. Mizes, K.-G. Loh, R. J. D. Miller, S. K. Ahuja, and E. F. Grabowski, Appl. Phys. Lett. 59, 2901 (1991). 41. T. R. Albrecht, S. Akamine, T. E. Carver, and C. F. Quate, J. Vac. Sci. Technol. A 8, 3386 (1990). 42. O. Wolter, T. Bayer, and J. Greschner, J. Vac. Sci. Technol. B 9, 1353 (1991).

460

L. A. Nagahara

43. P. I. Oden, L. A. Nagahara, J. J. Graham, J. Pan, N. J. Tao, Y. Li, T. G. Thundat, J. A. DeRose, and S. M. Lindsay, Ultramicroscopy 42–44, 580 (1992). 44. L. A. Nagahara, T. Ohmori, K. Hashimoto, and A. Fujishima, J. Vac. Sci. Technol. A 11, 763 (1993). 45. C. D. Frisbie, L. F. Rozsnyai, A. Noy, M. S. Wrighton, and C. M. Lieber, Science 265, 2071 (1994). 46. S. Akari, D. Horn, H. Heller, and W. Schrepp, Adv. Mater. 7, 549 (1995). 47. E. W. van der Vegte and G. Hadziioannou, Langmuir 13, 4357 (1997). 48. A. Noy, C. H. Sanders, D. V. Vezenov, S. S. Wong, and C. M. Lieber, Langmuir 14, 1508 (1998). 49. P. Lustenberger, H. Rohrer, R. Christoph, and H. Siegenthaler, J. Electroanal. Chem. 243, 225 (1988). 50. K. Itaya and E. Tomita, Surf. Sci. 201, L507 (1988). 51. A. Arvia, Surf. Sci. 181, 78 (1987). 52. H. Siegenthaler and R. Christoph, in Scanning Tunneling Microscopy and Related Methods (Kluwer Academic, Dordrecht, 1990), p. 315. 53. R. Sonnenfeld, J. Schneir, and P. K. Hansma, in Modern Aspects of Electrochemistry (Plenum, New York, 1990) p. 1. 54. T. R. I. Cataldi, I. G. Blackham, G. A. D. Briggs, J. B. Pethica, and H. A. O. Hill, J. Electroanal. Chem. 290, 1 (1990). 55. H. Siegenthaler, in Scanning Microscopy II: Futher Applications and Related Scanning Techniques (Springer-Verlag, Berlin, 1992), p. 7. 56. P. A. Christensen, Chem. Soc. Rev. 21, 197 (1992). 57. D. M. Kolb, R. J. Nichols, and R. J. Behm, in Electrified Interfaces in Physics, Chemistry and Biology (C. Kluwer Academic, 1992), p. 315. 58. M. J. Weaver and X. P. Gao, Annu. Rev. Phys. Chem. 44, 459 (1993). 59. M. J. Weaver, J. Phys. Chem. 100, 13079 (1996). 60. A. J. Bard, H. D. Abruna, C. E. D. Chidsey, L. R. Faulkner, S. W. Feldberg, K. Itaya, M. Majda, O. Melroy, R. W. Murray, M. D. Porter, M. P. Soriaga, and H. S. White, J. Phys. Chem. 97, 7147 (1993). 61. A. A. Gewirth and H. Siegenthaler, Nanoscale Probes of the Solid/Liquid Interface (Kluwer Academic, Dordrecht, 1995). 62. A. A. Gewirth and B. K. Niece, Chem. Rev. 97, 1129 (1997). 63. J. Wang, A. J. Davenport, H. S. Isaacs, and B. M. Ocko, Science 255, 1416 (1992). 64. D. M. Kolb, Prog. Surf. Sci. 51, 109 (1996). 65. C. Wöll, S. Chiang, R. J. Wilson, and P. H. Lippel, Phys. Rev. B 39, 7988 (1989). 66. J. V. Barth, H. Brune, G. Ertl, and R. J. Behm, Phys. Rev. B 42, 1407 (1990). 67. N. J. Tao and S. M. Lindsay, J. Appl. Phys. 70, 5141 (1991). 68. X. Gao, A. Hamelin, and M. J. Weaver, J. Chem. Phys. 95, 6993 (1991). 69. N. J. Tao and S. M. Lindsay, J. Phys. Chem. 96, 5213 (1992). 70. N. J. Tao and S. M. Lindsay, Surf. Sci. Lett. 274, L546 (1992). 71. P. I. Oden, N. J. Tao, and S. M. Lindsay, J. Vac. Sci. Technol. B 11, 137 (1993). 72. H. Angerstein-Kozlowska, B. E. Conway, A. Hamelin, and L. Stoicoviciu, Electrochim. Acta 31, 1051 (1986). 73. H. Angerstein-Kozlowska, B. E. Conway, A. Hamelin, and L. Stoicoviciu, J. Electroanal. Chem. 228, 429 (1987). 74. H. Angerstein-Kozlowska, B. E. Conway, K. Tellefsen, and B. Barnett, Electrochim. Acta 34, 1045 (1989). 75. B. E. Conway, Prog. Surf. Sci. 49, 331 (1995).

SPM as an analysis tool 461 76. D. J. Trevor, D. E. D. Chidsey, and D. N. Loiacono, Phys. Rev. Lett. 62, 929 (1989). 77. H. Honbo, S. Sugawara, and K. Itaya, Anal. Chem. 62, 2424 (1990). 78. X. Gao and M. J. Weaver, J. Electroanal. Chem. 367, 259 (1994). 79. M. A. Schneeweiss, D. M. Kolb, D.-Z. Liu, and D. Mandler, Can. J. Chem. 75, 1703 (1997). 80. O. M. Magnussen, J. Hotlos, R. J. Nichols, and D. M. Kolb, Phys. Rev. Lett. 63, 2929 (1990). 81. S. Manne, P. K. Hansma, J. Massie, V. B. Elings, and A. A. Gewirth, Science 251, 183 (1991). 82. N. Ikemiya, S. Miyaoka, and S. Hara, Surf. Sci. 311, L641 (1994). 83. T. Hachiya, H. Honbo, and K. Itaya, J. Electroanal. Chem. 315, 275 (1991). 84. N. Batina, T. Will, and D. M. Kolb, Faraday Disc. 94, 93 (1992). 85. M. H. Hölzle and D. M. Kolb, Ber. Bunsenges. Phys. Chem. 98, 330 (1994). 86. B. C. Schardt, S.-L. Yau, and F. Rinaldi, Science 243, 1050 (1989). 87. T. Yamada, N. Batina, and K. Itaya, Surf. Sci. 335, 204 (1995). 88. T. Yamada, N. Batina, and K. Itaya, J. Phys. Chem. 99, 8817 (1995). 89. N. Batina, T. Yamada, and K. Itaya, Langmuir 11, 4568 (1995). 90. Y. Nagatani, T. Hayashi, T. Yamada, and K. Itaya, Jpn. J. Appl. Phys. 36, 720 (1996). 91. A. Ulman, Introduction to Ultrathin Organic Films from Langmuir-Blodgett to Self-Assembly (Academic Press, San Diego, CA, 1991). 92. L. H. Dubois and R. G. Nuzzo, Annu. Revs. Phys. Chem. 437 (1992). 93. G. E. Poirier and E. D. Pylant, Science 272, 1145 (1996). 94. G. E. Poirier, Chem. Rev. 97, 1117 (1997). 95. G. E. Poirier, Langmuir 13, 2019 (1997). 96. K. Edinger, M. Grunze, and C. Wöll, Ber. Bunsenges. Phys. Chem. 101, 1811 (1997). 97. M. Hara, H. Sasabe, and W. Knoll, Thin Solid Films 273, 66 (1996). 98. M. H. Dishner, J. C. Hemminger, and F. J. Feher, Langmuir 13, 2318 (1997). 99. J. Pan, N. Tao, and S. M. Lindsay, Langmuir 9, 1556 (1993). 100. H. Ohno, L. A. Nagahara, S. Gwo, W. Mizutani, and H. Tokumoto, Jpn. J. Appl. Phys. 35, L512 (1996). 101. H. Ohno, L. A. Nagahara, S. Gwo, W. Mizutani, and H. Tokumoto, Mol. Cryst. Liq. Cryst. 295, 189 (1997). 102. G. Bar, S. Rubin, T. N. Taylor, B. I. Swanson, T. A. Zawodzinski, Jr., J. T. Chow, and J. P. Ferraris, J. Vac. Sci. Technol. A 14, 1794 (1996). 103. Y. Zhou, H. Fan, T. Fong, and G. P. Lopez, Langmuir 14, 660 (1998). 104. S.-S. Wong, H. Takano, and M. D. Porter, Anal. Chem. 70, 5209 (1998). 105. C. E. H. Berger, K. O. van der Werf, R. P. H. Kooyman, G. B. de Grooth, and J. Greve, Langmuir 11, 4188 (1995). 106. R. M. Overney, D. P. Leta, L. J. Fetters, Y. Liu, M. H. Rafailovich, and J. Skoklov, J. Vac. Sci. Technol. B 14, 1276 (1996). 107. G. Haugstad, W. L. Gladfelter, and R. R. Jones, J. Vac. Sci. Technol. A 14, 1864 (1996). 108. G. Bar, S. Rubin, A. N. Parikh, B. I. Swanson, T. A. Zawodzinski, Jr., and M. H. Whangbo, Langmuir 13, 373 (1997). 109. M. Radmacher, R. W. Tillmann, M. Fritz, and H. E. Gaub, Science 257, 1900 (1992). 110. G. M. Shedd and P. E. Russell, Nanotechnology 1, 67 (1990). 111. C. F. Quate, in Highlights in Condensed Matter Physics and Future Prospects (Plenum, New York, 1991), p. 573.

462

L. A. Nagahara

112. C. R. K. Marrian, Technology of Proximal Probe Lithography (SPIE, Bellingham, 1993). 113. P. Avouris, Atomic and Nanometer-Scale Modification of Materials: Fundamentals and Applications (Kluwer, Dordrecht, 1993). 114. L. A. Nagahara, H. Ohno, and H. Tokumoto, J. Photopolym. Sci. Technol. 8, 669 (1995). 115. C. Schoenenberger and N. Kramer, Microelectron. Eng. 32, (1996). 116. R. M. Nyffenegger and R. M. Penner, Chem. Rev. 97, 1195 (1997). 117. D. M. Eigler and E. K. Schweizer, Nature 344, 524 (1990). 118. I.-W. Lyo and P. Avouris, Science 253, 173 (1991). 119. T. Thundat, L. A. Nagahara, P. I. Oden, S. M. Lindsay, M. A. George, and W. S. Glaunsinger, J. Vac. Sci. Technol. 8, 3537 (1990). 120. J. A. Dagata, J. Schneir, H. H. Harary, C. J. Evans, M. T. Postek, and J. Bennett, Appl. Phys. Lett. 56, 2001 (1990). 121. L. A. Nagahara, in Proceedings of the First International Symposium on Electrochemical Microfabrication (The Electrochemical Society, Pennginton, NJ, 1991), p. 254. 122. L. A. Nagahara, P. I. Oden, A. Majumdar, J. P. Carrejo, J. Graham, and J. Alexander, SPIE 1639, 171 (1992). 123. A. Majumdar, P. I. Oden, J. P. Carrejo, L. A. Nagahara, J. J. Graham, and J. Alexander, Appl. Phys. Lett. 61, 2293 (1992). 124. E. A. Dobisz, H. W. P. Koops, F. K. Perkins, C. R. K. Marrian, and S. L. Brandow, J. Vac. Sci. Technol. B 14, 4148 (1996). 125. E. A. Dobisz, H. W. P. Koops, and F. K. Perkins, Appl. Phys. Lett. 68(25), 3653 (1996). 126. T. M. Mayer, D. P. Adams, and B. M. Marder, J. Vac. Sci. Technol. B 14, 2438 (1996). 127. L. A. Nagahara, T. Thundat, and S. M. Lindsay, Appl. Phys. Lett. 57, 270 (1990). 128. T. Thundat, L. A. Nagahara, and S. M. Lindsay, J. Vac. Sci. Technol. A 8, 539 (1990). 129. R. García, M. Calleja, and F. Pérez-Murano, Appl. Phys. Lett. 72, 2295 (1998). 130. J. H. Ye, F. Pérez-Murano, N. Barniol, G. Abadal, and X. Aymerich, J. Vac. Sci. Technol. B 13, 1423 (1995). 131. J. H. Ye, F. Pérez-Murano, N. Barniol, G. Abadal, and X. Aymerich, J. Phys. Chem. 99, 17650 (1995). 132. S. C. Minne, H. T. Soh, P. Flueckiger, and C. F. Quate, Appl. Phys. Lett. 66, 703 (1995). 133. S. C. Minne, P. Flueckiger, H. T. Soh, and C. F. Quate, J. Vac. Sci. Technol. B 13, 1380 (1995). 134. G. Abadal, F. Pérez-Murano, N. Barniol, X. Borrisé, and X. Aymerich, Ultramicroscopy 66, 133 (1996). 135. D. Wang, L. Tsau, and K. L. Wang, Appl. Phys. Lett. 67, 1295 (1995). 136. H. Sugimura, T. Uchida, N. Kitamura, and H. Masuhara, Jpn. J. Appl. Phys. 32, L553 (1993). 137. H. Sugimura, T. Uchida, N. Kitamura, and H. Masuhara, Appl. Phys. Lett. 63, 1288 (1993). 138. H. Sugimura, N. Kitamura, and H. Masuhara, Jpn. J. Appl. Phys. 33, L143 (1994). 139. H. Sugimura, T. Uchida, N. Kitamura, and H. Masuhara, J. Vac. Sci. Technol. B 12, 2884 (1994). 140. H. Sugimura, T. Yamamoto, N. Nakagiri, M. Miyashita, and T. Onuki, Appl. Phys. Lett. 65, 1569 (1994). 141. H. Sugimura, T. Uchida, N. Kitamura, and H. Masuhara, J. Phys. Chem. 98, 4352 (1994).

SPM as an analysis tool 463 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179.

H. Sugimura and N. Nakagiri, J. Vac. Sci. Technol. B 13, 1933 (1995). H. Sugimura and N. Nakagiri, Jpn. J. Appl. Phys. 34, 3406 (1995). H. Sugimura and N. Nakagiri, Jpn. J. Appl. Phys. 34, L698 (1995). H. Sugimura, N. Nakagiri, and N. Ichinose, Appl. Phys. Lett. 66, 3686 (1995). H. Sugimura and N. Nakagiri, Nanotechnology 6, 29 (1995). H. Sugimura and N. Nakagiri, Appl. Phys. Lett. 66, 1430 (1995). H. Sugimura and N. Nakagiri, Langmuir 11, 3623 (1995). H. Sugimura and N. Nakagiri, Thin Solid Films 281–282, 572 (1996). H. Sugimura and N. Nakagiri, J. Vac. Sci. Technol. A 14, 1223 (1996). H. Sugimura, H. Okiguchi, and N. Nakagiri, Jpn. J. Appl. Phys. 35, 3749 (1996). H. Sugimura, K. Okiguchi, N. Nakagiri, and M. Miyashita, J. Vac. Sci. Technol. B 14, 4140 (1996). H. Sugimura and N. Nakagiri, Thin Solid Films 273, 245 (1996). H. Sugimura and N. Nakagiri, J. Am. Chem. Soc. 119, 9226 (1997). H. Sugimura and N. Nakagiri, J. Vac. Sci. Technol. B 15, 1394 (1997). N. Barniol, F. Pérez-Murano, and X. Aymerich, Appl. Phys. Lett. 61, 462–464 (1992). F. Pérez-Murano, N. Barniol, and X. Aymerich, J. Vac. Sci. Technol. B 11, 651 (1993). J. Servat, P. Gorostiza, F. Sanz, F. Pérez-Murano, N. Barniol, G. Abadal, and X. Aymerich, J. Vac. Sci. Technol. A 14, 1208 (1996). G. Abadal, F. Pérez-Murano, N. Barniol, and X. Aymerich, Thin Solid Films 317, 493 (1998). E. S. Snow, W. H. Juan, S. W. Pang, and P. M. Campbell, Appl. Phys. Lett. 66, 1729 (1995). K. Matsumoto, M. Ishii, K. Segawa, Y. Oka, B. J. Vartanian, and J. S. Harris, Appl. Phys. Lett. 68(1), 34 (1996). K. Matsumoto, Physica B 227, 92–94 (1996). J.-I. Shirakashi, M. Ishii, K. Matsumoto, N. Miura, and M. Konagai, Jpn. J. Appl. Phys. 35, L1524 (1996). K. Matsumoto, Proc. IEEE 85, 621 (1997). M. Yasutake, Y. Ejiri, and T. Hattori, Jpn. J. Appl. Phys. 32, L1021 (1993). T. Hattori, Y. Ejiri, K. Saito, and M. Yasutake, J. Vac. Sci. Technol. A 12(4), 2586 (1994). L. Tsau, D. Wang, and K. L. Wang, Appl. Phys. Lett. 64, 2133 (1994). D. Wang, L. Tsau, and K. L. Wang, Appl. Phys. Lett. 65, 1415 (1994). P. M. Campbell, E. S. Snow, and P. J. McMarr, Appl. Phys. Lett. 66, 1388 (1995). A. E. Gordon, R. T. Fayfield, D. D. Litfin, and T. K. Higman, J. Vac. Sci. Technol. 13, 2805 (1995). T. Teuschler, K. Mahr, S. Miyazaki, M. Hundhausen, and L. Ley, Appl. Phys. Lett. 67, 3144 (1995). T. Teuschler, K. Mahr, S. Miyazaki, M. Hundhausen, and L. Ley, Appl. Phys. Lett. 66, 2499 (1995). K. Araki, K. Morimoto, K. Morita, M. Niwa, and Y. Hirai, Jpn. J. Appl. Phys. 35, 6679 (1996). H. C. Day and D. R. Allee, Nanotechnology 7, 106 (1996). D. Stievenard, P. A. Fontaine, and E. Dubois, Appl. Phys. Lett. 70, 3272 (1997). P. Avouris, T. Hetel, and R. Martel, Appl. Phys. Lett. 71, 285 (1997). S. L. Brandow, J. M. Calvert, E. S. Snow, and P. M. Campbell, J. Vac. Sci. Technol. A 15, 1455 (1997). F. P. Zamborini and R. M. Crooks, J. Am. Chem. Soc. 120, 9700 (1998). W. Mizutani, T. Ishida, and H. Tokumoto, Langmuir 14, 7197 (1998).

464 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204.

205.

206.

L. A. Nagahara R. Ullmann, T. Will, and K. D. M., Ber. Bensenges Phys. Chem. 99, 1414 (1995). D. M. Kolb, R. Ullmann, and T. Will, Science 275, 1097 (1997). G. E. Engelmann, J. C. Ziegler, and D. M. Kolb, Surf. Sci. 401, L420 (1998). D. M. Kolb, R. Ullmann, and J. C. Ziegler, Electrochimica Acta 43, 2751 (1998). S. Iijima, Nature 354, 56 (1991). T. W. Ebbesen and P. M. Ayayan, Nature 358, 220 (1992). H. Dai, J. H. Hafner, A. G. Rinzler, D. T. Colbert, and R. E. Smalley, Nature 384, 147 (1996). S. S. Wong, J. D. Harper, P. T. Lansbury, Jr., and C. M. Lieber, J. Am. Chem. Soc. 120, 603 (1998). S. S. Wong, A. T. Woolley, E. Joselevich, C. L. Cheung, and C. M. Lieber, J. Am. Chem. Soc. 120, 8557 (1998). S. S. Wong, A. T. Woolley, T. W. Odom, J.-L. Huang, P. Kim, D. M. Vezenov, and C. M. Lieber, Appl. Phys. Lett. 73, 3465 (1998). C. M. Lieber, Solid State Commun. 107, 607 (1998). S. S. Wong, E. Joselevich, A. T. Woolley, C. L. Cheung, and C. M. Lieber, Nature 394, 52 (1998). H. Dai, N. Franklin, and J. Han, Appl. Phys. Lett. 73, 1508 (1998). Y. Akama, E. Nishimura, A. Sakai, and H. Murakami, J. Vac. Sci. Technol. A8, 429 (1990). T. Fujii, M. Suzuki, M. Miyashita, M. Yamaguchi, T. Onuki, H. Nakamura, T. Matsubara, H. Yamada, and K. Nakayama, J. Vac. Sci. Technol. B 9, 666 (1991). D. J. Keller and C. Chin-Chung, Surf. Sci. 268, 333 (1992). L. A. Nagahara, A. Manivannan, H. Yanagi, M. Toriida, M. Ashida, Y. Maruyama, K. Hashimoto, and A. Fujishima, J. Vac. Sci. Technol. A 11, 781 (1993). National Technology Roadmap for Semiconductors (Semiconductor Industry Association, San Jose, CA, 2001). K. Wilder and C. F. Quate, J. Vac. Sci. Technol. B 16, 3864 (1998). S. W. Park, H. T. Soh, C. F. Quate, and S.-I. Park, Appl. Phys. Lett. 67, 2415 (1995). B. W. Chui, T. D. Stowe, T. W. Kenny, H. J. Mamin, B. D. Terris, and D. Rugar, Appl. Phys. Lett. 69, 2767 (1996). S. C. Minne, G. Yaralioglu, S. R. Manalis, J. D. Adams, J. Zesch, A. Atalar, and C. F. Quate, Appl. Phys. Lett. 72, 2340 (1998). G. Binnig, M. Despont, U. Drechsler, W. Häberle, M. Lutwyche, P. Vettiger, H. J. Mamin, B. W. Chui, and T. W. Kenny, Appl. Phys. Lett. 74, 1329 (1999). S. C. Minne, J. D. Adams, G. Yaralioglu, S. R. Manalis, A. Atalar, and C. F. Quate, Appl. Phys. Lett. 73, 1742 (1998). H. P. Lang, R. Berger, C. Andreoli, J. Brugger, M. Despont, P. Vettiger, C. Gerber, J. K. Gimzewski, J. P. Ramseyer, E. Meyer, and H. J. Güntherodt, Appl. Phys. Lett. 72, 383 (1998). C. R. K. Marrian, E. A. Dobisz, and J. M. Calvert, in Atomic and Nanometer-Scale Modification of Materials: Fundamentals and Applications (Kluwer Academic Publishers, The Netherlands, 1993), p. 139. C. R. K. Marrian and E. S. Snow, Microelectron. Eng. 32, 173 (1996).

18 Microelectrode techniques for characterization of advanced materials for battery and sensor applications Matsuhiko Nishizawa and Isamu Uchida 18.1

Introduction

The microelectrode can be defined as an electrode having dimensions of micrometer or less (nm). The importance of the use of microelectrodes has already been widely accepted in the field of electroanalytical chemistry, since a microelectrode (or its array) has many advantages for rapid and sensitive voltammetric measurements.1–4 In addition to such electroanalytical use, microelectrode-based systems have become of interest in recent years for the characterization of solid materials. We review here some of the recent topics on the use of microelectrodes for materials characterization, especially for the materials in battery and sensor applications. Results are presented for conducting polymers and for active materials of lithium ion batteries, based upon our data. It will be demonstrated that the microelectrode-based characterization techniques bring unique information, which will contribute to the design of high performance micro devices. The electrodes we used can be classified into types: microdisk and microbandarray electrodes. The microdisk electrode (Figure 18.1(a)) was prepared by shielding the side-body of a metal or carbon fiber by glass or organic films. If we use a glass capillary, we can routinely make a microdisk electrode having a total radius (including the shielding material) of less than 1 m. We applied such

(a)

Microdisk electrode

(b) Interdigitated microarray electrode Sputter-deposited metal film Electroactive film

Metal or carbon filament Shield Insulating substrate

Figure 18.1 Schematic drawings of microelectrodes; (a) microdisk electrode, (b) interdigitated microarray electrode.

466

M. Nishizawa and I. Uchida

a microdisk electrode to the characterization of redox behavior in micro particle levels. The microband-array electrode in this work is the interdigitated-type array (Figure 18.1(b)). Photolithography-based microfabrication technique was successfully applied to the electrode preparation. We can easily prepare an array electrode having a dimension of less than 10 m. Microarray electrodes are useful tools for the in situ measurement of the material’s conductivity during the redox reactions, as will be explained later.

18.2

Single particle measurement

We and other groups have studied a single living cell using capillary microelectrodes,5–11 which is now one of the trends in bioelectrochemistry. We expanded the possibility of such a measurement system to the characterization of advanced solid materials for battery application.12–17 We have to take into account the pioneering work of Bursell and Bjornbom,18 in which they applied microelectrode technique to characterize single carbon agglomerate particles in an alkaline solution. Scholz and Meyer have recently reviewed the voltammetry of solid particles immobilized on conventional (macro) electrode surfaces.19 Since most of the battery active materials are usually obtained in powder form, their electrochemical behavior has been evaluated so far by using composite electrodes consisting of the powdery active material, organic binder and conductive additives. The study of such composite film electrodes, which is an actual form in the present batteries, should be practically important. However, the composite electrode should be treated as a “complicated” porous electrode; furthermore, the obtained result (especially the kinetic parameters) should be an average of many particles having different sizes and shapes. In order to eliminate uncertainties at the composite electrodes, we developed a microelectrode technique to investigate a single particle of battery active materials, in which a filament microelectrode is attached to a single particle as a current collector. This technique allows us to obtain the electrochemical behavior of the sample particle itself, without any dilution due to the additives. Furthermore, the resistive polarization (iR drop) is almost negligible due to the small amplitude of reaction current at a single particle, so that we can get high resolution voltammograms which should be useful for discussion of the mechanism of insertion/extraction reactions from both kinetic and thermodynamic standpoints. 18.2.1

Measurement system

Commercially available metal filaments, whose diameters are around 25 m, can be successfully used as the microdisk electrode. In order to ensure the electrochemical inertness of the microelectrode itself, the electrode material should be chosen by taking account of the potential range applied. Furthermore, a proper flexibility is also required for the electrode material since the insertion materials will change their volume by insertion reaction. We used a Pt-Rh filament for a cathode material of lithium ion batteries such as transition metal oxides, and

Microelectrode techniques 467 a Mo filament for an anode material such as carbon. They were coated with a thin film of Teflon to minimize the background current. Figure 18.2 shows SEM images of particles of battery active materials that we used. Mesocarbon microbeads (MCMB) is a commercial product (Osaka gas Co.) that is actually used in some lithium ion battery as an anode material.20 As seen in Figure 18.2(a), MCMB is ideally spherical. The cathode material can also be made as a sphere. The SEM image shown in Figure 18.2(b) is of LiMn2O4 particles (Nittuki Chemical Co.). The spherical shape of materials is desirable for the quantitative analysis of transient behavior of the material, as will be explained later. Besides these commercial products, we have prepared various oxides in a laboratory scale with a sol–gel process called citrate method. However, in this case, the resulting particle is not always the dense sphere, rather the porous aggregate of small crystals.

(a)

(b)

Figure 18.2 SEM images of particles of (a) mesocarbon microbeads (Osaka Gas Co.) and (b) LiMn2O4 (Niki Chemical Co.).

468

M. Nishizawa and I. Uchida Dry air Dry bag CCD camera Cell

Controller

Vibration isolator

Z CE

Microelectrode

WE

Particle X

Y

Glass frit Li foil

Figure 18.3 Illustration of the measurement system for the single particle measurements.

The electrochemical setup is shown in Figure 18.3. The microfilament electrodes were mounted on an X–Y–Z micropositioner and manipulated to make electrical contact with a particle spread on a disk of glass frit or a separator sheet soaked with electrolyte solution, such as 1 M LiClO4/PC EC. Manipulation of the electrode was conducted under microscope observation. The monitoring of the open circuit voltage (OCV) was valuable for detecting the electrical contact between the particle and the electrode. The cell assembly including the X–Y–Z micropositioner was placed in a dry box filed with dry air (70ºC dew point). 18.2.2

Single particle behavior of advanced battery materials

We have studied a variety of active materials for lithium ion batteries, including the cathode materials of LiCoO2,14 LiMn2O4,14 LiNiO215 and the anode materials of carbons.16,17 Figure 18.4(a) shows a cyclic voltammogram of a LiMn2O4 particle (20 m diameter) prepared by citrate process, representing two well-defined peaks (peak ␣ and ␤) in both anodic and cathodic potential scans with excellent reversibility. The peak currents for both peak ␣ and ␤ were linearly increased with scan rates up to 1 mV s1, indicating these scan rates were slow enough to

Microelectrode techniques 469 establish the equilibrium of the lithium content over the particle at any potential. Lithium extraction/insertion occurs at both peaks with the redox reaction of Mn3/4 in the solid-state phase. According to the precise analysis of XRD spectra by Ohzuku et al.21 at the peak ␣ a homogeneous reaction would take place, while two cubic phases coexist at the peak ␤. The unusual sharpness of the peak ␤ is thought to be owing to this two-phase reaction process. It should be emphasized that the high-resolution voltammogram, shown in Figure 18.4(a), can not be obtained by a usual composite electrode. For example, Figure 18.4(b) shows a typical voltammogram of a LiMn2O4 composite electrode containing 5 wt% polyvinylidene fluoride (PVDF) and 5 wt% acetylene black (AcB), taken at the scan rate of 0.05 mV s1. In spite of the slower potential scan rate, the composite electrode did not show remarkable CV peaks, probably, owing to the resistive polarization within the film. The optimization of the procedure of the film preparation should improve the electrochemical response of the film. The single particle voltammetry allow us to omit such efforts. The other advantage of the single particle measurement is the possible direct observation of the material’s morphology. Figure 18.5 shows CVs of LiNiO2 particle, taken in 1 M LiClO4/PC EC at the scan rate of 3 mV s1. This material shows a few couples of current peaks of lithium ion insertion/extraction at around 1

(a)

I (nA)

0.5 0

–0.5 –1 3.6

3.8

4 E/V vs Li/Li+

4.2

3.8

4 E/V vs Li/Li+

4.2

(b) 20

I (µA)

10 0 –10 –20 3.6

Figure 18.4 CVs in 1 M LiClO4/PC EC of (a) a particle of LiMn2O4 (at 0.2 mV s1) and (b) a composite film consisting of 90 wt% LiMn2O4, 5 wt% PVDF and 5 wt% acetylene black (at 0.05 mV s1).

470

M. Nishizawa and I. Uchida 60 40

I (nA)

20 0

–20 –40 –60

3

3.4

3.8 4.2 E/V vs Li/Li+

4.6

5

Figure 18.5 CVs for a LiNiO2 particle (30 m diameter) at 3 mV s1 in 1 M LiClO4/ PC EC.

4.0 V vs Li. This reaction is basically reversible, as shown by the dotted CV. However, we observed also the solid CV curve with c.50% frequency; the electrochemical response suddenly disappeared around 4.8 V. From the in situ microscope observation,15 we found that the LiNiO2 particle broke into small parts at this potential, causing the detachment from the filament microelectrode. It has been reported that 0.3 Å shrinking occurs within the hexagonal-layered structure of Li1xNiO2 during the lithium extraction (anodic potential scan).22 We believe that this crystallographic shrinking would cause stress in crystal, resulting in the mechanical breaking of the particle. It should be emphasized that such particle breaking can never be recognized directly by other than this microelectrode-based single particle measurement. If the particle is a conductive, dense sphere, we can apply a spherical diffusion model to analyze its kinetic responses.16 This is also an important advantage of the single particle measurement. Figure 18.6(a) shows a series of CVs of mesocarbon microbeads (MCMB, Osaka Gas Co.) graphitized at 2800ºC, taken at a scan rate of 10 V s1 with various negative potential limits. Four redox couples are clearly recognized, which is characteristic of lithium intercalation/ de-intercalation with formation of the graphite-intercalation compound (GIC) with lithium ions. No significant change in CV characteristics was observed during these successive experiments, assuring the stability of the microelectrode technique including the electrical contact between the microelectrode and the MCMB particle. Diffusion coefficient (D) is one of the most important kinetic parameters for ion-insertion compounds because the kinetics of a solid-state diffusion process often limits the total reaction rate, which corresponds directly to the available current in battery applications. For the spherical diffusion process in

Microelectrode techniques 471 (a)

4 2

I (nA)

0 –2 –4

Dapp (cm2 s–1)

(b)

0

0.05

0

0.05

0.1

0.15

0.2

0.25

0.1

0.15

0.2

0.25

10–8

10–9

10–10

10–11 E/V vs Li/Li+

Figure 18.6 CV (at 10 V s1) and the variation of Dapp taken for a MCMB particle (30 m diam.) heat-treated at 2800ºC. Measurements were conducted in 1 M LiClO4/PC EC. (From Ref. 16.)

which the reaction advances from the surface to the center of a sphere, the current response upon a potential step (chronoammperogram) will follow the equation:23 i (2nFAD C/a) exp[(␲2D/a2) t]

(t >> a2/␲2D)

where a and A are the radius and the surface area of sphere respectively, and C is the variation of lithium concentration in the sphere during the potential step. Thus we can calculate the Dapp value from the linear slope of log (i) vs t without information about C. Figure 18.6(b) shows the plot of log (Dapp) vs electrode potential at which the 10 mV potential steps was applied. It was found that the Dapp varies the order between 109 and 1011 cm2 s1 over the course of lithium intercalation at a MCMB. By comparing Figure 18.6(a) and (b), it was also found that the plot of Dapp shows minimum correlating to the intercalation peaks on CV. Similar results have been reported recently for a highly oriented pyrolitic graphite (HOPG)24 and for an ultrathin monoparticulate film of graphite powder.25,26 The Dapp values reported for a composite film was just a monotonic

472

M. Nishizawa and I. Uchida W2

(a) Ni-band electrode (W1)

MCMB particle SiO2 substrate ∆V Bipotentiostat (b) 100

2nd–5th cycle

I (µA)

50

iw2

0 iw1

–50

–100

0

0.5

1 1.5 Potential/V vs Li/Li+

2

Figure 18.7 (a) Illustrated configuration for in situ measurement of the conductance change with Mo microdisk electrode and a Ni-band electrode. Electrochemical components such as reference and counter electrodes are not shown. (b) Current responses at W2 (Mo microdisk) and at W1 (Ni-band) attached to the MCMB particle heat-treated at 1000ºC. The potential scan rate was 0.2 mV s1, and the potential difference between W1 and W2 was 7 mV. (From Ref. 17.)

decrease with lithium insertion,27,28 indicating that the single particle measurement is suitable for precise investigation of the kinetic parameters. Recently, we developed a new measurement system shown in Figure 18.7(a), in order to study the DC conductance of a single particle in situ.17 A single Ni-band was prepared by sputtering on an insulating substrate, and an MCMB particle was then pressed against the band electrode (W1) by a Mo microdisk electrode (W2). By means of a bipotentiostat, the potentials of the two working electrodes (W1 and W2) were scanned, maintaining 7 mV potential difference (V ) between them, and the conductance of the particle was evaluated from the derived ohmic current (i). Figure 18.7(b) shows typical current responses recorded at

Microelectrode techniques 473 1

for MCMB heat-treated at 1000ºC. The mesophase carbons 0.2 mV s heat-treated at temperatures below 2000ºC are known to contain many imperfections such as turbostatic disorders and unorganized parts, and thus the lithium insertion reaction proceeds without the formation of any staging structure characteristic of graphite intercalation compounds. Since the faradaic current at a single MCMB was in the nA range, the observed current (A range) should be dominantly the ohmic current derived by 7 mV bias voltage. Indeed, the current profiles at W1 and W2 are completely symmetrical to each other, thus confirming that the observed currents are almost purely the ohmic currents flowing through the MCMB particle, and therefore the current profile can be treated as the variation of conductance of the particle. The conductance profile showed a reversible change upon the lithium insertion, and was stable up to at least 10 cycles. The data shown here can be defined as the pure conductance behavior of MCMB itself, and this is a reference data to understand the conductance behavior of MCMB-based composite, as described later. Besides the lithium ion battery materials discussed above, we have been studying hydrogen-storage alloys, which are the anode of Nickel–Hydrogen battery.12,13 AB5 type alloys (LaNi5, LaNi4.5Al0.5 and LmNi4.0Co0.4Mn0.3Al0.3) and Laves phase alloys (ZrNi1.4, ZrMn0.6Co0.1V0.2Ni1.2 and ZrMn0.4Cr0.4Ni1.2) were studied, and the resulting CV behavior and Dapp of hydrogen atom were comparable with reported values. Recently, Lundqvist and Lindbergh studied the diffusion coefficient and phase-transfer rate parameter in a single particle of LaNi5 and MmNi3.6Co0.8Mn0.4Al0.3 using a carbon fiber microelectrode.29

18.3

Microarray electrode for in situ conductance measurement

Electrical conductivity is a fundamental but important parameter of the materials especially taking account of the application to electrochemical devices. Most of the electroactive materials change their conductivity during the redox reaction because the electrochemical electron transfer should vary the electronic structure of the material. Usually, the materials conductivity is measured upon taking the materials out of the electrochemical cell; that is, ex situ measurement. However, since electroactive materials are often sensitive to atmosphere, the in situ measurement is ideally required. The Wrighton group30,31 achieved the in situ conductivity measurement by using the interdigitated array electrode (IDA electrode), and we applied the technique to the characterization of a variety of materials. Battery active materials,17,32–35 conducting polymers36–39 and fullerene films40,41 were studied. In addition to the fundamental characterization, we expanded the IDA system to the conductometric sensor devices.42–45 18.3.1

Systems for simultaneous measurement of voltammogram and conductivity

The IDA electrode was fabricated by photolithography with a sputter-deposited Pt or Ni film on a thermally oxidized silicon wafer, as to have two sets of

474

M. Nishizawa and I. Uchida (a) Interdigitated microarray electrode Sputter-deposited metal film

Electroactive film

Insulating susbstrate (b) In situ micro-conductometry Ref

Aux Electroactive film

iredox i⍀

Band electrode

Fixed ∆V (~5 mV) Bipotentiostat

Figure 18.8 (a) Schematic drawing of the IDA electrode coated with an electroactive film. (b) Electrochemical apparatus for in situ conductivity measurements.

comb-type arrays; each band-electrode element of 10 m wide, 2.4 mm long and 0.1 m thick, separated by typically 10 m from its adjacent elements. Electroactive films were prepared onto the IDA electrode substrate in such a way as to bridge band arrays, as illustrated in Figure 18.8(a). As for the techniques for film preparation, we have employed sputtering, electrodeposition, doctor blade technique and so forth. The apparatus for electrochemical in situ conductivity measurement is illustrated in Figure 18.8(b). By means of a bipotentiostat, the potential of the couple of the band electrode (W1 and W2) were controlled simultaneously, while at the same time keeping a constant, small potential difference (e.g. 5 mV). The electrical conductivity of the film was calculated from the ohmic current flowing through the film. When the electrode potentials were swept, the resulting current response was composed of both a faradaic current and an ohmic current. If we assume that the magnitudes of the faradaic currents (iF) at both arrays are equal, the ohmic current (i) can be extracted from the total currents observed at W1 (iW1) and W2 (iW2) by i (iW1 iW2)/2

iF (iW1 iW2)/2

Microelectrode techniques 475 and thus the conductivity of the film was estimated from the following equation ␴ iw/nldV

S cm1

where w is the gap width of arrays, n the total number of the band electrode, l the length of the electrode, d the thickness of the film, and V the potential difference between the arrays. The ionic conductivity of an electrolyte solution is negligible in this DC method. Note, however, our measurement is a two-point probe method, therefore an absolute value of the obtained conductivity should not be discussed. Rather, this method is suited to the study of dynamic relative changes in the conductivity which is taking place with redox reactions. Besides the above fundamental characterization of in situ conductivity, the apparatus shown in Figure 18.8 possesses the ability to be used for device applications. If the conductivity of the film is dependent on an external chemical perturbation such as a pH-change, it can work as a conductometric chemical sensor. In the experiments for this type of application, the electrode potential was fixed, and the response of ohmic current (conductivity) was measured. 18.3.2

Characterization of active materials for lithium ion batteries

The electrical conductivity of materials is an important characteristic for the design of high performance batteries. We have been studying the dynamic changes in DC conductivity in situ using IDA electrodes for a variety of cathode materials of lithium ion battery, such as LiV2O5, LiMn2O4, LiCoO2.32–35 We will present here our recent work concerning LiCoO2,35 which is one of the most promising cathode active materials for lithium secondary batteries, and the battery system based on the LiCoO2/carbon couple has been used commercially in many types of electronic equipment. A LiCoO2 film of 0.2 m thick was deposited onto the IDA electrode by RF-sputtering and annealed in air at 700ºC for 15 min. Figure 18.9 shows CVs (solid lines) and in situ potential-conductivity profiles (dashed lines), which were measured simultaneously. The results for the first and the sixth cycles are presented. During the initial extraction of lithium, the conductivity of the film increased more than four orders of magnitude and saturated around the main CV peak at 3.9 V. The drastic conductivity change observed before the CV peak is known theoretically to be due to the metalinsulator transition (MIT) resulting from the possible direct overlap of cobalt t2g orbital across a shared octahedral edge.46 Such drastic conductivity-change would require careful attention in using the LiCoO2 as a battery active material. However, we found that the increased conductivity did not revert to the initial low value in the following reduction process, and that the conductivity retained the value within an order during the successive charging/discharging. We also conducted the conductivity measurements under open-circuit condition in order to understand above results as a function of x in Li1xCoO2. It was found that the conductivity increases drastically when a very small amount of lithium was extracted (x 0.1) on the first charging, and the conductivity of the film cannot revert to the initial insulating state because of the difficulty in the re-insertion of

476

M. Nishizawa and I. Uchida 2

1

0

10–5

I (µA)

Conductivity (S cm–1)

10–3

–1 10–7 3.3

3.5

3.7 E/V vs Li/Li+

3.9

4.1

Figure 18.9 CV (solid lines) and potential-conductivity profiles (dashed lines) of LiCoO2 film, taken in 1 M LiClO4/PC at a scan rate of 0.1 mV s1. The first and the sixth cycles are presented. (From Ref. 35.)

this small amount of lithium. Although the scientific explanation can not be established at present, electrochemical processes accompanied by growth of insulating phases are kinetically unfavorable in general. Anyway, the insulating nature of the LiCoO2, which may be serious for the battery application, was not found to be regenerated within the time scale of battery applications. We have already shown, in Figure 18.7, the in situ conductance behavior of MCMB (1000ºC) single particle. Taking into consideration the stable conductance behavior of MCMB particle itself, we next studied the MCMB composite film which is an actual form of MCMB in battery applications.17 The average diameter of the MCMBs used for making composite films was 6 m. Figure 18.10 shows the (a) in situ conductance-profiles and (b) CVs of a MCMB/PVDF (100 : 5.5 in weight) composite film, recorded at a scan rate of 0.2 mV s1. Note that a conductive additive such as acetylene black is absent in this composite. The conductance-profile is “basically” the same as that of the MCMB particle; the conductance increased with lithium insertion. However, the magnitude of filmconductance was decreased drastically by discharge/charge cycles, and reached to only 4% of the initial conductance during the 7 cycles. As seen in Figure 18.10(b), the CV response of the film weakened in such a way as to synchronize with the decrease in conductance. The observed decrease in conductance and CV activity of the composite film may be explained by supposing an electrical isolation of MCMB particles. During the lithium insertion/extraction reaction, the MCMB particles would repeat expansion and contraction in their volume to some extent, causing a change of microstructure in the composite. If a particle is once isolated from the MCMB network within the composite, this particle can never take part in redox events. Similarly, the total conductance of the composite film would be decreased due to such internal defect formation in the conduction path. For

Microelectrode techniques 477

Conductance (mS)

(a)

(b)

12 10 8 6 4 2 0

0

0.5 1 1.5 Potential/V vs Li/Li+

2

0.5 1 1.5 Potential/V vs Li/Li+

2

I (µA)

20 0 –20

– 40 0

1st 3rd

5th 7th

Figure 18.10 (a) Potential-conductance profiles and (b) cyclic voltammograms of a composite film (MCMB:PVDF 100:5.5 in weight) measured simultaneously using the IDA electrode. The film thickness was c.20 m. The scan rate of electrode potential was 0.2 mV s1. The potential difference between arrays (W1 and W2) was 7 mV. (From Ref. 17.)

a practical use of disordered carbons, a conductive agent such as AcB is often added to the composite, in order to improve the relatively low conductivity of the carbon material itself and prevent the conductivity decrease shown here. We confirmed that the addition of 27.5% AcB improved the conductance profile greatly, so that more than 60% of the initial conductance was maintained even after the 7 cycles. The volume change of MCMB particle would also occur in this case. However, since the AcB would serve as a flexible conducting wire, only few particles would be isolated electrically. It is worth noting that these empirically accepted tendencies were represented here by our microelectrode technique from a more scientific standpoint. The present approach is naturally applicable to other components of lithium ion battery. Since most of the cathode oxide material such as LiMn2O4 is semiconductor, the effective use of conductive additives should be critically important in making its composite film. 18.3.3

Conducting polymers for sensor application

Electropolymerized conducting polymer films have been extensively investigated as materials for solid-state and electrochemical devices. Since the electropolymerization can be applied to micro-area plating, this technique is

478

M. Nishizawa and I. Uchida

useful to build up the micro devices. It has been found that the conductivity of most of the conducting polymers such as polyaniline (PAn) and polypyrrole (PPy) changes its value as functions of both the electrode potential (redox state) and pH. We have been investigating such properties of conducting polymers in situ using the IDA electrode. It was found that the conductivity response time was dramatically dependent on the film thickness, indicating that the formation of a thinner film on the IDA substrate was the key to fabricate a smart sensor. We found that a hydrophobic pretreatment of the gap part of IDA was effective to bridge the array with an electropolymerized ultrathin polymer film.37–39 Hydrophobic pretreatment of the IDA substrate was carried out by silanization typically in a 20 mM octadecyltriethoxysilane/benzene solution. Figure 18.11 shows SEM images of electropolymerized PPy at the untreated (A) and treated (B) IDA. A sudden current increase due to the ohmic current between the adjacent array electrodes was used as an indication of the interconnection. The electropolymerizations were terminated immediately after the interconnection of arrays with the PPy film. The images indicate that the hydrophobic surface obviously promotes the lateral polymer growth along the substrate surface. This promotion effect would be accounted by adsorption of pyrrole monomers and selective deposition of intermediate oligomers on the hydrophobic surfaces. The polymerization anisotropy (the ratio of lateral growth rate to vertical growth rate) was found to be c.25.37,38 Furthermore, the presence of an anionic surfactant such as dodecylsulfate greatly enhances the promotion effect, showing the polymerization anisotropy of 170.39 Dodecylsulfate molecules would assemble on the hydrophobic surface and work as an effective dopant of the growing PPy film, resulting in the amplification of the promotion effect of the hydrophobic surface. These surface treatments were also effective to prepare an ultrathin polyaniline film.38 As shown in Figure 18.12(a), we found that the conductivity of the PPy film varies with the pH of the solution.43,44 In the pH range, 3–11, pH-conductivity

(a)

(b)

10 µm

Electrode

Polypyrrole

Electrode Contact point

Polypyrrole

Contact line

Figure 18.11 SEM images of PPy films at the microarray electrodes. (a) Untreated electrode (total polymerization charge, 11.91 mC). (b) Hydrophobically pretreated electrode (total charge, 0.98 mC). (From Ref. 37.)

Microelectrode techniques 479 plots show sigmoidal shapes with an apparent inflection point which shifts to the low pH side with increase in the oxidation level of PPy, indicating that the de-protonation reaction easily occurs in the highly oxidized PPy. We have applied this pH-sensitive device (ultrathin PPy-coated IDA) to a penicillin sensor by coating the device with a penicillinase membrane.43,44 The penicillinase membrane was prepared by using a cross-linking agent, glutaraldehyde, and the film thickness of the resulting membrane was c.5 m in dry state. Penicillinase catalyzes hydrolysis of penicillin to penicilloic acid and acidifies the PPy ultrathin film. Therefore, the ohmic current (I) increases in the presence of penicillin. The sensing characteristic of the penicillin sensor was studied at 0.2V vs SCE using a flow-cell with a carrier of 0.1 M KNO3 5 mM phosphate buffer (pH 6.9). Figure 18.12(b) shows the repeated I responses upon flowing of various concentrations of penicillin solutions for 1 min at the flow rate of 2.0 ml min1. The

1.25 Normalized conductivity

(a)

VG = –200 mV vs SCE VG = 0 mV vs SCE VG = 200 mV vs SCE

1 0.75 0.5 0.25 0

2

2.5

5

6

7 pH

8

9

10

11

12

2

2

7

1

1.5 ∆ I⍀ (µA)

4

∆ I⍀ (µA)

(b)

3

5

0 0 1 2 3 4 5 6 7 8 9 Penicillin (mM) 3

1

8 mM

6 4

2

0.5 0.3

0

0.1

0.7

1

0.5

–0.5 0

20

40

60 Time (min)

80

100

120

Figure 18.12 (a) Variation of the normalized electrical conductivity of PPy as functions of pH and the electrode potential. (b) I responses of the penicillin sensor upon injection of various concentrations of penicillin solutions. (From Ref. 43.)

480

M. Nishizawa and I. Uchida

insert is the plot of variation of the peak height with the concentration of the penicillin solution. The peak height increases with the concentration of the penicillin solution up to 7 mM, but tends to saturate at higher concentrations since the overall electrochemical process is now controlled by the kinetics of the enzyme reaction. Since many enzymes bring about pH changes through their catalytic reactions, the principle shown here will be widely applicable to the fabrication of enzyme-based microelectrochemical devices which detect biologically important substrates. The device shown above is based on the detection of a pH change induced by enzyme reactions. On the other hand, the oxidation/reduction of conducting polymers via enzyme reaction is another possible principle of the conductometric biosensor. We reported previously that the microarray electrode coated with the copolymer of pyrrole and N-methylpyrrole containing diaphorase worked as a switching device, which showed “on ⇒ off ” response to the presence of NADH.42 Bartlett et al. have expanded this idea to the fabrication of sophisticated bio-electrochemical devices responsive to glucose47–49 and NADH.50 These devices are based on the conductivity-change of polymer film induced by the redox reaction of polymer itself.

18.4

Summary

Some of the recent topics on the use of microelectrodes for the materials characterization have been overviewed based mainly upon our results. As described, microelectrode-based techniques bring lots of novel and unique information on advanced materials for battery and sensor applications. It is certain that such electrochemical microsystems will gain greater importance in future. In addition, the IDA system can be directly applied to conductometric micro devices.

Acknowledgements The authors would like to thank their colleagues, Prof K. Itaya, Prof T. Matsue, Prof T. Nishina, Prof T. Abe, Dr T. Itoh, and more than ten graduates for their help in abetting researches described in this review.

References 1. M. Fleischmann, S. Pons, D. Rolison and P. Schmdt, Ultramicroelectrodes, Datatech Systems, Morganrown, NC (1987). 2. R. M. Wightman, Anal. Chem. 53, 1125 (1981). 3. K. Aoki, Electroanal. 5, 627 (1993). 4. T. Matsue, Trends Anal. Chem. 12, 100 (1993). 5. I. Uchida, T. Abe, T. Itabashi and T. Matsue, Chem. Lett. 1227 (1990). 6. T. Matsue, S. Koike, T. Abe, T. Itabashi and I. Uchida, Biochim. Biophys. Acta 1101, 69 (1992). 7. T. Matsue, S. Koike and I. Uchida, Biochem. Biophys. Res. Commun. 197, 1283 (1993).

Microelectrode techniques 481 8. T. Abe, Y. Y. Lau and A. G. Ewing, J. Am. Chem. Soc. 113, 7421 (1991). 9. Y. Y. Lau, T. Abe and A. G. Ewing, Anal. Chem. 64, 1702 (1992). 10. T. J. Schroeder, J. A. Jankowski, K. T. Kawagoe, R. M. Wightman, C. Lefrou and C. Amatore, Anal. Chem. 64, 3077 (1992). 11. T. J. Schroeder, J. A. Jankowski, J. Senyshyn, R. W. Holz and R. M. Wightman, J. Biol. Chem. 269, 17215 (1994). 12. H. Ura, T. Nishina and I. Uchida, J. Electroanal. Chem. 369, 169 (1995). 13. T. Nishina, H. Ura and I. Uchida, J. Electrochem. Soc. 144, 1273 (1997). 14. I. Uchida, H. Fujiyoshi and S. Waki, J. Power Sources 68, 139 (1997). 15. S. Waki, K. Dokko, T. Matsue and I. Uchida, Denki Kagaku 65, 954 (1997). 16. M. Nishizawa, R. Hashitani, T. Itoh, T. Matsue and I. Uchida, Electrochem. Solid-State Lett. 1, 10 (1998). 17. M. Nishizawa, H. Koshika and I. Uchida, J. Phys. Chem. 103, 192 (1999). 18. M. Bursell and P. Bjornbom, J. Electrochem. Soc. 137, 363 (1990). 19. F. Scholz and B. Meyer, in Electroanalytical Chemistry (A. J. Bard and I. Rubinstein, eds) Vol. 20, Marcel Dekker, New York (1998). 20. A. Mabuchi, K. Tokumitsu, H. Fujimoto and T. Kasuh, J. Electrochem. Soc. 142, 1041 (1995). 21. T. Ohzuku, M. Kitagawa and T. Hirai, J. Electrochem. Soc. 137, 769 (1990). 22. T. Ohzuku and A. Ueda, J. Electrochem. Soc. 141, 2972 (1994). 23. W. Jost, Diffusion in Solids, Liquids, Gases, Academic Press, New York (1960). 24. Z. Ogumi and M. Inaba, Bull. Chem. Soc. Jpn. 71, 521 (1998). 25. M. D. Levi and D. A. Aurbach, J. Phys. Chem. 101, 4641 (1997). 26. M. D. Levi, E. A. Levi and D. A. Aurbach, J. Electroanal. Chem. 421, 89 (1997). 27. M. Morita, N. Nishimura and Y. Matsuda, Electrochim. Acta 38, 1721 (1993). 28. M. Jean, C. Desnoyer, A. Tranchant and R. Messina, J. Electrochem. Soc. 142, 2122 (1995). 29. A. Lundqvist and G. Lindbergh, J. Electrochem. Soc. 145, 3740 (1998). 30. H. S. White, G. P. Kittlesen and M. S. Wrighton, J. Am. Chem. Soc. 106, 5375 (1984). 31. M. J. Natan and M. S. Wrighton, in Progress in Inorganic Chemistry, Vol. 37, Wiley, New York (1989). 32. M. Shibuya, S. Yamamura, T. Matsue and I. Uchida, Chem. Lett. 749 (1995). 33. M. Shibuya, T. Nishina, T. Matsue and I. Uchida, J. Electrochem. Soc. 143, 3157 (1996). 34. S. Yamamura, H. Koshika, M. Nishizawa, T. Matsue and I. Uchida, J. Solid State Electrochem. 2, 211 (1998). 35. M. Nishizawa, S. Yamamura, T. Itoh and I. Uchida, Chem. Commun. 1631 (1998). 36. M. Nishizawa, T. Sawaguchi, T. Matsue and I. Uchida, Synth. Met. 45, 241 (1991). 37. M. Nishizawa, M. Shibuya, T. Sawaguchi, T. Matsue and I. Uchida, J. Phys. Chem. 95, 9042 (1991). 38. M. Nishizawa, Y. Miwa, T. Matsue and I. Uchida, J. Electrochem. Soc. 140, 1650 (1993). 39. M. Nishizawa, Y. Miwa, T. Matsue and I. Uchida, J. Electranal. Chem. 371, 273 (1994). 40. M. Nishizawa, T. Matsue and I. Uchida, J. Electroanal. Chem. 353, 329 (1993). 41. M. Nishizawa, K. Tomura, T. Matsue and I. Uchida, J. Electroanal. Chem. 379, 233 (1994). 42. T. Matsue, M. Nishizawa, T. Sawaguchi and I. Uchida, J. Chem. Soc., Chem. Commun. 1029 (1991).

482

M. Nishizawa and I. Uchida

43. M. Nishizawa, T. Matsue and I. Uchida, Anal. Chem. 64, 2642 (1992). 44. M. Nishizawa, T. Matsue and I. Uchida, Sens. Actuators B 13–14, 53 (1993). 45. Y. Miwa, M. Nishizawa, T. Matsue and I. Uchida, Bull. Chem. Soc. Jpn. 67, 2864 (1994). 46. P. G. Bruce, Chem. Commun. 1818 (1997). 47. P. N. Bartlett and P. R. Birkin, Anal Chem. 65, 1118 (1994). 48. P. N. Bartlett and P. R. Birkin, Anal. Chem. 66, 1552 (194). 49. P. N. Bartlett, J. H. Wang and W. James, Analyst, 123, 387 (1998). 50. P. N. Bartlett, J. H. Wang and E. N. K. Wallance, Chem. Commun. 359 (1996).

Part V

Biological systems

19 Microsystems for biosensing nucleic acids and immuno proteins Tadashi Matsunaga and Tae-Kyu Lim

19.1

Introduction

In recent years microsystem technologies have become increasingly important, owing to the extraordinary rapid developments in sensors, actuators, and micromechanical tools.1 Powerful microsystem technologies require different types of miniaturized sensors to produce integrated analytical, monitoring, and process control systems. Rapid development in microsystem technology has produced miniaturized sensors, and integration requires development of similarly miniaturized systems capable of utilizing these. Miniaturization of analytical biochemical systems can improve the reliability and speed of analyses, reduce the necessary sample size, and decrease consumption of expensive reagents. Microsensors exhibit wide application potentials, ranging from medical diagnosis,2–5 biotechnology,6,7 and biological research,8 to environmental monitoring.9–10 The field of nucleic acid biosensing, including DNA microarrays or DNA chips, is undergoing rapid progress. DNA microarrays enable the rapid, costeffective, and simultaneous screening of genome DNA for numerous sequence variations. The term DNA microarray generally refers to a grid-like array of nucleic acid species on flat solid supporting structure, and is typically a glass or silicon chip device. Various kinds of DNA microarrays have been developed for assessing variations in DNA or RNA.11–13 Analysis of samples using DNA microarrays is fast becoming a standard approach in molecular biology research14–15 and clinical diagnosis.16–18 Microarrays have been used for investigation of the microbial genome in gene expression analysis19,20 and quantification of target microbial populations in environmental analysis.21 Electrochemical microfabrication is relatively new, and is rapidly becoming the technology of choice in the electronics and microsystems industries.22,23 The successful implementation of electrochemical processes in the manufacturing industry is possible as a result of progress in fundamental research in electrochemistry, electrochemical materials science, and electrochemical engineering. Hence, we describe several technologies for microsystems. These microsystems have been applied to biosensing of nucleic acids and immuno proteins.

486

T. Matsunaga and T.-K. Lim

19.2

Microsystem for biosensing nucleic acid

The development of DNA analysis systems and methods with enhanced speed, sensitivity, and throughput will be important for the completion of the Human Genome Project and for the subsequent utilization of the sequence information generated. Magnetic capture hybridization of nucleic acids has attracted considerable attention owing to the ease with which specific target DNA can be isolated from heterogenous samples. This simple method, furthermore, permits rapid concentration of target DNA from a bulk sample. It would therefore allow sensitive detection of target DNA from environmental samples, and blood and other clinical samples. For example, using magnetic capture hybridization, the detection of verotoxigenic Escherichia coli from artificially contaminated ground beef 24 and probing of prokaryotic mRNA for the neutral protease of Bacillus cereus from liquid culture25 have been reported. Bacterial magnetic particles (BMPs) isolated from the magnetic bacterium, Magnetospirillum magneticum AMB-1 are small in size (50–100 nm), and disperse very well, because they are covered with a stable lipid membrane.26 BMPs have been used in an enzyme immunoassay for the highly sensitive and rapid detection of immunoglobulin G(IgG).27,28 In this section, DNA microarrays and the magnetic capture hybridization technique are combined to provide a new technology, the MAG-microarray, in which BMPs are used for the detection of target DNA by magnetic concentration and high throughput analysis. 19.2.1

Identification of cyanobacterial DNA

Cyanobacteria are the dominant oxygenic phototrophs in the marine environment, and have adapted to survive under a border range of conditions. Studies in environmental microbiology are often limited by the inability to identify and directly quantify the enormous diversity of natural populations, because isolation and cultivation of microbes from natural environment are sometimes impossible. Molecular approaches overcome this problem. 16S rDNA is currently considered to be the most promising approach for the phylogenic classification of cyanobacteria.29–33 By comparative analysis of cyanobacterial 16S rDNA, oligonucleotide probes were developed to distinguish between different cyanobacterial 16S rRNA sequences.34 A method for the detection of cyanobacterial DNA was demonstrated, using designed oligonucleotide probes conjugated BMPs, and the MAG-microarray. 19.2.1.1

Microarray fabrication and magnetic separation robot

Initially, a photomask was prepared to create micro-wells on the surfaces of silicon wafers. The design of this microarray MAG-microarray is shown in Figure 19.1(a). The microarray has ninety-six micro-wells, and has the dimension of 100 100 m. Chromium and gold were sputtered at each etching step; the silicon wafer was washed with deionized water, and dried. Photo-resist was removed

Microsystems for biosensing 487 (a)

100 µm 500 µm

100 µm

10 µm

870 µm

100 µm

10 µm

10 µm

1310 µm

100 µm

100 µm 500 µm

100 µm

10 µm

Microarray Magnetic concentration (b)

100 µm 6.3 mm

MAG-microarray Micro-titer plate Target DNA captured by bacterial magnetic particles

Magnet

Figure 19.1 Detection of target DNA using bacterial magnetic particles on MAG microarray. (a) Design of a MAG-microarray; (b) concept of detection using MAG-microarray.

from the surface of the silicon wafer, and silicon was subsequently etched with 30% KOH at 70ºC for 5 min. A silicon dioxide layer was subsequently formed by immersion of the silicon wafer sequentially in the following: (i) concentrated nitric acid solution at 115ºC for 10 min, (ii) a solution containing 7 ml sulfuric acid and

488

T. Matsunaga and T.-K. Lim

Figure 19.2 Photograph of a magnetic separation robot.

4 ml hydrogen peroxide. As shown in Figure 19.1(b), target DNA captured by BMPs is concentrated on the MAG-microarray. The magnetic separation robot (Figure 19.2) composed of a stage on which to place the disposable tubes, tips and reagents, and a liquid dispensing unit with a magnet. The tip consisted of a reservoir to aspirate the reaction mixture, a thin tip-end, and an intermediate section. In the intermediate section, magnetic separation is performed during the sucking or dispensing of a suspension of magnetic particles by placing a magnet at the side of the tip. Resuspension of the trapped magnetic particles is easily performed by sucking and dispensing the resuspension buffer with the magnet removed from tip. The liquid dispenser is the same as other commercial dispensers, except that it includes a magnetic separator. 19.2.1.2

Determination of cyanobacterial genus using genus-specific oligonucleotide probes on an MAG-microarray

Genus-specific oligonucleotide probes were designed on the basis of sequence polymorphism at the 16S rDNA gene locus as characterized by multiple sequence alignment analysis using the 148 cyanobacterial strains listed in Table 19.1. Sequence alignment of 16S rDNA was performed using GenBank (Release 107). Figure 19.3 shows the flow chart for detection using the MAG-microarray. Using the magnetic separation robot (model SX-6g, Precision System Science, Tokyo,

Microsystems for biosensing 489 Table 19.1 Cyanobacterial genera and number of strains used in this study Genus

Number of strains

NKBG culture collection Anabaena Nostoc Oscillatoria Synechococcus DNA Data Bank of Japan Anabaena Calothrix Leptolyngbya Microcystis Nostoc Oscillatoria Spirulina Synechococcus Phormidium Prochlorococcus Prochlorothrix Trichodesmium Total 12 genera

11 8 1 9 8 1 4 30 15 6 5 22 5 13 3 7 148

Magnetic concentration Bacterial magnetic particle conjugated genus-specific oligonucleotide probe

Spotting

DIG-labeled 16S rDNA Micro-well Magnet

MAG-microarray Magnetic-capture hybridization

AP

Substrate (Atto PhosTM)

AP AP conjugated anti-DIG antibody Fluorescence detection

Self-operating magnetic separation system

Figure 19.3 Procedure for detection of cyanobacterial DNA on MAG-microarray using genus-specific oligonucleotide probes conjugated to bacterial magnetic particles.

Japan), the DIG-labeled PCR product of 16S rDNA (20 ng) was denatured at 95ºC for 10 min, and then hybridized with oligonucleotide probes immobilized on BMPs (100 g) at 60ºC for 5 min in 100 l of hybridization buffer (10 mM Tris HCl, 2.5 mM MgCl2, 100 mM KCl; pH 8.3).

490

T. Matsunaga and T.-K. Lim

The 16S rDNA gene locus of 29 strains was partially sequenced and aligned. Multiple sequence alignment analysis of cyanobacterial 16S rDNA gene locus (148 database strains) was inspected to identify any region likely to include genus-specific sequence polymorphism. Genus-specific probes for Anabaena, Microcystis, Nostoc, Oscillatoria, and Synechococcus spp. were mostly observed in the V7 and V8 variable regions of 16S rDNA. Oligonucleotide probes for Anabaena, Microcystis, Nostoc and Synechococcus spp. (17–23 mer) were designed around the sequence polymorphism of the V7 region of 16S rDNA (Figure 19.4). For Anabaena and Nostoc spp., two detection probes for each genus were designed, since these genera are difficult to discriminate. Nostoc spp. had the same sequence alignment as oligonucleotide probes for Anabaena spp. Therefore, it might be possible to distinguish between these two cultures on the basis of the difference in fluorescence of probes for Nostoc spp. The sequence polymorphism in the V7 region was insufficient for design of a genus-specific probe for Oscillatoria spp. The probe for Oscillatoria spp. was designed on the basis of a region downstream of the V8 region at 801–823 nt (in E. coli 16S rDNA nt). The regions used for designing the probes contained two or three nucleotide differences between the genera. No sequence alignment matches were found for any of the probes, when checked against other microorganisms. Detection of 16S rDNA from five axenic cyanobacterial cells of Anabaena spp. PCC7120, Microcystis spp. NIES-98, Nostoc spp. NKBG038601, Oscillatoria spp. NKBG091600 and Synechococcus spp. PCC7942 was per-formed using MAGmicroarray (Figure 19.5). Fluorescence was only observed when using complementary combinations of amplified 16S rDNA and genus-specific probes. No fluorescence was observed in analysis using E. coli and P. cepacia as the sample. The probe Anabaena2 has the same alignment as the 648–666 region of 16S rDNA in Nostoc spp. NKBG038601. The probe Anabaena1 was designed including W (A or T) and V (A, C or G), because this region has sequence variation in various Anabaena species. One of the nucleotide sequences of this region was similar to that of 16S rDNA in Nostoc spp. NKBG038601. Therefore, the probes Anabaena1 and Anabaena2 detected the 16S rDNA of Nostoc spp. On the other hand, the probes Nostoc1 and Nostoc2 have nucleotide differences with the same regions of 16S rDNA in Anabaena spp. PCC7120, respectively. As a result, the PCR products of Anabaena PCC7120 16S rDNA were detected by neither Nostoc1 nor Nostoc2. Therefore, Nostoc spp. NKBG091600 and Anabaena spp. PCC7120 were discriminated using this method. It was shown that genus-specific detection of five selected genera of cyanobacterial DNA was successfully performed using the MAG-microarray with the designed oligonucleotide probes conjugated with BMPs, using the magnetic separation robot. Ability of the MAG-microarray system to determine the genus of unknown cyanobacterial samples correctly was investigated as follows. A unicellular culture was observed by microscopy and identified to be a Synechococcus spp. 16S rDNA from this strain was analyzed using the MAG-microarray system. Fluorescence was only observed when using the genus-specific oligonucleotide probe for Synechococcus spp. (Figure 19.6). The data obtained using the

200

300

400 V6

810

570

820

580

830

V7

600 V8

800

900

1000

1100

600

610

630

640

1200

1400

1500

1600

650

660

670

RSR-2 5⬘-TAAGGAGGTGATCCAACCGC-3⬘

1300

Figure 19.4 Design of genus-specific probes. (a) Locations of PCR primers. RSF-1 and RSR-2 are universal primer pairs for amplification of whole region of 16S rDNA. CYA106F and CYA781R(a)(b) are oxyphotobacterium-specific primer pairs; (b) sequence alignment of the partial sequence from cyanobacterial 16S rDNA of Anabaena spp., Microcystis spp., Nostoc spp., Oscillatoria spp., and Synechococcus spp. Major sequences for each genus were determined by comparative sequence analysis of cyanobacterial 16S rDNA listed in Table 19.1. Location of genus-specific detection probes is indicated with underline; (c) detection probes. Polymorphic sites are indicated with the corresponding degenerated symbols: RA or G; YC or T; WA or T; KG or T; MA or C; SG or C; HA, C, or T; VA, C, or G; DA, G, or T; and NA, C, G, or T.

620

CYA781R(a)+(b) (a)5⬘-GACTACTGGGGTATCTAATCCCATT-3⬘ (b)5⬘-GACTACAGGGGTATCTAATCCCTTT-3⬘

700

Anabaena (1) AGGTGGCATTGTWWGTVT (2) GGGCAGAGGGAATTCCTGG Microcystis TCAAATCAGGTTGCTAA Nostoc (1) CAATGTAAGTCTGCTGT (2) CATAGCTAGAGTACGT Oscillatoria GATGGAGACTAGGTGTDGCCTGT Synechococcus AGTCTGYTGTCAAAGCCY

(c) Detection probes

Anabaena Microcystis Nostoc Oscillatoria Synechococcus

800

Microcystis Nostoc Oscillatoria Synechococcus

560 590

500

CYA106F 5⬘-CGGACGGGTGAGTAACGCGTGA-3⬘

100

RSF-1 5⬘-AGAGTTTGATCATGGCT-3⬘

0

(b) Anabaena

(a)

Probe pattern

Anabaena1

Anabaena2

Microcystis

Nostoc1

Nostoc2

Oscillatoria

Synechococcus

(a)

(b)

(c)

(d)

(e)

(f)

(g)

Figure 19.5 Detection of cyanobacterial DNA on MAG-microarray using BMPs conjugated genus-specific oligonucleotide probes. Oligonucleotide probes used for detection are given as probe pattern. Cyanobacterial strains used for analysis are follows. (a) Anabaena spp. PCC7120; (b) Microcystis spp. NIES 98; (c) Nostoc spp. NKBG038601; (d) Oscillatoria spp. NKBG091600; (e) Synechococcus spp. PCC7942; (f ) E. coli K-12; (g) Pseudomonas cepacia JCM5506.

Anabaena1

Anabaena2

Microcystis

Nostoc1

Nostoc2

Oscillatoria

Synechococcus

Figure 19.6 Detection of unknown cyanobacterial samples using MAG-microarray. The sample was already isolated and maintained for the NKBG culture collection.

Microsystems for biosensing 493 MAG-microarray system were consistent with microscopic identification of the strain as a Synechococcus. This method using MAG-microarray has several advantages over conventional hybridization methods, in that the use of BMPs enables automated magnetic capture hybridization using magnetic separation robot and magnetic separation and concentration of target DNA is easily achieved. Furthermore, the MAG-microarray does not need a complicated step for the immobilization of oligonucleotides on the microarray, and small quantities of BMP conjugated with oligonucleotide probes are sufficient for detection. In this study, ninety-six micro-wells were fabricated in the MAG-microarray and the volume of each well was 200 pl. As volume of each well on the microtiter plate was 100 l, the volume in the microarray was reduced 5 105 times. This means that the suspension of BMPs binding target DNA will theoretically be magnetically concentrated 5 105 times. Therefore, high sensitivity will be expected in future work.

19.2.2

Discrimination between Atlantic and Pacific subspecies of northern bluefin tuna (Thunnus thynnus)

Exact representation of the species and geographic origin of marine products is required, especially with respect to overseas trade. The northern bluefin tuna (Thunnus thynnus) is the fish, which is most highly valued for sashimi in the Japanese market, which may have resulted in an increase in illegal fishing and trading, without concern for the decrease in stocks. At present, the fishing season, the catch and the trading of the Atlantic subspecies (T. thynnus thynnus) are rigorously regulated. However, discrimination of this species from the Pacific subspecies (T. thynnus orientalis) is virtually impossible when several diagnostic external and internal morphological characteristics are removed, and/or when the fish are filleted. Therefore, genetic species identification appears to be useful. Although protein electrophoresis was performed for tuna species identification,35,36 this failed to discriminate between the Atlantic and Pacific northern bluefin tunas. Chow and coworkers36,37 performed PCR-based nucleotide sequence and RFLP analyses on a DNA fragment (ATCO) flanking the mitochondrial ATPase and cytochrome oxidase subunit III genes of all tuna species, for which they reported distinct differences in the restriction profiles and nucleotide sequences between Atlantic and Pacific northern bluefin tunas. There are however drawbacks, as the nucleotide sequences reported were incomplete, and the analytical procedure was time-consuming, which would limit its use in the field. We have analyzed entire nucleotide sequences of the ATCO fragments of these two subspecies to find DNA sequences which enable the design of specific DNA probes.

494

T. Matsunaga and T.-K. Lim

19.2.2.1

DNA sequence analysis of mitochondrial ATPase and cytochrome oxidase III genes (ATCO) region

Samples of Atlantic northern bluefin tuna (T. thynnus thynnus) and the Pacific northern bluefin tuna (T. thynnus orientalis) were obtained from the laboratory collection at the National Research Institute of Far Seas Fisheries, Shimizu, Japan. Standard phenol–chloroform extraction was used to extract crude DNA from frozen or ethanol-preserved muscle tissue, and DNA samples were dissolved in TE buffer prior to PCR amplification. The PCR primer set to amplify a flanking region between mitochondrial ATPase and cytochrome oxidase III genes (ATCO region) are from Chow and coworkers,36,37 and their amplification conditions were also used. The primer nucleotide sequences were; 5 -CTTCGACCAATTTATGAGCCC-3 (L8562) and 5 -GCCATATCGTAGCCCTTTTTG-3 (H9432). For nucleotide sequence analysis, amplified PCR fragments were electrophoresed on 1% agarose gel, excised, and cloned into pUC18, which was subsequently used to transform E. coli DH5 ␣MCR. Clones containing target fragment were screened and sequenced using an automatic DNA sequencing machine (DSQ-1000L, Shimadzu). The amplified fragments were 927 bp in length and no length differences were observed between individuals. The aligned nucleotide sequences of 885 bp, excluding priming sites of eight individuals of the Pacific (NBTP) and four of the

Figure 19.7 Nucleotide sequence alignment of 885 bp fragments (ATCO) flanking between mitochondrial DNA ATPase and cytochrome oxidase subunit III genes of eight individuals of Pacific northern bluefin tuna (NBTP) and four of Atlantic subspecies (NBTA). R, W, and Y indicate A or G, A or T, and C or T, respectively. Position and priming directions of two internal primers (NR1R and NR2F) are indicated by underlines with arrowhead, and positions of NR1 and NR2 probes are boxed. Nucleotide sequences of an individual in each subspecies are available in GenBank under accession numbers xxxxxxx and xxxxxxx.

Microsystems for biosensing 495 Atlantic (NBTA) subspecies, are shown in Figure 19.7. Two regions (NR1 and NR2, boxed in Figure 19.7) were adopted for designing specific DNA probes. NR1 and NR2 were 21- and 29-mer, respectively, and contained one and three nucleotide differences between the subspecies. The positions and priming directions of two internal primers (NR1R and NR2F) are indicated with arrows. NR1R and NR2F were designed to amplify two fragments containing NR1 and NR2 regions by combining with L856237 and H943236 primers, respectively. 19.2.2.2

Identification of tuna using DNA–BMPs

Relative luminescence unit

The DNA probes were immobilized on BMPs by streptavidin–biotin conjugation, and subjected to magnetic-capture hybridization with digoxigenin (DIG)-labeled fragments amplified using the internal primers. PCR attempts using two primer sets, L8562/NR1R and NR2F/H9432, amplified 150 bp and 406 bp fragments, respectively. These DIG-labeled PCR products were subjected to hybridization reaction with DNA–BMPs. Figure 19.8 shows the result of studies in which hybridization between subspecies-specific probes (NR1/NBTP, NR1/NBTA, NR2/NBTP, and NR2/NBTA) immobilized on BMPs and target PCR products amplified from NBTP or NBTA genome was evaluated. The luminescence intensities from hybridization between probes and non-complementary fragments were projected as value relative to those obtained from the probes and the complementary fragments. When amplified fragmenta containing the NR1 region of NBTP was hybridized with the NR1/NBTA probe, the relative luminescence

1.00 0.75 0.50 0.25 0

NBTP NBTA

NR2 NBTA probe NR2 NBTP probe NR1 NBTA probe NR1 NBTP probe

Figure 19.8 Luminescence intensities obtained by hybridization between subspecies specific probes (NR1/NBTP, NR1/NBTA, NR2/NBTP, and NR2/NBTA) immobilized on BMP and DIG labeled target DNA fragments from NBTP and NBTA genome. Luminescence intensities from hybridization with non-complementary sequences were presented as relative units to those obtained with complementary sequences.

496

T. Matsunaga and T.-K. Lim

intensities were reduced to 0.81 0.16. When amplified fragments containing the NR1 region of NBTA were hybridized with the NR1/NBTP probe, the relative luminescence intensities were reduced to 0.84 0.05. Likewise, in hybridization between the NR2 fragment of NBTP and NR2/NBTA probe, the luminescence intensities were reduced to 0.69 0.05, and those between the NR2 fragment of NBTA and NR2/NBTP were reduced to 0.56 0.05. The luminescence intensities between the probes and the non-complementary fragments were apparently lower than those between the probes and the complementary fragments, even when there was only a single nucleotide difference. The number of nucleotide difference between the probes and the target fragment, as well as those sizes, may be significant for producing luminescence intensity difference. These data suggest that this system using DNA–BMPs may be useful for discriminating between these two subspecies, by recognizing single nucleotide difference. Further investigation on hybridization condition, length of target sequence and the positional effects of nucleotide substitution are necessary for the optimization of this detection system. The magnetic capture hybridization technique using BMPs is useful for further development of an automated system for high-throughput analysis, which may make the species identification procedure more rapid and less costly. Nucleotide sequence analysis for the other tuna species is underway, and should enable better DNA probes to be designed, and it is hoped that these will enable discrimination between all tuna species.

19.3

Electrochemical microsystems for biosensing immuno proteins

Miniaturization in the microelectronic and microsystem industries provides new opportunities and challenges, involving issues connected with trends in nanoscale structuring, fabrication of high aspect ratio structures, new functional alloys, multidimensional interconnections, and automated large-scale processes, including additive control and recycling of electrolytes. Here, outlined is the use of a single tip amperometric biosensor for immuno protein detection using human whole blood, and the experimental evidence is examined, leading to the conclusion that electrochemical detection methods can be used. The microstructure for micro liquid handling, including a flow immunoassay system, is also discussed. The microsystems have been employed for determination of allergen and human chorionic gonadotropin (HCG). 19.3.1

Electrochemical detection of allergen using the interdigitated array microelectrode

The numbers of patients exhibiting allergic responses such as pollenosis, asthma, and atopic dermatitis have increased in the recent years. Identification of allergen is important for prevention and cure. Skin tests,38,39 which include prick and scratch tests and food challenge,40,41 are commonly used to determine potential

Microsystems for biosensing 497 allergens. However, these in vivo tests are dangerous, because they sometimes cause anaphylaxis. The radio immunosorbent test (RIST),39,42 radioallergosorbent test (RAST),39,43 and a few other approved methods44–46 have also been used for in vitro diagnosis by detection of immunoglobulin E (IgE). IgE is a significant factor in the allergic reaction and can be measured at high sensitivity. However, these methods are time-consuming and require complicated procedures and specialized equipment. Furthermore, there are some reports indicating that the amount of IgE antibody does not correlate with the subjective symptoms.47–50 Therefore, a safe, simple, and rapid clinical method for the detection of allergens is required. Mast cells and basophils possess high-affinity receptors for the Fc fragment of IgE. When allergens contact these cells, cross linking with cell receptors occur, leading to the binding of two or more IgE sites. Binding triggers the secretion of chemical mediators such as histamine and serotonin. Chemical mediators cause the dilation of blood vessels, increasing their permeability. This phenomenon is responsible for many of the clinical manifestations associated with allergic reactions. Therefore, detection methods using chemical mediators have attracted attention, because the reaction can be reproduced in vitro. Electrochemical methods for detection of allergic reaction using RBL-1 cells51,52 and leukocytes,53 based on cyclic voltammetry, have been developed in our laboratory. Serotonin, a chemical mediator secreted when allergens stimulate mast cells or basophils, is used as a marker in this system because it can be electrochemically oxidized at a potential around 300 mV. Single-tip amperometric biosensors have been developed for measuring glucose concentrations and neurotransmitter concentrations.54,56 Niwa et al.57 have shown that the use of an interdigitated array microelectrode results in a significant decrease in detection limit, due to amplification of the limiting current by redox cycling. This section includes a description of the first example of an electrochemical method for the detection of allergen in whole blood sample by measuring of the current increase using an array microelectrode. 19.3.1.1

Interdigitated array microelectrode

The interdigitated array microelectrode consists of 50 pairs of microbands and two square electrodes for the detection of serotonin in whole blood (Figure 19.9). One of the square electrodes was plated with silver/silver chloride and used as a reference electrode. The overall size of this array microelectrode was 20 mm long and 10 mm wide, but the actual electrochemical cell sensing area is 2 2 mm. The width of the microbands was 5 m, and the gap between fingers 5 m. The microelectrode was sealed with polytetrafluoroethylene tape, except for the electrochemical cell area. The array microelectrode was rinsed with water before use. The electrode was placed in phosphate-buffered saline (PBS; 1.5 mM KH2PO4, 7.3 mM Na2HPO4, 137 mM NaCl, 2.7 mM KCl, pH 7.4) and 10 scans were taken at a scan rate of 120 mV/s in the range of 0–700 mV vs the silver/ silver chloride (Ag/AgCl) electrode, in order to inactivate the electro-reactive substance.

498

T. Matsunaga and T.-K. Lim

1 cm

Figure 19.9 Photograph of interdigitated array microelectrode. Material of electrode: Au, number of microband pairs: 50, bandwidth: 10 m, gap between microbands: 5 m, length of electrode: 2 mm.

19.3.1.2

Electrochemical detection of serotonin in whole blood

Cyclic voltammetry of whole blood with and without serotonin was performed using the interdigitated array microelectrode. Blood containing 1 g/ml serotonin produces an anodic peak around 350 mV, with a peak current of 22 nA (Figure 19.10). In contrast, cyclic voltammetry taken using blood without serotonin did not produce a corresponding peak. The detection of allergens using whole blood from individuals allergic to cedar pollen was examined by applying a 350 mV constant potential. Allergens were added to whole blood and the current was monitored during incubation. When the individual had an allergy to cedar pollen, the current in the blood increased with incubation time. A maximum current increase of 6.6 nA was obtained after 40 min incubation time (Figure 19.11). The current increase in blood from a person non-allergic to cedar pollen was below 1 nA. Allergen detection is therefore possible by measuring the current increase in allergen-exposed blood after 40 min. To determine the specificity of this detection method, the current increase was measured 40 min after addition of cedar pollen and mite allergen (Table 19.2). The IgE antibody in blood samples was measured as control. The detection of an allergic reaction in the blood sample of a person with an allergy to cedar pollen and mites was tested. The current increase observed with the addition of both allergens was 8.6 nA with cedar pollen and 13.3 nA with mite allergen. In whole blood samples, the amount of IgE was 4.70 UA/ml with cedar pollen and 60.0 UA/ml from mite allergen. Another person not allergic to cedar pollen but

Microsystems for biosensing 499

Current (µA)

0.40

0.20 B

A 0.00 0

200

400

600

Potential (mV vs Ag/Agcl)

Figure 19.10 Cyclic voltammograms of whole blood samples with PBS using a microelectrode. A: without serotonin, B: with serotonin (1 g/ml). 8.0

Current increase (nA)

6.0

4.0 2.0 0.0 –2.0

0

20 40 Time (min)

60

Figure 19.11 Time course of current increase after allergen addition using a whole blood sample from a person who is allergic (●) and unresponsive (嘷) to cedar pollen allergen.

allergic to mite allergen was examined. The current increase in this person’s blood was 0.6 nA after addition of cedar pollen and 6.8 nA after addition of mite allergen. Moreover, a person non-allergic to both cedar pollen and mite allergen was tested. For this person’s sample the current increase was only 0.7 and 0.4 nA for cedar pollen and mite allergen, respectively. Slight current increases were observed in this case. The results obtained using this electrochemical method correlated

500

T. Matsunaga and T.-K. Lim Table 19.2 Comparison between electrochemical allergen detection and amount of IgE antibody in blood Sex (age)

Allergen

Current increase (nA)

Amount of IgE antibody (UA/ml)

Female (30)

Cedar pollen Mite Cedar pollen Mite Cedar pollen Mite

8.6 13.3 0.6 6.8 0.7 0.4

4.70 60.0 0.34 48.06 0.34 0.34

Male (24) Male (23)

below the detection limit.

well with IgE antibody concentrations in blood samples. This suggests that specific allergen detection is possible using interdigitated array microelectrodes. The reproducibility of allergen detection was examined using whole blood samples from two people, one allergic to cedar pollen and the other not. The coefficients of variation were within 0.6–1.9%, and show good correlation (n 5). 19.3.1.3

Diagnosis of cedar pollen allergy

The relationship between IgE antibody concentration in blood serum and current increase was investigated. Whole blood samples from 17 subjects were used, and the IgE antibody concentration was measured (Table 19.3). Among 17 samples, the results obtained from the electrochemical detection method were consistent with clinical diagnosis obtained by IgE concentrations in 13 samples. The subjects were split into four groups (A–D) on the basis of our results, IgE concentration, and symptoms. In group B (diagnosis; positive) and D (diagnosis; negative), the results obtained using the electrochemical method correlate with diagnosis determined by IgE antibody concentrations. Results obtained using the electrochemical detection method that did not correlate with IgE antibody concentrations were included in groups A and C. In group A, the background current (current without addition of the allergen protein fraction) was higher than in other groups (about 5–10 nA) and the current increase on addition of the allergen protein fraction was 0.7 and 1.2 nA. The amount of IgE was already high in this group. The differences in diagnosis might be explained by the season. Experiments were performed in the middle of the cedar pollen allergy season, and there was therefore secretion of serotonin before the test. Therefore, background currents were high and current increase was below the level for a positive diagnosis. Group C consisted of two subjects with IgE below 0.34 UA/ml (diagnosis; negative) who displayed symptoms. The current increase was approximately 2–3 nA (diagnosis; positive) using the electrochemical detection method. Since allergic symptoms are related to secretion of chemical mediators, electrochemical methods might possibly be used to detect allergic reactions from other pathways.

Microsystems for biosensing 501 Table 19.3 Correlation of electrochemical detection for cedar pollen and amount of IgE antibody Group

Sex (age)

Current increase (nA)

Amount of IgE (UA/ml)

Subjective symptom

A

Female (36) Male (28) Male (35) Male (29) Male (27) Female (29) Female (33) Female (25) Male (23) Female (26) Male (24) Male (25) Male (22) Male (23) Male (24) Female (23) Male (24)

0.7 1.2 3.9 14.6 4.5 3.2 8.7 4.1 9.5 3.1 2.3 0.2 1.1 0.7 0.3 4.8 0.7

39.0 19.5 11.1 10.3 10.7 7.40 4.70 2.20 1.30 0.34 0.34 0.34 0.34 0.34 0.34 0.34 0.34

B

C D

Notes

pseudopositive symptoms. below the detection limit.

It was shown that sensitive diagnosis of cedar pollen allergy was possible using this electrochemical detection method. The detection method correlates with inhalation and loading tests. Therefore, detection of allergen in whole blood samples can be performed by measuring the chemical mediator serotonin using an array microelectrode. Detection of allergen was possible by monitoring the current increase. In this electrochemical detection method, 20 l of whole blood is sufficient for detection of the allergic reaction. This method may be suitable for use as a self-care type allergen detection system that does not need special medical equipment and physicians. The advantage of this method is its safety, as it does not require the radioisotopes or in vivo tests used in other detection methods. 19.3.2

Miniaturization of flow immunoassay system based on different isoelectric points

Immunoassays coupled to a flow injection analysis system (FIA) are powerful analytical tools for measuring low levels of analytes such as antibodies, drugs, or hormones. During the last few years, flow injection immunoassays using small immunoaffinity columns, referred to as immunoreactors, have been developed. In an immunoreactor, the antibody on the ligand-binder is covalently coupled to a rigid support. Antibodies,58,59 protein A,60 protein G,61 and concanavalin A62,63 have been used as ligand binders and Sepharose,64,65 non-porous silica66 and

502

T. Matsunaga and T.-K. Lim

Biomag 4100 beads67 have been used as the solid phases. However, the problem with using antibodies (or antigen) immobilized on a solid-phase is the difficulty in controlling the concentration of reactants. The concentration of antigen (or antibody) cannot be easily adjusted by conventional methods. Furthermore, for the successful application of automated flow systems, affinity-binding sites have to be regenerated. Regeneration is achieved using appropriate reagents, such as solutions with low or high pH and/or high salt concentrations. However, regeneration damages and release immobilized antibodies, resulting in reduced binding capacities. Antibodies are grafted randomly on the solid phase, further limiting binding capacity. Furthermore, each antibody is specific for a particular antigen, limiting the range of analytes that can be assayed. We have developed an automated flow immunoassay system using an ion exchange column to detect food allergens,68,69 anti-double-stranded DNA antibody in systemic lupus erythematosus serum,70,71 and HCG. Antibody–antigen complexes were subsequently separated from unreacted antibody on an ion exchange column, on the basis of differences in the isoelectric point (pI).72 This method does not require antibody (or antigen) immobilization onto a solid phase, and overcomes regeneration limitations found in conventional methods. Furthermore, recovered free antibodies could be reused in the immunoassay without problems. Recently, enzyme electrodes making use of electron transfer mediators have received considerable attention.73–76 Ferrocene derivatives are fast reversible redox mediators, and are used in many biosensors. These electrodes have been applied to reagent-less biosensors and electrochemical enzyme flow immunoassay.77 However, enzyme flow immunoassays using immunocolumns require complicated procedures for continual use because of a separate amperometric detector where enzyme-generated electroactive products are oxidized or reduced at the electrode surface. During the last few years, rapid developments in microsystem technologies have produced miniaturized sensors.78–80 Integration of these new sensors requires the development of miniaturized systems capable of utilizing them. In this section, we describe an attempt to miniaturize the flow immunoassay system using an ion exchange capillary column and ferrocene-conjugated IgG (FG-IgG). This approach is simple and offers shorter assay times than with other electrochemical immunoassay systems, because it does not require enzyme-generated electroactive products. Furthermore, it significantly reduces the quantity of reagents required. 19.3.2.1

Preparation of the ferrocene-conjugated IgG and capillary column

We prepared FC-IgG using EDC (1-ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride), for continuous detection of HCG (Figure 19.12). This approach is simple and offers shorter assay times than with other electrochemical immunoassay systems because it does not require enzyme-generated electroactive products. Furthermore, it significantly reduces the quantity of reagents required. When cyclic voltammetry was carried out in the ferrocene solution, a redox

Microsystems for biosensing 503

CH3

CH3

CH3

N (CH2)3 C

+

O

Fe

N

OH

C N CH2 CH3

C pH 7.3

CH3 CH3

N

Fe

CH3

(CH2)3

N

NH

(CH2)3

O C O N

NH2

C Fe

CH2 CH3

NH O

+

NH C O NH CH2 CH3

Figure 19.12 Immobilization of ferrocene on IgG using EDC.

1250

Change in current (µA)

1000

750

500

250

0 0

8 10 12 2 4 6 Number of conjugated FCs per IgG

Figure 19.13 Relationship between the number of ferrocenes and changes in oxidative current. The current, 395 mV vs Ag/AgCl obtained by cyclic voltammetry was plotted against the number of conjugated ferrocenes. Each conjugate was adjusted to 5 g/ml.

potential appeared at 395 and 300 mV. Similar electrochemical behaviors have been observed with ferrocenemonocarboxylic acid.81 These results suggest that electrochemical oxidation of ferrocenemonocarboxylate is catalyzed at a potential of 395 mV vs Ag/AgCl. Therefore, application of a 395 mV potential is appropriate for the detection of Fc-IgG and HCG immunocomplexes. Analysis of the IgG solution by atomic absorption spectroscopy shows that prior to modification there are zero iron atoms per IgG molecule and after modification there are 11 per molecule. The oxidative current increased with increasing numbers of ferrocene on the IgG antibody, as shown in Figure 19.13. A cation exchange capillary column (762 m i.d. 10 cm) was prepared using a SELF PACK kit (Perseptive Biosystems, Framingham, USA) previously

504

T. Matsunaga and T.-K. Lim

Figure 19.14 Photograph of a prepared cation exchange capillary column using PEEK tubing.

described (Figure 19.14). The downstream end of a polyetheretherketone (PEEK) capillary column was fitted with a commercially available in-line filter containing metal frits (PerSeptive Biosystems Inc., Frammingham, MA) 1 mm in diameter and 0.5 mm thick installed in 1/16 inches o.d. PEEK rings. The other end of the PEEK capillary column was connected to the slurry reservoir (stainless tube; 3 mm i.d., 60 mm length). The packing slurry was prepared by adding 20 mg of the cation exchange particles (POROS 10 SP; sulfopropyl type, PerSeptive Biosystems Inc.) to 100 ml distilled water. This mixture was stirred for 2 min. This material is a perfusive support with a 10 m particle diameter modified with a polyhydroxylated polymer made functional by sulfopropyl groups. After filling the slurry reservoir, the slurry was pumped into the PEEK capillary column using distilled water at a pressure of 1000 psi. When packing is complete, the capillary column is flushed for another 5 min at 1000 psi using a HPLC pump (LC-6A, Shimadzu, Kyoto, Japan). The pump is then switched off and the pressure allowed to become zero before disconnecting the column from the slurry reservoir. The column could be reused forty times. After forty times, the response was approximately 85% of the first use. 19.3.2.2

Continual electrochemical flow immunoassay analysis using capillary column

Analysis of the isoelectric points for HCG (antigen), Fc-IgG antibody and antigen–antibody complexes were carried out by isoelectric gel electrophoresis using the Pharmacia PhastsystemTM. Three species were found to have isoelectric points at 7.0 (Fc-IgG), 4.7 (HCG), and 5.6 (HCG–Fc-IgG complex). The antigen– antibody complex arising from the immunoreaction between Fc-IgG and HCG was separated from unreacted Fc-IgG using optimized conditions, with

Microsystems for biosensing 505 a cation exchange capillary column equilibrated at pH 6.0 with 50 mM malonate buffer. The electrochemical flow immunoassay system is shown in Figure 19.15. HCG (5 l) was mixed with 5 l of 5 g/ml Fc-IgG. The mixture was loaded into an injection loop, incubated for 30 min, and passed into the cation exchange capillary column (sulfopropyl type, 762 m i.d. 10 cm), with malonate buffer (pH 6.0) at a flow rate of 40 l/min. The catalytic current of Fc-IgG in the eluent was measured using a three-electrode flow cell (BAS Inc., Japan) equipped with a glassy carbon electrode, an Ag/AgCl reference electrode, and a platinum counter-electrode. Using the capillary electrochemical flow immunoassay equipment, various concentrations of HCG was injected into the system with malonate buffer eluent under optimal analysis conditions. The response from a series of Fc-IgG and HCG complex injections is shown in Figure 19.16. The signal from this complex increased with HCG concentration. Furthermore, reversible, reproducible, and sensitive responses were obtained, indicating that the liquid-phase immunological sensing system was working. Injections of Fc-IgG alone produced no signal, indicating that the antibody was bound to the cation exchange resin. A typical calibration response for HCG is shown in Figure 19.17. A linear relationship was observed between the signal and HCG concentration between 0 and 2000 mIU/ml. The correlation coefficient was 0.997 within this range. This method was faster (1.3 min), simpler to use, and gave better precision (1.8%) than with conventional microtiter plates ELISA.82–84 Reproducibility of the HCG

(a)

Pump Buffer

Injector

Cation exchange capillary column

Electrochemical detector

10 cm

Working electrode 3 cm 762 µm

Ferrocence-IgG (Fc-IgG)

HCG

Counter electrode 1.5 cm

(b)

Ag/Agcl Waste reference electrode

HCG–Fc-IgG complex

Figure 19.15 Schematic diagram of an electrochemical capillary flow immunoassay system (a) and the principle of electrochemical capillary flow immunoassay (b). The reaction mixture consists of HCG, Fc-IgG and HCG–Fc-IgG complex. The mixture is passed through a capillary column that selectively traps Fc-IgG. The eluted species pass through the flow cell, and the current associated with the HCG–Fc-IgG antibody complex is monitored.

506

T. Matsunaga and T.-K. Lim 1200 (A) 2 min

Current (µA)

1000 800

(B) (C)

600 (D) 400 (E)

200 0

Time

Figure 19.16 Continual electrochemical flow immunoassay analysis. (A) 2000, (B) 1200, (C) 1000, (D) 500, and (E) 0 mIU/mL HCG were reacted with 5 g/mL Fc-IgG, and injected into the system. Arrows indicate sample injections.

measurement was examined at four different concentrations (20, 100, 500 and 1000 mIU/ml). Coefficients of variation were within 1.2–2.1%, and showed good correlation (R 0.996, n 10). The eluent from the capillary column was analyzed by native polyacrylamide gel electrophoresis. The eluent gel electrophoresis band showed a size difference compared with Fc-IgG (data not shown). This band was identified as antibody– antigen complexes determined by the molecular weight of Fc-IgG and HCG. Furthermore, free Fc-IgG retained by the cation exchange resin was detected after elution with 50 mM malonate buffer supplemented with 0.5 M sodium chloride. The cation exchange capillary column was regenerated (i.e. bound Fc-IgG removed from the capillary column) by elution with malonate buffer (pH 6.0) containing 0.5 M NaCl, with continual monitoring of HCG–Fc-IgG complexes. This resulted in the appearance of one peak after 1.3 min. The extract was analyzed by native polyacrylamide gel electrophoresis. A band was identified equal in size to Fc-IgG. The ability to use recycled Fc-IgG for the HCG immunoassay was investigated. The mixture of HCG (5 l, 1000 mIU/ml) and recycled Fc-IgG (5 l, 5 g/ml) was used for the electrochemical flow immunoassay and the signals were monitored. Our data suggests that the signals from recycled HCG–Fc-IgG antibody complexes were similar to those obtained using new HCG–Fc-IgG antibody. Fc-IgG recovered may be reused up to eight times without significant effects on efficiency (Figure 19.17). After eight times, the response was approximately 70% that of the first use. The capillary column was rapidly regenerated with malonate buffer containing 0.5 M NaCl, enabling the efficient reuse of Fc-IgG. These results indicate that this electrochemical flow immunoassay system is suitable for diagnosis of pregnancy in clinical practice.

Microsystems for biosensing 507

1000

Current (µA)

800 600 10 8

400

6 4

200

2 0

0

0

0

5

10

15

400 800 1200 1600 Concentration of HCG (mIU/mL)

20

2000

Figure 19.17 Relationship between current and HCG concentration. Fc-IgG; 5 g/ml. R 0.997 (0–2000 mIU/ml).

Table 19.4 Specificity of the electrochemical capillary flow immunoassay system Sample

Current (A)

None HCG HCG Bovin serum albumin (BSA) HCG IgG BSA IgG BSA IgG

0.0 590.5 570.2 576.4 2.1 0.8 1.2

Note: HCG concentration: 1 IU/ml; concentration of protein: 45 pmol/ml; Fc-IgG concentration: 5 g/ml.

The specificity of the electrochemical flow immunoassay for HCG was investigated by assaying a variety of other proteins (Table 19.4). This system generated a significant signal only with HCG.

19.4

Conclusions

Genomics,85 proteomics,86 immunology,87 neuroscience,88 and drug investigations89 now require smaller quantities of analytes to be separated from large sample numbers. Rapid development of microsystem technologies is producing miniaturized sensors, and integration of these new sensors requires development of miniaturized systems capable of utilizing them.

508

T. Matsunaga and T.-K. Lim

The development of a MAG-microarray enables automated sample preparation employing magnetic separation robot and bacterial magnetic particles. The method for detection of cyanobacterial DNA was demonstrated using designed oligonucleotide probes conjugated BMPs and the MAG-microarray. After each reaction and washing steps, the standard deviation of the amount of BMPs spotted onto the MAG-microarray was 8%. This variation might be the primary cause of stronger fluorescence intensity in Nostoc than with Anabaene when using the Anabaena2 probe. The time required for hybridization is considerably shorter (5 min) with the bead system than with the array system, because the beads have a large surface area (reaction area). This microarray is reusable simply by washing with sonication. Re-analysis of target DNA is also possible by recovery of target DNA. The large number of wells can be fabricated in an array for high-throughput analysis. Histamine, leukotrienes,90,91 and tryptase92 are reported chemical mediators used for allergen detection with the glass microfiber-based basophil histamine release test,93,94 which is now in clinical use. However, this method also requires complicated procedures for detection. Therefore, a safe, simple, and rapid clinical method for the detection of allergens is required. Serotonin in whole blood can be detected by measuring the current when a constant potential of 350 mV is applied. The current increase correlated with serotonin concentration in whole blood. Detection of allergen was possible by monitoring the current increase. Using this electrochemical detection method, 20 l of whole blood is sufficient for detection of the allergic reaction. This method may be suitable for use as a self-care detection system that does not need special medical equipment or physicians. The advantage of this method is its safety, as radioisotopes and in vivo tests are not required. It has been shown that HCG was detected using a miniaturized microsystem. This approach is simple and offers shorter assay times compared with other electrochemical immunoassay systems, because it does not require enzymatically generated electroactive products. Free Fc-IgG bound to the cation exchange capillary column was easily eluted with malonate buffer containing sodium chloride. Analysis of serum from an ectopic pregnancy patient resulted in a concentration of less than 1000 mIU/mL HCG.95 These results indicate this immunosystem is suitable for diagnosis of pregnancy and/or gonadal tumors in clinical practice. This capillary electrochemical flow immunoassay requires only minute quantities of serum, and generates reproducible results.

References 1. B. Wagner, Endosc. Surg. Allied. Technol. 3, 204 (1995). 2. D. Pyo, J. Biochem. Biophys. Methods 43, 113 (2000). 3. R. Andrews, R. Mah, A. Aghevli, K. Freitas, A. Galvagni, M. Guerrero, R. Papsin, C. Reed, and D. Stassinopoulos, Stereotact. Funct. Neurosurg. 73, 1 (1999). 4. B. Wolf, M. Brischwein, W. Baumann, R. Ehret, T. Henning, M. Lehmann, and A. Schwinde, Tumour. Biol. 19, 374 (1998).

Microsystems for biosensing 509 5. M. Brischwein, W. Baumann, R. Ehret, A. Schwinde, M. Kraus, and B. Wolf, Naturwissenschaften. 83, 193 (1996). 6. A. K. Singh, A. W. Flounders, J. V. Volponi, C. S. Ashley, K. Wally, and J. S. Schoeniger, Biosen. Bioelectr. 14, 703 (1999). 7. H. Tu, J. Xue, X. Cao, W. Zhang, and L. Jin, Analyst 125, 163 (2000). 8. B. Wolf, M. Brischwein, W. Baumann, R. Ehret, and M. Kraus, Biosen. Bioelectr. 13, 501 (1998). 9. W. A. Groves and E. T. Zellers, Am. Ind. Hyg. Assoc. J. 57, 1103 (1996). 10. A. Satoh, Sensors and Actuators A: Physical 72, 160 (1999). 11. All reviews in, Nat. Genet. No.1 21, 1 (1999). 12. J. G. Hacia, L. C. Brody, and F. S. Collins, Mol. Psychiatry. 3, 483 (1998). 13. M. Schena, DNA Microarrays (Oxford University Press Inc., New York, 1999). 14. D. Shalon, Pathol. Biol. (Paris) 46, 107 (1998). 15. P. Toronen, M. Kolehmainen, G. Wong, and E. Castren, FEBS Lett. 451, 142 (1999). 16. M. A. Behr, M. A. Wilson, W. P. Gill, H. Salamon, G. K. Schoolnik, S. Rane, and P. M. Small, Science 284, 1520 (1999). 17. M. Chee, R. Yang, E. Hubbell, A. Berno, X. C. Huang, D. Stern, J. Winkler, D. J. Lockhart, M. S. Morris, and S. P. Fodor, Science 274, 610 (1996). 18. J. DeRisi, L. Penland, P. O. Brown, M. L. Bittner, P. S. Meltzer, M. Ray, Y. Chen, Y. A. Su, and J. M. Trent, Nat. Genet. 14, 457 (1996). 19. D. A. Lashkari, J. L. DeRisi, J. H. McCusker, A. F. Namath, C. Gentile, S. Y. Hwang, P. O. Brown, and R. W. Davis, Proc. Natl. Acad. Sci. USA 94, 13057 (1997). 20. D. Shalon, S. J. Smith, and P. O. Brown, Genome. Res. 6, 639 (1996). 21. D. Y. Guschin, B. K. Mobarry, D. Proudnikov, D. A. Stahl, B. E. Rittmann, and A. D. Mirzabekov, Appl. Environ. Microbiol. 63, 2397 (1997). 22. J. W. Schultze and V.T. Sakova, Electrochem. Acta 44, 3605 (1999). 23. M. Datta and D. Landolt, Electrochem. Acta 45, 2535 (2000). 24. J. Chen, R. Johnson, and M. Griffiths, Appl. Environ. Microbiol. 64, 147 (1998). 25. H. J. Bach, A. Hartmann, J. T. Trevors, and J. C. Munch, J. Microbiol. Methods 37, 187 (1999). 26. T. Matsunaga and S. Kamiya, Appl. Microbiol. Biotechnol. 26, 3238 (1987). 27. T. Matsunaga, M. Kawasaki, X. Yu, N. Tsujimura, and N. Nakamura, Anal. Chem. 68, 3551 (1996). 28. T. Matsunaga, R. Sato, S. Kamiya, T. Tanaka, and H. Takeyama, J. Magn. Magn. Mater. 194, 126 (1999). 29. R. W. Castenholz and J. B. Waterbury, Bergey’s Manual of Systematic Bacteriology (Williams & Wilkins Co., Baltimore, MD, 1989), pp.1710–1728. 30. F. Garcia-Pichel, L. Prufert-Bebout, and G. Muyzer, Appl. Environ. Microbiol. 62, 3284 (1996). 31. B. Nelissen, Y. Van de Peer, A. Wilmotte, and R. De Wachter, Mol. Biol. Evol. 12, 1166 (1995). 32. U. Nubel, F. Garcia-Pichel, and G. Muyzer, Appl. Environ. Microbiol. 63, 3327 (1997). 33. A. Wilmotte, S. Turner, Y. Van de Peer, and N. R. Pace, J. Phycol. 28, 828 (1992). 34. K. Rudi, O. M. Skulberg, F. Larsen, and K. S. Jakobsen, Appl. Environ. Microbiol. 63, 2593 (1997). 35. G. D. Sharp and S. Pirages, In: The Physiological Ecology of Tunas (Sharp and Dizon eds, 1978), pp. 41–78. 36. S. Chow and H. Kishino, J. Mol. Evol. 41, 741 (1995). 37. S. Chow and S. Inoue, Nat. Res. Inst. Far. Seas Fish 30, 207 (1993).

510

T. Matsunaga and T.-K. Lim

38. S. Dreborg, R. Einarsson, and J. L. Longbottom, The Chemistry and Standardization of Allergens (Blackwell Scientific, London, 1986), 4th edn, Ch.10, pp. 1–28. 39. Y. Green-Graif and P. W. Ewan, Clin. Allergy 17, 431 (1987). 40. P. Baehler, Z. Chad, C. Gurbindo, A. P. Bonin, L. Bouthillier, and E. G. Seidman, Clin. Exp. Allergy 26, 254 (1996). 41. U. Bengtsson, U. Nilsson-Balknas, L. A. Hanson, and S. Ahlstedt, Gut. 39, 130 (1996). 42. J. Merrett and T. G. Merrett, Clin. Allergy 17, 409 (1987). 43. A. Norgard, C. Bindslev-Jensen, P. Stahl, and L. K. Poulsen, Allergy 50, 636 (1995). 44. I. Iwamoto, H. Yamazaki, A. Kimura, K. Ochiai, and N. Nakagawa, J. Japan. Allergy 39, 1374 (1990). 45. S. Mita, N. Oda and M. Adachi, J. Japan. Allergy 44, 1394 (1995). 46. S. Sumimoto, M. Kawai, Y. Kasajima, and T. Hamoto, J. Japan. Allergy 39, 1416 (1990). 47. M. Daeron, C. Bonnerot, S. Latour, and W. H. Fridman, J. Immunol. 149, 1365 (1992). 48. L. M. Lichtenstein, A. Kagey-Sobotka, J. M. White, and R. G. Hamilton, J. Immunol. 148, 3929 (1992). 49. H. C. Oettgen, T. R. Martin, A. Whyshaw-Boris, C. Deng, J. M. Drazen, and P. Leder, Nature 370, 367 (1994). 50. F. Shakib and S. J. Smith, Clin. Exp. Allergy 24, 270 (1994). 51. T. Matsunaga, A. Shigematsu, and N. Nakamura, Anal. Chem. 61, 2471 (1989). 52. N. Nakamura, S. Kumazawa, and T. Matsunaga, Appl. Microbiol. Biotechnol. 43, 622 (1995). 53. N. Nakamura, I. Inoue, Y. Kitazima, T. Matsunaga, T. Chiba, and T. Honda, Biosen. Bioelectr. 6, 431 (1991). 54. T. Matsumoto, M. Furusawa, and N. Ito, Chem. Sens. 12, 17 (1996). 55. O. Niwa, Y. Xu, H. Brian, and W. R. Heineman, Anal. Chem. 65, 1559 (1993). 56. O. Niwa, M. Morita, and H. Tabei, Electroanalysis 6, 237 (1994). 57. O. Niwa, M. Morita, and H. Tabei, Anal. Chem. 62, 447 (1990). 58. D. Leibler, A. Rabinkov, and M. Wilchek, J. Mol. Recognit. 9, 375 (1996). 59. P. Hammerl, A. Hartl, and J. Thalhamer, J. Immunol. Methods 165, 59 (1993). 60. E. Shpigel, A. Goldlust, A. Eshel, I. K. Ber, G. Efroni, Y. Singer, I. Levy, M. Dekel, and O. Shoseyov, Biotechnol. Appl. Biochem. 31, 197 (2000). 61. C. Kasper, L. Meringova, R. Freitag, and T. Tennikova, J. Chromatogr. A 798, 65 (1998). 62. A. F. Bergold and P. W. Carr, Anal. Chem. 61, 1117 (1989). 63. K. Isosaki, N. Seno, I. Matsumoto, T. Koyama, and S. Moriguchi, J. Chromatogr. 597, 123 (1992). 64. A. Berneman, B. Guilbert, S. Eschrich, and S. Avrameas, Mol. Immunol. 30, 1499 (1993). 65. K. E. Meyers, P. A. Kinniry, R. Kalluri, E. G. Neilson, and M. P. Madaio, Kidney Int. 53, 402 (1998). 66. Q. M. Mao, I. G. Prince, and M. T. Hearn, J. Chromatogr. A, 691, 273 (1995). 67. W. Stocklein and R. D. Schmid, Anal. Chim. Acta 234, 83 (1990). 68. T. K. Lim, N. Nakamura, and T. Matsunaga, Anal. Chim. Acta 354, 29 (1997). 69. T. K. Lim, N. Nakamura, and T. Matsunaga, Anal. Chim. Acta 370, 207 (1998). 70. T. K. Lim, Y. Komoda, N. Nakamura, and T. Matsunaga, Anal. Chem. 71, 1298 (1999). 71. T. K. Lim, N. Nakamura, and T. Matsunaga, Biotech. Bioengi. 68, 571 (2000). 72. T. K. Lim, N. Nakamura, and T. Matsunaga, Denki Kagaku (presently Electrochemistry) 63, 1154 (1995).

Microsystems for biosensing 511 73. B. Serra, A. J. Reviejo, C. Parrado, and J. M. Pingarron, Biosen. Bioelectr. 14, 505 (1999). 74. X. Cui, D. Jiang, P. Diao, J. Li, R. Tong, and X. Wang, Bioelectrochem. Bioenerg. 48, 243 (1999). 75. K. Yokoyama and Y. Kayanuma, Anal. Chem. 70, 3368 (1998). 76. T. Kohno, S. Asai, Y. Iribe, I. Hosoi, K. Shibata, and K. Ishikawa, J. Neurosci. Methods 81, 199 (1998). 77. C. Valat, B. Limoges, D. Huet, and J. L. Romette, Anal. Chim. Acta 404, 187 (2000). 78. P. E. Michel, G. C. Fiaccabrino, N. F. de Rooij, and M. Koudelka-Hep, Anal. Chim. Acta 392, 95 (1999). 79. H. A. Clark, R. Kopelman, R. Tjalkens, and M. A. Philbert, Anal. Chem. 71, 4837 (1999). 80. J. J. Gooding and E. A. H. Hall, Anal. Chem. 70, 3131 (1998). 81. M. Okochi, N. Nakamura, and T. Matsunaga, Electrochim. Acta 44, 3795 (1999). 82. N. Venkatesh and G. S. Murthy, J. Immuno. Methods 199, 167 (1996). 83. L. R. Witherspoon, S. E. Shuler, G. F. Joseph, E. F. Baird, H. R. Neely, and R. E. Sonnemaker, Clin. Chem. 38, 887 (1992). 84. S. Madersbacher, G. Untergasser, R. Gerth, M. Hermann, D. S. Schwäzler, and P. Berger, Exp. Clin. Endocrinol. Diabetes 106, 61 (1998). 85. M. Chee, R. Yang, E. Hubbell, A. Berno, X. C. Huang, D. Stern, J. Winkler, D. J. Lockhart, M. S. Morris, and S. P. A. Fodor, Science 274, 610 (1997). 86. S. P. A. Fodor, J. L. Read, M. C. Pirrung, L. Stryer, A. T. Lu, and D. Solas, Science 251, 767 (1991). 87. L. B. Koutny, D. Schmalzing, T. A. Taylor, and M. Fuchs, Anal. Chem. 68, 18 (1996). 88. J. Keller, H. Strasburger, D. T. Cerutti, and B. A. Sabel, J. Neurosci. Methods 97, 103 (2000). 89. S. Wissing, D. Q. Craig, S. A. Barker, and W. D. Moore, Int. J. Pharm. 199, 141 (2000). 90. M. Sasagawa, J. Japan. Allergy 39, 1556 (1990). 91. H. Yamamoto, K. Kuramitsu, I. Houya, H. Maruo, K. Sakata, K. Tabe, I. Kimura, K. Inoue, M. Shibasaki, K. Yamamoto, Y. Sakamoto, and H. Matsuo, J. Japan. Allergy 45, 1106 (1996). 92. L. B. Schwartz, T. R. Bradford, C. Rouse, A. M. Irani, G. Rasp, J. K. Van der Zwan, and P. W. Van der Linden, J. Clin. Immunol. 14, 190 (1994). 93. E. Iwasaki, M. Yamaura, Y. Miyabayashi, K. Yamaguchi, M. Zaitsu, K. Fujimaki, and M. Baba, J. Japan. Allergy 43, 609 (1994). 94. K. Tabe, F. Kaneko, N. Kawamura, M. Ootomo, Y. Maeda, T. Hayakawa, M. Hasegawa, K. Akiyama, H. Mita, T. Shida, and T. Miyamoto, J. Japan. Allergy 43, 527 (1994). 95. B. Cacciatore, U-H. Stenman, and P. Ylostalo, Br. J. Obstet. Gynaecol. 97, 904 (1990).

20 New microelectrode arrays for biosensing and membrane electroporation Eberhard Neumann and Katja Tönsing

20.1

Introduction

It is very impressive to note that electrochemical metal and carbon electrodes, manufactured with electrochemical microsystem technologies, find increasingly new applications not only in conventional physical-electrochemistry but also in cell biology, biotechnology and biomedical research for new clinical electrotherapies. The developmental trends in the technical devices largely aim at arrays of electrodes and further miniaturizing of these arrays as well as to include automatic computer-aided control of analytical and preparative electrode functions. This essay on new trends selects two representative cases for new electrode devices in cell biology and medical research, heading at new diagnostic and therapeutic clinical applications. One example is from the field of biosensing neurotransmitter substances and the other one refers to new developments of electrode systems for the electrochemical delivery of drugs and genes to biological cell aggregates and tissue by the powerful method of membrane electroporation.1 In both cases addressed to, the new trends include to use the electrical feedback control of electrode arrays for biosensing processes as well as for the extent and duration of tissue electroporation. In line with the impressive advances in medical microsurgery, where increasingly smaller organ targets become accessible, microelectrode systems have become a continuous technical challenge for bioanalytical microelectrode arrays and, as discussed here in some detail, for the new field of the electroporative delivery of effector substances like drugs and genes, using miniaturized electrochemical electrode arrays.

20.2

Microsensing cholinergic effector substances

There are numerous attempts to develop biosensors for neurotransmitters using the natural receptor proteins of the cellular transduction cascades. Some of these neurotransmitter receptor proteins are molecularly well characterized. Indeed, the nicotinic acetylcholine receptor (AcChoR) serves as a paradigm for the structure and the function of signal transduction proteins, converting the binding of an extracellular ligand molecule into electrical crossmembrane currents of alkali and alkaline earth metal ions. The ion-channel function of the membrane-bound

Biosensing and electroporation 513 AcChoR has been explored in great detail with the Neher–Sakmann patch-clamp technique.2 20.2.1

Membrane-bound sensory proteins

It should be noted that the function of the AcChoR as a channel protein is intrinsically coupled to a membrane. For practical reasons of using the AcChoR as an electro-chemical biosensor, the AcChoR containing lipid membrane has to be fixed. Moreover, a stable reusable mechanical configuration must also guarantee the stability of the membrane system. The new trends in manufacturing membrane-bound biosensors are to use solid supported lipid bilayers.3 For the electrochemical monitoring of transmembrane ion currents, the bilayer has to be fixed by a spacer to keep a certain distance to the solid surface. An example for such a tethered bilayer4 is the peptide support lipid bilayers.5 The principle of such a sensor configuration for the reconstituted AcChoR is shown in Figure 20.1. It was established that the AcChoR in such a tethered membrane is stable and binds cholinergic ligands and inhibitor molecules in practically the same manner as known from solution biochemistry.6 In solution the AcChoR exists as a part of a mixed ternary micelle of lipids and detergent molecules. Preparatively, this type of AcChoR biosensor is readily formed from lipid bilayer vesicles. The vesicles spontaneously associate with the thio-peptide-lipid monolayers coupled to a thin gold layer electrode via Au–S bounds.6

AcCho+

6

+ S S

+

–

5

U0 Uin

3

– Au

Figure 20.1 Principle of the cholinergic sensory system based on solid supported lipid bilayers containing the sensing acetylcholine receptor dimer species (Mr ≈ 580 000). The lipid membrane is covalently coupled to a gold surface via Au–S bonds of thiopeptide lipids in the monolayer facing the solid support. The approximate microdimensions of the spacings are 3 nm, 5 nm bilayer thickness and about 6 nm receptor vestibules. The perspective for the electrical circuit, to record the successful sensing of acetylcholine (AcCho) ions in terms of Na/K current events (double arrows across the proteins), is similar to a patch-clamp circuit.

514

E. Neumann and K. Tönsing

(a) Normal, [AcCho] = 0

10 pA 100 ms

(b) Normal, [AcCho] = 2 µM

10 pA 100 ms

(c) Hyperphosphorylated, [AcCho] = 2 µM

10 pA 100 ms

(d) 10 pA 100 ms

Hyperphosphorylated, [AcCho] = 0

0

0.5

1.0 t (s)

Figure 20.2 Recordings (current/time traces) of single channel events of the lipid vesiclereconstituted Torpedo AcChoR dimer under patch-clamp conditions. Buffer: [KCl] 100 mM, [CaCl2] 0.5 mM, [HEPES] 10 mM, pH 7.4, 0.01% NaN3, at T 295 K (22ºC) and [AcCho] as indicated. (a) Normal AcChoR dimer at V 80 mV, no AcCho; (b) Normal AcChoR dimer at V 80 mV, [AcCho] 2 M; (c) Hyperphosphorylated AcChoR dimer at V 100 mV, [AcCho] 2 M; (d) Hyperphosphorylated AcChoR dimer at V 100 mV, no AcCho present.

It is noted that the AcChoR bilayer vesicles are not only the source for the sensor protein but also the important control system for the proper functioning of the reconstituted AcChoR protein. The vesicles can be prepared large enough (a few micrometre in diameter) such that the patch-clamp technique can be applied. Proper functioning is then indicated by ligand-induced single current events (Figure 20.2), which can be blocked by inhibitors. 20.2.2

Electrochemistry of the nicotinic AcChoR

With the help of such vesicle patches many details of the molecular-electric properties of the AcChoR have been explored.7,8 For instance, it was established that the electric conductance (G) of the so-called single channel events, which are

Biosensing and electroporation 515 caused by the dimer species (Mr ⬇ 580 000) of the Torpedo californica AcChoR, is twice as large as those originating from the monomer species (Mr ⬇ 290 000). The monomer fragment, which can be isolated from electric tissue under chemically reducing conditions, is usually called the functional receptor unit, with a pentameric subunit composition of ␣2␤␥. The dimer is the -disulfide-bridged species (␣2␤␥–␥␤␣2) and appears to be the natural AcChoR complex. Both the dimer and the monomer fragment give rise to single channel events (Figure 20.2). The patch clamp K-ion conductance of the dimer reconstituted in lipid vesicles decreases with increasing Ca2-concentration in the range 0.1 [Ca2]/mM 2. This feature can be quantitatively rationalized by Ca2 binding to negatively charged sites of the AcChoR causing charge reversal which, in turn, reduces the normal K-ion accumulation in the channel vestibules.8 See Figure 20.3 and Table 20.1.

Vex ex

Vin

in

Figure 20.3 Cross section of the Torpedo AcChoR dimer (Mr ⬇580 000) drawn on the basis of cryoelectron microscopic data of the monomer species by Unwin, J. Mol. Biol. 229, 1101–1124 (1993). On one of the monomer fragments (Mr ⬇290 000) a possible location of the three anionic rings elaborated by Imoto et al., Nature 324, 670–674 (1986) is specified. Potential sites for the sialic acid residues (嘷 ) and the additional (hyper-)phosphate groups ( ) as quantified by Kukol and Neumann, Eur. Biophys. J. 27, 618–625 (1998) are introduced. The symbols denote the gegenions accumulated near the anionic groups in the vestibules. In the right-hand side monomer fragment the dimensions of the cylinder model are given: ␾ 2 nm, ᐉex 6.6 nm, ᐉin 2.0 nm with Aex ␲ . ᐉex ␾ 42 nm2 and Ain 13 nm2. See also Table 20.1.

32 5 39 5 33 5 29 5

96 6 84 6

99 6 114 8

Normal Desialyated Hyperphosphorylated Inward current Outward current

3 0.8 2 0.7 3 0.7 3.6 0.7

␣

35 3 40 4

28 2 20 2

␺0 (mV)

0.17 0.03

0.13 0.05 0.094 0.005

␴/(e0 . nm2)

14.3 4.4*

11 8

N(Vex)

18.7

14.2 10.3

Ntot

Note: Data evaluation of the reaction CaR 7 RCa2 where R refers to an anionic Ca2-binding site, KCa is the Ca2-dissociation equlibrium constant and G0/G [K in ]/[Kex] the vestibular accumulation factor. The parameters were obtained with Eqns (2) and (3) of the reference.8 Conditions are the same as those given in Figure 20.2. The surface potential ␺0, the surface charge density ␴ and number of excess anionic groups N(Vex) in the external receptor vestibule of cylinder volume Vex, cylinder internal surface Aex 42 nm2, and N(Vin) in the cytosolic vestibule of Vin and surface Ain 13 nm2, with Vtot Vex Vin, Atot Aex Ain and Ntot N(Vin) N(Vex), all refer to the electrostatic cylinder model; see Figure 20.3. The value 4.4* refers to the cytosolic vestibule, N(Vin).

0.33 0.05 0.27 0.04

0.34 0.05 1.4 0.5

G (pS) KCa(mM)

G0 (pS)

AcChoR dimer

Table 20.1 Limit conductances G0 and G , at [Ca2] 0 and [Ca2] → , of the Torpedo AcChoR dimer and electrostatic model parameters at T 295 K (22 C)

Biosensing and electroporation 517 Artificial enzymatic cleavage of the sialic acid residues (up to 20 2 per dimer) reduces the K-accumulation factor ␣ G0/G from ␣ 3 0.8 of the normal AcChoR to ␣ 2 0.7 for the desialyated AcChoR. Desialysation also decreases the Ca2-sensitivity of the conductance from G0 96.6 6 pS at [Ca2] → 0 of the normal AcChoR to G0 84.2 6 pS. Endogenous hyperphosphorylation (to upto 28 4 phosphates per dimer) enhances the vestibular K-accumulation to ␣ 3.6 0.7, without affecting the Ca2-dissociation equilibrium constant KCa 0.34 0.05 mM at 295 K (22ºC). Most interestingly, even in the absence of AcCho, the hyperphosphorylated AcChoR dimer exhibits spontaneously longlasting open channel events (␶ 200 50 ms). At [AcCho] 2 M there are two open states (␶1 20 10 ms, ␶2 140 60 ms) whereas the normal AcChoR dimer has only one open state (␶ 6 4 ms). Physiologically important is that (i) the sialic acid and phosphate residues render the AcChoR conductance sensitive to control by divalent ions and (ii) the channel behavior of the hyperphosphorylated AcChoR without AcCho appears to indicate pathophysiologically high phosphorylation activity of the cell leading, among others, to myasthenic syndromes. For the sensitivity of the AcChoR as an electrochemical biosensor it is worthwhile to note that at higher protein concentrations the AcChoR forms oligo-channel events with high conductance values.7–9 20.2.3

Specific sensory features of the AcChoR

The various specific channel properties of the AcChoR may be used for the specific identification in the biosensing of the various cholinergic ligands with the potentially different effects on the Ca2/Mg2-sensitivity, mean open times or hyperpolarizations. Another interesting feature of the Torpedo AcChoR is the tendency to desensitize after prolonged exposure to AcCh. The channel pattern changes characteristically during desensitization, in particular the frequency of the current events decreases. For the practical use of AcChoR as a biosensor it is important to remove the bound AcCh from the desensitized receptor states. This can be very efficiently achieved by the presence of the enzyme acetylcholinesterase (AcChoE). Due to the rather high AcCho binding affinity of the receptor, K ≈ 107–109 M,7–9 simple washing-out of the AcCho by buffer solution is relatively slow and may take many minutes. For rapid re-use of the AcChoR sensor, the sensing system should be complemented by a mobile, perhaps polymer coupled esterase. Hence, besides the reference electrode the sensing device should contain a capillary for the supply of a Ca2- or Mg2-salt solution and a mobile AcChoE finger for the regeneration of a renewed activatable AcChoR. The trends are to apply electrically controlled pumping microdevices for the supply and removal of the various diagnostic solutions in the nanoliter range of the interior of the global sensor electrode. Clearly, the trend here is to have the sensor protein membrane as the key part of a multicapillary electrode array. The actual solid-supported membrane-bound

518

E. Neumann and K. Tönsing + Micropumping system

U0 –

Reference electrode

Glass capillary

Protective layer Target surface

Figure 20.4 Scheme for the capillary array of the acetylcholine receptor sensory system. The fluid in which acetylcholine and congeners is to be biosensed, is sucked in the micro measuring compartment, containing the solid supported receptor-membrane, through a thin protection layer of permeable polymer network. Two inner capillaries serve as devices for solution supply and for the reference electrode. A micropump system for sucking in and pushing out of the sampling solutions is part of the automatic sampling device.

sensory element will be of submicrometer dimensions, where the global overall capillary array of now several millimeters diameter is the target of further miniaturization down to submillimeters. Electrochemically, the trend is to indicate the sensing result as patch-clamp current events. During current recording the clamp voltage should be varied in steps such that the conductance can be determined in a reliable way. The concentration range of AcCho to be covered is from 0.1 to 1 mM. The final goal is to manufacture a multicapillary electrode array (Figure 20.4) which is suited to detect released AcCho in small areas such as nerve-muscle synapses. Of course, in order to monitor the neurotransmitter release, the natural esterases have to be blocked by inhibitors. With respect to sensitivity and signal-to-noise ratio, a further trend seems to go in the direction of preparing membranes with high AcChoR content. The resulting oligochannels with high conductance values7,9 guarantee a very high signal-to-noise ratio.

20.3

Capillary electrode array for clinical electroporation

The observation that biological membranes are rendered transiently permeable by short high-voltage pulses (without cell damage) leading to crossmembrane material exchange,10 has opened a new field of membrane biophysics with impressive applications in cell biology, biotechnology and, recently, also in clinical medicine. The electrical pulses are applied in two types of electrode

Biosensing and electroporation 519 configurations: either conventionally with flat condenser plates or with thin needle electrode arrays which are stuck into the solid tissue across the skin.11 The electric field between the needle pins is, however, very inhomogeneous. This, at the present state of the art, poses some practical problems of local tissue damage. These problems must be technically overcome in the future, perhaps by a special needle array. 20.3.1

Membrane electroporation

The membrane electroporation is an electric method to render dielectric membranes porous and permeable, transiently and reversibly, by electric voltage pulses. The concept of membrane electroporation as a structural phenomenon leading, among others, to membrane permeability changes and transport phenomena is depicted in Figure 20.5. Starting from the original applications to

(a)

(b)

+

–

(c)

Figure 20.5 Scheme for the electroporative material exchange; (a) biological cells in a suspension containing, in this example, positively charged effector substances which in the absence of permeabilizing electric pulses cannot enter the cells; (b) membrane electroporation in a rectangular electric field pulse and crossmembrane exchange of substances. Uncharged or dipolar effector substances can enter the interior only by diffusion through the electroporated membrane of both pole caps; (c) after the pulse the exchanged external solution contains the cells with the internalized effector substance.

520

E. Neumann and K. Tönsing

release non-invasively intracellular components10 or to introduce foreign genes into single cells in suspension,1 the new trends nearly all refer to medical–clinical applications. For instance, it appears that membrane electroporation soon will be used to treat human cancer patients: specifically, to deliver various types of genes coding for effector substances such as tumor-therapeutically active interleukines or apoptotically effective proteins such as p53 and p73. Further on, the electroporative chemotherapy, using chemotherapeutica such as bleomycin in very low concentrations, which are otherwise ineffective, has reached an encouraging state of development.12 The results of clinical trials clearly indicate the trends of new applications in human tumor therapy. Physically, electrooptical and conductometrical data of unilamellar liposomes as cell models showed that the electric field causes not only membrane pores but also shape deformations of the liposomes.13,14 It appears that membrane electroporation and shape deformation are strongly coupled and mutually affecting each other.15 The primary field effect of membrane electroporation and cell deformation triggers a cascade of numerous secondary phenomena, such as pore enlargement and transport of small and large molecules across the electroporated membrane. 20.3.1.1

Electropore formation and resealing

As judged from the electrically indicated transport of salt ions, the external voltage pulse must induce transient, yet long-lived membrane pores.14,15 Certainly, the stationary pores kept open by the transmembrane field can only be small ( 1 nm diameter) in order to prevent discharging of the membrane interface by ionic conduction. However, additional lateral tension due to long-lasting electrical Maxwell stress on the vesicle or cell may lead to further pore enlargements. The special structural order of a long-lived, potential permeation site may be modeled by the so-called inverted or hydrophilic (HI) pore (Figure 20.6).16–19 20.3.1.2

Dynamic condenser and interfacial polarization

It is recalled that a lipid membrane is a highly dynamic phase of mobile lipid molecules, hydrophobically held together by the aqueous environment. In the electric field, the redistribution of ions on both sides of the membrane dielectrics is equivalent to electric condenser plates at constant potential. Such a charged condenser with both mobile interior and mobile environment favors the entrance of water molecules to produce pores (P), which may be hydrophobic (HO) and/or hydrophilic (HI) pores, with higher dielectric constant ( ⬇ 80) compared with the lower dielectric constant ( L ⬇ 2) of the replaced lipids (state C, Figure 20.6). An important feature of the great efficiency of membrane electroporation is the enormous amplification of the external field E by Maxwell–Wagner ionic interfacial polarization. The principle is depicted in Figure 20.7(a) for a spherical membrane and the electric potential drop ␸ is schematically shown for a biological cell in Figure 20.7(b). In detail, the induced membrane field is given by

Biosensing and electroporation 521

C

E

P

Em

HO

HI

Figure 20.6 Model for the chemical state transitions C 7 P and C 7 HO 7 HI; molecular rearrangements of the lipids in the pore edges of the lipid vesicle membrane. C denotes the closed bilayer state and P refers to porated states. The external electric field causes ionic interfacial polarization of the membrane dielectrics which is analogous to condenser plates (, ). Em Eind is the membrane field induced by the external field E, leading to water entrance in the membrane to produce pores (P); cylindrical hydrophobic (HO) pores or inverted hydrophilic (HI) pores. In the pore edge of the HI pore states the lipid molecules are turned to minimize the hydrophobic contact with water. Note that in the open ionic condenser the ion density adjacent to the aqueous pore (␧W) is larger than in the interface lipid (␧L)/medium due to ␧W >> ␧L.

Eind ␸ind/d 1.5 (a/d)Ef (␭m) |cos ␪| with a the cell radius, d the membrane thickness, f (␭m) 1 the conductivity factor and ␪ the positional angle (Figure 20.7).18–19 The field amplification factor 1.5a/d is especially significant for large objects with thin dielectric membranes. For typical values such as a 10 m and d 5 nm, we have a field amplification of (3a/2d) 3 . 103. For elongated cells like bacteria aligned by the field in the direction of E, the contribution of Eind at the pole caps, where |cos ␪| 1, amounts to Eind (L/2d)E, where the amplification factor (L/2d) is proportional to the bacterium length L.18 20.3.2

Thermodynamics of membrane electroporation

Electrochemically pore formation and resealing as modeled in Figure 20.6, is seen as a state transition of the lipids in the pore edge. Classical physical chemistry covers the processes in terms of equilibrium (K) and rate coefficients (ki). For the reaction scheme C 7 P, the degree fp of membrane electroporation is given by fp [P]/([P] [C]) K/(1 K), with K [P]/[C] k1/k1. For liposomes k1 is in the order of 106 s1 and is strongly field-dependent. Further details are documented elsewhere.17–19 The global equilibrium constant K of the poration–resealing process, C 7 P, is directly related to the standard value of the transformed reaction Gibbs energy O O ˆ— ˆ— by K exp(rG /RT),20 where RT is the molar thermal energy. The rG O O ˆ O– (P) G ˆ— ˆ— G (C), between the two states molar work potential difference rG

522 E. Neumann and K. Tönsing

E

(a)

Em

+

+ + + +

– –

+ – + –

– –

+ – + –

a

r –

d

(b)

∆m (180°) ∆m (0°)

X

Figure 20.7 Interfacial membrane polarization of a cell of radius a. (a) Cross section of a spherical membrane in the external field E; (b) the profile of the electrical potential ␸ across the solution and the cell membranes of thickness d in E as a function of distance x; note that ␸m (␪ 180) ␸ind ␸s ␸nat and ␸m (␪ 0) ␸ind ␸s ␸nat, where ␸ind is the drop in the induced membrane potential in the direction of E; ␸s ␸ins ␸out s the surface potential difference at zero external field and ␸nat the natural (diffusion) potential difference at zero external field, also called resting potential. Here, ␸ nat > 0 is shown.

C and P in the presence of an electric field generally comprises chemical and physical terms.17 The dominant contribution of the field effect is the electric ˆ pol 兰r M dEm, where the electric reaction moment polarization term rG rM Mm(P) Mm(C) refers to the difference in the molar dipole moments Mm of state C and P, respectively. The field-induced reaction moment in the electrochemical model is given by:20 rM NAVprP, where VP ␲ · r2p · d is the average (induced) pore volume of the assumed cylindrical pore. Inspired by the physical analysis by Abidor et al.16 we define the chemical reaction polarization as:21 r P ␧0(␧W ␧L)Em. The difference ␧W ␧L in the dielectric constants of water and of lipids, respectively, refers to the replacement of lipids by water at constant Em. Since ␧W >> ␧L, the formation of aqueous pores is strongly favored in the presence of a cross-membrane potential difference ␸m, in particular at

Biosensing and electroporation 523 high E, when the contribution of ␸ind in ␸m is large. See the Figure 20.7. Explicitly, at the angle ␪, and in the boundaries Em 0 and Em, r p . NA . f 2(␭m) . cos2 ␪ . E2/d.417,21 If the 兰r M dEm (9/8) . ␲ . ␧0 . a2 . (␧W ␧L) . 2 relation between K and Em (3a/2d) f(␭m) . E|cos␪| can be formulated as K K0 exp[兰r M dE/(RT)], where K0 refers to E 0, the mean pore radius rp can be cal– culated in a simple way from the field dependence of K or of f p (the average degree of poration). Typically, at ␸ind 0.42 V and pulse duration tE 10 s, the mean pore radius amounts to rp 0.35 0.05 nm.21 20.3.3

Electroporative transport of ionic macromolecules

It was found that the electrochemical pore model is not only suited to describe pore formation and resealing but also to cover shape deformations and salt ion efflux from salt-filled liposomes.13,14 The transport kinetics of larger macromolecules such as drugs and DNA is more complicated than the transport of small salt ions and occurs in several temporally distinct stages. Obviously, the transport is greatly facilitated if there is at first adsorption of the macromolecules to the membrane surface.18,22 For charged macromolecules, adsorption is followed by electrophoretic penetration into the surface of electroporated membrane patches. Further steps are the after-field diffusion, dissociation from the internal membrane surface and, finally, binding with cell components in the cell interior.22,23 It is instructive to compare the field strength dependence of the average degree – – ( f p) of electroporation with that ( f perm) of a secondary, transport process such as – salt ion efflux (see Figure 20.8), where f p !I/!0. In conclusion, the theory of membrane electroporation in terms of a chemical phase transition (C 7 P) coupled to shape deformation and transport processes has been developed to such a degree that analytical expressions are available for the optimization of the electroporation techniques in biotechnology and medicine, in particular, in the new fields of electroporative drug delivery and gene therapy. The new trends here are electroporative gene vaccinations which is certainly a great challenge for modern medicine. 20.3.4

Electroporation electrode arrays

The design of improved electrodes for voltage pulse applications and, in the case of needle electrodes, for the electroporative delivery of drugs and genes, is also guided by the progress in electroporative transport theory. A new development is a concentric pin array around a center needle electrode serving also as the capillary for the supply of effector substances to compacted cell aggregates and tissue (see Figure 20.9). For analytical and control purposes, recording of conductivity or impedance during and after the pulse period with simple instrumentation has become a technical challenge for common use in cell biology as well as for clinical trials in electroporative tumor therapy and gene therapy. The ultimate goal is the provision of

4.10–3

0.06 ∆I 0 0.03

fp 2.10–3

0

0

5 E (MV m–1)

10

0

Figure 20.8 The average fraction f¯p of the electroporated membrane area, (■) at a large NaCl concentration difference (in the vesicle interior [NaCl]in 0.2 M), in the medium [NaCl]out 0.2 mM, osmotically balanced with 0.284 M sucrose), (▲) at equal concentrations ([NaCl]in [NaCl]out 0.2 mM), smoothly increases with the field strength E, whereas the massive conductivity increase ␭I/␭0, (●) of the suspension of the salt filled vesicles of radius a 160 30 nm (␭0 7.5 S cm1, T 293 K (20C),19) indicates an apparent threshold value Ethr 7 MV m1. The ratio f¯p S(tE)/Sm, where S(tE) Np(tE) · ␲ · r2p and Sm is the total vesicle surface, was calculated from the electrooptic relaxations, yielding characteristic rate parameters of the electroporation – resealing cycle in its coupling to ion transport.

Delivery of effector substances

+ – Pulse-generator

Center needle electrode Skin tumor

1 mm

Figure 20.9 Scheme of an electroporation needle electrode array to be used in skin tumor therapy.

Biosensing and electroporation 525 an automatic clinical electroporator machine as standard device for electroporative treatments of patients.

Acknowledgment We gratefully acknowledge the careful processing of the manuscript by M. Hofer and financial support of the Ministerium für Schule und Weiterbildung, Wissenschaft und Forschung des Landes Nordrhein-Westfalen, Projekt: FV ELMINOS IV.4, and the Deutsche Forschungsgemeinschaft, DFG grant Ne 227/9 to E.N.

References 1. E. Neumann, M. Schaefer-Ridder, Y. Wang and P. H. Hofschneider, EMBO J. 1, 841–845 (1982). 2. O. P. Hamil, A. Marty, E. Neher, B. Sakmann and F. J. Sigworth, Plügers Arch. 391, 85–100 (1981). 3. E. Sackmann, Applications. Science 271, 43–48 (1996). 4. H. Lang, C. Duschl and H. Vogel, Langmuir 10, 197–210 (1994). 5. R. Naumann, A. Jonczyk, C. Hampel, H. Ringsdorf, W. Knoll, N. Bunjes and P. Gräber, Bioelectrochem. Bioenerg. 42, 241–247 (1997). 6. E. K. Schmidt, T. Liebermann, M. Kreiter, A. Jonczyk, R. Naumann, A. Offenhäuser, E. Neumann, A. Kukol, A. Maelicke and W. Knoll, Biosensors & Bioelectronics 13, 585–591 (1998). 7. E. Neumann, J. Weber and T. Schürholz, Archives Physiol. Biochem. 104, 731–744 (1996). 8. A. Kukol and E. Neumann, Eur. Biophys. J. 27, 618–625 (1998). 9. T. Schürholz and J. Weber, E. Neumann, Bioelectrochem. Bioenerg. 21, 71–81 (1989). 10. E. Neumann and K. Rosenheck, J. Membrane Biol. 10, 279–290 (1972). 11. L. M. Mir, S. Orlowski, J. Belehradek, Jr., J. Teissie, M. P. Rols, G. Sersa, D. Miklavcic, R. Gilbert and R. Heller, Bioelectrochem. Bioenerg. 38, 203–207 (1995). 12. R. Heller, M. Jaroszeski, L. Glass, J. Messina, D. Rapaport, R. CeConti, N. Fenske, R. Gilbert, L. M. Mir and D. Reintgen, Cancer 77, 964–971 (1996). 13. S. Kakorin, E. Redeker and E. Neumann, Eur. Biophys. J. 27, 43–53 (1998). 14. S. Kakorin and E. Neumann, Ber. Bunsenges. Phys. Chem. 102, 670–675 (1998). 15. E. Neumann and S. Kakorin, Radio. Oncol. 32, 7–17 (1998). 16. I. G. Abidor, V. B. Arakelyan, L. V. Chernomordik, Y. A. Chizmadzhev, V. P. Pastuchenko and M. R. Tarasevich, Bioelectrochem. Bioenerg. 6, 37–52 (1979). 17. E. Neumann and S. Kakorin, Curr. Opin. Colloid. Interface Sci. 1, 790–799 (1996). 18. E. Neumann, Bioelectrochem. Bioenerg. 28, 247–267 (1992). 19. E. Neumann, S. Kakorin, K. Toensing, Bioelectrochem. Bioenerg. 48, 3–16 (1999). 20. E. Neumann, Electroporation and Electrofusion in Cell Biology (Plenum Press, New York, 1989), pp. 61–82. 21. S. Kakorin, S. P. Stoylov and E. Neumann, Biophys. Chem. 58, 109–116 (1996). 22. E. Neumann, S. Kakorin, I. Tsoneva, B. Nikolova and T. Tomov, Biophys. J. 71, 868–977 (1996). 23. K. Toensing, S. Kakorin, E. Neumann, P. Budde and J. Frey, Biophys. J. 74, 98–108 (1998).

21 Multi-barrelled ion-selective microelectrodes Measurements of cell volume, membrane potential, and intracellular ion concentrations in invertebrate nerve cells Paul Wilhelm Dierkes, Susanne Neumann, Anja Müller, Dorothee Günzel, and Wolf-Rüdiger Schlue 21.1

Introduction

In the last decades, ion-selective microelectrodes have often been used in living cells to measure the free concentrations of various ions together with the membrane potential.1,2 In addition to neutral carrier-based microelectrodes for the measurement of K, Na, H, Mg2, and Ca2, classical ion exchanger microelectrodes are available for the investigation of K and Cl. Moreover, ion-selective microelectrodes are applicable for monitoring cell volume by recording the intracellular concentration of an ion species that serves as a volume marker.3 If the investigated cells are sufficiently large, multi-barrelled microelectrodes with a tip diameter less than 1 m, can be employed, which allows for the simultaneous recording of up to three different ion species.4,5 This potentiometric approach appears to be helpful for a detailed description of cell volume maintenance and regulation, because changes in cell volume can be correlated directly with changes in the intracellular ion concentrations, which might support conclusions concerning the underlying ionic mechanisms. The plasma membranes of animal cells are highly permeable to water. Since the movement of water across the cell membranes is primarily determined by the osmotic gradient between the intracellular and extracellular medium, any disturbance of the osmotic balance should result in the uptake or loss of water and hence a change in cell volume.6 Furthermore, cell volume changes might be caused by perturbations in the extracellular osmolarity or ionic composition, by energy depletion, as well as by stimulation or inhibition of ion transport systems in the cell membrane.6–8 Because cell shrinkage or swelling might impair normal cell function, many cells have regulatory mechanisms to keep their volume constant.6,8,9 Volume regulatory mechanisms may be either acute, being activated immediately after the onset of the osmotic perturbation, or delayed. Acute volume regulation is often mediated by transport systems for inorganic electrolytes such as K, Cl, and Na, while delayed regulation involves the production or degradation of larger membrane-impermeable osmolytes inside the cells.

Cell volume measurements 527 A decrease in cell volume can be counteracted by a Na influx and an increase by a K efflux, each coupled to a Cl flux to maintain electroneutrality. Most transport mechanisms involved in acute volume regulation seem to be activated only if the cell volume departs from the steady-state level by more than a critical amount.8 The transport systems involved in acute cell volume regulation may be modulated by alterations in the intracellular concentrations of H, Mg2, or Ca2.6,10,11

21.2 21.2.1

Ion-selective microelectrodes Construction of multi-barrelled ion-selective microelectrodes

Multi-barrelled ion-selective microelectrodes consist of glass capillaries with two, three, or four separate channels that are pulled out to form a single fine tip (see Figure 21.1). The outer diameter of the whole electrode tip is less than 1 m, which is small enough to be inserted into single cells.5 The tips of the prospective ion-selective channels are filled with so-called liquid membranes, which consist of an organic electrolyte or neutral ligand dissolved in a hydrophobic solvent. In principle, the organic electrolyte or neutral ligand selectively binds a single ion species, which leads to a potential difference across the liquid membrane that is determined by the concentration of the respective ion at two sides of the membrane. Because glass surfaces are highly hydrophilic, water would displace the liquid membrane from the tip of an untreated microelectrode. Therefore, the glass surface is made hydrophobic by coating it with organic groups, a process which is called silanization.12 After the tips have been filled with the liquid membranes, the ion-selective barrels are backfilled with appropriate electrolyte solutions (see Table 21.1). The reference channel used for recording the membrane potential is not silanized. This channel is filled with an electrolyte solution, the composition of which is such as to minimize interference with the ion-selective channels. The electrical potentials of all channels are measured against the potential of an extracellular reference electrode, using potentiometers with an input resistance of 1015 . The intracellular ion concentrations are calculated from the so-called ionic potentials (EIon), that is, the potential differences between the ion-selective barrels and the reference barrel. A detailed description of the construction of multi-barrelled ion-selective microelectrodes and suitable experimental set-ups is found elsewhere.1,13,14 21.2.2

Calibration procedure and microelectrode characteristics

The selectivity of ion-selective microelectrodes is limited. As a consequence, microelectrodes selective for K, Cl, Na, or Mg2 do not behave linearly in the physiological range, and therefore these electrodes must be calibrated with a series of calibration solutions. The influence of possibly interfering ions is minimized by using solutions with an ionic background that simulates the intracellular conditions. In contrast, H-selective electrodes behave linearly over a wide pH

528

Dierkes et al. (a)

(b)

(c)

(d)

1

5

3 4 2

Figure 21.1 Schematic drawing of double-barrelled (a,b), triple-barrelled (c), fourbarrelled and (d) ion-selective microelectrodes. Multi-barrelled ion-selective microelectrodes were pulled from borosilicate glass capillaries (1). The doublebarrelled microelectrodes were manufactured either from a theta-type capillary (Ø 2 mm, (a) or two single-barrelled capillaries (Ø 1.5 mm, (b). The triple-barrelled microelectrodes (c) were constructed from one theta-type capillary and a single-barrelled capillary and the four-barrelled microelectrodes (d) from two theta-type capillaries. The tips of the ion-selective barrels of the microelectrodes were filled with the liquid ion-sensor membrane (2) and then backfilled with appropriate electrolyte solutions (3) (see Table 21.1). The reference barrel for recording the membrane potential was also filled with a suitable electrolyte solution (4). All barrels were connected to the potentiometers via chlorinated silver wires (5).

range and must be calibrated only at two different pH values within the physiological range. In the experiments described in this chapter, electrode calibration was performed before and after every experiment. The calibration curves were constructed by plotting EIon against the negative logarithm of the ion concentration (pIon; see Figure 21.2). Since H-selective microelectrodes show a linear relation between EH and pH, the calibration values were fitted by means of the Nernst equation. Microelectrodes for K, Cl, Na, and Mg2 give a Nernst response only at relatively high ion concentrations.13 At lower concentrations, the calibration curves become progressively flatter, and eventually the electrode potential is

K-selective barrel (1) Neutral carrier valinomycin15 (2) Classical ion exchanger Corning477317 (potassium tetrakis (p-chlorophenylborate))12,15 H -selective barrel Neutral carrier ETH1907 (4-nonadecyl-pyridine)16 Mg2-selective barrel Neutral carrier ETH5214 (N,N"-octamethylene-bis(N -heptyl-N -methylmalonamide))17 Na-selective barrel Neutral carrier ETH227 (N,N ,N"-triheptyl-N,N ,N"trimethyl-4,4 ,4"-propylidintris-(3-oxabutyramide))18 Cl-selective barrel 5,10,15,20-tetraphenyl-21H,23H-porphin manganese(III) chloride19

Ionophores in the ion-selective barrel

0.7 mM 0.8 mM — 0.07 mM

1.1 mM

0.16 mM

53 mV (pH 6.2–7.7) 31 mV (0.5–10 mM) 48 mV (2.5–50 mM)

49 mV (2.5–50 mM)

Detection limit

49 mV (10–100 mM) 51 mV (10–100 mM)

Slope per decade (concentration range)

Table 21.1 Characteristics of multi-barrelled ion-selective microelectrodes

100 mM KCl

100 mM NaCl

100 mM MgCl2

Calibration solution pH 7.67

100 mM KCl 100 mM KCl

Backfill

Triple-barrelled

Four-barrelled

Four-barrelled

Four-barrelled

Triple-barrelled Triple-barrelled

Electrode type

530

Dierkes et al. 300 ETPA (cie)

250 ETMA (cie)

200

EIon (mV)

150 ECl

100 50 0

EK (cie) EK (val) EMg

–50 –100

EH

ENa –150

1

2

3

4

5

6

7

8

plon

Figure 21.2 Calibration curves. The ion potentials of various microelectrodes were plotted against the negative logarithm of the respective ion concentration (pIon). The ion potential measured for the respective electrode in standard leech saline (NaCl: 85 mM, KCl: 4 mM, CaCl2: 2 mM, MgCl2: 1 mM, HEPES (N-[2-hydroxyethyl]piperazine-N -[2-ethansulfonic acid]): 10 mM, pH 7.4 adjusted with NaOH) was set to 0. For the recording of EK either the cie, Corning477317, or valinomycin (val) were used. ETMA and ETPA were recorded by means of the cie. For the ion sensors used to record EMg, ENa, EH, or ECl, as well as for the backfill solutions see Table 21.1. The continuous lines were calculated by the use of the Nicolsky–Eisenman equation; in case of EH the Nernst equation was used. Broken lines illustrate the determination of the detection limit (arrow) of an ion-selective microelectrode. Each data set summarizes 8–22 microelectrode calibrations.

no longer affected by the ion concentration. The relation between the electrode potential and pIon can be described by the Nikolsky–Eisenman equation. The detection limit of an ion-selective microelectrode is defined by the intersection of the two asymptotes given by the Nikolsky–Eisenman equation,13,20 which describes the relation between the electrode potential and pIon in the presence of interfering ions (see Figure 21.2). The experimentally determined detection limit is affected mainly by the presence of interfering ions, but also by the membrane resistance and the tip diameter of the electrode channels.21 The characteristics of the multi-barrelled ion-selective microelectrodes used in this study are summarized in Table 21.1. Slope and detection limit of triple- and four-barrelled microelectrodes did not deviate significantly from the values obtained with double-barrelled microelectrodes. K-selective microelectrodes

Cell volume measurements 531 (a)

(b)

200 Elon (cie)

10 min

[TMA+] (mM) 5

150 Elon (cie) (mV)

150 E (mV)

200

100 50 Eref

0

4 0

5

50

0.3

0

100 50 25

1

100

0 4 (mM) 0 (mM)

[K+]

–50

0 0

1

[TMA+]

2

3

4

5

6

pK

Figure 21.3 K sensitivity of the classical ion-exchanger in the presence of TMA. (a) Original recording with a double-barrelled ion-selective microelectrode. Varying the K concentration in the range from 0 to 100 mM in the presence of 5 mM TMA had virtually no effect on the ion potential (EIon(cie); Eref reference potential); (b) K calibration curves in the presence of different TMA concentrations. Without TMA or with a TMA background of 0.3 mM, EIon(cie) showed a clear sensitivity to K. In the presence of TMA concentrations of 1 mM or more, EIon(cie) was virtually independent of K.

based on the classical ion-exchanger (cie) are less sensitive to K than to quaternary ammonium ions, which in this study were used as volume markers.3,22 Calibrations for tetramethylammonium (TMA) and tetrapropylammonium (TPA) were carried out with a background of 80 mM K, which is close to the intracellular K concentration ([K]i). The mean slope was 63 mV per decade for TMA (concentration range 0.5–25 mM) and 71 mV for TPA (0.1–2.5 mM), and the detection limits were 0.16 or 0.004 mM, respectively. The data in Figure 21.3 show that the K sensitivity of these electrodes was virtually abolished if the TMA concentration was 1 mM or more.

21.3

Application of ion-selective microelectrodes in biological tissues

Ion-selective microelectrodes are well suited for the direct, local, and continuous measurement of ion concentrations in vitro and in vivo under diverse experimental conditions. The measurements can be carried out within different compartments of biological tissues, such as extracellular spaces, cytosol, or even in organelles. Furthermore, with multi-barrelled ion-selective microelectrodes the concentration of several ion species can be monitored simultaneously, provided that the cells are sufficiently large and robust enough to endure electrode penetration. 21.3.1

Leech ganglia

The central nervous system of the leech (Hirudo medicinalis L.) is a particularly suitable preparation for monitoring ion movements between the different

532

Dierkes et al.

compartments within a nervous system. As in every central nervous system, one has to distinguish roughly between a neuronal, a glial, and an extracellular compartment, which are all accessible to ion-selective microelectrodes. Investigations in identified neurones and glial cells with conventional and ion-selective microelectrodes revealed a variety of ion transport systems in the plasma membranes (see Refs 1, 23, 24). The expression of these systems is different in glial cells and neurones, and there are also differences between the diverse types of neurones as well. The probably best characterized cells in the leech central nervous system are the large Retzius neurones (about 80 m in diameter), which tolerate long-term penetration by ion-selective microelectrodes with no discernible damage. Since some of the ion transport systems in leech Retzius neurones are likely to be involved in volume regulation (see Refs 6, 8, 9), these cells appear to be well suited for an investigation of the mechanisms of cell volume homoeostasis and regulation. 21.3.1.1

Loading of leech nerve cells with volume markers

There are various techniques for loading cells with volume markers, the applicability of which depends on the biological tissue under investigation.3,25–27 Most simply, the volume markers are applied via the bathing solution and taken up by the cells, a procedure that has been successfully used in leech glial cells which possess an uptake mechanism for quaternary ammonium ions such as choline or TMA.26 The effect of applying TPA or TMA via the bathing solution in leech Retzius neurones is shown in Figure 21.4(a). In this experiment, a triple-barrelled ion-selective microelectrode was inserted into a Retzius neurone, the two ionselective barrels being filled with the neutral carrier valinomycin and with the classical ion exchanger, respectively. Extracellular TPA application induced a small increase in the ion potential measured with the classical ion exchanger (Ecie), which is due to the uptake of TPA by the cell, because the ion potential measured with valinomycin (Eval) was unaffected indicating that [K]i remained constant. In comparison, bath application of TMA induced a much larger increase in Ecie which reflects a marked uptake of TMA, since again Eval was only slightly altered. The results suggest that Retzius neurones have a specific uptake system that discriminates between TMA and TPA. The membrane depolarization induced by TMA was probably due to the activation of cation channels coupled to TMA-sensitive nicotinic acetylcholine receptors,23,28 compare.29 In comparison to leech glial cells, TMA uptake by Retzius neurones was rather slow leading to a maximum increase in the intracellular TMA concentrations ([TMA]i) of 0.5 mM/min. TMA loading was performed over 10–20 min in order to achieve [TMA]i between 2 and 10 mM that were used in the experiments on cell volume regulation. Other leech neurones, such as anterior pagoda (AP), pressure (P), and noxious (N) neurones, showed similar uptake rates as Retzius neurones. After TMA washout, [TMA]i decreased at rates 5 to 10-fold smaller than the uptake rates. Since TPA was poorly taken up by the cells, the substance was iontophoretically injected via a single-barrelled microelectrode filled with the chloride salt of

Cell volume measurements 533

Em (mV)

130 120 110 100 90 80 70 60 50

5 min

–20

2 min

–30 – 40 300 Ecie

250 ∆ [TMA ]i = 2.1 mM +

∆ [TPA+]i = 0.008 mM

Ecie

Elon (mV)

Em (mV)

(b) –40 –50 –60 –70

Elon (mV)

(a)

200 150

∆ [TPA+]i = 0.07 mM

100 Eval 5 mM TPA+

5 mM TMA+

Eval

50 + 1nA

Figure 21.4 Loading Retzius neurones with the volume markers TMA and TPA. Intracellular recordings from two different cells with triple-barrelled microelectrodes. The ion-selective barrels were filled with the cie or with valinomycin. (a) Bath application of 5 mM TPA induced a small [TPA]i increase, as reflected by the slight increase in Ecie. A much larger Ecie increase was induced by applying 5 mM TMA, which indicates a marked increase in [TMA]i. The effects on Ecie were not caused by changes in [K]i, because Eval was not or only slightly changed. In the presence of TMA, Em depolarized reversibly, probably due to the activation of acetylcholine receptor-coupled cation channels. (b) In this experiment, a single-barrelled microelectrode filled with the chloride salt of TPA (100 mM) was used for the iontophoretic injection of TPA. Application of a depolarizing current induced a marked increase in Ecie, whereas Eval was not affected. After the end of current injection (1 nA for 30 s), Ecie increased further and then partially recovered. Finally, a virtually stable plateau was reached corresponding to a [TPA]i of 0.07 mM. The delay between the stop of the current injection and the maximum TPA increase probably reflects the time required for TPA diffusion from the injection electrode to the ion-selective electrode.

TPA (Figure 21.4(b)). The effects of the injection on Em, Ecie, and Eval were simultaneously monitored by means of a triple-barrelled ion-selective microelectrode. A positive injection current induced a membrane depolarization and a large increase in Ecie due to an increase in the intracellular TPA concentration ([TPA]i), because Eval was not or only slightly influenced. Ecie reached its maximum about 20 s after the end of current injection, which might be due to the TPA diffusion from the injection electrode to the ion-selective electrode. Subsequently, Ecie partially recovered, probably due to TPA diffusion into the fine axonal branches, and then reached a plateau. In comparison to TMA, the loss of TPA was markedly smaller, and [TPA]i was virtually constant during an experiment. By recording [TMA]i or [TPA]i relative changes in volume can be monitored.3,26 Cell shrinkage induces an increase in [TMA]i or [TPA]i, and cell

534

Dierkes et al.

swelling a decrease. Provided that the cell membrane is impermeable to the volume marker, the relative cell volume is given by the following equation:26 Vrel 100 ·

[Ion]ctr. [Ion]exp.

(1)

where Vrel relative cell volume in (%) and [Ion] [TMA]i or [TPA]i before (ctr. control) and during experimental (exp.) alterations. This equation can be directly applied in the case of TPA, because its loss was negligibly small in the course of an experiment (Figure 21.4(b)). In case of TMA, however, a significant loss occurred during the experiment. The TMA loss was calculated by assuming an exponential time course and was used for the correction of the TMA signal (Figure 21.5(a)). This procedure gave reliable results, because cell volume changes measured either with TMA or TPA did not significantly deviate from one another (Table 21.2). The finding that the TMA loss followed an exponential time course suggests a homogeneous distribution of the volume marker in the cytoplasm; if TMA had been taken up by intracellular organelles, a more complicated time course would be expected. 21.3.1.2

Cell volume changes induced by changing the osmolarity of the extracellular solution

The osmotic behaviour of leech neurones was investigated by measuring cell volume changes resulting from exposure to anisosmotic extracellular solutions. If the cells behave like perfect osmometers the following equation should be valid: Vt ␲0 V0 ␲t

(2)

where V0, Vt cell volume and ␲0, ␲t extracellular osmotic pressure before (time 0), and during (time t) experimental alterations. To determine to what extent leech neurones behave like ideal osmometers, the osmolarity of the bathing solution was changed by elevating or reducing the extracellular NaCl concentration. The data in Table 21.2 show that in general the observed volume changes were smaller than would be expected from Eqn (2). When the changes in osmolarity were moderate the differences were not significant, but significant differences were observed when the osmolarity was altered more strongly. Thus, leech neurones behave like ideal osmometers only within a rather limited range of extracellular osmolarities. Similar results have been obtained previously in snail neurones and murine neuroblastoma cells.25,30 The effect of raising the extracellular osmolarity on cell volume, [K]i, and [Cl]i is shown in Figure 21.5. The three parameters markedly changed within 1–2 min after raising the osmolarity and then remained constant, which suggests that during the 5-min exposure to the hyperosmotic solution no volume regulation occurred. Similar results were obtained in other preparations, in which volume regulatory processes were observed only during long-lasting exposure to

Cell volume measurements 535

5 min

–40

[TPA+]i (mM)

5 4

–40

3 2 1.6 1.4 1.2 1.0

0.20 0.16 0.12 1.6 1.4 1.2 1.0 120

Vrel (%)

120 100 80 60

100 80 60

+85 mM NaCl

–40 –50

5 min

–60 –70

0.24

Relative [TPA+]i increase

Relative [TMA+]i increase [TMA+]i (mM)

5 min

–50

–50

Vrel (%)

–30

Em (mM)

–30

(c) –20

Relative Relative [Cl–]i (mM) [K+] increase [K+]i (mM) [Cl–]i increase i

–20

Em (mM)

(b) Em (mM)

(a)

+85 mM NaCl

130 110 90 2.2 1.8 1.4 1.0 15 10 5 2.2 1.8 1.4 1.0 +85 mM NaCl

Figure 21.5 Effect of increasing the extracellular osmolarity by adding 85 mM NaCl to the physiological solution on Em, [TMA]i, [TPA]i, [Cl]i, and [K]i. Recordings from three different Retzius neurones. (a,b) Effect on Em and [TMA]i or [TPA]i recorded with double-barrelled ion-selective microelectrodes filled with the cie. Before raising the osmolarity of the bathing solution, the neurones were loaded with the volume markers as illustrated in Figure 21.4. Upon exposure to the hyperosmotic solution, [TMA]i rapidly increased to a constant level, indicating a stable decrease in cell volume. After return to physiological solution, [TMA]i and hence the cell volume rapidly recovered (a). The drift in the [TMA]i recording, which reflects the loss of TMA during the experiment, was fitted by a simple exponential function (dotted line). The fit parameters were used to correct the relative [TMA]i changes for the TMA loss. In contrast to TMA, the loss of TPA was negligibly small (b). The effects on [TPA]i induced by the hyperosmotic bathing solution were comparable to those on [TMA]i. The relative cell volume was calculated according to Eqn (1). (c) Effect on Em, [Cl]i, and [K]i recorded with a triple-barrelled ion-selective microelectrode. For the recording of EK valinomycin and for ECl 5,10,15,20-tetraphenyl-21H,23H-porphin manganese(III) chloride were used. The increase in bath osmolarity induced an increase in [K]i which was consistent with the drop in cell volume (see (a,b)). The change in [Cl]i was much larger, which indicates a marked uptake of Cl.

anisosmotic solutions.25,30 The relative change in [K]i corresponded closely to the change in cell volume (~35%), while the changes in [Na]i and [Cl]i were distinctly larger (~80%; Figure 21.5(c)). This observation indicates an uptake of NaCl from the external medium and hence an increase in the intracellular

536

Dierkes et al.

Table 21.2 Changes in cell volume of leech neurones measured with ion-selective microelectrodes Experimental condition

␲o /␲t (bath osmolarity in mOsm)

Vt /Vo with TMA as a volume marker

Vt /Vo with TPA as a volume marker

Physiological solution 85 mM NaCl 40 mM NaCl 20 mM NaCl 20 mM NaCl 40 mM NaCl 59 mM NaCl

1 (191) 0.55 (347) 0.73 (261) 0.86 (223) 1.20 (159) 1.68 (114) 2.42 (79)

1 0.64 0.07 (n 8)** 0.80 0.04 (n 4)* 0.91 0.07 (n 9) 1.15 0.14 (n 9) 1.32 0.06 (n 3)** 1.67 0.26 (n 4)**

1 0.66 0.07 (n 6)* n.d. n.d. n.d. n.d. n.d.

Note n.d. not determined, asterisks: data that differ significantly from an ideal osmotic behaviour (*p 0.05, **p 0.01).

(a)

(b)

50 µm

Figure 21.6 Morphometric determination of cell volume changes in Retzius neurones. The neurones were marked by iontophoretic injection of a fluorescent dye (Fura-2) and fluorescence micrographs were taken upon superfusion with a hyperosmotic ((a): 85 mM NaCl) or a hyposmotic bathing solution ((b): 59 mM NaCl). The area of the cross section of the cell was determined planimetrically and cell volume was calculated under the assumption of a spherical cell shape.

osmolyte content. Similarly, in hyposmotic solution, the relative changes in [K]i, [Na]i, and [Cl]i (not shown) were significantly larger than the changes in cell volume, indicating a loss of intracellular KCl or NaCl and hence a decrease in the intracellular osmolyte content. The experiments show that the cells change their intracellular osmolyte content under anisosmotic conditions, thereby reducing the osmotic gradient between the intracellular and extracellular medium, which also explains the non-ideal osmotic behaviour of the cells.

Cell volume measurements 537 The cell volume changes detected by ion-selective microelectrodes were verified by morphometric measurements. For this purpose, Retzius neurones were marked by iontophoretic injection of fluorescent dyes (e.g. Fura-2) and fluorescence micrographs were taken under the same experimental conditions as in the potentiometric measurements. The area of the cross section of the cells in the different solutions was determined planimetrically and the cell volume was estimated under the assumption of a spherical cell shape (Figure 21.6). In hyperosmotic solution obtained by adding 85 mM NaCl, the cells shrank on average to 72 9% (n 9) of their volume in physiological solution; in hyposmotic solution obtained by omitting 59 mM NaCl, the cells swelled to 157 31% (n 11). 21.3.1.3

Changes in cell volume under isosmotic conditions

[K+]i (mM) Em (mV)

3.0 2.5 2.0 1.5 1.0

Vrel (%)

–20

[TMA+]i (mM)

Also under isosmotic conditions the cell volume may be affected by various factors.6 For instance, in the central nervous system of vertebrates, an isosmotic elevation in the extracellular K concentration ([K]o) has been shown to induce cell swelling.26,31,32 In Retzius neurones, the elevation of [K]o from 4 to 30 mM 5 min

–40 –60 90 70 50

150 140 130 120 110 100 90 30 mM K+

Figure 21.7 Effect of increasing the extracellular K concentration on Em and the relative cell volume in a Retzius neurone. Intracellular recording with a triplebarrelled microelectrode filled with valinomycin for the recording of [K]i and the cie for the recording of [TMA]i. Raising the extracellular K concentration from 4 to 30 mM (Na replaced by K) reversibly induced a marked membrane depolarization and a decrease in [TMA]i reflecting cell swelling, while [K]i remained unaffected. The cell was TMA loaded via the bathing solution as illustrated in Figure 21.4.

5 7.1 7.3 7.5

75 65

25 15

85

25 20 15

–10 Em (mV)

0.3

–40 –50 –60

(c) –20

[Na+]i (mM)

0.4

–20 –30

40

Vrel (%)

–40 –50 –60 0.5

(b) 10 min Em (mV)

–20 –30

K (mM)

pHi [Na]i (mM) [Mg2+]i (mM)

Em (mV)

(a)

Dierkes et al.

[Cl–]i (mM)

538

–30 –40 –50

20 0 120 110 100 90

100 µm Kainate

Figure 21.8 Effect of the glutamatergic agonist kainate on various intracellular ion concentrations and cell volume in Retzius neurones. Intracellular recordings from of kainate-induced changes in Em, [Mg2]i, [Na]i, [K]i, [Cl]i, and pHi measured with a four-barrelled microelectrode (a) or triple-barrelled microelectrodes (b,c). Application of 100 M kainate induced a marked membrane depolarization, an increase in [Mg2]i, [Cl]i, and [Na]i, a decrease in [K]i as well as an intracellular acidification. Furthermore, an increase in cell volume was recorded by the use of TMA as a volume marker. Recordings from three different cells. TMA loading performed via the bathing solution (see Figure 21.4). For the ionophores used to detect the various ions see Table 21.1.

(Na replaced by K) caused a cell swelling by 16–40%, as reflected by the decrease in [TMA]i (Figure 21.7). [K]i and [Na]i increased only slightly by 3–13%, whereas [Cl]i increased by 130%. Considering the increase in cell volume, the cells markedly increased their osmolyte content, primarily by the uptake of KCl and to a lesser extent by the uptake of NaCl. If the cell volume would have remained constant, the osmolarity of the cytosol had increased by about 50 mOsm. The resulting change in the osmotic gradient between the intracellular and extracellular medium seems to be responsible for the induction of cell swelling. In the central nervous system of vertebrates, glutamate acts as the major excitatory neurotransmitter, mediating fast synaptic transmission by activating ligand-gated cation channels of the AMPA/kainate type.33 Long-term exposure to relatively high levels of extracellular glutamate has been shown to cause cell swelling, which might be involved in the cell death often observed under pathophysiological conditions. Also in the leech central nervous system glutamate acts as the major excitatory neurotransmitter.24,34 In Retzius neurones, application of the glutamatergic agonist kainate (100 M) induced a marked membrane depolarization, an increase in [Mg2]i, [Cl]i, and [Na]i, a decrease in [K]i as well as an intracellular acidification (Figure 21.8 24,34). Furthermore, an increase in

Cell volume measurements 539 cell volume by maximally 13 10% (n 5) occurred. Considering the increase in cell volume, we calculate that the cellular content of K was only slightly altered. In contrast, the content of Na and Cl increased by 80–100%, which would raise the osmolarity of the cytosol by about 25 mOsm, if the cell volume remained unchanged. These results indicate that the kainate-induced cell swelling was mainly due to the uptake of NaCl.

21.4

Conclusions

In summary, the results described in this chapter illustrate that the simultaneous recording of cell volume, membrane potential, and intracellular ion concentrations with multi-barrelled ion-selective microelectrodes is applicable in leech neurones. The cell volume of leech neurones was altered by changing bath osmolarity or the extracellular ionic composition as well as by activating ligand-gated ion channels. The finding that leech neurones change their intracellular osmolyte content in response to anisosmotic solutions indicates possible pathways by which the cells may minimize volume changes caused by a perturbation. In the central nervous system, an increase in network activity is often accompanied by an increase in [K]o and elevated extracellular levels of neurotransmitter, both of which induce cell swelling and may disturb normal cellular function. Since a variety of effects on cell volume described for vertebrate preparations also occur in the leech central nervous system, this preparation may be useful for a more detailed description of cell volume maintenance and regulation, because changes in cell volume can be correlated directly with changes in the intracellular ion concentrations measured by ion-selective microelectrodes.

References 1. W.-R. Schlue, W. Kilb, and D. Günzel, Electrochim. Acta 42, 3197 (1997). 2. R. C. Thomas, Ion-sensitive Intracellular Microelectrodes (Academic Press, London, New York, San Francisco, 1978). 3. L. Reuss, Proc. Natl. Acad. Sci. USA 82, 6014 (1985). 4. D. Günzel, S. Durry, and W.-R. Schlue, Pflügers Arch. 435, 65 (1997). 5. D. Günzel, A. Müller, S. Durry, and W.-R. Schlue, Electrochim. Acta 44, 3785 (1999). 6. F. Lang, G. L. Busch, M. Ritter, H. Völkl, S. Waldegger, E. Gulbins, and D. Häussinger, Physiol. Rev. 78, 247 (1998). 7. M. L. McManus and K. B. Churchwell, Clinical Significance of Cellular Osmoregulation (in: K. Strange, Cellular and Molecular Physiology of Cell Volume Regulation, 1994), pp. 63–77. 8. W. C. O’Neill, Am. J. Physiol. 276, C995 (1999). 9. B. Sarkadi and J. C. Parker, Biochim. Biophys. Acta 1071, 407 (1991). 10. J. K. Foskett and J. E. Melvin, Science 244, 1582 (1989). 11. B. Nilius, J. Prenen, and G. Droogmans, Pflügers Arch. 436, 742 (1998). 12. W. Walker, Anal. Chem. 43, 89A (1971). 13. D. Ammann, Ionselective Microelectrodes, Springer-Verlag, Berlin, Heidelberg (1986).

540

Dierkes et al.

14. D. Günzel and W.-R. Schlue, J. Physiol. 491, 595 (1996). 15. M. Oehme and W. Simon, Anal. Chim. Acta 86, 21 (1976). 16. P. Chao, D. Ammann, U. Oesch, W. Simon, and F. Lang, Pflügers Arch. 411, 216 (1988). 17. Z. Hu, T. Bührer, M. Müller, B. Rusterholz, M. Rouilly, and W. Simon, Anal. Chem. 61, 574 (1989). 18. R. A. Steiner, M. Oehme, D. Amman, and W. Simon, Anal. Chem. 51, 351 (1979). 19. Y. Kondo, T. Bührer, K. Seiler, E. Frömter, and W. Simon, Pflügers Arch. 414, 663 (1989). 20. G. G. Guibault, R. A. Durst, M. S. Frant, H. Freiser, E. H. Hansen, T. S. Light, E. Pungor, G. Rechnitz, N. M. Rice, T. J. Rohn, W. Simon, and J. D. R. Thomas, Pure Appl. Chem. 48, 127 (1976). 21. U. E. Spichiger and A. Fakler, Electrochim. Acta 42, 3137 (1997). 22. E. Neher and H. D. Lux, J. Gen. Physiol. 61, 385 (1973). 23. K. Ballanyi and W.-R. Schlue, Glia 2, 330–345 (1989). 24. P. W. Dierkes, P. Hochstrate, and W.-R. Schlue, J. Neurophysiol. 75, 2312 (1996). 25. F. J. Alvarez-Leefmans, S. M. Gamino, and L. Reuss, J. Physiol. 458, 603 (1992). 26. K. Ballanyi, P. Grafe, G. Serve, and W.-R. Schlue, Glia 3, 151 (1990). 27. G. Serve, W. Endres, and P. Grafe, Pflügers Arch. 411, 410 (1988). 28. G. N. Woodruff, R. J. Walker, and L. C. Newton, Comp. Gen. Pharmacol. 2, 101 (1971). 29. A. Bigiani and M. Pellegrino, J. Exp. Biol. 151, 423 (1990). 30. J. Altamirano, M. S. Brodwick, and F. J. Alvarez-Leefmans, J. Gen. Physiol. 112, 145 (1998). 31. N. R. Kreisman and J. C. LaManna, J. Neurophysiol. 82, 320 (1999). 32. W. Walz, Can. J. Physiol. Pharmacol. 65, 1051 (1987). 33. D. T. Monaghan, R. J. Bridges, and C. W. Cotman, Annu. Rev. Pharmacol. Toxicol. 29, 365 (1989). 34. R. Dörner, K. Ballanyi, and W.-R. Schlue, Brain Res. 523, 111 (1990).

22 Nerve cells and lipid vesicles on silicon chips – considerations on ionoelectronic sensors Peter Fromherz

22.1

Introduction

Biological sensors are made of membrane proteins. An extracellular chemical signal is translated – directly or mediated by an intracellular second messenger – into a conformational change that opens an ion channel across the membrane. It is suggestive to develop synthetic ionoelectronic sensors where the specificity of protein receptors is combined with the universality of microelectronics, more precisely where the ion current through a membrane is detected directly by a microelectronic device on a silicon chip. A suitable tool for the noninvasive detection of membrane currents is an open field-effect transistor (FET) with a membrane attached to its exposed gate.1–20 A diagram is shown in Figure 22.1. When a chemical signal opens ion channels in the membrane of an attached cell or lipid vesicle, ionic and capacitive current flows through the attached membrane. The current passes along the narrow cleft between membrane and chip to the bath kept at ground potential. It gives rise to a voltage drop in the cleft. The voltage change on the gate of the transistor affects the electrons flowing in the silicon from source to drain. The ionoelectronic coupling is achieved without charge transport across the interface, that is, without electrochemical reactions. An ionoelectronic sensor based on a membrane–transistor junction relies on three elements: (i) Ion channels that are opened directly or indirectly by a chemical signal. (ii) A membrane that is in close contact to the surface with a high resistance of the cleft. (iii) A thermodynamic force that drives current through the membrane. We may envisage three approaches to build such a device: (i) The attachment of an intact living cell with natural or genetically transfected receptors. (ii) The attachment of a membrane fragment or a vesicle taken from a natural or transfected cell. (iii) The attachment of an artificial lipid membrane or vesicle with reconstituted receptor molecules. This chapter reviews several elementary studies on the properties of cell– silicon and membrane–silicon junctions that may be useful for the future design of ionoelectronic sensors on the basis of a membrane–transistor junction. It is restricted to the approaches chosen in our own laboratory and is written as a guide through our results published at different places. First, the basic physics of the system is considered in terms of the planar core-coat conductor. Then crucial

542

P. Fromherz

Cell/vesicle

D

S

Chip

Figure 22.1 Cartoon of an ionoelectronic sensor with a cell or a vesicle on an open FET. A thin cleft of electrolyte separates the attached membrane from the silicon dioxide of the silicon chip. Note: the width of the cleft is 1–100 nm at a diameter of the cell or vesicle of 10–100 m. A chemical signal in the solution opens receptor channels in the free and in the attached membrane. Ionic current flows through the free and attached membrane, driven by a suitable thermodynamic force. The resulting superposition of ionic and capacitive current through the attached membrane flows along the narrow cleft and gives rise there to a voltage drop. The change of extracellular voltage in the cleft plays the role of a gate voltage for the open FET and modulates the electronic current from source (S) to drain (D) in the silicon chip.

parameters are determined, the distance of membrane and chip, the conductances in the junction and the distribution of ion channels. Finally, the role of action potentials is discussed as amplifiers of receptor signals.

22.2

Planar core-coat conductor

When a cell or a lipid vesicle is attached to oxidized silicon, the membrane and the solid are separated by a cleft of electrolyte as illustrated in Figure 22.1. As we shall see, the width of the cleft is 1–100 nm at a diameter of 10–100 m of a cell or a vesicle. Thus, a sandwich structure is formed of silicon, silicon dioxide, cleft, membrane and cytoplasm. This extended layered system is a planar electrical

Neurochip and sensorics 543 core-coat conductor: the conductive cleft is insulated by the thin films of silicon dioxide and membrane from the conductive phases of silicon and cytoplasm, respectively.5,6,19 Such a planar core-coat conductor is a homologous to a cylindrical electrical cable. The electrical properties of the distributed electrical system are determined by local charge conservation. They are described by a twodimensional partial differential equation, which is homologous to Kelvin’s cable equation.5,6,19 In this section a simple equivalent circuit is considered – the pointcontact model – which accounts for the crucial features of the extended core-coat conductor after appropriate parameterization.1,4,6,11 On this basis a fundamental coupling relation for ionoelectronic sensors is derived. 22.2.1

Point-contact model

The equivalent circuit of the planar core-coat conductor in a cell–silicon junction is shown in Figure 22.2. Capacitances are assigned to the membrane and to the oxide in the junction. One or several ion conductances in the attached membrane are driven by Nernst potentials. The extended cleft is represented by an ohmic conductance. The free part of the cell or vesicle is described by a capacitance and one or several ion conductances, too. The ionic and capacitive currents in the circuit determine the intracellular voltage VM and the extracellular voltage VJ in the cleft. The voltage VJ in the junction controls the transistor. It plays the same role as the gate-voltage on the metalized gate of a common MOS-FET. 22.2.2

Coupling relation

We apply Kirchhoff’s law to the node in the junction of Figure 22.2, and obtain Eqn (1) for the voltage VJ, with the membrane capacitance per unit area cM, the ion conductance giJM per unit area of the membrane in the junction, the reversal voltage V0i and a cleft conductance gJ per unit area of the junction. The approximation of Eqn (1) is valid for weak coupling, that is, for small values of VJ and dVJ/dt at a modest electrode capacitance cOX i gJVJ gJM (VM V0i) cM

dVM dt

(1)

The properties of the planar core-coat conductor are “squeezed” into the cleft conductance gJ per unit area of the junction according to Eqn (2) with the distance dJ of membrane and substrate, with the specific resistance ␳J of the electrolyte in the cleft and with the radius aJ of a circular junction.6,8 5␲ dJ 1 gJ ␳ J ␲ a2

(2)

J

We may eliminate the capacitive current in Eqn (1) by taking into account Kirchhoff’s law for the intracellular node of Figure 22.2.1 The capacitive current through the total membrane is balanced by the total ion current through the free

544

P. Fromherz

IINJ

Vm

VJ S

D

B

Figure 22.2 Equivalent circuit of the planar core-coat conductor in a cell–transistor junction (point-contact model). The width of the cleft between membrane and chip is blown up. Source (S), drain (D) and bulk silicon (B) are kept at bias voltages with respect to the bath. The intracellular voltage VM and the extracellular voltage VJ depend on the capacitances of the attached and free membrane, on one or several ion conductances – driven by Nernst-type batteries – on the stray capacitance of the chip and on the conductance of the cleft. A substance in the bath which opens ion channels affects conductances in the attached and free membrane. The resulting change of the voltage VJ is detected by the transistor. An injection current IINJ can be applied by an impaled or fused micropipette.

and the attached areas of the membrane according to Eqn (3), with the specific ion conductances giFM and giJM in the two regions and with the ratio ␤ of the areas of attached and free membrane. Again the approximation of Eqn (3) is valid for weak coupling, that is, for small values of VJ and dVJ/dt (1 ␤) cM

dVM (giFM ␤giJM) (VM V0i) dt

(3)

Inserting Eqn (3) into Eqn (1), taking into account Eqn (2), we obtain the coupling relation for an ionoelectronic sensor according to Eqn (4) VJ

i giFM ␳Ja2J gJM (VM V0i) 5dJ 1 ␤

(4)

The relation shows that a large signal on the gate requires: i a small distance dJ of membrane and substrate; ii a large radius aJ of the contact; iii an enhanced or depleted conductance of the receptor channels in the attached membrane with giJM giFM ⴝ 0; and iv an electrochemical driving force VM V0i.

Neurochip and sensorics 545

22.3

Distance of membrane and chip

In this section we consider the distance of membranes and silicon dioxide on silicon chips. First, the method of fluorescence interference contrast (FLIC) microscopy21–25 is summarized, then two measurements are described for an attached lipid vesicle and for a neuronal cell.

22.3.1

FLIC-microscopy

FLIC-microscopy takes advantage of the standing modes of light in front of the reflecting surface of silicon.21 The membrane of a cell or vesicle is stained with a fluorescent dye. It is attached to a silicon chip with microscopic terraces of silicon dioxide.22 The fluorescence intensity depends on the distance of the membrane from the silicon surface, given by the height dOX of the terraces and the width dJ of the cleft. The standing modes of light affect the probability per unit time Pex(dOX, dJ) for the excitation of a dye molecule illuminated in a microscope, the probability per unit time Pem(dOX, dJ) for the detection of an emitted photon in a microscope, and the total decay probability per unit time kfl(dOX, dJ) of the fluorescent state due to electromagnetic interactions. The three functions are computed by classical electrodynamics taking into account the optics of a microscope and an optical model for the junction with its five layers – bulk silicon, silicon dioxide, extracellular cleft, membrane and bulk electrolyte.21–23 The experimental intensities on the terraces of height dOX are fitted according to Eqn (5) – where knr is the rate constant of nonradiative deactivation – with three free parameters, the distance dJ between membrane and oxide, a scaling factor a and a constant background b.21,25 Jfl(dOX, dJ) a

22.3.2

Pex(dOX, dJ)·Pem(dOX, dJ) b kfl(dOX, dJ) knr

(5)

Lipid membrane and cell membrane

Fluorescence micrographs of a giant lipid vesicle (palmitoyl-oleoyl-phosphatidylcholine : dioleoyl-phosphatidyl-methyester 2 : 1) and of an astrocyte from the hippocampus of rat brain are shown in Figure 22.3. We used a chip that was covered with 2.5 m 2.5 m oxide terraces. There were sixteen terraces of different height in a unit cell. Vesicles were sedimented on a chip coated with polylysine,15,26 astrocytes were cultured on a chip coated with laminin.23 Lipid bilayer and plasma membrane were stained with the lipoid cyanine dye DiI.21 In both systems the fluorescence intensity was modulated in register with the oxide terraces. The patterns, however, were quite different. For a lipid vesicle the fluorescence was low on the lowest terrace (no. 1) as expected for close adhesion: the electromagnetic field in the visible range has almost a node on the surface of silicon. Quite in contrast the fluorescence of an astrocyte was high on the lowest terrace no. 1. This observation indicates immediately that a wide cleft separates membrane and oxide in cell adhesion. The fluorescence intensities are plotted vs the

546

P. Fromherz (a)

(b)

Figure 22.3 FLIC microscopy on a silicon chip with terraces of silicon dioxide. A unit cell of the chip surface consists of 4 4 quadratic oxide terraces – each 2.5 m 2.5 m – with a height of 30–320 nm. Unit cells are sketched as inserts with the position of the steps (order of increasing height). (a) Giant lipid vesicle of palmitoyl-oleoyl-phosphatidyl-choline and dioleoyl-phosphatidylmethylester (molar ratio 2 : 1) on a chip coated with poly-lysine. The fluorescence on the thinnest oxide no. 1 is low; (b) astrocyte from rat brain on a chip coated with laminin. The fluorescence on the thinnest oxide no. 1 is high. Lipid membrane and cell were stained with the cyanine dye DiI.

height of the terraces in Figure 22.4. By fitting Eqn (5) we obtained a distance between membrane and oxide dJ 1.1 nm of for the lipid membrane and dJ 109 nm for the astrocyte, in the average of many measurements.15,23 The close adhesion of the vesicle is due to electrostatic attraction between the negatively charged membrane and the oxide coated with a positively charged polyelectrolyte. Without electrostatic attraction, the distance of lipid membranes is enhanced to about 50 nm due to the repulsion by membrane undulations.26 The large distance of the cell is caused by steric repulsion of the filamenteous laminin molecules.23,26 We never achieved distances closer than 40 nm for cells – such as astrocytes, neurons, fibroblasts and HEK 293 cells – whatever coating we applied.

Neurochip and sensorics 547

Fluorescence (au)

1

Vesicle Astrocyte 0

0

100 200 Oxide thickness (nm)

300

Figure 22.4 Fluorescence intensity of the attached membrane vs the height of the oxide terraces for a negatively charged lipid vesicle on poly-lysine (dots) and for an astrocyte on laminin (open circles). The experimental data are fitted by an electromagnetic theory.

We may estimate the cleft conductance gJ per unit area according to Eqn (2), assuming that the specific resistance in the cleft is ␳J 100 cm as in a typical physiological electrolyte. For a lipid vesicle with aJ 20 M and dJ 1 nm we obtain gJ 1.25 mS/cm2. For a neuron or a HEK 293 cell with aJ 10 m and dJ 40 nm the conductance is gJ 2000 mS/cm2. Apparently, cells have a serious drawback compared to vesicles with respect to ionoelectronic coupling, due to their smaller diameter and their larger separation.

22.4

Conductance of junction

The conductance of the cleft between membrane and chip and the conductance of the attached membrane can be determined by ac-techniques. An ac-voltage is applied to the cell, to the bath or to the chip, and the response of the extracellular voltage VJ in the cleft is measured in amplitude and phase. The complex transfer function of voltage is evaluated in terms of the circuit of Figure 22.2, or in terms of the complete two-dimensional Kelvin-type equation for the planar core-coat conductor. Here two examples are considered, one with intracellular stimulation4–6 and one with bath stimulation.14,15,19 The responses in the junction were observed with transistors. In the case of chip stimulation, the response can be detected with a fluorescent voltage-sensitive dye in the attached and free membrane.27 22.4.1

Intracellular ac-stimulation

An individual nerve cell from the leech was attached to a transistor coated with poly-lysine as shown in Figure 22.5. The membrane was fused with a micropipette

548

P. Fromherz

100 µm

Figure 22.5 Retzius neurons from leech attached to a silicon chip with a 4 4 array of FETs. Scale bar 100 m. An open gate is visible in the right lower corner with the drain on the left and the common source for all transistors on the right. The cell membrane of the neuron in the center is fused with a patch-pipette (from above) to apply intracellular ac-voltage. A second microelectrode (from the right) is impaled to determine the actual intracellular voltage.

(whole-cell patch-clamp) to apply an intracellular ac-voltage VM( f )of frequency f. In some cases the actual intracellular voltage was checked with an impaled microelectrode,4 in some cases we corrected the voltage-drop in the pipette on the basis of additional impedance measurements.6 The ac-response of the source–drain current was recorded, and the extracellular ac-voltage VJ( f ) on the gate was evaluated in amplitude and phase. We observed two types of neuron–silicon junctions5,6 as shown in Figure 22.6: (i) In an A-type junction the amplitude |h| |VJ/VM| of the signal transfer was weak at low frequencies. It increased above 1000 Hz. The phase ␸h increased to 90 around 10–100 Hz and decreases above 1000 Hz. (ii) In a B-type junction the coupling was efficient at low frequencies with vanishing phase shift. The amplitude increases, the phase increases and decreases around 1000 Hz. Considering the circuit of Figure 22.2, it is apparent that in a B-junction the conductance of the attached membrane must be large, similar the conductance of the cleft, whereas in an A-junction the coupling is dominated by the capacitive current through the attached membrane.6

Neurochip and sensorics 549 (a)

1.0

Amplitude B

|h| 0.5 A 0.0 1

(b)

90

10

Phase

100 f (Hz)

1000

100 f (Hz)

1000

A

h 45

B 0 1

10

Figure 22.6 Complex transfer function h VJ/VM of ac-voltage from a Retzius neuron to a transistor versus frequency f, with (a) amplitude |h| and (b) phase ␸h. The squares refer to an A-type junction with a low membrane conductance, the circles refer to a B-type junction with a high conductance of the atttached membrane.

We fitted the data of Figure 22.6 by the spectral transfer function of the pointcontact model.5,6 The amplitude of h VJ( f ) / VM( f ) can be written as a superposition of the limits h0 and h at low and high frequency according to Eqn (6a) with ␻ 2␲f.6 The transition is determined by a characteristic time constant ␶J of the junction. The increase of the phase is governed by the time constant of the attached membrane ␶JM, the decrease by ␶J according to Eqn (6b). gJM is the total ionic conductance per unit area of attached membrane, cOX is the capacitance per unit area of oxide 兩h兩2

(␻␶J)2 2 1 2 h h 0 1 (␻␶J)2 1 (␻␶J)2

tan ␸h h0

1 1

(␻␶JM)

␶ 1␶ 冣冢 ␻␶ J

J

JM

gJM cM cOX cM , ¬ h , ¬ ␶J gJM gJ cM cOX gJM gJ

(6a) (6b)

550

P. Fromherz

The data of the A-junction in Figure 22.6 were fitted with ␶JM 14 ms and ␶J 25 s, whereas the time constants fitted to the data of the B-junction were ␶JM 130 s and ␶J 66 s. Assuming specific capacitances cM 5 F/cm2 and cOX 0.3 F/cm2 for membrane and oxide, we obtained for the A-junction a membrane conductance gJM ⬇0.4 mS/cm2 and a cleft conductance gJ ⬇200 mS/cm2, for the B-junction gJM ⬇40 mS/cm2 and gJ ⬇40 mS/cm2.6 A cleft conductance in the order of gJ 100 mS/cm2 is compatible with a cleft of radius aJ 15 m and width dJ 50 m which is filled with electrolyte of ␳J 100 cm according to Eqn (2). Most interesting is the variability of the membrane conductance by two orders of magnitude. We found that the transition from low to high conductance could be induced reversibly by a mechanical deformation of the cell.9 It is likely that the effect relies on a reversible opening and closing of ion channels. The increase of membrane conductance is correlated with a decrease of cleft conductance.6,9 A causal relation is suggested by the mechanism of selfgated ion channels.28 Whatever the explanation may be, the observation indicates that a substrate can affect seriously the electrical properties of an attached membrane. 22.4.2

Extracellular ac-stimulation

As a second example we consider the extracellular ac-stimulation of a vesicle– silicon junction by a modulation VE( f ) of the bath potential.14,15,19 Negatively charged giant vesicles were attached to a linear array of FETs coated with polylysine.15 An example is shown in Figure 22.7. The voltage profiles VJ(x, f ) along the junction were observed by evaluating the source–drain currents of all transistors beneath the vesicle in amplitude and phase.19 The spatial profile of the amplitude of the transfer function h(x, f ) VJ(x, f )/VE( f ) is shown in Figure 22.8(a).15 At low frequency the whole cleft follows the bath potential. The capacitive current through the free and attached membrane becomes effective at higher frequencies – first in the center of the junction. It leads to a hammock-like profile in a transition region around 2 Hz. At very high frequencies the capacitances of membrane and oxide determine the current flow. An evaluation of the data requires the complete theory of the planar core-coat conductor.19 An approximate solution for the transfer function h(a, f ) VJ(a, f )/VE( f ) along the radial coordinate a of a circular junction of radius aJ is given by Eqn (7) with the modified Bessel function I0, the reciprocal electrotonic length constant ␥ of the core-coat conductor and the effective admittance yJM per unit area of attached membrane – assuming that the electrical properties of free and attached membrane are identical

冤

冥

VJ(a, ␻) I0( a) yJM I0( a) 1 VE(␻) I0( aJ) yJM i␻ cOX I0( aJ) gJM i␻cM ␳J 2 ( yJM i␻cOX), ¬ yJM 1␤ dJ

(7)

Neurochip and sensorics 551

10 µm

Figure 22.7 Giant lipid vesicle attached to a linear array of 96 transistors. Scale bar 10 m. The drains are directed downwards, the gates are between the “noses” of local field in the center of the picture. The vesicle is made of palmitoyl-oleoyl-phosphatidyl-choline and dioleoyl-phosphatidylmethylester (molar ratio 2 : 1) and is attached to the silicon dioxide with poly-lysine.

We obtained a good agreement of experiment and theory with a very low membrane conductance gJM 1 S/cm2 and a very high sheet resistance of the cleft rJ ␳J/dJ 130 G as illustrated in Figure 22.8(b), at aJ 25 m, cM 0.6 F/cm2, cOX 0.35 F/cm2 and ␤ 0.7. In the average of many measurements the sheet resistance was rJ 100 G at a membrane conductance gJM 1 S/cm2. The electrical quality of the supported bilayer of an attached lipid vesicle is similar to a free standing black-lipid membrane. Using Eqn (2) we calculate the specific cleft conductance of the point-contact model gJ 8 S/cm2. With dJ 1.1 nm the specific resistance in the cleft is ␳J 11 k cm. This value is by almost two orders of magnitude higher than the resistance of the bulk electrolyte. It seems that the close distance of membrane and oxide lowers the number of mobile ionic charge carriers. The experiment reveals the fascinating features of supported lipid vesicles for the design of ionoelectronic sensors, with respect to the conductance of the cleft and of the lipid matrix. There are two advantages of extracellular ac-stimulation: (i) The cell is not damaged by the pipette. (ii) High frequencies are not shunted by the pipette such

552

P. Fromherz

(a)

(b)

1.0

1.0

0.8 0.6 0.4 0.1

0.2

1

0.8 0.6

|h |

0.4 0.1

1000

–20

–10

10 0 ) m µ x(

z)

100

20

f (H

z)

f (H

10

0.2

1

0.0

|h |

0.0

10 100 1000

–20

–10

0

10

20

m) a (µ

Figure 22.8 Spatial profiles of the amplitude |h| |VJ/VE| of voltage transfer from bulk electrolyte to the junction of a lipid vesicle vs frequency. (a) experimental profiles along the position x of the transistors; (b) theoretical profiles along the radius a of a circular junction.

that even junctions with weak coupling – with a short time constant ␶J – can be studied. However, a profile of voltage-transfer has to be evaluated to determine the conductance of membrane and cleft,14,19 whereas recording with a single transistor and evaluation with the point-contact model are sufficient for intracellular ac-stimulation.

22.5

Distribution of channels

An ionoelectronic sensor requires an inhomogeneous distribution of the ion channels that contribute to the signal, according to the coupling relation of Eqn (4). We checked whether such an accumulation or depletion of ion channels actually exists in the region of adhesion, using cultured nerve cells from the hippocampus of rat brain. First, we consider the implications of a voltage-clamp experiment with transistors,1 then we describe the results concerning voltage-gated potassium channels in rat neurons.13 22.5.1

Patch-clamp and transistor

We keep a cell at constant intracellular voltage by appropriate current injection through a pipette as indicated in Figure 22.2. The injection current IINJ reflects the average ion conductance gMi per unit area of the cell membrane according to Eqn (8a). Here, only a single conductance is taken into account. The extracellular voltage VJ in the junction – as obtained from the response of a transistor – probes the conductance giJM in the attached membrane according to Eqn (8b). IINJ /AM gMi(VM V0i)

(8a)

gJVJ

(8b)

giJM(VM

V0i)

Neurochip and sensorics 553 To evaluate the ion conductances from IINJ and VJ, we need the total area AM of the membrane and the cleft conductance gJ per unit area, respectively. These calibration factors can be obtained from ac-measurements (impedance and voltagetransfer), assuming a certain value for the membrane capacitance cM per unit area.6,12,13

22.5.2

Potassium channels in hippocampal neurons

Nerve cells from the hippocampus of rat brain were cultured on a chip as illustrated by Figure 22.9.13 A cell on a transistor was selected and contacted with a micropipette fused to the plasma membrane12 (whole-cell configuration of patchclamp). A special voltage-protocol was applied to the cell to activate potassium conductances as indicated in Figure 22.10(a). The sodium channels were blocked by TTX. The average potassium conductance gKM of the membrane was obtained from the injection current according to Eqn (8a). Two components can be distinguished in Figure 22.10(b): a transient signal (A-type) and a stationary signal (K-type). The local potassium conductance gKJM in the junction is drawn in Figure 22.10(c), as obtained from the transistor record according to Eqn (8b). There is no transient A-type conductance in the junction. Thus, the A-type component of the average conductance has to be assigned exclusively to the free membrane: there is a depletion of fast inactivating potassium channels in the area of adhesion with a ratio gKJM/gKFM 0. On the other hand, Figure 22.10(c) shows that the stationary K-type conductance is enhanced in the junction as compared

10 µm

Figure 22.9 Nerve cells from the hippocampus of rat brain cultured in defined medium on a linear array of transistors. Scanning electronmicrograph. Scale bar 10 m.

554

P. Fromherz +20

(a) –50

(c)

1 mS/cm2

(b)

1 mS/cm2

–110

50 ms

50 ms

Figure 22.10 Voltage-clamp experiment with hippocampal neurons on transistor. (a) Protocol of intracellular voltage (in mV); (b) average potassium conductance gKM per unit area of the cell membrane, as obtained from the pipette current; (c) local potassium conductance gKJM per unit area of the attached cell membrane, as given by the transistor record (average of 19 records).

to the average membrane conductance of Figure 22.10(b) with a ratio gKJM/ gKM 5. Assuming a ratio ␤ 0.1 of the area of attached and free membrane, we estimate a factor gKJM/gKFM 12 for the accumulation for the slowly inactivating potassium channels. The experiment with rat neuron shows that ion channels in the region of adhesion can be active. In particular it demonstrates that accumulation and as well as depletion of ion channels in the region of adhesion actually exists. Whether the functional properties of the channels in the attached and free membrane are identical, could not be checked. The problem with the experiment was, that the transistor signals were rather weak at conductances gKJM 艐 1 mS/cm2 and gJ ⬇1000 mS/cm2 of membrane and cleft. The extracellular voltage could be measured only by extensive signal averaging due to the noise of the transistors. The stability of the system was not sufficient to determine the complete voltagedependent gating. Recently we transfected HEK 293 cells with the human slow-poke potassium channel. We obtained transistor recordings under voltage-clamp without signal averaging. Preliminary data indicate that the functional properties of the channel (voltage-gating, calcium activation) are fully conserved in the area of adhesion.18

22.6

Action potentials

When a chemical signal is applied to a solution, the response of an ionoelectronic sensor with a cell or a vesicle on a chip may be slow due to diffusion. It may be difficult to distinguish the actual response from drift effects that originate in the cell and vesicle, in the junction or in the transistor. One feasibility to overcome

Neurochip and sensorics 555 the problem is the transformation of the slow intracellular receptor potential into an action potential or a train of action potentials that is recorded by the transistor. Three approaches may be considered: (i) Natural cells are used that are excitable and express receptor channels. (ii) Excitable cells are transfected with receptors. (iii) A cell is transfected with receptors and with sodium and potassium channels that give rise to an action potential. Here, we consider some problems with the extracellular recording of action potentials1,11,13 and describe transistor recordings of neurons from the leech.11 22.6.1

Extracellular recording

The waveform of the extracelluar voltage VJ(t) in response to an action potential differs from the waveform of the intracellular voltage VM(t) as a function of time t. We obtain the relation of the two transients on the basis of the pointcontact model shown in Figure 22.2, taking into account membrane conductances for sodium and potassium and a leak conductance. We obtain from Kirchhoff’s law for the two nodes in the junction and in the cell the approximations of Eqns (9) and (10) with i Na, K, leak – generalizations of Eqns (1) and (3) – that hold for weak coupling with small values of VJ and dVJ/dt and modest capacitance cOX1 gJVJ 兺giJM (VM V0i ) cM i

(1 ␤)cM

dVM dt

dVM 兺(giFM ␤ giJM) (VM V0i) dt i

(9) (10)

Elimination of the capacitive current by insertion of Eqn (10) into Eqn (9) leads to Eqn (11), in analogy to Eqn (4) gJVJ 兺(giJM giFM) (VM V0i )

(11)

i

Extracellular recording is possible only, if the cell membrane is inhomogeneous with respect to the conductances involved in the generation of an action potential, that is, with different conductances per unit area for sodium, potassium or for leak current in the attached and free membrane.1,11,12 We consider three limiting cases: (i) Complete depletion of ion conductances in the junction area. (ii) Enhancement of the leak conductance in the junction. (ii) Accumulation of the sodium and potassium channels in the attached membrane. A-type record. When the ion conductance of the junction is negligible, Eqn (9) is reduced to Eqn (12) and the extracellular voltage resembles the first derivative of the intracellular voltage. The intracellular voltage is driven by the ion currents through the free membrane cM dVM VJ g J dt

(12)

556

P. Fromherz

B-type record. When the leak conductance is enhanced in the junction at a homogeneous distribution of sodium and potassium channels, Eqn (11) is reduced to Eqn (13), where the leak conductance in the free membrane is neglected. The extracellular voltage reflects the shape of the intracellular voltage, which is controlled by the sodium and potassium currents gleak JM VJ g VM J

(13)

C-type record. When the sodium and potassium conductances are enhanced or depleted in the junction by the same factor ␮J, we obtain Eqn (14) by inserting Na K K the relations gNa JM ␮J gFM and gJM ␮J gFM into Eqns (9) and (10), neglecting the leak conductance VJ

1 ␮J cM dVM 1 ␤␮J gJ dt

(14)

With ␮J 0 we obtain the case of complete depletion where the extracellular voltage is proportional to the first derivative of the intracellular voltage. For accumulation, however, with ␮J 1, the sign on the right-hand side of Eqn (14) is inverted, the extracellular voltage resembles the inverted first derivative of the intracellular voltage. 22.6.2

Neuron-transistor

We recorded action potentials from nerve cells of leech,3,7,9,11,17 snail16,20 and rat.12 Figure 22.11(a) shows a leech neuron with a transistor beneath the cell body.9 An action potential was elicited by current injection through an impaled pipette. We obtained the transistor record shown in Figure 22.12(a).9 The extracellular voltage resembled the first derivative of the action potential. The experiment corresponds to the A-type record of Eqn (12). The attached membrane is depleted in ion conductances. When the neuron was pressed onto the chip by lowering the pipette, the record switched to the waveform shown in Figure 22.12(b).9 The extracellular voltage resembled the action potential itself. The experiment corresponds to the B-type record of Eqn (13). There is an enhanced leak conductance in the attached membrane. The A-type and B-type records correspond to the transfer spectra of A-type and B-type in Figure 22.6. Figure 22.11(b) shows a leech neuron with an array of transistors beneath the axon stump that protrude from the cell body.11 The transistor record is shown in Figure 22.12(c).11 The extracellular voltage resembles the inverted first derivative of the action potential. The experiment corresponds to the C-type record of Eqn (14). Sodium and potassium channels are accumulated in the junction area. Sometimes more complicated records were observed with neurons from leech, snail and rat. In those cases a selective depletion and accumulation of sodium and potassium channels has to be taken into account.11,12,16

(a)

(b)

50 µm

50 µm

Figure 22.11 Retzius neurons from leech on transistors. (a) A cell body is attached to a single transistor. It is impaled by a microelectrode to elicit an action potential by current injection; (b) the axon stump of a cell is attached to an array of transistors. The picture was taken before impaling the microelectrode. Scale bars 50 m.

(b)

VM (mV)

(a)

(c)

20

0

0

–20

0

–20 –20

–40 – 40

–40 –60 –60

– 60

–80 –80 2

0.4

4

VJ (mV)

1

0

2

0.2

0

0.0

–2

–0.2

–1 –4 –2

0

5

10

15

Time (ms)

20

–0.4 0

5

10

15

Time (ms)

20

0

5

10

15

20

Time (ms)

Figure 22.12 Transistor records of action potentials in Retzius neurons. Upper row: intracellular voltage. Lower row: extracellular voltage on the gate of a transistor. (a) Cell body on transistor with A-type coupling; (b) cell body on transistor with B-type coupling; (c) axon stump on transistor with C-type coupling.

558

P. Fromherz

Whether receptor potentials are strong enough to elicit action potentials in the configuration of a cell-transistor has to be investigated yet. In any case the design of a quantitative ionoelectronic sensor on this basis may be difficult due to the non-trivial relation of intracellular and extracellular voltage.

22.7

Outlook

The various studies summarized in this chapter reveal the kind of physics and chemistry that is involved in the design of ionoelectronic sensors. They indicate aspects that may be studied in the near future, problems that are more difficult to solve and some long-term projects. In the short term, receptor channels will be expressed in HEK 293 cells or other suitable host cells to create a prototype of an ionoelectronic sensor.18 A crucial task will be the development of suitable transistors with a lower noise.29 On the other hand a systematic variation of cell adhesion24 may lead to a reduced width of the cleft, with lower conductance and improved coupling. Reconstitution of receptor channels into giant lipid vesicles – at a sufficient density – will be difficult. The intermediate approach with membrane fragments from a cell may be more adequate. The yield of coupled assemblies may be enhanced by the application of chips with thousands of contact sites.20 A major problem will be the reproducible quantitative calibration. The crucial parameters of coupling – cleft conductance, distribution of ion channels, driving force – have to be constant during a measurement. Devices with stimulation and recording from the chip may be useful in this respect.7,17,30 A future challenge may be the combination of a receptor cell with an excitable cell, coupled through a synapse as in natural sensorics. Even the integration of natural or designed neuronal nets may be envisaged for amplification and processing.16,17,31

Acknowledgments The work is based on ideas presented in a lecture “On the Fundaments of Biosensorics” in October 1994 at a symposium in Haus Lämmerbuckel, Schwäbische Alb, thanks to an invitation by Daimler-Benz, AG. It was supported generously by the Max-Planck-Gesellschaft and by the Bundesministerium für Bildung und Forschung.

References 1. 2. 3. 4. 5. 6. 7. 8.

P. Fromherz, Eur. Biophys. J. 28, 254 (1999). P. Fromherz, Ber. Bunsenges. Phys. Chem. 100, 1093 (1996). P. Fromherz, A. Offenhäusser, T. Vetter, and J. Weis, Science 252, 1290 (1991). P. Fromherz, C.O. Müller, and R. Weis, Phys. Rev. Lett. 71, 4079 (1993). R. Weis, B. Müller, and P. Fromherz, Phys. Rev. Lett. 76, 327 (1996). R. Weis and P. Fromherz, Phys. Rev. E 55, 877 (1997). A. Stett, B. Müller, and P. Fromherz, Phys. Rev. E 55, 1779 (1997). S. Vassanelli and P. Fromherz, Appl. Phys. A 65, 85 (1997).

Neurochip and sensorics 559 9. M. Jenkner and P. Fromherz, Phys. Rev. Lett. 79, 4705 (1997). 10. M. Rentschler and P. Fromherz, Langmuir 14, 547 (1998). 11. R. Schätzthauer and P. Fromherz, Eur. J. Neurosci. 10, 1956 (1998). 12. S. Vassanelli and P. Fromherz, Appl. Phys. A 66, 459 (1998). 13. S. Vassanelli and P. Fromherz, J. Neurosci. 19, 6767 (1999). 14. V. Kiessling, B. Müller, and P. Fromherz, Langmuir 16, 3515 (2000). 15. P. Fromherz, V. Kiessling, K. Kottig, and G. Zeck, Appl. Phys. A 69, 571 (1999). 16. M. Jenkner, B. Müller, and P. Fromherz, Biol. Cybern. 84, 239 (2001). 17. M. Ulbrich and P. Fromherz, Adv. Mater. 13, 344 (2001). 18. B. Straub, E. Meyer, and P. Fromherz, Nature Biotechn. 19, 121 (2001). 19. V. Kiessling and P. Fromherz, Langmuir (submitted). 20. M. Brenner, A. vom Felde, R. Thewes, and P. Fromherz (unpublished). 21. A. Lambacher and P. Fromherz, Appl. Phys. A 63, 207 (1996). 22. D. Braun and P. Fromherz, Appl. Phys. A 65, 341 (1997). 23. D. Braun and P. Fromherz, Phys. Rev. Lett. 81, 5241 (1998). 24. Y. Iwanaga, D. Braun, and P. Fromherz, Eur. Biophys. J. 30, 17 (2001). 25. A. Lambacher and P. Fromherz, J. Opt. Soc. B 19, 1435 (2002). 26. G. Zeck and P. Fromherz, Langmuir (submitted). 27. D. Braun and P. Fromherz, Phys. Rev. Lett. 86, 2905 (2001). 28. P. Fromherz, Phys. Rev. Lett. 78, 4131 (1997). 29. M. Voelker and P. Fromherz (unpublished). 30. P. Fromherz and A. Stett, Phys. Rev. Lett. 75, 1670 (1995). 31. G. Zeck and P. Fromherz, Proc. Natl. Acad. Sci. USA 98, 10457 (2001).

Index

AB5 type alloys, 473 AcB (acetylene black), 469 accelerated life testing, 346 AcChoE (acetylcholinesterase), 517 AcChoR (acetylcholine receptor), 512 acetic acid, 110 acetylcholine, 532 acetylcholine receptor (AcChoR), 512 acetylcholinesterase (AcChoE), 517 acetylene black (AcB), 469 ACF cloth, 273 acidic air pollutants, 336 ac measurements, 553 ac technique, 437 action potentials, 554 adhesion force microscopy (AFM), 440 adhesive forces, 437 advantages of extracellular ac-simulation, 551 AFM lithography, 450 Airy’s formula, 125 AlF3-doped LiF, 195 amorphous carbon tip, 455 Anabaena, 490 anisotropic conductive film, 244 anisotropic conductive particle, 240 anisotropy micro-ellipsometry (AME), 21, 124, 131, 133 antibodies, 501 aqueous-type, 272 arrays, 9, 10 aspect ratio, 9 ATCO, 493 atmospheric corrosion, 331 atomic adsorption spectroscopy (AAS), 446

atomic beam holography, 218 atomic force microscope (AFM), 14, 67, 106, 180, 431 atomic/molecular manipulation, 430 attractive force, 437 Au lattice, 443 Avrami equation, 162 bacterial magnetic particles (BMPs), 486 bacteria manipulation, 7 bath simulation, 547 Becker, 273 Beer’s law, 374 beetle/chip biosensor, 400 Bessel function, 550 bimetallic (galvanic) corrosion, 336 “Binnig, Quate, and Gerber”, 436 bioelectronic interface, 405 BioFET (biologically sensitive field-effect transistor), 384 biologically sensitive field-effect transistor (BioFET), 384 biological material, 115 biosensing, 512 biosensors, 5, 116, 384 bipotentiostat, 44, 68 blood sensors, 116 BMPs (Bacterial magnetic particles), 486 boundary element method, 78 bump, 229 Burton–Cabrera–Frank model, 87 Butler–Gärtner model, 127 calixarene, 188, 192 capacitive current, 45–6 capillary-based microcell, 109

562

Index

capillary cell, 11 capillary electrochromatography (CEC), 378 capillary electrophoresis (CE), 372 capillary electrophoretic system, 26 capillary force, 439, 441 carbon fiber cloth, 273 carbon nanotube (CNT), 455 cardiac pacemaker, 288 catalytic poisoner, 236 cathodic reaction, 336, 342 cell resistance, 105 cell-transistor junction, 544 certified reference materials (CRM), 377 channel formation, 10 channel width, 259 characterization of systems, 8 chemical force microscopy (CFM), 441, 457 chemically sensitive field-effect transistor (ChemFET), 384 chemical sensors, 384 chemoreceptor, 405 cholinergic ligand, 513 chromaffin cell, 380 chronoamperometry, 32, 58 chronopotentiometry, 376 CMOS, 199, 416 CNT probe, 457 coarse particles, 331, 347 coarse particulate testing, 348 computer generated hologram (CGH), 218 Concao, 395 Concap (constant capacitance) mode, 388 conductance of junction, 547 conducting wires, 453 conduction tracks, 354 conductive polymer, 84 constant current imaging, 98 constant height imaging, 97 contact angle, 110 contact-mode, 436 contamination, 452 continuous dropping method, 241 copper plating, 237 core-coat conductor, 542, 550 corrosion, 23, 81, 329–30, 338, 350; failures, 341; mechanisms, 335, 365;

potential, 336; resistant, 339, 353; system, 330; testing, 346 Cottrell equation, 36 counter-current micro heat exchanger, 261 counter electrode, 104 cracking, 356 critical overpotential concept, 178 crystal ajustment agent, 228 crystallographic faces, 442 current, range of, 12 C/V measuring technique, 386 cyanide-free gold plating, 224 cyanobacteria, 486 cycle life, 300 cyclic voltammetry, 39 Damascene copper electroplating, 178 dc plating, 238 de Broglie, 213 Debye–Hückel theory, 52 deep reactive ion etching (DRIE), 326 definition of EMST, 7 deionized (DI) water, 353 DEM-451, 192 Demaille, 74 deposition velocities, 333 dialysis, 372 diffuse-charge model, 271 diffusion-limited current, 50 dimethyl amine borane (DMAB), 235 “2, 2 -dipyridine”, 226 direct current flow, 92 direct curret (DC) plating, 237 direct electroless nickel plating, 235 direct nickel plating, 234 disk electrode, 41, 51 displacement gold plating, 240 disproportionation reaction, 226 dissolution of ionic crystals, 87 dissolved oxygen concentration (DO), 232 distribution of channels, 552 disulfitoaurate(I)-complex, 249 DMAB (dimethyl amine borane), 235 DNA, 325, 457, 485 double layer capacity, 13, 45 drift effects, 554 dual-disk electrode, 44 duplex layers, 353 duty ratio, 229

Index 563 EB-induced deposition, 197 edge structure, 280 EJ-MOSFET, 201 elecron-tunnelling mechanism, 413 electric double-layer capacitor (EDLC), 271 electric osmosis, 375 electro dischange machining (EDM), 254 electroanalysis, 5 electrochemical capacitor, 271 electrochemical detection methods, 373, 500 electrochemical lab on chip, 26 electrochemical machining (ECM), 7, 16, 19, 26 electrochemical material science (EMS), 4 electrochemical microreactor, 26 electrochemical microsystem, 66 electrochemical micro system technology (EMST), 3 electrochemical -TAS, 26 electrochemical sensors, 350 electrochemical surface treatment, 19 electrochemiluminescence (ECL), 55 electrochemistry, 6 electrode interface, 435 electrodeposition, 19 electrogenerated chemiluminescence, 99 electroless deposition, 21 electroless nickel film, 233 electroless nickel plating, 232 electroless plating, 236 electrolysis, 375 electrolyte-insulator-semiconductor (EIS), 384 electron back scattering diffraction (EBSD), 21 electron beam deposition (EBD), 455 electron beam lithography (EBL), 188, 457 electron-hopping, 414 electronic density of states, 433 electronic Petri dishes, 421 electronic tunnel current, 12 electron transfer, 67, 88 electron-transfer rate, 413 electron transfer reaction (ETR), 10, 17, 123 electron-transfer shuttles, 413

electroplating, 236 electroporation, 512 electropore, 520 ellipsometry, 150 energy dispersive spectrometer, 225 energy-dispersive X-ray analyzer (EDX), 19 enzyme field-effect transistor (EnFET), 384 enzyme-linked immunosorbent assay (ELISA), 86, 505 equilibrium perturbation mode, 72 equivalent circuit, 544 Escherichia coli, 486 ethylene carbonate (EC), 314 ethylene oxide, 262 Euler angle, 21 evolution of heat, 106 excimer laser, 188 Eximer laser ablation, 263 exocytosys events, 380 external corrosion, 361 extracellular ac-simulation, 550 extracellular recording, 555 faradaic current, 45, 435 faradaic processes, 107 fast electron-transfer reaction, 409 fast-scan voltammetry (FSV), 32 feedback mode, 68 ferrocene-conjugated IgG (FG-IgG), 502 FIB deposition, 213 FIB etching, 210 Fick’s law, 33 field amplification factor, 521 field-effect transistor (FET), 384, 541, 542 field strength, 13 filling ratio, 239 fine particle, 349 fingerprint analysis, 381 FLIC-microscopy, 545 flow diagram, 23 flow injection analysis system (FIA), 321, 501 flow rate, 259 flow-stream detection methods, 373 flow systems, 5 flow-through concept, 113 fluorescence, 545 fluorescence measurements, 372

564

Index

flux residues, 342 focused ion beam (FIB), 206 focussed signal, primary and secondary effects, 22 force-distance plot, 438 force modulation microscopy (FMM), 440 Foturan glass panes, 114 four-electrode potentiostat, 68 Frank-van der Merwe growth mode, 166 frictional force microscopy (FFM), 439 FSV (fast-scan voltammetry), 32 fuel cell, 26, 286 fundamentals, 3 galvanic coupling, 343 galvanostatically, 33 gaseous pollutants, 334 gas-sensitive field-effect transistor (GasFET), 384 Gaussian statistics, 377 GenBank, 488 generation-collection mode, 69, 80 geometric blocking, 21 Gibbs energy, 521 Gibbs–Thomson equation, 158 Gibbs–Thomson relation, 172 Gibbs–Wulff–Kaischew theorem, 157 glucose oxidase (GOD), 94, 411 glutamate, 538 glycerol, 259 Gouy–Chapmann, 271 graphite-intercalation compound (GIC), 470 half-wave potential, 42 haneycomb style, 108 Hazard Card, 293, 294 hearing aids, 295 HEK 293 cells, 554, 558 Helmholtz, 271 Helmholtz layer, 136 hexagon flow, 259 higher filling ratios, 280 highly oriented pyrolytic graphite (HOPG), 433, 471 high performance liquid chromatography (HPLC), 372, 378 high resolution SECM image, 99 high throwing power bath, 238

hippocampal neurons, 553 hippocampus, 553 histamine, 508 holographic lithography, 214 homogeneous electron transfer reaction, 55 Hooke’s law, 438 hopping droplet, 111 Horrocks, 74 human chorionic gonadotropin (HCG), 496 hybrid electronic tongue, 381 hydrodynamics, 16 hypophosphite, 232 IgE (immunoglobulin E), 497 IgG (immunoglobulin G), 486 immiscible electrolyte solutions (ITIES), 87 immobilised enzymes, 86 immunoglobulin E (IgE), 497 immunoglobulin G (IgG), 486 inductively coupled plasma (ICP-MS), 377 inhibitor, 513 inorganic resist, 195 in situ method, 14 in situ SPM, 446 insulated STM-tip, 22 integrated circuit (IC), 224 interdigitated array (IDA), 480 interdigitated array electrode (IDA electrode), 473–4 internal corrosion, 362 intervening barrier layer, 339 intracellular simulation, 547 intracellular voltage, 548 in vitro, 531 in vivo, 531 ionic species, 332 ionoelectronic sensors, 541, 542 ion-selective field-effect transistor (ISFET), 384 ion-selective microelectrode, 526 ion-sensitive electrode (ISE), 384 ion sensitive membrane electrode, 117 ion transfer, 84, 89 ion-transfer feedback mode, 71 ion transfer reaction (ITR), 17, 123, 136 iR drop, 44, 50, 466

Index 565 isoelectric point (pI), 502 isolated-antenna BioFET, 391 jump-to-contact, 453 KAu(CN)2, 224 Kelvin-type equation, 547 Kirchhoff’s law, 543, 555 lab-on-chip, 5, 380 laboratory high throughput screening system, 246 Lambert–Beer’s law, 128 laminar flow, 17 Langmuir–Blodgett (LB) film, 446, 448 large-scale integrated (LSI), 224 laser beam, 106 laser induced fluorescence, 380 laser-induced fluorescence measurements, 376 Laser-LIGA process, 263 laser-pulled micropipette tip, 67 lateral force microscopy (LFM), 439 Laves phase alloys, 473 leak conductance, 556 leech ganglia, 531 leveling agent, 238 LiCoO2, 475 LIF(AlF3), 195 lift-off, 417 LIGA-EDM method, 254 LIGA process, 4, 26, 108 LIGA technique, 156, 246 Li-ion batteries, 303 Li-ion button cell, 296 LiMn2O4, 475, 477 linear diffusion reaction, 235 lipid vesicles, 541 Lipon, 303 liquid metal ion (LMI), 206 lithium/iodine-PVP batteries, 288 lithography, 21, 449 LiV2O5, 475 locodynamic experiment, 137 low leakage current, 280 low pressure chemical vapor deposition (LPCVD), 352 macroscopic dimensions, 381

magnetic ac mode, 437 magnetic force image, 431 magnetic force microscopy (MFM), 439 magnetic separation robot, 488 Magnetospirillum magneticum, 486 manipulate, 449 Marcus’ theory, 413, 415 materials, combination of, 18 materials properties, 17 matrix-matched calibrations, 376 Maxwell–Wagner, 520 -EDM (micro electro dischange machining), 254 membrane conductance, 550 membrane electrode assembly (MEA), 23 membrane electroporation, 519 membrane potential, 526 mesocarbon microbeads (MCMB), 467, 476 metal deposition, 93 metal-insulator transition (MIT), 475 metal-insulator-semiconductor (MIS), 386 metal microstructure, 247 metal migration, 338 metal organic chemical vapor deposition (MOCVD), 447 metal oxide semiconductor field-transistor (MOSFET), 181, 199, 389, 543 microanalysis, 371 microbatteries, 116 micro bubble, 231 microbump, 224 micro capacitors, 278 microcell, 10 microchannel, 245 microconnection, 224 microcoulometer, 26 Microcystis, 490 micro-dialysis, 410 microelectroanalytical methods, 377 microelectrochemical methods, 374 micro-electrochemical system, 409 micro electrochemistry, 4 microelectrode, 10, 465, 557, 409 microelectrode array, 10, 512 micro electro dischange machining (-EDM), 254 micro electro mechanical system (MEMS), 321

566

Index

micro electroporation, 5 microellipsommetry, 106 microfabrication, 91 micro flow system, 5 microfuel cells, 317 microgalvanic, 4 micro gap cell, 245 micro heat exchanger, 260 micro/miniaturized total analysis system (-TAS), 321 micromixing device, 255 micropower batteries, 287 micropower fuel cells, 316 micropower lithium cells, 293 micropump, 323 microreactors, 116, 245 micro-reflection spectroscopy, 132 microscopic electrochemistry, 9 microsensor, 115, 485 micro-total analysis systems (-TAS), 5, 372, 377, 380, 422 microvalve, 323 migration, 336 mixed potential concept, 172 mixed potential control, 170 mixing units, 257 mobile carrier, 110 mobile mask, 108 -TAS (micro total analysis system), 321 multi-barrelled, 527 multilamination, 263 multi-sensor system, 409 NADH, 480 Nafion, 316, 405 nanocell, 11 nanode, 38 nanoelectrode ensembles (NEE), 34 nano indentator, 14 nanolithography, 430 Nanotechnology, 3 National Research Institute of Far Seas Fisheries, 494 natural sensorics, 558 NBTA, 495 NBTP, 494 near-field scanning optical microscopy (NSOM), 454 needle type electrode, 416

negative sample bias, 434 Neher–Sackmann, 6, 513 Nernst equation, 528 Nernstian pH response, 395 Nernst potential, 165 Nernst-type batteries, 544 nerve cells, 541 neuron, 117 neuron–silicon junctions, 548 neurotransmitter, 410, 512, 538 NiCd batteries, 295 nickel displacement plating, 233 nickel displacement process, 232 nicotinic AcChoR, 514 Nikolsky–Eisenman equation, 530 nl-droplet method, 124 NMOS, 199 NMOS FET, 200 NMOS transistor, 200 nodular deposition, 236 non-contact mode, 436 nonlinear diffusion reaction, 235 Nostoc, 490 nucleation, 9 numbering up, 245 numerical simulation, 75 odour-binding proteins (ODP), 406 oligonucleotide, 489 online monitoring of the electrolyte, 248 optical/chemical methods, 122 optical detection, 327 optical/electrical laser methods, 125 optical flow microcells, 114 optical microcell, 11 optical/optical methods, 122 organic coatings, 345 organic-type, 272 Oscillatoria, 490 osmolarity, 534 overpotential deposition (OPD), 166 oxidation–reduction cycle (OCR), 444 oxygen reference electrode, 418 oxygen-plasma, 439 palladium chloride, 241 parallel-plate condenser model, 271 particulate matter, 331 passivation layers, 350

Index 567 patch-clamp, 552 patch-clamp technique, 6, 513 PCR-baced nucleotide sequence, 493 Pd cluster, 173 penetration experiments, 71 penicillin microsensor, 399 penicillin sensor, 386 penicillinase, 385, 479 periodical reverse (PR) plating, 237 periodic systems, 9 phase boundary, 106 pH-control, 117 phosphating process, 23 pH-sensitive electrode, 10 phosphoric acid, 226 photocurrent, 154 photocurrent spectra, 129, 134 photocurrent spectroscopy, 125 photodiode, 439 photo electrochemical reaction, 17 photopotential, 135 photoresist films, 350 photoresist micro electrochemistry, 124 photoresist microelectrodes, 132, 142 pH sensor, 385 physical coating technique, 252 piezoelectric scanner, 433 piezo oscillator, 436 pinholes, 360 pitting, 336, 362 planar transducer, 416 planer capillary electrophoresis (chip CE), 321 plane structure, 280 plasma-enhanced chemical vapor deposition (PECVD), 351, 390 plasma spraying method, 273 pneumatic, 324 point-contact model, 549 polyacetylene, 175 polyaniline (PAn), 175, 478 polydimethylsiloxane (PDMS), 326 polyetheretherketone (PEEK), 504 polymerase chain reaction (PCR), 323 poly(methyl methacrylate) (PMMA), 247, 326 polypyrrole (PPy), 175, 478 polythiophene, 175

polyvinilidene fluoride (PVDF), 469 pore size distributions, 278 porous EIS sensors, 390 porous silicon, 19, 385 positive holes, 387 positive sample bias, 434 potassium channels, 553 potential dop, 105 potential error, 105 potential window, 444 potentiodynamic, 33 potentiometric sensors, 385 potentiostatic, 33 Pourbaix diagram, 24 powerful fabrication process, 246 power-type capacitors, 283 primary batteries, 286 PR plating, 237 Propylene Carbonate (PC), 314 prosthetic group, 411 pulsed laser deposition process (PLD), 385 pulse plating, 16, 229 PWBs, 224 quantum mechanics, 433 quasi-direct communication, 415 quasi-microcells, 107 radio immunosorbent test (RIST), 497 Raman spectroscopy, 122 RC cell, 58 reactive ion etching, 247 reagentless biosensors, 414 redox cycling, 45 redox hydrogel, 414 redox system, 46, 52 reference electrode, 104 reflection spectroscopy, 122 reoxidation current, 45 residual Faradaic current, 436 resolution of time and space, 14 Retzius neurons, 532, 548, 557 Reynolds number, 16 RFLP, 493 RNA, 485 rotating-disk electrode (RDE), 40, 50, 55 SAL601, 188 scaling down, up, 23

568

Index

scanning auger microscopy, 19 scanning droplet cells, 11, 113 scanning electrochemical microscope (SECM), 4, 11, 22, 32, 59, 67, 115, 142, 180 scanning Kelvin probes, 106 scanning near-field optical microscopy (SNOM), 454 scanning probe lithography (SPL), 457 scanning probe microscope (SPM), 66, 91, 166, 429 scanning tunnel microscope (STM), 10, 28, 67, 106, 117, 180 SCEs, 202 Schottky barrier, 169 Schottky–Mott equation, 148 SECM apparatus, 73 SECM imaging, 72 secondary batteries, 286 selective corrosion, 336 self assembled monolayer (SAM), 94, 415, 439, 446 self-decomposition, 241 self-discharge, 291, 300 self-supporting micro-sieve, 251 Semiconductor Industry Association (SIA), 457 semiconductors, 83 sensors, 115 Sepharose, 501 SG/TC, 70 Si(111)-77, 433 sialic acid, 517 Si-based transducer, 396 signal-to-noise ratio, 518 silicone, 110 silicon-tin oxynitride (SiTON), 305 SIMOX, 204 single electron transistor (SET), 181 single molecule electrochemistry, 96 single Ni-band, 472 single-walled nanotube (SWNT), 455 Si-SET, 204 small coin-type capacitors, 282 SnSiO3, 306 soft gold plating, 224 spatial resolution, 430 specific capacitance, 278 spectro electrochemistry, 13

SPM probe, 441 steady-state voltammetry, 32, 51 Stern model, 271 stop-and-go technique, 111 Stranski–Krastanov growth mode, 166 stress corrotion cracking (SCC), 337 sulfuric acid, 110 superfilling, 178 surface reconstruction, 442 swelling, 361 SWNT probe, 457 synapic gap, 380 synapse, 558 synchrotron radiation, 188 Synechococcus, 490 tapping mode, 437 TE buffer, 494 TG/SC, 70 thermal disarray, 271 thermal field emission (TFE), 189 thin-film solid state Li batteries, 303 Thomas algorithm, 76 three dimensional flow circuit, 423 three-dimensional microstructure, 247 tip, 73 tip position modulation, 98 tip ultramicroelectrode, 59 Ti/TiO2 system, 124 topographic image, 431 topography, 430, 449 Torpedo Californica AcChoR, 515 transconductance, 400 transfer function, 549 transfer of neutral molecules, 90 TTF-TCNQ, 414 Tubular-type capacitors, 282 tunneling current, 433 UHV, 432 UHV-STM, 433 ultimate microcells, 105 ultramicroelectrode (UME), 38, 50, 58, 66 ultrasonic vibration, 231 UME (using microjet directed), 37 under potential deposition (UPD), 166, 444, 453

Index 569 UV-laser illumination, 154 UV-laser scan, 125, 132 valinomycin, 532 van der Waals force, 431 vapour condensation cell, 11 via-filling, 224, 236 via-hole, 237 video microscope, 111 Villermaux, 256 Volmer–Weber growth mode, 166

voltage-clamp, 554 voltammetry, 32 whole-beetle BioFET, 391 working electrode, 104 Wulff point, 157 Young’s modules, 449, 457 Zeldovich factor, 160 ZEP, 188, 191, 520 zincate, 232