Applied Scanning Probe Methods IV: Industrial Applications (NanoScience and Technology) (v. 4)

NanoScience and Technology NanoScience and Technology Series Editors: P. Avouris B. Bhushan K. von Klitzing H. Sakaki...

Author: Bharat Bhushan | Harald Fuchs

55 downloads 714 Views 8MB 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

NanoScience and Technology

NanoScience and Technology Series Editors: P. Avouris B. Bhushan K. von Klitzing H. Sakaki R. Wiesendanger The series NanoScience and Technology is focused on the fascinating nano-world, mesoscopic physics, analysis with atomic resolution, nano and quantum-effect devices, nanomechanics and atomic-scale processes. All the basic aspects and technology-oriented developments in this emerging discipline are covered by comprehensive and timely books. The series constitutes a survey of the relevant special topics, which are presented by leading experts in the field. These books will appeal to researchers, engineers, and advanced students.

Semiconductor Spintronics and Quantum Computation Editors: D.D. Awschalom, N. Samarth, D. Loss Nano-Optoelectonics Concepts, Physics and Devices Editor: M. Grundmann Noncontact Atomic Force Microscopy Editors: S. Morita, R. Wiesendanger, E. Meyer Nanoelectrodynamics Electrons and Electromagnetic Fields in Nanometer-Scale Structures Editor: H. Nejo Single Organic Nanoparticles Editors: H. Masuhara, H. Nakanishi, K. Sasaki Epitaxy of Nanostructures By V.A. Shchukin, N.N. Ledentsov and D. Bimberg Applied Scanning Probe Methods I Editors: B. Bhushan, H. Fuchs, S. Hosaka Nanostructures Theory and Modeling By C. Delerue and M. Lannoo Nanoscale Characterisation of Ferroelectric Materials Scanning Probe Microscopy Approach Editors: M. Alexe and A. Gruverman

Magnetic Microscopy of Nanostructures Editors: H. Hopster and H.P. Oepen Silicon Quantum Integrated Circuits Silicon-Germanium Heterostructure Devices: Basics and Realisations By E. Kasper, D.J. Paul The Physics of Nanotubes Fundamentals of Theory, Optics and Transport Devices Editors: S.V. Rotkin and S. Subramoney Single Molecule Chemistry and Physics An Introduction By C. Wang, C. Bai Atomic Force Microscopy, Scanning Nearﬁeld Optical Microscopy and Nanoscratching Application to Rough and Natural Surfaces By G. Kaupp Applied Scanning Probe Methods II Scanning Probe Microscopy Techniques Editors: B. Bhushan, H. Fuchs Applied Scanning Probe Methods III Characterization Editors: B. Bhushan, H. Fuchs Applied Scanning Probe Methods IV Industrial Applications Editors: B. Bhushan, H. Fuchs

Bharat Bhushan Harald Fuchs (Eds.)

Applied Scanning Probe Methods IV Industrial Applications

With 176 Figures Including 1 Color Figure

123

Editors: Professor Bharat Bhushan Nanotribology Laboratory for Information Storage and MEMS/NEMS (NLIM) 650 Ackerman Road, Suite 255, The Ohio State University Columbus, OH 43202-1107, USA e-mail: [email protected]

Professor Dr. Harald Fuchs Center for Nanotechnology (CeNTech) and Institute of Physics University of Münster, Gievenbecker Weg 11, 48149 Münster, Germany e-mail: [email protected]

Series Editors: Professor Dr. Phaedon Avouris IBM Research Division, Nanometer Scale Science & Technology Thomas J. Watson Research Center, P.O. Box 218 Yorktown Heights, NY 10598, USA

Professor Bharat Bhushan Nanotribology Laboratory for Information Storage and MEMS/NEMS (NLIM) 650 Ackerman Road, Suite 255, The Ohio State University Columbus, OH 43202-1107, USA

Professor Dr., Dres. h. c. Klaus von Klitzing Max-Planck-Institut für Festkörperforschung, Heisenbergstrasse 1 70569 Stuttgart, Germany

Professor Hiroyuki Sakaki University of Tokyo, Institute of Industrial Science, 4-6-1 Komaba, Meguro-ku Tokyo 153-8505, Japan

Professor Dr. Roland Wiesendanger Institut für Angewandte Physik, Universität Hamburg, Jungiusstrasse 11 20355 Hamburg, Germany DOI 10.1007/b137428 ISSN 1434-4904 ISBN-10 3-540-26912-6 Springer Berlin Heidelberg New York ISBN-13 978-3-540-26912-0 Springer Berlin Heidelberg New York Library of Congress Control Number: 2003059049 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media. springer.com © Springer-Verlag Berlin Heidelberg 2006 Printed in Germany The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Product liability: The publishers cannot guarantee the accuracy of any information about dosage and application contained in this book. In every individual case the user must check such information by consulting the relevant literature. Typesetting and production: LE-TEX Jelonek, Schmidt & Vöckler GbR, Leipzig Cover design: design & production, Heidelberg Printed on acid-free paper 2/3100/YL - 5 4 3 2 1 0

Foreword

The Nobel Prize of 1986 on Scanning Tunneling Microscopy signaled a new era in imaging. The scanning probes emerged as a new instrument for imaging with a precision sufﬁcient to delineate single atoms. At ﬁrst there were two – the Scanning Tunneling Microscope, or STM, and the Atomic Force Microscope, or AFM. The STM relies on electrons tunneling between tip and sample whereas the AFM depends on the force acting on the tip when it was placed near the sample. These were quickly followed by the Magnetic Force Microscope, MFM, and the Electrostatic Force Microscope, EFM. The MFM will image a single magnetic bit with features as small as 10 nm. With the EFM one can monitor the charge of a single electron. Prof. Paul Hansma at Santa Barbara opened the door even wider when he was able to image biological objects in aqueous environments. At this point the sluice gates were opened and a multitude of different instruments appeared. There are signiﬁcant differences between the Scanning Probe Microscopes or SPM, and others such as the Scanning Electron Microscope or SEM. The probe microscopes do not require preparation of the sample and they operate in ambient atmosphere, whereas, the SEM must operate in a vacuum environment and the sample must be cross-sectioned to expose the proper surface. However, the SEM can record 3D image and movies, features that are not available with the scanning probes. The Near Field Optical Microscope or NSOM is also member of this family. At this time the instrument suffers from two limitations; 1) most of the optical energy is lost as it traverses the cut-off region of the tapered ﬁber and 2) the resolution is insufﬁcient for many purposes. We are conﬁdent that NSOM’s with a reasonable optical throughput and a resolution of 10 nm will soon appear. The SNOM will then enter the mainstream of scanning probes.

VI

Foreword

In the Harmonic Force Microscope or HFM, the cantilever is driven at the resonant frequency with the amplitude adjusted so that the tip impacts the sample on each cycle. The forces between tip and sample generate multiple harmonics in the motion of the cantilever. The strength of these harmonics can be used to characterize the physical properties of the surface. It is interesting to note that this technology has spawned devices of a different kind. In one instance, the tip is functionalized in a way that allows the attachment of a single protein. Withdrawing the tip from a surface stretches the attached molecule and measures the elastic properties of single protein molecules. In another the surface tension on the surface of the cantilever is modiﬁed with a self-assembled monolayer of molecules such as thiols. The slight bending of the beam is easily detected with the components developed for use in the scanning probes. This system is used to detect the presence not only of the monomolecular layers but also of single molecules attached to the initial self-assembled monolayer. The extensive material in this ﬁeld means that the variety of topics is larger than can be accommodated in four volumes. The Editors, Profs. Bhushan and Fuchs, must have great powers of persuasion for they have done a remarkable job in collecting this set of paper in a relatively short period of time. The collection will become a milestone in the ﬁeld of scanning probes. c. f. quate Leland T. Edwards Professor (Research) of Engineering Stanford University Stanford, California Co-inventer of AFM in 1985

Preface

The rapidly increasing activities in nanoscience and nanotechnology supported by sizable national programs has led to a variety of efforts in the development and understanding of scanning probe techniques as well as their applications to industrial and medical environments. Beyond imaging, scanning probe techniques representing the eyes of nanotechnology allows us to investigate surfaces and interfaces close to surfaces at the nanometer scale and below, thus providing information about structure, mechanical, electronic, and magnetic properties. It became apparent during the collection phase of Vol. I in 2003 that many more activities exist which deserve presentation. Therefore, this three volume set was prepared in order to display the wide breadth of this ﬁeld and also to provide an excellent compendium for recent developments in this area. The response of colleagues and research groups being asked to contribute has been very positive, such that we decided, together with the publisher, to rapidly move on in these areas. It became possible to collect excellent contributions displaying ﬁrst hand information from leading laboratories worldwide. The present volumes II–IV cover three main areas: scanning probe microscopy (SPM) techniques (Vol. II); characterization (basic aspects, research, Vol. III); and industrial applications (Vol. IV). Volume II includes overviews on sensor technology based on SPM probes, high harmonic dynamic force microscopy, scanning ion conduction microscopy, spin polarized STM, dynamic force microscopy and spectroscopy, quantitative nanomechanical measurements in biology, scanning micro deformation microscopy, electrostatic force and force gradient microscopy and nearﬁeld optical microscopy. This volume also includes a contribution on nearﬁeld probe methods such as the scanning focus ion beam technique which is an extremely valuable tool for nanofabrication including scanning probes. Volume III includes the application of scanning probe methods for the characterization of different materials, mainly in the research stage, such as applications of SPM on living cells at high resolution, macromolecular dynamics, organic supramolecular structures under UHV conditions, STS on organic and inorganic low dimensional systems, and ferroelectric materials, morphological and tribological characterization of rough surfaces, AFM for contact and wear simulation, analysis of fullerene like nanoparticles and applications in the magnetic tape industry. The more relevant industrial applications are described in Vol. IV, which deals with scanning probe lithography for chemical, biological and engineering applications, nanofabrication with self-assembled monolayers by scanned probe lithography, fabrication of nanometer scale structures by local oxidation, template effects of

VIII

Preface

molecular assemblies, microfabricated cantilever arrays, nanothermomechanics and applications of heated atomic force microscope cantilevers. Certainly, the distinction between basic research ﬁelds of scanning probe techniques and the applications in industry are not sharp, as becomes apparent in the distribution of the individual articles in the different parts of these volumes. On the other hand, this clearly reﬂects an extremely active research ﬁeld which strengthens the cooperation between nanotechnology and nanoscience. The success of the series is solely based on the efforts and the huge amount of work done by the authors. We gratefully acknowledge their excellent contributions in a timely manner which helps to inform scientists in research and industry about latest achievements in scanning probe methods. We also would like to thank Dr. Marion Hertel, Senior Editor Chemistry, and Mrs. Beate Siek of Springer Verlag for their continuous support, without which this volume could never make it efﬁciently to market. January, 2006

Prof. Bharat Bhushan, USA Prof. Harald Fuchs, Germany

Contents – Volume IV

23

Scanning Probe Lithography for Chemical, Biological and Engineering Applications Joseph M. Kinsella, Albena Ivanisevic . . . . . . . . . . . . . . . .

1

23.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2

23.2

Modeling of the DPN Process . . . . . . . . . . . . . . . . . . . .

4

23.3 23.3.1 23.3.2 23.3.3 23.3.4

Patterning of Biological and Biologically Active Molecules . DNA Patterning . . . . . . . . . . . . . . . . . . . . . . . . . Protein Patterning . . . . . . . . . . . . . . . . . . . . . . . . Peptide Patterning . . . . . . . . . . . . . . . . . . . . . . . . Patterning of Templates for Biological Bottom-Up Assembly

. . . . .

. . . . .

. . . . .

7 8 10 13 15

23.4 23.4.1 23.4.2 23.4.3 23.4.4 23.4.5 23.4.6 23.4.7 23.4.8 23.4.9 23.4.10 23.4.11

Chemical Patterning . . . . . . . . . . . . Thiols . . . . . . . . . . . . . . . . . . . ω-Substituted Thiols . . . . . . . . . . . Silanes and Silazanes . . . . . . . . . . . Deposition of Solid Organic Inks . . . . . Polymers . . . . . . . . . . . . . . . . . . Polyelectrolytes . . . . . . . . . . . . . . Dendrimers . . . . . . . . . . . . . . . . Deposition of Supramolecular Materials . Deposition of Metals . . . . . . . . . . . Deposition of Solid-State Materials . . . Deposition of Magnetic Materials . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

17 17 18 19 20 21 23 23 24 25 26 27

23.5

Engineering Applications of DPN . . . . . . . . . . . . . . . . . .

28

23.6

Future Challenges and Applications . . . . . . . . . . . . . . . . .

30

23.7

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

X

24

Contents – Volume IV

Nanotribological Characterization of Human Hair and Skin Using Atomic Force Microscopy (AFM) Bharat Bhushan, Carmen LaTorre . . . . . . . . . . . . . . . . . .

35

24.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

35

24.2 24.2.1 24.2.2

Human Hair, Skin, and Hair Care Products . . . . . . . . . . . . . Human Hair and Skin . . . . . . . . . . . . . . . . . . . . . . . . . Hair Care: Cleaning and Conditioning Treatments, and Damaging Processes . . . . . . . . . . . . . . . . . . . . . . .

39 39

24.3 24.3.1 24.3.2

Experimental Techniques . . . . . . . . . . . . . . . . . . . . . . . Experimental Procedure . . . . . . . . . . . . . . . . . . . . . . . . Hair and Skin Samples . . . . . . . . . . . . . . . . . . . . . . . .

51 53 57

24.4 24.4.1

. .

59

. .

59

. .

70

. .

78

24.4.5

Results and Discussion . . . . . . . . . . . . . . . . . . . . . . Surface Roughness, Friction, and Adhesion for Various Ethnicities of Hair . . . . . . . . . . . . . . . . . . Surface Roughness, Friction, and Adhesion for Virgin and Chemically Damaged Caucasian Hair (with and without Commercial Conditioner Treatment) . . . . . Surface Roughness, Friction, and Adhesion for Hair Treated with Various Combinations of Conditioner Ingredients Investigation of Directionality Dependence and Scale Effects on Friction and Adhesion of Hair . . . . . . . . . . . . . . . . . Surface Roughness and Friction of Skin . . . . . . . . . . . . .

. . . .

85 98

24.5

Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

98

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

102

Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

103

24.4.2

24.4.3 24.4.4

25

46

Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography Jayne C. Garno, James D. Batteas . . . . . . . . . . . . . . . . . .

105

25.1 25.1.1 25.1.2 25.1.3 25.1.4

SPM-Based Methods of Lithography . . Bias-Induced Nanofabrication . . . . . . Force-Induced Nanofabrication of SAMs Dip-Pen Nanolithography (DPN) . . . . . Automated Scanning Probe Lithography .

. . . . .

105 107 108 110 111

25.2 25.2.1 25.2.2 25.2.3

Patterning with Self-Assembled Monolayers . . . . . . . . . . . . Structure of SAMs . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples of SAM Nanopatterns Generated by Force-Induced SPL Nanofabrication of SAMs by DPN and Bias-Induced SPL . . . . .

112 112 114 118

25.3

Directed Fabrication of Polymeric Structures . . . . . . . . . . . .

120

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

Contents – Volume IV

XI

25.4

Fabrication of Metallic Structures . . . . . . . . . . . . . . . . . .

122

25.5 25.5.1 25.5.2 25.5.3

Nanoscale Patterning of Proteins . . . . . . . . . . . . Protein Arrays Generated by DPN . . . . . . . . . . . Applying Bias-Induced SPL for Protein Nanopatterns Protein Immobilization on SAMs Generated by Force-Induced SPL . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

126 127 128

. . . . . . .

129

Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . .

130

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

131

25.6

26

Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography Marta Tello, Fernando García, Ricardo García . . . . . . . . . . .

137

26.1

Introduction to AFM Nanolithographies . . . . . . . . . . . . . . .

137

26.2

Basic Local Oxidation Aspects . . . . . . . . . . . . . . . . . . . .

138

26.3

Mechanism and Kinetics . . . . . . . . . . . . . . . . . . . . . . .

141

26.4

Feature Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

143

26.5

Applications I: Patterning, Data Storage and Template Growth . .

146

26.6

Applications II: Nanoelectronic Devices . . . . . . . . . . . . . . .

151

26.7

Parallel Oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . .

154

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

155

27

Template Effects of Molecular Assemblies Studied by Scanning Tunneling Microscopy (STM) Chen Wang, Chunli Bai . . . . . . . . . . . . . . . . . . . . . . . .

159

27.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

159

27.2

27.2.2 27.2.3 27.2.4

Single Guest Molecule Immobilization with Assembled Molecular Networks . . . . . Hydrogen Bonded Supramolecular Networks and Single Molecule Inclusions . . . . . . . . Van der Waals Interaction Stabilized Networks Metal-Organic Coordination Networks . . . . Covalently Bonded Molecular Grids . . . . . .

. . . . . . . . . . .

160

. . . .

. . . .

160 163 165 166

27.3 27.3.1 27.3.2

Intralayer Heterogeneous Molecular Arrays . . . . . . . . . . . . . Hydrogen Bond Stabilized Heterogeneous Lamellae . . . . . . . . Van der Waals Interaction Stabilized Intralayer Arrays . . . . . . .

166 167 168

27.4 27.4.1

Interlayer Effect on Molecular Adsorption and Assemblies . . . . Site Selective Adsorption . . . . . . . . . . . . . . . . . . . . . . .

171 172

27.2.1

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

XII

Contents – Volume IV

27.4.2 27.4.3

Molecular Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . Directional Assembly of Nanoparticle Arrays . . . . . . . . . . . .

176 177

27.5

Future Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . .

179

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

179

28

Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection Hans Peter Lang, Martin Hegner, Christoph Gerber . . . . . . . .

183

28.1 28.1.1 28.1.2 28.1.3 28.1.4

Introduction . . . . . . . . . Sensors . . . . . . . . . . . . Cantilevers . . . . . . . . . . Cantilever Operating Modes Cantilever Arrays . . . . . .

. . . . .

183 183 184 186 192

28.2 28.2.1 28.2.2

Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement Chamber . . . . . . . . . . . . . . . . . . . . . . . . Cantilever Functionalization . . . . . . . . . . . . . . . . . . . . .

196 196 198

28.3 28.3.1 28.3.2

Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Artiﬁcial Nose for Detection of Perfume Essences . . . . . . . . . Label-Free DNA Hybridization Detection . . . . . . . . . . . . . .

203 204 206

28.4

Applications and Outlook . . . . . . . . . . . . . . . . . . . . . . .

209

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

210

29

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices B. Gotsmann, U. Dürig . . . . . . . . . . . . . . . . . . . . . . . .

215

29.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

215

29.2 29.2.1 29.2.2 29.2.3 29.2.4

Heat Transfer Mechanisms . . . . . . . . . . . . . . Heat Generation in Microcantilevers . . . . . . . . . Heat Transfer Through Air and Silicon . . . . . . . Heat Transfer Through Radiation . . . . . . . . . . . Heat Transfer Through a Tip-Surface Point Contact

. . . . .

215 216 217 222 224

29.3

Momentum Transfer Through Air . . . . . . . . . . . . . . . . . .

227

29.4 29.4.1 29.4.2 29.4.3 29.4.4 29.4.5 29.4.6 29.4.7

Thermomechanical Nanoindentation of Polymers . . . General Considerations . . . . . . . . . . . . . . . . . Indentation Experiments . . . . . . . . . . . . . . . . Interlude: Carbon Nanotube Tips . . . . . . . . . . . . Interlude: Thermal Force and Indentation Formation . Interlude: Rim Formation on Polymer Samples . . . . Indentation Kinetics and the Indentation Mechanism . Interlude: Thermo-Nano-Mechanics Without a Heater

229 229 230 232 234 234 236 239

. . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . . . . .

Contents – Volume IV

XIII

29.5

Thermomechanical Nanowear Testing . . . . . . . . . . . . . . . .

241

29.6 29.6.1 29.6.2

Application to Data-Storage Devices . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scaling Challenges for Nanoindentation of Polymers . . . . . . . .

243 243 245

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

248

30

Applications of Heated Atomic Force Microscope Cantilevers Brent A. Nelson, William P. King . . . . . . . . . . . . . . . . . . .

251

30.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

251

30.2 30.2.1 30.2.2 30.2.3

Physical and Environmental Sensing . . . . . . . . . . . . Pressure Sensing . . . . . . . . . . . . . . . . . . . . . . . Thermal Conductivity Mapping and Subsurface Imaging . Topographical Detection . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

252 252 253 258

30.3 30.3.1 30.3.2 30.3.3 30.3.4

Chemical Sensing Applications . . . . . . . Calorimetry . . . . . . . . . . . . . . . . . Mass Detection . . . . . . . . . . . . . . . Time-of-Flight Scanning Force Microscopy Explosives Detection . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

261 261 262 263 264

30.4 30.4.1 30.4.2

Data Storage and Lithography . . . . . . . . . . . . . . . . . . . . Data Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lithography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

264 265 269

30.5

Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . .

272

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

272

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

Contents – Volume II

1

Higher Harmonics in Dynamic Atomic Force Microscopy Robert W. Stark, Martin Stark . . . . . . . . . . . . . . . . . . . .

1

1.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

1.2 1.2.1 1.2.2 1.2.3 1.2.4 1.2.5 1.2.6 1.2.7 1.2.8

Multimodal Model of the Microcantilever . . . Overview . . . . . . . . . . . . . . . . . . . . . Modal Analysis . . . . . . . . . . . . . . . . . Tip–Sample Interaction . . . . . . . . . . . . . State Space Formulation . . . . . . . . . . . . Dynamics: Linearized Tip–Sample Interaction Poles and Zeros . . . . . . . . . . . . . . . . . Dynamics: Nonlinear Interaction . . . . . . . . Optical Readout . . . . . . . . . . . . . . . . .

. . . . . . . . .

4 4 5 7 9 11 13 16 20

1.3

Higher Harmonic Imaging . . . . . . . . . . . . . . . . . . . . . .

23

1.4 1.4.1 1.4.2 1.4.3

Spectroscopy: Distinguishing Two Polymers Overview . . . . . . . . . . . . . . . . . . . . Experimental Details . . . . . . . . . . . . . Signal Analysis . . . . . . . . . . . . . . . .

. . . .

27 27 28 28

1.5

Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

33

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

33

2

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

. . . . . . . . .

. . . .

Atomic Force Acoustic Microscopy Ute Rabe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37

2.1 2.1.1 2.1.2 2.1.3

Introduction . . . . . . . . . . . . . . . . . . . . . Near-ﬁeld Acoustic Microscopy . . . . . . . . . . Scanning Probe Techniques and Nanoindentation . Vibration Modes of AFM Cantilevers . . . . . . .

. . . .

. . . .

. . . .

38 39 40 41

2.2 2.2.1 2.2.2 2.2.3

Linear Contact-resonance Spectroscopy Using Flexural Modes Flexural Vibrations of Clamped-free Beams . . . . . . . . . . . The Point-mass Model . . . . . . . . . . . . . . . . . . . . . . Experiments with Clamped-free Beams . . . . . . . . . . . . .

. . . .

. . . .

42 44 47 48

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

XVI

Contents – Volume II

2.3

Contact Forces as Linear Springs and Dashpots . . . . . . . . . . .

51

2.4 2.4.1 2.4.2 2.4.3

Characteristic Equation of the Surface-coupled Beam Discussion of the Characteristic Equation . . . . . . . Inﬂuence of an Additional Mass . . . . . . . . . . . . Roots of the Characteristic Equation with Damping . .

. . . .

55 58 61 63

2.5

Forced Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . .

64

2.6

Imaging and Contrast Inversion . . . . . . . . . . . . . . . . . . .

70

2.7

Sensitivity of the Flexural Modes . . . . . . . . . . . . . . . . . . .

73

2.8 2.8.1

Quantitative Evaluation . . . . . . . . . . . . . . . . . . . . . . . . Experiments for Quantitative Evaluation . . . . . . . . . . . . . . .

76 78

2.9

Nonlinear Forces . . . . . . . . . . . . . . . . . . . . . . . . . . .

82

2.10

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

83

A A.1 A.2 A.3

Appendix . . . Deﬁnitions . . UAFM-mode . AFAM-mode .

. . . .

84 84 85 86

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

88

3

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Scanning Ion Conductance Microscopy Tilman E. Schäffer, Boris Anczykowski, Harald Fuchs . . . . . . .

91

3.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

91

3.2 3.2.1 3.2.2 3.2.3

Fundamental Principles Basic Setup . . . . . . Nanopipettes . . . . . . Electrodes . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

92 92 95 96

3.3 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5

Ion Currents Through Nanopipettes . . Background Theory . . . . . . . . . . . Simple Analytical Model . . . . . . . . Finite Element Modeling . . . . . . . . Experimental Current-Distance Curves Imaging with Ion Current Feedback . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

97 97 97 99 101 102

3.4 3.4.1 3.4.2

Advanced Techniques . . . . . . . . . . . . . . . . . . . . . . . . . Modulation Methods . . . . . . . . . . . . . . . . . . . . . . . . . Applications in Bioscience . . . . . . . . . . . . . . . . . . . . . .

103 104 106

3.5 3.5.1 3.5.2 3.5.3

Combination with Other Scanning Techniques Combination with Atomic Force Microscopy . Application in Material Science . . . . . . . . Combination with Shear Force Microscopy . .

107 108 108 111

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Contents – Volume II

XVII

3.5.4

Application in Bioscience . . . . . . . . . . . . . . . . . . . . . . .

114

3.6

Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

115

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

116

4

Spin-Polarized Scanning Tunneling Microscopy Wulf Wulfhekel, Uta Schlickum, Jürgen Kirschner . . . . . . . . . .

121

4.1 4.1.1 4.1.2 4.1.3

Introduction . . . . . . . . . . . . . . . . . . . The Resolution Problem in Magnetic Imaging Magnetism and Spin . . . . . . . . . . . . . . . The Tunneling Magnetoresistance Effect . . .

. . . .

. . . .

121 121 122 122

4.2 4.2.1 4.2.2 4.2.3

The Principle of Spin-polarized Scanning Tunneling Microscopy The Constant Current Mode . . . . . . . . . . . . . . . . . . . . The Spectroscopic Mode . . . . . . . . . . . . . . . . . . . . . . Differential Magnetic Imaging Mode . . . . . . . . . . . . . . .

. . . .

124 125 125 126

4.3

Experimental Set-up . . . . . . . . . . . . . . . . . . . . . . . . . .

127

4.4 4.4.1 4.4.2

Ferromagnetic Domains and Domain Walls . . . . . . . . . . . . . Ultra-sharp Domain Walls in Co(0001) . . . . . . . . . . . . . . . Asymmetric Néel Caps in Fe(001) . . . . . . . . . . . . . . . . . .

128 129 131

4.5 4.5.1 4.5.2

Antiferromagnets in Contact with Ferromagnets . . . . . . . . . . Mn on Fe(001) and Topologically Induced Frustrations . . . . . . The Layered Antiferromagnet Cr on Fe(001) . . . . . . . . . . . .

133 133 136

4.6 4.6.1 4.6.2

Bulk Versus Surface: Which Electronic States Cause the Spin Contrast? . . . . . . . . . Voltage Dependence of the TMR Effect in Co(0001) . . . . . . . . Voltage Dependence of the TMR Effect in Cr/Fe(001) . . . . . . .

137 137 139

4.7

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

140

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

140

5

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Dynamic Force Microscopy and Spectroscopy Ferry Kienberger, Hermann Gruber, Peter Hinterdorfer . . . . . .

143

5.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

144

5.2

Scanning Probe Microscopy . . . . . . . . . . . . . . . . . . . . .

145

5.3

Dynamic Force Microscopy Imaging . . . . . . . . . . . . . . . . .

146

5.4 5.4.1 5.4.2 5.4.3

Force Spectroscopy Principles . . . . . Theory . . . . . . . Applications . . . .

149 149 151 153

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

XVIII

Contents – Volume II

5.5

Combined Imaging and Spectroscopy . . . . . . . . . . . . . . . .

158

5.6

Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . .

161

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

161

6

Sensor Technology for Scanning Probe Microscopy and New Applications Egbert Oesterschulze, Leon Abelmann, Arnout van den Bos, Rainer Kassing, Nicole Lawrence, Gunther Wittstock, Christiane Ziegler . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

6.1

Introductory Remarks . . . . . . . . . . . . . . . . . . . . . . . . .

165

6.2 6.2.1

Material Aspects of Probe Fabrication . . . . . . . . . . . . . . . . Mechanical Properties of Cantilever Probes . . . . . . . . . . . . .

166 167

6.3 6.3.1 6.3.2

Scanning Near-Field Optical Microscopy . . . . . . . . . . . . . . Principle of Near-Field Optics . . . . . . . . . . . . . . . . . . . . Probes for Scanning Near-Field Optical Microscopy (SNOM) . . .

174 174 175

6.4 6.4.1

Probes for Ultrafast Scanning Probe Microscopy . . . . . . . . . . Improved Sampling Technique . . . . . . . . . . . . . . . . . . . .

179 181

6.5 6.5.1 6.5.2

Functionalized Tips . . . . . . . . . . . . . . . . . . . . . . . . . . Tip Modiﬁcation . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

182 182 183

6.6 6.6.1 6.6.2

Scanning Electrochemical Microscopy . . . . . . . . . . . . . . . . Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

186 186 189

6.7 6.7.1 6.7.2 6.7.3 6.7.4

Tips for Magnetic Force Microscopy . Ideal Tip Shape . . . . . . . . . . . . Hand-Made Tips . . . . . . . . . . . . Coating AFM Tips . . . . . . . . . . Tip Planes: The CantiClever Concept

. . . . .

192 192 193 194 195

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

197

7

7.1 7.1.1 7.1.2 7.1.3 7.1.4

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

Quantitative Nanomechanical Measurements in Biology Małgorzata Lekka, Andrzej J. Kulik . . . . . . . . . . . . . . . . . Stiffness of Biological Samples . . . . . . . Cell Structure . . . . . . . . . . . . . . . . Determination of Young’s Modulus . . . . Brief Overview of the Application of AFM to Studies of Living Cells . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . .

205

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

205 205 208

. . . . . . . . . . . . . . . . . . . . . . . . . .

217 222

Contents – Volume II

7.2 7.2.1 7.2.2 7.2.3

XIX

. . . .

224 225 229 236

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

237

8

Friction Force Microscopy . . . . . . . . Friction and Chemical Force Microscopy Applications of FFM/CFM . . . . . . . . Summary . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Scanning Microdeformation Microscopy: Subsurface Imaging and Measurement of Elastic Constants at Mesoscopic Scale Pascal Vairac, Bernard Cretin . . . . . . . . . . . . . . . . . . . .

241

8.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

241

8.2

Review and Physical Background of Near-Field Acoustic Microscopes . . . . . . . . . . . . . . . . . . . . . . Review of Near-Field Microscopes . . . . . . . . . . . . . . . Physical Basis for Near-Field Acoustics and the Scale Effect Mechanical Approach . . . . . . . . . . . . . . . . . . . . . . Models of Subsurface Sensing Using Acoustic Waves and Surface Bending . . . . . . . . . . . . . . . . . . . . . .

. . . .

242 242 244 247

. . .

252

8.3.1 8.3.2 8.3.3

Imaging and Measurement with Scanning Microdeformation Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . Conﬁguration . . . . . . . . . . . . . . . . . . . . . . . . . . Application to Subsurface Imaging . . . . . . . . . . . . . . . Characterization of Local Mechanical Constants . . . . . . .

. . . .

. . . .

. . . .

254 254 256 259

8.4 8.4.1 8.4.2 8.4.3

Speciﬁc Application . . . Thin Film Measurements Shape Memory Alloy . . Viscosimetry . . . . . . .

. . . .

. . . .

. . . .

260 260 264 267

8.5

Ultimate Metrology: Measurements at the Mechanical Noise Level

274

8.6

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

278

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

279

8.2.1 8.2.2 8.2.3 8.2.4 8.3

9

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Electrostatic Force and Force Gradient Microscopy: Principles, Points of Interest and Application to Characterisation of Semiconductor Materials and Devices Paul Girard, Alexander Nikolaevitch Titkov . . . . . . . . . . . . .

283

9.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

285

9.2 9.2.1 9.2.2

Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principles of Surface-voltage Measurements . . . . . . . . . . . .

285 286 287

XX

9.2.3

Contents – Volume II

9.2.4

Detection of Strong Local Electrical Effect via the “Topographic” Data . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

294 296

9.3 9.3.1 9.3.2 9.3.3 9.3.4 9.3.5

Observation and Interpretation DC Observations . . . . . . . Ω Observations . . . . . . . . 2Ω Observations . . . . . . . Surface Voltage Observations . Guidelines for Interpretation .

. . . . . .

297 299 300 300 302 302

9.4 9.4.1 9.4.2 9.4.3

Future Opportunities . . . . . . . . . . . . . . . Interest in the KFGM Method . . . . . . . . . . Spatially Resolved Observations . . . . . . . . . Another Way to Estimate the Maximum Possible Spatial Resolution . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

304 304 309

. . . . . . . . . .

311

9.5 9.5.1 9.5.2

Some Applications . . . . . . . . . . . . . . . . . . . . . . . . . . Applications Under Ambient Conditions . . . . . . . . . . . . . . Vacuum or UHV Applications . . . . . . . . . . . . . . . . . . . .

313 314 316

9.6

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

316

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

318

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

10

Polarization-Modulation Techniques in Near-Field Optical Microscopy for Imaging of Polarization Anisotropy in Photonic Nanostructures Pietro Giuseppe Gucciardi, Ruggero Micheletto, Yoichi Kawakami, Maria Allegrini . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321

10.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

321

10.2 10.2.1

Polarimetric Imaging . . . . . . . . . . . . . . . . . . . . . . . . . The Jones Formalism . . . . . . . . . . . . . . . . . . . . . . . . .

322 325

10.3

Electromagnetic Field Diffracted by a SNOM Aperture . . . . . .

327

10.4 10.4.1 10.4.2 10.4.3

Experimental Implementations . . . . . . . . . . . . Static Polarization SNOM . . . . . . . . . . . . . . Polarization-Modulation SNOM: Illumination Mode Polarization-Modulation SNOM: Collection Mode .

. . . .

. . . .

. . . .

. . . .

. . . .

333 333 337 342

10.5 10.5.1 10.5.2 10.5.3 10.5.4 10.5.5

Applications of SNOM Polarimetry . . . . . . . . . . . . . Polarization Responses of Photonic Waveguides . . . . . . Measuring Stress-Induced Birefringence . . . . . . . . . . . Polarization Anisotropy in Mesoscale-Structured Materials Polarization Anisotropy in Polymers . . . . . . . . . . . . . Polarization Anisotropy in Photoluminescence Emission . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

344 345 348 349 351 354

10.6

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

357

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

357

. . . .

. . . .

. . . .

Contents – Volume II

11

XXI

Focused Ion Beam as a Scanning Probe: Methods and Applications Vittoria Raffa, Piero Castrataro, Arianna Menciassi, Paolo Dario .

361

11.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

361

11.2 11.2.1 11.2.2 11.2.3 11.2.4

Description of the System . . . . . System Overview . . . . . . . . . Liquid Metal Ion Source (LMIS) . Ion Optics . . . . . . . . . . . . . Dual Beam Systems . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

362 362 363 364 365

11.3 11.3.1 11.3.2 11.3.3 11.3.4 11.3.5

FIB Processes . . . . . . Imaging . . . . . . . . . Milling . . . . . . . . . . Gas-Assisted Etching . . Gas-Assisted Deposition Ion Beam Lithography .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

367 367 372 376 377 379

11.4 11.4.1 11.4.2 11.4.3

Main Applications . . . . . . . . . . FIB as an Analytical Technique . . FIB in the Semiconductor Industry . Micromachining . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

380 381 389 401

11.5

Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . .

408

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

409

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413

Contents – Volume III

12

Atomic Force Microscopy in Nanomedicine Dessy Nikova, Tobias Lange, Hans Oberleithner, Hermann Schillers, Andreas Ebner, Peter Hinterdorfer . . . . . . .

1

12.1

AFM in Biological Sciences . . . . . . . . . . . . . . . . . . . . .

1

12.2 12.2.1 12.2.2 12.2.3 12.2.4 12.2.5

Plasma Membrane Preparation for AFM Imaging . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . Plasma Membrane Preparation . . . . . . . . . . . . . . . Atomic Force Microscopy . . . . . . . . . . . . . . . . . Molecular Volume Measurements of Membrane Proteins . AFM Imaging . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

4 4 5 7 7 7

12.3 12.3.1 12.3.2 12.3.3

AFM Imaging of CFTR in Oocyte Membranes Introduction . . . . . . . . . . . . . . . . . . . Does the CFTR Form Functional Assemblies? Two CFTRs are Better Than One . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

10 11 11 13

12.4 12.4.1 12.4.2 12.4.3 12.4.4

Single Antibody–CFTR Recognition Imaging . Introduction . . . . . . . . . . . . . . . . . . . Tethering of Antibodies to AFM Tips . . . . . AFM Imaging and Recognition . . . . . . . . . A Single Antibody Sees a Single CFTR . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

16 16 17 17 17

12.5 12.5.1 12.5.2 12.5.3

Single Cell Elasticity: Probing for Diseases Introduction . . . . . . . . . . . . . . . . . Force–Mapping AFM . . . . . . . . . . . . Can One Protein Change Cell Elasticity? .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

19 19 20 21

12.6

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

24

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

24

. . . .

. . . .

XXIV

13

Contents – Volume III

Scanning Probe Microscopy: From Living Cells to the Subatomic Range Ille C. Gebeshuber, Manfred Drack, Friedrich Aumayr, Hannspeter Winter, Friedrich Franek . . . . . . . . . . . . . . . . .

27

13.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27

13.2 13.2.1 13.2.2

Cells In Vivo as Exempliﬁed by Diatoms . . . . . . . . . . . . . . Introduction to Diatoms . . . . . . . . . . . . . . . . . . . . . . . . SPM of Diatoms . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 28 30

13.3

Interaction of Large Organic Molecules . . . . . . . . . . . . . . .

33

13.4 13.4.1 13.4.2

Nanodefects on Atomically Flat Surfaces . . . . . . . . . . . . . . Ion Bombardment of Highly Oriented Pyrolytic Graphite (HOPG) Bombardment of Single Crystal Insulators with Multicharged Ions

37 38 42

13.5 13.5.1 13.5.2

Subatomic Features . . . . . . . . . . . . . . . . . . . . . . . . . . Atom Orbitals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Electron Spin Detection with AFM and STM . . . . . . . .

45 45 47

13.6

Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . .

50

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51

14

Surface Characterization and Adhesion and Friction Properties of Hydrophobic Leaf Surfaces and Nanopatterned Polymers for Superhydrophobic Surfaces Zachary Burton, Bharat Bhushan . . . . . . . . . . . . . . . . . .

55

14.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

14.2 14.2.1 14.2.2 14.2.3 14.2.4

Experimental Details . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . Samples . . . . . . . . . . . . . . . . . . Roughness Factor . . . . . . . . . . . . . Test Matrix for Nanopatterned Polymers .

. . . . .

58 58 59 61 62

14.3 14.3.1 14.3.2

Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . Hydrophobic Leaf Surfaces . . . . . . . . . . . . . . . . . . . . . . Nanopatterned Polymers . . . . . . . . . . . . . . . . . . . . . . .

63 64 74

14.4

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

79

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

81

15

15.1

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

Probing Macromolecular Dynamics and the Influence of Finite Size Effects Scott Sills, René M. Overney . . . . . . . . . . . . . . . . . . . . .

83

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

84

Contents – Volume III

XXV

15.2

The Glass Transition and Molecular Mobility . . . . . . . . . . . .

85

15.3 15.3.1 15.3.2 15.3.3 15.3.4 15.3.5 15.3.6

Macromolecular Probing Techniques . . . . . . . . . . . . . Static Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . Modulated Contacts . . . . . . . . . . . . . . . . . . . . . . . Calibration of Lateral Forces in Scanning Probe Microscopy . Shear Modulation Force Microscopy (SM-FM) . . . . . . . . Friction Force Microscopy (FFM) . . . . . . . . . . . . . . . Tribological Models for FFM . . . . . . . . . . . . . . . . . .

. . . . . . .

90 90 92 93 97 98 99

15.4 15.4.1 15.4.2 15.4.3

Internal Friction and Dynamics near the Glass Transition . . . . . Molecular Relaxations . . . . . . . . . . . . . . . . . . . . . . . . Structural Heterogeneity and Relaxation near the Glass Transition Cooperative Molecular Motion During the Glass Transition . . . .

103 103 105 107

15.5 15.5.1 15.5.2 15.5.3 15.5.4

Constraints and Structural Modiﬁcations near Interfaces Interfacial Plasticization . . . . . . . . . . . . . . . . . . Dewetting Kinetics . . . . . . . . . . . . . . . . . . . . Disentanglement Barriers . . . . . . . . . . . . . . . . . Interfacial Glass Transition Proﬁles . . . . . . . . . . .

. . . . .

. . . . .

109 109 110 111 113

15.6 15.6.1 15.6.2 15.6.3

Mechanical Operations in Nanoscopic Polymer Systems Indentation Contact Mechanics . . . . . . . . . . . . . . Rim Formation During Indentation . . . . . . . . . . . . Strain Shielding and Conﬁned Plasticity in Nanoscopic Polymer Systems . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

115 116 120

. . . . . .

122

Closing Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . .

126

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

127

15.7

16

. . . . .

. . . . .

. . . . . . .

. . . . .

. . . . . . .

. . . . .

Investigation of Organic Supramolecules by Scanning Probe Microscopy in Ultra-High Vacuum Laurent Nony, Enrico Gnecco, Ernst Meyer . . . . . . . . . . . . .

131

16.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

131

16.2 16.2.1 16.2.2 16.2.3

Methods . . . . . . . . . . . . . . . . . . . Organic Molecular Beam Epitaxy (OMBE) Scanning Tunneling Microscopy (STM) . . Atomic Force Microscopy (AFM) . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

132 132 134 137

16.3 16.3.1 16.3.2 16.3.3 16.3.4 16.3.5 16.3.6 16.3.7

Molecules . . . . . . . . . . Fullerenes . . . . . . . . . . Porphyrins . . . . . . . . . . Phthalocyanines . . . . . . . Perylene Derivatives . . . . . Lander Molecules . . . . . . PVBA Molecules . . . . . . Decacyclene and Derivatives

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

142 142 142 142 144 144 144 144

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

XXVI

Contents – Volume III

16.4 16.4.1 16.4.2

Molecules on Metals . . . . . . . . . . . . . . . . . . . . . . . . . STM Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . Non-Contact AFM Investigations . . . . . . . . . . . . . . . . . .

145 145 157

16.5 16.5.1 16.5.2

Molecules on Semiconductor Surfaces . . . . . . . . . . . . . . . . STM Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . Non-Contact AFM Investigations . . . . . . . . . . . . . . . . . .

162 162 165

16.6 16.6.1 16.6.2

Molecules on Insulating Surfaces . . . . . . . . . . . . . . . . . . . STM Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . Non-contact AFM Investigations . . . . . . . . . . . . . . . . . . .

167 168 169

16.7 16.7.1 16.7.2

Manipulation of Single Molecules . . . . . . . . . . . . . . . . . . STM Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . Non-contact AFM Investigations . . . . . . . . . . . . . . . . . . .

171 171 175

16.8

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

175

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

176

17

One- and Two-Dimensional Systems: Scanning Tunneling Microscopy and Spectroscopy of Organic and Inorganic Structures Luca Gavioli, Massimo Sancrotti . . . . . . . . . . . . . . . . . . .

183

17.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

183

17.2

Basic Principles of STM and STS . . . . . . . . . . . . . . . . . .

185

17.3 17.3.1 17.3.2

Inorganic Overlayers . . . . . . . . . . . . . . . . . . . . . . . . . 1D Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2D Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

188 188 196

17.4 17.4.1 17.4.2

Molecular Overlayers . . . . . . . . . . . . . . . . . . . . . . . . . 1D Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2D Overlayers . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

201 202 208

17.5

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

212

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

212

18

Scanning Probe Microscopy Applied to Ferroelectric Materials Oleg Tikhomirov, Massimiliano Labardi, Maria Allegrini . . . . . .

217

18.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

217

18.2

Development of Scanning Probe Techniques for Ferroelectrics . .

217

18.3 18.3.1 18.3.2 18.3.3

Scanning Force Microscopy Non-Contact Mode . . . . . Contact Mode . . . . . . . . Voltage-Modulated SFM . .

220 220 221 222

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Contents – Volume III

XXVII

18.3.4 18.3.5 18.3.6 18.3.7 18.3.8

Resonance Modes of EFM Lateral Force . . . . . . . . Frontal Force . . . . . . . . Second Harmonic . . . . . Tapping Mode . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

224 228 232 233 234

18.4 18.4.1 18.4.2 18.4.3 18.4.4 18.4.5

Scanning Optical Microscopy . . . Pure Optical Microscopy . . . . . . Scanning Electrooptic Microscopy . Near-Field Electrooptic Microscopy Micro-Spectroscopic Techniques . . Second Harmonic Microscopy . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

235 235 237 242 244 245

18.5 18.5.1 18.5.2 18.5.3 18.5.4 18.5.5 18.5.6 18.5.7

Applications to Ferroelectrics . . . . . . Imaging of Domains and Domain Walls Writing Patterns . . . . . . . . . . . . . Phase Transitions . . . . . . . . . . . . Morphotropic Phase Boundary . . . . . Relaxors . . . . . . . . . . . . . . . . . Thin Films . . . . . . . . . . . . . . . . Artiﬁcial Nanostructures . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

247 247 248 249 250 251 251 252

18.6

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

253

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

254

19

. . . . .

. . . . .

. . . . .

. . . . .

Morphological and Tribological Characterization of Rough Surfaces by Atomic Force Microscopy Renato Buzio, Ugo Valbusa . . . . . . . . . . . . . . . . . . . . . .

261

19.1.1 19.1.2 19.1.3

Characterization of Surface Roughness by Atomic Force Microscopy . . . . . . . . . . . . . . . . . . . . . Statistical Methods for Stationary Random Surfaces . . . . . . . . Statistical Methods for Fractal Surfaces . . . . . . . . . . . . . . . Estimation of Morphological Parameters from AFM Topographies

263 264 266 270

19.2 19.2.1 19.2.2 19.2.3

Modeling Contact Mechanics for Rough Surfaces . . Early Phenomenological Contact Theories . . . . . Contact Mechanics Theories for Fractal Roughness . On the Molecular Origins of Amontons’ Law . . . .

. . . .

272 273 277 284

19.3 19.3.1

. . . .

286

19.3.2 19.3.3

Investigations of Multi-Asperity Contacts by AFM . . . . . AFM Characterization of Surface Roughness for Tribological Purposes . . . . . . . . . . . . . . . . . . . Contact Mechanics Investigations at the Nanometer Scale . Contact Mechanics Investigations on the Micrometer Scale

. . . . . . . . . . . .

286 288 291

19.4

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

293

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

294

19.1

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

XXVIII

20

Contents – Volume III

AFM Applications for Contact and Wear Simulation Nikolai K. Myshkin, Mark I. Petrokovets, Alexander V. Kovalev . .

299

20.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

299

20.2

Scale Factor in Tribology . . . . . . . . . . . . . . . . . . . . . . .

299

20.3 20.3.1 20.3.2 20.3.3 20.3.4 20.3.5

AFM as a Tool of Contact Simulation . . . Contact of Rough Surfaces . . . . . . . . . Rough Contact with Adhesion . . . . . . . Multilevel Contact Models . . . . . . . . . Simulation of Contact Using AFM Images Nanomechanical Probing of Soft Layers . .

. . . . . .

300 300 303 307 309 312

20.4 20.4.1 20.4.2

AFM in Wear Simulation . . . . . . . . . . . . . . . . . . . . . . . Nanoscratching and Nanowear with AFM Tip . . . . . . . . . . . . Wear Simulation in AFM Contact Mode . . . . . . . . . . . . . . .

316 317 320

20.5

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

323

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

324

21

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

AFM Applications for Analysis of Fullerene-Like Nanoparticles Lev Rapoport, Armen Verdyan . . . . . . . . . . . . . . . . . . . .

327

21.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

327

21.2 21.2.1 21.2.2

Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . Friction Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . AFM Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . .

328 328 329

21.3

Characterization of Fullerene-Like Solid Lubricant Nanoparticles .

330

21.4

Friction of Solid Lubricant Films . . . . . . . . . . . . . . . . . . .

331

21.5

Friction and Wear of the Surfaces Lubricated with Oil + IF Nanoparticles . . . . . . . . . . . . . . . . . . . . . .

333

21.6

Friction of IF Nanoparticles Under Severe Contact Conditions . .

336

21.7

Mechanisms of Friction of the IF Nanoparticles . . . . . . . . . .

339

21.8

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

341

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

341

22

Scanning Probe Methods in the Magnetic Tape Industry James K. Knudsen . . . . . . . . . . . . . . . . . . . . . . . . . . .

343

22.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

343

22.2

Atomic Force Microscopy . . . . . . . . . . . . . . . . . . . . . .

345

Contents – Volume III

XXIX

22.2.1 22.2.2 22.2.3

Topographic Characterization of the Magnetic Tape . . . . . . . . Topographic Characterization of Heads . . . . . . . . . . . . . . . Tape Roughness Analysis . . . . . . . . . . . . . . . . . . . . . . .

345 349 351

22.3 22.3.1 22.3.2 22.3.3

Magnetic Force Microscopy . . . . . . . . . . . . Methodology . . . . . . . . . . . . . . . . . . . . . Characterization of the Magnetic Tape with MFM Characterization of Heads with MFM . . . . . . .

. . . .

358 358 359 364

22.4

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

367

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

367

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371

Contents – Volume I

Part I

Scanning Probe Microscopy

1

Dynamic Force Microscopy André Schirmeisen, Boris Anczykowski, Harald Fuchs . . . . . . .

3

Interfacial Force Microscopy: Selected Applications Jack E. Houston . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41

Atomic Force Microscopy with Lateral Modulation Volker Scherer, Michael Reinstädtler, Walter Arnold . . . . . . . .

75

Sensor Technology for Scanning Probe Microscopy Egbert Oesterschulze, Rainer Kassing . . . . . . . . . . . . . . . .

117

Tip Characterization for Dimensional Nanometrology John S. Villarrubia . . . . . . . . . . . . . . . . . . . . . . . . . .

147

2 3 4 5

Part II

Characterization

6

Micro/Nanotribology Studies Using Scanning Probe Microscopy Bharat Bhushan . . . . . . . . . . . . . . . . . . . . . . . . . . . .

171

Visualization of Polymer Structures with Atomic Force Microscopy Sergei Magonov . . . . . . . . . . . . . . . . . . . . . . . . . . . .

207

Displacement and Strain Field Measurements from SPM Images Jürgen Keller, Dietmar Vogel, Andreas Schubert, Bernd Michel . .

253

7 8 9

AFM Characterization of Semiconductor Line Edge Roughness Ndubuisi G. Orji, Martha I. Sanchez, Jay Raja, Theodore V. Vorburger . . . . . . . . . . . . . . . . . . . . . . . . . 277

10

Mechanical Properties of Self-Assembled Organic Monolayers: Experimental Techniques and Modeling Approaches Redhouane Henda . . . . . . . . . . . . . . . . . . . . . . . . . . .

303

XXXII

11 12

Contents – Volume I

Micro-Nano Scale Thermal Imaging Using Scanning Probe Microscopy Li Shi, Arun Majumdar . . . . . . . . . . . . . . . . . . . . . . . .

327

The Science of Beauty on a Small Scale. Nanotechnologies Applied to Cosmetic Science Gustavo Luengo, Frédéric Leroy . . . . . . . . . . . . . . . . . . .

363

Part III Industrial Applications 13 14 15

16

SPM Manipulation and Modifications and Their Storage Applications Sumio Hosaka . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

389

Super Density Optical Data Storage by Near-Field Optics Jun Tominaga . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

429

Capacitance Storage Using a Ferroelectric Medium and a Scanning Capacitance Microscope (SCM) Ryoichi Yamamoto . . . . . . . . . . . . . . . . . . . . . . . . . . .

439

Room-Temperature Single-Electron Devices formed by AFM Nano-Oxidation Process Kazuhiko Matsumoto . . . . . . . . . . . . . . . . . . . . . . . . .

459

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

List of Contributors – Volume IV

Chunli Bai National Center for Nanoscience and Technology Beijing 100080, P.R. China e-mail: [email protected] James D. Batteas National Institute of Standards and Technology Surface and Microanalysis Science Division 100 Bureau Drive, Stop 8372, Gaithersburg, MD 20899 e-mail: [email protected] Bharat Bhushan Nanotribology Laboratory for Information Storage and MEMS/NEMS (NLIM) Ohio State University, Columbus, OH 43210, USA e-mail: [email protected] U. Dürig IBM Research GmbH, Zurich Research Laboratory Säumerstrasse 4, CH-8803 Rüschlikon, Switzerland e-mail: [email protected] Fernando García Instituto de Microelectrónica de Madrid, CSIC C/ Isaac Newton 8, 28760, Tres Cantos, Madrid, Spain e-mail: [email protected] Ricardo García Instituto de Microelectrónica de Madrid, CSIC C/ Isaac Newton 8, 28760, Tres Cantos, Madrid, Spain e-mail: [email protected] Jayne C. Garno Department of Chemistry, Louisiana State University 232 Choppin Hall, Baton Rouge, LA 70803 e-mail: [email protected]

XXXIV

List of Contributors – Volume IV

Christoph Gerber National Competence Center of Research in Nanoscale Science, Basel Klingelbergstrasse 82, CH-4056 Basel, Switzerland e-mail: [email protected] B. Gotsmann IBM Research GmbH, Zurich Research Laboratory Säumerstrasse 4, CH-8803 Rüschlikon, Switzerland e-mail: [email protected] Martin Hegner National Competence Center of Research in Nanoscale Science, Basel Klingelbergstrasse 82, CH-4056 Basel, Switzerland e-mail: [email protected] Albena Ivanisevic Purdue University, Weldon School of Biomedical Engineering 500 Central Drive, West Lafayette, Indiana 47907-2022 e-mail: [email protected] Joseph M. Kinsella Purdue University, Department of Biomedical Engineering 500 Central Drive, West Lafayette, Indiana 47907-2022 e-mail: [email protected] Hans Peter Lang National Competence Center of Research in Nanoscale Science, Basel Klingelbergstrasse 82, CH-4056 Basel, Switzerland IBM Zurich Research Laboratory, Säumerstrasse 4, CH-8803 Rüschlikon, Switzerland e-mail: [email protected] Carmen LaTorre Owens Corning, Insulating Systems Business 2790 Columbus Road, Route 16 (Bldg 20-1), Granville, OH 43023, USA e-mail: [email protected] William P. King Woodruff School of Mechanical Engineering, Georgia Institute of Technology 771 Ferst Drive N.W., Atlanta, GA 30332-0405 e-mail: [email protected] Brent A. Nelson Woodruff School of Mechanical Engineering, Georgia Institute of Technology 771 Ferst Drive N.W., Atlanta, GA 30332-0405 e-mail: [email protected]

List of Contributors – Volume IV

Marta Tello Instituto de Microelectrónica de Madrid, CSIC C/ Isaac Newton 8, 28760, Tres Cantos, Madrid, Spain e-mail: [email protected] Chen Wang National Center for Nanoscience and Technology Beijing 100080, P.R. China e-mail: [email protected]

XXXV

List of Contributors – Volume II

Leon Abelmann Systems and Materials for Information storage group MESA + Research Institute P.O. Box 217, 7500 AE Enschede, The Netherlands e-mail: [email protected] Maria Allegrini INFM and Dipartimento di Fisica “Entrico Fermi”, Università di Pisa Largo Bruno Pontecorvo 3, I-56127 Pisa, Italy e-mail: [email protected] Boris Anczykowski nanoAnalytics GmbH, Gievenbecker Weg 11, 48149 Münster, Germany e-mail: [email protected] Piero Castrataro Scuola Superiore Sant’Anna, Polo Sant’Anna Valdera – CRIM Lab Viale Rinaldo Piaggio, 34, 56025 Pontedera (PI), Italy e-mail: [email protected] Bernard Cretin Department LPMO, FEMTO-ST Institute, UMR CNRS 6174 32 avenue de l’Observatoire, 25044 Besançon Cedex, France e-mail: [email protected] Paolo Dario Scuola Superiore Sant’Anna, Polo Sant’Anna Valdera – CRIM Lab Viale Rinaldo Piaggio, 34, 56025 Pontedera (PI), Italy e-mail: [email protected] Harald Fuchs Center for Nanotechnology (CeNTech) and Institute of Physics University of Münster, Gievenbecker Weg 11, 48149 Münster, Germany e-mail: [email protected] Paul Girard LAIN, UMR CNRS 5011, CC 082, Université de Montpellier II Place E. Bataillon, 34095 Montpellier Cedex 5, France e-mail: [email protected]

XXXVIII

List of Contributors – Volume II

Hermann Gruber Institute for Biophysics, J. Kepler University Altenbergerstr. 69, A-4040 Linz, Austria e-mail: [email protected] Pietro Giuseppe Gucciardi CNR-Istituto per i Processi Chimico-Fisici Via La Farina 237, I-98123 Messina, Italy e-mail: [email protected] Peter Hinterdorfer Institute for Biophysics, J. Kepler University Altenbergerstrasse 69, A-4040 Linz, Austria e-mail: [email protected] Rainer Kassing University of Kassel Institute for Microstructure Technologies and Analytics, IMA Technological Physics Heinrich-Plett-Str. 40, D-34132 Kassel, Germany e-mail: [email protected] Yoichi Kawakami Department of Electronic Science, Graduate School of Engineering Kyoto University, Nishikyo-ku, Katsura, 615-8510 Kyoto, Japan e-mail: [email protected] Ferry Kienberger Institute for Biophysics, J. Kepler University Altenbergerstr. 69, A-4040 Linz, Austria e-mail: [email protected] Jürgen Kirschner Max-Planck-Institut für Mikrostrukturphysik Weinberg 2, 06120 Halle, Germany e-mail: [email protected] Andrzej J. Kulik Ecole Polytechnique Fédérale de Lausanne EPFL – IPMC – NN 1015 Lausanne, Switzerland e-mail: Andrzej.Kulik@epﬂ.ch Nicole Lawrence (geb. Schwendler) Technische Universität Kaiserslautern Erwin-Schrödinger Strasse, 67663 Kaiserslautern, Germany e-mail: [email protected]

List of Contributors – Volume II

Małgorzata Lekka The Henryk Niewodniczañski Institute of Nuclear Physics Polish Academy of Sciences Radzikowskiego 152, 31-342 Kraków, Poland e-mail: [email protected] Arianna Menciassi Scuola Superiore Sant’Anna, Polo Sant’Anna Valdera – CRIM Lab Viale Rinaldo Piaggio, 34, 56025 Pontedera (PI), Italy e-mail: [email protected] Ruggero Micheletto Department of Electronic Science, Graduate School of Engineering Kyoto University, Nishikyo-ku, Katsura, 615-8510 Kyoto, Japan e-mail: [email protected] Egbert Oesterschulze Universität Kaiserslautern, Fachbereich Physik Physik und Technologie der Nanostrukturen Erwin-Schrödinger Straße, 67653 Kaiserslautern, Germany e-mail: [email protected] Ute Rabe Fraunhofer Institute for Nondestructive Testing, IZFP, Bldg. 37 D-66123 Saarbrücken, Germany e-mail: [email protected] Vittoria Raffa Scuola Superiore Sant’Anna, Polo Sant’Anna Valdera – CRIM Lab Viale Rinaldo Piaggio, 34, 56025 Pontedera (PI), Italy e-mail: [email protected] Tilman E. Schäffer Center for Nanotechnology (CeNTech) and Institute of Physics University of Münster, Gievenbecker Weg 11, 48149 Münster, Germany e-mail: [email protected] Uta Schlickum Max-Planck-Institut für Mikrostrukturphysik Weinberg 2, 06120 Halle, Germany e-mail: uta.schlickum@epﬂ.ch Martin Stark Ecole Polytechnique Fédérale de Lausanne Institut des Sciences et Ingénierie Chimiques Laboratory of Ultrafast Laser Spectroscopy 1015 Lausanne, Switzerland e-mail: Martin.Stark@epﬂ.ch

XXXIX

XL

List of Contributors – Volume II

Robert W. Stark Ludwig-Maximilians-Universität München and Center for NanoScience (CeNS) Dept. Earth and Environmental Sciences, Section Crystallography Theresienstr. 41, 80333 München, Germany e-mail: [email protected] Alexander N. Titkov Ioffe Physico-Technical Institute, 26 Polytecknicheskaya 194021 St Petersburg, Russia e-mail: [email protected] Pascal Vairac Department LPMO, FEMTO-ST Institute, UMR CNRS 6174 32 avenue de l’Observatoire, 25044 Besançon Cedex, France e-mail: [email protected] Arnout van den Bos Systems and Materials for Information storage group MESA + Research Institute P.O. Box 217, 7500 AE Enschede, The Netherlands e-mail: [email protected] Gunther Wittstock Carl von Ossietzky Universität Oldenburg Carl von Ossietzky Str. 9–11, 26129 Oldenburg, Germany e-mail: [email protected] Wulf Wulfhekel Max-Planck-Institut für Mikrostrukturphysik Weinberg 2, 06120 Halle, Germany e-mail: [email protected] Christiane Ziegler Technische Universität Kaiserslautern Erwin-Schrödinger Strasse, 67663 Kaiserslautern, Germany e-mail: [email protected]

List of Contributors – Volume III

Maria Allegrini INFM and Dipartimento di Fisica “Enrico Fermi”, Università di Pisa Largo Bruno Pontecorvo 3, 56127 Pisa, Italy e-mail: [email protected] Friedrich Aumayr Institut für Allgemeine Physik, Technische Universität Wien Wiedner Hauptstraße 8-10/134, A 1040 Wien, Austria e-mail: [email protected] Bharat Bhushan Nanotribology Laboratory for Information Storage and MEMS/NEMS (NLIM) 650 Ackerman Road, Suite 255, The Ohio State University Columbus, OH 43202-1107, USA e-mail: [email protected] Zachary Burton Shell Global Solutions (US) Inc. 3333 Highway 6 South, Houston, TX 77082-3101, USA e-mail: [email protected] Renato Buzio National Institute for Physics of Matter INFM Via Dodecaneso 33, 16146 Genova, Italy e-mail: buzio@ﬁsica.unige.it Manfred Drack GrAT Center for Appropriate Technology, Technische Universität Wien Wiedner Hauptstraße 8-10/965, A 1040 Wien, Austria e-mail: [email protected] Andreas Ebner Institute for Biophysics, J. Kepler University Altenbergerstr. 69, A-4040 Linz, Austria e-mail: [email protected]

XLII

List of Contributors – Volume III

Friedrich Franek Austrian Center of Competence for Tribology Viktor Kaplan-Straße 2, A 2700 Wiener Neustadt, Austria Institut für Sensor- und Aktuatorsysteme, Technische Universität Wien Floragasse 7/2, A 1040 Wien, Austria e-mail: [email protected] Luca Gavioli INFM and Dipartimento di Matematica e Fisica Università Cattolica del Sacro Cuore via dei Musei 41, I-25121 Brescia, Italy e-mail: [email protected] Ille C. Gebeshuber Austrian Center of Competence for Tribology Viktor Kaplan-Straße 2, A 2700 Wiener Neustadt, Austria Institut für Allgemeine Physik, Technische Universität Wien Wiedner Hauptstraße 8-10/134, A 1040 Wien, Austria e-mail: [email protected] Enrico Gnecco National Center of Competence in Research in Nanoscale Science University of Basel 4056 Basel, Switzerland e-mail: [email protected] Peter Hinterdorfer Institute for Biophysics, J. Kepler University Altenbergerstr. 69, A-4040 Linz, Austria e-mail: [email protected] James K. Knudsen 3328 York Bay, Woodbury, MN 55125 e-mail: [email protected] Alexander V. Kovalev Tribology Department Metal-Polymer Research Institute of Belarus National Academy of Sciences Kirov st. 32A, Gomel, 246652, Belarus e-mail: [email protected] Massimiliano Labardi INFM, Largo Bruno Pontecorvo 3, 56127 Pisa, Italy e-mail: [email protected] Tobias Lange Institute of Physiology II Robert-Koch Str. 27b, 48149 Muenster, Germany e-mail: [email protected]

List of Contributors – Volume III

Ernst Meyer National Center of Competence in Research in Nanoscale Science University of Basel 4056 Basel, Switzerland e-mail: [email protected] Nikolai K. Myshkin Tribology Department Metal-Polymer Research Institute of Belarus National Academy of Sciences Kirov st. 32A, Gomel, 246652, Belarus e-mail address: [email protected] Dessy Nikova Institute of Physiology II Robert-Koch Str. 27b, 48149 Muenster, Germany e-mail: [email protected] Laurent Nony L2MP, Equipe Nanostructuration, Faculté des Sciences de Saint-Jérôme 13397 Marseille, France e-mail: [email protected] Hans Oberleithner Institute of Physiology II Robert-Koch Str. 27b, 48149 Muenster, Germany e-mail: [email protected] René M. Overney Department of Chemical Engineering, University of Washington Box 351750, Seattle, WA 98195-1750, USA e-mail: [email protected] Mark I. Petrokovets Tribology Department Metal-Polymer Research Institute of Belarus National Academy of Sciences Kirov st. 32A, Gomel, 246652, Belarus e-mail: [email protected] Lev Rapoport Department of Science, Holon Academic Institute of Technology 52 Golomb St., Holon 58102, Israel e-mail: [email protected] Massimo Sancrotti INFM and Dipartimento di Matematica e Fisica Università Cattolica del Sacro Cuore via dei Musei 41, I-25121 Brescia, Italy Laboratorio TASC-INFM Strada Statale 14, km. 163.5, Basovizza, I-34012 Trieste, Italy e-mail: [email protected]

XLIII

XLIV

List of Contributors – Volume III

Hermann Schillers Institute of Physiology II Robert-Koch Str. 27b, 48149 Muenster, Germany e-mail: [email protected] Scott Sills Department of Chemical Engineering, University of Washington Box 351750, Seattle, WA 98195-1750, USA e-mail: [email protected] Oleg Tikhomirov Institute of Solid State Physics, Chernogolovka 142432, Russia INFM, Largo Bruno Pontecorvo 3, 56127 Pisa, Italy e-mail: [email protected] Ugo Valbusa National Institute for Physics of Matter INFM and Dipartimento di Fisica dell’Università degli Studi di Genova Via Dodecaneso 33, 16146 Genova, Italy e-mail: valbusa@ﬁsica.unige.it Armen Verdyan Department of Science, Holon Academic Institute of Technology 52 Golomb St., Holon 58102, Israel e-mail: [email protected] Hannspeter Winter Institut für Allgemeine Physik, Technische Universität Wien Wiedner Hauptstraße 8-10/134, A 1040 Wien, Austria e-mail: [email protected]

23 Scanning Probe Lithography for Chemical, Biological and Engineering Applications Joseph M. Kinsella · Albena Ivanisevic

List of Abbreviations AFM SPL STM SAM ODT DPN UHV MHA RH MFN ELP Trx-ELP HIV-1 APTES His-tagged PAH Cys-CPMV LFM NPRW HMDS MPTMS OPA FV MEH-PPV SPAN PPy HRP E-DPN EDOT PAMAM NHS β-CD pz BaFe MFM µCP PDMS

atomic force microscope scanning probe lithography scanning tunneling microscope self-assembled monolayer octadecanethiol dip-pen nanolithography ultrahigh vacuum mercaptohexadecanoic acid relative humidity meniscus force nanografting elastin-like polypeptide thioredoxin-ELP fusion protein human immunodeﬁciency virus type 1 aminopropyltriethoxy silane histidine-tagged poly(allylamine hydrochloride) cow pea mosaic virus genetically modiﬁed with cysteine residues lateral force microscopy nanopen reader and writer hexamethyldisilazane 3 -mercaptopropyltrimethoxysilane octadecylphosphonic acid force volume poly[2-methoxy-5-(2 -ethylhexyloxy)-1,4-phenylenevinylene] self-doped sulfonated polyaniline polypyrrole horseradish peroxidase electrochemical DPN 3,4-ethylenedioxythiophene poly(amidoamine) dendrimer N-hydroxysuccinimide β-cyclodextrin porphyrazines barium hexaferrite magnetic force microscopy microcontact printing polydimethylsiloxane

2

J.M. Kinsella · A. Ivanisevic

23.1 Introduction Taking a brief look back at lithographic and fabrication methods prior to 1959, it is interesting to note that the state of the art transistor in 1958 was 1 cm wide, while conventional photolithography by 2000 has reduced the size of similar structures to ∼ 100 nm [1]. Conventional methods of fabrication are estimated to peak around 2010 when feature sizes will be less than 25 nm. To achieve devices with these dimensions alternative techniques will have to be employed, including E-beam and ion beam lithography, extreme ultraviolet photolithography, mask phasing and others. Many alternatives to these techniques have been developed based on scanning probe lithography (SPL) methods that may soon prove to be viable production methods. This chapter will focus solely on such scanning probe techniques and applications thereof. In particular, we will explore the current state of research involved with employing these methods for biological and chemical modiﬁcations of surfaces on the nanoscale. The ﬁrst reports of lithography using scanning probe methods date back to 1987 at the IBM Almaden labs, where researchers discovered the ability to position single atoms using STM [2–4]. These proof-of-concept experiments were only possible under ultrahigh vacuums, were done at very low temperatures and were painstakingly slow. Shortly after these experiments, researchers showed that STM could be used as a means to physically remove areas of octadecanthiolate SAMs from a gold surface by repeated scanning under ambient conditions [5]. Throughout the early to mid 1990s several groups explored techniques that involved the removal or modiﬁcations of SAMs using scanning probe tips [6–8]. These techniques will be expounded upon in subsequent sections of this chapter; it is also worth noting at this juncture that several thorough reviews of self-assembled monolayers have been written [9–11]. In 1995, the ﬁrst experiments were performed showing that physisorbed organic molecules on an atomic force microscope probe could be transferred to an underlying substrate [12]. The report showed that ODT (HS–(CH2 )17 –CH3 ) could be transferred to mica by scanning at very low forces and that the resulting patterns were stable and of a homogenous height proﬁle. In 1999, the potential of this technique was stretched further than previously believed possible when controllable patterning of thiols on gold was achieved [13]. The importance of this report was in the control of ink deposition and the quality of the SAMs formed. The researchers showed that line widths as small as 30 nm and dots with diameters as small as 460 nm could be reproducibly formed. The group immediately noticed the parallels between this technique and the long known dip-pen writing technique and, therefore, appropriately named their method dip-pen nanolithography, Fig. 23.1. DPN is an emerging technique that offers a means of fabrication in which molecular level control exists. One key feature that makes DPN such an attractive technique is its inherent versatility in selecting ink molecules. In principle, given the proper substrate, any chosen molecule can be either directly or indirectly bound to the surface in a site speciﬁc fashion. The present range of compatible inks varies from very large biomolecules, such as proteins, to small hydrocarbons. The range of substrates that have been shown to be compatible with this method to date includes metals, semiconductors, insulators and mica. The direct patterning of many

23 Chem, Bio, and Eng Applications of SPL

3

Fig. 23.1. A schematic representation of the DPN procedure. An AFM coated with physisorbed molecules is brought into contact with the substrate. Once in close proximity with the surface, a water meniscus is formed between the tip and the sample through capillary action. The water meniscus acts as a localized reaction vessel and is used to transport the physisorbed ink onto to the surface. The writing direction of the tip, as well as other environmental factors, is used to control the formation of the pattern generated. Figure reprinted with permission from [13]

of these materials was not previously possible using standard lithographic techniques. Photolithography, while of great importance in the semiconductor industry, is of little use for patterning of organic or biological molecules due to the destructive nature of the UV light used for exposure of resists. Soft lithography methods such as microcontact printing are also compatible with various molecular inks. Currently the resolution of soft lithography methods is limited due to their dependence on conventional lithography or micromachining to produce the master stamp. This chapter highlights the importance of dip-pen nanolithography and related procedures for the fabrication of both traditional and non-traditional structures on the nanoscale. The chapter begins by presenting the fundamental mechanisms that control the transfer of ink. Theoretical studies describe the role of several parameters that are important for the patterning process: humidity, temperature, type of ink, and surface choice. The rest of this chapter is separated according to the types of inks used and the applications associated with them. Biological inks are currently being extensively studied in nanoscience applications and research, and are described ﬁrst. These inks can be utilized in nanoscience as a method to fabricate templates for higher order bottom-up assemblies. Biology and medicine itself can greatly beneﬁt from techniques such as DPN, because it provides a fabrication method capable of creating high-throughput nanoarrays and templates. Such patterned surfaces can be used to study fundamental problems associated with biological or biochemical interactions. In subsequent sections, the use of chemical inks is reviewed. We describe in detail the various types of chemicals that have already been patterned. Many of the inks that have been used to date are already used in commercial applications on the bulk scale. The engineering aspects of DPN are also presented in the concluding section. Many of the engineering applications described are possible by choosing from the

4

J.M. Kinsella · A. Ivanisevic

materials that have already been used as inks. We summarize some of the advances in microfabrication of cantilevers that have transformed the scanning probe lithography methods from serial technologies to parallel ones.

23.2 Modeling of the DPN Process The interest in the use of DPN for nanofabrication has spurred many researchers to determine a fundamental understanding of the mechanism involved in the transfer of ink from the AFM tip. In the initial DPN report, variables such as relative humidity, scan speed, scan size and time were considered to be the main factors in the transfer process. The true method itself is still not completely understood and highly debated, although it is agreed that the ultimate potential of the technique depends on a quantitative understanding of the ink transport mechanism. We will brieﬂy attempt to objectively describe the work done by many researchers on this subject in an effort to describe the fundamental science and engineering that is involved in DPN, as well as to convey the complexity of the process from a scientiﬁc standpoint. The ﬁrst quantitative efforts made to understand the transport properties of DPN were reported in 2001 [14]. Their intent was to model and experimentally determine the diffusion and adsorption of SAM patterns using DPN. The model was based on the view that the molecules diffused down a concentration gradient from a high density region around the area of the tip to a lower density region that consists of the unoccupied binding sites of the substrate directly outside the patterned regions. This model assumes that the physisorbed ink molecules on the tip act as a diffusion point source when in static contact with a two-dimensional substrate. This model also assumes that the molecules diffuse out onto a region in which all of the available binding sites are occupied from previously deposited molecules. Furthermore, the molecules are assumed to continue to travel outward until they ﬁnd an unoccupied site on the substrate and irreversibly bind to it. With these assumptions, the experiments model the dynamics of the self-assembly of the ink as a random walk on the substrate with the ﬂow of ink as a constant ﬂux. The results of the simulations showed that when the tip remained ﬁxed the quality of the diffused SAM that formed was dependent on the number of molecules deposited. The rate of deposition during the simulation proved to have no effect when the same number of molecules was deposited. The circularity of the patterns formed however was found to increase when the number of deposited molecules was increased. When modeling patterns that were formed by a moving tip it was determined that there is a signiﬁcant dependence between the tip speed and the deposition rate. In a follow up study the role of the meniscus formation on the transport properties of the ink was examined [15]. Monte Carlo simulations were employed to examine the behavior of the meniscus for various geometries and energies. A variety of surfaces are assumed to all be completely wettable at any temperature. Also, the tip is assumed to be an ellipse and the water is modeled as a two-dimensional lattice gas that is conﬁned between the tip and the substrate. Some of the variables that were examined include the wettability of the tip, which is highly dependent on the

23 Chem, Bio, and Eng Applications of SPL

5

wettability of the ink physisorbed to it, the radius of curvature of the tip, the distance of the tip from the sample surface, as well as the interaction strength between both ﬂuid and tip, and ﬂuid and surface. The results of this simulation showed that the formation of the meniscus is thermodynamically controlled and the width of the meniscus can decrease in a number of situations, including increased tip curvature, increased separation of the tip-substrate and decreased wettability of the tip. The degree of saturation was also shown to be a critical component of the width of the meniscus and the simulation had shown that it can be a tunable property over a very large range. This is important because the limit of resolution of DPN patterns will depend on the minimum possible width for which a stable meniscus can be formed. The simulation showed that the smallest possible meniscus that could be formed would be ∼ 2 nm, but that such a small meniscus would be prone to instabilities, thereby inferring that the actual limit of resolution for any reproducible large scale pattern would have to be larger than this. Also in 2002, the deposition of ODT was reported in waterless environmental conditions using DPN [16]. The radius of the ODT pattern was experimentally measured as a function of contact time and relative humidity. It was also assumed in this work that the tip will act as an inﬁnite reservoir of ink, which implies a constant concentration, not ﬂux. Modeling the tip as a constant concentration allows for the extraction of kinetic data. It is also assumed that the diffusion rate is independent of the concentration and would be Fickian. The ﬁnal assumption made was that, when the tip is removed from the surface, diffusion will cease at all areas except the periphery of the pattern. The researchers obtained data on deposition as a function of time. The data was ﬁt to a radial diffusion model and allowed one to determine the surface diffusion coefﬁcient of ODT onto clean gold to be 8400±2300 nm2 s−1 . Prior models based on constant ﬂuxes indicated that the radius would increase with t 1/2 . The data reported by Sheehan et al. did not ﬁt these models. Humidity experiments were performed in air and under a nitrogen environment to precisely control the saturation of water on the substrate and, therefore, the ability of the meniscus to be formed during deposition. Repeated experimentation showed that the deposition of the ODT to the gold surface was independent of humidity and that the meniscus was likely to have no role in the ink transfer. Experiments performed on atomically ﬂat NaCl substrates have given further insight on the formation of the water meniscus between the AFM tip and the sample surface at different relative humidity conditions [17]. The meniscus formation was studied under humidity conditions ranging from 70% to 0%. An ultrahigh vacuum (UHV) chamber was used and the humidity dependence utilizing either hydrophobic or hydrophilic AFM contact mode tips was recorded. Tips were made hydrophilic by etching them with H2 SO4 and a H2 O2 solution. The hydrophobic tips were prepared by adsorbing ODT or dodecylamine on the tip surface. Theoretically, the hydrophilic tips should increase the meniscus size due to the large capillary forces between the tip and sample. The hydrophobic tips should minimize these forces, thus reducing or eliminating the meniscus formation. When a meniscus forms between the tip and NaCl substrate, the water meniscus causes the NaCl to dissolve and leaves a pit on the surface. Using the hydrophilic tips, it was noted that the NaCl substrate was etched by the formation of a water meniscus under all relative humidity conditions, excluding those under UHV. Under RH conditions exceeding 60%, it was noted that

6

J.M. Kinsella · A. Ivanisevic

the substrate could not be etched due to the large water layer that extended across the entirety of the NaCl surface. The adhesion forces of the tip to the NaCl substrate were measured as a function of the atmospheric water conditions as well. This experiment was done to exclude the possibility that the pit formation is induced by a mechanical deformation of the sample during the experiments. It was noted that the adhesion force decreased with decreased humidity conditions. This result indicates that the capillary forces were negligible at low RH values. When the hydrophobic tips were studied, it was found that the substrate would still be dissolved under low RH values. In these cases the pits formed were not always circular. This observation was attributed to the variable meniscus shapes and sizes formed when coated tips were used. The study concluded that unless the atmospheric water conditions are controlled using a UHV chamber, a water meniscus will always form to some degree between the tip and the sample. Similar experiments have been performed studying the transport phenomenon with respect to the relative humidity and temperature using ODT and MHA (SH–(CH2 )15 –COOH) [18]. The humidity dependence of the molecular transport of ODT to the surface corroborates the previous ﬁndings that there is no dependence on the transport for ODT and minor humidity dependence for MHA. The qualitative effects of a water vapor environment versus an ethanol environment were also studied to determine whether the ink-meniscus solubility would alter the deposition. It was observed that lines and dots of ODT in an ethanolic environment had poorer edge resolution compared with patterns formed in comparable water vapor environments. From the results of these experiments, it was concluded that the transport mechanism is heavily dependent on the solubility of the ink. With hydrophobic inks such as ODT and MHA, the dependence on humidity is negligible, but for ionic salts and highly charged ink there remains a strong dependency on the presence of water vapor to form a meniscus between the tip and substrate. In the transfer of the nonpolar inks it was determined that the transport was controlled by a simple surface diffusion mechanism in which a dependence on thermal energy exists. A recent report highlights the temperature dependence found in patterning of ODT and MHA inks [19]. Patterns were formed nine times at nine different temperatures ranging from 22–33 ◦ C. Holding all other variables constant and increasing the temperature systematically led to an increase in the dot diameter of the MHA and ODT patterns. It was established that an increase in temperature would affect a number of important variables, including the number of molecules solvated, the diffusion rate across the meniscus, and the adsorbate diffusion onto the surface. Similarly, the effect of the relative humidity (RH) was studied by varying the RH between 5–100%. Again nine humidity conditions were studied by depositing nine dot patterns of each molecule. The role that humidity played in the transfer of the two molecules differed. For ODT increasing the RH and maintaining a constant temperature caused a decrease in the dot sizes. The opposite effect was seen when MHA was patterned at increased humidity and ﬁxed temperatures. From this data it was concluded that the formation of a larger meniscus at elevated humidity acts as a blocking layer for the insoluble ODT and a transport layer for the more soluble MHA. These experiments were in agreement with the previous work on the effects of humidity and meniscus formation for these two inks.

23 Chem, Bio, and Eng Applications of SPL

7

23.3 Patterning of Biological and Biologically Active Molecules Several exciting discoveries have recently been made in the use of DPN for patterning biological molecules, viruses and bacteria. Research in the life sciences and medicine has already been revolutionized by the development of microarrays and assays for diagnostic screening. Microarrays are devices typically containing patterns of differing proteins or DNA strands that react speciﬁcally towards complementary oligonucleotide strands, short organic markers, ligands or speciﬁc antibodies in solution. These devices are widely used for research, clinical diagnostic testing, screening and proﬁling. A typical microarray is formed by robotic spotting or photolithographic techniques, which limit their minimum feature size to ∼ 20 µm [20]. In comparison, using DPN to pattern the molecule of interest on a surface with even poor resolution capabilities of hundreds of nanometers would increase the areal density of arrays by a 10,000 to 100,000 fold [21]. With these capabilities, the realization of whole genome single chip screening can not only be achieved, but can be done using a chip that is only centimeters in size. In comparison a present day whole genome chip hindered by current resolution limits would be on the orders of meters. Another value of shrinking such devices lies in the reduced volume of analyte necessary for proper analysis. With a reduced volume of analyte the time of analysis is also decreased, which offers the possibility of on the spot detection for both clinical and research purposes. Time to purify and/or amplify the analyte would not be necessary in such a high-throughput device. In addition to decreasing analyte concentration and increasing testing time, the value of shrinking the microarray to dimensions that are on the same order of magnitude as biological molecules lies in providing a way to study the interactions of several important biochemical reactions in situ. This includes the interactions of receptor-ligand and whole cell-ligand, DNA–DNA, DNA–RNA, RNA–RNA, peptide-RNA and several other important interactions. The study of these interactions on an artiﬁcial surface that properly mimics their in vitro state will give researchers much insight on the fundamental mechanisms that control such reactions. Knowledge of these mechanisms is not only important from a fundamental biological or biomedical standpoint, but also for a number of applications that rely on speciﬁc interactions. Proof-of-concept experiments have shown that patterning of several biologically important molecules on a single chip is possible, although there is still much work to be done on detection methods. AFM based detection is a slow and tedious process at present, but it can be performed in situ allowing brilliant insight on the basic mechanisms of interactions. Integrated microelectronics are also being explored in a rapidly expanding ﬁeld as a characterization method that offers very sensitive detection capabilities. Detection methods lie beyond the scope of this section. Here we will focus on the current state of patterning, surface chemistries and the possibilities offered by DPN with regard to patterning biomolecules. Some examples of molecules that have already been patterned using DPN include: DNA oligonucleotides, proteins and peptides. These molecules have been written both directly and indirectly to a number of surfaces, including silicon ox-

8

J.M. Kinsella · A. Ivanisevic

ide, nickel oxide, gold and mica. The success of direct writing approaches was demonstrated using thiolated DNA or cysteine residue containing proteins and peptides on gold surfaces. Indirect approaches can rely on surface modiﬁcations prior to writing and can be used very efﬁciently for the speciﬁc binding of the molecule to the modiﬁed surface. Indirect methods can be used on any properly modiﬁed surface and offer enormous potential in the modiﬁcation of surfaces that prove difﬁcult in direct writing. An example of the indirect patterning of proteins that is typical in standard protein arrays is the modiﬁcation of glass or oxidized silicon with an aldehyde terminus and subsequent interaction of the amine terminus of the protein or a lysine side chain with the aldehyde via a hydrolysis reaction. One drawback to the indirect method of patterning is that typically only one type of biomolecule can be patterned. To overcome this, multiple steps involving patterning and adsorption can be employed, although the process is timeconsuming. 23.3.1 DNA Patterning It has been shown that thiolated oligonucleotides and 5 terminal acrylamide functionalized oligonucleotides can be patterned via DPN onto gold and silica surfaces, respectively [22]. This report not only was groundbreaking in that it was the ﬁrst to use DPN for the direct-write fabrication of biomolecular recognition arrays, but it was also one of the earliest reports of using DPN on a substrate other than gold. The report also notes that using standard silicon nitride tips results in poor patterning. To overcome this limitation, the tips were initially coated with 3-aminopropyltrimethoxy silane in order to make their surface more wettable. The wettable tip surface was more favorable for the physisorption of DNA ink molecules and contributed to more evenly distributed patterning on the surface. To test that the patterned oligonucleotide sequences retained their biological recognition, the patterned surfaces were treated with oligonucleotide modiﬁed gold nanoparticles. It was observed that complementary Watson–Crick base pairing bound the particles to sites were the oligonucleotides were patterned exclusively. Multiple DNA types were also patterned and incubated with their complementary strands, which were modiﬁed with ﬂuorophores for detection purposes. The results of these experiments showed that the binding of the complementary strands was speciﬁc and that even after rinsing the ﬂuorophore modiﬁed complementary strand off of the substrate with water, the patterned spots remained and further testing revealed that they had retained their activity. Oligonucleotides immobilized onto surfaces can also serve as templates to organize the assembly of multidimensional and multicomponent structures using solution phase bottom-up assembly [23]. This was shown by patterning DNA indirectly through the use of an MHA ink and subsequent coupling of the DNA to the MHA pattern using alkylamine-modiﬁed oligonucleotide strands. To generate patterns of different sequences of DNA on the surface a multistep fabrication scheme is necessary due to the indirect nature of this approach. To do so, patterning of MHA and subsequent adsorption of the DNA was performed using different DNA strands each time and repeated as many times as necessary. Once the desired numbers of different DNA strands were immobilized, they were then used as templates to bind materials

23 Chem, Bio, and Eng Applications of SPL

9

that contain the complementary sequences in a speciﬁc manner. In this experiment, two different DNA strands were chosen and their complementary strands were immobilized onto either 13 nm or 30 nm gold nanoparticles. When the nanoparticles were allowed to interact with the surface, it was noted that the interaction of the surface immobilized strands and the particle bound complementary strands would guide the assembly of the structures in a controlled fashion. Other scanning probe lithography techniques have been used to immobilize DNA onto surfaces. MFN is a method that employs the removal of a layer of molecular resist on a surface through the shear force provided by the surface tension of a thin ﬁlm of the solution present during scanning that is to be patterned [24]. This technique is similar to nanografting, where a high contact force is applied to remove a monolayer from a substrate [25]. Since the force applied to remove the resist in MNF is the surface tension supplied by the inking solution, the need for AFM feedback control is unnecessary and the cantilever can scan large areas very rapidly while removing the resist. It was also noted that the surface tension created by an aqueous thin ﬁlm on the surface signiﬁcantly exceeded the force that is typically applied by AFM feedback controls. Depending on the tip radius, the wettability of the surface and the volume of ﬂuid in the ﬁlm, the applied force due to the surface tension of the thin ﬁlm could approach micronewton ranges. Typical conditions for the applied force in contact imaging, DPN or traditional nanografting are only in the nanonewton range. These large forces provided pressures that exceeded the yield strength of the underlying gold substrate and, therefore, would mechanically ﬂatten the gold area that was patterned by MFN. This technique was successfully employed to pattern single strands of DNA onto gold substrates. Line patterns were formed with scan rates ranging from 20–160 µm/s. The patterning of the DNA occurred simultaneously with the grafting due to the presence of DNA in the solution used to provide the surface tension. The lithographically deﬁned DNA on the surface was incubated with different sizes of nanoparticles containing complementary strands to those patterned to verify its viability. Another process that closely resembles both DPN and robotic stamping methods has been developed using an array of silicon based microcantilevers to mechanically spot DNA and proteins on glass surfaces [26]. The theory for the technique relies on the same principles as DPN, although the array of cantilevers is designed with microreservoirs, microchannels and aluminum electrodes that can be used to actively control the height of the liquid rise when the arrays are dipped into an inking solution. This increases the liquid loading efﬁciency of the tips and allows for the tips to be cleanable unlike conventional coated tips in DPN experiments. To create a microarray, the liquids to be spotted are placed on the surface and the cantilever array is brought into proximity. The cantilevers are then dipped into the solutions and a small electric ﬁeld is applied for a short time. The applied voltages induce electrowetting on the surface of the cantilevers due to the changes in surface charge distribution, thereby forcing the liquid to spread across the cantilever and into the embedded microreservoirs. Not only were these cantilever arrays successful in the patterning of DNA, but they also were shown to be able to fabricate protein microarrays. The size of the features spotted was in the micron range. With optimization of the cantilever design, contact force and contact time this technique may contend with DPN in terms of feature size.

10

J.M. Kinsella · A. Ivanisevic

23.3.2 Protein Patterning Proteins offer a challenge to patterning by DPN based on their relatively large size, high molecular weights and intricate three-dimensional structure. The typical protein patterned for diagnostic microarrays is the antibody. Antibodies are Y shaped proteins that have average dimensions of ∼5–15 nm. The conformation of the antibody on the surface is also critical because the top two chains of the molecule (light chains) contain the variable regions and determine the speciﬁc biorecognition properties. If these regions of the protein are not exposed to the analyte, then subsequent interaction is severely hindered or rendered impossible. Because of the importance of the higher order structure of the protein with respect to its recognition abilities, the transfer process of the molecule to the surface must be done in a way that will not cause the protein to denature. Intact cytoplasmic proteins at physiological pH’s are often highly charged molecules with hydrophilic amino acids exposed towards the outside to solvate the molecule and make it thermodynamically stable. Adsorption can, therefore, occur through electrostatic interactions between the substrate and the protein or through covalent interactions of a modiﬁed substrate and an exposed amino acid of the protein. Other proteins may contain a combination of both hydrophilic and hydrophobic residues on their surfaces. Such proteins may diffuse at different rates or interact differently with surfaces. In the theory section, we addressed the concept of simple surface diffusion of the inking molecules from the tip onto the underlying substrate. The models generated were ﬁt to data generated by inks that were short organic molecules such as ODT, MHA or ionic salts. The composition of proteins and their interactions with surfaces differ greatly when compared to these inks. Several considerations regarding these limitations need to be taken into account when designing DPN experiments with protein inks. In the simple diffusion model, the molecules that were modeled diffused outward from the tip until they bound onto the next available binding site. The diffusion is impeded in the patterning of proteins due to their high molecular weights according to classic Fickian diffusion. Steric factors and/or electrostatic repulsions between adsorbed proteins and those being deposited may create difﬁculty in the ability to generate controlled patterns. The physisorption of the protein onto the tip is also a problem that may follow a mechanism very different to that of smaller hydrophobic molecules. Proteins are susceptible to denaturation upon physisorption onto standard AFM cantilevers by the surface forces present. The minimization of denaturation, however, is imperative when attempting to fabricate an operable protein array. To overcome the possibility of denaturing of the protein, modiﬁcations of the tip itself have been employed. Coating of the tip with a low surface energy SAM prior to inking has proven to be successful. Several different types of SAMs can be employed to coat the tip as long as they are both hydrophilic and biocompatible. Formation of the meniscus can also be increased when a more wettable tip is used, resulting in more reliable patterning. Typical protein patterning experiments are also performed at a heightened humidity to further enhance formation of the water meniscus. The initial work indicating that DPN could be a viable method to generate protein patterns was performed by writing thiolated collagen and collagen-like peptides on

23 Chem, Bio, and Eng Applications of SPL

11

a gold surface [27]. The spatial control over the collagen patterns generated ranged from ∼800 nm line widths to 30–40 nm line widths. At smaller line widths the level of organization of the patterned collagen was not observed to match that typically found in native collagen structures. The helical architecture of native collagen was easily observed for larger pattern sizes. Bulk deposition of the inking solution onto the surface failed to generate the higher order structures, which led researchers to conclude that the actual mechanism of formation of the triple helices found in native collagen was facilitated by the writing method. Reactions with collagen-speciﬁc primary antibodies were employed to test the biological activity of the patterned collagen. The patterned structures interacted with the antibody, indicating their retention of biological function. Experimental work has also been done to show that antibodies containing a cysteine residue, which contains a thiolated side chain, can be patterned onto gold under high humidity conditions [28]. Other work has been done patterning proteins onto glass, metals and oxidized silicon. In the case of the gold substrate, the patterned antibodies retained biological activity and interacted strongly with their appropriate antigens. Patterned antibodies on aldehyde modiﬁed silicon oxide and on negatively charged oxidized silicon were also shown to retain their activity even when the patterned area was as small as 55 nm diameter dots [29], Fig. 23.2. Experiments on silicon are important because this is the surface of choice in both conventional array technologies and in microelectronics. Nickel surfaces are also often used in biological arrays due to the strong afﬁnity for histidine residues. This would imply that in principle DPN can be used to directly write proteins or polypeptides that contain histidine residues in their amino acid sequences onto nickel surfaces. N-terminal histidine-tagged ubiquitin and thioredoxin protein arrays have been fabricated on nickel oxide surfaces using this interaction [30]. Feature sizes as small as 80 nm and as large as 500 nm were constructed. The activity of the patterned proteins was shown to be maintained after writing by staining against ﬂuorescent antibodies. Protein nanoarrays have also been formed via the indirect assembly of proteins using DPN patterning of templates to guide the assembly of the proteins onto the substrate [31]. Many proteins will have a high afﬁnity for carboxylic acids at physiological pH levels. Patterning of MHA onto gold using DPN has been used as an indirect method to immobilize a variety of proteins under these conditions. The patterning of MHA is simple and well-deﬁned in comparison to the direct patterning of proteins. The inability to immobilize a number of differing proteins using an indirect method, without several cycles of subsequent patterning followed by immobilization, is a severe limitation to the effectiveness of this method. Examples of indirect patterning using MHA include immobilizing IgG [31, 32], Lysozyme [31], thioredoxin-ELP fusion proteins [33] and modiﬁed monoclonal antibodies against HIV-1 [34] onto gold substrates. By passivating the bare gold surface after MHA patterning, the non-speciﬁc binding of these proteins to the unpatterned areas of the surface can be minimized. The height proﬁles of MHA patterns after adsorption of the proteins showed that one or two protein molecules would adsorb to each individual MHA spot. Importantly, the bound proteins retained their activity once adsorbed. Cell adhesion was also studied by adsorbing retronectin onto the MHA patterns. Retronectin is a protein that can cause certain cell types to adhere via

12

J.M. Kinsella · A. Ivanisevic

Fig. 23.2. (a–c) Antirabbit IgG proteins patterned on negatively charged silicon oxide surfaces using DPN imaged by ﬂuorescent microscopy. The antirabbit IgG proteins were labeled using Alexa 594 as a ﬂuorescent marker. (d) Protein patterns formed on aldehyde modiﬁed surfaces of antirabbit IgG. The upper line consists of dot patterns of the antirabbit IgG stained with Alexa 594. The lower line is dot patterns of antihuman IgG stained by Alexa 488. This demonstrates the ability of DPN to generate patterns of multiple inks on the same surface. The functionality of the deposited proteins is tested against speciﬁc probes in this reference demonstrating the potential of DPN in the functionalization of surfaces for protein nanoarray manufacturing. Figure reprinted with permission from [29]

the speciﬁc interaction of its RGD sequence with cytoplasmic integrins on the cellular membrane. It was found that the cells would attach to the patterns of retronectin almost exclusively in comparison to the bare gold surface. The morphology of the cells attached to retronectin differed from those that were attached to the gold as well. Such ﬁndings indicate that the use of DPN patterned proteins can be useful not only for diagnostic and biosensing applications, but also in cellular studies and biomaterials. The indirect patterning method used to capture the thioredoxin-ELP fusion protein (trx-ELP) relied on the patterning of MHA using DPN and the subsequent immobilization of elastin-like polypeptide (ELP) sequences onto the patterned MHA [33]. The surface bound ELP could be used to capture or release the trx-ELP depending on the environmental conditions, such as the ionic strength of the buffer or temperature. This use of indirect patterning and biorecognition between the ELP polypeptide and the trx-ELP is an example of higher order patterning in the nanoscale using biologically inspired bottom-up assembly procedures. Indirect capture of proteins has also been achieved using the patterning of MHA followed by the binding of biotin, streptavidin and then biotinylated proteins [35]. This recognition-based assembly of proteins onto gold relies on the small-molecule ligand biotin that shows a very high afﬁnity for the streptavidin molecule. The

23 Chem, Bio, and Eng Applications of SPL

13

biotin-streptavidin assembly procedure is important due to the ubiquity of existing biotin-tagged molecules. The high afﬁnity of the molecules permits selectivity of the patterning that results in almost no non-speciﬁc binding when non-fouling SAM is used to passivate the underlying substrate. 23.3.3 Peptide Patterning Peptides have been of much interest to nanoscience research recently. Solid state synthesis techniques have made the production of peptides cheap and efﬁcient. As proteomics and enzymology uncover the active sites of large proteins, it is typically discovered that the function relies on only a fraction of the amino acids that make up the entire protein. These amino acid sequences can be reproduced in the form of peptides and used for catalytic, recognition or sensing purposes. Like for larger proteins the conformation of peptides is critical in their function and its integrity must be preserved. Many researchers have modiﬁed bulk surfaces using peptides and have shown that upon monolayer formation the peptides remain active [36–38]. An example of this would be the modiﬁcation of gold using an RGD peptide [37]. The RGD sequence is one of many that have been shown to actively attach integrin proteins found on cell membranes to promote adhesion of speciﬁc cell types. Such work is important in biomaterial science or tissue engineering applications were implants may be engineered to actively attach cells onto their surface in order to depress immune responses and/or integration of the implant with surrounding tissues. A class of TAT peptides derived from the TAT protein has been patterned onto SAMs modiﬁed SiOx surfaces using DPN [39], Fig. 23.3. The TAT protein has been isolated from the human immunodeﬁciency virus type 1 (HIV-1) and shown to function in cell permeation and viral replication. The TAT peptide that was patterned

Fig. 23.3. AFM images (a) height and (b) lateral force microscopy (LFM) of DPN generated lines of TAT peptide on SMPB modiﬁed silicon oxide surfaces. Figure reproduced with permission from [39]

14

J.M. Kinsella · A. Ivanisevic

has been shown to interact with speciﬁc sequences of RNA. The modiﬁcation of the bulk SiOx surface prior to patterning included silanization of the surface using aminopropyltriethoxy silane (APTES) and subsequent coupling of a heterobifunctional cross-linker SMPB. The TAT peptide covalently binds to the cross-linker through a terminal cysteine residue that contains a thiol group. To successfully write the peptide onto the modiﬁed surface it was important to maintain high humidity conditions and apply minimum force with the tip so that the underlying bulk surface modiﬁcations would not be affected. The covalently bound peptide was shown to retain its recognition properties when incubated with a speciﬁc RNA sequence. Directed immobilization of peptides via the thiol bond of a cysteine residue onto gold surfaces has also been achieved [40]. Contrary to most studies using DPN, the deposition of these peptides was performed using tapping-mode AFM. Tappingmode AFM is typically used in the imaging of biological or “soft” surfaces in order to minimize damage to the adsorbed molecules. One of the key factors to depositing the peptides on the surface using tapping-mode DPN was to increase the drive amplitude of the probe to 2–5 times the value typically used for imaging purposes, scanning at slower speeds (0.2–1 Hz). Even at the elevated drive amplitudes used for the patterning, the force between the tip and the sample is still considerably less than the tip–sample forces found in contact mode imaging or deposition. The amplitude of the coated tapping-mode cantilever was found to be dampened by the presence of the physisorbed peptides when compared to the amplitude of the bare tapping-mode cantilever. The veriﬁcation of the transfer was shown using line scans acquired through tapping-mode images to determine if the height of the generated pattern was higher than the surrounding non-patterned areas. The immunolabeling of the peptides after deposition failed to show any ﬂuorescent signal indicating an inactivity of the adsorbed peptide. This may be attributed to the quenching of the ﬂuorescence by the gold background or the conformation of the patterned peptides. A follow-up study using an electrochemical variation on the DPN method (E-DPN) was performed in tapping-mode to pattern both His-tagged proteins and peptides onto a nickel surface [41]. The electrochemical dip-pen nanolithography resembles the typical DPN technique although the transfer of the ink is directed by the presence of an externally applied electric ﬁeld. A negative bias was applied at the tip during patterning and the nickel surface was held at ground potential. This causes the nickel surface to ionize once the meniscus is formed between the tip and the sample. The choice of the histidine-tagged proteins and peptide was made due to their strong afﬁnity towards Ni2+ , which was formed at the surface during the ionization process. The potential was held between −10 and +10 volts, the drive amplitude was increased to 5–10 times the value that would be used for imaging and slow scan speeds (∼ 0.5 Hz) were employed. When no voltage was applied it was noted that no deposition had occurred. This was due to the lack of nickel ions on the surface for the deposited peptides to bind to. The optimal tip voltage for the controlled transfer of the peptides was determined to be in the range of −2 to −3 V. A low positive bias did not induce the ionization of the nickel, therefore, no deposition was observed. The optimal transfer of the proteins also occurred when the tip voltage was held at −2 V. The patterned proteins displayed ﬂuorescence once tagged, indicating their

23 Chem, Bio, and Eng Applications of SPL

15

viability once immobilized. Height scans of the peptides indicated that the peptide was in an upright position with the terminal histidine residue bound to the nickel surface. 23.3.4 Patterning of Templates for Biological Bottom-Up Assembly Bottom-up assembly is a highly researched area of nanotechnology and science. Many of the roots of self-assembly mechanisms used in nanoscience have origins or inspirations based on mechanisms found in natural in biological processes. Biological self-assembly often relies on non-covalent interactions such as hydrogen bonding, hydrophobic/hydrophilic interactions, van der Waals forces and electrostatic interactions in the building of three-dimensional structures. Understanding these mechanisms has allowed the patterning of materials that interact using these forces to selectively guide the assembly or the placement of biological molecules. A basic example of this was seen in the development of arrays using both DNA and proteins. In the case of DNA the H-bonding base pairs of the complementary oligonucleotides in a solution will interact speciﬁcally with the strands patterned on the surface. For proteins, the antibodies interact through these secondary forces to tightly bind speciﬁc antigens. Although these structures are complex, their interactions can be exploited on surfaces or in solution to fabricate three-dimensional architectures. Biological macromolecules and biological entities such as bacteria and virions can also be indirectly patterned using non-covalent interactions. Patterning of polyelectrolytes, which will be explained in further detail later, can create speciﬁc areas on the substrate that are highly charged. Charged areas on the surface can be used in the electrostatic assembly of oppositely charged biological molecules. As an example, PAH can be utilized to impart a pattern of positive charge. DNA bears a net negative charge at physiological pH due to the phosphate backbone. When stretched onto a negatively charged hydrophilic surface that contains positive PAH patterns, the placement of the DNA will be guided by electrostatic interactions [42], Fig. 23.4. Much work has also been done using DNA itself as a template for the bottom up assembly of metals, polyelectrolytes and nanoparticles using electrostatic interactions. Another important biological entity that has gained much interest in nanoscience is the virus. Typical virus capsids are ∼10 nm in size and can be considered macromolecules in the sense that they are composed entirely of protein molecules. Virions have also been the subject of studies in genetic engineering and controlling features such as capsid protein composition is easily accomplished. The genetically modiﬁed cow pea mosaic virus that contains a cysteine residue (Cys-CPMV) at speciﬁc spots on the capsid has been used to speciﬁcally adsorb onto patterns of covalently modiﬁed gold surfaces produced via DPN [43], Fig. 23.5. To this end, an amine terminated thiol was patterned directly onto the gold surface. The remaining unpatterned areas were then passivated using a protein resistant SAM. The reactive terminal amine was then used to interact with a maleimide cross-linker, while the protein resistant SAM was not. The cross-linker is highly speciﬁc in its interactions and reacts with the thiol of the cysteine residue on the Cys-CPMV. This reaction formed a covalent

16

J.M. Kinsella · A. Ivanisevic

Fig. 23.4. AFM images of DNA stretched onto DPN generated patterns of the polyelectrolyte poly(allylammonium) hydrochloride (PAH). (a) Height image. (b) Phase image. The DPN generated patterns of PAH are clearly visible as vertical lines in the phase image. They do not appear in the height image due to the small height of the deposited polyelectrolyte when patterned using DPN. Figure reprinted with permission from [42]

Fig. 23.5. AFM height image of genetically modiﬁed cowpeamosaic virus (CPMV). Inset of ﬁgure (a) shows a model of the virus surface. The surface of the virus particle is speciﬁc modiﬁed, through genetic engineering, to express cysteine labels. These cysteine residues are present at each of the ﬁve corners of the pentagon structures on the capsid of the CPMV. (b) AFM height images of the virus assembled onto templates. Inset shows high resolution of the virus coated templates. (c) Lines of virus assemblies formed through nanografting of the surface and subsequent chemical template assembly. (d) High resolution AFM imaging of the virions assembled on the chemoselective template lines of (c). Figure reprinted with permission from [43]

23 Chem, Bio, and Eng Applications of SPL

17

link between the virus and the maleimide modiﬁed DPN pattern. It was also reported that, due to the large attractive interaction that the virions displayed for one another, stacking occurred on the patterned areas. Similar work has been done to capture cysteine modiﬁed virus particles that employed the writing of a mixed monolayer onto gold [44]. The mixed monolayer consisted of two chemically distinct dialkyl disulﬁde molecules. The ﬁrst contained a maleimide group that would interact speciﬁcally with the virus while the second contained a penta(ethylene glycol) group that worked to inhibit the non-speciﬁc adsorption of proteins onto the patterned surface. The density of available maleimide groups on the surface was determined by controlling the fraction of the maleimide molecule in the inking solution.

23.4 Chemical Patterning Decades of research in surface science have given nanoscience a predetermined beginning to take advantage of. The original DPN work involving patterning of ODT and MHA soon escalated into the patterning of various other small organic molecules, organics, nanoparticles, metal ions, polyelectrolytes, polymers, salts, supramolecular complexes and sol-gels. An equal number of surfaces have been used in these experiments ranging from metals to insulators and semiconductors. 23.4.1 Thiols The original work of DPN focused on the patterning of ODT and MHA on gold, which was not surprising since these thiolated molecules have been well-studied as bulk SAMs on gold surfaces [11]. On a gold surface assumed to be atomically ﬂat, the formation of a monolayer follows a stepwise mechanism. The initial step is the adsorption of the molecules to the surface. During this step, the molecules typically lie ﬂat across the surface. When the coverage of the surface is nearing capacity, the ﬂat lying molecules are forced to reorient into positions that allow the remaining surface vacancies to be occupied; this step is referred to as nucleation. Nucleation does not occur simultaneously across the surface, but instead typically occurs in dense patches. The orientation of the molecules in these patches is one where the carbon chains align at a speciﬁc angle relative to the surface. This angle between the adsorbed molecules and the surface is typically close to but not exactly normal. This speciﬁc orientation allows for the dense formation of the monolayer. This non-perpendicular arrangement also infers that the height of the monolayer on the surface is not the same as the length of the molecule adsorbed but will be slightly less. DPN patterns of thiolated molecules on gold have been proven to follow a similar adsorption mechanism as that which occurs during bulk adsorption of the same molecules on gold [45]. The morphology of the patterns generated with alkylthiols on gold can be controlled by tuning the writing speed [46]. This evidence provides further proof that the mechanism of formation of the adsorbed monolayer patterned by DPN is similar to

18

J.M. Kinsella · A. Ivanisevic

that of the bulk SAM formation mechanism. The control over the morphology can be explained by the inﬂuence of the time that the adsorbed molecules are exposed to the meniscus used for writing. It was found that the relative height of the patterned monolayers would decrease as the writing speed was increased. This implies that the initial deposition of molecules occurs and the orientation of the molecules is ﬂat on the surface. The quick writing speed would move the meniscus away from the deposited molecules too quickly and will not be allowed to induce their reorientation. The lack of reorientation of the deposited molecules is due to the lack of molecules competing for the same surface sites and a shortage of solvent due to the absence of the meniscus. Under these quick writing conditions an incomplete monolayer is formed. 23.4.2 ω-Substituted Thiols The interfacial properties of the DPN patterned monolayer of a thiolated molecule can be controlled by employing a molecule with a speciﬁc terminal functionality opposite the thiol end of the molecule (ω-terminated thiols). Alternatively, a mixed monolayer can also be formed through chemical modiﬁcation of the monolayer after patterning [47]. A number of ω-terminated thiols exist and those typically used for SAM formation are the amino, carboxylic acid and hydroxyl terminated varieties. These molecules provide versatile platforms for further bottom-up directed nanofabrication schemes. Such control over the interfacial properties of the surface can be seen by comparing the properties of ODT and MHA. Octadecanethiol is a long thiolated hydrocarbon that terminates with a methyl group. Mercaptohexanoic acid is a thiolated hydrocarbon that is of similar length as ODT but contains a terminal carboxylic acid. When a full monolayer of either molecule is allowed to adsorb on the surface the wetting properties of the differing surfaces should be different due to the difference in the hydrophobicity between the two molecules. This difference can be examined using lateral force microscopy (LFM) to image patterns formed by the different molecules [48]. The force between the hydrophobic ODT monolayer and the tip will be much less than the force between the more hydrophilic MHA monolayer and the tip when compared to the surrounding unpatterned gold surface. The process depends on the quality of the monolayer formed during patterning. A pattern of MHA written with a faster speed will appear to have the same attraction to the tip as a completely formed monolayer pattern of ODT. Such an observation is thought to be due to the orientation of the molecules and the nature of the terminal chemical groups. We previously stated that thiolated molecules that are typically used in DPN patterning are the amino, carboxylic acid and hydroxyl terminated thiols. Once patterned, the terminus of the molecule can be used as a template for higher order nanofabrication schemes. In earlier sections, we discussed the fabrication of a virus speciﬁc pattern using DPN fabricated templates of amine terminated thiols [44]. Additionally, we also discussed how the carboxylic acid terminated MHA has been used as a template for the indirect patterning of proteins [31]. The orthogonal assembly of charged nanostructures has been formed using redox-active ferrocenylalkylthiol inks and polyanionic oligonucleotide-modiﬁed

23 Chem, Bio, and Eng Applications of SPL

19

gold particles [49]. Two different ferrocenylalkylthiols were chosen, (Fc(CH2 )11 SH and Fc(C=O)(CH2 )11 SH), due to their non-overlapping redox processes. The inks were patterned on gold with line widths as small as 55 nm. An electrochemical cell was then employed using the patterned substrate as the working electrode to apply potentials that would oxidize the generated patterns. The oxidized patterns could be used for the concomitant assembly of the polyanionic particles. Due to the large difference in the redox potentials, the guided assembly of the particles could be controlled using linear sweep voltammetry. Thiols have also been used to form mixed monolayers on a surface using traditional SAM formation procedures. The properties of a mixed monolayer can be controlled by the types of thiolated molecules (different ω-terminated thiols, different number of methylene units) that are used and the stoichiometric concentrations of each type of thiolated molecule present. In classic formation of SAMs, the molecules were allowed to compete with each other for surface binding sites until complete coverage was attained. A method of directed formation of mixed monolayer patterns was discovered using a technique that employs a scanning probe lithography method called the nanopen reader and writer (NPRW) [50]. This method is similar to DPN in that a coated tip is used to transfer molecules. What is different from DPN is that typically the surface is already modiﬁed with one type of SAM. The coated tip is used to do two tasks simultaneously. The ﬁrst task is to remove the SAM below the area of scanning by using a high contact force. The high contact force shears off the adsorbed monolayer. The second task is to ﬁll-in the newly exposed bare area of the substrate with the molecule physisorbed onto the tip. The resulting substrate will have a monolayer of patterns with resolutions comparable to DPN. 23.4.3 Silanes and Silazanes Silanes and silazanes are important in the ﬁeld of solid state microfabrication. They are often used as adhesion layers to promote the ﬁlm formation of photoresists onto surfaces. Neither compound reacts with gold surfaces, but the silane and silazane ends of the molecules can interact with surface oxide groups to become immobilized. Several of the commonly used trichlorosilanes and trialkoxysilanes undergo polymerization reactions when exposed to water. Hexamethyldisilazane (HMDS) was used to pattern both silicon oxide and gallium arsenide surfaces by DPN [51]. This silazane is a common adhesion layer promoter ﬁlm and is not likely to polymerize when in contact with the meniscus during the transfer procedure. DPN generated patterns smaller than 100 nm were achieved on both types of semiconductor surfaces. It was reported that the transfer of the silazane ink was much slower than the reported transfer rates of thiols on gold. The rate of transfer was altered by the increase in the local temperature of the tip and substrate. This data indicates that temperature may be an important variable in the transfer of certain inks in the DPN process. Despite the difﬁculties associated with patterning of alkoxysilanes and trichlorosilanes, it has been shown that under a controlled environment that impedes the polymerization of these molecules patterning is possible [52, 53]. These are important molecules because standard microarray technology employs them as the starting

20

J.M. Kinsella · A. Ivanisevic

SAM molecules on glass and silicon oxide surfaces. Successfully patterning these molecules on substrates is an important advancement for the fabrication of nanoarrays and biosensors. The molecule patterned was 3 -mercaptopropyltriethoxysilane (MPTMS) on glass [52]. The functional groups of other silanes tend to interact with oxide surfaces and, therefore, interfere with the patterning, while the thiol group is relatively inert towards glass surfaces. The patterning was performed under humidity conditions from ∼ 0% to ambient humidity in order to limit the polymerization of the silanes. It was found that at low relative humidity values MPTMS could be controllably patterned without signiﬁcant polymerization. The proposed mechanism of reaction for silanes and silazanes on glass and silicon oxide involves the presence of adsorbed water in order to covalently bind the molecule to the surface. The silanes were also shown to remain active after patterning by coupling the thiol group of the adsorbed silane with biotin and Cy3-streptavidin. The coupled complex displayed ﬂuorescence on the areas that were speciﬁcally patterned. 23.4.4 Deposition of Solid Organic Inks AFM cantilevers fabricated with embedded resistive heating elements have been employed to deposit organic inks. The organic inks were chosen to exist as solids at room temperatures [54]. A resistive heater was used to increase the local temperature

Fig. 23.6. (a) AFM height image of OPA patterns generated using thermal dip-pen nanolithography (tDPN) on a mica surface by scanning for 256 s on a 500 nm × 500 nm area. The melting point of OPA is 99 ◦ C. Patterning below this temperature did not deposit any ink on the surface. As the temperature approached and exceeded the melting point of the ink deposition occurred. (b) AFM friction image of the same surface. Image reprinted with permission from [54]

23 Chem, Bio, and Eng Applications of SPL

21

of the tip to the melting point of the solid ink physisorbed onto it. Upon heating the ink melts and deposition onto the surface occurs. The use of the heating element allows for controlled deposition of the ink that can be stopped without withdrawal of the tip from the substrate. The tip can also be used in its “cold” state to image the patterned area without risk of further ink deposition. The ink used in the proof-ofconcept study was OPA, which has a melting temperature of 99 ◦ C, Fig. 23.6. This ink can form SAMs on mica, stainless steel, aluminum and metal oxides in its liquid state. Deposition of the OPA onto mica was studied at several different localized tip temperatures. Controllable patterning of lines that were 100 nm in width was routinely accomplished. 23.4.5 Polymers Advances in polymer synthesis have accelerated greatly due to the increase in applications of polymers in nanoscience [55]. The electrical, optical, catalytic and mechanical properties of polymers are dictated by their design and synthesis, which opens up a host of possible applications and opportunities. Polymers may be synthesized to interact with biological cells, tissues or organs in predetermined ways and are currently being explored for uses such as drug delivery, material coatings and implants. Polymers can either be directly immobilized onto surfaces using covalent bonding, electrostatic forces or van der Waals interactions, or they can be deposited in their monomer state and polymerized in situ. By employing the direct writing of polymers on surfaces, one can create controlled feature sizes and placement. In principle, due to the high molecular weights and low vapor pressures, the resolution of polymer inks should be lower than that encountered with thiol inks. Using DPN to pattern luminescent conductive polymers into nanowires onto glass has been demonstrated using MEH-PPV as ink [56], Fig. 23.7. This polymer displays luminescent properties that make it an attractive candidate for the fabrication of nano-LEDs. Nanowires of conductive polymers such as MEH-PPV can also be

Fig. 23.7. (a) Scanning confocal microscopy images of a series of MEH-PPV lines generated using DPN on glass. The scanning speed used to generate the lines decrease from the top line down. With the decrease in scanning speed, the ﬂuorescence intensity is also seen to decrease. (b) LFM of MHA lines patterned on gold. Reprinted with permission from [56]

22

J.M. Kinsella · A. Ivanisevic

used to study the basic principles of photoluminescence and electrical transport in an idealized one-dimensional system. The polymer wires that were deposited had line widths as small as 30 nm. Like the patterning of more traditional chemical inks, such as ODT, it was determined that the line width was inversely proportional to the scanning speed. The electrostatic attraction between conductive polymers and surfaces has itself been employed as the driving force to fabricate assemblies of nanoscale features using DPN [57]. To this end, silicon substrates were appropriately modiﬁed to impart either a net positive charge or a net negative charge onto the surface. The oppositely charged polymers were used as inks and the electrostatic interactions between the ink and the substrate were used to immobilize the polymer patterns. SPAN, a negatively charged polymer, and doped PPy, a positively charged polymer, were used as the polymer inks. It was noted that the SPAN polymer was not signiﬁcantly deposited onto the negative substrate in control experiments and the PPy polymer was not signiﬁcantly deposited onto the positively charged substrates. These experiments concluded that the mechanism causing the transfer was the long-range electrostatic attraction between the surface charges and the charge of the physisorbed ink on the tip as patterning was occurring. Electrochemical analysis of the deposited polymer patterns conﬁrmed that the transfer occurred successfully. The previous applications involved the transfer of the polymer to the surface after polymerization occurred. The ability to transfer monomers onto the surface and to polymerize them in situ is also possible. This technique can be advantageous because many of the polymers are insoluble in water while their monomers are soluble. Therefore, controlling the deposition of the polymers may be challenging using a water meniscus. A drawback to the in situ polymerization is that controlled doping of the polymers may be made difﬁcult, which limits their functionality in applications. Proof-of-concept experiments have demonstrated that a monomer ink can be patterned onto SiO2 substrates [58]. The chosen ink was required for the acidpromoted polymerization of pyrrole. The deposition was dependent on the degree of polymerization that the ink had undergone prior to writing. The monomer was also deposited into an electrode gap so that I–V measurements could be performed to characterize the conductance of the patterned material. A dependence of the resistance on the relative light intensity was reported, which indicates the presence of photoluminescent properties that would be expected in polymerized pyrrole. In an effort to reduce the environmentally damaging products used in the synthesis of polyaniline polymers and its derivatives, HRP was introduced as a catalyst to induce the free-radical polymerization reaction. Deposition of 4-aminothiolphenol using DPN on gold and subsequent HRP catalyzed reaction to form polymeric conducting wires was accomplished [59]. 4-aminothiolphenol is a monomer that forms the polymer structure poly(4-aminothiolphenol). The line widths and heights changed upon polymerization, indicating the patterned monomers had polymerized upon exposure to HRP. This and spectroscopic evidence of the polymerization led to the conclusion that the HRP catalyzed polymerization reaction occurred on the monomer patterns. E-DPN has been employed to write poly(thiophene) polymer nanowires on semiconducting and insulating surfaces [60]. This method employs an electrochemical in situ polymerization reaction between the tip and sample during patterning. The

23 Chem, Bio, and Eng Applications of SPL

23

monomer of interest EDOT was employed as the ink. During patterning, a negative bias voltage is applied to the tip resulting in polymerization. The polymer was then directly deposited as in typical DPN experiments. When no voltage was applied to the system, no polymer was deposited. The control over the morphology of the patterns was determined to be dictated by the standard conditions used to control deposition via DPN, as well as the applied voltages. The ability to pattern these polymers on both semiconducting and insulating surfaces allows the possibility of integration into novel device structures. 23.4.6 Polyelectrolytes Polyelectrolytes are a class of macromolecules that have been extensively researched in nanoscience due to their ability to self-assemble when deposited onto charged surfaces. The self-assembly of these molecules has been used in a method called layer by layer assembly. Polyelectrolytes do not covalently link as polymers do, but instead rely on electrostatic interactions to form ordered complexes. Layer by layer assembly exploits the electrostatic interactions of polyelectrolytes by forming layer upon layer of oppositely charged thin ﬁlms on surfaces. Since several polyelectrolytes are commercially available, it is possible to alternate not only the charge, but also the composition of several different layers during assembly. Polyelectrolytes have been used on several types of substrates and on particle surfaces. We have discussed the use of polyelectrolytes that have been patterned onto SiOx as templates for the guided assembly of stretched long DNA molecules [42]. An interesting study comparing the mechanical properties of polyelectrolyte patterns created using both DPN and microcontact printing has been carried out [61]. The polyelectrolyte PAH was patterned using DPN onto SiOx . Microcontact printing was then performed to generate micron sized lines on top of the DPN patterns. By direct comparison of tapping-mode images, it was noted that the DPN patterns are visible only by phase contrast imaging due to their relatively small heights, while the microcontact printed patterns can be discerned both in phase and height mode images. FV imaging was performed to discern whether the mechanical properties of the two patterns generated varied. From the FV experiments, it was deduced that the relative hydrophilicity of the DPN generated PAH patterns was greater than that of the microcontact printing PAH patterns. 23.4.7 Dendrimers Dendrimers are an important class of macromolecules that offer researchers the ﬂexibility of polymers but are often easier to control. Dendrimers are branched three-dimensional molecules that can be designed with speciﬁc surface functionalities through chemical modiﬁcations. These molecules have attracted extensive research interest in sensing, electronics, luminescent and drug delivery applications. Starburst PAMAM has shown promise in the transport of DNA across cell membranes [62]. The high degree of ﬂexibility in the surface functionality makes

24

J.M. Kinsella · A. Ivanisevic

dendrimers the perfect candidates for both direct and indirect writing experiments on several surfaces. Direct writing has been performed with amino terminated dendrimers onto silicon oxide surfaces using DPN [63]. These experiments were already successful when attempted by microcontact printing and the resulting patterns were determined to be stable [64]. It should be noted that dendrimers are macromolecules that have high molecular weights that, in principle, impede transport during patterning. To further understand the transfer kinetics, the patterning of ﬁve generations of starburst PAMAM was performed, each of differing molecular weights. It was noted that the deposition of the higher molecular weight generations of PAMAM was slower than the lower molecular weight generations, as expected from diffusion models. PAMAM has also been patterned onto NHS ester SAMs modiﬁed gold [65]. The esters exposed by SAMs allow for covalent attachment to the PAMAM by amide bond formation. Direct DPN patterning onto glass and mica was shown to be difﬁcult due to a lack of stabilizing interactions between the surface and ink. When PAMAM ink was covalently coupled to NHS, stable patterns below 100 nm could be routinely fabricated. 23.4.8 Deposition of Supramolecular Materials Supramolecular compounds have been of recent interest due to their ﬂexibility in chemical, electronic and recognition functionalities. The recognition functionality is displayed by the speciﬁcity of host-guest interactions between calixarenes host

Fig. 23.8. LFM images (left) of calixarene patterns generated using DPN. The ﬁrst row shows images of calixarenes patterned onto β-cyclodextrin (β-CD) SAMs on gold. The second row shows images of calixarenes patterned onto 11-sulfanyl-1-undecanol (OH) SAMs on gold. The patterns generated on β-CD terminated SAMs were shown to remain intact even after washing in both water and NaCl solutions. The patterns generated on OH SAMs rinsed off when washed in water and NaCl solutions. The image on the right shows an LFM image of 60 ± 20 nm lines of calixarene molecules generated on gold. Reprinted with permission from [66]

23 Chem, Bio, and Eng Applications of SPL

25

molecules and univalent guest molecules in solution. These interactions can be used for sensing, electronic and optical purposes through the introduction of speciﬁc guest molecules. As a proof-of-concept experiment calixarene inks were patterned on gold surfaces modiﬁed with β-cyclodextrin (β-CD) or onto 11-sulfanyl-1-undecanol SAMs, [66] Fig. 23.8. Patterns of the calixarene were generated on the β-CD SAM which remained stable even after washing. The patterns generated on the 11-sulfanyl-1-undecanol SAMs were easily removed from the surface by rinsing. The resolution of the patterns was sub 100 nm on both surfaces. Heteroatom-functionalized porphyrazines synthesized to be terminated on mercaptides have also been patterned on gold surfaces using DPN [67]. These experiments employed two types of pz molecules designed to adsorb either perpendicular to the surface or lie ﬂat on top of the surface. To this end, the pz molecule was either synthesized with disulﬁde linkers on the same side (perpendicular arrangement) or with disulﬁde linkers on opposite sides (ﬂat lying). Electrochemical measurements conﬁrmed that in solution the redox potential of the molecules was identical. The redox potential varied based on orientation when SAMs of each molecule were formed. The change in the redox potential induced by the orientation of the molecules may be useful in the design of molecular devices. 23.4.9 Deposition of Metals Inorganic materials are important inks for DPN based on their many important applications, including electronics, optics and catalysis. Electronic solid state fabrication procedures typically involve thermal evaporation, e-beam evaporation or sputtering of a desired metal or alloy to form a thin ﬁlm of the metal on the surface [1]. This is typically followed by an etching step to remove the metal from unwanted areas of the substrate. The power of DPN for the deposition of metals is that this technique can be used in either bottom-up assembly or top-down fabrication. The assembly of inorganic nanostructures using the bottom-up approach relies on techniques that are similar to those we have already seen, although incorporation of metal or nanoparticle inks instead of thiols, silanes or biomolecules is employed. The top-down method can be achieved by depositing molecules as a resist onto a thin ﬁlm of metal and subsequently etching away the unpatterned areas. Both approaches have been demonstrated and the top-down approach for fabricating metal nanostructures will be covered in greater detail in the section on engineering applications of DPN. An important metal in modern IC device structures is gold. Gold is often used for interconnects and contacts on microelectronic devices [1]. DPN has been used to deposit gold onto silicon substrates with sub 100-nm resolution [68]. The mechanism of the feature formation was dependent on the use of an HAuCl4 ink deposited onto an HF etched Si substrate. The Au(III) metal is reduced to an insoluble form, Au(0), by the silicon surface. Once patterned, the reduced metal nanostructures are stable to washing with both aqueous and organic solvents, heating to below 500 ◦ C, and HF etching. It was found that the patterns underwent a phase transition above 500 ◦ C and formed aggregates of gold on the surface above this temperature. These ﬁndings indicate the utility of the generated Au patterns and their ability to withstand some abrasive conditions typical of solid state fabrication procedures.

26

J.M. Kinsella · A. Ivanisevic

Using E-DPN, several other metal patterns have been generated on silicon surfaces [69]. The underlying mechanism was the reduction of the metal salts using a voltage applied to the tip during transport to create an electrochemical cell out of the meniscus. Platinum, gold, silver, copper, palladium and germanium have all been patterned using this technique. The compositions of these features was also tested by HF etching and thermally induced phase transitions to ensure that the applied voltage was depositing the metal and not oxidizing the substrate. The platinum features displayed catalytic activity when reacted with ethylene, which was characteristic of bulk platinum surfaces. The use of DPN to deposit metal clusters or nanoparticles onto surfaces has also been of much recent interest. Metal nanoparticles exhibit optical, electronic and catalytic properties which may be size dependent and differ from their complementary bulk materials. For electronic applications, the positioning of metal nanoparticles with high registration is of great importance in fabricating working structures. Early work prior to the advent of DPN technology showed that the tip of the atomic force microscope could be used to controllably manipulate the positions of nanoparticles on surfaces [70–73]. To this end, particles on the surface were pushed with the tip in contact mode, resulting in the movement of the particle as directed by the motion of the cantilever. This technique remains serial and the movement of large numbers of particles is time-consuming because only single particles or small aggregates of clusters are moved. DPN can be used to generate patterns of particle inks on a number of different surfaces through both direct writing and indirect templating procedures [74]. The NPRW system employed to fabricate precise mixed monolayers has also been used to successfully generate 3D patterns of thiol passivated gold nanoparticles [75]. The physisorbed gold particles on the tip are transferred to the surface during scanning while SAM resist on the surface is simultaneously removed due to the shear forces applied. Single monolayer deposition of the particles was achieved when a slow scan rate was employed. 23.4.10 Deposition of Solid-State Materials In typical top-down techniques, the underlying semiconductor material is etched away in a step-wise process to fabricate devices of speciﬁc geometries, orientations and chemical compositions. The procedures to fabricate these devices are parallel processes that generate structures on a wafer scale level. It is imperative that patterns of materials typically used by the semiconductor industry be successfully patterned in order to successfully achieve bottom-up approaches to nanofabrication of electrical circuitry. To meet this goal, direct patterning of solid state materials using inorganic precursor salts with amphiphilic block copolymer surfactants has been employed to generate patterns of Al2 O3 , SiO2 and SnO2 on silicon and silicon oxide substrates [76, 77]. Bulk surface depositions of these materials have been performed using sol-gel chemistry on semiconductor-oxide surfaces. The patterned composite hydrolyzes on the underlying substrate forming the desired inorganic nanostructure. To remove the organic copolymer, the surfaces were heated to 400 ◦ C, causing the organic

23 Chem, Bio, and Eng Applications of SPL

27

residues to combust. It was noted in these experiments that the diffusion of the ink qualitatively acted in a similar way to thiol inks. 23.4.11 Deposition of Magnetic Materials Magnetic materials are often found in applications such as high density recording media, magnetoelectronics and biotechnology. In the case of high density recording media, shrinking these features to the nanoscale would increase the information density by orders of magnitude over the current limits. Conventional fabrication of these devices is challenging due to current etching procedures. Because these devices shrink, using top-down methods becomes ever more difﬁcult. Patterning by DPN is an alternative that can allow nanoscale control over feature size and interfeature distances, which are both important for the standard operation of such devices. The magnetic material BaFe is considered a “hard” magnet and is useful for high density recording applications. Patterns of BaFe have been generated on silicon oxide substrates with sub 100 nm resolution using a sol-based ink containing a mixture of precursors necessary for the synthesis of BaFe [78]. After patterning, the substrates were thermally treated to yield polygonal BaFe particles with a mean diameter of ∼ 35 nm. The arrays patterned displayed characteristics of single domains or at most just a few domains. The elemental composition was characterized using various spectroscopic analysis techniques and the magnetic properties were displayed using MFM. The direct writing and indirect templating of magnetic nanoparticles with nanoscale precision has been demonstrated using DPN. In direct writing experiments, Fe2 O3 nanoparticles with an 11 nm diameter were deposited onto both silicon and mica [79], Fig. 23.9 The generated patterns exhibited sharp edges, which indicated little lateral diffusion and a high control over the positioning of the particles. Indirect templating experiments relied on the DPN patterning of MHA and electrostatic interaction as the driving force to generate arrays of iron oxide nanoparticles [80]. Iron oxide nanoparticles capped with tetramethylammonium hydroxide were synthesized with a ∼ 10 nm diameter. Features were routinely generated with less than 100 nm diameters on gold. The cross-sectional topography of the templated structures indicated that magnetic particles formed single layers on the MHA patterns.

Fig. 23.9. Contact mode AFM height image of γ -Fe2 O3 nanoparticle patterns generated by DPN on a mica surface. Control over the position of the particles during patterning without substantial lateral diffusion can be seen. Figure reprinted with permission from [79]

28

J.M. Kinsella · A. Ivanisevic

23.5 Engineering Applications of DPN Several of the inks described in the previous sections can be used in a wide range of engineering applications. Biomolecules are commonplace in catalysis and fermentations. Biomolecules also provide a potential source as components for the fabrication of 3D electrical or mechanical architectures. Many of the chemicals patterned also have inherent properties that make them useful in applied technologies. The interfacial control achieved over surfaces through the patterning of thiols, the photochemical properties of patterned silanes, silazanes and other polymers, as well as the electrical properties of deposited metals and sol-based inks are all useful for engineering purposes. The deposition of these important materials on the nanoscale also allows researchers to study the fundamental properties of the materials and gain further insight regarding their potential. Integration of DPN with existing methods for creating microstructures is an area of research that is actively being explored and may function to increase both the recognition of DPN and to concurrently improve these existing technologies. Typical photolithography approaches depend on the use of a photomask to fabricate structures from the top down. The photomask can be fabricated on various types of fused silica, while the opaque layer is typically chromium [1]. Any defect in the chromium structure of the mask will be translated directly on the fabricated features resulting in defects. Typically, these defects are repaired by additional deposition of the metal when found. As device size shrinks to sub 100 nm resolution, it is extremely important that the opaque layer contains zero defects. DPN can be employed in the repair of such small mask defects through directed site speciﬁc deposition of chromium onto defective sites. Other non-conventional soft lithography microfabrication methods can also beneﬁt from the integration of DPN. One such method is microcontact printing. µCP is a soft lithography method that involves the transfer of a molecular ink to a substrate using an elastomeric stamp. The process has the same versatility of inks and substrates as DPN. The resolution of µCP and its derivatives is typically on the microscale, although by using edge-transfer methods or controlling the diffusion of the ink, one can decrease the resolution to the nanoscale [81]. One advantage of the technique is that the physical size of the stamp is not limited and patterns are formed in parallel. However, one of the problems typically encountered when µCP is used is the inability to pattern multiple types of inks with high registration, since the technique relies on the physical stamping of the ink to the surface to deﬁne the patterns. Once the pattern is generated on the substrate, traditional SAM techniques can be used to passivate bare areas on the surface, although this typically only allows the patterning of two distinct regions on the surface: the µCP generated pattern and the area passivated. To overcome this limitation, a procedure similar to DPN has been used to ink the features of the elastomeric stamp used for µCP with different molecules [82]. This method of integrating scanning probe methods and µCP has allowed the transfer of multiple ink types using a single stamp. We have also noted the generation of polyelectrolyte layers on silicon using ﬁrst DPN and subsequently µCP [61]. This technique can be used to generate both microstructures and nanostructures on the same surface. The alignment and registra-

23 Chem, Bio, and Eng Applications of SPL

29

tion between the deposited using the two different lithographic techniques remains challenging. The direct comparison of the chemical, mechanical and electrical properties of materials deposited through these two techniques can be used to answer fundamental questions regarding the differing nature of the inks when existing in the microscale versus the nanoscale. Researchers have also combined the two techniques of DPN and µCP by fabricating dip-pen nanolithography stamp tips. To this end, standard AFM tips were coated with the elastomer PDMS and allowed to cure [83], Fig. 23.10. Patterning of MHA, ODT and cystamine on gold was displayed, as well as G6-OH dendrimer patterning on gold and SiOx . It was noted that the PDMS coated tip absorbs the ink and the solvent used during tip coating. Structures of MHA and ODT were patterned on sub 100 nm scales almost approaching the limits of DPN generated patterns and exceeding the limits of those patterns generated by µCP. Efforts have also been made to generate cantilevers from PDMS elastomers [84]. Such cantilevers have been used to generate sub-500 nm patterns of ODT on gold, although the lack of reﬂective metal coatings on the cantilever make control over the AFM feedback difﬁcult. The deposition of biomolecules, nanoparticles, organic molecules and magnetic materials on microfabricated devices using DPN has also been used to study the nanoscopic properties of these materials. The registration and resolution of the DPN technique has allowed researchers to deposit inks in electrode gap junctions, allowing one to directly probe the electrical properties of the ink. Electrical characterization of these materials on the nanoscale is important on a fundamental level when considering their uses for future nanodevices, (bio)arrays, sensors or single molecule electronics. DPN has also been used to pattern catalysts in desired locations in order to grow nanowires [85]. Speciﬁcally, Ni(NO3 )2 patterns were generated directly onto SiO2 substrates. After patterning, the substrates were exposed to elemental gallium and ammonia to induce the growth of gallium nitride (GaN) nanowires through a vapor– liquid–solid (VLS) mechanism. Nanowires several microns long were grown from

Fig.23.10.Optical micrograph of a chip containing two scanning probe contact printing tips. Inset: Scanning electron micrograph of scanning probe contact printing tip composed of a PDMS tip and a polyimide cantilever. Image reprinted with permission from [84]

30

J.M. Kinsella · A. Ivanisevic

the catalyst islands. Metal electrodes were evaporated onto the wires after growth and the electrical characteristics of the wires were determined to match their expected values. Existing top-down technologies relying on wet chemical etchants and wet chemical oxidants may also be interfaced with DPN to control the depletion or oxidation of site-speciﬁc areas of a substrate. Wet chemical oxidants have been employed as inks and used to controllably fabricate heterostructure nanowire materials. Further control over the degree of oxidation of the native gallium nitride (GaN) nanowires has been demonstrated by employing E-DPN to tune the deposition of the oxidant [86]. Speciﬁcally, KOH ink was employed to site-speciﬁcally oxidize native GaN nanowires forming GaN/gallium oxide heterostructures. Once the E-DPN induced oxidation occurred, the resulting gallium oxide regions of the nanowires could be etched using an acidic etchant solution. Devices containing the GaN nanowires were fabricated to test the electronic properties of the native wires, the heterostructured wires and the acid etched wires. It was noted that the heterostructured wire was far less conductive that the native GaN wire. This was attributed to the behavior of the gallium oxide as a barrier to electron transport through the wire. After etching of the gallium oxide, a signiﬁcant decrease in the current was always observed. Exploitation of the bias voltage applied during patterning, the dwell time of the tip and the relative humidity can all be used to form oxide structures that are of different diameters on the GaN wires. By controlling these parameters, one can control the degree of heterostructures and thereby control electronic properties of the individual wires. Several reports of using DPN to pattern resist layers on a number of surfaces have been published recently [87]. ODT patterns have been used as resists on gold and palladium surfaces to protect areas of the substrate from wet chemical etching. MHA has also been successful as a resist on gold and silver substrates. Using these methods, arrays of gold structures functionalized with oligonucleotides have been formed. MHA ink was deposited as the etch resist onto a gold surface to form the arrays. The Au was etched resulting in nanostructures protected by the MHA patterns. The samples were irradiated with UV light for 10 hrs and rinsed with water to remove the MHA. The result was bare gold nanostructures that could be functionalized with disulﬁde modiﬁed oligonucleotides. The controlled formation of nanogaps between gold electrodes has also been demonstrated using MHA inks. Nanogaps with distances less than 50 nm can be routinely fabricated using this method.

23.6 Future Challenges and Applications Shortly after the discovery of DPN, it was noted that in order to make the technique a production scale method, its serial nature would have to be overcome. To this end, arrays of cantilevers that could be used to generate patterns in parallel had to be fabricated. The proof-of-concept experiments employed standard contact mode tips that contained multiple cantilevers adjacent to one another as the multi-tip array. This array of tips was shown to be able to write both single ink types and multiple ink types in parallel with the alignment that was determined by the dimensions of the cantilevers. Since then, several groups have fabricated tip arrays of different tip

23 Chem, Bio, and Eng Applications of SPL

31

geometries and writing capabilities. To further enhance the parallel development of DPN, cantilevers containing integrated microﬂuidic inking channels have also been developed [88]. These initial experiments were soon followed by the fabrication of probe arrays to directly address the parallel writing potential of DPN. Prior to the advent of DPN researchers had developed methods to fabricate arrays of multiple tips in an effort to increase the scan size and speed of imaging for AFM [89]. The probe arrays fabricated for the purpose of DPN differed from the arrays meant for imaging. Factors in fabrication of the DPN arrays included the force constants, length of the probe, sharpness of the tips and the tip to tip spacing. Two types of arrays were developed in these experiments: the ﬁrst was a 32 pen array that could be used to fabricate structures of ODT with line widths averaging 260 nm. The second array was fabricated with sharper tips and 8 pens and achieved line widths of ODT ranging from 80–100 nm. Recent experiments in this area have focused on fabricating arrays of individually addressable cantilevers. Thermal bimorph actuated arrays were fabricated where each individual cantilever contained its own actuator [90]. When a cantilever was heated using an embedded electrical heater, the thermal mismatch strain between the chromium/gold and silicon nitride layers of the cantilever would cause the probe to deﬂect away from the surface. When the heating element was switched off, the cantilever would approach the surface and begin writing. The vapor-deposited ODT ink arrays of 10 individually addressable cantilevers enabled the simultaneous generation of the numerals 0–9 on a gold surface with sub 100 nm resolutions.

23.7 Conclusions In the few short years that the DPN technique for nanofabrication has been utilized, a number of different platforms with different applications have been actively explored. Many of these applications existed before the use of DPN, although no reproducible methods to segue these technologies into the nanoscale were available. The potential of DPN to revolutionize ﬁelds ranging from basic biology, chemistry, materials science, physics and engineering has yet to be fully reached. Advances in the parallel patterning and in atomic force microscopy in general will help this technique to achieve its full potential.

References 1. Campbell S (2001) Science and Engineering of Microfabrication. 3rd edn. Oxford University Press, New York 2. Becker RS, Golovchenko JA, Swartzentruber BS (1987) Nature 325:419 3. Eigler DM, Schweizer EK (1990) Nature 344:524 4. Crommie MF, Lutz CP, Eigler DM (1993) Science 262:218 5. Kim YT, Bard AJ (1992) Langmuir 8:1096 6. Liu G-Y, Salmeron MB (1994) Langmuir 10:367 7. Xiao X-D, Liu G-Y, Charych DH, Salmeron M (1995) Langmuir 11:1600

32

J.M. Kinsella · A. Ivanisevic

8. Kelley SO, Barton JK, Jackson NM, McPherson LD, Potter AB, Spain EM, Allen MJ, Hill MG (1998) Langmuir 14:6781 9. Whitesides GM, Mathias JP, Seto CT (1991) Science 254:1312 10. Dubois LH, Nuzzo RG (1992) Annu Rev Phys Chem 43:437 11. Ulman A (1996) Chem Rev 96:1533 12. Jaschke M, Butt H (1995) Langmuir 11:1061 13. Piner RD, Zhu J, Xu F, Hong S, Mirkin CA (1999) Science 283:661 14. Jang J, Hong S, Schatz GC, Ratner MA (2001) J Chem Phys 115:2721 15. Jang J, Schatz GC, Ratner MA (2002) J Phys Chem 116:3875 16. Sheehan PE, Whitman LJ (2002) Phys Rev Lett 88:156104 17. Rozhok S, Sun P, Piner R, Lieberman M, Mirkin CA (2004) J Phys Chem B 108:7814 18. Schwartz PV (2002) Langmuir 18:4041 19. Rozhok S, Piner R, Mirkin CA (2003) J Phys Chem B 107:751 20. Fodor SPA (1997) Science 277:393 21. Ginger DS, Zhang H, Mirkin CA (2004) Angew Chem Int Ed 43:30 22. Demers LM, Ginger DS, Park SJ, Li Z, Chung SW, Mirkin CA (2002) Science 296:1836 23. Demers LM, Park S-J, Taton TA, Li Z, Mirkin CA (2001) Angew Chem Int Ed 40:3071 24. Schwartz PV (2001) Langmuir 17:5971 25. Xu S, Liu GY (1997) Langmuir 13:127 26. Belaubre P, Guirardel M, Garcia G, Pourciel JB, Leberre V, Dagkessamanskaia A, Trevisiol E, Francois JM, Bergaud C (2003) Appl Phys Lett 82:3122 27. Wilson DL, Martin R, Hong S, Golomb MC, Mirkin CA, Kaplan DL (2001) Proc Natl Acad Sci USA 98:13660 28. Lee KB, Lim JH, Mirkin CA (2003) J Am Chem Soc 125:5588 29. Lim JH, Ginger DS, Lee KB, Heo J, Nam JM, Mirkin CA (2003) Angew Chem Int Ed 42:2309 30. Nam J, Han S, Lee K, Liu X, Ratner M, Mirkin CA (2004) Angew Chem Int Ed 43:1246 31. Lee KB, Park SJ, Mirkin CA, Smith JC, Mrksich M (2002) Science 295:1702 32. Zhang H, Lee K-B, Li Z, Mirkin CA (2003) Nanotechnology 14:1113 33. Hyun J, Lee W, Nath N, Chilkoti A, Zauscher S (2004) J Am Chem Soc 126:7330 34. Lee K-B, Kim E-Y, Mirkin CA, Wolinsky SM (2004) Nano Lett 4:1869 35. Hyun J, Ahn SJ, Lee WK, Chilkoti A, Zauscher S (2002) Nano Lett 2:1203 36. Boateng S, Lateef SS, Crot C, Motlagh D, Desai T, Samarel AM, Russell B, Hanley L (2002) Adv Mater 14:461 37. Xiao S-J, Textor M, Spencer ND (1998) Langmuir 14:5507 38. Kinsella JM, Ivanisevic A (2004) Appl Surf Sci 243:7 39. Cho Y, Ivanisevic A (2004) J Phys Chem B 108:15223 40. Agarwal G, Sowards LA, Naik RR, Stone MO (2003) J Am Chem Soc 125:580 41. Agarwal G, Naik RR, Stone MO (2003) J Am Chem Soc 125:7408 42. Nyamjav D, Ivanisevic A (2003) Adv Mater 15:1805 43. Cheung CL, Camarero JA, Woods BW, Lin T, Johnson JE, Yoreo JJD (2003) J Am Chem Soc 125:6848 44. Smith JC, Lee KB, Wang Q, Finn MG, Johnson JE, Mrksich M, Mirkin CA (2003) Nano Lett 3:883 45. Hong S, Zhu J, Mirkin CA (1999) Langmuir 15:7897 46. Barsotti RJ, O’Connell MS, Stellacci F (2004) Langmuir 20:4795 47. Laibinis PE, Whitesides GM, Allara DL, Tao YT, Parikh AN, Nuzzo RG (1991) J Am Chem Soc 113:7152 48. Zhou Y, Fan H, Fong T, Lopez GP (1998) Langmuir 14:660 49. Ivanisevic A, Im J-H, Lee K-B, Park S-J, Demers LM, Watson KJ, Mirkin CA (2001) J Am Chem Soc 123:12424

23 Chem, Bio, and Eng Applications of SPL 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 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.

33

Amro NA, Xu S, Liu G-Y (2000) Langmuir 16:3006 Ivanisevic A, Mirkin CA (2001) J Am Chem Soc 123:7887 Jung H, Kulkarni R, Collier CP (2003) J Am Chem Soc 125:12096 Pena DJ, Raphael MP, Byers JM (2003) Langmuir 19:9028 Sheehan PE, Whitman LJ, King WP, Nelson BA (2004) Appl Phys Lett 85:1589 Loiacono MJ, Granstrom EL, Frisbie CD (1998) J Phys Chem B 102:1679 Noy A, Miller AE, Klare JE, Weeks BL, Woods BW, DeYoreo JJ (2002) Nano Lett 2:109 Lim JH, Mirkin CA (2002) Adv Mater 14:1474 Ming S, Aslam M, Fu L, Wu N, Dravid V (2004) Appl Phys Lett 84:4200 Xu P, Kaplan DL (2004) Adv Mater 16:628 Maynor BW, Filocamo SF, Grinstaff MW, Liu J (2002) J Am Chem Soc 124:522 Nyamjav D, Ivanisevic A (2004) Chem Mater 16:5216 Bielinska AU, Kukowska-Latallo JF, Johnson J, Tomalia DA, Jr JRB (1996) Nuc Acids Res 24:2176 McKendry R, Huck WTS, Weeks B, Fiorini M, Abell C, Rayment T (2002) Nano Lett 2:713 Li H, Kang D-J, Blamire MG, Huck WTS (2002) Nano Lett 2:347 Degenhart GH, Dordi B, Schonherr H, Vansco GJ (2004) Langmuir 20:6216 Auletta T, Dordi B, Mulder A, Sartori A, Onclin S, Bruinink C, Peter M, Nijhuis C, Beijleveld H, Schonherr H, Vansco GJ, Casnati A, Ungaro R, Ravoo BJ, Huskens J, Reinhoudt D (2004) Angew Chem Int Ed 43:369 Vesper BJ, Salaita K, Zong H, Mirkin CA, Barrett AGM, Hoffman BM (2004) J Am Chem Soc 126:16653 Maynor BW, Li Y, Liu J (2001) Langmuir 17:2575 Li Y, Maynor BW, Liu J (2001) J Am Chem Soc 123:2105 Schaefer DM, Reifenberger R, Patil A, Andres RP (1995) Appl Phys Lett 66:1012 Junno T, Deppert K, Montelius L, Samuelson L (1995) Appl Phys Lett 66:3627 Resch R, Baur C, Bugacov A, Koel BE, Madhukar A, Requicha AAG, Will P (1998) Langmuir 14:6613 Resch R, Lewis D, Meltzer S, Montoya N, Koel BE, Madhukar A, Requicha AAG, Will P (2000) Ultramicroscopy 82:135 Liu S, Maoz R, Sagiv J (2004) Nano Lett 4:845 Garno JC, Yang Y, Amro NA, Cruchon-Dupeyrat S, Chen S, Liu G-Y (2003) Nano Lett 3:389 Su M, Liu X, Li SY, Dravid VP, Mirkin CA (2002) J Am Chem Soc 124:1560 Su M, Li S, Dravid VP (2003) J Am Chem Soc 125:9930 Fu L, Liu X, Zhang Y, Dravid V, Mirkin CA (2003) Nano Lett 3:757 Gundiah G, John NS, Thomas PJ, Kulkarni GU, Rao CNR (2004) Appl Phys Lett 84:5341 Liu X, Fu L, Hong S, Dravid VP, Mirkin CA (2002) Adv Mater 14:231 Cherniavskaya O, Adzic A, Knutson C, Gross BJ, Zang L, Liu R, Adams DM (2002) Langmuir 18:7029 Gerding JD, Willard DM, Orden AV (2005) J Am Chem Soc 127:1106 Zhang H, Elghanian R, Amro NA, Disawal S, Eby R (2004) Nano Lett 4:1649 Wang X, Ryu KS, Bullen DA, Zou J, Zhang H, Mirkin CA, Liu C (2003) Langmuir 19:8951 Li J, Lu C, Maynor B, Huang S, Liu J (2004) Chem Mater 16:1633 Maynor BW, Li J, Lu C, Liu J (2004) J Am Chem Soc 126:6409 Zhang H, Li Z, Mirkin CA (2002) Adv Mater 14:1472 Ryu K, Wang X, Shaikh K, Bullen D, Goluch E, Zou J, Liu C, Mirkin C (2004) Appl Phys Lett 85:136 Bullen D, Chung S-W, Wang X, Zou J, Mirkin CA, Liu C (2004) Appl Phys Lett 84:789 Wang X, Bullen DA, Zou J, Liu C (2004) J Vac Sci Technol B 22:2563

24 Nanotribological Characterization of Human Hair and Skin Using Atomic Force Microscopy (AFM) Bharat Bhushan · Carmen LaTorre

Human hair is a nanocomposite biological ﬁber. Maintaining the health, feel, shine, color, softness, and overall aesthetics of the hair is highly desired. Hair care products such as shampoos and conditioners, along with damaging processes such as chemical dyeing and permanent wave treatments, affect the maintenance and grooming process and are important to study because they alter many hair properties. Nanoscale characterization of the morphological, frictional, and adhesive properties (tribological properties) of hair are essential to evaluate and develop better cosmetic products, and to advance the understanding of biological and cosmetic science. The atomic/friction force microscope (AFM/FFM) has recently become an important tool for studying the micro/nanoscale properties of human hair. This chapter presents a comprehensive review of tribological properties of various hair and skin as a function of ethnicity, damage, conditioning treatment, and various environments. Nanotribological properties such as roughness, friction, adhesion, and wear are presented, as well as investigations of scale effects and directionality dependence on friction and adhesion.

24.1 Introduction Everybody wants beautiful, healthy hair and skin. For most people, grooming and maintenance of hair and skin is a daily process. The demand for products that improve the look and feel of these surfaces has created a huge industry for hair and skin care. Beauty care technology has advanced the cleaning, protection, and restoration of desirable hair and skin properties by altering the hair surface. For many years, especially in the second half of the twentieth century, scientists have focused on the physical and chemical properties of hair to consistently develop products that alter the health, feel, shine, color, softness, and overall aesthetics of the hair. Table 24.1 displays common products and processes involved in hair care. Hair care products such as shampoos and conditioners aid the maintenance and grooming process. Mechanical processes such as combing, cutting, and blowdrying serve to style the hair. Chemical products and processes such as chemical dyes, colorants, bleaches, and permanent wave treatments enhance the appearance and hue of the hair. Of particular interest is how all these common hair care items deposit onto and change hair properties, since these properties are closely tied to product performance. The fact that companies like Procter & Gamble, L’Oreal, and Unilever have hair care product sales consistently measured in billions of dollars

B. Bhushan · C. LaTorre

36 Table 24.1. Common hair care products/processes and their functions Product/process

Functions

Shampoos Conditioners

Clean the hair and skin of oils Repair hair damage and make the hair easier to comb; prevent ﬂyaway; add feel, shine, softness Combing/cutting/blowdrying Style the hair Chemical dyes/colorants/bleaches Enhance or change the color and look of hair Permanent wave treatments Change the style and look of the hair

(http://www.pg.com; http://www.loreal.com; http://unilever.com) suggests that understanding the science behind human hair has more than just purely academic beneﬁts. The tribology of the hair changes as a function of the various hair care products and processes. Figure 24.1 illustrates schematically various functions, along with the macroscale and micro/nanoscale mechanisms behind these interactions, that make surface roughness, friction, and adhesion very important to hair and skin [LaTorre and Bhushan, 2005a]. Desired features and corresponding tribological attributes of conditioners are listed in Table 24.2 [LaTorre et al., 2006]. Friction is the most relevant parameter to hair care. Our perception of a soft, silky feel comes from our ability to glide a comb or skin over the hair ﬁbers. For a smooth wet and dry feel, friction between hair and skin should be minimized in wet and dry environments, respectively. For a good feel with respect to bouncing and shaking of the hair during walking or running, friction between hair ﬁbers and groups of hair ﬁbers should be low. The friction one feels during combing is a result of interactions between hair and the comb material (generally plastic), and this too needs to be low to easily maintain, sculpt, and comb the hair. Adhesion is also important. To minimize entanglement, the adhesive force (the force required to separate the hair ﬁbers) needs to be low. In other cases, a certain level of adhesion may be acceptable and is often a function of the hair style. For individuals seeking “hair alignment”, where hair ﬁbers lie ﬂat and parallel to each other, a small amount of adhesive force between ﬁbers may be desired. For more complex and curly styles, even higher adhesion between ﬁbers may be optimal. Table 24.2. Desired features and corresponding tribological attributes of conditioners Desired hair feature

Tribological attributes

Smooth feel in wet and dry environments Shaking and bouncing during daily activities Easy combing and styling

Low friction between hair and skin in respective environments Low friction between hair ﬁbers and groups of hair Low friction between hair and comb (plastic) and low adhesion. Note: more complex styles may require higher adhesion between ﬁbers

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

37

Fig. 24.1. Schematics illustrating various functions with associated macroscale and micro/nanoscale mechanisms of hair and skin friction during feel or touch, shaking and bouncing of the hair, combing, and entanglement [LaTorre and Bhushan, 2005a]

38

B. Bhushan · C. LaTorre

Early research into human hair was done primarily on the chemical and physical properties of the hair ﬁber itself. Key topics dealt with the analysis of chemical composition in the ﬁber, microstructure, and hair growth, to name a few. Mechanical properties were also of interest. Most of the mechanical property studies of human hair were on the macroscale and used conventional methods, such as tension, torsion and bending tests [Robbins, 1994; Feughelman, 1997; Swift, 1999, 2000; Barnes and Roberts, 2000; Jachowitz and McMullen, 2002]. Efforts were also made to study the effects of environmental and chemical damage and treatment, such as dyeing, bleaching, and polymer application; these topics have remained a mainstream area of investigation due to the availability and formation of new chemicals and conditioning ingredients. Tribology has generally been studied via the macroscale friction force of hair. As a matter of fact, much of the tribological work performed by the hair care industry today still focuses on the measurement of macroscale friction, particularly between a skin replica and a hair swatch of interest [Robbins, 1994]. The intrinsic differences of the hair as a function of ethnicity eventually became a concern as well. For instance, research has shown that African–American hair has higher resistance to combing, higher static charge, and lower moisture content than Caucasian hair [Syed et al., 1996]. Because of differences like these, a growing number of hair care products speciﬁcally tailored for ethnic hair care have been developed and sold with commercial success. Modern research since the late 1990s has been primarily concerned with using micro/nanoscale experimental methods such as atomic force/friction force microscopies (AFM/FFM) and nanoindentation to answer the complex questions surrounding the structure and behavior of the hair. Nanoscale characterization of the cellular structure, mechanical properties, and tribological properties of hair are essential to evaluate and develop better cosmetic products, and to advance the understanding of composite biological systems, cosmetic science, and dermatology. AFM/FFM have been used to effectively study the structure of the hair surface and cross-section. AFM provides the potential for being able to see the cellular structure and molecular assembly of hair, for determining various properties of hair, such as elastic modulus and viscoelastic properties, and for investigating the physical behavior of various cellular structures of hair in various environments [Chen and Bhushan, 2005a,b]. As a non-invasive technique, AFM has been used to evaluate the effect of hair treatment and can be operated in ambient conditions in order to study the effect of environment on various physical properties. The nanoindenter has been used to characterize the nanomechanical behavior of the hair surface and cross-section using nanoindentation and nanoscratch techniques [Wei et al., 2005; Wei and Bhushan, 2006]. AFM has also been used to study nanotribological properties of hair and skin and the effects of hair care products on hair [LaTorre and Bhushan, 2005a,b; LaTorre et al., 2006]. Roughness parameters have been measured to compare changes due to damaging processes. The nanoscale friction force has been measured to understand damage or conditioner distribution and its effect on hair tribology. Adhesive force mapping has proven useful to observe the conditioner distribution as well. It is known that adhesion, friction, and lubrication of many interfaces are scaledependent [Bhushan, 1999b, 2005; Bhushan et al., 2004]. In the case of hair, the magnitude of the nanoscale coefﬁcient of friction in the paper by LaTorre and Bhushan [2005a] was lower than the macroscale values found in that by Bhushan

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

39

et al. [2005] for the same hair sets. It was also found that there are both similarities and differences when comparing the trends for various hair types at both scales. Also, directionality effects of friction were observed on both scales, but via slightly different mechanisms due to the size of the contacts involved [LaTorre and Bhushan, 2006]. It is thus recognized that scale effects are an important aspect of studying the tribology of hair. However, no microscale tribological data for hair exists in literature. This is unfortunate because many interactions between hair–skin, hair– comb, and hair-hair contact takes place at microasperities ranging from a few µm to hundreds of µm. For instance, the skin on one’s ﬁnger is not microscopically smooth, but is instead patterned with bumps and ridges that contact the hair surface. Also, adhesion may be very relevant on the microscale because often many small patches of cuticle scales (5–10 µm long) from two different ﬁbers may be in contact at the same time. Thus, to further the scale effect study and to bridge the gap between the macroscale and nanoscale data, as well as to gain a full understanding of the mechanisms behind the trends, it is worthwhile to look at hair tribology on the microscale as well. This chapter presents a comprehensive study of various hair and skin nanotribological properties as a function of ethnicity, damage, conditioning treatment, and various environments. Nanotribological properties including surface roughness, friction, adhesion, and wear are presented, as well as investigations of scale effects and directionality dependence on friction and adhesion.

24.2 Human Hair, Skin, and Hair Care Products 24.2.1 Human Hair and Skin Figure 24.2 shows a schematic of a human hair ﬁber with its various layers of cellular structure [Zviak, 1986; Robbins, 1994; Feughelman, 1997; Jolles et al., 1997]. Hair ﬁbers (about 50 to 100 µm in diameter) consist of the cuticle and cortex, and in some cases medulla in the central region. All are composed of dead cells, which are mainly ﬁlled with keratin protein. Table 24.3 displays a summary of the chemical species of hair [Chen and Bhushan, 2005a]. Depending on its moisture content, human hair consists of approximately 65–95% proteins, which are condensation polymers of amino acids. The remaining constituents are water, lipids (structural and free), pigment, and trace elements. Among numerous amino acids in human hair, cystine is one of the most important amino acids. The distinct cystine content of various cellular structures of human hair results in a signiﬁcant effect on their physical properties. Cystine has the capacity to cross-link the protein by its intermolecular disulﬁde linkages. A high cystine content corresponds to rich disulﬁde cross-links, leading to high mechanical properties. In addition to disulﬁde bonds, hair is also rich in peptide bonds and the abundant CO- and NH-groups present give rise to hydrogen bonds between groups of neighboring chain molecules. The species responsible for color in hair is the pigment melanin, which is located in the cortex of the hair in granular form.

B. Bhushan · C. LaTorre

40

Fig. 24.2. Schematic of hair ﬁber structure and cuticle sublamellar structure Table 24.3. Summary of chemical species presented in human hair Keratin protein

65–95% (R: functional group)

(Amino acids)

Cystine Lipids

Structural and free

18-MEA

Water

Up to 30%

Pigment and trace elements

Melanin

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

41

An average head contains over 100,000 hair follicles, which are the cavities in the skin surface from which hair ﬁbers grow. Each follicle grows about 20 new hair ﬁbers in a lifetime. Each ﬁber grows for several years until it falls out and is replaced by a new ﬁber. Hair typically grows at a rate on the order of 10 mm/month. 24.2.1.1 The Cuticle The cuticle consists of ﬂat overlapping cells (scales). The cuticle cells are attached at the root end and they point forward the tip end of the hair ﬁber, like tiles on a roof. Each cuticle cell is approximately 0.3 to 0.5 µm thick and the visible length of each cuticle cell is approximately 5 to 10 µm. The cuticle in human hair is generally 5 to 10 scales thick. Each cuticle cell consists of various sublamellar layers (the epicuticle, the A-layer, the exocuticle, the endocuticle and inner layer) and the cell membrane complex (see Fig. 24.2). Table 24.4 displays the various layers of the cuticle, their respective cystine levels [Robbins, 1994], and other details. The outer layer is the epicuticle, which is covered with a thin layer of covalently attached lipid 18-Methyl Eicosanoic Acid (18-MEA) (see Table 24.3). The A-layer is a component of high cystine content (∼ 30%) and located on the outer-facing aspect of each cell. The A-layer is highly cross-linked,which gives this layer considerable mechanical toughness and chemical resilience, and the swelling in water is presumed to be minimal. The exocuticle, which is immediately adjacent to the A-layer, is also of high cystine content (∼ 15%). On the inner facing aspect of each cuticle cell there is a thin layer of material known as the inner layer. Between the exocuticle and inner layer is the endocuticle, which is low in cystine (∼ 3%). The cell membrane complex itself is a lamellar structure, which consists of the inner β-layer, the δ-layer and the outer β-layer. Table 24.4. Various layers of the cuticle and their details Cuticle layer

Cystine component Details

Epicuticle

∼ 12%

18-MEA lipid layer attached to outer epicuticle contributes to lubricity of the hair

A-layer Exocuticle

∼ 30% ∼ 15%

Highly cross-linked Mechanically tough Chemically resilient

Endocuticle

∼ 3%

Inner layer

–

Cell membrane ∼ 2% complex (CMC)

Lamellar structure Consists of inner β-layer, δ-layer, and outer β-layer

42

B. Bhushan · C. LaTorre

Figure 24.3a shows SEM images of virgin Caucasian, Asian and African hair [Wei et al., 2005]. It can be seen that the Asian hair is the thickest (about 100 µm), followed by African hair (about 80 µm) and Caucasian hair (about 50 µm). The visible cuticle cell is about 5 to 10 µm long for the three hair types. A listing of various cross-section dimensional properties is presented in Table 24.5 [Wei et al., 2005]. While Caucasian hair and Asian hair typically have a similar cross-sectional

Fig. 24.3. (a) SEM images of various hair types. (b) SEM images of virgin Caucasian hair at three locations. (c) SEM images of Caucasian, virgin and treated hair [Wei et al., 2005]

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

43

Table 24.5. Variation in cross-sectional dimensions of human hair Shape

Maximum Minimum Ratio diameter diameter D1 /D2 (D1 ) (µm) (D2 ) (µm)

Caucasian Nearly oval 74 Asian Nearly round 92 African Oval-ﬂat 89

47 71 44

1.6 1.3 2.0

Number of Cuticle scale cuticle scales thickness (µm) 6–7 5–6 6–7

0.3–0.5 0.3–0.5 0.3–0.5

Average length of visible cuticle scale: about 5 to 10 µm

Fig. 24.4. AFM images of various virgin hair types [LaTorre and Bhushan, 2005a]

44

B. Bhushan · C. LaTorre

shape (Asian hair being the most cylindrical), African hair has a highly elliptical shape. African hair is much curlier and wavier along the hair ﬁber axis than Caucasian or Asian hair. Figure 24.3b shows the SEM images of virgin Caucasian hair at three locations: near scalp, middle and near tip. Three magniﬁcations were used to show the signiﬁcant differences. The hair near scalp had complete cuticles, while no cuticles were found on the hair near tip. This may be because the hair near the tip experienced more mechanical damage during its life than the hair near the scalp. The hair in the middle experienced intermediate damage, i.e. one or more cuticle scales were worn away, but many cuticles stayed complete. If some substructures of one cuticle scale, like A-layer or A-layer and exocuticle (see Fig. 24.2) are gone, or even worse, one or several cuticle scales are worn away, it is impossible to heal the hair biologically because hair ﬁbers are composed of dead cells. However, it is possible to physically “repair” the damaged hair by using conditioner, a functions of which is to cover or ﬁll the damaged area of the cuticles. Figure 24.3c shows the high magniﬁcation SEM images of virgin and treated Caucasian hair. The endocuticles (shown by arrows) were found in both hair types. In order for the conditioner to physically repair the hair, it is expected to cover the endocuticles. In the case of severely damaged hair, for example, an edge of one whole cuticle scale worn away, the conditioner may ﬁll that damaged edge. In the SEM image of the treated hair in Fig. 24.3c, the substance that stayed near the cuticle edge is probably the conditioner (shown by an arrow). Figure 24.4 shows the AFM images of various virgin hair types, along with the section plots [LaTorre and Bhushan, 2005a]. The arrows point to the position from where the section plots were taken. Each cuticle cell is nearly parallel to the underlying cuticle cell, and they all have similar angles to the hair axis, forming a tile-like hair surface structure. The visible cuticle cell is approximately 0.3 to 0.5 µm thick and about 5 to 10 µm long for all three hair types. 24.2.1.2 The Cortex and Medulla The cortex contains cortical cells and intercellular binding material, or the cell membrane complex. The cortical cells are generally 1 to 6 µm thick and 100 µm long, and run longitudinally along the hair ﬁber axis and take up the majority of the inner hair ﬁber composition [Randebrook, 1964]. The macroﬁbrils (about 0.1 to 0.4 µm in diameter) comprise a major portion of the cortical cells. Each macroﬁbril consists of intermediate ﬁlaments (about 7.5 nm in diameter), previously called microﬁbrils, and the matrix. The intermediate ﬁlaments are low in cystine (∼ 6%), and the matrix is rich in cystine (∼ 21%). The cell membrane complex consists of cell membranes and adhesive material that binds the cuticle and cortical cells together. The intercellular cement of the cell membrane complex is primarily nonkeratinous protein, and is low in cystine content (∼ 2%). The medulla of human hair, if present, generally makes up only a small percentage of the mass of the whole hair, and is believed to contribute negligibly to the mechanical properties of human hair ﬁbers. Figure 24.5a shows the SEM images of a cross-section of virgin hair [Wei et al., 2005] and Fig. 24.5b shows the TEM images of a cross-section of human hair [Swift, 1997].

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

45

Fig. 24.5. (a) SEM images of virgin hair cross-section (in the ﬁgure EXO, END, and CMC stand for exocuticle, endocuticle, and cell membrane complex, respectively) [Wei et al., 2005]. (b) TEM of hair cross-section

24.2.1.3 Skin Skin covers and protects our bodies. The skin at the forehead and scalp areas is of most interest when dealing with human hair, since most of the hair care products are developed speciﬁcally for head hair. The skin of the hand and ﬁngers is also of importance because the “feel” of hair is often sensed by physically touching the ﬁbers with these regions. In general, skin is composed of three main parts: epidermis, dermis, and subcutaneous tissue (L’Oreal); see Fig. 24.6. The epidermis contains four distinct cellular layers: basal, spinous, granular, and horny. In the basal layer, melanocytes deliver the pigment melanin to keratinocytes. Keratinocyte cells that have been corniﬁed are referred to as corneocytes [Pugliese,

46

B. Bhushan · C. LaTorre

Fig. 24.6. Schematic image of human skin structure with different layers: dermis, epidermis, and horny layer (L’Oreal)

1996]. Hexagonally shaped corneocyte cells compose the horny layer, or stratum corneum. The stratum corneum is the outer layer of the skin; at about 15-µm thick, it acts as a mechanical, thermal, and chemical barrier from environmental factors and contamination. The complex organization of corneocytes and the intercellular matrix contribute to the success of the barrier [Wertz and Downing, 1989]. In fact, Wertz et al. [1989] developed a structural model that observes the matrix as a lamellar phase composition of various lipids that provide a glue-like system to provide a barrier effect. The dermis structure is known for its ability to handle most of the physical stresses imposed on the skin, and takes up roughly 90% of the mass [Pugliese, 1996]. The dermis is divided into an outermost papillary layer and underlying reticular layer. 24.2.2 Hair Care: Cleaning and Conditioning Treatments, and Damaging Processes 24.2.2.1 Cleaning and Conditioning Treatments: Shampoo and Conditioner Shampoos are used primarily to clean the hair and scalp of dirt and other greasy residue that can build up after time. Shampoos also have many secondary functions including controlling dandruff, reducing irritation, and even conditioning. Conditioners, on the other hand, are used primarily to give the hair a soft, smooth feel that

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

47

results in easier hair combing. Secondary functions include preventing “ﬂyaway” hair due to static electricity, giving the hair a shiny appearance, and protecting the hair from further damage by forming a thin coating over the ﬁbers. Further developments in marketing and aesthetic factors (brand name, fragrance, feel, and color of the shampoos and conditioners) have created new market segments. In many instances, these factors have become primary reasons for use. 24.2.2.1.1 Shampoo: Constitution and Main Functions The following discussion is based on Gray [2001, 2003]. As stated above, shampoos serve various cleaning functions for the hair and scalp. In the past, typical shampoos were mainly soap-based products. However, soaps did not have very good lathering capability, and often left a residual “scum” layer on the hair that was undesirable and could not be rinsed off. In modern shampoos, advances in chemistry and technology have made it possible to replace the soap bases with complex formulas of cleansing agents, conditioning agents, functional additives, preservatives, aesthetic additives and even medically active ingredients. Table 24.6 shows the most common ingredients of shampoos and their functions. In most modern shampoos, the primary cleansing agents are surfactants. Dirt and greasy residue are removed from the hair and scalp by these surfactants, making them the most important part of the shampoo. Surfactants have great lathering capabilities and rinse off very easily; see Table 24.6 for a full list of features. Surfactant molecules have two different ends, one that is negatively charged and soluble in water (unable to mix with greasy matter), and another that is soluble in greasy matter (unable to mix with water). In general, surfactants clean the hair by the following process. Surfactant molecules encircle the greasy matter on the hair surface. The molecule end, which is soluble in greasy matter, buries itself in the grease, which leaves the water soluble molecule end to face outward with a negative charge. Since the hair ﬁbers are negatively charged as well, the two negative charges repel each other. Thus, the greasy matter is easily removed from the hair surface and rinsed off. Table 24.6. Components of common shampoos and their functions Shampoo component Functions Cleansing agents

Conditioning agents Functional additives Preservatives Aesthetic additives Medically active ingredients

– Produce lather to trap greasy matter, and prevent redeposition – Remove dirt and grease from hair and scalp – Stabilize the mixture and help keep the ingredient network together – Thicken the shampoo to the desired viscosity Condition the hair Control the viscosity and pH levels of the shampoo Prevent decomposition and contamination of the shampoo Enhance color, scent, and luminescence of the shampoo Aid treatment of dandruff or hair loss

48

B. Bhushan · C. LaTorre

24.2.2.1.2 Conditioner: Constitution and Main Functions As stated earlier, many shampoos have certain levels of conditioning agents that mimic the functions of a full conditioner product. Conditioner molecules contain cationic surfactants that give a positive electrical charge to the conditioner. The negative charge of the hair is attracted to the positively charged conditioner molecules, which results in conditioner deposition on the hair; see Fig. 24.7. This is especially true for damaged hair, since damaging processes result in hair ﬁbers being even more negatively charged. The attraction of the conditioner to hair results in a reduction of static electricity on the ﬁber surfaces, and consequently a reduction in the “ﬂy away” behavior. The conditioner layer also ﬂattens the cuticle scales against each other, which improves shine and color. The smooth feel resulting from conditioner use enables easier combing and detangling in both wet and dry conditions (see Table 24.2). Conditioner consists of a gel network chassis (cationic surfactants, fatty alcohols, and water) for superior wet feel and a combination of conditioning actives (cationic surfactants, fatty alcohols, and silicones) for superior dry feel. Figure 24.8 shows the transformation of the cationic surfactants and fatty alcohol mixture into the resulting

Fig. 24.7. Negatively charged hair and the deposition of positively charged conditioner on the cuticle surface [LaTorre et al., 2006]

Fig. 24.8. Conditioner formation from emulsion to gel network

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

49

gel network, which is a frozen lamellar liquid crystal gel phase. The process starts as an emulsion of the surfactants and alcohols in water. The materials then go through a strictly controlled heating and cooling process: the application of heat causes the solid compounds to melt, and the solidiﬁcation process enables a setting of the lamellar assembly molecules in a fully extended conformation, creating a lamellar gel network. When this network interacts with the hair surface, the high concentration of fatty alcohols makes them the most deposited ingredient group, followed by the silicones and cationic surfactants. Typical deposition levels for cationic surfactants, fatty alcohol, and silicone are around 500–800 ppm, 1000–2000 ppm, and 200 ppm, Table 24.7. Combinations of conditioner ingredients and their beneﬁts towards wet and dry feel Gel network chassis for superior wet feel Key ingredients Beneﬁts – Cationic surfactant – Fatty alcohols – Water

– Creamy texture – Ease of spreading – Slippery application feel – Soft rinsing feel

Combination of “conditioning actives” for superior dry feel Key Ingredients Beneﬁts – Silicones – Fatty alcohols – Cationic surfactant

– Moist – Soft – Dry combing ease

Table 24.8. Chemical structure and purpose/function of conditioner ingredients Ingredient

Chemical structure

Purpose/Function

Water Cationic Stearamidopropyl surfactants dimethylamine Behenyl amidopropyl dimethylamine glutamate (BAPDMA) Behentrimonium chloride (BTMAC) CH3 (CH2 )21 N(Cl)(CH3 )3

Fatty alcohols

Stearyl alcohol (C18 OH) Cetyl alcohol (C16 OH)

Silicones

PDMS blend (dimethicone)

– Aids formation of lamellar gel network – Lubricates and static control agent

– Lubricates and moisturizes – Aids formation of lamellar gel network along with cationic surfactant – Primary source of lubrication – Gives hair a soft and smooth feel

B. Bhushan · C. LaTorre

50

respectively. Typical concentrations are approximately 2–5 wt. %, 5–10 wt. %, and 1–10 wt. %, respectively [LaTorre et al., 2006]. The beneﬁts of conditioners are shown in Table 24.7 [LaTorre et al., 2006]. The wet feel beneﬁts are creamy texture, ease of spreading, slippery feel while applying, and soft rinsing feel. The dry feel beneﬁts are moistness, softness, and easier dry combing. Each of the primary conditioner ingredients also have speciﬁc functions and roles that affect the performance of the entire product. Table 24.8 displays the functions of the major conditioner ingredients and their chemical structure [LaTorre et al., 2006]. Cationic surfactants are critical to the forming of the lamellar gel network in conditioners, and also act as a lubricant and static control agent, since their positive charge aids in counteracting the negative charge of the hair ﬁbers. Fatty alcohols are used to lubricate and moisturize the hair surface, along with forming the gel network. Finally, silicones are the main source of lubrication in the conditioner formulation. 24.2.2.2 Damaging Processes In a previous section, we discussed some of the products that aid in “treating” the hair. There are other hair care products and processes that, while creating a desired look or style to the hair, also bring about signiﬁcant damage to the ﬁbers (see Table 24.1). Most of these processes occur on some type of periodic schedule, whether it be daily (combing the hair), or monthly (haircut and coloring at a salon). In general, hair ﬁber damage occurs most readily by mechanical or chemical means, or by a combination of both (chemo-mechanical). Mechanical Damage Mechanical damage occurs on a daily basis for many individuals. The damage results from large physical forces or temperatures, which degrade and wear the outer cuticle layers. Common causes are – – – –

combing (scratching and wearing of the cuticle layers) scratching (usually with ﬁngernails around the scalp) cutting (affects the areas surrounding the ﬁber tips) blowdrying (high temperatures thermally degrade the surface of the hair ﬁbers)

Permanent Wave Treatment Permanent wave treatments have not changed much with the invention of the Cold Wave. Generally speaking, the Cold Wave uses mercaptans (typically thioglycolic acid) to break down disulﬁde bridges and style the hair without much user interaction (at least in the period soon after the perm application) [Gray, 2001]. The Cold Wave process does not need increased temperatures (no thermal damage to the hair), but generally consists of a reduction period (whereby molecular reorientation to the cuticle and cortex occurs via a disulﬁde-mercaptan interchange pathway [Robbins,

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

51

1994]), followed by rinsing, setting of the hair to the desired style, and ﬁnally neutralization to decrease the mercaptan levels and stabilize the style. Chemical Relaxation Commonly used as a means of straightening hair (especially in highly curved, tightly curled African hair), this procedure uses an alkaline agent, an oil phase, and a water phase of a high viscosity emulsion to relax and reform bonds in extremely curly hair. A large part of the ability to sculpt the hair to a desired straightness comes from the breakage of disulﬁde bonds of the ﬁbers. Coloring and Dyeing Hair coloring and dyeing have become extremely successful hair care procedures, due in part to “over-the-counter” style kits that allow home hair care without professional assistance. The most common dyes are para dyes, which contain paraphenylenediamine (PPD) solutions accompanied by conditioners and antioxidants. Hydrogen peroxide (H2 O2 ) is combined with the para dyes to effectively create tinted, insoluble molecules that are contained within the cortex and are not small enough to pass through the cuticle layers, leaving a desired color to the hair. Due to the levels of hydrogen peroxide, severe chemical damage can ensue in the cuticle and cortex. Bleaching Like dyeing, bleaching consists of using hydrogen peroxide to tint the hair. However, bleaching can only lighten the shade of hair color, as the H2 O2 releases oxygen to bind hair pigments [Gray, 2001]. Bleaching may also be applied to limited areas of the hair (such as in highlights) to create a desired look. The chemical damage brought on by bleaching leads to high porosity and severe wear of the cuticle layer.

24.3 Experimental Techniques To date, most information about the detailed structure of human hair was obtained from scanning electron microscope (SEM) and transmission electron microscope (TEM) observations [Swift, 1991; Robbins, 1994; Swift, 1997; Wei et al., 2005]. SEM requires the hair sample to be covered with a very thin layer of a conductive material and needs to be operated under vacuum during both metallization and measurements. Surface metallization and vacuum exposure could potentially induce modiﬁcations to the surface details. Figure 24.5a and b shows typical images of human hair obtained by SEM and TEM, respectively. TEM examinations provide ﬁne detailed internal structure of human hair. However, thin sections of 50–100 nm thickness and heavy metal compounds staining treatment are required for TEM examinations. The cutting of these thin sections with the aid of an ultra-microtome is not an easy task. Moreover, since neither SEM nor TEM techniques can measure

52

B. Bhushan · C. LaTorre

the physical properties (mechanical, tribological, etc.) of various cellular structures of the human hair of interest, nor do they allow ambient imaging conditions, many outstanding issues remain unanswered. AFM has been commonly used for characterization of tribological and mechanical properties of surfaces [Bhushan, 1999a,b, 2002, 2004, 2005]. As a non-invasive technique, AFM can evaluate the effect of sample treatment, since it can be used without requiring any speciﬁc treatment. Most importantly, it can be operated in ambient conditions in order to study the effect of environment. AFM has the potential of being able to show the cellular structure and molecular assembly of hair, for determining various properties of hair, such as friction, adhesion, wear, elastic stiffness and viscoelastic properties, and for investigating the physical behavior of various cellular structures of hair in various environments. To date, most of AFM studies on hair ﬁbers were focused on the surface topographic imaging [Smith and Swift, 2002; LaTorre and Bhushan, 2005a] and friction, adhesion and wear properties [LaTorre and Bhushan, 2005a,b; LaTorre et al., 2006]. A schematic diagram of an AFM imaging a hair ﬁber is shown in Fig. 24.9. AFM/FFM uses a sharp tip with a radius of approximately 30–50 nm. This signiﬁcant reduction in tip-to-sample interaction compared to the macroscale allows the simulation of single asperity contact to give detailed surface information. While there are several AFM operating modes, the so-called “contact mode” is most relevant for studying tribological properties because it allows simultaneous measurement of surface roughness and friction force. When skin comes in contact with hair, actual contact occurs over a large number of asperities. During relative motion, friction and adhesion are governed by the surface interactions that occur at these asperities. To date, much of the work in

Fig. 24.9. Schematic diagram of AFM operation with human hair sample

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

53

the industry has focused on the measurement of macroscale friction, particularly between a skin replica and a hair swatch of interest [Robbins, 1994]. Figure 24.10 shows schematics of typical macroscale and nanoscale test apparatuses.

Fig. 24.10. Comparison of macroscale and micro/nanoscale friction test apparatuses

24.3.1 Experimental Procedure 24.3.1.1 Specimen Mounting Hair specimens were mounted onto AFM sample pucks using Liquid Paper® correction ﬂuid. A thin layer of the ﬂuid was brushed onto the puck, and when the ﬂuid hardened into a tacky state, the hair sample was carefully placed onto it. The Liquid Paper® dries quickly to keep the hair ﬁrmly in place. An optical microscope was

54

B. Bhushan · C. LaTorre

used to preliminarily image the specimen to ensure none of the Liquid Paper® was deposited on the hair surface. Synthetic materials were attached to AFM sample pucks using double-sided adhesive tape. 24.3.1.2 Nanotribological Property Measurements Surface roughness, friction force, and adhesive force measurements were performed using a commercial AFM system (MultiMode Nanoscope IIIa, Digital Instruments, Santa Barbara, CA) in ambient conditions (22 ± 1 ◦ C, 50 ± 5% relative humidity) [LaTorre and Bhushan, 2005a,b; LaTorre et al., 2006]. Square pyramidal Si3 N4 tips of nominal 30–50 nm radius attached to the end of a soft Si3 N4 cantilever beam (spring constant of 0.06 N/m) were used for surface roughness and nanoscale friction force measurements. This softer cantilever was used to minimize damage to the hair. After engagement of the tip with the cuticle surface, the tip was scanned perpendicular to the longitudinal axis of the ﬁber. The tip was centered over the cross-section in order to be at the very top of the ﬁber, so as to negate effects caused by the AFM tip hitting the sides of the hair and adding error to the measurements. In order to minimize scanning artifacts, a scan rate of 1 Hz was used for all measurements. Topographical images to characterize the shape and structure of the various hair types were taken at 5 × 5, 10 × 10, and 20 × 20 µm2 scans at a normal load of 5 nN. These scan sizes were suitable for capturing the surface features of multiple scales and scale edges of the cuticle. To characterize roughness, 2 × 2 µm2 scans of the cuticle surface without edges were used. Surface roughness images shown in this chapter were processed using the ﬁrst-order “planeﬁt” command available in the AFM software, which eliminates tilt in the image. Roughness data, as well as friction force data, were taken after the planeﬁt command was employed. The ﬁrst-order “ﬂatten” command was also used on friction force mappings to eliminate scanning artifacts and present a cleaner image. Friction force mapping of the scan area was collected simultaneously with roughness mapping. Figure 24.10 (bottom diagram) and Fig. 24.11 show the AFM tip scanning over the hair surface for untreated hair and hir treated with conditioner, respectively. The effects of the conditioner can be examined by comparing the friction information. A quantitative measure of the friction force was calculated by ﬁrst calibrating the force based on a method by Bhushan [Bhushan, 1999a, 2002]. The normal load was varied from 5 nN to 45 nN in roughly 5 nN increments, and a friction force measurement was taken at each increment. By plotting the friction force as a function of the normal force, the average coefﬁcient of friction was determined from the slope of the least squares ﬁt line of the data. Adhesive force measurements were made with square pyramidal Si3 N4 tips attached to the end of a Si3 N4 cantilever beam (spring constant of 0.58 N/m), using the force calibration plot technique. In this technique, the AFM tip is brought into contact with the sample by extending the piezo vertically, then retracting the piezo and calculating the force required to separate the tip from the sample. This method is described in detail by Bhushan [Bhushan, 1999a,b, 2002, 2004].

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

55

Fig. 24.11. Interactions between the AFM tip and conditioner on the cuticle surface for treated hair [LaTorre and Bhushan, 2005b]

The cantilever deﬂection is plotted on the vertical axis against the Z-position of the piezo scanner in a force calibration plot, Fig. 24.12. As the piezo extends (from right to left in the graph), it approaches the tip and does not show any deﬂection while in free air. The tip then approaches within a few nanometers (point A) and becomes attached to the sample via attractive forces at very close range (point B). After initial contact, any extension of the piezo results in a further deﬂection of the tip. Once the piezo reaches the end of its designated ramp size, it retracts to its starting position (from left to right). The tip goes beyond zero deﬂection and enters the adhesive regime. At point C, the elastic force of the cantilever becomes

Fig. 24.12. Typical force calibration plot for Caucasian virgin hair. Contact between the tip and hair occurs at point B. At point C, the elastic force of the cantilever becomes equivalent to the adhesive force, causing the cantilever to snap back to D

56

B. Bhushan · C. LaTorre

equivalent to the adhesive force, causing the cantilever to snap back to point D. The horizontal distance between A and D multiplied by the stiffness of the cantilever results in a quantitative measure of the adhesive force [Bhushan, 1999a,b, 2002, 2005]. The force calibration plot allows for the calculation of an adhesive force at a distinct point on the sample surface. Consequently, by taking a force calibration plot at discrete sampling intervals over an entire scan area, a resulting adhesive force mapping (also known as force-volume mapping) can be created to display the variation in adhesive force over the surface [Bhushan and Dandavate, 2000]. In this work, plots were taken at 64 × 64 distinct points over a scan area of 2 × 2 µm2 for all hair types and ethnicities. Current Digital Instruments software for the Multimode AFM does not allow direct calculation of the adhesive force from the force-volume maps. Thus, a custom program coded in Matlab was used to display the forcevolume maps. The adhesive force for each force calibration plot was obtained by multiplying the spring constant with the horizontal distance (in the retract mode) traveled by the piezo from the point of zero applied load to the point where the tip snaps off. 24.3.1.3 Relative Humidity, Temperature, Soaking, and Durability Measurements A humidity-temperature detector was used to monitor the humidity inside a Plexiglas test chamber enclosing the AFM system. An experimental setup was used to control the humidity inside the chamber. Measurements were taken at nominal relative humidities of 5, 50, and 80%. Hair ﬁbers were placed at each humidity for approximately 2 hours prior to measurements. A homemade thermal stage was used to conduct temperature effect measurements at 20, 37, 50, and 80 ◦ C. Hair ﬁbers were exposed to each temperature condition for approximately 30 minutes prior to testing. Soak tests were performed as follows. A dry hair ﬁber was taken from a swatch, and a sample was cut from the ﬁber (approximately 7 mm long) for coefﬁcient of friction measurements. An adjacent sample was also taken from the ﬁber and placed in a small beaker ﬁlled with de-ionized water. The sample was subjected to the aqueous environment for 5 minutes, which is representative of a typical exposure time when showering/bathing, then immediately analyzed with AFM. It should be noted that hair becomes saturated when wet in about 1 minute and remains saturated if kept in a humid environment. It was determined from unpublished results that if the wet hair was exposed to the ambient environment for more than 20 minutes while in the AFM, the hair became dry and the coefﬁcient of friction became similar to that of dry hair. Thus, coefﬁcient of friction measurements were made within a 20 minute time frame for each sample. In order to simulate scratching that can occur on the surface of the hair and its effect on the friction force on the cuticle surface, a durability test was conducted using a stiff silicon tip (spring constant of 40 N/m). A load of 10 µN was used on a 2 µm section of the cuticle. A total of 1000 cycles were performed at 2 Hz. Measurements were conducted using an AFM. The friction force signal was recorded with respect to cycling time.

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

57

24.3.1.4 Microscale and Macroscale Friction and Adhesion Measurements for the Study of Scale Effects Microscale coefﬁcient of friction and adhesive force measurements were performed using a commercial AFM system (MultiMode Nanoscope IIIa, Digital Instruments, Santa Barbara, CA) in ambient conditions (22 ± 1 ◦ C, 50 ± 5% relative humidity). For microscale data, a 4 µm radius silica ball was mounted on a Si3 N4 cantilever (spring constant of 0.58 N/m). A quantitative measure of friction force was calculated as described previously, except that the normal load was varied over a larger range (0–180 nN with microscale tip, compared to 0–45 nN with nanoscale tip). Adhesive force measurements were made with the microscale AFM tips using the force calibration plot technique, as described previously. For the macroscale coefﬁcient of friction, the tests were conducted using a ﬂaton-ﬂat tribometer under reciprocating motion, with measurement techniques similar to those described in Bhushan et al. [2005]. Macroscale coefﬁcient of friction data for virgin and virgin treated hair were taken from Bhushan et al. [2005]. The values for other hair types were taken in part from the indexed coefﬁcient of friction values presented in LaTorre and Bhushan [2005b], which were transformed into actual values by the following procedure. A comparison was made between nanoscale data for chemo-mechanically damaged and chemically damaged hair, and a 15% difference was found. The same % difference was assumed on the macroscale, such that the value for chemo-mechanically damaged hair presented in Bhushan et al. [2005] was used to calculate a value for the chemically damaged hair coefﬁcient of friction. From there, the indexed values for the chemically damaged treated (one cycle and 3 cycle) hair were scaled based on the newly found chemically damaged hair value. 24.3.2 Hair and Skin Samples For the research reported in this chapter, all hair samples were received from Procter & Gamble (Cincinnati, OH) and Procter & Gamble Far East (Kobe, Japan) and prepared as per Appendix A. Samples arrived as hair swatches approximately 0.3 m long. Although the exact location from the root is unknown, it is estimated that hair samples used for testing were between 0.1 to 0.2 m from the scalp. Table 24.9 presents a list of all samples used. The main hair categories of interest are: virgin (untreated), virgin (treated with one cycle of commercial conditioner), chemo-mechanically damaged (untreated), chemically damaged (untreated), and chemically damaged (treated with 1 or three cycles of commercial conditioner or a matrix of conditioner ingredients). Virgin samples are considered to be baseline specimens and are deﬁned as having an intact cuticle and absence of chemical damage. Chemo-mechanically damaged hair ﬁbers have been exposed to one or more cycles of coloring and permanent wave treatment, washing, and drying, as well as combing (to contribute mechanical damage), which are representative of common hair management and alteration. In the case of African damaged hair samples, chemical damage occurred only by chemical straightening. Chemically damaged ﬁbers have not been exposed to the combing sequence in their prepa-

B. Bhushan · C. LaTorre

58 Table 24.9. Hair and skin samples Sample

Type

Caucasian hair

– Virgin – Virgin, treated (1 cycle commercial conditioner) – Chemo-mechanically damaged – Chemically damaged – Chemically damaged, treated (1 cycle commercial conditioner) – Chemically damaged, treated (3 cycles commercial conditioner) – Chemically damaged, treated (various ingredients, see Table 24.10)

Asian hair

– Virgin – Virgin, treated (1 cycle commercial conditioner) – Chemo-mechanically damaged

African hair

– Virgin – Virgin, treated (1 cycle commercial conditioner) – Chemo-mechanically damaged

Synthetic materials

– Artiﬁcial collagen ﬁlm (hair) – Polyurethane ﬁlm (skin) – Human skin (putty replica)

ration. All treated hair samples were treated with either one or three rinse/wash cycles of a conditioner similar to a Procter & Gamble commercial product, or were treated with various combinations of surfactant, fatty alcohol, and silicone types and deposition levels, presented in the matrix of Table 24.10 [LaTorre et al., 2006]. Two different types of cationic surfactants were used: behentrimonium chloride (BTMAC) and behenyl amidopropyl dimethylamine (BAPDMA). Only one group of fatty alcohols were used for all samples. In the last set of samples, two different silicones were used: a PDMS (blend with high MW) silicone and an amino silicone. Collagen ﬁlm is typically used as a synthetic hair material for testing purposes. Polyurethane ﬁlms represent synthetic human skin. They are cast from human skin and have a similar surface energy, which also makes them suitable test specimens when real skin cannot be used. To characterize the surface roughness of human skin, it was also replicated using a two-part silicone elastomer putty (DMR-503 Replication Putty, Dynamold, Inc., Fortworth, TX). The thickness of the ﬁlm was approximately 3 mm. In order to simulate hair conditioner-skin contact in AFM experiments, it is important to have the contact angle and surface energy of an AFM tip close to that of skin. Table 24.11 shows the contact angles and surface energies of materials important to the nanocharacterization of the hair samples: Untreated human hair; PDMS, which is used in conditioners; skin, which comes into contact with hair; and Si3 N4 ﬁlm, which in the form of an AFM tip is used for nanotribological measurements [LaTorre et al., 2006]. The static contact angle of Si3 N4 ﬁlm with high purity deionized water was measured in air by a sessile-drop method using a contact angle goniometer (Model 100, Rame-Hart Inc., Mountain Lakes, NJ, USA). 5 µL of

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

59

Silicones (weight %)

Fatty alcohols (weight %)

Cationic surfactants (weight %)

Table 24.10. Matrix of hair samples treated with various combinations of ingredients

Ingredient

Damaged treated (commercial conditioner)

(BTMAC surfactant, no silicone)

(PDMS blend silicone, low deposition)

(PDMS blend silicone, high deposition)

(Amino silicone, low deposition)

(Amino silicone, high deposition)

(BAPDMA surfactant, no silicone)

Behentrimonium chloride (BTMAC)

−

×

×

×

×

×

−

Behenyl amidopropyl dimethylamine (BAPDMA)

−

−

−

−

−

−

×

Stearamidopropyl dimethylamine (SAPDMA)

×

−

−

−

−

−

−

Fatty alcohols

×

×

×

×

×

×

×

PDMS blend (Dimethicone)

×

−

−

−

−

−

−

PDMS blend (with blend of high MW)

−

−

Low

High

−

−

−

Amino silicone

−

−

−

−

Low

High

−

Deposition levels for conditioner ingredients on hair Cationic surfactant deposition level

Similar to commercial conditioner

×

×

×

×

×

×

Fatty alcohol deposition level

Similar to commercial conditioner

×

×

×

×

×

×

Silicone deposition level

Similar to commercial conditioner

None

Low

High

Low

High

None

deionized water was applied on two samples using a micropipette, and three contact angle measurements were taken on each sample and averaged.

24.4 Results and Discussion 24.4.1 Surface Roughness, Friction, and Adhesion for Various Ethnicities of Hair Topographical images of Caucasian, Asian, and African hair were taken up to scan sizes of 20 × 20 µm2 , as shown in Fig. 24.13. Lighter areas of the images correspond to higher topography, and darker areas correspond to lower topography. Only virgin and chemo-mechanically damaged hair are shown in Fig. 24.13, because virgin treated samples closely resemble virgin hair samples. One can see the variation in cuticle structure even in virgin hair. Cracking and miscellaneous damage at the cuticle edges is evident at both virgin and chemo-mechanically damaged conditions.

B. Bhushan · C. LaTorre

60

Table 24.11. Contact angle and surface energy of relevant materials associated with nanotribological characterization of hair

Human hair

– untreated – damaged brown coloring blond coloring

PDMS (bulk) Human skin

– forehead – forearm – ﬁnger

Si3 N4 ﬁlm

Contact angle (◦ )

Surface energy (N/m)

103a 100b

0.024a 0.028b

60b 55b

0.038b 0.047b

105c

0.020d

55e 88e 84f 74f 58g (before soap-washing) 104g (after soap-washing)

0.043e 0.038e 0.029f 0.024f 0.027g

35h

0.047i

a

Molina et al. [2001] LaTorre et al. [2006] c Bhushan and Burton [2005] d Jalbert et al. [1993] e Lerebour et al. [2000] f Schott [1971] g Ginn et al. [1968] h LaTorre et al. [2006] i Yanazawa [1984] b

In virgin hair, the damage is likely to be caused by mechanical damage resulting from daily activities such as washing, drying, and combing. Most of the virgin cuticle scales that were observed, however, were relatively intact. Long striations similar to scratches, and “scale edge ghosts” (outlines of a former overlying cuticle scale edge left on the underlying scale before it was broken away) were found on the surface. In some instances, the areas surrounding the cuticle edges appeared to show residue or debris on the surface, which were most likely remnants of a previous cuticle or the underside of the cuticle edges that were now exposed (such as the endocuticle). Caucasian and Asian virgin hair displayed a similar surface structure, while the African hair samples showed more signs of endocuticular remains along the scale edges. One can also see more curvature in the cuticle scales of African hair, which is attributed to its elliptical cross-sectional shape and curliness, which can partially uplift the scales in different places. With respect to chemo-mechanically damaged hair, it is observed that several regions seem to exist in these hair samples, ranging from intact cuticle scales to high levels of wear on the surface. In many cases, these regions occur side by side. This wide variation in the chemo-mechanically damaged cuticle structure results in a wider range of tribological properties on the micro/nanoscale for these ﬁbers. Caucasian and Asian chemo-mechanically damaged

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

61

Fig. 24.13. Surface roughness of virgin and chemo-mechanically damaged Caucasian, Asian, and African hair at various scan sizes [LaTorre and Bhushan, 2005a]

hair showed more worn away cuticle scales than in chemo-mechanically damaged African hair, which showed mostly endocuticle remnants. This is most likely due to the different effects that chemical straightening has on the hair versus multiple cycles of perming the hair.

62

B. Bhushan · C. LaTorre

A more focused look into roughness and friction on the cuticle surface can be found by comparing Caucasian, Asian, and African virgin and chemo-mechanically damaged hair, Fig. 24.14a and b, and virgin and virgin treated hair, Fig. 24.15a and b. Virgin hair was used as the baseline to compare variations in roughness and friction force against modiﬁed hair (chemo-mechanically damaged or virgin treated). Scan

Fig. 24.14a. Surface roughness and friction images for virgin and chemo-mechanically damaged Caucasian, Asian hair at 5 and 10 µm2 scan sizes. Shown above each image is a cross-section taken at the corresponding arrows to demonstrate roughness and friction force information

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

63

Fig. 24.14b. Surface roughness and friction force images for virgin and chemo-mechanically damaged African hair at 5 and 10 µm2 scan sizes. Shown above each image is a cross-section taken at the corresponding arrows to demonstrate roughness and friction force information [LaTorre and Bhushan, 2005a]

sizes of 5 × 5 µm2 and 10 × 10 µm2 are displayed. Above each AFM and FFM image there are cross-sectional plots of the surface (taken at the accompanying arrows) corresponding to surface roughness or friction force, respectively. From the surface roughness images, the step heights of one or more cuticle edges can be clearly seen. Step heights range from approximately 0.3 to 0.5 µm. If the surface is assumed to have Gaussian height distribution and an exponential autocorrelation function, then the surface can be statistically characterized by just two parameters: a vertical descriptor, height standard deviation σ, and a spatial descriptor, correlation distance β ∗ [Bhushan, 1999a,b, 2002]. The standard deviation σ is the square root of the arithmetic mean of the square of the vertical deviation from the mean line. The correlation length can be referred to as the length at which two data points on a surface proﬁle can be regarded as being independent, thus serving as a randomness measure [Bhushan, 2002]. Table 24.12 displays these roughness parameters for each ethnicity as a function of hair type (virgin, chemo-mechanically damaged, and virgin treated) [LaTorre and Bhushan, 2005a]. Virgin hair was shown to generally have the lowest roughness values, with virgin treated hair closely resembling virgin hair. Chemo-mechanically damaged hair showed a signiﬁcantly higher standard deviation of surface height. This variation is expected because of the nonuniformity of the mechanical and chemical damage that occurs throughout a whole head of hair, as well as in each individual ﬁber. This is in agreement with the images of chemo-mechanically damaged hair shown previously, where regions of intact cuticle and severe degradation of the surface are intermingled. The trends observed

64

B. Bhushan · C. LaTorre

Fig. 24.15a. Surface roughness and friction images for virgin and virgin treated Caucasian, Asian hair at 5 and 10 µm2 scan sizes. Shown above each image is a cross-section taken at the corresponding arrows to show roughness and friction force information

for standard deviation were not as evident for the correlation length β ∗ . For each ethnicity, chemo-mechanically damaged and virgin treated hair showed similar β ∗ values. From Fig. 24.14a and b, friction forces are generally seen to be higher on chemo-mechanically damaged hair than on virgin hair. Although friction forces

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

65

Fig. 24.15b. Surface roughness and friction force images for virgin and virgin treated African hair at 5 and 10 µm2 scan sizes. Shown above each image is a cross-section taken at the corresponding arrows to demonstrate roughness and friction force information [LaTorre and Bhushan, 2005a] Table 24.12. Surface roughness, coefﬁcient of friction, and adhesive force values of virgin, chemo-mechanically damaged, and virgin treated (1 cycle commercial conditioner) hair at each ethnicity Virgin hair

Chemo-mechanically damaged hair

Virgin treated hair (commercial conditioner)

Caucasian Asian African

Surface roughness parameters (σ (nm), β ∗ (µm)) σ (nm) β ∗ (µm) σ (nm) β ∗ (µm) 12 ± 8 0.61 ± 0.3 17 ± 10 1.0 ± 0.3 9.7 ± 4 0.73 ± 0.3 33 ± 15 0.94 ± 0.3 12 ± 5 0.92 ± 0.3 21 ± 16 0.78 ± 0.3

σ (nm) 12 ± 4 7.1 ± 0.1 11 ± 4

Caucasian Asian African

Average coefﬁcient of friction µ 0.02 ± 0.01 0.13 ± 0.05 0.03 ± 0.01 0.13 ± 0.04 0.04 ± 0.02 0.14 ± 0.08

0.03 ± 0.01 0.06 ± 0.04 0.05 ± 0.01

Caucasian Asian African

Adhesive force Fm (nN) 25 16 31 18 35 38

32 79 63

β ∗ (µm) 0.90 ± 0.3 0.97 ± 0.3 0.89 ± 0.2

were similar in magnitude, it was observed that the friction force on the cuticle surface of chemo-mechanically damaged hair showed a much larger variance, which contributed to the higher friction values. Another contribution to the higher friction

B. Bhushan · C. LaTorre

66

could be that the tiny peaks developed after damage also create a ratchet mechanism on a nanoscale, which affects the friction between the AFM tip and the surface. These peaks could then add to the friction signal. The damage of the hair by chemical and mechanical means has shown high reproducibility in the laboratory in terms of structure alteration, which explains the similar friction properties regardless of ethnicity for chemo-mechanically damaged hair. With virgin and virgin treated hair, however, it is unknown what prior mechanical damage and sun exposure the ﬁbers have seen, and this depends largely on the individuals. Thus, across ethnicity there is variability in friction force for those hair samples. Perhaps the most notable difference between virgin and virgin treated hair ﬁbers can be seen in the friction force mappings of Fig. 24.15a and b. Although quite comparable in surface roughness, a close examination of the virgin treated hair surface shows an increase in friction force, usually only surrounding the bottom edge of the cuticle. This was unlike virgin hair, where friction generally remained constant along the surface, and unlike chemo-mechanically damaged hair, where there was large variability that was random over the entire surface. Figure 24.16 presents friction force curves as a function of normal load for Caucasian virgin, chemo-mechanically damaged, and virgin treated hair to further illustrate the previous discussion. One can see a relatively linear relationship between the data points for each type of hair sample. When plotted in such a way, the coefﬁcient of friction is determined by the slope of the least squares ﬁt line through the data. If this line is extended to intercept the horizontal axis, then a value for adhesive force can also be calculated, since friction force F is governed by the relationship F = µ(W + Fa ) ,

(24.1)

where µ is the coefﬁcient of friction, W is the applied normal load, and Fa is the adhesive force [Bhushan, 1999a, 2002]. One explanation for the increase in friction force of virgin treated hair on the micro/nanoscale is that during tip contact, meniscus forces between the tip and the conditioner/cuticle become large as the tip rasters over the surface, causing an increase in the adhesive force. This adhesive force is of the same magnitude as the normal load, which makes the adhesive force contribution to friction rather signiﬁcant. Thus, at sites where conditioner is accumulated on the surface (namely around the cuticle scale edges), the friction force actually increases. On the macroscale, however, the adhesive force is much lower in magnitude than the applied normal load, so the adhesive force contribution to friction is negligible over the hair swatch. As a result, virgin treated hair shows a decrease in friction force on the macroscale, which is opposite to the micro/nanoscale trend. However, the friction and adhesion data on the micro/nanoscale are useful, because they relate to the presence of conditioner on the cuticle surface and allow one to obtain an estimate of conditioner distribution. It is also observed from Fig. 24.16 that while chemo-mechanically damaged hair displays a higher friction force on the application of the normal load, and consequently a higher coefﬁcient of friction, chemo-mechanically damaged hair friction is not as strongly dependent on adhesive force contribution as the virgin and virgin

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

67

Fig. 24.16. Average coefﬁcient of friction values for virgin, chemo-mechanically damaged, and treated hair at each ethnicity. Error bars represent ±1σ on the average coefﬁcient value. The bottom plot shows friction force vs. normal load curves displaying typical values for virgin, damaged, and virgin treated Caucasian hair [LaTorre and Bhushan, 2005a]

treated hair. Average values for µ were calculated and are compiled in Fig. 24.17 and Table 24.12 for all hair ethnicities and types [LaTorre and Bhushan, 2005a]. Error bars represent ±1σ on the average coefﬁcient value. The coefﬁcients of friction for virgin, chemo-mechanically damaged, and virgin treated Caucasian hair are 0.02, 0.13, and 0.03, respectively. For virgin, chemo-mechanically damaged, and virgin treated Asian hair, the coefﬁcients of friction are 0.03, 0.13, and 0.06, respectively. Finally, virgin, chemo-mechanically damaged, and virgin treated African hair coefﬁcients of friction are 0.04, 0.14, and 0.05, respectively. Chemo-mechanically damaged hair presents the highest coefﬁcient of friction, but also displays the largest standard deviation, due to the large variations in chemical and mechanical damage that each hair or hair bundle experiences. The coefﬁcient of friction of virgin treated

68

B. Bhushan · C. LaTorre

Fig. 24.17. Surface roughness, coefﬁcient of friction, and adhesive force data for virgin, chemomechanically damaged, and virgin treated hair at each ethnicity [LaTorre and Bhushan, 2005a]

hair is slightly larger than that of virgin hair for all ethnicities. While the coefﬁcient of friction is similar in virgin and virgin treated hair, the adhesive force contribution to friction for Caucasian virgin treated hair is higher than in Caucasian virgin hair, when calculated according to the method described above. However, this was not always the trend for Asian and African virgin treated hair samples. It should be noted that since in friction force measurement the tip moves laterally over the surface, this might cause a smearing out of the conditioner layer, which accounts for the inconsistent trend. In the adhesive force mappings described in the next section, where determination of the adhesive force does not depend on this lateral movement, all virgin treated hair samples showed higher adhesion than their virgin hair counterparts.

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

69

Fig. 24.18. Force-volume maps of virgin, chemo-mechanically damaged, and virgin treated hair at each ethnicity. Examples of the individual force calibration plots, which make up the FV maps, are presented for Caucasian hair of each type [LaTorre and Bhushan, 2005a]

A force calibration plot (FCP) technique and resulting adhesive force maps (commonly called force-volume maps) can be used to understand the adhesive forces between the AFM tip and the sample [Bhushan, 1999b, 2005; Bhushan and

70

B. Bhushan · C. LaTorre

Dandavate, 2000; Liu and Bhushan, 2003]. Shown in Fig. 24.18 are FV maps and an example of the individual force calibration plots from which the maps were created. The adhesive force distribution for chemo-mechanically damaged hair was shown to be comparable to virgin hair adhesive force values, but slightly lower. A signiﬁcant increase in the adhesive force over the entire mapping was found in all cases of virgin treated hair as compared to virgin hair, especially in Asian and African hair. Conditioner distribution can be seen from these images. This technique shows promise of being very useful in further study of the distribution of materials and hair care products on the surface of the hair. A typical value for the adhesive force of each FV map was calculated. Values are shown in the plot of Fig. 24.17 presented earlier, along with surface roughness and coefﬁcient of friction data for all hair samples. Adhesive force values are also tabulated in Table 24.12. 24.4.2 Surface Roughness, Friction, and Adhesion for Virgin and Chemically Damaged Caucasian Hair (with and without Commercial Conditioner Treatment) The hair surface is negatively charged and can be damaged by a variety of chemical (permanent hair waving, chemical relaxation, coloring, bleaching) and mechanical (combing, blowdrying) factors [Robbins, 1994; Bolduc and Shapiro, 2001; Gray, 2001]. Figure 24.19 shows the transformation and wear of the cuticles scales before

Fig. 24.19. The effect of damage to the cuticle scales and the deposition of conditioner on the cuticle surface. The cross-section of the hair with and without conditioner is shown below [LaTorre et al., 2006]

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

71

and after damage. Chemical damage causes parts of the scales to fracture and reveal underlying cuticle remnants. Conditioner application provides a protective coating to the hair surface to prevent future damage. Shown in Fig. 24.20 are surface roughness and friction force plots for virgin, virgin treated, chemically damaged, chemically damaged treated (one cycle conditioner), and chemically damaged treated (three cycles conditioner). Above each AFM and FFM image are cross-sectional plots of the surface (taken at the accompanying arrows) corresponding to surface roughness and friction force, respectively. Although virgin and virgin treated hair are quite comparable in surface roughness maps, examination of the treated hair surface shows an increase in friction force, especially in the area surrounding the scale edge bottom level. These frictional patterns observed in treated hair were not like anything observed in the virgin or chemically damaged cases. Images of all hair types have shown friction variation due to edge contributions and cuticle mechanical damage that has left only remnants of the cuticle sublayer (such as the endocuticle). Further investigation of the corresponding

Fig. 24.20. Surface roughness and friction force images for virgin, virgin treated, damaged, damaged treated (one cycle conditioner), and damaged treated (three cycles conditioner) hair at 5 µm scan sizes [LaTorre and Bhushan, 2005b]

72

B. Bhushan · C. LaTorre

treated hair roughness images showed that this increase in friction was not due to a signiﬁcant change in surface roughness either. One explanation for the increase in friction force of treated hair on the micro/nanoscale is that during tip contact, meniscus forces between the tip and the conditioner/cuticle become large as the tip rasters over the surface, causing an increase in the adhesive force. This adhesive force is of the same magnitude as the normal load, which makes the adhesive force contribution to friction rather signiﬁcant. Thus, at sites where conditioner is accumulated on the surface (namely around the cuticle scale edges), friction force actually increases. In general, friction forces are higher on chemically damaged hair than on virgin hair. Although friction forces were similar in magnitude, it was observed that the friction force on the cuticle surface of chemically damaged hair showed a much larger variance, which contributed to the higher friction values. Chemically damaged treated hair shows a much stronger afﬁnity to the conditioner. It is well-known that the cuticle surface of hair is negatively charged. This charge becomes even more negative with the application of chemical damage to the hair. As a result, the positively charged particles of the conditioner have an even stronger attraction to the chemically damaged surface, which explains the increased presence of conditioner (and corresponding higher friction forces) when compared to virgin treated hair. With the application of three conditioner cycles on chemically damaged treated hair, the friction force is still higher near the cuticle edge, however it is also increased all over the cuticle surface, showing a more uniform placement of the conditioner. Figure 24.21 shows adhesive force maps for the various hair types, which gives a measurement of adhesive force variation on the surface. Virgin treated hair shows a higher adhesive force than virgin hair due to the meniscus effects that come about from AFM tip interaction with the conditioner on the cuticle surface [Bhushan, 1999a,b, 2002]. The same trend is true and even more evident for chemically damaged treated hair compared to chemically damaged hair. A possible reason that one cycle of conditioner on chemically damaged hair showed a higher average adhesive force than three cycles could be because the three cycles generally place the conditioner more uniformly on the surface rather than accumulating it mostly near the bottom surface near the cuticle edge, which is where the adhesive force maps were generally taken. Nevertheless, the increased adhesive force shown in the plots is a clear indication of conditioner present on the hair surface, and its localization can be observed. Figure 24.22 presents surface roughness, coefﬁcient of friction, and adhesive force plots for the various virgin and chemically damaged hair types discussed above. The data is also presented in Table 24.13 [LaTorre and Bhushan, 2005b]. Surface roughness for human hair is characterized by a vertical descriptor, height standard deviation σ, and a spatial descriptor, correlation distance β ∗ [Bhushan, 1999a,b, 2002]. Virgin and virgin treated hair showed similar σ values, while β ∗ was higher in virgin treated hair. Chemically damaged hair and both types of chemically damaged treated hair showed similar roughness values, although σ was higher for the treated cases. The chemically damaged hair presented in this work is different from the chemo-mechanically damaged hair studied in LaTorre and Bhushan [2005a]. It seems that chemically damaging the surface does not lead to as much wear and surface roughness increase as the combination of both chemical and mechanical

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

73

Fig. 24.21. Adhesive force maps displaying variations in the adhesive force on the cuticle surface of various hair types [LaTorre and Bhushan, 2005b]

Fig. 24.22. Surface roughness, coefﬁcient of friction, and adhesive force plots for various hair types [LaTorre and Bhushan, 2005b]

B. Bhushan · C. LaTorre

74

Table 24.13. Nanotribological parameters for virgin, chemically damaged, and treated hair Surface roughness, coefﬁcient of friction, and adhesive force for virgin and chemically damaged hair, with and without conditioner treatment Hair type Surface roughness Coefﬁcient Adhesive force σ (nm) β ∗ (µm) of friction (nN) Virgin Virgin, treated Chemically damaged Damaged, treated (1 cycle) Damaged, treated (3 cycles)

12 ± 8 12 ± 4 8.4 ± 2 13 ± 4 11 ± 2

0.61 ± 0.2 0.90 ± 0.3 0.83 ± 0.2 0.75 ± 0.3 0.80 ± 0.2

0.02 ± 0.01 0.03 ± 0.01 0.13 ± 0.06 0.05 ± 0.02 0.04 ± 0.02

25 ± 5 32 ± 5 39 ± 0.5 66 ± 0.7 54 ± 33

damage. Thus, it should be noted (and it is understandable) that there are differences between the chemo-mechanically and chemically damaged hair. The coefﬁcients of friction of virgin and virgin treated hair are similar, but slightly higher for the treated cases. Chemically damaged hair shows a much higher coefﬁcient of friction, and with more variation in the values, since the chemical damage varies throughout each individual ﬁber. An interesting ﬁnding was that, contrary to the virgin and virgin treated hair results, the coefﬁcient of friction for chemically damaged hair decreased with application of the conditioner treatment (both one and three cycles). One possible explanation is that because the stronger negative charge on chemically damaged hair results in a better attraction of the conditioner, this leads to a higher adhesive force but, more importantly, a lower shear strength on the surface. This creates an overall effect of lubrication and ultimately lowers the coefﬁcient of friction. 24.4.2.1 Effect of Relative Humidity, Temperature, Soaking, and Durability Measurements Figure 24.23 displays the effect of relative humidity on the friction force and the adhesive force. The coefﬁcient of friction remained relatively constant for virgin and virgin treated hair. However, chemically damaged hair experienced a large increase in the coefﬁcient of friction at high humidity, while chemically damaged treated hair experienced the opposite trend. This clearly shows that heavy moisture in the air plays affects the frictional properties of chemically damaged hair. When combined with conditioner, a lubricating effect once again dominates as the water helps form a liquid layer that is more easily sheared. In terms of the adhesive force, most samples showed a decrease in adhesive force with high humidity. It is expected that as water builds up on a surface, meniscus effects diminish and as a result do not readily contribute to the adhesive force. Thus, the adhesive force is expected to decrease at very high humidity. Figure 24.24 displays the effect of temperature on the friction force and the adhesive force. The coefﬁcient of friction generally decreased with increasing temperature. As the hair ﬁber heats up, conditioner present on the surface decreases in viscosity, causing a thinner ﬁlm and a lower friction force. The lower friction force ultimately leads to lower coefﬁcient of friction values. The adhesive force was shown to decrease with increasing temperature as well. This was especially evident

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

75

Fig.24.23.Effect of relative humidity on nanotribological properties of various hair types [LaTorre and Bhushan, 2005b]

for treated hair ﬁbers, whereas a large adhesive force at room temperature decreased rapidly to adhesion values similar to non-treated ﬁbers. It is most likely that at higher temperatures the thinning conditioner layer causes a reduced surface tension, which directly relates to the drop in the adhesive force. Virgin, chemically damaged, and chemically damaged treated hair samples were soaked in de-ionized water for ﬁve minutes. Their corresponding coefﬁcient of friction was measured and compared to coefﬁcient of friction values for dry samples that were adjacent to the wet samples on the respective hair ﬁber. Figure 24.25 shows the results for two hair samples of each hair type. Virgin hair exhibits a decrease

76

B. Bhushan · C. LaTorre

Fig. 24.24. Effect of temperature on nanotribological properties of various hair types [LaTorre and Bhushan, 2005b]

in the coefﬁcient of friction after soaking. Virgin hair is more hydrophobic (see Table 24.11), so more water is present on the surface and this results in a lubrication effect after soaking. Chemically damaged hair tends to be hydrophilic due to the chemical degradation of the cuticle surface, and this results in absorption of water after soaking. This softens the hair, which leads to higher friction, even with conditioner treatment. This is yet another indication that virgin and chemically damaged hair types have signiﬁcantly different surface properties which in many cases results in opposite trends for their nanoscale tribological properties. The adhesive force for

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

77

Fig. 24.25. Effect of soaking in de-ionized water on coefﬁcient of friction and adhesive force for virgin, chemically damaged, and chemically damaged treated (one cycle commercial) hair [LaTorre et al., 2006]

virgin hair remained approximately the same before and after soaking, while it decreased for chemically damaged and chemically damaged treated hair after soaking. Figure 24.26 shows the durability effects on the friction force for various hair types. Above the graph are pictures of unworn and worn virgin hair, with the cuticle edge serving as a reference point. Before testing, the surface is relatively smooth and void of any large debris or wear. After 1000 cycles at approximately 10 µN load with a stiff silicon AFM tip, however, the interaction has caused degradation and wear (scratch) marks on the cuticle scale. This is the type of wear one could potentially see if hair were to come in contact with sand from a day spent at the beach, among other activities. Virgin hair shows an obvious increase in friction force signal as the scratch mark digs further into the surface. By this time, the lubricious lipid layer on the surface of the virgin cuticle has been worn away and the friction force comes close to the magnitude of chemically damaged hair friction force at

78

B. Bhushan · C. LaTorre

Fig. 24.26. Durability study of friction force change as a function of AFM tip cycling for various hair types. The images above the plot signify before and after comparisons of a cuticle surface subjected to cycling at a 10 µN load [LaTorre and Bhushan, 2005b]

the onset of cycling. When conditioner is applied to virgin hair, however, the wear does not show up as an increase in the friction force. Thus, conditioner serves as a protective covering to virgin hair and helps protect the tribological properties when wear ensues. 24.4.3 Surface Roughness, Friction, and Adhesion for Hair Treated with Various Combinations of Conditioner Ingredients Figure 24.27 displays the representative surface roughness and friction force maps for chemically damaged hair and the seven different applied treatments (see Table 24.10 presented earlier). When conditioner is applied to the surface, a pattern of high friction is shown in the area surrounding the bottom edge of the cuticle. Likewise, the application of a BTMAC surfactant (without silicone deposition) or a BAPDMA surfactant (without silicone deposition) results in similar friction features. This is believed to be an area of conditioner accumulation, which causes increased friction due to meniscus effects. Friction maps for PDMS blend of silicone (at both low and

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

79

Fig. 24.27. Surface roughness and friction force maps of Caucasian chemically damaged hair with various treatments [LaTorre et al., 2006]

80

B. Bhushan · C. LaTorre

high deposition levels) added to the BTMAC surfactant do not show this increase as readily, suggesting that this type of silicone is not a contributor of high friction force on the nanoscale. This can be due to the fact that a PDMS type silicone is fairly mobile on the surface and thus does not cause the same meniscus effects as the AFM tip rasters through it. Also, the surface energy of the PDMS silicone is believed to be lower than that of the cationic surfactants. As a result, the meniscus force affecting the friction between the AFM tip and silicones in the conditioner is lower than that of conditioner without silicones. Thus, the overall friction force will be lower. For high deposition levels of amino silicone, however, there is a high variation of friction force and a more distributed layer. The amino group typically is less mobile and harder to move, which accounts for a different slip plane ﬂow than PDMS silicone. Figure 24.28a displays the adhesive force maps for chemically damaged hair and the seven different treatments. As shown in the legend, a lighter area corresponds to a higher tip pull-off force (adhesive force). Chemically damaged untreated hair has a relatively low adhesive force, and is more or less consistent over the hair surface. In nearly all cases, addition of a conditioner treatment caused an increase in meniscus forces, which in turn increased the adhesive pull-off force between the AFM tip and the sample. Observing the chemically damaged treated hair, the uneven distribution of the conditioner layer is seen. This uneven distribution is also most evident for the amino silicone images, in which the less mobile silicone brings about a distinguishable change in adhesion over the surface. For the PDMS blend silicone, it is seen that adhesion over the surface is much more consistent than the amino silicone, where various areas of high adhesion occur. It is important to note that while the adhesive force maps presented are representative images for each treatment, the adhesive force varies signiﬁcantly when treatments are applied to the hair surface. Figure 24.28b shows histograms of all adhesive force data for chemically damaged, chemically damaged treated, PDMS blend silicones (high deposition), and amino silicones (high deposition). Chemically damaged treated hair shows a much larger range of adhesive force values and a normal distribution, which suggests that the conditioner layer is normally distributed. The histogram for PDMS blend silicone treatment shows a normal distribution at the larger adhesive force values, but also shows another peak at low adhesion values. Amino silicone treated hair follows a normal distribution, but it is interesting to note the distinct groupings of the adhesion values and the spacing between them. This is further evidence that the amino silicone groups most likely attach immediately to the hair surface and are less mobile than PDMS silicone, causing distinct regions of high and low adhesion values over the cuticle surface. Figure 24.29 displays a summary of the data collected for all chemically damaged hair samples and their treatments. Table 24.14 reviews some of the observations and corresponding mechanisms that help to explain the trends found [LaTorre et al., 2006]. The application of the commercial conditioner to the chemically damaged hair caused a decreased coefﬁcient of friction and a large increase in adhesive force. The decreased coefﬁcient of friction may be explained by the fact that the chemically damaged hair accumulates much of the positively charged conditioner on the surface due to its highly negative charge, which in turn makes it easier to shear the liquid on the surface, causing a lower coefﬁcient of friction. However, the nanoscale pull-

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

Fig. 24.28a. Adhesive force maps for damaged hair with various treatments

81

82

B. Bhushan · C. LaTorre

Fig. 24.28b. Adhesive force histograms for various hair types [LaTorre et al., 2006]

off force (adhesive force) is much larger than on the untreated hair because of meniscus effects. In general, the adhesive force varied widely, but typically showed a signiﬁcant increase with the presence of conditioner. As discussed previously, this is a clear sign that meniscus effects are inﬂuencing the pull-off force between the tip and the sample. In most cases, the macroscale and microscale coefﬁcient of friction followed the same trend, in which a decrease was observed with the addition of the PDMS blend or amino silicones to the BTMAC surfactant. The silicones are typically used as a major source of lubrication and thus give the conditioner more mobility on the hair surface compared to just surfactants and fatty alcohols. The inverse trend was seen only for the amino silicone group. The amino silicones have a strong electrostatic attraction to the negatively charged hair surface, which in turn creates higher binding forces and less mobility. The dampened mobility of the amino silicone at high deposition levels, with respect to hair surface and tip, may account for this wide variation in coefﬁcient of friction and large adhesive force values. In terms of adhesive force, it was previously observed in Fig. 24.28a that the amino silicone treatments showed much more distinct regions of higher and lower adhesion compared to PDMS blend silicones. This nonuniform amino silicone thickness distribution on hair is also believed to be caused by the inhibited mobility, as the molecules immediately attach to the hair at contact and do not redistribute as

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

83

Fig. 24.29. Coefﬁcient of friction, adhesive force, and surface roughness plots for damaged hair with various treatments [LaTorre et al., 2006]

a uniform coating. The increased polarity of the amino silicones compared to the PDMS blend can also be a major contributor of the higher friction and adhesion at high deposition levels.

84

B. Bhushan · C. LaTorre

Fig. 24.29. (continued)

If looking at the hair with only the BTMAC surfactant and fatty alcohols, and then adding low deposition levels of PDMS blend (with high MW weight) or amino silicone, it is seen that the coefﬁcient of friction decreases, while the adhesion remains approximately the same. If the deposition level is increased, it is observed that the PDMS blend is still lower, but now the coefﬁcient of friction for the amino silicone is about the same as in the BTMAC only samples. The mobility of the conditioner layer is again a major issue, as this easier mobility accounts for the lower coefﬁcient of friction. However, as is seen when there is a large amount of the amino silicones on the hair surface, the mobility ceases and the coefﬁcient of friction becomes high again. The BAPDMA surfactant typically showed a higher adhesive force than the BTMAC surfactant. However, there may be an increased accumulation of the BAPDMA surfactant, such that the conditioner layer is more easily sheared by the tip, as is the case for chemically damaged treated hair. The inherent differences in surfactant composition and how they interact with the chemically damaged hair surface most likely account for the coefﬁcient of friction differences.

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

85

Table 24.14. Observations and corresponding mechanisms regarding the coefﬁcient of friction and the adhesion for various hair treatments Observation Damaged vs. damaged treated hair Damaged hair shows a decrease in the coefﬁcient of friction but an increase in the adhesion from the application of commercial conditioner

Mechanism The conditioner layer deposited on the surface of the damaged hair results in a lower shear strength, which in turn lowers the coefﬁcient of friction, while meniscus effects increase the pull-off (adhesive) force between the tip and hair sample

PDMS blend vs. amino silicone (at high deposition) Amino silicones interact strongly with nega- A stronger electrostatic attraction exists, tively charged hair surface, especially at high which results in stronger binding forces deposition levels (which leads to higher adhesion) for high deposition amino silicone Amino silicone thickness distribution on hair Less mobility with amino silicones, so molis less uniform than with the PDMS blend ecules attach to hair at contact and do not redistribute easily Surfactant vs. surfactant plus addition of silicone Adhesion remained approximately the same, Mobility becomes easier with addition of while the coefﬁcient of friction decreased with PDMS blend silicone, which leads to a lower addition of PDMS blend silicone and amino coefﬁcient of friction. At high deposition of silicone (low deposition). The coefﬁcient of amino silicones, mobility ceases and the cofriction remained about the same at high de- efﬁcient of friction becomes high again position amino silicone BTMAC surfactant vs. BAPDMA surfactant The BTMAC surfactant has lower adhesion Interaction of surfactants with damaged hair but a higher coefﬁcient of friction than the surface causes inherent differences in the coBAPDMA surfactant efﬁcient of friction and adhesion. BAPDMA has both amino and amine groups, which increases polarity

With respect to roughness, the vertical standard deviation decreased slightly with most treatments, although standard deviations were similar. The spatial parameter increased slightly with treatments, but the variation also became extremely high. 24.4.4 Investigation of Directionality Dependence and Scale Effects on Friction and Adhesion of Hair 24.4.4.1 Directionality Dependence Figure 24.10, shown previously, displays friction test apparatuses used for macroscale and micro/nanoscale coefﬁcient of friction measurements, as discussed earlier. The

86

B. Bhushan · C. LaTorre

macroscale friction test apparatu, consists of taking a hair ﬁber swatch and running a block of synthetic skin over the ﬁbers. AFM coefﬁcient of friction measurements take place on a single ﬁber. With the microscale AFM tip (4 µm radius), several cuticle scales may come into contact at the same time, depending on their length, causing the scale edges to become a signiﬁcant factor in the resulting coefﬁcient of friction values. This is especially evident when scanning in the tip-to-root direction of the hair (referred to as “against cuticle”). With the nanoscale AFM tip (30–50 nm radius), single asperity contact can be simulated, and the cuticle edges can be avoided by limiting the scan area to a small section of the surface. Figure 24.30 displays SEM micrographs and AFM height maps of virgin hair at various scales. The top SEM micrograph shows the hair ﬁber so that the full diameter is observed. During macroscale friction testing, many hair ﬁbers will come into contact with the upper sample (generally synthetic skin) at the same time. The

Fig. 24.30. Comparison of hair topography at different scales

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

87

next micrograph reveals a better magniﬁcation of the cuticle scales. The scan length here is representative of that used for microscale coefﬁcient of friction measurements. The 20 × 20 µm2 AFM height map and the corresponding cross-sectional plot show that cuticle scale heights vary from layer to layer, and range from approximately 300 to 500 nm. With a 5 × 5 µm2 map we can focus on one cuticle scale edge. Most visible cuticle scale lengths are between 5 and 10 µm. For nanoscale coefﬁcient of friction measurements, a 2 µm scan length is generally used, so as to avoid these scale edges affecting the data. However, if directionality dependence is to be studied, a 5 µm scan line is sufﬁcient. Human hair has been shown in the past to have a directionality friction effect on the macroscale, making it easier to travel over the hair surface from root-to-tip than in the opposite direction due to the tile-like orientation of cuticles [Robbins, 1994]. The outer surface of human hair is composed of numerous cuticle scales running along the ﬁber axis, generally stacked on top of each other. As previously discussed, the heights of these step changes are approximately 300 nm. These large changes in topography make the cuticle an ideal surface for investigating the directionality effects of friction. The ﬁrst row of Fig. 24.31a displays a low resolution SEM micrograph and a friction proﬁle for macroscale coefﬁcient of friction measurements. Note that these data are taken for measurements where multiple ﬁbers are in contact with the synthetic skin upper specimen at the same time, and do not correspond directly with the SEM micrograph. With an applied normal load of 50 mN and 3 mm travel, it is observed that the friction force produced when scanning from root-to-tip (referred to as “along cuticle”) is lower than that when scanning against cuticle. This is a direct consequence of the literally thousands of scale edges that come into contact with the synthetic skin. When traveling against cuticle, these edges act as tiny resistors to motion as they are forced backwards and uplifted from their interface with the underlying cuticle layers. The resistance to motion of so many cuticle edges at the same time becomes “additive” and results in higher values of friction, corresponding to a higher coefﬁcient of friction than when traveling along cuticle. For along cuticle travel, these edges are forced downward against the underlying cuticle layers, so that the resistance effect of these edges is limited, which results in lower friction values. The second row of Fig. 24.31a shows an AFM height map corresponding to the microscale friction proﬁle shown to its right. Due to the size of the hair, it was only possible to capture a rectangular height map as shown. It is evident that the 100 µm travel results in the involvement of several cuticle scales. The applied normal load for the microscale friction proﬁle (about 20 nN) is signiﬁcantly reduced from that of the macroscale value, which consequently yields much lower friction forces. In the along cuticle direction, we can see small ﬂuctuations in the friction data over the scan distance. These are caused by local variations in surface roughness, system noise, and changes due to traveling over the scale edges. However, when scanning against the cuticle, distinctly large spikes in the data are observed at roughly 5–10 µm intervals. This is clearly the effect of the scale edges coming into contact with the AFM microscale tip and causing local collisions and ratcheting of the tip. Because of the sign convention of the AFM that causes a reversal in the sign when traversing the opposite direction, this signal is now observed to be highly negative. These edge effects are hence the primary item responsible for the higher friction and the coefﬁcient of friction observed in the against cuticle direction.

88

B. Bhushan · C. LaTorre

Fig. 24.31. (a) Directionality effects of hair friction on various scales. (b) Microscale coefﬁcient of friction data for various hair types showing directionality effects

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

89

Directionality effects on nanoscale hair friction have previously been reported by LaTorre and Bhushan [2005a]. As shown in the bottom row of Fig. 24.31a, the scan size of 5 × 2.5 µm2 provides one cuticle scale edge to be studied (the height map is rectangular only to be consistent with the microscale image in the middle row). As the tip rasters in the along cuticle direction, a small decrease in the friction force is observed as the tip follows the scale edge on a downward slope. When the tip comes back in the against cuticle direction, colliding with the scale edge and climbing up the sharp peak results in a high friction signal. As discussed above, because of the sign convention of the AFM/FFM that causes a reversal in the sign when traversing the opposite direction, this signal is now observed to be highly negative. The interesting difference between the two proﬁles lies in the fact that the magnitude of the decrease in friction when going up the step is much larger than the magnitude of the friction when the tip is going down the step, yet both signals are in the same direction. It is thus shown that the cuticle edge provides a local ratchet and collision mechanism that increases the friction signal at that point and clearly shows the directionality dependence caused by edge effects. Coefﬁcient of friction data showing directionality dependence on the macroscale for various virgin, damaged, and treated hair types can be found in Bhushan et al. [2005]. Figure 24.31b and Table 24.15 present a summary of the microscale coefﬁcient of friction data for the various hair samples. In most cases, the coefﬁcient of friction has more than doubled when scanning in the against cuticle direction. A more in depth discussion on the coefﬁcient of friction trends between the different hair types will follow. For now, however, it is important to realize that there is a strong directionality dependence on coefﬁcient of friction data for hair, especially on the microscale. On the nanoscale, while directionality dependence of friction force has been studied, actual coefﬁcient of friction data is generally only measured on a small scan area that does not include the cuticle edge, so that only the true cuticle surface is involved [LaTorre and Bhushan, 2005a]. Hence, nanoscale coefﬁcient of friction directionality data similar to Fig. 24.31b are not shown.

Table 24.15. Coefﬁcient of friction values showing directionality dependence on the microscale. This dependence has also been observed on the nanoscale and macroscale Hair sample

Microscale coefﬁcient of friction Along cuticle Against cuticle

Virgin (untreated) Virgin treated (one cycle commercial conditioner) Chemically damaged (untreated) Chemically damaged treated (one cycle commercial conditioner) Chemically damaged treated (3 cycle commercial conditioner)

0.08 0.07 0.16

0.19 0.17 0.31

0.08

0.29

0.06

0.20

90

B. Bhushan · C. LaTorre

24.4.4.2 Scale Effects Given the fact that the “directionality effect” is now observed to be universal for all types of hair and on all scales, and since the along cuticle direction is more relevant to our daily life (i.e. combing), we shall now focus on coefﬁcient of friction data in the along cuticle direction. As described earlier, microscale and nanoscale coefﬁcients of friction are taken as the slope of the least squares ﬁt line of a friction force vs. the normal force data curve. Figure 24.32a shows these types of raw data curves for various representative hair samples using both microscale (top plot) and nanoscale (bottom plot) AFM tips. The nanoscale data were taken from the raw data used in LaTorre and Bhushan [2005b].

Fig. 24.32a. Friction force vs. normal force curves for microscale and nanoscale coefﬁcients of friction

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

91

One can see a relatively linear relationship between the data points for each type of hair sample. If the least squares ﬁt lines of Fig. 24.32a are extended to intercept the horizontal axis (indicated by the dotted lines), then a value for the adhesive force can also be calculated, since friction force F is governed by the relationship of (3.1). These adhesive force values serve as average adhesive force values over the course of the full scan proﬁle, and differ slightly from the force calibration plots of Fig. 24.32b. From Fig. 24.32a, it is observed in both plots that treated hair ﬁbers, whether virgin or chemically damaged, have much higher average adhesive force values as compared to their untreated counterparts. Chemically damaged hair is observed to have the highest coefﬁcient of friction (highest slope) on both the microscale and nanoscale. As explained in LaTorre and Bhushan [2005a,b] and LaTorre et al. [2006], chemical damage to the hair causes the outer lubricious layer of the cuticle to wear off, resulting in an increased coefﬁcient of friction. Force calibration plots yield adhesive force values at a single point and are considered to be more relevant for measurement of the pull-off force between the tip and hair surface. Since it has been previously shown that treating both virgin and chemically damaged hair with conditioner results in large increases in the adhesive force using both microscale and nanoscale AFM tips, we will focus only on the force calibration plots of the virgin and virgin treated hair to discuss mechanisms for this trend. The ﬁrst plot in Fig. 24.32b shows a typical virgin hair force calibration plot with the microscale AFM tip. The cantilever deﬂection is plotted on the vertical axis against the Z-position of the piezo scanner. As the piezo extends (from right to left in the graph), it approaches the tip and does not show any deﬂection while in free air.

Fig. 24.32b. Adhesive force comparison of virgin and virgin treated hair using microscale and nanoscale AFM tips and force calibration plot technique

B. Bhushan · C. LaTorre

92

The tip then approaches within a few nanometers (point A) and becomes attached to the sample via attractive forces at a very close range (point B). After initial contact, any extension of the piezo results in a further deﬂection of the cantilever and tip. Once the piezo reaches the end of its designated ramp size, it retracts to its starting position (from left to right). The tip goes beyond zero deﬂection and enters the adhesive regime. At point C, the elastic force of the cantilever becomes equivalent to the adhesive force, causing the cantilever to snap back to point D. The horizontal distance between A and D multiplied by the stiffness of the cantilever results in a quantitative measure of the adhesive force [Bhushan, 1999a,b, 2002, 2005]. We can see that on the application of one cycle conditioner treatment to the virgin hair, the microscale adhesive force jumps to about 230 nN. This is clearly the effect of the meniscus forces brought about by the presence of the conditioner layer on the cuticle surface that interacts with the tip [LaTorre and Bhushan, 2005a,b; Chen and Bhushan, 2005b]. With the nanoscale tip (bottom row of Fig. 24.32b), an increase in the adhesive force is again seen with conditioner treatment. It is important to notice that the adhesive force values on the microscale are always larger than those on the nanoscale for a given hair. To explain the scale dependency of the adhesive force, we can model the hair-conditioner-tip interaction as a sphere close to a surface with a continuous liquid ﬁlm [Chen and Bhushan, 2005b]. The adhesive force Fa , is the force needed to pull the sample away from the tip (which is the same as the adhesive force calculated with force calibration plots). Fa is the sum of van der Waals force Fvdw mediated by the adsorbed water or conditioner layer and the meniscus force Fm due to the Laplace pressure (Fa = Fvdw + Fm ). The meniscus force Fm is given by Fm = 2πRγ(1 + cos θ) ,

(24.2)

where R is the tip radius, γ is the surface tension of the conditioner, and θ is the contact angle between the tip and conditioner [Bhushan, 2002]. The increase in the adhesive force calculated by force calibration plots with the microscale AFM tip compared to the nanoscale AFM tip is in large part due to the increased radius R of the microscale ball, which consequently induces larger Fm . Figure 24.33 and Table 24.16 display the coefﬁcient of friction and the adhesive force data on the macroscale, microscale, and nanoscale. Scale dependence is clearly observed. Macroscale data for virgin and virgin treated hair were taken from Bhushan et al. [2005]. The values for other hair were taken in part from the indexed coefﬁcient Table 24.16. Coefﬁcient of friction and the adhesive force of hair on various scales Hair sample Virgin (untreated) Virgin treated (1 cycle commercial conditioner) Chemically damaged (untreated) Chemically damaged treated (1 cycle commercial conditioner) Chemically damaged treated (3 cycles commercial conditioner)

Coefﬁcient of friction Macroscale Microscale Nanoscale

Adhesive force (nN) Macroscale Microscale Nanoscale

0.14

0.08 ± 0.01 0.04 ± 0.003

–

53

25

0.12 0.24

0.07 ± 0.03 0.04 ± 0.01 0.16 ± 0.003 0.12 ± 0.05

– –

221 66

32 40

0.14

0.08 ± 0.02 0.07 ± 0.01

–

191

66

0.13

0.06 ± 0.02 0.04 ± 0.02

–

97

54

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

93

Fig. 24.33. Summary of coefﬁcient of friction and adhesive force data on various scales

of friction values in LaTorre and Bhushan [2005b], which were transformed into actual values (as described previously). No adhesive force data are presented for the macroscale data, because the adhesive force contribution to friction is considered to be negligible compared to the applied normal load. On the microscale and nanoscale, however, the magnitude of the adhesive force is the same as that of the applied normal load, so they have signiﬁcant contributions on the coefﬁcient of friction data, and are thus presented. Microscale values are taken from the raw data represented in Fig. 24.32a and b. Nanoscale data were taken from LaTorre and Bhushan [2005b]. Macroscale coefﬁcient of friction (COF) data is shown in the top row of Fig. 24.33: Chemically damaged hair has higher coefﬁcient of friction than virgin

94

B. Bhushan · C. LaTorre

hair, 0.24 compared to 0.14. The coefﬁcient of friction decreases with the application of 1 cycle of conditioner for both virgin and chemically damaged hair. When three cycles of conditioner are applied to chemically damaged hair, there is only a slight decrease compared to one cycle application, 0.14 to 0.13. Thus, the data readily reveal that conditioner treatment decreases coefﬁcient for friction of hair. The main mechanism for this macroscale trend is that lubrication with a thin conditioner layer occurs over a large contact area, and thus the conditioner layer shears easily to create a lubricious effect [LaTorre and Bhushan, 2005b]. It is important to note that the magnitude of the coefﬁcient of friction values is higher on the macroscale than on the other scales for all hair types. Bhushan et al. [2004] have previously outlined several differences in operating conditions which can responsible for higher macroscale friction values. The one most relevant to our situation is that coefﬁcient of friction increases with an increase in the AFM tip radius. Nanoscale friction data is taken with a sharp AFM tip, while the macroscale and microscale tests have contacts that range from nanoasperities to much larger asperities that may be responsible for larger values of friction force on these scales. The combination of higher normal loads with a larger contact area (due to contact with multiple ﬁbers at the same time) may also be responsible for the increased coefﬁcient of friction on the macroscale. The coefﬁcient of friction trends are similar on the microscale. On the microscale, the virgin hair coefﬁcient of friction was 0.08, while application of one cycle conditioner decreased the coefﬁcient of friction only slightly to 0.07. From Fig. 24.33 it is important to note the large standard deviation on the virgin treated value. The corresponding adhesive force data in the same row are useful for a better understanding of this behavior. It is shown that the virgin treated hair has a large adhesive force contribution (due to meniscus effects caused by the conditioner layer), which has signiﬁcant effects on the variation of friction force, and consequently the coefﬁcient of friction. The chemically damaged hair has the largest coefﬁcient of friction of the set, 0.16. Application of 1 conditioner cycle brought the value down to 0.08, while it was even lower for 3 conditioner cycles, 0.06. The adhesive force increased signiﬁcantly with conditioner application; for virgin hair, the adhesive force increased from about 50 nN to 220 nN due to the meniscus effects that come about from AFM tip interaction with the conditioner on the cuticle surface. Likewise, the adhesive force for chemically damaged hair was about 65 nN and jumped to 190 nN and 100 nN for 1 and three cycles of conditioner treatment, respectively. A possible reason that one cycle of conditioner on damaged hair showed higher average adhesive force than three cycles could be because the three cycles generally place the conditioner more uniformly on the surface rather than accumulating it mostly on the bottom surface near the cuticle edge, which is where the adhesive force maps were generally taken. Nevertheless, the increased adhesive force shown in the plots is a clear indication of conditioner being present on the hair surface [LaTorre and Bhushan, 2005b]. On the nanoscale, the coefﬁcients of friction of virgin and virgin treated hair are similar, but slightly higher for the treated cases. This is opposite to the trend on the macroscale and slightly different to that observed on the microscale. These meniscus bridges require more force to break through than with untreated hair, which causes the coefﬁcient of friction to be similar to the untreated value, instead of experiencing the traditionally expected signiﬁcant decrease. Damaged hair shows a much higher

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

95

coefﬁcient of friction with more variation in the values, since the chemical damage varies throughout each individual ﬁber. Contrary to the virgin and virgin treated hair nanoscale results, the coefﬁcient of friction for damaged hair decreased with application of conditioner treatment (both 1 and 3 cycles), which agrees well with macroscale and microscale trends. There are several reasons that the nanoscale coefﬁcient of friction trends for virgin treated and chemically damaged treated hair are different. In order to serve as a reference for this discussion, Figs. 24.20 and 24.21 show surface roughness, friction force, and adhesive force maps for the various hair types (taken with a nanoscale AFM tip) [LaTorre and Bhushan, 2005b]. The effects of chemical damage play a large role. It is widely known that the cuticle surface of any hair is negatively charged. This charge becomes even more negative with the application of chemical damage to the hair. As a result, the positively charged particles of conditioner have an even stronger attraction to the chemically damaged surface, which explains the increased presence of conditioner when compared to virgin treated hair. With the application of three conditioner cycles on damaged treated hair, there is an even more uniform placement of the conditioner [LaTorre and Bhushan, 2005b; LaTorre et al., 2006]. This leads to a high adhesive force due to meniscus effects (similar to that of virgin treated hair) but more importantly, a lower shear strength on the surface. This creates an overall effect of lubrication as the tip travels across the cuticle surface and ultimately lowers the coefﬁcient of friction. Another reason for the difference in nanoscale trends between virgin and chemically damaged hair may have to do with the drastic differences in hydrophobicity of the two hair types. Virgin hair has been shown to be hydrophobic, with a contact angle of around 100◦ (see Table 24.11). Chemically damaged hair, however, is hydrophilic, with a contact angle of around 60◦ . The conditioner gel network is primarily composed of water, together with fatty alcohols, cationic surfactants, and silicones. Thus, the hydrophobicity of the hair will be relevant to not only how much conditioner is deposited, but also to how it diffuses into the hair and bonds to the hair surface. For virgin treated hair, the conditioner deposits in certain locations, especially near the cuticle edge, but due to the hydrophobicity of the cuticle does not spread out as readily as with chemically damaged hair. For chemically damaged hair, the conditioner spreads out a bit more uniformly and in more places over the cuticle surface due to both the hydrophilicity and the stronger negative charge that attracts more conditioner deposition. Thus, as the tip scans over the virgin treated surface, the conditioner does not smear as readily, causing the tip to have to break the tiny meniscus bridges formed with the conditioner. This results in an increased adhesive force, which contributes to a higher friction force. This ends ups increasing the coefﬁcient of friction to about the same level as the untreated virgin hair, instead of the reduction that is typically expected for a lubricated surface. In the case of chemically damaged hair, however, the conditioner layer is already more spread out, especially in the case of three cycles of treatment, as shown in the friction force maps of Fig. 24.20. As the tip scans over the surface, the overall effect is one of reduced shear strength, i.e. the conditioner layer, albeit not fully continuous, smears with tip travel and causes a reduced coefﬁcient of friction between the tip and the hair. With three cycles the conditioner thickness increases slightly [Chen and Bhushan, 2005b], and the layer is even more uniformly distributed over

96

B. Bhushan · C. LaTorre

the surface [LaTorre and Bhushan, 2005b], which causes a further reduction in the coefﬁcient of friction, much like the results seen on both the microscale and macroscale. Figure 24.34 schematically shows the mechanisms responsible for the reverse trends seen on the nanoscale for virgin treated hair. Table 24.17 summarizes the observations seen for both directionality dependence and scale effects. In the top diagram of Fig. 24.34, the thin layer of conditioner acts as a lubricant over the hair ﬁber, limiting the dry contact with the synthetic skin block and creating easier relative motion, which decreases coefﬁcient of friction compared to the untreated hair. This is true on the macroscale for both virgin and chemically damaged hair. On the microscale, the same trend is experienced; that is, the 4 µm radius of the AFM ball comes in contact with multiple cuticle scales at the same time, causing an overall lubrication effect for both virgin treated and chemically damaged treated hair as the thin conditioner layer is sheared to create easier relative motion. It is

Fig. 24.34. Schematic of various mechanisms responsible for scale dependence of hair friction and adhesion

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

97

Table 24.17. Observations and corresponding mechanisms regarding the coefﬁcient of friction and adhesion for various hair treatments Observation

Mechanism

Directionality dependence of the coefﬁcient of friction The coefﬁcient of friction is always higher in the against cuticle direction as compared to the along cuticle direction

Macroscale: Thousands of scale edges resist motion (creating higher friction forces) as they are forced backwards and uplifted. Microscale: Scale edges come in contact with the AFM microscale tip and cause local collisions and ratcheting of the tip, which creates higher friction Nanoscale: AFM nanoscale tip collides with the scale edge and has to climb up the edge, resulting in a high friction signal Scale effects

The coefﬁcient of friction values are largest on the macroscale, followed by microscale and then nanoscale values

The coefﬁcient of friction increases with an increase in the AFM tip radius. Nanoscale friction data are taken with a sharp AFM tip, while the macroscale and microscale tests have contacts ranging from nanoasperities to much larger asperities. The macroscale has contact with multiple ﬁbers at the same time

Adhesive force values are larger on the microscale than on the nanoscale

The increase in the adhesive force on the microscale is due to the increased radius of the microscale ball, which consequently induces larger meniscus force contribution to the adhesive force

On all scales, the coefﬁcient of friction decreases for chemically damaged treated hair compared to untreated. The same trend occurs for virgin treated hair on the macroscale and microscale, but the nanoscale value is same as for untreated hair

Macroscale: A thin layer of conditioner acts as a lubricant over the hair ﬁber (decreases the coefﬁcient of friction) (whether virgin hair or chemically damaged hair) Microscale: The same trend is experienced at macroscale because the AFM microscale tip comes in contact with multiple cuticle scales at the same time, causing an overall lubrication effect Nanoscale: Hydrophobicity of virgin hair causes different deposition of conditioner. The AFM tip has to break the tiny meniscus bridges when scanning the increasing adhesive force coefﬁcient of friction. For chemically damaged hair, a higher negative charge and a hydrophilic surface results in more uniform deposition and better smearing of the conditioner layer and a lower coefﬁcient of friction

On both microscale and nanoscale, the adhesive force increases for treated hair

Meniscus forces brought about by the presence of the conditioner layer increase the adhesive force

important to note, however, that the adhesive forces due to meniscus effects are of the same magnitude as the applied microscale normal load. On the nanoscale (bottom diagram of Fig. 24.34), we see different trends for virgin treated and chemically damaged treated hair. As discussed earlier, the hy-

B. Bhushan · C. LaTorre

98

drophobicity of the virgin hair causes a different deposition of conditioner. The AFM tip has to break the tiny meniscus bridges formed with the conditioner as it scans across the hair surface, which increases the adhesive force contribution and results in an increased coefﬁcient of friction. For hydrophilic chemically damaged hair, there is more uniform deposition and better smearing of the conditioner layer, which serves to lower the coefﬁcient of friction between the tip and the cuticle surface. 24.4.5 Surface Roughness and Friction of Skin Synthetic materials were also studied for surface roughness and friction force information, see Fig. 24.35a. While macroscale dimples could be seen on the surface of collagen ﬁlm, it was interesting to ﬁnd similar pits and dimples on the micro/nanoscale, consequently with a large variation in dimple size and depth. Polyurethane ﬁlms are shown to have quite different topography and friction forces, while their coefﬁcient of friction is very similar. Human skin shows a rougher texture with higher peaks, Fig. 24.35b. The roughness parameters for the collagen and polyurethane ﬁlms, and also for human skin, are presented in Table 24.18 [LaTorre and Bhushan, 2005a]. The vertical height standard deviation σ was approximately 3 times larger than that of virgin hair for both synthetic materials. However, the correlation length β ∗ was lower than what was typically observed in hair. The average coefﬁcient of friction for these synthetic materials are shown in Fig. 24.36, plotted next to virgin Caucasian hair as a reference. These values were calculated using the slope of the friction force curves, described previously. Both collagen and polyurethane ﬁlms displayed similar coefﬁcient of friction values of 0.22 and 0.24, respectively. Virgin hair displays a much lower coefﬁcient of friction than both materials, approximately eight times lower. Table 24.18. Surface roughness parameters σ, β ∗ for collagen, polyurethane ﬁlms, and human skin

Collagen (synthetic hair) Polyurethane (synthetic skin) Human skin

σ (nm)

β ∗ (µm)

36 ± 11 33 ± 6 80 ± 28

0.50 ± 0.1 0.71 ± 0.1 0.59 ± 0.2

24.5 Closure The AFM contact mode has been used to perform nanotribological studies on various hair types and skin. The friction force and the resulting coefﬁcient of friction are seen to be higher on chemo-mechanically damaged hair than on virgin hair, due to

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

99

Fig. 24.35. (a) Surface roughness and friction force images for collagen and polyurethane ﬁlms at 5 and 10 µm2 scan sizes. Shown above each image is a cross-section taken at the corresponding arrows to demonstrate roughness and friction force information. (b) Surface roughness for human skin at 5 and 10 µm2 scan sizes [LaTorre and Bhushan, 2005a]

100

B. Bhushan · C. LaTorre

Fig. 24.36. Average coefﬁcient of friction values for collagen and polyurethane ﬁlms. Error bars represent ±1σ of the average coefﬁcient value [LaTorre and Bhushan, 2005a]

the increase in surface roughness and a change in surface properties that results from exposure to chemical damage. Generally speaking, the average coefﬁcient of friction is similar between virgin and virgin treated hair of each ethnicity. However, in virgin treated hair, there is an increase in friction forces around the cuticle edges and surrounding area. It is currently believed that the increase in friction force is due in part to an increase in meniscus effects, which increase the adhesive force contribution to friction at sites where conditioner is deposited or accumulated on the surface, namely around the cuticle scale edges. This adhesive force is of the same magnitude as the normal load, which makes the adhesive force contribution to friction rather signiﬁcant. Studies using the force calibration plot technique showed a decrease in the adhesive force with damaged hair, and signiﬁcantly higher adhesive force for treated hair. This increase on the micro/nanoscale is most likely due to meniscus force contributions from the accumulation and localization of a conditioner layer on the hair surface. Thus, the presence of conditioner can be detected by this increasing adhesive force. Chemically damaged treated hair shows a much stronger afﬁnity to conditioner than virgin hair. The negative charge of hair ﬁbers becomes even more negative with the application of chemical damage to the hair. As a result, the positively charged particles of conditioner have even stronger attraction to the chemically damaged surface, and this results in an increased presence of conditioner (and corresponding higher friction forces) when compared to virgin treated hair. With the application of three conditioner cycles on chemically damaged treated hair, the friction force increases all over the cuticle surface, showing a more uniform placement of the conditioner. Contrary to the virgin and virgin treated hair results, the coefﬁcient of friction for chemically damaged hair decreased with the application of commercial conditioner treatment (both one and three cycles). One possible explanation is that because the stronger negative charge on damaged hair results in better attraction of conditioner, this leads to a higher adhesive force but, more importantly, a lower shear strength on the surface. Environmental effects were studied for various hair types. The coefﬁcient of friction generally decreased with increasing temperature. After soaking hair in de-

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

101

ionized water, virgin hair exhibits a decrease in the coefﬁcient of friction after soaking. Virgin hair is more hydrophobic (based on the contact angle data), so more water is present on the surface and this results in a lubrication effect after soaking. Chemically damaged hair tends to be hydrophilic due to the chemical degradation of the cuticle surface, and this results in absorption of water after soaking. This softens the hair, which leads to higher friction, even with conditioner treatment. Durability tests show that once conditioner is applied to virgin hair, wear does not show up as an increase in the friction force. Thus, conditioner serves as a protective covering to virgin hair and helps to protect the tribological properties when wear ensues. In most cases, a decrease in the coefﬁcient of friction was observed on chemically damaged hair with the addition of the PDMS blend or amino silicones to the BTMAC surfactant. The silicones are typically used as a major source of lubrication and thus give the conditioner more mobility on the hair surface compared to just surfactants and fatty alcohols. The inverse trend was seen only for the amino silicone group at high deposition. The dampened mobility of the amino silicone at high deposition levels, with respect to hair surface and tip, may account for this wide variation in the coefﬁcient of friction. At high deposition levels, the amino silicones showed much more distinct regions of high and low friction and adhesion, which shows that there is less mobility of these molecules and much less redistribution as they coat the hair. Investigations of scale effects and directionality dependence on friction and adhesion have been studied. On the macroscale, when traveling against cuticle, edges resist motion (creating higher friction forces) as they are forced backwards and uplifted from their interface with the underlying cuticle layers. On the microscale, when scanning against the cuticle, distinctly large spikes in the data are observed at roughly 5–10 µm intervals. This is clearly the effect of the scale edges coming in contact with the AFM microscale tip and causing local collisions and ratcheting of the tip, which creates higher friction. On the nanoscale, the AFM nanoscale tip collides with the scale edge and has to climb up the edge, resulting in a high friction signal. The coefﬁcient of friction values are largest on the macroscale, followed by microscale and then nanoscale values. In general, coefﬁcient of friction increases with an increase in the AFM tip radius. Nanoscale friction data are taken with a sharp AFM tip, while the macroscale and microscale tests have contacts that range from nanoasperities to much larger asperities that may be responsible for larger values of friction force on these scales. Adhesive force values are larger on the microscale than on the nanoscale. The increase in the adhesive force calculated by force calibration plots with the microscale AFM tip compared to the nanoscale AFM tip is in large part due to the increased radius of the microscale ball, which consequently induces a larger meniscus force contribution to the adhesive force. On all scales, the coefﬁcient of friction decreases for chemically damaged treated hair compared to untreated hair. The same trend occurs for virgin treated hair on the macroscale and microscale, but not on the nanoscale. On the macroscale, the thin layer of conditioner acts as a lubricant over the hair ﬁber, which decreases the coefﬁcient of friction compared to the untreated hair (whether virgin or chemically damaged). On the microscale, the same trend is experienced because the AFM microscale tip comes into contact with multiple cuticle scales at the same time, causing an overall lubrication effect. On the nanoscale, the hydrophobicity of virgin hair causes different deposition of conditioner. The AFM tip has to break the tiny

B. Bhushan · C. LaTorre

102

meniscus bridges formed with the conditioner as it scans across the hair surface, which increases the adhesive force contribution and results in an increased coefﬁcient of friction. For chemically damaged hair, there is a higher negative charge and a hydrophilic surface, which results in more uniform deposition and better smearing of the conditioner layer, which serves to lower the coefﬁcient of friction between the tip and the cuticle surface.

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. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.

Barnes HA, Roberts GP (2000) Int J Cosmet Sci 22:259 Bhushan B (1999a) Principles and Applications of tribology. Wiley, New York Bhushan B (1999b) Handbook of Micro/Nanotribology. 2nd Ed. CRC Press, Boca Raton Bhushan B (2002) Introduction to Tribology. Wiley, New York Bhushan B (ed) (2004) Springer Handbook of Nanotechnology. Springer, Berlin Heidelberg New York Bhushan B (2005) Nanotribology and Nanomechanics – an Introduction. Springer, Berlin Heidelberg New York Bhushan B, Burton Z (2005) Nanotechnology 16:467 Bhushan B, Dandavate C (2000) J Appl Phys 87:1201 Bhushan B, Liu H, Hsu SM (2004) ASME J Tribol 126:583 Bhushan B, Wei G, Haddad P (2005) Wear 259:1012 Bolduc C, Shapiro J (2001) Clinics Dermatology 19:431 Chen NH, Bhushan B (2005) J Microscopy 220:96 Chen NH, Bhushan B (2006) J Microscopy (in presss) Feughelman A (1997) Mechanical Properties and Structure of Alpha-Keratin Fibres: Wool, Human Hair and Related Fibres. University of New South Wales Press, Sydney Ginn ME, Noyes CM, Jungermann E (1968) J Interface Sci 26:146 Gray J (2001) Clinics in Dermatology 19:227 Gray J (2003) The World of Hair. http://www.pg.com/science/haircare/hair_twh_toc.htm Jachowicz J, McMullen R (2002) J Cosmet Sci 53:345 Jalbert C, Koberstein JT, Yilgor I, Gallagher P, Krukonis V (1993) Macromolecules 26:3069 Jollès P, Zahn H, Höcker H (eds) Formation and Structure of Human Hair. Birkhäuser, Berlin LaTorre C, Bhushan B (2005) Ultramicroscopy 105:155 LaTorre C, Bhushan B (2005) J Vac Sci Technol A 23:1034 LaTorre C, Bhushan B (2006) Ultramicroscopy (in press) LaTorre C, Bhushan B, Torgerson PM, Yang J (2006) J Cosmetic Sci (in press) Lerebour G, Cupferman S, Cohen C, Bellon-Fontaine MN (2000) Skin Res Technol 6:245 Liu H, Bhushan B (2003) Ultramicroscopy 97:321 Molina R, Comelles F, Julia MR, Erra P (2001) J Colloid Interface Sci 237:40 Pugliese PT (1996) Physiology of the Skin. Allured Publishing, Carol Stream Randebrook RJ (1964) Soc Cosmet Chem 15:691 Robbins C (1994) Chemical and Physical Behavior of Human Hair. Springer, Berlin Heidelberg New York Schott H (1971) J Pharm Sci 60:1893 Smith JR, Swift JA (2002) J Microscopy 206:182 Swift JA (1991) Int J Cosmet Sci 13:143 Swift JA (1997) In: Jolles P, Zahn H, Hocker H (eds) Formation and Structure of Human Hair. Birkhauser, Berlin, p 149 Swift JA (1999) Int J Cosmet Sci 21:227

24 Nanotribological Characterization Of Human Hair And Skin Using AFM

103

36. Swift JA (2000) Int J Cosmet Sci 51:37 37. Syed AN, Kuhajda A, Ayoub H, Ahmad K, Frank EM (1996) African–American Hair: Its physical Properties and Differences relative to Caucasian Hair. In: Hair Care (cosmetics & toiletries applied research series). Allured Publishing, Carol Stream 38. Wei G, Bhushan B (2006) Ultramicroscopy (in press) 39. Wei G, Bhushan B, Torgerson PM (2005) Ultramicroscopy 105:248 40. Wertz PW, Downing DT (1989) In: Swarbrick J, Guy RH (eds) Transdermal Drug Delivery. Marcel Dekker, New York 41. Wertz PW, Madison KC, Downing DT (1989) J Invest Dermatol 92:109 42. Yanazawa H (1984) Colloids Surf 9:133 43. Zviak C (ed.) (1986) The Science of Hair Care. Marcel Dekker, New York

Appendix This appendix outlines the steps involved in washing hair swatches with shampoo and/or conditioner. – Shampoo treatments Shampoo treatments consisted of applying a commercial shampoo evenly down a hair swatch with a syringe. Hair was lathered for 30 seconds, rinsed with tap water for 30 seconds, then repeated. The amount of shampoo used for each hair swatch was 0.1 cm3 shampoo per gram of hair. Swatches were hung to dry in an environmentally controlled laboratory, and then wrapped in aluminum foil. – Conditioner treatments 0.1 cm3 of commercial conditioner was applied per gram of hair. The conditioner was applied in a downward direction (scalp to tip) thoroughly throughout the hair swatch for 30 seconds, and then allowed to sit on the hair for another 30 seconds. The swatch was then rinsed thoroughly for 30 seconds. Swatches were hung to dry in an environmentally controlled laboratory, and then wrapped in aluminum foil.

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography Jayne C. Garno · James D. Batteas

List of Abbreviations and Symbols AFM SAM SPM SPL STM NPRW DPN IgG FFM LFM 16-MHA 11-MUD ODT OTS RH

atomic force microscopy self-assembled monolayer scanning probe microscopy scanning probe lithography scanning tunneling microscopy nanopen reader and writer dip-pen nanolithography immunoglobulin G frictional force microscopy lateral force microscopy 16-mercaptohexadecanoic acid 11-mercaptoundecanol octadecanethiol octadecyltrichlorosilane relative humidity

25.1 SPM-Based Methods of Lithography The capabilities of scanning probe lithography (SPL) have introduced a new era of surface chemistry, providing tools for precise manipulation of atoms and molecules [1–9]. SPL also offers nanoscale control to construct spatially deﬁned regions of surfaces with reactive or adhesive ligands for the subsequent attachment of polymers [10], inorganic materials [11–14], or biomolecules [15–19]. This chapter will focus on applications of SPL with self-assembled monolayers (SAMs). Selfassembled monolayers (SAMs) have emerged as a ﬂexible and convenient way to designate chosen functionalities on a surface in selected spatial areas. SAMs have attracted research attention because of promising applications in biosensing [20–24], corrosion inhibition [25–27], lubrication [28–32], surface modiﬁcation [33–35], and molecular device fabrication [36, 37]. Precisely engineered nanostructures of SAMs provide a means for the exploration of chemical reactions under well-deﬁned environments. Although not yet practical for high throughput applications and manufacturing, SPL studies provide a platform for the fabrication of nanoscale test structures.

J.C. Garno · J.D. Batteas

106

With the invention and on-going development of scanning probe instruments, such as the scanning tunneling microscope (STM) [38] and the atomic force microscope (AFM) [39], changes to surfaces were observed when applying too much force using AFM, or with STM if the bias exceeded certain threshold voltages. It became apparent that the electric ﬁelds generated and charges injected by the STM tip, as well as the pressure at the tip-surface contact exerted by an AFM tip often changed areas of the sample surface. Researchers began experimenting to selectively control these alterations. In a sense, SPL is caused by intentionally damaging selected areas of the surface. What all SPL methods have in common is that an SPM tip is used as a tool for nanofabrication, as well as characterization of surfaces. Thus, the shape and dimensions of the tip dictate the detailed resolution of written nanostructures, with fabricated features as small as 5 nm having been reported, and line dimensions routinely achieve 10 nm [40, 41]. Steady progress has been made in manufacturing more uniform tips of consistent geometries and physical properties. The same SPM tips used for fabricating surfaces may also be used to characterize the morphology of nanopatterned surfaces, since AFM and STM are well-established methods for visualizing surfaces with high resolution [42–45]. In addition to SPL, scanning probe instruments can also be applied to investigate structure, chemical reactions and physical properties [46–50]. Examples include the exploration of chemical and biochemical reactions and fundamental investigations of tip-surface interactions, chemical structures, and material properties at the molecular level [51–53]. Table 25.1. Comparison of SPL methods used for nanofabrication of SAMs SPL method

Pen

Paper

Mechanism

Surface Chemistry

Bias-induced biased STM or nanofabrication AFM tip in air

conductive or semi-conductive substrate

surface oxidation oxidized SAMs

Bias-induced replacement

biased STM or AFM tip in SAM solution

conductive or semi-conductive substrate

displacement of SAMs under elevated bias

Nanoshaving

bare AFM tip

SAMs including silanes, thiols

force & sweeping uncovered areas tip motion of substrate

Nanografting

bare AFM tip in thiol SAMs a SAM solution

force & solution replacement

Nanopen reader & writer

thiol-coated AFM tip

thiol SAMs

force & molecular diverse functional replacement groups of SAMs

Dip-pen nanolithography

thiol-coated AFM tip

clean, uncoated surface

meniscus liquid transfer

diverse functional groups of SAMs

diverse functional groups of SAMs

diverse functional groups of SAMs & other materials

A variety of SPL approaches have been developed for fabrication using SAMs, such as bias-induced nanofabrication [54, 55], tip-directed displacement under force [41], and liquid transfer (dip-pen nanolithography) [56]. There are several

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

107

excellent reviews that describe recent advances for SPL nanofabrication methods [41, 57–61]. A useful analogy for describing SPL methods with SAMs is a pen that writes with molecules (ink) on various surfaces (paper). Table 25.1 summarizes the differences between the nanofabrication mechanisms and surface chemistry for several commonly applied SPL methods used for nanopatterning SAMs. These methods can be generally categorized based on the writing mechanisms: nanofabricated patterns are created by changes in bias voltage or mechanical force applied to the tip, or by capillary transfer of molecules from a coated tip to a substrate. 25.1.1 Bias-Induced Nanofabrication When an electric ﬁeld is applied between a conductive SPM tip and sample, at certain bias voltages the exposed surface can undergo chemical or physical changes. The surface changes may be a consequence of ohmic heating, which induces evaporation or desorption of surface layers in UHV [41, 62], or in ambient conditions the changes may result from electrochemistry at either the tip or sample that occurs from electric ﬁeld effects [63]. Figure 25.1 displays the general principle of bias-induced lithography. Typically, short (microsecond to millisecond) pulses of tunneling current or bias voltages are applied between a conductive SPM tip placed near the sample surface. The lateral dimensions of the surface changes are determined by the duration and magnitude of the electric ﬁeld, as well as by the tip geometry. Often, with biasinduced oxidation, the changes produced by an electric ﬁeld do not induce changes in molecular heights, and thus are not detected by topographic images. However, SPM imaging modes, which display contrast between different terminal groups, such as force modulation, lateral force imaging, conductive probe (i.e. current), or electrical force images, can clearly distinguish between the modiﬁed and unmodiﬁed areas of nanopatterned substrates. Requirements for bias-induced nanofabrication include a conductive or semiconductive substrate and a conductive SPM probe. To prepare conductive AFM

Fig.25.1.Overview of bias-induced lithography

108

J.C. Garno · J.D. Batteas

tips, a thin ﬁlm of metal, usually gold, is sputter-coated onto the surface of probes pre-coated with a binding layer of Cr conductive tips and cantilevers of silicon that exhibit sufﬁcient electrical conductivity for bias-induced modiﬁcation of surfaces have recently become commercially available. Other commercial probes include, for example, coatings of cobalt, diamond-like carbon, doped diamond, platinum, platinum/iridium, tungsten carbide, titanium nitride, and nickel. 25.1.2 Force-Induced Nanofabrication of SAMs Nanofabrication of SAMs can be accomplished by applying mechanical force to the monolayer using an AFM tip during scans. Diagrams of three such SPL methods that use force, nanoshaving, nanografting, and nanopen reader and writer (NPRW) [41] are presented in Fig. 25.2. These SPL methods do not require conductive tips or substrates, rather, the mechanism of displacement relies on applying an increased force to the AFM tip, sufﬁcient to penetrate through a matrix layer and scrape away selected areas of the matrix SAM. Of course, if too much force is applied, the tip and/or substrate can be damaged, so it is critical to determine the minimum threshold force for nanofabrication with each experiment. Commercially available Si3 N4 tips are fairly robust for these nanofabrication methods, and hundreds of patterns can be written within a single experiment without detecting changes in image quality, which would be indicative of tip damage.

Fig. 25.2. Force-induced nanofabrication with SAMs. (a) nanoshaving; (b) nanografting; (c) Nanopen reader and writer

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

109

For nanoshaving, as shown in Fig. 25.2a, fabrication is accomplished by exerting a high local pressure on an AFM tip, thus pushing the tip to penetrate through SAM to contact the underlying gold surface. During scanning, this pressure causes the displacement of SAM adsorbates following the scanning track of the tip to create a nanopattern. Holes and trenches are fabricated with one to several scans, to produce an area of uncovered substrate. Analogously to sweeping, nanoshaving usually requires several scans of the same area to completely remove SAM adsorbates. After returning to low force, images and cursor measurements can be acquired. The thickness of molecular layers can be measured by referencing the substrate as a baseline. Depending on the matrix material, the displaced SAM molecules are either pushed to the edges of the nanofabricated patterns or may dissolve in the imaging solvent. Nanoshaving has been conducted with systems such as thiol-terminated SAMs on gold [41, 64], alkylsilanes on mica [65–67], and porphyrin layers on glass substrates [68]. Nanografting combines adsorbate displacement with the self-assembly of thiols on gold in solution (Fig. 25.2b) [69]. In nanografting, the solution for imaging contains a different thiol than the matrix, which adsorbs onto the newly exposed gold surface during nanoshaving. As the AFM tip scans at high force through the monolayer, the matrix adsorbates are removed and immediately replaced by new molecules from the solution. The resulting nanografted structures can then be characterized by imaging at low force using the same AFM tip. By comparing the height of the surrounding matrix with the known thickness of an ordered patterned SAM, the height or depth of nanografted patterns measured from cursor proﬁles can give an accurate local measurement of layer thickness with angstrom precision. Nanografting can create both positive and negative height patterns depending on the difference in chain length between the nanopatterns and matrix layer. Edge resolution of ∼ 1 nm can be routinely obtained with nanografting. Nanopatterns of multiple thiol molecules can be fabricated serially, by exchanging imaging solutions. A drawback of nanografting is that the exchange between solution molecules and the surface matrix SAM takes place for some systems of alkanethiols. Self-exchange is an important consideration for nanografting longer chain thiols into a shorter chain matrix layer, thus it is important to use dilute (0.01–0.1 mM) solutions for nanografting. Exchange is often detected within 2–4 hours (depending on the age of the matrix SAM) when molecules from solution adsorb onto defect sites and at step edges. For NPRW (Fig. 25.2c) the AFM tip is ﬁrst coated with thiol molecules as “ink”, and is then used as a “pen” for writing [70]. Nanopatterns are made by applying higher force to the AFM tip during scanning, which enables simultaneous displacement of the matrix SAM and writing with ink from the coated tip. An alkanethiol matrix layer is used as the “paper” because the hydrophobic surface of methylterminated molecules resists the deposition of ink throughout the scanning areas. At low force, the ink remains on the tip, which serves to improve image resolution. For NPRW, because of the resist properties of an alkanethiol matrix “paper”, the same tip can be used to image or “read” the nanopatterns under low force, with very high spatial resolution. NPRW has been used successfully for writing nanopatterns of ω-functionalized alkanethiol SAMs [70], thiol-coated nanoparticles [11] and for writing arenethiol SAMs [71].

110

J.C. Garno · J.D. Batteas

Regardless of whether one chooses to use nanoshaving, nanografting, or NPRW for surface modiﬁcation, it is important to emphasize that these SPL methods work best for nanosized structures. For patterning larger areas, greater than 500 nm, for example, the removal of matrix SAMs is not efﬁcient and build-up of displaced materials can occur. Therefore, the SPL methods presented in Fig. 25.2 are best applied for exquisite, nanoscale experimental designs, (< 200 nm) providing the best spatial precision attainable for AFM-based lithography [40]. Commercially available soft Si3 N4 cantilevers, which have force constants ranging from ∼ 0.03 N/m to ∼ 0.5 N/m, have mostly been used for nanofabrication by mechanical force. When imaging in liquid, the total force applied typically is less than 1 nN to prevent damage to surface layers. The fabrication forces used for nanoshaving, nanografting and NPRW typically range from ∼ 2 nN to 10 nN, depending on the system under investigation, as well as the tip geometry. 25.1.3 Dip-Pen Nanolithography (DPN) DPN is the most predominant method of AFM-based nanofabrication techniques using SAMs. In DPN an AFM tip is typically coated with an alkanethiol ink to write on clean gold substrates or “paper” [56, 72]. In DPN, the ink molecules migrate from the coated AFM tip through a capillary meniscus to the substrate by diffusion (Fig. 25.3). DPN uses an AFM tip as a “pen”, molecules with a chemical afﬁnity for the substrate are the “ink”, and the substrate serves as the “paper”. In DPN, capillary transport of molecules from the AFM tip to the substrate is used to directly write patterns. Additional force is not applied to the AFM tip or pen for DPN. DPN has been used to generate individual lines, dots, grids and arrays of alkanethiols [73,74]. When an AFM tip is used in air to image a surface, the narrow gap between the tip and surface forms a tiny capillary meniscus from the condensation of water.

Fig. 25.3. Schematic representation of the principles of DPN. The size of the water meniscus formed between the coated AFM tip and the Au substrate, and the duration of contact affects the resolution of nanopatterning. Reproduced from [56] with permission

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

111

In DPN, the water meniscus is used to transport organic molecules from the tip to the surface. The resolution of DPN depends on several parameters, such as the geometry of the AFM tip, the humidity of the ambient environment, as well as on the duration which the inked tip is placed in contact with the surface – typically on the order of ∼ 1 s to 10 s. The size of the water meniscus that bridges the tip and substrate depends on the relative humidity [75, 76]. With typical commercial cantilevers, DPN can provide linewidths down to ∼ 15 nm and ∼ 5 nm spatial resolution. 25.1.4 Automated Scanning Probe Lithography DPN has been transformed from a serial to a parallel process through the use of cantilever arrays [77, 78] (Fig. 25.4). The commercial introduction of multiple-tip cantilevers and advances in programming for SPM instruments provides capabilities for wider applications of SPL technology. Automated DPN applications include nanofabrication with small molecules, metal salts, dyes, nanoparticles, polymers, DNA and proteins [79]. Quate and coworkers have shown that as many as 50 tips can be used at once with improvements in both imaging and patterning speeds [80]. In addition to dedicated systems for nanofabrication, almost all commercial SPM controllers can now be programmed to enable automated scanning probe lithography (ASPL) to rapidly and uniformly create desired surface arrangements of nanopatterns [81]. Typical systems include the capabilities of controlling the length, direction, speed, bias and the applied force of the scanning motion, which is analogous to a pen-plotter. Automated SPL offers tremendous advantages for the speed and reproducibility of nanopatterning. ASPL can produce highly sophisticated pattern arrangements and geometries, with greater precision and reproducibility for the alignment, spacing and shapes of nanopatterns. Precise automated lithography using AFM is limited by several physical parameters, which cause imperfections at the nanometer scale. Distortions in writing nanopatterns can occur because of the non-linear response of the piezoelectric tube scanner, and twisting of the cantilever when too much force is applied to the tip. Tube-shaped scanners have a parabolic distortion at the edges of their maximum XY lateral ranges; therefore, for open-loop scanners it is critical to choose fabrication

Fig. 25.4. A multi-pen probe array of 32 AFM cantilevers (SEM micrograph). The insert provides a view of a single tip. Reproduced from [77] with permission

112

J.C. Garno · J.D. Batteas

locations within the linear range of the scanner. The response of the piezo to changes in voltages is not perfect, piezoelectric materials relax slowly, causing creep and hysteresis. These effects can be countered by optimizing the scanning speed, and by using separate computer statements to control movements individually for the x and y directions, rather than using vectorial displacements. Nanopattern geometries also are usually distorted when applying too much force to an AFM tip. High forces cause the tip to twist, and shift the alignment of nanopatterns, resulting in spurious marks in undesired areas as the tip is translated between patterns. The optimized force is found to vary with the sharpness of the cantilever, the imaging media and the surface material to be fabricated. With ASPL, a rapid systematic approach can be used to address this problem. Using a simple lithography routine, the load or bias can be successively increased during fabrication for each element of an array of square frames or holes. After imaging the fabricated array, one can easily observe that when the applied force or bias is too low, patterns are not fabricated. As the writing parameters are incrementally increased for each pattern, the ﬁrst nanopattern fabricated within the array pinpoints the threshold value for the optimized force/bias. The minimum force is a crucial parameter with force-induced nanofabrication methods for alleviating distortion caused by twisting, and for preventing damage to ultrasharp tips. With an optimized force setting as many as 200 nanopatterns have been rapidly produced by a single AFM tip, within a 4 h time frame.

25.2 Patterning with Self-Assembled Monolayers Molecular self-assembly is a simple method for reproducibly preparing organic ﬁlms with controlled thickness and surface properties. The ability to control the nature and density of functional groups presented at the surface of SAMs makes them excellent materials for SPM-based nanofabrication methods. The formation, characterization, and properties of SAMs have been described and reviewed previously [82–86]. SAMs offer a simple way to produce well-ordered, crystalline structures at the molecular scale, and present a diversity of chemically well-deﬁned terminal functional groups for nanopatterns. Also, changing the length of the chains between the head and tail groups provides further structural control for experimental designs. 25.2.1 Structure of SAMs Close-packed monolayers of SAMs can be readily prepared with high reproducibility to present a variety of functional groups such as alkyls, amides, esters, alcohols, and nitriles on gold surfaces. Typically, SAMs are formed by soaking gold thin ﬁlms in dilute (0.1–1.0 mM) thiol solutions of 2-butanol or ethanol. The substrates can remain in a thiol solution for 1–7 days at room temperature to ensure the formation of mature monolayers. √ √ Alkanethiols on Au(111) form a close packed, commensurate ( 3 × 3)R30◦ lattice on Au(111) surfaces [28, 85, 87–89], shown schematically in Fig. 25.5. In

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

113

Fig. 25.5. Surface structure of alkanethiols/Au(111). (a) Side view. (b) Top-view. Sulfur (gray circles) binds at the triple hollow sites of Au(111)

surface assemblies of alkanethiol SAMs, according to studies by infrared (IR), near-edge X-ray absorption ﬁne structure (NEXAFS) spectroscopy, and grazing incidence X-ray diffraction (GIXD), the alkyl chains of the thiol molecules are tilted approximately 30° with respect to the surface normal [32,88,90–92], as indicated in Fig. 25.5a. The sulfur atoms of the alkanethiol molecules are considered to bind at the triple hollow sites of Au(111) lattices [82]. STM and AFM studies have conﬁrmed the long-range order and periodicity of alkanethiol monolayers, and have enabled a direct view of defects such as domain boundaries, etch pits, steps and dislocations within such ﬁlms [82, 93]. Visualization of the intimate details of surface topography is a powerful asset for SPM investigations with SAMs. Typical examples of AFM and STM images of dodecanethiol are presented in Fig. 25.6. Viewed from the perspective of SPM images, the molecular landscapes may appear somewhat rough, because at the atomic scale most surfaces are not truly smooth and ﬂat, and contain defects. However, we keep in mind that the height of gold steps is around 0.25 nm and the overall surface roughness of the underlying gold substrates for these samples is less than ∼ 1 nm.

Fig. 25.6. Topographic views of dodecanethiol/Au(111). (a) AFM topograph. (b) STM view. (c) STM molecular lattice image

114

J.C. Garno · J.D. Batteas

Alkylsilane-based molecules can be used to form SAMs on silicon (with a native oxide), mica or glass substrates (Fig. 25.7) and both DPN [94, 95] and nanoshaving [13, 65] have been applied using these systems. The properties of alkylsilane assemblies are quite different from SAMs of alkanethiols. Similar to alkanethiol SAMs, the chain lengths and terminal moieties can be tailored for ﬂexibility in meeting experimental requirements. Substrates on which silane SAMs have been prepared include silicon oxide, aluminum oxide, quartz, glass, mica, zinc selenide, germanium oxide and gold [86]. SAMs of alkylchlorosilanes, alkylalkoxysilanes, and alkylaminosilanes require hydroxylated surfaces to form polysiloxane, which is connected to surface silanol groups (–SiOH) via a network of Si–O–Si bonds. High-quality silane SAMs are not as simple to produce as thiol SAMs because of the need to carefully control the presence of water in solution. Reproducibility is a problem, since the quality of the monolayers formed is very sensitive to reaction conditions. Silane monolayers on mica typically consist of domains separated by boundaries. Within domains, silane molecules form structures without long-range order or periodicity [67, 96–98]. The headgroups of silane SAMs cross-link into a Si–O network, and the chains tilt ∼ 15◦ from the surface normal [67, 97].

Fig. 25.7. Molecular model of an alkylsiloxane monolayer on Si/SiO2

25.2.2 Examples of SAM Nanopatterns Generated by Force-Induced SPL In addition to high-resolution visualization of the morphology of surfaces, combining AFM imaging with SPL provides highly local measurements of the height of molecular layers with angstrom precision [40]. Since the thicknesses for many n-alkanethiols are well-known, comparing the height difference between the matrix and nanopatterned thiols yields thickness measurements for new molecules [64, 71]. For example, Fig. 25.8a displays a single square nanopattern (200 × 200 nm2 ) of dodecanethiol nanografted within a matrix layer of glucose-terminated thiol [99]. A fairly ﬂat location at the central area of an Au(111) terrace was chosen for writing

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

115

the nanopattern. The height difference between the dodecanethiol nanopattern and the glucosylamide thiol is shown by the cursor proﬁle of Fig. 25.8b to be ∼ 0.70 nm. The thickness of an ordered ﬁlm of dodecanethiol is known to be ∼ 1.54 nm. Therefore, the thickness of the glucosyl amide thiol layer is ∼ 0.84 nm. The fully-extended theoretical length of glucosyl amide thiol molecules is approximately 1.2 nm. The molecular structure is shown in Fig. 25.8c. Using NPRW, another force-induced SPL example of writing square nanopatterns of SAMs (200 × 200 nm2 ) is presented in Fig. 25.9 [71]. The matrix layer is 2ﬂuoro-4-phenylethynyl-1-[(4-acetylthio)-phenylethynyl] benzene (F-OPE). The two nanopatterns of dodecanethiol display a negative height in comparison to the matrix layer of F-OPE (Fig. 25.9a). The corresponding lateral force image (Fig. 25.9b) exhibits sharp contrast between the nanopatterned regions and the F-OPE matrix, clearly resolving the area that was patterned. Nanopatterns of dodecanethiol were found to be ∼ 0.3 nm shorter than the surrounding F-OPE matrix, yielding the height of the F-OPE monolayer as ∼ 1.8 nm. The AFM images were acquired with a dodecanethiol coated AFM tip, however features of the underlying monoatomic steps of Au(111) and surface defects are still clearly resolved when imaging in air. The improved resolution results from coating tips with a functionality that sharpens the image contrast [70]. As previously demonstrated, even after force-induced

Fig. 25.8. Example of a nanografted pattern of dodecanethiol written within a glucosyl amide thiol SAM. (a) Topographic view of the 200 × 200 nm2 nanopattern. (b) Representative cursor proﬁle across the nanopattern. (c) Molecular representation of the matrix SAM and nanopattern

116

J.C. Garno · J.D. Batteas

Fig. 25.9. AFM lithography via NPRW in air. (a) Square nanopatterns of dodecanethiol (200 × 200 nm2 ) written in a matrix of ﬂuorinated OPE using NRPW. (b) The corresponding lateral force image displays dark contrast for the nanopatterned areas of dodecanethiol. (c) Schematic of the height difference observed for the nanopatterns. Reproduced from [71] with permission

nanofabrication of SAMs, AFM tips can still achieve resolution of the molecular lattice [40, 69, 70]. The contrast in lateral force images (Fig. 25.9b) results from distinct differences in tip-surface interactions between the terminal groups of the nanopatterns compared to the terminal groups of the matrix SAM. The tip experienced different torsion in the light and dark areas, which is generally interpreted as changes in friction. Friction is deﬁned as the force opposing motion, and for nanoscale frictional measurements, the frictional force between the tip and sample results in torsional twisting of the cantilever as it is rastered over the surface. Friction between two surfaces depends on both chemical and mechanical interactions. The technique for measuring these forces is called lateral force or frictional force microscopy (LFM or FFM). As the AFM probe moves over a surface, changes in the chemical composition of local areas can cause changes in the torsion of the cantilever on which the probe is mounted. Thus, the torsion of the cantilever is proportional to the friction between the probe and surface. Dramatic differences in lateral and vertical AFM resolution can make it difﬁcult to view the intricate details of nanopatterns with topographic images spanning wide areas. Typically, the height differences of nanopatterns are only a few angstroms in vertical height, in comparison to areas spanning tens to hundreds of nanometers in lateral dimension. Since the imaging mechanism for LFM is dependent on chemistry, frictional contrast can clearly distinguish nanostructures for much larger areas, even spanning micron-sized regions. Figure 25.10 illustrates the differences in topographic (Fig. 25.10a) and frictional contrast (Fig. 25.10b) for 16 nanopatterns written within a 1 × 1 µm2 area. The nanografted patterns were written with hexadecanethiol into a matrix SAM of glucosylamide disulﬁde [99] and the images were acquired in ethanol. For the large scan area of Fig. 25.10a, only the bottom row of nanopatterns is clearly resolved in the topographic view. However, the frictional force image clearly displays the location and dimensions of the 4 × 4 array of circular designs. This set of nanopatterns was written in about four minutes using automated SPL, and demonstrates the potential for writing sophisticated designs. A zoom-in view of the topography for four of the nanopatterns is shown in Fig. 25.10c. The designs are

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

117

Fig.25.10.A 4×4 array of complex circular nanostructures. (a) Wide-area topograph (1×1 µm2 ), only the bottom rows can be distinguished within this large scale area. (b) LFM images clearly display the contrast for nanopatterns, even over areas as large as 1 × 1 µm. (c) Zoom-in view (600 × 600 nm2 ) of four nanopatterns

drawn by writing four rings (50 nm diameter) with adjoining sides. The width of the rings is ∼ 8 nm, inscribed by outlining each circle three times with the AFM tip. The rows and columns are regularly and uniformly spaced at 100 nm increments, illustrating the exquisite precision achievable for pattern alignment even when using an open-loop scanner. Topographic images of several example arrays of SAM nanopatterns that can be produced by nanografting are shown in Fig. 25.11. These AFM images demonstrate the exquisite resolution and selectivity achievable using SPL, in which the designs and geometry of nanopatterned surfaces are only limited by imagination. The ﬁrst example (Fig. 25.11a) discloses a wide view of a 4 × 5 array of negative height patterns written within a hexadecanethiol SAM matrix. The 150 × 150 nm2 glucosyl disulﬁde patterns are uniformly spaced at 150 nm distances. The uppermost row is not clearly resolved with topographic imaging within this large scale area of 1.6 × 1.6 µm2 . Automated SPL achieves precise control of the pattern geometry, spacing and alignment.

Fig. 25.11. Examples of automated nanografting with SAMs. (a) Topographic view of an array of glucosyl disulﬁde patterns written within a hexadecanethiol SAM matrix. (b) Array of square (100 × 100 nm2 ) nanopatterns of hexadecanethiol grafted into a glucosyl disulﬁde SAM. The nanopatterns are ∼ 0.8 nm taller than the matrix. (c) Rings written with glucose-terminated thiol are shallower than the hexadecanethiol matix

118

J.C. Garno · J.D. Batteas

Nanopatterns are moved closer together in the second example (Fig. 25.11b), which displays a 5 × 5 array of positive height patterns (100 × 100 nm2 ) of hexadecanethiol nanografted into a glucosyl disulﬁde SAM. The nanopatterns are ∼ 1 nm taller than the matrix glucosyl disulﬁde SAM. The interpattern spacing is 50 nm for this example, and the writing density for each nanopatterned element was identical, evidencing the reproducibility achievable with SPL. Figure 25.11c exhibits an even denser packing of elements within an array for rings of glucosyl disulﬁde written into a hexadecanethiol matrix SAM. This example displays a zoom-in view of a 5 × 5 array of negative height ring structures, written with decreasingly smaller diameters (100, 80, 60, 40, 20 nm). The spacing between rings is ∼ 40 nm along the rows, and the distance between rows varied from 50– 100 nm. The width of the rings averages 10 nm. The area viewed is 0.5 × 0.5 µm2 , and each ring-shaped element along a row is drawn with successively smaller sizes, thus the height differences of the surface topography are more apparent with close-up views of surface morphology. The ability to chemically functionalize surfaces with different terminal groups on the nanometer length scale opens diverse possibilities for using these nanostructures as templates for further construction of three-dimensional molecular nanoassemblies. Nanopatterned SAMs provide functional templates for anchoring other materials such as polymers, metals, and proteins, and several examples will be described in the next sections of this chapter. 25.2.3 Nanofabrication of SAMs by DPN and Bias-Induced SPL DPN was ﬁrst used to pattern alkanethiol SAMs when introduced in 1999, and since then has been applied for many other ink-substrate combinations. Various SAM inks used for DPN include nonanedithiols [100], octadecanethiol [56, 72, 73, 76– 78, 101–103], 16-mercaptohexadecanoic acid [72, 73, 78, 102–106], and ferrocenylalkylthiols [103,107]. Nanoscale features such as dots, arrays and grids produced by DPN are shown in Fig. 25.12. Because there are too many systems studied by this approach to discuss each in detail here, we direct readers to a detailed overview of DPN studies as reported by Mirkin and co-workers [79]. There are several examples of bias-induced nanofabrication of SAMs. STMbased lithography was effected by applying voltage pulses to a SAM covered gold substrate immersed in 1,4-dioxane containing the molecules to be patterned [108]. After pulsing, small areas of the surface were exposed for adsorption of a new molecule. Molecules of an insulating dodecanethiol matrix were replaced in situ with conjugated molecules of 2 -ethyl-4:1 -ethynylphenyl-4 :1 ethynylphenyl-1,4 thioacetylbenzene. The nanostructures were approximately 10 nm in dimension, corresponding to ∼ 400 conjugated molecules. Bias-induced addition lithography has been accomplished by Maoz et al., who used an AFM conducting tip to induce electrooxidation of silane SAMs on silicon [109]. The oxidized surface groups were then used for site-selective selfassembly of a second silane monolayer with desired functional groups. Gorman and co-workers has demonstrated an STM-based replacement lithography (Fig. 25.13) in which at elevated bias voltages the matrix layer was replaced

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

119

Fig. 25.12. Lateral force images of molecular dots, arrays and grids prepared by DPN using ODT or 16-MHA as ink. (a) A gold substrate after an ODT-coated AFM tip was placed in contact with the substrate for 2, 4, and 16 min (left to right). (b) Dots of 16-MHA written on an Au substrate. The AFM tip was held on the Au substrate for 10, 20, and 40 s (left to right). (c) An array of dots generated by placing an ODT-coated tip in contact with the surface for ∼ 20 s. (d) A grid of lines 100 nm in width and 2 µm in length. Reproduced from [56] with permission

with different molecules from solution in a nonpolar medium [54, 62]. SAMs terminated with electroactive headgroups ferrocene and galvinol were nanopatterned, and then the electronic properties of the new thiolate SAM was measured in situ using the same STM tip [110]. Nanopatterns of diacyl 2,6-diaminopyridine decanethiolate were further functionalized by chemical recognition of complementary ferroceneuracil and dodecyl uracil to form electroactive or inactive molecular assemblies within a matrix decanethiol monolayer [111].

Fig. 25.13. Examples of bias-induced replacement lithography using STM. Patterns of the letters NCSU in which (A) dodecanethiol replaced a decanethiol SAM, written at +3.8 V; (B) decanethiol replaced selected areas of a dodecanethiol SAM written at +3.6 V. Reproduced from [54] with permission

120

J.C. Garno · J.D. Batteas

25.3 Directed Fabrication of Polymeric Structures To fabricate complex 3D nanostructures using the bottom-up approach, nanopatterned SAMs with selected terminal functionalities can be used as templates for further reactions, such as polymerization. This approach applies a chemical ampliﬁcation of the constructed nanopatterns to provide further 3D control of the surface height and functionality of the assemblies. The methods applied for nanofabrication of polymeric materials include DPN, bias-induced lithography, as well as forceinduced SPL methods. A diverse range of polymeric materials have been nanopatterned, ranging from dendrimers to conductive polymers, to form well-organized discrete structures of monomers and polymers at the nanometer scale. DPN was applied by Mirkin and co-workers to fabricate arrays of polymer brushes using ring-opening metathesis polymerization (ROMP) at nanometer length scales [112]. One DPN approach with ROMP used a DPN-patterned template to initiate polymerization, which was passivated with decanethiol (Fig. 25.14). Mirkin and co-workers also reported a second method applying ROMP in which DPN was used to directly pattern various monomers on substrates for catalyst activation. Xu and Kaplan applied a nanopatterned templating approach with peroxidase-catalyzed polymerization of phenols on structures of 4-amino-thiophenol patterned on gold by DPN [113]. Monomers have been used as ink with DPN to construct polymer nanopatterns. Solutions containing pyrrole in tetrahydrofuran (acid-promoted polymerization of pyrrole) were written by DPN to fabricate nanopatterns of a conducting

Fig. 25.14. Polymer brush nanostructures prepared surface-initiated ROMP using DPN. (a) AFM topograph of polymer brush lines. (b) Cross-sectional trace for a selected line from (a). (c) AFM image of a dot array of polymer brushes. (d) Cross-section for a selected line from (c). Reproduced from [112] with permission

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

121

polymer [114]. Dendrimers were used as ink on DPN tips pre-coated with a silicone elastomer [115]. Conducting polymers of poly(aniline sulfonic acid) and polypyrrole were used as inks for DPN on charged substrates, in which electrostatic adsorption was the driving force for ink transport [116]. Other applications of DPN include writing nanopatterns of polymerized 3,4-ethylenedioxythiophene [117] to fabricate nanostructures down to ∼ 30 nm linewidths. DPN has also been applied for writing poly[2-methoxy-5-2 -ethylhexyl)oxy-1,4-phenylenevinylene] to form lines of lightemitting polymeric nanowires [118]. Bias-induced SPL was used by Sotzing and co-workers to convert nanosized areas of a ﬁlm of a polymer precursor into a conducting polymer [119]. The electrochemical process used a conductive AFM tip as the working electrode to apply sufﬁcient potential at the tip-surface contact to initiate cross-linking of the polymer precursor via solid-state oxidation. Nanoshaving has been applied by Kaholek et al. to pattern polymer brushes using atom transfer radical polymerization, (Fig. 25.15) to prepare surface-attached polymer brushes of controlled dimensions [120]. A matrix layer of methylterminated octadecanethiol was used as a resist. Areas of the resist SAM were removed by nanoshaving, and the freshly exposed gold surfaces were backﬁlled by self-assembly of a thiolated initiator, ω-mercaptoundecyl bromoisobutyrate (BrC(CH3 )2 )COO(CH2 )11 )SH) to form an initiator pattern. The patterns were then exposed to a polymerizing solution of N-isopropylacrylamide in water for 60 min to form nanopatterned poly(N-isopropylacrylamide) brushes. Nanografting was applied to fabricate designed 3D nanostructures by selective surface reactions of octadecyltrichlorosilane (OTS) with surface terminal hy-

Fig. 25.15. Line patterns of pNIPAAM generated by ﬁrst removing a thiol-resist by nanoshaving under large normal forces (∼ 50 nN) using an AFM tip, followed by surface-initiated polymerization of NIPAAM for 60 min using a backﬁlled, covalently attached thiol initiator. The panels display contact mode AFM height images (20 µm × 20 µm) and corresponding height proﬁles of the nanopatterns for images acquired at room temperature in (a) air, (b) MQ-grade water, and (c) a mixture of MeOH/water (1 : 1, v : v). The labels 1–5 indicate the time (minutes) required for nanoshaving the lines. Reproduced from [120] with permission

122

J.C. Garno · J.D. Batteas

droxyl groups, as shown in Fig. 25.16 [10]. Hydroxyl-terminated SAM nanopatterns (11-mercaptoundecanol) written by nanografting were used to anchor alkylsilane molecules for pattern transfer. First, nanostructures of functionalized thiol SAMs were produced using nanografting. Then, a new reactant was introduced to attach to the reactive termini. Octadecanethiol was used as a resist SAM to prevent nonspeciﬁc attachment of silanes. The hydroxyl termini reacted with OTS to form tailored 3D structures. Trichlorosilane headgroups are known to react with hydroxyl groups, to form a network of Si–O bridges.

Fig. 25.16. Nanografting was used to generate 3D nanostructures. (a) Topographic view of the mercaptoundecanol//Au(111) SAM matrix (600 × 600 nm2 ). (b) After a square frame (300 × 300 nm2 ) of octadecanethiol was nanografted, the nanopattern is 0.7 ± 0.2 nm taller than the matrix. (c) The same area after 3 min of immersion in 10 mM octadecyltrichlorosilane in decahydronaphthalene. The height of the matrix areas increased by 1.7 ± 0.5 nm. (d) A combined cursor plot reveals the changes in height. Reproduced from [10] with permission

25.4 Fabrication of Metallic Structures The ability to precisely construct nanoscopic metal-molecule-metal junctions is important for the development of molecule-based electronic and optoelectronic de-

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

123

vices [121–124]. Nanopatterned SAMs with various terminal functionalities can serve as templates for directing the deposition of metals or metal nanoparticles to fabricate complex 3D nanostructures from the “bottom-up”. Elegant examples of chemistry-directed assembly of nanoparticles on SAMs has been reported by Sagiv and co-workers using nanopatterned organosilane monolayers as templates for assembling gold clusters [13, 125]. Tip-mediated local oxidation of the surface –CH3 groups of an octadecyltrichlorosilane layer on silicon was used to write nanopatterns with –COOH functions. Using chemical derivatization, the –COOH groups were then used to generate disulﬁde, thiol, and amine functionalities as templates for guiding the assembly of gold nanoparticles (Fig. 25.17). Liu and co-workers also used silane monolayers for the site-selective assembly of gold nanoparticles, in which bias-induced local oxidation of the silicon substrate was followed by SAM deposition for assembling Au nanoparticles [14]. DPN methods have also been applied for patterning metal nanoparticles, in which thiol SAMs were used either as resists [12] or as the linker groups [126] for attaching nanoparticles.

Fig. 25.17. Using template-guided hierarchical self-assembly and bias-induced lithography, Liu et al. generated nanoscale art using the outline of a work by Picasso. The upper friction image (center) shows the monolayer after bias-induced lithography inscribed a pattern of –COOH groups on the top surface of an OTS/Si SAM. The pattern was inscribed using a tip-surface bias of 8.5 V. The upper topography image (right) displays the bilayer pattern produced by the self-assembly of a nonadecenyltrichlorosilane (NTS) overlayer on the –COOH patterns. The bottom images display the ﬁnal metal-organic composite nanostructure after the self-assembly of 2–6 nm [Au-citrate] particles on the amino-terminated bilayer template. A detailed scan of the marked area in the full-size image is shown in the bottom right panel. Reproduced from [13] with permission

124

J.C. Garno · J.D. Batteas

Two SPL methods using force-induced nanofabrication of SAMs have been developed to precisely position gold nanoparticles on surfaces [11]. The gold nanoparticles were encapsulated with alkanethiol shells, which were modiﬁed via placeexchange reactions to include an outer shell of mixed thiol composition comprised of alkanethiol and alkanedithiol molecules [127]. The surface-active nanoparticles could then be anchored to gold surfaces via sulfur-gold chemisorption. One method is based on nanoshaving followed by selective adsorption of mixed-shell nanoparticles. Regions of a resistive alkanethiol matrix SAM are shaved by the AFM tip under high force in a solution containing mixed monolayer shell nanoparticles. Nanoparticles then adsorb onto the exposed areas deﬁned by nanoshaving. The second method (Fig. 25.18) is based on NPRW, in which the AFM tip is coated with surface-active nanoparticles instead of using thiols as ink. Under low force AFM images can be acquired, since the nanoparticles remain on the tip and do not deposit on the resistive SAM matrix. When a higher load is applied, areas of the SAM matrix are uncovered and nanoparticles are written following the scanning track of the AFM tip. For both methods, nanostructures can be characterized in situ using the same tip at reduced load to resolve individual particles within the nanopatterns. The methyl-terminated matrix SAM effectively resists the nonspeciﬁc binding of nanoparticles, and prevents lateral diffusion of nanoparticles. The thiol groups on the nanoparticle surfaces offer an advantage for chemisorptive attachment to gold surfaces. By combining SPL with electroless metal deposition, metal-molecule-metal junctions can be constructed, in which the dimensions and surface organization are dictated by the placement of anchoring surface groups. Electroless deposition of metals has previously been applied as a practical approach to the controlled deposition of metals onto surfaces [128–131]. Nanoscale metal-molecule-metal junctions were constructed by electroless deposition of copper onto 16-MHA nanopatterns, constructed via nanografting. The steps of patterning, copper deposition and subsequent interrogation of the resulting structures were all conducted in situ. The

Fig. 25.18. Nanopattern of thiol-modiﬁed nanoparticles written directly using NPRW. (a) AFM topographic view of a decanethiol SAM (550 × 550 nm2 ), where a 150 × 450 nm2 rectangle of surface-active nanoparticles was written. (b) Cursor proﬁle shows the height of the nanopattern. Reproduced from [11] with permission

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

125

key steps for fabricating copper nanocontacts are illustrated in Fig. 25.19. First, nanopatterns of 16-MHA were written within a matrix layer of 11-MUD, which was found to resist metal deposition. Next, the 16-MHA solution in the AFM sample cell was replaced by rinsing the cell successively with ethanol and water, followed by an alkaline plating solution (CuSO4 ·5H2 O, C4 H5 O6 Na·H2 O, and CH2 O at pH 12.8). Upon introduction of the solution, electroless deposition of copper occurred via an autocatalytic reaction in which metal ions (Cu2+ ) in solution are reduced by formaldehyde in the plating solution, and deposit speciﬁcally on the patterned acid regions of the surface as metal layers (Cu0 ) [132]. The deposition of several nanometers of copper occurs fairly rapidly, within 30 min. Exchanging the plating solution in the sample cell with water and ethanol halts metal deposition.

Fig. 25.19. Process of electrodeposition of copper on a nanografted pattern. (a) Topographic and (a ) LFM image of an 11-MUD SAM/Au(111). (b) Same area after nanografting a 200×200 nm2 pattern of 16-MHA. (b ) LFM image of nanopatterned area. (c) After electroless deposition of copper. (c ) Cursor proﬁle after copper deposition

126

J.C. Garno · J.D. Batteas

For the nanopatterned example, ﬁrst an area was chosen for nanografting (Fig. 25.19a). In the topographic image, the selected area of the 11-MUD matrix SAM displays a spiral staircase arrangement of nine Au(111) steps, with a few dark defect scars. The corresponding LFM image (Fig. 25.19a ) does not exhibit any distinguishable contrast differences. After nanografting a 200 × 200 nm2 pattern of 16-MHA (Fig. 25.19b), an outlined frame is vaguely distinguishable in the topographic image. However, the nanopattern area is readily apparent in the corresponding LFM image (Fig. 25.19b ), because of the differences in terminal groups between the 16-MHA pattern (carboxylate) and 11-MUD matrix (hydroxyl) areas. After introducing the metal plating solution, the resulting nanopattern is shown in Fig. 25.19c, the color scale has been saturated to allow us to reference the surrounding terraces of concentric steps as a landmark for this in situ experiment. According to the cursor proﬁle of Fig. 25.19c , the height of this copper nanopattern is 7 ± 1 nm. The surrounding matrix areas of 11-MUD did not display any copper deposits. Using this approach, the architecture of the copper nanostructures is tunable in all dimensions by controlling the written 2D pattern size of the fabricated structures, the surface density of reactive acid groups, and the concentration of reactants in the electroless plating solution.

25.5 Nanoscale Patterning of Proteins Protein patterning is a critical technology for the integration of biological molecules into miniature bioelectronic and sensing devices. Protein micropatterning has been applied for biosensors and biochips [133–136]. Direct applications of protein patterning include biosensing, medical implants, control of cell adhesion and growth, and fundamental studies of cell biology [137–140]. Understanding the interactions of protein binding to substrates or antibodies is crucial for developing workable technologies for biosensing. In current research efforts, protein patterning has been accomplished at the micrometer level using microcontact printing [141–146], photolithography [147–149], and ion beam lithography [150]. Recently, PDMS stamps (typically used for microcontact printing) were used as channels to direct small volumes of protein solutions into networks to create protein patterns on various surfaces [151, 152]. Thus, capabilities for micrometer scale spatial arrangements of biomolecules have been developed, and are being further reﬁned. Some of the ﬁrst studies using nanopatterned SAMs for protein immobilization were conducted in 1999 by Liu and co-workers using nanografting [153]. Since then, a growing number of investigators have taken advantage of the ﬂexibility of SAM chemistry and SPL nanopatterning as a tool for probing the surface chemistry of biomolecular interactions. Immobilized biomolecules on a surface serve as the receptor and in some cases as the signal transducer in biosensors. Therefore, the placement of biological ligands in precisely deﬁned locations can increase the density of sensor elements and lead to improved detection limits, and molecular-level control of the surface reactivity [154–158]. As a proof-of-concept, Lee et al. have reported a nanometer-scale antibody array to test for the presence of the human immunodeﬁciency virus type 1 (HIV-1) in blood samples prepared by DPN (Fig. 25.20).

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

127

Fig. 25.20. Nanoscale protein assay generated by DPN for detection of HIV-1 p24 antigen (0.2 pg/mL). (a) Array of anti-p24 IgG protein. The height trace is consistent with a monolayer of anti-p24 IgG (6.4 ± 0.9 nm (n = 10)). (b) After p24 binding to anti-p24 IgG, an average height increase of 2.3 ± 0.6 nm (n = 10) is observed. (c) After ampliﬁcation with anti-p24 IgG coated gold nanoparticles (20 nm), an average topographic change of 20.3 ± 1.9 nm (n = 10) is observed. Reproduced from [159] with permission

With a nanoarray-based approach, the three-component sandwich assay exceeded the limit of detection of conventional enzyme-linked immunosorbent assay (ELISA) based immunoassays by 1000-fold [159]. Tools for nanofabrication will provide important contributions in developing biochip and biosensing technologies, as well as supply basic research in protein– protein interactions. It can be anticipated that array-based technologies in proteomics including protein-based biochip and biosensing devices will signiﬁcantly advance biotechnology, clinical diagnostics, tissue engineering, and targeted drug delivery [160–164]. Methods of high-throughput protein analysis offer immense potential for fast, direct and quantitative detection, including the possibility of screening thousands of proteins within a single sample to test for protein, ligand, and drug interactions. AFM provides direct views of how the surface morphology and geometry of nanoengineered surfaces inﬂuence protein binding. The conditions can be chosen to meet environmental requirements, such as ambient or cooled temperatures, aqueous buffered media and the reagents can be scaled to dilute micromolar concentration ranges. The orientation of proteins on surfaces is determined by factors such as the type of binding and the positions and composition of external residues on the protein surface. The nanopatterned terminal moieties mediate the type of binding, such as through covalent, electrostatic, speciﬁc interactions, or molecular recognition. 25.5.1 Protein Arrays Generated by DPN Nanoscale protein arrays have been generated on SAMs using DPN to write nanopatterned SAM templates for protein adsorption. Lee et al. reported using DPN for immobilizing lysozyme and IgG on 16-MHA nanopatterns passivated with

128

J.C. Garno · J.D. Batteas

11-mercaptoundecyl-tri(ethylene glycol) [165]. The arrays were then used to study biological recognition processes with antibodies and cells. Hyun et al. used DPN to write nanopatterns of 16-MHA, which were then conjugated with biotin derivatives to mediate molecular recognition-based immobilization of biotinylated proteins [19]. Hyun et al. applied DPN for investigations with elastin-like polypeptide nanostructures written on 16-MHA arrays. By coating AFM tips with two components, proteins and either thiotic acid [16] or 2-[methoxypoly(ethyleneoxy)propyl]trimethoxysilane [166], Lim et al. developed a DPN strategy for directly patterning arrays of IgG and lysozyme (Fig. 25.21). The protein nanopatterns were then tested for biorecognition properties with anti-IgGs. Using a tip inked with DNase I, Hyun et al. applied DPN to write enzyme patterns on a substrate functionalized with an oligonucleotide-terminated SAM [167]. The DNase I enzyme was observed to selectively digested the immobilized oligonucleotides on the nanopatterned areas.

Fig. 25.21. Fluorescence images of protein patterns formed by DPN. Structures of antirabbit IgG (labeled with Alexa 594) written on a negatively charged SiO2 surface using DPN. (a) Letters. (b) Dot arrays, contact time 5 s. (c) Dot arrays, contact time = 3 s. (d) Using an aldehyde-derivatized SiO2 surface, dot arrays of two proteins were written by DPN. Reproduced from [166] with permission

25.5.2 Applying Bias-Induced SPL for Protein Nanopatterns Little work has been done using biased induced SPL for protein patterning by SPL, although piezo-resistive patterning reported by Bonnell and co-workers shows promise for patterning of proteins on PZT materials such as SrTiO3 [168]. Bias-induced SPL was applied to oxidize the headgroups of monolayers of α-hepta(ethylene glycol)

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

129

methyl ω-undecenyl ether on Si(111) substrates by Gu et al. [15]. The nanopatterned templates were then used to attach avidin and biotin-bovine serum albumin in phosphate buffered saline. 25.5.3 Protein Immobilization on SAMs Generated by Force-Induced SPL In the initial investigations of protein immobilization on nanografted SAMs, WaduMesthrige et al. used ω-functionalized alkanethiol SAMs to mediate electrostatic and covalent binding of IgG and lysozyme [153]. The reactivity and stability of protein nanopatterns was investigated in subsequent reports, as well as the retention of speciﬁc binding activity of the immobilized proteins for antibodies [17, 169]. Protein patterns sustained washing with buffer and surfactant solutions, and were stable for at least 40 hours of AFM imaging. The smallest feature yet produced by SPL is a 2 × 4 nm2 nanografted dot (of 32 thiol molecules) and a 10 × 150 nm2 line containing three proteins. A key advantage of the nanografting protocol is the capability to conduct experiments in situ, viewing the successive changes in surface topography after the steps of nanopatterning SAMs, rinsing, introducing buffers and proteins. The patterns are not dried in air, and remain in a carefully controlled environment. Nanografting has been applied by several investigators to write nanopatterns for protein immobilization. Three differently charged linkers were nanografted within a hexa(ethylene glycol) functionalized alkanethiol resist SAM by Zhou et al. [170]. The adsorption of lysozyme, rabbit IgG and bovine carbonic anhydrase (II) onto the nanografted patterns was studied in situ at a variety of pH values. Using forceinduced SPL methods of nanografting and nanoshaving, Kenseth et al. [171] fabricated nanopatterns of IgG using a matrix of (undec-11-mercapto-1-yl) triethylene glycol methyl ether. Nanostructured templates prepared by force-induced SPL and DPN were demonstrated by Cheung et al. using amine-terminated alkanethiols as linkers for attaching the cow pea mosaic virus [172]. Jang et al. applied nanografting to immobilize enzymes such as acetylcholinase esterase on adsorbed aldehyde molecules, and then tested the catalytic activity for the immobilized enzymes [173]. Examples of nanografted patterns used for electrostatic immobilization of lysozymes are shown in Fig. 25.22. Five nanopatterns (200 × 200 nm2 ) of 16MHA were grafted into a glycol-terminated SAM matrix (Fig. 25.22a). After rinsing and 30 min incubation in a solution of lysozyme (0.1 mg/mL), the height of the nanopatterns increased with the adsorption of protein, with high selectivity for the carboxylate-terminated nanopattern (Fig. 25.22b). The cursor proﬁle (Fig. 25.22c) displays the difference in height of the nanopatterns before and after protein adsorption. Lysozyme is considered to be ellipsoidal, with dimensions approximately 4.5 × 3 × 3 nm3 from X-ray crystallographic studies [174], which correlates well with the measured change in pattern heights ranging from 3–4 nm. Nanografting was applied to directly pattern designed metalloproteins via chemisorption onto gold by Case et al. [175]. The bundle protein structure was designed to present the C-termini of three helices, terminated with D-cysteine residues for assembly in a vertical orientation, normal to the Au(111) substrate.

130

J.C. Garno · J.D. Batteas

Fig. 25.22. In situ immobilization of lysozyme on 16-MHA nanopatterns. (a) T-shaped arrangement of ﬁve nanopatterns (200×200 nm2 ) of 16-MHA written within a glycol-terminated matrix SAM. (b) Same nanopatterns after protein adsorption. (c) Cursor proﬁle across the three central patterns indicate the height changes after protein adsorption. Courtesy of Joonyeong Kim, NIST

New nanoscale studies with proteins facilitated by SPL will aid the development of new and better approaches for immobilization and bioconjugation chemistries, which are key technologies used in manufacturing biochip and biosensing surfaces. Nanoscale studies can be applied to reﬁne critical parameters used to link and organize proteins on surfaces of biochips and biosensors. Scientiﬁc developments in this ﬁeld may lead to new technologies in areas such as bioinformatics, medical diagnostics, and drug discovery.

25.6 Conclusions and Outlook The ability to study and control processes on the nanometer scale is of interest in both fundamental and applied research. Engineered surfaces offer molecular-level control of the spacing and composition of nanopatterned elements, for anchoring and then

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

131

reacting with molecules in solution. Nanopatterned arrays of SAMs can be fabricated using SPL with precise control over chemical functionality, shape, dimension, and spacing on the nanometer length scale. Using wet chemical approaches to nanofabrication provides new methods that are amenable to a range of materials. Importantly, these approaches offer the possibility of precisely positioning biomolecules due to the ability to pattern materials under ambient or controlled physiological conditions. At present, SPL is used as a research tool in laboratories instead of as a manufacturing tool for high throughput applications. This chapter provides an overview of the many possible extensions of SPL for the design and construction of nanostructures. Nanomaterials such as proteins, nanoparticles and SAMs have strong commercial potential, due to size-dependent properties [176–178]. These materials can be applied towards the development of a new generation of chemical and biosensors, biochips, and molecular electronic devices. For smaller, faster and cheaper computers, the trend in manufacture for the next generation of computing and electronic devices is to push towards nanometer-size features. Nanoscale processes in manufacturing are expected to have an even greater impact than present-day microfabrication technologies [179]. At present, AFM-based lithography with SAMs offers a new platform for directly investigating changes that occur on surfaces during chemical reactions, as a tool for manipulating surface terminal groups at local scale. Chemical self-assembly is an attractive method for reversibly constructing supramolecular systems with welldeﬁned properties. We anticipate new directions in applications such as biosensing, biomimetic surfaces for drug delivery and molecule-based electronics. Acknowledgements. The authors gratefully acknowledge support from the NIST Competence Building Program in Molecular Electronics and the Advanced Technology Program. JCG acknowledges support from a National Research Council Postdoctoral Fellowship and Louisiana State University College of Basic Sciences startup funds.

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

Cavallini M, Biscarini F, Leon S, Zerbetto F, Bottari G, Leigh DA (2003) Science 299:531 Eigler DM, Schweizer EK (1990) Nature 344:524 Zeppenfeld P, Lutz CP, Eigler DM (1992) Ultramicroscopy 42–44:128 Crommie MF, Lutz CP, Eigler DM (1993) Science 262:218 Jung TA, Schlittler RR, Gimzewski JK, Tang H, Joachim C (1996) Science 271:181 Avouris P (1995) Acc Chem Res 28:95 Lyo I, Avouris P (1991) Science 253:173 Salling CT, Lagally MG (1994) Science 265:50 Stipe BC, Rezaei MA, Ho W, Gao S, Persson M, Lundqvist BI (1997) Phys Rev Lett 78:4410 Liu J-F, Cruchon-Dupeyrat S, Garno JC, Frommer J, Liu G-Y (2002) Nano Lett 2:937 Garno JC, Yang Y, Amro NA, Cruchon-Dupeyrat S, Chen S, Liu G-Y (2003) Nano Lett 3:389 Zhang H, Mirkin CA (2004) Chem Mater 16:1480 Liu S, Maoz R, Sagiv J (2004) Nano Lett 4:845 Li Q, Zheng J, Liu Z (2003) Langmuir 19:166 Gu J, Yam CM, Li S, Cai C (2004) J Am Chem Soc 126:8098

132

J.C. Garno · J.D. Batteas

16. 17. 18. 19. 20.

Lee K-B, Lim J-H, Mirkin CA (2003) J Am Chem Soc 125:5588 Wadu-Mesthrige K, Amro NA, Garno JC, Xu S, Liu G-Y (2001) Biophys J 80:1891 Amro NA, Liu G-Y (2002) Proc Natl Acad Sci USA 99:5165 Hyun J, Ahn SJ, Lee WK, Chilkoti A, Zauscher S (2002) Nano Lett 2:1203 Dimilla PA, Folkers JP, Biebuyck HA, Harter R, Lopez GP, Whitesides GM (1994) J Am Chem Soc 116:2225 Crooks RM, Ricco AJ (1998) Acc Chem Res 31:219 Flink S, van Veggel F, Reinhoudt DN (2000) Adv Mater 12:1315 Willner I, Katz E (2000) Angew Chem Int Ed 39:1180 Dubois LH, Nuzzo RG (1992) Annu Rev Phys Chem 43:437 Chailapakul O, Sun L, Xu C, Crooks RM (1993) J Am Chem Soc 115:12459 Whitesides GM, Laibinis PE (1990) Langmuir 6:87 Laibinis PE, Whitesides GM (1992) J Am Chem Soc 114:9022 Porter MD, Bright TB, Allara DL, Chidsey CED (1987) J Am Chem Soc 109:3559 Bain CD, Troughton EB, Yao T-T, Evall J, Whitesides GM (1989) J Am Chem Soc 111:321 Troughton EB, Bain CD, Whitesides GM, Nuzzo RG, Allara DL, Porter MD (1988) Langmuir 4:365 Chidsey CED, Loiacono DN (1990) Langmuir 6:682 Nuzzo RG, Dubois LH, Allara DL (1990) J Am Chem Soc 112:558 Xia YN, Whitesides GM (1998) Annu Rev Mater Sci 28:153 Xia YN, Rogers JA, Paul KE, Whitesides GM (1999) Chem Rev 99:1823 Liu G-Y, Xu S, Qian Y (2000) Acc Chem Res 33:457 Tour JM (2000) Acc Chem Res 33:791 Joachim C, Gimzewski JK, Aviram A (2000) Nature 408:541 Binnig G, Rohrer H, Gerber C (1982) Phys Rev Lett 49:57 Binnig G, Quate CF, Gerber C (1986) Phys Rev Lett 56:930 Xu S, Miller S, Laibinis PE, Liu G-Y (1999) Langmuir 15:7244 Liu G-Y, Xu S, Qian Y (2000) Acc Chem Res 33:457 Chen CJ (1993) Introduction to Scanning Tunneling Microscopy. Oxford University Press, New York Sarid D (1991) Scanning Force Microscopy, with Applications to Electric, Magnetic and Atomic Forces. Oxford University Press, New York Frommer J (1992) Angew Chem Int Ed Engl 31:1298 Hamers RJ (1996) J Phys Chem 100:13103 Louder DR, Parkinson BA (1995) Anal Chem 67:297A Bottomley LA, Coury JE, First PN (1996) Anal Chem 68:185R Marti O, Ribi HO, Drake B, Albrecht TR, Quate CF, Hansma PK (1988) Science 239:50 Burnham NA, Gremaud G, Kulik AJ, Gallo P-J, Oulevey F (1996) J Vac Sci Technol B 14:1308 Vinckier A, Semenza G (1998) FEBS Lett 430:12 Kiridena W, Jain V, Kuo PK, Liu G-Y (1997) Surf Interface Anal 25:383 Jourdan JS, Cruchon-Dupeyrat SJ, Huang Y, Kuo PK, Liu G-Y (1999) Langmuir 15:6495 Xu S, Laibinis PE, Liu G-Y (1998) J Am Chem Soc 120:9356 Gorman CB, Carroll RL, He Y, Tian F, Fuierer R (2000) Langmuir 16:6312 Liu S, Maoz R, Sagiv J (2004) Nano Lett 4:845 Piner RD, Zhu J, Xu F, Hong SH, Mirkin CA (1999) Science 283:661 Nyffenegger RM, Penner RM (1997) Chem Rev 97:1195 Kramer S, Fuierer RR, Gorman CB (2003) Chem Rev 103:4367 McCarty GS, Weiss PS (1999) Chem Rev 99:1983 Smith RK, Lewis PA, Weiss PS (2004) Prog Surf Sci 75:1

21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography 61. 62. 63. 64. 65. 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.

133

Wouters D, Schubert US (2004) Angew Chem Int Ed 43:2480 Lewis MS, Gorman CB (2004) J Phys Chem B 108:8581 Li Y, Maynor BW, Liu J (2001) J Am Chem Soc 123:2105 Brower TL, Garno JC, Ulman A, Liu G-Y, Yan C, Golzhauser A, Grunze M (2002) Langmuir 18:6207 Jourdan JS, Cruchon-Dupeyrat SJ, Huan Y, Kuo PK, Liu GY (1999) Langmuir 15:6495 Kiredena W, Jain V, Kuo P, Liu G-Y (1997) Surf Interface Anal 25:383 Xiao X-D, Liu GY, Charych DH, Salmeron M (1995) Langmuir 11:1600 Milic T, Garno JC, Smeureanu G, Batteas JD, Drain CM (2004) Langmuir 20:3974 Xu S, Liu GY (1997) Langmuir 13:127 Amro NA, Xu S, Liu G-Y (2000) Langmuir 16:3006 Hacker C, Zangmeister C, Garno JC, Batteas JD, Richter LA, van Zee R (2004) Langmuir 20:6195 Hong S, Zhu J, Mirkin CA (1999) Langmuir 15:7897 Hong S, Mirkin CA (2000) Science 288:1808 Mirkin CA, Hong S, Demers L (2001) Chem Phys Chem 2:37 Piner RD, Mirkin CA (1997) Langmuir 13:6864 Sheehan PE, Whitman LJ (2002) Phys Rev Lett 88:1561041 Zhang M, Bullen D, Chung S-W, Hong S, Ryu KS, Fan Z, Mirkin CA, Liu C (2002) Nanotechnology 13:212 Hong S, Mirkin CA (1999) Science 286:523 Ginger DS, Zhang H, Mirkin CA (2004) Angew Chem Int Ed 43:30 Minne SC, Adams JD, Yaralioglu G, Manalis SR, Atalar A, Quate CF (1998) Appl Phys Lett 73:1742 Cruchon-Dupeyrat S, Porthun S, Liu G-Y (2001) Appl Surf Sci 175–176:636 Poirier GE (1997) Chem Rev 97:1117 Dubois LH, Nuzzo RG (1992) Ann Rev Phys Chem 43:437 Schreiber F (2000) Prog Surf Sci 65:151 Ulman A (1991) An Introduction to Ultrathin Organic Films from Langmuir–Blodgett to Self-Assembly. Academic Press, New York Ulman A (1996) Chem Rev 96:1533 Fenter P, Eberhardt A, Eisenberger P (1994) Science 266:1216 Fenter P, Eberhardt A, Liang KS, Eisenberger P (1997) J Chem Phys 106:1600 Schreiber F (2004) J Phys: Condens Matter 16:R881 Nuzzo RG, Korenic EM, Dubois LH (1990) J Chem Phys 93:767 Himmelhaus M, Gauss I, Buck M, Eisert F, Woll C, Grunze M (1998) J Electron Spectrosc Relat Phenom 92:139 Dannenberger O, Weiss K, Himmel HJ, Jager B, Buck M, Woll C (1997) Thin Solid Films 307:183 Delamarche E, Michel B, Biebuyck HA, Gerber C (1996) Adv Mater 8:719 Ivanisevic A, Mirkin CA (2001) J Am Chem Soc 123:7887 Jung H, Kulkarni R, Collier CP (2003) J Am Chem Soc 125:12096 Tilman N, Ulman A, Schildkraut JS, Penner TL (1988) J Am Chem Soc 111:6136 Peanasky J, Schneider HM, Granick S (1995) Langmuir 11:953 Schwartz DK, Steinberg S, Israelachvili J, Zasadzinski JAN (1992) Phys Rev Lett 63:3354 Kadalbajoo M, Suda H, Opdahl A, Garno JC, Batteas JD, Tarlov MJ, DeShong P (2005) to be submitted to Langmuir Pena DJ, Raphael MP, Byers JM (2003) Langmuir 19:9028 Wang X, Ryu KS, Bullen DA, Zou J, Zhang H, Mirkin CA, Liu C (2003) Langmuir 19:8951 Rozhok S, Piner R, Mirkin CA (2003) J Phys Chem B 107:751 Ivanisevic A, McCumber KV, Mirkin CA (2002) J Am Chem Soc 124:11997

134

J.C. Garno · J.D. Batteas

104. 105. 106. 107.

Zhang H, Chung S-W, Mirkin CA (2003) Nano Lett 3:43 Weeks BL, Noy A, Miller AE, DeYoreo JJ (2002) Phys Rev Lett 88:2555051 Peterson EJ, Weeks BL, DeYoreo JJ, Schwartz PV (2004) J Phys Chem B 108:15206 Ivanisevic A, Im J-H, Lee K-B, Park S-J, Demers LM, Watson KJ, Mirkin CA (2001) J Am Chem Soc 123:12424 Chen J, Reed MA, Asplund CL, Cassell AM, Myrick ML, Rawlett AM, Tour JM, Van Patten PG (1999) Appl Phys Lett 75:624 Maoz R, Cohen SR, Sagiv J (1999) Adv Mater 11:55 Gorman CB, Carroll RL, Fuierer RR (2001) Langmuir 17:6923 Credo GM, Boal AK, Das K, Galow TH, Rotello VM, Feldheim DL, Gorman CB (2002) J Am Chem Soc 124:9036 Liu X, Guo S, Mirkin CA (2003) Angew Chem Int Ed 42:4785 Xu P, Kaplan D (2004) Adv Mater 16:628 Su M, Aslam M, Fu L, Wu N, Dravid VP (2004) Appl Phys Lett 84:4200 Zhang H, Elghanian R, Amro NA, Disawal S, Eby R (2004) Nano Lett 4:1649 Lim J-H, Mirkin CA (2002) Adv Mater 14:1474 Maynor BW, Filocamo SF, Grinstaff MW, Liu J (2002) J Am Chem Soc 124:522 Noy A, Miller AE, Klare JE, Weeks BL, Woods BW, DeYoreo JJ (2002) Nano Lett 2:109 Jang S-Y, Marquez M, Sotzing GA (2004) J Am Chem Soc 126:9476 Kaholek M, Lee WK, LaMattina B, Caster KC, Zauscher S (2004) Nano Lett 4:373 Nitzan A, Ratner MA (2003) Science 300:1384 Cahen D, Hodes G (2002) Adv Mater 14:789 Rawlett AM, Hopson TJ, Nagahara LA, Tsui RK, Ramachandran GK, Lindsay SM (2002) Appl Phys Lett 81:3043 Kagan CR, Ratner MA (2004) MRS Bull 29:376 Liu S, Maoz R, Schmid G, Sagiv J (2002) Nano Lett 2:1055 Barsotti RJ, O’Connell MS, Stellacci F (2004) Langmuir 20:4795 Templeton AC, Hostetler MJ, Warmoth EK, Chen S, Hartshorne CM, Krishnamurthy VM, Forbes MDE, Murray RW (1998) J Am Chem Soc 120:4548 Mbindyo JKN, Mallouk TE, Mattzela JB, Kratochvilova I, Razavi B, Jackson TN, Mayer TS (2002) J Am Chem Soc 124:4020 American Electroplaters & Surface Finishers Society (1990) Electroless Plating: Fundamentals and Applications, Orlando, FL Zangmeister CD, van Zee RD (2003) Langmuir 19:8065 Ho PK-H, Filas RW, Abusch-Magder D, Bao Z (2002) Langmuir 18:9625 Dressic WJ, Dulcey CS, Georger JH, Calabrese GS, Calvert JM (1994) J Electrochem Soc 141:210 O’Brien JC, Jones VW, Porter MD (2000) Anal Chem 72:703 Delamarche E, Sundarababu G, Biebuyck H, Michel B, Gerber C, Sigrist H, Wolf H, Ringsdorf H, Xanthopoulos N, Mathieu HJ (1996) Langmuir 12:1997 Rowe CA, Tender LM, Feldstein MJ, Golden JP, Scruggs SB, MacCraith BD, Cras JJ, Ligler FS (1999) Anal Chem 71:3846 Jones VW, Kenseth JR, Porter MD (1998) Anal Chem 70:1233 Blawas AS, Reichert WM (1998) Biomaterials 19:595 Scouten WH, Luong JH, Brown RS (1995) Trends Biotechnol 13:178 Zhang S, Yan L, Altman M, Lassle M, Nugent H, Frankel F, Lauffenburger DA, Whitesides GM, Rich A (1999) Biomaterials 20:1213 Ravenscroft MS, Bateman KE, Shaffer KM, Schessler HM, Jung DR, Schneider TW, Montgomery CB, Custer TL, Schaffner AE, Liu QY, Li YX, Barker JL, Hickman JJ (1998) J Am Chem Soc 120:12169 Bernard A, Delamarche E, Schmid H, Michel B, Bosshard HR, Biebuyck H (1998) Langmuir 14:2225

108. 109. 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.

25 Nanofabrication with Self-Assembled Monolayers by Scanning Probe Lithography

135

142. James CD, Davis RC, Kam L, Craighead HG, Isaacson M, Turner JN, Shain W (1998) Langmuir 14:741 143. Kane RS, Takayama S, Ostuni E, Ingber DE, Whitesides GM (1999) Biomaterials 20:2363 144. Yang Z, Chilkoti A (2000) Adv Mater 12:413 145. Bernard A, Renault JP, Michel B, Bosshard HR, Delamarche E (2000) Adv Mater 12:1067 146. Lahiri J, Ostuni E, Whitesides GM (1999) Langmuir 15:2055 147. Blawas AS, Oliver TF, Pirrung MC, Reichert WM (1998) Langmuir 14:4243 148. Nicolau DV, Taguchi T, Taniguchi H, Yoshikawa S (1998) Langmuir 14:1927 149. Dontha N, Nowall WB, Kuhr WG (1997) Anal Chem 69:2619 150. Bergman AA, Buijs J, Herbig J, Mathes DT, Demarest JJ, Wilson CD, Reimann CT, Baragiola RA, Hull R, Oscarsson SO (1998) Langmuir 14:6785 151. Delamarche E, Bernard A, Schmid H, Bietsch A, Michel B, Biebuyck H (1998) J Am Chem Soc 120:500 152. Patel N, Sanders GHW, Shakesheff KM, Cannizzaro SM, Davies MC, Langer R, Roberts CJ, Tendler SJB, Williams PM (1999) Langmuir 15:7252 153. Wadu-Mesthrige K, Xu S, Amro N, Liu G-Y (1999) Langmuir 15:8580 154. Marvin JS, Corcoran EE, Hattangadi NA, Zhang JV, Gere SA, Hellinga HW (1997) Proc Natl Acad Sci USA 94:4366 155. Eggers M, Hogan M, Reich RK, Lamture J, Ehrlich D, Hollis M, Kosicki B, Powdrill T, Beattie K, Smith S, Varma R, Gangadharan R, Mallik A, Burke B, Wallace D (1994) BioTechniques 17:516 156. Bergveld P (1996) Sens Actuators A 56:65 157. Polzius R, Schneider T, Bier FF, Bilitewski U, Koschinski W (1996) Biosens Bioelectron 11:503 158. Kunz RE (1997) Sens Actuators, B 38–39:13 159. Lee K-B, Kim E-Y, Mirkin CA, Wolinisky SM (2004) Nano Lett 4:1869 160. Wilson DS, Nock S (2001) Curr Opin Chem Biol 6:81 161. Walter G, Bussow K, Lueking A, Glokler J (2002) Trends in Molecular Medicine 8:250 162. Templin MF, Stoll D, Schrenk M, Traub PC, Vohringer CF, Joos TO (2002) Trends Biotechnol 20:160 163. Huang R-P (2001) J Immun Methods 255:1 164. Schweitzer B, Kingsmore SF (2002) Curr Opinion Chem Biotechnol 13:14 165. Lee K-B, Park S-J, Mirkin CA, Smith JC, Mrksich M (2002) Science 295:1702 166. Lim J-H, Ginger DS, Lee K-B, Heo J, Nam J-M, Mirkin CA (2003) Angew Chem Int Ed 42:2309 167. Hyun J, Kim J, Craig SL, Chilkoti A (2004) J Am Chem Soc 126:4770 168. Kalinin SV, Bonnell DA, Alvarez T, Lei X, Hu Z, Shao R, Ferris JH (2004) Adv Mater 16:795 169. Liu G-Y, Amro NA (2002) Proc Natl Acad Sci 99:5165 170. Zhou D, Wang X, Birch L, Rayment T, Abell C (2003) Langmuir 19:10557 171. Kenseth JR, Harnish JA, Jones VW, Porter MD (2001) Langmuir 17:4105 172. Cheung CL, Camarero JA, Woods BW, Lin T, Johnson JE, DeYoreo JJ (2003) J Am Chem Soc 125:6848 173. Jang C-H, Stevens BD, Phillips R, Calter MA, Ducker WA (2003) Nano Lett 3:691 174. Blake CCF, Koenig DF, Mair GA, North ACT, Phillips DC, Sarma VR (1965) Nature 206:757 175. Case MA, McLendon GL, Hu Y, Vanderlick TK, Scoles G (2003) Nano Lett 3:425 176. Martin CR, Mitchell DT (1998) Anal Chem: 70:322A–327A 177. Whetten RL, Shaﬁgullin MN, Khoury JT, Schaaff TG, Vezmar I, Alvarez MM, Wilkinson A (1999) Acc Chem Res 32:397 178. Wang ZL (1998) Adv Mater 10:513 179. Salling CT (1996) J Vac Sci Technol B 14:1322

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography Marta Tello · Fernando García · Ricardo García

26.1 Introduction to AFM Nanolithographies Shortly after the invention and development of the scanning tunneling microscope (STM) and the atomic force microscope (AFM) [1, 2], it was demonstrated that these microscopes could be used to modify surfaces at atomic and nanometer scales [3–6]. Since then, different methods have been proposed that use either AFM or STM as a manipulation tool to achieve nanometer-scale feature size. However, the use of STM as a lithographic tool is rather limited because of either ultra-high-vacuum operation or sample conductivity requirements. AFM on the other hand, as shown in the present book, provides atomic resolution images in vacuum, air or liquid environments. Furthermore, it can image conductive and insulating samples alike. In this chapter we describe some fundamental and applied aspects of so-called local oxidation nanolithography. This method is based on the spatial conﬁnement of an oxidation reaction underneath an AFM tip. However, before introducing the details of the method and because of the large variety of scanning probe-based modiﬁcation methods [7–10], we consider it pertinent to provide a brief overview of some of the more established AFM-based modiﬁcation methods. The most straightforward way to use AFM to modify a surface is by producing mechanical indentations on a substrate. This is achieved by pressing the tip of the AFM towards the sample and exerting forces usually in the micronewton range [11]. In this way, feature sizes of 10 nm diameter have been produced in a photoresist layer deposited over GaAs–AlGaAs heterostructures [12]. After the indentation, the pattern was transferred to the GaAs substrate by selective wet etching. Rugar et al. combined mechanical pressure with heat to generate indentations on a resist ﬁlm (polymethyl metacrylate (PMMA)) [13]. The tip was heated by using a laser. By varying the laser pulse and the mechanical force exerted by the tip on the PMMA ﬁlm, holes of different sizes were obtained. Binnig et al. were able to produce holes or pits of about 40 nm in width by using ultrathin PMMA layers deposited onto a hard substrate [14]. They also demonstrated that thermomechanical writing was compatible with parallel operation by using two-dimensional arrays of cantilevers. This method allowed them to achieve storage densities of 500 Gb/in2 . An interesting feature of thermomechanical writing is that the modiﬁcations are erasable. Heating either the whole substrate or locally by passing a hot tip over a speciﬁc region allows one to erase the information.

138

Marta Tello · F. García · R. García

Another modiﬁcation method that uses mechanical forces to modify a surface is nanografting or nanoshaving of molecules [15, 16]. In this method, a rigid substrate is functionalised with a self-assembled monolayer (SAM). Then, the AFM tip is pressed against the sample. Above a certain force threshold, the tip is able to remove the molecules leaving the bare substrate (nanoshaving). This region could be covered again by molecules by subsequent immersion in a solution containing a different type of self-assembling molecules. Then, the new molecules adsorb on the free spaces (nanografting). This leaves the sample with two different zones. These two regions can be distinguished by topographical contrast or by friction force images. Feature sizes of about 10 nm have been claimed [15]. Mass transport from the AFM tip to the sample surface has also been used to generate patterns. In this method, the regular silicon tip of the AFM is covered with a thin gold layer. Under certain conditions, the application of a voltage pulse between the metallic tip and the sample produces the formation of a gold nanocluster on the substrate [17–21]. Mamin and co-workers suggested that the deposition process was ﬁeld-assisted [17], although other authors suggested a current induced process [18]. By operating AFM in the non-contact mode, gold nanowires of 40 nm in width were fabricated [21]. The nanowires showed about one order of magnitude higher resistivities than the bulk gold due to grain and surface scattering effects. Dip pen nanolithography (DPN) uses the AFM tip to selectively deposit selfassembling molecules. The tip is dipped into a solution of self-assembling molecules. When the tip is brought into contact with the substrate, a liquid meniscus is spontaneously created between them [22]. The molecules diffuse within the meniscus towards the surface where they form a local self-assembled monolayer [23–25]. It is possible to deposit different kinds of molecules on the substrate by changing the solvent; this is called multiple ink DPN. Feature sizes of 10 nm have been claimed [26]. The rest of the chapter is organized as follows. First, we introduce some of the fundamental aspects of the local oxidation process. This is followed by a section discussing the mechanism and the kinetics of the oxidation process. A discussion on the minimum feature size follows. Finally, we describe the fabrication and performance of several types of data storage, electronic and mechanical devices fabricated by local oxidation nanolithograpphy.

26.2 Basic Local Oxidation Aspects In 1990, Dagata et al. showed that the application of a bias voltage between an STM tip and a silicon surface modiﬁed the surface. The modiﬁcations protruded from the bare substrate and coincided with the regions scanned by the tip. Further spectroscopy analysis revealed the presence of oxygen in the protrusions [27, 28]. Similar results were obtained by operating STM in a liquid cell and with a tantalum substrate [29]. Soon afterwards, AFM was applied to locally oxidize silicon [30] and since then it has been the instrument of choice to perform local oxidation experiments [31–33]. In local oxidation, a voltage pulse is applied between a tip (negative electrode) and the substrate (positive electrode). The instrument is placed in a chamber were there

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

139

is a water vapour pressure of 30–50% (relative humidity). The voltage pulse in the presence of water induces the formation of the sample oxide. The process of local oxidation can be compared to conventional anodic oxidation. The tip of the AFM acts as the cathode of the electrochemical cell, while the sample is the anode where the oxidation takes place, and the water meniscus formed between tip and sample provides the oxyanion species. Figure 26.1 shows a standard experimental set-up for local oxidation. The AFM is kept inside a chamber with inlets for dry and water-saturated nitrogen to allow the control of the relative humidity of the environment. Dry N2 is bubbled into an Erlenmeyer containing water. The N2 exits saturated with water vapour and is introduced into the chamber where the AFM is kept. In this way, it is possible to control the relative humidity of the environment, which is a key parameter in the local oxidation process. Local oxidation experiments can be performed with both contact and non-contact AFMs. The ultimate feature size deﬁnition is achieved in the non-contact AFM operation (see the next section) [34]. Figure 26.2 shows the scheme of the local oxidation process in the non-contact amplitude modulation mode of AFM [35]. Figure 26.2a shows the AFM tip oscillating at a certain amplitude in close proximity to the sample. The humidity in the chamber is kept in the 30–40% range, which assures the presence of a few monolayers of water adsorbed on the sample surface [22]. The application of a voltage pulse (above a certain threshold value) induces the formation of a water meniscus joining the tip and sample [36–38], as shown in Fig. 26.2b. In contact AFM oxidation, the proximity between the tip and the sample, and the nanometer scale character of the

Fig. 26.1. Standard experimental set-up for local oxidation nanolithography. The AFM is kept inside a chamber with inlets for dry and water saturated N2 to control the relative humidity

140

Marta Tello · F. García · R. García

Fig. 26.2. (a–d) Scheme for the local oxidation process in the non-contact AM-AFM mode. A voltage pulse is applied between the AFM tip and the sample. In the presence of water, the application of the pulse induces the formation of a water meniscus that supplies the oxyanions needed for the oxidation and conﬁnes the reaction within its limits. (e) Example of a local oxidation nanolithography pattern

interface allows the spontaneous formation of a water meniscus [22]. The meniscus has a double role; on the one hand it provides the oxyanions needed to create the oxide and, on the other, it conﬁnes the reaction within its limits [39] (Fig. 26.2c). The chemical nature of the modiﬁcations has been identiﬁed by several spectroscopy techniques such as Auger, SIMS and XPS, and on several materials such as Si and GaAs [27,40,41]. Regarding local silicon oxides, HF etching and TEM imaging have shown that the oxide grows both above and below the initial surface. Approximately 40% of the total thickness of the oxides remains below the sample surface [4, 42]. The size of the local oxides is controlled by both the voltage pulse strength and the duration (see the next section) (Fig. 26.2d). An example of the writing possibilities of local oxidation on silicon is shown in Fig. 26.2e. The width of the letters is about 100 nm. The ﬁrst surface modiﬁed by local oxidation was a hydrogen-passivated Si(111) surface [27]. Soon afterwards, other passivated silicon surfaces such as Si(100) and Si(110), as well as silicon with a native oxide layer were also modiﬁed. In addition to silicon, many other semi-conducting and metallic materials have been

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

141

locally oxidized, such as GaAs, InAs and InP [40–44], Nb [45], Ti [46, 47], Al [48], Cr [49], Ta [29], TiN [50], SiC [51], Si3 N4 [52], SrTiO3 [53] or YBa2 Cu3 O7−y [54]. Furthermore, Sagiv and co-workers have demonstrated the potential of this technique to oxidize self-assembled monolayers by converting methyl terminated groups of an OTS monolayer into carboxyl terminated groups [55–57]. The above experiments were performed with the standard local oxidation set up, i.e., a conductive AFM operated in an environment that allows either the spontaneous or the ﬁeld-induced formation of a water meniscus. However, local oxidation experiments have been performed with different conditions. For example, Tonneau and co-workers have demonstrated the oxidation under an ozone enriched atmosphere [58]. This shows that the role of the water meniscus was to supply the oxyanions. Local oxidation has also been performed with a scanning electron microscope [59]. There, the electron beam promoted the dissociation of residual water vapour from the chamber background and the subsequent oxidation of the region exposed to the electron beam. Furthermore, we have also demonstrated that local modiﬁcations could be achieved by using liquids other than water, such as ethyl and isopropyl alcohols [60]. A substantial enhancement of the aspect-ratio of the motifs formed on an Si(100) has been reported, although spatially resolved photoelectron spectroscopy has revealed that in the above case, the structures are no longer silicon oxides [61].

26.3 Mechanism and Kinetics The study of the local oxidation mechanisms and kinetics has been the focus of an intense activity [34, 36, 62–71], although it can be said that a kinetic model that incorporates all the relevant factors and gives a comprehensive explanation of the experimental data is still missing. Nonetheless, the dominant factors inﬂuencing the oxidation kinetics are well-established. These are the voltage pulse duration and strength, the relative humidity, the doping of the substrate, the tip–sample distance and the cantilever force constant. The tip-water meniscus-substrate interface can be seen as an electrochemical cell where an oxidation-reaction process takes place. The proposed reactions in this system for a substrate M are [62]: M + nH2 O → MOn + 2nH+ + 2n e− , −

(26.1) −

M + 2nH2 O + 2n e → nH2 + 2n OH + M . n

(26.2)

The oxide MOn is created in the sample-anode and in the tip-cathode H2 is liberated by the reaction: 2H+ + 2 e− → H2 .

(26.3)

The experimental study of the dependence of the oxide height and width on the voltage pulse duration and strength allows one to derive the kinetics of the oxidation reaction. Figure 26.3 shows a semi-logarithmic plot of the height (a) and width (b)

Marta Tello · F. García · R. García

142 Fig. 26.3. (a) Height dependence with the oxidation time at different voltages. (b) Width dependence with the oxidation time at different voltages

of the oxides versus the oxidation time at different voltages (8, 10, 12, 16, 20 and 24 V) [70]. The data can be adjusted by the following ﬁts [64, 65, 70]: h = h 0 (V ) + h 1 (V ) ln t , w = w0 (V ) + w1 (V ) ln t ,

(26.4) (26.5)

where h is the height of the oxides, w is the width, V is the applied voltage, t is the oxidation time, and h 0 , h 1 , w0 and w1 are voltage-dependent parameters. The results also show an almost linear dependence of the height and width of the oxides with the applied voltage. The dependence of the height and width of the oxides with the oxidation time indicates a process with a strong intial growth that rapidly saturates. Oxidation models of metals and semiconductors start with the ideas of Cabrera and Mott [72]. They proposed a diffusion model where the oxyanions have to overcome several diffusion barriers to pass from one interstitial site to the next. This diffusion model gives an inverse logarithmic relationship between the height of the oxides and the oxidation time: h1 h 1 ut , (26.6) = ln h h 2L where h 1 (V ) ∼ 10−5 −10−6 cm, h L is a critical thickness, u = u 0 exp(−W/kT ) ∼ 104 cm/s, and W is the diffusion barrier that the ions have to overcome. Although the Cabrera and Mott model gives expressions that relate the oxide thickness with the oxidation time in terms of material parameters, it does not explain the direct logarithmic dependence reported in local oxidation experiments. A direct log dependence

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

143

can be explained based on Uhlig’s model [73]. There, the oxidation of metal ﬁlms was explained by assuming that the rate controlling step in the reaction involves the interaction of the oxyanions with charged defects created within the oxide. The charged defects lead to the creation of a space charge region that diminishes the effective applied voltage and thus, the oxide growth rate. The following expression is derived from this model: h(t) = k (V ) log(ku t + 1) .

(26.7)

Dagata and coworkers realized that the substitution of t for t 0.4 provided a good ﬁt to the experiments for oxidation times t ≤ 500 ms, where the oxide growth rate decayed rapidly and the reaction saturated [68]. Other authors have proposed power laws to adjust the experimental data, such as [63, 69]: h ∼ V(t/t0 )γ ,

(26.8)

where γ ranges from 0.1 to 0.4. Dubois and Bubendorff derived a power law dependency by incorporating the diffusion process suggested by Cabrera and Mott with the presence of a space charge region [69]: h = hb

1/(δ+1) t , t0

(26.9)

where h b depends on the oxide concentration, t0 depends on the applied voltage and δ is a dimensionless parameter characteristic of the material. As we have seen, this relationship is in agreement with the experimental results. However, a single exponent does not ﬁt the entire time axis. Because of this, Dagata et al. proposed a model that considers two possible oxidation paths. In the ﬁrst one, the oxidation reaction takes place directly, i.e., the metal reacts with the oxyanions and creates the oxide at the surface [68]. This was called the transient regime and could be ﬁtted with a t 0.4 dependence. In the second path, the oxidation occurs through an intermediate process controlled by the built-up space charge. There, the oxyanions interact with the trapped charges before reaching the substrate and so the growth rate diminishes. This second part was adjusted by t 0.167 and provides a good ﬁt to the data for longer oxidation times.

26.4 Feature Size Local oxidation has generated a variety of very high resolution patterns. For example, García et al. have fabricated several periodic arrays of dots. In one case, an array composed of 4864 dots spaced 40 nm apart was generated. The dot size measured at the top ranged from 4 to 8 nm and at the base of the dot ranged from 20 to 27 nm [36]. Regular silicon tips were used to fabricate the above dots. Cooper and co-workers used carbon nanotube tips to generate an array of oxides with a spacing of 20 nm,

144

Marta Tello · F. García · R. García

which implies a feature width at the base below 20 nm [74]. Gotoh et al. also used carbon nanotube tips to fabricate parallel lines on a Ti surface with a full width at half maximum (FWHM) of about 7 nm [75]. Parallel arrays of lines have also been fabricated with regular silicon tips. Figure 26.4a shows an array of interdigitated lines fabricated by the application of a sequence of voltage pulses of 21 V and 1 ms. Each line is composed of 285 silicon oxide dots. A zoom of the region encircled by the white square in (a) can be seen in Fig. 26.4b. The separation of the lines, as revealed by the Fourier transform of this line, is 43 nm. Another example of local oxidation to pattern very high density arrays is shown in Fig. 26.5a. The lines are 13 nm apart, which is up to date with the writing of this chapter, the higher density record reached by local oxidation nanolithography. Proper measurement of feature sizes by AFM requires some further analysis. Direct lateral AFM measurements involve tip-sample convolution effects. For this reason, in many local oxidation experiments FWHM is usually provided. To determine precisely the geometry of local oxidation dots, we have studied the inﬂuence of the size and shape of the tip in the ﬁnal width. To that aim, we have fabricated local silicon oxide structures in the proximity of an object of known size and shape, in this case, sexithiophene monolayer (T6) islands [76]. Figure 26.6 shows a cross-section of two silicon oxide lines patterned in the proximity of a T6 island. These islands have a height of about 2.7 nm and can be considered to have vertical walls. The analysis of the AFM images shows that the silicon oxide motifs present a trapezoidal shape. The slope of the oxides (0.09) is smaller than the

Fig. 26.4. (a) Array of interdigitated lines fabricated by the applications of pulses of 21 V and 1 ms. (b) High resolution image of the white square in (a). (c) Cross-section along the white line in (b). The cross-section reveals a separation of 43 nm between the lines

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

145

Fig. 26.5. Array of parallel lines made by local oxidation nanolithography by applying pulses of 24.7 V and 80 µs. (b) The lines are spaced at 13 nm, as revealed by the averaged cross-section

slope of the T6 islands (0.23). Because the local oxide motifs and the T6 islands have a similar height, the above results imply that AFM images provide a faithful representation of the size and shape of the local oxide motifs. The measured slopes for the oxides imply a minimum feature size of 14 nm for a motif protuding 1 nm from the surface, although this does not preclude the fabrication of periodic arrays with feature sizes below 14 nm, as shown above. The trapezoidal geometry is fully consistent with transmission electron microscopy cross-sections of local anodic oxides [42].

Fig. 26.6. Cross-section along two silicon oxide lines fabricated in the proximity of a T6 island. The slopes of the motifs are smaller than the T6. This indicates that there are no tip-sample convolution effects while imaging the local oxides

146

Marta Tello · F. García · R. García

26.5 Applications I: Patterning, Data Storage and Template Growth Parallel arrays of lines or dots represent the most common patterns fabricated by local oxidation (Figs. 26.4–26.5, 26.7). By patterning arrays of both dots and parallel lines, Quate and co-workers demonstrated the potential to pattern 1 cm2 regions [77]. Nonetheless, other geometries are also possible. For example, in Fig. 26.8 an image of ﬁve concentric circles is shown. Each circle is silicon oxide that protrudes from the bare silicon substrate. The diameter of the inner circle is 50 nm. Gwo and coworkers used local oxidation to generate a two-dimensional array of hexagonal structures [78]. Pérez-Murano et al. and other groups have used local oxidation to pattern some key features of nanoelectromechanical systems, such as resonators or cantileverbased sensors. Metallic wires and elecrodes were patterned by Sagiv et al. in a slighly different way. They applied local oxidation nanolithography to modify directly a selfassembled monolayer. Instead of oxidizing the substrate, they oxidize the terminal group of the monolayer [55, 56, 79–81]. To fabricate the gold wires and electrodes they ﬁrst deposited a self-assembled monolayer of OTS (n-octadecyltrichlorosilane) and oxidized the desired structures with AFM by changing the methyl groups into carboxilic groups. An NTS (nonadecenyltrichlorosilane) treatment follows to attach another monolayer with terminal –CH=CH2 groups on the modiﬁed parts. By further chemical treatment, the ethylenic groups are converted into a thiol terminated layer. Finally Au55 clusters are attached to the thiol groups, leaving the gold pattern in the regions previously modiﬁed with AFM. In Fig. 26.9, some of the gold motifs can be seen in topographic (left) and phase (right) images. In Fig. 26.9e–f, a gold nanowire connected to an electrode has been patterned. The high density packing showed by some the above images suggested to use local oxidation for data storage. For example, in Fig. 26.10a, a matrix of 4684 dots has been patterned on a silicon substrate. A zoom in Fig. 26.10b of the white square of (a) reveals the homogeneity of the structures. The dots are 1 nm high, 20 nm wide and their separation is 40 nm, which implies areal densities of about 1 Tb/in2 . The

Fig. 26.7. (a) Array of parallel lines separated by 106 nm, the average width is 50 nm. (b) Zoom of the white square in (a). (c) Cross-section along the dashed line in (b). The height of the islands is 2.7 nm

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

147

Fig. 26.8. Image of ﬁve concentric circles patterned by local oxidation nanolithography. The diameter of the inner circle is 50 nm. The image demonstrates the reproducibility and accuracy of the technique

Fig. 26.9. Example of constructive nanolithography. (a–d) Topography (left) and phase (right) images of [Au55 ] wires. (e–f) Wires connected to contact pads. Image courtesy of [80]

π number written with an areal density of 1 Tb/in2 is shown in Fig. 26.11. π has been written in binary code with a precision of 20 decimals. The presence of one oxide dot can be interpreted as a binary “1”, while the absence of a dot stands for “0”. By using atomically ﬂat titanium substrates Cooper and coworkers reached areal densities of 1.6 Tb/in2 [74]. Their results are shown in Fig. 26.12, where a 500 nm × 500 nm area of oxide dots has been patterned in a titanium substrate. Despite these very high densities, writing by local oxidation is rather slow because of the sequential character of the AFM operation, which limits the technological impact of the above results.

148

Marta Tello · F. García · R. García

Fig. 26.10. (a) Array of 4864 dots fabricated by local oxidation nanolithography. (b) Zoom of the region enclosed by the white square in (a)

Fig. 26.11. π written in binary code with a precision of 20 decimals. Each dot represents a binary 1 and the absence of a dot represents 0. The width of the dots is 30 nm, which allows a storage density of 1 Tb/in2

Recently, several groups used local oxidation patterns as templates for the growth of a variety of materials, such as silicon [82], GaAs dots [83], proteins [84] or conjugated molecules [85]. Gwo and coworkers fabricated an array of dots oxidized by

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

149

AFM. After removing the native and anodic oxides with a plasma treatment, they deposited epitaxial silicon. The experiments showed that the silicon only grew in the regions modiﬁed by AFM [82, 86]. Sugimura and Nakagiri fabricated several stripes by local oxidation nanolithography and deposited metallic Au in the unpatterned regions [87]. Moreover, the selective bonding of silane layers in regions with OH− groups was used to bind SAMs in the zones modiﬁed by local oxidation nanolithography [88, 89]. For example, we have applied local oxide patterns to fabricate conjugated molecular tracks and wires made of sexithiophene (T6) molecules. Sexithiophene is a rigid molecule composed by six thiophene rings that in certain growth conditions on silicon surfaces creates dendritic shaped monolayer islands, such as the

Fig. 26.12. 500 nm × 500 nm area patterned by local oxidation nanolithography in a ﬂat titanium substrate. The storage density is 1.6 Tb/in2 . Image courtesy of [74]

Fig. 26.13. (a) Single T6 island on a bare silicon substrate. (b) Cross-section along the line marked by the arrow in (a). The height of the island is 2.6 nm, which corresponds to the length of the molecule

150

Marta Tello · F. García · R. García

one shown in Fig. 26.13. The height of the islands corresponds to the length of the molecules, ∼ 2.6 nm, which means that T6 has a vertical orientation in the substrate [90]. The goal is to introduce some anisotropy in the growth of T6 by using nanometer-size silicon oxide templates fabricated by local oxidation. Figures 26.14 and 26.15 illustrate the concept. A long track made of T6 molecules spans the entire length of the lines (Fig. 26.14). The transition between isotropic to anisotropic growth due to the preferential interaction with the local oxide patterns is clearly shown in Fig. 26.15, where two T6 ﬁngers are formed when the T6 molecules reached the patterned area. Figure 26.15b shows a cross-section along the black line in (a). The islands present the same vertical orientation outside and inside the silicon oxide patterns so the template growth does not seem to affect the structure.

Fig. 26.14. Example of templated growth of T6. Image of a T6 island growing inside a silicon oxide pattern. The island grows following the direction of the silicon oxide lines creating channels several microns long and hundreds of nanometers wide. The minimum width of the T6 channel is 100 nm

Fig. 26.15. (a) Example of T6 template growth along silicon oxide stripes. (b) The crosssection reveals that T6 keeps the vertical orientation inside and outside the silicon oxide patterns

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

151

Iwasaki and co-workers developed several protocols for the template growth of ferritin molecules onto local silicon oxides. There, the nanopatterns were used both as positive and negative templates, i.e., molecules deposited preferentially inside (outside) the motifs [84].

26.6 Applications II: Nanoelectronic Devices One relevant application of local oxidation nanolithography is in the fabrication of mesocopic and nanometer-scale electronic devices, such as ﬁeld-effect transistors [91, 92], single electron transistors [93], Josephson junctions [94] or quantum rings [95]. In some cases, the local oxides are an active part of the device, while in others the oxides are used as masks for further processing. Usually, some parts of the device are made by other lithographic techniques, such as optical or electron-beam lithography, while local oxidation is used to pattern the smallest parts of the device. In 1994, Campbell and co-workers fabricated silicon wires by a combination of local oxidation nanolithography and chemical etching [96]. They used a thin silicon ﬁlm isolated by a dielectric ﬁlm from the substrate. A local oxide line was patterned on the ﬁlm, then this line was used as a mask for chemical etching of the unmodiﬁed silicon. The oxide line protected the etching of the silicon underneath and this resulted in the formation of a silicon nanowire [96, 97]. Based in this method, the same group reported the fabrication of a ﬁeld effect transistor [92]. Quate and co-workers also used local oxide lines as masks to fabricate metal-oxide semiconductor FETs [91]. The versatility of local oxidation nanolithography was clearly shown by Snow and co-workers who implemented two different devices, aluminum constrictions where quantized conductance was observed [99] and tunnel junctions to create a metal-oxide-metal transistor. In the last case, the barrier had a 30 nm width and the device showed gate modulation at 300 K [100]. Single electron transistors (SETs) fabricated by local oxidation have been the focus of Matsumoto and co-workers. SETs are sensitive to the variation of a single electronic charge. Their operation is based on the Coulomb blockade phenomenon [101]. A regular transistor SET has three terminals: source, drain and gate. The major difference is the existence of a small metallic island between the source and the drain. The operation of SET depends on the effectiveness of the insulating barrier separating the metallic island from the electrodes. Matsumoto and coworkers fabricated a niobium SET by local oxidation nanolithography that operated at 100 K [102]. Operating SET at room temperature requires good insulating barriers with critical sizes below 10 nm. By using ultraﬂat titanium substrates and sample annealing, Matsumoto and co-workers improved the properties of the local oxide barriers [93,103,104]. In Fig. 26.16a SET operated at room-temperature is shown [75]. The scheme of this SET can be seen in Fig. 26.16b. The Coulomb oscillations of the SET at room-temperature appear in Fig. 26.16c. The periods of Coulomb oscillation are about 1 V, which give an estimate of the gate capacitance of 1.6 × 10−19 F. Bouchiat et al. applied local oxidation to fabricate Josephson junctions and superconducting quantum interference devices (SQUIDs) on Nb substrates. Local oxidation was used to produce constrictions with sizes between 30–100 nm in width and

152 Fig. 26.16. (a) Room temperature SET fabricated in Nb. The tunneling barriers of the island are patterned by local oxidation nanolithography. (b) Scheme of the SET. (c) Coulomb oscillations of the room temperature SET. Image courtesy of [75]

Marta Tello · F. García · R. García

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

153

200–1000 nm in length [94]. Quantum mechanical devices have been fabricated on several semiconductor heterostructures [105, 106]. For example, the modiﬁcation of AlGaAs/GaAs heterostructures containing a two-dimensional electron gas (2DEG) has been used to fabricate quantum rings such as the one shown in Fig. 26.17a [95].

Fig. 26.17. (a) Quantum ring fabricated by local oxidation nanolithography. The current ﬂows from the source to the drain. The in-plane gates (qpc1a, qpc1b, qpc2a, qpc2b, pg1 and pg2) are used to tune the point contacts and two arms of the ring. (b) Schematic of the ring with the dimensions. Image courtesy of [95]

Fig. 26.18. (a) Measurement of Coulomb blockade resonances at ﬁxed magnetic ﬁeld. The current is measured as a function of a voltage applied to both plunger gates (pg1 and 2) simultaneously. (b) The evolution of such sweeps with magnetic ﬁeld results in the additional spectrum shown. The regions of high current (lighter ones) mark conﬁgurations in which a bound state in the ring aligns with the Fermi level in the source and the drain. The Aharonov–Bohm period expected from the ring geometry is indicated by the thin white horizontal lines. (c) Magnetic ﬁeld sweep for constant plunger gate voltage Vpg = 218 mV (dashed line in (b). This peak shows a maximum in amplitude for B = 0, whereas other peaks (Vpg = 270 mV) display a minimum. Image courtesy of [95]

154

Marta Tello · F. García · R. García

Local oxidation nanolithography is used to pattern the lines of the device and to deplete the electron gas below the oxidized regions. The current passes from source to drain and different gates are used to tune the quantum point contacts (qpc1a, qpc1b, qpc2a, qpc2b) and the current ﬂowing through the ring (pg1 and pg2). The dimensions of the structure are presented in Fig. 26.17b. After oxidation, the sample has been covered by a metallic electrode and tuned into the Coulomb blockade regime. In Fig. 26.18b, the current through the ring as a function of the gate voltage and the magnetic ﬁeld is plotted. In Fig. 26.18a, the Coulomb blockade oscillations at a ﬁxed magnetic ﬁeld can be seen. From this plot, the charging energy can be measured as E c = e2 /CΣ ≈ 190 µeV, CΣ being the total capacitance of the ring. In Fig. 26.18b, the Coulomb blockade peaks oscillate as a function of the magnetic ﬁeld with a period of 75 mT, which corresponds to the value of the Aharonov–Bohm period for this ring. The oscillations are shown in more detail in Fig. 26.18c for a ﬁxed plunger gate voltage.

26.7 Parallel Oxidation One of the limitations of local oxidation nanolithography is the sequential character inherent to the AFM operation. The tip has to be moved from one place to another to make the features in a stepwise fashion. This makes the technique very slow for technological applications. To overcome this drawback, Quate and co-workers demonstrated the compatibility of local oxidation with a parallel of arrays of cantilevers, where each tip works independently for reading or writing purposes [77]. A 1 cm2 area was patterned with parallel oxide lines [77].

Fig. 26.19a–c. Schematic representation of the three main steps of parallel local oxidation

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography

155

Fig. 26.20. AFM image of a 15 µm × 15 µm area patterned by parallel local oxidation

Sagiv and coworkers and Cavallini et al. patterned larger areas by using rigid metallic stamps instead of AFM tips. Sagiv et al. used a TEM grid in contact with a surface covered with a self-assembled monolayer and applied a voltage pulse to oxidize the terminal methyl groups of SAM [81]. The friction images showed that micrometer-size grid features were replicated on the monolayer. Cavallini and co-workers showed that submicrometer local oxides could be imprinted by using stamps [108]. They used a metallic stamp, in this case a metallized digital video disk (DVD) polymeric support to directly modify a silicon substrate. Figure 26.19 shows a scheme of the process. The sample and the stamp where brought into mechanical contact and the application of a voltage pulse between the stamp and the surface in the presence of water vapor produced silicon oxides motifs that replicated the features of the stamp. This method allowed 5 × 6 mm2 regions to be patterned. A topographic AFM image of a small section of the patterned region is presented in Fig. 26.20. The oxides were obtained after the application of a pulse of 15 V during 15 s. Ramsier and co-workers showed that the same method can be used to pattern mm2 regions in zirconium nitride ﬁlms [109].

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Binnig G, Rohrer H (1982) Helv Phys Acta 55:726 Binnig G, Quate CF, Gerber Ch (1986) Phys Rev Lett 5556:930 Eigler DM, Schweizer EK (1990) Nature 344:524 Dagata JA, Schneir J, Harary HH, Evans CJ, Postek MT, Bennet J (1990) Appl Phys Lett 56:2001 Lyo IW, Avouris P (1991) Science 253:173 García R (1994) Appl Phys Lett 64:1162 Quate CF (1997) Surf Sci 980:299–300 Binnig H, Rohrer H (1999) Rev Mod Phys 71:S324 García R (2002) Alternative Lithography. Kluwer/Plenum, New York Wouters D, Schubert US (2004) Angew Chem Int Ed 43:2480

156

Marta Tello · F. García · R. García

11. Colton RJ, Engel A, Frommer JE, Gaub HE, Gewirth AA, Guckenberger R, Rabe J, Heckl WM, Parkinson B (1998) Procedures in Scanning Probe Microscopies. Wiley, New York 12. Wendel M, Lorenz H, Kotthaus JP (1995) Appl Phys Lett 67:3732 13. Mamin HJ, Rugar D (1992) Appl Phys Lett 61:1003 14. Binnig G, Despont M, Drechsler U, Häberle W, Lutwyche MI, Vettiger P, Mamin HJ, Chui BW, Kenny TW (1999) Appl Phys Lett 74:1329 15. Xu S, Liu GY (1997) Langmuir 12:127 16. Xu S, Miller S, Laibinis PE, Liu GY (1999) Langmuir 15:7244 17. Mamin HJ, Guethner PH, Rugar D (1990) Phys Rev Lett 65:2118 18. Pascual JI, Méndez J, Gómez-Herrero J, Baró AM, García N, Binh VT (1993) Phys Rev Lett 71:1852 19. Hosaka S, Koyanagi H (1994) Jpn J Appl Phys part 2 33:L1358 20. Ramsperger U, Uchihashi T, Nejoh H (2001) Appl Phys Lett 78:85 21. Calleja M, Tello M, Anguita J, Garcia F, Garcia R (2001) Appl Phys Lett 79:2471 22. Israelachvili JN (1992) Intermolecular and Surface Forces. Academic Press, London 23. Piner RD, Zhu J, Xu F, Hong S, Mirkin CA (1999) Science 283:661 24. Noy A, Miller AE, Klare JE, Weeks BL, Woods BW, DeYoreo JJ (2002) Nano Lett 2:109 25. Sheehan PE, Whitman LJ (2004) Appl Phys Lett 85:1589 26. Ginger DS, Zhang H, Mirkin CA (2004) Angew Chem Int Ed 43:30 27. Dagata JA, Schneir J, Harary HH, Bennet J, Tseng W (1991) J Vac Sci Technol B 9:11384 28. Dagata JA, Tseng W, Bennet J, Schneir J, Harary HH (1991) J Appl Phys 70:3661 29. Thundat T, Nagahara LA, Oden PI, Lindsay SM, George MA, Glaunsinger WS (1990) J Vac Sci Technol A 8:3527 30. Day HC, Allee DR (1993) Appl Phys Lett 62:2691 31. García R, Calleja M, Pérez-Murano F (1998) Appl Phys Lett 72:2295 32. Pérez-Murano F, Abadal G, Varonil N, Aymerich X, Servat J, Gorostiza P, Sanz F (1995) J Appl Phys 78:6797 33. Snow ES, Campbell PM (1995) Science 270:1639 34. Tello M, García R (2001) Appl Phys Lett 79:424 35. Garcia R, San Paulo A (1999) Phys Rev B 60:4961 36. Garcia R, Calleja M, Rohrer H (1999) J Appl Phys 86;1898 37. Gómez-Moñivas S, Sáenz JJ, Calleja M, García R (2003) Phys Rev Lett 91:056101 38. García-Martín A, García R (submitted) 39. Calleja M, Tello M, García R (2002) J Appl Phys 92:5539 40. Okada Y, Iuchi Y, Kawabe M, Harris Jr JS (2000) J Appl Phys 88:1136 41. Lazzarino M, Jun S, Ressel B, Prince KC, Ascoli C (2002) Appl Phys Lett 81:2842 42. Morimoto K, Pérez-Murano F, Dagata J (2000) Appl Surf Sci 158:205 43. Sasa S, Ikeda T, Akahori M, Kajiuchi A, Inoue M (1998) Jpn J Appl Phys 38:1064 44. M. Tello (2004) Doctoral Thesis. Universidad Autónoma de Madrid, Spain 45. Shirakashi J, Ishii, M, Matsumoto K, Miura N, Konagai M (1996) Jpn J Appl Phys 35:L1524 46. Sugimura H, Uchida T, Kitamura N, Masuhara H (1993) Appl Phys Lett 63:1288 47. Rolandi M, Suez I, Dai H, Fréchet JM (2004) Nano Lett 4:889 48. Abadal G, Boisen A, Davis ZJ, Hansen O, Grey F (1999) Appl Phys Lett 74:3206 49. Wang D, Tsau L, Wang KL, Chow P (1995) Appl Phys Lett 67:1295 50. Gwo S, Yeh CL, Chen PF, Chou YC, Chen TT, Chao TS, Hu SF, Huang TY (1999) Appl Phys Lett 74:1090 51. Xie XN, Chung HJ, Sow CH, Wee TS (2004) Appl Phys Lett 84:4914 52. Chien FSS, Chou YC, Chen TT, Hsieh WF, Chao TS, Gwo S (2001) J Apl Phys 89:2465 53. Pellegrino L, Pallechi I, Marré D, Belingeri E, Siri AS (2002) Appl Phys Lett 81:3849 54. Song I, Kim BM, Park G (2000) Appl Phys 76:601

26 Fabrication of Nanometer-Scale Structures by Local Oxidation Nanolithography 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 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.

157

Maoz R, Cohen SR, Sagiv J (1999) Adv Mater 11:55 Hoeppener S, Maoz R, Cohen SR, Chi L, Fuchs H, Sagiv J (2002) Adv Mater 14:1036 Wouters D, Schubert US (2003) Langmuir 19:9033 Marchi F, Bouchiat V, Dallaporta H, Safarov V, Tonneau D, Doppelt P (1998) J Vac Sci Technol B 16:2952 Wei YY, Eres G (2000) Appl Phys Lett 76:194 Tello M, García R (2003) Appl Phys Lett 83:2339 Tello M, García R, Martín-Gago JA, Martín-González MS, Martínez NF, Aballe L, Barinov A, Gregoratti L (2005) Adv Mater 17:1480 Sugimura H, Nakagiri N (1995) Jpn J Appl Phys 34:3406 Teuschler T, Mahr K, Miyazaki S, Hundhausen M, Levy L (1995) Appl Phys Lett 67:3144 Avouris P, Hertel T, Martel R (1997) Appl Phys Lett 71:287 Stiévenard D, Fontaine PA, Dubois E (1997) Appl Phys Lett 70:3272 Pérez-Murano F, Birkelund K, Morimoto K, Dagata JA (1999) Appl Phys Lett 75:199 Calleja M, Anguita J, García R, Birkelund K, Pérez-Murano F, Dagata J (1999) Nanotechnology 1:10 Dagata JA, Pérez-Murano F, Abadal G, Morimoto K, Inoue T, Itoh J, Yokoyama H (2000) Appl Phys Lett 76:2710 Dubois E, Bubendorff JL (2000) J Appl Phys 87:8148 Calleja M, García R (2000) Appl Phys Lett 76:3427 Dagata JA, Pérez-Murano F, Martín C, Kuramochi H, Yokoyama H (2004) J Appl Phys 96:2386 Cabrera N, Mott NF (1948) Rep Prog Phys 12:163 Uhlig HH (1956) Acta Metall 4:541 Cooper EB, Manalis SR, Fang H, Dai H, Matsumoto K, Minne SC, Hunt T, Quate CF (1999) Appl Phys Lett 75:3566 Gotoh Y, Matsumoto K, Maeda T, Cooper EB, Manalis SR, Fang H, Minne SC, Hunt T, Dai H, Harris J, Quate CF (2000) J Vac Sci Technol A 18:1321 Tello M, García F, García R (2002) J Appl Phys 92:4075 Minne SC, Adams JD, Yaralioglu G, Manalis SR, Atalar A, Quate CF (1998) Appl Phys Lett 73:1742 Chien FSS, Wu CL, Chou YC, Chen TT, Gwo S, Hsieh WF (1999) Appl Phys Lett 73:2429 Maoz R, Frydman E, Cohen SR, Sagiv J (2000) Adv Mater 12:725 Liu S, Maoz R, Schmid G, Sagiv J (2002) Nanoletters 2:1055 Hoeppener S, Maoz R, Sagiv J (2003) Nanoletters 3:761 Yasuda T, Yamasaki S, Gwo S (2000) Appl Phys Lett 77:3917 Martín-Sánchez J, Gónzalez Y, González L, Tello M, García R, Granados D, García JM, Briones F (2005) J Crystal Growth 284:313 Yoshinobu T, Suzuki J, Kurooka H, Moon WC, Iwasaki H (2003) Electrochim Acta 48:3131 García R, Tello M, Moulin JF, Biscarini F (2004) Nanolett 4:1115 Gwo S, Yasuda T, Yamasaki S (2001) J Vac Sci Technol A 19:1806 Sugimura H, Nakagiri N (1995) Appl Phys Lett 66:1430 Sugimura H, Nakagiri N, Ichinose N (1995) Appl Phys Lett 66:3686 Komeda T, Namba K, Nishioka Y (1997) Appl Phys Lett 70:3398 Biscarini F, Zamboni R, Samori P, Ostoja P, Taliani C (1995) Phys Rev B 52:14868 Minne SC, Soh HT, Flueckiger Ph, Quate CF (1995) Appl Phys Lett 66:703 Campbell PM, Snow ES, McMarr PJ (1995) Appl Phys Lett 66:1388 Matsumoto K, Gotoh Y, Maeda T, Dagata JA, Harris JS (2000) J Vac Sci Technol A 18:1321 Bouchiat V, Faucher M, Thirion C, Wernsdorfer W, Fournier T, Pannetier B (2001) Appl Phys Lett 79:123

158

Marta Tello · F. García · R. García

95. Fuhrer A, Lüscher S, Ihn T, Heinzel T, Ensslin K, Wegscheider W, Bichler M (2001) Nature 413:822 96. Campbell PM, Snow ES, McMarr PJ (1994) Solid State Electron 37:583 97. Clément N, Tonneau D, Dallaporta H, Bouchiat V, Fraboulet D, Mariole D, Gautier J, Safarov V (2002) Physica E 13:999 98. Abadal G, Boisen A, Davis ZJ, Hansen O, Grey F (1999) Appl Phys Lett 74:3206 99. Snow ES, Park D, Campbell PM (1996) Appl Phys Lett 69:269 100. Snow ES, Campbell PM, Buot FA, Park D, Marrian CRK, Magno R (1998) Appl Phys Lett 72:3071 101. Devoret MH, Esteve D, Urbina C (1992) Nature 360:547 102. Shirakasi J, Matsumoto K, Miura N, Konagai M (1997) Jpn J Appl Phys 36:L1257 103. Matsumoto K, Gotoh Y, Maeda T, Dagata J, Harris J (1999) Jpn J Appl Phys 38:477 104. Shirakashi J, Matsumoto K, Miura N, Konagai M (1998) Appl Phys Lett 72:1893 105. Held R, Lüscher S, Heinzel T, Ensslin K, Wegscheider W (1999) Appl Phys Lett 75:1134 106. Curson NJ, Nemutudi R, Appleyard NJ, Pepper M, Ritchie DA, Jones GAC (2001) Appl Phys Lett 78:3466 107. Lutwyche M, Andreoli C, Binnig G, Brüger J, Drechsler U, Haeberle W, Rohrer H, Rothuzen H, Vettiger P, Yaralioglu G, Quate CF (1999) Sens Actuators A 73:89 108. Cavallini M, Mei P, Biscarini F, García R (2003) Appl Phys Lett 83:5286 109. Farkas N, Comer JR, Zhang G, Evans EA, Ramsier RD, Wight S, Dagata JA (2004) Appl Phys Lett 85:5691

27 Template Effects of Molecular Assemblies Studied by Scanning Tunneling Microscopy (STM) Chen Wang · Chunli Bai

27.1 Introduction Self-assembly of nanometer-sized building blocks into designed molecular architectures represents one of the major goals of supramolecular chemistry and material science [1], with potential applications in molecular devices, such as for molecular information storage devices or organically functionalized surfaces. The driving mechanisms for the assembling processes are, common to many other studies, intermolecular interactions and molecule-substrate interactions. The molecular assembling processes could proceed in diverse conditions such as low temperatures, ambient conditions and liquid–solid interfaces. The STM technique is especially suitable for studying such processes on surfaces. The capability of submolecular resolution by STM can provide important insight on the surface heterogeneity, structural stability, and guest-host interaction of the assemblies. The molecularly decorated surfaces can develop new properties, such as selective adsorption, in association with the functional groups within the assemblies. Interactions such as covalent, metal-ligand binding, hydrogen bond and electrostatic interactions have been successfully used for the design of supramolecular architectures. The hydrogen bonds have the advantage of selectivity and directionality, which are especially important in building biological nanostructures. For other interactions, such as van der Waals and hydrophobic interaction, the lack of directional selectivity makes them generally less desirable in constructing molecular structures. Extensive studies of self-assembled molecular structures have greatly enhanced the knowledge of the optical, electrical, frictional and other properties of molecular materials. The expanding interest and potential applications of the vast variety of molecular structures have stimulated many studies in this general ﬁeld. In this work, we present some of the recent progress on the template effects of surface-bound self-assembled molecular structures. The motivations could be reﬂected in the efforts of gaining better control of the functional selectivity of the self-assembled molecular structures. Template-assisted processing of materials has been a ubiquitous practice using lithographically prepared templates [2, 3] or SAM templates [4]. We will focus on the perspective of using molecular assemblies as templates to build low dimensional molecular structures. The length scale of the molecular templates is on the order of a few nanometers. The preparation conditions include ultra-high vacuum (UHV), low temperatures, ambient conditions and solvent conditions. The underlying mechanisms are the interplays of the adsorption and diffusion barriers inherent to molecular templates. As will be demonstrated, the

160

C. Wang · C. Bai

rich variety of organic functional groups may lead to a wide selection of possible templates. The construction of 2D molecular nanostructures could be achieved at clean metallic and semiconductor surfaces, or at passivated surfaces covered with inert spacer molecules. The assembly behaviour on those surfaces could be very different as the result of surface chemical compositions. The effect is reﬂected through the changes in the magnitude of the adsorption and diffusion barriers. With the presence of various functional groups periodically distributed in the buffer layer, one could encounter anisotropic diffusion barriers that are dependent on the functional groups in the molecular assembling. We will discuss the in-plane co-adsorption and buffered adsorptions of binary molecular systems, respectively, as they represent two typical conditions encountered in molecular templated adsorption and assembly.

27.2 Single Guest Molecule Immobilization with Assembled Molecular Networks The assembled molecular networks with cavities of various size and geometry provide a viable approach to explore guest-host interactions by immobilising guest species at a single molecular level. Such an effort could be expanded into the generalised area of host structure construction and physicochemical properties of the resulting complexes [5, 6]. 27.2.1 Hydrogen Bonded Supramolecular Networks and Single Molecule Inclusions As one of the most useful interactions, hydrogen bond is widely used and studied in molecular self-assembly. The hydrogen bonding conﬁguration of 1,3,5benzenetricarboxylic acid (trimesic acid, TMA) adsorbed on a Cu(100) surface was observed to depend on temperature [7]. At low temperatures (around 200 K), ˚ is the stabilised a quasi-honeycomb-like network with a core diameter about 20 A structure, while the striped structure prevails at room temperature. The distortion of the hexagonal molecular lattice is due to the preferential adsorption at the hollow sites of the four-fold lattice of the Cu(100) substrate [7]. Figure 27.1 shows two types of TMA molecular networks that have been identiﬁed on a graphite surface at 25 K. In addition to the hexagonal honeycomb lattice, the “ﬂower” structure is also stabilized by hydrogen bonding with two different sized cavities [8]. Single TMA molecules could be observed as guest species in the cavities of the existing networks. The entrapped guest TMA molecules were suggested to be stabilized by two hydrogen bonds to the host network as shown in Fig. 27.2. Little effect on the host lattice geometry from the guest molecule can be observed. Recent studies showed that coronene and C60 molecules can also be immobilized within the TMA networks [9, 10]. In addition, an appreciable migration of C60 molecules can be induced at a relatively high tunneling current (approximately 150 pA), providing evidence of mobility at the liquid–solid interface [9].

27 Template Effects of Molecular Assemblies

161

Fig. 27.1. (a) and (b) show two types, honeycomb and ﬂower, of TMA networks observed on a graphite surface at 25 K. (c) and (d) show the corresponding structural models. Reproduced with permission from [8]

A large area of a hydrogen bonded molecular network formed by perylenetetracarboxylic-di-imide (PTCDI) √ and √ 1,3,5-triazine-2,4,6-triamine (melamine) was obtained on an Ag/Si(111)– 3 × 3, R30◦ surface under UHV conditions [11]. Such ˚ supramolecular network adopts a hexogonal pattern with a lattice constant of 34.6 A. The open pores can be packed in with up to seven C60 molecules (Fig. 27.3). The adsorption registry √ of C√60 heptomers is clearly resolved and differs from that of C60 on Ag/Si(111)– 3 × 3 R30◦ . This is a reﬂection of the conﬁnement effect due to the host lattice structure. In addition, C60 can also adsorb directly on top of PTCDI and melamine molecules, leading to a replicated lattice structure. A monolayer of 1,3,5-tris (carboxymethoxy) benzene (TCMB) shows 2D hexagonal networks formed by a hydrogen bond, while a monolayer of 1,3,5–tris (10-carboxydecyloxy) benzene (TCDB) shows 2D tetragonal networks on highly

162

C. Wang · C. Bai

Fig. 27.2. (a–d) Inclusion of single TMA molecules in the TMA networks. (e) Structural model of a guest molecule trapped inside the host network. Reproduced with permission from [8]

oriented pyrolytic graphite (HOPG) [12]. The inclusion effect of hydrogen-bonded two-dimensional networks of TCDB was demonstrated on the surface of HOPG in ambient conditions [13]. With the TCDB network as the host structure, and copper(II) phthalocyanine (CuPc), coronene, decacyclene and pentacene as guest molecules, the host-guest architectures were achieved (Fig. 27.4). An appreciable variation of the lattice dimension was observed as the result of the guest-host interaction. The control of adsorption site and geometry of organic molecules in self-assembled monolayers was established by this method.

27 Template Effects of Molecular Assemblies

163

Fig. 27.3. (a) STM image of the molecular network formed by PTCDI and melamine √ √ hexagonal adsorbed on an Ag/Si(111)– 3 × 3R30◦ surface. The bright clusters correspond to seven C60 molecules within a network cavity as illustrated in (b). Reproduced with permission from [11]

Fig. 27.4. (a) Guest-host structure formed by a TCDB host lattice and coronene guest molecules. The scan size is 11.2 nm × 11.2 nm. (b) Molecular model for the TCDB network connected by hydrogen bonds. Reproduced with permission from [13]. Copyright 2004 American Chemical Society

27.2.2 Van der Waals Interaction Stabilized Networks Characteristic hexagonal and quasi-quadratic structures derived from the ﬁrst generation of n-alkoxy-substituted stilbenoid dendrimers have been observed. The assembly structures were stabilized by the interdigitated alkoxy chains, and the lattice geometry changed appreciably as the alkoxy chain length was increased to more than sixteen carbon numbers [14]. Quasi-quadratic molecular lattices can also be constructed by interdigitated alkylated copper phthalocyanines [15]. The trapping effect of the 2D assembly of octaalkoxyl-substituted phthalocyanine (PcOC8) for individual molecules of phthalocyanine, porphyrins, and calix[8]arene was observed in Fig. 27.5 [16]. It was shown that single molecules are trapped in quadratic rather than hexagonal lattices, and domain boundaries are the preferential trapping sites as compared with the sites within the domains as shown in Fig. 27.6 [16]. The observed trapping behaviors could ﬁnd

164

C. Wang · C. Bai

Fig. 27.5. (a) Immobilization of single molecules inside the quadratic lattices of alkylated phthalocyanines. No distortion of the host lattice geometry is observed. (b) Structural model of the guest molecule entrapment. (c) Side view of the assembly with the guest molecule. Reproduced with permission from [16]. Copyright 2002 American Chemical Society

Fig. 27.6. Examples of the entrapment of single calixarene[8] molecules at different sites of the assembled host lattices shown in both (a) low and (b) high magniﬁcation STM images. (c) Lattice spacing expansion effect due to the guest molecule insertion. [Reproduced with permission from [16]. Copyright 2002 American Chemical Society

27 Template Effects of Molecular Assemblies

165

analogues in the impurity segregation phenomena of point defects, dislocations, and grain-boundary in solid-state materials. The molecular inclusion induced spacing increase between host molecular rows is presented in Fig. 27.6c. It is evident that, with the increase of the intercalating molecular size, the distances of neighboring PcOC8 rows also increase correspondingly. This could be an indication of the ﬂexibility of the PcOC8 lattice. The insertion of molecules in the initially packed lattices led to enhanced repulsive interaction between molecules. On the other hand, the overlapped of alkyl parts could readjust to accommodate additional molecules and compensate the associated increment of repulsions. The molecular networks have thus shown certain degrees of ﬂexibility in trapping different sized and shaped single molecules. 27.2.3 Metal-Organic Coordination Networks A number of two-dimensional molecular grid structures have been demonstrated using coordination polymers. Such networks can be designed using various coordination ligands, leading to different geometries and cavity size. Typical examples can be found in the structures from 4,4 -bipyridine [17–19], carboxylates [20, 21] with metal ligands. It was shown that large areas of molecular networks can be obtained by coadsorption of Fe and carboxylates on a Cu(100) surface [20]. The geometry of the networks can be adjusted using different linker molecules. The STM images revealed both the organic linker molecules and the metal ligand atoms as shown in Fig. 27.7. The guest-host and guest-substrate interactions can be studied from the thermal stabilities of the guest molecules (C60 in this work). Such metal-organic supramolecules have been widely pursued as porous materials for gas separation and

Fig. 27.7. STM images of metal-organic network formed by 4,1 ,4 ,1 -terphenyl1,4 -dicarboxylic acid (TDA) and Fe atoms. (a) Large area image of the network (scale bar 10 nm). (b) High resolution image of the network overlapped with the molecular structure. Reproduced with permission from [20]

166

C. Wang · C. Bai

storage. The direct visualization of the inclusion of guest molecules into such a host network could lead to useful insight on the guest-host interactions in the presence of substrate. 27.2.4 Covalently Bonded Molecular Grids In separate studies, two-dimensional square and hexagonal supramolecular networks can been obtained by using covalent, cation bonding and hydrogen bonding [22]. An example of covalently bonded square lattice was demonstrated using a lanthanum sandwich complex of tetrapyridylporphyrin and p-xylylene as shown in Fig. 27.8 [22]. It can be noted that the effort to acquire capabilities of designing two-dimensional molecular networks has made much progress in recent years. Supported by the knowledge of molecular engineering, novel molecular structures with diverse functions can be expected.

Fig. 27.8. Covalently bonded rectangular network by (a) p-xylylene and (b) lanthanum sandwich complex of tetrapyridylporphyrin. (c) Molecular structure of corresponding to the STM image in (d). Reproduced with permission from [22]

27.3 Intralayer Heterogeneous Molecular Arrays The 2D assembly process of multi-component molecular systems is an interesting topic that may lead to studies of engineering novel molecular nanostructures. Re-

27 Template Effects of Molecular Assemblies

167

ported attempts of adsorption of alkanes with other species normally lead to phase separation [23]. Little attempt has been made to achieve multi-component molecular nanostructures based upon van der Waals or electrostatic interactions due to the reasons discussed above. The reported results indicate that the micro-phase separation is prevalent. Several recent reports have revealed uniform heterogeneous assembly structures. The results suggest these combinatory systems could provide a wide range of molecular interactions, and are worthy of systematic study. A wide variety of self-assembled monolayers of organic molecules on solid substrate surfaces have been studied in the past decade [24–27]. With the aid of STM and many other surface characterization techniques, a rich variety of ordered two-dimensional (2D) monolayers of monosubstituted alkane derivatives have been observed. The molecular arrays are governed by various soft bonds [28], including van der Waals interaction (such as the case of simple alkanes) [25] and hydrogen bonds (such as alcohols and acids) [27, 29, 30], electrostatic interaction (such as cationic surfactants) [31] and dipolar interaction (such as aldehyde) [32]. These results have enabled further studies on multi-component assemblies and the outcome could be beneﬁcial to the engineering of functional molecular nanostructures. 27.3.1 Hydrogen Bond Stabilized Heterogeneous Lamellae For monosubstituted n-alkane derivatives, such as thiol, amines and alcohols, a headto-head connection through hydrogen bonding is commonly formed, showing a 60◦ angle between the molecular axis and the directions of lamellae [29, 33, 34]. When two functional groups are involved, such as disubstituted alkane derivatives, they can induce the formation of superamolecular self-assembly. For example, 1,2-dihydroxyoctadencane forms a lamella-type structure with a 65◦ angle of alkyl chains relative to the lamella [35], while 1,14-tetradecanediol forms a supramolecular arrangement with a herringbone angle of 120◦ between alkyl chains [36]. 16-hydroxyhexadecanoic acid (HO(CH2 )15 COOH) and 15-hydroxypentadecanoic acid self-assemble into the network through multiple hydrogen bonding showing the “odd-even” effect [37]. These observations indicate that the molecular packing arrangements are inﬂuenced by the competitive and collaborative interactions between these different functional groups. It is of genuine interest to understand the interactions between functional groups (especially hydrogen bonding) in the pursuit of novel supramolecular structures. One example of the hydrogen bond stabilized heterogeneous molecular assemblies could be seen as the isomeric compounds of 4,4 -bipyridyl (4bpy) and 2,2 -bipyridyl (2bpy) co-adsorbed with stearic acid on an HOPG surface [38]. Uniformly distributed single molecular arrays have been observed as in Fig. 27.9. 4bpy and 2bpy are isomeric compounds, and the only difference is the positions of the nitrogen atoms. When bpy molecules are mixed with stearic acid, the nitrogen atoms of bpy can form hydrogen bonds with the –COOH groups of stearic acid, and hence the positions of the nitrogen atoms determine the hydrogen bond structure. It is observed that in the 4bpy–stearic acid system, the angle of the 4bpy to the stearic acid molecular axis is ∼ 165◦ , whereas in the 2bpy–stearic acid system, the angle of the 2bpy to the stearic acid molecular axis is ∼ 125◦ . The distances between the

168

C. Wang · C. Bai

Fig. 27.9. The 4bpy arrays stabilized by hydrogen bonding with the co-adsorbed stearic acids. Reproduced with permission from [34]

two neighboring “bright” rows in the STM image are different in these two systems. Thus, by using hydrogen bond conﬁguration, one can distinguish the 4bpy molecule from its isomeric 2bpy molecule. In addition, the enantiomeric molecular patterns in both 4bpy and 2bpy systems can be obtained [38, 39]. There have also been reports on hydrogen bond stabilized molecular arrays on metal surfaces, such as the chiral molecule of 4-[trans-2-(pyrid-4-yl-vinyl)] benzoic acid on Au(111) and Ag(111) surfaces [40]. The periodicity of the array structures was determined by the elbow sites on the Au(111) surface, and the repulsive inter-array interactions on Ag(111) surface. It was demonstrated that by adjusting the dipole-dipole interaction via functional groups of cyanophenyl porphyrin, monomer (H2-TBPP, without cyano substitution), and trimer (CTBPP), tetramer (cis-BCTBPP) as well as 1D molecular wire structure (trans-BCTBPP) can be achieved [41]. 27.3.2 Van der Waals Interaction Stabilized Intralayer Arrays In the situation where the adsorption barriers have similar magnitude for template molecules and guest molecules, both species will adsorb at the same basal plane. Depending on the thermodynamic equilibrium, both heterogeneous assembly or homogeneous phase separated domains could be obtained. The determining factors are entropic force and intermolecular interactions. Heterogeneous two-dimensional patterns of phthalocyanines were reported via co-adsorption of binary molecular species [42–45]. The formation of such heterogeneous assemblies is directly associated with the intermolecular interactions, and more interestingly, the assembly of these ordered structures is based on electrostatic or van der Waals interaction, which lacks directional selectivity. It is further plausible to expect that the molecular assemblies may serve as templates for constructing 2D heterogeneous molecular structures.

27 Template Effects of Molecular Assemblies

169

Fig. 27.10. Molecular arrays of phthalocyanine using templates of C18 H37 SH. (a) Large view and (b) high resolution STM images of the assembly. (c) and (d) The effect of the molar ratio on the assembly structures. (e) Structural model of the heterogeneous molecular array structure. Reproduced with permission from [44]. Copyright 2001 American Chemical Society

Single molecular arrays of Pc formed with templates of alkane derivatives have been obtained and characterized by STM [44, 45]. Figure 27.10 illustrates single molecular arrays of Pc obtained with the templates of n-octadecyl mercaptan

170

C. Wang · C. Bai

(C18SH), 1-iodooctadecane (C18I), 1-bromooctadecane (C18Br), 1-chlorooctadecane (C18Cl) and octadecyl cyanate (C18CN). Three different phases can be identiﬁed depending on the molar ratio of Pc to alkane derivatives [44]. By adjusting the molar ratio to about 1 : 3, a uniform array-like assembly can become dominant on the surface [44]. In addition, the co-deposition of Pc with other alkane derivatives such as octadecanol and stearic acid has not led to the single-molecular-array-like assembly. It was, therefore, conjectured that the interaction between Pc and the end group of the alkane derivative and the strength of interaction between functional groups of alkane derivatives play important roles in the assembly process. Rigorous elucidations of this process need quantitative simulations. In another study using decanethiol lamellae on Au(111) as the template [46,47], bimolecular arrays of C60 were obtained [48]. The C60 molecules were adsorbed atop thiol groups in the stripes of decanethiols as shown in Fig. 27.11. Since self-assembled structures represent thermodynamic minima, they are formed by reversible association of individual molecules, and the interplay of enthalpy and entropy in the assembly process is more important than in the synthesis process based on the formation of covalent bonds. Accurate simulations of the assembly behavior require considerations of not only the interaction between the adsorbates and substrate, but also of the interactions between similar as well as dissimilar molecules. The entropy contribution during the transition from the solution state to assembly of structures should also be considered [49]; this brings great complexity to the theoretical simulation of the system. Similar studies could be found in the lamella ordering of alternating layers in the binary mixtures of hard spheres (such as polyethylene oxide (PEO), polyethylene glycol (PEG) polystyrene latex (PS) spheres) and linear rods (such as nematic and smectic liquid crystal molecules, the ﬁlamentous bacteriophage fd virus or the tobacco mosaic virus (TMV)) [50–54]. The driving force for the lamella ordering phase has been attributed to entropic maximization with the assumption of hard-sphere interaction (i.e. zero interaction when two spheres are separated). Similarly, a binary mixture of different sized spheres has also been observed to assemble into regular array-like

Fig. 27.11. (a) Bimolecular C60 arrays assembled at position of the thiol groups. (b) Proposed structural model. Reproduced with permission from [48]. Copyright 2001 American Institute of Physics

27 Template Effects of Molecular Assemblies

171

structures as the result of entropy maximization [55]. Such an entropy-driven mechanism could qualitatively explain the observed surface-bound binary molecular array structures. In addition, the observed effect of the terminal functional groups of the linear molecules indicate that the presumed hard-sphere interaction should be modiﬁed to accommodate the results of the molecular heterogeneous structures. Since the concentrations of the solutions used in the above STM experiments and the molecular structures are nearly identical, the contributions of entropy associated with the formation of self-assembled structures to the free energy may be considered comparable. Therefore, one can focus on the enthalpy effect that can be determined by intermolecular interaction, including the interaction between similar molecules and dissimilar molecules, i.e. the energy change by bringing molecules together into a particular assembly pattern. It can be considered that the observed single molecular arrays may provide an alternative approach for constructing single molecular wires, in addition to the nicely demonstrated synthetic routes [56]. This approach could assemble hundreds of Pc molecules into highly directional arrays. As the separation between adjacent molecules is typically less than the mean free path of conductive electrons, it is believed that the assembled molecular arrays can essentially serve as conductive molecular wires when properly connected. Because of the rich physicochemical properties of phthalocyanine and fullerene molecules, the obtained molecular arrays are of interest in the potential application for molecular devices. For example, it could be envisioned that by altering the length of the carbon chains of the functionalized alkanes and the structure of Pc molecules, molecular wires of different properties and width can be prepared.

27.4 Interlayer Effect on Molecular Adsorption and Assemblies Molecule-substrate interaction has been extensively studied and is still a vibrant topic after decade of extensive studies. Molecular adsorption induced effects, such as molecular diffusion, and surface restructuring, have been well-documented in the case of molecular adsorption on metal and semiconductor substrates [57–60]. There have been a very limited number of experimental studies dedicated to the microscopic effects of molecular adsorption on an organic molecular support [61–63]. The lack of information on the microscopic adsorption on organic substrates may mainly be due to the lack of experimental capabilities suitable for such systems. With the wealth of functionalities associated with the molecular assemblies, it is conceivable that such studies will be highly rewarding. The information obtained is likely to be of great importance to the construction of molecular nanostructures in both 2D and 3D. Considering the different polarity and electronic properties of the functional groups from the family of alkane derivatives, the monolayers of alkane derivatives may provide ideal templates for the investigation of the adsorption behavior of organic molecules. Compared with the nanofabricated surface, this alkane-derivativemodiﬁed surface is organic rather than metallic or semiconducting. The relative polarity and the ratio of the polar/non-polar area may be modiﬁed easily by changing the functional groups and the chain length of the alkane derivatives.

172

C. Wang · C. Bai

The interlayer interaction is an important factor for determining the overlayer structures. The organic–organic heterogeneous interface is generally associated with weak interactions, such as the van der Waals force. The intermolecular interactions are the dominant factors such as in the case of PTCDA on thiol SAM [61]. The introduction of various functional groups to the buffer layer offers possibilities to further explore the effect of molecular interfacial interactions on the assembly structures. 27.4.1 Site Selective Adsorption 27.4.1.1 Simple Alkane Lamellae When pre-covered on the support surface, the molecular lamella structures inherently introduce the heterogeneous adsorption sites and anisotropic diffusion barriers in association with the functional groups. The presence of heterogeneous adsorption sites may result in the selective adsorption of single molecules. The adsorbed species would also experience the anisotropic diffusion barrier and organize in a restricted manner. A schematic illustration is given in Fig. 27.12 showing the functional group induced variation of the diffusion barrier. Site-selective adsorption of copper phthalocyanine (CuPc) has been observed on top organic surfaces of monolayers of various alkane derivatives (stearic acid, 1-octadecanol and 1-iodooctaecane) adsorbed on the HOPG surface. STM studies have shown that the alkane derivatives form templates, which direct the adsorption of the CuPc. This selective adsorption behavior is attributed to a dominating preference of the CuPc for adsorption on the hydrocarbon-chain portions of the supporting layers [64]. Two representative adsorption sites were considered in the molecular mechanics simulations schematically shown in Fig. 27.13, i.e. site II in which CuPc adsorb on the top of the trough linked by head-to-head functional groups, and site I in which CuPc adsorb on the alkyl moiety. The calculated results indicate that the system potentials on site I are higher than those on site II by above 21 kJ/mol for three alkane derivative systems. This may be caused by two factors. First the trough linked by ˚ in width and the number of atoms in the head-to-head functional groups is about 3 A trough is less than at other sites of the organic monolayers. Consequently, the van der

Fig. 27.12. Schematic of surface diffusion barriers on top of molecular assemblies

27 Template Effects of Molecular Assemblies

173

Fig. 27.13. Molecular mechanics simulation of different adsorption sites a top alkane lamellae

Waals interaction between the adsorbed CuPc and the underlying organic substrate decreases when CuPc adsorbs on the top of the trough. Moreover, the trough is linked by atoms and groups with large polarity, and the electrostatic repulsion between the π ring of CuPc and functional groups will be stronger than that between CuPc and alkyl. These two factors drive the selective adsorption of CuPc on alkyl moiety to achieve the most stable adsorption state. The selectivity is dependent on the relevant functional groups and could vary among alkane derivatives. When phthalocyanine molecules coadsorbed with stearic acid, isolated and paired Pc molecules were detected on top of the stearic acid monolayer, as seen in Fig. 27.14 [64]. As observed in the experiments, these Pc molecules all located on top of the alkane part of the stearic acid lamellae, Pc molecules adsorbed on the regions were ascribed to the location of carboxyl groups were never detected.

174

C. Wang · C. Bai

Fig. 27.14. (a) Isolated CuPc selectively adsorbed on the hydrocarbon-chain portion of stearic acid. (b) CuPc adsorbed on the monolayer of 1-octadecanol [64]. Copyright 2004 American Chemical Society

27.4.1.2 Alkylated Amino Acid Molecular Templates Site selective adsorption of urea molecules was observed on the lamella templates of double-alkyl amino acid in which unsaturated hydrogen bonds are available for adsorption [65]. The unprotected amino acid groups were found to be the preferential adsorption sites for urea molecules, as shown in Fig. 27.15. Each amino acid group

Fig. 27.15. (a) and (b) Selective adsorption of urea molecules at the sites of amino groups. (c) Schematic of site selective adsorption of guest molecules atop molecular templates. Reproduced with permission from [65]

27 Template Effects of Molecular Assemblies

175

can adsorb either one or two urea molecules, reﬂected by localized clustering of the adsorbates at the adsorption sites corresponding to the position of amino acids (Fig. 27.15). Such a selective adsorption effect is not observable for the lamella structure of single-alkyl amino acid, due to the dimer formation that saturates the hydrogen bonds of amino acids. However, such saturation of the amino acid groups can be avoided by introducing fatty acids (C23 H47 COOH) as the matrix molecule [128]. The alkyl substituted amino acids are randomly distributed within the matrix lamella of C23 H47 COOH. The amino acid groups are not saturated in this case and are available for interaction with adsorbates such as palladium(II) acetate and urea. 27.4.1.3 Tridodecyl Amine (TDA) Templates The template effects have been illustrated with the lamella of tridodecyl amine (TDA) molecules on the adsorption, diffusion and assembly structures of copper phthalocyanine. The conformation of the nitrogen atom in the amine group of TDA is tetrahedrical, in which the nitrogen atom sits on one acme of this tetrahedron (Fig. 27.16). Since the C–N bond is dipolar, amine molecules are also dipolar, in which the nitrogen is partially negatively charged. Thus, when amine molecules

Fig. 27.16. (a) High resolution STM image of the TDA lamellae structure and (b) proposed packing model of the lamellae structure. (c) Structural schematics illustrating the side and top views of the binding of a benzoic acid molecule to the TDA template. (d) STM image of benzoic acid adsorbed on TDA lamellae [66]. Copyright 2003 American Chemical Society

176

C. Wang · C. Bai

adsorb onto an inert surface of graphite, there exists a net dipole moment pointed near perpendicular to the surface. Benzoic acid was found exclusively adsorbed onto the TDA assembly at the sites of amino groups [66]. This shows the possibility of using alkane derivative lamellae as the template to direct adsorption and assembly of other organic molecules. By co-adsorption of TDA with copper phthalocyanine, isolated CuPc molecules and clusters were stabilized at the alkane part of the lamellae (Fig. 27.17) [67]. When co-adsorbed with TDA at low Pc:TDA ratio, CuPc dimers are most commonly observed, with a smaller population of quadrimers and hexamers. From the large scale view in Fig. 27.14, one can observe two CuPc molecules located on both sides of the amine group of the TDA lamellae. The CuPc dimers appear to adsorb on top of the alkane part of the TDA lamellae. The lateral diffusion of the single CuPc molecules, as well as clusters of adsorbed CuPc molecules were found exclusively along the direction of the TDA lamellae. Such a highly directional diffusion behavior can be direct evidence of the onedimensional template effect of TDA lamellae. Such effects have never been observed on the lamellae of simple alkanes, possibly due to the lack of functional groups that could establish sufﬁcient diffusion barriers for the overlayer adsorbates. This concept can be generalized to the construction of molecular templates for novel molecular nanostructures.

Fig. 27.17. Isolated CuPc molecules observed on the lamellae of TDA. Arrows indicate the migration of the molecules in consecutive scans [67]

27.4.2 Molecular Arrays In addition to the above described isolated adsorbates, well-ordered double rowed CuPc molecular bands were developed at higher CuPc coverage as the result of the molecular template of TDA lamellae. A bimolecular CuPc band can be readily

27 Template Effects of Molecular Assemblies

177

observed in Fig. 27.18 [67]. These bimolecular bands can extend to hundreds of nanometers without appreciable distortion. As measured from the STM images, the repeating period of this assembling structure is consistent with the 3.37 ± 0.05 nm width of the TDA lamellae, which that suggests the ordering of CuPc molecules is induced by the TDA lamellae. High resolution STM observation reveals that the arrangement of CuPc in these double molecular bands is the same as that in the 2D crystals: joggled together with each other, showing an intermolecular distance of 1.45 nm. There is an apparent dark trough between these bimolecular molecular bands. The width of this trough is about 0.7 nm larger than the 0.3–0.4 nm trough that separates the TDA lamellae. A molecular model of this band structure has been proposed as shown in Fig. 27.18c.

Fig. 27.18. Large scale view (a) and high resolution image (b) of CuPc double molecular bands. (c) is the molecular model corresponding to the structure shown in (b) [67]

27.4.3 Directional Assembly of Nanoparticle Arrays Assembling of molecular-sized components is a promising approach for the construction of nanostructures. These components could include organic molecules and monolayer protected metal clusters. For metal nanoparticles, 2D ordered monolayers have also been reported [68,69]. Monolayer protected clusters (MPC) are considered to represent a new class of specimen due to their high stability in air and solvent, and also their derivatization ﬂexibility [70]. Although ligand-capped gold clusters have been known for a number of years, the practical formation of stable, isolable monolayer-protected clusters has only been recently demonstrated [71]. The preparation methodology was greatly advanced in the following years [72–76]. The alkanethiolated MPCs differ from conventional colloids and nanoparticles in that they can be repeatedly isolated and re-dissolved in common organic solvents without irreversible aggregation. The Schiffrin reaction can endure considerable modiﬁcations to protect both the ligand structures and the core metal of MPCs. Reports of MPCs with Ag and alloy (Au/Ag, Au/Cu, Au/Ag/Cu, Au/Pt, Au/Pd and Au/Ag/Cu/Pd) cores and various differently structured ligands have appeared [77, 78]. STM observations indicate that ultra-small thiol capped Au55 clusters and nickel clusters (Ni-MPC) can assemble into ordered arrays on top of the monolayer of

178

C. Wang · C. Bai

Fig. 27.19. (a) Lamella structure of 1-dodecanethiol. (b,c) The ordered assemblies of Ni-MPCs along the template of 1-dodecanethiol lamellae. Possible multi-layer structures are marked by dark arrows. Reproduced with permission from [80]

Fig. 27.20. (a) Assembling structures of Ni-MPC on top of the lamellae of C34 H70 , possible multilayer structures are indicated by dark arrows. (b) C34 H70 lamellae visualized after removal of the Ni-MPC assembly. Insert: High-resolution image of the monolayer structure of C34 H70 . Reproduced with permission from [80]

27 Template Effects of Molecular Assemblies

179

thiol and alkanes (Figs. 27.19 and 27.20) [79, 80]. Assembling nanoparticles into ordered structures is an interesting topic, because of its relevance to the construction of nano-devices. The current result demonstrates a possible way toward the ordered array of nanoparticles using the template effect of alkane monolayers and will be helpful in achieving the goal of the construction of nano-electronic devices. The presence of heterogeneous adsorption sites would result in selective adsorption and assembling of nanoparticles.

27.5 Future Perspectives There are many functions of molecular templates that have yet to be explored, such as the sensor or catalytic behavior of the molecular assemblies, molecular devices based on heterogeneous assemblies (donor, acceptor, p-, n-types), assemblies of molecular magnets, etc. It is conceivable that with the rich variety of functional groups that can be incorporated in the molecular structures, the pursuit of templates with novel physicochemical properties could be fruitful. Efforts in the theoretical analysis of the assembling processes are being made to gain a deeper insight on the driving mechanisms.

References 1. Lehn JM (1997) Supramolecular Chemistry: Concepts and Perspectives. VCH, Weinheim 2. Kim SO, Solak HH, Stoykovich MP, Ferrier NJ, De Pablo JJ, Nealey PF (2003) Nature 424:411; Register RA (2003) Nature 424:378 3. Boltau M, Walheim S, Mlynek J, Krausch G, Steiner U (1998) Nature 391:877 4. Gleiche M, Chi LF, Fuchs H (2000) Nature 403:173 5. Nassimbeni LR (2003) Acc Chem Res 36:631 6. Swiegers GF, Malefetse TJ (2000) Chem Rev 100:3483 7. Dmitriev A, Lin N, Weckesser J, Barth JV, Kern K (2002) J Phys Chem B 106:6907 8. Griessl SJH, Lackinger M, Edelwirth M, Hietschold M, Heckl WM (2002) Single Mol 3:25 9. Griessl SJH, Lackinger M, Jamitzky F, Markert T, Hietschold M, Heckl WM (2004) Langmuir 20:9403 10. Griessl SJH, Lackinger M, Jamitzky F, Markert T, Hietschold M, Heckl WM (2004) J Phys Chem B 108:11556 11. Theobald JA, Oxtoby NS, Phillips MA, Champness NR, Beton PH (2003) Nature 424:1029 12. Lu J, Zeng QD, Wang C, Zheng QY, Wan LJ, Bai CL (2002) J Mater Chem 12:2856 13. Lu J, Lei SB, Zeng QD, Kang SZ, Wang C, Wan LJ, Bai CL (2004) J Phys Chem B 108:5161 14. Xu SD, Zeng QD, Lu J, Wang C, Wan LJ, Bai CL (2003) Surf Sci 538:L451 15. Qiu XH, Wang C, Zeng QD, Xu B, Yin SX, Wang HN, Xu SD, Bai CL (2000) J Am Chem Soc 122:5550 16. Liu YH, Lei SB, Yin SX, Xu SL, Zheng QY, Zeng QD, Wang C, Wan LJ, Bai CL (2002) J Phys Chem B 106:12569 17. Biradha K, Hongo Y, Fujita M (2000) Angew Chem Int Ed 39:3843 18. De Feyter S, Abdel-Mottaleb MMS, Schuurmans N, Verkuijl BJV, van Esch JH, Feringa BL, De Schryver FC (2004) Chem Eur J 10:1124 19. Kaes C, Katz A, Hosseini MW (2000) Chem Rev 100:3553

180

C. Wang · C. Bai

20. Stepanow S, Lingenfelder M, Dmitriev A, Spillmann H, Delvigne E, Lin N, Deng XB, Cai CZ, Barth JV, Kern K (2004) Nat Mater 3:229 21. Eddaoudi M, Moler DB, Li H, Chen B, Reineke TM, O’Keeffe M, Yaghi OM (2001) Acc Chem Res 34:319 22. for example, Michl J, Magnera TF (2002) Proc Natl Acad Sci USA 99:4788 23. for example, Baker RT, Mougous JD, Brackley A, Patrick DL (1999) Langmuir 15:4884 24. Smith DPE, Horber H, Gerber C, Binnig G (1989) Science 245:43 25. Rabe JP, Buchholtz S (1991) Science 253:424 26. Giancarlo LC, Flynn GW (1998) Annu Rev Phys Chem 49:297 27. De Feyter S, Gesquiere A, Abdel-Mottaleb MM, Grim PCM, De Schryver FC, Meiners C, Sieffert M, Valiyaveettil S, Mullen K (2000) Acc Chem Res 33:520 28. De Feyter S, De Schryver FC (2003) Chem Soc Rev 32:139 29. Claypool CL, Faglioni F, Goddard III WA, Gray WB, Lewis NS, Marcus RA (1997) J Phys Chem B101:5978 30. Hibino M, Sumi A, Tsuchiya H, Hatta I (1998) J Phys Chem B 102:4544 31. Xu SL, Wang C, Zeng QD, Wu P, Wang ZG, Yan HK, Bai CL (2002) Langmuir 18:657 32. Xu SL, Zeng QD, Wu P, Qiao YH, Wang C, Bai CL (2003) Appl Phys A 76:209 33. Cyr DM, Venkataraman B, Flynn GW, Black A, Whitesides GM (1996) J Phys Chem 100:13747 34. Xu QM, Wan LJ, Bai CL (2001) Surf Interface Anal 32:256 35. Qian P, Nanjo H, Yokoyama T, Suzuki TM (1998) Chem Lett 1133 36. Elbel N, Roth W, Günther E, von Seggern H (1994) Surf Sci 303:424 37. Wintgens D, Yablon DG, Flynn GW (2003) J Phys Chem B 107:173 38. Xu B, Yin SX, Wang C, Zeng QD, Qiu XH, Bai CL (2001) Surf Interface Anal 32:245 39. Qian P, Nanjo H, Yokoyama T, Suzuki TM (1999) Chem Comm 1197 40. Weckesser J, De Vita A, Barth JV, Cai C, Kern K (2001) Phys Rev Lett 87:96101 41. Yokoyama T, Yokoyama S, Kamikado T, Okuno Y, Mashiko S (2001) Nature 413:619 42. Hipps KW, Scudiero L, Barlow DE, Cooke MP (2002) J Am Chem Soc 124:2126 43. Scudiero L, Hipps KW, Barlow DE (2003) J Phys Chem B 107:2903 44. Lei SB, Wang C, Yin SX, Bai CL (2001) J Phys Chem B 105:12272 45. Lei SB, Yin SX, Wang C, Wan LJ, Bai CL (2002) Chem Mater 14:2837 46. Toerker M, Staub R, Fritz T, Schmitz-Hübsch T, Sellam F, Leo K (2000) Surf Sci 445:100 47. Poirier GE (1997) Chem Rev 97:1117 48. Zeng CG, Wang B, Li B, Wang HQ, Hou JG (2001) Appl Phys Lett 79:1685 49. Whitesides GM, Mathias JP, Seta CT (1991) Science 254:1312 50. Koda T, Numajiri M, Ikeda S (1996) J Phys Soc Jpn 65:3551 51. Aams M, Dogic Z, Keller SL, Fraden S (1998) Nature 393:349 52. Dogic Z, Frenkel D, Fraden S (2000) Phys Rev E 62:3925 53. van der Schoot P (2002) J Chem Phys 117:3537; ibid (2000) J Chem Phys 112:9132 54. Lee J, Thompson RB, Jasnow D, Balazs AC (2002) Phys Rev Lett 89:155503 55. Bartlett P, Ottewill RH, Pusey PN (1992) Phys Rev Lett 68:3801 56. Aratani N, Osuka A, Kim YH, Jeong DH, Kim D (2000) Angew Chem Int Ed 39:1458; Tsuda A, Osuka A (2001) Science 293:79; Tour JM (2000) Acc Chem Res 33:791 57. Forrest SR (1997) Chem Rev 97:1793 58. Rosei F, Schunack M, Jiang P, Gourdon A, Lægsgaard E, Stensgaard I, Joachim C, Besenbacher F (2002) Science 296:328 59. Briner BG, Doering M, Rust HP, Bradshaw AM (1997) Science 278:257 60. Mitsui T, Rose MK, Fomin E, Ogletree DF, Salmeron M (2002) Science 297:1850 61. Gerstenberg MC, Schreiber F, Leung TYB, Bracco G, Forrest SR, Scoles G (2000) Phys Rev B 61:7678 62. Samori P, Severin N, Simpson CC, Müllen K, Rabe JP (2002) J Am Chem Soc 124:9454

27 Template Effects of Molecular Assemblies 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80.

181

Xu B, Yin SX, Wang C, Qui XH, Zeng QD, Bai CL (2000) J Phys Chem B 104:10502 Lei SB, Yin SX, Wang C, Wan LJ, Bai CL (2004) J Phys Chem B 108:224 Hoeppener S, Wonnemann J, Chi LF, Erker G, Fuchs H (2003) Chem Phys Chem 4:490 Lei SB, Wang C, Wan LJ, Bai CL (2003) Langmuir 19:9759 Lei SB, Wang C, Wan LJ, Bai CL (2004) J Phys Chem B 108:1173 for example, Sastry M, Rao M, Ganesh K (2002) Acc Chem Res 35:847 Sampaio JF, Beverly KC, Heath JR (2001) J Phys Chem B 105:8797 Templeton AL, Wuelﬁng WP, Murray RW (2000) Acc Chem Res 33:27 Brust M, Walker M, Bethell D, Schffrin DJ, Whyman R (1994) Chem Commun 801 Chen S, Murray RW (1999) Langmuir 15:682 Wuelﬁng WP, Zamborini FP, Templeton AC, Wen X, Yoon H, Murray RW (2001) Chem Mater 13:87 Templeton AC, Chen S, Gross SM, Murray RW (1999) Langmuir 15:66 Johnson SR, Evans SD, Brydson R (1998) Langmuir 14:6639 Buining PA, Humbel BM, Philipse AP, Verkleij AJ (1997) Langmuir 13:3921 Collier CP, Saykally RJ, Shiang JJ, Henrichs SE, Heath JR (1997) Science 277:1978 Hostetler MJ, Zhong CJ, Yen BHK, Anderegg J, Gross SM, Evans ND, Porter M, Murray RW (1998) J Am Chem Soc 120:9396 Hoeppener S, Chi LF, Fuchs H (2002) Nano Lett 2:459 Lei SB, Wang C, Yin SX, Wan LJ, Bai CL (2003) Chem Phys Chem 4:1114

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection Hans Peter Lang · Martin Hegner · Christoph Gerber

List of Abbreviations AFM MW PCA PEEK PDMS PSD RIE SFM VCSEL µFN

atomic force microscopy molecular weight principal component analysis poly-etheretherketone poly-dimethyl-siloxane position sensitive detector reactive ion etching scanning force microscopy vertical-cavity surface-emitting laser microﬂuidic network

28.1 Introduction 28.1.1 Sensors A sensor is a device that detects, or senses, a signal. A sensor is also a transducer, i.e. it transforms one form of energy into another or responds to a physical parameter. Transducers can be electrochemical (pH probe), electromechanical (piezoelectric actuator, quartz, strain gauge), electroacoustic (gramophone pickup, microphone), photoelectric (photodiode, solar cell), electromagnetic (antenna, photocell, tape or harddisk head for storage applications), magnetic (Hall effect sensor), electrostatic (electrometer), thermoelectric (thermocouple, thermo-resistors), electrical (capacitor, resistor) or mechanical (deﬂection sensors). Sensors either directly indicate a state or a value (mercury thermometer), or use an indicator that indirectly displays the state or value by an analog or digital converter (display, computer). (Bio-)chemical sensors convert changes of physical or chemical parameters due to the presence of molecules in the environment into a recordable signal (Fig. 28.1).

184

H.P. Lang · M. Hegner · C. Gerber

Fig. 28.1. Schematic of general sensor transduction. Analyte molecules present in the environment are recognized by the sensing layer of the sensor. Either recognition is very speciﬁc, i.e. recognition sites on immobilized ligand molecules recognize the analyte molecules, or recognition is partially speciﬁc, e.g. the analyte molecules diffuse into a polymer layer. The binding event is then transduced into a recordable signal via a variety of transduction mechanisms. The acquired signal is then further ampliﬁed and processed

28.1.2 Cantilevers The term cantilever is understood here as a microfabricated rectangular bar-shaped structure that is longer than it is wide and has a thickness that is much smaller than its length or width. A cantilever is a horizontal structural element supported only at one end on a chip body; the other end is free. At the free end, a microfabricated sharp tip might be attached for use as a local probe to scan a surface. Cantilever beams have been applied since the mid 1980s as sensitive structures to measure interatomic forces in the piconewton range using a technique called scanning force microscopy (SFM) or atomic force microscopy (AFM) [1]. In this method, a cantilever with a sharp tip is scanned across a conductive or nonconductive surface by the use of an x-y-z actuator system (e.g. a piezoelectric scanner). Scanning is performed in a pattern of adjacent parallel lines to cover a rectangular or square area of the sample surface. Typical scanning speeds range from 1 nm/s to several micrometers per second, depending on the size of the area to be scanned. The tip can either be in direct contact with the surface (contact mode) or oscillated to interact with the surface only for a short time during the oscillation cycle (dynamic mode, noncontact mode). Common to SFM methods is the interaction of the cantilever tip with the surface. This interaction can be used to control a feedback loop intended to keep the force or force gradient between cantilever tip and sample surface constant. By recording the correction signal that has to be applied to the z-actuation drive to keep the interaction between tip and sample surface constant, a topography image of the

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

185

sample surface can be obtained. SFM methods are nowadays well-established in scientiﬁc research, education and to a certain extent also in the industry. Beyond imaging of surfaces, other applications of cantilevers have been demonstrated. The use of a sharp tip at the cantilever apex is not required for application as the (bio-)chemical sensor described here. Because of their ﬂexibility, cantilevers can not only be used for probing the surface proﬁle of a sample, but also to monitor processes taking place on the surface of the beam [2]. The use of cantilever beams as sensors, clamped at one end and freestanding at the other, allows the adsorption of molecules to be observed with unprecedented accuracy, because the formation of molecule layers on the cantilever surface will generate surface stress. This eventually results in a bending of the cantilever, provided the adsorption preferentially occurs on one surface of the cantilever. Adsorption is controlled by coating one surface (e.g. the upper surface) of a cantilever with a thin layer of a material that shows afﬁnity to molecules in the environment (sensor surface). This surface of the cantilever is referred to as the “functionalized” surface. The other surface of the cantilever (e.g. the lower surface) may be left uncoated or be coated with a passivation layer, i.e. a chemical surface that does not exhibit signiﬁcant afﬁnity to the molecules in the environment to be detected (see Fig. 28.2). To facilitate the establishment of functionalized surfaces, a metal layer is often evaporated onto the surface designed as sensor surface. Metal surfaces, e.g. a gold surface, may be utilized to covalently bind a monolayer that represents the chemical surface sensitive to the molecules to be detected from environment. A typical example is a monolayer of thiol molecules covalently bound to a gold surface. The gold layer is also favorable for use as a reﬂection layer, if the bending of the cantilever is read out via an optical beamdeﬂection method.

Fig. 28.2. Schematic sideview representation of a functionalized cantilever: (1) cantilever beam with a typical thickness of 1 µm, typically made from silicon, (2) chip body with a typical thickness of 500 µm, (3) 2 nm thick evaporated titanium layer required for adherence of (4) a 20 nm thick evaporated gold layer, (5) functional sensing layer, e.g. 1 nm thick self-assembled thiol monolayer, (6) passivation layer (several nanometers thick), e.g. a silanized polyethyleneglycol layer, (7) target molecules from the environment, (8) area in which target molecules have docked to the functionalized surface

H.P. Lang · M. Hegner · C. Gerber

186

28.1.3 Cantilever Operating Modes 28.1.3.1 Static Mode The continuous bending of a cantilever as a function of molecular coverage with molecules is referred to as an operation in the “static mode”, see Fig. 28.3. Adsorption of molecules onto the functional layer generates stress at the interface between the functional layer and the forming molecular layer. The stress is transduced towards the site at which the molecules of the functional layer are attached to the cantilever surface. Because the forces within the functional layer try to keep the distance between molecules constant, the cantilever beam responds with bending due to its extreme ﬂexibility. The resulting surface stress change is calculated according to Stoney’s formula [3]: ∆σ = Et 2 /[4R(1 − ν)] ,

(28.1)

where E is Young’s modulus (E Si = 1.3 × 1011 N/m2 for Si(100)), t the thickness of the cantilever, ν Poisson’s ratio (νSi = 0.25), and R the bending radius of the cantilever. The static-mode operation can be performed in various environments. In its simplest embodiment, molecules from gaseous environment adsorb on the functionalized sensing surface and form a molecular layer (Fig. 28.3a), provided

Fig. 28.3. (a) Static-mode operation. An adsorbing layer of molecules produces a change in surface stress on the upper functionalized surface of the cantilever, resulting in bending of the cantilever. (b) Diffusion of target molecules from the environment into a polymer layer induces swelling of the polymer layer and thereby a bending of the cantilever, because the swollen polymer layer expands more than the silicon cantilever. (c) Static-mode operation in liquids. Target molecules from the liquid environment are molecularly recognized by binding sites on the functionalized sensing layer on the upper surface of the cantilever

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

187

that the molecules exhibit some afﬁnity to the surface. Polymer sensing layers show a partial sensitivity, because molecules from the environment diffuse into the polymer layer at different rates, mainly depending on the size and solubility of the molecules in the polymer layer (Fig. 28.3b). A wide range of hydrophilic/hydrophobic polymers can be selected, differing in their suitability for polar/unpolar molecules. Thus, the polymers can be chosen according to what the applications demand. Static-mode operation in liquids, however, usually requires rather speciﬁc sensing layers, based on molecular recognition, such as DNA hybridization or antigenantibody recognition (Fig. 28.3c). 28.1.3.2 Dynamic Mode Because the bending of the cantilever is a direct result of the adsorbtion of molecules onto the cantilever surface, it is rather difﬁcult to obtain reliable information on the amount of molecules adsorbed, as the surface coverage is basically not known. In addition, molecules on the surface might be exchanged with molecules from the environment in a dynamic equilibrium. However, mass changes can be determined accurately by operation of a cantilever actuated at its eigenfrequency. The eigenfrequency is equal to the resonance frequency of an oscillating cantilever if the elastic properties of the cantilever remain unchanged during the molecule adsorption process and damping effects are negligible. This operation mode is called the dynamic mode (e.g., the use as a microbalance, Fig. 28.4a). Owing to mass addition on the cantilever surface, the cantilever’s eigenfrequency will shift to a lower value. The mass change on a rectangular cantilever is calculated [4] according to (28.2) ∆m = (k/4π 2 ) × 1/ f 12 − 1/ f 02 , where f 0 is the eigenfrequency before the mass change occurs, and f 1 the eigenfrequency after the mass change. The spring constant k of the cantilever is obtained using k = Et 3 w/4l 3 ,

(28.3)

where l, w, and t denote the length, width, and thickness of the cantilever, respectively. Mass-change determination can be combined with varying environment temperature conditions (Fig. 28.4b) to obtain the method introduced in the literature as “micromechanical thermogravimetry” [5]. The sample to be investigated has to be mounted at the apex of the cantilever. Its mass should not exceed several hundred nanograms. In the case of adsorption/desorption/decomposition processes, mass changes in the picogram range can be observed in real time by tracking the resonance-frequency shift. Dynamic mode operation in liquid environment poses problems, such as high damping of the cantilever oscillation due to high viscosity of the surrounding media.

H.P. Lang · M. Hegner · C. Gerber

188 Fig. 28.4. (a) Cantilever operation in dynamic mode. The cantilever is excited at its resonance frequency. Tracking the resonance frequency allows one to determine mass changes. (b) Micromechanical thermogravimetry. Mass changes are measured in dependence of temperature as an external parameter. (c) Detection of biochemical processes involving mass changes due to adsorption and binding of molecules

This results in a low quality factor Q of the oscillation, and the resonance frequency shift is difﬁcult to track with high resolution. The quality factor is deﬁned as Q = 2∆ f/ f 0 .

(28.4)

Whereas in air a frequency resolution of below 1 Hz is easily achieved, resolution values of about 20 Hz are already considered very good for measurements in a liquid environment. 28.1.3.2.1 Operation in a Large Damping Environment In the case of damping or changes of the elastic properties of the cantilever during the experiment, e.g. a stiffening or softening of the spring constant by adsorption of a molecule layer, the measured resonance frequency will not be exactly the same as the eigenfrequency, and the mass derived from the frequency shift will be inaccurate. In a medium, the vibration of a cantilever is described by the model of a driven damped harmonic oscillator: m∗

d2 x dx +γ + kx = F cos(2π ft) , 2 dt dt

(28.5)

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

189

where m ∗ = const.(m c + m l ) is the effective mass of the cantilever (for a rectangular cantilever the constant is 0.25). Especially in liquids, the mass of the co-moved liquid m l adds signiﬁcantly to the mass of the cantilever m c . The term γ dx dt is the drag force due to damping, F cos(2π ft) is the driving force executed by the piezooscillator, and k is the spring constant of the cantilever. If no damping is present, the eigenfrequencies of the various oscillation modes of a bar-shaped cantilever are calculated according to: k αn2 fn = , (28.6) 2π 3(m c + m l ) where f n are the eigenfrequencies in the n-th mode, αn are constants depending on the mode: α1 = 1.8751, α2 = 4.6941, αn = π(n − 0.5); k is the spring constant of the cantilever, m c the mass of the cantilever, and m l the mass of the medium surrounding the cantilever, e.g. liquid [6, 7]. If mass is added to the cantilever due to adsorption, the effective mass is calculated according to m ∗ = const.(m c + m l + ∆m) , where ∆m is the additional mass adsorbed. Typically, the co-moved mass liquid is much larger than the adsorbed mass. Figure 28.5 clearly shows that the resonance frequency is only equal eigenfrequency if no damping is present. With damping, the frequency at the peak of the resonance curve occurs is no longer identical with that at

Fig. 28.5. (a) Resonance curve with no damping (0), and increasing damping (1)– (3). The undamped curve with resonance frequency f 0 shows a very high amplitude, whereas the resonance peak amplitude decreases with damping. This also involves a shift in resonance frequencies f 1 to f 3 to lower values. (b) Corresponding phase curves showing no damping (0), and increasing damping (1)–(3). The step-like phase jump at resonance of the undamped resonance gradually broadens with increasing damping

(28.7) of the to the which which

H.P. Lang · M. Hegner · C. Gerber

190

the turning point of the phase curve occurs. For example, resonance curve 2 with damping γ2 shows its maximum amplitude at frequency f 2 . The corresponding phase would be ϕres (γ2 ), which is not π/2. If direct resonance frequency tracking or a phase-locked loop is used to determine the frequency of the oscillating cantilever, then only its resonance frequency is detected, but not its eigenfrequency. Remember that the eigenfrequency, and not the resonance frequency, is required to determine mass changes. 28.1.3.3 Heat Mode If a cantilever is coated with metal layers, thermal expansion differences in cantilever and coating layer will further inﬂuence cantilever bending as a function of temperature. This mode of operation is referred to as ‘heat mode’ and causes cantilever bending because of differing thermal expansion coefﬁcients in the sensor layer and cantilever materials [2] (Fig. 28.6): ∆z = 1.25 × (α1 − α2 ) × (t1 + t2 )/t22 κ × l 3 P/(α1 t1 + α2 t2 )w .

(28.8)

Here α1 , α2 are the thermal expansion coefﬁcients of the cantilever and coating materials, t1 , t2 the material thicknesses, P is the total power generated on the cantilever, and κ a geometry parameter of the cantilever device.

Fig. 28.6. (a) Cantilever heat mode. If a material with a different thermal expansion coefﬁcient is evaporated as a thin layer onto the surface of a cantilever, then the cantilever bends when the external temperature is changed. (b) The heat change does not have to be generated by a change of temperature, but may also originate from heat production or drain during exothermal or endothermal reactions taking place on the cantilever surface. Exothermal processes are, for example, catalytic reactions. (c) If a sample is attached to the cantilever apex, a calorimetric experiment can be performed on the sample

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

191

Heat changes are either caused by external inﬂuences (change in temperature, Fig. 28.6a), occur directly on the surface by exothermal, e.g. catalytic, reactions (Fig. 28.6b), or are due to material properties of a sample attached to the apex of the cantilever (micromechanical calorimetry, Fig. 28.6c). The sensitivity of the cantilever heat mode is orders of magnitude higher than that of traditional calorimetric methods performed on milligram samples, as it only requires nanogram amounts of sample and achieves nanojoules [8] to picojoules [9, 10] of sensitivity. These three measurement modes have established cantilevers as versatile tools to perform experiments in nanoscale science with very small amounts of material. 28.1.3.4 Other Measurement Modes 28.1.3.4.1 Photothermal Spectroscopy When a material adsorbs photons, a fraction of energy is converted into heat. This photothermal heating can be measured as a function of the light wavelength to provide optical absorption data of the material (Fig. 28.7). The interaction of light with a bimetallic cantilever creates heat on the cantilever surface, resulting in a bending of the cantilever [11]. Such bimetallic-cantilever devices are capable of detecting heat ﬂows due to an optical heating power of 100 pW, which is two orders of magnitude better than in conventional photothermal spectroscopy.

Fig. 28.7. Bending of a bimetallic cantilever due to heat production by incident photons

28.1.3.4.2 Electrochemistry A cantilever coated with a metallic layer (measurement electrode) on one side is placed in an electrolytic medium, e.g. a salt solution, together with a metallic reference electrode, usually made of a noble metal (Fig. 28.8). Variations of the voltage between measurement and reference electrode induce electrochemical processes on the measurement electrode (cantilever), e.g. adsorption or desorption of ions from the electrolyte solution onto the measurement electrode. These processes lead to a bending of the cantilever due to changes in surface stress and in the electrostatic forces.

192

H.P. Lang · M. Hegner · C. Gerber

Fig.28.8.A cantilever coated on one side with a metallic layer that is connected to a voltage source is immersed in an electrolytic ﬂuid (E). The other pol e of the voltage source is connected to a reference electrode (R). Depending on the polarity and the magnitude of the voltage applied to the cantilever and reference electrode, ions from the electrolyte solution adsorb onto or desorb from the cantilever

28.1.3.4.3 Detection of Electrostatic and Magnetic Forces The detection of electrostatic and magnetic forces is possible if charged or magnetic particles are deposited on the cantilever [12–14] (Fig. 28.9). If the cantilever is brought into the vicinity of electrostatic charges or magnetic particles, attractive or repulsion forces occur according to the polarity of the charges or magnetic particles present on the cantilever. These forces will result in an upwards or downwards bending of the cantilever. The magnitude of bending depends on the distribution of the charged or magnetic particles on both the cantilever and the surrounding environment according to the laws of electromagnetism.

Fig. 28.9. (a) Cantilever bending due to electrostatic forces. (b) Cantilever bending due to magnetic forces

28.1.4 Cantilever Arrays 28.1.4.1 Disadvantages of Single-Cantilever Measurements The bending response of a single cantilever is often inﬂuenced by various undesired effects, such as thermal drift and unspeciﬁc reactions taking place on the uncoated

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

193

cantilever surface, resulting in additional cantilever bending. To avoid such effects, we have introduced measurements using reference cantilevers [15], i.e. cantilever sensors that will not react with the target analyte molecules (Fig. 28.10). As the difference in signals from the reference and sensor cantilevers shows the net cantilever response, even small sensor responses can be extracted from large cantilever deﬂections without being predominated by undesired effects.

Fig. 28.10. (a) Single cantilever: no thermal-drift compensation is possible, both surfaces have to be chemically well-deﬁned. One of the surfaces, e.g. the lower one, has to be passivated, otherwise the cantilever response will be convoluted with undesired effects originating from uncontrolled reactions taking place on the lower surface. (b) Dual cantilevers: one of them is used as the sensor cantilever (coated e.g. on the upper side with a molecule layer exhibiting afﬁnity to the molecules to be detected), the other as the reference cantilever, coated with a passivation layer on the upper surface (exhibiting no afﬁnity to the molecules to be detected). Thermal drifts are cancelled out if difference responses, i.e. differences in deﬂections of sensor and reference cantilevers are taken. Alternatively, both cantilevers are used as sensor cantilevers (sensor layer on the upper surfaces), and the lower surface is passivated. (c) Cantilever array: several cantilevers are used either as sensor cantilevers or as reference cantilevers so that multiple difference signals are possible simultaneously. Thermal drift is cancelled out as one surface of all cantilevers, e.g. the lower one, is left uncoated or coated with the same passivation layer

28.1.4.2 Cantilever Array Designs Here, cantilever array designs are presented that have been used in the cantilever array sensor groups of IBM’s Zurich Research Laboratory in Rüschlikon (Switzerland) and of the Institute of Physics, University of Basel (Switzerland). Accordingly, designs used by other research groups will not be discussed. The ﬁrst design in 1996 (Fig. 28.11a) consisted of a thick chip-body part (silicon, 500 µm thick, shown in grey in the ﬁgure), a doped epitaxial silicon layer for the cantilevers (1 µm thick, white), and a silicon top layer (2 µm thick, dark grey) for labeling purposes. The structures deﬁning the chip body, the cantilevers and the top labeling layers are transferred to the silicon surfaces by means of photolithographic masks. The chips are thinned down from the lower side in an anisotropic wetetching process using the different doping levels of the cantilevers as an etch stop. A second wet-etch step is applied from the top to release the cantilevers from the top and to etch the marks that are visible from the top. The uniformity of the cantilevers in one chip turned out to be within a few percent of the average resonance frequency. The wet-etch process steps exert considerable mechanical stress on the cantilever during production, so that defects or microcracks on some of the cantilever arrays in the wafer cannot be excluded. As all cantilevers are connected through the

194

H.P. Lang · M. Hegner · C. Gerber

Fig. 28.11. (a) Cantilever array design, IBM 1996. (b) Cantilever array design, IBM 2000. (c) Cantilever array design, IBM 2003. Scanning electron micrographs on the left, side and top views on the right. The hatched area is a planned modiﬁcation (not shown in the photo). Its purpose is to provide a rigid reference cantilever (full wafer thickness)

same membrane, crosstalk between responses from individual cantilevers cannot be completely excluded, although it was found to be only a few percent of the individual cantilever deﬂection signals. To circumvent such crosstalk, the 2000 design was developed, which features special support structures for each cantilever, so that each cantilever is individually attached to the chip body (Fig. 28.11b). Whereas the chip body and the support structures are wet-etched from the lower side, the cantilevers are deﬁned using dry-etching by reactive ion etching (RIE). This method produces high-aspect-ratio structures (grooves with sidewall angles larger than 85◦ ). To avoid time-critical wet-etching processes and the risk of damage to the cantilevers during the production process, the 2003 design relies entirely on dry-etching processes. The wall steepness has been increased to 89◦ by the use of improved RIE equipment, so that the etching from the lower side can also be performed by dryetching, even if as much as 500 µm of silicon have to be etched off from the lower side. To further avoid crosstalk and to facilitate functionalization of the cantilevers, they are ﬁxed to an extended (500-µm-long) bar-shaped structure (Fig. 28.11c).

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

195

28.1.4.3 Uniformity of Cantilever Arrays Even if cantilever arrays consist of identical cantilevers by design, the properties might differ slightly from one cantilever to the next. For this reason, several test procedures are suggested to ascertain the uniformity of the cantilever arrays utilized in experiments. 28.1.4.3.1 Resonance-Frequency Investigation The resonance frequency of a cantilever is basically determined by the material it consists of as well as its length, width and thickness. Measurement of the resonance frequencies in an uncoated cantilever array typically yields an absolute difference of less than 10 Hz at a typical resonance frequency of 4 kHz (1-µm-thick, 500-µm-long, 100-µm-wide cantilevers), which corresponds to a resonance frequency deviation of less than 0.25%. 28.1.4.3.2 Comparison of Responses to a Heat Pulse One way to verify the homogeneity of gold-coated cantilevers is to observe the bending response of a bimetallic cantilever when a heat pulse is applied using a Peltier heater beneath the cantilever array. Figure 28.12 shows the response of eight cantilevers in an array to a heat pulse of 60 s duration (double arrow). The deviation in the deﬂection of the individual cantilevers is less than 9% of the maximum deﬂection observed. This gives a rough estimate of what is expected of cantilever responses in sensing experiments. However, the thermal (bimetallic) response of cantilevers may differ from the (bio-)chemical response, because the sensing layer might cause additional differences between cantilevers.

Fig. 28.12. Heat pulse of 0.71 V and 0.32 A applied for a duration of 60 s to an array of goldcoated cantilevers to test their uniformity. The two horizontal lines indicate smallest and largest cantilever deﬂections obtained from the heat pulse (data by J. Zhang, Univ. Basel, Switzerland)

196

H.P. Lang · M. Hegner · C. Gerber

28.2 Experimental Setup 28.2.1 Measurement Chamber 28.2.1.1 Designs for Gas and Liquid Environments A measurement setup for cantilever arrays consists of four major parts: 1. the measurement chamber containing the cantilever array, 2. an optical or piezoresistive system to detect the cantilever deﬂection (e.g. laser sources, collimation lenses and a position-sensitive detector (PSD), 3. electronics to amplify, process and acquire

Fig. 28.13. Schematic of measurement setups for (a) a gaseous (artiﬁcial nose) and (b) a liquid environment (biochemical sensor)

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

197

the signals from the PSD, and 4. a gas- or liquid-handling system to reproducibly inject samples into the measurement chamber and purge the chamber. Figure 28.13 shows the schematic setup of the experiments performed in (a) a gaseous and (b) a liquid (biochemical) environment. The cantilever sensor array is located in an analysis chamber of 3–90 µl in volume, which has inlet and outlet ports for gases or liquids. The cantilever deﬂection is determined by means of an array of eight vertical-cavity surface-emitting lasers (VCSELs) arranged at a linear pitch of 250 µm that emit at a wavelength of 760 nm into a narrow cone of 5 to 10◦ . The light of each VCSEL is collimated and focused onto the apex of the corresponding cantilever by a pair of achromatic doublet lenses, 12.5 mm in diameter. This size was selected so that all eight laser beams pass through the lenses close to its center in order to minimize scattering, chromatic and spherical aberration artifacts. The light is then reﬂected off the gold-coated surface of the cantilever and hits the surface of a PSD. PSDs are light-sensitive photo-potentiometer-like devices that produce photocurrents at two opposing electrodes. The magnitude of the photocurrents linearly depends on the distance of the impinging light spot from the electrodes. Thus the position of an incident light beam can be determined with micrometer precision. The photocurrents are transformed into voltages and ampliﬁed in a preampliﬁer (see Fig. 28.14). As only one PSD is used, the eight lasers cannot be switched on simultaneously. Therefore, a time-multiplexing procedure is used to switch the lasers on and off sequentially at typical intervals of 10–100 ms (see Fig. 28.15). The resulting

Fig. 28.14. Schematic circuit drawing of a current-to-voltage converter and preampliﬁer that produces difference and sum signals. The PSD is biased at +15 V to decrease its capacitance and enhance its response. The photocurrents I1 and I2 are fed into a current-to-voltage (I–V ) converter stage to obtain voltages V1 and V2 . These are further processed to yield a difference signal V2 − V1 , which is used to measure cantilever deﬂection, and a sum signal V1 + V2 , which measures the total intensity of light on the PSD and serves as a reference value to keep the intensity of the VCSELs constant (courtesy of A. Tonin, Univ. Basel, Switzerland)

198

H.P. Lang · M. Hegner · C. Gerber

Fig. 28.15. Schematic circuit drawing of the VCSEL driver to switch on and off the eight VCSELs and to keep their light intensity constant. The light of a VCSEL impinges on the PSD and produces photocurrents I1,2 that are converted into voltages V1,2 in the preampliﬁer (I to V ). The sum signal V1 + V2 is fed into a voltage comparator, which compares the actual value of the sum signal with a preset reference value. The difference between reference and actual value is then used as input to a PI controller that regulates the current to drive the VCSELs at constant intensity (V to I). A multiplexer switches all eight VCSELs on and off sequentially (courtesy of A. Tonin, Univ. Basel, Switzerland)

deﬂection signal is digitized and stored together with time information on a personal computer (PC), which also controls the multiplexing of the VCSELs as well as the switching of the valves and mass ﬂow controllers used for setting the composition ratio of the analyte mixture. The measurement setup for liquids (Fig. 28.13b) consists of a poly-etheretherketone (PEEK) liquid cell, which contains the cantilever array and is sealed by a viton O-ring and a glass plate. The VCSELs and the PSD are mounted on a metal frame around the liquid cell. After preprocessing the position of the deﬂected light beam in a current-to-voltage converter and ampliﬁer stage (Fig. 28.14), the signal is digitized in an analog-to-digital converter and stored on a PC. The liquid cell is equipped with inlet and outlet ports for liquids. They are connected via 0.18 mm of i.d. teﬂon tubing to individual thermally-equilibrated glass containers, in which the biochemical liquids are stored. A six-position valve allows the inlet to the liquid chamber to be connected to each of the liquid-sample containers separately. The liquids are pulled (or pushed) through the liquid chamber by means of a syringe pump connected to the outlet of the chamber. A Peltier element is situated very close to the lumen of the chamber to allow temperature regulation within the chamber. The entire experimental setup is housed in a temperature-controlled box regulated with an accuracy of 0.01 K to the target temperature. 28.2.2 Cantilever Functionalization To serve as sensors, cantilevers have to be coated with a sensor layer that is either highly speciﬁc, i.e. is able to recognize target molecules in a key-lock process,

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

199

or partially speciﬁc, so that the sensor information from several cantilevers yields a pattern that is characteristic of the target molecules. To provide a platform for speciﬁc functionalization, the upper surface of these cantilevers is generally coated with 2 nm of titanium and 20 nm of gold, which yields a reﬂective surface and an interface for attaching functional groups of probe molecules, e.g. for anchoring molecules with a thiol group to the gold surface of the cantilever. Such thin metal layers are believed not to contribute signiﬁcantly to the bending due to surface-stress changes, because the temperature is kept constant. Many examples of molecular adsorption on cantilevers are described in the literature, for example, the adsorption of alkyl thiols on gold [16, 17], the detection of mercury vapor and relative humidity [18], dye molecules [19], monoclonal antibodies [20], sugar and proteins [21], solvent vapors [22–24], fragrance vapors [25], as well as the pH-dependent response of carboxy-terminated alkyl thiols [26], label-free DNA hybridization detection [27, 28], and biomolecular recognition of proteins relevant in cardiovascular diseases [29]. 28.2.2.1 Functionalization Methods There is a large variety of both simple and more advanced methods to coat a cantilever with material. The requirement is to develop a fast, reproducible and reliable method for the functionalization of one or both of the surfaces of a cantilever separately. 28.2.2.1.1 Simple Methods Obvious methods to coat a cantilever include thermal or electron-beam-assisted evaporation of material, electro-spray or other deposition methods. The disadvantage of these methods is that inherently they do not provide sufﬁcient resolution to coat individual cantilevers in an array, unless shadow masks are used. Such masks need to be accurately aligned with the cantilever structures, which is a time-consuming process. Other methods to coat cantilevers use manual placement of particles onto the cantilever [2,5,8,19,30], which requires skillful handling of tiny samples. Cantilevers can also be coated by directly pipetting solutions of the probe molecules onto the cantilevers [22], or by means of air-brush spraying and shadow masks to coat the cantilevers separately [24]. All these methods only have limited reproducibility if a larger number of cantilever arrays has to be coated. As manual alignment is required, these methods are very time-consuming. 28.2.2.1.2 Microﬂuidic Networks A step towards reliable and reproducible coating procedures is achieved by the use of microﬂuidic networks (µFN) [31]. µFN are structures of channels and wells, etched

200

H.P. Lang · M. Hegner · C. Gerber

several tens to hundreds of micrometers deep into silicon-wafers. The wells can be ﬁlled easily using a laboratory pipette, so that the ﬂuid with the probe molecules to be attached to the cantilever is guided through the channels towards openings at a pitch matched to the distance between individual cantilevers in the array (see Fig. 28.16). The cantilever array is then approached to the opening by means of an x-y-z positioner, with an accuracy of better than 10 µm, and introduced into the open channels of the µFN that are ﬁlled with a solution of the probe molecules. The incubation of the cantilever array in the channels of the µFN takes from a few seconds (self-assembly of alkanethiol monolayers) to several tens of minutes (coating with protein solutions). To prevent evaporation of the solutions, the channels are covered by a slice of PDMS. In addition, the microﬂuidic network can be placed in a container ﬁlled with saturated vapor of the solvent used for the probe molecules. The obtained functional layers on the cantilevers are of good and reproducible quality, and the coating process is considerable faster than with the above-described simple techniques.

Fig. 28.16. Microﬂuidic network for functionalization of cantilever arrays. (a) Schematic, (b) side view, (c) photo of the actual device. (1) Cantilever array, (2) reservoir wells, (3) microﬂuidic network with channels, (4) PDMS cover to avoid evaporation of the liquid, (5) pipette to ﬁll the reservoir well. Figure courtesy of A. Bietsch, Univ. Basel, Switzerland

28.2.2.1.3 Array of Dimension-Matched Glass Capillaries As the positioning of the cantilever array relative to the µFN still requires manual alignment, we have developed a device in which the cantilever array is placed rigidly

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

201

Fig. 28.17. Setup for cantilever array functionalization using an array of dimension-matched microcapillaries. (a) Schematic, (b) side view, (c) photo. (1) Cantilever array, (2) glass capillaries, (3) tubing with liquid reservoir in a larger piece of tubing to compensate for liquid losses due to evaporation. Figure courtesy of A. Bietsch, Univ. Basel, Switzerland

in a ﬁt with an accuracy of better than 50 µm and ﬁxed with a clamp (Fig. 28.17). Because the µFN needs to be cleaned thoroughly after each functionalization process, we replaced the µFN channels by an array of dimension-matched disposable glass capillaries. The outer diameter of the glass capillaries is 240 µm so that they can be placed neatly next to each other to accommodate the pitch of the cantilevers in the array (250 µm). Their inner diameter is 150 µm, providing sufﬁcient room to insert the cantilevers (width: 100 µm) safely.

Fig. 28.18. Cantilever array (a) before and (b) after insertion into the open ends of an array of microcapillaries. The different shades of grey in the capillaries are due to the ﬁlling of the capillaries with various dyes. Note that the capillaries can only be moved until they touch the wafer-thick ﬁnger structures (see Fig. 28.11c). Courtesy of A. Bietsch, Univ. Basel, Switzerland

202

H.P. Lang · M. Hegner · C. Gerber

A detailed view of how the cantilevers in an array are inserted into an array of glass capillaries is shown in Fig. 28.18. This method has been successfully applied for the deposition of a variety of materials onto cantilevers, e.g. for polymer solutions [24], self-assembled monolayers [26], thiol-functionalized single-stranded DNA oligonucleotides [28], and protein solutions [29]. 28.2.2.1.4 Inkjet Spotting Even if the capillary-array technique yields reliable and reproducible layers that can be deposited within minutes, it is not ideal for coating a large number of cantilever arrays. Therefore, we applied the inkjet printing technique to establish a rapid and general method to coat cantilever arrays [32, 33] (Fig. 28.19). This technique is scalable to large arrays and can also coat arbitrary structures in noncontact manner. An x-y-z positioning system allows a sharp nozzle (capillary diameter: 70 µm) to be positioned with an accuracy of approx. 10 µm over a cantilever. Individual droplets (diameter: 60–80 µm, volume 0.1–0.3 nl) can be dispensed individually by use of a piezo-driven ejection system in the inkjet nozzle. When the droplets are

Fig. 28.19. Cantilever array functionalization by inkjet spotting. (a) Schematic, (b) photo of the actual device, (c) closeup. (1) Cantilever array, (2) inkjet nozzle, (3) x-y-z positioning unit. Courtesy of A. Bietsch, Univ. Basel, Switzerland

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

203

spotted with a pitch smaller than 0.1 mm, they merge and form continuous ﬁlms. By adjusting the number of droplets deposited on cantilevers, the resulting ﬁlm thickness can be controlled precisely. The inkjet-spotting technique allows a cantilever to be coated within seconds and yields very homogeneous, reproducibly deposited layers of well-controlled thickness. Successful coating of self-assembled alkanethiol monolayers, polymer solutions, self-assembled DNA single-stranded oligonucleotides [33], and protein layers has been demonstrated. In conclusion, inkjet spotting has turned out to be a very efﬁcient method for functionalization that is not restricted to cantilevers, but can easily and rapidly coat even arbitrarily shaped sensors reproducibly and reliably [34, 35]. 28.2.2.2 Functionalization Procedure 28.2.2.2.1 Polymer-Coated Cantilever Arrays Polymer layers were deposited on a gold-coated cantilever array by inkjet spotting. Various commercial polymers were dissolved in solvents (5 mg/ml). Eight droplets of solution were dispensed onto the upper surface of one of the cantilevers of the array by inkjet spotting [32, 33], allowed to dry and form a homogeneous polymer layer of a few hundred nanometers in thickness. The following polymers were used: CMC: carboxymethylcellulose sodium salt in water (all concentrations 5 mg/ml); PEG: polyethylene glycol 6000 in water; PEI: polyethyleneimine in water; PSS: poly(sodium 4-styrene sulfonate), MW 70,000 in water; PAAM: poly(allylamine hydrochloride), MW 15,000 in water; PVP: poly(2-vinylpyridine) standard, MW 64,000 in ethanol; PVA: polyvinylalcohol 10-98 in dimethylsulfoxide; PMMA: polymethymethacrylate, MW 15,000 in methyl-isobutyl-ketone. 28.2.2.2.2 DNA-Coated Cantilever Arrays For measurements in the liquid phase, functional monolayers were deposited by self-assembly of thiolated molecule layers, e.g. DNA oligonucleotide layers for hybridization experiments in liquid. Various oligonucleotide layers were applied in parallel and under identical conditions to the individual cantilever sensors by means of microcapillaries (see Fig. 28.17 and Fig. 28.18), ﬁlled with a 40-µM solution of 3 - or 5 -thiolated probe DNA in triethyl ammonium acetate buffer for 20 min. The coated arrays were rinsed in sodium saline citrate (ssc) 5× buffer and dried in nitrogen [28].

28.3 Measurements Measurements using cantilever array sensors can either be highly speciﬁc to detect molecules that match a receptor site on the cantilever coating or partially speciﬁc,

204

H.P. Lang · M. Hegner · C. Gerber

i.e. several cantilever sensors respond to the presence of a vapor, but only the pattern of responses of all eight cantilevers contains sufﬁcient information to characterize a sample vapor. In the following, two experiments are presented. The ﬁrst demonstrates the partial sensitivity of polymer-coated cantilevers as a kind of “artiﬁcial nose” [36] in a gaseous environment, the other shows the high speciﬁcity of a DNA hybridization recognition reaction in liquid. 28.3.1 Artiﬁcial Nose for Detection of Perfume Essences A cantilever array consisting of eight differently coated cantilevers can be employed as an “artiﬁcial nose” to characterize vapors. The key elements of an artiﬁcial nose [36] are 1. chemical sensors composed of a physical transducer and a chemical interface layer or a receptor domain with partial speciﬁcity; 2. an appropriate patternrecognition system to recognize simple or complex odors (pattern classiﬁer), and 3. a sampling system to reproducibly perform measurements. The sensor array is exposed to the same sample, and produces individual responses, as well as a pattern of responses. The main advantage of an artiﬁcial nose is that it is reproducible, does not wear out, and can be placed in environments that are harmful to humans. Here we use polymer-coated cantilevers as chemical sensors prepared as described above. Detection of vapors proceeds via diffusion of the vapor molecules into the polymer, resulting in a swelling of the polymer and bending of the cantilever. The bending is speciﬁc to the interaction between the solvent vapor and polymer time- and magnitude-wise. To demonstrate the capability of the cantilever array artiﬁcial nose, we injected vapor (20 ml/min) from the headspace above 0.1 ml of ethanolic solution of perfume essence samples (lemon, wood, ﬂower; main components: ethanol 35%, water 14%, dipropylene glycol 50%, fragrances < 1%) in a stream of dry nitrogen. Figure 28.20 shows the deﬂection traces from eight polymercoated cantilevers upon injection of perfume essence vapor during 60 s (starting at t = 100 s). The cantilever responses of three samples are shown (two wood essence and one lemon essence sample). A “ﬁngerprint” of the perfume essence sample is generated by reducing the deﬂection curves of all eight cantilevers to a discrete set of cantilever amplitudes at several points in time, e.g. at 110 s, 130 s, and 150 s. This yields in total 8 × 3 = 24 cantilever deﬂection amplitudes that account for a measurement data set. This data set is then evaluated using principal-component-analysis (PCA) techniques [37], which extracts the most dominant deviations in the responses for the various perfume essence vapors. The largest differences in signal amplitudes of the ﬁngerprint patterns are plotted in a two-dimensional graph, whereby an individual measurement (i.e. a set of 24 cantilever magnitudes) represents a single point in the PCA space. The axes refer to projections of the multidimensional datasets into two dimensions (principal components). This procedure is targeted at maximum distinction performance between analytes, i.e. several measurements of the same analyte should yield a cluster in principal-component space, whereas measurements of differing analytes should produce well-separated clusters of measurements.

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection Fig. 28.20. Bending patterns of polymercoated cantilever sensors upon exposure to perfume essence vapor. (a) Wood ﬂavor (ﬁrst sample), (b) wood ﬂavor (second sample), (c) lemon ﬂavor. “Purge” means ﬂushing the measurement chamber with dry nitrogen gas at a rate of 20 ml/min, “Inj.” denotes the injection of dry nitrogen gas saturated with the vapor from the headspace of a vial ﬁlled with 100 µl of ethanolic solution of perfume essence. The cantilever deﬂection responses are offset for clarity reasons

205

206

H.P. Lang · M. Hegner · C. Gerber

Fig. 28.21. PCA plot of perfume essence samples. Measurements of ﬂower, wood and lemon perfume essences reveal three clearly distinguishable clusters

The PCA evaluation of the cantilever-sensor response curves is shown in Fig. 28.21. Clear clustering is observed for the perfume essences tested, demonstrating the successful recognition, as well as the selectivity of the method. However, the “artiﬁcial nose” can only recognize sample vapors that have been measured before. Therefore, it is a characterization rather than a chemical-analysis tool. 28.3.2 Label-Free DNA Hybridization Detection The main advantage of cantilever array sensors is the possibility that measurements of differences in the responses of sensor and reference cantilevers can be evaluated. Measuring the deﬂection of only one cantilever will yield misleading results that might give rise to an incorrect interpretation of the cantilever-deﬂection trace [38]. Therefore, at least one of the cantilevers (the sensor cantilever) is coated with a sensitive layer that exhibits an afﬁnity to the molecules to be detected, whereas other cantilevers are coated with a molecular layer that does not show an afﬁnity to them (reference cantilevers). The biochemical system to be investigated here involves a DNA hybridization experiment in liquid using a thiolated 12-mer oligonucleotide sequence from the Bio B biotin synthetase gene (EMBL accession number: J04423). We selected three surface-bound probes, Bio B1 (5 -SH-C6 -ACA TTG TCG CAA-3 , C6 is a spacer), Bio B2 (5 -SH-C6 -TGC TGT TTG AAG-3 ) and Bio B6 (5 -SHC6 -TCA GGA ACG CCT-3 ), which are immobilized by thiol binding onto the gold-coated upper surface of a cantilever in an array (Fig. 28.22a). The target complements are called Bio B1C, B2C and B6C are diluted in 5x ssc buffer at 100 pM concentration. Upon injection of the matching sequence to Bio B1, i.e. Bio B1C, the sensor cantilever coated with Bio B1 will bend, whereas the reference cantilever coated with Bio B2 will not bend (Fig. 28.22b). After thorough rinsing with an unbinding agent, the cantilever coated with Bio B1 will bend back to its initial position (Fig. 28.22c). The bending is due to the formation of surface stress during the hybridization process because of steric crowding (double-stranded DNA requires more space than single-stranded DNA).

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

207

Fig. 28.22. Schematic of the cantilever array DNA hybridization sensor. (a) Each cantilever is functionalized with a self-assembled monolayer of thiolated oligonucleotides. (b) Upon injection of the complementary sequence to the oligonucleotide sequence shown in the middle, hybridization takes place and the cantilever bends downwards. (c) After thorough rinsing with an unbinding agent, the cantilever in the middle returns to its initial position. The schematics are courtesy of Jürgen Fritz, currently at the International University of Bremen, Germany

The actual experiment proceeds as follows (see Fig. 28.23a): First, the liquid cell with the functionalized cantilever array is ﬁlled with ssc buffer. After a stable base line has been achieved, ssc buffer is injected after 4 min for 3 min. The cantilevers deﬂect, but once the injection is over, a stable baseline is reached again. At 18 min, the target Bio B1C is injected, which is supposed to hybridize with the Bio B1 probe, but not with the Bio B2 or the Bio B6 probe. Both cantilevers deﬂect, but the deﬂection magnitude of the Bio B1-coated cantilever is larger than that of the Bio B2-coated cantilever. Finally, at 37 min, ssc buffer is injected again and a stable baseline is reached. From the deﬂection data shown in Fig. 28.23a, it is clear that no conclusive result can be obtained from individual cantilever responses only, as both the sensor and the reference cantilevers bend. However, a clear 20 nm deﬂection signal is observed when calculating the difference in deﬂection responses from probes Bio B1 (sensor) and reference Bio B2 (Fig. 28.23b), or the difference in deﬂection responses from probes Bio B1 and reference Bio B6. The differential deﬂection magnitudes obtained are 25 nm (B1–B2) or 30 nm (B1–B6), respectively.

208 Fig. 28.23. (a) Deﬂection traces of sensor (functionalized with DNA oligonucleotide sequence Bio B1) and reference cantilevers (functionalized with DNA oligonucleotide sequences Bio B2 and Bio B6, respectively). (b) Differences B1 to B2 and B1 to B6 of the bending responses of the sensor cantilever B1 and the reference cantilevers B2 and B6. The dotted lines are guides to the eye. The hybridization of B1 with B1C yields a difference signal of 25 (B1–B2) or 30 nm (B1–B6). (c) Difference in responses of the two reference probes. B2–B6 only yields a small signal of 5 nm. “Buffer” indicates ﬂushing the cell with 5x ssc buffer; “100 pM B1C” means injection of the complement to Bio B1 at a concentration of 100 pM in ssc buffer. Data courtesy of J. Zhang, University of Basel, Switzerland

H.P. Lang · M. Hegner · C. Gerber

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

209

The difference in deﬂection responses between two reference cantilevers yields no signal or only a very small signal that can be attributed to some unspeciﬁc binding of B1C to one of the reference probes, supposedly to B2, as the difference B2–B6 yields a positive signal of less than 5 nm, see Fig. 28.23c. We conclude that it is absolutely mandatory to use at least two cantilevers in an experiment, a reference cantilever and a sensor cantilever, to be able to cancel out undesired artifacts such as thermal drift or unspeciﬁc adsorption.

28.4 Applications and Outlook The ﬁeld of cantilever sensors has been very active in recent years [39–41]. Chief topics published 2003 and 2004 in the literature include the following studies: fabrication of silicon, piezoresistive [42, 43] or polymer [44] cantilevers, detection of vapors and volatile compounds, e.g. mercury vapor [45], HF vapor [46,47], chemical vapors [48,49], as well as the development of gas sensors utilizing the piezoresistive method [50]. Pd based sensors for hydrogen [51], deuterium and tritium [52, 53] have been reported, as well as sensors based on hydrogels [54–57] or zeolites [58]. A humidity sensor is suggested in [59]. Many articles investigate the topics of detection of explosives [60–65], pathogens [66], nerve agents [67], viruses [68], bacteria, e.g. E. coli [69–71], and pesticides like dichlorodiphenyltrichloroethane (DDT) [72]. Issues of detection of environmental pollutants are discussed in [73,74]. A chemical vapor sensor based on the bimetal technique is described in [75]. Several papers study the mechanism of static cantilever bending [76–78] and how the transduction can be improved by surface modiﬁcation of the sensor layer on the cantilever [79,80], or how its properties can be improved [81–83]. In two publications, electrochemical redox reactions are measured with cantilevers [84, 85]. A large number of papers are focused on biochemical applications using cantilevers, e.g. to detect DNA [86–89], proteins [90,91], prostate-speciﬁc antigen (PSA) [92], peptides using antibodies [93, 94] and living cells [95]. Most groups use the static deﬂection mode for measurements in liquids, whereas others use piezoresistive cantilevers [96] or the heat mode (calorimeter biosensor) [97]. The understanding of the chemistry [98] and the mechanism [99] of biosensing is of big importance, as is a strategy to improve the transduction mechanism [100]. Medical applications involve diagnostics [101], drug discovery [102] and detection of substances like glucose [103, 104]. Mass measurement in dynamic mode is investigated in a variety of publications, e.g. down to the attogram regime in ultra-high vacuum [105] using Doppler vibrometer readout of the oscillation [106]. Several authors study the modeling of resonant sensors [107–114] and exploit higher modes [115], as well as torsional or lateral resonance modes for improved mass sensing in dynamic mode [116, 117]. To improve the oscillation [118], the quality factor can be enhanced electronically [119, 120]. Effects of temperature and pressure on the oscillation should also be taken into account [121]. Other researchers use piezoresonators integrated into the cantilever structure [122]. Ellipsometry combined with cantilever readouts has been suggested [123] as a combined technique. Cantilevers have been used to calibrate AFM cantilevers [124, 125]. Nanowire electrodes attached to cantilevers in an array can be used for local

H.P. Lang · M. Hegner · C. Gerber

210

multiprobe resistivity measurements [126]. Two-dimensional microcantilever arrays have been proposed for multiplexed biomolecular analysis [127, 128]. We conclude that cantilever-sensor array techniques are very powerful and highly sensitive tools to study physisorption and chemisorption processes, as well as to determine material-speciﬁc properties such as heat transfer during phase transitions. Experiments in liquid environments have provided new insight on such complex biochemical reactions as the hybridization of DNA or molecular recognition in antibody/antigen systems or proteomics. Future developments point towards technological applications, e.g. to ﬁnding new ways to characterize real-world samples such as clinical samples. For example, the development of medical diagnosis tools requires the improvement of the sensitivity of a large number of genetic tests to be performed with small amounts of single donor-blood or body-ﬂuid samples. On the other hand, from a scientiﬁc point of view, the challenge lies in optimizing cantilever sensors to improve their sensitivity to the ultimate limit: the detection of individual molecules. Acknowledgements. We thank R. McKendry (University College London, London, U.K.), J. Zhang, A. Bietsch, V. Barwich, M. Ghatkesar, F. Huber, J.-P. Ramseyer, A. Tonin, H.R. Hidber, E. Meyer and H.-J. Güntherodt (University of Basel, Basel, Switzerland) for valuable contributions and discussions, as well as U. Drechsler, M. Despont, H. Schmid, E. Delamarche, H. Wolf, R. Stutz, R. Allenspach, and P.F. Seidler (IBM Research, Zurich Research Laboratory, Rüschlikon, Switzerland). This project was partially funded by the National Center of Competence in Research in Nanoscience (Basel, Switzerland), the Swiss National Science Foundation and the Commission for Technology and Innovation (Bern, Switzerland).

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

Binnig G, Quate CF, Gerber C (1986) Phys Rev Lett 56:930 Gimzewski JK, Gerber C, Meyer E, Schlittler RR (1994) Chem Phys Lett 217:589 Stoney GG (1909) Proc R Soc London Ser A 82:172 Thundat T, Warmack RJ, Chen GJ, Allison DP (1994) Appl Phys Lett 64:2894 Berger R, Lang HP, Gerber C, Gimzewski JK, Fabian JH, Scandella L, Meyer E, Güntherodt H-J (1998) Chem Phys Lett 294:363 Greenspoon JE (1961) J Acoust Soc America 33:1485 Zhang J, O’Shea S (2003) Sens Actuators B 94:65 Berger R, Gerber C, Gimzewski JK, Meyer E, Güntherodt H-J (1996) Appl Phys Lett 69:40 Bachels T, Schäfer R (1999) Chem Phys Lett 300:177 Bachels T, Tiefenbacher F, Schäfer R (1999) J Chem Phys 110:10008 Barnes JR, Stephenson RJ, Welland ME, Gerber C, Gimzewski JK (1994) Nature 372:79 Puers R, Lapadatu D (1996) Sens Actuators A 56:203 Huang JM, Liew KM, Wong CH, Rajendran S, Tan MJ, Liu AQ (2001) Sens Actuators A 93:273 Fricke J, Obermaier C (1993) J Micromech Microeng 3:100 Lang HP, Berger R, Andreoli C, Brugger J, Despont M, Vettiger P, Gerber C, Gimzewski JK, Ramseyer J-P, Meyer E, Güntherodt H-J (1998) Appl Phys Lett 72:383 Berger R, Delamarche E, Lang HP, Gerber C, Gimzewski JK, Meyer E, Güntherodt H-J (1997) Science 276:2021 Berger R, Delamarche E, Lang HP, Gerber C, Gimzewski JK, Meyer E, Güntherodt H-J (1998) Appl Phys A 66:S55

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

211

18. Wachter EA, Thundat T (1995) Rev Sci Instrum 66:3662 19. Scandella L, Binder G, Mezzacasa T, Gobrecht J, Berger R, Lang HP, Gerber C, Gimzewski JK, Koegler JH, Jansen JC (1998) Micropor Mesopor Mater 21:403 20. Raiteri R, Nelles G, Butt H-J, Knoll W, Skladal P (1999) Sens Actuators B 61:213 21. Moulin AM, O’Shea SJ, Welland ME (2000) Ultramicroscopy 82:23 22. Lang HP, Berger R, Battiston FM, Ramseyer J-P, Meyer E, Andreoli C, Brugger J, Vettiger P, Despont M, Mezzacasa T, Scandella L, Güntherodt H-J, Gerber C, Gimzewski JK (1998) Appl Phys A 66:S61 23. Lang HP, Baller MK, Berger R, Gerber C, Gimzewski JK, Battiston FM, Fornaro P, Ramseyer J-P, Meyer E, Güntherodt H-J (19999) Anal Chim Acta 393:59 24. Baller MK, Lang HP, Fritz J, Gerber C, Gimzewski JK, Drechsler U, Rothuizen H, Despont M, Vettiger P, Battiston FM, Ramseyer J-P, Fornaro P, Meyer E, Güntherodt H-J (2000) Ultramicroscopy 82:1 25. Battiston FM, Ramseyer J-P, Lang HP, Baller MK, Gerber C, Gimzewski JK, Meyer E, Güntherodt H-J (2001) Sens Actuators B 77:122 26. Fritz J, Baller MK, Lang HP, Strunz T, Meyer E, Güntherodt H-J, Delamarche E, Gerber C, Gimzewski JK (2000) Langmuir 16:9694 27. Fritz J, Baller MK, Lang HP, Rothuizen H, Vettiger P, Meyer E, Güntherodt H-J, Gerber C, Gimzewski JK (2000) Science 288:316 28. McKendry R, Zhang J, Arntz Y, Strunz T, Hegner M, Lang HP, Baller MK, Certa U, Meyer E, Güntherodt H-J, Gerber C (2002) Proc Nat Acad Sci USA 99:9783 29. Arntz Y, Seelig JD, Lang HP, Zhang J, Hunziker P, Ramseyer J-P, Meyer E, Hegner M, Gerber C (2003) Nanotechnology 14:86 30. Berger R, Gerber C, Lang HP, Gimzewski JK (1997) Microelectron Eng 35:373 31. Wolf H, Juncker, D, Michel B, Hunziker P, Delamarche E (2004) Biosens Bioelectron 10:1193 32. Bietsch A, Hegner M, Lang HP, Gerber C (2004) Langmuir 20:5119 33. Bietsch A, Zhang J, Hegner M, Lang, HP, Gerber C (2004) Nanotechnology 15:873 34. Lange D, Hagleitner C, Hierlemann A, Brand O, Baltes H (2002) Anal Chem 74:3084 35. Savran CA, Burg TP, Fritz J, Manalis SR (2003) Appl Phys Lett 83:1659 36. Gardner JW, Bartlett PN (1992) Sensors and Sensory Systems for an Electronic Nose. Kluwer, Dordrecht 37. Timm, NH (2002) Springer Texts in Statistics – Applied Multivariate Analysis. Springer, Berlin Heidelberg New York 38. Lang HP, Hegner M, Meyer E, Gerber C (2002) Nanotechnology 13:R29 39. Ziegler C (2004) Anal Bioanal Chem 379:946 40. Lavrik NV, Sepaniak MJ, Datskos PG (2004) Rev Sci Instrum 75:2229 41. Majumdar A (2002) Disease Markers 18:167 42. Tang YJ, Fang J, Yan XD, Ji HF (2004) Sens Actuators B 97:109 43. Forsen E, Nilsson SG, Carlberg P, Abadal G, Perez-Murano F, Esteve J, Montserrat J, Figueras E, Campabadal F, Verd J, Montelius L, Barniol N, Boisen A (2004) Nanotechnology 15:S628 44. McFarland AW, Poggi MA, Bottomley LA, Colton JS (2004) Rev Sci Instrum 75:2756 45. Rogers B, Manning L, Jones M, Sulchek T, Murray K, Beneschott B, Adams JD, Hu Z, Thundat T, Cavazos H, Minne SC (2003) Rev Sci Instrum 74:4899 46. Mertens J, Finot E, Nadal MH, Eyraud V, Heintz O, Bourillot E (2004) Sens Actuators B 99:58 47. Tang YJ, Fang J, Xu XH, Ji HF, Brown GM, Thundat T (2004) Anal Chem 76:2478 48. Yu XM, Zhang DC, Wang CS, Du XF, Wang XB, Ruan XB (2003) Chinese Phys Lett 20:1637 49. Abedinov N, Popov C, Yordanov Z, Ivanov T, Gotszalk T, Grabiec P, Kulisch W, Rangelow IW, Filenko D, Shirshov Y (2003) J Vac Sci Technol B 21:2931

212

H.P. Lang · M. Hegner · C. Gerber

50. Zhou J, Li P, Zhang S, Huang YP, Yang PY, Bao MH, Ruan G (2003) Microelectron Eng 69:37 51. Misna TE, McCorkle D, Warmack B (2003) Sens Actuators B 88:120 52. Fabre A, Finot E, Demoment J, Contreras S (2003) Ultramicroscopy 97:425 53. Fabre A, Finot E, Demoment M, Contreras S (2003) J Alloys Compd 356:376 54. Zhang YF, Ji HF, Brown GM, Thundat T (2003) Anal Chem 75:4773 55. Hilt JZ, Gupta AK, Bashir R, Peppas NA (2003) Biomed Microdevices 5:177 56. Liu K, Ji HF (200) Anal Sci 20:9 57. Zhang YF, Ji HF, Snow D, Sterling R, Brown GM (2004) Instrum Sci Technol 32:361 58. Zhou J, Li P, Zhang S, Long YC, Zhou F, Huang YP, Yang PY, Bao MH (2003) Sens Actuators B 94:337 59. Lee CY, Lee GB (2003) J Micromech Microeng 13:620 60. Muralidharan G, Wig A, Pinnaduwage LA, Hedden D, Thundat T, Lareau RT (2003) Ultramicroscopy 97:433 61. Pinnaduwage LA, Boiadjiev V, Hawk JE, Thundat T (2003) Appl Phys Lett 83:1471 62. Pinnaduwage LA, Thundat T, Gehl A, Wilson SD, Hedden DL, Lareau RT (2004) Ultramicroscopy 100:211 63. Pinnaduwage LA, Thundat T, Hawk JE, Hedden DL, Britt R, Houser EJ, Stepnowski S, McGill RA, Bubb D (2004) Sens Actuators B 99:223 64. Pinnaduwage LA, Wig A, Hedden DL, Gehl A, Yi D, Thundat T, Lareau RT (2004) J Appl Phys 95:5871 65. Pinnaduwage LA, Yi D, Tian F, Thundat T, Lareau RT (2004) Langmuir 20:2690 66. Weeks BL, Camarero J, Noy A, Miller AE, Stanker L, De Yoreo JJ (2003) Scanning 25:297 67. Yang YM, Ji HF, Thundat T (2003) J Am Chem Soc 125:1124 68. Gunter RL, Delinger WG, Manygoats K, Kooser A, Porter TL (2003) Sens Actuators A 107:219 69. Zhang J, Ji HF (2004) Anal Sci 20:585 70. Gfeller KY, Nugaeva N, Hegner M (2005) Biosens Bioelectron 21:528 71. Gfeller KY, Nugaeva N, Hegner M (2005) Appl Env Microbiol 71:2626 72. Alvarez M, Calle A, Tamayo J, Lechuga, LM, Abad A, Montoya A (2003) Biosens Bioelectron 18:649 73. Yan XD, Tang YJ, Ji HF, Lvov Y, Thundat T (2004) Instrum Sci Technol 32:175 74. Cherian S, Gupta RK, Mullin BC, Thundat T (2003) Biosens Bioelectron 19:411 75. Adams JD, Parrott G, Bauer C, Sant T, Manning L, Jones M, Rogers B, McCorkle D, Ferrell TL (2003) Appl Phys Lett 83:3428 76. Khaled ARA, Vafai K (2004) J Micromech Microeng 14:1220 77. Zhang Y, Ren Q, Zhao YP (2004) J Phys D 37:2140 78. Jeon S, Thundat T (2004) Appl Phys Lett 85:1083 79. Headrick JJ, Sepaniak MJ, Lavrik NV, Datskos PG (2003) Ultramicroscopy 97:417 80. Bottomley LA, Poggi MA, Shen SX (2004) Anal Chem 76:5685 81. Porter TL, Eastman MP, Macomber C, Delinger WG, Zhine R (2003) Ultramicroscopy 97:365 82. Kooser A, Gunter RL, Delinger WD, Porter TL, Eastman MP (2004) Sens Actuators B 99:474 83. Bumbu GG, Kircher G, Wolkenhauer M, Berger R, Gutmann JS (2004) Macromol Chem Phys 205:1713 84. Quist F, Tabard-Cossa V, Badia A (2003) J Phys Chem B 107:10691 85. Tian F, Pei JH, Hedden DL, Brown GM, Thundat T (2004) Ultramicroscopy 100:217 86. Su M, Li SU, Dravid, VP (2003) Appl Phys Lett 82:3562 87. Gunter RL, Zhine R, Delinger WG, Manygoats K, Kooser A, Porter TL (2004) IEEE Sensors J 4:430

28 Microfabricated Cantilever Array Sensors for (Bio-)Chemical Detection

213

88. Alvarez M, Carrascosa LG, Moreno M, Calle A, Zaballos A, Lechuga LM, Martinez-A C, Tamayo J (2004) Langmuir 20:9663 89. Gunter RL, Zhine R, Delinger WG, Manygoats K, Kooser A, Porter TL (2004) IEEE Sensors J 4:430 90. Lee JH, Yoon KH, Hwang KS, Park J, Ahn S, Kim TS (2004) Biosens Bioelectron 20:269 91. Lee JH, Kim TS, Yoon KH (2004) Appl Phys Lett 84:3187 92. Hwang KS, Lee JH, Park J, Yoon DS, Park JH, Kim TS (2004) Lab Chip 4:547 93. Kim BH, Mader O, Weimar U, Brock R, Kern DP (2003) J Vac Sci Technol B 21:1472 94. Kooser A, Manygoats K, Eastman MP, Porter TL (2003) Biosens Bioelectron 19:503 95. Saif MTA, Sager CR, Coyer S (2003) Annals Biomed Eng 31:950 96. Reasmussen PA, Thaysen J, Hansen O, Eriksen SC, Boisen A (2003) Ultramicroscopy 97:371 97. Arakawa ET, Lavrik NV, Rajic S, Datskos PG (2003) Ultramicroscopy 97:459 98. Yan XD, Lvov Y, Ji HF, Singh A, Thundat T (2003) Org Biomolec Chem 1:460 99. Liu F, Zhang Y, Ou-Yang ZC (2003) Biosens Bioelectron 18:655 100. Khaled ARA, Vafai K, Yang M, Zhang X, Orkan CS (2003) Sens Actuators B 94:103 101. Kumar S, Bajpai RP, Bharadwaj LM (2003) IETE Techn Rev 20:361 102. Zhang YF, Venkatachalan SP, Xu H, Xu XH, Joshi P, Ji HF, Schulte M (2004) Biosens Bioelectron 19:1473 103. Pei JH, Tian F, Thundat T (2004) Anal Chem 76:292 104. Yan XD, Ji HF, Lvov Y (2004) Chem Phys Lett 396:34 105. Ono T, Li XX, Miyashita H, Esashi M (2003) Rev Sci Instrum 74:1240 106. Ono T, Esashi M (2004) Meas Sci Technol 15:1977 107. Dufour I, Fadel L (2003) Sens Actuators B 91:353 108. Yum K, Wang ZY, Suryavanshi AP, Yu MF (2004) J Appl Phys 96:3933 109. Fadel L, Dufour I, Lochon F, Français O (2004) Sens Actuators B 102:73 110. Ryu WH, Chung YC, Choi DK, Yoon CS, Kim CK, Kim YH (2004) Sens Actuators B 97:98 111. Jeon S, Braiman Y, Thundat T (2004) Rev Sci Instrum 75:4841 112. Adams JD, York D, Whisman N (2004) Rev Sci Instrum 75:2903 113. Fadel L, Lochon F, Dufour I, Français O (2004) J Micromech Microeng 14:S23 114. Ren Q, Zhao YP (2004) In: Microsystem technologies, vol 10. Springer, Berlin Heidelberg New York, p 307 115. Kar AK, George MA (2003) J Appl Phys 94:4626 116. Sharos LB, Raman A, Crittenden S, Reifenberger R (2004) Appl Phys Lett 84:4638 117. Ilic B, Craighead HG, Krylov S, Senaratne W, Ober C, Neuzil P (2004) J Appl Phys 95:3694 118. Passian A, Muralidharan G, Mehta A, Simpson H, Ferell TL, Thundat T (2003) Ultramicroscopy 97:391 119. Tamayo J, Alvarez M, Lechuga LM (2003) Sens Actuators B 89:33 120. Tamayo J, Lechuga LM (2003) Appl Phys Lett 82:2919 121. Mertens J, Finot E, Thundat T, Falve A, Nadal MH, Eyraud V, Bourillot E (2003) Ultramicroscopy 97:119 122. Kim SJ, Hong KI, Choi DK (2003) J Appl Phys Pt 1 42:1475 123. Godin M, Laroche O, Tabard-Cossa V, Beaulieu LY, Grütter P, Williams PJ (2003) Rev Sci Instrum 74:4902 124. Gibson CT, Weeks BL, Abell C, Rayment T, Myhra S (2003) Ultramicroscopy 97:113 125. Hu ZY, Seeley T, Kossek S, Thundat T (2004) Rev Sci Instrum 75:400 126. Lin R, Boggild P, Hansen O (2004) J Appl Phys 96:2895 127. Khanafer K, Khaled ARA, Vafai K (2004) J Micromech Microeng 14:1328 128. Yue M, Lin H, Dedrick DE, Satyanarayana S, Majumdar A, Bedekar AS, Jenkins JW, Sundaram S (2004) J Microelectromech Syst 13:290

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices B. Gotsmann · U. Dürig

29.1 Introduction The interplay between thermal and mechanical properties of solids is a fascinating topic of research in materials science and certainly has numerous applications. In particular, mechanical polymer properties show both complex and interesting dependencies on temperature. There is a clear trend of extending such research down to the nanoscale, which is nurtured by the necessity to understand nanoscale properties in order to tailor materials for nanoscale applications. In our case, research is driven by its application for data storage [1]. Polymers are an obvious choice for thermomechanical storage materials: Their mechanical properties can be reversibly switched by the application of heat. Thermomechanical indentation formation is believed to be a suitable write mechanism of data storage. The read/write head, which at the same time is the analysis tool, is derived from atomic force microscopy (AFM). The microfabricated silicon cantilevers/tips are equipped with a heating element, making the AFM tip a manipulation and imaging tool with variable temperature. In particular, using heated probes we add another degree of freedom to nanoindentation experiments on polymers using standard probes [2–5], comparable to the case when the sample is heated [6]. Other applications of heated probes range from microthermal analysis [7], thermal imaging [8], lithography [9] to nanoscience [10]. As in other areas, it is also observed that by going from the macro- to the micro- and nanoscale, interface and size effects become increasingly important. Sections 29.2 and 29.3 of this chapter analyze heat and momentum transfer between the tool (cantilever/tip) and a solid surface. In Sect. 29.4 thermomechanical nanoindentation experiments on polymers are described, and Sect. 29.5 introduces thermomechanics in nanowear experiments. Finally, the application to data storage is discussed in Sect. 29.6.

29.2 Heat Transfer Mechanisms Let us start the discussion by looking at the layout of our tool: the main building block is a normal silicon AFM cantilever, see Fig. 29.1. A part of the cantilever is designed to be a heater with an area of about 4 × 6 µm. This heater is phosphorusdoped with 5 × 1017 cm−3 , i.e. much lower than the highly doped cantilever legs

B. Gotsmann · U. Dürig

216 Fig.29.1.Scanning electron micrograph of a cantilever with integrated heater (artiﬁcially contrasted) and tip

(1020 cm−3 ), and can thus be heated by applying an electrical current through the cantilever legs and the heater. The tip is manufactured to sit directly on the heater area, which extends beyond the tip area. Therefore, heat transported through the air-gap between the heater and the sample has to be taken into account. In fact, most of the heat is transported that way and through the cantilever legs, while only a fraction of the heat can be transported through the tip-surface contact into the sample. In the following, we will look at heat transport through the air gap by air conduction and convection, through the silicon cantilever, through the air gap by radiation and, ﬁnally, through the tip and the tip-surface interface. First, however, we describe the basic measurement and calibration technique. 29.2.1 Heat Generation in Microcantilevers All our experiments rely on a simple method of measuring the heater temperature and thermal resistance between the heater and the ambient. The electrical resistance of our cantilevers is dominated by the electrical resistance of the heater region. This electrical resistance depends rather strongly on the temperature. For the doping level used, the resistance is known to have a maximum at 550 ◦ C [11]. Temperature calibration of the heater is done by measuring an I–V curve, see Fig. 29.2. Typically, we have a resistor in series and measure the voltage drop across cantilever and resistor separately. The power needed to reach the maximum resistance PRmax is determined. We can safely assume that all the power dissipated in the cantilever leads to heating of the heater. Then a plot of resistance versus power, R(P), can be rescaled to R(Theater ) using Theater = RT + P · (550 ◦ C − RT )/PRmax = RT + Rth · P , where RT is the room temperature. The implicit assumption here is that the thermal resistance of the system Rth is independent of the heater temperature Theater . On checking that assumption with independent means of measuring the heater temperature, we concluded that the resulting systematic error is far below the measurement errors, fabrication tolerances and scatter. Using this calibration method, we estimate an absolute error of about 30% for a temperature difference ∆T = Theater − RT . Relative measurements

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

217

Fig. 29.2. I–V curve and derived P–R and T –R curves of an integrated resistive heater for calibration and measurement of the heater temperature

and temperature changes, however, can be conducted with ∼ 0.1 K resolution of temperature. The thermal resistance of the system Rth strongly depends on the different heat transport paths, which may vary with changing experimental parameters. We can differentiate between the heat paths by systematically varying their relative contributions. In the following paragraphs, paragraphs will discuss the various cooling mechanisms relevant for our system. 29.2.2 Heat Transfer Through Air and Silicon To analyze the heat-transfer mechanism in our heated cantilever structures, we consider an inﬁnitely long rectangular strip with width w and thickness t, which is placed at a distance h above a substrate (see Fig. 29.3). The strip contains a heater of length , which absorbs power from a source. The unheated sections to the left

Fig. 29.3. Schematic of a heater with and without cantilever legs with the important dimensions shown

B. Gotsmann · U. Dürig

218

and the right of the heater are termed legs. We assume that the width of the strip is larger than the width of the air gap, viz., w ≥ h. Under these conditions, the heat ﬂux is dominated by thermal diffusion within the strip and heat ﬂux across the air gap, whereas the heat ﬂux into the free air volume above the strip can be neglected. Note also that for dimensions smaller than a few millimeters, convective heat transport is negligible in comparison to normal thermal conduction through the air gap1 . The thermal response of the strip is described by the following differential equation τ

∂ 2 T(x, t) ∂T(x, t) Pe (t)h − T(x, t) + = λ2 , ∂t ∂x 2 κair w

(29.1)

where T(x, t) denotes the temperature difference of the strip with respect to the substrate. Note that we assume that the strip is thin and wide and that correspondingly we can assume a homogeneous temperature distribution orthogonal to the strip axis. The ﬁrst two terms correspond to the standard thermal diffusion equation. The third term describes the heat ﬂux through the air gap between strip and substrate, where we approximate the heat ﬂux by j = T(x, t)κair /h and where κair denotes the thermal conductance of the air gap between strip and substrate. The fourth term corresponds to the heat source, where Pe (t) denotes the heater power, which is assumed to be uniformly distributed over the heater region. Note that (29.1) is equivalent to the one-dimensional Schroedinger equation in quantum mechanics. Solutions can be represented by superpositions of plane waves of the form T(x, t) = Tkω ,ω exp(iωt − ikω x) ,

(29.2)

with kω = ±

i√ 1 + iωτ . λ

(29.3)

The thermal properties of the strip are governed by two scale parameters: (i) a thermal relaxation length κS ht , (29.4) λ= κair which describes the length over which the heat from the heater spills over into the unheated legs, and (ii) a thermal relaxation time τ=

1

λ2 , DS

(29.5)

For normal gases, convective heat transport is negligible if the so-called Grashof number Gr ∆ log(T )gν−2 3 is less than 1000, where ∆ log(T ) is of order unity, g = 9.81 m s−2 is the acceleration of gravity, ν 1.5 × 10−5 m2 s−1 is the kinematic viscosity and is a typical thermal length scale of the object.

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

219

which describes the time it takes to establish thermal equilibrium, or alternatively, the time it takes for a heat pulse to penetrate a distance of λ into the legs. Here, κ S and D S are the thermal conductance and the thermal diffusivity of the strip, respectively. For silicon, typical values for λ and τ are 1/2 1/2 t h (29.6) λ = 10 µm 0.5 µm 0.1 µm and τ = 5 µs

h t . 0.5 µm 0.1 µm

(29.7)

Note that the scale parameters are independent of the width of the strip, w. In the steady state, viz., constant heater power and constant height, the temperature distribution in the legs falls off exponentially as the distance from the heater increases: T(x) = T(/2) exp −(|x| − /2)/λ , (29.8) where T(/2) denotes the temperature at the heater boundary. The temperature distribution in the heater is cosh(x/λ) Pe h 1− . (29.9) T(x) = κair w cosh(/2λ) + sinh(/2λ) Correspondingly, one obtains a maximum temperature at the center of the heater of Pe h , with κair weff cosh(/2λ) + sinh(/2λ) 1 = .

+ 2λ cosh(/2λ) + sinh(/2λ) − 1 1 + (/2λ)2

T0 = eff

(29.10)

For long heaters, λ, the maximum temperature is inversely proportional to the heater length and the temperature at the boundary is exactly T0 /2. For short heaters, ≤ λ, the maximum temperature saturates at a value corresponding to an effective heater length of 2λ to account for the spillover of the heat into the legs. For vanishingly short heaters, the temperature at the heater boundary approaches T0 . To summarize, we can say that the heater length should be smaller than λ in order to achieve the highest temperature in a small volume for a given heater power. Next we consider time-dependent heater excitation. The thermal response at the center of the heater produced by a Fourier component Pe (ω) is [12] T0 (ω) =

Pe (ω)h 1 κair w 1 + iωτ

T0 (ω) =

1 Pe (ω)h √ 2κair wλ 1 + iωτ

for λ ,

(29.11)

and for ≤ λ .

(29.12)

B. Gotsmann · U. Dürig

220

The response of a long heater is equivalent to a ﬁrst-order low-pass ﬁlter with a cut-off frequency of 1/τ. As the corresponding impulse response is proportional to exp(−t/τ), no temperature pulses with a duration shorter than τ can be achieved. For a short heater, however, the situation is different. Here, the effective heater length decreases with increasing frequency, giving rise to a high-frequency dispersion that is only of the order 1/2. As a consequence, temperature pulses with a duration comparable to that of the excitation pulse can be obtained even for pulses shorter than τ. Also, the maximum temperature scales inversely as the square root of the pulse length. These properties render short heaters very useful for efﬁciently producing fast heat pulses in the microsecond domain, even though the overall relaxation time of the heater-leg structure is more than one order of magnitude slower in typical microfabricated silicon devices. This property has been explicitly exploited in our studies of the thermomechanical indentation kinetics discussed in Sect. 29.4. Thermoelectric height sensing exploits the fact that a small variation of the air gap gives rise to a corresponding change of the heater temperature. The modulation of the heater temperature (assuming constant heater power) can readily be computed from (29.10) for a quasi-static excitation, that is for a height modulation that is slow compared with τ: δh h 1 δh δT0 = T0 2 h

δT0 = T0

for λ , for ≤ λ .

and (29.13)

It can be seen that a short heater strip sensor provides only half the temperature response signal with respect to one with a long heater section. This sensitivity penalty arises from the fact that the legs play the predominant part in the heat transport to the substrate and that the back action on the heater is increasingly less efﬁcient for increasingly more distant leg sections. A detailed analysis of the full dynamic response properties is given in an unpublished communication by Duerig (2005). It sufﬁces to say that in essence the sensor acts as a ﬁrst-order low-pass ﬁlter with a cut-off frequency of 1/τ, irrespective of the heater length. So far we have considered only inﬁnitely long legs, which can never be realized in practical devices. For long heaters, λ, the leg length does not play a role and can in fact be any length practical for the device. The leg length, leg , however, has a decisive inﬂuence on the response properties of a short heater if the legs are comparable to or shorter than λ. Without going into detailed mathematics (unpublished communication by Duerig (2005)), we can assess the effect using simple physical arguments. First we note that the response time, which is governed by the time it takes for heat pulse to propagate to the heat sink, scales as leg 2 for leg ≤ λ . (29.14) τ(leg ) ∝ λ Thermal height sensing relies on heat conduction through the air gap. Hence, the temperature modulation should scale as the heater temperature times the ratio of the heat dissipated through the air gap Q˙ air divided by the total heat carried away by the legs. For a short heater as considered here, the total heat is just the heater power Pe .

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

221

The heat dissipated through the air gap in turn scales in proportion to leg /λ times the heater temperature T0 . As the heat loss due to air conduction becomes negligible for short legs, the heater temperature for constant power is simply proportional to the leg length: T0 ∝

leg . λ

(29.15)

Whence we ﬁnd that the temperature modulation induced by a small change of the air gap height scales as δT0 ∝ T0

Q˙ air ∝ T0 Pe

leg λ

2

∝ Pe

leg λ

3 .

(29.16)

Thus making the legs longer than λ is crucial for obtaining an efﬁcient thermomechanical height sensor. As a ﬁnal point, we would like to address size effects associated with thermal conduction. The scale parameter here is the mean free path, th , of the carriers of the heat energy, molecules in the air gap and phonons in the Si heater-leg strip. The mean free path is on the order of 50 to 100 nm both in air and in Si, and thus of a similar order of magnitude as at least one of the characteristic physical dimensions in the problem, viz., the height of the air gap and the thickness of the Si strip. The mean free path expresses the distance a carrier travels before it is able to exchange energy with other carriers by scattering. A boundary is a signiﬁcant perturbation of the environment and has scattering properties that differ equally signiﬁcantly from those of the carrier ensemble. The net effect is that heat transport is hampered both across and along the boundary. Let us ﬁrst look at the transport across the boundary, which is the relevant scenario for the thermal transport in the air gap. We treat the carrier ensemble as an ideal gas of particles. Furthermore, we assume local thermal equilibrium in the gas and correspondingly we consider only a small ﬁrst-order correction to a Maxwell– Boltzmann velocity distribution of the particles in order to account for a thermal gradient in the gas [13]. At the boundary, however, local equilibrium is violated, as it turns out, because the boundary surface must be at a different temperature from that of the gas in order to fulﬁll energy and particle conservation. The temperature difference is proportional to the heat ﬂux across the boundary, and therefore, we can introduce a thermal boundary resistance per unit area given by Duerig in an unpublished communication (2005) Rboundary

2 − a 2th (T ) , a κ(T )

(29.17)

where T denotes the local gas temperature at the boundary, κ(T ) is the thermal conductance at T , and a is the so-called accommodation coefﬁcient, which is in essence a fudge factor accounting for the interface scattering. It can be interpreted as the fraction of gas particles that completely thermalize (acquire the temperature of the boundary) during scattering, whereas 1 − a is the fraction of elastically reﬂected particles. Typical values of the accommodation coefﬁcient for realistic surfaces are

B. Gotsmann · U. Dürig

222

on the order of 0.7, yielding Rboundary 3th /κ. Hence, the net effect of the thermal boundary resistance is to replace the physical interface by a virtual interface such that the width of the gap is increased by three times the mean free path. Hence, the effective width of the air gap is h eff h + 2 th (Ts ) + th (Ts + T ) , (29.18) where Ts denotes the absolute temperature of the substrate. For air, the mean free path is 60 nm at room temperature, increasing to 100 nm at 200 ◦ C. Hence, for typical operating temperatures, the boundary resistance has a signiﬁcant effect on the effective thermal conductivity of the air gap, even if the latter is as much as 1 µm in size. Now we consider transport parallel to the boundary, which is the relevant scenario for thermal transport in the Si strip. We treat the phonons again as ideal gas. The scattering of phonons at the interface is a complicated process, which can be abstracted analogously to the scattering of air molecules. A phonon can either thermalize at the interface with a probability a, or be elastically scattered with a probability 1 − a. Elastically scattered phonons are indistinguishable from phonons in the gas and thus do not affect the thermal transport properties. Thermalized phonons, however, lose their memory and no longer contribute to the thermal transport (the ﬁrst-order correction to the Boltzmann distribution is lost in the scattering process). An elementary calculation yields that the total heat ﬂux is reduced by an amount that is equivalent to narrowing the dimensions, width and thickness, of the strip according to weff w

− 2ath . (29.19) teff t Typical values for the phonon mean free path are on the order of 100 nm [14] and a 1. Hence, for thin strips with a thickness of a few 100 nm, interface scattering gives rise to a substantial decrease of the thermal conductance. For very thin strips with t th , the conductance should scale as t/th [14]. Finally we note that heater strips can function as excellent experimental setups for measuring size effects of thermal transport via relaxation time measurements (see (29.4) and (29.5)), and thus direct heat ﬂux sensing, which typically is difﬁcult to perform, can be eliminated. 29.2.3 Heat Transfer Through Radiation In the preceding Sect. 29.2.2 heat transfer by thermal radiation has been neglected. In order to justify this a posteriori, we can ﬁrst consider thermal black-body radiation involving propagating electromagnetic waves, readily described by the Stefan– Boltzmann equation: S=

π 2 kB4 4 T1 − T24 . 3 2 60h c

Here, the cooling power per area, S, is expressed in terms of the Boltzmann constant, kB , the Planck constant, h, the speed of light, c, and the temperatures T1 and T2 of the two surfaces facing each other.

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

223

For a heater temperature of 100 K above RT , this corresponds to a thermal resistance of ∼ 6 × 108 K/W for effective heater dimensions of (9 µm)2 . Compared with the total thermal resistance of a heater under ambient conditions of about 1 × 105 K/W, the contribution of the black-body radiation is negligible. The overall thermal resistance is dominated by air conduction. However, in our experimental case, the Stefan–Boltzmann equation is only an approximation: At submicron distances between heater and sample surface near ﬁeld effects have to be taken into account. Such effects have a long history of theoretical analysis, see for example, [15–19]. Experimentally, the effect appears somewhat difﬁcult to pin down and observations have only rarely been claimed [20–23]. It has been theoretically predicted that heat transport by evanescent thermal radiation will depend strongly on the materials involved, with a strong distance dependence (1/d 2 for most cases) and a weakened temperature dependence (T 2 for most cases), compared with the Stefan–Boltzmann law [16]. According to theory, we can expect that the near-ﬁeld effect for a polymer surface is very small. For a silicon surface it can be signiﬁcantly higher, depending on the doping [17–19]. Nevertheless, the contribution of cooling by thermal radiation under vacuum conditions becomes signiﬁcant, not so much because of an increased overall thermal resistance without air conduction of around 3 × 105 K/W (now dominated by conduction through the cantilever legs), but rather because we can use the distance dependence to demonstrate the existence of near-ﬁeld cooling. In air the distance dependence is dominated by air conduction, whereas in vacuum the conduction through the cantilever legs does not depend on the heater-sample distance. On approaching the surface (before tip-surface contact), we can thus vary the contribution from radiation until the tip touches the surface and we have to take thermal conductance through the tip into account. As seen in Fig. 29.4, the radiation can be observed clearly and the magnitude and distance dependence are in accordance with theoretical predictions: around 150 W/(K m2 ) at about 500 nm distance (tip height). A detailed discussion will be published elsewhere.

Fig. 29.4. Thermal resistance of a heated cantilever in vacuum as a function of lever-surface clearance. A silicon sample was used. The heater temperature is around 575 ◦ C

B. Gotsmann · U. Dürig

224

29.2.4 Heat Transfer Through a Tip-Surface Point Contact As already seen in Fig. 29.4, the heat transport through the tip-surface contact can be readily measured under vacuum conditions. The background of thermal conductance through the cantilever legs and the radiation can be determined from the data obtained with the tip not in contact and then subtracted. As shown in Fig. 29.5, the thermal resistance of the tip-surface contact, Rts , can be extracted. In Fig. 29.6, Rts is separated from the background, and is itself a sum of several thermal resistances: The thermal resistance of tip and interface, as well as the polymer spreading resistance: Rts = Rtip + Rint + Rspread . The thermal resistance of the tip, Rtip , originates from the conductance of phonons within the silicon tip and from the layer of native oxide covering it. In the tip the resistivity increases with respect to bulk silicon because of enhanced phonon scattering with boundary surfaces [14]. The thermal resistance of the silicon tip can be estimated using predictions for the thermal resistivity of silicon nanowires as a function of diameter [24]. Integrating the expression for the varying diameter down to the interface of ∼ 10 nm of a cone-shaped tip with a typical opening angle of 50◦ diameter yields a thermal resistance of the order of 106 –107 K/W due to phonon scattering. This is in agreement with ﬁnite element calculations [25]. In addition, the effect of the oxide cap has to be considered and is added to the tip’s thermal resistance. The value of this contribution can be estimated as an

Fig. 29.5. Thermal resistance of a heated cantilever/tip as a function of displacement and contact with an 80-nm-thick SU8 ﬁlm on a silicon substrate. Out of contact, the displacement translates into a distance change between heater and sample. In contact, the displacement translates into a load force of the tip pressed onto the polymer. The corresponding temperature change in the heater is about 1.5 K in total around the heater temperature of 315 ◦ C. From the difference between the thermal resistance out of contact (∼ 1.077 MK/W) and the thermal resistance in contact (1.071–1.073 MK/W) the thermal resistance due to heat transport through the tip-surface contact is calculated to be 2–3 × 108 K/W (see Fig. 29.7)

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

225

Fig. 29.6. Schematic representation of the thermal resistances involved in heat transfer between heater and polymer

independent thermal resistance of its own using approximate dimensions, or as a mesoscopic link enabling “phonon tunneling” between the silicon cone and the sample [26]. The uncertainty remains large and we arrive at 107 –108 K/W for the oxide cap. Accordingly, the total value of Rtip might be dominated by the oxide cap and we, therefore, expect also 107 –108 K/W for Rtip . The thermal resistance of the interface Rint similarly deﬁes an accurate prediction, as the interface resistance usually depends strongly on the quality of the interface and the contact pressure. In our case, we can at least assume a simple case of a single 0 , which depends asperity contact characterized by a contact radius a and a constant rint on the materials involved and the pressure: Rint =

0 rint . πa2

From contact models, we can deduce that a is on the order of a few nanometers. If an ideal silicon tip is assumed, Rint could, therefore, be treated simply as a scattering site 0 for phonons in silicon: rint = 2.1 × 10−9 K m2 /W [25]. Experimental values using 0 can be found for silicon-polymer large areas (rather than a tip-surface contact) of rint interfaces and are in the range of 10−8 to 10−7 K m2 /W [27, 28]. This, however, will only be a lower limit, because under high pressures such as in our case, the values can be higher. In summary, we expect Rint = 3 × 107 −1 × 109 K/W for a ∼ 5 nm. Finally, the spreading resistance within the polymer itself is probably understood best. For a polymer ﬁlm on a silicon substrate, we can use the approximate solution [29]: ⎞ ⎛ 1 1 2 ⎠. Rspread = − log ⎝ k 4kpol a 2πkpol t 1 + pol ksub

Here, kpol and t are the thermal conductance and the thickness of the polymer ﬁlm, respectively. ksub denotes the thermal conductivity of the substrate (silicon) on which the ﬁlm is spun. The second term is only a correction of a few percent for the thickness of our polymer ﬁlms of approximately 100 nm.

B. Gotsmann · U. Dürig

226

A complication in our case is the fact that kpol is a strong function of pressure [30]. The value of 0.2–0.3 W/(K m) is typical for polymers, but can be increased by up to a factor of 2 under a pressure of 1 GPa. Since the stresses under the tip are transient, we have to resort to estimating an effective pressure-increased thermal conductivity of 0.3–0.6 W/(K m). Using a contact radius of 5 nm, we obtain an estimate of Rspread ∼ (0.8−1.6) × 108 K/W. From the above discussion, we conclude that all three contributions to Rth will play a role. In order to quantify the different contributions to Rts , we can again vary the individual contributions. First, we can use the force variation on the tip-surface contact during an experiment as shown in Fig. 29.5, then the contact radius a can be calculated using well-established mechanical contact models [31]. Further, as an additional criteria we can require that Rtip remain the same when varying the sample and that it can be assumed independent of a. With this strategy, we use the variation of sample material and the dependence on the contact size to obtain ﬁts for the separate contributions to Rts . For example, from the thermal resistance Rth in Fig. 29.5, we obtain experimental values of the various contributions (see Fig. 29.7). The results are in agreement with the above theoretical estimations and allow us to narrow down their range. In this example, the relative proportions depend on the load force (contact radius) and yield 30–50% for Rtip , ∼ 40% for Rspread , and 20–30% for Rint . We deﬁne a heating efﬁciency c as the rise in polymer surface temperature divided by the rise in heater temperature and obtain: c=

Rspread ≈ 40% . Rts

From experiments using different tips and other methods (see below), we can summarize that c ∼ 0.3−0.7 for nearly all cases (only polymers). Of the few experimental data available in the literature for direct comparison, measurements by Shi and Majumdar [32] under ambient conditions using AFM tips are maybe the most similar. The results were carefully interpreted, and it was discussed to what extent thermal conduction through a water meniscus, which is

Fig. 29.7. Experimental thermal resistance Rth of a typical tip–sample contact on a polymer (SU8) ﬁlm, and ﬁt (dash-dotted) containing the three components Rtip (dotted), Rint (solid) and Rspread (dashed). For the given tip size (radius 13.5 nm), the contact radius varies between 3.5 and 6.5 nm. The thermal resistance of the tip was estimated using a reference experiment on silicon

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

227

always present under ambient conditions, plays a role. In our measurement, this heat path can be avoided due to vacuum conditions, and even under ambient conditions the rather high tip temperatures render the existence of a water meniscus unlikely.

29.3 Momentum Transfer Through Air Heat transfer through air is accompanied by momentum transfer. At the air-cantilever interface heat is transferred through the interaction of air molecules with the solid silicon surface. This interaction is inelastic in the sense that the molecules change their kinetic energy by being scattered from the surface. In our case, the air molecules going towards the cantilever/heater will generally be cooler than the heater surface and accommodate a higher temperature through the interaction. At the same time there will also be momentum transfer: If, on average, the surface heats scattering molecules, then it will give additional momentum to the molecules, and the cantilever will be recoiled. Such effects lead to thermal forces, a subtle effect which has been studied for more than a century now [33] and is best known from the resulting transport phenomenon “thermophoresis”. In general, a particle suspended in a gas will be forced in the opposite direction of the temperature gradient within the gas. The gas particles interacting with the particle’s surface from the hotter side transfer more momentum to the particle than the slightly cooler gas particles hitting the surface from the other side. The side of our cantilever facing the substrate is subjected to cooler gas than the backside of the cantilever. Nevertheless, the resulting force that we can observe for a heated cantilever is always directed away from the surface, in the opposite direction than would be expected from thermophoresis. Although in thermophoresis a particle will generally be transported away from a heated surface, we have found that the particle (i.e. our cantilever) is actually attracted to a heated surface, an effect similar to the so-called “negative thermophoresis” predicted for particles with high thermal conductivity [34]. Clearly, the simple picture of thermophoresis cannot be readily applied to our case for two reasons: Firstly, the cantilever is not a thermally “passive” particle but rather a heat source with a given temperature. Secondly, also the interaction of the gas particles with the substrate has to be taken into account: Air molecules hitting the cantilever surface on the sample side carry some history of the interaction with the substrate surface. In our setup, the mean free path of molecules in air (λ ∼ 60 nm) is only a few times larger than the typical dimension of the problem, the distance between heater (i.e. cantilever) and substrate surface of approximately d ∼ 500 nm. The Knudsen number Kn = λ/d ∼ 0.1−0.2 categorizes the setup in the socalled “transition regime”, the regime between continuum and the free molecular limit. An exact prediction of the resulting force for such Knudsen numbers requires a numerical treatment. To simplify, we can look at the solution for the free molecular limit. It has been shown [35] that the arising force can be de-

228

B. Gotsmann · U. Dürig

scribed well by using an equation adapted from the original expression by Passion et al. [36]:

as τs + acb (1 − as )τc acb τc + as (1 − acb )τs F = cAeff + as + acb − as acb as + acb − as acb √ − (1 − act )τb + act τc − τb . Here, F denotes the force and Aeff the effective area. The accommodation factors for momentum transfer are denoted by as , act , and acb for the substrate surface, cantilever top-side (facing the surface) and cantilever backside, respectively. For convenience, the temperatures of cantilever, surface and surrounding gas are normalized to ambient temperature, τc , τs and τb , respectively. In the free molecular regime, the prefactor c would simply be P/2, with P being the ambient pressure. Here, we have to take into account the effect of the ﬁnite Knudsen number and hence a reduction by about 50% to P/4 [36,37]. The air facing the backside of the cantilever is modeled as a wall similar to the substrate surface with an accommodation factor ab of unity, see Fig. 29.8. This accounts for the interparticle collisions and allows us to have a closed expression for F. The most straightforward way to demonstrate the effect is to approach a heated cantilever towards the substrate, as shown in Fig. 29.9. It can be clearly seen that the force is repulsive and increases on approaching the surface. Its transient nature stems mainly from the variation of the Knudson number upon approach, i.e. the variation of the relation between distance and mean free path. The closest distance we have in such a curve is given by the contact of the tip with the substrate, resulting in a distance between heater and substrate of the tip height of approx. 500 nm. At a ﬁxed distance, the force depends almost linearly on the temperature difference between heater and substrate. At this contact distance we obtain forces on the order 0.25 nN/K, making the effect relevant for a quantitative interpretation of the indentation experiments at low loads described below.

Fig. 29.8. Schematic model geometry for the thermal force with temperatures and accommodation coefﬁcients

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

229

Fig. 29.9. Example of thermal force arising on heated cantilever near a surface. In (a) the force is plotted as a function of distance. The interference on the raw signal (solid line) can be subtracted (dashed line). In (b) the thermal force is plotted versus temperature at a constant heater-surface distance given by the tip height of about 500 nm

In summary, thermal forces can be understood and are subtracted from measurements shown in other sections. The thermal forces arising on heated cantilevers are surprisingly large, and will, therefore, probably again play a role in micromechanical device applications.

29.4 Thermomechanical Nanoindentation of Polymers 29.4.1 General Considerations When using the experimental tool described in the preceding sections, a large parameter space is available to study the thermomechanical properties of polymers on a local scale. The heater temperature can be controlled between RT and about 1000 ◦ C, resulting in a polymer temperature between RT and about 500 ◦ C. Forces from nanonewtons to micronewtons can be applied, heating/loading times can range from 1 s down to 10−6 s, and tip shapes with curvature radii ranging from 50–5 nm allow lateral resolutions below 10 nm. For the indentation experiments, the following technical aspects need to be considered. During the short writing events (typically 10 µs), the heater did not reach thermal equilibrium. Therefore for heat calibration (see above), the temperature at the end of the 10 µs was used, unless stated otherwise. In the experiments shown below, different reading schemes were used, which were shown to yield identical results: the reading using the integrated heater (see Sect. 29.2.2) and the commonly used optical-deﬂection measurement.

230

B. Gotsmann · U. Dürig

In AFM the load force is usually controlled by a piezo-displacement, i.e. a displacement of the cantilever holder, which translates into a load force by the Hookian law. Although this method is very accurate and well-understood, it is not yet ideal for our experiments. Low spring constants (∼ 0.055 N/m) are required for data-storage application, making the translation method difﬁcult. Moreover, we require very fast load rates, but the speed of the lever and the piezo limit the maximum speed that we can achieve. Technically (see Sect. 29.3), controlling the displacement can be difﬁcult for cantilever arrays. For these reasons, we use electrostatic forces to induce the load. Applying a bias voltage between a conducting substrate (i.e. a doped Si wafer below the polymer ﬁlm) and the cantilever results in an electrostatic force, which, provided that fairly low voltages (0–15 V) are used, has an inﬂuence on the cantilever design. Firstly, the tip height has to be comparatively small, which fortunately is supported by our reading scheme. We use tip heights of 400–700 nm. Secondly, a large capacitive area can help reduce the voltages further. For cantilevers that are not overly ﬂoppy, the following equation holds: F ∝ · V2 , and we typically obtain a force response of ∼ 2 nN/V2 . This force is calibrated in a separate experiment in which the same bias is applied in a dc fashion while measuring the pull-off force. The increase of the pull-off force due to the bias is assumed to be identical to the loading force during writing using bias voltages of the same magnitude and on a faster time scale. The application of heat, as outlined in Sect. 29.2, also requires applying a voltage to the lever. This has to be taken into account in determining the force response. 29.4.2 Indentation Experiments The two most important and most easily varied parameters are the load force and the heater temperature. We will look at these two ﬁrst, before discussing the time scale and tip-shape dependence. To investigate how the indentation size depends on the writing parameters, we wrote arrays of indentations in which the applied force and temperature of the heater were systematically varied along the rows and columns. An example is given in Fig. 29.10 using a thin ﬁlm of polymethylmethacrylate (PMMA) spun-cast onto a buffer consisting of highly cross-linked epoxy (SU8). Each indentation in this real-space image can be attributed to a load force and heater temperature. It can be seen that for any given load force, the indentation size depends strongly on heater temperature. Below a certain threshold temperature, no indentation can be obtained, above the threshold the indentation size (the most relevant are the depth and the outer diameter) increases rapidly with either increasing temperature or load force. We deﬁne the writing threshold as the minimum heater temperature needed to produce an indentation that appears about 1 nm deep in a subsequent AFM image. Since this writing threshold deﬁnition does not involve the analysis of deeper indentations, it allows us to relate indentation results to various experimental parameters independent of complex contact mechanics, in which the change of material properties during the indentation process has to be taken into account. Initially, there is

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

231

Fig. 29.10. Indentation experiment of PMMA. A grid of indentations was written with spacings of 120 nm. Each indentation event used different heat and load, but the same duration of 10−5 s. The heat voltage was varied on one axis, the load voltage on the other. The resulting forces and temperatures are plotted as axes to the AFM image of the indentations

thus no need for a detailed analysis of the nanoscale heat transfer and the thermomechanical deformation, which occur simultaneously and inﬂuence each other. Using this writing threshold deﬁnition, we can determine a threshold heater temperature Tthresh for any given load force F. Such a function Tthresh (F) is characteristic of the polymer material and the writing parameters’ time and tip shape. Of course, it is also possible to deﬁne a threshold force as a function of temperature Fthresh (T ). Let us ﬁrst investigate the inﬂuence of the tip shape. For this purpose, we have performed writing threshold measurements for two different tips on the same material (PMMA) as shown in Fig. 29.11. In both cases, the function Tthresh (F) is linear within the accuracy of the measurements. It can be seen that a sharper tip requires less heating at larger load forces. This can readily be understood by regarding the limiting case of no heating at all. In that limiting case, the load force needed to make an indentation should depend strongly on the tip sharpness, as is known from nanoindentation experiments. The other limit of low forces now leads to a situation in which the sharp tip actually encounters more difﬁculties to write, in the sense that more heat is needed to make an indentation. In this limit, the heat transfer through the tip-surface interface becomes decisive. As discussed in Sect. 29.2.4, it becomes more difﬁcult to heat the polymer above its glass transition temperature through the strongly reduced interface of a sharp tip. The heating efﬁciency c is thus a strong function of tip-sharpness. To extract information on the polymer material, we made measurements with a single tip on different materials. In other words, it is advisable to make a reference measurement on a known material for any given tip. This allows us to determine relative changes in hardness as can be seen when comparing the load force needed to indent the polymer with an unheated tip [2–5]. The other limit (low forces) can be used to relate the thermomechanical indentation to the glass-transition temperature of polymers. As shown in Fig. 29.12, the

232

B. Gotsmann · U. Dürig

Fig. 29.11. Writing threshold Tthresh (F ) obtained from experiments similar to the one shown in Fig. 29.10. The data were obtained using the same sample and 15 µs heat/load times. The tip shapes for the two sets of data were obtained by scanning electron microscopy. (From [38], © by AIP)

writing threshold temperature then depends linearly on the glass-transition temperature of the various polymers. When comparing the results in Fig. 29.12 with the heat-transfer measurements discussed above (see Sect. 29.2.4), we found that part of the difference of the heater temperature Tthresh to Tg can be attributed to the heat efﬁciency less than unity. As will be discussed below, a second part is due to the fast time scale. Both effects appear to be very similar for most polymers. The strong correlation of Tthresh to Tg , however, suggests that the glass temperature is an important parameter to differentiate polymers. In summary, the temperature, load and tip-shape are among the most important experimental parameters, as they probe the hardness and glass temperature of the polymer.

Fig. 29.12. Correlation of writing threshold heater temperature (i.e. at 10 µs heat pulse, low loading force (< 20 nN) for a standard-sized tip) to the known glass-transition temperature of different polymers, such as poly-α-Me-styrene, Polymethylmethacrylate, SU8, polystyrene with varying content of benzocylcobutene cross-linker, polysulfone, and polyimide

29.4.3 Interlude: Carbon Nanotube Tips In view of the application of the technique to data storage (as discussed below), an optimum tip-shape would be required to achieve a high density and to reduce tip

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

233

Fig. 29.13. SEM image of MWCNT tip attached to a silicon cantilever/tip (grayscale inverted). (From [38], © by AIP)

wear. Naturally, a rod-shaped or tube-shaped tip is the most promising. We have used multi-walled carbon nanotube (MWCNT) tips attached to our heated silicon tips [38]. The fabrication of the tip is described in detail in a chapter in [39]. With the tip shown in Fig. 29.13, writing threshold experiments similar to those shown in Fig. 29.10 were performed. The MWCNT tip had a diameter of about 20 nm from which a tip radius of 10 nm can be estimated. The shapes of the indentations resemble those made with conventional tips. However, a quantitative analysis of the writing threshold shown in Fig. 29.14 reveals a signiﬁcantly lower writing temperature than for conventional silicon tips of similar sharpness. Extrapolated to the limit of no heating, the MWCNT tip has a similar minimum writing force as a silicon tip of similar sharpness. In this limit, the hardness of the polymer and the size of the tip govern the writing parameters. In the other limit of low writing forces, the minimum temperature needed to make an indentation is signiﬁcantly lower. This limit is governed by the heat transfer from the heater through the tip into the polymer sample. A comparison of the two silicon tips with the MWCNT tip suggests that the MWCNT tip writes more easily because of improved heat transfer rather than because improved tip geometry. This improved heat transfer can be attributed to an improvement in the heat transfer through the interface, because the carbon tip is likely to have a lower phonon mismatch with the polymer than the silicon tip.

Fig. 29.14. Writing threshold graph as in Fig. 29.11 for two silicon tips compared to the MWCNT tip. (From [38], © by AIP)

234

B. Gotsmann · U. Dürig

Furthermore, as discussed above, a large contribution to the thermal resistance stems from the silicon tip apex itself. This could be partly bypassed with the MWCNT, which was mounted on the sidewall of the tip. 29.4.4 Interlude: Thermal Force and Indentation Formation An experimental difﬁculty in the low force limit of the writing threshold experiment is the existence of a signiﬁcant repulsive thermal force between the heated lever and the sample (see Sect. 29.3). This manifests itself, for example, in a reduction of the usable parameter space. If the bias voltage (which is applied to provide the load force) is too low, then (depending on the cantilever design) it is not possible to write indentations at all, because the thermal force lifts the tip out of contact. An example of force-temperature indentation characteristics written under such conditions is shown in Fig. 29.15.

Fig. 29.15. (a) Writing parameter window comparable to Fig. 29.10 on poly(styrene94.6% BCB5.4% ). The real-space spacing between individual indentations is 120 nm and 150 nm in the vertical and horizontal directions, respectively. The heat voltage was varied from row to row between 290 and 600 ◦ C approximately linearly. The sample bias voltage was varied from column to column between 12 and 254 nN; however the ﬁrst column yielding indentations has nominally 34 nN. Below that value, the minimum heat required to leave an indentation (“writing threshold”) is larger than the maximum heat allowed to ensure that the tip stays in contact. A second threshold (“thermal force threshold”) can be assigned, at which the tip is lifted out of contact through the thermal force. In (b) the area where the two thresholds intersect is shown enlarged. Here the indentations are spaced at 150 nm distance in both directions, the load variation was approximately 1 nN between the rows, and the temperature variation was about 14.5 ◦ C

29.4.5 Interlude: Rim Formation on Polymer Samples Indentations on polymer samples often are accompanied by a pile-up or rim formation around the actual indentation. Usually, the pile-up around an indentation

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

235

consists of material moved from the indented region to the surrounding area. If compression is considered, this rim material should be less than or equal to the volume of the indentation. There are some AFM-imaging-related artifacts in determining the corresponding topography, which is of similar size to the probing tip. However, these artifacts are well-understood and their effects can be estimated. Moreover, during imaging, the tip might not penetrate right to the bottom of the remaining indentation if its sidewalls have relaxed even slightly. This might play a role in the interpretation of the measured dimensions of nanoindentations. Even if we consider such effects of imaging artifacts and possible relaxation, we sometimes have situations were the pile-up rim reaches a much higher volume than the indentation, see for example Fig. 29.16. This suggests that the rim not only consists of displaced material, but has also undergone a change in the material properties, i.e. a reduction of the density or an increase of free volume. This becomes clear in the following observation. If we increase the writing temperature (at low loads) on a PMMA ﬁlm to a value higher than the writing threshold, we can observe that the tip eventually penetrates the PMMA layer until it stops at the SU8 buffer layer (Fig. 29.17). On increasing the heat even further, the indentation itself will no longer change (depth limited), but the rim keeps increasing. This phenomenon can be explained by the rapid quenching after the indentation process. At the end of the indentation process, the temperature in the polymer is expected to be quenched below Tg within less than 1 µs (see below), which leaves the polymer in a state of increased speciﬁc volume [40]. Further relevant effects inﬂuencing rim formation lies are in the interaction with the substrate and kinetic effects during indentation [41]. It is interesting to note that this swelling effect strongly correlates with the molecular weight, Mw , of the PMMA chains. For high molecular weights (Mw ∼ hundreds of kDa) or cross-linked samples, the effect is small, becoming increasingly predominant for Mw < 100 kDa. Generally, for Mw of less than 2–3 times the entanglement molecular weight (5–10 kDa for PMMA), indentations are actually difﬁcult to obtain because, similar as in Fig. 29.16b), the rim predominates the

Fig. 29.16. Cross-section through a line of a writing threshold example similar to Fig. 29.10 for a polysulfone ﬁlm. The load force was ﬁxed at 200 nN. The temperature of the heater at the end of the pulse has been increased from 430 to 610 ◦ C in steps of about 10.6 ◦ C. (a) The load was sufﬁcient to form a plastic indentation even if the polymer was not heated enough to reach the softening temperature. (b) By increasing the heater temperature, a swelling of the polymer occurs, which works against the indentation and leads to an erasure of previously written “cold” bits. (c) As this process continues, the thermomechanical formation of indentations begins to predominate until, ﬁnally, normal thermomechanical indentation occurs. (From [1], © by IEEE)

236

B. Gotsmann · U. Dürig

Fig. 29.17. Cross-section of indentations made on a PMMA ﬁlm using low loads and varying heater temperatures. The volume of the pile-up region around the indentation grows much more strongly with temperature than the depth, which is limited by the ﬁnite thickness of the ﬁlm. This indicates an increase of speciﬁc volume of the polymer in the rim region

indentation. Indentations made without heating or made on cross-linked samples, however, do not show this enhanced rim formation and can, therefore, be used to analyze the indentation mechanism by regarding the shape [41, 42]. This observation can be used to illustrate the concept of erasing by the interaction of a freshly written indentation (bit) with a neighboring indent, see Fig. 29.18. This effect is used for erasing in the probe-storage applications mentioned below. Note that, although this is not clear from Fig. 29.18, the surface roughness can be conserved despite erasing of an area [1].

Fig. 29.18. Indentations in a PMMA ﬁlm with varying pitch. The depth of the indentations is 15 nm, about the thickness of the PMMA layer. The indentations on the left-hand side were written ﬁrst, then a second series of indentations were made with decreasing distance to the ﬁrst series. (From [1], © 2002 IEEE)

29.4.6 Indentation Kinetics and the Indentation Mechanism The indentation experiments described above pinpoint some of the most important parameters of thermomechanical nanoindentation. The indentation formation is gov-

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

237

erned by the hardness and the glass-transition temperature of the polymer, as well as the tip geometry and heat-transfer properties of the cantilever/tip. The writing threshold experiment results in a relationship of Tthresh (F) (or F(Tthresh )) that is linear within the uncertainties given. From this observation, an indentation mechanism can be suggested based on the requirement that for any given temperature the force has to be high enough so that the stress in the polymer reaches the yield stress. The yield stress σ y can be observed macroscopically [43,44] to be a nearly linear function of temperature with σ y ∼ 0 for T = Tg . In a simple model, indentation occurs when the maximum stress in the polymer σmax exceeds the yield stress σ y (T ); at the threshold these stresses are equal. The maximum stress is an approximate linear function of F and thus we can write F(Tthresh ) ∝ σ y (T ). This model of yielding is further supported by the analysis of the indentation shapes as a function of indentation parameters (unpublished communication by Altebaeumer). However, such a simple model of yielding contradicts the initial model of “rubbery indentation” proposed earlier [1]. In the latter, the material is not assumed to undergo yielding, but is deformed elastically (like rubber) in the heated state at a temperature above Tg . After cooling to below Tg , the indentation is “frozen in” and the “loaded rubber springs” are kinetically hindered from relaxing. The “rubbery indentation” mechanism can only work above Tg , where the polymer is not in its glassy but in its rubbery state. However, we observe a linear Tthresh (F) all the way down to heater temperatures equivalent to RT , i.e. we ﬁnd writing temperatures lower than the tabulated values of Tg . This appears to be an argument against the “rubbery mechanism”; we must consider, however, that in any case indentation formation is done under signiﬁcant stresses. It can be argued that the stresses can lower the glass-transition temperature. A glass-transition temperature as a function of stress (Tg (σ)) would translate into Tthresh (F) in our experiment. Although at compressive stresses Tg always increases, it is expected that under shear or tensile stress the underlying alpha-transition is eased. The glass-transition, which can be regarded as the intersect between an Arrhenius-type activation energy below Tg , and a WLF kinetics above Tg [40], must shift if the activation energy is lowered. Such a stress-induced lowering of the activation energy under Tg is commonly used to describe stress-relaxation processes in polymers [45]. It might appear unreasonable to distinguish between “rubbery indentation” and yielding when the underlying physics appears to be the same, only being explained using different semantics. However, they produce very different predictions. In the yielding model, the indentation kinetics is essentially Arrhenius-type with a single activation energy [57]. In the “rubbery model” the kinetics would be more suitably described by WLF kinetics generally applied for polymers above Tg [40]. We note that a theory on yielding by Robertson [58] predicts WLF kinetics below Tg under shear, but this theory was found to be useful only near Tg [59]. Thus, to distinguish between “rubbery indentation” and yielding, the indentation kinetics should be investigated. Whereas Arrhenius kinetics would obey Ea 1 1 exp = , τ τ0 kb T

B. Gotsmann · U. Dürig

238

WLF follows T=

log(τref /τ) × T∞ − c1 × Tref . log(τref /τ) − c1

Here, τ, T , and kB are the indentation time, indentation temperature of the polymer and the Boltzmann constant, respectively. The activation energy E a and the WLF parameters τref , Tinf , Tref and c1 are the ﬁt parameters. An example of such an experiment is given in Fig. 29.19. A good ﬁt with WLF can be obtained. The initial experiment on PMMA (Fig. 29.19 and [1]) was extended to highly cross-linked thermosets on the one hand, and the time scale was expanded down by one order of magnitude to 1 µs on the other hand. In a cross-linked system viscous ﬂow can no longer account for the indentation. An example of measured indentation kinetics for a cross-linked polymer (SU8) is given in Fig. 29.20. The experimental limitation is the ﬁnite time constant of the heater within the cantilever: below about 100 µs, the mean heater temperature starts to differ from the maximum temperature reached at the end of the heating pulse. Note, however, that the thermal time constant of the polymer is expected to be clearly below 1 µs, so that the polymer temperature will follow the heater temperature with a constant efﬁciency c. For convenience, both the mean and the average heater temperatures are plotted for a writing threshold example in Fig. 29.20. It becomes clear that both temperatures can be used for a WLF ﬁt of similar quality. The fact that the results cannot be explained with a single activation energy supports the model of rubbery bit-writing. Although the overall results cannot be ﬁtted with an Arrhenius function, one could in principle make Arrhenius ﬁts for the longer heating times. That would suggest a switch in mechanism from yielding for longer to “rubbery indentation” for shorter time scales. Future experiments will aim at distinguishing between these two options. It remains clear, however, that at faster times, at which the writing threshold experiments T(F) were made, the data supports the rubbery model. Note that the WLF shape does not readily tell us which material property is to be made responsible for the indentation. It has been shown that viscosity alone could account for the effect in PMMA [46], but for the highly cross-linked SU8 some function of storage and loss modulus of the polymer is more likely. Interestingly, in this example the indentation temperatures clearly

Fig. 29.19. Indentation kinetics plot for a PMMA sample. The writing threshold heater temperature was measured at different heating times for a given load of 3 nN. A ﬁt to WLF kinetics yields the parameters T∞ = 59 ◦ C, c1 = 10, Tref = 120 ◦ C. Data from [1]

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

239

Fig. 29.20. Indentation kinetics plot for an SU8 sample. The writing-threshold heater temperature was measured at different heating times for a given load of 200 nN. As the time constant of the heating inﬂuences the result, both the temperature at the end of the heating pulse (maximum temperature) and the mean temperature are plotted. A ﬁt to WLF kinetics yields the following ranges for the parameters: T∞ = 62−55 ◦ C, c1 = 10.0, Tref = 87−75 ◦ C, for the mean temperature, and T∞ = 18−33 ◦ C, c1 = 9.55, Tref = 60−70 ◦ C, for the maximum temperature. In both cases τref = 600 s and the heating efﬁciency c ∼ 70%

reach below Tg (∼ 220 ◦ C) as a high load force was used, supporting the conjecture of a stress-induced lowering of Tg . 29.4.7 Interlude: Thermo-Nano-Mechanics Without a Heater In thermomechanical indentation, the highest speed possible and the lowest possible writing temperature of the indentation process are of technological interest, because they relate directly to data rate and power consumption of a storage device. In the thermomechanical experiments, the bit-writing time is clearly limited by the heater characteristics; most of the heat and energy are dissipated in the cantilever and the air, and only a fraction of the heat is used to heat the polymer. For typical cantilever designs, heating times below 1 µs are very inefﬁcient. Nevertheless, using our standard cantilever designs, indentation formation was achieved with heat pulses of 300 ns. The theoretical limit of indentation time and energy for a given polymer temperature are given by the time constant of heat dissipation in the polymer and the heat capacity. If a volume of about V = (10 nm)3 has to be heated to T = RT + ∆T , we can estimate a heating energy of E = V × ∆T × ρ × c p = 10−16 −10−15 J, using a density ρ and a heat capacity c p of typically 1 g/cm3 and 1500 J/(kg K), respectively, and a temperature rise ∆T of 200 K. The mechanical work done by pressing the tip into the polymer is on the same order of magnitude.

240

B. Gotsmann · U. Dürig

The time it takes until a heated volume cools off again can be estimated to be around 1 ns or less. Clearly this is experimentally inaccessible using resistive heaters in cantilevers. The application of load can be much faster: In the experiments we can apply the load within less than 1 µs easily when the tip rests in contact and the electrostatic potential is applied. An example is given in Fig. 29.21a. The indentations are about twice as deep when the voltage is applied as those written with the voltage being ramped up slowly. For the interpretation we consider an adiabatic heating process. Let us assume that the work of compression induces an energy increase. Using the Hertz-model of the tip-surface interaction and typical parameter values for polymers and our tips (tip radius ∼ 10 nm, reduced elastic modulus 6.7 × 109 Pa, loading force ∼ 400 nN), the work of compression can be calculated to be about 10−15 J. The compressed volume is about that of a cylinder having a radius equal to the contact radius of the tip-surface interface and a length of about the tip radius, i.e. the penetration depth of the pressure ﬁeld under the tip. Together with a typical heat capacity of 1500 J/(K m), we calculate a temperature increase of about 250 K. This temperature increase, however, only lasts as long as the time the heat needs to diffuse to the surrounding polymer. With a heat-diffusion constant of polymers of about 10−6 –10−8 m2 /s, we estimate about 1–100 ns dwell time. Using the measured time temperature correspondence from Fig. 29.20, we can estimate a corresponding temperature rise of 25 K on the scale of 10−5 s, which is typically used in our experiments. In this time scale we know that for writing small indentations, a heating of about 0.25 K equals the effect of 1 nN of heating. Hence we have a heat source equivalent to a force increase of ∼ 100 nN. Simply speaking, applying 400 nN fast is equal to applying 500 nN slowly. This is compatible with our data of the depth increase of the indentations as a function of heat or load increase above the writing threshold. The adiabatic effect can even be ampliﬁed by providing additional kinetic energy. In the example shown in Fig. 29.21b, we accelerated the cantilever/tip from about

Fig. 29.21. (a) Indentations made with 400 nN load on an SU8 sample. The load was applied via an electrostatic potential, both slowly and fast. (b) Same experiments with additional kinetic impact effect

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

241

200 nm using the same electrostatic force, which results in a more dramatic effect of indentation enhancement. In that experiment the momentum during the impact is much larger than in the reference experiment in Fig. 29.21a leading to both a larger adiabatic heating effect and a larger net force during indentation.

29.5 Thermomechanical Nanowear Testing For decades wear of polymer surfaces has been examined from various perspectives. This phenomenon combines technological interest with the nature and science of the viscoelastic properties of polymers. As wear is a rather complex topic in general, efforts have been made to perform experiments under deﬁned conditions. The socalled nanowear of a nanometer-sized AFM tip on a surface constitutes a particular way of simplifying wear experiments to the extreme case of a single asperity contact. Whereas macroscopic wear experiments measure mainly material loss and debris formation or scratch resistance, nanowear experiments exhibit a very rich variety of manifestations. For non-crosslinked materials, a ripple formation is typically observed [47]. This ripple pattern can be regarded as a precursor state for abrasive wear and can be identiﬁed even before wear debris forms. Hence nanowear studies could help us to understand macroscopic wear. However, such initial wear can be sufﬁcient to disturb the performance of potential nano- and micromechanical devices. Moreover, initial wear is important to gain a deeper understanding of the physical mechanisms of macroscopic wear. A typical AFM wear experiment is reproduced in Fig. 29.22, and shows an initial wear state characterized by the formation of ripple structures normal to the fast scanning direction. Such ripples have typical heights and spacings of 1–30 and 50–300 nm, respectively, for typical scanning speeds of 1–100 µm/s, loading forces of 10–500 nN and tip radii of 10–50 nm (see [48, 49] and references therein). All reports show a rise of speciﬁc volume in the ripples, making these ripples more compliant [50]. Although the exact wear mechanism remains speculative, a large number of studies have been undertaken to look at the inﬂuence of tip shape, load force, scan repetitions, sliding speed, tip and polymer temperature. The dependence of the rate

Fig. 29.22. Example of the ripple structure of nanowear of polystyrene. The wear was induced by scanning the ﬁeld 30 times with a heated tip (256 ◦ C heater temperature). The length and width of the worn region is 2.5 µm. The gray scale covers 25 nm. (From [49], © by ACS)

242

B. Gotsmann · U. Dürig

of ripple formation on the temperature points to a viscoelastic process linked to WLF polymer kinetics [51]. Nanowear tests are, therefore, a means to study the mobility of polymer chains at the surface, similar to friction experiments [56]. Here, we only show the ripple formation and wear as it is controlled with a heated tip. The heating of the polymer mediated by the tip is local and, therefore, the analysis can be highly localized and limited to the interaction time. In contrast, experiments using heated samples [48, 51, 52] have to take into account not only the build-up of the ripples, but also the decay of the ripples over time during formation and imaging. To elucidate the effect of temperature on the different wear modes, we discuss an experiment in which the tip temperature is continuously increased along the slow scan axis, thereby reducing the three-dimensional position-temperature parameter space by one dimension. We then obtain a two-dimensional real-space image of surface wear, in which each line perpendicular to the fast scan axis corresponds to a certain tip temperature and all relevant wear properties corresponding to that temperature can be inferred from inspecting this line. One example of such a reduced temperature-position wear image measured on a polystyrene sample and imaged after four scan repetitions is shown in Fig. 29.23a. It is convenient to project the fast scan axis by taking the arithmetic mean of the

Fig. 29.23. (a) Worn area of a polystyrene surface after scanning with a variably heated tip. The length and width of the worn region are 5 and 2.5 µm, respectively. The ﬁeld was scanned four times with a scan velocity of 50 µm/s. The gray scale covers 17 nm. The wear modes can be classiﬁed in three regimes: A ripple pattern with an amplitude depending on the number of scans, tip temperature, and load (i); a strongly enhanced rippling in a transition regime (ii); and an abrasive wear mode (iii). (b) Averaged line scan along the slow scan axis for a similar experiment as in (a) but worn with only 5 µm/s. (From [49], © by ACS)

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

243

topographic height in the worn region. The resulting one-dimensional graph is shown in Fig. 29.23b, where the temperature axis corresponds to the vertical slow scan axis in Fig. 29.23a. Consistent with the concept of time-temperature equivalence, i.e. that the effect produced by increasing the number of scans at a ﬁxed temperature is equivalent to the effect produced by increasing the temperature for a ﬁxed number of scans, the ripple amplitude increases with increasing temperature in regime (i) until saturation is attained. From such measurements, we can extract an activation energy for ripple formation under the given conditions (tip shape, sliding speed, load force etc.). Using Arrhenius kinetics we infer an energy barrier on the order of 0.4 eV close to, but below, the glass-transition temperature and a load susceptibility of 70 meV/20 nN. The transition from rippling to pileup wear occurs in a narrow window (regime ii) at ∼ 237 ◦ C heater temperature, possibly a manifestation of the glass transition occurring in the polymer. Here, the amplitude, period, and volume of the ripples increase drastically, somewhat reminiscent of mode softening in second-order phase transitions. The transition is clearly driven by temperature, which also means that, according to time-temperature superposition, more heat is needed for shorter interaction times. Accordingly, we observe that the heater temperature in the transition regime increases with the scan velocity, namely to 253 and 274 ◦ C for 3.3 and 10 µ/s, respectively. Above the transition temperature, pileup wear sets in more or less abruptly (regime iii). This wear mode is “abrasive” in the sense that the height of the polymer ﬁlm is reduced by 1.2 nm per scan. Interestingly, the depth of the wear trough is independent of the temperature. The observed pileup suggests that the rubbed-down material is deposited at the end of the scan ﬁeld. Analyzing the depth of the scan ﬁeld and the amount of pileup at the edges, we arrive at a picture that is difﬁcult to reconcile with a continuously abrasive wear mode but indicates that the rub-down results from a massaging action of the tip on the polymer.

29.6 Application to Data-Storage Devices 29.6.1 Introduction Data storage is a key element in current and future information technologies. The ever-increasing demand for more storage capacity in an ever-shrinking form factor, as well as the pressure to decrease the price per storage capacity ($/MByte) have been the driving force behind worldwide research efforts and development activities. For many decades, silicon-based semiconductor memory chips and magnetic hard drives (HDD) have dominated the data-storage market. The latter have achieved storage-density increases of about 60–100% per year, at more or less constant cost. However, both technologies are approaching their limits in terms of current concepts, materials, and fabrication techniques. For semiconductor memories, such as DRAM, SRAM, and Flash, the challenges are predominantly in lithography and very thin gate oxide materials. For HDDs, the main challenge is the so-called superparamagnetic

244

B. Gotsmann · U. Dürig

effect, which predicts that the bit stability over temperature and time will decrease with increasing densities, because the volume of the magnetic grains tends to shrink with the bit size. The invention of scanning-probe techniques (SPT), such as scanning tunneling (STM), atomic force (AFM) and scanning near-ﬁeld optical (SNOM) microscopy, gave birth to the ﬁeld of nanometer- and atomic-scale imaging. SPT even made atomic- and molecular-scale manipulation possible, leading to quite complex atomic functionality such as atomic/molecular switches and logic building blocks. Originally a tool for nanoscience, SPT is currently being investigated for its potential use in data storage. For this purpose, various storage media, such as phase-change, magnetic, ferroelectric and polymeric media, are being considered. A major drawback of SPT for data storage is the low data rate of a single probe, which is limited by the achievable mechanical resonance frequencies of the microfabricated cantilevers and their write/read speed. The latter is strongly inﬂuenced by the storage medium and by the write/read functionality required by the probe. Single-probe storage approaches currently operate, at best, on the low microsecond time scale, whereas semiconductor and HDD technologies operate at least three orders of magnitude faster. The drawback of low speed/data rate can be overcome by employing large two-dimensional (2D) arrays of cantilevers, fabricated using micro/nano electromechanical system (MNEMS) techniques and operating in parallel, and to have each cantilever perform write/read/erase operations in an individual storage ﬁeld (see, for example, [1]). In this concept, data is stored at ultrahigh density with competitive data rates and access times. Heatable AFM cantilevers/tips, as described in detail in

Fig. 29.24. Schematic of the key elements of a scanned probe storage device. Individual heatable cantilevers are bound to a CMOS electronics chip, so that each cantilever has its own electronic cell. The microfabricated scanner is driven by a voice-coil mechanism. For this, permanent magnets are ﬁxed to the scanner while a coil sits on a base plate carrying the CMOS chip and the levers. A mass balancing scheme ensures that during the scanning motion the centre of mass remains ﬁxed. The polymer media is deposited onto the scanned area. The non-moving part of the scanner chip is connected to the CMOS chip via a spacer structure ensuring that only the tips are in or near contact with the polymer media. Such an assembly could have more than 4000 tips and would ﬁt into the form factor of a secure-digital card (© IEEE 2004 [55])

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

245

Sect. 29.2, serve as the individual read/write heads. The high areal storage density and small form factor make this concept very attractive as a potential future storage technology in mobile applications, offering gigabytes of capacity and low power consumption at data rates of megabytes per second. This device concept, previously referred to as “The Millipede”, is illustrated in Fig. 29.24. It basically consists of a large 2D array of cantilevers and a MEMSbased x/y storage-medium scanner [1]. The electromagnetically actuated scanner moves the storage medium with nanometer precision in both the x- and the ydirection underneath the cantilever-array chip, so that each tip reads and writes in its own storage ﬁeld. Independent and parallel cantilever operation is achieved by a vertically interconnected CMOS read/write channel array chip. The MEMS cantilever array and other essential components of such a device, such as a MEMS-based electromagnetically-activated microscanner, or the recording technology are described elsewhere [1, 53, 54]. Here, our aim is to discuss the basics of thermomechanical nanoindentation in view of the technological challenges of a storage device. 29.6.2 Scaling Challenges for Nanoindentation of Polymers Advantages and trade-offs of using nanoindentation of polymers as a write mechanism for data storage can be illustrated in terms of the scaling challenges applicable to storage technologies in general. Density: Among the main motivators for trying to introduce different types of probe storage devices into the market is the high areal storage density, and, in particular, the fact that it is not limited by lithography. In the case of nanoindentation of polymers, the density limits are basically given by the tip size and the homogeneity of the polymer ﬁlm. Polymers can be amorphous down to the nanometer scale, and thus no limits will appear for the latter until individual indentations are spaced only few nanometers apart, leading at least up to ∼ 10 Tb/in2 . The ultimate limit for mechanical storage could be viewed as mechanical switching of individual molecules. An example of indentations written with a spacing of 18 nm is shown in Fig. 29.25, demonstrating an areal density of 2 Tb/in2 . However, we note that to

Fig. 29.25. AFM image of data stored by indentations at a pitch of 18 nm, leading to a storage density of ∼ 2 Tb/in2

246

B. Gotsmann · U. Dürig

validate the method as a technology candidate, a density demonstration with error rate determination is essential. This has been done successfully with a density of more than 600 Gb/in2 [55]. Data Rate: As mentioned above, the low data rate of scanned probes is one of the weak points of the approach. Even if it can be overcome by parallelization, a high speed of individual read/write events is desirable. The demonstrations mentioned above of a write/read event in the range of around 10 µs are not to be considered at the cutting edge of the technology yet. Megahertz heating rates of cantilevers/heaters have been demonstrated elsewhere, and the fastest thermomechanical indentations written are below 1 µs (see Sect. 29.4.7). Even if the heating time would be limited to 100 ns, fast data writing could be achieved using ultrafast load force pulses. In terms of the mechanics of polymer indentation, writing the physical limit is unclear. From our experiments we conclude that at least down to 1 µs it will not present a problem and the polymer dynamics follows WLF kinetics. At some limit below 1 µs it will be difﬁcult to provide the polymer chains with sufﬁcient mobility by heating it up without breaking of the chains. Long-Term Stability: The long-term stability of written data is given by the polymers’ mobility below the glass-transition temperature. This mobility is fundamentally given by the activation energy of a backbone wiggle, i.e. the so-called alpha-relaxation. Depending on the polymer, this can be as be as much as several electron volts, and would thus be high enough for typical lifetime requirements. Power Consumption: The power consumption of writing, as limited by the heat required to increase the temperature of the polymer, is ultimately around 10−15 J/bit, as discussed above. The challenge seems to be an effective heat generation within the cantilever and the heat transport through the tip-surface contact. In both these aspects, current cantilever/tip designs are not yet at the physical limits. We note that orders of magnitude above the apparent limit of 10−15 J/bit, the power consumption of scanner and CMOS chip will predominate. In any case, an independent optimization of the heater for writing and reading appears advantageous. A corresponding cantilever design is shown in Fig. 29.26. Tip/Media Wear: Technologically nanowear of the polymer during reading or writing is one of the possible limitations for a device lifetime. Read/write endurance is currently under intense investigation. Here we point out two strategies to minimize media wear.

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

247

Fig.29.26.Scanning electron micrograph of a cantilever structure as used in prototyping. The three different electrical connections are used to separately heat a read and write resistor carrying the tip. Heating the write resistor by applying a voltage to the middle electrode leads to a simultaneous application of a capacitive force between a platform structure and the substrate under the polymer media. The overall ﬂexural stability of the cantilever is ensured by having all of it except the hinges relatively thick (© IEEE 2004 [55])

One strategy is to increase wear resistance, without interfering with the thermomechanical basis of nanoindentation, is to use cross-linking within the polymer. For example SU8, a highly cross-linked epoxy, does not fail during reading or overwriting on the same area for at least 104 times. Alternatively, it helps to minimize the interaction of the probing tip by reducing the load force and tip temperature during reading with respect to writing. For minimizing the load force, a low stiffness of the cantilever is advantageous. If higher forces are needed to indent the softened polymer, they can be switched on by application of a bias voltage and the resulting electrostatic attraction. In AFM, the force is usually controlled by ramping the cantilever holder away from or towards the sample. This is technically too complicated for implementation in a probe array. A much easier way is switching the force induced by an electrostatic potential between substrate and cantilever. Similarly, heating of the tip is only required during writing and not during reading. For this, reading should be done with an unheated tip, which can be realized by implementing a separate heater-sensor not attached to the tip into the cantilever. A cantilever design comprising these operations is shown in Fig. 29.26. The tip lifetime could be arbitrarily high, if rod- or tube-shaped tips are used, because then wear would not result in a change of the sharpness. Apart from that, harder materials promise 1000-fold improvement over silicon if this should turn out to be necessary. Both these approaches are being persued at various research establishments around the world. In summary, SPT and polymers have great scaling potential for several generations of a storage device.

B. Gotsmann · U. Dürig

248

Acknowledgements. This chapter contains otherwise unpublished data obtained by Martin Hinz and Mark A. Lantz, for which we are very grateful. Progress in the ﬁeld of polymer media was enabled by our colleagues J. Frommer, C. Hawker, J. Hedrick, E. Hagberd, H. Ito, V. Lee, H. Wolf, who made or characterized the polymers described here. Understanding of the indentation experiments is based on the work of T. Altebäumer, G. Binnig, G. Cross, W. Häberle, D. Jubin, W. P. King, A. Kleiner, A. Knoll, J. Mamin, H. Pozidis, and D. Wiessmann. For designing and making the cantilevers we thank M. Despont, U. Drechsler, R. Rothuizen, and R. Stutz. It is further a pleasure to acknowledge the help, support and contributions from T. Albrecht, P. Bächtold, C. Bolliger, R. Bradshaw, G. Cherubini, A. Dholakia, E. Elefteriou, C. Hagleithner, S. Hild, O. Marti, R. Miller, A. Pantazi, A. Sebastian, P. F. Seidler, P. Vettiger, C. Wade, and J. Windeln.

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. 26. 27. 28.

Vettiger P et al (2202) IEEE Trans Nanotechnol 1:39 Briscoe BJ (1998) J Phys D: Appl Phys 31:2395 VanLandingham MR et al (2001) Macromol Symp 167:15 Klapperich C et al (2001) Trans ASME 123:624 Fischer-Cripps AC (2002) Nanoindentation. Springer, Berlin Heidelberg New York Hinz M, Kleiner A, Hild S, Marti O, Dürig U, Gotsmann B, Drechsler U, Albrecht TR, Vettiger P (2004) Eur Pol J 40(5):957 Pollock HM, Hammiche A (2001) J Phys D: Appl Phys 34:R23 Shi L, Majumbdar A (2004) In: Bhushan B, Fuchs H, Hosake, S (eds) Applied Scanning Probe Methods. Springer, Berlin Heidelberg New York Joon Hyung Bae, Ono T, Konoma C, Esashi M (2003) Proceedings of the 16th Annual International Conference on Micro Electro Mechanical Systems, MEMS-03, Kyoto IEEE, p 24, ISSN:1084–6999 Cahil DC et al (2003) J Appl Phys 93:793 Sze SM (1981) Physics of Semiconductor Devices. Wiley, New York Dürig U (2005) unpublished Kennard EH (1938) Kinetic Theory of Gases. McGraw-Hill, New York Chen G (2000) Int J Therm Sci 39:471 Polder D, Van Hove M (1971) Phys Rev B 4:3303 Loomis JJ, Maris HJ (1994) Phys Rev B 50:18517 Volokitin AI, Persson BNJ (2004) Phys Rev B 69:045417 Pendry JB (1999) J Phys: Cond Matter 11:6621 Mulet J-P et al (2002) Microscale Thermophys Eng 6:209 Hargreaves CM (1973) Philips Res Suppl 5 Xu JB et al (1994) J Appl Phys 76:7209 Mueller-Hirsch W et al (1999) J Vac Sci Technol A 17:1205 DiMatteo RS et al (2001) Appl Phys Lett 79:26 Volz S, Chen G (1999) Appl Phys Lett 75:2056 King WP (2002) Thermomechanical Formation of Polymer Nanostructures. PhD Dissertation, Stanford University Patton KR, Geller MR (2001) Phys Rev B 64:155320 Hu C, Kiene M, Ho PS (2001) Appl Phys Lett 79:4121 Govorkov S et al (1997) Rev Sci Instrum 68:3828

29 Nano-Thermomechanics: Fundamentals and Application in Data Storage Devices

249

29. Yovanovich MM, Culham JR, Teerstra P (1998) IEEE Trans Components, Packaging and Manufacturing Techn, Part A 21:168 30. Ross RG et al (1984) Rep Prog Phys 47:1347 31. Cappella B, Dietler G (1999) Surf Sci Rep 34:1 32. Shi L, Majumdar A (2002) J Heat Transfer 124:329 33. Zheng F (2002) Adv Coll Int Sci 97:255 34. Sone Y, Akoi K (1981) In: Fisher SS (ed) Rareﬁed Gas Dynamics. Progress in Astronautics and Aeronautics, vol 74. AIAA, New York, p 489 35. Gotsmann B, Dürig U (2005) Appl Phys Lett 87:194102 36. Passian A et al (2003) Phys Rev Lett 90:124503, Passian A et al (2002) J Appl Phys 92:6326 37. Loyalka SK (0977) J Chem Phys 66:4935 38. Lantz M, Gotsmann B, Dürig U, Vettiger P, Nakayama Y, Yoshikazu S, Tetsuo T, Tokumoto H (2003) Appl Phys Lett 83:1266 39. Shimizu T, Tokumoto H, Akita S, Nakayama Y (2000) Surf Sci 486:L455 40. Ferry JD (1980) Viscoelastic Properties of Polymers. Wiley, New York 41. Sills S, Overney RM, Gotsmann B, Frommer J (2004) Tribol Lett 1:9 42. T Altebaeumer, unpublished 43. Kody RS, Lesser AJ (1997) J Mat Sci 32:5637 44. Brooks NWJ, Duckett RA, Ward IM (1998) J Pol Sci B 36:2177 45. van Krevelen DW (1997) Properties of Polymers. Elsevier, Amsterdam 46. Mackay ME (2005) IEEE Trans Nanotechnol 5:641 47. Leung OM, Goh C (1992) Science 25:64 48. Schmidt RH, Haugstad G, Gladfelter WL (2003) Langmuir 19:10390 49. Gotsmann B, Dürig U (2004) Langmuir 20:1495 50. Iwata F, Matsumoto T, Sasaki A (2000) Nanotechnology 11:10 51. Schmidt RH, Haugstad G, Gladfelter WL (1999) Langmuir 15:317 52. Wang XP, Loy MMT, Xiao X (2002) Nanotechnology 13:478 53. Eleftheriou E et al (2003) IEEE Trans Magn 39:938 54. Eleftheriou E et al (2003) In: Proceedings 29th VLDB Conference, Berlin 55. Pozidis H, Häberle W, Wiesmann D, Drechsler U, Despont M, Albrecht TR, Eleftheriou E (2004) IEEE Trans Magn 40:2531 56. Sills S, Overney RM (2003) Phys Rev Lett 91:095501 57. Ree T, Eyring H (1955) J Appl Phys 26:793 58. Robertson RE (1966) J Chem Phys 44:3950 59. Argon AS, Bessonov MI (1977): Pol Eng Sci 17:174

30 Applications of Heated Atomic Force Microscope Cantilevers Brent A. Nelson · William P. King

Abbreviations AFM DPN PMMA SEM SThM tDPN TOF-MS

atomic force microscope dip-pen nanolithography poly(methyl methacrylate) scanning electron microscope scanning thermal microscopy thermal dip-pen nanolithography time-of-ﬂight mass spectrometer

30.1 Introduction Heated atomic force microscope (AFM) cantilevers can provide mechanical actuation, sensitive measurement of heat ﬂow, and thermal processing through local heating. Mechanical actuation by temperature-induced bending has been used for sensing temperature, inducing oscillation, controlling individual cantilevers in arrays, and actuating during high speed AFM imaging. Local measurement of heat ﬂow with cantilevers having both heaters and thermometers can yield information about thermal properties or the topography of a substrate. Thermal processing of materials on the cantilever has been used for calorimetry and chemical detection, while thermal processing of substrates has used various phase transitions to perform lithography, surface characterization, and data storage. Heat sources for AFM cantilevers can be either internal or external to the cantilever. Internal heat sources have been electrically-resistive heaters made from metal or doped silicon. External heat sources have included focused lasers, and this review also includes global heating from furnaces or proximal heaters. The transduction mechanisms for heated AFM cantilever applications have been either mechanical or electrical. Mechanical responses have included shifts in cantilever resonance due to mass changes, used during mass and chemical sensing, and cantilever deﬂections due to temperature-induced bending, used during calorimetry of on-cantilever samples. Electrical responses have been important in lithography and in thermal property and topographical mapping, using the temperature-sensitive resistance of on-cantilever resistors as a thermometer.

B.A. Nelson · W.P. King

252 Table 30.1. Use and Applications of heated Probes Use

Applications

Actuation

High-speed imaging (Sect. 30.2.3.2) Mass/chemical sensing (Sect. 30.3.2.2) Multi-functional cantilevers (Sects. 30.3.3, 30.4.2)

Heat ﬂow sensing

Thermal property measurement (Sect. 30.2.2) Subsurface imaging (Sect. 30.2.2) Topographical imaging (Sects. 30.2.3.3, 30.4.1)

Substrate thermal processing

Calorimetry (Sect. 30.3.1) Data storage (Sect. 30.4.1) Lithography (Sect. 30.4.2)

On-cantilever thermal processing

Oxidative tip sharpening (Sect. 30.2.3.1) Calorimetry (Sect. 30.3.1) Thermogravimetry (Sect. 30.3.2.1) Explosives detection (Sect. 30.3.4)

30.2 Physical and Environmental Sensing 30.2.1 Pressure Sensing When two surfaces at different temperatures are separated by a distance close to the mean free path of a surrounding gas, the molecular collisions on the surfaces depend upon the two temperatures of the surfaces. The temperature-dependence of these molecular collisions inﬂuences the apparent pressure between the surfaces and is often referred to as a Knudsen force. The Knudsen number Kn is the ratio of the gas mean free path λ to the gap width d. In an ideal gas, the Knudsen number can be expressed by λ η πR0 T , (30.1) Kn = = d Pd 2Mm where R0 is the universal gas constant, η, T , P and Mm are the viscosity, temperature, pressure and molecular mass of the medium, respectively. Knudsen forces are maximal in the transitional region, signiﬁed by 0.01 < Kn < 0.5, where both surface and intermolecular interactions are signiﬁcant. A temperature rise in an AFM cantilever can occur either due to heating from an interrogating laser or from resistive heating in the cantilever, if it is electrically active, as is the case for piezoresistive cantilevers [1, 2]. For AFM operation at low pressure, the mean free path of the surrounding gas molecules is comparable to the cantilever-substrate distance, resulting in a Knudsen number near unity. The combination of these two factors can result in cantilever deﬂections due to Knudsen forces during AFM under vacuum.

30 Applications of Heated Atomic Force Microscope Cantilevers

253

Several experiments characterized the conditions under which Knudsen forces could induce measurable cantilever deﬂection and suggested how this approach might be used to sense pressure [3–6]. In one experiment, the cantilever was placed inside a bell jar under vacuum and heated with a laser while the cantilever deﬂection was measured with a capacitive sensor [4]. By varying the pressure inside the system, the authors studied the dependence of the Knudsen forces on pressure from the cantilever deﬂection due to the imposed laser heating. Shown in Fig. 30.1, the magnitude of the Knudsen forces exhibited the expected peak in the transitional regime and a linear dependence at pressures below the peak for air, Ar, He, N2 , O2 , and CO2 [3]. A later experiment demonstrated an increase of the Knudsen forces as the cantilever temperature was increased at constant pressure by varying the power of the heating laser [6]. The linearity and sensitivity of the pressure-dependence of cantilever deﬂection below the transitional regime demonstrates the opportunity to use heated cantilevers for pressure sensing.

Fig. 30.1. Knudsen force variation with environmental pressure [3]. The signal on the y-axis is a measure of the amplitude of the cantilever vibration at the frequency at which the heating laser is chopped

30.2.2 Thermal Conductivity Mapping and Subsurface Imaging Scanning thermal microscopy (SThM) has been used for high spatial resolution thermometry and for measuring thermal properties of surfaces [7]. Some SThM techniques use passive thermocouple probes for temperature measurement, but this review focuses on resistive temperature sensors and their use as active heating sources for measuring thermal properties of surfaces. This section discusses (i) fabrication of scanning thermal probes, (ii) DC heating of scanning thermal probes, (iii) AC heating of scanning thermal probes, and (iv) modeling of scanning thermal probes. 30.2.2.1 Fabrication of Scanning Thermal Probes The ﬁrst probe with an integrated resistive heater designed for simultaneous topographical and thermal imaging was developed in 1994 [8]. Figure 30.2 shows this

254

B.A. Nelson · W.P. King

Fig.30.2.Thermal probes that have been fabricated for thermal property mapping include (a) Wollaston wire probes [8], (b) platinum ﬁlament probes grown by focused electron beam [11], (c) batch-fabricated palladium resistor probes [10], and (d) ﬂexible polyimide probes [12]

probe, made from a Wollaston process wire, consisting of a 75 µm Ag sheath and a 5 µm Pt core. The wire was bent at a sharp angle and the silver was etched away at the apex of the bend, yielding a 200 µm long section of exposed Pt wire to form the tip. The exposed Pt wire acts as a temperature-sensitive resistor or a resistive heater, depending on whether the current passed through the wire is low or high.

30 Applications of Heated Atomic Force Microscope Cantilevers

255

Low electrical currents enable measurement of resistance with negligible heating, whereas high electrical currents cause signiﬁcant heating while still enabling resistance measurement. A mirror afﬁxed across the wire legs allowed probe deﬂection to be monitored with the standard optical lever technique to measure topography. Although an effective localized heat source, this ﬁrst wire probe had two major drawbacks. First, Wollaston wire probes yield poor lateral resolution, near 1.5 µm [9], which is much worse than lateral resolution limits in topographical imaging. The wire probes also require individual assembly. Addressing both challenges, a later technique used batch fabrication micromachining technology and multiple-level electron beam lithography to pattern a tapered paladium wire across the apex of a cantilever, shown schematically in Fig. 30.2 [10]. Taper widths of 150 nm were typical with electrical resistance in the range of 250–1000 Ω. Another technique developed to improve upon the lateral resolution of the Wollaston probes used a focused electron beam to deposit a Pt ﬁlament onto an AFM cantilever with a patterned Al layer to electrically contact the ﬁlament [11]. The ﬁlaments, shown in Fig. 30.2, included four-legged structures that enabled four-point electrical measurements and enhanced the mechanical rigidity and thermal sensitivity. Platinum tips deposited at the apex of the ﬁlament yielded a tip radius of ∼ 20 nm. The above-described probes were not well-suited for use in aqueous and biological environments because of limited thermal isolation, exposed electrical connections, and relatively high spring constants. To address these limitations, Li et al. sandwiched a thin ﬁlm Ni/W resistor between two layers of polyimide, shown schematically in Fig. 30.2 [12, 13]. The low thermal conductivity of the polyimide provided thermal isolation from the ambient medium, while the low stiffness of the material enabled imaging of soft biological samples. Electrical isolation was achieved by also coating the cantilever attachment base in polyimide. This probe generated thermal conductivity maps of a HeLa tumor cell in air and also metal lines on a Si substrate in liquid. 30.2.2.2 DC Heating of Scanning Thermal Probes Heat ﬂow from a heated probe can be related to its temperature by Q = G Tprobe − T∞ = G∆T ,

(30.2)

where Q is the heat ﬂow from the probe, Tprobe is the probe temperature, T∞ is the ambient temperature, and G is the thermal conductance from the heater. If ∆T is held constant by heating the probe to a constant temperature, then changes to G during imaging will also cause changes to Q [9, 14–17]. Alternatively, if Q is held constant by supplying a constant power to the probe, then changes to G during imaging will change the probe temperature. In either case, when the heater is located at the tip, the primary source of variation in the thermal conductance stems from variation in the thermal conductivity of the substrate, although there is a secondary contribution from the topography, which is discussed in Sects. 30.2.3.3 and 30.4.1. By recording the required power to maintain constant cantilever temperature or by recording the temperature during supply of constant power during imaging, a qualitative map of

256

B.A. Nelson · W.P. King

Fig. 30.3. Thermal images of subsurface copper particles at depths, from left to right, of 400 nm, 1 µm, and 4 µm [9]

the thermal conductivity of a surface can be generated. This technique can also detect thermal conductance variation due to subsurface inhomogeneities. The ﬁrst such thermal conductivity map and subsurface image was generated by Nonnenmacher and Wickramasinghe in 1992 [15]. Using a laser-heated AFM cantilever and measuring the cantilever temperature from the electrical contact potential between the tip and the scanned sample, the technique detected a SiO2 structure buried underneath a tungsten ﬁlm. The potential for subsurface imaging was more thoroughly examined by experiments with a Wollaston probe heated to 20 ◦ C above ambient to perform thermal imaging of copper particles buried at different depths [9]. The probe demonstrated a depth of vision of a few microns. Figure 30.3 shows thermal images of the subsurface particles. The Wollaston probe also characterized carbon ﬁbres within carbon composites, yielding signiﬁcantly improved contrast over the topographical image, shown in Fig. 30.4 [16]. The thermal images gave information that was used to determine the processing history of unknown carbon composite samples.

Fig. 30.4. Thermal (left) and topographical (right) images of carbon–carbon composites [16]

30.2.2.3 AC Heating of Scanning Thermal Probes When an AC current induces periodic heating, the amplitude and phase lag of the temperature response in the cantilever yield information about the thermal diffusivity

30 Applications of Heated Atomic Force Microscope Cantilevers

257

of the surface, compared to DC heating which yields information aobut thermal conductivity [10,14,18]. AC heating also permits lock-in techniques that can improve the signal-to-noise ratio over that obtained using DC heating. Additionally, the temperature-dependence of the phase response can provide information about phase transitions, as discussed in Sect. 24.1. In the ﬁrst use of AC heating for thermal property mapping, a chopped laser heated a thermocouple probe to illuminate grain boundaries during imaging of a topographically ﬂat diamond surface [18]. In another study, the phase lag between heating and thermal response of a Wollaston probe created the polymer blend thermal image shown in Fig. 30.5 [14]. In an additional method for mapping thermal properties, later research monitored cantilever deﬂection while a high frequency AC current induced thermal expansion of the substrate during imaging [10]. This method differs from scanning joule expansion microscopy (SJEM) in that the probe is heated rather than the sample [19]. The sample expansion was related to the thermal diffusivity, which modulates the amplitude of the temperature variation, and also the coefﬁcient of thermal expansion, which determines the degree of expansion for a given temperature variation. Contrast in such expansion images indicated variations in both the thermal diffusivity and the thermal expansion coefﬁcient.

Fig. 30.5. Thermal DC heating (left) and AC phase (right) images of a poly(methyl methacrylate) (PMMA)/chlorinated polyethylene (CPE) blend taken by a heated Wollaston wire probe [14]. The PMMA has a higher thermal conductivity than the CPE, and corresponds to the bright islands. The scan size was 100 µm × 100 µm

30.2.2.4 Modeling Scanning Thermal Probes In spite of the advances in thermal property mapping and subsurface imaging, results have remained largely qualitative and not quantitative. One primary challenge is for Wollaston probes, where theoretical estimates show that a signiﬁcant portion of the supplied power is lost through the silver sheath to the surrounding air, resulting in nonlinearity between substrate thermal conductivity and required heating power to

258

B.A. Nelson · W.P. King

maintain constant probe temperature [7,20]. One study attempted to develop a semiempirical correlation relating substrate thermal conductivity to the ratio of the power required to maintain constant probe temperature in contact with the substrate to the power required when out of contact with the substrate [21]. Although the empiricallyﬁt model matched the general trend of power ratio variation with substrate thermal conductivity, the model could not give accurate determination of thermal conductivity from a given power ratio. Another approach for quantifying substrate response to heated probes developed an approximate analytical model for the temperature distribution in the substrate during DC and AC heating, enabling rapid results for variable experimental conditions [22]. This model, however, neglected conduction from the sides of the tip, and power dissipation through the probe leads. The lack of accurate models that enable quantitative thermal measurements is the primary shortcoming in active-probe SThM. 30.2.3 Topographical Detection Heated probes have seen multiple uses for topographical imaging. The three primary applications have been in (i) tip sharpening, (ii) thermal cantilever actuation, and (iii) topographic imaging through thermal proximity sensing. 30.2.3.1 Tip Sharpening Cantilever tip sharpness and adhesive interactions determine lateral resolution in AFM. Silicon cantilever tips can be sharpened by oxidizing the silicon, then removing the oxide. Oxidation is usually performed in a furnace, but oxides can also be grown on cantilever tips by localized heating [23–25]. A technique to do this used local heat from an overbiased piezoresistive cantilever under low oxygen pressure to oxidize, and then used local heat under vacuum to desorb the oxide [23]. The tip was ﬁrst cleaned by Ar ion sputtering under vacuum to remove native oxide and contaminants. The tip was then heated under O2 pressure of 5 × 10−5 torr to form a thin protective oxide layer, after which introduction into laboratory air for 10 min caused no adverse effects from contamination. The cantilever was then placed back under vacuum and again heated to desorb the oxide layer, resulting in a sharp, clean Si tip. Oxide desorption was later improved through stimulation with an electron beam [25]. After cleaning, the authors demonstrated enhanced imaging resolution reduced noise when measuring the resonance frequency shift of the cantilever as a function of tip-substrate separation, although no effort was made to quantify changes to the tip sharpness. 30.2.3.2 Thermal Actuation A limitation in conventional atomic force microscopy is the image acquisition speed, which is primarily limited by the ∼ 600 Hz bandwidth of the vertical piezotube

30 Applications of Heated Atomic Force Microscope Cantilevers

259

scanner. Cantilever arrays can increase throughput but require deﬂection sensing and actuation for each cantilever. To address these challenges, piezoelectric and thermal bimorph vertical actuators have been fabricated directly onto individual cantilevers, enabling faster response times times due to smaller actuated masses and allowing individual cantilever actuation [26–31]. Thermal bimorph actuators use layers of materials with dissimilar coefﬁcients of thermal expansion to induce bending during heating. On-cantilever piezoelectric actuation has demonstrated a bandwidth of 33 kHz, a signiﬁcant improvement over piezotube vertical actuation [32]. However the improvement in bandwidth comes at the expense of total vertical range as the cantilever resonant frequency scales linearly with the cantilever thickness while the maximum vertical range scales with the inverse of the thickness [29]. To achieve both wide bandwidth and large vertical range, cantilevers were fabricated with both piezoelectric actuators for responding to small amplitude, high frequency topographical variation and thermal actuators for large scale, low frequency variation, achieving a 15 kHz imaging bandwidth at a vertical range of a few microns [29]. Figure 30.6 shows a constant force image taken with one of these cantilevers at 2 mm s−1 tip velocity with 2.5 µm peak to valley height. The cantilevers with integrated piezoelectric and thermal actuators were not well suited for array operation due to the electrical connections required for multiple actuators on each cantilever and the need for external deﬂection sensing. Oncantilever sensing and actuation was achieved using CMOS processing to fabricate a cantilever with integrated thermal actuation and piezoresistive deﬂection sensing [26]. Although the on-chip sensing and actuation enabled scaling up to large cantilever arrays, the effort was not focused on high speed, so to improve the dynamic performance of thermal actuation an active ﬁlter was added, achieving a 5 kHz imaging bandwidth [27]. Because the thermomechanical response of the cantilever occurs over a ﬁnite amount of time, a high voltage can be initially applied to the actuator to more rapidly induce a deﬂection, and then the voltage can

Fig. 30.6. AFM images taken at high speed using thermal actuation for vertical control. Left: AFM image of metal lines and contact holes with over 2 µm variation, taken at 2 mm s−1 using parallel thermal and piezoelectric vertical actuation [29]. Middle: AFM image of 5 µm square pits with 180 nm step height taken at .62 mm s−1 using a tuned boost ﬁlter and purely thermal actuation [27]. Right: alternate-contact mode AFM image of a 40 nm tall chromium pattern with 1 µm period imaged at .1 mm s−1 tip velocity [31]

260

B.A. Nelson · W.P. King

be stepped down to a lower value to maintain that deﬂection. The initially high voltage causes the cantilever to deﬂect more quickly compared to if the lower voltage were held constant the whole time. The boost ﬁlter achieved this temporary overshoot effect by amplifying high frequency components of the actuator driving voltage. Figure 30.6 shows an image of a 180 nm square pit taken at .62 mm s−1 with a properly tuned ﬁlter. To further enable improvements to AFM imaging throughput, CMOS processing was used to fabricate a cantilever array with multiplexing, thermal actuation and piezoresistive wheatstone bridge deﬂection sensing integrated on-chip [30]. An analog controller with a zero in its transfer function to cancel the pole in the transfer function of the thermal actuator enabled high imaging bandwidth, but due to variation among cantilevers, a digital controller with individualized parameters for each cantilever was recommended for pseudo-parallel multiplexed scanning. With the analog controller optimized for a single cantilever, a surface with 800 nm of vertical range was imaged at .6 mm s−1 . To minimize contact forces while still achieving on-cantilever actuation for maximum imaging bandwidth, alternate-contact mode imaging with integrated thermal actuation has also been demonstrated [31]. An AC current matched to the resonant frequency of the cantilever excited oscillation, while a DC current actuated the cantilever for tracking topography. Figure 30.6 shows alternate-contact mode images taken at .1 mm s−1 using the phase lag between oscillation current and cantilever deﬂection as the control signal. 30.2.3.3 Topographic Imaging Through Thermal Proximity Sensing As mentioned in Sect. 30.2.2.2, topography can create artifacts in thermal images. When the tip of a warm cantilever follows the contours of a surface, changes in the gap between the warm cantilever and the substrate induce changes in the thermal conductance, which affects the heat transfer out of the probe. This is identical to the situation discussed in Sect. 30.2.2.2, except that the variation in thermal conductance is due to topographical variation rather than substrate thermal conductivity variation, and thermal images therefore represent information about topography. In this way, the temperature of a heated cantilever can be used as the control signal for topographical imaging. The use of proximity-dependant thermal conductance for surface proﬁling was ﬁrst demonstrated with a heated thermocouple tip elevated off the substrate by 100 nm or more [33]. The approach demonstrated 3 nm height resolution and 100 nm lateral resolution. A similar technique with a heated cantilever in contact with the substrate was later used to simplify the readback mechanism in the thermomechanical data storage application described in more detail in Sect. 30.4.1 [34]. Since the data storage technology already used an integrated resistive heater to form data bits, relying on thermal topographical imaging eliminated the need for additional deﬂection-sensing elements. The integrated topographical sensing enabled scaling to a 32 × 32 array [35] and may yield several orders of magnitude improvement over piezoresistive detection [36–38]. Section 30.4.1 gives further discussion of thermal imaging of topography.

30 Applications of Heated Atomic Force Microscope Cantilevers

261

30.3 Chemical Sensing Applications Heated cantilevers can measure chemical as well as physical properties. This section discusses (i) calorimetry, (ii) mass detection, (iii) time-of-ﬂight scanning force microscopy, and (iv) explosives detection. 30.3.1 Calorimetry Calorimetric measurements using heated AFM cantilevers can be divided into two groups: those that perform calorimetry on small masses situated on the cantilever itself, and those that use the cantilever as an active heat source to perform calorimetry on substrates [10, 14, 39–44]. The thermal mass of the cantilever is small enough to detect phase transitions of very small amounts of material due to the energetics of the enthalpy change associated with the transition. An AFM cantilever was sensitive enough to detect rotational phase transitions of 7.8 ng of C21 H44 from discontinuities in the temperature variation of the cantilever on which the sample was situated during constant heating by a proximal heater [40]. The rotational transition was also detected for C24 H50 and C23 H48 [41]. An empirical calibration constant was suggested to attempt to quantitatively estimate the phase transition enthalpy change from the time-dependent temperature response of the cantilever during heating, although the work was still highly qualitative. When the cantilever is used as an active heater to probe large substrates, substrate phase transitions can be sensed through either the phase lag between an AC heating current and the cantilever temperature response, or through changes in the volumetric expansion and mechanical compliance of the substrate. When an AC heating current is applied to a heated probe, changes in the heat capacity that occur during phase transitions cause the phase lag between the cantilever heating current and temperature response to change sharply. A Wollaston

Fig. 30.7. The derivative of the phase lag between heating and sensing plotted against the probe temperature for three separate locations on a quenched poly(ethylene terephthatate) sample. The events, associated with glass transition, recrystallization, and melting, are reproducible, and correspond to similar peaks in conventional differential scanning calorimetry [39]

262

B.A. Nelson · W.P. King

probe heated with an AC heating current superimposed onto a temperature ramp yielded phase vs. temperature plots with sharp spikes attributed to glass transition, recrystallization, and melting temperatures for various polymer blend substrates, shown in Fig. 30.7 [39]. Substrate calorimetry with heated probes has also been carried out by holding the probe in contact with the substrate and then monitoring probe deﬂection while ramping the probe temperature. As the substrate softens, the high initial loading force causes the probe to penetrate into the substrate, which causes a discontinuity in the deﬂection signal of the probe. Recrystallization causes observable discontinuities in the thermal expansion behavior. This technique has identiﬁed glass transition, recrystallization, and melting temperatures of various polymers [10], and a similar experimental technique examined the softening temperature of carbon ﬁbers to determine the depth of oxidative stabilization [44]. 30.3.2 Mass Detection The mass of an AFM cantilever is sufﬁciently small that its mass-dependent resonant frequency is sensitive to picogram changes in mass, such as those due to dehydration, adsorption or sublimation. Heat can be used either to induce these changes in mass, or to actuate the cantilever so that the resonant frequency can be monitored [2, 45–49]. 30.3.2.1 Thermogravimetry Thermogravimetry involves the monitoring of sample weight loss as a function of temperature. Utilizing the mass sensitivity of the cantilever, an overbiased piezoresistive cantilever performed thermogravimetry on 420 ng of copper-sulfatepentahydrate (CuSO4 · 5H2 O), detecting discrete dehydration steps during heating [2]. The resistivity of the piezoresistor is temperature-dependent, allowing its

Fig.30.8.Resonance frequency (top) and derivative of the calculated mass loss with respect to temperature (bottom) for a cantilever with 420 ng of CuSO4 · 5H2 O glued to its platform. During heating, the sample dehydrates and the mass is decreased, causing the resonant frequency to increase [2]

30 Applications of Heated Atomic Force Microscope Cantilevers

263

use both as a heater and as a thermometer after calibration. Using a piezoelectric actuator to oscillate the cantilever, dehydration of the copper sulfate caused two discrete jumps in the resonant frequency during heating, shown in Fig. 30.8. In an effort to improve Joule-heated cantilevers for thermogravimetry, ﬁnite element analysis modeled a cantilever design with a slit near the apex to create a thermal and electrical constriction, resulting in a more uniform temperature on the sample platform and an improved heating efﬁciency [45]. 30.3.2.2 Actuation Adsorption of ambient chemicals onto functionalized cantilevers can induce measurable cantilever bending or resonant frequency shifts, making cantilevers highly sensitive chemical sensors [50]. One study used thermal actuation to measure changes in the resonant frequency of a cantilever under different concentrations of ethanol, n-octane, ethyl acetate, and toluene [47]. The cantilever was coated with poly(etherurethane) to make it chemically sensitive to the chosen organic analytes, and adsorption of the analytes from the air caused an increase in mass of the cantilever, inducing a resonance frequency shift. Figure 30.9 shows that the frequency shift is linearly dependent on analyte concentration, with a distinct linear coefﬁcient for each analyte.

Fig. 30.9. Cantilever resonant frequency shift due to adsorption of, from left to right, toluene, n-octane, ethyl acetate, and ethanol [47]

30.3.3 Time-of-Flight Scanning Force Microscopy To enable determination of chemical compositions of surfaces during AFM, cantilever thermal actuation was used to integrate a time-of-ﬂight mass spectrometer (TOF-MS) with AFM for combined topographical and chemical analysis [51]. The cantilever, shown in Fig. 30.10, was mounted onto a custom microstage and was thermally actuated between imaging and extraction positions. In the imaging position, the cantilever was pointed downwards to image the surface and grab chemical

264

B.A. Nelson · W.P. King

Fig. 30.10. Schematic (left) and SEM image (right) of TOF scanning force microscope apparatus. Thermal actuation switches the cantilever from an imaging position to the extraction electrode where adsorbed ions are desorbed into a TOF mass spectrometer [52]

species off the substrate. In the extraction position, the thermal bimorph switched the cantilever up to the extraction electrode where chemical compounds on the tip were ionized and accelerated towards the TOF-MS to determine the chemical species picked up during imaging. The proximity between the tip and extraction electrode in this system signiﬁcantly reduced the required voltage for desorption and ionization over standard TOF-MS. Piezoresistors monitored cantilever deﬂection during imaging, as the presence of the microstage prevented the use of optical techniques. The technique detected Si, C, and Pt using a Pt coated tip [52]. 30.3.4 Explosives Detection In addition to detecting the energetics of reversible phase transitions, cantilevers can also detect the heat released by deﬂagration of trace quantities of explosives [1]. Using an overbiased piezoresistive cantilever as a heat source, detection of deﬂagration occurred only at sufﬁciently large heating pulses and only after exposure to trinitrotoluene (TNT). The deﬂagration signal was uninﬂuenced by common interfering substances such as water, acetone, ethanol, and gasoline [53, 54]. The heat released by the exothermic deﬂagration raised the cantilever temperature, inducing additional bending during heating pulses as compared to the bending response of the cantilever without TNT during identical heating pulses, demonstrated in Fig. 30.11. After deﬂagration, the cantilever was clear of TNT and no further deﬂagration events occurred during subsequent heating. The degree of cantilever bending was linearly proportional to the mass of adsorbed TNT.

30.4 Data Storage and Lithography Because of its ability to form and detect nanometer-scale structures, scanning probe microscopy has been considered a candidate technology for advanced data storage or nanolithography. The practical data density of magnetic disks will likely plateau in

30 Applications of Heated Atomic Force Microscope Cantilevers

265

Fig. 30.11. Cantilever bending response during a heating pulse with (solid line) and without (dotted line) adsorbed TNT on the surface. The hump at the beginning of the voltage pulse is caused by the exothermic deﬂagration event, and increases linearly with the mass of adsorbed TNT [53]

the range of 100–200 Gbit in−2 due to the superparamagnetic effect [55], and optical lithography could reach resolution limits near 50 nm [56]. Numerous scanning probe techniques have been suggested as possible alternatives to these limits [34, 57–61]. AFM offers advantages over STM in that it is somewhat faster in single-probe operation, can be used on nonconducting surfaces, and is more easily parallelized. 30.4.1 Data Storage The ﬁrst demonstration of surface modiﬁcation using a heated AFM cantilever tip employed a heating laser focused on a silicon nitride cantilever in contact with a thick ﬁlm of PMMA [62]. Under a tip loading force of 0.1–1 µN, laser heating pulses of 15 mW and 0.3–100 µs heated the cantilever and cantilever tip, locally melting the polymer in contact with the tip. The tip penetrated into the polymer surface, forming a melted indentation of diameter 100 nm–1 µm. The spatial resolution of the writing was governed by the tip sharpness, heating pulse duration, and cantilever load force. The same cantilever imaged the formed indents using the standard optical lever technique. Reading was demonstrated at 100 kHz, with the 500 kHz upper bound on reading limited by the cantilever mechanical time constant. In general, a mechanical probe can follow surface contours at rate of three to four times its mechanical resonant frequency [63]. A related study examined thermomechanical recording using a tapered optical ﬁber instead of an AFM cantilever, with the goal of increasing speed and decreasing mechanical wear [64]. The ﬁber optical stylus produced data

266

B.A. Nelson · W.P. King

bit indentations as small as 50 nm×74 nm, but was not able to read these indentations with sufﬁcient signal-to-noise, although for bit sizes of 150 nm×300 nm data reading was possible. For both of these studies, the instrumentation required for controlling position and performing writing and reading was more complex than was practical for data storage applications. One simpliﬁcation of the required instrumentation was to use an AFM cantilever with an integrated heating element. In a demonstration of using piezoresistive cantilevers for thermomechanical writing, short 40 mW electrical pulses delivered to a piezoresistive cantilever raised the cantilever temperature to above 700 ◦ C [65]. The sharp silicon cantilever tip produced indents as small as 200 nm, but the 300– 500 µs cantilever cooling time was quite long due to the large size of the heated region. The fabrication of heated AFM cantilevers designed speciﬁcally for data storage reduced the cantilever heating time [66]. Figure 30.12 shows one of these cantilevers, which had a small heating element fabricated at the cantilever free end, very close to the cantilever tip. Because the heater element was small and relatively isolated, 1–10 µs heating and cooling time constants were possible. At the time, it was thought that a data storage device might use the heated AFM cantilevers for thermomechanical writing and use either other piezoresistive cantilevers or a combination of both heating and piezoresistive sensors for reading [66]. Other groups eventually achieved this combination [43, 67, 68]. Several nearly consecutive improvements in the operation of AFM cantilevers with integrated heaters signiﬁcantly improved their promise for practical use: an approach for high-density data bit writing, data reading using thermal probes, and fabrication of two-dimensional cantilever arrays. High-density thermomechanical writing was accomplished by writing indentations into a 40 nm thick PMMA layer prepared on top of a second layer of crosslinked epoxy [34]. Unlike previous thermomechanical data writing in polymers of a thickness much greater than the tip height [62, 64, 65], this conﬁguration limited tip penetration and thus also limited the width of the formed indent. The underlying epoxy layer minimized the wear incurred on the tip during data bit writing. Figure 30.13 shows thermomechanical data writing in a thin polymer ﬁlm, and Fig. 30.14 shows these indentations. In this polymer ﬁlm stack, data bit spatial periodicity was as small as 40 nm, corresponding to a data density of 500 Gbit in−2 . Using this same approach, data bit indents as small as 23 nm were later written [69].

Fig. 30.12. Atomic force microscope cantilevers fabricated with a heater integrated near the cantilever tip [66]. The heater size was approximately 5 × 10 µm2 and the tip had a sharpness of better than 30 nm

30 Applications of Heated Atomic Force Microscope Cantilevers

267

Fig. 30.13. Thermomechanical writing and thermal reading in a thin polymer layer with a heated AFM cantilever tip [34]. In writing mode, the AFM cantilever is heated above the melting temperature of the polymer and indents into the polymer ﬁlm. In reading mode, the cantilever detects topography by measuring thermal impedance from the cantilever

The development of data reading using heated cantilevers was an essential simpliﬁcation for scaling up the data storage cantilevers into arrays. In the thermal reading mode, the AFM cantilever was heated such that it was warm, but not so hot that it deformed the polymer. As the warm cantilever scanned over the contours of the data-written polymer substrate, changes in thermal impedance between the cantilever and substrate induced measurable changes in the electrical resistance of the cantilever, illustrated in Fig. 30.13. While not speciﬁcally designed for reading, the heated AFM cantilevers shown in Fig. 30.12 were adequate for data reading using this technique [34], and were used to create the image in Fig. 30.14. Vertical displacement sensitivity was reported to be in the range of 10−6 –10−5 nm−1 . Modeling and measurements later showed that heat transfer across the heater-substrate air gap, rather than heat transfer down the cantilever tip, governs this thermal reading mechanism [37]. However, the temperature rise in the tip allowed the thermomechanical writing. Therefore, because the heat transfer mechanisms for writing and reading were relatively independent, it was possible to optimize both the writing and the reading performance of the cantilever simultaneously [38]. Eventually an extremely small heater region was explored to reduce heating time [70], and the write/read operations were split into two separate heaters [71]. Simulations showed that the thermal cantilever was more sensitive than a similarly-sized piezoresistive cantilever by at least two orders of magnitude [36]. While it has not been studied in detail, the

Fig. 30.14. Thermally-read image of nanometer-scale indentations written into a thin polymer ﬁlm [34]

268

B.A. Nelson · W.P. King

heated AFM cantilever could become a highly sensitive metrology tool. While AFM cantilevers with both integrated heaters and piezoresistive elements [43, 67, 68] may become valuable for other applications, the ﬁnding that the thermal cantilever is an excellent metrology tool made piezoresistor integration unnecessary for the success of this data storage technology. Motivated by the need for high-speed data throughput, AFM cantilevers for data storage were the ﬁrst type of cantilever to be fabricated in a two-dimensional array [72]. The use of an array for thermomechanical data storage was made possible through the dual read/write functionality of the heated AFM cantilever. Without the capability for integrated reading, an array of AFM cantilevers would require laser interrogation of every cantilever, optical interferometry of the entire chip, or additional electronics for individual piezoresistors on each cantilever in the array. Arrays of heated AFM cantilevers were ﬁrst made in a 5 × 5 cantilever conﬁguration [72] and eventually expanded to a 32 × 32 cantilever conﬁguration [73, 74], known as the “Millipede”. Figure 30.15 shows the Millipede cantilever and cantilever array. Operation of the cantilever array demonstrated data writing above 600 Gbit in−2 at a single-cantilever writing rate near 100 kHz [71, 75]. The development of AFM-based data storage using arrays of heated AFM cantilevers has made an impact on nanoscience and nanotechnology by enabling novel nanoscale thermal transport and temperature measurements, for example writing thermomechanical indentations using carbon nanotubes afﬁxed to the cantilever tip [76], or studying nanoscale mechanical deformation and wear using highly local heating from the tip [77]. The technology developed could also have broad impact on the packaging of microelectromechanical systems [78]. An overview of lessons learned in the Millipede project is available [79].

Fig. 30.15. Photo of the 32 × 32 Millipede arrays of heater-cantilevers (top), and scanning electron microscope images of the arrayed cantilevers (bottom left), and a single cantilever (bottom right) [73, 74]

30 Applications of Heated Atomic Force Microscope Cantilevers

269

30.4.2 Lithography Heated AFM cantilevers have made several recent contributions to lithography. In particular, cantilever heating can be used to modulate writing from a heated AFM cantilever tip. In dip-pen nanolithography (DPN), a chemically-coated AFM cantilever tip locally transfers chemicals from the tip to a substrate much like painting with water colors, shown schematically in Fig. 30.16. This local chemical transfer can produce features as small as 10 nm [61]. One challenge for DPN is that, whenever a chemically-coated tip is in contact with a surface, chemicals will transfer from the tip to the surface, or even from the surface to the tip [80–82]. This phenomenon can present contamination challenges, as metrology using the same tip that is used for deposition is difﬁcult or impossible without transporting more chemical from the tip to the surface or smearing the deposited chemical. Heated AFM cantilevers have provided two strategies for overcoming this challenge. First, thermal bimorph cantilevers have actuated the AFM tip into and out of contact with a surface to modulate DPN writing [83–85]. Second, heated AFM cantilevers have been used for “thermal” dip pen nanolithography (tDPN), where the chemical coating is solid at room temperature but melts at elevated temperature [86], enabling writing to be turned on and off with cantilever heating. In the thermal bimorph approach of actuating cantilever tip-substrate contact, the cantilevers were made of silicon nitride and coated with a chrome/gold heater. When cold, the cantilevers were relatively ﬂat. Appropriate positioning of the cantilever could bring the tips into contact with the substrate while heating caused the cantilever to bend such that the tips could break contact. Figure 30.17 shows the actuation concept for control of tip-substrate contact and an illustration of the fabricated cantilever. A nominal single-cantilever heating power of 2 mW produced a tip deﬂection of 10 µm in ∼ 1 ms. An array of ten cantilevers was made, also shown in Fig. 30.17. As a demonstration of meeting the requirement for control of tip-substrate

Fig. 30.16. In dip-pen nanolithography, a chemically-coated AFM cantilever scans over a surface, locally inking where the cantilever tip has contacted the surface [61]

270

B.A. Nelson · W.P. King

Fig. 30.17. Thermal bimorph cantilevers for dip-pen nanolithography [83]. Top left: when cold, the cantilever was in contact with the surface and when heated the cantilever became curled, bringing the tip out of contact with the surface. Bottom left: the cantilever was made of silicon nitride coated with a chrome-gold heater. Top right: the cantilever was made in a one-dimensional array of ten cantilevers. Bottom right: the cantilever array wrote out numerical digits by scanning the probe array in a ﬁgure “8” pattern and modulating tip-substrate cantilever through thermal bimorph actuation. The DPN-written digits in the bottom write image are approximately 1.5 µm across

contact, the cantilevers were coated with octadecane thiol (ODT) and brought into contact with a gold substrate while the cantilever array traced a ﬁgure “8”. The cantilever tips were actuated into and out of contact with the surface, such that each of the ten cantilevers wrote one of the digits from 1 to 10. Figure 30.17 shows an excerpt of these written digits. In the ﬁrst demonstration of thermal dip-pen nanolithography (tDPN), a cantilever much like the data storage cantilever described in the previous section was coated with octadecylphosphonic acid (OPA), a material that is solid at room temperature but melts at 99 ◦ C [86]. The tip operated like a nanometer-scale soldering iron, with no material depositing from the tip to the surface with the cold cantilever tip in contact with the substrate. Figure 30.18 shows the ﬁrst tDPN experiment, where the cantilever tip was in contact with and scanned over a mica surface in four 0.5 µm squares at four different temperatures. When the cantilever was not close to 99 ◦ C, no material was deposited from the cantilever tip to the mica surface. When the cantilever was heated above the melting temperature of the OPA, material transferred from the tip to the surface, with the deposition rate increasing at higher

30 Applications of Heated Atomic Force Microscope Cantilevers

271

Fig. 30.18. Thermal dip-pen nanolithography [86]. Left: a heatable AFM cantilever is coated with a material that is solid at room temperature and melts at high temperature, such that the material is deposited from the tip only during heating. Right: results of the ﬁrst tDPN experiment. The top image is topography and the bottom image is friction. Four locations were scanned, each at a different temperature. Deposition occurred only when the cantilever was heated above the melting temperature of the material coating the cantilever tip

temperatures. Post-deposition metrology with the same cantilever showed the OPA deposition in both the topography signal and the lateral force signal, also shown in Fig. 30.18. The ability to turn DPN writing on and off, either by actuating the cantilever or by selectively melting the material coating the tip, could become a signiﬁcant advancement for DPN. In particular, these techniques enable arrays of cantilevers to become practical, as an array could be scanned over a large area without contaminating the entire surface. The ability to turn writing on and off by arbitrarily modulating tip-surface contact allows a number of materials to be written in a small space and in a short time, without the need to change cantilevers. Additionally, the use of a heated cantilever to deposit melted material could increase the number of materials used in DPN, including low melting temperature metal solders. Several other lithographical approaches have also used heated AFM cantilevers. Femtosecond laser pulses were optically focused onto the tip of a cantilever to cause ablation on 25 nm thick gold ﬁlms, achieving 10 nm lateral resolution [87]. The optics required in this technique will make large-scale utilization difﬁcult, however. An

272

B.A. Nelson · W.P. King

AFM cantilever heated by a resistively heated AFM probe holder melted crystalline nanospheres within a block copolymer [88], demonstrating the ability to melt single 24 nm crystalline domains. In an application using the ultracompliant polyimide probe shown in Fig. 30.2, heat supplied by the probe induced localized cross-linking in a commercial photoresist, demonstrating maskless photoresist-based lithography at a resolution of 450 nm [89]. Because of the high compliance of the probe, array operation without mechanical feedback was demonstrated with no apparent damage to the tip or deformation to the photoresist substrate.

30.5 Summary and Conclusions Heated AFM cantilevers have been used for thermal property measurement, microsystems actuation, and thermal processing, but applications using these capabilities have only begun to realize their full potential. Many physical, chemical, and biological phenomena depend upon temperature, and the most interesting measurements are likely yet to be demonstrated. For example, few precision force measurements have been made with heated AFM cantilevers, even though they are outstanding force transducers. Additionally, no investigations that we are aware of have explored the effects of heated probes as highly localized heat sources in biological or biochemical systems. The most pressing unresolved issue in all of these applications, and for future applications, is the lack of precision quantitative measurements with well-understood uncertainty. The overall impact of heated AFM cantilever probes would be significantly enhanced by further quantitative investigation of heat ﬂow in AFM probes, high-resolution temperature calibration, temperature determination at tip-substrate contacts, and possibly standardization of these across heated AFM cantilever probe types. Acknowledgements. The authors thank the members of their research group at Georgia Institute of Technology. BAN was supported by an NDSEG Fellowship and WPK was supported by NSF CAREER.

References 1. Pinnaduwage LA, Gehl A, Hedden DL, Muralidharan G, Thundat T, Lareau RT, Sulchek T, Manning L, Rogers B, Jones M, Adams JD (2003) Nature 425:474 2. Berger R, Lang H, Gerber C, Gimzewski J, Fabian J, Scandella L, Meyer E, Guntherodt H (1998) Chem Phys Lett 294:363 3. Passian A, Warmack RJ, Ferrell TL, Thundat T (2003) Phys Rev Lett 90:124503 4. Passian A, Warmack RJ, Wig A, Farahi RH, Meriaudeau F, Ferrell TL, Thundat T (2003) Ultramicroscopy 97:401 5. Passian A, Wig A, Meriaudeau F, Ferrell TL, Thundat T (2002) J Appl Phys 92:6326 6. Lereu AL, Passian A, Warmack RJ, Ferrell TL, Thundat T (2004) Appl Phys Lett 84:1013 7. Majumdar A (1999) Annu Rev Mater Sci 29:505 8. Pylkki RJ, Moyer PJ, West PE (1994) Jpn J Appl Phys 33:3785

30 Applications of Heated Atomic Force Microscope Cantilevers

273

9. Hammiche A, Pollock HM, Song M, Hourston DJ (1996) Meas Sci Technol 7:142 10. Hammiche A, Bozec L, Conroy M, Pollock HM, Mills G, Weaver JMR, Price DM, Reading M, Hourston DJ, Song M (2000) J Vac Sci Technol B 18:1322 11. Edinger K, Gotszalk T, Rangelow IW (2001) J Vac Sci Technol B 19:2856 12. Li M-H, Gianchandani YB (2003) Sens Actuators A 104:236 13. Li M-H, Gianchandani YB (2000) J Vac Sci Technol B 18:3600 14. Hammiche A, Hourston DJ, Pollock HM, Reading M, Song M (1996) J Vac Sci Technol B 14:1486 15. Nonnenmacher M, Wickramasinghe HK (1992) Appl Phys Lett 61:168 16. Blanco C, Appleyard SP, Rand B (2002) J Microsc 205:21 17. Lee J-H, Gianchandani YB (2004) Rev Sci Instrum 75:1222 18. Oesterschulze E, Stopka M (1996) Microelectron Eng 31:241 19. Varesi J, Majumdar A (1998) Appl Phys Lett 72:37 20. Gomes S, Thannoy N, Grossel P, Depasse F, Bainier C, Charraut D (2001) Int J Thermal Sci 40:949 21. Lefevre S, Volz S, Saulnier J-B, Fuentes C (2003) Rev Sci Instrum 74:2418 22. Depasse F, Grossel P, Gomes S (2003) J Phys D: Appl Phys 36:204 23. Arai T, Tomitori M (1998) Appl Phys A 66:S319 24. Arai T, Tomitori M (1999) Appl Surf Sci 144–145:501 25. Tomitori M, Arai T (1999) Appl Surf Sci 140:432 26. Akiyama T, Staufer U, de Rooij NF, Lange D, Hagleitner C, Brand O, Baltes H, Tonin A, Hidber HR (2000) J Vac Sci Technol B 18:2669 27. Akiyama T, Staufer U, de Rooij NF (2002) Rev Sci Instrum 73:2643 28. Franks W, Lange D, Lee S, Hierlemann A, Spencer N, Baltes H (2002) Ultramicroscopy 91:21 29. Sulchek T, Minne SC, Adams JD, Fletcher DA, Atalar A, Quate CF, Adderton DM (1999) Appl Phys Lett 75:1637 30. Volden T, Zimmermann M, Lange D, Brand O, Baltes H (2004) Sens Actuators A 115:516 31. Pedrak R, Ivanov T, Ivanova K, Gotszalk T, Abedinov N, Rangelow IW, Edinger K, Tomerov E, Schenkel T, Hudek P (2003) J Vac Sci Technol B 21:3102 32. Manalis SR, Minne SC, Atalar A, Quate CF (1996) Rev Sci Instrum 67:3294 33. Williams CC, Wickramasinghe HK (1986) Appl Phys Lett 49:1587 34. Binnig G, Despont M, Drechsler U, Haberle W, Lutwyche M, Vettiger P, Mamin HJ, Chui BW, Kenny TW (1999) Appl Phys Lett 74:1329 35. Lutwyche MI, Despont M, Drechsler U, Durig U, Haberle W, Rothuizen H, Stutz R, Widmer R, Binnig GK, Vettiger P (2000) Appl Phys Lett 77:3299 36. King WP, Kenny TW, Goodson KE (2004) Appl Phys Lett 85:2086 37. King WP, Kenny TW, Goodson KE, Cross G, Despont M, Durig U, Rothuizen H, Binnig GK, Vettiger P (2001) Appl Phys Lett 78:1300 38. King WP, Kenny TW, Goodson KE, Cross GLW, Despont M, Durig UT, Rothuizen H, Binnig G, Vettiger P (2002) J Microelectromech Syst 11:765 39. Hammiche A, Reading M, Pollock HM, Song M, Hourston DJ (1996) Rev Scientiﬁc Instruments 67:4268 40. Berger R, Gerber C, Gimzewski J, Meyer E, Guntherodt H (1996) Appl Phys Lett 69:40 41. Nakagawa Y, Schafer R, Guntherodt H-J (1998) Appl Phys Lett 73:2296 42. Nakagawa Y, Schafer R (1999) Angew Chem Int Ed 38:1083 43. Abedinov N, Grabiec P, Gotszalk T, Ivanov T, Voigt J, Rangelow IW (2001) J Vac Sci Technol A – Vac Surf Films 19:2884 44. Blanco C, Lu S, Appleyard SP, Rand B (2003) Carbon 41:165

274

B.A. Nelson · W.P. King

45. Fabian J-H, Scandella L, Fuhrmann H, Berger R, Mezzacasa T, Musil C, Gobrecht J, Meyer E (2000) Ultramicroscopy 82:69 46. Ono T, Esashi M (2004) Meas Sci Technol 15:1977 47. Hierlemann A, Lange D, Hagleitner C, Kerness N, Koll A, Brand O, Baltes H (2000) Sens Actuators B 70:2 48. Hagleitner C, Hierlemann A, Lange D, Kummer A, Kerness N, Brand O, Baltes H (2001) Nature 414:293 49. Grigorov AV, Davis ZJ, Rasmussen P, Boisen A (2004) Microelectron Eng 73–74:881 50. Lavrik NV, Sepaniak MJ, Datskos PG (2004) Rev Sci Instrum 75:2229 51. Lee DW, Despont M, Drechsler U, Gerber C, Vettiger P, Wetzel A, Bennewitz R, Meyer E (2003) Microelectron Eng 67–68:635 52. Lee D, Wetzel A, Bennewitz R, Meyer E, Despont M, Vettiger P, Gerber C (2004) Appl Phys Lett 84:1558 53. Pinnaduwage LA, Wig A, Hedden DL, Gehl A, Yi D, Thundat T, Lareau RT (2004) J Appl Phys 95:5871 54. Pinnaduwage LA, Thundat T, Gehl A, Wilson SD, Hedden DL, Lareau RT (2004) Ultramicroscopy 100:211 55. Ruigrok JJM, Coehoorn R, Cumpson SR, Kesteren HW (2000) J Appl Phys 87:5389 56. Marrian CRK, Tennet DM (2003) J Vac Sci Technol A 21:S207 57. Snow ES, Campbell PM (1995) Science 270:1639 58. Cooper EB, Manalis SR, Fang H, Dai H, Matsumoto K, Minne SC, Hunt T, Quate CF (1999) Appl Phys Lett 75:3566 59. Wilder K, Quate C, Adderton D, Bernstein R, Elings V (1998) Appl Phys Lett 73:2527 60. Eigler DM, Schweizer EK (1990) Nature 344:524 61. Piner RD, Zhu J, Xu F, Hong S, Mirkin CA (1999) Science 283:661 62. Mamin HJ, Rugar D (1992) Appl Phys Lett 61:1003 63. Ried RP, Mamin HJ, Terris BD, Fan LS, Rugar D (1997) J Microelectromech Syst 6:294 64. Hoen S, Mamin HJ, Rugar D (1994) Appl Phys Lett 64:267 65. Mamin HJ (1996) Appl Phys Lett 69:433 66. Chui BW, Stowe TD, Ju YS, Goodson KE, Kenny TW, Mamin HJ, Terris BD, Ried RP (1998) J Microelectromech Syst 7:69 67. Lee CS, Nam HJ, Kim YS, Jin WH, Cho SM, Bu JU (2003) Appl Phys Lett 83:4839 68. Yang ZX, Li XX, Wang YL, Bao HF, Liu M (2004) Microelectron J 35:479 69. Vettiger P, Cross G, Despont M, Drechsler U, Duerig U, Gotsmann B, Haberle W, Lantz MA, Rothuizen HE, Stutz R, Binnig GK (2002) IEEE Trans Nanotech 1:39 70. Drechsler U, Burer N, Despont M, Durig U, Gotsmann B, Robin F, Vettiger P (2003) Microelectron Eng 67–8:397 71. Pantazi A, Lantz MA, Cherubini G, Pozidis H, Eleftheriou E (2004) Nanotechnology 15:S612 72. Vettiger P, Brugger J, Despont M, Drechsler U, Durig U, Haberle W, Lutwyche M, Rothuizen H, Stutz R, Widmer R, Binnig G (1999) Microelectron Eng 46:11 73. Despont M, Brugger J, Drechsler U, Dürig U, Häberle W, Lutwyche M, Rothuizen H, Stutz R, Widmer R, Rohrer H, Binnig GK, Vettiger P (2000) Sens Actuators A 80:100 74. Vettiger P, Despont M, Drechsler U, Durig U, Haberle W, Lutwyche MI, Rothuizen HE, Stutz R, Widmer R, Binnig GK (2000) IBM J Research and Development 44:323 75. Eleftheriou E, Antonakopoulos T, Binnig GK, Cherubini G, Despont M, Dholakia A, Durig U, Lantz MA, Pozidis H, Rothuizen HE, Vettiger P (2003) IEEE Trans Magnetics 39:938 76. Lantz MA, Gotsmann B, Durig UT, Vettiger P, Nakayama Y, Shimizu T, Tokumoto H (2003) Appl Phys Lett 83:1266 77. Gotsmann B, Durig U (2004) Langmuir 20:1495

30 Applications of Heated Atomic Force Microscope Cantilevers

275

78. Despont M, Drechsler U, Yu R, Pogge HB, Vettiger P (2004) J Microelectromech Syst 13:895 79. Vettiger P, Binnig G (2003) Sci Am 288:46 80. Schwartz PV (2002) Langmuir 18:4041 81. Sheehan PE, Whitman LJ (2002) Phys Rev Lett 88:156104 82. Rozhok S, Piner R, Mirkin CA (2003) J Phys Chem B 107:751 83. Bullen D, Wang XF, Zou J, Chung SW, Mirkin CA, Liu C (2004) J Microelectromech Syst 13:594 84. Bullen D, Chung SW, Wang XF, Zou J, Mirkin CA, Liu C (2004) Appl Phys Lett 84:789 85. Zhang M, Bullen D, Chung SW, Hong S, Ryu KS, Fan ZF, Mirkin CA, Liu C (2002) Nanotechnology 13:212 86. Sheehan PE, Whitman LJ, King WP, Nelson BA (2004) Appl Phys Lett 85:1589 87. Chimmalgi A, Choi TY, Grigoropoulos CP, Komvopoulos K (2003) Appl Phys Lett 82:1146 88. Vasilev C, Heinzelmann H, Reiter G (2004) J Polymer Sci: Part B: Polymer Phys 42:1312 89. Basu AS, McNamara S, Gianchandani YB (2004) J Vac Sci Technol B 22:3217

Subject Index

abalone nacre, II 110 activation energy, III 89 active near-ﬁeld optical probes, II 178 actuation, IV 251 adhesion, III 272, III 282, III 300, III 303–305, III 309, III 313, III 316, III 323 adhesion corrected, III 99 adhesion length, III 283 adhesion meter (CAM), III 306 adhesion paradox, III 283 adhesives, III 31 adsorbed layers, III 284 Al2 O3 , III 42 alkane derivative, IV 170, IV 171 n-alkane derivatives, IV 167 amino acid, IV 174 Amontons’ law, III 272 amplitude modulation, II 1 analytical technique, II 381 anharmonic signals, II 23 antenna structures, II 177 aperture, III 236 aperture-based near-ﬁeld optical probes, II 175 apertureless near-ﬁeld microscope, III 235 apertureless SNOM, III 236, III 242 array of silicon cantilever, II 169 artifacts, III 242 artiﬁcial hip joints, III 286 asperity, III 346–348, III 352, III 354–357, III 367 asphalt, III 287 atom orbitals, III 45 atomic force microscopy (AFM), II 1, II 143–147, II 149, II 152, II 155, II 158, II 159, II 161, II 165, III 28, IV 251 AFM tips, III 288 AFM topographies, III 270 AFM based lithography, IV 105

blunt AFM tips, III 289 dynamic AFM, II 1 α atoms, III 41 β atoms, III 41 autocorrelation function (ACF), III 357 auxiliary electrode, II 187 average, III 353 barium titanate, III 240, III 250 barrier-hopping ﬂuctuations, III 100 basic principles of STM and STS, III 185 bearing ratio, III 355 benthic, III 30 bias-induced nanofabrication, IV 107 bidirectional optical lever, III 225 binary mixture, IV 170 biochemical sensors, II 185 biocompatibility, III 286 biology, III 283 biomineralization, II 109 biomolecules, II 183 bipyridyl, IV 167 bitter pattern, III 359 blind friction calibration, III 93 boron carbide, III 286 boundary conditions, II 6 bow-tie antenna, II 177 BSE, III 34 buckling, III 227 calender, III 348 calendering, III 345, III 348 calibration of lateral forces, III 93 calix[8]arene, IV 163 calorimetry, IV 252 CAM, III 306 canticlever concept, II 195 cantilever bending, III 222 dynamics, II 33

278 probes, II 165 cantor set, III 280 capacitance, III 231 capacitive forces, III 233 capillary neck, III 103 carbon, II 170 clusters, III 270, III 272 nanotube, II 170 nitride, III 286 thin ﬁlms, III 270 carrier lifetime, II 181 cavity model of elastic-plastic indentation, III 117 cells, III 287 in vivo, III 28 channelling contrast, II 367 chaperones, III 34 characterization of the magnetic tape with MFM, III 359 charge carrier transport, III 109 chemical force mapping, II 183 chemical force microscopy (CFM), II 183 chemical functionalization, II 183 chemical mapping, II 184 chromatic aberration, II 364 cluster-assembled, III 269 co-adsorbed, IV 167 collocated systems, II 6 colloidal probes, III 291 commensurate, III 288 compliance, II 167 conﬁned polymer systems, III 85 confocal scanning optical microscopy (CSOM), III 238, III 245 time-resolved CSOM, III 241 contact area, III 299, III 302, III 308, III 309, III 311 junctions, III 274 mechanics, III 90 mode, III 220, III 221, III 225 potential, III 222 potential difference (CPD), III 223, III 231 pressure, III 117 problem, III 303 stiffness, II 8 contamination needles, II 193 continuum models, III 288 contrast, III 229 mechanisms, II 367 converse piezoelectric effect, III 220

Subject Index cooperative molecular motion, III 107 cooperatively rearranging regions (CRRs), III 103 cooperatively rearranging regions (CRRs), III 85 coronene, IV 162 correlation length, III 357 corrosion phenomena, II 189 Coulomb explosion, III 43 creep models, III 100 Creutzfeldt-Jakob disease, III 33 vCJD, III 34 critical wavelength, II 193 cryoelectron microscopy, III 35 current vs. distance (IZ) curve, II 97, II 101 cut-off effect, II 175 data storage devices, II 169 Deborah number, III 105 decacyclene, IV 162 defect characterization, II 388 deformation, III 300, III 306, III 309, III 316, III 323 dendrimers, IV 23 Derjaguin–Muller–Toporov (DMT), II 7 detachment stress, III 282 detection of higher eigenmodes, II 20 device fabrication, II 395 device under test (DUT), II 179 dewetting kinetics, III 110 diamond, II 167 probes, II 171 diamondlike, III 286 diatoms, III 28 dielectric permittivity, III 233 diffusional shielding, II 188 dimensional constraints, III 109 dip-pen nanolithography (DPN), IV 2, IV 105, IV 251 Dirac’s delta function, II 181 direct writing, IV 8 directional diffusion, IV 176 disentanglement barriers, III 111 dissipation, III 274 lengths, III 108 dissipative, III 231 distribution of heights, III 264 DLC coating, III 322, III 323 DMT model, III 303 DNA, IV 8 domains, III 229

Subject Index boundary, III 227, III 234 contrast, III 221–223, III 225, III 235–237, III 239, III 245, III 250 structure, III 218, III 235–237, III 242, III 244, III 245, III 247, III 251 walls, III 218, III 219, III 221, III 230, III 236, III 237, III 245, III 246, III 248, III 249, III 251 dual beam systems, II 365 dynamic force microscopy (DFM), II 143–147, II 158 dynamic-contact electrostatic force microscopy, III 222 eigenfrequency, II 167 eigenmode, II 3, II 5 eigenvector, II 3 elastic–plastic materials, III 120 elasto–plastic deformation, III 279 electrocatalysis, II 189 electrochemical DPN (E DPN), IV 22 electrochemical microscopy, II 166 electrode, II 96 electron beam deposited tips, II 170, II 193 electrooptic contrast, III 234 modulation, III 238 electrostatic force, III 222 interaction, II 8 electrostatic force microscopy (EFM), III 219, III 220 energy barrier, II 152–155 ergodic, III 285 errors, III 367 correction, III 344, III 346 rates, III 348 Escherichia coli, III 35 etched, III 269 Eunotia sudetica, III 31 evanescent, II 175 waves, II 174 Eyring model, III 99 far-ﬁeld optics, II 174 ferroelectric, III 217, III 238, III 240, III 241, III 244, III 251 ferroelectric random access memory (FeRAM), III 218, III 252 ﬁlms, III 246, III 249 ﬁnite element modeling, II 99

279 ﬂat tips, III 291 ﬂexural mode, III 225 focused ion beam (FIB), II 194, II 361 force spectroscopy, II 143, II 144, II 149, II 152, II 153, II 155, II 156, II 158, II 161 force–distance curve, II 185 force-induced nanofabrication, IV 108 Fourier coefﬁcient, II 1 Fourier optics, II 174 fractal, III 266, III 357 dimension, III 266 morphology, III 291 surfaces, III 277 fractures, III 267 free volume, III 88 free cantilever, II 2 friction, III 221, III 227, III 248 coefﬁcient, III 104, III 272 force, III 272 friction-velocity analyses, III 104 internal, III 103 monomeric, III 103 theories, III 274 friction force microscopy (FFM), III 85, III 98 frictional dissipation, III 103 fullerenes, III 286 functional head group, II 183 functionalized probes for biological applications, II 166 functionalized tips, II 182 gallium arsenide, II 167, II 173 gas-assisted deposition, II 377 GASH, III 221 Gaussian distribution, III 354–356 geofractals, III 266 glass transition, III 85 graphite, III 41 GroEL, III 35 GroEL GroES complex, III 35 GroES, III 35 guest-host interactions, IV 160 GW model, III 303, III 308, III 311 hard disk recording, II 194 hardness tester, III 292 harmonic, II 1, II 3 signal, II 18 head-tape interface, III 350, III 352, III 353, III 357

280 spacing, III 352, III 354, III 355, III 357 Heaviside function Θ(t − t0 ), II 181 height–height correlation function, III 267 Hertzian, III 275, III 288 hetero epitaxial nucleation, II 110 heterogeneous dynamics, III 85 glass formers, III 103 heterogeneous molecular arrays, IV 166 high coercivity tip, II 194 high spatial frequencies, II 193 high spatial frequency near-ﬁeld components, II 175 high-speed imaging, IV 252 higher harmonic images, II 25 higher harmonics, II 16 highly oriented pyrolytic graphite (HOPG), III 38 hollow atoms, III 41 host network, IV 160 hydrocarbons, III 284 hydrogen bond, IV 160 IC diagnostics: destructive and nondestructive analysis, II 394 IC failure mode analysis, II 391 immobilization of peptides, IV 14 implants, III 286 in situ polymerization, IV 22 in-plane gate (IPG) transistor, II 397 inclusion effect, IV 162 indentation, III 291 hardness, III 274 indirect patterning, IV 8 inorganic materials, IV 25 inorganic overlayers, III 188 instability, II 16 insulating surfaces, III 38 interfacial boundaries, III 109 constraints, III 84 energy, III 282 glass transition proﬁles, III 113 plasticization, III 109 reactions, II 186 Tg proﬁle, III 125 intermittent contact mode, II 1, III 234 intermolecular forces, II 149 intralayer array, IV 168 intramolecular forces, II 143, II 149, II 156, II 158, II 161 ion

Subject Index bombardment, III 38 column, II 362 current, II 97 optics, II 364 ion beam, II 408 ion beam lithography, II 379 ion-blasted, III 269 JKR model, III 303 Kelvin probe force microscopy, II 3 kinetic sputtering, III 38 kurtosis, III 265, III 352, III 354, III 355, III 357, III 367 kuru, III 34 lamella templates, IV 174 lamella-type structure, IV 167 Langevin equation, III 101 Laplace transformation, II 13 large organic molecules, III 33 Larmor frequency, III 47 laser scanning microscopy, III 236 lateral contact stiffness calibration, III 93 forces, III 228 twisting, III 225 lateral force microscopy (LFM), IV 18 layered compounds, III 288 LB ﬁlms, III 288 leakage current, II 93 light lever detection, II 22 light lever readout, II 20 light-emitting diodes, III 109 linear creep model, III 100 linker molecule, II 183 lithium niobate, III 219, III 238, III 249 living cells, III 27 LMIS, II 362 lock-in, III 225 low moment tip, II 194 low temperature grown (LT) GaAs, II 179 luminescence spectroscopy, III 245 machined, III 269 magnetic charges, II 192 material, II 195, IV 27 microscopy, II 166 particles, III 343, III 345, III 348, III 349, III 359, III 361 recording tape, III 343

Subject Index resonance, III 48 resonance imaging, III 48 storage devices, III 286 tape, III 343 magnetic force microscopy (MFM), II 192, III 345, III 358–367, IV 27 magnetic resonance force microscopy, III 48 magnetization, II 192 magnetoresistive, III 344, III 350 magnetoresistive head (MRH), II 380 manufactured metal surfaces, III 267 mask defects, IV 28 material aspects, II 166 material characterization, II 386 material contrast, II 367 mediator, II 188 meniscus force nanografting (MFN), IV 9 mesoscopic contacts, III 293 metal, III 274, III 288 metal-in-gap (MIG), III 343 method of reduced variables, III 104 mica, III 288 micro-Brillouin, III 245 micro-Raman, III 245, III 248 microarrays, IV 7 microcontact Printing (µCP), IV 28 microelectromechanical systems (MEMS), II 167, III 265 micromachining, II 401, III 286 micromechanical properties, III 30 micromotor, III 286 MIG, III 364, III 365 head, III 365, III 366 milling, II 361, II 407 millions of nanometric oscillators, II 169 millipede project, II 169 mineral bridges, II 110 mixed monolayer, IV 19 mobile atoms, III 284 modal harmonic distortion, II 32 modulated contacts, III 92 modulus-matched interface, III 124 molecular assembled, III 269 dynamics (MD), III 284 electronic devices, III 109 friction, III 102 mobility, III 84, III 85 networks, IV 160 overlayers, III 201 recognition, II 149, II 156, II 161

281 relaxations, III 85 templates, IV 159 molecular recognition force spectroscopy, II 158 monolayer protected clusters, IV 177 monomers, IV 22 morphology, III 263, III 274 morphotropic phase boundary (MPB), III 219, III 250 mother of pearl, II 109 multiple degree of freedom (MDOF), II 9 multiply charged ions, III 38 multiscale, III 266 nacre, II 109 nanodefects, III 37 nanoelectromechanical systems (NEMS), III 85, III 103 nanofabrication, IV 103 nanografting, IV 105, IV 109 nanoimpact studies, III 123 nanoindentation, III 312, III 316, III 317 nanomanipulation, II 409 nanometric oscillators, II 169 nanoparticle inks, IV 25 NanoPen Reader and Writer (NPRW), IV 19, IV 109 nanopipette, II 95 nanoroughness, III 286 nanoscience technology, II 165 nanoscopic constraints, III 85 nanoscratching, III 316, III 317 nanoshaving, IV 109 nanostructuring, III 45 nanotube tips, III 272 nanotubes, III 286 nanowear, III 316–318 Navicula seminulum, III 31 near-ﬁeld electrooptic microscopy, III 242 near-ﬁeld optics, II 166, II 174 near-ﬁeld scanning optical microscopy, III 219 noise, III 345, III 357, III 360–362, III 367 nominal contact area, III 272 non-Markovian behavior, III 102 noncollocated systems, II 6 noncontact, III 220, III 223 nonlinearity, II 16 nonminimum phase, II 6 nonspeciﬁc tip modiﬁcation, II 182 object spatial frequencies, II 174

282 oligonucleotides, IV 8 optical aberrations, III 240 optical microscopy, III 235 order–disorder transitions, III 87 organic thin ﬁlm transistors, III 109 output feedback, II 11 matrix, II 10 2D overlayers, III 208 paraelectric, III 238, III 240, III 241, III 251 parametric excitation, II 33 patch clamp technique, II 107 PEG, II 151, II 158 pentacene, IV 162 phase separation, IV 167 phase transition, III 219, III 236, III 238, III 240, III 244, III 249, III 251, III 252 photoconductive switch, II 179 phthalocyanine, IV 163, IV 168 physicochemical parameters, II 186 piezoelectric force microscopy, III 219 piezoelectric SPM, III 247 piezoresistive cantilevers, IV 252 piezoresistive Wheatstone bridge, II 169 piezoresponse, III 243 piezoresponse force microscopy (PFM), III 220 plastic, III 274 failure, III 280, III 281 ﬂow, III 117 yield, III 116 plasticity index ψ, III 276 pointed tapered metal-coated waveguide, II 175 poisson equation, II 99 polarized light microscopy, III 235 pole zero, II 13 pole-tip recession (PTR), III 350, III 351 poly(ethylene glycol) (PEG), II 150 poly(styrol) (PS), II 28 polyelectrolytes, IV 23 polymer, IV 21 coatings, III 315 polymeric, III 288 polysilicon, III 286 polystyrene, III 103 porous sample, II 94 porphyrins, IV 163 potential barrier, III 102 potential sputtering, III 38

Subject Index power spectrum, III 269 preferential adsorption, IV 160 projection mask technique, II 172 prostheses, III 286 proteins, IV 10 arrays, IV 127 immobilization, IV 129 nanoarrays, IV 11 nanopatterns, IV 128 unfolding, II 157 pull-off force, III 282 pump/probe experiment, II 179 pyroelectric probe, III 236 quality factor, III 223 quantum computing, III 47 quartz microbalance, III 43 ramped creep model, III 100 randomness, III 264 Rayleigh’s criterion, II 175 reactive ion plasma etching, II 177 real area of contact, III 273, III 355, III 356, III 358 recognition image, II 143, II 159–161 reconstruction, II 33 3D, II 382 recording head, III 343, III 344, III 364 redox couple, II 187 relaxation, III 102, III 104 relaxor, III 219, III 251 resolving, II 175 power, II 174 resonance modes, III 224 resonant frequencies, II 5 resonant modes, III 232 rheological boundary layers, III 109 rheological gradients, III 119 rim formation during indentation, III 120 rms roughness, III 353, III 354, III 357 road pavements, III 288 rock surfaces, III 287 root locus map, II 13, II 15 roughness, III 264, III 299–301, III 303, III 304, III 309–311, III 316, III 324, III 344–347, III 351–353, III 355, III 357, III 358, III 360, III 367 rubber, III 288 saturation, III 364, III 365 scanning electrochemical microscopy, II 186 scanning electron microscope (SEM), II 366

Subject Index scanning electrooptic microscopy, III 237, III 248 scanning force spectroscopy (SFS), II 183, II 184 scanning ion conductance microscope (SICM), II 91 scanning near-ﬁeld optical microscopy (SNOM), II 174, III 236 scanning probe lithography, IV 103 scanning thermal microscopy, IV 251 scanning tunneling microscopy, II 165, III 28 sealing effects, III 283 second harmonic, III 233, III 245, III 248 secondary ion mass spectroscopy (SIMS), II 367 self-afﬁne surface, III 267 self-assembled monolayers (SAMs), III 288, III 306, III 313, III 315, IV 162 self-lubricating coatings, III 317 self-similar, III 266 semiconductor device characterization and nanofabrication, II 166 semiconductor industry, II 389 shear bands, III 120 shear force microscopy, II 111 shear modulation force microscopy (SM FM), III 85, III 97 shear strength, III 274 silanes, IV 19 silazanes, IV 19 silicon, II 167, II 168 silicon (111)-(7 × 7) surface, III 46 silicon-carbide, III 286 silver/silver chloride electrode, II 96 single degree of freedom (SDOF), II 3 single electron spin, III 48 detection, III 47 single molecular array, IV 169 single molecule level, III 34 single molecules, IV 163 single-crystalline diamond, II 171 site-selective adsorption, IV 172 skewness, III 264, III 354, III 355, III 357, III 367 slope-detection method, III 223 small cantilever, III 35 solid-state materials, IV 26 speciﬁc tip modiﬁcation, II 183 spectrum, II 27 spherical aberration, II 365 sputter-deposited, III 269

283 squeezing effect, II 92, II 100 standalone cantilever probes, II 169 state space, II 4, II 9 static contacts, III 90 static friction, III 284 statistics, III 277 steady-state diffusion-limited reaction, II 187 stearic acid, IV 167, IV 173 stiffness, III 292 strain hardening, III 121 rate effects, III 123 shielding, III 124 softening, III 120 stress distribution, III 280 stroboscopic mode, III 241 structural analysis, II 388 structural anisotropy, III 125 structural heterogeneity, III 105 structure function, III 268 structures 1D, III 188, III 202 2D, III 196 subatomic features, III 45 subatomic range, III 27 subharmonics, II 19 substrate constraints, III 122 subwavelength aperture, II 175 superposition of friction-velocity isotherms, III 104 supramolecular architectures, IV 159 compounds, IV 24 surface charge density, III 231 surface energy, III 306 surface force apparatus (SFA), III 273 surface forces, III 309 switching, III 241, III 248, III 249, III 252 system matrix, II 9 tape drive, III 343, III 346, III 348, III 351, III 362 tape roughness analysis, III 351 tapping-mode, II 1, II 108, III 225, III 234 TEM -SEM sample preparation, II 384 template, IV 168 texture, III 299, III 300 TGS, III 223, III 225, III 242, III 245, III 250 the glass transition, III 85 thermogravimetry, IV 252

284 thermomechanical data storage (TDS), III 115 thin ﬁlms, III 218, III 219, III 237, III 238, III 240, III 251, III 267 head, III 343, III 366 recording, III 362 recording media, III 361, III 362 thiol, IV 167 third-body, III 273, III 284 tip modiﬁcation, II 182 tip–sample interaction, II 7, II 12 tips, II 193 tire friction, III 283 top-to-bottom nanostructure fabrication, II 176 topographic characterization of heads, III 349 topographic characterization of the magnetic tape, III 345 topographic contrast, II 367 topography, III 299, III 301, III 307, III 318, III 323 topothesy, III 268 total harmonic, II 32 total harmonic distortion (THD), II 17 transfer function, II 13, II 22 transmission electron microscopy (TEM), II 371 transmission minima, II 5, II 12 transmission zeros, II 13 transport phenomena, II 186 transport properties of DPN, IV 4 tribological models for FFM, III 99

Subject Index tribology, III 263 truncated, II 5 model, II 15 twisting, III 227 two-level model, III 303 two-level roughness, III 303, III 307 ultrafast electrical ﬁeld sampling, II 166 ultrafast scanning probe microscopy, II 179 ultrahigh vacuum, III 42 ultramicroelectrode (UME), II 186 unbinding force, II 149, II 151–154 van der Waals interaction, II 7, IV 163 VCSEL, II 178 virus, IV 15 viscoelastic materials, III 121 vitriﬁcation, III 106 voltage modulation, III 220 voltage-modulated scanning force microscopy (VM SFM), III 222 water, III 284 waveguide properties, II 175 wet-chemical surface modiﬁcation, II 186 Williams–Lendel–Ferry (WLF), III 89 WLF behavior, III 104 writing, III 219, III 248 X ray crystallography, III 35 yield stress, III 280 Young’s modulus, III 276