Frontmatter Handbook of Micro/Nanotribology. Ed. Bharat Bhushan Boca Raton: CRC Press LLC, 1999 1999 by CRC Press LLC 1999 by CRC Press LLC Acquiring Editor: Cindy CarelliProject Editor: Andrea DembyMarketing Manager: Jane StarkCover design: Dawn BoydPrePress: Carlos EsserManufacturing: Lisa Spreckelson Library of Congress Cataloging-in-Publication Data Handbook of micro/nanotribology / edited by Bharat Bhushan. -- 2nd ed. p. cm.Includes bibliographical references and index.ISBN 0-8493-8402-8 (alk. paper)1. Tribology--Handbooks, manuals, etc. I. Bhushan, Bharat, 1949- .TJ1075.H245 1999621.8 9--dc21 98-24466 CIPThis book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted withpermission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publishreliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materialsor for the consequences of their use.Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical,including photocopying, microlming, and recording, or by any information storage or retrieval system, without priorpermission in writing from the publisher.All rights reserved. Authorization to photocopy items for internal or personal use, or the personal or internal use ofspecic clients, may be granted by CRC Press LLC, provided that $.50 per page photocopied is paid directly to CopyrightClearance Center, 27 Congress Street, Salem, MA 01970 USA. The fee code for users of the Transactional Reporting Serviceis ISBN 0-8493-8402-8/98/$0.00+$.50. The fee is subject to change without notice. For organizations that have been granteda photocopy license by the CCC, a separate system of payment has been arranged.The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for creating newworks, or for resale. Specic permission must be obtained in writing from CRC Press LLC for such copying.Direct all inquiries to CRC Press LLC, 2000 Corporate Blvd., N.W., Boca Raton, Florida 33431. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are only used for identication and explanation, without intent to infringe. 1999 by CRC Press LLCNo claim to original U.S. Government worksInternational Standard Book Number 0-8493-8402-8Library of Congress Card Number 98-24466Printed in the United States of America 1 2 3 4 5 6 7 8 9 0 1999 by CRC Press LLC Foreword The invention of the scanning tunneling microscope has led to an explosion ofa family that is now called scanning probe microscopes (SPMs). The mostpopular instrument in this family is the atomic force microscope (AFM).According to some estimates, sales of SPMs in 1993 were about 100 millionU.S. dollars (about 2,000 units installed to date) worldwide. The biggest portionof this results from AFM sales, although the rst ideas and preliminary resultswere introduced to the scientic community only a few years ago (1986). Thewhole eld of SPM is not very old (the rst operation of STM was in 1981)and still is in a rapidly evolving state. The scientic industrial applicationsinclude quality control in the semiconductor industry and related research,molecular biology and chemistry, medical studies, materials science, and the eld of information storagesystems.Soon after the invention of the AFM it was discovered that part of the information in the imagesresulted from friction and that the instrument could be used as a tool for tribology. In general, SPMsare now used intensively in this eld. Researchers can image single lubricant molecules and their agglom-eration and measure surface topography, adhesion, friction, wear, lubricant lm thickness, and mechan-ical properties all on a micrometer to nanometer scale.With the advent of more powerful computers, atomic-scale simulations have been conducted oftribological phenomena. Simulations have been able to predict the observed phenomena. Developmentof the eld of micro/nanotribology has attracted numerous physicists and chemists. This is a eld Ipersonally know very little about. I am, however, very excited that SPMs have had such an immenseimpact on the eld of tribology.I congratulate Professor Bharat Bhushan in helping to develop this eld of micro/nanotribology. The Handbook of Micro/Nanotribology is very timely and I expect that it will be well received by the interna-tional scientic community. With best wishes. Prof. Dr. Gerd Binnig IBM Research DivisionMunich, GermanyNobel Laureate Physics, 1986 1999 by CRC Press LLC Preface Second Edition, 1999 The rst edition of the Handbook of Micro/Nanotribology was published in the Spring of 1995. Soon afterits publication, the rst-of-a-kind monograph became a reference book for the novice, as well as experts,in the emerging eld of micro/nanotribology. Since the eld is evolving very rapidly, we felt that themonograph needed a second edition.The second edition is totally revised. The scope of the rst edition has been expanded. In the rstpart, Basic Studies, two new chapters on AFM Instrumentation and Tips and Surface Forces and Adhesionhave been added. In the second part, Applications, four new chapters on Design and Construction ofMagnetic Storage Devices, Microdynamic Devices and Systems, Mechanical Properties of Materials inMicrostructure Technology, and Micro/Nanotribology and Micro/Nanomechanics of MEMS Deviceshave been added. The content of each original chapter has been revised and broadened. In many cases,new authors were invited to contribute chapters on topics covered in the rst edition. Of the 16 chaptersin the book 11 in Basic Studies and ve in Applications 11 are written by new contributors, whereasve chapters have been thoroughly revised by the original authors.The organization of the Handbook is straightforward. It is divided into two parts: Part I covers thebasic studies and Part II encompasses design, construction, and applications to magnetic storage devicesand MEMS. The introduction chapter starts out with a denition and the evolution of micro/nanotri-bology, description of various measurement techniques, and description of various industrial applicationswhere the study of tribology, on a nanoscale, is critical. After this chapter, the subject matter is presentedas follows: an overview of AFM instrumentation and tips; current understanding of surface physics anddescription of methods used to physically and chemically characterize solid surfaces; roughness charac-terization and static contact models using fractal analysis; surface forces and adhesion; introduction ofsliding at the interface and study of friction on an atomic scale; study of scratching and wear as a resultof sliding, applications to nanofabrication/nanomachining, as well as nano/picoindentation; study oflubricants used to minimize friction and wear; surface forces and microrheology of thin liquid lms;measurement of nanomechanical properties of surfaces and thin lms; and atomic-scale simulations ofinterfacial phenomena.Part II includes material in the following order: design and construction of magnetic storage devices;microdynamic devices and systems; micro/nanotribology and micro/nano mechanics of magnetic storagedevices; mechanical properties of materials in microstructure technology; and micro/nanotribology andmicro/nanomechanics of MEMS devices.The Handbook is intended for graduate students of tribology and researchers who are active or intendto become so in this eld. This book should serve as an excellent text for a graduate course in micro/nan-otribology. For a reduced scope of the course, Chapters 2, 3, and 10 (Part I) and Chapters 12 and 13(Part II) can be eliminated. For a more concise course, Chapter 4 may also be eliminated. 1999 by CRC Press LLC I would like to thank the authors for their excellent contributions in a timely manner. My secretary,Kathleen Tucker, patiently typed six of the chapters contributed by me to this book. Last, but not least,I wish to thank my wife, Sudha, my son, Ankur, and my daughter, Noopur, who have been very forbearingduring the preparation of this book. Bharat Bhushan Powell, Ohio 1999 by CRC Press LLC Preface First Edition, 1995 Tribology is the science and technology of two interacting surfaces in relative motion and of relatedsubjects and practices. The popular equivalent is friction, wear, and lubrication. The advent of newtechniques to measure surface topography, adhesion, friction, wear, lubricant-lm thickness, andmechanical properties all on a micro to nanometer scale, to image lubricant molecules and availabilityof supercomputers to conduct atomic-scale simulations has led to development of a new eld referredto as Microtribology, Nanotribology, Molecular Tribology, or Atomic-Scale Tribology. This eld is con-cerned with experimental and theoretical investigations of processes ranging from atomic and molecularscales to microscale, occurring during adhesion, friction, wear, and thin-lm lubrication at slidingsurfaces. These studies are needed to develop fundamental understanding of interfacial phenomena ona small scale and to study interfacial phenomena in micro- and nano structures used in magnetic storagesystems, microelectromechanical systems (MEMS) and other industrial applications. The componentsused in micro- and nano structures are very light (on the order of few micrograms) and operate undervery light loads (on the order of few micrograms to few milligrams). As a result, friction and wear(nanoscopic wear) of lightly loaded micro/nano components are highly dependent on the surface inter-actions (few atomic layers). These structures are generally lubricated with molecularly thin lms. Micro-and nanotribological techniques are ideal to study friction and wear processes of micro/nano structures.These studies are also valuable in the fundamental understanding of interfacial phenomena in macro-structures to provide a bridge between science and engineering. Friction and wear on micro- andnanoscales have been found to be generally smaller compared to that at macroscales. Therefore, micro-nanotribological studies may identify regimes for ultra-low friction and zero wear.The eld of tribology is truly interdisciplinary. Until recently, it has been dominated by mechanicaland chemical engineers who have conducted macro tests to predict friction and wear lives in machinecomponents and devised new lubricants to minimize friction and wear. Development of the eld ofmicro/nanotribology has attracted many more physicists and chemists who have signicantly contributedto the fundamental understanding of friction and wear processes on an atomic scale. Thus, tribology isnow studied by both engineers and scientists. The micro/nanotribology eld is growing rapidly and ithas become fashionable to call oneself a tribologist. Since 1991, international conferences and courseshave been organized on this new eld of micro/nanotribology.We felt a need to develop a handbook of micro/nano tribology . This rst-of-a-kind monographpresents the state-of-the-art of the micro/nanotribology eld by the leading international researchers. Ineach chapter we start with macroconcepts leading to microconcepts. We assume that the reader is notexpert in the eld of microtribology, but has some knowledge of macrotribology. It covers characterizationof solid surfaces, various measurement techniques and their applications, and theoretical modeling ofinterfaces.The organization of the handbook is straightforward. The book is divided into two parts. Part I coversthe basic studies and Part II covers applications to magnetic storage devices and MEMS. Part I includes 1999 by CRC Press LLC ten chapters. The introduction chapter starts out with a denition and evolution of micro/nanotribology,description of various measurement techniques, and description of various industrial applications wherethe study of tribology, on a nanoscale, is critical. Following the introduction chapter, subject matter ispresented in the following order: description of methods used to physically and chemically characterizesolid surfaces, roughness characterization and static contact models using fractal analysis, introductionof sliding at the interface and study of friction, study of wear as a result of sliding, study of lubricantsused to minimize friction and wear, imaging of lubricant molecules, surface forces and microrheologyof thin liquid lms, measurement of micromechanical properties of surfaces and thin lms, and atomic-scale simulations of interfacial phenomena. Part II includes three chapters. Subject matter is presentedin the following order: micro/nanotribology and micro/nano mechanics of magnetic storage devices andMEMS, role of particulate contaminants in tribology and applications to magnetic storage devices andMEMS, and modeling of hydrodynamic lubrication with molecularly thin gas lms and applications toMEMS.The handbook is intended for graduate students of tribology and research workers who are active orintend to become active in this eld. This book should serve as an excellent text for a graduate coursein micro/nanotribology. For a reduced scope of the course, Chapters 2 and 9 from Part I and Chapters 11to 13 from Part II can be eliminated. For a further reduction, Chapter 3 from Part I can also be eliminated.I would like to thank the authors for their excellent contributions in a timely manner. And I wish tothank my wife, Sudha, my son, Ankur, and my daughter, Noopur, who have been very forbearing duringthe preparation of this book. Bharat Bhushan Powell, Ohio 1999 by CRC Press LLC The Editor Bharat Bhushan, Ph.D., a pioneer in tribology and the mechanics of magneticstorage devices, is an internationally recognized expert in the general elds ofconventional tribology and micro/nanotribology and one of the most prolicauthors. He is presently an Ohio Eminent Scholar and The Howard D. Win-bigler Professor in the Department of Mechanical Engineering and the Directorof Computer Microtribology and Contamination Laboratory at The Ohio StateUniversity, Columbus.He received a Masters in mechanical engineering from the MassachusettsInstitute of Technology, a Masters in mechanics and a Ph.D. in mechanicalengineering from the University of Colorado at Boulder, an MBA from Rens-selaer Polytechnic Institute at Troy, NY, a Doctor Technicae from the Universityof Trondheim, Norway, and a Habilitate Doctor of Technical Sciences from theWarsaw University of Technology in Poland. He is a registered professional engineer (mechanical).He has authored ve technical books, 22 handbook chapters, more than 350 technical papers in refereedjournals, more than 60 technical reports, edited more than 24 books, and holds seven U.S. patents. Heis founding editor-in-chief of the World Scientic Advances in Information Storage Systems Series, TheCRC Press Mechanics and Materials Science Series , and the Journal of Information Storage and ProcessingSystems . He has given more than 175 invited presentations on ve continents, including several keynoteaddresses at major international conferences. He is an accomplished organizer. He organized the rstsymposium on Tribology and Mechanics of Magnetic Storage Systems in 1984 and the rst internationalsymposium on Advances in Information Storage Systems in 1990. He continues to organize and chairthe AISS symposia annually. He founded an ASME Information Storage and Processing Systems Divisionand is the founding chair. His biography has been listed in over two dozen Whos Who books, including Whos Who in the World. He has received more than a dozen awards for his contributions to science andtechnology from professional societies, industry, and U.S. government agencies. He is a foreign memberof the Byelorussian Academy of Engineering and Technology, the Russian Engineering Academy and theUkrainian Academy of Transportation, a senior member of IEEE, and a member of STLE, NSPE, SigmaXi, and Tau Beta Pi.Dr. Bhushan has worked for the Department of Mechanical Engineering at the Massachusetts Instituteof Technology, Cambridge, MA; Automotive Specialists, Denver, CO; the R & D Division of MechanicalTechnology, Inc., Latham, NY; the Technology Services Division of SKF Industries, Inc., King of Prussia,PA; the General Products Division Laboratory of IBM Corporation, Tucson, AZ; the Almaden ResearchCenter of IBM Corporation, San Jose, CA; and the Department of Mechanical Engineering at theUniversity of California, Berkeley.He is married and has two children. His hobbies include music, photography, hiking, and traveling. 1999 by CRC Press LLC Contributors Phillip B. Abel, Ph.D. NASA Lewis Research CenterCleveland, Ohio Alan Berman, Ph.D. Tape Operations Seagate TechnologyCosta Mesa, California Bharat Bhushan, Ph.D. Department of Mechanical EngineeringOhio State UniversityColumbus, Ohio Donald Brenner, Ph.D. Department of Materials Science and EngineeringNorth Carolina State UniversityRaleigh, North Carolina Nancy A. Burnham, Ph.D. Department of PhysicsIGA Ecole PolytechniqueLausanne, Switzerland Jaime Colchero, Ph.D. Dep. Fisica de la Mat. CondensadaUniversidad Autonoma de MadridMadrid, Spain Fredric Ericson, Ph.D. Department of Materials ScienceUppsala UniversityUppsala, Sweden John Ferrante, Ph.D. Department of PhysicsCleveland State UniversityCleveland, Ohio Judith A. Harrison, Ph.D. Chemistry DepartmentUnited States Naval AcademyAnnapolis, Maryland Jacob N. Israelachvili, Ph.D. Department of Chemical EngineeringUniversity of CaliforniaSanta Barbara, California Hiroshi Kano, Ph.D. Recording Media DivisionSony CorporationSakuragi Tagajo City, Japan Hirofumi Kondo, Ph.D. Recording Media DivisionSony CorporationSakuragi Tagajo City, Japan Andrzej J. Kulik, Ph.D. Department of PhysicsIGA - Ecole PolytechniqueLausanne, Switzerland Othmar Marti, Ph.D. Abteilung fuer Experimentelle PhysikUniversitaet UlmUlm, Germany Ernst Meyer, Ph.D. Institut fur PhysikBasel, Switzerland Richard S. Muller, Ph.D. Berkeley Sensor & Actuator CenterUniversity of CaliforniaBerkeley, California Hiroyuki Osaki, Ph.D. Recording Media DivisionSony CorporationSakuragi Tagajo City, Japan Norio Saito, Ph.D. Recording Media DivisionSony CorporationSakuragi Tagajo City, Japan Jan-ke Schweitz, Ph.D. Department of Materials ScienceUppsala UniversityUppsala, Sweden Steven J. Stuart, Ph.D. Chemistry DepartmentUnited States Naval AcademyAnnapolis, Maryland Hiroshi Takino, Ph.D. Recording Media DivisionSony CorporationSakuragi Tagajo City, Japan Department of Mechanical Engineering ITU Istanbul, TurkeyAbdullah Aslamaci 1999 by CRC Press LLC Contents Part I Basic Studies 1 IntroductionMeasurement Techniques and Applications Bharat Bhushan 2 AFM Instrumentation and Tips Othmar Marti 3 Surface Physics in Tribology John Ferrante and Phillip B. Abel 4 Characterization and Modeling of Surface Roughness and Contact Mechanics Arun Majumdar and Bharat Bhushan 5 Surface Forces and Adhesion Nancy Burnham and Adrzej A. Kulik 6 Friction on an Atomic Scale Jaime Colchero, Ernst Meyer, and Othmar Marti 7 Microscratching/Microwear Nanofabrication/Nanomachining, and Nano/pico-indentation Using Atomic Force Microscopy Bharat Bhushan 8 Boundary Lubrication Studies Using Atomic Force/Friction Force Microscopy Bharat Bhushan 9 Surface Forces and Microrheology of Molecularly Thin Liquid Films Alan Berman and Jacob N. Israelachvili 10 Nanomechanical Properties of Solid Surfaces and Thin Films Bharat Bhushan 11 Atomic-Scale Simulations of Tribological and Related Phenomena Judith A. Harrison, Steven J. Stuart, and Donald W. Brenner Part II Applications 12 Design and Construction of Magnetic Storage Devices Hirofumi Kondo, Hiroshi Takino, Hiroyuki Osaki, Norio Saito, and Hiroshi Kano 1999 by CRC Press LLC 13 Microdynamic Systems in the Silicon Age Richard S. Muller 14 Micro/Nanotribology and Micro/Nanomechanics of Magnetic Storage Devices Bharat Bhushan 15 Mechanical Properties of Materials in Microstructure Technology Fredric Ericson and Jan-ke Schweitz 16 Micro/Nanotribology and Micro/Nanomechanics of MEMS Devices Bharat Bhushan Bhushan, B. Introduction - Measurement Techniques and Applications Handbook of Micro/Nanotribology. Ed. Bharat Bhushan Boca Raton: CRC Press LLC, 1999 1999 by CRC Press LLC Part I Basic Studies 1999 by CRC Press LLC 1 Introduction MeasurementTechniques and Applications Bharat Bhushan 1.1 History of Tribology and Its Signicance to Industry1.2 Origins and Signicance of Micro/Nanotribology1.3 Measurement Techniques Scanning Tunneling Microscope Atomic Force Microscope Friction Force Microscope Surface Force Apparatus Vibration Isolation 1.4 Magnetic Storage and MEMS Components Magnetic Storage Devices MEMS 1.5 Role of Micro/Nanotribology in Magnetic Storage Devices, MEMS, and Other MicrocomponentsReferencesIn this chapter, we rst present the history of macrotribology and micro/nanotribology and their indus-trial signicance. Next, we describe various measurement techniques used in micro/nanotribologicalstudies, then present the examples of magnetic storage devices and microelectromechanical systems(MEMS) where micro/nanotribological tools and techniques are essential for interfacial studies. Finally,we present examples of why micro/nanotribological studies are important in magnetic storage devices,MEMS, and other microcomponents. 1.1 History of Tribology and Its Signicance to Industry Tribology is the science and technology of two interacting surfaces in relative motion and of relatedsubjects and practices. The popular equivalent is friction, wear, and lubrication. The word tribology ,coined in 1966, is derived from the Greek word tribos meaning rubbing, thus the literal translation wouldbe the science of rubbing (Jost, 1966). It is only the name tribology that is relatively new, because interestin the constituent parts of tribology is older than recorded history (Dowson, 1979). It is known thatdrills made during the Paleolithic period for drilling holes or producing re were tted with bearingsmade from antlers or bones, and potters wheels or stones for grinding cereals, etc., clearly had a 1999 by CRC Press LLC requirement for some form of bearings (Davidson, 1957). A ball-thrust bearing dated about AD 40 wasfound in Lake Nimi near Rome.Records show the use of wheels from 3500 BC , which illustrates our ancestors concern with reducingfriction in translationary motion. The transportation of large stone building blocks and monumentsrequired the know-how of frictional devices and lubricants, such as water-lubricated sleds. Figure 1.1illustrates the use of a sledge to transport a heavy statue by Egyptians circa 1880 BC (Layard, 1853). Inthis transportation, 172 slaves are being used to drag a large statue weighing about 600 kN along a woodentrack. One man, standing on the sledge supporting the statue, is seen pouring a liquid into the path ofmotion; perhaps he was one of the earliest lubrication engineers. (Dowson, 1979, has estimated that eachman exerted a pull of about 800 N. On this basis the total effort, which must at least equal the frictionforce, becomes 172 800 N. Thus, the coefcient of friction is about 0.23.) A tomb in Egypt that wasdated several thousand years BC provides the evidence of use of lubricants. A chariot in this tomb stillcontained some of the original animal-fat lubricant in its wheel bearings.During and after the glory of the Roman empire, military engineers rose to prominence by devisingboth war machinery and methods of fortication, using tribological principles. It was the renaissanceengineer-artist Leonardo da Vinci (14521519), celebrated in his days for his genius in military construc-tion as well as for his painting and sculpture, who rst postulated a scientic approach to friction.Leonardo introduced, for the rst time, the concept of coefcient of friction as the ratio of the frictionforce to normal load. In 1699, Amontons found that the friction force is directly proportional to thenormal load and is independent of the apparent area of contact. These observations were veried byCoulomb in 1781, who made a clear distinction between static friction and kinetic friction.Many other developments occurred during the 1500s, particularly in the use of improved bearingmaterials. In 1684, Robert Hooke suggested the combination of steel shafts and bell-metal bushes aspreferable to wood shod with iron for wheel bearings. Further developments were associated with thegrowth of industrialization in the latter part of the 18th century. Early developments in the petroleumindustry started in Scotland, Canada, and the U.S. in the 1850s (Parish, 1935; Dowson, 1979).Although the essential laws of viscous ow had earlier been postulated by Newton, scientic under-standing of lubricated bearing operations did not occur until the end of the nineteenth century. Indeed,the beginning of our understanding of the principle of hydrodynamic lubrication was made possible bythe experimental studies of Tower (1884), the theoretical interpretations of Reynolds (1886), and relatedwork by Petroff (1883). Since then, developments in hydrodynamic bearing theory and practice wereextremely rapid in meeting the demand for reliable bearings in new machinery.Wear is a much younger subject than friction and bearing development, and it was initiated on alargely empirical basis. FIGURE 1.1 Egyptians using lubricant to aid movement of Colossus, El-Bersheh, circa 1800 BC . 1999 by CRC Press LLC Since the beginning of the 20th century, from enormous industrial growth leading to demand forbetter tribology, our knowledge in all areas of tribology has expanded tremendously (Holm, 1946; Bowdenand Tabor, 1950, 1964).Tribology is crucial to modern machinery which uses sliding and rolling surfaces. Examples of pro-ductive wear are writing with a pencil, machining, and polishing. Examples of productive friction arebrakes, clutches, driving wheels on trains and automobiles, bolts, and nuts. Examples of unproductivefriction and wear are internal combustion and aircraft engines, gears, cams, bearings, and seals. Accordingto some estimates, losses resulting from ignorance of tribology amount in the U.S. to about 6% of itsgross national product or about $200 billion per year, and approximately one third of world energyresources in present use appear as friction in one form or another. In attempting to comprehend asenormous an amount as $200 billion, it is helpful to break it down into specic interfaces. It is believedthat about $10 billion (5% of the total resources wasted at the interfaces) are wasted at the headmediuminterfaces in magnetic recording. Thus, the importance of friction reduction and wear control cannot beoveremphasized for economic reasons and long-term reliability. According to Jost (1966, 1976), the U.K.could save approximately 500 million per year, and the U.S. could save in excess of $16 billion per yearby better tribological practices. The savings are both substantial and signicant, and these savings canbe obtained without the deployment of large capital investment.The purpose of research in tribology is understandably the minimization and elimination of lossesresulting from friction and wear at all levels of technology where the rubbing of surfaces are involved.Research in tribology leads to greater plant efciency, better performance, fewer breakdowns, and sig-nicant savings. 1.2 Origins and Signicance of Micro/Nanotribology The advent of new techniques to measure surface topography, adhesion, friction, wear, lubricant lmthickness, and mechanical properties, all on a micro- to nanometer scale, and to image lubricant mole-cules and the availability of supercomputers to conduct atomic-scale simulations has led to developmentof a new eld referred to as microtribology, nanotribology, molecular tribology, or atomic-scale tribology(Bhushan et al., 1995a; Bhushan, 1997, 1998a). This eld is concerned with experimental and theoreticalinvestigations of processes ranging from atomic and molecular scales to microscales, occurring duringadhesion, friction, wear, and thin-lm lubrication at sliding surfaces. The differences between the con-ventional or macrotribology and micro/nanotribology are contrasted in Figure 1.2. In macrotribology,tests are conducted on components with relatively large mass under heavily loaded conditions. In thesetests, wear is inevitable and the bulk properties of mating components dominate the tribological perfor-mance. In micro/nanotribology, measurements are made on components, at least one of the matingcomponents, with relatively small mass under lightly loaded conditions. In this situation, negligible wearoccurs and the surface properties dominate the tribological performance.The micro/nanotribological studies are needed to develop fundamental understanding of interfacialphenomena on a small scale and to study interfacial phenomena in micro - and nanostructures used in FIGURE 1.2 Comparisons between macrotribology and micro/nanotribology. 1999 by CRC Press LLC magnetic storage systems, MEMS, and other industrial applications. The components used in micro-and nanostructures are very light (on the order of a few micrograms) and operate under very light loads(on the order of a few micrograms to a few milligrams). As a result, friction and wear (on a nanoscale)of lightly loaded micro/nanocomponents are highly dependent on the surface interactions (few atomiclayers). These structures are generally lubricated with molecularly thin lms. Micro- and nanotribologicaltechniques are ideal for studying the friction and wear processes of micro- and nanostructures. Althoughmicro/nanotribological studies are critical to study micro- and nanostructures, these studies are alsovaluable in the fundamental understanding of interfacial phenomena in macrostructures to provide abridge between science and engineering. At interfaces of technological innovations, contact occurs atmultiple asperity contacts. A sharp tip of a tip-based microscope sliding on a surface simulates a singleasperity contact, thus allowing high-resolution measurements of surface interactions at a single asperitycontact. Friction and wear on micro- and nanoscales have been found to be generally small comparedto that at macroscales. Therefore, micro/nanotribological studies may identify regimes for ultralowfriction and near zero wear.To give a historical perspective of the eld, the scanning tunneling microscope (STM) developed byDr. Gerd Binnig and his colleagues in 1981 at the IBM Zurich Research Laboratory, Forschungslabor, isthe rst instrument capable of directly obtaining three-dimensional images of solid surfaces with atomicresolution (Binnig et al., 1982). Binnig and Rohrer received a Nobel prize in physics in 1986 for theirdiscovery. STMs can only be used to study surfaces which are electrically conductive to some degree.Based on their STM design in 1985, Binnig et al. developed an atomic force microscope (AFM) to measureultrasmall forces (less than 1 N) present between the AFM tip surface and the sample surface (Binniget al., 1986a, 1987). AFMs can be used for measurement of all engineering surfaces which may be eitherelectrically conducting or insulating. AFM has become a popular surface proler for topographic mea-surements on micro- to nanoscale (Bhushan and Blackman, 1991; Oden et al., 1992; Ganti and Bhushan,1995; Poon and Bhushan, 1995; Koinkar and Bhushan, 1997a; Bhushan et al., 1997c). Mate et al. (1987)were the rst to modify an AFM in order to measure both normal and friction forces, and this instrumentis generally called friction force microscope (FFM) or lateral force microscope (LFM). Since then, anumber of researchers have used the FFM to measure friction on micro- and nanoscales (Erlandssonet al., 1988a,b; Kaneko, 1988; Blackman et al., 1990b; Cohen et al., 1990; Marti et al., 1990; Meyer andAmer, 1990b; Miyamoto et al., 1990; Kaneko et al., 1991; Meyer et al., 1992; Overney et al., 1992; Germannet al., 1993; Bhushan et al., 1994ae, 1995ag, 1997ab; Frisbie et al., 1994; Ruan and Bhushan, 1994ac;Koinkar and Bhushan, 1996ac, 1997a,c; Bhushan and Sundararajan, 1998). By using a standard or asharp diamond tip mounted on a stiff cantilever beam, AFMs can be used for scratching, wear, andmeasurements of elastic/plastic mechanical properties (such as indentation hardness and modulus ofelasticity) (Burnham and Colton, 1989; Maivald et al., 1991; Hamada and Kaneko, 1992; Miyamoto et al.,1991, 1993; Bhushan, 1995; Bhushan et al., 1994be, 1995af, 1996, 1997a,b; Koinkar and Bhushan,1996a,b, 1997b,c; Kulkarni and Bhushan, 1996a,b, 1997; DeVecchio and Bhushan, 1997).AFMs and their modications have also been used for studies of adhesion (Blackman et al., 1990a;Burnham et al., 1990; Ducker et al., 1992; Hoh et al., 1992; Salmeron et al., 1992, 1993; Weisenhorn et al.,1992; Burnham et al., 1993a,b; Hues et al., 1993; Frisbie et al., 1994; Bhushan and Sundararajan, 1998),electrostatic force measurements (Martin et al., 1988; Yee et al., 1993), ion conductance and electrochem-istry (Hansma et al., 1989; Manne et al., 1991; Binggeli et al., 1993), material manipulation (Weisenhornet al., 1990; Leung and Goh, 1992), detection of transfer of material (Ruan and Bhushan, 1993), thin-lm boundary lubrication (Blackman et al., 1990a,b; Mate and Novotny, 1991; Mate, 1992; Meyer et al.,1992; OShea et al., 1992; Overney et al., 1992; Bhushan et al., 1995f,g; Koinkar and Bhushan, 1996bc),to measure lubricant lm thickness (Mate et al., 1989, 1990; Bhushan and Blackman, 1991; Koinkar andBhushan, 1996c), to measure surface temperatures (Majumdar et al., 1993; Stopta et al., 1995), formagnetic force measurements including its application for magnetic recording (Martin et al., 1987b;Rugar et al., 1990; Schonenberger and Alvarado, 1990; Grutter et al., 1991, 1992; Ohkubo et al., 1991;Zuger and Rugar, 1993), and for imaging crystals, polymers, and biological samples in water (Drakeet al., 1989; Gould et al., 1990; Prater et al., 1991; Haberle et al., 1992; Hoh and Hansma, 1992). STMs 1999 by CRC Press LLC have been used in several different ways. They have been used to image liquids such as liquid crystalsand lubricant molecules on graphite surfaces (Foster and Frommer, 1988; Smith et al., 1989, 1990; Andohet al., 1992), to manipulate individual atoms of xenon (Eigler and Schweizer, 1990) and silicon (Lyo andAvouris, 1991), in formation of nanofeatures by localized heating or by inducing chemical reactionsunder the STM tip (Abraham et al., 1986; Silver et al., 1987; Albrecht et al., 1989; Mamin et al., 1990;Utsugi, 1990; Hosoki et al., 1992; Kobayashi et al., 1993), and nanomachining (Parkinson, 1990). AFMshave also been used for nanofabrication (Majumdar et al., 1992; Bhushan et al., 1994be, Bhushan, 1995,1997; Boschung et al., 1994; Tsau et al., 1994) and nanomachining (Delawski and Parkinson, 1992).Instruments that are able to measure tunneling current and forces simultaneously are being custombuilt (Specht et al., 1991; Anselmetti et al., 1992). Coupled AFM/STM measurements are made to dis-tinguish between the topography of a sample and its electronic structure. Another aim is to determinethe role of pressure in the tunnel junction in obtaining STM images.Surface force apparatuses (SFA) are used to study both static and dynamic properties of the molecularlythin liquid lms sandwiched between two molecularly smooth surfaces. Tabor and Winterton (1969) andlater Israelachvili and Tabor (1972) developed apparatuses for measuring the van der Waals forces betweentwo molecularly smooth mica surfaces as a function of separation in air or vacuum. These techniqueswere further developed for making measurements in liquids or controlled vapors (Israelachvili andAdams, 1978; Klein, 1980; Tonck et al., 1988; Georges et al., 1993). Israelachvili et al. (1988), Homola(1989), Gee et al. (1990), Homola et al. (1990, 1991), Klein et al. (1991), and Georges et al. (1994)measured the dynamic shear response of liquid lms. Recently, new friction attachments were developedwhich allow for two surfaces to be sheared past each other at varying sliding speeds or oscillatingfrequencies, while simultaneously measuring both the friction forces and normal forces between them(Van Alsten and Granick, 1988, 1990a,b; Peachey et al., 1991; Hu et al., 1991). The distance between twosurfaces can also be independently controlled to within 0.1 nm and the force sensitivity is about 10 8 N.The SFAs are being used to study the rheology of molecularly thin liquid lms; however, the liquidunder study has to be conned between molecularly smooth, optically transparent surfaces with radii ofcurvature on the order of 1 mm (leading to poorer lateral resolution as compared with AFMs). SFAsdeveloped by Tonck et al. (1988) and Georges et al. (1993, 1994) use an opaque and smooth ball with alarge radius (~3 mm) against an opaque and smooth at surface. Only AFMs/FFMs can be used to study engineering surfaces in the dry and wet conditions with atomic resolution .The interest in the micro/nanotribology eld grew from magnetic storage devices and its applicabilityto MEMS is clear. In this chapter, we rst describe various measurement techniques, and then we presentthe examples of magnetic storage devices and MEMS where micro/nanotribological tools and techniquesare essential for interfacial studies. We then present examples of why micro/nanotribological studies areimportant in magnetic storage devices, MEMS, and other microcomponents. 1.3 Measurement Techniques The family of instruments based on STMs and AFMs are called scanning probe microscopes (SPMs).These include STM, AFM, FFM (or LFM), scanning magnetic microscopy (SMM) (or magnetic forcemicroscopy, MFM), scanning electrostatic force microscopy (SEFM), scanning near-eld optical micros-copy (SNOM), scanning thermal microscopy (SThM), scanning chemical force microscopy (SCFM),scanning electrochemical microscopy (SEcM), scanning Kelvin probe microscopy (SKPM), scanningchemical potential microscopy (SCPM), scanning ion conductance microscopy (SICM), and scanningcapacitance microscopy (SCM). The family of instruments which measures forces (e.g., AFM, FFM, SMM,and SEFM) are also referred to as scanning force microscopics (SFM). Although these instruments offeratomic resolution and are ideal for basic research, they are also used for cutting-edge industrial applica-tions which do not require atomic resolution. Commercial production of SPMs started with STM in1988 by Digital Instruments, Inc., and has grown to over $100 million in 1993 (about 2000 units installedto 1993) with an expected annual growth rate of 70%. For comparisons of SPMs with other microscopes,see Table 1.1 (Aden, 1994). The numbers of these instruments are equally divided among the U.S., Japan, 1999 by CRC Press LLCand Europe with the following industry/university and government laboratory splits: 50/50, 70/30, and30/70, respectively. According to some estimates, over 3000 users of SPMs exist with $400 million insupport. It is clear that research and industrial applications of SPMs are rapidly expanding.STMs, AFMs, and their modications can be used at extreme magnications ranging from 103 to 109in x-, y-, and z-directions for imaging macro- to atomic dimensions with high-resolution informationand for spectroscopy. These instruments can be used in any sample environment such as ambient air(Binnig et al., 1986a), various gases (Burnham et al., 1990), liquid (Marti et al., 1987; Drake et al., 1989;Binggeli et al., 1993), vacuum (Binnig et al., 1982; Meyer and Amer, 1988), low temperatures (Coombsand Pethica, 1986; Kirk et al., 1988; Giessibl et al., 1991; Albrecht et al., 1992; Hug et al., 1993), and hightemperatures. Imaging in liquid allows the study of live biological samples, and it also eliminates watercapillary forces present in ambient air present at the tipsample interface. Low-temperature (liquidhelium temperatures) imaging is useful for the study of biological and organic materials and the studyof low-temperature phenomena such as superconductivity or charge density waves. Low-temperatureoperation is also advantageous for high-sensitivity force mapping due to the reduction in thermalvibration. These instruments are used for proximity measurements of magnetic, electrical, chemical,optical, thermal, spectroscopy, friction, and wear properties. Their industrial applications include micro-circuitry and semiconductor industry, information storage systems, molecular biology, molecular chem-istry, medical devices, and materials science.1.3.1 Scanning Tunneling MicroscopeThe principle of electron tunneling was proposed by Giaever (1960). He envisioned that if a potentialdifference is applied to two metals separated by a thin insulating lm, a current will ow because of theability of electrons to penetrate a potential barrier. To be able to measure a tunneling current, the twometals must be spaced no more than 10 nm apart. Binnig et al. (1982) introduced vacuum tunnelingcombined with lateral scanning. The vacuum provides the ideal barrier for tunneling. The lateral scanningallows one to image surfaces with exquisite resolution, lateral less than 1 nm and vertical less than 0.1 nm,sufcient to dene the position of single atoms. The very high vertical resolution of STM is obtainedbecause the tunnel current varies exponentially with the distance between the two electrodes, that is, themetal tip and the scanned surface. Typically, tunneling current decreases by a factor of 2 as the separationis increased by 0.2 nm. Very high lateral resolution depends upon the sharp tips. Binnig et al. (1982)overcame two key obstacles for damping external vibrations and for moving the tunneling probe in closeproximity to the sample; their instrument is called the STM. Todays STMs can be used in the ambientenvironment for atomic-scale imaging of surfaces. Excellent reviews on this subject are presented by Pohl(1986), Hansma and Tersoff (1987), Sarid and Elings (1991), Durig et al. (1992), Frommer (1992),Guntherodt and Wiesendanger (1992), Wiesendanger and Guntherodt (1992), Bonnell (1993), Marti andAmrein (1993), Stroscio and Kaiser (1993), and Anselmetti et al. (1995) and the following dedicatedissues of the Journal of Vacuum Science Technolology (B9, 1991, pp. 4011211) and Ultramicroscopy (Vol.4244, 1992).TABLE 1.1 Comparison of Various Conventional Microscopes with SPMsOptical Confocal SEM/TEM SPMMagnication 103104107109Instrument price, U.S. $ 10k 30k 250k 100kTechnology age 200 yrs 10 yrs 30 yrs 9 yrsApplications Ubiquitous New and unfolding Science and technology Cutting-edgeMarket 1993 $800M $80M $400M $100MGrowth rate 10% 30% 10% 70%Data provided by Topometrix. 1999 by CRC Press LLCThe principle of STM is straightforward. A sharp metal tip (one electrode of the tunnel junction) isbrought close enough (0.3 to 1 nm) to the surface to be investigated (second electrode) that, at aconvenient operating voltage (10 mV to 1 V), the tunneling current varies from 0.2 to 10 nA, which ismeasurable. The tip is scanned over a surface at a distance of 0.3 to 1 nm, while the tunneling currentbetween it and the surface is sensed. The STM can be operated in either the constant-current mode orthe constant-height mode, Figure 1.3. The left-hand column of Figure 1.3 shows the basic constantcurrent mode of operation. A feedback network changes the height of the tip z to keep the currentconstant. The displacement of the tip given by the voltage applied to the piezoelectric drives then yieldsa topographic picture of the surface. Alternatively, in the constant-height mode, a metal tip can be scannedacross a surface at nearly constant height and constant voltage while the current is monitored, as shownin the right-hand column of Figure 1.3. In this case, the feedback network responds only rapidly enoughto keep the average current constant (Hansma and Tersoff, 1987). A current mode is generally used foratomic-scale images. This mode is not practical for rough surfaces. A three-dimensional picture [z(x, y)]of a surface consists of multiple scans [z(x)] displayed laterally from each other in the y direction. Itshould be noted that if different atomic species are present in a sample, the different atomic specieswithin a sample may produce different tunneling currents for a given bias voltage. Thus, the height datamay not be a direct representation of the topography of the surface of the sample.1FIGURE 1.3 Scanning tunneling microscope can be operated in either the constant-current or the constant-heightmode. The images are of graphite in air. (From Hansma, P. K. and Tersoff, J. (1987), J. Appl. Phys., 61, R1R23. Withpermission.)1In fact, Marchon et al. (1989) STM imaged sputtered diamond-like carbon lms in barrier-height mode bymodulating the tip-to-surface distance, with lock-in detection of the tunneling current. The local barrier-heightmeasurements give information on the local values of the work function, thus providing chemical information, inaddition to the topographic map. 1999 by CRC Press LLC1.3.1.1 Binnig et al.s DesignFigure 1.4 shows a schematic of one of Binnig and Rohrers designs for operation in an ultrahigh vacuum(Binnig et al., 1982; Binnig and Rohrer, 1983). The metal tip was xed to rectangular piezodrives Px, Py,and Pz made out of commercial piezoceramic material for scanning. The sample is mounted on either asuperconducting magnetic levitation or two-stage spring system to achieve the stability of a gap widthof about 0.02 nm. The tunnel current JT is a sensitive function of the gap width d; that is, JTVTexp(A1/2d), where VT is the bias voltage, is the average barrier height (work function) and A ~ 1 if is measured in eV and d in . With a work function of a few eV, JT changes by an order of magnitudefor every angstrom change of h. If the current is kept constant to within, for example, 2%, then the gaph remains constant to within 1 pm. For operation in the constant-current mode, the control unit (CU)applies a voltage Vz to the piezo Pz such that JT remains constant when scanning the tip with Py and Pxover the surface. At the constant-work functions , Vz(Vx, Vy) yields the roughness of the surface z(x, y)directly, as illustrated at a surface step at A. Smearing the step, (lateral resolution) is on the order of(R)1/2, where R is the radius of the curvature of the tip. Thus, a lateral resolution of about 2 nm requirestip radii on the order of 10 nm. A 1-mm-diameter solid rod ground at one end at roughly 90 yieldsoverall tip radii of only a few hundred nanometers, but with closest protrusion of rather sharp microtipson the relatively dull end yielding a lateral resolution of about 2 nm. In situ sharpening of the tips bygently touching the surface brings the resolution down to the 1-nm range; by applying high elds (onthe order of 108 V/cm) during, for example, half an hour, resolutions considerably below 1 nm could bereached. Most experiments were done with tungsten wires either ground or etched to a radius typicallyin the range of 0.1 to 10 m. In some cases, in situ processing of the tips was done for further reductionof tip radii.1.3.1.2 Commercial STMsThere are a number of commercial STMs available on the market. Digital Instruments, Inc., located inSanta Barbara, CA introduced the rst commercial STM, the Nanoscope I, in 1987. In the NanoscopeIII STM for operation in ambient air, the sample is held in position while a piezoelectric crystal in theform of a cylindrical tube scans the sharp metallic probe over the surface in a raster pattern while sensingand outputting the tunneling current to the control station, Figure 1.5 (Anonymous, 1992b). The digitalsignal processor (DSP) calculates the desired separation of the tip from the sample by sensing thetunneling current owing between the sample and the tip. The bias voltage applied between the sampleand the tip encourages the tunneling current to ow. The DSP completes the digital feedback loop byoutputting the desired voltage to the piezoelectric tube. The STM operates in both the constant-heightand constant-current modes depending on a parameter selection in the control panel. In the constant-current mode, the feedback gains are set high, the tunneling tip closely tracks the sample surface, andFIGURE 1.4 Principle of operation of the STM made by Binnig and Rohrer (1983). 1999 by CRC Press LLCthe variation in the tip height required to maintain constant tunneling current is measured by the changein the voltage applied to the piezotube. In the constant-height mode, the feedback gains are set low, thetip remains at a nearly constant height as it sweeps over the sample surface, and the tunneling currentis imaged. The following description of the instrument is almost exclusively based on Anonymous(1992b).Physically, the Nanoscope STM consists of three main parts: the head which houses the piezoelectrictube scanner for three-dimensional motion of the tip and the preamplier circuit (FET input amplier)mounted on top of the head for the tunneling current, the base on which the sample is mounted, andthe base support, which supports the base and head, Figure 1.6A. The assembly is connected to a controlsystem that controls the operation of the microscope. The base accommodates samples up to 10 20 mmand 10 mm in thickness. The different scanning heads mount magnetically on the tripod formed by thefront, coarse-adjust screws and the rear, nd-adjust screws. Optional scan heads for the STM include 0.7(for atomic resolution), 12, 75, and 125 m square.The scanning head controls the three-dimensional motion of tip. The removable head consists of apiezotube scanner, about 12.7 mm in diameter, mounted into an Invar shell used to minimize verticalthermal drifts because of good thermal match between the piezotube and the Invar. The piezotube hasseparate electrodes for X, Y, and Z which are driven by separate drive circuits. The electrode conguration(Figure 1.6B) provides X and Y motions, which are perpendicular to each other, minimizes horizontaland vertical coupling, and provides good sensitivity. The vertical motion of the tube is controlled by theZ-electrode which is driven by the feedback loop. The X and Y scanning motions are each controlled bytwo electrodes which are driven by voltages of the same magnitudes, but opposite signs. These electrodesare called Y, X, +Y, and +X. Applying complimentary voltages allows a short, stiff tube to provide agood scan range without large voltages. The motion of the tip due to external vibrations is proportionalto the square of the ratio of vibrational frequency to the resonant frequency of the tube. Therefore, tominimize the tip vibrations, the resonant frequencies of the tube are high, about 60 kHz in the verticaldirection and about 40 kHz in the horizontal direction. The tip holder is a stainless steel tube with a300-m inner diameter for 250-m-diameter tips, mounted in ceramic in order to keep the mass on theend of the tube low. The tip is mounted either on the front edge of the tube (to keep mounting masslow and resonant frequency high) (Figure 1.5) or the center of the tube for large-range scanners, namely75 and 125 m (to preserve the symmetry of the scanning). This commercial STM will accept any tipwith a 250-m-diameter shaft. The piezotube requires XY calibration which is carried out by imagingan appropriate calibration standard. Cleaved graphite is used for the small-scan length head while two-dimensional grids (a gold-plated ruling) can be used for longer-range heads.The Invar base holds the sample in position, supports the head, and provides XY motion for thesample, Figure 1.6C. A spring-steel sample clip with two thumbscrews holds the sample in place. An XYtranslation stage built into the base allows the sample to be repositioned under the tip. Three precisionFIGURE 1.5 Principle of operation of a commercial STM, a sharp tip attached to a piezoelectric tube scanner isscanned on a sample. 1999 by CRC Press LLCscrews arranged in a triangular pattern support the head and provide coarse and ne adjustment of thetip height. The base support consists of the base support ring and the motor housing. The base supportring cradles the base allowing access to the adjustment screws. The stepper motor enclosed in the motorhousing allows the tip to be engaged and withdrawn from the surface automatically.For measurements, the sample is placed under the sample-holding clip, with about half the sampleextending forward of the wire using an appropriate scanner, and a tip is inserted in the tip-holding tubemounted on the piezotube. The tip is gripped with a tweezer near the sharp end and the blunt end ofthe tip is inserted into the tip holder. For the tip to be held in the tube, it is necessary to put a smallbend in the tip before it is completely inserted. Next, the scanning head is placed on the three magneticFIGURE 1.6 Schematics of a commercial STM made by Digital Instruments, Inc.: (A) front view, (B) generalelectrode conguration for piezoelectric tube scanner, and (C) front and top view of the STM base. (From Anonymous(1992), Nanoscope III Scanning Tunneling Microscope, Instruction Manual, Courtesy of Digital Instruments, Inc.,Santa Barbara, CA, 1992.) 1999 by CRC Press LLCballs mounted on the threaded screws in the base. The tip is lowered with the coarse-adjustment screwsuntil there is only a slight gap, less than 0.25 mm (the tip will be damaged if brought into contact)between the end of the tip and its reected image visible on the sample. Next the scan parameters areset, the motor is turned on, which engages the tip, and the scanning is initiated to form a desired imageof the sample surface.Samples to be imaged with STM must be conductive enough to allow a few nanoamperes of currentto ow from the bias voltage source to the area to be scanned. In many cases, nonconductive samplescan be coated with a thin layer of a conductive material to facilitate imaging. The bias voltage and thetunneling current depend on the sample. Usually they are set at a standard value for engagement andne-tuned to enhance the quality of the image. The scan size depends on the sample and the features ofinterest. A maximum scan rate of 122 Hz can be used. The maximum scan rate is usually related to thescan size. Scan rate above 10 Hz is used for small scans (typically 60 Hz for atomic-scale imaging witha 0.7-m scanner). The scan rate should be lowered for large scans, especially if the sample surfaces arerough or contain large steps. Moving the tip quickly along the sample surface at high scan rates withlarge scan sizes will usually lead to a tip crash. Essentially, the scan rate should be inversely proportionalto the scan size (typically 2 to 4 Hz for 1 m, 0.5 to 1 Hz for 12 m, and 0.2 Hz for 125 m scan sizes).Scan rate in length/time is equal to scan length divided by the scan rate in hertz. For example, for 10 10 m scan size scanned at 0.5 Hz, the scan rate is 20 m/s. Typically, 256 256 data formats are mostcommonly used. The lateral resolution at larger scans is approximately equal to scan length divided by 256.Figure 1.7 shows an example of an STM image of freshly cleaved, highly oriented pyrolytic graphite(HOPG) surface taken at room temperature and ambient pressure (Binnig et al., 1986b; Park and Quate,1986; Ruan and Bhushan, 1994b).1.3.1.2.1 Electrochemical STM (ECSTM)Electrochemical STM (ECSTM) allows the performance of the electrochemical reactions on the STM. Itincludes a microscope base with an integral potentiostat, a short head with a 0.7-m scan range, and adifferential preamp and the software required to operate the potentiostat and display the result ofelectrochemical reaction.1.3.1.2.2 Stand-Alone STMThe stand-alone STMs are available to scan large samples which rest directly on the sample. From DigitalInstruments, Inc., it is available in 12- and 75-m scan ranges. It is similar to the standard STM exceptthe sample base has been eliminated. Two coarse- and one ne-adjustment screws used to position thetip manually relative to the sample surface are mounted in the head shell.1.3.1.3 Tip ConstructionThe STM cantilever should have a sharp metal tip with a low aspect ratio (tip length/tip shank) tominimize exural vibrations. Ideally, the tip should be atomically sharp, but, in practice, most tipFIGURE 1.6(C) 1999 by CRC Press LLCpreparation methods produce a tip which is rather ragged and consists of several asperities with the oneclosest to the surface responsible for tunneling. STM cantilevers with sharp tips are typically fabricatedfrom metal wires of tungsten (W), platinumiridium (PtIr), or gold (Au) and sharpened by grinding,cutting with a wire cutter or razor blade, eld emission/evaporator, ion milling, fracture, or electrochem-ical polishing/etching (Ibe et al., 1990). The two most commonly used tips are made from either a PtIr(80/20) alloy or tungsten wire. Iridium is used to provide stiffness. The PtIr tips are generally mechan-ically formed and are readily available. The tungsten tips are etched from tungsten wire with an electro-chemical process, for example, by using 1 mol KOH solution with a platinum electrode in anelectrochemical cell at about 30 V. In general, PtIr tips provide better atomic resolution than tungstentips, probably due to the lower reactivity of Pt, but tungsten tips are more uniformly shaped and mayperform better on samples with steeply sloped features. The wire diameter used for the cantilever istypically 250 m with the radius of curvature ranging from 20 to 100 nm and a cone angle ranging from10 to 60, Figure 1.8. The wire can be bent in an L shape, if so required for use in the instrument. Forcalculations of normal spring constant and natural frequency of round cantilevers, see Sarid and Elings(1991).Controlled geometry (CG) Pt-Ir probes are commercially available, Figure 1.9. These probes areelectrochemically etched from Pt-Ir (80/20) wire and polished to a specic shape which is consistentfrom tip to tip. Probes have a full cone angle of approximately 15 and a tip radius of less than 50 nm.For imaging of deep trenches (>0.25 m) and nanofeatures, focused ion beam (FIB) milled CG probeswith an extremely sharp tip radius (0.25 m). Because of exing of the probe, it is unsuitable for imagingstructures at the atomic level since the exing of the probe can create image artifacts. For atomic-scaleimaging, an FIB-milled probe is used that is relatively stiff yet allows for closely spaced topography. Theseprobes start out as conventional Si3N4 pyramidal probes, but the pyramid is FIB milled until a small coneshape is formed which has a high aspect ratio that is 0.2 to 0.3 m in length. The milled probes allownanostructure resolution without sacricing rigidity. These probes are manufactured by Materials Ana-lytical Services, Raleigh, NC.Ruger et al. (1992) and Sidles and Rugar (1993) developed a cantilever beam which allowed themeasurements of forces on the order of 1016 to 1018 N. They used an Si3N4 beam 10 m in length andabout 20 nm thick with a spring constant of about 105 N/m.Binh and Garcia (1991, 1992) made nanocantilevers with high resonant frequencies on the order oftens of kilohertz and a spring constant of around 1 N/m from W and Cu. They produced these with anextremely thin round cantilever with a spherical tip at its end for measurement of very small van derWaals forces (Garcia and Binh, 1992). For a cantilever beam radius, cantilever length, and the ball tipradius of 10, 200, and 150 nm, respectively, made of W, they achieved a resonant frequency of 60 kHzand a spring constant of 0.3 N/m. 1.3.3 Friction Force Microscope (FFM)1.3.3.1 Mate et al.s DesignThe rst FFM was developed by Mate et al. (1987) at IBM Almaden Research Center, San Jose, CA. Intheir setup, the sample was mounted on three orthogonal piezoelectric tubes (25 mm long), two of which(x-, y-axes) raster the sample in the surface plane with the third (z) moving the sample toward and awayfrom the tip, Figure 1.27A. A tungsten wire (12 mm long, 0.25 or 0.5 mm in diameter) was used as thecantilever, one end (the free end) of which was bent at a right angle and was electrochemically etchedin NaOH solution to obtain a sharp point (150 to 300 nm in radius) to serve as the tip. The springFIGURE 1.25 Schematic of rectangular cantilever stainless steel beam with three-sided pyramidal natural diamondtip. 1999 by CRC Press LLCconstants were 150 and 2500 N/m for the 0.25- and 0.5-mm-diameter wires, respectively. A laser beamwas used to monitor cantilever deection in the lateral direction. A light beam was focused on the edgeof the lever by a microscope objective. The interference pattern between the reected and the referencebeams reected from the tungsten wire and an optical at was projected on a photodiode to measurethe instantaneous deection of the lever. Normal force was approximated by using a calibrated piezo-electric tube extension. The force was determined by multiplying the cantilever deection by the springconstant of the cantilever in the lateral direction. Later, Erlandsson et al. (1988b) used two independentlaser beams to monitor cantilever deections in the normal and lateral directions to measure normaland friction forces, Figure 1.27B. They included another laser beam in the lateral direction (Figure 1.27B)to their original AFM design shown in Figure 1.13. Mate et al. (1987) measured the friction of a tungstentip sliding against a freshly cleaved HOP graphite by pushing the tip against the sample in the z-directionat desired loads (ranging from 7.5 to 56 N) by moving the sample back and forth parallel to the surfaceplane at a velocity of 40 nm/s, and repeating the scanning by stepping the sample (for three-dimensionalproling). Erlandsson et al. (1988b) measured the atomic-scale friction of muscovite mica. Germannet al. (1993) measured the atomic-scale friction of diamond surfaces.1.3.3.2 Kaneko et al.s DesignThe second type of FFM was developed by Kaneko and his co-workers (Kaneko, 1988; Kaneko et al.,1991). Their earlier design is shown in Figure 1.28 (Kaneko, 1988). A diamond tip was held by a parallel-leaf spring unit (length = 10 mm, width = 1 mm, thickness = 20 m, spring constant = 3N/m). Thesample was mounted on another Parallel-leaf spring unit (length = 10 mm, width = 3 mm, thickness =20 to 30 m, spring constant 9 to 30 N/m). These parallel-leaf springs have greater torsional rigidity thansingle-leaf springs with the same spring constants. Thus deection errors caused by tip movement arereduced by using parallel-leaf springs (Miyamoto et al., 1990). A piezoelectric tripod (16 m stroke at100 V) was used for loading the sample against the tip as well as moving it in the two directions in thesample plane. A focusing error-detection-type optical head (resolution 22fxxxP

1999 by CRC Press LLCEquation 2.98 shows this effect for the fast scanning direction. In frequency the sampling theorem meansthat the digitizing rate needs to be at least twice as high as the highest frequency in the signal. An exampledemonstrates this. If one wants to measure something with a 1-nm periodicity at a scanning speed of1 m/s at least 2000 samples per second are needed, better yet 12,000 samples per second. In thetime/frequency domain it is easy to prevent aliasing. Just use a low-pass lter with the maximum allowablefrequency (the Nyquist frequency) in front of the analog-to-digital converter. It prevents aliasing. If thesampling theorem is violated, then nothing will be seen.2.9.5 Typical SetupsThe analysis of the sampling theorem suggests that a typical setup for the analog-to-digital conversionshould look like Figure 2.33 or Figure 2.34. The difference between the two gures lies in the fact thatfor an analog feedback circuit one has to digitize the topographical signal, which is at the output of thefeedback controller, whereas in a digital feedback circuit the computer calculates the z-signal from theerror signal. In both cases, one must match the cutoff frequency of the low-pass lters to the conversionrate of the digital-to-analog (D/A) or analog-to-digital (A/D) converters.The precision of the converters limits the resolution; 16 bits are equal to one part in 65,536. Byobserving the sampling theorem, this means that the smallest object must be 32,768 times smaller thanFIGURE 2.32 Effect of the sampling theorem. The successive lines from the top to the bottom are the original andsampled data. The original contains the following objects (from the left side): sine with period 40 (from 0 to 80),sine with period 20 (from 80 to 120), sine with period 10 (from 120 to 160), sine with period 5 (from 160 to 190),a square wave with period 18 (190 to 256), a square wave with period 10 (256 to 286), a square wave with period 4(286 to 310), a single square pulse of width 12 between 310 and 356, a triangle of period 40 between 356 and 396,a triangle of period 20 between 400 and 420, a triangle of period 12 between 430 and 442, and, nally, delta pulses(1 wide) at 458, 469, 480, 491, and 509. 1999 by CRC Press LLCthe scanning range. Table 2.3 shows a compilation of the resolution as a function of the most commonmaximal scanning ranges and some D/A (or A/D) converter resolutions. Since the 16-bit converters arevery common (because of the CDaudio market), most instruments use this resolution. A 100-m scanneris capable of resolving a little bit more than 1 nm.2.9.6 Data RepresentationThe visualization and interpretation of images from SPMs is intimately connected to the processing ofthese images.An ideal scanning probe microscope is a noise-free device that images a sample with perfect tips ofknown shape and has perfect linear scanning piezos. In reality, AFMs are not that ideal. The scanningdevice in AFMs is affected by distortions. To do quantitative measurements, such as determining the unitcell size, these distortions have to be measured on test substances and have to be corrected for. TheFIGURE 2.33 Typical setup of a data acquisition system with an analog feedback circuit.FIGURE 2.34 Typical setup of a data acquisition system with a digital feedback system.TABLE 2.3 Sampling Resolution (in nanometers) as a Function of the Scan Range and the Converter Resolution8 Bit 12 Bit 16 Bit 18 Bit 20 Bit1 nm 0.008 4.9E-04 3.1E-05 7.6E-06 1.9E-0610 nm 0.078 4.9E-03 3.1E-04 7.6E-05 1.9E-05100 nm 0.78 0.05 0.003 7.63E-04 1.91E-041 m 7.81 0.49 0.031 0.008 0.00210 m 78.13 4.88 0.31 0.08 0.02100 m 781.25 48.83 3.05 0.76 0.19250 m 1953.13 122.07 7.63 1.91 0.48 1999 by CRC Press LLCdistortions are both linear and nonlinear. Linear distortions mainly result from imperfections in themachining of the piezotranslators causing cross talk from the z-piezo to the x- and y-piezos, and viceversa. Among the linear distortions, there are two kinds which are very important: rst, scanning piezosinvariably have different sensitivities along the different scan axes due to the variation of the piezo materialand uneven sizes of the electrode areas. Second, the same reasons might cause the scanning axis to notbe orthogonal. Furthermore, the plane in which the piezoscanner moves for constant z is hardly evercoincident with the sample plane. Hence, a linear ramp is added to the sample data. This ramp is especiallybothersome when the height z is displayed as an intensity map, also called a top-view display.The nonlinear distortions are harder to deal with. They can affect AFMs for a variety of reasons. First,piezoelectric ceramics do have a hysteresis loop, much like ferromagnetic materials. The deviations ofpiezoceramic materials from linearity increase with increasing amplitude of the driving voltage. Themechanical position for one voltage depends on the voltages applied to the piezo before. Hence, to getthe best position accuracy, one should always approach a point on the sample from the same direction.Another type of nonlinear distortion of the images occurs when the scan frequency approaches the upperfrequency limit of the x- and y-drive ampliers or the upper frequency limit of the feedback loop(z-component). This distortion, due to the feedback loop, can only be minimized by reducing the scanfrequency. On the other hand, there is a simple way to reduce distortions due to the x- and y-piezo driveampliers. To keep the system as simple as possible, one normally uses a triangular waveform for drivingthe scanning piezos. However, triangular waves contain frequency components at multiples of the scanfrequency. If the cutoff frequency of the x- and y-drive electronics or of the feedback loop is too closeto the scanning frequency (two to three times the scanning frequency), the triangular drive voltage isrounded off at the turning points. This rounding error causes, rst, a distortion of the scan linearity and,second, through phase lags, the projection of part of the backward scan onto the forward scan. This typeof distortion can be minimized by carefully selecting the scanning frequency and by using driving voltagesfor the x- and y-piezos with waveforms like trapezoidal waves, which are closer to a sine wave. The valuesmeasured for x, y, or z are affected by noise. The origin of this noise can be either electronic, somedisturbances, or a property of the sample surface due to adsorbates. In addition to this incoherent noise,interference with mains and other equipment nearby might be present. Depending on the type of noise,one can lter it in the real space or in Fourier space. The most important part of image processing is tovisualize the measured data. Typical AFM data sets can consist of many thousands to over a millionpoints per plane. There may be more than one image plane present. The AFM data represents a topo-graphy in various data spaces.We use data from a combined measurement of topography, stiffness, and adhesion to outline thedifferent points of data processing. Figure 2.35a shows the topography, Figure 2.35b the local stiffnessimage, and Figure 2.35c the adhesion image acquired simultaneously, as described in (Marti et al., 1997).The same data can be rendered as a series of consecutive cross sections. Figure 2.36 is an example. Linerenderings can be excellent to judge the general form of the topography, independent of color settings.The original data set contains 256 lines with 256 points each. Since it is not possible to draw 256 distinctlines, the data in Figure 2.36 show every sixth line. Along the lines, all the points are drawn.Similarly, the data can also be rendered in a wire mesh fashion (Figure 2.37). Again, a line is drawnat every sixth point both in the horizontal and vertical direction. It is especially suitable for monochromedisplay systems with only two colors. The number of scan lines that can be displayed is usually well below100 and the display resolution along the fast scanning axis x is much better than along y.If the computer display is capable of displaying at least 64 shades of gray, then top-view images canbe created (Figure 2.38). In these images, the position on the screen corresponds to the position on thesample and the height is coded as a shade of gray. Usually, the convention is that the brighter a point,the higher it is. The number of points that can be displayed is only limited by the number of pixelsavailable. This view of the data is excellent for measuring distances between surface features. Periodicstructures show up particularly well on such a top view. The human eye is not capable of distinguishingmore than 64 shades of gray. If the average z-height of the tip varies from one side of the image to theother, then the interesting features usually have too little contrast. Hence, contrast equalization is needed. 1999 by CRC Press LLCFor data being affected by a large background slope, it is often still possible to detect some features inthe line scan view. Some researchers prefer a simultaneous display of both line scan images and top-viewimages to get the most information in the shortest time. Top views use much less calculation time thanline scan images. Hence, computerized fast data acquisition systems usually display the data as a top viewrst.Figure 2.35 shows in a detailed way what the topography of the sample is. However, the ne detailsin the low-lying parts are obscured. There are simply not enough shades which the eye can distinguish.The display can be enhanced by adding some illumination to it. In illuminated data sets one depicts thecosine of the angle between the surface normal and some predened direction, the illumination direction.This technique is a powerful tool to enhance the appearance of a data set, but it can be abused! Changingthe direction of the light source, as shown in Figure 2.39, can obscure some undesired features.The effect of the illumination is similar to displaying the magnitude of the gradient of the samplesurface along the direction to the light source. Features perpendicular to the illumination cannot be seen.FIGURE 2.35 A rubber sample: (a) shows the topography, (b) the local stiffness, and (c) the adhesion. 1999 by CRC Press LLCMultiple light sources or extended light sources diminish this effect, but the illumination is much morecomplicated to calculate. If not shown in conjunction with some other display method, one is not ableto judge the validity of such an image.FIGURE 2.36 A line rendering of the topography (Figures 2.35).FIGURE 2.37 A wire mesh rendering of the topography of a rubber sample (Figure 2.35). 1999 by CRC Press LLCFIGURE 2.38 Variation of the light added to a topography. (a) is the topography without any light (the same asFigure 2.35). (c) is the same data, but rendered with a light source from the right at 50 elevation. (b) has 20% lightadded, (c) 40%, (d) 60%, (e) 80%, and (f) 100%. 1999 by CRC Press LLCFigure 2.40 shows the inuence of the elevation angle on the displayed image. It is clear that anillumination from about 30 to 60 is optimal. One can combine top views or illuminated top views andwire mesh scan displays to form solid surface models of the sample surface. Such images are usually onlygenerated in the nal processing stage before publication, because they need quite a lot of computing time.Continuous shading techniques sometimes make it difcult to analyze the topography. By usingdiscontinuous color tables, one can create something similar to contour maps. An example is shown inFigure 2.41.Figure 2.42 shows a combination of the top view display and the wire mesh display, a three-dimensionalmodel where the height is coded as a shade of gray. As with the top view (Figure 2.35), one can vary theamount of illumination added to the data. Figures 2.42 through 2.47 show the inuence of the amountof illumination in the display.Data consisting of more than one channel can be packed into one three-dimensional image. Onechannel, usually the topography, is used to model the height, whereas the second channel is used toproduce the shades. Figure 2.48 shows the topography shaded with the stiffness. Figure 2.49 shows thesame data, but colored with the topography.FIGURE 2.39(ad) 1999 by CRC Press LLCAdditional information can be packed into an image by using color. Assume that an image has twoplanes of data. We can display the rst plane with shades of green and the second one with shades of redon top of each other. Where the magnitude of both planes is high, one gets an orange color; where bothare low, one gets black. But if the magnitude of one plane is larger than that of the other plane on onepixel, one gets red or green colors. This way, one can display the registry of two different quantities inthe same image.2.9.7 The Two-Dimensional Histogram MethodTo analyze quantitatively friction microscopy data, one would like to calculate local friction coefcients.Normally, one operates a friction microscope in the constant-force mode, recording the topography z(x,y)and the lateral force FL(x,y). If the normal force FN is kept constant, one can get a local friction coefcient(x,y) = FL(x,y)/FN. (x,y) is well dened, if the normal force FN is truly constant, if the calibrations ofboth forces are known, and, most crucially, if the zero points of the force scales are known and stable.FIGURE 2.39 Illuminated surface: (a) shows an illumination from the top (0) with an elevation of 50. (b) is from45 (upper right) (b) and all the other illuminations have an elevation of 50. (c) is at 90, (d) at 135, (e) at 180,(f) at 225, (g) at 270, and (h) at 315. 1999 by CRC Press LLCIn many force microscopes the scales can be calibrated rather accurately, for instance, using the procedureoutlined above, but the zero points are quite unreliable. The zeros usually do not drift very much, butit is hard to determine their exact position on the force scales.Figure 2.50 shows the two-dimensional histograms calculated with data from Figure 2.35. These his-tograms, rst used as a data reduction means in communications with satellites (Yaroslavski, 1985), showcorrelations between different data channels. In the case of friction data, the lateral force FL(x,y) and thenormal force FN(x,y) are used as an address to bins arranged in a two-dimensional array (Marti, 1993a,b;Marti et al., 1993). For every data point (x,y) the bin addressed by (FN(x,y), FL(x,y)) is incremented byone. The two-dimensional histogram then shows the statistical weight of every combination. For aconstant friction coefcient and a varying normal force, one obtains (FN(x,y), FL(x,y)) pairs located ona straight line, with the slope equal to the friction coefcient. Since the creation of a two-dimensionalhistogram is independent of the zero point values, it is not necessary to know the offsets. If differentmaterials are present on the sample surface and if they have friction coefcients of sufciently differentmagnitudes, then the two materials will fall into different groups of bins in the two-dimensional histo-gram. Displaying only points falling into one group of bins will select all the locations where one materialis present (Marti 1993a,b; Lthi et al., 1995; Meyer et al., 1996).FIGURE 2.40(ad) 1999 by CRC Press LLCAs was shown above, the two-dimensional histogram analysis is not limited to an analysis of frictionforce data. It could be applied to other problems in scanning probe microscopy, for instance, relatingspectroscopic data to the tunneling current or to determine the locations of excessive forces or tunnelingcurrents, to name a few.2.9.8 Some Common Image-Processing MethodsSince many commercial data acquisition systems use implicitly some kind of data processing, we willdescribe some of the effects of these procedures. Since the original data is commonly subject to slopeson the surface, most programs use some kind of slope correction. Figure 2.51 shows the effect of suchalgorithms. In Figure 2.51a there is the original data, without any correction. The least disturbing wayis to subtract a plane z(x,y) = axx + ayy + a0 from the data. The coefcients are determined by ttingz(x,y) to the data. As can be seen in Figure 2.51b, the main effect is to lower the right side and to lift upthe left side. A more severe option is to subtract a second order function such as z(x,y) = axxx2 + ayyy2 +axyxy + axx + ayy + a0. Again, the parameters are determined with a t. This function is appropriate forFIGURE 2.40 Inuence of the elevation angle on the displayed data in illumination mode. The data are illuminatedfrom the right. (a) has an elevation angle of 0 (from top). (b) is at 15, (c) at 30, (d) at 45, (e) at 60, (f) at 75,and (g) at 90 (parallel to the surface). 1999 by CRC Press LLCFIGURE 2.41 A discontinuous color table can generate a data display similar to contour maps. Here we have useda color table with 32 steps.FIGURE 2.42 Three-dimensional rendering of the data of Figure 2.35. The shading of the surface is purely topo-graphical. 1999 by CRC Press LLCFIGURE 2.43 Three-dimensional rendering of the data of Figure 2.35. The shading of the surface is 80% topo-graphical and 20% illumination.FIGURE 2.44 Three-dimensional rendering of the data of Figure 2.35. The shading of the surface is 60% topo-graphical and 40% illumination. 1999 by CRC Press LLCFIGURE 2.45 Three-dimensional rendering of the data of Figure 2.35. The shading of the surface is 40% topo-graphical and 60% illumination.FIGURE 2.46 Three-dimensional rendering of the data of Figure 2.35. The shading of the surface is 20% topo-graphical and 80% illumination. 1999 by CRC Press LLCFIGURE 2.47 Three-dimensional rendering of the data of Figure 2.35. The shading of the surface is purely illumi-nation.FIGURE 2.48 Three-dimensional rendering of the data of Figure 2.35. The topography channel determines theheight, whereas the stiffness channel gives the color. 1999 by CRC Press LLCalmost plane data, where the nonlinearity of the piezos caused such a distortion. It is obvious that thedata are now grossly distorted. Both algorithms have two disadvantages. They can only be used for datathat are already known and they can be calculation intensive.During data acquisition, only part of the data is known. Therefore, other algorithms which only workon single rows or columns are preferred. Figure 2.51d shows the effect of subtracting the average of arow from every point in the row. Data sets that have ne structure and no long-range variations of theslope can safely be subject to this algorithm. These data, however, are affected and change its character.Figure 2.51e shows the same procedure, but now applied to the columns. Figure 2.51f is an extension tosubtracting the data. Here not only the average value is subtracted, but also the slope of a line tted toone row. Hence, all the tilt is also taken out of the data. This is a dangerous function, since it cancompletely change the data appearance. However, most data acquisition programs use exactly this func-tion. Figure 2.51g, nally, is the same procedure applied to rows.Figure 2.51h shows the effect of an unsharp mask. This procedure rst calculates a low-pass lteredimage by averaging all the points in the neighborhood and then subtracting this value. As can be seen,the effect is a high-pass lter where all the slow variations are gone. While this technique is ideally suitedto nding out if there are short-range variations in the data.With data sets such as that in Figure 2.35 where low-lying data are almost not visible, one might betempted to apply a technique widely used in image processing: histogram equalization (Figure 2.52).This technique calculates a cumulative histogram of the z-values. The number as a function of the heightclass denes a function. The inverse of this function is then applied to the data. After this operation thehistogram has an equal number of points in the height classes. Since it is a nonlinear function, we donot recommend that it be used on AFM data.FIGURE 2.49 Three-dimensional rendering of the data of Figure 2.35. The topography channel determines theheight, whereas the adhesion channel gives the color. 1999 by CRC Press LLCAcknowledgmentsThe author thanks his colleagues and co-workers for many enlightening discussions. Special thanks goto Jaime Colchero, Michael Hipp, Eva Weilandt, Sabine Hild, Armin Rosa, Bernd Zink, Georg Krausch,Bernd Heise, Martin Pietralla, Gerd-Ingo Asbach, Roland Winkler, and Joachim Spatz. Many of theinsights have been gained while preparing lectures for very interesting students.FIGURE 2.50 Two-dimensional histograms between the different channels. (a) shows the correlation between thetopography (horizontal) and the local stiffness. (b) shows the correlation between the topography (horizontal) andthe adhesion. (c), nally, is the correlation between the local stiffness (horizontal) and the adhesion. 1999 by CRC Press LLCFIGURE 2.51(a-d) 1999 by CRC Press LLCFIGURE 2.51 The effect of different slope correction algorithms. (a) shows the original data. In (b) a plane wastted to the data and subtracted. (c) show the effect of subtracting a second-order surface from the data. (c) showsthe effect of setting the average of each row to 0. In (e) the same has been done with the columns. (f) shows theeffect of setting the average and the average slope of each row to 0. (g), nally, shows the same, but for the columns.(h) shows the effect of unsharp masking with an averaging area of about one fth of the image size. 1999 by CRC Press LLCReferencesAguilar, M., Pascual, P. J., and Santisteban, A. (1986). Scanning Tunneling Microscope Automation.IBM J. Res. Dev. 30(5), 525532.Aguilar, M., Garca, A., Pascual, P. J., Presa, J., and Santisteban, A., (1987). Computer System for ScanningTunneling Microscope Automation, Surf. Sci. 181, 191.Akamine, S. and Quate, C. F. (1992). Low Temper